2 * Wrc preprocessor lexical analysis
4 * Copyright 1999-2000 Bertho A. Stultiens (BS)
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21 * 24-Apr-2000 BS - Started from scratch to restructure everything
22 * and reintegrate the source into the wine-tree.
23 * 04-Jan-2000 BS - Added comments about the lexicographical
24 * grammar to give some insight in the complexity.
25 * 28-Dec-1999 BS - Eliminated backing-up of the flexer by running
26 * `flex -b' on the source. This results in some
27 * weirdo extra rules, but a much faster scanner.
28 * 23-Dec-1999 BS - Started this file
30 *-------------------------------------------------------------------------
31 * The preprocessor's lexographical grammar (approximately):
33 * pp := {ws} # {ws} if {ws} {expr} {ws} \n
34 * | {ws} # {ws} ifdef {ws} {id} {ws} \n
35 * | {ws} # {ws} ifndef {ws} {id} {ws} \n
36 * | {ws} # {ws} elif {ws} {expr} {ws} \n
37 * | {ws} # {ws} else {ws} \n
38 * | {ws} # {ws} endif {ws} \n
39 * | {ws} # {ws} include {ws} < {anytext} > \n
40 * | {ws} # {ws} include {ws} " {anytext} " \n
41 * | {ws} # {ws} define {ws} {anytext} \n
42 * | {ws} # {ws} define( {arglist} ) {ws} {expansion} \n
43 * | {ws} # {ws} pragma {ws} {anytext} \n
44 * | {ws} # {ws} ident {ws} {anytext} \n
45 * | {ws} # {ws} error {ws} {anytext} \n
46 * | {ws} # {ws} warning {ws} {anytext} \n
47 * | {ws} # {ws} line {ws} " {anytext} " {number} \n
48 * | {ws} # {ws} {number} " {anytext} " {number} [ {number} [{number}] ] \n
53 * expr := {expr} [+-*%^/|&] {expr}
54 * | {expr} {logor|logand} {expr}
56 * | {expr} ? {expr} : {expr}
62 * id := [a-zA-Z_][a-zA-Z0-9_]*
64 * anytext := [^\n]* (see note)
69 * | {arglist} , {id} ...
74 * | {anytext} ## {anytext}
78 * Note: "anytext" is not always "[^\n]*". This is because the
79 * trailing context must be considered as well.
81 * The only certain assumption for the preprocessor to make is that
82 * directives start at the beginning of the line, followed by a '#'
83 * and end with a newline.
84 * Any directive may be suffixed with a line-continuation. Also
85 * classical comment / *...* / (note: no comments within comments,
86 * therefore spaces) is considered to be a line-continuation
87 * (according to gcc and egcs AFAIK, ANSI is a bit vague).
88 * Comments have not been added to the above grammar for simplicity
89 * reasons. However, it is allowed to enter comment anywhere within
90 * the directives as long as they do not interfere with the context.
91 * All comments are considered to be deletable whitespace (both
92 * classical form "/ *...* /" and C++ form "//...\n").
94 * All recursive scans, except for macro-expansion, are done by the
95 * parser, whereas the simple state transitions of non-recursive
96 * directives are done in the scanner. This results in the many
97 * exclusive start-conditions of the scanner.
99 * Macro expansions are slightly more difficult because they have to
100 * prescan the arguments. Parameter substitution is literal if the
101 * substitution is # or ## (either side). This enables new identifiers
102 * to be created (see 'info cpp' node Macro|Pitfalls|Prescan for more
105 * FIXME: Variable macro parameters is recognized, but not yet
106 * expanded. I have to reread the ANSI standard on the subject (yes,
109 * The following special defines are supported:
110 * __FILE__ -> "thissource.c"
112 * __DATE__ -> "May 1 2000"
113 * __TIME__ -> "23:59:59"
114 * These macros expand, as expected, into their ANSI defined values.
116 * The same include prevention is implemented as gcc and egcs does.
117 * This results in faster processing because we do not read the text
118 * at all. Some wine-sources attempt to include the same file 4 or 5
119 * times. This strategy also saves a lot blank output-lines, which in
120 * its turn improves the real resource scanner/parser.
125 * Special flex options and exclusive scanner start-conditions
128 %option 8bit never-interactive
155 cident [a-zA-Z_][0-9a-zA-Z_]*
156 ul [uUlL]|[uUlL][lL]|[lL][uU]|[lL][lL][uU]|[uU][lL][lL]|[lL][uU][lL]
165 #include "wpp_private.h"
169 * Make sure that we are running an appropriate version of flex.
171 #if !defined(YY_FLEX_MAJOR_VERSION) || (1000 * YY_FLEX_MAJOR_VERSION + YY_FLEX_MINOR_VERSION < 2005)
172 #error Must use flex version 2.5.1 or higher (yy_scan_* routines are required).
175 #define YY_READ_BUF_SIZE 65536 /* So we read most of a file at once */
177 #define yy_current_state() YY_START
178 #define yy_pp_state(x) yy_pop_state(); yy_push_state(x)
181 * Always update the current character position within a line
183 #define YY_USER_ACTION pp_status.char_number+=ppleng;
186 * Buffer management for includes and expansions
188 #define MAXBUFFERSTACK 128 /* Nesting more than 128 includes or macro expansion textss is insane */
190 typedef struct bufferstackentry {
191 YY_BUFFER_STATE bufferstate; /* Buffer to switch back to */
192 pp_entry_t *define; /* Points to expanding define or NULL if handling includes */
193 int line_number; /* Line that we were handling */
194 int char_number; /* The current position on that line */
195 const char *filename; /* Filename that we were handling */
196 int if_depth; /* How many #if:s deep to check matching #endif:s */
197 int ncontinuations; /* Remember the continuation state */
198 int should_pop; /* Set if we must pop the start-state on EOF */
199 /* Include management */
200 include_state_t incl;
201 char *include_filename;
203 } bufferstackentry_t;
205 #define ALLOCBLOCKSIZE (1 << 10) /* Allocate these chunks at a time for string-buffers */
208 * Macro expansion nesting
209 * We need the stack to handle expansions while scanning
210 * a macro's arguments. The TOS must always be the macro
211 * that receives the current expansion from the scanner.
