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
129 %option noinput nounput
130 %option prefix="ppy_"
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]
160 #include "wine/port.h"
170 # define LLONG_MAX ((long long)0x7fffffff << 32 | 0xffffffff)
171 # define LLONG_MIN (-LLONG_MAX - 1)
174 # define ULLONG_MAX ((long long)0xffffffff << 32 | 0xffffffff)
177 #ifndef HAVE_UNISTD_H
178 #define YY_NO_UNISTD_H
181 #include "wpp_private.h"
185 * Make sure that we are running an appropriate version of flex.
187 #if !defined(YY_FLEX_MAJOR_VERSION) || (1000 * YY_FLEX_MAJOR_VERSION + YY_FLEX_MINOR_VERSION < 2005)
188 #error Must use flex version 2.5.1 or higher (yy_scan_* routines are required).
191 #define YY_READ_BUF_SIZE 65536 /* So we read most of a file at once */
193 #define yy_current_state() YY_START
194 #define yy_pp_state(x) yy_pop_state(); yy_push_state(x)
197 * Always update the current character position within a line
199 #define YY_USER_ACTION pp_status.char_number+=ppy_leng;
202 * Buffer management for includes and expansions
204 #define MAXBUFFERSTACK 128 /* Nesting more than 128 includes or macro expansion textss is insane */
206 typedef struct bufferstackentry {
207 YY_BUFFER_STATE bufferstate; /* Buffer to switch back to */
208 pp_entry_t *define; /* Points to expanding define or NULL if handling includes */
209 int line_number; /* Line that we were handling */
210 int char_number; /* The current position on that line */
211 const char *filename; /* Filename that we were handling */
212 int if_depth; /* How many #if:s deep to check matching #endif:s */
213 int ncontinuations; /* Remember the continuation state */
214 int should_pop; /* Set if we must pop the start-state on EOF */
215 /* Include management */
216 include_state_t incl;
217 char *include_filename;
218 } bufferstackentry_t;
220 #define ALLOCBLOCKSIZE (1 << 10) /* Allocate these chunks at a time for string-buffers */
223 * Macro expansion nesting
224 * We need the stack to handle expansions while scanning
225 * a macro's arguments. The TOS must always be the macro
226 * that receives the current expansion from the scanner.
228 #define MAXMACEXPSTACK 128 /* Nesting more than 128 macro expansions is insane */
230 typedef struct macexpstackentry {
231 pp_entry_t *ppp; /* This macro we are scanning */
232 char **args; /* With these arguments */
233 char **ppargs; /* Resulting in these preprocessed arguments */
234 int *nnls; /* Number of newlines per argument */
235 int nargs; /* And this many arguments scanned */
236 int parentheses; /* Nesting level of () */
237 int curargsize; /* Current scanning argument's size */
238 int curargalloc; /* Current scanning argument's block allocated */
239 char *curarg; /* Current scanning argument's content */
240 } macexpstackentry_t;
242 #define MACROPARENTHESES() (top_macro()->parentheses)
247 static void newline(int);
248 static int make_number(int radix, YYSTYPE *val, const char *str, int len);
249 static void put_buffer(const char *s, int len);
250 /* Buffer management */
251 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop);
252 static bufferstackentry_t *pop_buffer(void);
253 /* String functions */
254 static void new_string(void);
255 static void add_string(const char *str, int len);
256 static char *get_string(void);
257 static void put_string(void);
258 static int string_start(void);
259 /* Macro functions */
260 static void push_macro(pp_entry_t *ppp);
261 static macexpstackentry_t *top_macro(void);
262 static macexpstackentry_t *pop_macro(void);
263 static void free_macro(macexpstackentry_t *mep);
264 static void add_text_to_macro(const char *text, int len);
265 static void macro_add_arg(int last);
266 static void macro_add_expansion(void);
268 static void expand_special(pp_entry_t *ppp);
269 static void expand_define(pp_entry_t *ppp);
270 static void expand_macro(macexpstackentry_t *mep);
275 static int ncontinuations;
277 static int strbuf_idx = 0;
278 static int strbuf_alloc = 0;
279 static char *strbuffer = NULL;
280 static int str_startline;
282 static macexpstackentry_t *macexpstack[MAXMACEXPSTACK];
283 static int macexpstackidx = 0;
285 static bufferstackentry_t bufferstack[MAXBUFFERSTACK];
286 static int bufferstackidx = 0;
291 include_state_t pp_incl_state =
299 includelogicentry_t *pp_includelogiclist = NULL;
304 **************************************************************************
305 * The scanner starts here
306 **************************************************************************
311 * Catch line-continuations.
312 * Note: Gcc keeps the line-continuations in, for example, strings
313 * intact. However, I prefer to remove them all so that the next
314 * scanner will not need to reduce the continuation state.
316 * <*>\\\n newline(0);
320 * Detect the leading # of a preprocessor directive.
