]>
Commit | Line | Data |
---|---|---|
45b966db ZW |
1 | /* CPP Library - lexical analysis. |
2 | Copyright (C) 2000 Free Software Foundation, Inc. | |
3 | Contributed by Per Bothner, 1994-95. | |
4 | Based on CCCP program by Paul Rubin, June 1986 | |
5 | Adapted to ANSI C, Richard Stallman, Jan 1987 | |
6 | Broken out to separate file, Zack Weinberg, Mar 2000 | |
c5a04734 | 7 | Single-pass line tokenization by Neil Booth, April 2000 |
45b966db ZW |
8 | |
9 | This program is free software; you can redistribute it and/or modify it | |
10 | under the terms of the GNU General Public License as published by the | |
11 | Free Software Foundation; either version 2, or (at your option) any | |
12 | later version. | |
13 | ||
14 | This program is distributed in the hope that it will be useful, | |
15 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
17 | GNU General Public License for more details. | |
18 | ||
19 | You should have received a copy of the GNU General Public License | |
20 | along with this program; if not, write to the Free Software | |
21 | Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ | |
22 | ||
23 | #include "config.h" | |
24 | #include "system.h" | |
25 | #include "intl.h" | |
e38992e8 | 26 | #include "hashtab.h" |
45b966db ZW |
27 | #include "cpplib.h" |
28 | #include "cpphash.h" | |
29 | ||
ff2b53ef ZW |
30 | #define PEEKBUF(BUFFER, N) \ |
31 | ((BUFFER)->rlimit - (BUFFER)->cur > (N) ? (BUFFER)->cur[N] : EOF) | |
32 | #define GETBUF(BUFFER) \ | |
33 | ((BUFFER)->cur < (BUFFER)->rlimit ? *(BUFFER)->cur++ : EOF) | |
34 | #define FORWARDBUF(BUFFER, N) ((BUFFER)->cur += (N)) | |
35 | ||
36 | #define PEEKN(N) PEEKBUF (CPP_BUFFER (pfile), N) | |
37 | #define FORWARD(N) FORWARDBUF (CPP_BUFFER (pfile), (N)) | |
38 | #define GETC() GETBUF (CPP_BUFFER (pfile)) | |
39 | #define PEEKC() PEEKBUF (CPP_BUFFER (pfile), 0) | |
45b966db ZW |
40 | |
41 | static void skip_block_comment PARAMS ((cpp_reader *)); | |
42 | static void skip_line_comment PARAMS ((cpp_reader *)); | |
43 | static int maybe_macroexpand PARAMS ((cpp_reader *, long)); | |
44 | static int skip_comment PARAMS ((cpp_reader *, int)); | |
45 | static int copy_comment PARAMS ((cpp_reader *, int)); | |
46 | static void skip_string PARAMS ((cpp_reader *, int)); | |
47 | static void parse_string PARAMS ((cpp_reader *, int)); | |
48 | static U_CHAR *find_position PARAMS ((U_CHAR *, U_CHAR *, unsigned long *)); | |
49 | static int null_cleanup PARAMS ((cpp_buffer *, cpp_reader *)); | |
64aaf407 | 50 | static void null_warning PARAMS ((cpp_reader *, unsigned int)); |
45b966db | 51 | |
f2d5f0cc ZW |
52 | static void safe_fwrite PARAMS ((cpp_reader *, const U_CHAR *, |
53 | size_t, FILE *)); | |
1368ee70 ZW |
54 | static void output_line_command PARAMS ((cpp_reader *, cpp_printer *, |
55 | unsigned int)); | |
56 | static void bump_column PARAMS ((cpp_printer *, unsigned int, | |
57 | unsigned int)); | |
c5a04734 | 58 | static void expand_name_space PARAMS ((cpp_toklist *, unsigned int)); |
1368ee70 ZW |
59 | static void expand_token_space PARAMS ((cpp_toklist *)); |
60 | static void init_token_list PARAMS ((cpp_reader *, cpp_toklist *, int)); | |
9e62c811 ZW |
61 | static void pedantic_whitespace PARAMS ((cpp_reader *, U_CHAR *, |
62 | unsigned int)); | |
f2d5f0cc | 63 | |
c5a04734 ZW |
64 | #define auto_expand_name_space(list) \ |
65 | expand_name_space ((list), (list)->name_cap / 2) | |
66 | ||
45b966db ZW |
67 | /* Re-allocates PFILE->token_buffer so it will hold at least N more chars. */ |
68 | ||
69 | void | |
70 | _cpp_grow_token_buffer (pfile, n) | |
71 | cpp_reader *pfile; | |
72 | long n; | |
73 | { | |
74 | long old_written = CPP_WRITTEN (pfile); | |
75 | pfile->token_buffer_size = n + 2 * pfile->token_buffer_size; | |
76 | pfile->token_buffer = (U_CHAR *) | |
77 | xrealloc(pfile->token_buffer, pfile->token_buffer_size); | |
