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 grammer 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 never-interactive
152 cident [a-zA-Z_][0-9a-zA-Z_]*
153 ul [uUlL]|[uUlL][lL]|[lL][uU]|[lL][lL][uU]|[uU][lL][lL]|[lL][uU][lL]
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_USE_PROTOS
177 #define YY_READ_BUF_SIZE 65536 /* So we read most of a file at once */
179 #define yy_current_state() YY_START
180 #define yy_pp_state(x) yy_pop_state(); yy_push_state(x)
183 * Always update the current character position within a line
185 #define YY_USER_ACTION char_number+=ppleng;
188 * Buffer management for includes and expansions
190 #define MAXBUFFERSTACK 128 /* Nesting more than 128 includes or macro expansion textss is insane */
192 typedef struct bufferstackentry {
193 YY_BUFFER_STATE bufferstate; /* Buffer to switch back to */
194 pp_entry_t *define; /* Points to expanding define or NULL if handling includes */
195 int line_number; /* Line that we were handling */
196 int char_number; /* The current position on that line */
197 char *filename; /* Filename that we were handling */
198 int if_depth; /* How many #if:s deep to check matching #endif:s */
199 int ncontinuations; /* Remember the continuation state */
200 int should_pop; /* Set if we must pop the start-state on EOF */
201 /* Include management */
204 char *include_filename;
208 } bufferstackentry_t;
210 #define ALLOCBLOCKSIZE (1 << 10) /* Allocate these chunks at a time for string-buffers */
213 * Macro expansion nesting
214 * We need the stack to handle expansions while scanning
215 * a macro's arguments. The TOS must always be the macro
216 * that receives the current expansion from the scanner.
218 #define MAXMACEXPSTACK 128 /* Nesting more than 128 macro expansions is insane */
220 typedef struct macexpstackentry {
221 pp_entry_t *ppp; /* This macro we are scanning */
222 char **args; /* With these arguments */
223 char **ppargs; /* Resulting in these preprocessed arguments */
224 int *nnls; /* Number of newlines per argument */
225 int nargs; /* And this many arguments scanned */
226 int parentheses; /* Nesting level of () */
227 int curargsize; /* Current scanning argument's size */
228 int curargalloc; /* Current scanning argument's block allocated */
229 char *curarg; /* Current scanning argument's content */
230 } macexpstackentry_t;
232 #define MACROPARENTHESES() (top_macro()->parentheses)
237 static void newline(int);
238 static int make_number(int radix, YYSTYPE *val, char *str, int len);
239 static void put_buffer(char *s, int len);
240 /* Buffer management */
241 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop);
242 static bufferstackentry_t *pop_buffer(void);
243 /* String functions */
244 static void new_string(void);
245 static void add_string(char *str, int len);
246 static char *get_string(void);
247 static void put_string(void);
248 static int string_start(void);
249 /* Macro functions */
250 static void push_macro(pp_entry_t *ppp);
251 static macexpstackentry_t *top_macro(void);
252 static macexpstackentry_t *pop_macro(void);
253 static void free_macro(macexpstackentry_t *mep);
254 static void add_text_to_macro(char *text, int len);
255 static void macro_add_arg(int last);
256 static void macro_add_expansion(void);
258 static void expand_special(pp_entry_t *ppp);
259 static void expand_define(pp_entry_t *ppp);
260 static void expand_macro(macexpstackentry_t *mep);
265 static int ncontinuations;
267 static int strbuf_idx = 0;
268 static int strbuf_alloc = 0;
269 static char *strbuffer = NULL;
270 static int str_startline;
272 static macexpstackentry_t *macexpstack[MAXMACEXPSTACK];
273 static int macexpstackidx = 0;
275 static bufferstackentry_t bufferstack[MAXBUFFERSTACK];
276 static int bufferstackidx = 0;
278 static int pass_data=1;
284 * Trace the include files to prevent double reading.
285 * This save 20..30% of processing time for most stuff
286 * that uses complex includes.
288 * -1 Don't track or seen junk
289 * 0 New include, waiting for "#ifndef __xxx_h"
290 * 1 Seen #ifndef, waiting for "#define __xxx_h ..."
291 * 2 Seen #endif, waiting for EOF
293 int include_state = -1;
294 char *include_ppp = NULL; /* The define to be set from the #ifndef */
295 int include_ifdepth = 0; /* The level of ifs at the #ifdef */
296 int seen_junk = 0; /* Set when junk is seen */
297 includelogicentry_t *includelogiclist = NULL;
302 **************************************************************************
303 * The scanner starts here
304 **************************************************************************
309 * Catch line-continuations.
310 * Note: Gcc keeps the line-continuations in, for example, strings
311 * intact. However, I prefer to remove them all so that the next
312 * scanner will not need to reduce the continuation state.
314 * <*>\\\n newline(0);
318 * Detect the leading # of a preprocessor directive.