322 <INITIAL,pp_ignore>^{ws}*# pp_incl_state.seen_junk++; yy_push_state(pp_pp);
325 * Scan for the preprocessor directives
327 <pp_pp>{ws}*include{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_inc); return tINCLUDE;} else {yy_pp_state(pp_eol);}
328 <pp_pp>{ws}*define{ws}* yy_pp_state(yy_current_state() != pp_ignore ? pp_def : pp_eol);
329 <pp_pp>{ws}*error{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tERROR;
330 <pp_pp>{ws}*warning{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tWARNING;
331 <pp_pp>{ws}*pragma{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPRAGMA;
332 <pp_pp>{ws}*ident{ws}* yy_pp_state(pp_eol); if(yy_top_state() != pp_ignore) return tPPIDENT;
333 <pp_pp>{ws}*undef{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_ifd); return tUNDEF;} else {yy_pp_state(pp_eol);}
334 <pp_pp>{ws}*ifdef{ws}* yy_pp_state(pp_ifd); return tIFDEF;
335 <pp_pp>{ws}*ifndef{ws}* pp_incl_state.seen_junk--; yy_pp_state(pp_ifd); return tIFNDEF;
336 <pp_pp>{ws}*if{ws}* yy_pp_state(pp_if); return tIF;
337 <pp_pp>{ws}*elif{ws}* yy_pp_state(pp_if); return tELIF;
338 <pp_pp>{ws}*else{ws}* yy_pp_state(pp_endif); return tELSE;
339 <pp_pp>{ws}*endif{ws}* yy_pp_state(pp_endif); return tENDIF;
340 <pp_pp>{ws}*line{ws}* if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tLINE;} else {yy_pp_state(pp_eol);}
341 <pp_pp>{ws}+ if(yy_top_state() != pp_ignore) {yy_pp_state(pp_line); return tGCCLINE;} else {yy_pp_state(pp_eol);}
342 <pp_pp>{ws}*[a-z]+ ppy_error("Invalid preprocessor token '%s'", ppy_text);
343 <pp_pp>\r?\n newline(1); yy_pop_state(); return tNL; /* This could be the null-token */
344 <pp_pp>\\\r?\n newline(0);
345 <pp_pp>\\\r? ppy_error("Preprocessor junk '%s'", ppy_text);
346 <pp_pp>. return *ppy_text;
349 * Handle #include and #line
351 <pp_line>[0-9]+ return make_number(10, &ppy_lval, ppy_text, ppy_leng);
352 <pp_inc>\< new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_iqs);
353 <pp_inc,pp_line>\" new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_dqs);
354 <pp_inc,pp_line>{ws}+ ;
355 <pp_inc,pp_line>\n newline(1); yy_pop_state(); return tNL;
356 <pp_inc,pp_line>\\\r?\n newline(0);
357 <pp_inc,pp_line>(\\\r?)|(.) ppy_error(yy_current_state() == pp_inc ? "Trailing junk in #include" : "Trailing junk in #line");
360 * Ignore all input when a false clause is parsed
362 <pp_ignore>[^#/\\\n]+ ;
363 <pp_ignore>\n newline(1);
364 <pp_ignore>\\\r?\n newline(0);
365 <pp_ignore>(\\\r?)|(.) ;
368 * Handle #if and #elif.
369 * These require conditionals to be evaluated, but we do not
370 * want to jam the scanner normally when we see these tokens.
371 * Note: tIDENT is handled below.
374 <pp_if>0[0-7]*{ul}? return make_number(8, &ppy_lval, ppy_text, ppy_leng);
375 <pp_if>0[0-7]*[8-9]+{ul}? ppy_error("Invalid octal digit");
376 <pp_if>[1-9][0-9]*{ul}? return make_number(10, &ppy_lval, ppy_text, ppy_leng);
377 <pp_if>0[xX][0-9a-fA-F]+{ul}? return make_number(16, &ppy_lval, ppy_text, ppy_leng);
378 <pp_if>0[xX] ppy_error("Invalid hex number");
379 <pp_if>defined yy_push_state(pp_defined); return tDEFINED;
380 <pp_if>"<<" return tLSHIFT;
381 <pp_if>">>" return tRSHIFT;
382 <pp_if>"&&" return tLOGAND;
383 <pp_if>"||" return tLOGOR;
384 <pp_if>"==" return tEQ;
385 <pp_if>"!=" return tNE;
386 <pp_if>"<=" return tLTE;
387 <pp_if>">=" return tGTE;
388 <pp_if>\n newline(1); yy_pop_state(); return tNL;
389 <pp_if>\\\r?\n newline(0);
390 <pp_if>\\\r? ppy_error("Junk in conditional expression");
392 <pp_if>\' new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_sqs);
393 <pp_if>\" ppy_error("String constants not allowed in conditionals");
394 <pp_if>. return *ppy_text;
397 * Handle #ifdef, #ifndef and #undef
398 * to get only an untranslated/unexpanded identifier
400 <pp_ifd>{cident} ppy_lval.cptr = pp_xstrdup(ppy_text); return tIDENT;
402 <pp_ifd>\n newline(1); yy_pop_state(); return tNL;
403 <pp_ifd>\\\r?\n newline(0);
404 <pp_ifd>(\\\r?)|(.) ppy_error("Identifier expected");
407 * Handle #else and #endif.
410 <pp_endif>\n newline(1); yy_pop_state(); return tNL;
411 <pp_endif>\\\r?\n newline(0);
412 <pp_endif>. ppy_error("Garbage after #else or #endif.");
415 * Handle the special 'defined' keyword.
