PowerPC: Update NEWS with ppc64 STT_GNU_IFUNC support

[glibc.git] / manual / pattern.texi

@node Pattern Matching, I/O Overview, Searching and Sorting, Top
@c %MENU% Matching shell ``globs'' and regular expressions
@chapter Pattern Matching

@Theglibc{} provides pattern matching facilities for two kinds of
patterns: regular expressions and file-name wildcards.  The library also
provides a facility for expanding variable and command references and
parsing text into words in the way the shell does.

@menu
* Wildcard Matching::    Matching a wildcard pattern against a single string.
* Globbing::             Finding the files that match a wildcard pattern.
* Regular Expressions::  Matching regular expressions against strings.
* Word Expansion::       Expanding shell variables, nested commands,
                            arithmetic, and wildcards.
                            This is what the shell does with shell commands.
@end menu

@node Wildcard Matching
@section Wildcard Matching

@pindex fnmatch.h
This section describes how to match a wildcard pattern against a
particular string.  The result is a yes or no answer: does the
string fit the pattern or not.  The symbols described here are all
declared in @file{fnmatch.h}.

@comment fnmatch.h
@comment POSIX.2
@deftypefun int fnmatch (const char *@var{pattern}, const char *@var{string}, int @var{flags})
This function tests whether the string @var{string} matches the pattern
@var{pattern}.  It returns @code{0} if they do match; otherwise, it
returns the nonzero value @code{FNM_NOMATCH}.  The arguments
@var{pattern} and @var{string} are both strings.

The argument @var{flags} is a combination of flag bits that alter the
details of matching.  See below for a list of the defined flags.

In @theglibc{}, @code{fnmatch} might sometimes report ``errors'' by
returning nonzero values that are not equal to @code{FNM_NOMATCH}.
@end deftypefun

These are the available flags for the @var{flags} argument:

@table @code
@comment fnmatch.h
@comment GNU
@item FNM_FILE_NAME
Treat the @samp{/} character specially, for matching file names.  If
this flag is set, wildcard constructs in @var{pattern} cannot match
@samp{/} in @var{string}.  Thus, the only way to match @samp{/} is with
an explicit @samp{/} in @var{pattern}.

@comment fnmatch.h
@comment POSIX.2
@item FNM_PATHNAME
This is an alias for @code{FNM_FILE_NAME}; it comes from POSIX.2.  We
don't recommend this name because we don't use the term ``pathname'' for
file names.

@comment fnmatch.h
@comment POSIX.2
@item FNM_PERIOD
Treat the @samp{.} character specially if it appears at the beginning of
@var{string}.  If this flag is set, wildcard constructs in @var{pattern}
cannot match @samp{.} as the first character of @var{string}.

If you set both @code{FNM_PERIOD} and @code{FNM_FILE_NAME}, then the
special treatment applies to @samp{.} following @samp{/} as well as to
@samp{.} at the beginning of @var{string}.  (The shell uses the
@code{FNM_PERIOD} and @code{FNM_FILE_NAME} flags together for matching
file names.)

@comment fnmatch.h
@comment POSIX.2
@item FNM_NOESCAPE
Don't treat the @samp{\} character specially in patterns.  Normally,
@samp{\} quotes the following character, turning off its special meaning
(if any) so that it matches only itself.  When quoting is enabled, the
pattern @samp{\?} matches only the string @samp{?}, because the question
mark in the pattern acts like an ordinary character.

If you use @code{FNM_NOESCAPE}, then @samp{\} is an ordinary character.

@comment fnmatch.h
@comment GNU
@item FNM_LEADING_DIR
Ignore a trailing sequence of characters starting with a @samp{/} in
@var{string}; that is to say, test whether @var{string} starts with a
directory name that @var{pattern} matches.

If this flag is set, either @samp{foo*} or @samp{foobar} as a pattern
would match the string @samp{foobar/frobozz}.

@comment fnmatch.h
@comment GNU
@item FNM_CASEFOLD
Ignore case in comparing @var{string} to @var{pattern}.

@comment fnmatch.h
@comment GNU
@item FNM_EXTMATCH
@cindex Korn Shell
@pindex ksh
Recognize beside the normal patterns also the extended patterns
introduced in @file{ksh}.  The patterns are written in the form
explained in the following table where @var{pattern-list} is a @code{|}
separated list of patterns.

@table @code
@item ?(@var{pattern-list})
The pattern matches if zero or one occurrences of any of the patterns
in the @var{pattern-list} allow matching the input string.

@item *(@var{pattern-list})
The pattern matches if zero or more occurrences of any of the patterns
in the @var{pattern-list} allow matching the input string.

@item +(@var{pattern-list})
The pattern matches if one or more occurrences of any of the patterns
in the @var{pattern-list} allow matching the input string.

@item @@(@var{pattern-list})
The pattern matches if exactly one occurrence of any of the patterns in
the @var{pattern-list} allows matching the input string.

@item !(@var{pattern-list})
The pattern matches if the input string cannot be matched with any of
the patterns in the @var{pattern-list}.
@end table
@end table

@node Globbing
@section Globbing

@cindex globbing
The archetypal use of wildcards is for matching against the files in a
directory, and making a list of all the matches.  This is called
@dfn{globbing}.

You could do this using @code{fnmatch}, by reading the directory entries
one by one and testing each one with @code{fnmatch}.  But that would be
slow (and complex, since you would have to handle subdirectories by
hand).

The library provides a function @code{glob} to make this particular use
of wildcards convenient.  @code{glob} and the other symbols in this
section are declared in @file{glob.h}.

@menu
* Calling Glob::             Basic use of @code{glob}.
* Flags for Globbing::       Flags that enable various options in @code{glob}.
* More Flags for Globbing::  GNU specific extensions to @code{glob}.
@end menu

@node Calling Glob
@subsection Calling @code{glob}

The result of globbing is a vector of file names (strings).  To return
this vector, @code{glob} uses a special data type, @code{glob_t}, which
is a structure.  You pass @code{glob} the address of the structure, and
it fills in the structure's fields to tell you about the results.

@comment glob.h
@comment POSIX.2
@deftp {Data Type} glob_t
This data type holds a pointer to a word vector.  More precisely, it
records both the address of the word vector and its size.  The GNU
implementation contains some more fields which are non-standard
extensions.

@table @code
@item gl_pathc
The number of elements in the vector, excluding the initial null entries
if the GLOB_DOOFFS flag is used (see gl_offs below).

@item gl_pathv
The address of the vector.  This field has type @w{@code{char **}}.

@item gl_offs
The offset of the first real element of the vector, from its nominal
address in the @code{gl_pathv} field.  Unlike the other fields, this
is always an input to @code{glob}, rather than an output from it.

If you use a nonzero offset, then that many elements at the beginning of
the vector are left empty.  (The @code{glob} function fills them with
null pointers.)