213 #define MAXMACEXPSTACK 128 /* Nesting more than 128 macro expansions is insane */
215 typedef struct macexpstackentry {
216 pp_entry_t *ppp; /* This macro we are scanning */
217 char **args; /* With these arguments */
218 char **ppargs; /* Resulting in these preprocessed arguments */
219 int *nnls; /* Number of newlines per argument */
220 int nargs; /* And this many arguments scanned */
221 int parentheses; /* Nesting level of () */
222 int curargsize; /* Current scanning argument's size */
223 int curargalloc; /* Current scanning argument's block allocated */
224 char *curarg; /* Current scanning argument's content */
225 } macexpstackentry_t;
227 #define MACROPARENTHESES() (top_macro()->parentheses)
232 static void newline(int);
233 static int make_number(int radix, YYSTYPE *val, const char *str, int len);
234 static void put_buffer(const char *s, int len);
235 static int is_c_h_include(char *fname, int quoted);
236 /* Buffer management */
237 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop);
238 static bufferstackentry_t *pop_buffer(void);
239 /* String functions */
240 static void new_string(void);
241 static void add_string(const char *str, int len);
242 static char *get_string(void);
243 static void put_string(void);
244 static int string_start(void);
245 /* Macro functions */
246 static void push_macro(pp_entry_t *ppp);
247 static macexpstackentry_t *top_macro(void);
248 static macexpstackentry_t *pop_macro(void);
249 static void free_macro(macexpstackentry_t *mep);
250 static void add_text_to_macro(const char *text, int len);
251 static void macro_add_arg(int last);
252 static void macro_add_expansion(void);
254 static void expand_special(pp_entry_t *ppp);
255 static void expand_define(pp_entry_t *ppp);
256 static void expand_macro(macexpstackentry_t *mep);
261 static int ncontinuations;
263 static int strbuf_idx = 0;
264 static int strbuf_alloc = 0;
265 static char *strbuffer = NULL;
266 static int str_startline;
268 static macexpstackentry_t *macexpstack[MAXMACEXPSTACK];
269 static int macexpstackidx = 0;
271 static bufferstackentry_t bufferstack[MAXBUFFERSTACK];
272 static int bufferstackidx = 0;
274 static int pass_data=1;
279 include_state_t pp_incl_state =
287 includelogicentry_t *pp_includelogiclist = NULL;
292 **************************************************************************
293 * The scanner starts here
294 **************************************************************************
299 * Catch line-continuations.
300 * Note: Gcc keeps the line-continuations in, for example, strings
301 * intact. However, I prefer to remove them all so that the next
302 * scanner will not need to reduce the continuation state.
304 * <*>\\\n newline(0);
308 * Detect the leading # of a preprocessor directive.
310 <INITIAL,pp_ignore>^{ws}*# pp_incl_state.seen_junk++; yy_push_state(pp_pp);
313 * Scan for the preprocessor directives
315 <pp_pp>{ws}*include{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_inc); return tINCLUDE;} else {yy_pp_state(pp_eol);}
316 <pp_pp>{ws}*define{ws}* yy_pp_state(yy_current_state() != pp_ignore ? pp_def : pp_eol);
317 <pp_pp>{ws}*error{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tERROR;
318 <pp_pp>{ws}*warning{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tWARNING;
319 <pp_pp>{ws}*pragma{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPRAGMA;
320 <pp_pp>{ws}*ident{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPPIDENT;
321 <pp_pp>{ws}*undef{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_ifd); return tUNDEF;} else {yy_pp_state(pp_eol);}
322 <pp_pp>{ws}*ifdef{ws}* yy_pp_state(pp_ifd); return tIFDEF;
323 <pp_pp>{ws}*ifndef{ws}* pp_incl_state.seen_junk--; yy_pp_state(pp_ifd); return tIFNDEF;
324 <pp_pp>{ws}*if{ws}* yy_pp_state(pp_if); return tIF;
325 <pp_pp>{ws}*elif{ws}* yy_pp_state(pp_if); return tELIF;
326 <pp_pp>{ws}*else{ws}* yy_pp_state(pp_endif); return tELSE;
327 <pp_pp>{ws}*endif{ws}* yy_pp_state(pp_endif); return tENDIF;
328 <pp_pp>{ws}*line{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tLINE;} else {yy_pp_state(pp_eol);}
329 <pp_pp>{ws}+ if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tGCCLINE;} else {yy_pp_state(pp_eol);}
330 <pp_pp>{ws}*[a-z]+ pperror("Invalid preprocessor token '%s'", pptext);
331 <pp_pp>\r?\n newline(1); yy_pop_state(); return tNL; /* This could be the null-token */
332 <pp_pp>\\\r?\n newline(0);
333 <pp_pp>\\\r? pperror("Preprocessor junk '%s'", pptext);
334 <pp_pp>. return *pptext;
337 * Handle #include and #line
339 <pp_line>[0-9]+ return make_number(10, &pplval, pptext, ppleng);
340 <pp_inc>\< new_string(); add_string(pptext, ppleng); yy_push_state(pp_iqs);
341 <pp_inc,pp_line>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
342 <pp_inc,pp_line>{ws}+ ;
343 <pp_inc,pp_line>\n newline(1); yy_pop_state(); return tNL;
344 <pp_inc,pp_line>\\\r?\n newline(0);
345 <pp_inc,pp_line>(\\\r?)|(.) pperror(yy_current_state() == pp_inc ? "Trailing junk in #include" : "Trailing junk in #line");
348 * Ignore all input when a false clause is parsed
350 <pp_ignore>[^#/\\\n]+ ;
351 <pp_ignore>\n newline(1);
352 <pp_ignore>\\\r?\n newline(0);
353 <pp_ignore>(\\\r?)|(.) ;
356 * Handle #if and #elif.
