sync with luatex experimental. WARNING: new format ! Version update to 0.79.2 .

[luatex.git] / source / texk / web2c / bibtex.web

% This program is copyright (C) 1985, 1988, 2010 by Oren Patashnik;
% all rights are reserved.
% 
% This program, BibTeX, is available under the same terms as
% Donald Knuth's TeX program.
% 
% (Request to implementors: The WEB system provides for alterations via
% an auxiliary file; the master file should stay intact.)
% 
% See Appendix H of the WEB manual for hints on how to install this program.

% Version 0.98f was released in March 1985.
% Version 0.98g was released in April; it removed some system dependencies
%       (introducing term_in and term_out in place of just tty, and removing
%       some nonlocal goto's) and it gave context for certain parsing errors.
% Version 0.98h was released in April; it patched a bug in the output
%       line-breaking routine that can arise with some nonstandard style files.
% Version 0.98i was released in May; its main change split up the main program
%       and some procedures to help certain compilers cope with size
%       limitations, among other things changing error and warning macros so
%       they'd produce (much) less inline code; it also redefined the class of
%       legal style-file identifiers---although this affects only the bizarre
%       ones, it makes BibTeX's error messages more coherent; and it had many
%       minor changes, including about a 15% speed-up on TOPS-20.
% Version 0.99a was released in January 1988.  Its main changes: allowed the
%       inclusion of entire .bib files (rather than just those entries
%       \cited or \nocited); made the sorting algorithm stable; eliminated
%       any case conversion for file names; allowed concatenation in database
%       fields and string definitions; handled hyphenated names properly;
%       handled accented characters properly; implemented new empty$,
%       preamble$, text.length$, text.prefix$, and warning$ built-in functions;
%       allowed a new cross-referencing feature; and made many minor fixes,
%       including about a 40% speed-up on TOPS-20.
% Version 0.99b was released in February 1988.  It changed text.length$ and
%       text.prefix$ to not count braces as text characters, and it changed
%       text.prefix$ to add any necessary matching right braces.
% Version 0.99c was released in February 1988.  It removed two begin-end pairs
%       that, for convention only, surrounded entire modules, but that elicited
%       label-related complaints from some compilers.
% Version 0.99d was released in March 2010.  It made output lines breakable
%       only at white_space (so that, for example, URLs would not be broken).
%       Other known bugs (all minor) will be fixed in a subsequent release.
% Updated bibtex.web was released on 8 December 2010.  Still version
%       0.99d; this release clarified the license.

% Please report any bugs to biblio@@tug.org

% Although considerable effort has been expended to make the BibTeX program
% correct and reliable, no warranty is implied; the author disclaims any
% obligation or liability for damages, including but not limited to
% special, indirect, or consequential damages arising out of or in
% connection with the use or performance of this software.

% This program was written by Oren Patashnik, in consultation with Leslie
% Lamport, to be used with Lamport's LaTeX document preparation system.
% Some modules were taken from Knuth's TeX and TeXware with his permission.

% Here is TeX material that gets inserted after \input webmac
\def\hang{\hangindent 3em\indent\ignorespaces}
\font\ninerm=cmr9
\let\mc=\ninerm % medium caps for names like PASCAL
\def\PASCAL{{\mc PASCAL}}
\def\ph{{\mc PASCAL-H}}
\def\<#1>{$\langle#1\rangle$}
\def\section{\mathhexbox278}

\def\(#1){} % this is used to make section names sort themselves better
\def\9#1{} % this is used for sort keys in the index via @@:sort key}{entry@@>

% Note: WEAVE will typeset an upper-case `E' in a PASCAL identifier a
% bit strangely so that the `TeX' in the name of this program is typeset
% correctly; if this becomes a problem remove these three lines to get
% normal upper-case `E's in PASCAL identifiers
\def\drop{\kern-.1667em\lower.5ex\hbox{E}\kern-.125em} % middle of TeX
\catcode`E=13 \uppercase{\def E{e}}
\def\\#1{\hbox{\let E=\drop\it#1\/\kern.05em}} % italic type for identifiers

\font\sc=cmcsc10

\def\BibTeX{{\rm B\kern-.05em{\sc i\kern-.025em b}\kern-.08em
    T\kern-.1667em\lower.7ex\hbox{E}\kern-.125emX}}

\def\LaTeX{{\rm L\kern-.36em\raise.3ex\hbox{\sc a}\kern-.15em
    T\kern-.1667em\lower.7ex\hbox{E}\kern-.125emX}}

\def\title{\BibTeX\ }
\def\today{\ifcase\month\or
  January\or February\or March\or April\or May\or June\or
  July\or August\or September\or October\or November\or December\fi
  \space\number\day, \number\year}
\def\topofcontents{\null\vfill
 \def\titlepage{F}
 \centerline{\:\titlefont The {\:\ttitlefont \BibTeX} preprocessor}
 \vskip 15pt \centerline{(Version 0.99d---\today)} \vfill}
\pageno=\contentspagenumber \advance\pageno by 1



@* Introduction.
@^documentation@>
@^space savings@>
@^system dependencies@>
@^wizard@>
@!@:BibTeX}{\BibTeX@>
@!@:BibTeX documentation}{\BibTeX\ documentation@>
@:LaTeX}{\LaTeX@>
\BibTeX\ is a preprocessor (with elements of postprocessing as
explained below) for the \LaTeX\ document-preparation system.  It
handles most of the formatting decisions required to produce a
reference list, outputting a \.{.bbl} file that a user can edit to add
any finishing touches \BibTeX\ isn't designed to handle (in practice,
such editing almost never is needed); with this file \LaTeX\ actually
produces the reference list.

Here's how \BibTeX\ works.  It takes as input (a)~an \.{.aux} file
produced by \LaTeX\ on an earlier run; (b)~a \.{.bst} file (the style
file), which specifies the general reference-list style and specifies
how to format individual entries, and which is written by a style
designer (called a wizard throughout this program) in a
special-purpose language described in the \BibTeX\ documentation---see
the file {\.{btxdoc.tex}}; and (c)~\.{.bib} file(s) constituting a
database of all reference-list entries the user might ever hope to
use.  \BibTeX\ chooses from the \.{.bib} file(s) only those entries
specified by the \.{.aux} file (that is, those given by \LaTeX's
\.{\\cite} or \.{\\nocite} commands), and creates as output a \.{.bbl}
file containing these entries together with the formatting commands
specified by the \.{.bst} file (\BibTeX\ also creates a \.{.blg} log
file, which includes any error or warning messages, but this file
isn't used by any program).  \LaTeX\ will use the \.{.bbl} file,
perhaps edited by the user, to produce the reference list.

Many modules of \BibTeX\ were taken from Knuth's \TeX\ and \TeX ware,
with his permission.  All known system-dependent modules are marked in
the index entry ``system dependencies''; Dave Fuchs helped exorcise
unwanted ones.  In addition, a few modules that can be changed to make
\BibTeX\ smaller are marked in the index entry ``space savings''.

Megathanks to Howard Trickey, for whose suggestions future users and
style writers would be eternally grateful, if only they knew.

The |banner| string defined here should be changed whenever \BibTeX\
gets modified.

@d banner=='This is BibTeX, Version 0.99d' {printed when the program starts}


@
@^system dependencies@>
Terminal output goes to the file |term_out|, while terminal input
comes from |term_in|.  On our system, these (system-dependent) files
are already opened at the beginning of the program, and have the same
real name.

@d term_out == tty
@d term_in == tty


@
@^system dependencies@>
This program uses the term |print| instead of |write| when writing on
both the |log_file| and (system-dependent) |term_out| file, and it
uses |trace_pr| when in |trace| mode, for which it writes on just the
|log_file|.  If you want to change where either set of macros writes
to, you should also change the other macros in this program for that
set; each such macro begins with |print_| or |trace_pr_|.

@d print(#) == begin write(log_file,#); write(term_out,#); end
@d print_ln(#) == begin write_ln(log_file,#); write_ln(term_out,#); end
@d print_newline == print_a_newline
                                {making this a procedure saves a little space}
@#
@d trace_pr(#) == begin write(log_file,#); end
@d trace_pr_ln(#) == begin write_ln(log_file,#); end
@d trace_pr_newline == begin write_ln(log_file); end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_a_newline;
begin
write_ln(log_file);
write_ln(term_out);
end;


@
@^debugging@>
@^statistics@>
Some of the code below is intended to be used only when diagnosing the
strange behavior that sometimes occurs when \BibTeX\ is being
installed or when system wizards are fooling around with \BibTeX\
without quite knowing what they are doing. Such code will not normally
be compiled; it is delimited by the codewords
`$|debug|\ldots|gubed|$', with apologies to people who wish to
preserve the purity of English. Similarly, there is some conditional
code delimited by `$|stat|\ldots|tats|$' that is intended only for use
when statistics are to be kept about \BibTeX's memory/cpu usage,
and there is conditional code delimited by `$|trace|\ldots|ecart|$'
that is intended to be a trace facility for use mainly when debugging
\.{.bst} files.

@d debug == @{          { remove the `|@{|' when debugging }
@d gubed == @t@>@}      { remove the `|@}|' when debugging }
@f debug == begin
@f gubed == end
@#
@d stat == @{           { remove the `|@{|' when keeping statistics }
@d tats == @t@>@}       { remove the `|@}|' when keeping statistics }
@f stat == begin
@f tats == end
@#
@d trace == @{          { remove the `|@{|' when in |trace| mode }
@d ecart == @t@>@}      { remove the `|@}|' when in |trace| mode }
@f trace == begin
@f ecart == end


@
@^system dependencies@>
We assume that |case| statements may include a
default case that applies if no matching label is found,
since most \PASCAL\ compilers have plugged this hole in the language by
incorporating some sort of default mechanism. For example, the \ph\
compiler allows `|others|:' as a default label, and other \PASCAL s allow
syntaxes like `\ignorespaces|else|\unskip' or `\\{otherwise}' or
`\\{otherwise}:', etc. The definitions of |othercases| and |endcases|
should be changed to agree with local conventions.   Note that no semicolon
appears before |endcases| in this program, so the definition of |endcases|
should include a semicolon if the compiler wants one.  (Of course, if no
default mechanism is available, the |case| statements of \BibTeX\ will have
to be laboriously extended by listing all remaining cases. People who are
stuck with such \PASCAL s have in fact done this, successfully but not
happily!)

@d othercases == others:        {default for cases not listed explicitly}
@d endcases == @+end {follows the default case in an extended |case| statement}
@f othercases == else
@f endcases == end


@
Labels are given symbolic names by the following definitions, so that
occasional |goto| statements will be meaningful.  We insert the label
`|exit|:' just before the `\ignorespaces|end|\unskip' of a procedure
in which we have used the `|return|' statement defined below (and this
is the only place `|exit|:' appears).  This label is sometimes used
for exiting loops that are set up with the |loop| construction defined
below.  Another generic label is `|loop_exit|:'; it appears
immediately after a loop.

Incidentally, this program never declares a label that isn't actually used,
because some fussy \PASCAL\ compilers will complain about redundant labels.

@d exit=10              {go here to leave a procedure}
@d loop_exit=15         {go here to leave a loop within a procedure}
@d loop1_exit=16        {the first generic label for a procedure with two}
@d loop2_exit=17        {the second}


@
@^for loops@>
And |while| we're discussing loops: This program makes into |while|
loops many that would otherwise be |for| loops because of Standard
\PASCAL\ limitations (it's a bit complicated---standard \PASCAL\
doesn't allow a global variable as the index of a |for| loop inside a
procedure; furthermore, many compilers have fairly severe limitations
on the size of a block, including the main block of the program; so
most of the code in this program occurs inside procedures, and since
for other reasons this program must use primarily global variables, it
doesn't use many |for| loops).


@
@^program conventions@>
This program uses this convention: If there are several quantities in
a boolean expression, they are ordered by expected frequency (except
perhaps when an error message results) so that execution will be
fastest; this is more an attempt to understand the program than to
make it faster.


@
Here are some macros for common programming idioms.

@d incr(#) == #:=#+1    {increase a variable by unity}
@d decr(#) == #:=#-1    {decrease a variable by unity}
@d loop == @+ while true do@+   {repeat over and over until a |goto| happens}
@f loop == xclause
  {\.{WEB}'s |xclause| acts like `\ignorespaces|while true do|\unskip'}
@d do_nothing ==        {empty statement}
@d return == goto exit  {terminate a procedure call}
@f return == nil
@d empty=0              {symbolic name for a null constant}
@d any_value=0          {this appeases \PASCAL's boolean-evaluation scheme}



@* The main program.
@^system dependencies@>
@:LaTeX}{\LaTeX@>
This program first reads the \.{.aux} file that \LaTeX\ produces,
(\romannumeral1) determining which \.{.bib} file(s) and \.{.bst} file
to read and (\romannumeral2) constructing a list of cite keys in order
of occurrence.  The \.{.aux} file may have other \.{.aux} files nested
within.  Second, it reads and executes the \.{.bst} file,
(\romannumeral1) determining how and in which order to process the
database entries in the \.{.bib} file(s) corresponding to those cite
keys in the list (or in some cases, to all the entries in the \.{.bib}
file(s)), (\romannumeral2) determining what text to be output for each
entry and determining any additional text to be output, and
(\romannumeral3) actually outputting this text to the \.{.bbl} file.
In addition, the program sends error messages and other remarks to the
|log_file| and terminal.

@d close_up_shop=9998           {jump here after fatal errors}
@d exit_program=9999            {jump here if we couldn't even get started}

@p
@t\4@>@<Compiler directives@>@/
program BibTEX;                 {all files are opened dynamically}
label   close_up_shop,@!exit_program @<Labels in the outer block@>;
const   @<Constants in the outer block@>
type    @<Types in the outer block@>
var     @<Globals in the outer block@>@;
@<Procedures and functions for about everything@>@;
@<The procedure |initialize|@>
@#
begin
initialize;
print_ln(banner);@/
@<Read the \.{.aux} file@>;
@<Read and execute the \.{.bst} file@>;
close_up_shop:
@<Clean up and leave@>;
exit_program:
end.


@
@^overflow in arithmetic@>
@^system dependencies@>
If the first character of a \PASCAL\ comment is a dollar sign,
\ph\ treats the comment as a list of ``compiler directives'' that will
affect the translation of this program into machine language.  The
directives shown below specify full checking and inclusion of the \PASCAL\
debugger when \BibTeX\ is being debugged,
but they cause range checking and other
redundant code to be eliminated when the production system is being generated.
Arithmetic overflow will be detected in all cases.

@<Compiler directives@>=
@{@&$C-,A+,D-@}  {no range check, catch arithmetic overflow, no debug overhead}
@!debug @{@&$C+,D+@}@+ gubed            {but turn everything on when debugging}


@
@^bottom up@>
@^gymnastics@>
@^mooning@>
All procedures in this program (except for |initialize|) are grouped
into one of the seven classes below, and these classes are dispersed
throughout the program.  However: Much of this program is written top
down, yet \PASCAL\ wants its procedures bottom up.  Since mooning is
neither a technically nor a socially acceptable solution to the
bottom-up problem, this section instead performs the topological
gymnastics that \.{WEB} allows, ordering these classes to satisfy
\PASCAL\ compilers.  There are a few procedures still out of place
after this ordering, though, and the other modules that complete the
task have ``gymnastics'' as an index entry.

@<Procedures and functions for about everything@>=
@<Procedures and functions for all file I/O, error messages, and such@>@;
@<Procedures and functions for file-system interacting@>@;
@<Procedures and functions for handling numbers, characters, and strings@>@;
@<Procedures and functions for input scanning@>@;
@<Procedures and functions for name-string processing@>@;
@<Procedures and functions for style-file function execution@>@;
@<Procedures and functions for the reading and processing of input files@>


@
This procedure gets things started properly.

@<The procedure |initialize|@>=
procedure initialize;
var @<Local variables for initialization@>
begin
@<Check the ``constant'' values for consistency@>;
if (bad > 0) then
    begin
    write_ln (term_out,bad:0,' is a bad bad');
    goto exit_program;
    end;
@<Set initial values of key variables@>;
pre_def_certain_strings;@/
get_the_top_level_aux_file_name;
end;


@
@^space savings@>
@^system dependencies@>
These parameters can be changed at compile time to extend or reduce
\BibTeX's capacity.  They are set to accommodate about 750 cites when
used with the standard styles, although |pool_size| is usually the
first limitation to be a problem, often when there are 500 cites.

@<Constants in the outer block@>=
@!buf_size=1000; {maximum number of characters in an input line (or string)}
@!min_print_line=3; {minimum \.{.bbl} line length: must be |>=3|}
@!max_print_line=79; {the maximum: must be |>min_print_line| and |<buf_size|}
@!aux_stack_size=20; {maximum number of simultaneous open \.{.aux} files}
@!max_bib_files=20; {maximum number of \.{.bib} files allowed}
@!pool_size=65000; {maximum number of characters in strings}
@!max_strings=4000; {maximum number of strings, including pre-defined;
                                                        must be |<=hash_size|}
@!max_cites=750; {maximum number of distinct cite keys; must be
                                                        |<=max_strings|}
@!min_crossrefs=2; {minimum number of cross-refs required for automatic
                                                        |cite_list| inclusion}
@!wiz_fn_space=3000; {maximum amount of |wiz_defined|-function space}
@!single_fn_space=100; {maximum amount for a single |wiz_defined|-function}
@!max_ent_ints=3000; {maximum number of |int_entry_var|s
                                        (entries $\times$ |int_entry_var|s)}
@!max_ent_strs=3000; {maximum number of |str_entry_var|s
                                        (entries $\times$ |str_entry_var|s)}
@!ent_str_size=100; {maximum size of a |str_entry_var|; must be |<=buf_size|}
@!glob_str_size=1000; {maximum size of a |str_global_var|;
                                                        must be |<=buf_size|}
@!max_fields=17250; {maximum number of fields (entries $\times$ fields,
                                        about |23*max_cites| for consistency)}
@!lit_stk_size=100; {maximum number of literal functions on the stack}


@
@^space savings@>
@^system dependencies@>
These parameters can also be changed at compile time, but they're
needed to define some \.{WEB} numeric macros so they must be so
defined themselves.

@d hash_size=5000       {must be |>= max_strings| and |>= hash_prime|}
@d hash_prime=4253      {a prime number about 85\% of |hash_size| and |>= 128|
                                                and |< @t$2^{14}-2^6$@>|}
@d file_name_size=40    {file names shouldn't be longer than this}
@d max_glob_strs=10     {maximum number of |str_global_var| names}
@d max_glb_str_minus_1 = max_glob_strs-1  {to avoid wasting a |str_global_var|}


@
In case somebody has inadvertently made bad settings of the ``constants,''
\BibTeX\ checks them using a global variable called |bad|.

This is the first of many sections of \BibTeX\ where global variables are
defined.

@<Globals in the outer block@>=
@!bad:integer;          {is some ``constant'' wrong?}


@
Each digit-value of |bad| has a specific meaning.

@<Check the ``constant'' values for consistency@>=
bad := 0;
if (min_print_line < 3) then                    bad:=1;
if (max_print_line <= min_print_line) then      bad:=10*bad+2;
if (max_print_line >= buf_size) then            bad:=10*bad+3;
if (hash_prime < 128) then                      bad:=10*bad+4;
if (hash_prime > hash_size) then                bad:=10*bad+5;
if (hash_prime >= (16384-64)) then              bad:=10*bad+6;
if (max_strings > hash_size) then               bad:=10*bad+7;
if (max_cites > max_strings) then               bad:=10*bad+8;
if (ent_str_size > buf_size) then               bad:=10*bad+9;
if (glob_str_size > buf_size) then              bad:=100*bad+11;
                                                        {well, almost each}


@
A global variable called |history| will contain one of four values at
the end of every run: |spotless| means that no unusual messages were
printed; |warning_message| means that a message of possible interest
was printed but no serious errors were detected; |error_message| means
that at least one error was found; |fatal_message| means that the
program terminated abnormally. The value of |history| does not
influence the behavior of the program; it is simply computed for the
convenience of systems that might want to use such information.

@d spotless=0           {|history| value for normal jobs}
@d warning_message=1    {|history| value when non-serious info was printed}
@d error_message=2      {|history| value when an error was noted}
@d fatal_message=3      {|history| value when we had to stop prematurely}

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure mark_warning;
begin
if (history = warning_message) then
    incr(err_count)
  else if (history = spotless) then
    begin
    history := warning_message;
    err_count := 1;
    end;
end;
@#
procedure mark_error;
begin
if (history < error_message) then
    begin
    history := error_message;
    err_count := 1;
    end
  else  {|history = error_message|}
    incr(err_count);
end;
@#
procedure mark_fatal;
begin
history := fatal_message;
end;


@
For the two states |warning_message| and |error_message| we keep track
of the number of messages given; but since |warning_message|s aren't
so serious, we ignore them once we've seen an |error_message|.  Hence
we need just the single variable |err_count| to keep track.


@<Globals in the outer block@>=
@!history:spotless..fatal_message; {how bad was this run?}
@!err_count:integer;


@
The |err_count| gets set or reset when |history| first changes to
|warning_message| or |error_message|, so we don't need to initialize
it.

@<Set initial values of key variables@>=
history := spotless;



@* The character set.
@^ASCII code@>
(The following material is copied (almost) verbatim from \TeX.
Thus, the same system-dependent changes should be made to both programs.)

In order to make \TeX\ readily portable between a wide variety of
computers, all of its input text is converted to an internal seven-bit
code that is essentially standard ASCII, the ``American Standard Code for
Information Interchange.''  This conversion is done immediately when each
character is read in. Conversely, characters are converted from ASCII to
the user's external representation just before they are output to a
text file.

Such an internal code is relevant to users of \TeX\ primarily because it
governs the positions of characters in the fonts. For example, the
character `\.A' has ASCII code $65=@'101$, and when \TeX\ typesets
this letter it specifies character number 65 in the current font.
If that font actually has `\.A' in a different position, \TeX\ doesn't
know what the real position is; the program that does the actual printing from
\TeX's device-independent files is responsible for converting from ASCII to
a particular font encoding.

\TeX's internal code is relevant also with respect to constants
that begin with a reverse apostrophe.


@
Characters of text that have been converted to \TeX's internal form
are said to be of type |ASCII_code|, which is a subrange of the integers.

@<Types in the outer block@>=
@!ASCII_code=0..127;    {seven-bit numbers}


@
@^character set dependencies@>
@^system dependencies@>
The original \PASCAL\ compiler was designed in the late 60s, when six-bit
character sets were common, so it did not make provision for lower-case
letters. Nowadays, of course, we need to deal with both capital and small
letters in a convenient way, especially in a program for typesetting;
so the present specification of \TeX\ has been written under the assumption
that the \PASCAL\ compiler and run-time system permit the use of text files
with more than 64 distinguishable characters. More precisely, we assume that
the character set contains at least the letters and symbols associated
with ASCII codes @'40 through @'176; all of these characters are now
available on most computer terminals.

Since we are dealing with more characters than were present in the first
\PASCAL\ compilers, we have to decide what to call the associated data
type. Some \PASCAL s use the original name |char| for the
characters in text files, even though there now are more than 64 such
characters, while other \PASCAL s consider |char| to be a 64-element
subrange of a larger data type that has some other name.

In order to accommodate this difference, we shall use the name |text_char|
to stand for the data type of the characters that are converted to and
from |ASCII_code| when they are input and output. We shall also assume
that |text_char| consists of the elements |chr(first_text_char)| through
|chr(last_text_char)|, inclusive. The following definitions should be
adjusted if necessary.

@d text_char == char    {the data type of characters in text files}
@d first_text_char=0    {ordinal number of the smallest element of |text_char|}
@d last_text_char=127   {ordinal number of the largest element of |text_char|}

@<Local variables for initialization@>=
i:0..last_text_char;    {this is the first one declared}


@
The \TeX\ processor converts between ASCII code and
the user's external character set by means of arrays |xord| and |xchr|
that are analogous to \PASCAL's |ord| and |chr| functions.

@<Globals in the outer block@>=
@!xord: array [text_char] of ASCII_code;
  {specifies conversion of input characters}
@!xchr: array [ASCII_code] of text_char;
  {specifies conversion of output characters}


@
@^character set dependencies@>
@^system dependencies@>
Since we are assuming that our \PASCAL\ system is able to read and write the
visible characters of standard ASCII (although not necessarily using the
ASCII codes to represent them), the following assignment statements initialize
most of the |xchr| array properly, without needing any system-dependent
changes. On the other hand, it is possible to implement \TeX\ with
less complete character sets, and in such cases it will be necessary to
change something here.

@<Set initial values of key variables@>=
xchr[@'40]:=' ';
xchr[@'41]:='!';
xchr[@'42]:='"';
xchr[@'43]:='#';
xchr[@'44]:='$';
xchr[@'45]:='%';
xchr[@'46]:='&';
xchr[@'47]:='''';@/
xchr[@'50]:='(';
xchr[@'51]:=')';
xchr[@'52]:='*';
xchr[@'53]:='+';
xchr[@'54]:=',';
xchr[@'55]:='-';
xchr[@'56]:='.';
xchr[@'57]:='/';@/
xchr[@'60]:='0';
xchr[@'61]:='1';
xchr[@'62]:='2';
xchr[@'63]:='3';
xchr[@'64]:='4';
xchr[@'65]:='5';
xchr[@'66]:='6';
xchr[@'67]:='7';@/
xchr[@'70]:='8';
xchr[@'71]:='9';
xchr[@'72]:=':';
xchr[@'73]:=';';
xchr[@'74]:='<';
xchr[@'75]:='=';
xchr[@'76]:='>';
xchr[@'77]:='?';@/
xchr[@'100]:='@@';
xchr[@'101]:='A';
xchr[@'102]:='B';
xchr[@'103]:='C';
xchr[@'104]:='D';
xchr[@'105]:='E';
xchr[@'106]:='F';
xchr[@'107]:='G';@/
xchr[@'110]:='H';
xchr[@'111]:='I';
xchr[@'112]:='J';
xchr[@'113]:='K';
xchr[@'114]:='L';
xchr[@'115]:='M';
xchr[@'116]:='N';
xchr[@'117]:='O';@/
xchr[@'120]:='P';
xchr[@'121]:='Q';
xchr[@'122]:='R';
xchr[@'123]:='S';
xchr[@'124]:='T';
xchr[@'125]:='U';
xchr[@'126]:='V';
xchr[@'127]:='W';@/
xchr[@'130]:='X';
xchr[@'131]:='Y';
xchr[@'132]:='Z';
xchr[@'133]:='[';
xchr[@'134]:='\';
xchr[@'135]:=']';
xchr[@'136]:='^';
xchr[@'137]:='_';@/
xchr[@'140]:='`';
xchr[@'141]:='a';
xchr[@'142]:='b';
xchr[@'143]:='c';
xchr[@'144]:='d';
xchr[@'145]:='e';
xchr[@'146]:='f';
xchr[@'147]:='g';@/
xchr[@'150]:='h';
xchr[@'151]:='i';
xchr[@'152]:='j';
xchr[@'153]:='k';
xchr[@'154]:='l';
xchr[@'155]:='m';
xchr[@'156]:='n';
xchr[@'157]:='o';@/
xchr[@'160]:='p';
xchr[@'161]:='q';
xchr[@'162]:='r';
xchr[@'163]:='s';
xchr[@'164]:='t';
xchr[@'165]:='u';
xchr[@'166]:='v';
xchr[@'167]:='w';@/
xchr[@'170]:='x';
xchr[@'171]:='y';
xchr[@'172]:='z';
xchr[@'173]:='{';
xchr[@'174]:='|';
xchr[@'175]:='}';
xchr[@'176]:='~';@/
xchr[0]:=' '; xchr[@'177]:=' ';
  {ASCII codes 0 and |@'177| do not appear in text}


@
@^character set dependencies@>
@^system dependencies@>
Some of the ASCII codes without visible characters have been given symbolic
names in this program because they are used with a special meaning.  The
|tab| character may be system dependent.

@d null_code=@'0        {ASCII code that might disappear}
@d tab=@'11             {ASCII code treated as |white_space|}
@d space=@'40           {ASCII code treated as |white_space|}
@d invalid_code=@'177   {ASCII code that should not appear}


@
@^character set dependencies@>
@^system dependencies@>
@:TeXbook}{\sl The \TeX book@>
The ASCII code is ``standard'' only to a certain extent, since many
computer installations have found it advantageous to have ready access
to more than 94 printing characters. Appendix~C of {\sl The \TeX book\/}
gives a complete specification of the intended correspondence between
characters and \TeX's internal representation.

If \TeX\ is being used
on a garden-variety \PASCAL\ for which only standard ASCII
codes will appear in the input and output files, it doesn't really matter
what codes are specified in |xchr[1..@'37]|, but the safest policy is to
blank everything out by using the code shown below.

However, other settings of |xchr| will make \TeX\ more friendly on
computers that have an extended character set, so that users can type things
like `\.^^Z' instead of `\.{\\ne}'. At MIT, for example, it would be more
appropriate to substitute the code
$$\hbox{|for i:=1 to @'37 do xchr[i]:=chr(i);|}$$
\TeX's character set is essentially the same as MIT's, even with respect to
characters less than~@'40. People with extended character sets can
assign codes arbitrarily, giving an |xchr| equivalent to whatever
characters the users of \TeX\ are allowed to have in their input files.
It is best to make the codes correspond to the intended interpretations as
shown in Appendix~C whenever possible; but this is not necessary. For
example, in countries with an alphabet of more than 26 letters, it is
usually best to map the additional letters into codes less than~@'40.

@<Set initial values of key variables@>=
for i:=1 to @'37 do xchr[i]:=' ';
xchr[tab]:=chr(tab);


@
This system-independent code makes the |xord| array contain a suitable
inverse to the information in |xchr|. Note that if |xchr[i]=xchr[j]|
where |i<j<@'177|, the value of |xord[xchr[i]]| will turn out to be
|j| or more; hence, standard ASCII code numbers will be used instead
of codes below @'40 in case there is a coincidence.

@<Set initial values of key variables@>=
for i:=first_text_char to last_text_char do xord[chr(i)]:=invalid_code;
for i:=1 to @'176 do xord[xchr[i]]:=i;


@
Also, various characters are given symbolic names; all the ones this
program uses are collected here.  We use the sharp sign as the
|concat_char|, rather than something more natural (like an ampersand),
for uniformity of database syntax (ampersand is a valid character in
identifiers).

@d double_quote = """"          {delimits strings}
@d number_sign = "#"            {marks an |int_literal|}
@d comment = "%"                {ignore the rest of a \.{.bst} or \TeX\ line}
@d single_quote = "'"           {marks a quoted function}
@d left_paren = "("             {optional database entry left delimiter}
@d right_paren = ")"            {corresponding right delimiter}
@d comma = ","                  {separates various things}
@d minus_sign = "-"             {for a negative number}
@d equals_sign = "="            {separates a field name from a field value}
@d at_sign = "@@"               {the beginning of a database entry}
@d left_brace = "{"             {left delimiter of many things}
@d right_brace = "}"            {corresponding right delimiter}
@d period = "."                 {these are three}
@d question_mark = "?"          {string-ending characters}
@d exclamation_mark = "!"       {of interest in \.{add.period\$}}
@d tie = "~"                    {the default space char, in \.{format.name\$}}
@d hyphen = "-"                 {like |white_space|, in \.{format.name\$}}
@d star = "*"                   {for including entire database}
@d concat_char = "#"            {for concatenating field tokens}
@d colon = ":"                  {for lower-casing (usually title) strings}
@d backslash = "\"              {used to recognize accented characters}


@
These arrays give a lexical classification for the |ASCII_code|s;
|lex_class| is used for general scanning and |id_class| is used for
scanning identifiers.

@<Globals in the outer block@>=
@!lex_class: array [ASCII_code] of lex_type;
@!id_class: array [ASCII_code] of id_type;


@
Every character has two types of the lexical classifications.  The
first type is general, and the second type tells whether the character
is legal in identifiers.

@d illegal = 0          {the unrecognized |ASCII_code|s}
@d white_space = 1      {things like |space|s that you can't see}
@d alpha = 2            {the upper- and lower-case letters}
@d numeric = 3          {the ten digits}
@d sep_char = 4         {things sometimes treated like |white_space|}
@d other_lex = 5        {when none of the above applies}
@d last_lex = 5         {the same number as on the line above}
@#
@d illegal_id_char = 0  {a few forbidden ones}
@d legal_id_char = 1    {most printing characters}

@<Types in the outer block@>=
@!lex_type = 0..last_lex;@/
@!id_type = 0..1;


@
@^character set dependencies@>
@^system dependencies@>
Now we initialize the system-dependent |lex_class| array.  The |tab|
character may be system dependent.  Note that the order of these
assignments is important here.

@<Set initial values of key variables@>=
for i:=0 to @'177 do lex_class[i] := other_lex;
for i:=0 to @'37 do lex_class[i] := illegal;
lex_class[invalid_code] := illegal;
lex_class[tab] := white_space;
lex_class[space] := white_space;
lex_class[tie] := sep_char;
lex_class[hyphen] := sep_char;
for i:=@'60 to @'71 do lex_class[i] := numeric;
for i:=@'101 to @'132 do lex_class[i] := alpha;
for i:=@'141 to @'172 do lex_class[i] := alpha;


@
@^character set dependencies@>
@^system dependencies@>
And now the |id_class| array.

@<Set initial values of key variables@>=
for i:=0 to @'177 do id_class[i] := legal_id_char;
for i:=0 to @'37 do id_class[i] := illegal_id_char;
id_class[space] := illegal_id_char;
id_class[tab] := illegal_id_char;
id_class[double_quote] := illegal_id_char;
id_class[number_sign] := illegal_id_char;
id_class[comment] := illegal_id_char;
id_class[single_quote] := illegal_id_char;
id_class[left_paren] := illegal_id_char;
id_class[right_paren] := illegal_id_char;
id_class[comma] := illegal_id_char;
id_class[equals_sign] := illegal_id_char;
id_class[left_brace] := illegal_id_char;
id_class[right_brace] := illegal_id_char;


@
The array |char_width| gives relative printing widths of each
|ASCII_code|, and |string_width| will be used later to sum up
|char_width|s in a string.

@<Globals in the outer block@>=
@!char_width : array [ASCII_code] of integer;
@!string_width : integer;


@
@^character set dependencies@>
@^system dependencies@>
Now we initialize the system-dependent |char_width| array, for which
|space| is the only |white_space| character given a nonzero printing
width.  The widths here are taken from Stanford's June~'87
$cmr10$~font and represent hundredths of a point (rounded), but since
they're used only for relative comparisons, the units have no meaning.

@d ss_width = 500               {character |@'31|'s width in the $cmr10$ font}
@d ae_width = 722               {character |@'32|'s width in the $cmr10$ font}
@d oe_width = 778               {character |@'33|'s width in the $cmr10$ font}
@d upper_ae_width = 903         {character |@'35|'s width in the $cmr10$ font}
@d upper_oe_width = 1014        {character |@'36|'s width in the $cmr10$ font}

@<Set initial values of key variables@>=
for i:=0 to @'177 do char_width[i] := 0;
@#
char_width[@'40] := 278;
char_width[@'41] := 278;
char_width[@'42] := 500;
char_width[@'43] := 833;
char_width[@'44] := 500;
char_width[@'45] := 833;
char_width[@'46] := 778;
char_width[@'47] := 278;
char_width[@'50] := 389;
char_width[@'51] := 389;
char_width[@'52] := 500;
char_width[@'53] := 778;
char_width[@'54] := 278;
char_width[@'55] := 333;
char_width[@'56] := 278;
char_width[@'57] := 500;
char_width[@'60] := 500;
char_width[@'61] := 500;
char_width[@'62] := 500;
char_width[@'63] := 500;
char_width[@'64] := 500;
char_width[@'65] := 500;
char_width[@'66] := 500;
char_width[@'67] := 500;
char_width[@'70] := 500;
char_width[@'71] := 500;
char_width[@'72] := 278;
char_width[@'73] := 278;
char_width[@'74] := 278;
char_width[@'75] := 778;
char_width[@'76] := 472;
char_width[@'77] := 472;
char_width[@'100] := 778;
char_width[@'101] := 750;
char_width[@'102] := 708;
char_width[@'103] := 722;
char_width[@'104] := 764;
char_width[@'105] := 681;
char_width[@'106] := 653;
char_width[@'107] := 785;
char_width[@'110] := 750;
char_width[@'111] := 361;
char_width[@'112] := 514;
char_width[@'113] := 778;
char_width[@'114] := 625;
char_width[@'115] := 917;
char_width[@'116] := 750;
char_width[@'117] := 778;
char_width[@'120] := 681;
char_width[@'121] := 778;
char_width[@'122] := 736;
char_width[@'123] := 556;
char_width[@'124] := 722;
char_width[@'125] := 750;
char_width[@'126] := 750;
char_width[@'127] :=1028;
char_width[@'130] := 750;
char_width[@'131] := 750;
char_width[@'132] := 611;
char_width[@'133] := 278;
char_width[@'134] := 500;
char_width[@'135] := 278;
char_width[@'136] := 500;
char_width[@'137] := 278;
char_width[@'140] := 278;
char_width[@'141] := 500;
char_width[@'142] := 556;
char_width[@'143] := 444;
char_width[@'144] := 556;
char_width[@'145] := 444;
char_width[@'146] := 306;
char_width[@'147] := 500;
char_width[@'150] := 556;
char_width[@'151] := 278;
char_width[@'152] := 306;
char_width[@'153] := 528;
char_width[@'154] := 278;
char_width[@'155] := 833;
char_width[@'156] := 556;
char_width[@'157] := 500;
char_width[@'160] := 556;
char_width[@'161] := 528;
char_width[@'162] := 392;
char_width[@'163] := 394;
char_width[@'164] := 389;
char_width[@'165] := 556;
char_width[@'166] := 528;
char_width[@'167] := 722;
char_width[@'170] := 528;
char_width[@'171] := 528;
char_width[@'172] := 444;
char_width[@'173] := 500;
char_width[@'174] :=1000;
char_width[@'175] := 500;
char_width[@'176] := 500;



@* Input and output.
The basic operations we need to do are
(1)~inputting and outputting of text characters to or from a file;
(2)~instructing the operating system to initiate (``open'')
or to terminate (``close'') input or output to or from a specified file; and
(3)~testing whether the end of an input file has been reached.

@<Types in the outer block@>=
@!alpha_file=packed file of text_char;  {files that contain textual data}


@
@^system dependencies@>
Most of what we need to do with respect to input and output can be handled
by the I/O facilities that are standard in \PASCAL, i.e., the routines
called |get|, |put|, |eof|, and so on. But
standard \PASCAL\ does not allow file variables to be associated with file
names that are determined at run time, so it cannot be used to implement
\BibTeX; some sort of extension to \PASCAL's ordinary |reset| and |rewrite|
is crucial for our purposes. We shall assume that |name_of_file| is a variable
of an appropriate type such that the \PASCAL\ run-time system being used to
implement \BibTeX\ can open a file whose external name is specified by
|name_of_file|. \BibTeX\ does no case conversion for file names.

@<Globals in the outer block@>=
@!name_of_file:packed array[1..file_name_size] of char;
                         {on some systems this is a \&{record} variable}
@!name_length:0..file_name_size;
  {this many characters are relevant in |name_of_file| (the rest are blank)}
@!name_ptr:0..file_name_size+1;         {index variable into |name_of_file|}


@
@^system dependencies@>
@:PASCAL H}{\ph@>
The \ph\ compiler with which the present version of \TeX\ was prepared has
extended the rules of \PASCAL\ in a very convenient way. To open file~|f|,
we can write
$$\vbox{\halign{#\hfil\qquad&#\hfil\cr
|reset(f,@t\\{name}@>,'/O')|&for input;\cr
|rewrite(f,@t\\{name}@>,'/O')|&for output.\cr}}$$
The `\\{name}' parameter, which is of type `\ignorespaces|packed
array[@t\<\\{any}>@>] of text_char|', stands for the name of
the external file that is being opened for input or output.
Blank spaces that might appear in \\{name} are ignored.

The `\.{/O}' parameter tells the operating system not to issue its own
error messages if something goes wrong. If a file of the specified name
cannot be found, or if such a file cannot be opened for some other reason
(e.g., someone may already be trying to write the same file), we will have
|@!erstat(f)<>0| after an unsuccessful |reset| or |rewrite|.  This allows
\TeX\ to undertake appropriate corrective action.

\TeX's file-opening procedures return |false| if no file identified by
|name_of_file| could be opened.

@d reset_OK(#)==erstat(#)=0
@d rewrite_OK(#)==erstat(#)=0

@<Procedures and functions for file-system interacting@>=
function erstat(var f:file):integer; extern;    {in the runtime library}
@#@t\2@>
function a_open_in(var f:alpha_file):boolean;   {open a text file for input}
begin reset(f,name_of_file,'/O'); a_open_in:=reset_OK(f);
end;
@#
function a_open_out(var f:alpha_file):boolean;  {open a text file for output}
begin rewrite(f,name_of_file,'/O'); a_open_out:=rewrite_OK(f);
end;


@
@^system dependencies@>
Files can be closed with the \ph\ routine `|close(f)|', which should
be used when all input or output with respect to |f| has been
completed.  This makes |f| available to be opened again, if desired;
and if |f| was used for output, the |close| operation makes the
corresponding external file appear on the user's area, ready to be
read.

@<Procedures and functions for file-system interacting@>=
procedure a_close(var f:alpha_file);            {close a text file}
begin close(f);
end;


@
Text output is easy to do with the ordinary \PASCAL\ |put| procedure,
so we don't have to make any other special arrangements.
The treatment of text input is more difficult, however, because
of the necessary translation to |ASCII_code| values, and because
\TeX's conventions should be efficient and they should
blend nicely with the user's operating environment.


@
Input from text files is read one line at a time, using a routine
called |input_ln|. This function is defined in terms of global
variables called |buffer| and |last|.  The |buffer| array contains
|ASCII_code| values, and |last| is an index into this array marking
the end of a line of text.  (Occasionally, |buffer| is used for
something else, in which case it is copied to a temporary array.)

@<Globals in the outer block@>=
@!buffer:buf_type;      {usually, lines of characters being read}
@!last:buf_pointer;     {end of the line just input to |buffer|}


@
@^save space@>
@^space savings@>
@^system dependencies@>
The type |buf_type| is used for |buffer|, for saved copies of it, or
for scratch work.  It's not |packed| because otherwise the program
would run much slower on some systems (more than 25 percent slower,
for example, on a TOPS-20 operating system).  But on systems that are
byte-addressable and that have a good compiler, packing |buf_type|
would save lots of space without much loss of speed.  Other modules
that have packable arrays are also marked with a ``space savings''
index entry.

@<Types in the outer block@>=
@!buf_pointer = 0..buf_size;                    {an index into a |buf_type|}
@!buf_type = array[buf_pointer] of ASCII_code;  {for various buffers}


@
@^kludge@>
And while we're at it, we declare another buffer for general use.
Because buffers are not packed and can get large, we use |sv_buffer|
several purposes; this is a bit kludgy, but it helps make the stack
space not overflow on some machines.  It's used when reading the
entire database file (in the \.{read} command) and when doing
name-handling (through the alias |name_buf|) in the |built_in|
functions \.{format.names\$} and \.{num.names\$}.

@<Globals in the outer block@>=
@!sv_buffer : buf_type;
@!sv_ptr1 : buf_pointer;
@!sv_ptr2 : buf_pointer;
@!tmp_ptr,@!tmp_end_ptr : integer; {copy pointers only, usually for buffers}


@
@.BibTeX capacity exceeded@>
When something in the program wants to be bigger or something out
there wants to be smaller, it's time to call it a run.  Here's the
first of several macros that have associated procedures so that they
produce less inline code.

@d overflow(#)==begin           {fatal error---close up shop}
                print_overflow;
                print_ln(#:0);
                goto close_up_shop;
                end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_overflow;
begin
print ('Sorry---you''ve exceeded BibTeX''s ');
mark_fatal;
end;


@
@.this can't happen@>
When something happens that the program thinks is impossible,
call the maintainer.

@d confusion(#)==begin          {fatal error---close up shop}
                 print (#);
                 print_confusion;
                 goto close_up_shop;
                 end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_confusion;
begin
print_ln ('---this can''t happen');
print_ln ('*Please notify the BibTeX maintainer*');
mark_fatal;
end;


@
@:BibTeX capacity exceeded}{\quad buffer size@>
When a buffer overflows, it's time to complain (and then quit).

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure buffer_overflow;
begin
overflow('buffer size ',buf_size);
end;


@
@:BibTeX capacity exceeded}{\quad buffer size@>
The |input_ln| function brings the next line of input from the
specified file into available positions of the buffer array and
returns the value |true|, unless the file has already been entirely
read, in which case it returns |false| and sets |last:=0|.  In
general, the |ASCII_code| numbers that represent the next line of the
file are input into |buffer[0]|, |buffer[1]|, \dots, |buffer[last-1]|;
and the global variable |last| is set equal to the length of the line.
Trailing |white_space| characters are removed from the line
(|white_space| characters are explained in the character-set section%
---most likely they're blanks); thus, either |last=0| (in which case
the line was entirely blank) or |lex_class[buffer[last-1]]<>white_space|.
An overflow error is given if the normal actions of |input_ln| would
make |last>buf_size|.

Standard \PASCAL\ says that a file should have |eoln| immediately
before |eof|, but \BibTeX\ needs only a weaker restriction: If |eof|
occurs in the middle of a line, the system function |eoln| should return
a |true| result (even though |f^| will be undefined).