416 * This is necessary to get the identifier prior to any
419 <pp_defined>{cident} yy_pop_state(); ppy_lval.cptr = pp_xstrdup(ppy_text); return tIDENT;
421 <pp_defined>(\()|(\)) return *ppy_text;
422 <pp_defined>\\\r?\n newline(0);
423 <pp_defined>(\\.)|(\n)|(.) ppy_error("Identifier expected");
426 * Handle #error, #warning, #pragma and #ident.
427 * Pass everything literally to the parser, which
428 * will act appropriately.
429 * Comments are stripped from the literal text.
431 <pp_eol>[^/\\\n]+ if(yy_top_state() != pp_ignore) { ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL; }
432 <pp_eol>\/[^/\\\n*]* if(yy_top_state() != pp_ignore) { ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL; }
433 <pp_eol>(\\\r?)|(\/[^/*]) if(yy_top_state() != pp_ignore) { ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL; }
434 <pp_eol>\n newline(1); yy_pop_state(); if(yy_current_state() != pp_ignore) { return tNL; }
435 <pp_eol>\\\r?\n newline(0);
438 * Handle left side of #define
440 <pp_def>{cident}\( ppy_lval.cptr = pp_xstrdup(ppy_text); if(ppy_lval.cptr) ppy_lval.cptr[ppy_leng-1] = '\0'; yy_pp_state(pp_macro); return tMACRO;
441 <pp_def>{cident} ppy_lval.cptr = pp_xstrdup(ppy_text); yy_pp_state(pp_define); return tDEFINE;
443 <pp_def>\\\r?\n newline(0);
444 <pp_def>(\\\r?)|(\n)|(.) perror("Identifier expected");
447 * Scan the substitution of a define
449 <pp_define>[^'"/\\\n]+ ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL;
450 <pp_define>(\\\r?)|(\/[^/*]) ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL;
451 <pp_define>\\\r?\n{ws}+ newline(0); ppy_lval.cptr = pp_xstrdup(" "); return tLITERAL;
452 <pp_define>\\\r?\n newline(0);
453 <pp_define>\n newline(1); yy_pop_state(); return tNL;
454 <pp_define>\' new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_sqs);
455 <pp_define>\" new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_dqs);
458 * Scan the definition macro arguments
460 <pp_macro>\){ws}* yy_pp_state(pp_mbody); return tMACROEND;
462 <pp_macro>{cident} ppy_lval.cptr = pp_xstrdup(ppy_text); return tIDENT;
463 <pp_macro>, return ',';
464 <pp_macro>"..." return tELIPSIS;
465 <pp_macro>(\\\r?)|(\n)|(.)|(\.\.?) ppy_error("Argument identifier expected");
466 <pp_macro>\\\r?\n newline(0);
469 * Scan the substitution of a macro
471 <pp_mbody>[^a-zA-Z0-9'"#/\\\n]+ ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL;
472 <pp_mbody>{cident} ppy_lval.cptr = pp_xstrdup(ppy_text); return tIDENT;
473 <pp_mbody>\#\# return tCONCAT;
474 <pp_mbody>\# return tSTRINGIZE;
475 <pp_mbody>[0-9][^'"#/\\\n]* ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL;
476 <pp_mbody>(\\\r?)|(\/[^/*'"#\\\n]*) ppy_lval.cptr = pp_xstrdup(ppy_text); return tLITERAL;
477 <pp_mbody>\\\r?\n{ws}+ newline(0); ppy_lval.cptr = pp_xstrdup(" "); return tLITERAL;
478 <pp_mbody>\\\r?\n newline(0);
479 <pp_mbody>\n newline(1); yy_pop_state(); return tNL;
480 <pp_mbody>\' new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_sqs);
481 <pp_mbody>\" new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_dqs);
484 * Macro expansion text scanning.
485 * This state is active just after the identifier is scanned
486 * that triggers an expansion. We *must* delete the leading
487 * whitespace before we can start scanning for arguments.
489 * If we do not see a '(' as next trailing token, then we have
490 * a false alarm. We just continue with a nose-bleed...
492 <pp_macign>{ws}*/\( yy_pp_state(pp_macscan);
494 if(yy_top_state() != pp_macscan)
497 <pp_macign>{ws}*\\\r?\n newline(0);
498 <pp_macign>{ws}+|{ws}*\\\r?|. {
499 macexpstackentry_t *mac = pop_macro();
501 put_buffer(mac->ppp->ident, strlen(mac->ppp->ident));
502 put_buffer(ppy_text, ppy_leng);
507 * Macro expansion argument text scanning.
508 * This state is active when a macro's arguments are being read for expansion.