320 <INITIAL,pp_ignore>^{ws}*# seen_junk++; yy_push_state(pp_pp);
323 * Scan for the preprocessor directives
325 <pp_pp>{ws}*include{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_inc); return tINCLUDE;} else {yy_pp_state(pp_eol);}
326 <pp_pp>{ws}*define{ws}* yy_pp_state(yy_current_state() != pp_ignore ? pp_def : pp_eol);
327 <pp_pp>{ws}*error{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tERROR;
328 <pp_pp>{ws}*warning{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tWARNING;
329 <pp_pp>{ws}*pragma{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPRAGMA;
330 <pp_pp>{ws}*ident{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPPIDENT;
331 <pp_pp>{ws}*undef{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_ifd); return tUNDEF;} else {yy_pp_state(pp_eol);}
332 <pp_pp>{ws}*ifdef{ws}* yy_pp_state(pp_ifd); return tIFDEF;
333 <pp_pp>{ws}*ifndef{ws}* seen_junk--; yy_pp_state(pp_ifd); return tIFNDEF;
334 <pp_pp>{ws}*if{ws}* yy_pp_state(pp_if); return tIF;
335 <pp_pp>{ws}*elif{ws}* yy_pp_state(pp_if); return tELIF;
336 <pp_pp>{ws}*else{ws}* return tELSE;
337 <pp_pp>{ws}*endif{ws}* return tENDIF;
338 <pp_pp>{ws}*line{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tLINE;} else {yy_pp_state(pp_eol);}
339 <pp_pp>{ws}+ if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tGCCLINE;} else {yy_pp_state(pp_eol);}
340 <pp_pp>{ws}*[a-z]+ pperror("Invalid preprocessor token '%s'", pptext);
341 <pp_pp>\r?\n newline(1); yy_pop_state(); return tNL; /* This could be the null-token */
342 <pp_pp>\\\r?\n newline(0);
343 <pp_pp>\\\r? pperror("Preprocessor junk '%s'", pptext);
344 <pp_pp>. return *pptext;
347 * Handle #include and #line
349 <pp_line>[0-9]+ return make_number(10, &pplval, pptext, ppleng);
350 <pp_inc>\< new_string(); add_string(pptext, ppleng); yy_push_state(pp_iqs);
351 <pp_inc,pp_line>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
352 <pp_inc,pp_line>{ws}+ ;
353 <pp_inc,pp_line>\n newline(1); yy_pop_state(); return tNL;
354 <pp_inc,pp_line>\\\r?\n newline(0);
355 <pp_inc,pp_line>(\\\r?)|(.) pperror(yy_current_state() == pp_inc ? "Trailing junk in #include" : "Trailing junk in #line");
358 * Ignore all input when a false clause is parsed
360 <pp_ignore>[^#/\\\n]+ ;
361 <pp_ignore>\n newline(1);
362 <pp_ignore>\\\r?\n newline(0);
363 <pp_ignore>(\\\r?)|(.) ;
366 * Handle #if and #elif.
367 * These require conditionals to be evaluated, but we do not
368 * want to jam the scanner normally when we see these tokens.
369 * Note: tIDENT is handled below.
372 <pp_if>0[0-7]*{ul}? return make_number(8, &pplval, pptext, ppleng);
373 <pp_if>0[0-7]*[8-9]+{ul}? pperror("Invalid octal digit");
374 <pp_if>[1-9][0-9]*{ul}? return make_number(10, &pplval, pptext, ppleng);
375 <pp_if>0[xX][0-9a-fA-F]+{ul}? return make_number(16, &pplval, pptext, ppleng);
376 <pp_if>0[xX] pperror("Invalid hex number");
377 <pp_if>defined yy_push_state(pp_defined); return tDEFINED;
378 <pp_if>"<<" return tLSHIFT;
379 <pp_if>">>" return tRSHIFT;
380 <pp_if>"&&" return tLOGAND;
381 <pp_if>"||" return tLOGOR;
382 <pp_if>"==" return tEQ;
383 <pp_if>"!=" return tNE;
384 <pp_if>"<=" return tLTE;
385 <pp_if>">=" return tGTE;
386 <pp_if>\n newline(1); yy_pop_state(); return tNL;
387 <pp_if>\\\r?\n newline(0);
388 <pp_if>\\\r? pperror("Junk in conditional expression");
390 <pp_if>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
391 <pp_if>\" pperror("String constants not allowed in conditionals");
392 <pp_if>. return *pptext;
395 * Handle #ifdef, #ifndef and #undef
396 * to get only an untranslated/unexpanded identifier
398 <pp_ifd>{cident} pplval.cptr = xstrdup(pptext); return tIDENT;
400 <pp_ifd>\n newline(1); yy_pop_state(); return tNL;
401 <pp_ifd>\\\r?\n newline(0);
402 <pp_ifd>(\\\r?)|(.) pperror("Identifier expected");
405 * Handle the special 'defined' keyword.
406 * This is necessary to get the identifier prior to any
409 <pp_defined>{cident} yy_pop_state(); pplval.cptr = xstrdup(pptext); return tIDENT;
411 <pp_defined>(\()|(\)) return *pptext;
412 <pp_defined>\\\r?\n newline(0);
413 <pp_defined>(\\.)|(\n)|(.) pperror("Identifier expected");
416 * Handle #error, #warning, #pragma and #ident.
417 * Pass everything literally to the parser, which
418 * will act appropriately.