357 * These require conditionals to be evaluated, but we do not
358 * want to jam the scanner normally when we see these tokens.
359 * Note: tIDENT is handled below.
362 <pp_if>0[0-7]*{ul}? return make_number(8, &pplval, pptext, ppleng);
363 <pp_if>0[0-7]*[8-9]+{ul}? pperror("Invalid octal digit");
364 <pp_if>[1-9][0-9]*{ul}? return make_number(10, &pplval, pptext, ppleng);
365 <pp_if>0[xX][0-9a-fA-F]+{ul}? return make_number(16, &pplval, pptext, ppleng);
366 <pp_if>0[xX] pperror("Invalid hex number");
367 <pp_if>defined yy_push_state(pp_defined); return tDEFINED;
368 <pp_if>"<<" return tLSHIFT;
369 <pp_if>">>" return tRSHIFT;
370 <pp_if>"&&" return tLOGAND;
371 <pp_if>"||" return tLOGOR;
372 <pp_if>"==" return tEQ;
373 <pp_if>"!=" return tNE;
374 <pp_if>"<=" return tLTE;
375 <pp_if>">=" return tGTE;
376 <pp_if>\n newline(1); yy_pop_state(); return tNL;
377 <pp_if>\\\r?\n newline(0);
378 <pp_if>\\\r? pperror("Junk in conditional expression");
380 <pp_if>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
381 <pp_if>\" pperror("String constants not allowed in conditionals");
382 <pp_if>. return *pptext;
385 * Handle #ifdef, #ifndef and #undef
386 * to get only an untranslated/unexpanded identifier
388 <pp_ifd>{cident} pplval.cptr = pp_xstrdup(pptext); return tIDENT;
390 <pp_ifd>\n newline(1); yy_pop_state(); return tNL;
391 <pp_ifd>\\\r?\n newline(0);
392 <pp_ifd>(\\\r?)|(.) pperror("Identifier expected");
395 * Handle #else and #endif.
398 <pp_endif>\n newline(1); yy_pop_state(); return tNL;
399 <pp_endif>\\\r?\n newline(0);
400 <pp_endif>. pperror("Garbage after #else or #endif.");
403 * Handle the special 'defined' keyword.
404 * This is necessary to get the identifier prior to any
407 <pp_defined>{cident} yy_pop_state(); pplval.cptr = pp_xstrdup(pptext); return tIDENT;
409 <pp_defined>(\()|(\)) return *pptext;
410 <pp_defined>\\\r?\n newline(0);
411 <pp_defined>(\\.)|(\n)|(.) pperror("Identifier expected");
414 * Handle #error, #warning, #pragma and #ident.
415 * Pass everything literally to the parser, which
416 * will act appropriately.
417 * Comments are stripped from the literal text.
419 <pp_eol>[^/\\\n]+ if(yy_top_state() != pp_ignore) { pplval.cptr = pp_xstrdup(pptext); return tLITERAL; }
420 <pp_eol>\/[^/\\\n*]* if(yy_top_state() != pp_ignore) { pplval.cptr = pp_xstrdup(pptext); return tLITERAL; }
421 <pp_eol>(\\\r?)|(\/[^/*]) if(yy_top_state() != pp_ignore) { pplval.cptr = pp_xstrdup(pptext); return tLITERAL; }
422 <pp_eol>\n newline(1); yy_pop_state(); if(yy_current_state() != pp_ignore) { return tNL; }
423 <pp_eol>\\\r?\n newline(0);
426 * Handle left side of #define
428 <pp_def>{cident}\( pplval.cptr = pp_xstrdup(pptext); pplval.cptr[ppleng-1] = '\0'; yy_pp_state(pp_macro); return tMACRO;
429 <pp_def>{cident} pplval.cptr = pp_xstrdup(pptext); yy_pp_state(pp_define); return tDEFINE;
431 <pp_def>\\\r?\n newline(0);
432 <pp_def>(\\\r?)|(\n)|(.) perror("Identifier expected");
435 * Scan the substitution of a define
437 <pp_define>[^'"/\\\n]+ pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
438 <pp_define>(\\\r?)|(\/[^/*]) pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
439 <pp_define>\\\r?\n{ws}+ newline(0); pplval.cptr = pp_xstrdup(" "); return tLITERAL;
440 <pp_define>\\\r?\n newline(0);
441 <pp_define>\n newline(1); yy_pop_state(); return tNL;
442 <pp_define>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
443 <pp_define>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
446 * Scan the definition macro arguments
448 <pp_macro>\){ws}* yy_pp_state(pp_mbody); return tMACROEND;
450 <pp_macro>{cident} pplval.cptr = pp_xstrdup(pptext); return tIDENT;
451 <pp_macro>, return ',';
452 <pp_macro>"..." return tELIPSIS;
453 <pp_macro>(\\\r?)|(\n)|(.)|(\.\.?) pperror("Argument identifier expected");
454 <pp_macro>\\\r?\n newline(0);
457 * Scan the substitution of a macro
459 <pp_mbody>[^a-zA-Z0-9'"#/\\\n]+ pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
460 <pp_mbody>{cident} pplval.cptr = pp_xstrdup(pptext); return tIDENT;
461 <pp_mbody>\#\# return tCONCAT;
462 <pp_mbody>\# return tSTRINGIZE;
463 <pp_mbody>[0-9][^'"#/\\\n]* pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
464 <pp_mbody>(\\\r?)|(\/[^/*'"#\\\n]*) pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
465 <pp_mbody>\\\r?\n{ws}+ newline(0); pplval.cptr = pp_xstrdup(" "); return tLITERAL;
466 <pp_mbody>\\\r?\n newline(0);
467 <pp_mbody>\n newline(1); yy_pop_state(); return tNL;
468 <pp_mbody>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
469 <pp_mbody>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
472 * Macro expansion text scanning.