511 if(++MACROPARENTHESES() > 1)
512 add_text_to_macro(ppy_text, ppy_leng);
515 if(--MACROPARENTHESES() == 0)
521 add_text_to_macro(ppy_text, ppy_leng);
524 if(MACROPARENTHESES() > 1)
525 add_text_to_macro(ppy_text, ppy_leng);
529 <pp_macscan>\" new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_dqs);
530 <pp_macscan>\' new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_sqs);
531 <pp_macscan>"/*" yy_push_state(pp_comment); add_text_to_macro(" ", 1);
532 <pp_macscan>\n pp_status.line_number++; pp_status.char_number = 1; add_text_to_macro(ppy_text, ppy_leng);
533 <pp_macscan>([^/(),\\\n"']+)|(\/[^/*(),\\\n'"]*)|(\\\r?)|(.) add_text_to_macro(ppy_text, ppy_leng);
534 <pp_macscan>\\\r?\n newline(0);
537 * Comment handling (almost all start-conditions)
539 <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);
540 <pp_comment>[^*\n]*|"*"+[^*/\n]* ;
541 <pp_comment>\n newline(0);
542 <pp_comment>"*"+"/" yy_pop_state();
545 * Remove C++ style comment (almost all start-conditions)
547 <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]* {
548 if(ppy_text[ppy_leng-1] == '\\')
549 ppy_warning("C++ style comment ends with an escaped newline (escape ignored)");
553 * Single, double and <> quoted constants
555 <INITIAL,pp_macexp>\" pp_incl_state.seen_junk++; new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_dqs);
556 <INITIAL,pp_macexp>\' pp_incl_state.seen_junk++; new_string(); add_string(ppy_text, ppy_leng); yy_push_state(pp_sqs);
557 <pp_dqs>[^"\\\n]+ add_string(ppy_text, ppy_leng);
559 add_string(ppy_text, ppy_leng);
561 switch(yy_current_state())
568 if (yy_current_state()==RCINCL) yy_pop_state();
569 ppy_lval.cptr = get_string();
572 ppy_lval.cptr = get_string();
578 <pp_sqs>[^'\\\n]+ add_string(ppy_text, ppy_leng);
580 add_string(ppy_text, ppy_leng);
582 switch(yy_current_state())
587 ppy_lval.cptr = get_string();
593 <pp_iqs>[^\>\\\n]+ add_string(ppy_text, ppy_leng);
595 add_string(ppy_text, ppy_leng);
597 ppy_lval.cptr = get_string();
602 * This is tricky; we need to remove the line-continuation
603 * from preprocessor strings, but OTOH retain them in all
604 * other strings. This is because the resource grammar is
605 * even more braindead than initially analysed and line-
606 * continuations in strings introduce, sigh, newlines in
607 * the output. There goes the concept of non-breaking, non-
608 * spacing whitespace.
610 switch(yy_top_state())
620 add_string(ppy_text, ppy_leng);
624 <pp_iqs,pp_dqs,pp_sqs>\\. add_string(ppy_text, ppy_leng);
625 <pp_iqs,pp_dqs,pp_sqs>\n {
627 add_string(ppy_text, ppy_leng);
628 ppy_warning("Newline in string constant encounterd (started line %d)", string_start());
632 * Identifier scanning
634 <INITIAL,pp_if,pp_inc,pp_macexp>{cident} {
636 pp_incl_state.seen_junk++;
637 if(!(ppp = pplookup(ppy_text)))
639 if(yy_current_state() == pp_inc)
640 ppy_error("Expected include filename");
642 else if(yy_current_state() == pp_if)
644 ppy_lval.cptr = pp_xstrdup(ppy_text);
648 if((yy_current_state()==INITIAL) && (strcasecmp(ppy_text,"RCINCLUDE")==0)){
649 yy_push_state(RCINCL);
652 else put_buffer(ppy_text, ppy_leng);
655 else if(!ppp->expanding)
666 yy_push_state(pp_macign);
670 pp_internal_error(__FILE__, __LINE__, "Invalid define type %d\n", ppp->type);
673 else put_buffer(ppy_text, ppy_leng);
677 * Everything else that needs to be passed and
678 * newline and continuation handling
680 <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(ppy_text, ppy_leng);
681 <INITIAL,pp_macexp>{ws}+ put_buffer(ppy_text, ppy_leng);
682 <INITIAL>\n newline(1);
683 <INITIAL>\\\r?\n newline(0);
684 <INITIAL>\\\r? pp_incl_state.seen_junk++; put_buffer(ppy_text, ppy_leng);
687 * Special catcher for macro argmument expansion to prevent
688 * newlines to propagate to the output or admin.
690 <pp_macexp>(\n)|(.)|(\\\r?(\n|.)) put_buffer(ppy_text, ppy_leng);
692 <RCINCL>[A-Za-z0-9_\.\\/]+ {
693 ppy_lval.cptr=pp_xstrdup(ppy_text);
695 return tRCINCLUDEPATH;
701 new_string(); add_string(ppy_text,ppy_leng);yy_push_state(pp_dqs);
705 * This is a 'catch-all' rule to discover errors in the scanner
706 * in an orderly manner.
708 <*>. pp_incl_state.seen_junk++; ppy_warning("Unmatched text '%c' (0x%02x); please report\n", isprint(*ppy_text & 0xff) ? *ppy_text : ' ', *ppy_text);
711 YY_BUFFER_STATE b = YY_CURRENT_BUFFER;
712 bufferstackentry_t *bep = pop_buffer();
714 if((!bep && pp_get_if_depth()) || (bep && pp_get_if_depth() != bep->if_depth))
715 ppy_warning("Unmatched #if/#endif at end of file");
719 if(YY_START != INITIAL)
720 ppy_error("Unexpected end of file during preprocessing");
723 else if(bep->should_pop == 2)
725 macexpstackentry_t *mac;
729 ppy__delete_buffer(b);
734 **************************************************************************
736 **************************************************************************
748 *-------------------------------------------------------------------------
749 * Output newlines or set them as continuations
751 * Input: -1 - Don't count this one, but update local position (see pp_dqs)
752 * 0 - Line-continuation seen and cache output
753 * 1 - Newline seen and flush output
754 *-------------------------------------------------------------------------
756 static void newline(int dowrite)
758 pp_status.line_number++;
759 pp_status.char_number = 1;
767 for(;ncontinuations; ncontinuations--)
774 *-------------------------------------------------------------------------
775 * Make a number out of an any-base and suffixed string
777 * Possible number extensions:
780 * - "LL" long long int
782 * - "UL" unsigned long int
783 * - "ULL" unsigned long long int
784 * - "LU" unsigned long int
785 * - "LLU" unsigned long long int
789 * The sizes of resulting 'int' and 'long' are compiler specific.