419 * Comments are stripped from the literal text.
421 <pp_eol>[^/\\\n]+ if(yy_top_state() != pp_ignore) { pplval.cptr = xstrdup(pptext); return tLITERAL; }
422 <pp_eol>\/[^/\\\n*]* if(yy_top_state() != pp_ignore) { pplval.cptr = xstrdup(pptext); return tLITERAL; }
423 <pp_eol>(\\\r?)|(\/[^/*]) if(yy_top_state() != pp_ignore) { pplval.cptr = xstrdup(pptext); return tLITERAL; }
424 <pp_eol>\n newline(1); yy_pop_state(); if(yy_current_state() != pp_ignore) { return tNL; }
425 <pp_eol>\\\r?\n newline(0);
428 * Handle left side of #define
430 <pp_def>{cident}\( pplval.cptr = xstrdup(pptext); pplval.cptr[ppleng-1] = '\0'; yy_pp_state(pp_macro); return tMACRO;
431 <pp_def>{cident} pplval.cptr = xstrdup(pptext); yy_pp_state(pp_define); return tDEFINE;
433 <pp_def>\\\r?\n newline(0);
434 <pp_def>(\\\r?)|(\n)|(.) perror("Identifier expected");
437 * Scan the substitution of a define
439 <pp_define>[^'"/\\\n]+ pplval.cptr = xstrdup(pptext); return tLITERAL;
440 <pp_define>(\\\r?)|(\/[^/*]) pplval.cptr = xstrdup(pptext); return tLITERAL;
441 <pp_define>\\\r?\n{ws}+ newline(0); pplval.cptr = xstrdup(" "); return tLITERAL;
442 <pp_define>\\\r?\n newline(0);
443 <pp_define>\n newline(1); yy_pop_state(); return tNL;
444 <pp_define>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
445 <pp_define>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
448 * Scan the definition macro arguments
450 <pp_macro>\){ws}* yy_pp_state(pp_mbody); return tMACROEND;
452 <pp_macro>{cident} pplval.cptr = xstrdup(pptext); return tIDENT;
453 <pp_macro>, return ',';
454 <pp_macro>"..." return tELIPSIS;
455 <pp_macro>(\\\r?)|(\n)|(.)|(\.\.?) pperror("Argument identifier expected");
456 <pp_macro>\\\r?\n newline(0);
459 * Scan the substitution of a macro
461 <pp_mbody>[^a-zA-Z0-9'"#/\\\n]+ pplval.cptr = xstrdup(pptext); return tLITERAL;
462 <pp_mbody>{cident} pplval.cptr = xstrdup(pptext); return tIDENT;
463 <pp_mbody>\#\# return tCONCAT;
464 <pp_mbody>\# return tSTRINGIZE;
465 <pp_mbody>[0-9][^'"#/\\\n]* pplval.cptr = xstrdup(pptext); return tLITERAL;
466 <pp_mbody>(\\\r?)|(\/[^/*'"#\\\n]*) pplval.cptr = xstrdup(pptext); return tLITERAL;
467 <pp_mbody>\\\r?\n{ws}+ newline(0); pplval.cptr = xstrdup(" "); return tLITERAL;
468 <pp_mbody>\\\r?\n newline(0);
469 <pp_mbody>\n newline(1); yy_pop_state(); return tNL;
470 <pp_mbody>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
471 <pp_mbody>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
474 * Macro expansion text scanning.
475 * This state is active just after the identifier is scanned
476 * that triggers an expansion. We *must* delete the leading
477 * whitespace before we can start scanning for arguments.
479 * If we do not see a '(' as next trailing token, then we have
480 * a false alarm. We just continue with a nose-bleed...
482 <pp_macign>{ws}*/\( yy_pp_state(pp_macscan);
484 if(yy_top_state() != pp_macscan)
487 <pp_macign>{ws}*\\\r?\n newline(0);
488 <pp_macign>{ws}+|{ws}*\\\r?|. {
489 macexpstackentry_t *mac = pop_macro();
491 put_buffer(mac->ppp->ident, strlen(mac->ppp->ident));
492 put_buffer(pptext, ppleng);
497 * Macro expansion argument text scanning.
498 * This state is active when a macro's arguments are being read for expansion.
