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., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, 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]
166 #include "wpp_private.h"
170 * Make sure that we are running an appropriate version of flex.
172 #if !defined(YY_FLEX_MAJOR_VERSION) || (1000 * YY_FLEX_MAJOR_VERSION + YY_FLEX_MINOR_VERSION < 2005)
173 #error Must use flex version 2.5.1 or higher (yy_scan_* routines are required).
176 #define YY_READ_BUF_SIZE 65536 /* So we read most of a file at once */
178 #define yy_current_state() YY_START
179 #define yy_pp_state(x) yy_pop_state(); yy_push_state(x)
182 * Always update the current character position within a line
184 #define YY_USER_ACTION pp_status.char_number+=ppleng;
187 * Buffer management for includes and expansions
189 #define MAXBUFFERSTACK 128 /* Nesting more than 128 includes or macro expansion textss is insane */
191 typedef struct bufferstackentry {
192 YY_BUFFER_STATE bufferstate; /* Buffer to switch back to */
193 pp_entry_t *define; /* Points to expanding define or NULL if handling includes */
194 int line_number; /* Line that we were handling */
195 int char_number; /* The current position on that line */
196 const char *filename; /* Filename that we were handling */
197 int if_depth; /* How many #if:s deep to check matching #endif:s */
198 int ncontinuations; /* Remember the continuation state */
199 int should_pop; /* Set if we must pop the start-state on EOF */
200 /* Include management */
201 include_state_t incl;
202 char *include_filename;
204 } bufferstackentry_t;
206 #define ALLOCBLOCKSIZE (1 << 10) /* Allocate these chunks at a time for string-buffers */
209 * Macro expansion nesting
210 * We need the stack to handle expansions while scanning
211 * a macro's arguments. The TOS must always be the macro
212 * that receives the current expansion from the scanner.
214 #define MAXMACEXPSTACK 128 /* Nesting more than 128 macro expansions is insane */
216 typedef struct macexpstackentry {
217 pp_entry_t *ppp; /* This macro we are scanning */
218 char **args; /* With these arguments */
219 char **ppargs; /* Resulting in these preprocessed arguments */
220 int *nnls; /* Number of newlines per argument */
221 int nargs; /* And this many arguments scanned */
222 int parentheses; /* Nesting level of () */
223 int curargsize; /* Current scanning argument's size */
224 int curargalloc; /* Current scanning argument's block allocated */
225 char *curarg; /* Current scanning argument's content */
226 } macexpstackentry_t;
228 #define MACROPARENTHESES() (top_macro()->parentheses)
233 static void newline(int);
234 static int make_number(int radix, YYSTYPE *val, const char *str, int len);
235 static void put_buffer(const char *s, int len);
236 static int is_c_h_include(char *fname, int quoted);
237 /* Buffer management */
238 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop);
239 static bufferstackentry_t *pop_buffer(void);
240 /* String functions */
241 static void new_string(void);
242 static void add_string(const char *str, int len);
243 static char *get_string(void);
244 static void put_string(void);
245 static int string_start(void);
246 /* Macro functions */
247 static void push_macro(pp_entry_t *ppp);
248 static macexpstackentry_t *top_macro(void);
249 static macexpstackentry_t *pop_macro(void);
250 static void free_macro(macexpstackentry_t *mep);
251 static void add_text_to_macro(const char *text, int len);
252 static void macro_add_arg(int last);
253 static void macro_add_expansion(void);
255 static void expand_special(pp_entry_t *ppp);
256 static void expand_define(pp_entry_t *ppp);
257 static void expand_macro(macexpstackentry_t *mep);
262 static int ncontinuations;
264 static int strbuf_idx = 0;
265 static int strbuf_alloc = 0;
266 static char *strbuffer = NULL;
267 static int str_startline;
269 static macexpstackentry_t *macexpstack[MAXMACEXPSTACK];
270 static int macexpstackidx = 0;
272 static bufferstackentry_t bufferstack[MAXBUFFERSTACK];
273 static int bufferstackidx = 0;
275 static int pass_data=1;
280 include_state_t pp_incl_state =
288 includelogicentry_t *pp_includelogiclist = NULL;
293 **************************************************************************
294 * The scanner starts here
295 **************************************************************************
300 * Catch line-continuations.
301 * Note: Gcc keeps the line-continuations in, for example, strings
302 * intact. However, I prefer to remove them all so that the next
303 * scanner will not need to reduce the continuation state.
305 * <*>\\\n newline(0);
309 * Detect the leading # of a preprocessor directive.