790 * I depend on sizeof(int) > 2 here (although a relatively safe
792 * Long longs are not yet implemented because this is very compiler
793 * specific and I don't want to think too much about the problems.
795 *-------------------------------------------------------------------------
797 static int make_number(int radix, YYSTYPE *val, const char *str, int len)
806 ext[2] = toupper(str[len-1]);
807 ext[1] = len > 1 ? toupper(str[len-2]) : ' ';
808 ext[0] = len > 2 ? toupper(str[len-3]) : ' ';
810 if(!strcmp(ext, "LUL"))
812 ppy_error("Invalid constant suffix");
815 else if(!strcmp(ext, "LLU") || !strcmp(ext, "ULL"))
820 else if(!strcmp(ext+1, "LU") || !strcmp(ext+1, "UL"))
825 else if(!strcmp(ext+1, "LL"))
829 else if(!strcmp(ext+2, "L"))
833 else if(!strcmp(ext+2, "U"))
840 /* Assume as in the declaration of wrc_ull_t and wrc_sll_t */
841 #ifdef HAVE_LONG_LONG
845 val->ull = strtoull(str, NULL, radix);
846 if (val->ull == ULLONG_MAX && errno == ERANGE)
847 ppy_error("integer constant %s is too large\n", str);
853 val->sll = strtoll(str, NULL, radix);
854 if ((val->sll == LLONG_MIN || val->sll == LLONG_MAX) && errno == ERANGE)
855 ppy_error("integer constant %s is too large\n", str);
859 pp_internal_error(__FILE__, __LINE__, "long long constants not supported on this platform");
862 else if(is_u && is_l)
865 val->ulong = strtoul(str, NULL, radix);
866 if (val->ulong == ULONG_MAX && errno == ERANGE)
867 ppy_error("integer constant %s is too large\n", str);
870 else if(!is_u && is_l)
873 val->slong = strtol(str, NULL, radix);
874 if ((val->slong == LONG_MIN || val->slong == LONG_MAX) && errno == ERANGE)
875 ppy_error("integer constant %s is too large\n", str);
878 else if(is_u && !is_l)
882 ul = strtoul(str, NULL, radix);
883 if ((ul == ULONG_MAX && errno == ERANGE) || (ul > UINT_MAX))
884 ppy_error("integer constant %s is too large\n", str);
885 val->uint = (unsigned int)ul;
889 /* Else it must be an int... */
891 l = strtol(str, NULL, radix);
892 if (((l == LONG_MIN || l == LONG_MAX) && errno == ERANGE) ||
893 (l > INT_MAX) || (l < INT_MIN))
894 ppy_error("integer constant %s is too large\n", str);
901 *-------------------------------------------------------------------------
902 * Macro and define expansion support
904 * FIXME: Variable macro arguments.
905 *-------------------------------------------------------------------------
907 static void expand_special(pp_entry_t *ppp)
909 const char *dbgtext = "?";
910 static char *buf = NULL;
913 assert(ppp->type == def_special);
915 if(!strcmp(ppp->ident, "__LINE__"))
917 dbgtext = "def_special(__LINE__)";
918 new_buf = pp_xrealloc(buf, 32);
922 sprintf(buf, "%d", pp_status.line_number);
924 else if(!strcmp(ppp->ident, "__FILE__"))
926 dbgtext = "def_special(__FILE__)";
927 new_buf = pp_xrealloc(buf, strlen(pp_status.input) + 3);
931 sprintf(buf, "\"%s\"", pp_status.input);
934 pp_internal_error(__FILE__, __LINE__, "Special macro '%s' not found...\n", ppp->ident);
937 fprintf(stderr, "expand_special(%d): %s:%d: '%s' -> '%s'\n",
940 pp_status.line_number,
946 push_buffer(ppp, NULL, NULL, 0);
951 static void expand_define(pp_entry_t *ppp)
953 assert(ppp->type == def_define);
956 fprintf(stderr, "expand_define(%d): %s:%d: '%s' -> '%s'\n",
959 pp_status.line_number,
962 if(ppp->subst.text && ppp->subst.text[0])
964 push_buffer(ppp, NULL, NULL, 0);
965 yy_scan_string(ppp->subst.text);
969 static int curdef_idx = 0;
970 static int curdef_alloc = 0;
971 static char *curdef_text = NULL;
973 static void add_text(const char *str, int len)
980 if(curdef_idx >= curdef_alloc || curdef_alloc - curdef_idx < len)
982 new_alloc = curdef_alloc + ((len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1));
983 new_text = pp_xrealloc(curdef_text, new_alloc * sizeof(curdef_text[0]));
986 curdef_text = new_text;
987 curdef_alloc = new_alloc;
988 if(curdef_alloc > 65536)
989 ppy_warning("Reallocating macro-expansion buffer larger than 64kB");
991 memcpy(&curdef_text[curdef_idx], str, len);
995 static mtext_t *add_expand_text(mtext_t *mtp, macexpstackentry_t *mep, int *nnl)
1009 fprintf(stderr, "add_expand_text: exp_text: '%s'\n", mtp->subst.text);