473 * This state is active just after the identifier is scanned
474 * that triggers an expansion. We *must* delete the leading
475 * whitespace before we can start scanning for arguments.
477 * If we do not see a '(' as next trailing token, then we have
478 * a false alarm. We just continue with a nose-bleed...
480 <pp_macign>{ws}*/\( yy_pp_state(pp_macscan);
482 if(yy_top_state() != pp_macscan)
485 <pp_macign>{ws}*\\\r?\n newline(0);
486 <pp_macign>{ws}+|{ws}*\\\r?|. {
487 macexpstackentry_t *mac = pop_macro();
489 put_buffer(mac->ppp->ident, strlen(mac->ppp->ident));
490 put_buffer(pptext, ppleng);
495 * Macro expansion argument text scanning.
496 * This state is active when a macro's arguments are being read for expansion.
499 if(++MACROPARENTHESES() > 1)
500 add_text_to_macro(pptext, ppleng);
503 if(--MACROPARENTHESES() == 0)
509 add_text_to_macro(pptext, ppleng);
512 if(MACROPARENTHESES() > 1)
513 add_text_to_macro(pptext, ppleng);
517 <pp_macscan>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
518 <pp_macscan>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
519 <pp_macscan>"/*" yy_push_state(pp_comment); add_text_to_macro(" ", 1);
520 <pp_macscan>\n pp_status.line_number++; pp_status.char_number = 1; add_text_to_macro(pptext, ppleng);
521 <pp_macscan>([^/(),\\\n"']+)|(\/[^/*(),\\\n'"]*)|(\\\r?)|(.) add_text_to_macro(pptext, ppleng);
522 <pp_macscan>\\\r?\n newline(0);
525 * Comment handling (almost all start-conditions)
527 <INITIAL,pp_pp,pp_ignore,pp_eol,pp_inc,pp_if,pp_ifd,pp_endif,pp_defined,pp_def,pp_define,pp_macro,pp_mbody,RCINCL>"/*" yy_push_state(pp_comment);
528 <pp_comment>[^*\n]*|"*"+[^*/\n]* ;
529 <pp_comment>\n newline(0);
530 <pp_comment>"*"+"/" yy_pop_state();
533 * Remove C++ style comment (almost all start-conditions)
535 <INITIAL,pp_pp,pp_ignore,pp_eol,pp_inc,pp_if,pp_ifd,pp_endif,pp_defined,pp_def,pp_define,pp_macro,pp_mbody,pp_macscan,RCINCL>"//"[^\n]* {
536 if(pptext[ppleng-1] == '\\')
537 ppwarning("C++ style comment ends with an escaped newline (escape ignored)");
541 * Single, double and <> quoted constants
543 <INITIAL,pp_macexp>\" pp_incl_state.seen_junk++; new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
544 <INITIAL,pp_macexp>\' pp_incl_state.seen_junk++; new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
545 <pp_dqs>[^"\\\n]+ add_string(pptext, ppleng);
547 add_string(pptext, ppleng);
549 switch(yy_current_state())
556 if (yy_current_state()==RCINCL) yy_pop_state();
557 pplval.cptr = get_string();
560 pplval.cptr = get_string();
561 if (is_c_h_include(pplval.cptr, 1)) pass_data=0;
568 <pp_sqs>[^'\\\n]+ add_string(pptext, ppleng);
570 add_string(pptext, ppleng);
572 switch(yy_current_state())
577 pplval.cptr = get_string();
583 <pp_iqs>[^\>\\\n]+ add_string(pptext, ppleng);
585 add_string(pptext, ppleng);
587 pplval.cptr = get_string();
592 * This is tricky; we need to remove the line-continuation
593 * from preprocessor strings, but OTOH retain them in all
594 * other strings. This is because the resource grammar is
595 * even more braindead than initially analysed and line-
596 * continuations in strings introduce, sigh, newlines in
597 * the output. There goes the concept of non-breaking, non-
598 * spacing whitespace.