78 | CPP_SET_WRITTEN (pfile, old_written); | |
79 | } | |
80 | ||
81 | static int | |
82 | null_cleanup (pbuf, pfile) | |
83 | cpp_buffer *pbuf ATTRIBUTE_UNUSED; | |
84 | cpp_reader *pfile ATTRIBUTE_UNUSED; | |
85 | { | |
86 | return 0; | |
87 | } | |
88 | ||
89 | /* Allocate a new cpp_buffer for PFILE, and push it on the input buffer stack. | |
90 | If BUFFER != NULL, then use the LENGTH characters in BUFFER | |
91 | as the new input buffer. | |
92 | Return the new buffer, or NULL on failure. */ | |
93 | ||
94 | cpp_buffer * | |
95 | cpp_push_buffer (pfile, buffer, length) | |
96 | cpp_reader *pfile; | |
97 | const U_CHAR *buffer; | |
98 | long length; | |
99 | { | |
100 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
101 | cpp_buffer *new; | |
102 | if (++pfile->buffer_stack_depth == CPP_STACK_MAX) | |
103 | { | |
104 | cpp_fatal (pfile, "macro or `#include' recursion too deep"); | |
105 | return NULL; | |
106 | } | |
107 | ||
108 | new = (cpp_buffer *) xcalloc (1, sizeof (cpp_buffer)); | |
109 | ||
110 | new->if_stack = pfile->if_stack; | |
111 | new->cleanup = null_cleanup; | |
112 | new->buf = new->cur = buffer; | |
ff2b53ef | 113 | new->rlimit = buffer + length; |
45b966db | 114 | new->prev = buf; |
ff2b53ef | 115 | new->mark = NULL; |
45b966db ZW |
116 | new->line_base = NULL; |
117 | ||
118 | CPP_BUFFER (pfile) = new; | |
119 | return new; | |
120 | } | |
121 | ||
122 | cpp_buffer * | |
123 | cpp_pop_buffer (pfile) | |
124 | cpp_reader *pfile; | |
125 | { | |
126 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
127 | if (ACTIVE_MARK_P (pfile)) | |
128 | cpp_ice (pfile, "mark active in cpp_pop_buffer"); | |
129 | (*buf->cleanup) (buf, pfile); | |
130 | CPP_BUFFER (pfile) = CPP_PREV_BUFFER (buf); | |
131 | free (buf); | |
132 | pfile->buffer_stack_depth--; | |
133 | return CPP_BUFFER (pfile); | |
134 | } | |
135 | ||
f2d5f0cc ZW |
136 | /* Deal with the annoying semantics of fwrite. */ |
137 | static void | |
138 | safe_fwrite (pfile, buf, len, fp) | |
139 | cpp_reader *pfile; | |
140 | const U_CHAR *buf; | |
141 | size_t len; | |
142 | FILE *fp; | |
143 | { | |
144 | size_t count; | |
45b966db | 145 | |
f2d5f0cc ZW |
146 | while (len) |
147 | { | |
148 | count = fwrite (buf, 1, len, fp); | |
149 | if (count == 0) | |
150 | goto error; | |
151 | len -= count; | |
152 | buf += count; | |
153 | } | |
154 | return; | |
155 | ||
156 | error: | |
157 | cpp_notice_from_errno (pfile, CPP_OPTION (pfile, out_fname)); | |
158 | } | |
159 | ||
160 | /* Notify the compiler proper that the current line number has jumped, | |
161 | or the current file name has changed. */ | |
162 | ||
163 | static void | |
1368ee70 | 164 | output_line_command (pfile, print, line) |
45b966db | 165 | cpp_reader *pfile; |
f2d5f0cc | 166 | cpp_printer *print; |
1368ee70 | 167 | unsigned int line; |
45b966db | 168 | { |
1368ee70 | 169 | cpp_buffer *ip = cpp_file_buffer (pfile); |
f2d5f0cc ZW |
170 | enum { same = 0, enter, leave, rname } change; |
171 | static const char * const codes[] = { "", " 1", " 2", "" }; | |
172 | ||
173 | if (CPP_OPTION (pfile, no_line_commands)) | |
174 | return; | |
175 | ||
f2d5f0cc ZW |
176 | /* Determine whether the current filename has changed, and if so, |
177 | how. 'nominal_fname' values are unique, so they can be compared | |
178 | by comparing pointers. */ | |
179 | if (ip->nominal_fname == print->last_fname) | |
180 | change = same; | |
181 | else | |
45b966db | 182 | { |
f2d5f0cc ZW |
183 | if (pfile->buffer_stack_depth == print->last_bsd) |
184 | change = rname; | |
185 | else | |
45b966db | 186 | { |
f2d5f0cc ZW |
187 | if (pfile->buffer_stack_depth > print->last_bsd) |
188 | change = enter; | |
189 | else | |
190 | change = leave; | |
191 | print->last_bsd = pfile->buffer_stack_depth; | |
45b966db | 192 | } |
f2d5f0cc | 193 | print->last_fname = ip->nominal_fname; |
45b966db | 194 | } |