1010 add_text(mtp->subst.text, strlen(mtp->subst.text));
1015 fprintf(stderr, "add_expand_text: exp_stringize(%d): '%s'\n",
1017 mep->args[mtp->subst.argidx]);
1018 cptr = mep->args[mtp->subst.argidx];
1022 if(*cptr == '"' || *cptr == '\\')
1032 fprintf(stderr, "add_expand_text: exp_concat\n");
1033 /* Remove trailing whitespace from current expansion text */
1036 if(isspace(curdef_text[curdef_idx-1] & 0xff))
1041 /* tag current position and recursively expand the next part */
1043 mtp = add_expand_text(mtp->next, mep, nnl);
1045 /* Now get rid of the leading space of the expansion */
1046 cptr = &curdef_text[tag];
1047 n = curdef_idx - tag;
1050 if(isspace(*cptr & 0xff))
1058 if(cptr != &curdef_text[tag])
1060 memmove(&curdef_text[tag], cptr, n);
1061 curdef_idx -= (curdef_idx - tag) - n;
1066 if((mtp->next && mtp->next->type == exp_concat) || (mtp->prev && mtp->prev->type == exp_concat))
1067 exp = mep->args[mtp->subst.argidx];
1069 exp = mep->ppargs[mtp->subst.argidx];
1072 add_text(exp, strlen(exp));
1073 *nnl -= mep->nnls[mtp->subst.argidx];
1074 cptr = strchr(exp, '\n');
1078 cptr = strchr(cptr+1, '\n');
1080 mep->nnls[mtp->subst.argidx] = 0;
1083 fprintf(stderr, "add_expand_text: exp_subst(%d): '%s'\n", mtp->subst.argidx, exp);
1087 pp_internal_error(__FILE__, __LINE__, "Invalid expansion type (%d) in macro expansion\n", mtp->type);
1092 static void expand_macro(macexpstackentry_t *mep)
1098 pp_entry_t *ppp = mep->ppp;
1099 int nargs = mep->nargs;
1101 assert(ppp->type == def_macro);
1102 assert(ppp->expanding == 0);
1104 if((ppp->nargs >= 0 && nargs != ppp->nargs) || (ppp->nargs < 0 && nargs < -ppp->nargs))
1106 ppy_error("Too %s macro arguments (%d)", nargs < abs(ppp->nargs) ? "few" : "many", nargs);
1110 for(n = 0; n < nargs; n++)
1111 nnl += mep->nnls[n];
1114 fprintf(stderr, "expand_macro(%d): %s:%d: '%s'(%d,%d) -> ...\n",
1117 pp_status.line_number,
1124 for(mtp = ppp->subst.mtext; mtp; mtp = mtp->next)
1126 if(!(mtp = add_expand_text(mtp, mep, &nnl)))
1130 for(n = 0; n < nnl; n++)
1133 /* To make sure there is room and termination (see below) */
1136 /* Strip trailing whitespace from expansion */
1137 for(k = curdef_idx, cptr = &curdef_text[curdef_idx-1]; k > 0; k--, cptr--)
1139 if(!isspace(*cptr & 0xff))
1144 * We must add *one* whitespace to make sure that there
1145 * is a token-separation after the expansion.
1151 /* Strip leading whitespace from expansion */
1152 for(n = 0, cptr = curdef_text; n < k; n++, cptr++)
1154 if(!isspace(*cptr & 0xff))
1161 fprintf(stderr, "expand_text: '%s'\n", curdef_text + n);
1162 push_buffer(ppp, NULL, NULL, 0);
1163 /*yy_scan_bytes(curdef_text + n, k - n);*/
1164 yy_scan_string(curdef_text + n);
1169 *-------------------------------------------------------------------------
1170 * String collection routines
1171 *-------------------------------------------------------------------------
1173 static void new_string(void)
1177 ppy_warning("new_string: strbuf_idx != 0");
1180 str_startline = pp_status.line_number;
1183 static void add_string(const char *str, int len)
1190 if(strbuf_idx >= strbuf_alloc || strbuf_alloc - strbuf_idx < len)
1192 new_alloc = strbuf_alloc + ((len + ALLOCBLOCKSIZE-1) & ~(ALLOCBLOCKSIZE-1));
1193 new_buffer = pp_xrealloc(strbuffer, new_alloc * sizeof(strbuffer[0]));
1196 strbuffer = new_buffer;
1197 strbuf_alloc = new_alloc;
1198 if(strbuf_alloc > 65536)
1199 ppy_warning("Reallocating string buffer larger than 64kB");
1201 memcpy(&strbuffer[strbuf_idx], str, len);
1205 static char *get_string(void)
1207 char *str = pp_xmalloc(strbuf_idx + 1);
1210 memcpy(str, strbuffer, strbuf_idx);
1211 str[strbuf_idx] = '\0';
1218 static void put_string(void)
1220 put_buffer(strbuffer, strbuf_idx);
1226 static int string_start(void)
1228 return str_startline;
1233 *-------------------------------------------------------------------------
1235 *-------------------------------------------------------------------------
1237 static void push_buffer(pp_entry_t *ppp, char *filename, char *incname, int pop)
1240 printf("push_buffer(%d): %p %p %p %d\n", bufferstackidx, ppp, filename, incname, pop);