The @code{gl_offs} field is meaningful only if you use the
@code{GLOB_DOOFFS} flag.  Otherwise, the offset is always zero
regardless of what is in this field, and the first real element comes at
the beginning of the vector.

@item gl_closedir
The address of an alternative implementation of the @code{closedir}
function.  It is used if the @code{GLOB_ALTDIRFUNC} bit is set in
the flag parameter.  The type of this field is
@w{@code{void (*) (void *)}}.

This is a GNU extension.

@item gl_readdir
The address of an alternative implementation of the @code{readdir}
function used to read the contents of a directory.  It is used if the
@code{GLOB_ALTDIRFUNC} bit is set in the flag parameter.  The type of
this field is @w{@code{struct dirent *(*) (void *)}}.

This is a GNU extension.

@item gl_opendir
The address of an alternative implementation of the @code{opendir}
function.  It is used if the @code{GLOB_ALTDIRFUNC} bit is set in
the flag parameter.  The type of this field is
@w{@code{void *(*) (const char *)}}.

This is a GNU extension.

@item gl_stat
The address of an alternative implementation of the @code{stat} function
to get information about an object in the filesystem.  It is used if the
@code{GLOB_ALTDIRFUNC} bit is set in the flag parameter.  The type of
this field is @w{@code{int (*) (const char *, struct stat *)}}.

This is a GNU extension.

@item gl_lstat
The address of an alternative implementation of the @code{lstat}
function to get information about an object in the filesystems, not
following symbolic links.  It is used if the @code{GLOB_ALTDIRFUNC} bit
is set in the flag parameter.  The type of this field is @code{@w{int
(*) (const char *,} @w{struct stat *)}}.

This is a GNU extension.

@item gl_flags
The flags used when @code{glob} was called.  In addition, @code{GLOB_MAGCHAR}
might be set.  See @ref{Flags for Globbing} for more details.

This is a GNU extension.
@end table
@end deftp

For use in the @code{glob64} function @file{glob.h} contains another
definition for a very similar type.  @code{glob64_t} differs from
@code{glob_t} only in the types of the members @code{gl_readdir},
@code{gl_stat}, and @code{gl_lstat}.

@comment glob.h
@comment GNU
@deftp {Data Type} glob64_t
This data type holds a pointer to a word vector.  More precisely, it
records both the address of the word vector and its size.  The GNU
implementation contains some more fields which are non-standard
extensions.

@table @code
@item gl_pathc
The number of elements in the vector, excluding the initial null entries
if the GLOB_DOOFFS flag is used (see gl_offs below).

@item gl_pathv
The address of the vector.  This field has type @w{@code{char **}}.

@item gl_offs
The offset of the first real element of the vector, from its nominal
address in the @code{gl_pathv} field.  Unlike the other fields, this
is always an input to @code{glob}, rather than an output from it.

If you use a nonzero offset, then that many elements at the beginning of
the vector are left empty.  (The @code{glob} function fills them with
null pointers.)

The @code{gl_offs} field is meaningful only if you use the
@code{GLOB_DOOFFS} flag.  Otherwise, the offset is always zero
regardless of what is in this field, and the first real element comes at
the beginning of the vector.

@item gl_closedir
The address of an alternative implementation of the @code{closedir}
function.  It is used if the @code{GLOB_ALTDIRFUNC} bit is set in
the flag parameter.  The type of this field is
@w{@code{void (*) (void *)}}.

This is a GNU extension.

@item gl_readdir
The address of an alternative implementation of the @code{readdir64}
function used to read the contents of a directory.  It is used if the
@code{GLOB_ALTDIRFUNC} bit is set in the flag parameter.  The type of
this field is @w{@code{struct dirent64 *(*) (void *)}}.

This is a GNU extension.

@item gl_opendir
The address of an alternative implementation of the @code{opendir}
function.  It is used if the @code{GLOB_ALTDIRFUNC} bit is set in
the flag parameter.  The type of this field is
@w{@code{void *(*) (const char *)}}.

This is a GNU extension.

@item gl_stat
The address of an alternative implementation of the @code{stat64} function
to get information about an object in the filesystem.  It is used if the
@code{GLOB_ALTDIRFUNC} bit is set in the flag parameter.  The type of
this field is @w{@code{int (*) (const char *, struct stat64 *)}}.

This is a GNU extension.

@item gl_lstat
The address of an alternative implementation of the @code{lstat64}
function to get information about an object in the filesystems, not
following symbolic links.  It is used if the @code{GLOB_ALTDIRFUNC} bit
is set in the flag parameter.  The type of this field is @code{@w{int
(*) (const char *,} @w{struct stat64 *)}}.

This is a GNU extension.

@item gl_flags
The flags used when @code{glob} was called.  In addition, @code{GLOB_MAGCHAR}
might be set.  See @ref{Flags for Globbing} for more details.

This is a GNU extension.
@end table
@end deftp

@comment glob.h
@comment POSIX.2
@deftypefun int glob (const char *@var{pattern}, int @var{flags}, int (*@var{errfunc}) (const char *@var{filename}, int @var{error-code}), glob_t *@var{vector-ptr})
The function @code{glob} does globbing using the pattern @var{pattern}
in the current directory.  It puts the result in a newly allocated
vector, and stores the size and address of this vector into
@code{*@var{vector-ptr}}.  The argument @var{flags} is a combination of
bit flags; see @ref{Flags for Globbing}, for details of the flags.

The result of globbing is a sequence of file names.  The function
@code{glob} allocates a string for each resulting word, then
allocates a vector of type @code{char **} to store the addresses of
these strings.  The last element of the vector is a null pointer.
This vector is called the @dfn{word vector}.

To return this vector, @code{glob} stores both its address and its
length (number of elements, not counting the terminating null pointer)
into @code{*@var{vector-ptr}}.

Normally, @code{glob} sorts the file names alphabetically before
returning them.  You can turn this off with the flag @code{GLOB_NOSORT}
if you want to get the information as fast as possible.  Usually it's
a good idea to let @code{glob} sort them---if you process the files in
alphabetical order, the users will have a feel for the rate of progress
that your application is making.

If @code{glob} succeeds, it returns 0.  Otherwise, it returns one
of these error codes:

@vtable @code
@comment glob.h
@comment POSIX.2
@item GLOB_ABORTED
There was an error opening a directory, and you used the flag
@code{GLOB_ERR} or your specified @var{errfunc} returned a nonzero
value.
@iftex
See below
@end iftex
@ifinfo
@xref{Flags for Globbing},
@end ifinfo
for an explanation of the @code{GLOB_ERR} flag and @var{errfunc}.

@comment glob.h
@comment POSIX.2
@item GLOB_NOMATCH
The pattern didn't match any existing files.  If you use the
@code{GLOB_NOCHECK} flag, then you never get this error code, because
that flag tells @code{glob} to @emph{pretend} that the pattern matched
at least one file.