501 if(++MACROPARENTHESES() > 1)
502 add_text_to_macro(pptext, ppleng);
505 if(--MACROPARENTHESES() == 0)
511 add_text_to_macro(pptext, ppleng);
514 if(MACROPARENTHESES() > 1)
515 add_text_to_macro(pptext, ppleng);
519 <pp_macscan>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
520 <pp_macscan>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
521 <pp_macscan>"/*" yy_push_state(pp_comment); add_text_to_macro(" ", 1);
522 <pp_macscan>\n line_number++; char_number = 1; add_text_to_macro(pptext, ppleng);
523 <pp_macscan>([^/(),\\\n"']+)|(\/[^/*(),\\\n'"]*)|(\\\r?)|(.) add_text_to_macro(pptext, ppleng);
524 <pp_macscan>\\\r?\n newline(0);
527 * Comment handling (almost all start-conditions)
529 <INITIAL,pp_pp,pp_ignore,pp_eol,pp_inc,pp_if,pp_ifd,pp_defined,pp_def,pp_define,pp_macro,pp_mbody,RCINCL>"/*" yy_push_state(pp_comment);
530 <pp_comment>[^*\n]*|"*"+[^*/\n]* ;
531 <pp_comment>\n newline(0);
532 <pp_comment>"*"+"/" yy_pop_state();
535 * Remove C++ style comment (almost all start-conditions)
537 <INITIAL,pp_pp,pp_ignore,pp_eol,pp_inc,pp_if,pp_ifd,pp_defined,pp_def,pp_define,pp_macro,pp_mbody,pp_macscan,RCINCL>"//"[^\n]* {
538 if(pptext[ppleng-1] == '\\')
539 ppwarning("C++ style comment ends with an escaped newline (escape ignored)");
543 * Single, double and <> quoted constants
545 <INITIAL,pp_macexp>\" seen_junk++; new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
546 <INITIAL,pp_macexp>\' seen_junk++; new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
547 <pp_dqs>[^"\\\n]+ add_string(pptext, ppleng);
549 add_string(pptext, ppleng);
551 switch(yy_current_state())
559 if (yy_current_state()==RCINCL) yy_pop_state();
560 pplval.cptr = get_string();
566 <pp_sqs>[^'\\\n]+ add_string(pptext, ppleng);
568 add_string(pptext, ppleng);
570 switch(yy_current_state())
575 pplval.cptr = get_string();
581 <pp_iqs>[^\>\\\n]+ add_string(pptext, ppleng);
583 add_string(pptext, ppleng);
585 pplval.cptr = get_string();
590 * This is tricky; we need to remove the line-continuation
591 * from preprocessor strings, but OTOH retain them in all
592 * other strings. This is because the resource grammar is
593 * even more braindead than initially analysed and line-
594 * continuations in strings introduce, sigh, newlines in
595 * the output. There goes the concept of non-breaking, non-
596 * spacing whitespace.
598 switch(yy_top_state())
608 add_string(pptext, ppleng);
612 <pp_iqs,pp_dqs,pp_sqs>\\. add_string(pptext, ppleng);
613 <pp_iqs,pp_dqs,pp_sqs>\n {
615 add_string(pptext, ppleng);
616 ppwarning("Newline in string constant encounterd (started line %d)", string_start());
620 * Identifier scanning
622 <INITIAL,pp_if,pp_inc,pp_macexp>{cident} {
625 if(!(ppp = pplookup(pptext)))
627 if(yy_current_state() == pp_inc)
628 pperror("Expected include filename");
630 if(yy_current_state() == pp_if)
632 pplval.cptr = xstrdup(pptext);
636 if((yy_current_state()==INITIAL) && (strcasecmp(pptext,"RCINCLUDE")==0)){
637 yy_push_state(RCINCL);
640 else put_buffer(pptext, ppleng);
643 else if(!ppp->expanding)
654 yy_push_state(pp_macign);
658 internal_error(__FILE__, __LINE__, "Invalid define type %d\n", ppp->type);
664 * Everything else that needs to be passed and
665 * newline and continuation handling
667 <INITIAL,pp_macexp>[^a-zA-Z_#'"/\\\n \r\t\f\v]+|(\/|\\)[^a-zA-Z_/*'"\\\n \r\t\v\f]* seen_junk++; put_buffer(pptext, ppleng);
668 <INITIAL,pp_macexp>{ws}+ put_buffer(pptext, ppleng);
669 <INITIAL>\n newline(1);
670 <INITIAL>\\\r?\n newline(0);
671 <INITIAL>\\\r? seen_junk++; put_buffer(pptext, ppleng);
674 * Special catcher for macro argmument expansion to prevent
675 * newlines to propagate to the output or admin.
677 <pp_macexp>(\n)|(.)|(\\\r?(\n|.)) put_buffer(pptext, ppleng);
679 <RCINCL>[A-Za-z0-9_\.\\/]+ {
680 pplval.cptr=xstrdup(pptext);
682 return tRCINCLUDEPATH;
688 new_string(); add_string(pptext,ppleng);yy_push_state(pp_dqs);
692 * This is a 'catch-all' rule to discover errors in the scanner
693 * in an orderly manner.
695 <*>. seen_junk++; ppwarning("Unmatched text '%c' (0x%02x); please report\n", isprint(*pptext & 0xff) ? *pptext : ' ', *pptext);
698 YY_BUFFER_STATE b = YY_CURRENT_BUFFER;
699 bufferstackentry_t *bep = pop_buffer();
701 if((!bep && get_if_depth()) || (bep && get_if_depth() != bep->if_depth))
702 ppwarning("Unmatched #if/#endif at end of file");
706 if(YY_START != INITIAL)
707 pperror("Unexpected end of file during preprocessing");
710 else if(bep->should_pop == 2)
712 macexpstackentry_t *mac;
721 **************************************************************************
723 **************************************************************************
735 *-------------------------------------------------------------------------
736 * Output newlines or set them as continuations
738 * Input: -1 - Don't count this one, but update local position (see pp_dqs)
739 * 0 - Line-continuation seen and cache output
740 * 1 - Newline seen and flush output
741 *-------------------------------------------------------------------------
743 static void newline(int dowrite)
754 for(;ncontinuations; ncontinuations--)
761 *-------------------------------------------------------------------------
762 * Make a number out of an any-base and suffixed string
764 * Possible number extensions:
767 * - "LL" long long int
769 * - "UL" unsigned long int
770 * - "ULL" unsigned long long int
771 * - "LU" unsigned long int
772 * - "LLU" unsigned long long int
776 * The sizes of resulting 'int' and 'long' are compiler specific.