@<Procedures and functions for all file I/O, error messages, and such@>=
function input_ln(var f:alpha_file) : boolean;
                                {inputs the next line or returns |false|}
label loop_exit;
begin
last:=0;
if (eof(f)) then input_ln:=false
else
  begin
  while (not eoln(f)) do
    begin
    if (last >= buf_size) then
        buffer_overflow;
    buffer[last]:=xord[f^];
    get(f); incr(last);
    end;
  get(f);
  while (last > 0) do           {remove trailing |white_space|}
    if (lex_class[buffer[last-1]] = white_space) then
      decr(last)
     else
      goto loop_exit;
loop_exit:
  input_ln:=true;
  end;
end;



@* String handling.
\BibTeX\ uses variable-length strings of seven-bit characters.
Since \PASCAL\ does not have a well-developed string mechanism,
\BibTeX\ does all its string processing by home-grown
(predominantly \TeX's) methods.
Unlike \TeX, however, \BibTeX\ does not use a |pool_file| for
string storage; it creates its few pre-defined strings at run-time.

The necessary operations are handled with a simple data structure.
The array |str_pool| contains all the (seven-bit) ASCII codes in all
the strings \BibTeX\ must ever search for (generally identifiers
names), and the array |str_start| contains indices of the starting
points of each such string. Strings are referred to by integer
numbers, so that string number |s| comprises the characters
|str_pool[j]| for |str_start[s]<=j<str_start[s+1]|. Additional integer
variables |pool_ptr| and |str_ptr| indicate the number of entries used
so far in |str_pool| and |str_start|; locations |str_pool[pool_ptr]|
and |str_start[str_ptr]| are ready for the next string to be
allocated.  Location |str_start[0]| is unused so that hashing will
work correctly.

Elements of the |str_pool| array must be ASCII codes that can actually be
printed; i.e., they must have an |xchr| equivalent in the local
character set.

@<Globals in the outer block@>=
@!str_pool : packed array[pool_pointer] of ASCII_code;  {the characters}
@!str_start : packed array[str_number] of pool_pointer; {the starting pointers}
@!pool_ptr : pool_pointer;      {first unused position in |str_pool|}
@!str_ptr : str_number;         {start of the current string being created}
@!str_num : str_number;         {general index variable into |str_start|}
@!p_ptr1,@!p_ptr2 : pool_pointer;       {several procedures use these locally}


@
Where |pool_pointer| and |str_number| are pointers into |str_pool| and
|str_start|.

@<Types in the outer block@>=
@!pool_pointer = 0..pool_size;  {for variables that point into |str_pool|}
@!str_number = 0..max_strings;  {for variables that point into |str_start|}


@
These macros send a string in |str_pool| to an output file.

@d max_pop = 3  {---see the |built_in| functions section}
@#
@d print_pool_str(#) == print_a_pool_str(#)
                                {making this a procedure saves a little space}
@#
@d trace_pr_pool_str(#) == begin
                           out_pool_str(log_file,#);
                           end


@
@^kludge@>
@^system dependencies@>
@:this can't happen}{\quad Illegal string number@>
And here are the associated procedures.  Note: The |term_out| file is
system dependent.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure out_pool_str (var f:alpha_file; @!s:str_number);
var i:pool_pointer;
begin   {allowing |str_ptr <= s < str_ptr+max_pop| is a \.{.bst}-stack kludge}
if ((s<0) or (s>=str_ptr+max_pop) or (s>=max_strings)) then
    confusion ('Illegal string number:',s:0);
for i := str_start[s] to str_start[s+1]-1 do
    write(f,xchr[str_pool[i]]);
end;
@#
procedure print_a_pool_str (@!s:str_number);
begin
out_pool_str(term_out,s);
out_pool_str(log_file,s);
end;


@
@.WEB@>
Several of the elementary string operations are performed using \.{WEB}
macros instead of using \PASCAL\ procedures, because many of the
operations are done quite frequently and we want to avoid the
overhead of procedure calls. For example, here is
a simple macro that computes the length of a string.

@d length(#) == (str_start[#+1]-str_start[#])
                        {the number of characters in string number \#}


@
@:BibTeX capacity exceeded}{\quad pool size@>
Strings are created by appending character codes to |str_pool|.
The macro called |append_char|, defined here, does not check to see if the
value of |pool_ptr| has gotten too high; this test is supposed to be
made before |append_char| is used.

To test if there is room to append |l| more characters to |str_pool|,
we shall write |str_room(l)|, which aborts \BibTeX\ and gives an
error message if there isn't enough room.

@d append_char(#) ==            {put |ASCII_code| \# at the end of |str_pool|}
begin str_pool[pool_ptr]:=#; incr(pool_ptr);
end
@#
@d str_room(#) ==               {make sure that the pool hasn't overflowed}
  begin
  if (pool_ptr+# > pool_size) then
      pool_overflow;
  end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure pool_overflow;
begin
overflow('pool size ',pool_size);
end;


@
@:BibTeX capacity exceeded}{\quad number of strings@>
Once a sequence of characters has been appended to |str_pool|, it
officially becomes a string when the function |make_string| is called.
It returns the string number of the string it just made.

@<Procedures and functions for handling numbers, characters, and strings@>=
function make_string : str_number;      {current string enters the pool}
begin
if (str_ptr=max_strings) then
    overflow('number of strings ',max_strings);
incr(str_ptr);
str_start[str_ptr]:=pool_ptr;
make_string := str_ptr - 1;
end;


@
These macros destroy and recreate the string at the end of the pool.

@d flush_string == begin
                   decr(str_ptr);
                   pool_ptr := str_start[str_ptr];
                   end
@#
@d unflush_string == begin
                     incr(str_ptr);
                     pool_ptr := str_start[str_ptr];
                     end


@
This subroutine compares string |s| with another string that appears
in the buffer |buf| between positions |bf_ptr| and |bf_ptr+len-1|; the
result is |true| if and only if the strings are equal.

@<Procedures and functions for handling numbers, characters, and strings@>=
function str_eq_buf (@!s:str_number; var buf:buf_type;
                                        @!bf_ptr,@!len:buf_pointer) : boolean;
  {test equality of strings}
label exit;
var i : buf_pointer;    {running}
@!j : pool_pointer;     {indices}
begin
if (length(s) <> len) then      {strings of unequal length}
    begin
    str_eq_buf := false;
    return;
    end;
i := bf_ptr;
j := str_start[s];
while (j < str_start[s+1]) do
    begin
    if (str_pool[j] <> buf[i]) then
        begin
        str_eq_buf := false;
        return;
        end;
    incr(i);
    incr(j);
    end;
str_eq_buf := true;
exit:
end;


@
This subroutine compares two |str_pool| strings and returns true
|true| if and only if the strings are equal.

@<Procedures and functions for handling numbers, characters, and strings@>=
function str_eq_str (@!s1,@!s2:str_number) : boolean;
label exit;
begin
if (length(s1) <> length(s2)) then
    begin
    str_eq_str := false;
    return;
    end;
p_ptr1 := str_start[s1];
p_ptr2 := str_start[s2];
while (p_ptr1 < str_start[s1+1]) do
    begin
    if (str_pool[p_ptr1] <> str_pool[p_ptr2]) then
        begin
        str_eq_str := false;
        return;
        end;
    incr(p_ptr1);
    incr(p_ptr2);
    end;
str_eq_str:=true;
exit:
end;


@
@:BibTeX capacity exceeded}{\quad file name size@>
This procedure copies file name |file_name| into the beginning of
|name_of_file|, if it will fit.  It also sets the global variable
|name_length| to the appropriate value.

@<Procedures and functions for file-system interacting@>=
procedure start_name (@!file_name:str_number);
var p_ptr: pool_pointer;        {running index}
begin
if (length(file_name) > file_name_size) then
    begin
    print ('File=');
    print_pool_str (file_name);
    print_ln (',');
    file_nm_size_overflow;
    end;
name_ptr := 1;
p_ptr := str_start[file_name];
while (p_ptr < str_start[file_name+1]) do
    begin
    name_of_file[name_ptr] := chr (str_pool[p_ptr]);
    incr(name_ptr); incr(p_ptr);
    end;
name_length := length(file_name);
end;


@
@:BibTeX capacity exceeded}{\quad file name size@>
Yet another complaint-before-quiting.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure file_nm_size_overflow;
begin
overflow('file name size ',file_name_size);
end;


@
@:BibTeX capacity exceeded}{\quad file name size@>
This procedure copies file extension |ext| into the array
|name_of_file| starting at position |name_length+1|.  It also sets the
global variable |name_length| to the appropriate value.

@<Procedures and functions for file-system interacting@>=
procedure add_extension(@!ext:str_number);
var p_ptr: pool_pointer;        {running index}
begin
if (name_length + length(ext) > file_name_size) then
    begin
    print ('File=',name_of_file,', extension=');
    print_pool_str (ext); print_ln (',');
    file_nm_size_overflow;
    end;
name_ptr := name_length + 1;
p_ptr := str_start[ext];
while (p_ptr < str_start[ext+1]) do
    begin
    name_of_file[name_ptr] := chr (str_pool[p_ptr]);
    incr(name_ptr); incr(p_ptr);
    end;
name_length := name_length + length(ext);
name_ptr := name_length+1;
while (name_ptr <= file_name_size) do   {pad with blanks}
    begin
    name_of_file[name_ptr] := ' ';
    incr(name_ptr);
    end;
end;


@
@:BibTeX capacity exceeded}{\quad file name size@>
This procedure copies the default logical area name |area| into the
array |name_of_file| starting at position 1, after shifting up the
rest of the filename.  It also sets the global variable |name_length|
to the appropriate value.

@<Procedures and functions for file-system interacting@>=
procedure add_area(@!area:str_number);
var p_ptr: pool_pointer;        {running index}
begin
if (name_length + length(area) > file_name_size) then
    begin
    print ('File=');
    print_pool_str (area); print (name_of_file,',');
    file_nm_size_overflow;
    end;
name_ptr := name_length;
while (name_ptr > 0) do         {shift up name}
    begin
    name_of_file[name_ptr+length(area)] := name_of_file[name_ptr];
    decr(name_ptr);
    end;
name_ptr := 1;
p_ptr := str_start[area];
while (p_ptr < str_start[area+1]) do
    begin
    name_of_file[name_ptr] := chr (str_pool[p_ptr]);
    incr(name_ptr); incr(p_ptr);
    end;
name_length := name_length + length(area);
end;


@
This system-independent procedure converts upper-case characters to
lower case for the specified part of |buf|.  It is system independent
because it uses only the internal representation for characters.

@d case_difference = "a" - "A"

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure lower_case (var buf:buf_type; @!bf_ptr,@!len:buf_pointer);
var i:buf_pointer;
begin
if (len > 0) then
  for i := bf_ptr to bf_ptr+len-1 do
    if ((buf[i]>="A") and (buf[i]<="Z")) then
        buf[i] := buf[i] + case_difference;
end;


@
This system-independent procedure is the same as the previous except
that it converts lower- to upper-case letters.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure upper_case (var buf:buf_type; @!bf_ptr,@!len:buf_pointer);
var i:buf_pointer;
begin
if (len > 0) then
  for i := bf_ptr to bf_ptr+len-1 do
    if ((buf[i]>="a") and (buf[i]<="z")) then
        buf[i] := buf[i] - case_difference;
end;



@* The hash table.
All static strings that \BibTeX\ might have to search for, generally
identifiers, are stored and retrieved by means of a fairly standard
hash-table algorithm (but slightly altered here) called the method of
``coalescing lists''
(cf.\ Algorithm 6.4C in {\sl The Art of Computer Programming}).
Once a string enters the table, it is never removed.  The actual
sequence of characters forming a string is stored in the |str_pool|
array.

The hash table consists of the four arrays |hash_next|, |hash_text|,
|hash_ilk|, and |ilk_info|.  The first array, |hash_next[p]|, points
to the next identifier belonging to the same coalesced list as the
identifier corresponding to~|p|.  The second, |hash_text[p]|, points
to the |str_start| entry for |p|'s string. If position~|p| of the hash
table is empty, we have |hash_text[p]=0|; if position |p| is either
empty or the end of a coalesced hash list, we have
|hash_next[p]=empty|; an auxiliary pointer variable called |hash_used|
is maintained in such a way that all locations |p>=hash_used| are
nonempty.  The third, |hash_ilk[p]|, tells how this string is used (as
ordinary text, as a variable name, as an \.{.aux} file command, etc).
The fourth, |ilk_info[p]|, contains information specific to the
corresponding |hash_ilk|---for |integer_ilk|s: the integer's value;
for |cite_ilk|s: a pointer into |cite_list|; for |lc_cite_ilk|s: a
pointer to a |cite_ilk| string; for |command_ilk|s: a constant to be
used in a |case| statement; for |bst_fn_ilk|s: function-specific
information; for |macro_ilk|s: a pointer to its definition string; for
|control_seq_ilk|s: a constant for use in a |case| statement; for all
other |ilk|s it contains no information.  This |ilk|-specific
information is set in other parts of the program rather than here in
the hashing routine.

@d hash_base = empty + 1                {lowest numbered hash-table location}
@d hash_max = hash_base + hash_size - 1 {highest numbered hash-table location}
@d hash_is_full == (hash_used=hash_base) {test if all positions are occupied}
@#
@d text_ilk = 0         {a string of ordinary text}
@d integer_ilk = 1      {an integer (possibly with a |minus_sign|)}
@d aux_command_ilk = 2  {an \.{.aux}-file command}
@d aux_file_ilk = 3     {an \.{.aux} file name}
@d bst_command_ilk = 4  {a \.{.bst}-file command}
@d bst_file_ilk = 5     {a \.{.bst} file name}
@d bib_file_ilk = 6     {a \.{.bib} file name}
@d file_ext_ilk = 7     {one of \.{.aux}, \.{.bst}, \.{.bib}, \.{.bbl},
                                                                or \.{.blg}}
@d file_area_ilk = 8    {one of \.{texinputs:} or \.{texbib:}}
@d cite_ilk = 9         {a \.{\\citation} argument}
@d lc_cite_ilk = 10     {a \.{\\citation} argument converted to lower case}
@d bst_fn_ilk = 11      {a \.{.bst} function name}
@d bib_command_ilk = 12 {a \.{.bib}-file command}
@d macro_ilk = 13       {a \.{.bst} macro or a \.{.bib} string}
@d control_seq_ilk = 14 {a control sequence specifying a foreign character}
@d last_ilk = 14        {the same number as on the line above}

@<Types in the outer block@>=
@!hash_loc=hash_base..hash_max;         {a location within the hash table}
@!hash_pointer=empty..hash_max;         {either |empty| or a |hash_loc|}
@#
@!str_ilk=0..last_ilk;  {the legal string types}


@
@<Globals in the outer block@>=
@!hash_next : packed array[hash_loc] of hash_pointer;   {coalesced-list link}
@!hash_text : packed array[hash_loc] of str_number;     {pointer to a string}
@!hash_ilk : packed array[hash_loc] of str_ilk;         {the type of string}
@!ilk_info : packed array[hash_loc] of integer;         {|ilk|-specific info}
@!hash_used : hash_base..hash_max+1;    {allocation pointer for hash table}
@!hash_found : boolean;  {set to |true| if it's already in the hash table}
@!dummy_loc : hash_loc;  {receives |str_lookup| value whenever it's useless}


@
@<Local variables for initialization@>=
@!k:hash_loc;


@
Now it's time to initialize the hash table; note that |str_start[0]|
must be unused if |hash_text[k] := 0| is to have the desired effect.

@<Set initial values of key variables@>=
for k:=hash_base to hash_max do
    begin
    hash_next[k] := empty;
    hash_text[k] := 0;  {thus, no need to initialize |hash_ilk| or |ilk_info|}
    end;
hash_used := hash_max + 1;      {nothing in table initially}


@
Here is the subroutine that searches the hash table for a
(string,~|str_ilk|) pair, where the string is of length |l>=0| and
appears in |buffer[j..(j+l-1)]|.  If it finds the pair, it returns the
corresponding hash-table location and sets the global variable
|hash_found| to |true|.  Otherwise it sets |hash_found| to |false|,
and if the parameter |insert_it| is |true|, it inserts the pair into
the hash table, inserts the string into |str_pool| if not previously
encountered, and returns its location.  Note that two different pairs
can have the same string but different |str_ilk|s, in which case the
second pair encountered, if |insert_it| were |true|, would be inserted
into the hash table though its string wouldn't be inserted into
|str_pool| because it would already be there.

@d max_hash_value = hash_prime+hash_prime-2+127         {|h|'s maximum value}
@d do_insert == true            {insert string if not found in hash table}
@d dont_insert == false         {don't insert string}
@#
@d str_found = 40               {go here when you've found the string}
@d str_not_found = 45           {go here when you haven't}

@<Procedures and functions for handling numbers, characters, and strings@>=
function str_lookup(var buf:buf_type; @!j,@!l:buf_pointer; @!ilk:str_ilk;
                @!insert_it:boolean) : hash_loc;        {search the hash table}
label str_found,@!str_not_found;
var h:0..max_hash_value;        {hash code}
@!p:hash_loc;           {index into |hash_| arrays}
@!k:buf_pointer;        {index into |buf| array}
@!old_string:boolean;   {set to |true| if it's an already encountered string}
@!str_num:str_number;   {pointer to an already encountered string}
begin
@<Compute the hash code |h|@>;
p:=h+hash_base;         {start searching here; note that |0<=h<hash_prime|}
hash_found := false;
old_string := false;
loop
    begin
    @<Process the string if we've already encountered it@>;
    if (hash_next[p]=empty) then        {location |p| may or may not be empty}
        begin
        if (not insert_it) then goto str_not_found;
        @<Insert pair into hash table and make |p| point to it@>;
        goto str_found;
        end;
    p:=hash_next[p];            {old and new locations |p| are not empty}
    end;
str_not_found: do_nothing;      {don't insert pair; function value meaningless}
str_found: str_lookup:=p;
end;


@
@^for loops@>
@.WEB@>
The value of |hash_prime| should be roughly 85\% of |hash_size|, and
it should be a prime number
(it should also be less than $2^{14} + 2^{6} = 16320$ because of
\.{WEB}'s simple-macro bound).  The theory of hashing tells us to expect
fewer than two table probes, on the average, when the search is
successful.

@<Compute the hash code |h|@>=
begin
h := 0;         {note that this works for zero-length strings}
k := j;
while (k < j+l) do      {not a |for| loop in case |j = l = 0|}
    begin
    h:=h+h+buf[k];
    while (h >= hash_prime) do h:=h-hash_prime;
    incr(k);
    end;
end


@
Here we handle the case in which we've already encountered this
string; note that even if we have, we'll still have to insert the pair
into the hash table if |str_ilk| doesn't match.

@<Process the string if we've already encountered it@>=
begin
if (hash_text[p]>0) then                        {there's something here}
    if (str_eq_buf(hash_text[p],buf,j,l)) then  {it's the right string}
        if (hash_ilk[p] = ilk) then             {it's the right |str_ilk|}
            begin
            hash_found := true;
            goto str_found;
            end
          else
            begin                               {it's the wrong |str_ilk|}
            old_string := true;
            str_num := hash_text[p];
            end;
end


@
@^for loops@>
@:BibTeX capacity exceeded}{\quad hash size@>
This code inserts the pair in the appropriate unused location.

@<Insert pair into hash table and make |p| point to it@>=
begin
if (hash_text[p]>0) then                {location |p| isn't empty}
    begin
        repeat if (hash_is_full) then overflow('hash size ',hash_size);
        decr(hash_used);
        until (hash_text[hash_used]=0); {search for an empty location}
    hash_next[p]:=hash_used;
    p:=hash_used;
    end;                        {now location |p| is empty}
if (old_string) then            {it's an already encountered string}
    hash_text[p] := str_num
  else
    begin                       {it's a new string}
    str_room(l);                {make sure it'll fit in |str_pool|}
    k := j;
    while (k < j+l) do          {not a |for| loop in case |j = l = 0|}
        begin
        append_char(buf[k]);
        incr(k);
        end;
    hash_text[p] := make_string;                {and make it official}
    end;
hash_ilk[p] := ilk;
end


@
@^string pool@>
Now that we've defined the hash-table workings we can initialize the
string pool.  Unlike \TeX, \BibTeX\ does not use a |pool_file| for
string storage; instead it inserts its pre-defined strings into
|str_pool|---this makes one file fewer for the \BibTeX\ implementor
to deal with.  This section initializes |str_pool|; the pre-defined
strings will be inserted into it shortly; and other strings are
inserted while processing the input files.

@<Set initial values of key variables@>=
pool_ptr:=0; str_ptr:=1;        {hash table must have |str_start[0]| unused}
str_start[str_ptr]:=pool_ptr;


@
The longest pre-defined string determines type definitions used to
insert the pre-defined strings into |str_pool|.

@d longest_pds=12       {the length of `\.{change.case\$}'}

@<Types in the outer block@>=
@!pds_loc = 1..longest_pds;
@!pds_len = 0..longest_pds;
@!pds_type = packed array [pds_loc] of char;


@
The variables in this program beginning with |s_| specify the
locations in |str_pool| for certain often-used strings.  Those here
have to do with the file system; the next section will actually insert
them into |str_pool|.

@<Globals in the outer block@>=
@!s_aux_extension : str_number; {\.{.aux}}
@!s_log_extension : str_number; {\.{.blg}}
@!s_bbl_extension : str_number; {\.{.bbl}}
@!s_bst_extension : str_number; {\.{.bst}}
@!s_bib_extension : str_number; {\.{.bib}}
@!s_bst_area : str_number;      {\.{texinputs:}}
@!s_bib_area : str_number;      {\.{texbib:}}


@
@^important note@>
@^system dependencies@>
It's time to insert some of the pre-defined strings into |str_pool|
(and thus the hash table).  These system-dependent strings should
contain no upper-case letters, and they must all be exactly
|longest_pds| characters long (even if fewer characters are actually
stored).  The |pre_define| routine appears shortly.

Important notes: These pre-definitions must not have any glitches or
the program may bomb because the |log_file| hasn't been opened yet,
and |text_ilk|s should be pre-defined later, for
\.{.bst}-function-execution purposes.

@<Pre-define certain strings@>=
pre_define('.aux        ',4,file_ext_ilk);
s_aux_extension := hash_text[pre_def_loc];
pre_define('.bbl        ',4,file_ext_ilk);
s_bbl_extension := hash_text[pre_def_loc];
pre_define('.blg        ',4,file_ext_ilk);
s_log_extension := hash_text[pre_def_loc];
pre_define('.bst        ',4,file_ext_ilk);
s_bst_extension := hash_text[pre_def_loc];
pre_define('.bib        ',4,file_ext_ilk);
s_bib_extension := hash_text[pre_def_loc];
pre_define('texinputs:  ',10,file_area_ilk);
s_bst_area := hash_text[pre_def_loc];
pre_define('texbib:     ',7,file_area_ilk);
s_bib_area := hash_text[pre_def_loc];


@
This global variable gives the hash-table location of pre-defined
strings generated by calls to |str_lookup|.

@<Globals in the outer block@>=
@!pre_def_loc : hash_loc;


@
This procedure initializes a pre-defined string of length at most
|longest_pds|.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure pre_define (@!pds:pds_type; @!len:pds_len; @!ilk:str_ilk);
var i : pds_len;
begin
for i:=1 to len do
    buffer[i] := xord[pds[i]];
pre_def_loc := str_lookup(buffer,1,len,ilk,do_insert);
end;


@
These constants all begin with |n_| and are used for the |case|
statement that determines which command to execute.  The variable
|command_num| is set to one of these and is used to do the branching,
but it must have the full |integer| range because at times it can
assume an arbitrary |ilk_info| value (though it will be one of the
values here when we actually use it).

@d n_aux_bibdata = 0    {\.{\\bibdata}}
@d n_aux_bibstyle = 1   {\.{\\bibstyle}}
@d n_aux_citation = 2   {\.{\\citation}}
@d n_aux_input = 3      {\.{\\@@input}}
@#
@d n_bst_entry = 0      {\.{entry}}
@d n_bst_execute = 1    {\.{execute}}
@d n_bst_function = 2   {\.{function}}
@d n_bst_integers = 3   {\.{integers}}
@d n_bst_iterate = 4    {\.{iterate}}
@d n_bst_macro = 5      {\.{macro}}
@d n_bst_read = 6       {\.{read}}
@d n_bst_reverse = 7    {\.{reverse}}
@d n_bst_sort = 8       {\.{sort}}
@d n_bst_strings = 9    {\.{strings}}
@#
@d n_bib_comment = 0    {\.{comment}}
@d n_bib_preamble = 1   {\.{preamble}}
@d n_bib_string = 2     {\.{string}}

@<Globals in the outer block@>=
@!command_num : integer;


@
@^important note@>
Now we pre-define the command strings; they must all be exactly
|longest_pds| characters long.

Important note: These pre-definitions must not have any glitches or
the program may bomb because the |log_file| hasn't been opened yet.

@<Pre-define certain strings@>=
pre_define('\citation   ',9,aux_command_ilk);
ilk_info[pre_def_loc] := n_aux_citation;
pre_define('\bibdata    ',8,aux_command_ilk);
ilk_info[pre_def_loc] := n_aux_bibdata;
pre_define('\bibstyle   ',9,aux_command_ilk);
ilk_info[pre_def_loc] := n_aux_bibstyle;
pre_define('\@@input     ',7,aux_command_ilk);
ilk_info[pre_def_loc] := n_aux_input;
@#
pre_define('entry       ',5,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_entry;
pre_define('execute     ',7,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_execute;
pre_define('function    ',8,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_function;
pre_define('integers    ',8,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_integers;
pre_define('iterate     ',7,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_iterate;
pre_define('macro       ',5,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_macro;
pre_define('read        ',4,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_read;
pre_define('reverse     ',7,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_reverse;
pre_define('sort        ',4,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_sort;
pre_define('strings     ',7,bst_command_ilk);
ilk_info[pre_def_loc] := n_bst_strings;
@#
pre_define('comment     ',7,bib_command_ilk);
ilk_info[pre_def_loc] := n_bib_comment;
pre_define('preamble    ',8,bib_command_ilk);
ilk_info[pre_def_loc] := n_bib_preamble;
pre_define('string      ',6,bib_command_ilk);
ilk_info[pre_def_loc] := n_bib_string;



@* Scanning an input line.
This section describes the various |buffer| scanning routines.  The
two global variables |buf_ptr1| and |buf_ptr2| are used in scanning an
input line.  Between scans, |buf_ptr1| points to the first character
of the current token and |buf_ptr2| points to that of the next.  The
global variable |last|, set by the function |input_ln|, marks the end
of the current line; it equals 0 at the end of the current file.  All
the procedures and functions in this section will indicate an
end-of-line when it's the end of the file.

@d token_len == (buf_ptr2 - buf_ptr1)   {of the current token}
@d scan_char == buffer[buf_ptr2]        {the current character}

@<Globals in the outer block@>=
@!buf_ptr1:buf_pointer; {points to the first position of the current token}
@!buf_ptr2:buf_pointer; {used to find the end of the current token}


@
These macros send the current token, in |buffer[buf_ptr1]| to
|buffer[buf_ptr2-1]|, to an output file.

@d print_token == print_a_token {making this a procedure saves a little space}
@#
@d trace_pr_token == begin
                     out_token(log_file);
                     end


@
@^system dependencies@>
And here are the associated procedures.  Note: The |term_out| file is
system dependent.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure out_token (var f:alpha_file);
var i:buf_pointer;
begin
i := buf_ptr1;
while (i < buf_ptr2) do
    begin
    write(f,xchr[buffer[i]]);
    incr(i);
    end;
end;
@#
procedure print_a_token;
begin
out_token(term_out);
out_token(log_file);
end;


@
This function scans the |buffer| for the next token, starting at the
global variable |buf_ptr2| and ending just before either the single
specified stop-character or the end of the current line, whichever
comes first, respectively returning |true| or |false|; afterward,
|scan_char| is the first character following this token.

@<Procedures and functions for input scanning@>=
function scan1 (@!char1:ASCII_code) : boolean;
begin
buf_ptr1 := buf_ptr2;
                        {scan until end-of-line or the specified character}
while ((scan_char <> char1) and (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (buf_ptr2 < last) then
    scan1 := true
  else
    scan1 := false;
end;


@
This function is the same but stops at |white_space| characters as well.

@<Procedures and functions for input scanning@>=
function scan1_white (@!char1:ASCII_code) : boolean;
begin
buf_ptr1 := buf_ptr2;
        {scan until end-of-line, the specified character, or |white_space|}
while ((lex_class[scan_char] <> white_space) and (scan_char <> char1) and
                                                        (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (buf_ptr2 < last) then
    scan1_white := true
  else
    scan1_white := false;
end;


@
This function is similar to |scan1|, but stops at either of two
stop-characters as well as the end of the current line.

@<Procedures and functions for input scanning@>=
function scan2 (@!char1,@!char2:ASCII_code) : boolean;
begin
buf_ptr1 := buf_ptr2;
                        {scan until end-of-line or the specified characters}
while ((scan_char <> char1) and (scan_char <> char2) and (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (buf_ptr2 < last) then
    scan2 := true
  else
    scan2 := false;
end;


@
This function is the same but stops at |white_space| characters as well.

@<Procedures and functions for input scanning@>=
function scan2_white (@!char1,@!char2:ASCII_code) : boolean;
begin
buf_ptr1 := buf_ptr2;
        {scan until end-of-line, the specified characters, or |white_space|}
while ((scan_char <> char1) and (scan_char <> char2) and
                (lex_class[scan_char] <> white_space) and (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (buf_ptr2 < last) then
    scan2_white := true
  else
    scan2_white := false;
end;


@
This function is similar to |scan2|, but stops at either of three
stop-characters as well as the end of the current line.

@<Procedures and functions for input scanning@>=
function scan3 (@!char1,@!char2,@!char3:ASCII_code) : boolean;
begin
buf_ptr1 := buf_ptr2;
                        {scan until end-of-line or the specified characters}
while ((scan_char <> char1) and (scan_char <> char2) and
                                (scan_char <> char3) and (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (buf_ptr2 < last) then
    scan3 := true
  else
    scan3 := false;
end;


@
This function scans for letters, stopping at the first nonletter; it
returns |true| if there is at least one letter.

@<Procedures and functions for input scanning@>=
function scan_alpha : boolean;
begin
buf_ptr1 := buf_ptr2;
                                        {scan until end-of-line or a nonletter}
while ((lex_class[scan_char] = alpha) and (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (token_len = 0) then
    scan_alpha := false
  else
    scan_alpha := true;
end;


@
These are the possible values for |scan_result|; they're set by the
|scan_identifier| procedure and are described in the next section.

@d id_null = 0
@d specified_char_adjacent = 1
@d other_char_adjacent = 2
@d white_adjacent = 3

@<Globals in the outer block@>=
@!scan_result : id_null..white_adjacent;


@
This procedure scans for an identifier, stopping at the first
|illegal_id_char|, or stopping at the first character if it's
|numeric|.  It sets the global variable |scan_result| to |id_null| if
the identifier is null, else to |white_adjacent| if it ended at a
|white_space| character or an end-of-line, else to
|specified_char_adjacent| if it ended at one of |char1| or |char2| or
|char3|, else to |other_char_adjacent| if it ended at a nonspecified,
non|white_space| |illegal_id_char|.  By convention, when some calling
code really wants just one or two ``specified'' characters, it merely
repeats one of the characters.

@<Procedures and functions for input scanning@>=
procedure scan_identifier (@!char1,@!char2,@!char3:ASCII_code);
begin
buf_ptr1 := buf_ptr2;
if (lex_class[scan_char] <> numeric) then
                        {scan until end-of-line or an |illegal_id_char|}
    while ((id_class[scan_char] = legal_id_char) and (buf_ptr2 < last)) do
        incr(buf_ptr2);
if (token_len = 0) then
    scan_result := id_null
else if ((lex_class[scan_char] = white_space) or (buf_ptr2 = last)) then
    scan_result := white_adjacent
else if ((scan_char = char1) or (scan_char = char2) or (scan_char = char3))
                                                                        then
    scan_result := specified_char_adjacent
else
    scan_result := other_char_adjacent;
end;


@
The next two procedures scan for an integer, setting the global
variable |token_value| to the corresponding integer.

@d char_value == (scan_char - "0")      {the value of the digit being scanned}

@<Globals in the outer block@>=
@!token_value : integer;        {the numeric value of the current token}


@
This function scans for a nonnegative integer, stopping at the first
nondigit; it sets the value of |token_value| accordingly.  It returns
|true| if the token was a legal nonnegative integer (i.e., consisted
of one or more digits).

@<Procedures and functions for input scanning@>=
function scan_nonneg_integer : boolean;
begin
buf_ptr1 := buf_ptr2;
token_value := 0;
                                        {scan until end-of-line or a nondigit}
while ((lex_class[scan_char] = numeric) and (buf_ptr2 < last)) do
    begin
    token_value := token_value*10 + char_value;
    incr(buf_ptr2);
    end;
if (token_len = 0) then                 {there were no digits}
    scan_nonneg_integer := false
  else
    scan_nonneg_integer := true;
end;


@
This procedure scans for an integer, stopping at the first nondigit;
it sets the value of |token_value| accordingly.  It returns |true| if
the token was a legal integer (i.e., consisted of an optional
|minus_sign| followed by one or more digits).

@d negative == (sign_length = 1)        {if this integer is negative}

@<Procedures and functions for input scanning@>=
function scan_integer : boolean;
var sign_length : 0..1;         {1 if there's a |minus_sign|, 0 if not}
begin
buf_ptr1 := buf_ptr2;
if (scan_char = minus_sign) then        {it's a negative number}
    begin
    sign_length := 1;
    incr(buf_ptr2);                     {skip over the |minus_sign|}
    end
  else
    sign_length := 0;
token_value := 0;
                                        {scan until end-of-line or a nondigit}
while ((lex_class[scan_char] = numeric) and (buf_ptr2 < last)) do
    begin
    token_value := token_value*10 + char_value;
    incr(buf_ptr2);
    end;
if (negative) then
    token_value := -token_value;
if (token_len = sign_length) then       {there were no digits}
    scan_integer := false
  else
    scan_integer := true;
end;


@
This function scans over |white_space| characters, stopping either at
the first nonwhite character or the end of the line, respectively
returning |true| or |false|.

@<Procedures and functions for input scanning@>=
function scan_white_space : boolean;
begin
                                        {scan until end-of-line or a nonwhite}
while ((lex_class[scan_char] = white_space) and (buf_ptr2 < last)) do
    incr(buf_ptr2);
if (buf_ptr2 < last) then
    scan_white_space := true
  else
    scan_white_space := false;
end;


@
The |print_bad_input_line| procedure prints the current input line,
splitting it at the character being scanned: It prints |buffer[0]|,
|buffer[1]|, \dots, |buffer[buf_ptr2-1]| on one line and
|buffer[buf_ptr2]|, \dots, |buffer[last-1]| on the next (and both
lines start with a colon between two |space|s).  Each |white_space|
character is printed as a |space|.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_bad_input_line;
var bf_ptr : buf_pointer;
begin
print (' : ');
bf_ptr := 0;
while (bf_ptr < buf_ptr2) do
    begin
    if (lex_class[buffer[bf_ptr]] = white_space) then
        print (xchr[space])
      else
        print (xchr[buffer[bf_ptr]]);
    incr(bf_ptr);
    end;
print_newline;
print (' : ');
bf_ptr := 0;
while (bf_ptr < buf_ptr2) do
    begin
    print (xchr[space]);
    incr(bf_ptr);
    end;
bf_ptr := buf_ptr2;
while (bf_ptr < last) do
    begin
    if (lex_class[buffer[bf_ptr]] = white_space) then
        print (xchr[space])
      else
        print (xchr[buffer[bf_ptr]]);
    incr(bf_ptr);
    end;
print_newline;@/
bf_ptr := 0;
while ((bf_ptr < buf_ptr2) and (lex_class[buffer[bf_ptr]] = white_space)) do
    incr(bf_ptr);
if (bf_ptr = buf_ptr2) then
    print_ln ('(Error may have been on previous line)');
mark_error;
end;


@
This little procedure exists because it's used by at least two other
procedures and thus saves some space.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_skipping_whatever_remains;
begin
print ('I''m skipping whatever remains of this ');
end;



@* Getting the top-level auxiliary file name.
@^system dependencies@>
These modules read the name of the top-level \.{.aux} file.  Some
systems will try to find this on the command line; if it's not there
it will come from the user's terminal.  In either case, the name goes
into the |char| array |name_of_file|, and the files relevant to this
name are opened.

@d aux_found=41         {go here when the \.{.aux} name is legit}
@d aux_not_found=46     {go here when it's not}

@<Globals in the outer block@>=
@!aux_name_length : 0..file_name_size+1;        {\.{.aux} name sans extension}


@
@^system dependencies@>
@^user abuse@>
I mean, this is truly disgraceful.  A user has to type something in to
the terminal just once during the entire run.  And it's not some
complicated string where you have to get every last punctuation mark
just right, and it's not some fancy list where you get nervous because
if you forget one item you have to type the whole thing again; it's
just a simple, ordinary, file name.  Now you'd think a five-year-old
could do it; you'd think it's so simple a user should be able to do it
in his sleep.  But noooooooooo.  He had to sit there droning on and on
about who knows what until he exceeded the bounds of common sense, and
he probably didn't even realize it.  Just pitiful.  What's this world
coming to?  We should probably just delete all his files and be done
with him.  Note: The |term_out| file is system dependent.

@d sam_you_made_the_file_name_too_long == begin
                                          sam_too_long_file_name_print;
                                          goto aux_not_found;
                                          end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure sam_too_long_file_name_print;
begin
write (term_out,'File name `');
name_ptr := 1;
while (name_ptr <= aux_name_length) do
    begin
    write (term_out,name_of_file[name_ptr]);
    incr(name_ptr);
    end;
write_ln (term_out,''' is too long');
end;


@
@^system dependencies@>
@^user abuse@>
We've abused the user enough for one section; suffice it to
say here that most of what we said last module still applies.
Note: The |term_out| file is system dependent.

@d sam_you_made_the_file_name_wrong == begin
                                       sam_wrong_file_name_print;
                                       goto aux_not_found;
                                       end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure sam_wrong_file_name_print;
begin
write (term_out,'I couldn''t open file name `');
name_ptr := 1;
while (name_ptr <= name_length) do
    begin
    write (term_out,name_of_file[name_ptr]);
    incr(name_ptr);
    end;
write_ln (term_out,'''');
end;


@
@^system dependencies@>
This procedure consists of a loop that reads and processes a (nonnull)
\.{.aux} file name.  It's this module and the next two that must be
changed on those systems using command-line arguments.  Note: The
|term_out| and |term_in| files are system dependent.

@<Procedures and functions for the reading and processing of input files@>=
procedure get_the_top_level_aux_file_name;
label aux_found,@!aux_not_found;
var @<Variables for possible command-line processing@>@/
begin
check_cmnd_line := false;                       {many systems will change this}
loop
    begin
    if (check_cmnd_line) then
        @<Process a possible command line@>
      else
        begin
        write (term_out,'Please type input file name (no extension)--');
        if (eoln(term_in)) then                 {so the first |read| works}
            read_ln (term_in);
        aux_name_length := 0;
        while (not eoln(term_in)) do
            begin
            if (aux_name_length = file_name_size) then
                begin
                while (not eoln(term_in)) do    {discard the rest of the line}
                    get(term_in);
                sam_you_made_the_file_name_too_long;
                end;
            incr(aux_name_length);
            name_of_file[aux_name_length] := term_in^;
            get(term_in);
            end;
        end;
    @<Handle this \.{.aux} name@>;
aux_not_found:
    check_cmnd_line := false;
    end;
aux_found:                      {now we're ready to read the \.{.aux} file}
end;


@
@^system dependencies@>
The switch |check_cmnd_line| tells us whether we're to check for a
possible command-line argument.

@<Variables for possible command-line processing@>=
@!check_cmnd_line : boolean;    {|true| if we're to check the command line}


@
@^system dependencies@>
Here's where we do the real command-line work.  Those systems needing
more than a single module to handle the task should add the extras to
the ``System-dependent changes'' section.

@<Process a possible command line@>=
begin
do_nothing;             {the ``default system'' doesn't use the command line}
end


@
Here we orchestrate this \.{.aux} name's handling: we add the various
extensions, try to open the files with the resulting name, and
store the name strings we'll need later.

@<Handle this \.{.aux} name@>=
begin
if ((aux_name_length + length(s_aux_extension) > file_name_size) or@|
        (aux_name_length + length(s_log_extension) > file_name_size) or@|
        (aux_name_length + length(s_bbl_extension) > file_name_size)) then
    sam_you_made_the_file_name_too_long;
@<Add extensions and open files@>;
@<Put this name into the hash table@>;
goto aux_found;
end


@
Here we set up definitions and declarations for files opened in this
section.  Each element in |aux_list| (except for
|aux_list[aux_stack_size]|, which is always unused) is a pointer to
the appropriate |str_pool| string representing the \.{.aux} file name.
The array |aux_file| contains the corresponding \PASCAL\ |file|
variables.

@d cur_aux_str == aux_list[aux_ptr]  {shorthand for the current \.{.aux} file}
@d cur_aux_file == aux_file[aux_ptr]    {shorthand for the current |aux_file|}
@d cur_aux_line == aux_ln_stack[aux_ptr] {line number of current \.{.aux} file}

@<Globals in the outer block@>=
@!aux_file : array[aux_number] of alpha_file; {open \.{.aux} |file| variables}
@!aux_list : array[aux_number] of str_number;   {the open \.{.aux} file list}
@!aux_ptr : aux_number;         {points to the currently open \.{.aux} file}
@!aux_ln_stack : array[aux_number] of integer;  {open \.{.aux} line numbers}
@#
@!top_lev_str : str_number;     {the top-level \.{.aux} file's name}
@#
@!log_file : alpha_file;        {the |file| variable for the \.{.blg} file}
@!bbl_file : alpha_file;        {the |file| variable for the \.{.bbl} file}


@
Where |aux_number| is the obvious.

@<Types in the outer block@>=
@!aux_number = 0..aux_stack_size;       {gives the |aux_list| range}


@
@^system dependencies@>
We must make sure the (top-level) \.{.aux}, \.{.blg}, and \.{.bbl}
files can be opened.

@<Add extensions and open files@>=
begin
name_length := aux_name_length;         {set to last used position}
add_extension (s_aux_extension);        {this also sets |name_length|}
aux_ptr := 0;                           {initialize the \.{.aux} file stack}
if (not a_open_in(cur_aux_file)) then
    sam_you_made_the_file_name_wrong;
@#
name_length := aux_name_length;
add_extension (s_log_extension);        {this also sets |name_length|}
if (not a_open_out(log_file)) then
    sam_you_made_the_file_name_wrong;
@#
name_length := aux_name_length;
add_extension (s_bbl_extension);        {this also sets |name_length|}
if (not a_open_out(bbl_file)) then
    sam_you_made_the_file_name_wrong;
end


@
@:this can't happen}{\quad Already encountered auxiliary file@>
This code puts the \.{.aux} file name, both with and without the
extension, into the hash table, and it initializes |aux_list|.  Note
that all previous top-level \.{.aux}-file stuff must have been
successful.

@<Put this name into the hash table@>=
begin
name_length := aux_name_length;
add_extension (s_aux_extension);        {this also sets |name_length|}
name_ptr := 1;
while (name_ptr <= name_length) do
    begin
    buffer[name_ptr] := xord[name_of_file[name_ptr]];
    incr(name_ptr);
    end;
top_lev_str := hash_text[
                str_lookup(buffer,1,aux_name_length,text_ilk,do_insert)];
cur_aux_str := hash_text[
                str_lookup(buffer,1,name_length,aux_file_ilk,do_insert)];
                                {note that this has initialized |aux_list|}
if (hash_found) then
    begin
      trace
      print_aux_name;
      ecart@/
    confusion ('Already encountered auxiliary file');
    end;
cur_aux_line := 0;   {this finishes initializing the top-level \.{.aux} file}
end


@
Print the name of the current \.{.aux} file, followed by a |newline|.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_aux_name;
begin
print_pool_str (cur_aux_str);
print_newline;
end;



@* Reading the auxiliary file(s).
@^auxiliary-file commands@>
Now it's time to read the \.{.aux} file.  The only commands we handle
are \.{\\citation} (there can be arbitrarily many, each having
arbitrarily many arguments), \.{\\bibdata} (there can be just one, but
it can have arbitrarily many arguments), \.{\\bibstyle} (there can be
just one, and it can have just one argument), and \.{\\@@input} (there
can be arbitrarily many, each with one argument, and they can be
nested to a depth of |aux_stack_size|).  Each of these commands is
assumed to be on just a single line.  The rest of the \.{.aux} file is
ignored.

@d aux_done=31          {go here when finished with the \.{.aux} files}

@<Labels in the outer block@>=
,@!aux_done


@
We keep reading and processing input lines until none left.  This is
part of the main program; hence, because of the |aux_done| label,
there's no conventional |begin|-|end| pair surrounding the entire
module.

@<Read the \.{.aux} file@>=
print ('The top-level auxiliary file: ');
print_aux_name;
loop
    begin                       {|pop_the_aux_stack| will exit the loop}
    incr(cur_aux_line);
    if (not input_ln(cur_aux_file)) then        {end of current \.{.aux} file}
        pop_the_aux_stack
      else
        get_aux_command_and_process;
    end;
  trace
  trace_pr_ln ('Finished reading the auxiliary file(s)');
  ecart@/
aux_done:
last_check_for_aux_errors;


@
When we find a bug, we print a message and flush the rest of the line.
This macro must be called from within a procedure that has an |exit|
label.

@d aux_err_return == begin
                     aux_err_print;
                     return;            {flush this input line}
                     end

@d aux_err(#) == begin
                 print (#);
                 aux_err_return;
                 end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure aux_err_print;
begin
print ('---line ',cur_aux_line:0,' of file ');
print_aux_name;@/
print_bad_input_line;                   {this call does the |mark_error|}
print_skipping_whatever_remains;
print_ln ('command')
end;


@
@:this can't happen}{\quad Illegal auxiliary-file command@>
Here are a bunch of macros whose print statements are used at least
twice.  Thus we save space by making the statements procedures.  This
macro complains when there's a repeated command that's to be used just
once.