600 switch(yy_top_state())
610 add_string(pptext, ppleng);
614 <pp_iqs,pp_dqs,pp_sqs>\\. add_string(pptext, ppleng);
615 <pp_iqs,pp_dqs,pp_sqs>\n {
617 add_string(pptext, ppleng);
618 ppwarning("Newline in string constant encounterd (started line %d)", string_start());
622 * Identifier scanning
624 <INITIAL,pp_if,pp_inc,pp_macexp>{cident} {
626 pp_incl_state.seen_junk++;
627 if(!(ppp = pplookup(pptext)))
629 if(yy_current_state() == pp_inc)
630 pperror("Expected include filename");
632 if(yy_current_state() == pp_if)
634 pplval.cptr = pp_xstrdup(pptext);
638 if((yy_current_state()==INITIAL) && (strcasecmp(pptext,"RCINCLUDE")==0)){
639 yy_push_state(RCINCL);
642 else put_buffer(pptext, ppleng);
645 else if(!ppp->expanding)
656 yy_push_state(pp_macign);
660 pp_internal_error(__FILE__, __LINE__, "Invalid define type %d\n", ppp->type);
666 * Everything else that needs to be passed and
667 * newline and continuation handling
669 <INITIAL,pp_macexp>[^a-zA-Z_#'"/\\\n \r\t\f\v]+|(\/|\\)[^a-zA-Z_/*'"\\\n \r\t\v\f]* pp_incl_state.seen_junk++; put_buffer(pptext, ppleng);
670 <INITIAL,pp_macexp>{ws}+ put_buffer(pptext, ppleng);
671 <INITIAL>\n newline(1);
672 <INITIAL>\\\r?\n newline(0);
673 <INITIAL>\\\r? pp_incl_state.seen_junk++; put_buffer(pptext, ppleng);
676 * Special catcher for macro argmument expansion to prevent
677 * newlines to propagate to the output or admin.
679 <pp_macexp>(\n)|(.)|(\\\r?(\n|.)) put_buffer(pptext, ppleng);
681 <RCINCL>[A-Za-z0-9_\.\\/]+ {
682 pplval.cptr=pp_xstrdup(pptext);
684 return tRCINCLUDEPATH;
690 new_string(); add_string(pptext,ppleng);yy_push_state(pp_dqs);
694 * This is a 'catch-all' rule to discover errors in the scanner
695 * in an orderly manner.
697 <*>. pp_incl_state.seen_junk++; ppwarning("Unmatched text '%c' (0x%02x); please report\n", isprint(*pptext & 0xff) ? *pptext : ' ', *pptext);
700 YY_BUFFER_STATE b = YY_CURRENT_BUFFER;
701 bufferstackentry_t *bep = pop_buffer();
703 if((!bep && pp_get_if_depth()) || (bep && pp_get_if_depth() != bep->if_depth))
704 ppwarning("Unmatched #if/#endif at end of file");
708 if(YY_START != INITIAL)
709 pperror("Unexpected end of file during preprocessing");
712 else if(bep->should_pop == 2)
714 macexpstackentry_t *mac;
723 **************************************************************************
725 **************************************************************************
737 *-------------------------------------------------------------------------
738 * Output newlines or set them as continuations
740 * Input: -1 - Don't count this one, but update local position (see pp_dqs)
741 * 0 - Line-continuation seen and cache output
742 * 1 - Newline seen and flush output
743 *-------------------------------------------------------------------------
745 static void newline(int dowrite)
747 pp_status.line_number++;
748 pp_status.char_number = 1;
756 for(;ncontinuations; ncontinuations--)
763 *-------------------------------------------------------------------------
764 * Make a number out of an any-base and suffixed string
766 * Possible number extensions:
769 * - "LL" long long int
771 * - "UL" unsigned long int
772 * - "ULL" unsigned long long int
773 * - "LU" unsigned long int
774 * - "LLU" unsigned long long int
778 * The sizes of resulting 'int' and 'long' are compiler specific.
779 * I depend on sizeof(int) > 2 here (although a relatively safe
781 * Long longs are not yet implemented because this is very compiler
782 * specific and I don't want to think too much about the problems.
784 *-------------------------------------------------------------------------
786 static int make_number(int radix, YYSTYPE *val, const char *str, int len)
794 ext[2] = toupper(str[len-1]);
795 ext[1] = len > 1 ? toupper(str[len-2]) : ' ';
796 ext[0] = len > 2 ? toupper(str[len-3]) : ' ';
798 if(!strcmp(ext, "LUL"))
799 pperror("Invalid constant suffix");
800 else if(!strcmp(ext, "LLU") || !strcmp(ext, "ULL"))
805 else if(!strcmp(ext+1, "LU") || !strcmp(ext+1, "UL"))
810 else if(!strcmp(ext+1, "LL"))
814 else if(!strcmp(ext+2, "L"))
818 else if(!strcmp(ext+2, "U"))
824 pp_internal_error(__FILE__, __LINE__, "long long constants not implemented yet");
828 val->ulong = strtoul(str, NULL, radix);
831 else if(!is_u && is_l)
833 val->slong = strtol(str, NULL, radix);
836 else if(is_u && !is_l)
838 val->uint = (unsigned int)strtoul(str, NULL, radix);
842 /* Else it must be an int... */
843 val->sint = (int)strtol(str, NULL, radix);
849 *-------------------------------------------------------------------------
850 * Macro and define expansion support
852 * FIXME: Variable macro arguments.