777 * I depend on sizeof(int) > 2 here (although a relatively safe
779 * Long longs are not yet implemented because this is very compiler
780 * specific and I don't want to think too much about the problems.
782 *-------------------------------------------------------------------------
784 static int make_number(int radix, YYSTYPE *val, char *str, int len)
792 ext[2] = toupper(str[len-1]);
793 ext[1] = len > 1 ? toupper(str[len-2]) : ' ';
794 ext[0] = len > 2 ? toupper(str[len-3]) : ' ';
796 if(!strcmp(ext, "LUL"))
797 pperror("Invalid constant suffix");
798 else if(!strcmp(ext, "LLU") || !strcmp(ext, "ULL"))
803 else if(!strcmp(ext+1, "LU") || !strcmp(ext+1, "UL"))
808 else if(!strcmp(ext+1, "LL"))
812 else if(!strcmp(ext+2, "L"))
816 else if(!strcmp(ext+2, "U"))
822 internal_error(__FILE__, __LINE__, "long long constants not implemented yet");
826 val->ulong = strtoul(str, NULL, radix);
829 else if(!is_u && is_l)
831 val->slong = strtol(str, NULL, radix);
834 else if(is_u && !is_l)
836 val->uint = (unsigned int)strtoul(str, NULL, radix);
837 if(!win32 && val->uint > 65535)
839 pperror("Constant overflow");
844 /* Else it must be an int... */
845 val->sint = (int)strtol(str, NULL, radix);
846 if(!win32 && (val->sint < -32768 || val->sint > 32768))
849 * Note: test must be > 32768 because unary minus
850 * is handled as an expression! This can result in
851 * failure and must be checked in the parser.
853 pperror("Constant overflow");
860 *-------------------------------------------------------------------------
861 * Macro and define expansion support
863 * FIXME: Variable macro arguments.
864 *-------------------------------------------------------------------------
866 static void expand_special(pp_entry_t *ppp)
869 static char *buf = NULL;
871 assert(ppp->type == def_special);
873 if(!strcmp(ppp->ident, "__LINE__"))
875 dbgtext = "def_special(__LINE__)";
876 buf = xrealloc(buf, 32);
877 sprintf(buf, "%d", line_number);
879 else if(!strcmp(ppp->ident, "__FILE__"))
881 dbgtext = "def_special(__FILE__)";
882 buf = xrealloc(buf, strlen(input_name) + 3);
883 sprintf(buf, "\"%s\"", input_name);
885 else if(!strcmp(ppp->ident, "__DATE__"))
887 dbgtext = "def_special(__DATE__)";
888 buf = xrealloc(buf, 32);
889 strftime(buf, 32, "\"%b %d %Y\"", localtime(&now));
891 else if(!strcmp(ppp->ident, "__TIME__"))
893 dbgtext = "def_special(__TIME__)";
894 buf = xrealloc(buf, 32);
895 strftime(buf, 32, "\"%H:%M:%S\"", localtime(&now));
898 internal_error(__FILE__, __LINE__, "Special macro '%s' not found...\n", ppp->ident);
900 if(debuglevel & DEBUGLEVEL_PPLEX)
901 fprintf(stderr, "expand_special(%d): %s:%d: '%s' -> '%s'\n",
910 push_buffer(ppp, NULL, NULL, 0);
915 static void expand_define(pp_entry_t *ppp)
917 assert(ppp->type == def_define);
919 if(debuglevel & DEBUGLEVEL_PPLEX)
920 fprintf(stderr, "expand_define(%d): %s:%d: '%s' -> '%s'\n",
926 if(ppp->subst.text && ppp->subst.text[0])
928 push_buffer(ppp, NULL, NULL, 0);
929 yy_scan_string(ppp->subst.text);
933 static int curdef_idx = 0;
934 static int curdef_alloc = 0;
935 static char *curdef_text = NULL;
937 static void add_text(char *str, int len)
941 if(curdef_idx >= curdef_alloc || curdef_alloc - curdef_idx < len)
943 curdef_alloc += (len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1);
944 curdef_text = xrealloc(curdef_text, curdef_alloc * sizeof(curdef_text[0]));
945 if(curdef_alloc > 65536)
946 ppwarning("Reallocating macro-expansion buffer larger than 64kB");
948 memcpy(&curdef_text[curdef_idx], str, len);
952 static mtext_t *add_expand_text(mtext_t *mtp, macexpstackentry_t *mep, int *nnl)
965 if(debuglevel & DEBUGLEVEL_PPLEX)
966 fprintf(stderr, "add_expand_text: exp_text: '%s'\n", mtp->subst.text);
967 add_text(mtp->subst.text, strlen(mtp->subst.text));
971 if(debuglevel & DEBUGLEVEL_PPLEX)
972 fprintf(stderr, "add_expand_text: exp_stringize(%d): '%s'\n",
974 mep->args[mtp->subst.argidx]);
975 cptr = mep->args[mtp->subst.argidx];
979 if(*cptr == '"' || *cptr == '\\')
988 if(debuglevel & DEBUGLEVEL_PPLEX)