@comment glob.h
@comment POSIX.2
@item GLOB_NOSPACE
It was impossible to allocate memory to hold the result.
@end vtable

In the event of an error, @code{glob} stores information in
@code{*@var{vector-ptr}} about all the matches it has found so far.

It is important to notice that the @code{glob} function will not fail if
it encounters directories or files which cannot be handled without the
LFS interfaces.  The implementation of @code{glob} is supposed to use
these functions internally.  This at least is the assumptions made by
the Unix standard.  The GNU extension of allowing the user to provide
own directory handling and @code{stat} functions complicates things a
bit.  If these callback functions are used and a large file or directory
is encountered @code{glob} @emph{can} fail.
@end deftypefun

@comment glob.h
@comment GNU
@deftypefun int glob64 (const char *@var{pattern}, int @var{flags}, int (*@var{errfunc}) (const char *@var{filename}, int @var{error-code}), glob64_t *@var{vector-ptr})
The @code{glob64} function was added as part of the Large File Summit
extensions but is not part of the original LFS proposal.  The reason for
this is simple: it is not necessary.  The necessity for a @code{glob64}
function is added by the extensions of the GNU @code{glob}
implementation which allows the user to provide own directory handling
and @code{stat} functions.  The @code{readdir} and @code{stat} functions
do depend on the choice of @code{_FILE_OFFSET_BITS} since the definition
of the types @code{struct dirent} and @code{struct stat} will change
depending on the choice.

Beside this difference the @code{glob64} works just like @code{glob} in
all aspects.

This function is a GNU extension.
@end deftypefun

@node Flags for Globbing
@subsection Flags for Globbing

This section describes the standard flags that you can specify in the
@var{flags} argument to @code{glob}.  Choose the flags you want,
and combine them with the C bitwise OR operator @code{|}.

Note that there are @ref{More Flags for Globbing} available as GNU extensions.

@vtable @code
@comment glob.h
@comment POSIX.2
@item GLOB_APPEND
Append the words from this expansion to the vector of words produced by
previous calls to @code{glob}.  This way you can effectively expand
several words as if they were concatenated with spaces between them.

In order for appending to work, you must not modify the contents of the
word vector structure between calls to @code{glob}.  And, if you set
@code{GLOB_DOOFFS} in the first call to @code{glob}, you must also
set it when you append to the results.

Note that the pointer stored in @code{gl_pathv} may no longer be valid
after you call @code{glob} the second time, because @code{glob} might
have relocated the vector.  So always fetch @code{gl_pathv} from the
@code{glob_t} structure after each @code{glob} call; @strong{never} save
the pointer across calls.

@comment glob.h
@comment POSIX.2
@item GLOB_DOOFFS
Leave blank slots at the beginning of the vector of words.
The @code{gl_offs} field says how many slots to leave.
The blank slots contain null pointers.

@comment glob.h
@comment POSIX.2
@item GLOB_ERR
Give up right away and report an error if there is any difficulty
reading the directories that must be read in order to expand @var{pattern}
fully.  Such difficulties might include a directory in which you don't
have the requisite access.  Normally, @code{glob} tries its best to keep
on going despite any errors, reading whatever directories it can.

You can exercise even more control than this by specifying an
error-handler function @var{errfunc} when you call @code{glob}.  If
@var{errfunc} is not a null pointer, then @code{glob} doesn't give up
right away when it can't read a directory; instead, it calls
@var{errfunc} with two arguments, like this:

@smallexample
(*@var{errfunc}) (@var{filename}, @var{error-code})
@end smallexample

@noindent
The argument @var{filename} is the name of the directory that
@code{glob} couldn't open or couldn't read, and @var{error-code} is the
@code{errno} value that was reported to @code{glob}.

If the error handler function returns nonzero, then @code{glob} gives up
right away.  Otherwise, it continues.

@comment glob.h
@comment POSIX.2
@item GLOB_MARK
If the pattern matches the name of a directory, append @samp{/} to the
directory's name when returning it.

@comment glob.h
@comment POSIX.2
@item GLOB_NOCHECK
If the pattern doesn't match any file names, return the pattern itself
as if it were a file name that had been matched.  (Normally, when the
pattern doesn't match anything, @code{glob} returns that there were no
matches.)

@comment glob.h
@comment POSIX.2
@item GLOB_NOESCAPE
Don't treat the @samp{\} character specially in patterns.  Normally,
@samp{\} quotes the following character, turning off its special meaning
(if any) so that it matches only itself.  When quoting is enabled, the
pattern @samp{\?} matches only the string @samp{?}, because the question
mark in the pattern acts like an ordinary character.

If you use @code{GLOB_NOESCAPE}, then @samp{\} is an ordinary character.

@code{glob} does its work by calling the function @code{fnmatch}
repeatedly.  It handles the flag @code{GLOB_NOESCAPE} by turning on the
@code{FNM_NOESCAPE} flag in calls to @code{fnmatch}.

@comment glob.h
@comment POSIX.2
@item GLOB_NOSORT
Don't sort the file names; return them in no particular order.
(In practice, the order will depend on the order of the entries in
the directory.)  The only reason @emph{not} to sort is to save time.
@end vtable

@node More Flags for Globbing
@subsection More Flags for Globbing

Beside the flags described in the last section, the GNU implementation of
@code{glob} allows a few more flags which are also defined in the
@file{glob.h} file.  Some of the extensions implement functionality
which is available in modern shell implementations.

@vtable @code
@comment glob.h
@comment GNU
@item GLOB_PERIOD
The @code{.} character (period) is treated special.  It cannot be
matched by wildcards.  @xref{Wildcard Matching}, @code{FNM_PERIOD}.

@comment glob.h
@comment GNU
@item GLOB_MAGCHAR
The @code{GLOB_MAGCHAR} value is not to be given to @code{glob} in the
@var{flags} parameter.  Instead, @code{glob} sets this bit in the
@var{gl_flags} element of the @var{glob_t} structure provided as the
result if the pattern used for matching contains any wildcard character.

@comment glob.h
@comment GNU
@item GLOB_ALTDIRFUNC
Instead of the using the using the normal functions for accessing the
filesystem the @code{glob} implementation uses the user-supplied
functions specified in the structure pointed to by @var{pglob}
parameter.  For more information about the functions refer to the
sections about directory handling see @ref{Accessing Directories}, and
@ref{Reading Attributes}.

@comment glob.h
@comment GNU
@item GLOB_BRACE
If this flag is given the handling of braces in the pattern is changed.
It is now required that braces appear correctly grouped.  I.e., for each
opening brace there must be a closing one.  Braces can be used
recursively.  So it is possible to define one brace expression in
another one.  It is important to note that the range of each brace
expression is completely contained in the outer brace expression (if
there is one).