311 <INITIAL,pp_ignore>^{ws}*# pp_incl_state.seen_junk++; yy_push_state(pp_pp);
314 * Scan for the preprocessor directives
316 <pp_pp>{ws}*include{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_inc); return tINCLUDE;} else {yy_pp_state(pp_eol);}
317 <pp_pp>{ws}*define{ws}* yy_pp_state(yy_current_state() != pp_ignore ? pp_def : pp_eol);
318 <pp_pp>{ws}*error{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tERROR;
319 <pp_pp>{ws}*warning{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tWARNING;
320 <pp_pp>{ws}*pragma{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPRAGMA;
321 <pp_pp>{ws}*ident{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPPIDENT;
322 <pp_pp>{ws}*undef{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_ifd); return tUNDEF;} else {yy_pp_state(pp_eol);}
323 <pp_pp>{ws}*ifdef{ws}* yy_pp_state(pp_ifd); return tIFDEF;
324 <pp_pp>{ws}*ifndef{ws}* pp_incl_state.seen_junk--; yy_pp_state(pp_ifd); return tIFNDEF;
325 <pp_pp>{ws}*if{ws}* yy_pp_state(pp_if); return tIF;
326 <pp_pp>{ws}*elif{ws}* yy_pp_state(pp_if); return tELIF;
327 <pp_pp>{ws}*else{ws}* yy_pp_state(pp_endif); return tELSE;
328 <pp_pp>{ws}*endif{ws}* yy_pp_state(pp_endif); return tENDIF;
329 <pp_pp>{ws}*line{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tLINE;} else {yy_pp_state(pp_eol);}
330 <pp_pp>{ws}+ if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tGCCLINE;} else {yy_pp_state(pp_eol);}
331 <pp_pp>{ws}*[a-z]+ pperror("Invalid preprocessor token '%s'", pptext);
332 <pp_pp>\r?\n newline(1); yy_pop_state(); return tNL; /* This could be the null-token */
333 <pp_pp>\\\r?\n newline(0);
334 <pp_pp>\\\r? pperror("Preprocessor junk '%s'", pptext);
335 <pp_pp>. return *pptext;
338 * Handle #include and #line
340 <pp_line>[0-9]+ return make_number(10, &pplval, pptext, ppleng);
341 <pp_inc>\< new_string(); add_string(pptext, ppleng); yy_push_state(pp_iqs);
342 <pp_inc,pp_line>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
343 <pp_inc,pp_line>{ws}+ ;
344 <pp_inc,pp_line>\n newline(1); yy_pop_state(); return tNL;
345 <pp_inc,pp_line>\\\r?\n newline(0);
346 <pp_inc,pp_line>(\\\r?)|(.) pperror(yy_current_state() == pp_inc ? "Trailing junk in #include" : "Trailing junk in #line");
349 * Ignore all input when a false clause is parsed
351 <pp_ignore>[^#/\\\n]+ ;
352 <pp_ignore>\n newline(1);
353 <pp_ignore>\\\r?\n newline(0);
354 <pp_ignore>(\\\r?)|(.) ;
357 * Handle #if and #elif.
358 * These require conditionals to be evaluated, but we do not
359 * want to jam the scanner normally when we see these tokens.
360 * Note: tIDENT is handled below.
363 <pp_if>0[0-7]*{ul}? return make_number(8, &pplval, pptext, ppleng);
364 <pp_if>0[0-7]*[8-9]+{ul}? pperror("Invalid octal digit");
365 <pp_if>[1-9][0-9]*{ul}? return make_number(10, &pplval, pptext, ppleng);
366 <pp_if>0[xX][0-9a-fA-F]+{ul}? return make_number(16, &pplval, pptext, ppleng);
367 <pp_if>0[xX] pperror("Invalid hex number");
368 <pp_if>defined yy_push_state(pp_defined); return tDEFINED;
369 <pp_if>"<<" return tLSHIFT;
370 <pp_if>">>" return tRSHIFT;
371 <pp_if>"&&" return tLOGAND;
372 <pp_if>"||" return tLOGOR;
373 <pp_if>"==" return tEQ;
374 <pp_if>"!=" return tNE;
375 <pp_if>"<=" return tLTE;
376 <pp_if>">=" return tGTE;
377 <pp_if>\n newline(1); yy_pop_state(); return tNL;
378 <pp_if>\\\r?\n newline(0);
379 <pp_if>\\\r? pperror("Junk in conditional expression");
381 <pp_if>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
382 <pp_if>\" pperror("String constants not allowed in conditionals");
383 <pp_if>. return *pptext;
386 * Handle #ifdef, #ifndef and #undef
387 * to get only an untranslated/unexpanded identifier
389 <pp_ifd>{cident} pplval.cptr = pp_xstrdup(pptext); return tIDENT;
391 <pp_ifd>\n newline(1); yy_pop_state(); return tNL;
392 <pp_ifd>\\\r?\n newline(0);
393 <pp_ifd>(\\\r?)|(.) pperror("Identifier expected");
396 * Handle #else and #endif.
399 <pp_endif>\n newline(1); yy_pop_state(); return tNL;
400 <pp_endif>\\\r?\n newline(0);
401 <pp_endif>. pperror("Garbage after #else or #endif.");
404 * Handle the special 'defined' keyword.