1241 if(bufferstackidx >= MAXBUFFERSTACK)
1242 pp_internal_error(__FILE__, __LINE__, "Buffer stack overflow");
1244 memset(&bufferstack[bufferstackidx], 0, sizeof(bufferstack[0]));
1245 bufferstack[bufferstackidx].bufferstate = YY_CURRENT_BUFFER;
1246 bufferstack[bufferstackidx].define = ppp;
1247 bufferstack[bufferstackidx].line_number = pp_status.line_number;
1248 bufferstack[bufferstackidx].char_number = pp_status.char_number;
1249 bufferstack[bufferstackidx].if_depth = pp_get_if_depth();
1250 bufferstack[bufferstackidx].should_pop = pop;
1251 bufferstack[bufferstackidx].filename = pp_status.input;
1252 bufferstack[bufferstackidx].ncontinuations = ncontinuations;
1253 bufferstack[bufferstackidx].incl = pp_incl_state;
1254 bufferstack[bufferstackidx].include_filename = incname;
1260 /* These will track the ppy_error to the correct file and line */
1261 pp_status.line_number = 1;
1262 pp_status.char_number = 1;
1263 pp_status.input = filename;
1267 pp_internal_error(__FILE__, __LINE__, "Pushing buffer without knowing where to go to");
1271 static bufferstackentry_t *pop_buffer(void)
1273 if(bufferstackidx < 0)
1274 pp_internal_error(__FILE__, __LINE__, "Bufferstack underflow?");
1276 if(bufferstackidx == 0)
1281 if(bufferstack[bufferstackidx].define)
1282 bufferstack[bufferstackidx].define->expanding = 0;
1285 if(!bufferstack[bufferstackidx].should_pop)
1288 fprintf(ppy_out, "# %d \"%s\" 2\n", bufferstack[bufferstackidx].line_number, bufferstack[bufferstackidx].filename);
1290 /* We have EOF, check the include logic */
1291 if(pp_incl_state.state == 2 && !pp_incl_state.seen_junk && pp_incl_state.ppp)
1293 pp_entry_t *ppp = pplookup(pp_incl_state.ppp);
1296 includelogicentry_t *iep = pp_xmalloc(sizeof(includelogicentry_t));
1302 iep->filename = bufferstack[bufferstackidx].include_filename;
1304 iep->next = pp_includelogiclist;
1306 iep->next->prev = iep;
1307 pp_includelogiclist = iep;
1309 fprintf(stderr, "pop_buffer: %s:%d: includelogic added, include_ppp='%s', file='%s'\n", bufferstack[bufferstackidx].filename, bufferstack[bufferstackidx].line_number, pp_incl_state.ppp, iep->filename);
1312 free(bufferstack[bufferstackidx].include_filename);
1314 free(pp_incl_state.ppp);
1315 pp_incl_state = bufferstack[bufferstackidx].incl;
1318 pp_status.line_number = bufferstack[bufferstackidx].line_number;
1319 pp_status.char_number = bufferstack[bufferstackidx].char_number;
1320 pp_status.input = bufferstack[bufferstackidx].filename;
1321 ncontinuations = bufferstack[bufferstackidx].ncontinuations;
1325 printf("pop_buffer(%d): %p %p (%d, %d, %d) %p %d\n",
1327 bufferstack[bufferstackidx].bufferstate,
1328 bufferstack[bufferstackidx].define,
1329 bufferstack[bufferstackidx].line_number,
1330 bufferstack[bufferstackidx].char_number,
1331 bufferstack[bufferstackidx].if_depth,
1332 bufferstack[bufferstackidx].filename,
1333 bufferstack[bufferstackidx].should_pop);
1335 ppy__switch_to_buffer(bufferstack[bufferstackidx].bufferstate);
1337 if(bufferstack[bufferstackidx].should_pop)
1339 if(yy_current_state() == pp_macexp)
1340 macro_add_expansion();
1342 pp_internal_error(__FILE__, __LINE__, "Pop buffer and state without macro expansion state");
1346 return &bufferstack[bufferstackidx];
1351 *-------------------------------------------------------------------------
1352 * Macro nestng support
1353 *-------------------------------------------------------------------------
1355 static void push_macro(pp_entry_t *ppp)
1357 if(macexpstackidx >= MAXMACEXPSTACK)
1359 ppy_error("Too many nested macros");
1363 macexpstack[macexpstackidx] = pp_xmalloc(sizeof(macexpstack[0][0]));
1364 if(!macexpstack[macexpstackidx])
1366 memset( macexpstack[macexpstackidx], 0, sizeof(macexpstack[0][0]));
1367 macexpstack[macexpstackidx]->ppp = ppp;
1371 static macexpstackentry_t *top_macro(void)
1373 return macexpstackidx > 0 ? macexpstack[macexpstackidx-1] : NULL;
1376 static macexpstackentry_t *pop_macro(void)
1378 if(macexpstackidx <= 0)
1379 pp_internal_error(__FILE__, __LINE__, "Macro expansion stack underflow\n");
1380 return macexpstack[--macexpstackidx];