@d aux_err_illegal_another(#) == begin
                                 aux_err_illegal_another_print (#);
                                 aux_err_return;
                                 end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure aux_err_illegal_another_print (@!cmd_num : integer);
begin
print ('Illegal, another \bib');
case (cmd_num) of
    n_aux_bibdata : print ('data');
    n_aux_bibstyle : print ('style');
    othercases
        confusion ('Illegal auxiliary-file command')
endcases;
print (' command');
end;


@
This one complains when a command is missing its |right_brace|.

@d aux_err_no_right_brace == begin
                             aux_err_no_right_brace_print;
                             aux_err_return;
                             end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure aux_err_no_right_brace_print;
begin
print ('No "',xchr[right_brace],'"');
end;


@
This one complains when a command has stuff after its |right_brace|.

@d aux_err_stuff_after_right_brace == begin
                                      aux_err_stuff_after_right_brace_print;
                                      aux_err_return;
                                      end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure aux_err_stuff_after_right_brace_print;
begin
print ('Stuff after "',xchr[right_brace],'"');
end;


@
And this one complains when a command has |white_space| in its
argument.

@d aux_err_white_space_in_argument == begin
                                      aux_err_white_space_in_argument_print;
                                      aux_err_return;
                                      end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure aux_err_white_space_in_argument_print;
begin
print ('White space in argument');
end;


@
@^auxiliary-file commands@>
@:this can't happen}{\quad Unknown auxiliary-file command@>
We're not at the end of an \.{.aux} file, so we see if the current
line might be a command of interest.  A command of interest will be a
line without blanks, consisting of a command name, a |left_brace|, one
or more arguments separated by commas, and a |right_brace|.

@<Scan for and process an \.{.aux} command@>=
procedure get_aux_command_and_process;
label exit;
begin
buf_ptr2 := 0;                          {mark the beginning of the next token}
if (not scan1(left_brace)) then         {no |left_brace|---flush line}
    return;
command_num := ilk_info[
        str_lookup(buffer,buf_ptr1,token_len,aux_command_ilk,dont_insert)];
if (hash_found) then
    case (command_num) of
        n_aux_bibdata : aux_bib_data_command;
        n_aux_bibstyle : aux_bib_style_command;
        n_aux_citation : aux_citation_command;
        n_aux_input : aux_input_command;
        othercases
            confusion ('Unknown auxiliary-file command')
    endcases;
exit:
end;


@
Here we introduce some variables for processing a \.{\\bibdata}
command.  Each element in |bib_list| (except for
|bib_list[max_bib_files]|, which is always unused) is a pointer to the
appropriate |str_pool| string representing the \.{.bib} file name.
The array |bib_file| contains the corresponding \PASCAL\ |file|
variables.

@d cur_bib_str == bib_list[bib_ptr]     {shorthand for current \.{.bib} file}
@d cur_bib_file == bib_file[bib_ptr]    {shorthand for current |bib_file|}

@<Globals in the outer block@>=
@!bib_list : array[bib_number] of str_number;   {the \.{.bib} file list}
@!bib_ptr : bib_number;         {pointer for the current \.{.bib} file}
@!num_bib_files : bib_number;   {the total number of \.{.bib} files}
@!bib_seen : boolean;   {|true| if we've already seen a \.{\\bibdata} command}
@!bib_file : array[bib_number] of alpha_file; {corresponding |file| variables}


@
Where |bib_number| is the obvious.

@<Types in the outer block@>=
@!bib_number = 0..max_bib_files;        {gives the |bib_list| range}


@
@<Set initial values of key variables@>=
bib_ptr := 0;           {this makes |bib_list| empty}
bib_seen := false;      {we haven't seen a \.{\\bibdata} command yet}


@
@:auxiliary-file commands}{\quad \.{\\bibdata}@>
A \.{\\bibdata} command will have its arguments between braces and
separated by commas.  There must be exactly one such command in the
\.{.aux} file(s).  All upper-case letters are converted to lower case.

@<Procedures and functions for the reading and processing of input files@>=
procedure aux_bib_data_command;
label exit;
begin
if (bib_seen) then
    aux_err_illegal_another (n_aux_bibdata);
bib_seen := true;       {now we've seen a \.{\\bibdata} command}
while (scan_char <> right_brace) do
    begin
    incr(buf_ptr2);                     {skip over the previous stop-character}
    if (not scan2_white(right_brace,comma)) then
        aux_err_no_right_brace;
    if (lex_class[scan_char] = white_space) then
        aux_err_white_space_in_argument;
    if ((last > buf_ptr2+1) and (scan_char = right_brace)) then
        aux_err_stuff_after_right_brace;
    @<Open a \.{.bib} file@>;
    end;
exit:
end;


@
Here's a procedure we'll need shortly.  It prints the name of the
current \.{.bib} file, followed by a |newline|.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_bib_name;
begin
print_pool_str (cur_bib_str);
print_pool_str (s_bib_extension);
print_newline;
end;


@
This macro is similar to |aux_err| but it complains specifically about
opening a file for a \.{\\bibdata} command.

@d open_bibdata_aux_err(#) == begin
                              print (#);
                              print_bib_name;
                              aux_err_return;   {this does the |mark_error|}
                              end


@
@:BibTeX capacity exceeded}{\quad number of \.{.bib} files@>
Now we add the just-found argument to |bib_list| if it hasn't already
been encountered as a \.{\\bibdata} argument and if, after appending
the |s_bib_extension| string, the resulting file name can be opened.

@<Open a \.{.bib} file@>=
begin
if (bib_ptr = max_bib_files) then
    overflow('number of database files ',max_bib_files);
cur_bib_str := hash_text[
                str_lookup(buffer,buf_ptr1,token_len,bib_file_ilk,do_insert)];
if (hash_found) then    {already encountered this as a \.{\\bibdata} argument}
    open_bibdata_aux_err ('This database file appears more than once: ');
start_name (cur_bib_str);
add_extension (s_bib_extension);
if (not a_open_in(cur_bib_file)) then
    begin
    add_area (s_bib_area);
    if (not a_open_in(cur_bib_file)) then
        open_bibdata_aux_err ('I couldn''t open database file ');
    end;
  trace
  trace_pr_pool_str (cur_bib_str);
  trace_pr_pool_str (s_bib_extension);
  trace_pr_ln (' is a bibdata file');
  ecart@/
incr(bib_ptr);
end


@
Here we introduce some variables for processing a \.{\\bibstyle}
command.

@<Globals in the outer block@>=
@!bst_seen : boolean;   {|true| if we've already seen a \.{\\bibstyle} command}
@!bst_str : str_number;         {the string number for the \.{.bst} file}
@!bst_file : alpha_file;        {the corresponding |file| variable}


@
And we initialize.

@<Set initial values of key variables@>=
bst_str := 0;           {mark |bst_str| as unused}
bst_seen := false;      {we haven't seen a \.{\\bibstyle} command yet}


@
@:auxiliary-file commands}{\quad \.{\\bibstyle}@>
A \.{\\bibstyle} command will have exactly one argument, and it will
be between braces.  There must be exactly one such command in the
\.{.aux} file(s).  All upper-case letters are converted to lower case.

@<Procedures and functions for the reading and processing of input files@>=
procedure aux_bib_style_command;
label exit;
begin
if (bst_seen) then
    aux_err_illegal_another (n_aux_bibstyle);
bst_seen := true;               {now we've seen a \.{\\bibstyle} command}
incr(buf_ptr2);                 {skip over the |left_brace|}
if (not scan1_white(right_brace)) then
    aux_err_no_right_brace;
if (lex_class[scan_char] = white_space) then
    aux_err_white_space_in_argument;
if (last > buf_ptr2+1) then
    aux_err_stuff_after_right_brace;
@<Open the \.{.bst} file@>;
exit:
end;


@
@:this can't happen}{\quad Already encountered style file@>
Now we open the file whose name is the just-found argument appended
with the |s_bst_extension| string, if possible.

@<Open the \.{.bst} file@>=
begin
bst_str := hash_text[
                str_lookup(buffer,buf_ptr1,token_len,bst_file_ilk,do_insert)];
if (hash_found) then
    begin
      trace
      print_bst_name;
      ecart@/
    confusion ('Already encountered style file');
    end;
start_name (bst_str);
add_extension (s_bst_extension);
if (not a_open_in(bst_file)) then
    begin
    add_area (s_bst_area);
    if (not a_open_in(bst_file)) then
        begin
        print ('I couldn''t open style file ');
        print_bst_name;@/
        bst_str := 0;                           {mark as unused again}
        aux_err_return;
        end;
    end;
print ('The style file: ');
print_bst_name;
end


@
Print the name of the \.{.bst} file, followed by a |newline|.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_bst_name;
begin
print_pool_str (bst_str);
print_pool_str (s_bst_extension);
print_newline;
end;


@
Here we introduce some variables for processing a \.{\\citation}
command.  Each element in |cite_list| (except for
|cite_list[max_cites]|, which is always unused) is a pointer to the
appropriate |str_pool| string.  The cite-key list is kept in order of
occurrence with duplicates removed.

@d cur_cite_str == cite_list[cite_ptr]  {shorthand for the current cite key}

@<Globals in the outer block@>=
@!cite_list : packed array[cite_number] of str_number;  {the cite-key list}
@!cite_ptr : cite_number;       {pointer for the current cite key}
@!entry_cite_ptr : cite_number; {cite pointer for the current entry}
@!num_cites : cite_number;      {the total number of distinct cite keys}
@!old_num_cites : cite_number;  {set to a previous |num_cites| value}
@!citation_seen : boolean;      {|true| if we've seen a \.{\\citation} command}
@!cite_loc : hash_loc;          {the hash-table location of a cite key}
@!lc_cite_loc : hash_loc;       {and of its lower-case equivalent}
@!lc_xcite_loc : hash_loc;      {a second |lc_cite_loc| variable}
@!cite_found : boolean;         {|true| if we've already seen this cite key}
@!all_entries : boolean;        {|true| if we're to use the entire database}
@!all_marker : cite_number;     {we put the other entries in |cite_list| here}


@
Where |cite_number| is the obvious.

@<Types in the outer block@>=
@!cite_number = 0..max_cites;   {gives the |cite_list| range}


@
@<Set initial values of key variables@>=
cite_ptr := 0;          {this makes |cite_list| empty}
citation_seen := false; {we haven't seen a \.{\\citation} command yet}
all_entries := false;   {by default, use just the entries explicitly named}


@
@^case mismatch@>
@^entire database inclusion@>
@^whole database inclusion@>
@:LaTeX}{\LaTeX@>
@:auxiliary-file commands}{\quad \.{\\citation}@>
A \.{\\citation} command will have its arguments between braces and
separated by commas.  Upper/lower cases are considered to be different
for \.{\\citation} arguments, which is the same as the rest of \LaTeX\
but different from the rest of \BibTeX.  A cite key needn't exactly
case-match its corresponding database key to work, although two cite
keys that are case-mismatched will produce an error message.
(A {\sl case mismatch\/} is a mismatch, but only because of a case
difference.)

A \.{\\citation} command having \.{*} as an argument indicates that
the entire database will be included (almost as if a \.{\\nocite}
command that listed every cite key in the database, in order, had been
given at the corresponding spot in the \.{.tex} file).

@d next_cite = 23       {read the next argument}

@<Procedures and functions for the reading and processing of input files@>=
procedure aux_citation_command;
label next_cite,@!exit;
begin
citation_seen := true;          {now we've seen a \.{\\citation} command}
while (scan_char <> right_brace) do
    begin
    incr(buf_ptr2);             {skip over the previous stop-character}
    if (not scan2_white(right_brace,comma)) then
        aux_err_no_right_brace;
    if (lex_class[scan_char] = white_space) then
        aux_err_white_space_in_argument;
    if ((last > buf_ptr2+1) and (scan_char = right_brace)) then
        aux_err_stuff_after_right_brace;
    @<Check the cite key@>;
next_cite:
    end;
exit:
end;


@
@^kludge@>
We must check if (the lower-case version of) this cite key has been
previously encountered, and proceed accordingly.  The alias kludge
helps make the stack space not overflow on some machines.

@d ex_buf1== ex_buf             {an alias, used only in this module}

@<Check the cite key@>=
begin
  trace
  trace_pr_token;
  trace_pr (' cite key encountered');
  ecart@/
@<Check for entire database inclusion (and thus skip this cite key)@>;
tmp_ptr := buf_ptr1;
while (tmp_ptr < buf_ptr2) do
    begin
    ex_buf1[tmp_ptr] := buffer[tmp_ptr];
    incr(tmp_ptr);
    end;
lower_case (ex_buf1, buf_ptr1, token_len);      {convert to `canonical' form}
lc_cite_loc := str_lookup(ex_buf1,buf_ptr1,token_len,lc_cite_ilk,do_insert);
if (hash_found) then    {already encountered this as a \.{\\citation} argument}
    @<Cite seen, don't add a cite key@>
  else
    @<Cite unseen, add a cite key@>;
                                {it's a new cite key---add it to |cite_list|}
end


@
Here we check for a \.{\\citation} command having \.{*} as an
argument, indicating that the entire database will be included.

@<Check for entire database inclusion (and thus skip this cite key)@>=
begin
if (token_len = 1) then
  if (buffer[buf_ptr1] = star) then
    begin
      trace
      trace_pr_ln ('---entire database to be included');
      ecart@/
    if (all_entries) then
        begin
        print_ln ('Multiple inclusions of entire database');
        aux_err_return;
        end
      else
        begin
        all_entries := true;
        all_marker := cite_ptr;
        goto next_cite;
        end;
    end;
end


@
@^case mismatch errors@>
We've previously encountered the lower-case version, so we check that
the actual version exactly matches the actual version of the
previously-encountered cite key(s).

@<Cite seen, don't add a cite key@>=
begin
  trace
  trace_pr_ln (' previously');
  ecart@/
dummy_loc := str_lookup(buffer,buf_ptr1,token_len,cite_ilk,dont_insert);
if (not hash_found) then                {case mismatch error}
    begin
    print ('Case mismatch error between cite keys ');
    print_token;
    print (' and ');
    print_pool_str (cite_list[ilk_info[ilk_info[lc_cite_loc]]]);
    print_newline;
    aux_err_return;
    end;
end


@
@:this can't happen}{\quad Cite hash error@>
Now we add the just-found argument to |cite_list| if there isn't
anything funny happening.

@<Cite unseen, add a cite key@>=
begin
  trace
  trace_pr_newline;
  ecart@/
cite_loc := str_lookup(buffer,buf_ptr1,token_len,cite_ilk,do_insert);
if (hash_found) then
    hash_cite_confusion;
check_cite_overflow (cite_ptr);
cur_cite_str := hash_text[cite_loc];
ilk_info[cite_loc] := cite_ptr;
ilk_info[lc_cite_loc] := cite_loc;
incr(cite_ptr);
end


@
@:this can't happen}{\quad Cite hash error@>
Here's a serious complaint (that is, a bug) concerning hash problems.
This is the first of several similar bug-procedures that exist only
because they save space.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure hash_cite_confusion;
begin
confusion ('Cite hash error');
end;


@
@^fetish@>
@:BibTeX capacity exceeded}{\quad number of cite keys@>
Complain if somebody's got a cite fetish.  This procedure is called
when were about to add another cite key to |cite_list|.  It assumes
that |cite_loc| gives the potential cite key's hash table location.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure check_cite_overflow (@!last_cite : cite_number);
begin
if (last_cite = max_cites) then
    begin
    print_pool_str (hash_text[cite_loc]);
    print_ln (' is the key:');
    overflow('number of cite keys ',max_cites);
    end;
end;


@
@:auxiliary-file commands}{\quad \.{\\\AT!input}@>
An \.{\\@@input} command will have exactly one argument, it will
be between braces, and it must have the |s_aux_extension|.
All upper-case letters are converted to lower case.

@<Procedures and functions for the reading and processing of input files@>=
procedure aux_input_command;
label exit;
var aux_extension_ok : boolean;         {to check for a correct file extension}
begin
incr(buf_ptr2);                         {skip over the |left_brace|}
if (not scan1_white(right_brace)) then
    aux_err_no_right_brace;
if (lex_class[scan_char] = white_space) then
    aux_err_white_space_in_argument;
if (last > buf_ptr2+1) then
    aux_err_stuff_after_right_brace;
@<Push the \.{.aux} stack@>;
exit:
end;


@
@:BibTeX capacity exceeded}{\quad number of \.{.aux} files@>
We must check that this potential \.{.aux} file won't overflow the
stack, that it has the correct extension, that we haven't encountered
it before (to prevent, among other things, an infinite loop).

@<Push the \.{.aux} stack@>=
begin
incr(aux_ptr);
if (aux_ptr = aux_stack_size) then
    begin
    print_token; print (': ');
    overflow('auxiliary file depth ',aux_stack_size);
    end;
aux_extension_ok := true;
if (token_len < length(s_aux_extension)) then@/
    aux_extension_ok := false   {else |str_eq_buf| might bomb the program}
else if (not str_eq_buf(s_aux_extension, buffer,
        buf_ptr2-length(s_aux_extension), length(s_aux_extension))) then
    aux_extension_ok := false;
if (not aux_extension_ok) then
    begin
    print_token;
    print (' has a wrong extension');
    decr(aux_ptr);
    aux_err_return;
    end;
cur_aux_str := hash_text[
                str_lookup(buffer,buf_ptr1,token_len,aux_file_ilk,do_insert)];
if (hash_found) then
    begin
    print ('Already encountered file ');
    print_aux_name;
    decr(aux_ptr);
    aux_err_return;
    end;
@<Open this \.{.aux} file@>;
end


@
We check that this \.{.aux} file can actually be opened, and then open it.

@<Open this \.{.aux} file@>=
begin
start_name (cur_aux_str);       {extension already there for \.{.aux} files}
name_ptr := name_length+1;
while (name_ptr <= file_name_size) do   {pad with blanks}
    begin
    name_of_file[name_ptr] := ' ';
    incr(name_ptr);
    end;
if (not a_open_in(cur_aux_file)) then
    begin
    print ('I couldn''t open auxiliary file ');
    print_aux_name;
    decr(aux_ptr);
    aux_err_return;
    end;
print ('A level-',aux_ptr:0,' auxiliary file: ');
print_aux_name;
cur_aux_line := 0;
end


@
Here we close the current-level \.{.aux} file and go back up a level,
if possible, by decrementing |aux_ptr|.

@<Procedures and functions for the reading and processing of input files@>=
procedure pop_the_aux_stack;
begin
a_close (cur_aux_file);
if (aux_ptr=0) then
    goto aux_done
  else
    decr(aux_ptr);
end;


@
@^gymnastics@>
That's it for processing \.{.aux} commands, except for finishing the
procedural gymnastics.

@<Procedures and functions for the reading and processing of input files@>=
@<Scan for and process an \.{.aux} command@>


@
We must complain if anything's amiss.

@d aux_end_err(#) == begin
                     aux_end1_err_print;
                     print (#);
                     aux_end2_err_print;
                     end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure aux_end1_err_print;
begin
print ('I found no ');
end;
@#
procedure aux_end2_err_print;
begin
print ('---while reading file ');
print_aux_name;
mark_error;
end;


@
Before proceeding, we see if we have any complaints.

@<Procedures and functions for the reading and processing of input files@>=
procedure last_check_for_aux_errors;
begin
num_cites := cite_ptr;          {record the number of distinct cite keys}
num_bib_files := bib_ptr;       {and the number of \.{.bib} files}
if (not citation_seen) then
    aux_end_err ('\citation commands')
  else if ((num_cites = 0) and (not all_entries)) then
    aux_end_err ('cite keys');
if (not bib_seen) then
    aux_end_err ('\bibdata command')
  else if (num_bib_files = 0) then
    aux_end_err ('database files');
if (not bst_seen) then
    aux_end_err ('\bibstyle command')
  else if (bst_str = 0) then
    aux_end_err ('style file');
end;



@* Reading the style file.
This part of the program reads the \.{.bst} file, which consists of a
sequence of commands.  Each \.{.bst} command consists of a name (for
which case differences are ignored) followed by zero or more
arguments, each enclosed in braces.

@d bst_done=32          {go here when finished with the \.{.bst} file}
@d no_bst_file=9932     {go here when skipping the \.{.bst} file}

@<Labels in the outer block@>=
,@!bst_done,@!no_bst_file


@
The |bbl_line_num| gets initialized along with the |bst_line_num|, so
it's declared here too.

@<Globals in the outer block@>=
@!bbl_line_num : integer;       {line number of the \.{.bbl} (output) file}
@!bst_line_num : integer;       {line number of the \.{.bst} file}


@
This little procedure exists because it's used by at least two other
procedures and thus saves some space.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_ln_num_print;
begin
print ('--line ',bst_line_num:0,' of file ');
print_bst_name;
end;


@
When there's a serious error parsing the \.{.bst} file, we flush the
rest of the current command; a blank line is assumed to mark the end
of a command (but for the purposes of error recovery only).  Thus,
error recovery will be better if style designers leave blank lines
between \.{.bst} commands.  This macro must be called from within a
procedure that has an |exit| label.

@d bst_err_print_and_look_for_blank_line_return ==
                begin
                bst_err_print_and_look_for_blank_line;
                return;
                end

@d bst_err(#) == begin          {serious error during \.{.bst} parsing}
                 print (#);
                 bst_err_print_and_look_for_blank_line_return;
                 end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_err_print_and_look_for_blank_line;
begin
print ('-');
bst_ln_num_print;
print_bad_input_line;                   {this call does the |mark_error|}
while (last <> 0) do                    {look for a blank input line}
    if (not input_ln(bst_file)) then    {or the end of the file}
        goto bst_done
      else
        incr(bst_line_num);
buf_ptr2 := last;                       {to input the next line}
end;


@
When there's a harmless error parsing the \.{.bst} file (harmless
syntactically, at least) we give just a |warning_message|.

@d bst_warn(#) == begin         {non-serious error during \.{.bst} parsing}
                  print (#);
                  bst_warn_print;
                  end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_warn_print;
begin
bst_ln_num_print;
mark_warning;
end;


@
Here's the outer loop for reading the \.{.bst} file---it keeps reading
and processing \.{.bst} commands until none left.  This is part of the
main program; hence, because of the |bst_done| label, there's no
conventional |begin|-|end| pair surrounding the entire module.

@<Read and execute the \.{.bst} file@>=
if (bst_str = 0) then   {there's no \.{.bst} file to read}
    goto no_bst_file;   {this is a |goto| so that |bst_done| is not in a block}
bst_line_num := 0;      {initialize things}
bbl_line_num := 1;      {best spot to initialize the output line number}
buf_ptr2 := last;       {to get the first input line}
loop
    begin
    if (not eat_bst_white_space) then   {the end of the \.{.bst} file}
        goto bst_done;
    get_bst_command_and_process;
    end;
bst_done: a_close (bst_file);
no_bst_file: a_close (bbl_file);


@
This \.{.bst}-specific scanning function skips over |white_space|
characters (and comments) until hitting a nonwhite character or the
end of the file, respectively returning |true| or |false|.  It also
updates |bst_line_num|, the line counter.

@<Procedures and functions for input scanning@>=
function eat_bst_white_space : boolean;
label exit;
begin
loop
    begin
    if (scan_white_space) then          {hit a nonwhite character on this line}
        if (scan_char <> comment) then  {it's not a comment character; return}
            begin
            eat_bst_white_space := true;
            return;
            end;
    if (not input_ln(bst_file)) then    {end-of-file; return |false|}
        begin
        eat_bst_white_space := false;
        return;
        end;
    incr(bst_line_num);
    buf_ptr2 := 0;
    end;
exit:
end;


@
It's often illegal to end a \.{.bst} command in certain places, and
this is where we come to check.

@d eat_bst_white_and_eof_check(#) ==
        begin
        if (not eat_bst_white_space) then
            begin
            eat_bst_print;
            bst_err (#);
            end;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure eat_bst_print;
begin
print ('Illegal end of style file in command: ');
end;


@
We must attend to a few details before getting to work on this
\.{.bst} command.

@<Scan for and process a \.{.bst} command@>=
procedure get_bst_command_and_process;
label exit;
begin
if (not scan_alpha) then
    bst_err ('"',xchr[scan_char],'" can''t start a style-file command');
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
command_num := ilk_info[
        str_lookup(buffer,buf_ptr1,token_len,bst_command_ilk,dont_insert)];
if (not hash_found) then
    begin
    print_token;
    bst_err (' is an illegal style-file command');
    end;
@<Process the appropriate \.{.bst} command@>;
exit:
end;


@
@^style-file commands@>
@:this can't happen}{\quad Unknown style-file command@>
Here we determine which \.{.bst} command we're about to process, and
then go to it.

@<Process the appropriate \.{.bst} command@>=
case (command_num) of
    n_bst_entry : bst_entry_command;
    n_bst_execute : bst_execute_command;
    n_bst_function : bst_function_command;
    n_bst_integers : bst_integers_command;
    n_bst_iterate : bst_iterate_command;
    n_bst_macro : bst_macro_command;
    n_bst_read : bst_read_command;
    n_bst_reverse : bst_reverse_command;
    n_bst_sort : bst_sort_command;
    n_bst_strings : bst_strings_command;
    othercases confusion ('Unknown style-file command')
endcases


@
We need data structures for the function definitions, the entry
variables, the global variables, and the actual entries corresponding
to the cite-key list.  First we define the classes of `function's
used.  Functions in all classes are of |bst_fn_ilk| except for
|int_literal|s, which are of |integer_ilk|; and |str_literal|s, which
are of |text_ilk|.

@d built_in = 0         {the `primitive' functions}
@d wiz_defined = 1      {defined in the \.{.bst} file}
@d int_literal = 2      {integer `constants'}
@d str_literal = 3      {string `constants'}
@d field = 4            {things like `author' and `title'}
@d int_entry_var = 5    {integer entry variable}
@d str_entry_var = 6    {string entry variable}
@d int_global_var = 7   {integer global variable}
@d str_global_var = 8   {string global variable}
@d last_fn_class = 8    {the same number as on the line above}


@
@:this can't happen}{\quad Unknown function class@>
Here's another bug report.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure unknwn_function_class_confusion;
begin
confusion ('Unknown function class');
end;


@
@:this can't happen}{\quad Unknown function class@>
Occasionally we'll want to |print| the name of one of these function
classes.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_fn_class (@!fn_loc : hash_loc);
begin
case (fn_type[fn_loc]) of
    built_in : print ('built-in');
    wiz_defined : print ('wizard-defined');
    int_literal : print ('integer-literal');
    str_literal : print ('string-literal');
    field : print ('field');
    int_entry_var : print ('integer-entry-variable');
    str_entry_var : print ('string-entry-variable');
    int_global_var : print ('integer-global-variable');
    str_global_var : print ('string-global-variable');
    othercases unknwn_function_class_confusion
endcases;
end;


@
@:this can't happen}{\quad Unknown function class@>
This version is for printing when in |trace| mode.

@<Procedures and functions for all file I/O, error messages, and such@>=
  trace
  procedure trace_pr_fn_class (@!fn_loc : hash_loc);
  begin
  case (fn_type[fn_loc]) of
    built_in : trace_pr ('built-in');
    wiz_defined : trace_pr ('wizard-defined');
    int_literal : trace_pr ('integer-literal');
    str_literal : trace_pr ('string-literal');
    field : trace_pr ('field');
    int_entry_var : trace_pr ('integer-entry-variable');
    str_entry_var : trace_pr ('string-entry-variable');
    int_global_var : trace_pr ('integer-global-variable');
    str_global_var : trace_pr ('string-global-variable');
    othercases unknwn_function_class_confusion
  endcases;
  end;
  ecart


@
Besides the function classes, we have types based on \BibTeX's
capacity limitations and one based on what can go into the array
|wiz_functions| explained below.

@d quote_next_fn = hash_base - 1  {special marker used in defining functions}
@d end_of_def = hash_max + 1      {another such special marker}

@<Types in the outer block@>=
@!fn_class = 0..last_fn_class;          {the \.{.bst} function classes}
@!wiz_fn_loc = 0..wiz_fn_space;  {|wiz_defined|-function storage locations}
@!int_ent_loc = 0..max_ent_ints;        {|int_entry_var| storage locations}
@!str_ent_loc = 0..max_ent_strs;        {|str_entry_var| storage locations}
@!str_glob_loc = 0..max_glb_str_minus_1; {|str_global_var| storage locations}
@!field_loc = 0..max_fields;            {individual field storage locations}
@!hash_ptr2 = quote_next_fn..end_of_def; {a special marker or a |hash_loc|}


@
@^save space@>
@^space savings@>
@^system dependencies@>
We store information about the \.{.bst} functions in arrays the same
size as the hash-table arrays and in locations corresponding to their
hash-table locations.  The two arrays |fn_info| (an alias of
|ilk_info| described earlier) and |fn_type| accomplish this: |fn_type|
specifies one of the above classes, and |fn_info| gives information
dependent on the class.

Six other arrays give the contents of functions: The array
|wiz_functions| holds definitions for |wiz_defined| functions---each
such function consists of a sequence of pointers to hash-table
locations of other functions (with the two special-marker exceptions
above); the array |entry_ints| contains the current values of
|int_entry_var|s; the array |entry_strs| contains the current values
of |str_entry_var|s; an element of the array |global_strs| contains
the current value of a |str_global_var| if the corresponding
|glb_str_ptr| entry is empty, otherwise the nonempty entry is a
pointer to the string; and the array |field_info|, for each field of
each entry, contains either a pointer to the string or the special
value |missing|.

The array |global_strs| isn't packed (that is, it isn't |array| \dots\
|of packed array| \dots$\,$) to increase speed on some systems;
however, on systems that are byte-addressable and that have a good
compiler, packing |global_strs| would save lots of space without much
loss of speed.

@d fn_info == ilk_info          {an alias used with functions}
@#
@d missing = empty              {a special pointer for missing fields}

@<Globals in the outer block@>=
@!fn_loc : hash_loc;            {the hash-table location of a function}
@!wiz_loc : hash_loc;           {the hash-table location of a wizard function}
@!literal_loc : hash_loc;       {the hash-table location of a literal function}
@!macro_name_loc : hash_loc;    {the hash-table location of a macro name}
@!macro_def_loc : hash_loc;     {the hash-table location of a macro definition}
@!fn_type : packed array[hash_loc] of fn_class;
@!wiz_def_ptr : wiz_fn_loc;     {storage location for the next wizard function}
@!wiz_fn_ptr : wiz_fn_loc;      {general |wiz_functions| location}
@!wiz_functions : packed array[wiz_fn_loc] of hash_ptr2;
@!int_ent_ptr : int_ent_loc;    {general |int_entry_var| location}
@!entry_ints : array[int_ent_loc] of integer;
@!num_ent_ints : int_ent_loc;   {the number of distinct |int_entry_var| names}
@!str_ent_ptr : str_ent_loc;    {general |str_entry_var| location}
@!entry_strs : array[str_ent_loc] of
                                packed array[0..ent_str_size] of ASCII_code;
@!num_ent_strs : str_ent_loc;   {the number of distinct |str_entry_var| names}
@!str_glb_ptr : 0..max_glob_strs;       {general |str_global_var| location}
@!glb_str_ptr : array[str_glob_loc] of str_number;
@!global_strs : array[str_glob_loc] of array[0..glob_str_size] of ASCII_code;
@!glb_str_end : array[str_glob_loc] of 0..glob_str_size;        {end markers}
@!num_glb_strs : 0..max_glob_strs; {number of distinct |str_global_var| names}
@!field_ptr : field_loc;        {general |field_info| location}
@!field_parent_ptr,@!field_end_ptr : field_loc; {two more for doing cross-refs}
@!cite_parent_ptr,@!cite_xptr : cite_number;  {two others for doing cross-refs}
@!field_info : packed array[field_loc] of str_number;
@!num_fields : field_loc;       {the number of distinct field names}
@!num_pre_defined_fields : field_loc;   {so far, just one: \.{crossref}}
@!crossref_num : field_loc;     {the number given to \.{crossref}}
@!no_fields : boolean;          {used for |tr_print|ing entry information}


@
Now we initialize storage for the |wiz_defined| functions and we
initialize variables so that the first |str_entry_var|,
|int_entry_var|, |str_global_var|, and |field| name will be assigned
the number~0.  Note: The variables |num_ent_strs| and |num_fields|
will also be set when pre-defining strings.

@<Set initial values of key variables@>=
wiz_def_ptr := 0;
num_ent_ints := 0;
num_ent_strs := 0;
num_fields := 0;
str_glb_ptr := 0;
while (str_glb_ptr < max_glob_strs) do          {make |str_global_var|s empty}
    begin
    glb_str_ptr[str_glb_ptr] := 0;
    glb_str_end[str_glb_ptr] := 0;
    incr(str_glb_ptr);
    end;
num_glb_strs := 0;



@* Style-file commands.
@^style-file commands@>
There are ten \.{.bst} commands: Five (\.{entry}, \.{function},
\.{integers}, \.{macro}, and \.{strings}) declare and define
functions, one (\.{read}) reads in the \.{.bib}-file entries, and four
(\.{execute}, \.{iterate}, \.{reverse}, and \.{sort})
manipulate the entries and produce output.

The boolean variables |entry_seen| and |read_seen| indicate whether
we've yet encountered an \.{entry} and a \.{read} command.  There must
be exactly one of each of these, and the \.{entry} command, as well as
any \.{macro} command, must precede the \.{read} command.
Furthermore, the \.{read} command must precede the four that
manipulate the entries and produce output.

@<Globals in the outer block@>=
@!entry_seen : boolean; {|true| if we've already seen an \.{entry} command}
@!read_seen : boolean;  {|true| if we've already seen a \.{read} command}
@!read_performed : boolean; {|true| if we started reading the database file(s)}
@!reading_completed : boolean; {|true| if we made it all the way through}
@!read_completed : boolean; {|true| if the database info didn't bomb \BibTeX}


@
And we initialize them.

@<Set initial values of key variables@>=
entry_seen := false;
read_seen := false;
read_performed := false;
reading_completed := false;
read_completed := false;


@
@:this can't happen}{\quad Identifier scanning error@>
Here's another bug.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure id_scanning_confusion;
begin
confusion ('Identifier scanning error');
end;


@
@:this can't happen}{\quad Identifier scanning error@>
This macro is used to scan all \.{.bst} identifiers.  The argument
supplies the \.{.bst} command name.  The associated procedure simply
prints an error message.

@d bst_identifier_scan(#) ==
        begin
        scan_identifier (right_brace,comment,comment);
        if ((scan_result = white_adjacent) or
                                (scan_result = specified_char_adjacent)) then
            do_nothing
        else
            begin
            bst_id_print;
            bst_err (#);
            end;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_id_print;
begin
if (scan_result = id_null) then
    print ('"',xchr[scan_char],'" begins identifier, command: ')
else if (scan_result = other_char_adjacent) then
    print ('"',xchr[scan_char],'" immediately follows identifier, command: ')
else
    id_scanning_confusion;
end;


@
This macro just makes sure we're at a |left_brace|.

@d bst_get_and_check_left_brace(#) ==
        begin
        if (scan_char <> left_brace) then
            begin
            bst_left_brace_print;
            bst_err (#);
            end;
        incr(buf_ptr2);                 {skip over the |left_brace|}
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_left_brace_print;
begin
print ('"',xchr[left_brace],'" is missing in command: ');
end;


@
And this one, a |right_brace|.

@d bst_get_and_check_right_brace(#) ==
        begin
        if (scan_char <> right_brace) then
            begin
            bst_right_brace_print;
            bst_err (#);
            end;
        incr(buf_ptr2);                 {skip over the |right_brace|}
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_right_brace_print;
begin
print ('"',xchr[right_brace],'" is missing in command: ');
end;


@
This macro complains if we've already encountered a function to be
inserted into the hash table.

@d check_for_already_seen_function(#) ==
        begin
        if (hash_found) then  {already encountered this as a \.{.bst} function}
            begin
            already_seen_function_print (#);
            return;
            end;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure already_seen_function_print (@!seen_fn_loc : hash_loc);
label exit;     {so the call to |bst_err| works}
begin
print_pool_str (hash_text[seen_fn_loc]);
print (' is already a type "');
print_fn_class (seen_fn_loc);
print_ln ('" function name');
bst_err_print_and_look_for_blank_line_return;
exit:
end;


@
@:style-file commands}{\quad \.{entry}@>
An \.{entry} command has three arguments, each a (possibly empty) list
of function names between braces (the names are separated by one or
more |white_space| characters).  All function names in this and other
commands must be legal \.{.bst} identifiers.  Upper/lower cases are
considered to be the same for function names in these lists---all
upper-case letters are converted to lower case.  These arguments give
lists of |field|s, |int_entry_var|s, and |str_entry_var|s.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_entry_command;
label exit;
begin
if (entry_seen) then
    bst_err ('Illegal, another entry command');
entry_seen := true;             {now we've seen an \.{entry} command}
eat_bst_white_and_eof_check ('entry');
@<Scan the list of |field|s@>;
eat_bst_white_and_eof_check ('entry');
if (num_fields = num_pre_defined_fields) then
    bst_warn ('Warning--I didn''t find any fields');
@<Scan the list of |int_entry_var|s@>;
eat_bst_white_and_eof_check ('entry');
@<Scan the list of |str_entry_var|s@>;
exit:
end;


@
This module reads a |left_brace|, the list of |field|s, and a
|right_brace|.  The |field|s are those like `author' and `title.'

@<Scan the list of |field|s@>=
begin
bst_get_and_check_left_brace ('entry');
eat_bst_white_and_eof_check ('entry');
while (scan_char <> right_brace) do
    begin
    bst_identifier_scan ('entry');
    @<Insert a |field| into the hash table@>;
    eat_bst_white_and_eof_check ('entry');
    end;
incr(buf_ptr2);                 {skip over the |right_brace|}
end


@
@^secret agent man@>
Here we insert the just found field name into the hash table, record
it as a |field|, and assign it a number to be used in indexing into
the |field_info| array.

@<Insert a |field| into the hash table@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a field');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,do_insert);
check_for_already_seen_function (fn_loc);
fn_type[fn_loc] := field;@/
fn_info[fn_loc] := num_fields;  {give this field a number (take away its name)}
incr(num_fields);
end


@
This module reads a |left_brace|, the list of |int_entry_var|s,
and a |right_brace|.

@<Scan the list of |int_entry_var|s@>=
begin
bst_get_and_check_left_brace ('entry');
eat_bst_white_and_eof_check ('entry');
while (scan_char <> right_brace) do
    begin
    bst_identifier_scan ('entry');
    @<Insert an |int_entry_var| into the hash table@>;
    eat_bst_white_and_eof_check ('entry');
    end;
incr(buf_ptr2);                 {skip over the |right_brace|}
end


@
Here we insert the just found |int_entry_var| name into the hash table
and record it as an |int_entry_var|.  An |int_entry_var| is one that
the style designer wants a separate copy of for each entry.

@<Insert an |int_entry_var| into the hash table@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is an integer entry-variable');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,do_insert);
check_for_already_seen_function (fn_loc);
fn_type[fn_loc] := int_entry_var;@/
fn_info[fn_loc] := num_ent_ints;        {give this |int_entry_var| a number}
incr(num_ent_ints);
end


@
This module reads a |left_brace|, the list of |str_entry_var|s, and a
|right_brace|.  A |str_entry_var| is one that the style designer wants
a separate copy of for each entry.

@<Scan the list of |str_entry_var|s@>=
begin
bst_get_and_check_left_brace ('entry');
eat_bst_white_and_eof_check ('entry');
while (scan_char <> right_brace) do
    begin
    bst_identifier_scan ('entry');
    @<Insert a |str_entry_var| into the hash table@>;
    eat_bst_white_and_eof_check ('entry');
    end;
incr(buf_ptr2);                 {skip over the |right_brace|}
end


@
Here we insert the just found |str_entry_var| name into the hash
table, record it as a |str_entry_var|, and set its pointer into
|entry_strs|.

@<Insert a |str_entry_var| into the hash table@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a string entry-variable');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,do_insert);
check_for_already_seen_function (fn_loc);
fn_type[fn_loc] := str_entry_var;@/
fn_info[fn_loc] := num_ent_strs;        {give this |str_entry_var| a number}
incr(num_ent_strs);
end


@
A legal argument for an \.{execute}, \.{iterate}, or \.{reverse}
command must exist and be |built_in| or |wiz_defined|.
Here's where we check, returning |true| if the argument is illegal.

@<Procedures and functions for the reading and processing of input files@>=
function bad_argument_token : boolean;
label exit;
begin
bad_argument_token := true;     {now it's easy to exit if necessary}
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,dont_insert);
if (not hash_found) then                        {unknown \.{.bst} function}
    begin
    print_token;
    bst_err (' is an unknown function');
    end
else if ((fn_type[fn_loc] <> built_in) and
         (fn_type[fn_loc] <> wiz_defined)) then
    begin
    print_token;
    print (' has bad function type ');
    print_fn_class (fn_loc);
    bst_err_print_and_look_for_blank_line_return;
    end;
bad_argument_token := false;
exit:
end;


@
@:style-file commands}{\quad \.{execute}@>
An \.{execute} command has one argument, a single |built_in| or
|wiz_defined| function name between braces.  Upper/lower cases are
considered to be the same---all upper-case letters are converted to
lower case.  Also, we must make sure we've already seen a \.{read}
command.

This module reads a |left_brace|, a single function to be executed,
and a |right_brace|.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_execute_command;
label exit;
begin
if (not read_seen) then
    bst_err ('Illegal, execute command before read command');
eat_bst_white_and_eof_check ('execute');
bst_get_and_check_left_brace ('execute');
eat_bst_white_and_eof_check ('execute');
bst_identifier_scan ('execute');
@<Check the \.{execute}-command argument token@>;
eat_bst_white_and_eof_check ('execute');
bst_get_and_check_right_brace ('execute');
@<Perform an \.{execute} command@>;
exit:
end;


@
Before executing the function, we must make sure it's a legal one.  It
must exist and be |built_in| or |wiz_defined|.

@<Check the \.{execute}-command argument token@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a to be executed function');
  ecart@/
if (bad_argument_token) then
    return;
end


@
@:style-file commands}{\quad \.{function}@>
A \.{function} command has two arguments; the first is a
|wiz_defined| function name between braces.  Upper/lower cases are
considered to be the same---all upper-case letters are converted to
lower case.  The second argument defines this function.  It consists
of a sequence of functions, between braces, separated by |white_space|
characters.  Upper/lower cases are considered to be the same for
function names but not for |str_literal|s.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_function_command;
label exit;
begin
eat_bst_white_and_eof_check ('function');
@<Scan the |wiz_defined| function name@>;
eat_bst_white_and_eof_check ('function');
bst_get_and_check_left_brace ('function');
scan_fn_def(wiz_loc);           {this scans the function definition}
exit:
end;


@
This module reads a |left_brace|, a |wiz_defined| function name, and
a |right_brace|.

@<Scan the |wiz_defined| function name@>=
begin
bst_get_and_check_left_brace ('function');
eat_bst_white_and_eof_check ('function');
bst_identifier_scan ('function');
@<Check the |wiz_defined| function name@>;
eat_bst_white_and_eof_check ('function');
bst_get_and_check_right_brace ('function');
end


@
The function name must exist and be a new one; we mark it as
|wiz_defined|.  Also, see if it's the default entry-type function.

@<Check the |wiz_defined| function name@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a wizard-defined function');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
wiz_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,do_insert);
check_for_already_seen_function (wiz_loc);
fn_type[wiz_loc] := wiz_defined;
if (hash_text[wiz_loc] = s_default) then  {we've found the default entry-type}
    b_default := wiz_loc;       {see the |built_in| functions for |b_default|}
end


@
We're about to start scanning tokens in a function definition.  When a
function token is illegal, we skip until it ends; a |white_space|
character, an end-of-line, a |right_brace|, or a |comment| marks the
end of the current token.

@d next_token=25                {a bad function token; go read the next one}
@#
@d skip_token(#) == begin       {not-so-serious error during \.{.bst} parsing}
                    print (#);
                    skip_token_print;   {also, skip to the current token's end}
                    goto next_token;
                    end

@<Procedures and functions for input scanning@>=
procedure skip_token_print;
begin
print ('-');
bst_ln_num_print;
mark_error;
if (scan2_white(right_brace,comment)) then              {ok if token ends line}
    do_nothing;
end;


@
@^commented-out code@>
@^for a good time, try comment-out code@>
This macro is similar to the last one but is specifically for
recursion in a |wiz_defined| function, which is illegal; it helps save
space.

@d skip_recursive_token == begin
                           print_recursion_illegal;
                           goto next_token;
                           end

@<Procedures and functions for input scanning@>=
procedure print_recursion_illegal;
begin
  trace
  trace_pr_newline;
  ecart@/
print_ln ('Curse you, wizard, before you recurse me:');
print ('function ');
print_token;
print_ln (' is illegal in its own definition');
  @{
  print_recursion_illegal;
  @}@/
skip_token_print;                       {also, skip to the current token's end}
end;


@
Here's another macro for saving some space when there's a problem with
a token.

@d skip_token_unknown_function == begin
                                  skp_token_unknown_function_print;
                                  goto next_token;
                                  end

@<Procedures and functions for input scanning@>=
procedure skp_token_unknown_function_print;
begin
print_token;
print (' is an unknown function');
skip_token_print;                       {also, skip to the current token's end}
end;


@
And another.

@d skip_token_illegal_stuff_after_literal ==
                        begin
                        skip_illegal_stuff_after_token_print;
                        goto next_token;
                        end

@<Procedures and functions for input scanning@>=
procedure skip_illegal_stuff_after_token_print;
begin
print ('"',xchr[scan_char],'" can''t follow a literal');
skip_token_print;                       {also, skip to the current token's end}
end;


@
This recursive function reads and stores the list of functions
(separated by |white_space| characters or ends-of-line) that define
this new function, and reads a |right_brace|.