The string between the matching braces is separated into single
expressions by splitting at @code{,} (comma) characters.  The commas
themselves are discarded.  Please note what we said above about recursive
brace expressions.  The commas used to separate the subexpressions must
be at the same level.  Commas in brace subexpressions are not matched.
They are used during expansion of the brace expression of the deeper
level.  The example below shows this

@smallexample
glob ("@{foo/@{,bar,biz@},baz@}", GLOB_BRACE, NULL, &result)
@end smallexample

@noindent
is equivalent to the sequence

@smallexample
glob ("foo/", GLOB_BRACE, NULL, &result)
glob ("foo/bar", GLOB_BRACE|GLOB_APPEND, NULL, &result)
glob ("foo/biz", GLOB_BRACE|GLOB_APPEND, NULL, &result)
glob ("baz", GLOB_BRACE|GLOB_APPEND, NULL, &result)
@end smallexample

@noindent
if we leave aside error handling.

@comment glob.h
@comment GNU
@item GLOB_NOMAGIC
If the pattern contains no wildcard constructs (it is a literal file name),
return it as the sole ``matching'' word, even if no file exists by that name.

@comment glob.h
@comment GNU
@item GLOB_TILDE
If this flag is used the character @code{~} (tilde) is handled special
if it appears at the beginning of the pattern.  Instead of being taken
verbatim it is used to represent the home directory of a known user.

If @code{~} is the only character in pattern or it is followed by a
@code{/} (slash), the home directory of the process owner is
substituted.  Using @code{getlogin} and @code{getpwnam} the information
is read from the system databases.  As an example take user @code{bart}
with his home directory at @file{/home/bart}.  For him a call like

@smallexample
glob ("~/bin/*", GLOB_TILDE, NULL, &result)
@end smallexample

@noindent
would return the contents of the directory @file{/home/bart/bin}.
Instead of referring to the own home directory it is also possible to
name the home directory of other users.  To do so one has to append the
user name after the tilde character.  So the contents of user
@code{homer}'s @file{bin} directory can be retrieved by

@smallexample
glob ("~homer/bin/*", GLOB_TILDE, NULL, &result)
@end smallexample

If the user name is not valid or the home directory cannot be determined
for some reason the pattern is left untouched and itself used as the
result.  I.e., if in the last example @code{home} is not available the
tilde expansion yields to @code{"~homer/bin/*"} and @code{glob} is not
looking for a directory named @code{~homer}.

This functionality is equivalent to what is available in C-shells if the
@code{nonomatch} flag is set.

@comment glob.h
@comment GNU
@item GLOB_TILDE_CHECK
If this flag is used @code{glob} behaves like as if @code{GLOB_TILDE} is
given.  The only difference is that if the user name is not available or
the home directory cannot be determined for other reasons this leads to
an error.  @code{glob} will return @code{GLOB_NOMATCH} instead of using
the pattern itself as the name.

This functionality is equivalent to what is available in C-shells if
@code{nonomatch} flag is not set.

@comment glob.h
@comment GNU
@item GLOB_ONLYDIR
If this flag is used the globbing function takes this as a
@strong{hint} that the caller is only interested in directories
matching the pattern.  If the information about the type of the file
is easily available non-directories will be rejected but no extra
work will be done to determine the information for each file.  I.e.,
the caller must still be able to filter directories out.

This functionality is only available with the GNU @code{glob}
implementation.  It is mainly used internally to increase the
performance but might be useful for a user as well and therefore is
documented here.
@end vtable

Calling @code{glob} will in most cases allocate resources which are used
to represent the result of the function call.  If the same object of
type @code{glob_t} is used in multiple call to @code{glob} the resources
are freed or reused so that no leaks appear.  But this does not include
the time when all @code{glob} calls are done.

@comment glob.h
@comment POSIX.2
@deftypefun void globfree (glob_t *@var{pglob})
The @code{globfree} function frees all resources allocated by previous
calls to @code{glob} associated with the object pointed to by
@var{pglob}.  This function should be called whenever the currently used
@code{glob_t} typed object isn't used anymore.
@end deftypefun

@comment glob.h
@comment GNU
@deftypefun void globfree64 (glob64_t *@var{pglob})
This function is equivalent to @code{globfree} but it frees records of
type @code{glob64_t} which were allocated by @code{glob64}.
@end deftypefun


@node Regular Expressions
@section Regular Expression Matching

@Theglibc{} supports two interfaces for matching regular
expressions.  One is the standard POSIX.2 interface, and the other is
what @theglibc{} has had for many years.

Both interfaces are declared in the header file @file{regex.h}.
If you define @w{@code{_POSIX_C_SOURCE}}, then only the POSIX.2
functions, structures, and constants are declared.
@c !!! we only document the POSIX.2 interface here!!

@menu
* POSIX Regexp Compilation::    Using @code{regcomp} to prepare to match.
* Flags for POSIX Regexps::     Syntax variations for @code{regcomp}.
* Matching POSIX Regexps::      Using @code{regexec} to match the compiled
                                   pattern that you get from @code{regcomp}.
* Regexp Subexpressions::       Finding which parts of the string were matched.
* Subexpression Complications:: Find points of which parts were matched.
* Regexp Cleanup::              Freeing storage; reporting errors.
@end menu

@node POSIX Regexp Compilation
@subsection POSIX Regular Expression Compilation

Before you can actually match a regular expression, you must
@dfn{compile} it.  This is not true compilation---it produces a special
data structure, not machine instructions.  But it is like ordinary
compilation in that its purpose is to enable you to ``execute'' the
pattern fast.  (@xref{Matching POSIX Regexps}, for how to use the
compiled regular expression for matching.)

There is a special data type for compiled regular expressions:

@comment regex.h
@comment POSIX.2
@deftp {Data Type} regex_t
This type of object holds a compiled regular expression.
It is actually a structure.  It has just one field that your programs
should look at:

@table @code
@item re_nsub
This field holds the number of parenthetical subexpressions in the
regular expression that was compiled.
@end table

There are several other fields, but we don't describe them here, because
only the functions in the library should use them.
@end deftp

After you create a @code{regex_t} object, you can compile a regular
expression into it by calling @code{regcomp}.

@comment regex.h
@comment POSIX.2
@deftypefun int regcomp (regex_t *restrict @var{compiled}, const char *restrict @var{pattern}, int @var{cflags})
The function @code{regcomp} ``compiles'' a regular expression into a
data structure that you can use with @code{regexec} to match against a
string.  The compiled regular expression format is designed for
efficient matching.  @code{regcomp} stores it into @code{*@var{compiled}}.

It's up to you to allocate an object of type @code{regex_t} and pass its
address to @code{regcomp}.

The argument @var{cflags} lets you specify various options that control
the syntax and semantics of regular expressions.  @xref{Flags for POSIX
Regexps}.