989 fprintf(stderr, "add_expand_text: exp_concat\n");
990 /* Remove trailing whitespace from current expansion text */
993 if(isspace(curdef_text[curdef_idx-1] & 0xff))
998 /* tag current position and recursively expand the next part */
1000 mtp = add_expand_text(mtp->next, mep, nnl);
1002 /* Now get rid of the leading space of the expansion */
1003 cptr = &curdef_text[tag];
1004 n = curdef_idx - tag;
1007 if(isspace(*cptr & 0xff))
1015 if(cptr != &curdef_text[tag])
1017 memmove(&curdef_text[tag], cptr, n);
1018 curdef_idx -= (curdef_idx - tag) - n;
1023 if((mtp->next && mtp->next->type == exp_concat) || (mtp->prev && mtp->prev->type == exp_concat))
1024 exp = mep->args[mtp->subst.argidx];
1026 exp = mep->ppargs[mtp->subst.argidx];
1029 add_text(exp, strlen(exp));
1030 *nnl -= mep->nnls[mtp->subst.argidx];
1031 cptr = strchr(exp, '\n');
1035 cptr = strchr(cptr+1, '\n');
1037 mep->nnls[mtp->subst.argidx] = 0;
1039 if(debuglevel & DEBUGLEVEL_PPLEX)
1040 fprintf(stderr, "add_expand_text: exp_subst(%d): '%s'\n", mtp->subst.argidx, exp);
1044 internal_error(__FILE__, __LINE__, "Invalid expansion type (%d) in macro expansion\n", mtp->type);
1049 static void expand_macro(macexpstackentry_t *mep)
1055 pp_entry_t *ppp = mep->ppp;
1056 int nargs = mep->nargs;
1058 assert(ppp->type == def_macro);
1059 assert(ppp->expanding == 0);
1061 if((ppp->nargs >= 0 && nargs != ppp->nargs) || (ppp->nargs < 0 && nargs < -ppp->nargs))
1062 pperror("Too %s macro arguments (%d)", nargs < abs(ppp->nargs) ? "few" : "many", nargs);
1064 for(n = 0; n < nargs; n++)
1065 nnl += mep->nnls[n];
1067 if(debuglevel & DEBUGLEVEL_PPLEX)
1068 fprintf(stderr, "expand_macro(%d): %s:%d: '%s'(%d,%d) -> ...\n",
1078 for(mtp = ppp->subst.mtext; mtp; mtp = mtp->next)
1080 if(!(mtp = add_expand_text(mtp, mep, &nnl)))
1084 for(n = 0; n < nnl; n++)
1087 /* To make sure there is room and termination (see below) */
1090 /* Strip trailing whitespace from expansion */
1091 for(k = curdef_idx, cptr = &curdef_text[curdef_idx-1]; k > 0; k--, cptr--)
1093 if(!isspace(*cptr & 0xff))
1098 * We must add *one* whitespace to make sure that there
1099 * is a token-seperation after the expansion.
1105 /* Strip leading whitespace from expansion */
1106 for(n = 0, cptr = curdef_text; n < k; n++, cptr++)
1108 if(!isspace(*cptr & 0xff))
1114 if(debuglevel & DEBUGLEVEL_PPLEX)
1115 fprintf(stderr, "expand_text: '%s'\n", curdef_text + n);
1116 push_buffer(ppp, NULL, NULL, 0);
1117 /*yy_scan_bytes(curdef_text + n, k - n);*/
1118 yy_scan_string(curdef_text + n);
1123 *-------------------------------------------------------------------------
1124 * String collection routines
1125 *-------------------------------------------------------------------------
1127 static void new_string(void)
1131 ppwarning("new_string: strbuf_idx != 0");
1134 str_startline = line_number;
1137 static void add_string(char *str, int len)
1141 if(strbuf_idx >= strbuf_alloc || strbuf_alloc - strbuf_idx < len)
1143 strbuf_alloc += (len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1);
1144 strbuffer = xrealloc(strbuffer, strbuf_alloc * sizeof(strbuffer[0]));
1145 if(strbuf_alloc > 65536)
1146 ppwarning("Reallocating string buffer larger than 64kB");
1148 memcpy(&strbuffer[strbuf_idx], str, len);
1152 static char *get_string(void)
1154 char *str = (char *)xmalloc(strbuf_idx + 1);
1155 memcpy(str, strbuffer, strbuf_idx);
1156 str[strbuf_idx] = '\0';
1163 static void put_string(void)
1165 put_buffer(strbuffer, strbuf_idx);
1171 static int string_start(void)
1173 return str_startline;
1178 *-------------------------------------------------------------------------
1180 *-------------------------------------------------------------------------
1182 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop)
1185 printf("push_buffer(%d): %p %p %p %d\n", bufferstackidx, ppp, filename, incname, pop);
1186 if(bufferstackidx >= MAXBUFFERSTACK)
1187 internal_error(__FILE__, __LINE__, "Buffer stack overflow");
1189 memset(&bufferstack[bufferstackidx], 0, sizeof(bufferstack[0]));