f2d5f0cc ZW |
195 | /* If the current file has not changed, we can output a few newlines |
196 | instead if we want to increase the line number by a small amount. | |
197 | We cannot do this if print->lineno is zero, because that means we | |
198 | haven't output any line commands yet. (The very first line | |
199 | command output is a `same_file' command.) */ | |
200 | if (change == same && print->lineno != 0 | |
201 | && line >= print->lineno && line < print->lineno + 8) | |
45b966db | 202 | { |
f2d5f0cc | 203 | while (line > print->lineno) |
45b966db | 204 | { |
f2d5f0cc ZW |
205 | putc ('\n', print->outf); |
206 | print->lineno++; | |
45b966db | 207 | } |
f2d5f0cc | 208 | return; |
45b966db | 209 | } |
f2d5f0cc ZW |
210 | |
211 | #ifndef NO_IMPLICIT_EXTERN_C | |
212 | if (CPP_OPTION (pfile, cplusplus)) | |
213 | fprintf (print->outf, "# %u \"%s\"%s%s%s\n", line, ip->nominal_fname, | |
214 | codes[change], | |
215 | ip->system_header_p ? " 3" : "", | |
216 | (ip->system_header_p == 2) ? " 4" : ""); | |
217 | else | |
218 | #endif | |
219 | fprintf (print->outf, "# %u \"%s\"%s%s\n", line, ip->nominal_fname, | |
220 | codes[change], | |
221 | ip->system_header_p ? " 3" : ""); | |
222 | print->lineno = line; | |
223 | } | |
224 | ||
225 | /* Write the contents of the token_buffer to the output stream, and | |
226 | clear the token_buffer. Also handles generating line commands and | |
227 | keeping track of file transitions. */ | |
228 | ||
229 | void | |
230 | cpp_output_tokens (pfile, print) | |
231 | cpp_reader *pfile; | |
232 | cpp_printer *print; | |
233 | { | |
1368ee70 ZW |
234 | cpp_buffer *ip; |
235 | ||
f6fab919 ZW |
236 | if (CPP_WRITTEN (pfile) - print->written) |
237 | { | |
238 | if (CPP_PWRITTEN (pfile)[-1] == '\n' && print->lineno) | |
239 | print->lineno++; | |
240 | safe_fwrite (pfile, pfile->token_buffer, | |
241 | CPP_WRITTEN (pfile) - print->written, print->outf); | |
242 | } | |
1368ee70 ZW |
243 | |
244 | ip = cpp_file_buffer (pfile); | |
245 | if (ip) | |
246 | output_line_command (pfile, print, CPP_BUF_LINE (ip)); | |
247 | ||
f2d5f0cc | 248 | CPP_SET_WRITTEN (pfile, print->written); |
45b966db ZW |
249 | } |
250 | ||
1368ee70 ZW |
251 | /* Helper for cpp_output_list - increases the column number to match |
252 | what we expect it to be. */ | |
253 | ||
254 | static void | |
255 | bump_column (print, from, to) | |
256 | cpp_printer *print; | |
257 | unsigned int from, to; | |
258 | { | |
259 | unsigned int tabs, spcs; | |
260 | unsigned int delta = to - from; | |
261 | ||
262 | /* Only if FROM is 0, advance by tabs. */ | |
263 | if (from == 0) | |
264 | tabs = delta / 8, spcs = delta % 8; | |
265 | else | |
266 | tabs = 0, spcs = delta; | |
267 | ||
268 | while (tabs--) putc ('\t', print->outf); | |
269 | while (spcs--) putc (' ', print->outf); | |
270 | } | |
271 | ||
272 | /* Write out the list L onto pfile->token_buffer. This function is | |
273 | incomplete: | |
274 | ||
275 | 1) pfile->token_buffer is not going to continue to exist. | |
276 | 2) At the moment, tokens don't carry the information described | |
277 | in cpplib.h; they are all strings. | |
278 | 3) The list has to be a complete line, and has to be written starting | |
279 | at the beginning of a line. */ | |
280 | ||
281 | void | |
282 | cpp_output_list (pfile, print, list) | |
283 | cpp_reader *pfile; | |
284 | cpp_printer *print; | |
285 | const cpp_toklist *list; | |
286 | { | |
287 | unsigned int i; | |
288 | unsigned int curcol = 1; | |
289 | ||
290 | /* XXX Probably does not do what is intended. */ | |
291 | if (print->lineno != list->line) | |
292 | output_line_command (pfile, print, list->line); | |
293 | ||
294 | for (i = 0; i < list->tokens_used; i++) | |
295 | { | |
296 | if (list->tokens[i].type == CPP_VSPACE) | |
297 | { | |
298 | output_line_command (pfile, print, list->tokens[i].aux); | |