If you use the flag @code{REG_NOSUB}, then @code{regcomp} omits from
the compiled regular expression the information necessary to record
how subexpressions actually match.  In this case, you might as well
pass @code{0} for the @var{matchptr} and @var{nmatch} arguments when
you call @code{regexec}.

If you don't use @code{REG_NOSUB}, then the compiled regular expression
does have the capacity to record how subexpressions match.  Also,
@code{regcomp} tells you how many subexpressions @var{pattern} has, by
storing the number in @code{@var{compiled}->re_nsub}.  You can use that
value to decide how long an array to allocate to hold information about
subexpression matches.

@code{regcomp} returns @code{0} if it succeeds in compiling the regular
expression; otherwise, it returns a nonzero error code (see the table
below).  You can use @code{regerror} to produce an error message string
describing the reason for a nonzero value; see @ref{Regexp Cleanup}.

@end deftypefun

Here are the possible nonzero values that @code{regcomp} can return:

@table @code
@comment regex.h
@comment POSIX.2
@item REG_BADBR
There was an invalid @samp{\@{@dots{}\@}} construct in the regular
expression.  A valid @samp{\@{@dots{}\@}} construct must contain either
a single number, or two numbers in increasing order separated by a
comma.

@comment regex.h
@comment POSIX.2
@item REG_BADPAT
There was a syntax error in the regular expression.

@comment regex.h
@comment POSIX.2
@item REG_BADRPT
A repetition operator such as @samp{?} or @samp{*} appeared in a bad
position (with no preceding subexpression to act on).

@comment regex.h
@comment POSIX.2
@item REG_ECOLLATE
The regular expression referred to an invalid collating element (one not
defined in the current locale for string collation).  @xref{Locale
Categories}.

@comment regex.h
@comment POSIX.2
@item REG_ECTYPE
The regular expression referred to an invalid character class name.

@comment regex.h
@comment POSIX.2
@item REG_EESCAPE
The regular expression ended with @samp{\}.

@comment regex.h
@comment POSIX.2
@item REG_ESUBREG
There was an invalid number in the @samp{\@var{digit}} construct.

@comment regex.h
@comment POSIX.2
@item REG_EBRACK
There were unbalanced square brackets in the regular expression.

@comment regex.h
@comment POSIX.2
@item REG_EPAREN
An extended regular expression had unbalanced parentheses,
or a basic regular expression had unbalanced @samp{\(} and @samp{\)}.

@comment regex.h
@comment POSIX.2
@item REG_EBRACE
The regular expression had unbalanced @samp{\@{} and @samp{\@}}.

@comment regex.h
@comment POSIX.2
@item REG_ERANGE
One of the endpoints in a range expression was invalid.

@comment regex.h
@comment POSIX.2
@item REG_ESPACE
@code{regcomp} ran out of memory.
@end table

@node Flags for POSIX Regexps
@subsection Flags for POSIX Regular Expressions

These are the bit flags that you can use in the @var{cflags} operand when
compiling a regular expression with @code{regcomp}.

@table @code
@comment regex.h
@comment POSIX.2
@item REG_EXTENDED
Treat the pattern as an extended regular expression, rather than as a
basic regular expression.

@comment regex.h
@comment POSIX.2
@item REG_ICASE
Ignore case when matching letters.

@comment regex.h
@comment POSIX.2
@item REG_NOSUB
Don't bother storing the contents of the @var{matches-ptr} array.

@comment regex.h
@comment POSIX.2
@item REG_NEWLINE
Treat a newline in @var{string} as dividing @var{string} into multiple
lines, so that @samp{$} can match before the newline and @samp{^} can
match after.  Also, don't permit @samp{.} to match a newline, and don't
permit @samp{[^@dots{}]} to match a newline.

Otherwise, newline acts like any other ordinary character.
@end table

@node Matching POSIX Regexps
@subsection Matching a Compiled POSIX Regular Expression

Once you have compiled a regular expression, as described in @ref{POSIX
Regexp Compilation}, you can match it against strings using
@code{regexec}.  A match anywhere inside the string counts as success,
unless the regular expression contains anchor characters (@samp{^} or
@samp{$}).

@comment regex.h
@comment POSIX.2
@deftypefun int regexec (const regex_t *restrict @var{compiled}, const char *restrict @var{string}, size_t @var{nmatch}, regmatch_t @var{matchptr}[restrict], int @var{eflags})
This function tries to match the compiled regular expression
@code{*@var{compiled}} against @var{string}.

@code{regexec} returns @code{0} if the regular expression matches;
otherwise, it returns a nonzero value.  See the table below for
what nonzero values mean.  You can use @code{regerror} to produce an
error message string describing the reason for a nonzero value;
see @ref{Regexp Cleanup}.

The argument @var{eflags} is a word of bit flags that enable various
options.

If you want to get information about what part of @var{string} actually
matched the regular expression or its subexpressions, use the arguments
@var{matchptr} and @var{nmatch}.  Otherwise, pass @code{0} for
@var{nmatch}, and @code{NULL} for @var{matchptr}.  @xref{Regexp
Subexpressions}.
@end deftypefun

You must match the regular expression with the same set of current
locales that were in effect when you compiled the regular expression.

The function @code{regexec} accepts the following flags in the
@var{eflags} argument:

@table @code
@comment regex.h
@comment POSIX.2
@item REG_NOTBOL
Do not regard the beginning of the specified string as the beginning of
a line; more generally, don't make any assumptions about what text might
precede it.

@comment regex.h
@comment POSIX.2
@item REG_NOTEOL
Do not regard the end of the specified string as the end of a line; more
generally, don't make any assumptions about what text might follow it.
@end table

Here are the possible nonzero values that @code{regexec} can return:

@table @code
@comment regex.h
@comment POSIX.2
@item REG_NOMATCH
The pattern didn't match the string.  This isn't really an error.

@comment regex.h
@comment POSIX.2
@item REG_ESPACE
@code{regexec} ran out of memory.
@end table

@node Regexp Subexpressions
@subsection Match Results with Subexpressions

When @code{regexec} matches parenthetical subexpressions of
@var{pattern}, it records which parts of @var{string} they match.  It
returns that information by storing the offsets into an array whose
elements are structures of type @code{regmatch_t}.  The first element of
the array (index @code{0}) records the part of the string that matched
the entire regular expression.  Each other element of the array records
the beginning and end of the part that matched a single parenthetical
subexpression.

@comment regex.h
@comment POSIX.2
@deftp {Data Type} regmatch_t
This is the data type of the @var{matcharray} array that you pass to
@code{regexec}.  It contains two structure fields, as follows:

@table @code
@item rm_so
The offset in @var{string} of the beginning of a substring.  Add this
value to @var{string} to get the address of that part.