1383 static void free_macro(macexpstackentry_t *mep)
1387 for(i = 0; i < mep->nargs; i++)
1395 static void add_text_to_macro(const char *text, int len)
1397 macexpstackentry_t *mep = top_macro();
1399 assert(mep->ppp->expanding == 0);
1401 if(mep->curargalloc - mep->curargsize <= len+1) /* +1 for '\0' */
1404 int new_alloc = mep->curargalloc + (ALLOCBLOCKSIZE > len+1) ? ALLOCBLOCKSIZE : len+1;
1405 new_curarg = pp_xrealloc(mep->curarg, new_alloc * sizeof(mep->curarg[0]));
1408 mep->curarg = new_curarg;
1409 mep->curargalloc = new_alloc;
1411 memcpy(mep->curarg + mep->curargsize, text, len);
1412 mep->curargsize += len;
1413 mep->curarg[mep->curargsize] = '\0';
1416 static void macro_add_arg(int last)
1420 char **new_args, **new_ppargs;
1422 macexpstackentry_t *mep = top_macro();
1424 assert(mep->ppp->expanding == 0);
1426 new_args = pp_xrealloc(mep->args, (mep->nargs+1) * sizeof(mep->args[0]));
1429 mep->args = new_args;
1431 new_ppargs = pp_xrealloc(mep->ppargs, (mep->nargs+1) * sizeof(mep->ppargs[0]));
1434 mep->ppargs = new_ppargs;
1436 new_nnls = pp_xrealloc(mep->nnls, (mep->nargs+1) * sizeof(mep->nnls[0]));
1439 mep->nnls = new_nnls;
1441 mep->args[mep->nargs] = pp_xstrdup(mep->curarg ? mep->curarg : "");
1442 if(!mep->args[mep->nargs])
1444 cptr = mep->args[mep->nargs]-1;
1445 while((cptr = strchr(cptr+1, '\n')))
1449 mep->nnls[mep->nargs] = nnl;
1452 mep->curargalloc = mep->curargsize = 0;
1456 fprintf(stderr, "macro_add_arg: %s:%d: %d -> '%s'\n",
1458 pp_status.line_number,
1460 mep->args[mep->nargs-1]);
1462 /* Each macro argument must be expanded to cope with stingize */
1463 if(last || mep->args[mep->nargs-1][0])
1465 yy_push_state(pp_macexp);
1466 push_buffer(NULL, NULL, NULL, last ? 2 : 1);
1467 yy_scan_string(mep->args[mep->nargs-1]);
1468 /*mep->bufferstackidx = bufferstackidx; But not nested! */
1472 static void macro_add_expansion(void)
1474 macexpstackentry_t *mep = top_macro();
1476 assert(mep->ppp->expanding == 0);
1478 mep->ppargs[mep->nargs-1] = pp_xstrdup(mep->curarg ? mep->curarg : "");
1480 mep->curargalloc = mep->curargsize = 0;
1484 fprintf(stderr, "macro_add_expansion: %s:%d: %d -> '%s'\n",
1486 pp_status.line_number,
1488 mep->ppargs[mep->nargs-1] ? mep->ppargs[mep->nargs-1] : "");
1493 *-------------------------------------------------------------------------
1495 *-------------------------------------------------------------------------
1497 static void put_buffer(const char *s, int len)
1500 add_text_to_macro(s, len);
1502 fwrite(s, 1, len, ppy_out);
1507 *-------------------------------------------------------------------------
1508 * Include management
1509 *-------------------------------------------------------------------------
1511 void pp_do_include(char *fname, int type)
1515 includelogicentry_t *iep;
1517 for(iep = pp_includelogiclist; iep; iep = iep->next)
1519 if(!strcmp(iep->filename, fname))
1522 * We are done. The file was included before.
1523 * If the define was deleted, then this entry would have
1534 ppy_error("Empty include filename");
1538 /* Undo the effect of the quotation */
1541 if((ppy_in = pp_open_include(fname+1, type ? pp_status.input : NULL, &newpath)) == NULL)
1543 ppy_error("Unable to open include file %s", fname+1);
1547 fname[n-1] = *fname; /* Redo the quotes */
1548 push_buffer(NULL, newpath, fname, 0);
1549 pp_incl_state.seen_junk = 0;
1550 pp_incl_state.state = 0;
1551 pp_incl_state.ppp = NULL;
1554 fprintf(stderr, "pp_do_include: %s:%d: include_state=%d, include_ppp='%s', include_ifdepth=%d\n",
1555 pp_status.input, pp_status.line_number, pp_incl_state.state, pp_incl_state.ppp, pp_incl_state.ifdepth);
1556 ppy__switch_to_buffer(ppy__create_buffer(ppy_in, YY_BUF_SIZE));
1558 fprintf(ppy_out, "# 1 \"%s\" 1%s\n", newpath, type ? "" : " 3");
1562 *-------------------------------------------------------------------------
1563 * Push/pop preprocessor ignore state when processing conditionals
1565 *-------------------------------------------------------------------------
1567 void pp_push_ignore_state(void)
1569 yy_push_state(pp_ignore);
1572 void pp_pop_ignore_state(void)