@<Procedures and functions for input scanning@>=
procedure scan_fn_def (@!fn_hash_loc : hash_loc);
label next_token,@!exit;
type @!fn_def_loc = 0..single_fn_space; {for a single |wiz_defined|-function}
var singl_function : packed array[fn_def_loc] of hash_ptr2;
    @!single_ptr : fn_def_loc;  {next storage location for this definition}
    @!copy_ptr : fn_def_loc;    {dummy variable}
    @!end_of_num : buf_pointer; {the end of an implicit function's name}
    @!impl_fn_loc : hash_loc;   {an implicit function's hash-table location}
begin
eat_bst_white_and_eof_check ('function');
single_ptr := 0;
while (scan_char <> right_brace) do
    begin
    @<Get the next function of the definition@>;
next_token:
    eat_bst_white_and_eof_check ('function');
    end;
@<Complete this function's definition@>;
incr(buf_ptr2);                 {skip over the |right_brace|}
exit:
end;


@
@:BibTeX capacity exceeded}{\quad single function space@>
This macro inserts a hash-table location (or one of the two
special markers |quote_next_fn| and |end_of_def|) into the
|singl_function| array, which will later be copied into the
|wiz_functions| array.

@d insert_fn_loc(#) ==  begin
                        singl_function[single_ptr] := #;
                        if (single_ptr = single_fn_space) then
                            singl_fn_overflow;
                        incr(single_ptr);
                        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure singl_fn_overflow;
begin
overflow('single function space ',single_fn_space);
end;


@
There are five possibilities for the first character of the token
representing the next function of the definition: If it's a
|number_sign|, the token is an |int_literal|; if it's a
|double_quote|, the token is a |str_literal|; if it's a
|single_quote|, the token is a quoted function; if it's a
|left_brace|, the token isn't really a token, but rather the start of
another function definition (which will result in a recursive call to
|scan_fn_def|); if it's anything else, the token is the name of an
already-defined function.  Note: To prevent the wizard from using
recursion, we have to check that neither a quoted function nor an
already-defined-function is actually the currently-being-defined
function (which is stored at |wiz_loc|).

@<Get the next function of the definition@>=
case (scan_char) of
    number_sign : @<Scan an |int_literal|@>;
    double_quote : @<Scan a |str_literal|@>;
    single_quote : @<Scan a quoted function@>;
    left_brace : @<Start a new function definition@>;
    othercases @<Scan an already-defined function@>
endcases


@
An |int_literal| is preceded by a |number_sign|, consists of an
integer (i.e., an optional |minus_sign| followed by one or more
|numeric| characters), and is followed either by a |white_space|
character, an end-of-line, or a |right_brace|.  The array |fn_info|
contains the value of the integer for |int_literal|s.

@<Scan an |int_literal|@>=
begin
incr(buf_ptr2);                         {skip over the |number_sign|}
if (not scan_integer) then
    skip_token ('Illegal integer in integer literal');
  trace
  trace_pr ('#');
  trace_pr_token;
  trace_pr_ln (' is an integer literal with value ',token_value:0);
  ecart@/
literal_loc := str_lookup(buffer,buf_ptr1,token_len,integer_ilk,do_insert);
if (not hash_found) then
    begin
    fn_type[literal_loc] := int_literal;        {set the |fn_class|}
    fn_info[literal_loc] := token_value;        {the value of this integer}
    end;
if ((lex_class[scan_char]<>white_space) and (buf_ptr2<last) and
            (scan_char<>right_brace) and@| (scan_char<>comment)) then
    skip_token_illegal_stuff_after_literal;
insert_fn_loc (literal_loc);    {add this function to |wiz_functions|}
end


@
A |str_literal| is preceded by a |double_quote| and consists of all
characters on this line up to the next |double_quote|.  Also, there
must be either a |white_space| character, an end-of-line, a
|right_brace|, or a |comment| following (since functions in the
definition must be separated by |white_space|).  The array |fn_info|
contains nothing for |str_literal|s.

@<Scan a |str_literal|@>=
begin
incr(buf_ptr2);                         {skip over the |double_quote|}
if (not scan1(double_quote)) then
    skip_token ('No `',xchr[double_quote],''' to end string literal');
  trace
  trace_pr ('"');
  trace_pr_token;
  trace_pr ('"');
  trace_pr_ln (' is a string literal');
  ecart@/
literal_loc := str_lookup(buffer,buf_ptr1,token_len,text_ilk,do_insert);@/
fn_type[literal_loc] := str_literal;    {set the |fn_class|}
incr(buf_ptr2);                         {skip over the |double_quote|}
if ((lex_class[scan_char]<>white_space) and (buf_ptr2<last) and
        (scan_char<>right_brace) and@| (scan_char<>comment)) then
    skip_token_illegal_stuff_after_literal;
insert_fn_loc (literal_loc);            {add this function to |wiz_functions|}
end


@
A quoted function is preceded by a |single_quote| and consists of all
characters up to the next |white_space| character, end-of-line,
|right_brace|, or |comment|.

@<Scan a quoted function@>=
begin
incr(buf_ptr2);                                 {skip over the |single_quote|}
if (scan2_white(right_brace,comment)) then              {ok if token ends line}
    do_nothing;
  trace
  trace_pr ('''');
  trace_pr_token;
  trace_pr (' is a quoted function ');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,dont_insert);
if (not hash_found) then                        {unknown \.{.bst} function}
    skip_token_unknown_function
else
    @<Check and insert the quoted function@>;
end


@
Here we check that this quoted function is a legal one---the function
name must already exist, but it mustn't be the currently-being-defined
function (which is stored at |wiz_loc|).

@<Check and insert the quoted function@>=
begin
if (fn_loc = wiz_loc) then
    skip_recursive_token
else
    begin
      trace
      trace_pr ('of type ');
      trace_pr_fn_class (fn_loc);
      trace_pr_newline;
      ecart@/
    insert_fn_loc (quote_next_fn);      {add special marker together with}
    insert_fn_loc (fn_loc);             {this function to |wiz_functions|}
    end
end


@
@^kludge@>
@:this can't happen}{\quad Already encountered implicit function@>
This module marks the implicit function as being quoted, generates a
name, and stores it in the hash table.  This name is strictly internal
to this program, starts with a |single_quote| (since that will make
this function name unique), and ends with the variable |impl_fn_num|
converted to ASCII.  The alias kludge helps make the stack space not
overflow on some machines.

@d ex_buf2 == ex_buf            {an alias, used only in this module}

@<Start a new function definition@>=
begin
ex_buf2[0] := single_quote;
int_to_ASCII (impl_fn_num,ex_buf2,1,end_of_num);
impl_fn_loc := str_lookup(ex_buf2,0,end_of_num,bst_fn_ilk,do_insert);
if (hash_found) then
    confusion ('Already encountered implicit function');
  trace
  trace_pr_pool_str (hash_text[impl_fn_loc]);
  trace_pr_ln (' is an implicit function');
  ecart@/
incr(impl_fn_num);
fn_type[impl_fn_loc] := wiz_defined;@/
insert_fn_loc (quote_next_fn);  {all implicit functions are quoted}
insert_fn_loc (impl_fn_loc);    {add it to |wiz_functions|}
incr(buf_ptr2);                 {skip over the |left_brace|}
scan_fn_def (impl_fn_loc);      {this is the recursive call}
end


@
The variable |impl_fn_num| counts the number of implicit functions
seen in the \.{.bst} file.

@<Globals in the outer block@>=
@!impl_fn_num : integer;        {the number of implicit functions seen so far}


@
Now we initialize it.

@<Set initial values of key variables@>=
impl_fn_num := 0;


@
@:BibTeX capacity exceeded}{\quad buffer size@>
This module appends a character to |int_buf| after checking to make
sure it will fit; for use in |int_to_ASCII|.

@d append_int_char(#) == begin
                         if (int_ptr = buf_size) then
                             buffer_overflow;
                         int_buf[int_ptr]:=#;
                         incr(int_ptr);
                         end


@
This procedure takes the integer |int|, copies the appropriate
|ASCII_code| string into |int_buf| starting at |int_begin|, and sets
the |var| parameter |int_end| to the first unused |int_buf| location.
The ASCII string will consist of decimal digits, the first of which
will be not be a~0 if the integer is nonzero, with a prepended minus
sign if the integer is negative.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure int_to_ASCII (@!int:integer; var int_buf:buf_type;
                        @!int_begin:buf_pointer; var int_end:buf_pointer);
var int_ptr,@!int_xptr : buf_pointer;   {pointers into |int_buf|}
  @!int_tmp_val : ASCII_code;           {the temporary element in an exchange}
begin
int_ptr := int_begin;
if (int < 0) then       {add the |minus_sign| and use the absolute value}
    begin
    append_int_char (minus_sign);
    int := -int;
    end;
int_xptr := int_ptr;
repeat                          {copy digits into |int_buf|}
    append_int_char ("0" + (int mod 10));
    int := int div 10;
  until (int = 0);
int_end := int_ptr;             {set the string length}
decr(int_ptr);
while (int_xptr < int_ptr) do   {and reorder (flip) the digits}
    begin
    int_tmp_val := int_buf[int_xptr];
    int_buf[int_xptr] := int_buf[int_ptr];
    int_buf[int_ptr] := int_tmp_val;
    decr(int_ptr);
    incr(int_xptr);
    end
end;


@
An already-defined function consists of all characters up to the next
|white_space| character, end-of-line, |right_brace|, or |comment|.
This function name must already exist, but it mustn't be the
currently-being-defined function (which is stored at |wiz_loc|).

@<Scan an already-defined function@>=
begin
if (scan2_white(right_brace,comment)) then              {ok if token ends line}
    do_nothing;
  trace
  trace_pr_token;
  trace_pr (' is a function ');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,dont_insert);
if (not hash_found) then                        {unknown \.{.bst} function}
    skip_token_unknown_function
else if (fn_loc = wiz_loc) then
    skip_recursive_token
else
    begin
      trace
      trace_pr ('of type ');
      trace_pr_fn_class (fn_loc);
      trace_pr_newline;
      ecart@/
    insert_fn_loc (fn_loc);     {add this function to |wiz_functions|}
    end;
end


@
@:BibTeX capacity exceeded}{\quad wizard-defined function space@>
Now we add the |end_of_def| special marker, make sure this function will
fit into |wiz_functions|, and put it there.

@<Complete this function's definition@>=
begin
insert_fn_loc (end_of_def);  {add special marker ending the definition}
if (single_ptr + wiz_def_ptr > wiz_fn_space) then
    begin
    print (single_ptr + wiz_def_ptr : 0,': ');
    overflow('wizard-defined function space ',wiz_fn_space);
    end;
fn_info[fn_hash_loc] := wiz_def_ptr;            {pointer into |wiz_functions|}
copy_ptr := 0;
while (copy_ptr < single_ptr) do                {make this function official}
    begin
    wiz_functions[wiz_def_ptr] := singl_function[copy_ptr];
    incr(copy_ptr);
    incr(wiz_def_ptr);
    end;
end


@
@:style-file commands}{\quad \.{integers}@>
An \.{integers} command has one argument, a list of function names
between braces (the names are separated by one or more |white_space|
characters).  Upper/lower cases are considered to be the same for
function names in these lists---all upper-case letters are converted to
lower case.  Each name in this list specifies an |int_global_var|.
There may be several \.{integers} commands in the \.{.bst} file.

This module reads a |left_brace|, a list of |int_global_var|s, and a
|right_brace|.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_integers_command;
label exit;
begin
eat_bst_white_and_eof_check ('integers');
bst_get_and_check_left_brace ('integers');
eat_bst_white_and_eof_check ('integers');
while (scan_char <> right_brace) do
    begin
    bst_identifier_scan ('integers');
    @<Insert an |int_global_var| into the hash table@>;
    eat_bst_white_and_eof_check ('integers');
    end;
incr(buf_ptr2);                 {skip over the |right_brace|}
exit:
end;


@
Here we insert the just found |int_global_var| name into the hash
table and record it as an |int_global_var|.  Also, we initialize it by
setting |fn_info[fn_loc]| to 0.

@<Insert an |int_global_var| into the hash table@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is an integer global-variable');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,do_insert);
check_for_already_seen_function (fn_loc);
fn_type[fn_loc] := int_global_var;@/
fn_info[fn_loc] := 0;                           {initialize}
end


@
@:style-file commands}{\quad \.{iterate}@>
An \.{iterate} command has one argument, a single |built_in| or
|wiz_defined| function name between braces.  Upper/lower cases are
considered to be the same---all upper-case letters are converted to
lower case.  Also, we must make sure we've already seen a \.{read}
command.

This module reads a |left_brace|, a single function to be iterated,
and a |right_brace|.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_iterate_command;
label exit;
begin
if (not read_seen) then
    bst_err ('Illegal, iterate command before read command');
eat_bst_white_and_eof_check ('iterate');
bst_get_and_check_left_brace ('iterate');
eat_bst_white_and_eof_check ('iterate');
bst_identifier_scan ('iterate');
@<Check the \.{iterate}-command argument token@>;
eat_bst_white_and_eof_check ('iterate');
bst_get_and_check_right_brace ('iterate');
@<Perform an \.{iterate} command@>;
exit:
end;


@
Before iterating the function, we must make sure it's a legal one.  It
must exist and be |built_in| or |wiz_defined|.

@<Check the \.{iterate}-command argument token@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a to be iterated function');
  ecart@/
if (bad_argument_token) then
    return;
end


@
@:style-file commands}{\quad \.{macro}@>
A \.{macro} command, like a \.{function} command, has two arguments;
the first is a macro name between braces.  The name must be a legal
\.{.bst} identifier.  Upper/lower cases are considered to be the
same---all upper-case letters are converted to lower case.  The second
argument defines this macro.  It consists of a
|double_quote|-delimited string (which must be on a single line)
between braces, with optional |white_space| characters between the
braces and the |double_quote|s.  This |double_quote|-delimited string
is parsed exactly as a |str_literal| is for the \.{function} command.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_macro_command;
label exit;
begin
if (read_seen) then
    bst_err ('Illegal, macro command after read command');
eat_bst_white_and_eof_check ('macro');
@<Scan the macro name@>;
eat_bst_white_and_eof_check ('macro');
@<Scan the macro's definition@>;
exit:
end;


@
This module reads a |left_brace|, a macro name, and a |right_brace|.

@<Scan the macro name@>=
begin
bst_get_and_check_left_brace ('macro');
eat_bst_white_and_eof_check ('macro');
bst_identifier_scan ('macro');
@<Check the macro name@>;
eat_bst_white_and_eof_check ('macro');
bst_get_and_check_right_brace ('macro');
end


@
The macro name must be a new one; we mark it as |macro_ilk|.

@<Check the macro name@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a macro');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
macro_name_loc := str_lookup(buffer,buf_ptr1,token_len,macro_ilk,do_insert);
if (hash_found) then
    begin
    print_token;
    bst_err (' is already defined as a macro');
    end;
ilk_info[macro_name_loc]:=hash_text[macro_name_loc]; {default in case of error}
end


@
This module reads a |left_brace|, the |double_quote|-delimited string
that defines this macro, and a |right_brace|.

@<Scan the macro's definition@>=
begin
bst_get_and_check_left_brace ('macro');
eat_bst_white_and_eof_check ('macro');
if (scan_char <> double_quote) then
    bst_err ('A macro definition must be ',xchr[double_quote],'-delimited');
@<Scan the macro definition-string@>;
eat_bst_white_and_eof_check ('macro');
bst_get_and_check_right_brace ('macro');
end


@
A macro definition-string is preceded by a |double_quote| and consists
of all characters on this line up to the next |double_quote|.  The
array |ilk_info| contains a pointer to this string for the macro name.

@<Scan the macro definition-string@>=
begin
incr(buf_ptr2);                         {skip over the |double_quote|}
if (not scan1(double_quote)) then
    bst_err ('There''s no `',xchr[double_quote],''' to end macro definition');
  trace
  trace_pr ('"');
  trace_pr_token;
  trace_pr ('"');
  trace_pr_ln (' is a macro string');
  ecart@/
macro_def_loc := str_lookup(buffer,buf_ptr1,token_len,text_ilk,do_insert);@/
fn_type[macro_def_loc] := str_literal;  {set the |fn_class|}
ilk_info[macro_name_loc] := hash_text[macro_def_loc];
incr(buf_ptr2);                         {skip over the |double_quote|}
end


@
@^gymnastics@>
We need to include stuff for \.{.bib} reading here because that's done
by the \.{read} command.

@<Procedures and functions for the reading and processing of input files@>=
@<Scan for and process a \.{.bib} command or database entry@>


@
@:style-file commands}{\quad \.{read}@>
The \.{read} command has no arguments so there's no more parsing to
do.  We must make sure we haven't seen a \.{read} command before and
we've already seen an \.{entry} command.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_read_command;
label exit;
begin
if (read_seen) then
    bst_err ('Illegal, another read command');
read_seen := true;              {now we've seen a \.{read} command}
if (not entry_seen) then
    bst_err ('Illegal, read command before entry command');
sv_ptr1 := buf_ptr2;            {save the contents of the \.{.bst} input line}
sv_ptr2 := last;
tmp_ptr := sv_ptr1;
while (tmp_ptr < sv_ptr2) do
    begin
    sv_buffer[tmp_ptr] := buffer[tmp_ptr];
    incr(tmp_ptr);
    end;
@<Read the \.{.bib} file(s)@>;
buf_ptr2 := sv_ptr1;            {and restore}
last := sv_ptr2;
tmp_ptr := buf_ptr2;
while (tmp_ptr < last) do
    begin
    buffer[tmp_ptr] := sv_buffer[tmp_ptr];
    incr(tmp_ptr);
    end;
exit:
end;


@
@:style-file commands}{\quad \.{reverse}@>
A \.{reverse} command has one argument, a single |built_in| or
|wiz_defined| function name between braces.  Upper/lower cases are
considered to be the same---all upper-case letters are converted to
lower case.  Also, we must make sure we've already seen a \.{read}
command.

This module reads a |left_brace|, a single function to be iterated in
reverse, and a |right_brace|.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_reverse_command;
label exit;
begin
if (not read_seen) then
    bst_err ('Illegal, reverse command before read command');
eat_bst_white_and_eof_check ('reverse');
bst_get_and_check_left_brace ('reverse');
eat_bst_white_and_eof_check ('reverse');
bst_identifier_scan ('reverse');
@<Check the \.{reverse}-command argument token@>;
eat_bst_white_and_eof_check ('reverse');
bst_get_and_check_right_brace ('reverse');
@<Perform a \.{reverse} command@>;
exit:
end;


@
Before iterating the function in reverse, we must make sure it's a
legal one.  It must exist and be |built_in| or |wiz_defined|.

@<Check the \.{reverse}-command argument token@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a to be iterated in reverse function');
  ecart@/
if (bad_argument_token) then
    return;
end


@
@:style-file commands}{\quad \.{sort}@>
The \.{sort} command has no arguments so there's no more parsing to
do, but we must make sure we've already seen a \.{read} command.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_sort_command;
label exit;
begin
if (not read_seen) then
    bst_err ('Illegal, sort command before read command');
@<Perform a \.{sort} command@>;
exit:
end;


@
@:style-file commands}{\quad \.{strings}@>
A \.{strings} command has one argument, a list of function names
between braces (the names are separated by one or more |white_space|
characters).  Upper/lower cases are considered to be the same for
function names in these lists---all upper-case letters are converted to
lower case.  Each name in this list specifies a |str_global_var|.
There may be several \.{strings} commands in the \.{.bst} file.

This module reads a |left_brace|, a list of |str_global_var|s,
and a |right_brace|.

@<Procedures and functions for the reading and processing of input files@>=
procedure bst_strings_command;
label exit;
begin
eat_bst_white_and_eof_check ('strings');
bst_get_and_check_left_brace ('strings');
eat_bst_white_and_eof_check ('strings');
while (scan_char <> right_brace) do
    begin
    bst_identifier_scan ('strings');
    @<Insert a |str_global_var| into the hash table@>;
    eat_bst_white_and_eof_check ('strings');
    end;
incr(buf_ptr2);                 {skip over the |right_brace|}
exit:
end;


@
@:BibTeX capacity exceeded}{\quad number of string global-variables@>
Here we insert the just found |str_global_var| name into the hash
table, record it as a |str_global_var|, set its pointer into
|global_strs|, and initialize its value there to the null string.

@d end_of_string = invalid_code  {this illegal |ASCII_code| ends a string}

@<Insert a |str_global_var| into the hash table@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a string global-variable');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
fn_loc := str_lookup(buffer,buf_ptr1,token_len,bst_fn_ilk,do_insert);
check_for_already_seen_function (fn_loc);
fn_type[fn_loc] := str_global_var;@/
fn_info[fn_loc] := num_glb_strs;                {pointer into |global_strs|}
if (num_glb_strs = max_glob_strs) then
    overflow('number of string global-variables ',max_glob_strs);
incr(num_glb_strs);
end


@
@^gymnastics@>
That's it for processing \.{.bst} commands, except for finishing the
procedural gymnastics.  Note that this must topologically follow the
stuff for \.{.bib} reading, because that's done by the \.{.bst}'s
\.{read} command.

@<Procedures and functions for the reading and processing of input files@>=
@<Scan for and process a \.{.bst} command@>



@* Reading the database file(s).
This section reads the \.{.bib} file(s), each of which consists of a
sequence of entries (perhaps with a few \.{.bib} commands thrown in,
as explained later).  Each entry consists of an |at_sign|, an entry
type, and, between braces or parentheses and separated by |comma|s, a
database key and a list of fields.  Each field consists of a field
name, an |equals_sign|, and nonempty list of field tokens separated by
|concat_char|s.  Each field token is either a nonnegative number, a
macro name (like `jan'), or a brace-balanced string delimited by
either |double_quote|s or braces.  Finally, case differences are
ignored for all but delimited strings and database keys, and
|white_space| characters and ends-of-line may appear in all reasonable
places (i.e., anywhere except within entry types, database keys, field
names, and macro names); furthermore, comments may appear anywhere
between entries (or before the first or after the last) as long as
they contain no |at_sign|s.


@
These global variables are used while reading the \.{.bib} file(s).
The elements of |type_list|, which indicate an entry's type (book,
article, etc.), point either to a |hash_loc| or are one of two special
markers: |empty|, from which |hash_base = empty + 1| was defined,
means we haven't yet encountered the \.{.bib} entry corresponding to
this cite key; and |undefined| means we've encountered it but it had
an unknown entry type.  Thus the array |type_list| is of type
|hash_ptr2|, also defined earlier.  An element of the boolean array
|entry_exists| whose corresponding entry in |cite_list| gets
overwritten (which happens only when |all_entries| is |true|)
indicates whether we've encountered that entry of |cite_list| while
reading the \.{.bib} file(s); this information is unused for entries
that aren't (or more precisely, that have no chance of being)
overwritten.  When we're reading the database file, the array
|cite_info| contains auxiliary information for |cite_list|.  Later,
|cite_info| will become |sorted_cites|, and this dual role imposes the
(not-very-imposing) restriction |max_strings >= max_cites|.

@d undefined = hash_max + 1     {a special marker used for |type_list|}

@<Globals in the outer block@>=
@!bib_line_num : integer;       {line number of the \.{.bib} file}
@!entry_type_loc : hash_loc;    {the hash-table location of an entry type}
@!type_list : packed array[cite_number] of hash_ptr2;
@!type_exists : boolean;        {|true| if this entry type is \.{.bst}-defined}
@!entry_exists : packed array[cite_number] of boolean;
@!store_entry : boolean;        {|true| if we're to store info for this entry}
@!field_name_loc : hash_loc;    {the hash-table location of a field name}
@!field_val_loc : hash_loc;     {the hash-table location of a field value}
@!store_field : boolean;        {|true| if we're to store info for this field}
@!store_token : boolean;        {|true| if we're to store this macro token}
@!right_outer_delim : ASCII_code; {either a |right_brace| or a |right_paren|}
@!right_str_delim : ASCII_code; {either a |right_brace| or a |double_quote|}
@!at_bib_command : boolean;     {|true| for a command, false for an entry}
@!cur_macro_loc : hash_loc;     {|macro_loc| for a \.{string} being defined}
@!cite_info : packed array[cite_number] of str_number; {extra |cite_list| info}
@!cite_hash_found : boolean;    {set to a previous |hash_found| value}
@!preamble_ptr : bib_number;    {pointer into the |s_preamble| array}
@!num_preamble_strings : bib_number;    {counts the |s_preamble| strings}


@
This little procedure exists because it's used by at least two other
procedures and thus saves some space.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_ln_num_print;
begin
print ('--line ',bib_line_num:0,' of file ');
print_bib_name;
end;


@
When there's a serious error parsing a \.{.bib} file, we flush
everything up to the beginning of the next entry.

@d bib_err(#) == begin          {serious error during \.{.bib} parsing}
                 print (#);
                 bib_err_print;
                 return;
                 end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_err_print;
begin
print ('-');
bib_ln_num_print;
print_bad_input_line;                   {this call does the |mark_error|}
print_skipping_whatever_remains;
if (at_bib_command) then
    print_ln ('command')
  else
    print_ln ('entry');
end;


@
When there's a harmless error parsing a \.{.bib} file, we just give a
warning message.  This is always called after other stuff has been
printed out.

@d bib_warn(#) == begin         {non-serious error during \.{.bst} parsing}
                  print (#);
                  bib_warn_print;
                  end
@#
@d bib_warn_newline(#) == begin         {same as above but with a newline}
                          print_ln (#);
                          bib_warn_print;
                          end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_warn_print;
begin
bib_ln_num_print;
mark_warning;
end;


@
For all |num_bib_files| database files, we keep reading and processing
\.{.bib} entries until none left.

@<Read the \.{.bib} file(s)@>=
begin
@<Final initialization for \.{.bib} processing@>;
read_performed := true;
bib_ptr := 0;
while (bib_ptr < num_bib_files) do
    begin
    print ('Database file #',bib_ptr+1:0,': ');
    print_bib_name;@/
    bib_line_num := 0;          {initialize to get the first input line}
    buf_ptr2 := last;
    while (not eof(cur_bib_file)) do
        get_bib_command_or_entry_and_process;
    a_close (cur_bib_file);
    incr(bib_ptr);
    end;
reading_completed := true;
  trace
  trace_pr_ln ('Finished reading the database file(s)');
  ecart@/
@<Final initialization for processing the entries@>;
read_completed := true;
end


@
We need to initialize the |field_info| array, and also various things
associated with the |cite_list| array (but not |cite_list| itself).

@<Final initialization for \.{.bib} processing@>=
begin
@<Initialize the |field_info|@>;
@<Initialize things for the |cite_list|@>;
end


@
This module initializes all fields of all entries to |missing|, the
value to which all fields are initialized.

@<Initialize the |field_info|@>=
begin
check_field_overflow (num_fields*num_cites);
field_ptr := 0;
while (field_ptr < max_fields) do
    begin
    field_info[field_ptr] := missing;
    incr(field_ptr);
    end;
end


@
@^fetish@>
@:BibTeX capacity exceeded}{\quad total number of fields@>
Complain if somebody's got a field fetish.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure check_field_overflow (@!total_fields : integer);
begin
if (total_fields > max_fields) then
    begin
    print_ln (total_fields:0,' fields:');
    overflow('total number of fields ',max_fields);
    end;
end;


@
We must initialize the |type_list| array so that we can detect
duplicate (or missing) entries for cite keys on |cite_list|.  Also,
when we're to include the entire database, we use the array
|entry_exists| to detect those missing entries whose |cite_list| info
will (or to be more precise, might) be overwritten; and we use the
array |cite_info| to save the part of |cite_list| that will (might) be
overwritten.  We also use |cite_info| for counting cross~references
when it's appropriate---when an entry isn't otherwise to be included
on |cite_list| (that is, the entry isn't \.{\\cite}d or
\.{\\nocite}d).  Such an entry is included on the final |cite_list| if
it's cross~referenced at least |min_crossrefs| times.

@<Initialize things for the |cite_list|@>=
begin
cite_ptr := 0;
while (cite_ptr < max_cites) do
    begin
    type_list[cite_ptr] := empty;@/
    cite_info[cite_ptr] := any_value;  {to appeas \PASCAL's boolean evaluation}
    incr(cite_ptr);
    end;
old_num_cites := num_cites;
if (all_entries) then
    begin
    cite_ptr := all_marker;
    while (cite_ptr < old_num_cites) do
        begin
        cite_info[cite_ptr] := cite_list[cite_ptr];
        entry_exists[cite_ptr] := false;
        incr(cite_ptr);
        end;
    cite_ptr := all_marker;     {we insert the ``other'' entries here}
    end
  else
    begin
    cite_ptr := num_cites;      {we insert the cross-referenced entries here}
    all_marker := any_value;    {to appease \PASCAL's boolean evaluation}
    end;
end


@
Before we actually start the code for reading a database file, we must
define this \.{.bib}-specific scanning function.  It skips over
|white_space| characters until hitting a nonwhite character or the end
of the file, respectively returning |true| or |false|.  It also
updates |bib_line_num|, the line counter.

@<Procedures and functions for input scanning@>=
function eat_bib_white_space : boolean;
label exit;
begin
while (not scan_white_space) do         {no characters left; read another line}
    begin
    if (not input_ln(cur_bib_file)) then        {end-of-file; return |false|}
        begin
        eat_bib_white_space := false;
        return;
        end;
    incr(bib_line_num);
    buf_ptr2 := 0;
    end;
eat_bib_white_space := true;
exit:
end;


@
It's often illegal to end a \.{.bib} command in certain places, and
this is where we come to check.

@d eat_bib_white_and_eof_check ==
        begin
        if (not eat_bib_white_space) then
            begin
            eat_bib_print;
            return;
            end;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure eat_bib_print;
label exit;     {so the call to |bib_err| works}
begin
bib_err ('Illegal end of database file');
exit:
end;


@
And here are a bunch of error-message macros, each called more than
once, that thus save space as implemented.  This one is for when one
of two possible characters is expected while scanning.

@d bib_one_of_two_expected_err(#) ==
        begin
        bib_one_of_two_print (#);
        return;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_one_of_two_print (@!char1,@!char2:ASCII_code);
label exit;     {so the call to |bib_err| works}
begin
bib_err ('I was expecting a `',xchr[char1],''' or a `',xchr[char2],'''');
exit:
end;


@
This one's for an expected |equals_sign|.

@d bib_equals_sign_expected_err ==
        begin
        bib_equals_sign_print;
        return;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_equals_sign_print;
label exit;     {so the call to |bib_err| works}
begin
bib_err ('I was expecting an "',xchr[equals_sign],'"');
exit:
end;


@
This complains about unbalanced braces.

@d bib_unbalanced_braces_err ==
        begin
        bib_unbalanced_braces_print;
        return;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_unbalanced_braces_print;
label exit;     {so the call to |bib_err| works}
begin
bib_err ('Unbalanced braces');
exit:
end;


@
And this one about an overly exuberant field.

@d bib_field_too_long_err ==
        begin
        bib_field_too_long_print;
        return;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_field_too_long_print;
label exit;     {so the call to |bib_err| works}
begin
bib_err ('Your field is more than ',buf_size:0,' characters');
exit:
end;


@
This one is just a warning, not an error.  It's for when something
isn't (or might not be) quite right with a macro name.

@d macro_name_warning(#) ==
        begin
        macro_warn_print;
        bib_warn_newline (#);
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure macro_warn_print;
begin
print ('Warning--string name "');
print_token;
print ('" is ');
end;


@
@:this can't happen}{\quad Identifier scanning error@>
This macro is used to scan all \.{.bib} identifiers.  The argument
tells what was happening at the time.  The associated procedure simply
prints an error message.

@d bib_identifier_scan_check(#) ==
        begin
        if ((scan_result = white_adjacent) or
                                (scan_result = specified_char_adjacent)) then
            do_nothing
        else
            begin
            bib_id_print;
            bib_err (#);
            end;
        end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_id_print;
begin
if (scan_result = id_null) then
    print ('You''re missing ')
else if (scan_result = other_char_adjacent) then
    print ('"',xchr[scan_char],'" immediately follows ')
else
    id_scanning_confusion;
end;


@
This module either reads a database entry, whose three main components
are an entry type, a database key, and a list of fields, or it reads a
\.{.bib} command, whose structure is command dependent and explained
later.

@d cite_already_set = 22        {this gets around \PASCAL\ limitations}
@d first_time_entry = 26        {for checking for repeated database entries}

@<Scan for and process a \.{.bib} command or database entry@>=
procedure get_bib_command_or_entry_and_process;
label cite_already_set,@!first_time_entry,@!loop_exit,@!exit;
begin
at_bib_command := false;@/
@<Skip to the next database entry or \.{.bib} command@>;
@<Scan the entry type or scan and process the \.{.bib} command@>;
eat_bib_white_and_eof_check;
@<Scan the entry's database key@>;
eat_bib_white_and_eof_check;
@<Scan the entry's list of fields@>;
exit:
end;


@
This module skips over everything until hitting an |at_sign| or the
end of the file.  It also updates |bib_line_num|, the line counter.

@<Skip to the next database entry or \.{.bib} command@>=
while (not scan1(at_sign)) do                   {no |at_sign|; get next line}
    begin
    if (not input_ln(cur_bib_file)) then        {end-of-file}
        return;
    incr(bib_line_num);
    buf_ptr2 := 0;
    end


@
@:this can't happen}{\quad An at-sign disappeared@>
This module reads an |at_sign| and an entry type (like `book' or
`article') or a \.{.bib} command.  If it's an entry type, it must be
defined in the \.{.bst} file if this entry is to be included in the
reference list.

@<Scan the entry type or scan and process the \.{.bib} command@>=
begin
if (scan_char <> at_sign) then
    confusion ('An "',xchr[at_sign],'" disappeared');
incr(buf_ptr2);                                 {skip over the |at_sign|}
eat_bib_white_and_eof_check;
scan_identifier (left_brace,left_paren,left_paren);
bib_identifier_scan_check ('an entry type');
  trace
  trace_pr_token;
  trace_pr_ln (' is an entry type or a database-file command');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
command_num := ilk_info[
        str_lookup(buffer,buf_ptr1,token_len,bib_command_ilk,dont_insert)];
if (hash_found) then
    @<Process a \.{.bib} command@>
else
    begin                                       {process an entry type}
    entry_type_loc := str_lookup(
                        buffer,buf_ptr1,token_len,bst_fn_ilk,dont_insert);
    if ((not hash_found) or (fn_type[entry_type_loc]<>wiz_defined)) then@/
        type_exists := false  {no such entry type defined in the \.{.bst} file}
      else
        type_exists := true;
    end;
end


@
@^database-file commands@>
@:this can't happen}{\quad Unknown database-file command@>
Here we determine which \.{.bib} command we're about to process, then
go to it.

@<Process a \.{.bib} command@>=
begin
at_bib_command := true;
case (command_num) of
    n_bib_comment : @<Process a \.{comment} command@>;
    n_bib_preamble : @<Process a \.{preamble} command@>;
    n_bib_string : @<Process a \.{string} command@>;
    othercases bib_cmd_confusion
endcases;
end


@
@:this can't happen}{\quad Unknown database-file command@>
Here's another bug.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bib_cmd_confusion;
begin
confusion ('Unknown database-file command');
end;


@
@:database-file commands}{\quad \.{comment}@>
The \.{comment} command is implemented for SCRIBE compatibility.  It's
not really needed because \BibTeX\ treats (flushes) everything not
within an entry as a comment anyway.

@<Process a \.{comment} command@>=
begin
return;                 {flush comments}
end


@
@:database-file commands}{\quad \.{preamble}@>
The \.{preamble} command lets a user have \TeX\ stuff inserted (by the
standard styles, at least) directly into the \.{.bbl} file.  It is
intended primarily for allowing \TeX\ macro definitions used within
the bibliography entries (for better sorting, for example).  One
\.{preamble} command per \.{.bib} file should suffice.

A \.{preamble} command has either braces or parentheses as outer
delimiters.  Inside is the preamble string, which has the same syntax
as a field value: a nonempty list of field tokens separated by
|concat_char|s.  There are three types of field tokens---nonnegative
numbers, macro names, and delimited strings.

This module does all the scanning (that's not subcontracted), but the
\.{.bib}-specific scanning function
|scan_and_store_the_field_value_and_eat_white| actually stores the
value.

@<Process a \.{preamble} command@>=
begin
if (preamble_ptr = max_bib_files) then
    bib_err ('You''ve exceeded ',max_bib_files:0,' preamble commands');
eat_bib_white_and_eof_check;
if (scan_char = left_brace) then
    right_outer_delim := right_brace
else if (scan_char = left_paren) then
    right_outer_delim := right_paren
else
    bib_one_of_two_expected_err (left_brace,left_paren);
incr(buf_ptr2);                         {skip over the left-delimiter}
eat_bib_white_and_eof_check;
store_field := true;
if (not scan_and_store_the_field_value_and_eat_white) then
    return;
if (scan_char <> right_outer_delim) then
    bib_err ('Missing "',xchr[right_outer_delim],'" in preamble command');
incr(buf_ptr2);                         {skip over the |right_outer_delim|}
return;
end


@
@:database-file commands}{\quad \.{string}@>
The \.{string} command is implemented both for SCRIBE compatibility
and for allowing a user: to override a \.{.bst}-file \.{macro}
command, to define one that the \.{.bst} file doesn't, or to engage in
good, wholesome, typing laziness.

The \.{string} command does mostly the same thing as the
\.{.bst}-file's \.{macro} command (but the syntax is different and the
\.{string} command compresses |white_space|).  In fact, later in this
program, the term ``macro'' refers to either a \.{.bst} ``macro'' or a
\.{.bib} ``string'' (when it's clear from the context that it's not
a \.{WEB} macro).

A \.{string} command has either braces or parentheses as outer
delimiters.  Inside is the string's name (it must be a legal
identifier, and case differences are ignored---all upper-case letters
are converted to lower case), then an |equals_sign|, and the string's
definition, which has the same syntax as a field value: a nonempty
list of field tokens separated by |concat_char|s.  There are three
types of field tokens---nonnegative numbers, macro names, and
delimited strings.

@<Process a \.{string} command@>=
begin
eat_bib_white_and_eof_check;
@<Scan the string's name@>;
eat_bib_white_and_eof_check;
@<Scan the string's definition field@>;
return;
end


@
This module reads a left outer-delimiter and a string name.

@<Scan the string's name@>=
begin
if (scan_char = left_brace) then
    right_outer_delim := right_brace
else if (scan_char = left_paren) then
    right_outer_delim := right_paren
else
    bib_one_of_two_expected_err (left_brace,left_paren);
incr(buf_ptr2);                         {skip over the left-delimiter}
eat_bib_white_and_eof_check;
scan_identifier (equals_sign,equals_sign,equals_sign);
bib_identifier_scan_check ('a string name');
@<Store the string's name@>;
end


@
@^commented-out code@>
This module marks this string as |macro_ilk|; the commented-out code
will give a warning message when overwriting a previously defined
macro.

@<Store the string's name@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a database-defined macro');
  ecart@/
lower_case (buffer, buf_ptr1, token_len);       {ignore case differences}
cur_macro_loc := str_lookup(buffer,buf_ptr1,token_len,macro_ilk,do_insert);
ilk_info[cur_macro_loc] := hash_text[cur_macro_loc]; {default in case of error}
  @{
  if (hash_found) then                          {already seen macro}
      macro_name_warning ('having its definition overwritten');
  @}@/
end


@
This module skips over the |equals_sign|, reads and stores the list of
field tokens that defines this macro (compressing |white_space|), and
reads a |right_outer_delim|.

@<Scan the string's definition field@>=
begin
if (scan_char <> equals_sign) then
    bib_equals_sign_expected_err;
incr(buf_ptr2);                         {skip over the |equals_sign|}
eat_bib_white_and_eof_check;
store_field := true;
if (not scan_and_store_the_field_value_and_eat_white) then
    return;
if (scan_char <> right_outer_delim) then
    bib_err ('Missing "',xchr[right_outer_delim],'" in string command');
incr(buf_ptr2);                         {skip over the |right_outer_delim|}
end


@
@^kludge@>
The variables for the function
|scan_and_store_the_field_value_and_eat_white| must be global since
the functions it calls use them too.  The alias kludge helps make the
stack space not overflow on some machines.

@d field_vl_str == ex_buf       {aliases, used ``only'' for this function}
@d field_end == ex_buf_ptr      {the end marker for the field-value string}
@d field_start == ex_buf_xptr   {and the start marker}

@<Globals in the outer block@>=
@!bib_brace_level : integer;    {brace nesting depth (excluding |str_delim|s)}


@
@^gymnastics@>
Since the function |scan_and_store_the_field_value_and_eat_white|
calls several other yet-to-be-described functions (one directly and
two indirectly), we must perform some topological gymnastics.

@<Procedures and functions for input scanning@>=
@<The scanning function |compress_bib_white|@>@;
@<The scanning function |scan_balanced_braces|@>@;
@<The scanning function |scan_a_field_token_and_eat_white|@>


@
This function scans the list of field tokens that define the field
value string.  If |store_field| is |true| it accumulates (indirectly)
in |field_vl_str| the concatenation of all the field tokens,
compressing nonnull |white_space| to a single |space| and, if the
field value is for a field (rather than a string definition), removing
any leading or trailing |white_space|; when it's finished it puts the
string into the hash table.  It returns |false| if there was a serious
syntax error.

@<Procedures and functions for input scanning@>=
function scan_and_store_the_field_value_and_eat_white : boolean;
label exit;
begin
scan_and_store_the_field_value_and_eat_white := false;
                                        {now it's easy to exit if necessary}
field_end := 0;
if (not scan_a_field_token_and_eat_white) then
    return;
while (scan_char = concat_char) do      {scan remaining field tokens}
    begin
    incr(buf_ptr2);                     {skip over the |concat_char|}
    eat_bib_white_and_eof_check;
    if (not scan_a_field_token_and_eat_white) then
        return;
    end;
if (store_field) then
    @<Store the field value string@>;
scan_and_store_the_field_value_and_eat_white := true;
exit:
end;


@
Each field token is either a nonnegative number, a macro name (like
`jan'), or a brace-balanced string delimited by either |double_quote|s
or braces.  Thus there are four possibilities for the first character
of the field token: If it's a |left_brace| or a |double_quote|, the
token (with balanced braces, up to the matching |right_str_delim|) is
a string; if it's |numeric|, the token is a number; if it's anything
else, the token is a macro name (and should thus have been defined by
either the \.{.bst}-file's \.{macro} command or the \.{.bib}-file's
\.{string} command).  This function returns |false| if there was a
serious syntax error.

@<The scanning function |scan_a_field_token_and_eat_white|@>=
function scan_a_field_token_and_eat_white : boolean;
label exit;
begin
scan_a_field_token_and_eat_white := false; {now it's easy to exit if necessary}
case (scan_char) of
    left_brace :
        begin
        right_str_delim := right_brace;
        if (not scan_balanced_braces) then
            return;
        end;
    double_quote :
        begin
        right_str_delim := double_quote;
        if (not scan_balanced_braces) then
            return;
        end;
    "0", "1", "2", "3", "4", "5", "6", "7", "8", "9" :
        @<Scan a number@>;
    othercases
        @<Scan a macro name@>
endcases;
eat_bib_white_and_eof_check;
scan_a_field_token_and_eat_white := true;
exit:
end;


@
Now we come to the stuff that actually accumulates the field value to
be stored.  This module copies a character into |field_vl_str| if it
will fit; since it's so low level, it's implemented as a macro.

@d copy_char(#) == begin
                   if (field_end = buf_size) then
                       bib_field_too_long_err
                     else
                       begin
                       field_vl_str[field_end] := #;
                       incr(field_end);
                       end;
                   end


@
The \.{.bib}-specific scanning function |compress_bib_white| skips
over |white_space| characters within a string until hitting a nonwhite
character; in fact, it does everything |eat_bib_white_space| does, but
it also adds a |space| to |field_vl_str|.  This function is never
called if there are no |white_space| characters (or ends-of-line) to
be scanned (though the associated macro might be).  The function
returns |false| if there is a serious syntax error.

@d check_for_and_compress_bib_white_space ==
        begin
        if ((lex_class[scan_char]=white_space) or (buf_ptr2=last)) then
            if (not compress_bib_white) then
                return;
        end

@<The scanning function |compress_bib_white|@>=
function compress_bib_white : boolean;
label exit;
begin
compress_bib_white := false;            {now it's easy to exit if necessary}
copy_char (space);
while (not scan_white_space) do         {no characters left; read another line}
    begin
    if (not input_ln(cur_bib_file)) then        {end-of-file; complain}
        begin
        eat_bib_print;
        return;
        end;
    incr(bib_line_num);
    buf_ptr2 := 0;
    end;
compress_bib_white := true;
exit:
end;


@
This \.{.bib}-specific function scans a string with balanced braces,
stopping just past the matching |right_str_delim|.  How much work it
does depends on whether |store_field = true|.  It returns |false| if
there was a serious syntax error.

@<The scanning function |scan_balanced_braces|@>=
function scan_balanced_braces : boolean;
label loop_exit,@!exit;
begin
scan_balanced_braces := false;          {now it's easy to exit if necessary}
incr(buf_ptr2);                         {skip over the left-delimiter}
check_for_and_compress_bib_white_space;
if (field_end > 1) then
  if (field_vl_str[field_end-1] = space) then
    if (field_vl_str[field_end-2] = space) then {remove wrongly added |space|}
        decr(field_end);
bib_brace_level := 0;           {and we're at a non|white_space| character}
if (store_field) then
    @<Do a full brace-balanced scan@>
  else
    @<Do a quick brace-balanced scan@>;
incr(buf_ptr2);                         {skip over the |right_str_delim|}
scan_balanced_braces := true;
exit:
end;


@
This module scans over a brace-balanced string without keeping track
of anything but the brace level.  It starts with |bib_brace_level = 0|
and at a non|white_space| character.