1190 bufferstack[bufferstackidx].bufferstate = YY_CURRENT_BUFFER;
1191 bufferstack[bufferstackidx].define = ppp;
1192 bufferstack[bufferstackidx].line_number = line_number;
1193 bufferstack[bufferstackidx].char_number = char_number;
1194 bufferstack[bufferstackidx].if_depth = get_if_depth();
1195 bufferstack[bufferstackidx].should_pop = pop;
1196 bufferstack[bufferstackidx].filename = input_name;
1197 bufferstack[bufferstackidx].ncontinuations = ncontinuations;
1198 bufferstack[bufferstackidx].include_state = include_state;
1199 bufferstack[bufferstackidx].include_ppp = include_ppp;
1200 bufferstack[bufferstackidx].include_filename = incname;
1201 bufferstack[bufferstackidx].include_ifdepth = include_ifdepth;
1202 bufferstack[bufferstackidx].seen_junk = seen_junk;
1203 bufferstack[bufferstackidx].pass_data = pass_data;
1209 /* These will track the pperror to the correct file and line */
1212 input_name = filename;
1216 internal_error(__FILE__, __LINE__, "Pushing buffer without knowing where to go to");
1220 static bufferstackentry_t *pop_buffer(void)
1222 if(bufferstackidx < 0)
1223 internal_error(__FILE__, __LINE__, "Bufferstack underflow?");
1225 if(bufferstackidx == 0)
1230 if(bufferstack[bufferstackidx].define)
1231 bufferstack[bufferstackidx].define->expanding = 0;
1234 line_number = bufferstack[bufferstackidx].line_number;
1235 char_number = bufferstack[bufferstackidx].char_number;
1236 input_name = bufferstack[bufferstackidx].filename;
1237 ncontinuations = bufferstack[bufferstackidx].ncontinuations;
1238 if(!bufferstack[bufferstackidx].should_pop)
1241 fprintf(ppout, "# %d \"%s\" 2\n", line_number, input_name);
1243 /* We have EOF, check the include logic */
1244 if(include_state == 2 && !seen_junk && include_ppp)
1246 pp_entry_t *ppp = pplookup(include_ppp);
1249 includelogicentry_t *iep = xmalloc(sizeof(includelogicentry_t));
1252 iep->filename = bufferstack[bufferstackidx].include_filename;
1253 iep->next = includelogiclist;
1255 iep->next->prev = iep;
1256 includelogiclist = iep;
1257 if(debuglevel & DEBUGLEVEL_PPMSG)
1258 fprintf(stderr, "pop_buffer: %s:%d: includelogic added, include_ppp='%s', file='%s'\n", input_name, line_number, include_ppp, iep->filename);
1260 else if(bufferstack[bufferstackidx].include_filename)
1261 free(bufferstack[bufferstackidx].include_filename);
1265 include_state = bufferstack[bufferstackidx].include_state;
1266 include_ppp = bufferstack[bufferstackidx].include_ppp;
1267 include_ifdepth = bufferstack[bufferstackidx].include_ifdepth;
1268 seen_junk = bufferstack[bufferstackidx].seen_junk;
1269 pass_data = bufferstack[bufferstackidx].pass_data;
1275 printf("pop_buffer(%d): %p %p (%d, %d, %d) %p %d\n",
1277 bufferstack[bufferstackidx].bufferstate,
1278 bufferstack[bufferstackidx].define,
1279 bufferstack[bufferstackidx].line_number,
1280 bufferstack[bufferstackidx].char_number,
1281 bufferstack[bufferstackidx].if_depth,
1282 bufferstack[bufferstackidx].filename,
1283 bufferstack[bufferstackidx].should_pop);
1285 pp_switch_to_buffer(bufferstack[bufferstackidx].bufferstate);
1287 if(bufferstack[bufferstackidx].should_pop)
1289 if(yy_current_state() == pp_macexp)
1290 macro_add_expansion();
1292 internal_error(__FILE__, __LINE__, "Pop buffer and state without macro expansion state");
1296 return &bufferstack[bufferstackidx];
1301 *-------------------------------------------------------------------------
1302 * Macro nestng support
1303 *-------------------------------------------------------------------------
1305 static void push_macro(pp_entry_t *ppp)
1307 if(macexpstackidx >= MAXMACEXPSTACK)
1308 pperror("Too many nested macros");
1310 macexpstack[macexpstackidx] = xmalloc(sizeof(macexpstack[0][0]));
1312 macexpstack[macexpstackidx]->ppp = ppp;
1316 static macexpstackentry_t *top_macro(void)
1318 return macexpstackidx > 0 ? macexpstack[macexpstackidx-1] : NULL;
1321 static macexpstackentry_t *pop_macro(void)
1323 if(macexpstackidx <= 0)
1324 internal_error(__FILE__, __LINE__, "Macro expansion stack underflow\n");