299 | continue; | |
300 | } | |
301 | ||
302 | if (curcol < list->tokens[i].col) | |
303 | { | |
304 | /* Insert space to bring the column to what it should be. */ | |
305 | bump_column (print, curcol - 1, list->tokens[i].col); | |
306 | curcol = list->tokens[i].col; | |
307 | } | |
308 | /* XXX We may have to insert space to prevent an accidental | |
309 | token paste. */ | |
310 | safe_fwrite (pfile, list->namebuf + list->tokens[i].val.name.offset, | |
311 | list->tokens[i].val.name.len, print->outf); | |
312 | curcol += list->tokens[i].val.name.len; | |
313 | } | |
314 | } | |
315 | ||
f2d5f0cc ZW |
316 | /* Scan a string (which may have escape marks), perform macro expansion, |
317 | and write the result to the token_buffer. */ | |
45b966db ZW |
318 | |
319 | void | |
f2d5f0cc | 320 | _cpp_expand_to_buffer (pfile, buf, length) |
45b966db ZW |
321 | cpp_reader *pfile; |
322 | const U_CHAR *buf; | |
323 | int length; | |
324 | { | |
f2d5f0cc ZW |
325 | cpp_buffer *ip; |
326 | enum cpp_ttype token; | |
f6fab919 | 327 | U_CHAR *buf1; |
45b966db ZW |
328 | |
329 | if (length < 0) | |
330 | { | |
331 | cpp_ice (pfile, "length < 0 in cpp_expand_to_buffer"); | |
332 | return; | |
333 | } | |
334 | ||
f6fab919 ZW |
335 | /* Copy the buffer, because it might be in an unsafe place - for |
336 | example, a sequence on the token_buffer, where the pointers will | |
337 | be invalidated if we enlarge the token_buffer. */ | |
338 | buf1 = alloca (length); | |
339 | memcpy (buf1, buf, length); | |
340 | ||
45b966db | 341 | /* Set up the input on the input stack. */ |
f6fab919 | 342 | ip = cpp_push_buffer (pfile, buf1, length); |
45b966db ZW |
343 | if (ip == NULL) |
344 | return; | |
345 | ip->has_escapes = 1; | |
346 | ||
347 | /* Scan the input, create the output. */ | |
f2d5f0cc ZW |
348 | for (;;) |
349 | { | |
350 | token = cpp_get_token (pfile); | |
351 | if (token == CPP_EOF) | |
352 | break; | |
353 | if (token == CPP_POP && CPP_BUFFER (pfile) == ip) | |
354 | { | |
355 | cpp_pop_buffer (pfile); | |
356 | break; | |
357 | } | |
358 | } | |
45b966db ZW |
359 | } |
360 | ||
f2d5f0cc ZW |
361 | /* Scan until CPP_BUFFER (PFILE) is exhausted, discarding output. |
362 | Then pop the buffer. */ | |
363 | ||
364 | void | |
365 | cpp_scan_buffer_nooutput (pfile) | |
366 | cpp_reader *pfile; | |
367 | { | |
368 | cpp_buffer *buffer = CPP_BUFFER (pfile); | |
369 | enum cpp_ttype token; | |
370 | unsigned int old_written = CPP_WRITTEN (pfile); | |
371 | /* In no-output mode, we can ignore everything but directives. */ | |
372 | for (;;) | |
373 | { | |
374 | if (! pfile->only_seen_white) | |
375 | _cpp_skip_rest_of_line (pfile); | |
376 | token = cpp_get_token (pfile); | |
377 | if (token == CPP_EOF) | |
378 | break; | |
379 | if (token == CPP_POP && CPP_BUFFER (pfile) == buffer) | |
380 | { | |
381 | cpp_pop_buffer (pfile); | |
382 | break; | |
383 | } | |
384 | } | |
385 | CPP_SET_WRITTEN (pfile, old_written); | |
386 | } | |
387 | ||
388 | /* Scan until CPP_BUFFER (pfile) is exhausted, writing output to PRINT. | |
389 | Then pop the buffer. */ | |
390 | ||
391 | void | |
392 | cpp_scan_buffer (pfile, print) | |
393 | cpp_reader *pfile; | |
394 | cpp_printer *print; | |
395 | { | |
396 | cpp_buffer *buffer = CPP_BUFFER (pfile); | |
397 | enum cpp_ttype token; | |
398 | ||
399 | for (;;) | |
400 | { | |
401 | token = cpp_get_token (pfile); | |
402 | if ((token == CPP_POP && !CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
403 | || token == CPP_EOF || token == CPP_VSPACE | |
404 | /* XXX Temporary kluge - force flush after #include only */ | |
405 | || (token == CPP_DIRECTIVE | |
406 | && CPP_BUFFER (pfile)->nominal_fname != print->last_fname)) | |
407 | { | |
408 | cpp_output_tokens (pfile, print); | |
409 | if (token == CPP_EOF) | |
410 | return; | |
411 | if (token == CPP_POP && CPP_BUFFER (pfile) == buffer) | |