405 * This is necessary to get the identifier prior to any
408 <pp_defined>{cident} yy_pop_state(); pplval.cptr = pp_xstrdup(pptext); return tIDENT;
410 <pp_defined>(\()|(\)) return *pptext;
411 <pp_defined>\\\r?\n newline(0);
412 <pp_defined>(\\.)|(\n)|(.) pperror("Identifier expected");
415 * Handle #error, #warning, #pragma and #ident.
416 * Pass everything literally to the parser, which
417 * will act appropriately.
418 * Comments are stripped from the literal text.
420 <pp_eol>[^/\\\n]+ if(yy_top_state() != pp_ignore) { pplval.cptr = pp_xstrdup(pptext); return tLITERAL; }
421 <pp_eol>\/[^/\\\n*]* if(yy_top_state() != pp_ignore) { pplval.cptr = pp_xstrdup(pptext); return tLITERAL; }
422 <pp_eol>(\\\r?)|(\/[^/*]) if(yy_top_state() != pp_ignore) { pplval.cptr = pp_xstrdup(pptext); return tLITERAL; }
423 <pp_eol>\n newline(1); yy_pop_state(); if(yy_current_state() != pp_ignore) { return tNL; }
424 <pp_eol>\\\r?\n newline(0);
427 * Handle left side of #define
429 <pp_def>{cident}\( pplval.cptr = pp_xstrdup(pptext); pplval.cptr[ppleng-1] = '\0'; yy_pp_state(pp_macro); return tMACRO;
430 <pp_def>{cident} pplval.cptr = pp_xstrdup(pptext); yy_pp_state(pp_define); return tDEFINE;
432 <pp_def>\\\r?\n newline(0);
433 <pp_def>(\\\r?)|(\n)|(.) perror("Identifier expected");
436 * Scan the substitution of a define
438 <pp_define>[^'"/\\\n]+ pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
439 <pp_define>(\\\r?)|(\/[^/*]) pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
440 <pp_define>\\\r?\n{ws}+ newline(0); pplval.cptr = pp_xstrdup(" "); return tLITERAL;
441 <pp_define>\\\r?\n newline(0);
442 <pp_define>\n newline(1); yy_pop_state(); return tNL;
443 <pp_define>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
444 <pp_define>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
447 * Scan the definition macro arguments
449 <pp_macro>\){ws}* yy_pp_state(pp_mbody); return tMACROEND;
451 <pp_macro>{cident} pplval.cptr = pp_xstrdup(pptext); return tIDENT;
452 <pp_macro>, return ',';
453 <pp_macro>"..." return tELIPSIS;
454 <pp_macro>(\\\r?)|(\n)|(.)|(\.\.?) pperror("Argument identifier expected");
455 <pp_macro>\\\r?\n newline(0);
458 * Scan the substitution of a macro
460 <pp_mbody>[^a-zA-Z0-9'"#/\\\n]+ pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
461 <pp_mbody>{cident} pplval.cptr = pp_xstrdup(pptext); return tIDENT;
462 <pp_mbody>\#\# return tCONCAT;
463 <pp_mbody>\# return tSTRINGIZE;
464 <pp_mbody>[0-9][^'"#/\\\n]* pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
465 <pp_mbody>(\\\r?)|(\/[^/*'"#\\\n]*) pplval.cptr = pp_xstrdup(pptext); return tLITERAL;
466 <pp_mbody>\\\r?\n{ws}+ newline(0); pplval.cptr = pp_xstrdup(" "); return tLITERAL;
467 <pp_mbody>\\\r?\n newline(0);
468 <pp_mbody>\n newline(1); yy_pop_state(); return tNL;
469 <pp_mbody>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
470 <pp_mbody>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
473 * Macro expansion text scanning.
474 * This state is active just after the identifier is scanned
475 * that triggers an expansion. We *must* delete the leading
476 * whitespace before we can start scanning for arguments.
478 * If we do not see a '(' as next trailing token, then we have
479 * a false alarm. We just continue with a nose-bleed...
481 <pp_macign>{ws}*/\( yy_pp_state(pp_macscan);
483 if(yy_top_state() != pp_macscan)
486 <pp_macign>{ws}*\\\r?\n newline(0);
487 <pp_macign>{ws}+|{ws}*\\\r?|. {
488 macexpstackentry_t *mac = pop_macro();
490 put_buffer(mac->ppp->ident, strlen(mac->ppp->ident));
491 put_buffer(pptext, ppleng);
496 * Macro expansion argument text scanning.
497 * This state is active when a macro's arguments are being read for expansion.