@<Do a quick brace-balanced scan@>=
begin
while (scan_char <> right_str_delim) do {we're at |bib_brace_level = 0|}
    if (scan_char = left_brace) then
        begin
        incr(bib_brace_level);
        incr(buf_ptr2);                 {skip over the |left_brace|}
        eat_bib_white_and_eof_check;
        while (bib_brace_level > 0) do
            @<Do a quick scan with |bib_brace_level > 0|@>;
        end
    else if (scan_char = right_brace) then
        bib_unbalanced_braces_err
    else
        begin
        incr(buf_ptr2);                 {skip over some other character}
        if (not scan3 (right_str_delim, left_brace, right_brace)) then
            eat_bib_white_and_eof_check;
        end
end


@
This module does the same as above but, because |bib_brace_level > 0|, it
doesn't have to look for a |right_str_delim|.

@<Do a quick scan with |bib_brace_level > 0|@>=
begin   {top part of the |while| loop---we're always at a nonwhite character}
if (scan_char = right_brace) then
    begin
    decr(bib_brace_level);
    incr(buf_ptr2);                     {skip over the |right_brace|}
    eat_bib_white_and_eof_check;
    end
else if (scan_char = left_brace) then
    begin
    incr(bib_brace_level);
    incr(buf_ptr2);                     {skip over the |left_brace|}
    eat_bib_white_and_eof_check;
    end
else
    begin
    incr(buf_ptr2);                     {skip over some other character}
    if (not scan2 (right_brace, left_brace)) then
        eat_bib_white_and_eof_check;
    end
end


@
This module scans over a brace-balanced string, compressing multiple
|white_space| characters into a single |space|.  It starts with
|bib_brace_level = 0| and starts at a non|white_space| character.

@<Do a full brace-balanced scan@>=
begin
while (scan_char <> right_str_delim) do
  case (scan_char) of
    left_brace :
        begin
        incr(bib_brace_level);
        copy_char (left_brace);@/
        incr(buf_ptr2);                 {skip over the |left_brace|}
        check_for_and_compress_bib_white_space;@/
        @<Do a full scan with |bib_brace_level > 0|@>;
        end;
    right_brace :
        bib_unbalanced_braces_err;
    othercases
        begin
        copy_char (scan_char);
        incr(buf_ptr2);                 {skip over some other character}
        check_for_and_compress_bib_white_space;
        end
  endcases;
end


@
This module is similar to the last but starts with |bib_brace_level > 0|
(and, like the last, it starts at a non|white_space| character).

@<Do a full scan with |bib_brace_level > 0|@>=
begin
loop
  case (scan_char) of
    right_brace :
        begin
        decr(bib_brace_level);
        copy_char (right_brace);@/
        incr(buf_ptr2);                 {skip over the |right_brace|}
        check_for_and_compress_bib_white_space;
        if (bib_brace_level = 0) then
            goto loop_exit;
        end;
    left_brace :
        begin
        incr(bib_brace_level);
        copy_char (left_brace);@/
        incr(buf_ptr2);                 {skip over the |left_brace|}
        check_for_and_compress_bib_white_space;
        end;
    othercases
        begin
        copy_char (scan_char);
        incr(buf_ptr2);                 {skip over some other character}
        check_for_and_compress_bib_white_space;
        end
  endcases;
loop_exit:
end


@
@:this can't happen}{\quad A digit disappeared@>
This module scans a nonnegative number and copies it to |field_vl_str|
if it's to store the field.

@<Scan a number@>=
begin
if (not scan_nonneg_integer) then
    confusion ('A digit disappeared');
if (store_field) then
    begin
    tmp_ptr := buf_ptr1;
    while (tmp_ptr < buf_ptr2) do
        begin
        copy_char (buffer[tmp_ptr]);
        incr(tmp_ptr);
        end;
    end;
end


@
This module scans a macro name and copies its string to |field_vl_str|
if it's to store the field, complaining if the macro is recursive or
undefined.

@<Scan a macro name@>=
begin
scan_identifier (comma,right_outer_delim,concat_char);
bib_identifier_scan_check ('a field part');
if (store_field) then
    begin
    lower_case (buffer, buf_ptr1, token_len);   {ignore case differences}
    macro_name_loc := str_lookup(
                        buffer,buf_ptr1,token_len,macro_ilk,dont_insert);
    store_token := true;
    if (at_bib_command) then
      if (command_num = n_bib_string) then
        if (macro_name_loc = cur_macro_loc) then
            begin
            store_token := false;
            macro_name_warning ('used in its own definition');
            end;
    if (not hash_found) then
        begin
        store_token := false;
        macro_name_warning ('undefined');
        end;
    if (store_token) then
        @<Copy the macro string to |field_vl_str|@>;
    end;
end


@
The macro definition may have |white_space| that needs compressing,
because it may have been defined in the \.{.bst} file.

@<Copy the macro string to |field_vl_str|@>=
begin
tmp_ptr := str_start[ilk_info[macro_name_loc]];
tmp_end_ptr := str_start[ilk_info[macro_name_loc]+1];
if (field_end = 0) then
  if ((lex_class[str_pool[tmp_ptr]] = white_space) and (tmp_ptr < tmp_end_ptr))
                                                                        then
    begin               {compress leading |white_space| of first nonnull token}
    copy_char (space);
    incr(tmp_ptr);
    while ((lex_class[str_pool[tmp_ptr]] = white_space) and
                                                (tmp_ptr <  tmp_end_ptr)) do
        incr(tmp_ptr);
    end;                {the next remaining character is non|white_space|}
while (tmp_ptr < tmp_end_ptr) do
    begin
    if (lex_class[str_pool[tmp_ptr]] <> white_space) then
        copy_char (str_pool[tmp_ptr])
      else if (field_vl_str[field_end-1] <> space) then
        copy_char (space);
    incr(tmp_ptr);
    end;
end


@
@^ham and eggs@>
Now it's time to store the field value in the hash table, and store an
appropriate pointer to it (depending on whether it's for a database
entry or command).  But first, if necessary, we remove a trailing
|space| and a leading |space| if these exist.  (Hey, if we had some
ham we could make ham-and-eggs if we had some eggs.)

@<Store the field value string@>=
begin
if (not at_bib_command) then            {chop trailing |space| for a field}
  if (field_end > 0) then
    if (field_vl_str[field_end-1] = space) then
        decr(field_end);
if ((not at_bib_command) and (field_vl_str[0] = space) and (field_end > 0))
                                then    {chop leading |space| for a field}
    field_start := 1
  else
    field_start := 0;
field_val_loc := str_lookup(field_vl_str,field_start,field_end-field_start,
                                                        text_ilk,do_insert);
fn_type[field_val_loc] := str_literal;          {set the |fn_class|}
  trace
  trace_pr ('"');
  trace_pr_pool_str (hash_text[field_val_loc]);
  trace_pr_ln ('" is a field value');
  ecart@/
if (at_bib_command) then        {for a \.{preamble} or \.{string} command}
    @<Store the field value for a command@>
  else                                                  {for a database entry}
    @<Store the field value for a database entry@>;
end


@
@:this can't happen}{\quad Unknown database-file command@>
Here's where we store the goods when we're dealing with a command
rather than an entry.

@<Store the field value for a command@>=
begin
case (command_num) of
    n_bib_preamble :
        begin
        s_preamble[preamble_ptr] := hash_text[field_val_loc];
        incr(preamble_ptr);
        end;
    n_bib_string :
        ilk_info[cur_macro_loc] := hash_text[field_val_loc];
    othercases bib_cmd_confusion
endcases;
end


@
And here, an entry.

@<Store the field value for a database entry@>=
begin
field_ptr := entry_cite_ptr * num_fields + fn_info[field_name_loc];
if (field_info[field_ptr] <> missing) then
    begin
    print ('Warning--I''m ignoring ');
    print_pool_str (cite_list[entry_cite_ptr]);
    print ('''s extra "');
    print_pool_str (hash_text[field_name_loc]);
    bib_warn_newline ('" field');
    end
  else
    begin                       {the field was empty, store its new value}
    field_info[field_ptr] := hash_text[field_val_loc];
    if ((fn_info[field_name_loc] = crossref_num) and (not all_entries)) then
        @<Add or update a cross reference on |cite_list| if necessary@>;
    end;
end


@
@^kludge@>
@:this can't happen}{\quad Cite hash error@>
If the cross-referenced entry isn't already on |cite_list| we add it
(at least temporarily); if it is already on |cite_list| we update the
cross-reference count, if necessary.  Note that |all_entries| is
|false| here.  The alias kludge helps make the stack space not
overflow on some machines.

@d extra_buf == out_buf         {an alias, used only in this module}

@<Add or update a cross reference on |cite_list| if necessary@>=
begin
tmp_ptr := field_start;
while (tmp_ptr < field_end) do
    begin
    extra_buf[tmp_ptr] := field_vl_str[tmp_ptr];
    incr(tmp_ptr);
    end;
lower_case (extra_buf, field_start, field_end-field_start);
                                                {convert to `canonical' form}
lc_cite_loc := str_lookup(extra_buf,field_start,field_end-field_start,
                                                        lc_cite_ilk,do_insert);
if (hash_found) then
    begin
    cite_loc := ilk_info[lc_cite_loc];  {even if there's a case mismatch}
    if (ilk_info[cite_loc] >= old_num_cites) then  {a previous \.{crossref}}
        incr(cite_info[ilk_info[cite_loc]]);
    end
  else
    begin                                       {it's a new \.{crossref}}
    cite_loc := str_lookup(field_vl_str,field_start,field_end-field_start,
                                                        cite_ilk,do_insert);
    if (hash_found) then
        hash_cite_confusion;
    add_database_cite (cite_ptr);               {this increments |cite_ptr|}
    cite_info[ilk_info[cite_loc]] := 1; {the first cross-ref for this cite key}
    end;
end


@
This procedure adds (or restores) to |cite_list| a cite key; it is
called only when |all_entries| is |true| or when adding
cross~references, and it assumes that |cite_loc| and |lc_cite_loc| are
set.  It also increments its argument.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure add_database_cite (var new_cite : cite_number);
begin
check_cite_overflow (new_cite);                 {make sure this cite will fit}
check_field_overflow (num_fields*new_cite);
cite_list[new_cite] := hash_text[cite_loc];
ilk_info[cite_loc] := new_cite;
ilk_info[lc_cite_loc] := cite_loc;
incr(new_cite);
end;


@
And now, back to processing an entry (rather than a command).  This
module reads a left outer-delimiter and a database key.

@<Scan the entry's database key@>=
begin
if (scan_char = left_brace) then
    right_outer_delim := right_brace
else if (scan_char = left_paren) then
    right_outer_delim := right_paren
else
    bib_one_of_two_expected_err (left_brace,left_paren);
incr(buf_ptr2);                                 {skip over the left-delimiter}
eat_bib_white_and_eof_check;
if (right_outer_delim = right_paren) then       {to allow it in a database key}
    begin
    if (scan1_white(comma)) then                {ok if database key ends line}
        do_nothing;
    end
  else
    if (scan2_white(comma,right_brace)) then {|right_brace=right_outer_delim|}
        do_nothing;
@<Check for a database key of interest@>;
end


@
@^kludge@>
The lower-case version of this database key must correspond to one in
|cite_list|, or else |all_entries| must be |true|, if this entry is to
be included in the reference list.  Accordingly, this module sets
|store_entry|, which determines whether the relevant information for
this entry is stored.  The alias kludge helps make the stack space not
overflow on some machines.

@d ex_buf3 == ex_buf            {an alias, used only in this module}

@<Check for a database key of interest@>=
begin
  trace
  trace_pr_token;
  trace_pr_ln (' is a database key');
  ecart@/
tmp_ptr := buf_ptr1;
while (tmp_ptr < buf_ptr2) do
    begin
    ex_buf3[tmp_ptr] := buffer[tmp_ptr];
    incr(tmp_ptr);
    end;
lower_case (ex_buf3, buf_ptr1, token_len);      {convert to `canonical' form}
if (all_entries) then
    lc_cite_loc := str_lookup(ex_buf3,buf_ptr1,token_len,lc_cite_ilk,do_insert)
  else
    lc_cite_loc := str_lookup(ex_buf3,buf_ptr1,token_len,lc_cite_ilk,
                                                                dont_insert);
if (hash_found) then
    begin
    entry_cite_ptr := ilk_info[ilk_info[lc_cite_loc]];
    @<Check for a duplicate or \.{crossref}-matching database key@>;
    end;
store_entry := true;    {unless |(not hash_found) and (not all_entries)|}
if (all_entries) then
    @<Put this cite key in its place@>
  else if (not hash_found) then
    store_entry := false;       {no such cite key exists on |cite_list|}
if (store_entry) then
    @<Make sure this entry is ok before proceeding@>;
end


@
@:this can't happen}{\quad The cite list is messed up@>
It's illegal to have two (or more) entries with the same database key
(even if there are case differrences), and we skip the rest of the
entry for such a repeat occurrence.  Also, we make this entry's
database key the official |cite_list| key if it's on |cite_list| only
because of cross references.

@<Check for a duplicate or \.{crossref}-matching database key@>=
begin
if ((not all_entries) or (entry_cite_ptr < all_marker)
                                or (entry_cite_ptr >= old_num_cites)) then
    begin
    if (type_list[entry_cite_ptr] = empty) then
        begin
        @<Make sure this entry's database key is on |cite_list|@>;
        goto first_time_entry;
        end;
    end
else if (not entry_exists[entry_cite_ptr]) then
    begin
    @<Find the lower-case equivalent of the |cite_info| key@>;
    if (lc_xcite_loc = lc_cite_loc) then
        goto first_time_entry;
    end;@/
                                {oops---repeated entry---issue a reprimand}
if (type_list[entry_cite_ptr] = empty) then
    confusion ('The cite list is messed up');
bib_err ('Repeated entry');
first_time_entry:  {note that when we leave normally, |hash_found| is |true|}
end


@
An entry that's on |cite_list| only because of cross referencing must
have its database key (rather than one of the \.{crossref} keys) as
the official |cite_list| string.  Here's where we assure that.  The
variable |hash_found| is |true| upon entrance to and exit from this
module.

@<Make sure this entry's database key is on |cite_list|@>=
begin
if ((not all_entries) and (entry_cite_ptr >= old_num_cites)) then
    begin
    cite_loc := str_lookup(buffer,buf_ptr1,token_len,cite_ilk,do_insert);
    if (not hash_found) then
        begin                   {it's not on |cite_list|---put it there}
        ilk_info[lc_cite_loc] := cite_loc;
        ilk_info[cite_loc] := entry_cite_ptr;
        cite_list[entry_cite_ptr] := hash_text[cite_loc];@/
        hash_found := true;             {restore this value for later use}
        end;
    end;
end


@
@^kludge@>
@:this can't happen}{\quad A cite key disappeared@>
This module, a simpler version of the
|find_cite_locs_for_this_cite_key| function, exists primarily to
compute |lc_xcite_loc|.  When this code is executed we have
|(all_entries) and (entry_cite_ptr >= all_marker) and (not
entry_exists[entry_cite_ptr])|.  The alias kludge helps make the stack
space not overflow on some machines.

@d ex_buf4 == ex_buf            {aliases, used only}
@d ex_buf4_ptr == ex_buf_ptr    {in this module}

@<Find the lower-case equivalent of the |cite_info| key@>=
begin
ex_buf4_ptr := 0;
tmp_ptr := str_start[cite_info[entry_cite_ptr]];
tmp_end_ptr := str_start[cite_info[entry_cite_ptr]+1];
while (tmp_ptr < tmp_end_ptr) do
    begin
    ex_buf4[ex_buf4_ptr] := str_pool[tmp_ptr];
    incr(ex_buf4_ptr);
    incr(tmp_ptr);
    end;
lower_case (ex_buf4, 0, length(cite_info[entry_cite_ptr]));
                                                {convert to `canonical' form}
lc_xcite_loc := str_lookup(ex_buf4,0,length(cite_info[entry_cite_ptr]),
                                                lc_cite_ilk,dont_insert);
if (not hash_found) then
    cite_key_disappeared_confusion;
end


@
@:this can't happen}{\quad A cite key disappeared@>
Here's another bug complaint.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure cite_key_disappeared_confusion;
begin
confusion ('A cite key disappeared');
end;


@
@:this can't happen}{\quad Cite hash error@>
This module, which gets executed only when |all_entries| is |true|,
does one of three things, depending on whether or not, and where, the
cite key appears on |cite_list|: If it's on |cite_list| before
|all_marker|, there's nothing to be done; if it's after |all_marker|,
it must be reinserted (at the current place) and we must note that its
corresponding entry exists; and if it's not on |cite_list| at all, it
must be inserted for the first time.  The |goto| construct must stay
as is, partly because some \PASCAL\ compilers might complain if
``|and|'' were to connect the two boolean expressions (since
|entry_cite_ptr| could be uninitialized when |hash_found| is |false|).

@<Put this cite key in its place@>=
begin
if (hash_found) then
    begin
    if (entry_cite_ptr < all_marker) then
        goto cite_already_set           {that is, do nothing}
      else
        begin
        entry_exists[entry_cite_ptr] := true;
        cite_loc := ilk_info[lc_cite_loc];
        end;
    end
  else
    begin                               {this is a new key}
    cite_loc := str_lookup(buffer,buf_ptr1,token_len,cite_ilk,do_insert);
    if (hash_found) then
        hash_cite_confusion;
    end;@/
entry_cite_ptr := cite_ptr;
add_database_cite (cite_ptr);           {this increments |cite_ptr|}
cite_already_set:
end


@
@^case mismatch errors@>
@^commented-out code@>
We must give a warning if this entry~type doesn't exist.  Also, we
point the appropriate entry of |type_list| to the entry type just read
above.

For SCRIBE compatibility, the code to give a warning for a case
mismatch between a cite key and a database key has been commented out.
In fact, SCRIBE is the reason that it doesn't produce an error message
outright.  (Note: Case mismatches between two cite keys produce
full-blown errors.)

@<Make sure this entry is ok before proceeding@>=
begin
  @{
  dummy_loc := str_lookup(buffer,buf_ptr1,token_len,cite_ilk,dont_insert);
  if (not hash_found) then      {give a warning if there is a case difference}
    begin
    print ('Warning--case mismatch, database key "');
    print_token;
    print ('", cite key "');
    print_pool_str (cite_list[entry_cite_ptr]);
    bib_warn_newline ('"');
    end;
  @}@/
if (type_exists) then
    type_list[entry_cite_ptr] := entry_type_loc
  else
    begin
    type_list[entry_cite_ptr] := undefined;
    print ('Warning--entry type for "');
    print_token;
    bib_warn_newline ('" isn''t style-file defined');
    end;
end


@
This module reads a |comma| and a field as many times as it can, and
then reads a |right_outer_delim|, ending the current entry.

@<Scan the entry's list of fields@>=
begin
while (scan_char <> right_outer_delim) do
    begin
    if (scan_char <> comma) then
        bib_one_of_two_expected_err (comma,right_outer_delim);
    incr(buf_ptr2);                     {skip over the |comma|}
    eat_bib_white_and_eof_check;
    if (scan_char = right_outer_delim) then
        goto loop_exit;
    @<Get the next field name@>;
    eat_bib_white_and_eof_check;
    if (not scan_and_store_the_field_value_and_eat_white) then
        return;
    end;
loop_exit:
incr(buf_ptr2);                         {skip over the |right_outer_delim|}
end


@
This module reads a field name; its contents won't be stored unless it
was declared in the \.{.bst} file and |store_entry = true|.

@<Get the next field name@>=
begin
scan_identifier (equals_sign,equals_sign,equals_sign);
bib_identifier_scan_check ('a field name');
  trace
  trace_pr_token;
  trace_pr_ln (' is a field name');
  ecart@/
store_field := false;
if (store_entry) then
    begin
    lower_case (buffer, buf_ptr1, token_len);   {ignore case differences}
    field_name_loc := str_lookup(
                        buffer,buf_ptr1,token_len,bst_fn_ilk,dont_insert);
    if (hash_found) then
      if (fn_type[field_name_loc]=field) then@/
        store_field := true;  {field name was pre-defined or \.{.bst}-declared}
    end;
eat_bib_white_and_eof_check;
if (scan_char <> equals_sign) then
    bib_equals_sign_expected_err;
incr(buf_ptr2);                 {skip over the |equals_sign|}
end


@
This gets things ready for further \.{.bst} processing.

@<Final initialization for processing the entries@>=
begin
num_cites := cite_ptr;  {to include database and \.{crossref} cite keys, too}
num_preamble_strings := preamble_ptr;   {number of \.{preamble} commands seen}
@<Add cross-reference information@>;
@<Subtract cross-reference information@>;
@<Remove missing entries or those cross referenced too few times@>;
@<Initialize the |int_entry_var|s@>;
@<Initialize the |str_entry_var|s@>;
@<Initialize the |sorted_cites|@>;
end


@
@^child entry@>
@^cross references@>
@^nested cross references@>
@^parent entry@>
Now we update any entry (here called a {\it child\/} entry) that
cross~referenced another (here called a {\it parent\/} entry); this
cross~referencing occurs when the child's \.{crossref} field (value)
consists of the parent's database key.  To do the update, we replace
the child's |missing| fields by the corresponding fields of the
parent.  Also, we make sure the \.{crossref} field contains the
case-correct version.  Finally, although it is technically illegal to
nest cross~references, and although we give a warning (a few modules
hence) when someone tries, we do what we can to accommodate the
attempt.

@<Add cross-reference information@>=
begin
cite_ptr := 0;
while (cite_ptr < num_cites) do
    begin
    field_ptr := cite_ptr * num_fields + crossref_num;
    if (field_info[field_ptr] <> missing) then
      if (find_cite_locs_for_this_cite_key (field_info[field_ptr])) then
        begin
        cite_loc := ilk_info[lc_cite_loc];
        field_info[field_ptr] := hash_text[cite_loc];
        cite_parent_ptr := ilk_info[cite_loc];
        field_ptr := cite_ptr * num_fields + num_pre_defined_fields;
        field_end_ptr := field_ptr - num_pre_defined_fields + num_fields;
        field_parent_ptr := cite_parent_ptr * num_fields
                                                + num_pre_defined_fields;
        while (field_ptr < field_end_ptr) do
            begin
            if (field_info[field_ptr] = missing) then
                field_info[field_ptr] := field_info[field_parent_ptr];
            incr(field_ptr);
            incr(field_parent_ptr);
            end;
        end;
    incr(cite_ptr);
    end;
end


@
@^kludge@>
@^raisin@>
Occasionally we need to figure out the hash-table location of a given
cite-key string and its lower-case equivalent.  This function does
that.  To perform the task it needs to borrow a buffer, a need that
gives rise to the alias kludge---it helps make the stack space not
overflow on some machines (and while it's at it, it'll borrow a
pointer, too).  Finally, the function returns |true| if the cite key
exists on |cite_list|, and its sets |cite_hash_found| according to
whether or not it found the actual version (before |lower_case|ing) of
the cite key; however, its {\sl raison d'\^$\mkern-8mu$etre\/}
(literally, ``to eat a raisin'') is to compute |cite_loc| and
|lc_cite_loc|.

@d ex_buf5 == ex_buf            {aliases, used only}
@d ex_buf5_ptr == ex_buf_ptr    {in this module}

@<Procedures and functions for handling numbers, characters, and strings@>=
function find_cite_locs_for_this_cite_key (@!cite_str : str_number) : boolean;
begin
ex_buf5_ptr := 0;
tmp_ptr := str_start[cite_str];
tmp_end_ptr := str_start[cite_str+1];
while (tmp_ptr < tmp_end_ptr) do
    begin
    ex_buf5[ex_buf5_ptr] := str_pool[tmp_ptr];
    incr(ex_buf5_ptr);
    incr(tmp_ptr);
    end;
cite_loc := str_lookup(ex_buf5,0,length(cite_str),cite_ilk,dont_insert);
cite_hash_found := hash_found;
lower_case (ex_buf5, 0, length(cite_str));      {convert to `canonical' form}
lc_cite_loc := str_lookup(ex_buf5,0,length(cite_str),lc_cite_ilk,dont_insert);
if (hash_found) then
    find_cite_locs_for_this_cite_key := true
  else
    find_cite_locs_for_this_cite_key := false;
end;


@
@:this can't happen}{\quad Cite hash error@>
Here we remove the \.{crossref} field value for each child whose
parent was cross~referenced too few times.  We also issue any
necessary warnings arising from a bad cross~reference.

@<Subtract cross-reference information@>=
begin
cite_ptr := 0;
while (cite_ptr < num_cites) do
    begin
    field_ptr := cite_ptr * num_fields + crossref_num;
    if (field_info[field_ptr] <> missing) then
      if (not find_cite_locs_for_this_cite_key (field_info[field_ptr])) then
        begin                           {the parent is not on |cite_list|}
        if (cite_hash_found) then
            hash_cite_confusion;
        nonexistent_cross_reference_error;
        field_info[field_ptr] := missing;       {remove the \.{crossref} ptr}
        end
      else
        begin                           {the parent exists on |cite_list|}
        if (cite_loc <> ilk_info[lc_cite_loc]) then
            hash_cite_confusion;
        cite_parent_ptr := ilk_info[cite_loc];
        if (type_list[cite_parent_ptr] = empty) then
            begin
            nonexistent_cross_reference_error;@/
            field_info[field_ptr] := missing;   {remove the \.{crossref} ptr}
            end
          else
            begin                       {the parent exists in the database too}
            field_parent_ptr := cite_parent_ptr * num_fields + crossref_num;
            if (field_info[field_parent_ptr] <> missing) then
                @<Complain about a nested cross reference@>;
            if ((not all_entries) and (cite_parent_ptr >= old_num_cites) and
                        (cite_info[cite_parent_ptr] < min_crossrefs)) then@/
                field_info[field_ptr] := missing; {remove the \.{crossref} ptr}
            end;
        end;
    incr(cite_ptr);
    end;
end


@
This procedure exists to save space, since it's used twice---once for
each of the two succeeding modules.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bad_cross_reference_print (@!s:str_number);
begin
print ('--entry "');
print_pool_str (cur_cite_str);
print_ln ('"');
print ('refers to entry "');
print_pool_str (s);
end;


@
When an entry being cross referenced doesn't exist on |cite_list|, we
complain.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure nonexistent_cross_reference_error;
begin
print ('A bad cross reference-');
bad_cross_reference_print (field_info[field_ptr]);
print_ln ('", which doesn''t exist');
mark_error;
end;


@
We also complain when an entry being cross referenced has a
non|missing| \.{crossref} field itself, but this one is just a
warning, not a full-blown error.

@<Complain about a nested cross reference@>=
begin
print ('Warning--you''ve nested cross references');
bad_cross_reference_print (cite_list[cite_parent_ptr]);
print_ln ('", which also refers to something');
mark_warning;
end


@
We remove (and give a warning for) each cite key on the original
|cite_list| without a corresponding database entry.  And we remove any
entry that was included on |cite_list| only because it was
cross~referenced, yet was cross~referenced fewer than |min_crossrefs|
times.  Throughout this module, |cite_ptr| points to the next cite key
to be checked and |cite_xptr| points to the next permanent spot on
|cite_list|.

@<Remove missing entries or those cross referenced too few times@>=
begin
cite_ptr := 0;
while (cite_ptr < num_cites) do
    begin
    if (type_list[cite_ptr] = empty) then
        print_missing_entry (cur_cite_str)
    else if ((all_entries) or (cite_ptr < old_num_cites) or
                                (cite_info[cite_ptr] >= min_crossrefs)) then
        begin
        if (cite_ptr > cite_xptr) then
            @<Slide this cite key down to its permanent spot@>;
        incr(cite_xptr);
        end;
    incr(cite_ptr);
    end;
num_cites := cite_xptr;
if (all_entries) then
    @<Complain about missing entries whose cite keys got overwritten@>;
end


@
When a cite key on the original |cite_list| (or added to |cite_list|
because of cross~referencing) didn't appear in the database, complain.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_missing_entry (@!s:str_number);
begin
print ('Warning--I didn''t find a database entry for "');
print_pool_str (s);
print_ln ('"');
mark_warning;
end;


@
@:this can't happen}{\quad A cite key disappeared@>
@:this can't happen}{\quad Cite hash error@>
We have to move to its final resting place all the entry information
associated with the exact location in |cite_list| of this cite key.

@<Slide this cite key down to its permanent spot@>=
begin
cite_list[cite_xptr] := cite_list[cite_ptr];
type_list[cite_xptr] := type_list[cite_ptr];
if (not find_cite_locs_for_this_cite_key (cite_list[cite_ptr])) then
    cite_key_disappeared_confusion;
if ((not cite_hash_found) or (cite_loc <> ilk_info[lc_cite_loc])) then
    hash_cite_confusion;
ilk_info[cite_loc] := cite_xptr;@/
field_ptr := cite_xptr * num_fields;
field_end_ptr := field_ptr + num_fields;
tmp_ptr := cite_ptr * num_fields;
while (field_ptr < field_end_ptr) do
    begin
    field_info[field_ptr] := field_info[tmp_ptr];
    incr(field_ptr);
    incr(tmp_ptr);
    end;
end


@
We need this module only when we're including the whole database.
It's for missing entries whose cite key originally resided in
|cite_list| at a spot that another cite key (might have) claimed.

@<Complain about missing entries whose cite keys got overwritten@>=
begin
cite_ptr := all_marker;
while (cite_ptr < old_num_cites) do
    begin
    if (not entry_exists[cite_ptr]) then
        print_missing_entry (cite_info[cite_ptr]);
    incr(cite_ptr);
    end;
end


@
@:BibTeX capacity exceeded}{\quad total number of integer entry-variables@>
This module initializes all |int_entry_var|s of all entries to 0, the
value to which all integers are initialized.

@<Initialize the |int_entry_var|s@>=
begin
if (num_ent_ints*num_cites > max_ent_ints) then
    begin
    print (num_ent_ints*num_cites,': ');
    overflow('total number of integer entry-variables ',max_ent_ints);
    end;
int_ent_ptr := 0;
while (int_ent_ptr < num_ent_ints*num_cites) do
    begin
    entry_ints[int_ent_ptr] := 0;
    incr(int_ent_ptr);
    end;
end


@
@:BibTeX capacity exceeded}{\quad total number of string entry-variables@>
This module initializes all |str_entry_var|s of all entries to the
null string, the value to which all strings are initialized.

@<Initialize the |str_entry_var|s@>=
begin
if (num_ent_strs*num_cites > max_ent_strs) then
    begin
    print (num_ent_strs*num_cites,': ');
    overflow('total number of string entry-variables ',max_ent_strs);
    end;
str_ent_ptr := 0;
while (str_ent_ptr < num_ent_strs*num_cites) do
    begin
    entry_strs[str_ent_ptr][0] := end_of_string;
    incr(str_ent_ptr);
    end;
end


@
The array |sorted_cites| initially specifies that the entries are to
be processed in order of cite-key occurrence.  The \.{sort} command
may change this to whatever it likes (which, we hope, is whatever the
style-designer instructs it to like).  We make |sorted_cites| an alias
to save space; this works fine because we're done with |cite_info|.

@d sorted_cites == cite_info    {an alias used for the rest of the program}

@<Initialize the |sorted_cites|@>=
begin
cite_ptr := 0;
while (cite_ptr < num_cites) do
    begin
    sorted_cites[cite_ptr] := cite_ptr;
    incr(cite_ptr);
    end;
end



@* Executing the style file.
This part of the program produces the output by executing the
\.{.bst}-file commands \.{execute}, \.{iterate}, \.{reverse}, and
\.{sort}.  To do this it uses a stack (consisting of the two arrays
|lit_stack| and |lit_stk_type|) for storing literals, a buffer
|ex_buf| for manipulating strings, and an array |sorted_cites|
for holding pointers to the sorted cite keys (|sorted_cites| is an
alias of |cite_info|).

@<Globals in the outer block@>=
@!lit_stack : array[lit_stk_loc] of integer;    {the literal function stack}
@!lit_stk_type : array[lit_stk_loc] of stk_type; {their corresponding types}
@!lit_stk_ptr : lit_stk_loc;    {points just above the top of the stack}
@!cmd_str_ptr : str_number;     {stores value of |str_ptr| during execution}
@!ent_chr_ptr : 0..ent_str_size; {points at a |str_entry_var| character}
@!glob_chr_ptr : 0..glob_str_size; {points at a |str_global_var| character}
@!ex_buf : buf_type;            {a buffer for manipulating strings}
@!ex_buf_ptr : buf_pointer;     {general |ex_buf| location}
@!ex_buf_length : buf_pointer;  {the length of the current string in |ex_buf|}
@!out_buf : buf_type;           {the \.{.bbl} output buffer}
@!out_buf_ptr : buf_pointer;    {general |out_buf| location}
@!out_buf_length : buf_pointer; {the length of the current string in |out_buf|}
@!mess_with_entries : boolean;  {|true| if functions can use entry info}
@!sort_cite_ptr : cite_number;  {a loop index for the sorted cite keys}
@!sort_key_num : str_ent_loc;   {index for the |str_entry_var| \.{sort.key\$}}
@!brace_level : integer;        {the brace nesting depth within a string}


@
Where |lit_stk_loc| is a stack location, and where |stk_type| gives
one of the three types of literals (an integer, a string, or a
function) or a special marker.  If a |lit_stk_type| element is a
|stk_int| then the corresponding |lit_stack| element is an integer; if
a |stk_str|, then a pointer to a |str_pool| string; and if a |stk_fn|,
then a pointer to the function's hash-table location.  However, if the
literal should have been a |stk_str| that was the value of a field
that happened to be |missing|, then the special value
|stk_field_missing| goes on the stack instead; its corresponding
|lit_stack| element is a pointer to the field-name's string.  Finally,
|stk_empty| is the type of a literal popped from an empty stack.

@d stk_int = 0          {an integer literal}
@d stk_str = 1          {a string literal}
@d stk_fn = 2           {a function literal}
@d stk_field_missing = 3 {a special marker: a field value was missing}
@d stk_empty = 4        {another: the stack was empty when this was popped}
@d last_lit_type = 4    {the same number as on the line above}

@<Types in the outer block@>=
@!lit_stk_loc = 0..lit_stk_size;        {the stack range}
@!stk_type = 0..last_lit_type;          {the literal types}


@
And the first output line requires this initialization.

@<Set initial values of key variables@>=
out_buf_length := 0;


@
When there's an error while executing \.{.bst} functions, what we do
depends on whether the function is messing with the entries.
Furthermore this error is serious enough to classify as an
|error_message| instead of a |warning_message|.  These messages (that
is, from |bst_ex_warn|) are meant both for the user and for the style
designer while debugging.

@d bst_ex_warn(#) == begin              {error while executing some function}
                     print (#);
                     bst_ex_warn_print;
                     end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_ex_warn_print;
begin
if (mess_with_entries) then
    begin
    print (' for entry ');
    print_pool_str (cur_cite_str);
    end;
print_newline;
print ('while executing-');
bst_ln_num_print;
mark_error;
end;


@
When an error is so harmless, we print a |warning_message| instead of
an |error_message|.

@d bst_mild_ex_warn(#) == begin         {error while executing some function}
                          print (#);
                          bst_mild_ex_warn_print;
                          end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_mild_ex_warn_print;
begin
if (mess_with_entries) then
    begin
    print (' for entry ');
    print_pool_str (cur_cite_str);
    end;
print_newline;
bst_warn ('while executing');                   {This does the |mark_warning|}
end;


@
It's illegal to mess with the entry information at certain times;
here's a complaint for these times.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_cant_mess_with_entries_print;
begin
bst_ex_warn ('You can''t mess with entries here');
end;


@
This module executes a single specified function once.  It can't do
anything with the entries.

@<Perform an \.{execute} command@>=
begin
init_command_execution;
mess_with_entries := false;
execute_fn (fn_loc);
check_command_execution;
end


@
This module iterates a single specified function for all entries
specified by |cite_list|.

@<Perform an \.{iterate} command@>=
begin
init_command_execution;
mess_with_entries := true;
sort_cite_ptr := 0;
while (sort_cite_ptr < num_cites) do
    begin
    cite_ptr := sorted_cites[sort_cite_ptr];
      trace
      trace_pr_pool_str (hash_text[fn_loc]);
      trace_pr (' to be iterated on ');
      trace_pr_pool_str (cur_cite_str);
      trace_pr_newline;
      ecart@/
    execute_fn (fn_loc);
    check_command_execution;
    incr(sort_cite_ptr);
    end;
end


@
This module iterates a single specified function for all entries
specified by |cite_list|, but does it in reverse order.

@<Perform a \.{reverse} command@>=
begin
init_command_execution;
mess_with_entries := true;
if (num_cites > 0) then
    begin
    sort_cite_ptr := num_cites;
    repeat
        decr(sort_cite_ptr);
        cite_ptr := sorted_cites[sort_cite_ptr];
          trace
          trace_pr_pool_str (hash_text[fn_loc]);
          trace_pr (' to be iterated in reverse on ');
          trace_pr_pool_str (cur_cite_str);
          trace_pr_newline;
          ecart@/
        execute_fn (fn_loc);
        check_command_execution;
      until (sort_cite_ptr = 0);
    end;
end


@
This module sorts the entries based on \.{sort.key\$}; it is a stable
sort.

@<Perform a \.{sort} command@>=
begin
  trace
  trace_pr_ln ('Sorting the entries');
  ecart@/
if (num_cites > 1) then
    quick_sort (0, num_cites-1);
  trace
  trace_pr_ln ('Done sorting');
  ecart@/
end


@
These next two procedures (actually, one procedures and one function,
but who's counting) are subroutines for |quick_sort|, which follows.
The |swap| procedure exchanges the two elements its arguments point
to.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure swap (@!swap1,@!swap2 : cite_number);
var innocent_bystander : cite_number;   {the temporary element in an exchange}
begin
innocent_bystander := sorted_cites[swap2];
sorted_cites[swap2] := sorted_cites[swap1];
sorted_cites[swap1] := innocent_bystander;
end;


@
@:this can't happen}{\quad Duplicate sort key@>
The function |less_than| compares the two \.{sort.key\$}s indirectly
pointed to by its arguments and returns |true| if the first argument's
\.{sort.key\$} is lexicographically less than the second's (that is,
alphabetically earlier).  In case of ties the function compares the
indices |arg1| and |arg2|, which are assumed to be different, and
returns |true| if the first is smaller.  This function uses
|ASCII_code|s to compare, so it might give ``interesting'' results
when handling nonletters.

@d compare_return(#) == begin           {the compare is finished}
                        less_than := #;
                        return;
                        end

@<Procedures and functions for handling numbers, characters, and strings@>=
function less_than (@!arg1,@!arg2 : cite_number) : boolean;
label exit;
var char_ptr : 0..ent_str_size;         {character index into compared strings}
    @!ptr1,@!ptr2 : str_ent_loc;        {the two \.{sort.key\$} pointers}
    @!char1,@!char2 : ASCII_code;       {the two characters being compared}
begin
ptr1 := arg1*num_ent_strs + sort_key_num;
ptr2 := arg2*num_ent_strs + sort_key_num;
char_ptr := 0;
loop
    begin
    char1 := entry_strs[ptr1][char_ptr];
    char2 := entry_strs[ptr2][char_ptr];
    if (char1 = end_of_string) then
        if (char2 = end_of_string) then
            if (arg1 < arg2) then
                compare_return (true)
            else if (arg1 > arg2) then
                compare_return (false)
            else                                {|arg1 = arg2|}
                confusion ('Duplicate sort key')
        else                                    {|char2 <> end_of_string|}
            compare_return (true)
    else                                        {|char1 <> end_of_string|}
        if (char2 = end_of_string) then
            compare_return (false)
    else if (char1 < char2) then
        compare_return (true)
    else if (char1 > char2) then
        compare_return (false);
    incr(char_ptr);
    end;
exit:
end;


@
The recursive procedure |quick_sort| sorts the entries indirectly
pointed to by the |sorted_cites| elements between |left_end| and
|right_end|, inclusive, based on the value of the |str_entry_var|
\.{sort.key\$}.  It's a fairly standard quicksort (for example, see
Algorithm 5.2.2Q in {\sl The Art of Computer Programming}), but uses
the median-of-three method to choose the partition element just in
case the entries are already sorted (or nearly sorted---humans and
ASCII might have different ideas on lexicographic ordering); it is a
stable sort.  This code generally prefers clarity to assembler-type
execution-time efficiency since |cite_list|s will rarely be huge.

The value |short_list|, which must be at least |2*end_offset + 2| for
this code to work, tells us the list-length at which the list is small
enough to warrant switching over to straight insertion sort from the
recursive quicksort.  The values here come from modest empirical tests
aimed at minimizing, for large |cite_list|s (five hundred or so), the
number of comparisons (between keys) plus the number of calls to
|quick_sort|.  The value |end_offset| must be positive; this helps
avoid $n^2$ behavior observed when the list starts out nearly, but not
completely, sorted (and fairly frequently large |cite_list|s come from
entire databases, which fairly frequently are nearly sorted).

@d short_list = 10      {use straight insertion sort at or below this length}
@d end_offset = 4       {the index end-offsets for choosing a median-of-three}

@<Check the ``constant'' values for consistency@>=
if (short_list < 2*end_offset + 2) then bad:=100*bad+22;


@
Here's the actual procedure.

@d next_insert = 24     {now insert the next element}

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure quick_sort (@!left_end,@!right_end : cite_number);
label next_insert;
var left,@!right : cite_number;         {two general |sorted_cites| pointers}
    @!insert_ptr : cite_number;         {the to-be-(straight)-inserted element}
    @!middle : cite_number;     {the |(left_end+right_end) div 2| element}
    @!partition : cite_number;          {the median-of-three partition element}
begin
  trace
  trace_pr_ln ('Sorting ',left_end:0,' through ',right_end:0);
  ecart@/
if (right_end - left_end < short_list) then
    @<Do a straight insertion sort@>
  else
    begin
    @<Draw out the median-of-three partition element@>;
    @<Do the partitioning and the recursive calls@>;
    end;
end;


@
This code sorts the entries between |left_end| and |right_end| when
the difference is less than |short_list|.  Each iteration of the outer
loop inserts the element indicated by |insert_ptr| into its proper
place among the (sorted) elements from |left_end| through
|insert_ptr-1|.

@<Do a straight insertion sort@>=
begin
for insert_ptr := left_end+1 to right_end do
    begin
    for right := insert_ptr downto left_end+1 do
        begin
        if (less_than (sorted_cites[right-1], sorted_cites[right])) then
            goto next_insert;
        swap (right-1, right);
        end;
next_insert:
    end;
end


@
Now we find the median of the three \.{sort.key\$}s to which the three
elements |sorted_cites[left_end+end_offset]|,
|sorted_cites[right_end]-end_offset|, and
|sorted_cites[(left_end+right_end) div 2]| point (a nonzero
|end_offset| avoids using as the leftmost of the three elements the
one that was swapped there when the old partition element was swapped
into its final spot; this turns out to avoid $n^2$ behavior when the
list is nearly sorted to start with).  This code determines which of
the six possible permutations we're dealing with and moves the median
element to |left_end|.  The comments next to the |swap| actions give
the known orderings of the corresponding elements of |sorted_cites|
before the action.

@<Draw out the median-of-three partition element@>=
begin
left := left_end + end_offset;
middle := (left_end+right_end) div 2;
right := right_end - end_offset;
if (less_than (sorted_cites[left], sorted_cites[middle])) then
  if (less_than (sorted_cites[middle], sorted_cites[right])) then
                                        {|left < middle < right|}
        swap(left_end,middle)
    else if (less_than (sorted_cites[left], sorted_cites[right])) then
                                        {|left < right < middle|}
        swap(left_end,right)
      else                              {|right < left < middle|}
        swap(left_end,left)
  else if (less_than (sorted_cites[right], sorted_cites[middle])) then
                                        {|right < middle < left|}
        swap(left_end,middle)
    else if (less_than (sorted_cites[right], sorted_cites[left])) then
                                        {|middle < right < left|}
        swap(left_end,right)
      else                              {|middle < left < right|}
        swap(left_end,left);
end


@
This module uses the median-of-three computed above to partition the
elements into those less than and those greater than the median.
Equal \.{sort.key\$}s are sorted by order of occurrence (in
|cite_list|).

@<Do the partitioning and the recursive calls@>=
begin
partition := sorted_cites[left_end];
left := left_end + 1;
right := right_end;
repeat
    while (less_than (sorted_cites[left], partition)) do
        incr(left);
    while (less_than (partition, sorted_cites[right])) do
        decr(right);
                {now |sorted_cites[right] < partition < sorted_cites[left]|}
    if (left < right) then
        begin
        swap (left,right);
        incr(left);
        decr(right);
        end;
until (left = right+1); {pointers have crossed}
swap (left_end,right);{restoring the partition element to its |right|ful place}
quick_sort (left_end,right-1);
quick_sort (left,right_end);
end


@
@:BibTeX capacity exceeded}{\quad literal-stack size@>
@:this can't happen}{\quad Unknown literal type@>
Ok, that's it for sorting; now we'll play with the literal stack.
This procedure pushes a literal onto the stack, checking for stack
overflow.

@<Procedures and functions for style-file function execution@>=
procedure push_lit_stk (@!push_lt:integer; @!push_type:stk_type);
  trace
  var dum_ptr : lit_stk_loc;    {used just as an index variable}
  ecart@/
begin
lit_stack[lit_stk_ptr] := push_lt;
lit_stk_type[lit_stk_ptr] := push_type;
  trace
  for dum_ptr := 0 to lit_stk_ptr do
    trace_pr ('  ');
  trace_pr ('Pushing ');
  case (lit_stk_type[lit_stk_ptr]) of
    stk_int : trace_pr_ln (lit_stack[lit_stk_ptr]:0);
    stk_str : begin
              trace_pr ('"');
              trace_pr_pool_str (lit_stack[lit_stk_ptr]);
              trace_pr_ln ('"');
              end;
    stk_fn : begin
             trace_pr ('`');
             trace_pr_pool_str (hash_text[lit_stack[lit_stk_ptr]]);
             trace_pr_ln ('''');
             end;
    stk_field_missing : begin
                        trace_pr ('missing field `');
                        trace_pr_pool_str (lit_stack[lit_stk_ptr]);
                        trace_pr_ln ('''');
                        end;
    stk_empty : trace_pr_ln ('a bad literal--popped from an empty stack');
    othercases unknwn_literal_confusion
  endcases;
  ecart@/
if (lit_stk_ptr = lit_stk_size) then
    overflow('literal-stack size ',lit_stk_size);
incr(lit_stk_ptr);
end;


@
@^push the literal stack@>
This macro pushes the last thing, necessarily a string, that was
popped.  And this module, along with others that push the literal
stack without explicitly calling |push_lit_stack|, have an index entry
under ``push the literal stack''; these implicit pushes collectively
speed up the program by about ten percent.