412 | { | |
413 | cpp_pop_buffer (pfile); | |
414 | return; | |
415 | } | |
416 | } | |
417 | } | |
418 | } | |
419 | ||
45b966db ZW |
420 | /* Return the topmost cpp_buffer that corresponds to a file (not a macro). */ |
421 | ||
422 | cpp_buffer * | |
423 | cpp_file_buffer (pfile) | |
424 | cpp_reader *pfile; | |
425 | { | |
426 | cpp_buffer *ip; | |
427 | ||
428 | for (ip = CPP_BUFFER (pfile); ip; ip = CPP_PREV_BUFFER (ip)) | |
429 | if (ip->ihash != NULL) | |
430 | return ip; | |
431 | return NULL; | |
432 | } | |
433 | ||
1368ee70 ZW |
434 | /* Token-buffer helper functions. */ |
435 | ||
436 | /* Expand a token list's string space. */ | |
437 | static void | |
c5a04734 | 438 | expand_name_space (list, len) |
1368ee70 | 439 | cpp_toklist *list; |
c5a04734 ZW |
440 | unsigned int len; |
441 | { | |
442 | list->name_cap += len; | |
443 | list->namebuf = (unsigned char *) xrealloc (list->namebuf, list->name_cap); | |
1368ee70 ZW |
444 | } |
445 | ||
446 | /* Expand the number of tokens in a list. */ | |
447 | static void | |
448 | expand_token_space (list) | |
449 | cpp_toklist *list; | |
450 | { | |
451 | list->tokens_cap *= 2; | |
452 | list->tokens = (cpp_token *) | |
c5a04734 ZW |
453 | xrealloc (list->tokens - 1, (list->tokens_cap + 1) * sizeof (cpp_token)); |
454 | list->tokens++; /* Skip the dummy. */ | |
1368ee70 ZW |
455 | } |
456 | ||
c5a04734 ZW |
457 | /* Initialize a token list. We allocate an extra token in front of |
458 | the token list, as this allows us to always peek at the previous | |
459 | token without worrying about underflowing the list. */ | |
1368ee70 ZW |
460 | static void |
461 | init_token_list (pfile, list, recycle) | |
462 | cpp_reader *pfile; | |
463 | cpp_toklist *list; | |
464 | int recycle; | |
465 | { | |
c5a04734 ZW |
466 | /* Recycling a used list saves 3 free-malloc pairs. */ |
467 | if (!recycle) | |
1368ee70 | 468 | { |
c5a04734 ZW |
469 | /* Initialize token space. Put a dummy token before the start |
470 | that will fail matches. */ | |
471 | list->tokens_cap = 256; /* 4K's worth. */ | |
1368ee70 | 472 | list->tokens = (cpp_token *) |
c5a04734 ZW |
473 | xmalloc ((list->tokens_cap + 1) * sizeof (cpp_token)); |
474 | list->tokens[0].type = CPP_EOF; | |
475 | list->tokens++; | |
1368ee70 | 476 | |
c5a04734 | 477 | /* Initialize name space. */ |
1368ee70 | 478 | list->name_cap = 1024; |
1368ee70 | 479 | list->namebuf = (unsigned char *) xmalloc (list->name_cap); |
c5a04734 ZW |
480 | |
481 | /* Only create a comment space on demand. */ | |
482 | list->comments_cap = 0; | |
483 | list->comments = 0; | |
1368ee70 ZW |
484 | } |
485 | ||
c5a04734 ZW |
486 | list->tokens_used = 0; |
487 | list->name_used = 0; | |
488 | list->comments_used = 0; | |
9e62c811 ZW |
489 | if (pfile->buffer) |
490 | list->line = pfile->buffer->lineno; | |
1368ee70 ZW |
491 | list->dir_handler = 0; |
492 | list->dir_flags = 0; | |
493 | } | |
494 | ||
495 | /* Scan an entire line and create a token list for it. Does not | |
496 | macro-expand or execute directives. */ | |
497 | ||
498 | void | |
499 | _cpp_scan_line (pfile, list) | |
500 | cpp_reader *pfile; | |
501 | cpp_toklist *list; | |
502 | { | |
503 | int i, col; | |
504 | long written, len; | |
505 | enum cpp_ttype type; | |
9e62c811 | 506 | int space_before; |
1368ee70 ZW |
507 | |
508 | init_token_list (pfile, list, 1); | |
509 | ||
510 | written = CPP_WRITTEN (pfile); | |
511 | i = 0; | |
9e62c811 | 512 | space_before = 0; |
1368ee70 ZW |
513 | for (;;) |
514 | { | |
515 | col = CPP_BUFFER (pfile)->cur - CPP_BUFFER (pfile)->line_base; | |
516 | type = _cpp_lex_token (pfile); | |
517 | len = CPP_WRITTEN (pfile) - written; | |
518 | CPP_SET_WRITTEN (pfile, written); | |
519 | if (type == CPP_HSPACE) | |
9e62c811 ZW |
520 | { |
521 | if (CPP_PEDANTIC (pfile)) | |
522 | pedantic_whitespace (pfile, pfile->token_buffer + written, len); | |