500 if(++MACROPARENTHESES() > 1)
501 add_text_to_macro(pptext, ppleng);
504 if(--MACROPARENTHESES() == 0)
510 add_text_to_macro(pptext, ppleng);
513 if(MACROPARENTHESES() > 1)
514 add_text_to_macro(pptext, ppleng);
518 <pp_macscan>\" new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
519 <pp_macscan>\' new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
520 <pp_macscan>"/*" yy_push_state(pp_comment); add_text_to_macro(" ", 1);
521 <pp_macscan>\n pp_status.line_number++; pp_status.char_number = 1; add_text_to_macro(pptext, ppleng);
522 <pp_macscan>([^/(),\\\n"']+)|(\/[^/*(),\\\n'"]*)|(\\\r?)|(.) add_text_to_macro(pptext, ppleng);
523 <pp_macscan>\\\r?\n newline(0);
526 * Comment handling (almost all start-conditions)
528 <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);
529 <pp_comment>[^*\n]*|"*"+[^*/\n]* ;
530 <pp_comment>\n newline(0);
531 <pp_comment>"*"+"/" yy_pop_state();
534 * Remove C++ style comment (almost all start-conditions)
536 <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]* {
537 if(pptext[ppleng-1] == '\\')
538 ppwarning("C++ style comment ends with an escaped newline (escape ignored)");
542 * Single, double and <> quoted constants
544 <INITIAL,pp_macexp>\" pp_incl_state.seen_junk++; new_string(); add_string(pptext, ppleng); yy_push_state(pp_dqs);
545 <INITIAL,pp_macexp>\' pp_incl_state.seen_junk++; new_string(); add_string(pptext, ppleng); yy_push_state(pp_sqs);
546 <pp_dqs>[^"\\\n]+ add_string(pptext, ppleng);
548 add_string(pptext, ppleng);
550 switch(yy_current_state())
557 if (yy_current_state()==RCINCL) yy_pop_state();
558 pplval.cptr = get_string();
561 pplval.cptr = get_string();
562 if (is_c_h_include(pplval.cptr, 1)) pass_data=0;
569 <pp_sqs>[^'\\\n]+ add_string(pptext, ppleng);
571 add_string(pptext, ppleng);
573 switch(yy_current_state())
578 pplval.cptr = get_string();
584 <pp_iqs>[^\>\\\n]+ add_string(pptext, ppleng);
586 add_string(pptext, ppleng);
588 pplval.cptr = get_string();
593 * This is tricky; we need to remove the line-continuation
594 * from preprocessor strings, but OTOH retain them in all
595 * other strings. This is because the resource grammar is
596 * even more braindead than initially analysed and line-
597 * continuations in strings introduce, sigh, newlines in
598 * the output. There goes the concept of non-breaking, non-
599 * spacing whitespace.
601 switch(yy_top_state())
611 add_string(pptext, ppleng);
615 <pp_iqs,pp_dqs,pp_sqs>\\. add_string(pptext, ppleng);
616 <pp_iqs,pp_dqs,pp_sqs>\n {
618 add_string(pptext, ppleng);
619 ppwarning("Newline in string constant encounterd (started line %d)", string_start());
623 * Identifier scanning
625 <INITIAL,pp_if,pp_inc,pp_macexp>{cident} {
627 pp_incl_state.seen_junk++;
628 if(!(ppp = pplookup(pptext)))
630 if(yy_current_state() == pp_inc)
631 pperror("Expected include filename");
633 if(yy_current_state() == pp_if)
635 pplval.cptr = pp_xstrdup(pptext);
639 if((yy_current_state()==INITIAL) && (strcasecmp(pptext,"RCINCLUDE")==0)){
640 yy_push_state(RCINCL);
643 else put_buffer(pptext, ppleng);
646 else if(!ppp->expanding)
657 yy_push_state(pp_macign);
661 pp_internal_error(__FILE__, __LINE__, "Invalid define type %d\n", ppp->type);
664 else put_buffer(pptext, ppleng);
668 * Everything else that needs to be passed and
669 * newline and continuation handling
671 <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);
672 <INITIAL,pp_macexp>{ws}+ put_buffer(pptext, ppleng);
673 <INITIAL>\n newline(1);
674 <INITIAL>\\\r?\n newline(0);
675 <INITIAL>\\\r? pp_incl_state.seen_junk++; put_buffer(pptext, ppleng);
678 * Special catcher for macro argmument expansion to prevent
679 * newlines to propagate to the output or admin.
681 <pp_macexp>(\n)|(.)|(\\\r?(\n|.)) put_buffer(pptext, ppleng);
683 <RCINCL>[A-Za-z0-9_\.\\/]+ {
684 pplval.cptr=pp_xstrdup(pptext);
686 return tRCINCLUDEPATH;
692 new_string(); add_string(pptext,ppleng);yy_push_state(pp_dqs);
696 * This is a 'catch-all' rule to discover errors in the scanner
697 * in an orderly manner.