@d repush_string == begin
                    if (lit_stack[lit_stk_ptr] >= cmd_str_ptr) then
                        unflush_string;
                    incr(lit_stk_ptr);
                    end


@
@:this can't happen}{\quad Nontop top of string stack@>
This procedure pops the stack, checking for, and trying to recover
from, stack underflow.  (Actually, this procedure is really a
function, since it returns the two values through its |var|
parameters.)  Also, if the literal being popped is a |stk_str| that's
been created during the execution of the current \.{.bst} command, pop
it from |str_pool| as well (it will be the string corresponding to
|str_ptr-1|).  Note that when this happens, the string is no longer
`officially' available so that it must be used before anything else is
added to |str_pool|.

@<Procedures and functions for style-file function execution@>=
procedure pop_lit_stk (var pop_lit:integer; var pop_type:stk_type);
begin
if (lit_stk_ptr = 0) then
    begin
    bst_ex_warn ('You can''t pop an empty literal stack');@/
    pop_type := stk_empty;      {this is an error recovery attempt}
    end
  else
    begin
    decr(lit_stk_ptr);
    pop_lit := lit_stack[lit_stk_ptr];
    pop_type := lit_stk_type[lit_stk_ptr];
    if (pop_type = stk_str) then
      if (pop_lit >= cmd_str_ptr) then
        begin
        if (pop_lit <> str_ptr-1) then
            confusion ('Nontop top of string stack');
        flush_string;
        end;
    end;
end;


@
@:this can't happen}{\quad Illegal literal type@>
@:this can't happen}{\quad Unknown literal type@>
More bug complaints, this time about bad literals.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure illegl_literal_confusion;
begin
confusion ('Illegal literal type');
end;
@#
procedure unknwn_literal_confusion;
begin
confusion ('Unknown literal type');
end;


@
@:this can't happen}{\quad Illegal literal type@>
@:this can't happen}{\quad Unknown literal type@>
Occasionally we'll want to know what's on the literal stack.  Here we
print out a stack literal, giving its type.  This procedure should
never be called after popping an empty stack.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_stk_lit (@!stk_lt:integer; @!stk_tp:stk_type);
begin
case (stk_tp) of
    stk_int : print (stk_lt:0,' is an integer literal');
    stk_str : begin
              print ('"');
              print_pool_str (stk_lt);
              print ('" is a string literal');
              end;
    stk_fn : begin
             print ('`');
             print_pool_str (hash_text[stk_lt]);
             print (''' is a function literal');
             end;
    stk_field_missing : begin
                        print ('`');
                        print_pool_str (stk_lt);
                        print (''' is a missing field');
                        end;
    stk_empty : illegl_literal_confusion;
    othercases unknwn_literal_confusion
endcases;
end;


@
@:this can't happen}{\quad Illegal literal type@>
@:this can't happen}{\quad Unknown literal type@>
This procedure appropriately chastises the style designer; however, if
the wrong literal came from popping an empty stack, the procedure
|pop_lit_stack| will have already done the chastising (because this
procedure is called only after popping the stack) so there's no need
for more.

@<Procedures and functions for style-file function execution@>=
procedure print_wrong_stk_lit (@!stk_lt:integer; @!stk_tp1,@!stk_tp2:stk_type);
begin
if (stk_tp1 <> stk_empty) then
    begin
    print_stk_lit (stk_lt, stk_tp1);
    case (stk_tp2) of
        stk_int : print (', not an integer,');
        stk_str : print (', not a string,');
        stk_fn : print (', not a function,');
        stk_field_missing,
        stk_empty : illegl_literal_confusion;
        othercases unknwn_literal_confusion
    endcases;
    bst_ex_warn_print;
    end;
end;


@
@:this can't happen}{\quad Illegal literal type@>
@:this can't happen}{\quad Unknown literal type@>
This is similar to |print_stk_lit|, but here we don't give the
literal's type, and here we end with a new line.  This procedure
should never be called after popping an empty stack.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure print_lit (@!stk_lt:integer; @!stk_tp:stk_type);
begin
case (stk_tp) of
    stk_int : print_ln (stk_lt:0);
    stk_str : begin
              print_pool_str (stk_lt);
              print_newline;
              end;
    stk_fn : begin
             print_pool_str (hash_text[stk_lt]);
             print_newline;
             end;
    stk_field_missing : begin
                        print_pool_str (stk_lt);
                        print_newline;
                        end;
    stk_empty : illegl_literal_confusion;
    othercases unknwn_literal_confusion
endcases;
end;


@
This procedure pops and prints the top of the stack; when the stack is
empty the procedure |pop_lit_stk| complains.

@<Procedures and functions for style-file function execution@>=
procedure pop_top_and_print;
var stk_lt : integer;
  @!stk_tp : stk_type;
begin
pop_lit_stk (stk_lt,stk_tp);
if (stk_tp = stk_empty) then
    print_ln ('Empty literal')
  else
    print_lit (stk_lt,stk_tp);
end;


@
This procedure pops and prints the whole stack.

@<Procedures and functions for style-file function execution@>=
procedure pop_whole_stack;
begin
while (lit_stk_ptr > 0) do
    pop_top_and_print;
end;


@
At the beginning of a \.{.bst}-command execution we make the stack
empty and record how much of |str_pool| has been used.

@<Procedures and functions for style-file function execution@>=
procedure init_command_execution;
begin
lit_stk_ptr := 0;       {make the stack empty}
cmd_str_ptr := str_ptr; {we'll check this when we finish command execution}
end;


@
@:this can't happen}{\quad Nonempty empty string stack@>
At the end of a \.{.bst} command-execution we check that the stack and
|str_pool| are still in good shape.

@<Procedures and functions for style-file function execution@>=
procedure check_command_execution;
begin
if (lit_stk_ptr<>0) then
    begin
    print_ln ('ptr=',lit_stk_ptr:0,', stack=');
    pop_whole_stack;
    bst_ex_warn ('---the literal stack isn''t empty');
    end;
if (cmd_str_ptr<>str_ptr) then
    begin
      trace
      print_ln ('Pointer is ',str_ptr:0,' but should be ',cmd_str_ptr:0);
      ecart@/
    confusion ('Nonempty empty string stack');
    end;
end;


@
This procedure adds to |str_pool| the string from |ex_buf[0]| through
|ex_buf[ex_buf_length-1]| if it will fit.  It assumes the global
variable |ex_buf_length| gives the length of the current string in
|ex_buf|.  It then pushes this string onto the literal stack.

@<Procedures and functions for style-file function execution@>=
procedure add_pool_buf_and_push;
begin
str_room (ex_buf_length);               {make sure this string will fit}
ex_buf_ptr := 0;
while (ex_buf_ptr < ex_buf_length) do
    begin
    append_char (ex_buf[ex_buf_ptr]);
    incr(ex_buf_ptr);
    end;
push_lit_stk (make_string, stk_str);    {and push it onto the stack}
end;


@
@:BibTeX capacity exceeded}{\quad buffer size@>
These macros append a character to |ex_buf|.  Which is called depends
on whether the character is known to fit.

@d append_ex_buf_char(#) == begin
                            ex_buf[ex_buf_ptr] := #;
                            incr(ex_buf_ptr);
                            end
@#
@d append_ex_buf_char_and_check(#) ==
                            begin
                            if (ex_buf_ptr = buf_size) then
                                buffer_overflow;
                            append_ex_buf_char(#);
                            end


@
@:BibTeX capacity exceeded}{\quad buffer size@>
This procedure adds to the execution buffer the given string in
|str_pool| if it will fit.  It assumes the global variable
|ex_buf_length| gives the length of the current string in |ex_buf|,
and thus also gives the location of the next character.

@<Procedures and functions for style-file function execution@>=
procedure add_buf_pool (@!p_str : str_number);
begin
p_ptr1 := str_start[p_str];
p_ptr2 := str_start[p_str+1];
if (ex_buf_length+(p_ptr2-p_ptr1) > buf_size) then
    buffer_overflow;
ex_buf_ptr := ex_buf_length;
while (p_ptr1 < p_ptr2) do
    begin                       {copy characters into the buffer}
    append_ex_buf_char (str_pool[p_ptr1]);
    incr(p_ptr1);
    end;
ex_buf_length := ex_buf_ptr;
end;


@
This procedure actually writes onto the \.{.bbl}~file a line of output
(the characters from |out_buf[0]| to |out_buf[out_buf_length-1]|,
after removing trailing |white_space| characters).  It also updates
|bbl_line_num|, the line counter.  It writes a blank line if and only
if |out_buf| is empty.  The program uses this procedure in such a way
that |out_buf| will be nonempty if there have been characters put in
it since the most recent \.{newline\$}.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure output_bbl_line;
label loop_exit,@!exit;
begin
if (out_buf_length <> 0) then           {the buffer's not empty}
    begin
    while (out_buf_length > 0) do       {remove trailing |white_space|}
      if (lex_class[out_buf[out_buf_length-1]] = white_space) then
        decr(out_buf_length)
       else
        goto loop_exit;
loop_exit:
    if (out_buf_length = 0) then        {ignore a line of just |white_space|}
        return;
    out_buf_ptr := 0;
    while (out_buf_ptr < out_buf_length) do
        begin
        write (bbl_file, xchr[out_buf[out_buf_ptr]]);
        incr(out_buf_ptr);
        end;
    end;
write_ln (bbl_file);
incr(bbl_line_num);     {update line number}
out_buf_length := 0;    {make the next line empty}
exit:
end;


@
@:BibTeX capacity exceeded}{\quad output buffer size@>
This procedure adds to the output buffer the given string in
|str_pool|.  It assumes the global variable |out_buf_length| gives the
length of the current string in |out_buf|, and thus also gives the
location for the next character.  If there are enough characters
present in the output buffer, it writes one or more lines out to the
\.{.bbl} file.  It breaks a line only at a |white_space| character,
and when it does, it adds two |space|s to the next output line.

@<Procedures and functions for style-file function execution@>=
procedure add_out_pool (@!p_str : str_number);
label loop1_exit,loop2_exit;
var break_ptr : buf_pointer;    {the first character following the line break}
@!end_ptr : buf_pointer;        {temporary end-of-buffer pointer}
break_pt_found : boolean;       {a suitable |white_space| character}
unbreakable_tail : boolean;     {as it contains no |white_space| character}
begin
p_ptr1 := str_start[p_str];
p_ptr2 := str_start[p_str+1];
if (out_buf_length+(p_ptr2-p_ptr1) > buf_size) then
    overflow('output buffer size ',buf_size);
out_buf_ptr := out_buf_length;
while (p_ptr1 < p_ptr2) do
    begin                       {copy characters into the buffer}
    out_buf[out_buf_ptr] := str_pool[p_ptr1];
    incr(p_ptr1);
    incr(out_buf_ptr);
    end;
out_buf_length := out_buf_ptr;
unbreakable_tail := false;
while ((out_buf_length > max_print_line) and (not unbreakable_tail)) do
    @<Break that line@>;
end;


@
Here we break the line by looking for a |white_space| character,
backwards from |out_buf[max_print_line]| until
|out_buf[min_print_line]|; we break at the |white_space| and indent
the next line two |space|s.  The next module handles things when
there's no |white_space| character to break at.  (It seems that the
annoyances to the average user of a warning message when there's an
output line longer than |max_print_line| outweigh the benefits, so we
don't issue such warnings in the current code.)

@<Break that line@>=
begin
end_ptr := out_buf_length;
out_buf_ptr := max_print_line;
break_pt_found := false;
while ((lex_class[out_buf[out_buf_ptr]] <> white_space) and
                                        (out_buf_ptr >= min_print_line)) do
    decr(out_buf_ptr);
if (out_buf_ptr = min_print_line-1) then        {no |white_space| character}
    @<Break that unbreakably long line@>        {(if |white_space| follows)}
else
    break_pt_found := true;                     {hit a |white_space| character}
if (break_pt_found) then
    begin
    out_buf_length := out_buf_ptr;
    break_ptr := out_buf_length + 1;
    output_bbl_line;                    {output what we can}
    out_buf[0] := space;
    out_buf[1] := space;                {start the next line with two |space|s}
    out_buf_ptr := 2;
    tmp_ptr := break_ptr;
    while (tmp_ptr < end_ptr) do        {and slide the rest down}
        begin
        out_buf[out_buf_ptr] := out_buf[tmp_ptr];
        incr(out_buf_ptr);
        incr(tmp_ptr);
        end;
    out_buf_length := end_ptr - break_ptr + 2;
    end;
end


@
If there's no |white_space| character up through
|out_buf[max_print_line]|, we instead break the line at the first
following |white_space| character, if one exists.  And if, starting
with that |white_space| character, there are multiple consecutive
|white_space| characters, |out_buf_ptr| points to the last of them.
If no |white_space| character exists, we haven't found a viable break
point, so we don't break the line (yet).

@<Break that unbreakably long line@>=
begin
out_buf_ptr := max_print_line + 1;      {|break_pt_found| is still |false|}
while (out_buf_ptr < end_ptr) do
    if (lex_class[out_buf[out_buf_ptr]] <> white_space) then
        incr(out_buf_ptr)
    else
        goto loop1_exit;
loop1_exit:
if (out_buf_ptr = end_ptr) then
    unbreakable_tail := true            {because no |white_space| character}
else                            {at |white_space|, and |out_buf_ptr < end_ptr|}
    begin
    break_pt_found := true;
    while (out_buf_ptr+1 < end_ptr) do  {look for more |white_space|}
        if (lex_class[out_buf[out_buf_ptr+1]] = white_space) then
            incr(out_buf_ptr)           {which then points to |white_space|}
        else
            goto loop2_exit;
loop2_exit:
    end;
end


@
@^Tuesdays@>
@^windows@>
@:this can't happen}{\quad Unknown function class@>
This procedure executes a single specified function; it is the single
execution-primitive that does everything (except windows, and it takes
Tuesdays off).

@<|execute_fn| itself@>=
procedure execute_fn (@!ex_fn_loc : hash_loc);
@<Declarations for executing |built_in| functions@>
@!wiz_ptr : wiz_fn_loc;         {general |wiz_functions| location}
begin
  trace
  trace_pr ('execute_fn `');
  trace_pr_pool_str (hash_text[ex_fn_loc]);
  trace_pr_ln ('''');
  ecart@/
case (fn_type[ex_fn_loc]) of
    built_in : @<Execute a |built_in| function@>;
    wiz_defined : @<Execute a |wiz_defined| function@>;
    int_literal : push_lit_stk (fn_info[ex_fn_loc], stk_int);
    str_literal : push_lit_stk (hash_text[ex_fn_loc], stk_str);
    field : @<Execute a field@>;
    int_entry_var : @<Execute an |int_entry_var|@>;
    str_entry_var : @<Execute a |str_entry_var|@>;
    int_global_var : push_lit_stk (fn_info[ex_fn_loc], stk_int);
    str_global_var : @<Execute a |str_global_var|@>;
    othercases unknwn_function_class_confusion
endcases;
end;


@
To execute a |wiz_defined| function, we just execute all those
functions in its definition, except that the special marker
|quote_next_fn| means we push the next function onto the stack.

@<Execute a |wiz_defined| function@>=
begin
wiz_ptr := fn_info[ex_fn_loc];
while (wiz_functions[wiz_ptr] <> end_of_def) do
    begin
    if (wiz_functions[wiz_ptr] <> quote_next_fn) then
        execute_fn (wiz_functions[wiz_ptr])
      else
        begin
        incr(wiz_ptr);
        push_lit_stk (wiz_functions[wiz_ptr], stk_fn);
        end;
    incr(wiz_ptr);
    end;
end


@
This module pushes the string given by the field onto the literal
stack unless it's |missing|, in which case it pushes a special value
onto the stack.

@<Execute a field@>=
begin
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
  else
    begin
    field_ptr := cite_ptr*num_fields + fn_info[ex_fn_loc];
    if (field_info[field_ptr] = missing) then
        push_lit_stk (hash_text[ex_fn_loc], stk_field_missing)
      else
        push_lit_stk (field_info[field_ptr], stk_str);
    end
end


@
This module pushes the integer given by an |int_entry_var| onto the
literal stack.

@<Execute an |int_entry_var|@>=
begin
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
  else
    push_lit_stk (entry_ints[cite_ptr*num_ent_ints+fn_info[ex_fn_loc]],
                                                                stk_int);
end


@
This module adds the string given by a |str_entry_var| to |str_pool|
via the execution buffer and pushes it onto the literal stack.

@<Execute a |str_entry_var|@>=
begin
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
  else
    begin
    str_ent_ptr := cite_ptr*num_ent_strs + fn_info[ex_fn_loc];@/
    ex_buf_ptr := 0;                    {also serves as |ent_chr_ptr|}
    while (entry_strs[str_ent_ptr][ex_buf_ptr] <> end_of_string) do
                                        {copy characters into the buffer}
        append_ex_buf_char (entry_strs[str_ent_ptr][ex_buf_ptr]);
    ex_buf_length := ex_buf_ptr;
    add_pool_buf_and_push;              {push this string onto the stack}
    end;
end


@
This module pushes the string given by a |str_global_var| onto the
literal stack, but it copies the string to |str_pool| (character by
character) only if it has to---it {\it doesn't\/} have to if the
string is static (that is, if the string isn't at the top, temporary
part of the string pool).

@<Execute a |str_global_var|@>=
begin
str_glb_ptr := fn_info[ex_fn_loc];
if (glb_str_ptr[str_glb_ptr] > 0) then  {we're dealing with a static string}
    push_lit_stk (glb_str_ptr[str_glb_ptr],stk_str)
  else
    begin
    str_room(glb_str_end[str_glb_ptr]);
    glob_chr_ptr := 0;
    while (glob_chr_ptr < glb_str_end[str_glb_ptr]) do  {copy the string}
        begin
        append_char (global_strs[str_glb_ptr][glob_chr_ptr]);
        incr(glob_chr_ptr);
        end;
    push_lit_stk (make_string, stk_str);        {and push it onto the stack}
    end;
end



@* The built-in functions.
@^add a built-in function@>
@^biblical procreation@>
@^grade inflation@>
This section gives the all the code for all the built-in functions
(including pre-defined |field|s, |str_entry_var|s, and
|int_global_var|s, which technically aren't classified as |built_in|).
To modify or add one, we needn't go anywhere else (with one exception:
The constant |max_pop|, which gives the maximum number of literals
that any of these functions pops off the stack, is defined earlier
because it's needed earlier; thus, if we need to update it, which will
happen if some new |built_in| functions uses more than |max_pop|
literals from the stack, we'll have to go outside this section).
Adding a |built_in| function entails modifying (at least four of) the
five modules marked by ``add a built-in function'' in the index, in
addition to adding the code to execute the function.

These variables all begin with |b_| and specify the hash-table
locations of the |built_in| functions, except that |b_default| is
pseudo-|built_in|---either it will point to the no-op \.{skip\$} or to
the \.{.bst}-defined function \.{default.type}; it's used when an
entry has a type that's not defined in the \.{.bst} file.

@<Globals in the outer block@>=
@!b_equals : hash_loc;          {\.{=}}
@!b_greater_than : hash_loc;    {\.{>}}
@!b_less_than : hash_loc;       {\.{<}}
@!b_plus : hash_loc;            {\.{+} (this may be changed to an |a_minus|)}
@!b_minus : hash_loc;           {\.{-}}
@!b_concatenate : hash_loc;     {\.{*}}
@!b_gets : hash_loc;            {\.{:=} (formerly, |b_gat|)}
@!b_add_period : hash_loc;      {\.{add.period\$}}
@!b_call_type : hash_loc;       {\.{call.type\$}}
@!b_change_case : hash_loc;     {\.{change.case\$}}
@!b_chr_to_int : hash_loc;      {\.{chr.to.int\$}}
@!b_cite : hash_loc;            {\.{cite\$}}
@!b_duplicate : hash_loc;       {\.{duplicate\$}}
@!b_empty : hash_loc;           {\.{empty\$}}
@!b_format_name : hash_loc;     {\.{format.name\$}}
@!b_if : hash_loc;              {\.{if\$}}
@!b_int_to_chr : hash_loc;      {\.{int.to.chr\$}}
@!b_int_to_str : hash_loc;      {\.{int.to.str\$}}
@!b_missing : hash_loc;         {\.{missing\$}}
@!b_newline : hash_loc;         {\.{newline\$}}
@!b_num_names : hash_loc;       {\.{num.names\$}}
@!b_pop : hash_loc;             {\.{pop\$}}
@!b_preamble : hash_loc;        {\.{preamble\$}}
@!b_purify : hash_loc;          {\.{purify\$}}
@!b_quote : hash_loc;           {\.{quote\$}}
@!b_skip : hash_loc;            {\.{skip\$}}
@!b_stack : hash_loc;           {\.{stack\$}}
@!b_substring : hash_loc;       {\.{substring\$}}
@!b_swap : hash_loc;            {\.{swap\$}}
@!b_text_length : hash_loc;     {\.{text.length\$}}
@!b_text_prefix : hash_loc;     {\.{text.prefix\$}}
@!b_top_stack : hash_loc;       {\.{top\$}}
@!b_type : hash_loc;            {\.{type\$}}
@!b_warning : hash_loc;         {\.{warning\$}}
@!b_while : hash_loc;           {\.{while\$}}
@!b_width : hash_loc;           {\.{width\$}}
@!b_write : hash_loc;           {\.{write\$}}
@!b_default : hash_loc;         {either \.{skip\$} or \.{default.type}}
@#
  stat
  @!blt_in_loc : array[blt_in_range] of hash_loc; {for execution counts}
  @!execution_count : array[blt_in_range] of integer; {the same}
  @!total_ex_count : integer;           {the sum of all |execution_count|s}
  @!blt_in_ptr : blt_in_range;          {a pointer into |blt_in_loc|}
  tats@/


@
Where |blt_in_range| gives the legal |built_in| function numbers.

@<Types in the outer block@>=
@!blt_in_range = 0..num_blt_in_fns;


@
@^add a built-in function@>
These constants all begin with |n_| and are used for the |case|
statement that determines which |built_in| function to execute.

@d n_equals = 0         {\.{=}}
@d n_greater_than = 1   {\.{>}}
@d n_less_than = 2      {\.{<}}
@d n_plus = 3           {\.{+}}
@d n_minus = 4          {\.{-}}
@d n_concatenate = 5    {\.{*}}
@d n_gets = 6           {\.{:=}}
@d n_add_period = 7     {\.{add.period\$}}
@d n_call_type = 8      {\.{call.type\$}}
@d n_change_case = 9    {\.{change.case\$}}
@d n_chr_to_int = 10    {\.{chr.to.int\$}}
@d n_cite = 11          {\.{cite\$} (this may start a riot)}
@d n_duplicate = 12     {\.{duplicate\$}}
@d n_empty = 13         {\.{empty\$}}
@d n_format_name = 14   {\.{format.name\$}}
@d n_if = 15            {\.{if\$}}
@d n_int_to_chr = 16    {\.{int.to.chr\$}}
@d n_int_to_str = 17    {\.{int.to.str\$}}
@d n_missing = 18       {\.{missing\$}}
@d n_newline = 19       {\.{newline\$}}
@d n_num_names = 20     {\.{num.names\$}}
@d n_pop = 21           {\.{pop\$}}
@d n_preamble = 22      {\.{preamble\$}}
@d n_purify = 23        {\.{purify\$}}
@d n_quote = 24         {\.{quote\$}}
@d n_skip = 25          {\.{skip\$}}
@d n_stack = 26         {\.{stack\$}}
@d n_substring = 27     {\.{substring\$}}
@d n_swap = 28          {\.{swap\$}}
@d n_text_length = 29   {\.{text.length\$}}
@d n_text_prefix = 30   {\.{text.prefix\$}}
@d n_top_stack = 31     {\.{top\$}}
@d n_type = 32          {\.{type\$}}
@d n_warning = 33       {\.{warning\$}}
@d n_while = 34         {\.{while\$}}
@d n_width = 35         {\.{width\$}}
@d n_write = 36         {\.{write\$}}

@<Constants in the outer block@>=
@!num_blt_in_fns = 37;  {one more than the previous number}


@
@^add a built-in function@>
@^important note@>
It's time for us to insert more pre-defined strings into |str_pool|
(and thus the hash table) and to insert the |built_in| functions into
the hash table.  The strings corresponding to these functions should
contain no upper-case letters, and they must all be exactly
|longest_pds| characters long.  The |build_in| routine (to appear
shortly) does the work.

Important note: These pre-definitions must not have any glitches or the
program may bomb because the |log_file| hasn't been opened yet.

@<Pre-define certain strings@>=
build_in('=           ',1,b_equals,n_equals);
build_in('>           ',1,b_greater_than,n_greater_than);
build_in('<           ',1,b_less_than,n_less_than);
build_in('+           ',1,b_plus,n_plus);
build_in('-           ',1,b_minus,n_minus);
build_in('*           ',1,b_concatenate,n_concatenate);
build_in(':=          ',2,b_gets,n_gets);
build_in('add.period$ ',11,b_add_period,n_add_period);
build_in('call.type$  ',10,b_call_type,n_call_type);
build_in('change.case$',12,b_change_case,n_change_case);
build_in('chr.to.int$ ',11,b_chr_to_int,n_chr_to_int);
build_in('cite$       ',5,b_cite,n_cite);
build_in('duplicate$  ',10,b_duplicate,n_duplicate);
build_in('empty$      ',6,b_empty,n_empty);
build_in('format.name$',12,b_format_name,n_format_name);
build_in('if$         ',3,b_if,n_if);
build_in('int.to.chr$ ',11,b_int_to_chr,n_int_to_chr);
build_in('int.to.str$ ',11,b_int_to_str,n_int_to_str);
build_in('missing$    ',8,b_missing,n_missing);
build_in('newline$    ',8,b_newline,n_newline);
build_in('num.names$  ',10,b_num_names,n_num_names);
build_in('pop$        ',4,b_pop,n_pop);
build_in('preamble$   ',9,b_preamble,n_preamble);
build_in('purify$     ',7,b_purify,n_purify);
build_in('quote$      ',6,b_quote,n_quote);
build_in('skip$       ',5,b_skip,n_skip);
build_in('stack$      ',6,b_stack,n_stack);
build_in('substring$  ',10,b_substring,n_substring);
build_in('swap$       ',5,b_swap,n_swap);
build_in('text.length$',12,b_text_length,n_text_length);
build_in('text.prefix$',12,b_text_prefix,n_text_prefix);
build_in('top$        ',4,b_top_stack,n_top_stack);
build_in('type$       ',5,b_type,n_type);
build_in('warning$    ',8,b_warning,n_warning);
build_in('width$      ',6,b_width,n_width);
build_in('while$      ',6,b_while,n_while);
build_in('width$      ',6,b_width,n_width);
build_in('write$      ',6,b_write,n_write);


@
This procedure inserts a |built_in| function into the hash table and
initializes the corresponding pre-defined string (of length at most
|longest_pds|).  The array |fn_info| contains a number from 0 through
the number of |built_in| functions minus 1 (i.e., |num_blt_in_fns - 1|
if we're keeping statistics); this number is used by a |case|
statement to execute this function and is used for keeping execution
counts when keeping statistics.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure build_in (@!pds:pds_type; @!len:pds_len; var fn_hash_loc:hash_loc;
                                        @!blt_in_num:blt_in_range);
begin
pre_define (pds,len,bst_fn_ilk);@/
fn_hash_loc := pre_def_loc;     {the |pre_define| routine sets |pre_def_loc|}
fn_type[fn_hash_loc] := built_in;
fn_info[fn_hash_loc] := blt_in_num;
  stat
  blt_in_loc[blt_in_num] := fn_hash_loc;@/
  execution_count[blt_in_num] := 0; {initialize the function-execution count}
  tats@/
end;


@
This is a procedure so that |initialize| is smaller.

@<Procedures and functions for handling numbers, characters, and strings@>=
procedure pre_def_certain_strings;
begin
@<Pre-define certain strings@>@;
end;


@
These variables all begin with |s_| and specify the locations in
|str_pool| of certain often-used strings that the \.{.bst} commands
need.  The |s_preamble| array is big enough to allow an average of one
\.{preamble\$} command per \.{.bib} file.

@<Globals in the outer block@>=
@!s_null : str_number;          {the null string}
@!s_default : str_number;       {\.{default.type}, for unknown entry types}
@!s_t : str_number;             {\.{t}, for |title_lowers| case conversion}
@!s_l : str_number;             {\.{l}, for |all_lowers| case conversion}
@!s_u : str_number;             {\.{u}, for |all_uppers| case conversion}
@!s_preamble : array[bib_number] of str_number;
                                {for the \.{preamble\$} |built_in| function}


@
These constants all begin with |n_| and are used for the |case|
statement that determines which, if any, control sequence we're
dealing with; a control sequence of interest will be either one of the
undotted characters `\.{\\i}' or `\.{\\j}' or one of the foreign
characters in Table~3.2 of the \LaTeX\ manual.

@d n_i = 0              {\.{i}, for the undotted character \.{\\i}}
@d n_j = 1              {\.{j}, for the undotted character \.{\\j}}
@d n_oe = 2             {\.{oe}, for the foreign character \.{\\oe}}
@d n_oe_upper = 3       {\.{OE}, for the foreign character \.{\\OE}}
@d n_ae = 4             {\.{ae}, for the foreign character \.{\\ae}}
@d n_ae_upper = 5       {\.{AE}, for the foreign character \.{\\AE}}
@d n_aa = 6             {\.{aa}, for the foreign character \.{\\aa}}
@d n_aa_upper = 7       {\.{AA}, for the foreign character \.{\\AA}}
@d n_o = 8              {\.{o}, for the foreign character \.{\\o}}
@d n_o_upper = 9        {\.{O}, for the foreign character \.{\\O}}
@d n_l = 10             {\.{l}, for the foreign character \.{\\l}}
@d n_l_upper = 11       {\.{L}, for the foreign character \.{\\L}}
@d n_ss = 12            {\.{ss}, for the foreign character \.{\\ss}}


@
@^important note@>
@.default.type@>
Here we pre-define a few strings used in executing the \.{.bst} file:
the null string, which is sometimes pushed onto the stack; a string
used for default entry types; and some control sequences used to spot
foreign characters.  We also initialize the |s_preamble| array to
empty.  These pre-defined strings must all be exactly |longest_pds|
characters long.

Important note: These pre-definitions must not have any glitches or
the program may bomb because the |log_file| hasn't been opened yet,
and |text_ilk|s should be pre-defined here, not earlier, for
\.{.bst}-function-execution purposes.

@<Pre-define certain strings@>=
pre_define('            ',0,text_ilk);  s_null := hash_text[pre_def_loc];
fn_type[pre_def_loc] := str_literal;@/
pre_define('default.type',12,text_ilk); s_default := hash_text[pre_def_loc];
fn_type[pre_def_loc] := str_literal;@/
b_default := b_skip;    {this may be changed to the \.{default.type} function}
preamble_ptr := 0;                      {initialize the |s_preamble| array}
pre_define('i           ',1,control_seq_ilk);
ilk_info[pre_def_loc] := n_i;
pre_define('j           ',1,control_seq_ilk);
ilk_info[pre_def_loc] := n_j;
pre_define('oe          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_oe;
pre_define('OE          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_oe_upper;
pre_define('ae          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_ae;
pre_define('AE          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_ae_upper;
pre_define('aa          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_aa;
pre_define('AA          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_aa_upper;
pre_define('o           ',1,control_seq_ilk);
ilk_info[pre_def_loc] := n_o;
pre_define('O           ',1,control_seq_ilk);
ilk_info[pre_def_loc] := n_o_upper;
pre_define('l           ',1,control_seq_ilk);
ilk_info[pre_def_loc] := n_l;
pre_define('L           ',1,control_seq_ilk);
ilk_info[pre_def_loc] := n_l_upper;
pre_define('ss          ',2,control_seq_ilk);
ilk_info[pre_def_loc] := n_ss;


@
@^important note@>
@.crossref@>
@.entry.max\$@>
@.global.max\$@>
@.sort.key\$@>
Now we pre-define any built-in |field|s, |str_entry_var|s, and
|int_global_var|s; these strings must all be exactly |longest_pds|
characters long.  Note that although these are built-in functions, we
classify them (in the |fn_type| array) otherwise.

Important note: These pre-definitions must not have any glitches or
the program may bomb because the |log_file| hasn't been opened yet.

@<Pre-define certain strings@>=
pre_define('crossref    ',8,bst_fn_ilk);
fn_type[pre_def_loc] := field;@/
fn_info[pre_def_loc] := num_fields;     {give this |field| a number}
crossref_num := num_fields;
incr(num_fields);@/
num_pre_defined_fields := num_fields;   {that's it for pre-defined |field|s}
pre_define('sort.key$   ',9,bst_fn_ilk);
fn_type[pre_def_loc] := str_entry_var;
fn_info[pre_def_loc] := num_ent_strs;   {give this |str_entry_var| a number}
sort_key_num := num_ent_strs;
incr(num_ent_strs);@/
pre_define('entry.max$  ',10,bst_fn_ilk);
fn_type[pre_def_loc] := int_global_var;
fn_info[pre_def_loc] := ent_str_size;   {initialize this |int_global_var|}
pre_define('global.max$ ',11,bst_fn_ilk);
fn_type[pre_def_loc] := int_global_var;
fn_info[pre_def_loc] := glob_str_size;  {initialize this |int_global_var|}


@
@^add a built-in function@>
@:this can't happen}{\quad Unknown built-in function@>
This module branches to the code for the appropriate |built_in|
function.  Only three---{\.{call.type\$}}, {\.{if\$}}, and
{\.{while\$}}---do a recursive call.

@<Execute a |built_in| function@>=
begin
  stat          {update this function's execution count}
  incr(execution_count[fn_info[ex_fn_loc]]);
  tats@/
case (fn_info[ex_fn_loc]) of
    n_equals :          x_equals;
    n_greater_than :    x_greater_than;
    n_less_than :       x_less_than;
    n_plus :            x_plus;
    n_minus :           x_minus;
    n_concatenate :     x_concatenate;
    n_gets :            x_gets;
    n_add_period :      x_add_period;
    n_call_type :       @<|execute_fn|({\.{call.type\$}})@>;
    n_change_case :     x_change_case;
    n_chr_to_int :      x_chr_to_int;
    n_cite :            x_cite;
    n_duplicate :       x_duplicate;
    n_empty :           x_empty;
    n_format_name :     x_format_name;
    n_if :              @<|execute_fn|({\.{if\$}})@>;
    n_int_to_chr :      x_int_to_chr;
    n_int_to_str :      x_int_to_str;
    n_missing :         x_missing;
    n_newline :         @<|execute_fn|({\.{newline\$}})@>;
    n_num_names :       x_num_names;
    n_pop :             @<|execute_fn|({\.{pop\$}})@>;
    n_preamble :        x_preamble;
    n_purify :          x_purify;
    n_quote :           x_quote;
    n_skip :            @<|execute_fn|({\.{skip\$}})@>;
    n_stack :           @<|execute_fn|({\.{stack\$}})@>;
    n_substring :       x_substring;
    n_swap :            x_swap;
    n_text_length :     x_text_length;
    n_text_prefix :     x_text_prefix;
    n_top_stack :       @<|execute_fn|({\.{top\$}})@>;
    n_type :            x_type;
    n_warning :         x_warning;
    n_while :           @<|execute_fn|({\.{while\$}})@>;
    n_width :           x_width;
    n_write :           x_write;
    othercases confusion ('Unknown built-in function')
endcases;
end


@
@^add a built-in function@>
@^gymnastics@>
This extra level of module-pointing allows a uniformity of module
names for the |built_in| functions, regardless of whether they do a
recursive call to |execute_fn| or are trivial (a single statement).
Those that do a recursive call are left as part of |execute_fn|,
avoiding \PASCAL's forward procedure mechanism, and those that don't
(except for the single-statement ones) are made into procedures so
that |execute_fn| doesn't get too large.

@<Procedures and functions for style-file function execution@>=
@<|execute_fn|({\.{=}})@>@;
@<|execute_fn|({\.{>}})@>@;
@<|execute_fn|({\.{<}})@>@;
@<|execute_fn|({\.{+}})@>@;
@<|execute_fn|({\.{-}})@>@;
@<|execute_fn|({\.{*}})@>@;
@<|execute_fn|({\.{:=}})@>@;
@<|execute_fn|({\.{add.period\$}})@>@;
@<|execute_fn|({\.{change.case\$}})@>@;
@<|execute_fn|({\.{chr.to.int\$}})@>@;
@<|execute_fn|({\.{cite\$}})@>@;
@<|execute_fn|({\.{duplicate\$}})@>@;
@<|execute_fn|({\.{empty\$}})@>@;
@<|execute_fn|({\.{format.name\$}})@>@;
@<|execute_fn|({\.{int.to.chr\$}})@>@;
@<|execute_fn|({\.{int.to.str\$}})@>@;
@<|execute_fn|({\.{missing\$}})@>@;
@<|execute_fn|({\.{num.names\$}})@>@;
@<|execute_fn|({\.{preamble\$}})@>@;
@<|execute_fn|({\.{purify\$}})@>@;
@<|execute_fn|({\.{quote\$}})@>@;
@<|execute_fn|({\.{substring\$}})@>@;
@<|execute_fn|({\.{swap\$}})@>@;
@<|execute_fn|({\.{text.length\$}})@>@;
@<|execute_fn|({\.{text.prefix\$}})@>@;
@<|execute_fn|({\.{type\$}})@>@;
@<|execute_fn|({\.{warning\$}})@>@;
@<|execute_fn|({\.{width\$}})@>@;
@<|execute_fn|({\.{write\$}})@>@;
@<|execute_fn| itself@>


@
Now it's time to declare some things for executing |built_in|
functions only.  These (and only these) variables are used
recursively, so they can't be global.

@d end_while = 51       {stop executing the \.{while\$} function}

@<Declarations for executing |built_in| functions@>=
label end_while;
var r_pop_lt1,@!r_pop_lt2 : integer;    {stack literals for \.{while\$}}
@!r_pop_tp1,@!r_pop_tp2 : stk_type;     {stack types for \.{while\$}}


@
These are nonrecursive variables that |execute_fn| uses.  Declaring
them here (instead of in the previous module) saves execution time and
stack space on most machines.

@d name_buf == sv_buffer        {an alias, a buffer for manipulating names}

@<Globals in the outer block@>=
@!pop_lit1,@!pop_lit2,@!pop_lit3 : integer;     {stack literals}
@!pop_typ1,@!pop_typ2,@!pop_typ3 : stk_type;    {stack types}
@!sp_ptr : pool_pointer;                {for manipulating |str_pool| strings}
@!sp_xptr1,@!sp_xptr2 : pool_pointer;   {more of the same}
@!sp_end : pool_pointer;                {marks the end of a |str_pool| string}
@!sp_length,sp2_length : pool_pointer;  {lengths of |str_pool| strings}
@!sp_brace_level : integer;             {for scanning |str_pool| strings}
@!ex_buf_xptr,@!ex_buf_yptr : buf_pointer;      {extra |ex_buf| locations}
@!control_seq_loc : hash_loc;   {hash-table loc of a control sequence}
@!preceding_white : boolean;    {used in scanning strings}
@!and_found : boolean;          {to stop the loop that looks for an ``and''}
@!num_names : integer;          {for counting names}
@!name_bf_ptr : buf_pointer;    {general |name_buf| location}
@!name_bf_xptr,@!name_bf_yptr : buf_pointer;    {and two more}
@!nm_brace_level : integer;     {for scanning |name_buf| strings}
@!name_tok : packed array[buf_pointer] of buf_pointer; {name-token ptr list}
@!name_sep_char : packed array[buf_pointer] of ASCII_code; {token-ending chars}
@!num_tokens : buf_pointer;     {this counts name tokens}
@!token_starting : boolean;     {used in scanning name tokens}
@!alpha_found : boolean;        {used in scanning the format string}
@!double_letter,@!end_of_group,@!to_be_written : boolean;       {the same}
@!first_start : buf_pointer;    {start-ptr into |name_tok| for the first name}
@!first_end : buf_pointer;      {end-ptr into |name_tok| for the first name}
@!last_end : buf_pointer;       {end-ptr into |name_tok| for the last name}
@!von_start : buf_pointer;      {start-ptr into |name_tok| for the von name}
@!von_end : buf_pointer;        {end-ptr into |name_tok| for the von name}
@!jr_end : buf_pointer;         {end-ptr into |name_tok| for the jr name}
@!cur_token,@!last_token : buf_pointer; {|name_tok| ptrs for outputting tokens}
@!use_default : boolean;        {for the inter-token intra-name part string}
@!num_commas : buf_pointer;     {used to determine the name syntax}
@!comma1,@!comma2 : buf_pointer;        {ptrs into |name_tok|}
@!num_text_chars : buf_pointer; {special characters count as one}


@
The |built_in| function {\.{=}} pops the top two (integer or string)
literals, compares them, and pushes the integer 1 if they're equal, 0
otherwise.  If they're not either both string or both integer, it
complains and pushes the integer 0.

@<|execute_fn|({\.{=}})@>=
procedure x_equals;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> pop_typ2) then
    begin
    if ((pop_typ1 <> stk_empty) and (pop_typ2 <> stk_empty)) then
        begin
        print_stk_lit (pop_lit1,pop_typ1);
        print (', ');
        print_stk_lit (pop_lit2,pop_typ2);
        print_newline;
        bst_ex_warn ('---they aren''t the same literal types');
        end;
    push_lit_stk (0, stk_int);
    end
else if ((pop_typ1 <> stk_int) and (pop_typ1 <> stk_str)) then
    begin
    if (pop_typ1 <> stk_empty) then
        begin
        print_stk_lit (pop_lit1,pop_typ1);
        bst_ex_warn (', not an integer or a string,');
        end;
    push_lit_stk (0, stk_int);
    end
else if (pop_typ1 = stk_int) then
    if (pop_lit2 = pop_lit1) then
        push_lit_stk (1, stk_int)
      else
        push_lit_stk (0, stk_int)
else
    if (str_eq_str (pop_lit2,pop_lit1)) then
        push_lit_stk (1, stk_int)
      else
        push_lit_stk (0, stk_int);
end;


@
The |built_in| function {\.{>}} pops the top two (integer) literals,
compares them, and pushes the integer 1 if the second is greater than
the first, 0 otherwise.  If either isn't an integer literal, it
complains and pushes the integer 0.

@<|execute_fn|({\.{>}})@>=
procedure x_greater_than;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (0, stk_int);
    end
else if (pop_typ2 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int);
    push_lit_stk (0, stk_int);
    end
else
    if (pop_lit2 > pop_lit1) then
        push_lit_stk (1, stk_int)
      else
        push_lit_stk (0, stk_int);
end;


@
The |built_in| function {\.{<}} pops the top two (integer) literals,
compares them, and pushes the integer 1 if the second is less than the
first, 0 otherwise.  If either isn't an integer literal, it complains
and pushes the integer 0.

@<|execute_fn|({\.{<}})@>=
procedure x_less_than;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (0, stk_int);
    end
else if (pop_typ2 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int);
    push_lit_stk (0, stk_int);
    end
else
    if (pop_lit2 < pop_lit1) then
        push_lit_stk (1, stk_int)
      else
        push_lit_stk (0, stk_int);
end;


@
The |built_in| function {\.{+}} pops the top two (integer) literals
and pushes their sum.  If either isn't an integer literal, it
complains and pushes the integer 0.

@<|execute_fn|({\.{+}})@>=
procedure x_plus;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (0, stk_int);
    end
else if (pop_typ2 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int);
    push_lit_stk (0, stk_int);
    end
else
    push_lit_stk (pop_lit2+pop_lit1, stk_int);
end;


@
The |built_in| function {\.{-}} pops the top two (integer) literals
and pushes their difference (the first subtracted from the second).
If either isn't an integer literal, it complains and pushes the
integer 0.

@<|execute_fn|({\.{-}})@>=
procedure x_minus;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (0, stk_int);
    end
else if (pop_typ2 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int);
    push_lit_stk (0, stk_int);
    end
else
    push_lit_stk (pop_lit2-pop_lit1, stk_int);
end;


@
The |built_in| function {\.{*}} pops the top two (string) literals,
concatenates them (in reverse order, that is, the order in which
pushed), and pushes the resulting string back onto the stack.  If
either isn't a string literal, it complains and pushes the null
string.

@<|execute_fn|({\.{*}})@>=
procedure x_concatenate;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ2 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else
    @<Concatenate the two strings and push@>;
end;


@
@^push the literal stack@>
Often both strings will be at the top of the string pool, in which
case we just move some pointers.  Furthermore, it's worth doing some
special stuff in case either string is null, since empirically this
seems to happen about $20\%$ of the time.  In any case, we don't need
the execution buffer---we simple move the strings around in the string
pool when necessary.

@<Concatenate the two strings and push@>=
begin
if (pop_lit2 >= cmd_str_ptr) then
    if (pop_lit1 >= cmd_str_ptr) then
        begin
        str_start[pop_lit1] := str_start[pop_lit1+1];
        unflush_string;
        incr(lit_stk_ptr);
        end
    else if (length(pop_lit2) = 0) then
        push_lit_stk (pop_lit1, stk_str)
    else        {|pop_lit2| is nonnull, only |pop_lit1| is below |cmd_str_ptr|}
        begin
        pool_ptr := str_start[pop_lit2+1];
        str_room (length(pop_lit1));
        sp_ptr := str_start[pop_lit1];
        sp_end := str_start[pop_lit1+1];
        while (sp_ptr < sp_end) do
            begin
            append_char (str_pool[sp_ptr]);
            incr(sp_ptr);
            end;
        push_lit_stk (make_string, stk_str);    {and push it onto the stack}
        end
else
    @<Concatenate them and push when |pop_lit2 < cmd_str_ptr|@>;
end


@
@^push the literal stack@>
We simply continue the previous module.