853 *-------------------------------------------------------------------------
855 static void expand_special(pp_entry_t *ppp)
857 const char *dbgtext = "?";
858 static char *buf = NULL;
860 assert(ppp->type == def_special);
862 if(!strcmp(ppp->ident, "__LINE__"))
864 dbgtext = "def_special(__LINE__)";
865 buf = pp_xrealloc(buf, 32);
866 sprintf(buf, "%d", pp_status.line_number);
868 else if(!strcmp(ppp->ident, "__FILE__"))
870 dbgtext = "def_special(__FILE__)";
871 buf = pp_xrealloc(buf, strlen(pp_status.input) + 3);
872 sprintf(buf, "\"%s\"", pp_status.input);
875 pp_internal_error(__FILE__, __LINE__, "Special macro '%s' not found...\n", ppp->ident);
878 fprintf(stderr, "expand_special(%d): %s:%d: '%s' -> '%s'\n",
881 pp_status.line_number,
887 push_buffer(ppp, NULL, NULL, 0);
892 static void expand_define(pp_entry_t *ppp)
894 assert(ppp->type == def_define);
897 fprintf(stderr, "expand_define(%d): %s:%d: '%s' -> '%s'\n",
900 pp_status.line_number,
903 if(ppp->subst.text && ppp->subst.text[0])
905 push_buffer(ppp, NULL, NULL, 0);
906 yy_scan_string(ppp->subst.text);
910 static int curdef_idx = 0;
911 static int curdef_alloc = 0;
912 static char *curdef_text = NULL;
914 static void add_text(const char *str, int len)
918 if(curdef_idx >= curdef_alloc || curdef_alloc - curdef_idx < len)
920 curdef_alloc += (len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1);
921 curdef_text = pp_xrealloc(curdef_text, curdef_alloc * sizeof(curdef_text[0]));
922 if(curdef_alloc > 65536)
923 ppwarning("Reallocating macro-expansion buffer larger than 64kB");
925 memcpy(&curdef_text[curdef_idx], str, len);
929 static mtext_t *add_expand_text(mtext_t *mtp, macexpstackentry_t *mep, int *nnl)
943 fprintf(stderr, "add_expand_text: exp_text: '%s'\n", mtp->subst.text);
944 add_text(mtp->subst.text, strlen(mtp->subst.text));
949 fprintf(stderr, "add_expand_text: exp_stringize(%d): '%s'\n",
951 mep->args[mtp->subst.argidx]);
952 cptr = mep->args[mtp->subst.argidx];
956 if(*cptr == '"' || *cptr == '\\')
966 fprintf(stderr, "add_expand_text: exp_concat\n");
967 /* Remove trailing whitespace from current expansion text */
970 if(isspace(curdef_text[curdef_idx-1] & 0xff))
975 /* tag current position and recursively expand the next part */
977 mtp = add_expand_text(mtp->next, mep, nnl);
979 /* Now get rid of the leading space of the expansion */
980 cptr = &curdef_text[tag];
981 n = curdef_idx - tag;
984 if(isspace(*cptr & 0xff))
992 if(cptr != &curdef_text[tag])
994 memmove(&curdef_text[tag], cptr, n);
995 curdef_idx -= (curdef_idx - tag) - n;
1000 if((mtp->next && mtp->next->type == exp_concat) || (mtp->prev && mtp->prev->type == exp_concat))
1001 exp = mep->args[mtp->subst.argidx];
1003 exp = mep->ppargs[mtp->subst.argidx];
1006 add_text(exp, strlen(exp));
1007 *nnl -= mep->nnls[mtp->subst.argidx];
1008 cptr = strchr(exp, '\n');
1012 cptr = strchr(cptr+1, '\n');
1014 mep->nnls[mtp->subst.argidx] = 0;
1017 fprintf(stderr, "add_expand_text: exp_subst(%d): '%s'\n", mtp->subst.argidx, exp);
1021 pp_internal_error(__FILE__, __LINE__, "Invalid expansion type (%d) in macro expansion\n", mtp->type);
1026 static void expand_macro(macexpstackentry_t *mep)
1032 pp_entry_t *ppp = mep->ppp;
1033 int nargs = mep->nargs;
1035 assert(ppp->type == def_macro);
1036 assert(ppp->expanding == 0);
1038 if((ppp->nargs >= 0 && nargs != ppp->nargs) || (ppp->nargs < 0 && nargs < -ppp->nargs))
1039 pperror("Too %s macro arguments (%d)", nargs < abs(ppp->nargs) ? "few" : "many", nargs);
1041 for(n = 0; n < nargs; n++)
1042 nnl += mep->nnls[n];
1045 fprintf(stderr, "expand_macro(%d): %s:%d: '%s'(%d,%d) -> ...\n",
1048 pp_status.line_number,
1055 for(mtp = ppp->subst.mtext; mtp; mtp = mtp->next)
1057 if(!(mtp = add_expand_text(mtp, mep, &nnl)))
1061 for(n = 0; n < nnl; n++)
1064 /* To make sure there is room and termination (see below) */
1067 /* Strip trailing whitespace from expansion */
1068 for(k = curdef_idx, cptr = &curdef_text[curdef_idx-1]; k > 0; k--, cptr--)
1070 if(!isspace(*cptr & 0xff))
1075 * We must add *one* whitespace to make sure that there
1076 * is a token-separation after the expansion.