699 <*>. pp_incl_state.seen_junk++; ppwarning("Unmatched text '%c' (0x%02x); please report\n", isprint(*pptext & 0xff) ? *pptext : ' ', *pptext);
702 YY_BUFFER_STATE b = YY_CURRENT_BUFFER;
703 bufferstackentry_t *bep = pop_buffer();
705 if((!bep && pp_get_if_depth()) || (bep && pp_get_if_depth() != bep->if_depth))
706 ppwarning("Unmatched #if/#endif at end of file");
710 if(YY_START != INITIAL)
711 pperror("Unexpected end of file during preprocessing");
714 else if(bep->should_pop == 2)
716 macexpstackentry_t *mac;
725 **************************************************************************
727 **************************************************************************
739 *-------------------------------------------------------------------------
740 * Output newlines or set them as continuations
742 * Input: -1 - Don't count this one, but update local position (see pp_dqs)
743 * 0 - Line-continuation seen and cache output
744 * 1 - Newline seen and flush output
745 *-------------------------------------------------------------------------
747 static void newline(int dowrite)
749 pp_status.line_number++;
750 pp_status.char_number = 1;
758 for(;ncontinuations; ncontinuations--)
765 *-------------------------------------------------------------------------
766 * Make a number out of an any-base and suffixed string
768 * Possible number extensions:
771 * - "LL" long long int
773 * - "UL" unsigned long int
774 * - "ULL" unsigned long long int
775 * - "LU" unsigned long int
776 * - "LLU" unsigned long long int
780 * The sizes of resulting 'int' and 'long' are compiler specific.
781 * I depend on sizeof(int) > 2 here (although a relatively safe
783 * Long longs are not yet implemented because this is very compiler
784 * specific and I don't want to think too much about the problems.
786 *-------------------------------------------------------------------------
788 static int make_number(int radix, YYSTYPE *val, const char *str, int len)
796 ext[2] = toupper(str[len-1]);
797 ext[1] = len > 1 ? toupper(str[len-2]) : ' ';
798 ext[0] = len > 2 ? toupper(str[len-3]) : ' ';
800 if(!strcmp(ext, "LUL"))
801 pperror("Invalid constant suffix");
802 else if(!strcmp(ext, "LLU") || !strcmp(ext, "ULL"))
807 else if(!strcmp(ext+1, "LU") || !strcmp(ext+1, "UL"))
812 else if(!strcmp(ext+1, "LL"))
816 else if(!strcmp(ext+2, "L"))
820 else if(!strcmp(ext+2, "U"))
827 /* Assume as in the declaration of wrc_ull_t and wrc_sll_t */
828 #ifdef HAVE_LONG_LONG
831 val->ull = strtoull(str, NULL, radix);
836 val->sll = strtoll(str, NULL, radix);
840 pp_internal_error(__FILE__, __LINE__, "long long constants not supported on this platform");
843 else if(is_u && is_l)
845 val->ulong = strtoul(str, NULL, radix);
848 else if(!is_u && is_l)
850 val->slong = strtol(str, NULL, radix);
853 else if(is_u && !is_l)
855 val->uint = (unsigned int)strtoul(str, NULL, radix);
859 /* Else it must be an int... */
860 val->sint = (int)strtol(str, NULL, radix);
866 *-------------------------------------------------------------------------
867 * Macro and define expansion support
869 * FIXME: Variable macro arguments.
870 *-------------------------------------------------------------------------
872 static void expand_special(pp_entry_t *ppp)
874 const char *dbgtext = "?";
875 static char *buf = NULL;
877 assert(ppp->type == def_special);
879 if(!strcmp(ppp->ident, "__LINE__"))
881 dbgtext = "def_special(__LINE__)";
882 buf = pp_xrealloc(buf, 32);
883 sprintf(buf, "%d", pp_status.line_number);
885 else if(!strcmp(ppp->ident, "__FILE__"))
887 dbgtext = "def_special(__FILE__)";
888 buf = pp_xrealloc(buf, strlen(pp_status.input) + 3);
889 sprintf(buf, "\"%s\"", pp_status.input);
892 pp_internal_error(__FILE__, __LINE__, "Special macro '%s' not found...\n", ppp->ident);
895 fprintf(stderr, "expand_special(%d): %s:%d: '%s' -> '%s'\n",
898 pp_status.line_number,
904 push_buffer(ppp, NULL, NULL, 0);
909 static void expand_define(pp_entry_t *ppp)
911 assert(ppp->type == def_define);
914 fprintf(stderr, "expand_define(%d): %s:%d: '%s' -> '%s'\n",
917 pp_status.line_number,
920 if(ppp->subst.text && ppp->subst.text[0])
922 push_buffer(ppp, NULL, NULL, 0);
923 yy_scan_string(ppp->subst.text);
927 static int curdef_idx = 0;
928 static int curdef_alloc = 0;
929 static char *curdef_text = NULL;