1325 return macexpstack[--macexpstackidx];
1328 static void free_macro(macexpstackentry_t *mep)
1332 for(i = 0; i < mep->nargs; i++)
1343 static void add_text_to_macro(char *text, int len)
1345 macexpstackentry_t *mep = top_macro();
1347 assert(mep->ppp->expanding == 0);
1349 if(mep->curargalloc - mep->curargsize <= len+1) /* +1 for '\0' */
1351 mep->curargalloc += max(ALLOCBLOCKSIZE, len+1);
1352 mep->curarg = xrealloc(mep->curarg, mep->curargalloc * sizeof(mep->curarg[0]));
1354 memcpy(mep->curarg + mep->curargsize, text, len);
1355 mep->curargsize += len;
1356 mep->curarg[mep->curargsize] = '\0';
1359 static void macro_add_arg(int last)
1363 macexpstackentry_t *mep = top_macro();
1365 assert(mep->ppp->expanding == 0);
1367 mep->args = xrealloc(mep->args, (mep->nargs+1) * sizeof(mep->args[0]));
1368 mep->ppargs = xrealloc(mep->ppargs, (mep->nargs+1) * sizeof(mep->ppargs[0]));
1369 mep->nnls = xrealloc(mep->nnls, (mep->nargs+1) * sizeof(mep->nnls[0]));
1370 mep->args[mep->nargs] = xstrdup(mep->curarg ? mep->curarg : "");
1371 cptr = mep->args[mep->nargs]-1;
1372 while((cptr = strchr(cptr+1, '\n')))
1376 mep->nnls[mep->nargs] = nnl;
1379 mep->curargalloc = mep->curargsize = 0;
1382 if(debuglevel & DEBUGLEVEL_PPLEX)
1383 fprintf(stderr, "macro_add_arg: %s:%d: %d -> '%s'\n",
1387 mep->args[mep->nargs-1]);
1389 /* Each macro argument must be expanded to cope with stingize */
1390 if(last || mep->args[mep->nargs-1][0])
1392 yy_push_state(pp_macexp);
1393 push_buffer(NULL, NULL, NULL, last ? 2 : 1);
1394 yy_scan_string(mep->args[mep->nargs-1]);
1395 /*mep->bufferstackidx = bufferstackidx; But not nested! */
1399 static void macro_add_expansion(void)
1401 macexpstackentry_t *mep = top_macro();
1403 assert(mep->ppp->expanding == 0);
1405 mep->ppargs[mep->nargs-1] = xstrdup(mep->curarg ? mep->curarg : "");
1407 mep->curargalloc = mep->curargsize = 0;
1410 if(debuglevel & DEBUGLEVEL_PPLEX)
1411 fprintf(stderr, "macro_add_expansion: %s:%d: %d -> '%s'\n",
1415 mep->ppargs[mep->nargs-1]);
1420 *-------------------------------------------------------------------------
1422 *-------------------------------------------------------------------------
1424 static void put_buffer(char *s, int len)
1427 add_text_to_macro(s, len);
1430 fwrite(s, 1, len, ppout);
1436 *-------------------------------------------------------------------------
1437 * Include management
1438 *-------------------------------------------------------------------------
1440 int is_c_h_include(char *fname)
1442 int sl=strlen(fname);
1443 if (sl < 2) return 0;
1444 if ((toupper(fname[sl-1])!='H') && (toupper(fname[sl-1])!='C')) return 0;
1445 if (fname[sl-2]!='.') return 0;
1449 void do_include(char *fname, int type)
1453 includelogicentry_t *iep;
1455 for(iep = includelogiclist; iep; iep = iep->next)
1457 if(!strcmp(iep->filename, fname))
1460 * We are done. The file was included before.
1461 * If the define was deleted, then this entry would have
1471 pperror("Empty include filename");
1473 /* Undo the effect of the quotation */
1476 if((ppin = open_include(fname+1, type, &newpath)) == NULL)
1477 pperror("Unable to open include file %s", fname+1);
1479 fname[n-1] = *fname; /* Redo the quotes */
1480 push_buffer(NULL, newpath, fname, 0);
1484 if (is_c_h_include(newpath)) pass_data=0;
1487 if(debuglevel & DEBUGLEVEL_PPMSG)
1488 fprintf(stderr, "do_include: %s:%d: include_state=%d, include_ppp='%s', include_ifdepth=%d ,pass_data=%d\n", input_name, line_number, include_state, include_ppp, include_ifdepth,pass_data);
1489 pp_switch_to_buffer(pp_create_buffer(ppin, YY_BUF_SIZE));
1491 fprintf(ppout, "# 1 \"%s\" 1%s\n", newpath, type ? "" : " 3");
1495 *-------------------------------------------------------------------------
1496 * Push/pop preprocessor ignore state when processing conditionals
1498 *-------------------------------------------------------------------------
1500 void push_ignore_state(void)
1502 yy_push_state(pp_ignore);
1505 void pop_ignore_state(void)