1082 /* Strip leading whitespace from expansion */
1083 for(n = 0, cptr = curdef_text; n < k; n++, cptr++)
1085 if(!isspace(*cptr & 0xff))
1092 fprintf(stderr, "expand_text: '%s'\n", curdef_text + n);
1093 push_buffer(ppp, NULL, NULL, 0);
1094 /*yy_scan_bytes(curdef_text + n, k - n);*/
1095 yy_scan_string(curdef_text + n);
1100 *-------------------------------------------------------------------------
1101 * String collection routines
1102 *-------------------------------------------------------------------------
1104 static void new_string(void)
1108 ppwarning("new_string: strbuf_idx != 0");
1111 str_startline = pp_status.line_number;
1114 static void add_string(const char *str, int len)
1118 if(strbuf_idx >= strbuf_alloc || strbuf_alloc - strbuf_idx < len)
1120 strbuf_alloc += (len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1);
1121 strbuffer = pp_xrealloc(strbuffer, strbuf_alloc * sizeof(strbuffer[0]));
1122 if(strbuf_alloc > 65536)
1123 ppwarning("Reallocating string buffer larger than 64kB");
1125 memcpy(&strbuffer[strbuf_idx], str, len);
1129 static char *get_string(void)
1131 char *str = pp_xmalloc(strbuf_idx + 1);
1132 memcpy(str, strbuffer, strbuf_idx);
1133 str[strbuf_idx] = '\0';
1140 static void put_string(void)
1142 put_buffer(strbuffer, strbuf_idx);
1148 static int string_start(void)
1150 return str_startline;
1155 *-------------------------------------------------------------------------
1157 *-------------------------------------------------------------------------
1159 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop)
1162 printf("push_buffer(%d): %p %p %p %d\n", bufferstackidx, ppp, filename, incname, pop);
1163 if(bufferstackidx >= MAXBUFFERSTACK)
1164 pp_internal_error(__FILE__, __LINE__, "Buffer stack overflow");
1166 memset(&bufferstack[bufferstackidx], 0, sizeof(bufferstack[0]));
1167 bufferstack[bufferstackidx].bufferstate = YY_CURRENT_BUFFER;
1168 bufferstack[bufferstackidx].define = ppp;
1169 bufferstack[bufferstackidx].line_number = pp_status.line_number;
1170 bufferstack[bufferstackidx].char_number = pp_status.char_number;
1171 bufferstack[bufferstackidx].if_depth = pp_get_if_depth();
1172 bufferstack[bufferstackidx].should_pop = pop;
1173 bufferstack[bufferstackidx].filename = pp_status.input;
1174 bufferstack[bufferstackidx].ncontinuations = ncontinuations;
1175 bufferstack[bufferstackidx].incl = pp_incl_state;
1176 bufferstack[bufferstackidx].include_filename = incname;
1177 bufferstack[bufferstackidx].pass_data = pass_data;
1183 /* These will track the pperror to the correct file and line */
1184 pp_status.line_number = 1;
1185 pp_status.char_number = 1;
1186 pp_status.input = filename;
1190 pp_internal_error(__FILE__, __LINE__, "Pushing buffer without knowing where to go to");
1194 static bufferstackentry_t *pop_buffer(void)
1196 if(bufferstackidx < 0)
1197 pp_internal_error(__FILE__, __LINE__, "Bufferstack underflow?");
1199 if(bufferstackidx == 0)
1204 if(bufferstack[bufferstackidx].define)
1205 bufferstack[bufferstackidx].define->expanding = 0;
1208 pp_status.line_number = bufferstack[bufferstackidx].line_number;
1209 pp_status.char_number = bufferstack[bufferstackidx].char_number;
1210 pp_status.input = bufferstack[bufferstackidx].filename;
1211 ncontinuations = bufferstack[bufferstackidx].ncontinuations;
1212 if(!bufferstack[bufferstackidx].should_pop)
1215 fprintf(ppout, "# %d \"%s\" 2\n", pp_status.line_number, pp_status.input);
1217 /* We have EOF, check the include logic */
1218 if(pp_incl_state.state == 2 && !pp_incl_state.seen_junk && pp_incl_state.ppp)
1220 pp_entry_t *ppp = pplookup(pp_incl_state.ppp);
1223 includelogicentry_t *iep = pp_xmalloc(sizeof(includelogicentry_t));
1226 iep->filename = bufferstack[bufferstackidx].include_filename;
1228 iep->next = pp_includelogiclist;
1230 iep->next->prev = iep;
1231 pp_includelogiclist = iep;
1233 fprintf(stderr, "pop_buffer: %s:%d: includelogic added, include_ppp='%s', file='%s'\n", pp_status.input, pp_status.line_number, pp_incl_state.ppp, iep->filename);
1235 else if(bufferstack[bufferstackidx].include_filename)
1236 free(bufferstack[bufferstackidx].include_filename);
1238 if(pp_incl_state.ppp)
1239 free(pp_incl_state.ppp);
1240 pp_incl_state = bufferstack[bufferstackidx].incl;
1241 pass_data = bufferstack[bufferstackidx].pass_data;
1247 printf("pop_buffer(%d): %p %p (%d, %d, %d) %p %d\n",
1249 bufferstack[bufferstackidx].bufferstate,
1250 bufferstack[bufferstackidx].define,
1251 bufferstack[bufferstackidx].line_number,
1252 bufferstack[bufferstackidx].char_number,
1253 bufferstack[bufferstackidx].if_depth,
1254 bufferstack[bufferstackidx].filename,
1255 bufferstack[bufferstackidx].should_pop);
1257 pp_switch_to_buffer(bufferstack[bufferstackidx].bufferstate);
1259 if(bufferstack[bufferstackidx].should_pop)
1261 if(yy_current_state() == pp_macexp)
1262 macro_add_expansion();
1264 pp_internal_error(__FILE__, __LINE__, "Pop buffer and state without macro expansion state");
1268 return &bufferstack[bufferstackidx];