931 static void add_text(const char *str, int len)
935 if(curdef_idx >= curdef_alloc || curdef_alloc - curdef_idx < len)
937 curdef_alloc += (len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1);
938 curdef_text = pp_xrealloc(curdef_text, curdef_alloc * sizeof(curdef_text[0]));
939 if(curdef_alloc > 65536)
940 ppwarning("Reallocating macro-expansion buffer larger than 64kB");
942 memcpy(&curdef_text[curdef_idx], str, len);
946 static mtext_t *add_expand_text(mtext_t *mtp, macexpstackentry_t *mep, int *nnl)
960 fprintf(stderr, "add_expand_text: exp_text: '%s'\n", mtp->subst.text);
961 add_text(mtp->subst.text, strlen(mtp->subst.text));
966 fprintf(stderr, "add_expand_text: exp_stringize(%d): '%s'\n",
968 mep->args[mtp->subst.argidx]);
969 cptr = mep->args[mtp->subst.argidx];
973 if(*cptr == '"' || *cptr == '\\')
983 fprintf(stderr, "add_expand_text: exp_concat\n");
984 /* Remove trailing whitespace from current expansion text */
987 if(isspace(curdef_text[curdef_idx-1] & 0xff))
992 /* tag current position and recursively expand the next part */
994 mtp = add_expand_text(mtp->next, mep, nnl);
996 /* Now get rid of the leading space of the expansion */
997 cptr = &curdef_text[tag];
998 n = curdef_idx - tag;
1001 if(isspace(*cptr & 0xff))
1009 if(cptr != &curdef_text[tag])
1011 memmove(&curdef_text[tag], cptr, n);
1012 curdef_idx -= (curdef_idx - tag) - n;
1017 if((mtp->next && mtp->next->type == exp_concat) || (mtp->prev && mtp->prev->type == exp_concat))
1018 exp = mep->args[mtp->subst.argidx];
1020 exp = mep->ppargs[mtp->subst.argidx];
1023 add_text(exp, strlen(exp));
1024 *nnl -= mep->nnls[mtp->subst.argidx];
1025 cptr = strchr(exp, '\n');
1029 cptr = strchr(cptr+1, '\n');
1031 mep->nnls[mtp->subst.argidx] = 0;
1034 fprintf(stderr, "add_expand_text: exp_subst(%d): '%s'\n", mtp->subst.argidx, exp);
1038 pp_internal_error(__FILE__, __LINE__, "Invalid expansion type (%d) in macro expansion\n", mtp->type);
1043 static void expand_macro(macexpstackentry_t *mep)
1049 pp_entry_t *ppp = mep->ppp;
1050 int nargs = mep->nargs;
1052 assert(ppp->type == def_macro);
1053 assert(ppp->expanding == 0);
1055 if((ppp->nargs >= 0 && nargs != ppp->nargs) || (ppp->nargs < 0 && nargs < -ppp->nargs))
1056 pperror("Too %s macro arguments (%d)", nargs < abs(ppp->nargs) ? "few" : "many", nargs);
1058 for(n = 0; n < nargs; n++)
1059 nnl += mep->nnls[n];
1062 fprintf(stderr, "expand_macro(%d): %s:%d: '%s'(%d,%d) -> ...\n",
1065 pp_status.line_number,
1072 for(mtp = ppp->subst.mtext; mtp; mtp = mtp->next)
1074 if(!(mtp = add_expand_text(mtp, mep, &nnl)))
1078 for(n = 0; n < nnl; n++)
1081 /* To make sure there is room and termination (see below) */
1084 /* Strip trailing whitespace from expansion */
1085 for(k = curdef_idx, cptr = &curdef_text[curdef_idx-1]; k > 0; k--, cptr--)
1087 if(!isspace(*cptr & 0xff))
1092 * We must add *one* whitespace to make sure that there
1093 * is a token-separation after the expansion.
1099 /* Strip leading whitespace from expansion */
1100 for(n = 0, cptr = curdef_text; n < k; n++, cptr++)
1102 if(!isspace(*cptr & 0xff))
1109 fprintf(stderr, "expand_text: '%s'\n", curdef_text + n);
1110 push_buffer(ppp, NULL, NULL, 0);
1111 /*yy_scan_bytes(curdef_text + n, k - n);*/
1112 yy_scan_string(curdef_text + n);
1117 *-------------------------------------------------------------------------
1118 * String collection routines
1119 *-------------------------------------------------------------------------
1121 static void new_string(void)
1125 ppwarning("new_string: strbuf_idx != 0");
1128 str_startline = pp_status.line_number;
1131 static void add_string(const char *str, int len)
1135 if(strbuf_idx >= strbuf_alloc || strbuf_alloc - strbuf_idx < len)
1137 strbuf_alloc += (len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1);
1138 strbuffer = pp_xrealloc(strbuffer, strbuf_alloc * sizeof(strbuffer[0]));
1139 if(strbuf_alloc > 65536)
1140 ppwarning("Reallocating string buffer larger than 64kB");
1142 memcpy(&strbuffer[strbuf_idx], str, len);
1146 static char *get_string(void)
1148 char *str = pp_xmalloc(strbuf_idx + 1);
1149 memcpy(str, strbuffer, strbuf_idx);
1150 str[strbuf_idx] = '\0';
1157 static void put_string(void)
1159 put_buffer(strbuffer, strbuf_idx);
1165 static int string_start(void)