523 | space_before = 1; | |
524 | continue; | |
525 | } | |
0f89df67 ZW |
526 | else if (type == CPP_COMMENT) |
527 | /* Only happens when processing -traditional macro definitions. | |
528 | Do not give this a token entry, but do not change space_before | |
529 | either. */ | |
530 | continue; | |
1368ee70 ZW |
531 | |
532 | if (list->tokens_used >= list->tokens_cap) | |
533 | expand_token_space (list); | |
534 | if (list->name_used + len >= list->name_cap) | |
bb1ec1d7 | 535 | expand_name_space (list, list->name_used + len + 1 - list->name_cap); |
1368ee70 | 536 | |
9e62c811 ZW |
537 | if (type == CPP_MACRO) |
538 | type = CPP_NAME; | |
539 | ||
1368ee70 ZW |
540 | list->tokens_used++; |
541 | list->tokens[i].type = type; | |
542 | list->tokens[i].col = col; | |
c5a04734 | 543 | list->tokens[i].flags = space_before ? PREV_WHITESPACE : 0; |
9e62c811 | 544 | |
1368ee70 ZW |
545 | if (type == CPP_VSPACE) |
546 | break; | |
547 | ||
548 | list->tokens[i].val.name.len = len; | |
549 | list->tokens[i].val.name.offset = list->name_used; | |
550 | memcpy (list->namebuf + list->name_used, CPP_PWRITTEN (pfile), len); | |
551 | list->name_used += len; | |
552 | i++; | |
9e62c811 | 553 | space_before = 0; |
1368ee70 ZW |
554 | } |
555 | list->tokens[i].aux = CPP_BUFFER (pfile)->lineno + 1; | |
9e62c811 ZW |
556 | |
557 | /* XXX Temporary kluge: put back the newline. */ | |
558 | FORWARD(-1); | |
1368ee70 ZW |
559 | } |
560 | ||
561 | ||
45b966db ZW |
562 | /* Skip a C-style block comment. We know it's a comment, and point is |
563 | at the second character of the starter. */ | |
564 | static void | |
565 | skip_block_comment (pfile) | |
566 | cpp_reader *pfile; | |
567 | { | |
3a2b2c7a | 568 | unsigned int line, col; |
61474454 | 569 | const U_CHAR *limit, *cur; |
45b966db ZW |
570 | |
571 | FORWARD(1); | |
3a2b2c7a ZW |
572 | line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
573 | col = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
61474454 NB |
574 | limit = CPP_BUFFER (pfile)->rlimit; |
575 | cur = CPP_BUFFER (pfile)->cur; | |
576 | ||
577 | while (cur < limit) | |
45b966db | 578 | { |
61474454 NB |
579 | char c = *cur++; |
580 | if (c == '\n' || c == '\r') | |
45b966db ZW |
581 | { |
582 | /* \r cannot be a macro escape marker here. */ | |
583 | if (!ACTIVE_MARK_P (pfile)) | |
61474454 NB |
584 | CPP_BUMP_LINE_CUR (pfile, cur); |
585 | } | |
586 | else if (c == '*') | |
587 | { | |
588 | /* Check for teminator. */ | |
589 | if (cur < limit && *cur == '/') | |
590 | goto out; | |
591 | ||
592 | /* Warn about comment starter embedded in comment. */ | |
593 | if (cur[-2] == '/' && CPP_OPTION (pfile, warn_comments)) | |
594 | cpp_warning_with_line (pfile, CPP_BUFFER (pfile)->lineno, | |
595 | cur - CPP_BUFFER (pfile)->line_base, | |
596 | "'/*' within comment"); | |
45b966db | 597 | } |
45b966db | 598 | } |
61474454 NB |
599 | |
600 | cpp_error_with_line (pfile, line, col, "unterminated comment"); | |
601 | cur--; | |
602 | out: | |
603 | CPP_BUFFER (pfile)->cur = cur + 1; | |
45b966db ZW |
604 | } |
605 | ||
606 | /* Skip a C++/Chill line comment. We know it's a comment, and point | |
607 | is at the second character of the initiator. */ | |
608 | static void | |
609 | skip_line_comment (pfile) | |
610 | cpp_reader *pfile; | |
611 | { | |
612 | FORWARD(1); | |
613 | for (;;) | |
614 | { | |
615 | int c = GETC (); | |
616 | ||
617 | /* We don't have to worry about EOF in here. */ | |
618 | if (c == '\n') | |
619 | { | |
620 | /* Don't consider final '\n' to be part of comment. */ | |
621 | FORWARD(-1); | |
622 | return; | |
623 | } | |
624 | else if (c == '\r') | |
625 | { | |
626 | /* \r cannot be a macro escape marker here. */ | |
627 | if (!ACTIVE_MARK_P (pfile)) | |
628 | CPP_BUMP_LINE (pfile); | |
ae79697b | 629 | if (CPP_OPTION (pfile, warn_comments)) |
45b966db ZW |