1273 *-------------------------------------------------------------------------
1274 * Macro nestng support
1275 *-------------------------------------------------------------------------
1277 static void push_macro(pp_entry_t *ppp)
1279 if(macexpstackidx >= MAXMACEXPSTACK)
1280 pperror("Too many nested macros");
1282 macexpstack[macexpstackidx] = pp_xmalloc(sizeof(macexpstack[0][0]));
1283 memset( macexpstack[macexpstackidx], 0, sizeof(macexpstack[0][0]));
1284 macexpstack[macexpstackidx]->ppp = ppp;
1288 static macexpstackentry_t *top_macro(void)
1290 return macexpstackidx > 0 ? macexpstack[macexpstackidx-1] : NULL;
1293 static macexpstackentry_t *pop_macro(void)
1295 if(macexpstackidx <= 0)
1296 pp_internal_error(__FILE__, __LINE__, "Macro expansion stack underflow\n");
1297 return macexpstack[--macexpstackidx];
1300 static void free_macro(macexpstackentry_t *mep)
1304 for(i = 0; i < mep->nargs; i++)
1315 static void add_text_to_macro(const char *text, int len)
1317 macexpstackentry_t *mep = top_macro();
1319 assert(mep->ppp->expanding == 0);
1321 if(mep->curargalloc - mep->curargsize <= len+1) /* +1 for '\0' */
1323 mep->curargalloc += (ALLOCBLOCKSIZE > len+1) ? ALLOCBLOCKSIZE : len+1;
1324 mep->curarg = pp_xrealloc(mep->curarg, mep->curargalloc * sizeof(mep->curarg[0]));
1326 memcpy(mep->curarg + mep->curargsize, text, len);
1327 mep->curargsize += len;
1328 mep->curarg[mep->curargsize] = '\0';
1331 static void macro_add_arg(int last)
1335 macexpstackentry_t *mep = top_macro();
1337 assert(mep->ppp->expanding == 0);
1339 mep->args = pp_xrealloc(mep->args, (mep->nargs+1) * sizeof(mep->args[0]));
1340 mep->ppargs = pp_xrealloc(mep->ppargs, (mep->nargs+1) * sizeof(mep->ppargs[0]));
1341 mep->nnls = pp_xrealloc(mep->nnls, (mep->nargs+1) * sizeof(mep->nnls[0]));
1342 mep->args[mep->nargs] = pp_xstrdup(mep->curarg ? mep->curarg : "");
1343 cptr = mep->args[mep->nargs]-1;
1344 while((cptr = strchr(cptr+1, '\n')))
1348 mep->nnls[mep->nargs] = nnl;
1351 mep->curargalloc = mep->curargsize = 0;
1355 fprintf(stderr, "macro_add_arg: %s:%d: %d -> '%s'\n",
1357 pp_status.line_number,
1359 mep->args[mep->nargs-1]);
1361 /* Each macro argument must be expanded to cope with stingize */
1362 if(last || mep->args[mep->nargs-1][0])
1364 yy_push_state(pp_macexp);
1365 push_buffer(NULL, NULL, NULL, last ? 2 : 1);
1366 yy_scan_string(mep->args[mep->nargs-1]);
1367 /*mep->bufferstackidx = bufferstackidx; But not nested! */
1371 static void macro_add_expansion(void)
1373 macexpstackentry_t *mep = top_macro();
1375 assert(mep->ppp->expanding == 0);
1377 mep->ppargs[mep->nargs-1] = pp_xstrdup(mep->curarg ? mep->curarg : "");
1379 mep->curargalloc = mep->curargsize = 0;
1383 fprintf(stderr, "macro_add_expansion: %s:%d: %d -> '%s'\n",
1385 pp_status.line_number,
1387 mep->ppargs[mep->nargs-1]);
1392 *-------------------------------------------------------------------------
1394 *-------------------------------------------------------------------------
1396 static void put_buffer(const char *s, int len)
1399 add_text_to_macro(s, len);
1402 fwrite(s, 1, len, ppout);
1408 *-------------------------------------------------------------------------
1409 * Include management
1410 *-------------------------------------------------------------------------
1412 static int is_c_h_include(char *fname, int quoted)
1414 int sl=strlen(fname);
1415 if (sl < 2 + 2 * quoted) return 0;
1416 if ((toupper(fname[sl-1-quoted])!='H') && (toupper(fname[sl-1-quoted])!='C')) return 0;
1417 if (fname[sl-2-quoted]!='.') return 0;
1421 void pp_do_include(char *fname, int type)
1425 includelogicentry_t *iep;
1427 for(iep = pp_includelogiclist; iep; iep = iep->next)
1429 if(!strcmp(iep->filename, fname))
1432 * We are done. The file was included before.
1433 * If the define was deleted, then this entry would have
1443 pperror("Empty include filename");
1445 /* Undo the effect of the quotation */
1448 if((ppin = pp_open_include(fname+1, type ? pp_status.input : NULL, &newpath)) == NULL)
1449 pperror("Unable to open include file %s", fname+1);
1451 fname[n-1] = *fname; /* Redo the quotes */
1452 push_buffer(NULL, newpath, fname, 0);
1453 pp_incl_state.seen_junk = 0;
1454 pp_incl_state.state = 0;
1455 pp_incl_state.ppp = NULL;
1456 if (is_c_h_include(newpath, 0)) pass_data=0;
1460 fprintf(stderr, "pp_do_include: %s:%d: include_state=%d, include_ppp='%s', include_ifdepth=%d ,pass_data=%d\n",
1461 pp_status.input, pp_status.line_number, pp_incl_state.state, pp_incl_state.ppp, pp_incl_state.ifdepth, pass_data);
1462 pp_switch_to_buffer(pp_create_buffer(ppin, YY_BUF_SIZE));
1464 fprintf(ppout, "# 1 \"%s\" 1%s\n", newpath, type ? "" : " 3");
1468 *-------------------------------------------------------------------------
1469 * Push/pop preprocessor ignore state when processing conditionals
1471 *-------------------------------------------------------------------------
1473 void pp_push_ignore_state(void)
1475 yy_push_state(pp_ignore);
1478 void pp_pop_ignore_state(void)