@<Concatenate them and push when |pop_lit2 < cmd_str_ptr|@>=
begin
if (pop_lit1 >= cmd_str_ptr) then
    if (length(pop_lit2) = 0) then
        begin
        unflush_string;
        lit_stack[lit_stk_ptr] := pop_lit1;
        incr(lit_stk_ptr);
        end
    else if (length(pop_lit1) = 0) then
        incr(lit_stk_ptr)
    else        {both strings nonnull, only |pop_lit2| is below |cmd_str_ptr|}
        begin
        sp_length := length(pop_lit1);
        sp2_length := length(pop_lit2);
        str_room (sp_length + sp2_length);
        sp_ptr := str_start[pop_lit1+1];
        sp_end := str_start[pop_lit1];
        sp_xptr1 := sp_ptr + sp2_length;
        while (sp_ptr > sp_end) do              {slide up |pop_lit1|}
            begin
            decr(sp_ptr);
            decr(sp_xptr1);
            str_pool[sp_xptr1] := str_pool[sp_ptr];
            end;
        sp_ptr := str_start[pop_lit2];
        sp_end := str_start[pop_lit2+1];
        while (sp_ptr < sp_end) do              {slide up |pop_lit2|}
            begin
            append_char (str_pool[sp_ptr]);
            incr(sp_ptr);
            end;
        pool_ptr := pool_ptr + sp_length;
        push_lit_stk (make_string, stk_str);    {and push it onto the stack}
        end
else
    @<Concatenate them and push when |pop_lit1,pop_lit2 < cmd_str_ptr|@>;
end


@
@^push the literal stack@>
Again, we simply continue the previous module.

@<Concatenate them and push when |pop_lit1,pop_lit2 < cmd_str_ptr|@>=
begin
if (length(pop_lit1) = 0) then
    incr(lit_stk_ptr)
else if (length(pop_lit2) = 0) then
    push_lit_stk (pop_lit1, stk_str)
else            {both strings are nonnull, and both are below |cmd_str_ptr|}
    begin
    str_room (length(pop_lit1) + length(pop_lit2));
    sp_ptr := str_start[pop_lit2];
    sp_end := str_start[pop_lit2+1];
    while (sp_ptr < sp_end) do                  {slide up |pop_lit2|}
        begin
        append_char (str_pool[sp_ptr]);
        incr(sp_ptr);
        end;
    sp_ptr := str_start[pop_lit1];
    sp_end := str_start[pop_lit1+1];
    while (sp_ptr < sp_end) do                  {slide up |pop_lit1|}
        begin
        append_char (str_pool[sp_ptr]);
        incr(sp_ptr);
        end;
    push_lit_stk (make_string, stk_str);        {and push it onto the stack}
    end;
end


@
The |built_in| function {\.{:=}} pops the top two literals and assigns
to the first (which must be an |int_entry_var|, a |str_entry_var|, an
|int_global_var|, or a |str_global_var|) the value of the second;
it complains if the value isn't of the appropriate type.

@<|execute_fn|({\.{:=}})@>=
procedure x_gets;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_fn) then
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_fn)
else if ((not mess_with_entries) and
        ((fn_type[pop_lit1] = str_entry_var) or
         (fn_type[pop_lit1] = int_entry_var))) then
    bst_cant_mess_with_entries_print
else
    case (fn_type[pop_lit1]) of
        int_entry_var : @<Assign to an |int_entry_var|@>;
        str_entry_var : @<Assign to a |str_entry_var|@>;
        int_global_var : @<Assign to an |int_global_var|@>;
        str_global_var : @<Assign to a |str_global_var|@>;
        othercases begin
                   print ('You can''t assign to type ');
                   print_fn_class (pop_lit1);
                   bst_ex_warn (', a nonvariable function class');
                   end
    endcases;
end;


@
This module checks that what we're about to assign is really an
integer, and then assigns.

@<Assign to an |int_entry_var|@>=
if (pop_typ2 <> stk_int) then
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int)
  else
    entry_ints[cite_ptr*num_ent_ints+fn_info[pop_lit1]] := pop_lit2


@
@.String size exceeded@>
It's time for a complaint if either of the two (entry or global)
string lengths is exceeded.

@d bst_string_size_exceeded(#) == begin
                                  bst_1print_string_size_exceeded;
                                  print (#);
                                  bst_2print_string_size_exceeded;
                                  end

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure bst_1print_string_size_exceeded;
begin
print ('Warning--you''ve exceeded ');
end;
@#
procedure bst_2print_string_size_exceeded;
begin
print ('-string-size,');
bst_mild_ex_warn_print;
print_ln ('*Please notify the bibstyle designer*');
end;


@
@.entry string size exceeded@>
@:String size exceeded}{\quad entry string size@>
This module checks that what we're about to assign is really a
string, and then assigns.

@<Assign to a |str_entry_var|@>=
begin
if (pop_typ2 <> stk_str) then
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_str)
  else
    begin
    str_ent_ptr := cite_ptr*num_ent_strs + fn_info[pop_lit1];
    ent_chr_ptr := 0;
    sp_ptr := str_start[pop_lit2];
    sp_xptr1 := str_start[pop_lit2+1];
    if (sp_xptr1-sp_ptr > ent_str_size) then
        begin
        bst_string_size_exceeded (ent_str_size:0,', the entry');
        sp_xptr1 := sp_ptr + ent_str_size;
        end;
    while (sp_ptr < sp_xptr1) do
        begin                   {copy characters into |entry_strs|}
        entry_strs[str_ent_ptr][ent_chr_ptr] := str_pool[sp_ptr];
        incr(ent_chr_ptr);
        incr(sp_ptr);
        end;
    entry_strs[str_ent_ptr][ent_chr_ptr] := end_of_string;
    end
end


@
This module checks that what we're about to assign is really an
integer, and then assigns.

@<Assign to an |int_global_var|@>=
if (pop_typ2 <> stk_int) then
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int)
  else
    fn_info[pop_lit1] := pop_lit2


@
@.global string size exceeded@>
@:String size exceeded}{\quad global string size@>
This module checks that what we're about to assign is really a
string, and then assigns.

@<Assign to a |str_global_var|@>=
begin
if (pop_typ2 <> stk_str) then
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_str)
  else
    begin
    str_glb_ptr := fn_info[pop_lit1];
    if (pop_lit2 < cmd_str_ptr) then
        glb_str_ptr[str_glb_ptr] := pop_lit2
      else
        begin
        glb_str_ptr[str_glb_ptr] := 0;
        glob_chr_ptr := 0;
        sp_ptr := str_start[pop_lit2];
        sp_end := str_start[pop_lit2+1];
        if (sp_end - sp_ptr > glob_str_size) then
            begin
            bst_string_size_exceeded (glob_str_size:0,', the global');
            sp_end := sp_ptr + glob_str_size;
            end;
        while (sp_ptr < sp_end) do
            begin                       {copy characters into |global_strs|}
            global_strs[str_glb_ptr][glob_chr_ptr] := str_pool[sp_ptr];
            incr(glob_chr_ptr);
            incr(sp_ptr);
            end;
        glb_str_end[str_glb_ptr] := glob_chr_ptr;
        end;
    end
end


@
The |built_in| function {\.{add.period\$}} pops the top (string)
literal, adds a |period| to a nonnull string if its last
non|right_brace| character isn't a |period|, |question_mark|, or
|exclamation_mark|, and pushes this resulting string back onto the
stack.  If the literal isn't a string, it complains and pushes the
null string.

@<|execute_fn|({\.{add.period\$}})@>=
procedure x_add_period;
label loop_exit;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else if (length(pop_lit1) = 0) then     {don't add |period| to the null string}
    push_lit_stk (s_null, stk_str)
else
    @<Add the |period|, if necessary, and push@>;
end;


@
@^push the literal stack@>
Here we scan backwards from the end of the string, skipping
non|right_brace| characters, to see if we have to add the |period|.

@<Add the |period|, if necessary, and push@>=
begin
sp_ptr := str_start[pop_lit1+1];
sp_end := str_start[pop_lit1];
while (sp_ptr > sp_end) do                      {find a non|right_brace|}
    begin
    decr(sp_ptr);
    if (str_pool[sp_ptr] <> right_brace) then
        goto loop_exit;
    end;
loop_exit:
case (str_pool[sp_ptr]) of
    period,
    question_mark,
    exclamation_mark :
        repush_string;
    othercases
        @<Add the |period| (it's necessary) and push@>
endcases;
end


@
Ok guys, we really have to do it.

@<Add the |period| (it's necessary) and push@>=
begin
if (pop_lit1 < cmd_str_ptr) then
    begin
    str_room (length(pop_lit1)+1);
    sp_ptr := str_start[pop_lit1];
    sp_end := str_start[pop_lit1+1];
    while (sp_ptr < sp_end) do          {slide |pop_lit1| atop the string pool}
        begin
        append_char (str_pool[sp_ptr]);
        incr(sp_ptr);
        end;
    end
else                                    {the string is already there}
    begin
    pool_ptr := str_start[pop_lit1+1];
    str_room (1);
    end;
append_char (period);
push_lit_stk (make_string, stk_str);
end


@
The |built_in| function {\.{call.type\$}} executes the function
specified in |type_list| for this entry unless it's |undefined|, in
which case it executes the default function \.{default.type} defined
in the \.{.bst} file, or unless it's |empty|, in which case it does
nothing.

@<|execute_fn|({\.{call.type\$}})@>=
begin
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
  else
    if (type_list[cite_ptr] = undefined) then
        execute_fn (b_default)
    else if (type_list[cite_ptr] = empty) then
        do_nothing
    else
        execute_fn (type_list[cite_ptr]);
end


@
The |built_in| function {\.{change.case\$}} pops the top two (string)
literals; it changes the case of the second according to the
specifications of the first, as follows.  (Note: The word `letters' in
the next sentence refers only to those at brace-level~0, the top-most
brace level; no other characters are changed, except perhaps for
special characters, described shortly.)  If the first literal is the
string~\.{t}, it converts to lower case all letters except the very
first character in the string, which it leaves alone, and except the
first character following any |colon| and then nonnull |white_space|,
which it also leaves alone; if it's the string~\.{l}, it converts all
letters to lower case; if it's the string~\.{u}, it converts all
letters to upper case; and if it's anything else, it complains and
does no conversion.  It then pushes this resulting string.  If either
type is incorrect, it complains and pushes the null string; however,
if both types are correct but the specification string (i.e., the
first string) isn't one of the legal ones, it merely pushes the second
back onto the stack, after complaining.  (Another note: It ignores
case differences in the specification string; for example, the strings
\.{t} and \.{T} are equivalent for the purposes of this |built_in|
function.)

@d ok_pascal_i_give_up = 21

@<|execute_fn|({\.{change.case\$}})@>=
procedure x_change_case;
label ok_pascal_i_give_up;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ2 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    @<Determine the case-conversion type@>;
    ex_buf_length := 0;
    add_buf_pool (pop_lit2);
    @<Perform the case conversion@>;
    add_pool_buf_and_push;              {push this string onto the stack}
    end;
end;


@
First we define a few variables for case conversion.  The constant
definitions, to be used in |case| statements, are in order of probable
frequency.

@d title_lowers = 0     {representing the string \.{t}}
@d all_lowers = 1       {representing the string \.{l}}
@d all_uppers = 2       {representing the string \.{u}}
@d bad_conversion = 3   {representing any illegal case-conversion string}

@<Globals in the outer block@>=
@!conversion_type : 0..bad_conversion;  {the possible cases}
@!prev_colon : boolean;                 {|true| if just past a |colon|}


@
Now we determine which of the three case-conversion types we're
dealing with: \.{t},~\.{l}, or~\.{u}.

@<Determine the case-conversion type@>=
begin
case (str_pool[str_start[pop_lit1]]) of
    "t","T" : conversion_type := title_lowers;
    "l","L" : conversion_type := all_lowers;
    "u","U" : conversion_type := all_uppers;
    othercases conversion_type := bad_conversion
endcases;
if ((length(pop_lit1) <> 1) or (conversion_type = bad_conversion)) then
    begin
    conversion_type := bad_conversion;
    print_pool_str (pop_lit1);
    bst_ex_warn (' is an illegal case-conversion string');
    end;
end


@
This procedure complains if the just-encountered |right_brace| would
make |brace_level| negative.

@<Procedures and functions for name-string processing@>=
procedure decr_brace_level (@!pop_lit_var : str_number);
begin
if (brace_level = 0) then
    braces_unbalanced_complaint (pop_lit_var)
  else
    decr(brace_level);
end;


@
This complaint often arises because the style designer has to type
lots of braces.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure braces_unbalanced_complaint (@!pop_lit_var : str_number);
begin
print ('Warning--"');
print_pool_str (pop_lit_var);
bst_mild_ex_warn ('" isn''t a brace-balanced string');
end;


@
This one makes sure that |brace_level=0| (it's called at a point in a
string where braces must be balanced).

@<Procedures and functions for name-string processing@>=
procedure check_brace_level (@!pop_lit_var : str_number);
begin
if (brace_level > 0) then
    braces_unbalanced_complaint (pop_lit_var);
end;


@
Here's where we actually go through the string and do the case
conversion.

@<Perform the case conversion@>=
begin
brace_level := 0;       {this is the top level}
ex_buf_ptr := 0;        {we start with the string's first character}
while (ex_buf_ptr < ex_buf_length) do
    begin
    if (ex_buf[ex_buf_ptr] = left_brace) then
        begin
        incr(brace_level);
        if (brace_level <> 1) then
            goto ok_pascal_i_give_up;
        if (ex_buf_ptr + 4 > ex_buf_length) then
            goto ok_pascal_i_give_up
          else if (ex_buf[ex_buf_ptr+1] <> backslash) then
            goto ok_pascal_i_give_up;
        if (conversion_type = title_lowers) then
          if (ex_buf_ptr = 0) then
            goto ok_pascal_i_give_up
          else if ((prev_colon) and
                        (lex_class[ex_buf[ex_buf_ptr-1]] = white_space)) then
            goto ok_pascal_i_give_up;
        @<Convert a special character@>;
ok_pascal_i_give_up:
        prev_colon := false;
        end
    else if (ex_buf[ex_buf_ptr] = right_brace) then
        begin
        decr_brace_level (pop_lit2);
        prev_colon := false;
        end
    else
        if (brace_level = 0) then
            @<Convert a |brace_level = 0| character@>;
    incr(ex_buf_ptr);
    end;
check_brace_level (pop_lit2);
end


@
@^special character@>
We're dealing with a special character (usually either an undotted
`\i' or `\j', or an accent like one in Table~3.1 of the \LaTeX\
manual, or a foreign character like one in Table~3.2) if the first
character after the |left_brace| is a |backslash|; the special
character ends with the matching |right_brace|.  How we handle what's
in between depends on the special character.  In general, this code
will do reasonably well if there is other stuff, too, between braces,
but it doesn't try to do anything special with |colon|s.

@<Convert a special character@>=
begin
incr(ex_buf_ptr);                       {skip over the |left_brace|}
while ((ex_buf_ptr < ex_buf_length) and (brace_level > 0)) do
    begin
    incr(ex_buf_ptr);                   {skip over the |backslash|}
    ex_buf_xptr := ex_buf_ptr;
    while ((ex_buf_ptr < ex_buf_length) and
                (lex_class[ex_buf[ex_buf_ptr]] = alpha)) do
        incr(ex_buf_ptr);               {this scans the control sequence}
    control_seq_loc := str_lookup(ex_buf,ex_buf_xptr,ex_buf_ptr-ex_buf_xptr,
                                                control_seq_ilk,dont_insert);
    if (hash_found) then
        @<Convert the accented or foreign character, if necessary@>;
    ex_buf_xptr := ex_buf_ptr;
    while ((ex_buf_ptr < ex_buf_length) and (brace_level > 0) and
                                        (ex_buf[ex_buf_ptr] <> backslash)) do
        begin                   {this scans to the next control sequence}
        if (ex_buf[ex_buf_ptr] = right_brace) then
            decr(brace_level)
        else if (ex_buf[ex_buf_ptr] = left_brace) then
            incr(brace_level);
        incr(ex_buf_ptr);
        end;
    @<Convert a noncontrol sequence@>;
    end;
decr(ex_buf_ptr);               {unskip the |right_brace|}
end


@
@^control sequence@>
@:this can't happen}{\quad Unknown type of case conversion@>
A control sequence, for the purposes of this program, consists just of
the consecutive alphabetic characters following the |backslash|; it
might be empty (although ones in this section aren't).

@<Convert the accented or foreign character, if necessary@>=
begin
case (conversion_type) of
    title_lowers,
    all_lowers :
        case (ilk_info[control_seq_loc]) of
            n_l_upper,
            n_o_upper,
            n_oe_upper,
            n_ae_upper,
            n_aa_upper :
                lower_case (ex_buf, ex_buf_xptr, ex_buf_ptr-ex_buf_xptr);
            othercases
                do_nothing
        endcases;
    all_uppers :
        case (ilk_info[control_seq_loc]) of
            n_l,
            n_o,
            n_oe,
            n_ae,
            n_aa :
                upper_case (ex_buf, ex_buf_xptr, ex_buf_ptr-ex_buf_xptr);
            n_i,
            n_j,
            n_ss :
                @<Convert, then remove the control sequence@>;
            othercases
                do_nothing
        endcases;
    bad_conversion :
        do_nothing;
    othercases
        case_conversion_confusion
endcases;
end


@
@:this can't happen}{\quad Unknown type of case conversion@>
Another bug complaint.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure case_conversion_confusion;
begin
confusion ('Unknown type of case conversion');
end;


@
After converting the control sequence, we need to remove the preceding
|backslash| and any following |white_space|.

@<Convert, then remove the control sequence@>=
begin
upper_case (ex_buf, ex_buf_xptr, ex_buf_ptr-ex_buf_xptr);
while (ex_buf_xptr < ex_buf_ptr) do
    begin                       {remove preceding |backslash| and shift down}
    ex_buf[ex_buf_xptr-1] := ex_buf[ex_buf_xptr];
    incr(ex_buf_xptr);
    end;
decr(ex_buf_xptr);
while ((ex_buf_ptr < ex_buf_length) and
                (lex_class[ex_buf[ex_buf_ptr]] = white_space)) do
    incr(ex_buf_ptr);           {remove |white_space| trailing the control seq}
tmp_ptr := ex_buf_ptr;
while (tmp_ptr < ex_buf_length) do
    begin                       {more shifting down}
    ex_buf[tmp_ptr-(ex_buf_ptr-ex_buf_xptr)] := ex_buf[tmp_ptr];
    incr(tmp_ptr)
    end;
ex_buf_length := tmp_ptr - (ex_buf_ptr - ex_buf_xptr);
ex_buf_ptr := ex_buf_xptr;
end


@
@:this can't happen}{\quad Unknown type of case conversion@>
There are no control sequences in what we're about to convert,
so a straight conversion suffices.

@<Convert a noncontrol sequence@>=
begin
case (conversion_type) of
    title_lowers,
    all_lowers :
        lower_case (ex_buf, ex_buf_xptr, ex_buf_ptr-ex_buf_xptr);
    all_uppers :
        upper_case (ex_buf, ex_buf_xptr, ex_buf_ptr-ex_buf_xptr);
    bad_conversion :
        do_nothing;
    othercases
        case_conversion_confusion
endcases;
end


@
@:this can't happen}{\quad Unknown type of case conversion@>
This code does any needed conversion for an ordinary character; it
won't touch nonletters.

@<Convert a |brace_level = 0| character@>=
begin
case (conversion_type) of
    title_lowers :
        begin
        if (ex_buf_ptr = 0) then
            do_nothing
        else if ((prev_colon) and
                        (lex_class[ex_buf[ex_buf_ptr-1]] = white_space)) then
            do_nothing
        else
            lower_case (ex_buf, ex_buf_ptr, 1);
        if (ex_buf[ex_buf_ptr] = colon) then
            prev_colon := true
        else if (lex_class[ex_buf[ex_buf_ptr]] <> white_space) then
            prev_colon := false;
        end;
    all_lowers :
        lower_case (ex_buf, ex_buf_ptr, 1);
    all_uppers :
        upper_case (ex_buf, ex_buf_ptr, 1);
    bad_conversion :
        do_nothing;
    othercases
        case_conversion_confusion
endcases;
end


@
The |built_in| function {\.{chr.to.int\$}} pops the top (string)
literal, makes sure it's a single character, converts it to the
corresponding |ASCII_code| integer, and pushes this integer.  If the
literal isn't an appropriate string, it complains and pushes the
integer~0.

@<|execute_fn|({\.{chr.to.int\$}})@>=
procedure x_chr_to_int;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (0, stk_int);
    end
else if (length(pop_lit1) <> 1) then
    begin
    print ('"');
    print_pool_str (pop_lit1);
    bst_ex_warn ('" isn''t a single character');
    push_lit_stk (0, stk_int);
    end
else
    push_lit_stk (str_pool[str_start[pop_lit1]], stk_int);
                                        {push the (|ASCII_code|) integer}
end;


@
The |built_in| function {\.{cite\$}} pushes the appropriate string
from |cite_list| onto the stack.

@<|execute_fn|({\.{cite\$}})@>=
procedure x_cite;
begin
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
  else
    push_lit_stk (cur_cite_str, stk_str);
end;


@
@^push the literal stack@>
The |built_in| function {\.{duplicate\$}} pops the top literal from
the stack and pushes two copies of it.

@<|execute_fn|({\.{duplicate\$}})@>=
procedure x_duplicate;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    push_lit_stk (pop_lit1, pop_typ1);
    push_lit_stk (pop_lit1, pop_typ1);
    end
  else
    begin
    repush_string;
    if (pop_lit1 < cmd_str_ptr) then
        push_lit_stk (pop_lit1, pop_typ1)
      else
        begin
        str_room (length(pop_lit1));
        sp_ptr := str_start[pop_lit1];
        sp_end := str_start[pop_lit1+1];
        while (sp_ptr < sp_end) do
            begin
            append_char (str_pool[sp_ptr]);
            incr(sp_ptr);
            end;
        push_lit_stk (make_string, stk_str);    {and push it onto the stack}
        end;
    end;
end;


@
The |built_in| function {\.{empty\$}} pops the top literal and pushes
the integer 1 if it's a missing field or a string having no
non|white_space| characters, 0 otherwise.  If the literal isn't a
missing field or a string, it complains and pushes 0.

@<|execute_fn|({\.{empty\$}})@>=
procedure x_empty;
label exit;
begin
pop_lit_stk (pop_lit1,pop_typ1);
case (pop_typ1) of
    stk_str : @<Push 0 if the string has a non|white_space| char, else 1@>;
    stk_field_missing : push_lit_stk (1, stk_int);
    stk_empty : push_lit_stk (0, stk_int);
    othercases
        begin
        print_stk_lit (pop_lit1,pop_typ1);
        bst_ex_warn (', not a string or missing field,');
        push_lit_stk (0, stk_int);
        end
endcases;
exit:
end;


@
When we arrive here we're dealing with a legitimate string.  If it has
no characters, or has nothing but |white_space| characters, we push~1,
otherwise we push~0.

@<Push 0 if the string has a non|white_space| char, else 1@>=
begin
sp_ptr := str_start[pop_lit1];
sp_end := str_start[pop_lit1+1];
while (sp_ptr < sp_end) do
    begin
    if (lex_class[str_pool[sp_ptr]] <> white_space) then
        begin
        push_lit_stk (0, stk_int);
        return;
        end;
    incr(sp_ptr);
    end;
push_lit_stk (1, stk_int);
end


@
The |built_in| function {\.{format.name\$}} pops the top three
literals (they are a string, an integer, and a string literal, in that
order).  The last string literal represents a name list (each name
corresponding to a person), the integer literal specifies which name
to pick from this list, and the first string literal specifies how to
format this name, as described in the \BibTeX\ documentation.
Finally, this function pushes the formatted name.  If any of the types
is incorrect, it complains and pushes the null string.

@d von_found = 52               {for when a von token is found}

@<|execute_fn|({\.{format.name\$}})@>=
procedure x_format_name;
label loop1_exit,@!loop2_exit,@!von_found;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
pop_lit_stk (pop_lit3,pop_typ3);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ2 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ3 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit3,pop_typ3,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    ex_buf_length := 0;
    add_buf_pool (pop_lit3);
    @<Isolate the desired name@>;
    @<Copy name and count |comma|s to determine syntax@>;
    @<Find the parts of the name@>;
    ex_buf_length := 0;
    add_buf_pool (pop_lit1);
    figure_out_the_formatted_name;@/
    add_pool_buf_and_push;      {push the formatted string onto the stack}
    end;
end;


@
This module skips over undesired names in |pop_lit3| and it throws
away the ``and'' from the end of the name if it exists.  When it's
done, |ex_buf_xptr| points to its first character and |ex_buf_ptr|
points just past its last.

@<Isolate the desired name@>=
begin
ex_buf_ptr := 0;
num_names := 0;
while ((num_names < pop_lit2) and (ex_buf_ptr < ex_buf_length)) do
    begin
    incr(num_names);
    ex_buf_xptr := ex_buf_ptr;
    name_scan_for_and (pop_lit3);
    end;
if (ex_buf_ptr < ex_buf_length) then            {remove the ``and''}
    ex_buf_ptr := ex_buf_ptr - 4;
if (num_names < pop_lit2) then
    begin
    if (pop_lit2 = 1) then
        print ('There is no name in "')
      else
        print ('There aren''t ',pop_lit2:0,' names in "');
    print_pool_str (pop_lit3);
    bst_ex_warn ('"');
    end
end


@
This module, starting at |ex_buf_ptr|, looks in |ex_buf| for an
``and'' surrounded by nonnull |white_space|.  It stops either at
|ex_buf_length| or just past the ``and'', whichever comes first,
setting |ex_buf_ptr| accordingly.  Its parameter |pop_lit_var| is
either |pop_lit3| or |pop_lit1|, depending on whether
{\.{format.name\$}} or {\.{num.names\$}} calls it.

@<Procedures and functions for name-string processing@>=
procedure name_scan_for_and (@!pop_lit_var : str_number);
begin
brace_level := 0;
preceding_white := false;
and_found := false;
while ((not and_found) and (ex_buf_ptr < ex_buf_length)) do
  case (ex_buf[ex_buf_ptr]) of
    "a", "A" :
        begin
        incr(ex_buf_ptr);
        if (preceding_white) then
            @<See if we have an ``and''@>;      {if so, |and_found := true|}
        preceding_white := false;
        end;
    left_brace :
        begin
        incr(brace_level);
        incr(ex_buf_ptr);
        @<Skip over |ex_buf| stuff at |brace_level > 0|@>;
        preceding_white := false;
        end;
    right_brace :
        begin
        decr_brace_level (pop_lit_var);         {this checks for an error}
        incr(ex_buf_ptr);
        preceding_white := false;
        end;
    othercases
        if (lex_class[ex_buf[ex_buf_ptr]] = white_space) then
            begin
            incr(ex_buf_ptr);
            preceding_white := true;
            end
        else
            begin
            incr(ex_buf_ptr);
            preceding_white := false;
            end
  endcases;
check_brace_level (pop_lit_var);
end;


@
When we come here |ex_buf_ptr| is just past the |left_brace|, and when
we leave it's either at |ex_buf_length| or just past the matching
|right_brace|.

@<Skip over |ex_buf| stuff at |brace_level > 0|@>=
while ((brace_level > 0) and (ex_buf_ptr < ex_buf_length)) do
    begin
    if (ex_buf[ex_buf_ptr] = right_brace) then
        decr(brace_level)
    else if (ex_buf[ex_buf_ptr] = left_brace) then
        incr(brace_level);
    incr(ex_buf_ptr);
    end


@
When we come here |ex_buf_ptr| is just past the ``a'' or ``A'', and when
we leave it's either at the same place or, if we found an ``and'', at
the following |white_space| character.

@<See if we have an ``and''@>=
begin
if (ex_buf_ptr <= (ex_buf_length - 3)) then     {enough characters are left}
    if ((ex_buf[ex_buf_ptr] = "n") or (ex_buf[ex_buf_ptr] = "N")) then
        if ((ex_buf[ex_buf_ptr+1] = "d") or (ex_buf[ex_buf_ptr+1] = "D")) then
            if (lex_class[ex_buf[ex_buf_ptr+2]] = white_space) then
                begin
                ex_buf_ptr := ex_buf_ptr + 2;
                and_found := true;
                end;
end


@
When we arrive here, the desired name is in |ex_buf[ex_buf_xptr]|
through |ex_buf[ex_buf_ptr-1]|.  This module does its thing for
characters only at |brace_level = 0|; the rest get processed verbatim.
It removes leading |white_space| (and |sep_char|s), and trailing
|white_space| (and |sep_char|s) and |comma|s, complaining for each
trailing |comma|.  It then copies the name into |name_buf|, removing
all |white_space|, |sep_char|s and |comma|s, counting |comma|s, and
constructing a list of name tokens, which are sequences of characters
separated (at |brace_level=0|) by |white_space|, |sep_char|s or
|comma|s.  Each name token but the first has an associated
|name_sep_char|, the character that separates it from the preceding
token.  If there are too many (more than two) |comma|s, a complaint is
in order.

@<Copy name and count |comma|s to determine syntax@>=
begin
@<Remove leading and trailing junk, complaining if necessary@>;
name_bf_ptr := 0;
num_commas := 0;
num_tokens := 0;@/
token_starting := true;         {to indicate that a name token is starting}
while (ex_buf_xptr < ex_buf_ptr) do
    case (ex_buf[ex_buf_xptr]) of
        comma : @<Name-process a |comma|@>;
        left_brace : @<Name-process a |left_brace|@>;
        right_brace : @<Name-process a |right_brace|@>;
        othercases
            case (lex_class[ex_buf[ex_buf_xptr]]) of
                white_space : @<Name-process a |white_space|@>;
                sep_char : @<Name-process a |sep_char|@>;
                othercases @<Name-process some other character@>
            endcases
    endcases;
name_tok[num_tokens] := name_bf_ptr;    {this is an end-marker}
end


@
This module removes all leading |white_space| (and |sep_char|s), and
trailing |white_space| (and |sep_char|s) and |comma|s.  It complains
for each trailing |comma|.

@<Remove leading and trailing junk, complaining if necessary@>=
begin
while ((ex_buf_xptr < ex_buf_ptr) and
                        (lex_class[ex_buf[ex_buf_ptr]] = white_space) and
                        (lex_class[ex_buf[ex_buf_ptr]] = sep_char)) do
        incr(ex_buf_xptr);                      {this removes leading stuff}
while (ex_buf_ptr > ex_buf_xptr) do             {now remove trailing stuff}
    case (lex_class[ex_buf[ex_buf_ptr-1]]) of
        white_space,
        sep_char :
            decr(ex_buf_ptr);
        othercases
            if (ex_buf[ex_buf_ptr-1] = comma) then
                begin
                print ('Name ',pop_lit2:0,' in "');
                print_pool_str (pop_lit3);
                print ('" has a comma at the end');
                bst_ex_warn_print;
                decr(ex_buf_ptr);
                end
            else
                goto loop1_exit
    endcases;
loop1_exit:

end

@
Here we mark the token number at which this comma has occurred.

@<Name-process a |comma|@>=
begin
if (num_commas = 2) then
    begin
    print ('Too many commas in name ',pop_lit2:0,' of "');
    print_pool_str (pop_lit3);
    print ('"');
    bst_ex_warn_print;
    end
  else
    begin
    incr(num_commas);
    if (num_commas = 1) then
        comma1 := num_tokens
      else
        comma2 := num_tokens;                   {|num_commas = 2|}
    name_sep_char[num_tokens] := comma;
    end;
incr(ex_buf_xptr);
token_starting := true;
end


@
We copy the stuff up through the matching |right_brace| verbatim.

@<Name-process a |left_brace|@>=
begin
incr(brace_level);
if (token_starting) then
    begin
    name_tok[num_tokens] := name_bf_ptr;
    incr(num_tokens);
    end;
name_buf[name_bf_ptr] := ex_buf[ex_buf_xptr];
incr(name_bf_ptr);
incr(ex_buf_xptr);
while ((brace_level > 0) and (ex_buf_xptr < ex_buf_ptr)) do
    begin
    if (ex_buf[ex_buf_xptr] = right_brace) then
        decr(brace_level)
    else if (ex_buf[ex_buf_xptr] = left_brace) then
        incr(brace_level);
    name_buf[name_bf_ptr] := ex_buf[ex_buf_xptr];
    incr(name_bf_ptr);
    incr(ex_buf_xptr);
    end;
token_starting := false;
end


@
We don't copy an extra |right_brace|; this code will almost never be
executed.

@<Name-process a |right_brace|@>=
begin
if (token_starting) then
    begin
    name_tok[num_tokens] := name_bf_ptr;
    incr(num_tokens);
    end;
print ('Name ',pop_lit2:0,' of "');
print_pool_str (pop_lit3);
bst_ex_warn ('" isn''t brace balanced');
incr(ex_buf_xptr);
token_starting := false;
end


@
A token will be starting soon in a buffer near you, one way$\ldots$

@<Name-process a |white_space|@>=
begin
if (not token_starting) then
    name_sep_char[num_tokens] := space;
incr(ex_buf_xptr);
token_starting := true;
end


@
@^user abuse@>
or another.  If one of the valid |sep_char|s appears between tokens,
we usually use it instead of a |space|.  If the user has been silly
enough to have multiple |sep_char|s, or to have both |white_space| and
a |sep_char|, we use the first such character.

@<Name-process a |sep_char|@>=
begin
if (not token_starting) then
    name_sep_char[num_tokens] := ex_buf[ex_buf_xptr];
incr(ex_buf_xptr);
token_starting := true;
end


@
For ordinary characters, we just copy the character.

@<Name-process some other character@>=
begin
if (token_starting) then
    begin
    name_tok[num_tokens] := name_bf_ptr;
    incr(num_tokens);
    end;
name_buf[name_bf_ptr] := ex_buf[ex_buf_xptr];
incr(name_bf_ptr);
incr(ex_buf_xptr);
token_starting := false;
end


@
@:this can't happen}{\quad Illegal number of comma,s@>
Here we set all the pointers for the various parts of the name,
depending on which of the three possible syntaxes this name uses.

@<Find the parts of the name@>=
begin
if (num_commas = 0) then
    begin
    first_start := 0;
    last_end := num_tokens;
    jr_end := last_end;
    @<Determine where the first name ends and von name starts and ends@>;
    end
else if (num_commas = 1) then
    begin
    von_start := 0;
    last_end := comma1;
    jr_end := last_end;
    first_start := jr_end;
    first_end := num_tokens;
    von_name_ends_and_last_name_starts_stuff;
    end
else if (num_commas = 2) then
    begin
    von_start := 0;
    last_end := comma1;
    jr_end := comma2;
    first_start := jr_end;
    first_end := num_tokens;
    von_name_ends_and_last_name_starts_stuff;
    end
else
    confusion ('Illegal number of comma,s');
end


@
When there are no brace-level-0 |comma|s in the name, the von name
starts with the first nonlast token whose first brace-level-0 letter
is in lower case (for the purposes of this determination, an accented
or foreign character at brace-level-1 that's in lower case will do, as
well).  A module following this one determines where the von name ends
and the last starts.

@<Determine where the first name ends and von name starts and ends@>=
begin
von_start := 0;
while (von_start < last_end-1) do
    begin
    name_bf_ptr := name_tok[von_start];
    name_bf_xptr := name_tok[von_start+1];
    if (von_token_found) then
        begin
        von_name_ends_and_last_name_starts_stuff;
        goto von_found;
        end;
    incr(von_start);
    end;                        {there's no von name, so}
while (von_start > 0) do        {backtrack if there are connected tokens}
    begin
    if ((lex_class[name_sep_char[von_start]] <> sep_char) or
                        (name_sep_char[von_start] = tie)) then
        goto loop2_exit;
    decr(von_start);
    end;
loop2_exit:
von_end := von_start;
von_found:
first_end := von_start;
end


@
@^special character@>
It's a von token if there exists a first brace-level-0 letter (or
brace-level-1 special character), and it's in lower case; in this case
we return |true|.  The token is in |name_buf|, starting at
|name_bf_ptr| and ending just before |name_bf_xptr|.

@d return_von_found ==  begin
                        von_token_found := true;
                        return;
                        end

@<Procedures and functions for name-string processing@>=
function von_token_found : boolean;
label exit;
begin
nm_brace_level := 0;
von_token_found := false;               {now it's easy to exit if necessary}
while (name_bf_ptr < name_bf_xptr) do
    if ((name_buf[name_bf_ptr] >= "A") and
                        (name_buf[name_bf_ptr] <= "Z")) then
        return
    else if ((name_buf[name_bf_ptr] >= "a") and
                        (name_buf[name_bf_ptr] <= "z")) then
        return_von_found
    else if (name_buf[name_bf_ptr] = left_brace) then
        begin
        incr(nm_brace_level);
        incr(name_bf_ptr);
        if ((name_bf_ptr + 2 < name_bf_xptr) and
                                (name_buf[name_bf_ptr] = backslash)) then
            @<Check the special character (and |return|)@>
          else
            @<Skip over |name_buf| stuff at |nm_brace_level > 0|@>;
        end
    else
        incr(name_bf_ptr);
exit:
end;


@
@^special character@>
When we come here |name_bf_ptr| is just past the |left_brace|,
but we always leave by |return|ing.

@<Check the special character (and |return|)@>=
begin
incr(name_bf_ptr);                      {skip over the |backslash|}
name_bf_yptr := name_bf_ptr;
while ((name_bf_ptr < name_bf_xptr) and
                (lex_class[name_buf[name_bf_ptr]] = alpha)) do
    incr(name_bf_ptr);                  {this scans the control sequence}
control_seq_loc := str_lookup(name_buf,name_bf_yptr,name_bf_ptr-name_bf_yptr,
                                                control_seq_ilk,dont_insert);
if (hash_found) then
    @<Handle this accented or foreign character (and |return|)@>;
while ((name_bf_ptr < name_bf_xptr) and (nm_brace_level > 0)) do
    begin
    if ((name_buf[name_bf_ptr] >= "A") and
                        (name_buf[name_bf_ptr] <= "Z")) then
        return
    else if ((name_buf[name_bf_ptr] >= "a") and
                        (name_buf[name_bf_ptr] <= "z")) then
        return_von_found
    else if (name_buf[name_bf_ptr] = right_brace) then
        decr(nm_brace_level)
    else if (name_buf[name_bf_ptr] = left_brace) then
        incr(nm_brace_level);
    incr(name_bf_ptr);
    end;
return;
end


@
@:this can't happen}{\quad Control-sequence hash error@>
The accented or foreign character is either `\.{\\i}' or `\.{\\j}' or
one of the eleven alphabetic foreign characters in Table~3.2 of the
\LaTeX\ manual.

@<Handle this accented or foreign character (and |return|)@>=
begin
case (ilk_info[control_seq_loc]) of
    n_oe_upper,
    n_ae_upper,
    n_aa_upper,
    n_o_upper,
    n_l_upper :
        return;
    n_i,
    n_j,
    n_oe,
    n_ae,
    n_aa,
    n_o,
    n_l,
    n_ss :
        return_von_found;
    othercases
        confusion ('Control-sequence hash error')
endcases;
end


@
When we come here |name_bf_ptr| is just past the |left_brace|; when we
leave it's either at |name_bf_xptr| or just past the matching
|right_brace|.

@<Skip over |name_buf| stuff at |nm_brace_level > 0|@>=
while ((nm_brace_level > 0) and (name_bf_ptr < name_bf_xptr)) do
    begin
    if (name_buf[name_bf_ptr] = right_brace) then
        decr(nm_brace_level)
    else if (name_buf[name_bf_ptr] = left_brace) then
        incr(nm_brace_level);
    incr(name_bf_ptr);
    end


@
@^Casey Stengel would be proud@>
@^special character@>
@^Tuesdays@>
The last name starts just past the last token, before the first
|comma| (if there is no |comma|, there is deemed to be one at the end
of the string), for which there exists a first brace-level-0 letter
(or brace-level-1 special character), and it's in lower case, unless
this last token is also the last token before the |comma|, in which
case the last name starts with this token (unless this last token is
connected by a |sep_char| other than a |tie| to the previous token, in
which case the last name starts with as many tokens earlier as are
connected by non|tie|s to this last one (except on Tuesdays
$\ldots\,$), although this module never sees such a case).  Note that
if there are any tokens in either the von or last names, then the last
name has at least one, even if it starts with a lower-case letter.

@<Procedures and functions for name-string processing@>=
procedure von_name_ends_and_last_name_starts_stuff;
label exit;
begin                           {there may or may not be a von name}
von_end := last_end - 1;
while (von_end > von_start) do
    begin
    name_bf_ptr := name_tok[von_end-1];
    name_bf_xptr := name_tok[von_end];
    if (von_token_found) then
        return;
    decr(von_end);
    end;
exit:
end;


@
This module uses the information in |pop_lit1| to format the name.
Everything at |sp_brace_level = 0| is copied verbatim to the formatted
string; the rest is described in the succeeding modules.

@<Figure out the formatted name@>=
begin
ex_buf_ptr := 0;
sp_brace_level := 0;
sp_ptr := str_start[pop_lit1];
sp_end := str_start[pop_lit1+1];
while (sp_ptr < sp_end) do
    if (str_pool[sp_ptr] = left_brace) then
        begin
        incr(sp_brace_level);
        incr(sp_ptr);
        @<Format this part of the name@>;
        end
    else if (str_pool[sp_ptr] = right_brace) then
        begin
        braces_unbalanced_complaint (pop_lit1);
        incr(sp_ptr);
        end
    else
        begin
        append_ex_buf_char_and_check (str_pool[sp_ptr]);
        incr(sp_ptr);
        end;
if (sp_brace_level > 0) then
    braces_unbalanced_complaint (pop_lit1);
ex_buf_length := ex_buf_ptr;
end


@
When we arrive here we're at |sp_brace_level = 1|, just past the
|left_brace|.  Letters at this |sp_brace_level| other than those
denoting the parts of the name (i.e., the first letters of `first,'
`last,' `von,' and `jr,' ignoring case) are illegal.  We do two passes
over this group; the first determines whether we're to output
anything, and, if we are, the second actually outputs it.

@<Format this part of the name@>=
begin
sp_xptr1 := sp_ptr;
alpha_found := false;
double_letter := false;
end_of_group := false;
to_be_written := true;
while ((not end_of_group) and (sp_ptr < sp_end)) do
    if (lex_class[str_pool[sp_ptr]] = alpha) then
        begin
        incr(sp_ptr);
        @<Figure out what this letter means@>;
        end
    else if (str_pool[sp_ptr] = right_brace) then
        begin
        decr(sp_brace_level);
        incr(sp_ptr);
        end_of_group := true;
        end
    else if (str_pool[sp_ptr] = left_brace) then
        begin
        incr(sp_brace_level);
        incr(sp_ptr);
        skip_stuff_at_sp_brace_level_greater_than_one;
        end
    else
        incr(sp_ptr);
if ((end_of_group) and (to_be_written)) then    {do the second pass}
    @<Finally format this part of the name@>;
end


@
When we come here |sp_ptr| is just past the |left_brace|, and when we
leave it's either at |sp_end| or just past the matching |right_brace|.

@<Procedures and functions for name-string processing@>=
procedure skip_stuff_at_sp_brace_level_greater_than_one;
begin
while ((sp_brace_level > 1) and (sp_ptr < sp_end)) do
    begin
    if (str_pool[sp_ptr] = right_brace) then
        decr(sp_brace_level)
    else if (str_pool[sp_ptr] = left_brace) then
        incr(sp_brace_level);
    incr(sp_ptr);
    end;
end;


@
We won't output anything for this part of the name if this is a second
occurrence of an |sp_brace_level = 1| letter, if it's an illegal
letter, or if there are no tokens corresponding to this part.  We also
determine if we're we to output complete tokens (indicated by a double
letter).

@<Figure out what this letter means@>=
begin
if (alpha_found) then
    begin
    brace_lvl_one_letters_complaint;
    to_be_written := false;
    end
  else
    begin
    case (str_pool[sp_ptr-1]) of
        "f","F" : @<Figure out what tokens we'll output for the `first' name@>;
        "v","V" : @<Figure out what tokens we'll output for the `von' name@>;
        "l","L" : @<Figure out what tokens we'll output for the `last' name@>;
        "j","J" : @<Figure out what tokens we'll output for the `jr' name@>;
        othercases
                begin
                brace_lvl_one_letters_complaint;
                to_be_written := false;
                end
    endcases;
    if (double_letter) then
        incr(sp_ptr);
    end;
alpha_found := true;
end


@
At most one of the important letters, perhaps doubled, may appear at
|sp_brace_level = 1|.

@<Procedures and functions for name-string processing@>=
procedure brace_lvl_one_letters_complaint;
begin
print ('The format string "');
print_pool_str (pop_lit1);
bst_ex_warn ('" has an illegal brace-level-1 letter');
end;


@
Here we set pointers into |name_tok| and note whether we'll be dealing
with a full first-name tokens (|double_letter = true|) or
abbreviations (|double_letter = false|).

@<Figure out what tokens we'll output for the `first' name@>=
begin
cur_token := first_start;
last_token := first_end;
if (cur_token = last_token) then
    to_be_written := false;
if ((str_pool[sp_ptr] = "f") or (str_pool[sp_ptr] = "F")) then
    double_letter := true;
end


@
The same as above but for von-name tokens.