630 | cpp_warning (pfile, "backslash-newline within line comment"); |
631 | } | |
632 | } | |
633 | } | |
634 | ||
635 | /* Skip a comment - C, C++, or Chill style. M is the first character | |
636 | of the comment marker. If this really is a comment, skip to its | |
637 | end and return ' '. If this is not a comment, return M (which will | |
638 | be '/' or '-'). */ | |
639 | ||
640 | static int | |
641 | skip_comment (pfile, m) | |
642 | cpp_reader *pfile; | |
643 | int m; | |
644 | { | |
645 | if (m == '/' && PEEKC() == '*') | |
646 | { | |
647 | skip_block_comment (pfile); | |
648 | return ' '; | |
649 | } | |
650 | else if (m == '/' && PEEKC() == '/') | |
651 | { | |
652 | if (CPP_BUFFER (pfile)->system_header_p) | |
653 | { | |
654 | /* We silently allow C++ comments in system headers, irrespective | |
655 | of conformance mode, because lots of busted systems do that | |
656 | and trying to clean it up in fixincludes is a nightmare. */ | |
657 | skip_line_comment (pfile); | |
658 | return ' '; | |
659 | } | |
ae79697b | 660 | else if (CPP_OPTION (pfile, cplusplus_comments)) |
45b966db | 661 | { |
0f89df67 | 662 | if (! CPP_BUFFER (pfile)->warned_cplusplus_comments) |
45b966db | 663 | { |
0f89df67 ZW |
664 | if (CPP_WTRADITIONAL (pfile)) |
665 | cpp_pedwarn (pfile, | |
666 | "C++ style comments are not allowed in traditional C"); | |
667 | else if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile)) | |
668 | cpp_pedwarn (pfile, | |
669 | "C++ style comments are not allowed in ISO C89"); | |
670 | if (CPP_WTRADITIONAL (pfile) | |
671 | || (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile))) | |
672 | cpp_pedwarn (pfile, | |
45b966db ZW |
673 | "(this will be reported only once per input file)"); |
674 | CPP_BUFFER (pfile)->warned_cplusplus_comments = 1; | |
675 | } | |
676 | skip_line_comment (pfile); | |
677 | return ' '; | |
678 | } | |
679 | else | |
680 | return m; | |
681 | } | |
682 | else if (m == '-' && PEEKC() == '-' | |
ae79697b | 683 | && CPP_OPTION (pfile, chill)) |
45b966db ZW |
684 | { |
685 | skip_line_comment (pfile); | |
686 | return ' '; | |
687 | } | |
688 | else | |
689 | return m; | |
690 | } | |
691 | ||
692 | /* Identical to skip_comment except that it copies the comment into the | |
693 | token_buffer. This is used if !discard_comments. */ | |
694 | static int | |
695 | copy_comment (pfile, m) | |
696 | cpp_reader *pfile; | |
697 | int m; | |
698 | { | |
699 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
700 | const U_CHAR *limit; | |
701 | ||
702 | if (skip_comment (pfile, m) == m) | |
703 | return m; | |
704 | ||
705 | limit = CPP_BUFFER (pfile)->cur; | |
706 | CPP_RESERVE (pfile, limit - start + 2); | |
707 | CPP_PUTC_Q (pfile, m); | |
708 | for (; start <= limit; start++) | |
709 | if (*start != '\r') | |
710 | CPP_PUTC_Q (pfile, *start); | |
711 | ||
712 | return ' '; | |
713 | } | |
714 | ||
64aaf407 NB |
715 | static void |
716 | null_warning (pfile, count) | |
717 | cpp_reader *pfile; | |
718 | unsigned int count; | |
719 | { | |
720 | if (count == 1) | |
721 | cpp_warning (pfile, "embedded null character ignored"); | |
722 | else | |
723 | cpp_warning (pfile, "embedded null characters ignored"); | |
724 | } | |
725 | ||
45b966db ZW |
726 | /* Skip whitespace \-newline and comments. Does not macro-expand. */ |
727 | ||
728 | void | |
729 | _cpp_skip_hspace (pfile) | |
730 | cpp_reader *pfile; | |
731 | { | |
64aaf407 | 732 | unsigned int null_count = 0; |
45b966db | 733 | int c; |
64aaf407 | 734 | |
45b966db ZW |
735 | while (1) |
736 | { | |
737 | c = GETC(); | |
738 | if (c == EOF) | |
64aaf407 | 739 | goto out; |
45b966db ZW |
740 | else if (is_hspace(c)) |
741 | { | |
742 | if ((c == '\f' || c == '\v') && CPP_PEDANTIC (pfile)) | |
743 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
744 | c == '\f' ? "formfeed" : "vertical tab"); | |