@item rm_eo
The offset in @var{string} of the end of the substring.
@end table
@end deftp

@comment regex.h
@comment POSIX.2
@deftp {Data Type} regoff_t
@code{regoff_t} is an alias for another signed integer type.
The fields of @code{regmatch_t} have type @code{regoff_t}.
@end deftp

The @code{regmatch_t} elements correspond to subexpressions
positionally; the first element (index @code{1}) records where the first
subexpression matched, the second element records the second
subexpression, and so on.  The order of the subexpressions is the order
in which they begin.

When you call @code{regexec}, you specify how long the @var{matchptr}
array is, with the @var{nmatch} argument.  This tells @code{regexec} how
many elements to store.  If the actual regular expression has more than
@var{nmatch} subexpressions, then you won't get offset information about
the rest of them.  But this doesn't alter whether the pattern matches a
particular string or not.

If you don't want @code{regexec} to return any information about where
the subexpressions matched, you can either supply @code{0} for
@var{nmatch}, or use the flag @code{REG_NOSUB} when you compile the
pattern with @code{regcomp}.

@node Subexpression Complications
@subsection Complications in Subexpression Matching

Sometimes a subexpression matches a substring of no characters.  This
happens when @samp{f\(o*\)} matches the string @samp{fum}.  (It really
matches just the @samp{f}.)  In this case, both of the offsets identify
the point in the string where the null substring was found.  In this
example, the offsets are both @code{1}.

Sometimes the entire regular expression can match without using some of
its subexpressions at all---for example, when @samp{ba\(na\)*} matches the
string @samp{ba}, the parenthetical subexpression is not used.  When
this happens, @code{regexec} stores @code{-1} in both fields of the
element for that subexpression.

Sometimes matching the entire regular expression can match a particular
subexpression more than once---for example, when @samp{ba\(na\)*}
matches the string @samp{bananana}, the parenthetical subexpression
matches three times.  When this happens, @code{regexec} usually stores
the offsets of the last part of the string that matched the
subexpression.  In the case of @samp{bananana}, these offsets are
@code{6} and @code{8}.

But the last match is not always the one that is chosen.  It's more
accurate to say that the last @emph{opportunity} to match is the one
that takes precedence.  What this means is that when one subexpression
appears within another, then the results reported for the inner
subexpression reflect whatever happened on the last match of the outer
subexpression.  For an example, consider @samp{\(ba\(na\)*s \)*} matching
the string @samp{bananas bas }.  The last time the inner expression
actually matches is near the end of the first word.  But it is
@emph{considered} again in the second word, and fails to match there.
@code{regexec} reports nonuse of the ``na'' subexpression.

Another place where this rule applies is when the regular expression
@smallexample
\(ba\(na\)*s \|nefer\(ti\)* \)*
@end smallexample
@noindent
matches @samp{bananas nefertiti}.  The ``na'' subexpression does match
in the first word, but it doesn't match in the second word because the
other alternative is used there.  Once again, the second repetition of
the outer subexpression overrides the first, and within that second
repetition, the ``na'' subexpression is not used.  So @code{regexec}
reports nonuse of the ``na'' subexpression.

@node Regexp Cleanup
@subsection POSIX Regexp Matching Cleanup

When you are finished using a compiled regular expression, you can
free the storage it uses by calling @code{regfree}.

@comment regex.h
@comment POSIX.2
@deftypefun void regfree (regex_t *@var{compiled})
Calling @code{regfree} frees all the storage that @code{*@var{compiled}}
points to.  This includes various internal fields of the @code{regex_t}
structure that aren't documented in this manual.

@code{regfree} does not free the object @code{*@var{compiled}} itself.
@end deftypefun

You should always free the space in a @code{regex_t} structure with
@code{regfree} before using the structure to compile another regular
expression.

When @code{regcomp} or @code{regexec} reports an error, you can use
the function @code{regerror} to turn it into an error message string.

@comment regex.h
@comment POSIX.2
@deftypefun size_t regerror (int @var{errcode}, const regex_t *restrict @var{compiled}, char *restrict @var{buffer}, size_t @var{length})
This function produces an error message string for the error code
@var{errcode}, and stores the string in @var{length} bytes of memory
starting at @var{buffer}.  For the @var{compiled} argument, supply the
same compiled regular expression structure that @code{regcomp} or
@code{regexec} was working with when it got the error.  Alternatively,
you can supply @code{NULL} for @var{compiled}; you will still get a
meaningful error message, but it might not be as detailed.

If the error message can't fit in @var{length} bytes (including a
terminating null character), then @code{regerror} truncates it.
The string that @code{regerror} stores is always null-terminated
even if it has been truncated.

The return value of @code{regerror} is the minimum length needed to
store the entire error message.  If this is less than @var{length}, then
the error message was not truncated, and you can use it.  Otherwise, you
should call @code{regerror} again with a larger buffer.

Here is a function which uses @code{regerror}, but always dynamically
allocates a buffer for the error message:

@smallexample
char *get_regerror (int errcode, regex_t *compiled)
@{
  size_t length = regerror (errcode, compiled, NULL, 0);
  char *buffer = xmalloc (length);
  (void) regerror (errcode, compiled, buffer, length);
  return buffer;
@}
@end smallexample
@end deftypefun

@node Word Expansion
@section Shell-Style Word Expansion
@cindex word expansion
@cindex expansion of shell words

@dfn{Word expansion} means the process of splitting a string into
@dfn{words} and substituting for variables, commands, and wildcards
just as the shell does.

For example, when you write @samp{ls -l foo.c}, this string is split
into three separate words---@samp{ls}, @samp{-l} and @samp{foo.c}.
This is the most basic function of word expansion.

When you write @samp{ls *.c}, this can become many words, because
the word @samp{*.c} can be replaced with any number of file names.
This is called @dfn{wildcard expansion}, and it is also a part of
word expansion.

When you use @samp{echo $PATH} to print your path, you are taking
advantage of @dfn{variable substitution}, which is also part of word
expansion.

Ordinary programs can perform word expansion just like the shell by
calling the library function @code{wordexp}.

@menu
* Expansion Stages::            What word expansion does to a string.
* Calling Wordexp::             How to call @code{wordexp}.
* Flags for Wordexp::           Options you can enable in @code{wordexp}.
* Wordexp Example::             A sample program that does word expansion.
* Tilde Expansion::             Details of how tilde expansion works.
* Variable Substitution::       Different types of variable substitution.
@end menu

@node Expansion Stages
@subsection The Stages of Word Expansion

When word expansion is applied to a sequence of words, it performs the
following transformations in the order shown here:

@enumerate
@item
@cindex tilde expansion
@dfn{Tilde expansion}: Replacement of @samp{~foo} with the name of
the home directory of @samp{foo}.