1167 return str_startline;
1172 *-------------------------------------------------------------------------
1174 *-------------------------------------------------------------------------
1176 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop)
1179 printf("push_buffer(%d): %p %p %p %d\n", bufferstackidx, ppp, filename, incname, pop);
1180 if(bufferstackidx >= MAXBUFFERSTACK)
1181 pp_internal_error(__FILE__, __LINE__, "Buffer stack overflow");
1183 memset(&bufferstack[bufferstackidx], 0, sizeof(bufferstack[0]));
1184 bufferstack[bufferstackidx].bufferstate = YY_CURRENT_BUFFER;
1185 bufferstack[bufferstackidx].define = ppp;
1186 bufferstack[bufferstackidx].line_number = pp_status.line_number;
1187 bufferstack[bufferstackidx].char_number = pp_status.char_number;
1188 bufferstack[bufferstackidx].if_depth = pp_get_if_depth();
1189 bufferstack[bufferstackidx].should_pop = pop;
1190 bufferstack[bufferstackidx].filename = pp_status.input;
1191 bufferstack[bufferstackidx].ncontinuations = ncontinuations;
1192 bufferstack[bufferstackidx].incl = pp_incl_state;
1193 bufferstack[bufferstackidx].include_filename = incname;
1194 bufferstack[bufferstackidx].pass_data = pass_data;
1200 /* These will track the pperror to the correct file and line */
1201 pp_status.line_number = 1;
1202 pp_status.char_number = 1;
1203 pp_status.input = filename;
1207 pp_internal_error(__FILE__, __LINE__, "Pushing buffer without knowing where to go to");
1211 static bufferstackentry_t *pop_buffer(void)
1213 if(bufferstackidx < 0)
1214 pp_internal_error(__FILE__, __LINE__, "Bufferstack underflow?");
1216 if(bufferstackidx == 0)
1221 if(bufferstack[bufferstackidx].define)
1222 bufferstack[bufferstackidx].define->expanding = 0;
1225 pp_status.line_number = bufferstack[bufferstackidx].line_number;
1226 pp_status.char_number = bufferstack[bufferstackidx].char_number;
1227 pp_status.input = bufferstack[bufferstackidx].filename;
1228 ncontinuations = bufferstack[bufferstackidx].ncontinuations;
1229 if(!bufferstack[bufferstackidx].should_pop)
1232 fprintf(ppout, "# %d \"%s\" 2\n", pp_status.line_number, pp_status.input);
1234 /* We have EOF, check the include logic */
1235 if(pp_incl_state.state == 2 && !pp_incl_state.seen_junk && pp_incl_state.ppp)
1237 pp_entry_t *ppp = pplookup(pp_incl_state.ppp);
1240 includelogicentry_t *iep = pp_xmalloc(sizeof(includelogicentry_t));
1243 iep->filename = bufferstack[bufferstackidx].include_filename;
1245 iep->next = pp_includelogiclist;
1247 iep->next->prev = iep;
1248 pp_includelogiclist = iep;
1250 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);
1252 else if(bufferstack[bufferstackidx].include_filename)
1253 free(bufferstack[bufferstackidx].include_filename);
1255 if(pp_incl_state.ppp)
1256 free(pp_incl_state.ppp);
1257 pp_incl_state = bufferstack[bufferstackidx].incl;
1258 pass_data = bufferstack[bufferstackidx].pass_data;
1264 printf("pop_buffer(%d): %p %p (%d, %d, %d) %p %d\n",
1266 bufferstack[bufferstackidx].bufferstate,
1267 bufferstack[bufferstackidx].define,
1268 bufferstack[bufferstackidx].line_number,
1269 bufferstack[bufferstackidx].char_number,
1270 bufferstack[bufferstackidx].if_depth,
1271 bufferstack[bufferstackidx].filename,
1272 bufferstack[bufferstackidx].should_pop);
1274 pp_switch_to_buffer(bufferstack[bufferstackidx].bufferstate);
1276 if(bufferstack[bufferstackidx].should_pop)
1278 if(yy_current_state() == pp_macexp)
1279 macro_add_expansion();
1281 pp_internal_error(__FILE__, __LINE__, "Pop buffer and state without macro expansion state");
1285 return &bufferstack[bufferstackidx];
1290 *-------------------------------------------------------------------------
1291 * Macro nestng support
1292 *-------------------------------------------------------------------------
1294 static void push_macro(pp_entry_t *ppp)
1296 if(macexpstackidx >= MAXMACEXPSTACK)
1297 pperror("Too many nested macros");
1299 macexpstack[macexpstackidx] = pp_xmalloc(sizeof(macexpstack[0][0]));
1300 memset( macexpstack[macexpstackidx], 0, sizeof(macexpstack[0][0]));
1301 macexpstack[macexpstackidx]->ppp = ppp;
1305 static macexpstackentry_t *top_macro(void)
1307 return macexpstackidx > 0 ? macexpstack[macexpstackidx-1] : NULL;
1310 static macexpstackentry_t *pop_macro(void)
1312 if(macexpstackidx <= 0)