@<Figure out what tokens we'll output for the `von' name@>=
begin
cur_token := von_start;
last_token := von_end;
if (cur_token = last_token) then
    to_be_written := false;
if ((str_pool[sp_ptr] = "v") or (str_pool[sp_ptr] = "V")) then
    double_letter := true;
end


@
The same as above but for last-name tokens.

@<Figure out what tokens we'll output for the `last' name@>=
begin
cur_token := von_end;
last_token := last_end;
if (cur_token = last_token) then
    to_be_written := false;
if ((str_pool[sp_ptr] = "l") or (str_pool[sp_ptr] = "L")) then
    double_letter := true;
end


@
The same as above but for jr-name tokens.

@<Figure out what tokens we'll output for the `jr' name@>=
begin
cur_token := last_end;
last_token := jr_end;
if (cur_token = last_token) then
    to_be_written := false;
if ((str_pool[sp_ptr] = "j") or (str_pool[sp_ptr] = "J")) then
    double_letter := true;
end


@
This is the second pass over this part of the name; here we actually
write stuff out to |ex_buf|.

@<Finally format this part of the name@>=
begin
ex_buf_xptr := ex_buf_ptr;
sp_ptr := sp_xptr1;
sp_brace_level := 1;
while (sp_brace_level > 0) do
    if ((lex_class[str_pool[sp_ptr]] = alpha) and (sp_brace_level = 1)) then
        begin
        incr(sp_ptr);
        @<Figure out how to output the name tokens, and do it@>;
        end
    else if (str_pool[sp_ptr] = right_brace) then
        begin
        decr(sp_brace_level);
        incr(sp_ptr);
        if (sp_brace_level > 0) then
            append_ex_buf_char_and_check (right_brace);
        end
    else if (str_pool[sp_ptr] = left_brace) then
        begin
        incr(sp_brace_level);
        incr(sp_ptr);
        append_ex_buf_char_and_check (left_brace);
        end
    else
        begin
        append_ex_buf_char_and_check (str_pool[sp_ptr]);
        incr(sp_ptr);
        end;
if (ex_buf_ptr > 0) then
  if (ex_buf[ex_buf_ptr-1] = tie) then
    @<Handle a discretionary |tie|@>;
end


@
When we come here, |sp_ptr| is just past the letter indicating the
part of the name for which we're about to output tokens.  When we
leave, it's at the first character of the rest of the group.

@<Figure out how to output the name tokens, and do it@>=
begin
if (double_letter) then
    incr(sp_ptr);
use_default := true;
sp_xptr2 := sp_ptr;
if (str_pool[sp_ptr] = left_brace) then         {find the inter-token string}
    begin
    use_default := false;
    incr(sp_brace_level);
    incr(sp_ptr);
    sp_xptr1 := sp_ptr;
    skip_stuff_at_sp_brace_level_greater_than_one;
    sp_xptr2 := sp_ptr - 1;
    end;
@<Finally output the name tokens@>;
if (not use_default) then
    sp_ptr := sp_xptr2 + 1;
end


@
Here, for each token in this part, we output either a full or an
abbreviated token and the inter-token string for all but the last
token of this part.

@<Finally output the name tokens@>=
while (cur_token < last_token) do
    begin
    if (double_letter) then
        @<Finally output a full token@>
      else
        @<Finally output an abbreviated token@>;
    incr(cur_token);
    if (cur_token < last_token) then
        @<Finally output the inter-token string@>;
    end


@
@:BibTeX capacity exceeded}{\quad buffer size@>
Here we output all the characters in the token, verbatim.

@<Finally output a full token@>=
begin
name_bf_ptr := name_tok[cur_token];
name_bf_xptr := name_tok[cur_token+1];
if (ex_buf_length+(name_bf_xptr-name_bf_ptr) > buf_size) then
    buffer_overflow;
while (name_bf_ptr < name_bf_xptr) do
    begin
    append_ex_buf_char (name_buf[name_bf_ptr]);
    incr(name_bf_ptr);
    end;
end


@
@^special character@>
Here we output the first alphabetic or special character of the token;
brace level is irrelevant for an alphabetic (but not a special)
character.

@<Finally output an abbreviated token@>=
begin
name_bf_ptr := name_tok[cur_token];
name_bf_xptr := name_tok[cur_token+1];
while (name_bf_ptr < name_bf_xptr) do
    begin
    if (lex_class[name_buf[name_bf_ptr]] = alpha) then
        begin
        append_ex_buf_char_and_check (name_buf[name_bf_ptr]);
        goto loop_exit;
        end
    else if ((name_buf[name_bf_ptr] = left_brace) and
                                (name_bf_ptr + 1 < name_bf_xptr)) then
      if (name_buf[name_bf_ptr+1] = backslash) then
        @<Finally output a special character and exit loop@>;
    incr(name_bf_ptr);
    end;
loop_exit:
end


@
@^special character@>
@^user abuse@>
@:BibTeX capacity exceeded}{\quad buffer size@>
We output a special character here even if the user has been silly
enough to make it nonalphabetic (and even if the user has been sillier
still by not having a matching |right_brace|).

@<Finally output a special character and exit loop@>=
begin
if (ex_buf_ptr + 2 > buf_size) then
    buffer_overflow;
append_ex_buf_char (left_brace);
append_ex_buf_char (backslash);
name_bf_ptr := name_bf_ptr + 2;
nm_brace_level := 1;
while ((name_bf_ptr < name_bf_xptr) and (nm_brace_level > 0)) do
    begin
    if (name_buf[name_bf_ptr] = right_brace) then
        decr(nm_brace_level)
    else if (name_buf[name_bf_ptr] = left_brace) then
        incr(nm_brace_level);
    append_ex_buf_char_and_check (name_buf[name_bf_ptr]);
    incr(name_bf_ptr);
    end;
goto loop_exit;
end


@
@:BibTeX capacity exceeded}{\quad buffer size@>
Here we output either the \.{.bst} given string if it exists, or else
the \.{.bib} |sep_char| if it exists, or else the default string.  A
|tie| is the default space character between the last two tokens of
the name part, and between the first two tokens if the first token is
short enough; otherwise, a |space| is the default.

@d long_token = 3       {a token this length or longer is ``long''}

@<Finally output the inter-token string@>=
begin
if (use_default) then
    begin
    if (not double_letter) then
        append_ex_buf_char_and_check (period);
    if (lex_class[name_sep_char[cur_token]] = sep_char) then
        append_ex_buf_char_and_check (name_sep_char[cur_token])
    else if ((cur_token = last_token-1) or
                        (not enough_text_chars (long_token))) then
        append_ex_buf_char_and_check (tie)
    else
        append_ex_buf_char_and_check (space);
    end
  else
    begin
    if (ex_buf_length+(sp_xptr2-sp_xptr1) > buf_size) then
        buffer_overflow;
    sp_ptr := sp_xptr1;
    while (sp_ptr < sp_xptr2) do
        begin
        append_ex_buf_char (str_pool[sp_ptr]);
        incr(sp_ptr);
        end
    end;
end


@
@^special character@>
This function looks at the string in |ex_buf|, starting at
|ex_buf_xptr| and ending just before |ex_buf_ptr|, and it returns
|true| if there are |enough_chars|, where a special character (even if
it's missing its matching |right_brace|) counts as a single charcter.
This procedure is called only for strings that don't have too many
|right_brace|s.

@<Procedures and functions for name-string processing@>=
function enough_text_chars (@!enough_chars : buf_pointer) : boolean;
begin
num_text_chars := 0;
ex_buf_yptr := ex_buf_xptr;
while ((ex_buf_yptr < ex_buf_ptr) and (num_text_chars < enough_chars)) do
    begin
    incr(ex_buf_yptr);
    if (ex_buf[ex_buf_yptr-1] = left_brace) then
        begin
        incr(brace_level);
        if ((brace_level = 1) and (ex_buf_yptr < ex_buf_ptr)) then
          if (ex_buf[ex_buf_yptr] = backslash) then
            begin
            incr(ex_buf_yptr);                  {skip over the |backslash|}
            while ((ex_buf_yptr < ex_buf_ptr) and (brace_level > 0)) do
                begin
                if (ex_buf[ex_buf_yptr] = right_brace) then
                    decr(brace_level)
                else if (ex_buf[ex_buf_yptr] = left_brace) then
                    incr(brace_level);
                incr(ex_buf_yptr);
                end;
            end;
        end
    else if (ex_buf[ex_buf_yptr-1] = right_brace) then
        decr(brace_level);
    incr(num_text_chars);
    end;
if (num_text_chars < enough_chars) then
    enough_text_chars := false
  else
    enough_text_chars := true;
end;


@
If the last character output for this name part is a |tie| but the
previous character it isn't, we're dealing with a discretionary |tie|;
thus we replace it by a |space| if there are enough characters in the
rest of the name part.

@d long_name = 3                {a name this length or longer is ``long''}

@<Handle a discretionary |tie|@>=
begin
decr(ex_buf_ptr);                       {remove the previous |tie|}
if (ex_buf[ex_buf_ptr-1] = tie) then    {it's not a discretionary |tie|}
    do_nothing
else if (not enough_text_chars (long_name)) then {this is a short name part}
    incr(ex_buf_ptr)                    {so restore the |tie|}
else                                    {replace it by a |space|}
    append_ex_buf_char (space);
end


@
This is a procedure so that |x_format_name| is smaller.

@<Procedures and functions for name-string processing@>=
procedure figure_out_the_formatted_name;
label loop_exit;
begin
@<Figure out the formatted name@>;
end;


@
The |built_in| function {\.{if\$}} pops the top three literals (they
are two function literals and an integer literal, in that order); if
the integer is greater than 0, it executes the second literal, else it
executes the first.  If any of the types is incorrect, it complains
but does nothing else.

@<|execute_fn|({\.{if\$}})@>=
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
pop_lit_stk (pop_lit3,pop_typ3);
if (pop_typ1 <> stk_fn) then
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_fn)
else if (pop_typ2 <> stk_fn) then
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_fn)
else if (pop_typ3 <> stk_int) then
    print_wrong_stk_lit (pop_lit3,pop_typ3,stk_int)
else
    if (pop_lit3 > 0) then
        execute_fn (pop_lit2)
      else
        execute_fn (pop_lit1);
end


@
The |built_in| function {\.{int.to.chr\$}} pops the top (integer)
literal, interpreted as the |ASCII_code| of a single character,
converts it to the corresponding single-character string, and pushes
this string.  If the literal isn't an appropriate integer, it
complains and pushes the null string.

@<|execute_fn|({\.{int.to.chr\$}})@>=
procedure x_int_to_chr;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (s_null, stk_str);
    end
else if ((pop_lit1 < 0) or (pop_lit1 > 127)) then
    begin
    bst_ex_warn (pop_lit1:0,' isn''t valid ASCII');
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    str_room(1);
    append_char (pop_lit1);
    push_lit_stk (make_string, stk_str);
    end;
end;


@
The |built_in| function {\.{int.to.str\$}} pops the top (integer)
literal, converts it to its (unique) string equivalent, and pushes
this string.  If the literal isn't an integer, it complains and pushes
the null string.

@<|execute_fn|({\.{int.to.str\$}})@>=
procedure x_int_to_str;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    int_to_ASCII (pop_lit1, ex_buf, 0, ex_buf_length);@/
    add_pool_buf_and_push;              {push this string onto the stack}
    end;
end;


@
The |built_in| function {\.{missing\$}} pops the top literal and
pushes the integer 1 if it's a missing field, 0 otherwise.  If the
literal isn't a missing field or a string, it complains and pushes 0.
Unlike \.{empty\$}, this function should be called only when
|mess_with_entries| is true.

@<|execute_fn|({\.{missing\$}})@>=
procedure x_missing;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
else if ((pop_typ1 <> stk_str) and (pop_typ1 <> stk_field_missing)) then
    begin
    if (pop_typ1 <> stk_empty) then
        begin
        print_stk_lit (pop_lit1,pop_typ1);
        bst_ex_warn (', not a string or missing field,');
        end;
    push_lit_stk (0, stk_int);
    end
else
    if (pop_typ1 = stk_field_missing) then
        push_lit_stk (1, stk_int)
      else
        push_lit_stk (0, stk_int);
end;


@
The |built_in| function {\.{newline\$}} writes whatever has
accumulated in the output buffer |out_buf| onto the \.{.bbl} file.

@<|execute_fn|({\.{newline\$}})@>=
begin
output_bbl_line;
end


@
The |built_in| function {\.{num.names\$}} pops the top (string)
literal; it pushes the number of names the string represents---one
plus the number of occurrences of the substring ``and'' (ignoring case
differences) surrounded by nonnull |white_space| at the top brace
level.  If the literal isn't a string, it complains and pushes the
value 0.

@<|execute_fn|({\.{num.names\$}})@>=
procedure x_num_names;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (0, stk_int);
    end
else
    begin
    ex_buf_length := 0;
    add_buf_pool (pop_lit1);
    @<Determine the number of names@>;
    push_lit_stk (num_names, stk_int);
    end;
end;


@
This module, while scanning the list of names, counts the occurrences
of ``and'' (ignoring case differences) surrounded by nonnull
|white_space|, and adds 1.

@<Determine the number of names@>=
begin
ex_buf_ptr := 0;
num_names := 0;
while (ex_buf_ptr < ex_buf_length) do
    begin
    name_scan_for_and (pop_lit1);
    incr(num_names);
    end;
end


@
The |built_in| function {\.{pop\$}} pops the top of the stack but
doesn't print it.

@<|execute_fn|({\.{pop\$}})@>=
begin
pop_lit_stk (pop_lit1,pop_typ1);
end


@
The |built_in| function {\.{preamble\$}} pushes onto the stack the
concatenation of all the \.{preamble} strings read from the database
files.

@<|execute_fn|({\.{preamble\$}})@>=
procedure x_preamble;
begin
ex_buf_length := 0;
preamble_ptr := 0;
while (preamble_ptr < num_preamble_strings) do
    begin
    add_buf_pool (s_preamble[preamble_ptr]);
    incr(preamble_ptr);
    end;
add_pool_buf_and_push;          {push the concatenation string onto the stack}
end;


@
@^special character@>
The |built_in| function {\.{purify\$}} pops the top (string) literal,
removes nonalphanumeric characters except for |white_space| and
|sep_char| characters (these get converted to a |space|) and removes
certain alphabetic characters contained in the control sequences
associated with a special character, and pushes the resulting string.
If the literal isn't a string, it complains and pushes the null
string.

@<|execute_fn|({\.{purify\$}})@>=
procedure x_purify;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    ex_buf_length := 0;
    add_buf_pool (pop_lit1);
    @<Perform the purification@>;
    add_pool_buf_and_push;              {push this string onto the stack}
    end;
end;


@
@^special character@>
The resulting string has nonalphanumeric characters removed, and each
|white_space| or |sep_char| character converted to a |space|.  The next
module handles special characters.  This code doesn't complain if the
string isn't brace balanced.

@<Perform the purification@>=
begin
brace_level := 0;       {this is the top level}
ex_buf_xptr := 0;       {this pointer is for the purified string}
ex_buf_ptr := 0;        {and this one is for the original string}
while (ex_buf_ptr < ex_buf_length) do
    begin
    case (lex_class[ex_buf[ex_buf_ptr]]) of
        white_space,
        sep_char :
            begin
            ex_buf[ex_buf_xptr] := space;
            incr(ex_buf_xptr);
            end;
        alpha,
        numeric :
            begin
            ex_buf[ex_buf_xptr] := ex_buf[ex_buf_ptr];
            incr(ex_buf_xptr);
            end;
        othercases
            if (ex_buf[ex_buf_ptr] = left_brace) then
                begin
                incr(brace_level);
                if ((brace_level = 1) and
                                (ex_buf_ptr + 1 < ex_buf_length)) then
                  if (ex_buf[ex_buf_ptr+1] = backslash) then
                    @<Purify a special character@>;
                end
            else if (ex_buf[ex_buf_ptr] = right_brace) then
                if (brace_level > 0) then
                    decr(brace_level)
    endcases;
    incr(ex_buf_ptr);
    end;
ex_buf_length := ex_buf_xptr;
end


@
@^special character@>
Special characters (even without a matching |right_brace|) are
purified by removing the control sequences (but restoring the correct
thing for `\.{\\i}' and `\.{\\j}' as well as the eleven alphabetic
foreign characters in Table~3.2 of the \LaTeX\ manual) and removing
all nonalphanumeric characters (including |white_space| and
|sep_char|s).

@<Purify a special character@>=
begin
incr(ex_buf_ptr);                       {skip over the |left_brace|}
while ((ex_buf_ptr < ex_buf_length) and (brace_level > 0)) do
    begin
    incr(ex_buf_ptr);                   {skip over the |backslash|}
    ex_buf_yptr := ex_buf_ptr;  {mark the beginning of the control sequence}
    while ((ex_buf_ptr < ex_buf_length) and
                (lex_class[ex_buf[ex_buf_ptr]] = alpha)) do@/
        incr(ex_buf_ptr);               {this scans the control sequence}
    control_seq_loc := str_lookup(ex_buf,ex_buf_yptr,ex_buf_ptr-ex_buf_yptr,
                                                control_seq_ilk,dont_insert);
    if (hash_found) then
        @<Purify this accented or foreign character@>;
    while ((ex_buf_ptr < ex_buf_length) and (brace_level > 0) and
                                        (ex_buf[ex_buf_ptr] <> backslash)) do
        begin                   {this scans to the next control sequence}
        case (lex_class[ex_buf[ex_buf_ptr]]) of
            alpha,
            numeric :
                begin
                ex_buf[ex_buf_xptr] := ex_buf[ex_buf_ptr];
                incr(ex_buf_xptr);
                end;
            othercases
                if (ex_buf[ex_buf_ptr] = right_brace) then
                    decr(brace_level)
                else if (ex_buf[ex_buf_ptr] = left_brace) then
                    incr(brace_level)
        endcases;
        incr(ex_buf_ptr);
        end;
    end;
decr(ex_buf_ptr);               {unskip the |right_brace| (or last character)}
end


@
We consider the purified character to be either the first alphabetic
character of its control sequence, or perhaps both alphabetic
characters.

@<Purify this accented or foreign character@>=
begin
ex_buf[ex_buf_xptr] := ex_buf[ex_buf_yptr]; {the first alphabetic character}
incr(ex_buf_xptr);
case (ilk_info[control_seq_loc]) of
    n_oe,
    n_oe_upper,
    n_ae,
    n_ae_upper,
    n_ss :
        begin                                   {and the second}
        ex_buf[ex_buf_xptr] := ex_buf[ex_buf_yptr+1];
        incr(ex_buf_xptr);
        end;
    othercases
        do_nothing
endcases;
end


@
The |built_in| function {\.{quote\$}} pushes the string consisting of
the |double_quote| character.

@<|execute_fn|({\.{quote\$}})@>=
procedure x_quote;
begin
str_room(1);
append_char (double_quote);
push_lit_stk (make_string, stk_str);
end;


@
The |built_in| function {\.{skip\$}} is a no-op.

@<|execute_fn|({\.{skip\$}})@>=
begin
do_nothing;
end


@
The |built_in| function {\.{stack\$}} pops and prints the whole stack;
it's meant to be used for style designers while debugging.

@<|execute_fn|({\.{stack\$}})@>=
begin
pop_whole_stack;
end


@
@^push the literal stack@>
The |built_in| function {\.{substring\$}} pops the top three literals
(they are the two integers literals |pop_lit1| and |pop_lit2| and a
string literal, in that order).  It pushes the substring of the (at
most) |pop_lit1| consecutive characters starting at the |pop_lit2|th
character (assuming 1-based indexing) if |pop_lit2| is positive, and
ending at the |-pop_lit2|th character from the end if |pop_lit2| is
negative (where the first character from the end is the last
character).  If any of the types is incorrect, it complain and pushes
the null string.

@<|execute_fn|({\.{substring\$}})@>=
procedure x_substring;
label exit;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
pop_lit_stk (pop_lit3,pop_typ3);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ2 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_int);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ3 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit3,pop_typ3,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    sp_length := length(pop_lit3);
    if (pop_lit1 >= sp_length) then
      if ((pop_lit2 = 1) or (pop_lit2 = -1)) then
        begin
        repush_string;
        return;
        end;
    if ((pop_lit1 <= 0) or (pop_lit2 = 0) or (pop_lit2 > sp_length) or
                                        (pop_lit2 < -sp_length)) then
        begin
        push_lit_stk (s_null, stk_str);
        return;
        end
      else
        @<Form the appropriate substring@>;
    end;
exit:
end;


@
@^push the literal stack@>
This module finds the substring as described in the last section,
and slides it into place in the string pool, if necessary.

@<Form the appropriate substring@>=
begin
if (pop_lit2 > 0) then
    begin
    if (pop_lit1 > sp_length - (pop_lit2-1)) then
        pop_lit1 := sp_length - (pop_lit2-1);
    sp_ptr := str_start[pop_lit3] + (pop_lit2-1);
    sp_end := sp_ptr + pop_lit1;
    if (pop_lit2 = 1) then
      if (pop_lit3 >= cmd_str_ptr) then {no shifting---merely change pointers}
        begin
        str_start[pop_lit3+1] := sp_end;
        unflush_string;
        incr(lit_stk_ptr);
        return;
        end;
    end
else                                    {|-ex_buf_length <= pop_lit2 < 0|}
    begin
    pop_lit2 := -pop_lit2;
    if (pop_lit1 > sp_length - (pop_lit2-1)) then
        pop_lit1 := sp_length - (pop_lit2-1);
    sp_end := str_start[pop_lit3+1] - (pop_lit2-1);
    sp_ptr := sp_end - pop_lit1;
    end;
while (sp_ptr < sp_end) do                      {shift the substring}
    begin
    append_char (str_pool[sp_ptr]);
    incr(sp_ptr);
    end;
push_lit_stk (make_string, stk_str);            {and push it onto the stack}
end


@
The |built_in| function {\.{swap\$}} pops the top two literals from
the stack and pushes them back swapped.

@<|execute_fn|({\.{swap\$}})@>=
procedure x_swap;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if ((pop_typ1 <> stk_str) or (pop_lit1 < cmd_str_ptr)) then
    begin
    push_lit_stk (pop_lit1, pop_typ1);
    if ((pop_typ2 = stk_str) and (pop_lit2 >= cmd_str_ptr)) then
        unflush_string;
    push_lit_stk (pop_lit2, pop_typ2);
    end
else if ((pop_typ2 <> stk_str) or (pop_lit2 < cmd_str_ptr)) then
    begin
    unflush_string;                     {this is |pop_lit1|}
    push_lit_stk (pop_lit1, stk_str);
    push_lit_stk (pop_lit2, pop_typ2);
    end
else                                    {bummer, both are recent strings}
    @<Swap the two strings (they're at the end of |str_pool|)@>;
end;


@
We have to swap both (a)~the strings at the end of the string pool,
and (b)~their pointers on the literal stack.

@<Swap the two strings (they're at the end of |str_pool|)@>=
begin
ex_buf_length := 0;
add_buf_pool (pop_lit2);                {save the second string}
sp_ptr := str_start[pop_lit1];
sp_end := str_start[pop_lit1+1];
while (sp_ptr < sp_end) do              {slide the first string down}
    begin
    append_char (str_pool[sp_ptr]);
    incr(sp_ptr);
    end;
push_lit_stk (make_string, stk_str);    {and push it onto the stack}
add_pool_buf_and_push;                  {push second string onto the stack}
end


@
@^special character@>
The |built_in| function {\.{text.length\$}} pops the top (string)
literal, and pushes the number of text characters it contains, where
an accented character (more precisely, a ``special character''$\!$,
defined earlier) counts as a single text character, even if it's
missing its matching |right_brace|, and where braces don't count as
text characters.  If the literal isn't a string, it complains and
pushes the null string.

@<|execute_fn|({\.{text.length\$}})@>=
procedure x_text_length;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else
    begin
    num_text_chars := 0;
    @<Count the text characters@>;
    push_lit_stk (num_text_chars, stk_int);     {and push it onto the stack}
    end;
end;


@
@^special character@>
Here we determine the number of text characters in the string, where
an entire special character counts as a single text character (even if
it's missing its matching |right_brace|), and where braces don't count
as text characters.

@<Count the text characters@>=
begin
sp_ptr := str_start[pop_lit1];
sp_end := str_start[pop_lit1+1];
sp_brace_level := 0;
while (sp_ptr < sp_end) do
    begin
    incr(sp_ptr);
    if (str_pool[sp_ptr-1] = left_brace) then
        begin
        incr(sp_brace_level);
        if ((sp_brace_level = 1) and (sp_ptr < sp_end)) then
          if (str_pool[sp_ptr] = backslash) then
            begin
            incr(sp_ptr);               {skip over the |backslash|}
            while ((sp_ptr < sp_end) and (sp_brace_level > 0)) do
                begin
                if (str_pool[sp_ptr] = right_brace) then
                    decr(sp_brace_level)
                else if (str_pool[sp_ptr] = left_brace) then
                    incr(sp_brace_level);
                incr(sp_ptr);
                end;
            incr(num_text_chars);
            end;
        end
    else if (str_pool[sp_ptr-1] = right_brace) then
        begin
        if (sp_brace_level > 0) then
            decr(sp_brace_level);
        end
    else
        incr(num_text_chars);
    end;
end


@
@^special character@>
The |built_in| function {\.{text.prefix\$}} pops the top two literals
(the integer literal |pop_lit1| and a string literal, in that order).
It pushes the substring of the (at most) |pop_lit1| consecutive text
characters starting from the beginning of the string.  This function
is similar to {\.{substring\$}}, but this one considers an accented
character (or more precisely, a ``special character''$\!$, even if
it's missing its matching |right_brace|) to be a single text character
(rather than however many |ASCII_code| characters it actually
comprises), and this function doesn't consider braces to be text
characters; furthermore, this function appends any needed matching
|right_brace|s.  If any of the types is incorrect, it complains and
pushes the null string.

@<|execute_fn|({\.{text.prefix\$}})@>=
procedure x_text_prefix;
label exit;
begin
pop_lit_stk (pop_lit1,pop_typ1);
pop_lit_stk (pop_lit2,pop_typ2);
if (pop_typ1 <> stk_int) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_typ2 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit2,pop_typ2,stk_str);
    push_lit_stk (s_null, stk_str);
    end
else if (pop_lit1 <= 0) then
    begin
    push_lit_stk (s_null, stk_str);
    return;
    end
else
    @<Form the appropriate prefix@>;
exit:
end;


@
@^push the literal stack@>
This module finds the prefix as described in the last section, and
appends any needed matching |right_brace|s.

@<Form the appropriate prefix@>=
begin
sp_ptr := str_start[pop_lit2];
sp_end := str_start[pop_lit2+1];        {this may change}
@<Scan the appropriate number of characters@>;
if (pop_lit2 >= cmd_str_ptr) then       {no shifting---merely change pointers}
    pool_ptr := sp_end
else
    while (sp_ptr < sp_end) do          {shift the substring}
        begin
        append_char (str_pool[sp_ptr]);
        incr(sp_ptr);
        end;
while (sp_brace_level > 0) do           {add matching |right_brace|s}
    begin
    append_char (right_brace);
    decr(sp_brace_level);
    end;
push_lit_stk (make_string, stk_str);    {and push it onto the stack}
end


@
@^special character@>
This section scans |pop_lit1| text characters, where an entire special
character counts as a single text character (even if it's missing its
matching |right_brace|), and where braces don't count as text
characters.

@<Scan the appropriate number of characters@>=
begin
num_text_chars := 0;
sp_brace_level := 0;
sp_xptr1 := sp_ptr;
while ((sp_xptr1 < sp_end) and (num_text_chars < pop_lit1)) do
    begin
    incr(sp_xptr1);
    if (str_pool[sp_xptr1-1] = left_brace) then
        begin
        incr(sp_brace_level);
        if ((sp_brace_level = 1) and (sp_xptr1 < sp_end)) then
          if (str_pool[sp_xptr1] = backslash) then
            begin
            incr(sp_xptr1);             {skip over the |backslash|}
            while ((sp_xptr1 < sp_end) and (sp_brace_level > 0)) do
                begin
                if (str_pool[sp_xptr1] = right_brace) then
                    decr(sp_brace_level)
                else if (str_pool[sp_xptr1] = left_brace) then
                    incr(sp_brace_level);
                incr(sp_xptr1);
                end;
            incr(num_text_chars);
            end;
        end
    else if (str_pool[sp_xptr1-1] = right_brace) then
        begin
        if (sp_brace_level > 0) then
            decr(sp_brace_level);
        end
    else
        incr(num_text_chars);
    end;
sp_end := sp_xptr1;
end


@
The |built_in| function {\.{top\$}} pops and prints the top of the
stack.

@<|execute_fn|({\.{top\$}})@>=
begin
pop_top_and_print;
end


@
The |built_in| function {\.{type\$}} pushes the appropriate string
from |type_list| onto the stack (unless either it's |undefined| or
|empty|, in which case it pushes the null string).

@<|execute_fn|({\.{type\$}})@>=
procedure x_type;
begin
if (not mess_with_entries) then
    bst_cant_mess_with_entries_print
  else
    if ((type_list[cite_ptr] = undefined) or
                                (type_list[cite_ptr] = empty)) then
        push_lit_stk (s_null, stk_str)
      else
        push_lit_stk (hash_text[type_list[cite_ptr]], stk_str);
end;


@
The |built_in| function {\.{warning\$}} pops the top (string) literal
and prints it following a warning message.  This is implemented as a
special |built_in| function rather than using the {\.{top\$}} function
so that it can |mark_warning|.

@<|execute_fn|({\.{warning\$}})@>=
procedure x_warning;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str)
else
    begin
    print ('Warning--');
    print_lit (pop_lit1,pop_typ1);
    mark_warning;
    end;
end;


@
The |built_in| function {\.{while\$}} pops the top two (function)
literals, and keeps executing the second as long as the (integer)
value left on the stack by executing the first is greater than 0.  If
either type is incorrect, it complains but does nothing else.

@<|execute_fn|({\.{while\$}})@>=
begin
pop_lit_stk (r_pop_lt1,r_pop_tp1);
pop_lit_stk (r_pop_lt2,r_pop_tp2);
if (r_pop_tp1 <> stk_fn) then
    print_wrong_stk_lit (r_pop_lt1,r_pop_tp1,stk_fn)
else if (r_pop_tp2 <> stk_fn) then
    print_wrong_stk_lit (r_pop_lt2,r_pop_tp2,stk_fn)
else
    loop
        begin
        execute_fn (r_pop_lt2);                 {this is the \.{while\$} test}
        pop_lit_stk (pop_lit1,pop_typ1);
        if (pop_typ1 <> stk_int) then
            begin
            print_wrong_stk_lit (pop_lit1,pop_typ1,stk_int);
            goto end_while;
            end
        else
            if (pop_lit1 > 0) then
                execute_fn (r_pop_lt1)          {this is the \.{while\$} body}
              else
                goto end_while;
        end;
end_while:      {justifies this |mean_while|}
end


@
@^literal literal@>
@^special character@>
The |built_in| function {\.{width\$}} pops the top (string) literal
and pushes the integer that represents its width in units specified by
the |char_width| array.  This function takes the literal literally;
that is, it assumes each character in the string is to be printed as
is, regardless of whether the character has a special meaning to \TeX,
except that special characters (even without their |right_brace|s) are
handled specially.  If the literal isn't a string, it complains and
pushes~0.

@<|execute_fn|({\.{width\$}})@>=
procedure x_width;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    begin
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str);
    push_lit_stk (0, stk_int);
    end
else
    begin
    ex_buf_length := 0;
    add_buf_pool (pop_lit1);
    string_width := 0;
    @<Add up the |char_width|s in this string@>;
    push_lit_stk (string_width, stk_int);
    end
end;


@
We use the natural width for all but special characters, and we
complain if the string isn't brace-balanced.

@<Add up the |char_width|s in this string@>=
begin
brace_level := 0;                       {we're at the top level}
ex_buf_ptr := 0;                        {and the beginning of string}
while (ex_buf_ptr < ex_buf_length) do
    begin
    if (ex_buf[ex_buf_ptr] = left_brace) then
        begin
        incr(brace_level);
        if ((brace_level = 1) and (ex_buf_ptr + 1 < ex_buf_length)) then
            if (ex_buf[ex_buf_ptr+1] = backslash) then
                @<Determine the width of this special character@>
              else
                string_width := string_width + char_width[left_brace]
          else
            string_width := string_width + char_width[left_brace];
        end
    else if (ex_buf[ex_buf_ptr] = right_brace) then
        begin
        decr_brace_level (pop_lit1);
        string_width := string_width + char_width[right_brace];
        end
    else
        string_width := string_width + char_width[ex_buf[ex_buf_ptr]];
    incr(ex_buf_ptr);
    end;
check_brace_level (pop_lit1);
end


@
@^special character@>
We use the natural widths of all characters except that some
characters have no width: braces, control sequences (except for the
usual 13 accented and foreign characters, whose widths are given in
the next module), and |white_space| following control sequences (even
a null control sequence).

@<Determine the width of this special character@>=
begin
incr(ex_buf_ptr);                               {skip over the |left_brace|}
while ((ex_buf_ptr < ex_buf_length) and (brace_level > 0)) do
    begin
    incr(ex_buf_ptr);                   {skip over the |backslash|}
    ex_buf_xptr := ex_buf_ptr;
    while ((ex_buf_ptr < ex_buf_length) and
                (lex_class[ex_buf[ex_buf_ptr]] = alpha)) do@/
        incr(ex_buf_ptr);               {this scans the control sequence}
    if ((ex_buf_ptr < ex_buf_length) and (ex_buf_ptr = ex_buf_xptr)) then
        incr(ex_buf_ptr)                {this skips a nonalpha control seq}
      else
        begin
        control_seq_loc := str_lookup(ex_buf,ex_buf_xptr,
                        ex_buf_ptr-ex_buf_xptr,control_seq_ilk,dont_insert);
        if (hash_found) then
            @<Determine the width of this accented or foreign character@>;
        end;
    while ((ex_buf_ptr < ex_buf_length) and
                (lex_class[ex_buf[ex_buf_ptr]] = white_space)) do
    incr(ex_buf_ptr);                   {this skips following |white_space|}
    while ((ex_buf_ptr < ex_buf_length) and (brace_level > 0) and
                                        (ex_buf[ex_buf_ptr] <> backslash)) do
        begin                   {this scans to the next control sequence}
        if (ex_buf[ex_buf_ptr] = right_brace) then
            decr(brace_level)
        else if (ex_buf[ex_buf_ptr] = left_brace) then
            incr(brace_level)
        else
            string_width := string_width + char_width[ex_buf[ex_buf_ptr]];
        incr(ex_buf_ptr);
        end;
    end;
decr(ex_buf_ptr);                       {unskip the |right_brace|}
end


@
Five of the 13 possibilities resort to special information not present
in the |char_width| array; the other eight simply use |char_width|'s
information for the first letter of the control sequence.

@<Determine the width of this accented or foreign character@>=
begin
case (ilk_info[control_seq_loc]) of
    n_ss : string_width := string_width + ss_width;
    n_ae : string_width := string_width + ae_width;
    n_oe : string_width := string_width + oe_width;
    n_ae_upper : string_width := string_width + upper_ae_width;
    n_oe_upper : string_width := string_width + upper_oe_width;
    othercases
        string_width := string_width + char_width[ex_buf[ex_buf_xptr]]
endcases;
end


@
The |built_in| function {\.{write\$}} pops the top (string) literal
and writes it onto the output buffer |out_buf| (which will result in
stuff being written onto the \.{.bbl} file if the buffer fills up).  If
the literal isn't a string, it complains but does nothing else.

@<|execute_fn|({\.{write\$}})@>=
procedure x_write;
begin
pop_lit_stk (pop_lit1,pop_typ1);
if (pop_typ1 <> stk_str) then
    print_wrong_stk_lit (pop_lit1,pop_typ1,stk_str)
else
    add_out_pool (pop_lit1);
end;



@* Cleaning up.
@^clich\'e-\`a-trois@>
@^fat lady@>
@^turn out lights@>
@^Yogi@>
This section does any last-minute printing and ends the program.

@<Clean up and leave@>=
begin
if ((read_performed) and (not reading_completed)) then
    begin
    print ('Aborted at line ',bib_line_num:0,' of file ');
    print_bib_name;
    end;
trace_and_stat_printing;
@<Print the job |history|@>;
a_close (log_file);
{turn out the lights, the fat lady has sung; it's over, Yogi}
end


@
Here we print |trace| and/or |stat| information, if desired.

@<Procedures and functions for all file I/O, error messages, and such@>=
procedure trace_and_stat_printing;
begin
@#
  trace
  @<Print all \.{.bib}- and \.{.bst}-file information@>;
  @<Print all |cite_list| and entry information@>;
  @<Print the |wiz_defined| functions@>;
  @<Print the string pool@>;
  ecart@/
@#
  stat
  @<Print usage statistics@>;
  tats@/
@#
end;


@
This prints information obtained from the \.{.aux} file about the
other files.

@<Print all \.{.bib}- and \.{.bst}-file information@>=
begin
if (num_bib_files = 1) then
    trace_pr_ln ('The 1 database file is')
  else
    trace_pr_ln ('The ',num_bib_files:0,' database files are');
if (num_bib_files = 0) then
    trace_pr_ln ('   undefined')
  else
    begin
    bib_ptr := 0;
    while (bib_ptr < num_bib_files) do
        begin
        trace_pr ('   ');
        trace_pr_pool_str (cur_bib_str);
        trace_pr_pool_str (s_bib_extension);
        trace_pr_newline;
        incr(bib_ptr);
        end;
    end;
trace_pr ('The style file is ');
if (bst_str = 0) then
    trace_pr_ln ('undefined')
  else
    begin
    trace_pr_pool_str (bst_str);
    trace_pr_pool_str (s_bst_extension);
    trace_pr_newline;
    end;
end


@
In entry-sorted order, this prints an entry's |cite_list| string and,
indirectly, its entry type and entry variables.

@<Print all |cite_list| and entry information@>=
begin
if (all_entries) then
    trace_pr ('all_marker=',all_marker:0,', ');
if (read_performed) then
    trace_pr_ln ('old_num_cites=',old_num_cites:0)
  else
    trace_pr_newline;
trace_pr ('The ',num_cites:0);
if (num_cites = 1) then
    trace_pr_ln (' entry:')
  else
    trace_pr_ln (' entries:');
if (num_cites = 0) then
    trace_pr_ln ('   undefined')
  else
    begin
    sort_cite_ptr := 0;
    while (sort_cite_ptr < num_cites) do
        begin
        if (not read_completed) then    {we didn't finish the \.{read} command}
            cite_ptr := sort_cite_ptr
          else
            cite_ptr := sorted_cites[sort_cite_ptr];
        trace_pr_pool_str (cur_cite_str);
        if (read_performed) then
            @<Print entry information@>
          else
            trace_pr_newline;
        incr(sort_cite_ptr);
        end;
    end;
end


@
This prints information gathered while reading the \.{.bst} and
\.{.bib} files.

@<Print entry information@>=
begin
trace_pr (', entry-type ');
if (type_list[cite_ptr] = undefined) then
    undefined : trace_pr ('unknown')
else if (type_list[cite_ptr] = empty) then
    trace_pr ('--- no type found')
else
    trace_pr_pool_str (hash_text[type_list[cite_ptr]]);
trace_pr_ln (', has entry strings');
@<Print entry strings@>;
trace_pr ('  has entry integers');
@<Print entry integers@>;
trace_pr_ln ('  and has fields');
@<Print fields@>;
end


@
This prints, for the current entry, the strings declared by the
\.{entry} command.

@<Print entry strings@>=
begin
if (num_ent_strs = 0) then
    trace_pr_ln ('    undefined')
else if (not read_completed) then
    trace_pr_ln ('    uninitialized')
else
    begin
    str_ent_ptr := cite_ptr * num_ent_strs;
    while (str_ent_ptr < (cite_ptr+1)*num_ent_strs) do
        begin
        ent_chr_ptr := 0;
        trace_pr ('    "');
        while (entry_strs[str_ent_ptr][ent_chr_ptr] <> end_of_string) do
            begin
            trace_pr (xchr[entry_strs[str_ent_ptr][ent_chr_ptr]]);
            incr(ent_chr_ptr);
            end;
        trace_pr_ln ('"');
        incr(str_ent_ptr);
        end;
    end;
end


@
This prints, for the current entry, the integers declared by the
\.{entry} command.

@<Print entry integers@>=
begin
if (num_ent_ints = 0) then
    trace_pr (' undefined')
else if (not read_completed) then
    trace_pr (' uninitialized')
else
    begin
    int_ent_ptr := cite_ptr*num_ent_ints;
    while (int_ent_ptr < (cite_ptr+1)*num_ent_ints) do
        begin
        trace_pr (' ',entry_ints[int_ent_ptr]:0);
        incr(int_ent_ptr);
        end;
    end;
trace_pr_newline;
end


@
This prints the fields stored for the current entry.

@<Print fields@>=
begin
if (not read_performed) then
    trace_pr_ln ('    uninitialized')
  else
    begin
    field_ptr := cite_ptr * num_fields;
    field_end_ptr := field_ptr + num_fields;
    no_fields := true;
    while (field_ptr < field_end_ptr) do
        begin
        if (field_info[field_ptr] <> missing) then
            begin
            trace_pr ('    "');
            trace_pr_pool_str (field_info[field_ptr]);
            trace_pr_ln ('"');
            no_fields := false;
            end;
        incr(field_ptr);
        end;
    if (no_fields) then
        trace_pr_ln ('    missing');
    end;
end


@
This gives all the |wiz_defined| functions that appeared in the
\.{.bst} file.

@<Print the |wiz_defined| functions@>=
begin
trace_pr_ln ('The wiz-defined functions are');
if (wiz_def_ptr = 0) then
    trace_pr_ln ('   nonexistent')
  else
    begin
    wiz_fn_ptr := 0;
    while (wiz_fn_ptr < wiz_def_ptr) do
        begin
        if (wiz_functions[wiz_fn_ptr] = end_of_def) then
            trace_pr_ln (wiz_fn_ptr:0,'--end-of-def--')
        else if (wiz_functions[wiz_fn_ptr] = quote_next_fn) then
            trace_pr (wiz_fn_ptr:0,'  quote_next_function    ')
        else
            begin
            trace_pr (wiz_fn_ptr:0,'  `');
            trace_pr_pool_str (hash_text[wiz_functions[wiz_fn_ptr]]);
            trace_pr_ln ('''');
            end;
        incr(wiz_fn_ptr);
        end;
   end;
end


@
This includes all the `static' strings (that is, those that are also
in the hash table), but none of the dynamic strings (that is, those
put on the stack while executing \.{.bst} commands).

@<Print the string pool@>=
begin
trace_pr_ln ('The string pool is');
str_num := 1;
while (str_num < str_ptr) do
    begin
    trace_pr (str_num:4, str_start[str_num]:6,' "');
    trace_pr_pool_str (str_num);
    trace_pr_ln ('"');
    incr(str_num);
    end;
end


@
@^statistics@>
These statistics can help determine how large some of the constants
should be and can tell how useful certain |built_in| functions are.
They are written to the same files as tracing information.

@d stat_pr == trace_pr
@d stat_pr_ln == trace_pr_ln
@d stat_pr_pool_str == trace_pr_pool_str

@<Print usage statistics@>=
begin
stat_pr ('You''ve used ',num_cites:0);
if (num_cites = 1) then
    stat_pr_ln (' entry,')
  else
    stat_pr_ln (' entries,');
stat_pr_ln ('            ',wiz_def_ptr:0,' wiz_defined-function locations,');
stat_pr_ln ('            ',str_ptr:0,' strings with ',str_start[str_ptr]:0,
                                                        ' characters,');
blt_in_ptr := 0;
total_ex_count := 0;
while (blt_in_ptr < num_blt_in_fns) do
    begin
    total_ex_count := total_ex_count + execution_count[blt_in_ptr];
    incr(blt_in_ptr);
    end;
stat_pr_ln ('and the built_in function-call counts, ', total_ex_count:0,
                                                        ' in all, are:');
blt_in_ptr := 0;
while (blt_in_ptr < num_blt_in_fns) do
    begin
    stat_pr_pool_str (hash_text[blt_in_loc[blt_in_ptr]]);
    stat_pr_ln (' -- ',execution_count[blt_in_ptr]:0);
    incr(blt_in_ptr);
    end;
end


@
@^bunk, history@>
@^system dependencies@>
@:this can't happen}{\quad History is bunk@>
Some implementations may wish to pass the |history| value to the
operating system so that it can be used to govern whether or not other
programs are started. Here we simply report the history to the user.

@<Print the job |history|@>=
case (history) of
    spotless : do_nothing;
    warning_message : begin
                      if (err_count = 1) then
                          print_ln ('(There was 1 warning)')
                        else
                          print_ln ('(There were ',err_count:0,' warnings)');
                      end;
    error_message : begin
                    if (err_count = 1) then
                        print_ln ('(There was 1 error message)')
                      else
                        print_ln ('(There were ',err_count:0,
                                                        ' error messages)');
                    end;
    fatal_message : print_ln ('(That was a fatal error)');
    othercases begin
               print ('History is bunk');
               print_confusion;
               end
endcases



@* System-dependent changes.
@^system dependencies@>
This section should be replaced, if necessary, by changes to the program
that are necessary to make \BibTeX\ work at a particular installation.
It is usually best to design your change file so that all changes to
previous sections preserve the section numbering; then everybody's version
will be consistent with the printed program. More extensive changes,
which introduce new sections, can be inserted here; then only the index
itself will get a new section number.



@* Index.
@.this can't happen@>
Here is where you can find all uses of each identifier in the program,
with underlined entries pointing to where the identifier was defined.
If the identifier is only one letter long, however, you get to see only
the underlined entries. All references are to section numbers instead of
page numbers.

This index also lists a few error messages and other aspects of the
program that you might want to look up some day. For example, the
entry for ``system dependencies'' lists all sections that should
receive special attention from people who are installing \TeX\ in a
new operating environment. A list of various things that can't happen
appears under ``this can't happen''$\!$.