@item
Next, three different transformations are applied in the same step,
from left to right:

@itemize @bullet
@item
@cindex variable substitution
@cindex substitution of variables and commands
@dfn{Variable substitution}: Environment variables are substituted for
references such as @samp{$foo}.

@item
@cindex command substitution
@dfn{Command substitution}: Constructs such as @w{@samp{`cat foo`}} and
the equivalent @w{@samp{$(cat foo)}} are replaced with the output from
the inner command.

@item
@cindex arithmetic expansion
@dfn{Arithmetic expansion}: Constructs such as @samp{$(($x-1))} are
replaced with the result of the arithmetic computation.
@end itemize

@item
@cindex field splitting
@dfn{Field splitting}: subdivision of the text into @dfn{words}.

@item
@cindex wildcard expansion
@dfn{Wildcard expansion}: The replacement of a construct such as @samp{*.c}
with a list of @samp{.c} file names.  Wildcard expansion applies to an
entire word at a time, and replaces that word with 0 or more file names
that are themselves words.

@item
@cindex quote removal
@cindex removal of quotes
@dfn{Quote removal}: The deletion of string-quotes, now that they have
done their job by inhibiting the above transformations when appropriate.
@end enumerate

For the details of these transformations, and how to write the constructs
that use them, see @w{@cite{The BASH Manual}} (to appear).

@node Calling Wordexp
@subsection Calling @code{wordexp}

All the functions, constants and data types for word expansion are
declared in the header file @file{wordexp.h}.

Word expansion produces a vector of words (strings).  To return this
vector, @code{wordexp} uses a special data type, @code{wordexp_t}, which
is a structure.  You pass @code{wordexp} the address of the structure,
and it fills in the structure's fields to tell you about the results.

@comment wordexp.h
@comment POSIX.2
@deftp {Data Type} {wordexp_t}
This data type holds a pointer to a word vector.  More precisely, it
records both the address of the word vector and its size.

@table @code
@item we_wordc
The number of elements in the vector.

@item we_wordv
The address of the vector.  This field has type @w{@code{char **}}.

@item we_offs
The offset of the first real element of the vector, from its nominal
address in the @code{we_wordv} field.  Unlike the other fields, this
is always an input to @code{wordexp}, rather than an output from it.

If you use a nonzero offset, then that many elements at the beginning of
the vector are left empty.  (The @code{wordexp} function fills them with
null pointers.)

The @code{we_offs} field is meaningful only if you use the
@code{WRDE_DOOFFS} flag.  Otherwise, the offset is always zero
regardless of what is in this field, and the first real element comes at
the beginning of the vector.
@end table
@end deftp

@comment wordexp.h
@comment POSIX.2
@deftypefun int wordexp (const char *@var{words}, wordexp_t *@var{word-vector-ptr}, int @var{flags})
Perform word expansion on the string @var{words}, putting the result in
a newly allocated vector, and store the size and address of this vector
into @code{*@var{word-vector-ptr}}.  The argument @var{flags} is a
combination of bit flags; see @ref{Flags for Wordexp}, for details of
the flags.

You shouldn't use any of the characters @samp{|&;<>} in the string
@var{words} unless they are quoted; likewise for newline.  If you use
these characters unquoted, you will get the @code{WRDE_BADCHAR} error
code.  Don't use parentheses or braces unless they are quoted or part of
a word expansion construct.  If you use quotation characters @samp{'"`},
they should come in pairs that balance.

The results of word expansion are a sequence of words.  The function
@code{wordexp} allocates a string for each resulting word, then
allocates a vector of type @code{char **} to store the addresses of
these strings.  The last element of the vector is a null pointer.
This vector is called the @dfn{word vector}.

To return this vector, @code{wordexp} stores both its address and its
length (number of elements, not counting the terminating null pointer)
into @code{*@var{word-vector-ptr}}.

If @code{wordexp} succeeds, it returns 0.  Otherwise, it returns one
of these error codes:

@table @code
@comment wordexp.h
@comment POSIX.2
@item WRDE_BADCHAR
The input string @var{words} contains an unquoted invalid character such
as @samp{|}.

@comment wordexp.h
@comment POSIX.2
@item WRDE_BADVAL
The input string refers to an undefined shell variable, and you used the flag
@code{WRDE_UNDEF} to forbid such references.

@comment wordexp.h
@comment POSIX.2
@item WRDE_CMDSUB
The input string uses command substitution, and you used the flag
@code{WRDE_NOCMD} to forbid command substitution.

@comment wordexp.h
@comment POSIX.2
@item WRDE_NOSPACE
It was impossible to allocate memory to hold the result.  In this case,
@code{wordexp} can store part of the results---as much as it could
allocate room for.

@comment wordexp.h
@comment POSIX.2
@item WRDE_SYNTAX
There was a syntax error in the input string.  For example, an unmatched
quoting character is a syntax error.
@end table
@end deftypefun

@comment wordexp.h
@comment POSIX.2
@deftypefun void wordfree (wordexp_t *@var{word-vector-ptr})
Free the storage used for the word-strings and vector that
@code{*@var{word-vector-ptr}} points to.  This does not free the
structure @code{*@var{word-vector-ptr}} itself---only the other
data it points to.
@end deftypefun

@node Flags for Wordexp
@subsection Flags for Word Expansion

This section describes the flags that you can specify in the
@var{flags} argument to @code{wordexp}.  Choose the flags you want,
and combine them with the C operator @code{|}.

@table @code
@comment wordexp.h
@comment POSIX.2
@item WRDE_APPEND
Append the words from this expansion to the vector of words produced by
previous calls to @code{wordexp}.  This way you can effectively expand
several words as if they were concatenated with spaces between them.

In order for appending to work, you must not modify the contents of the
word vector structure between calls to @code{wordexp}.  And, if you set
@code{WRDE_DOOFFS} in the first call to @code{wordexp}, you must also
set it when you append to the results.

@comment wordexp.h
@comment POSIX.2
@item WRDE_DOOFFS
Leave blank slots at the beginning of the vector of words.
The @code{we_offs} field says how many slots to leave.
The blank slots contain null pointers.

@comment wordexp.h
@comment POSIX.2
@item WRDE_NOCMD
Don't do command substitution; if the input requests command substitution,
report an error.

@comment wordexp.h
@comment POSIX.2
@item WRDE_REUSE
Reuse a word vector made by a previous call to @code{wordexp}.
Instead of allocating a new vector of words, this call to @code{wordexp}
will use the vector that already exists (making it larger if necessary).

Note that the vector may move, so it is not safe to save an old pointer
and use it again after calling @code{wordexp}.  You must fetch
@code{we_pathv} anew after each call.

@comment wordexp.h
@comment POSIX.2
@item WRDE_SHOWERR
Do show any error messages printed by commands run by command substitution.
More precisely, allow these commands to inherit the standard error output
stream of the current process.  By default, @code{wordexp} gives these
commands a standard error stream that discards all output.