1313 pp_internal_error(__FILE__, __LINE__, "Macro expansion stack underflow\n");
1314 return macexpstack[--macexpstackidx];
1317 static void free_macro(macexpstackentry_t *mep)
1321 for(i = 0; i < mep->nargs; i++)
1332 static void add_text_to_macro(const char *text, int len)
1334 macexpstackentry_t *mep = top_macro();
1336 assert(mep->ppp->expanding == 0);
1338 if(mep->curargalloc - mep->curargsize <= len+1) /* +1 for '\0' */
1340 mep->curargalloc += (ALLOCBLOCKSIZE > len+1) ? ALLOCBLOCKSIZE : len+1;
1341 mep->curarg = pp_xrealloc(mep->curarg, mep->curargalloc * sizeof(mep->curarg[0]));
1343 memcpy(mep->curarg + mep->curargsize, text, len);
1344 mep->curargsize += len;
1345 mep->curarg[mep->curargsize] = '\0';
1348 static void macro_add_arg(int last)
1352 macexpstackentry_t *mep = top_macro();
1354 assert(mep->ppp->expanding == 0);
1356 mep->args = pp_xrealloc(mep->args, (mep->nargs+1) * sizeof(mep->args[0]));
1357 mep->ppargs = pp_xrealloc(mep->ppargs, (mep->nargs+1) * sizeof(mep->ppargs[0]));
1358 mep->nnls = pp_xrealloc(mep->nnls, (mep->nargs+1) * sizeof(mep->nnls[0]));
1359 mep->args[mep->nargs] = pp_xstrdup(mep->curarg ? mep->curarg : "");
1360 cptr = mep->args[mep->nargs]-1;
1361 while((cptr = strchr(cptr+1, '\n')))
1365 mep->nnls[mep->nargs] = nnl;
1368 mep->curargalloc = mep->curargsize = 0;
1372 fprintf(stderr, "macro_add_arg: %s:%d: %d -> '%s'\n",
1374 pp_status.line_number,
1376 mep->args[mep->nargs-1]);
1378 /* Each macro argument must be expanded to cope with stingize */
1379 if(last || mep->args[mep->nargs-1][0])
1381 yy_push_state(pp_macexp);
1382 push_buffer(NULL, NULL, NULL, last ? 2 : 1);
1383 yy_scan_string(mep->args[mep->nargs-1]);
1384 /*mep->bufferstackidx = bufferstackidx; But not nested! */
1388 static void macro_add_expansion(void)
1390 macexpstackentry_t *mep = top_macro();
1392 assert(mep->ppp->expanding == 0);
1394 mep->ppargs[mep->nargs-1] = pp_xstrdup(mep->curarg ? mep->curarg : "");
1396 mep->curargalloc = mep->curargsize = 0;
1400 fprintf(stderr, "macro_add_expansion: %s:%d: %d -> '%s'\n",
1402 pp_status.line_number,
1404 mep->ppargs[mep->nargs-1]);
1409 *-------------------------------------------------------------------------
1411 *-------------------------------------------------------------------------
1413 static void put_buffer(const char *s, int len)
1416 add_text_to_macro(s, len);
1419 fwrite(s, 1, len, ppout);
1425 *-------------------------------------------------------------------------
1426 * Include management
1427 *-------------------------------------------------------------------------
1429 static int is_c_h_include(char *fname, int quoted)
1431 int sl=strlen(fname);
1432 if (sl < 2 + 2 * quoted) return 0;
1433 if ((toupper(fname[sl-1-quoted])!='H') && (toupper(fname[sl-1-quoted])!='C')) return 0;
1434 if (fname[sl-2-quoted]!='.') return 0;
1438 void pp_do_include(char *fname, int type)
1442 includelogicentry_t *iep;
1444 for(iep = pp_includelogiclist; iep; iep = iep->next)
1446 if(!strcmp(iep->filename, fname))
1449 * We are done. The file was included before.
1450 * If the define was deleted, then this entry would have
1460 pperror("Empty include filename");
1462 /* Undo the effect of the quotation */
1465 if((ppin = pp_open_include(fname+1, type ? pp_status.input : NULL, &newpath)) == NULL)
1466 pperror("Unable to open include file %s", fname+1);
1468 fname[n-1] = *fname; /* Redo the quotes */
1469 push_buffer(NULL, newpath, fname, 0);
1470 pp_incl_state.seen_junk = 0;
1471 pp_incl_state.state = 0;
1472 pp_incl_state.ppp = NULL;
1473 if (is_c_h_include(newpath, 0)) pass_data=0;
1477 fprintf(stderr, "pp_do_include: %s:%d: include_state=%d, include_ppp='%s', include_ifdepth=%d ,pass_data=%d\n",
1478 pp_status.input, pp_status.line_number, pp_incl_state.state, pp_incl_state.ppp, pp_incl_state.ifdepth, pass_data);
1479 pp_switch_to_buffer(pp_create_buffer(ppin, YY_BUF_SIZE));
1481 fprintf(ppout, "# 1 \"%s\" 1%s\n", newpath, type ? "" : " 3");
1485 *-------------------------------------------------------------------------
1486 * Push/pop preprocessor ignore state when processing conditionals
1488 *-------------------------------------------------------------------------
1490 void pp_push_ignore_state(void)
1492 yy_push_state(pp_ignore);
1495 void pp_pop_ignore_state(void)