@comment wordexp.h
@comment POSIX.2
@item WRDE_UNDEF
If the input refers to a shell variable that is not defined, report an
error.
@end table

@node Wordexp Example
@subsection @code{wordexp} Example

Here is an example of using @code{wordexp} to expand several strings
and use the results to run a shell command.  It also shows the use of
@code{WRDE_APPEND} to concatenate the expansions and of @code{wordfree}
to free the space allocated by @code{wordexp}.

@smallexample
int
expand_and_execute (const char *program, const char **options)
@{
  wordexp_t result;
  pid_t pid
  int status, i;

  /* @r{Expand the string for the program to run.}  */
  switch (wordexp (program, &result, 0))
    @{
    case 0:                     /* @r{Successful}.  */
      break;
    case WRDE_NOSPACE:
      /* @r{If the error was @code{WRDE_NOSPACE},}
         @r{then perhaps part of the result was allocated.}  */
      wordfree (&result);
    default:                    /* @r{Some other error.}  */
      return -1;
    @}

  /* @r{Expand the strings specified for the arguments.}  */
  for (i = 0; options[i] != NULL; i++)
    @{
      if (wordexp (options[i], &result, WRDE_APPEND))
        @{
          wordfree (&result);
          return -1;
        @}
    @}

  pid = fork ();
  if (pid == 0)
    @{
      /* @r{This is the child process.  Execute the command.} */
      execv (result.we_wordv[0], result.we_wordv);
      exit (EXIT_FAILURE);
    @}
  else if (pid < 0)
    /* @r{The fork failed.  Report failure.}  */
    status = -1;
  else
    /* @r{This is the parent process.  Wait for the child to complete.}  */
    if (waitpid (pid, &status, 0) != pid)
      status = -1;

  wordfree (&result);
  return status;
@}
@end smallexample

@node Tilde Expansion
@subsection Details of Tilde Expansion

It's a standard part of shell syntax that you can use @samp{~} at the
beginning of a file name to stand for your own home directory.  You
can use @samp{~@var{user}} to stand for @var{user}'s home directory.

@dfn{Tilde expansion} is the process of converting these abbreviations
to the directory names that they stand for.

Tilde expansion applies to the @samp{~} plus all following characters up
to whitespace or a slash.  It takes place only at the beginning of a
word, and only if none of the characters to be transformed is quoted in
any way.

Plain @samp{~} uses the value of the environment variable @code{HOME}
as the proper home directory name.  @samp{~} followed by a user name
uses @code{getpwname} to look up that user in the user database, and
uses whatever directory is recorded there.  Thus, @samp{~} followed
by your own name can give different results from plain @samp{~}, if
the value of @code{HOME} is not really your home directory.

@node Variable Substitution
@subsection Details of Variable Substitution

Part of ordinary shell syntax is the use of @samp{$@var{variable}} to
substitute the value of a shell variable into a command.  This is called
@dfn{variable substitution}, and it is one part of doing word expansion.

There are two basic ways you can write a variable reference for
substitution:

@table @code
@item $@{@var{variable}@}
If you write braces around the variable name, then it is completely
unambiguous where the variable name ends.  You can concatenate
additional letters onto the end of the variable value by writing them
immediately after the close brace.  For example, @samp{$@{foo@}s}
expands into @samp{tractors}.

@item $@var{variable}
If you do not put braces around the variable name, then the variable
name consists of all the alphanumeric characters and underscores that
follow the @samp{$}.  The next punctuation character ends the variable
name.  Thus, @samp{$foo-bar} refers to the variable @code{foo} and expands
into @samp{tractor-bar}.
@end table

When you use braces, you can also use various constructs to modify the
value that is substituted, or test it in various ways.

@table @code
@item $@{@var{variable}:-@var{default}@}
Substitute the value of @var{variable}, but if that is empty or
undefined, use @var{default} instead.

@item $@{@var{variable}:=@var{default}@}
Substitute the value of @var{variable}, but if that is empty or
undefined, use @var{default} instead and set the variable to
@var{default}.

@item $@{@var{variable}:?@var{message}@}
If @var{variable} is defined and not empty, substitute its value.

Otherwise, print @var{message} as an error message on the standard error
stream, and consider word expansion a failure.

@c ??? How does wordexp report such an error?
@c WRDE_BADVAL is returned.

@item $@{@var{variable}:+@var{replacement}@}
Substitute @var{replacement}, but only if @var{variable} is defined and
nonempty.  Otherwise, substitute nothing for this construct.
@end table

@table @code
@item $@{#@var{variable}@}
Substitute a numeral which expresses in base ten the number of
characters in the value of @var{variable}.  @samp{$@{#foo@}} stands for
@samp{7}, because @samp{tractor} is seven characters.
@end table

These variants of variable substitution let you remove part of the
variable's value before substituting it.  The @var{prefix} and
@var{suffix} are not mere strings; they are wildcard patterns, just
like the patterns that you use to match multiple file names.  But
in this context, they match against parts of the variable value
rather than against file names.

@table @code
@item $@{@var{variable}%%@var{suffix}@}
Substitute the value of @var{variable}, but first discard from that
variable any portion at the end that matches the pattern @var{suffix}.

If there is more than one alternative for how to match against
@var{suffix}, this construct uses the longest possible match.

Thus, @samp{$@{foo%%r*@}} substitutes @samp{t}, because the largest
match for @samp{r*} at the end of @samp{tractor} is @samp{ractor}.

@item $@{@var{variable}%@var{suffix}@}
Substitute the value of @var{variable}, but first discard from that
variable any portion at the end that matches the pattern @var{suffix}.

If there is more than one alternative for how to match against
@var{suffix}, this construct uses the shortest possible alternative.

Thus, @samp{$@{foo%r*@}} substitutes @samp{tracto}, because the shortest
match for @samp{r*} at the end of @samp{tractor} is just @samp{r}.

@item $@{@var{variable}##@var{prefix}@}
Substitute the value of @var{variable}, but first discard from that
variable any portion at the beginning that matches the pattern @var{prefix}.

If there is more than one alternative for how to match against
@var{prefix}, this construct uses the longest possible match.

Thus, @samp{$@{foo##*t@}} substitutes @samp{or}, because the largest
match for @samp{*t} at the beginning of @samp{tractor} is @samp{tract}.

@item $@{@var{variable}#@var{prefix}@}
Substitute the value of @var{variable}, but first discard from that
variable any portion at the beginning that matches the pattern @var{prefix}.

If there is more than one alternative for how to match against
@var{prefix}, this construct uses the shortest possible alternative.

Thus, @samp{$@{foo#*t@}} substitutes @samp{ractor}, because the shortest
match for @samp{*t} at the beginning of @samp{tractor} is just @samp{t}.

@end table