64aaf407 NB |
745 | else if (c == '\0') |
746 | null_count++; | |
45b966db ZW |
747 | } |
748 | else if (c == '\r') | |
749 | { | |
750 | /* \r is a backslash-newline marker if !has_escapes, and | |
751 | a deletable-whitespace or no-reexpansion marker otherwise. */ | |
752 | if (CPP_BUFFER (pfile)->has_escapes) | |
753 | { | |
754 | if (PEEKC() == ' ') | |
755 | FORWARD(1); | |
756 | else | |
757 | break; | |
758 | } | |
759 | else | |
760 | CPP_BUMP_LINE (pfile); | |
761 | } | |
762 | else if (c == '/' || c == '-') | |
763 | { | |
764 | c = skip_comment (pfile, c); | |
765 | if (c != ' ') | |
766 | break; | |
767 | } | |
768 | else | |
769 | break; | |
770 | } | |
771 | FORWARD(-1); | |
64aaf407 NB |
772 | out: |
773 | if (null_count) | |
774 | null_warning (pfile, null_count); | |
45b966db ZW |
775 | } |
776 | ||
777 | /* Read and discard the rest of the current line. */ | |
778 | ||
779 | void | |
780 | _cpp_skip_rest_of_line (pfile) | |
781 | cpp_reader *pfile; | |
782 | { | |
783 | for (;;) | |
784 | { | |
785 | int c = GETC(); | |
786 | switch (c) | |
787 | { | |
788 | case '\n': | |
789 | FORWARD(-1); | |
790 | case EOF: | |
791 | return; | |
792 | ||
793 | case '\r': | |
794 | if (! CPP_BUFFER (pfile)->has_escapes) | |
795 | CPP_BUMP_LINE (pfile); | |
796 | break; | |
797 | ||
798 | case '\'': | |
799 | case '\"': | |
800 | skip_string (pfile, c); | |
801 | break; | |
802 | ||
803 | case '/': | |
804 | case '-': | |
805 | skip_comment (pfile, c); | |
806 | break; | |
807 | ||
808 | case '\f': | |
809 | case '\v': | |
810 | if (CPP_PEDANTIC (pfile)) | |
811 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
812 | c == '\f' ? "formfeed" : "vertical tab"); | |
813 | break; | |
814 | ||
815 | } | |
816 | } | |
817 | } | |
818 | ||
819 | /* Parse an identifier starting with C. */ | |
820 | ||
821 | void | |
822 | _cpp_parse_name (pfile, c) | |
823 | cpp_reader *pfile; | |
824 | int c; | |
825 | { | |
826 | for (;;) | |
827 | { | |
828 | if (! is_idchar(c)) | |
829 | { | |
830 | FORWARD (-1); | |
831 | break; | |
832 | } | |
833 | ||
834 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
835 | cpp_pedwarn (pfile, "`$' in identifier"); | |
836 | ||
837 | CPP_RESERVE(pfile, 2); /* One more for final NUL. */ | |
838 | CPP_PUTC_Q (pfile, c); | |
839 | c = GETC(); | |
840 | if (c == EOF) | |
841 | break; | |
842 | } | |
45b966db ZW |
843 | return; |
844 | } | |
845 | ||
846 | /* Parse and skip over a string starting with C. A single quoted | |
847 | string is treated like a double -- some programs (e.g., troff) are | |
848 | perverse this way. (However, a single quoted string is not allowed | |
849 | to extend over multiple lines.) */ | |
850 | static void | |
851 | skip_string (pfile, c) | |
852 | cpp_reader *pfile; | |
853 | int c; | |
854 | { | |
3a2b2c7a | 855 | unsigned int start_line, start_column; |
64aaf407 | 856 | unsigned int null_count = 0; |
45b966db | 857 | |
3a2b2c7a ZW |
858 | start_line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
859 | start_column = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
45b966db ZW |
860 | while (1) |
861 | { | |
862 | int cc = GETC(); | |
863 | switch (cc) | |
864 | { | |
865 | case EOF: | |
866 | cpp_error_with_line (pfile, start_line, start_column, | |
867 | "unterminated string or character constant"); | |
868 | if (pfile->multiline_string_line != start_line | |
869 | && pfile->multiline_string_line != 0) | |
870 | cpp_error_with_line (pfile, | |
871 | pfile->multiline_string_line, -1, | |
872 | "possible real start of unterminated constant"); | |
873 | pfile->multiline_string_line = 0; | |
64aaf407 | 874 | goto out; |
45b966db | 875 | |
64aaf407 NB |
876 | case '\0': |
877 | null_count++; | |
878 | break; | |
879 | ||
45b966db ZW |
880 | case '\n': |
881 | CPP_BUMP_LINE (pfile); | |
882 | /* In Fortran and assembly language, silently terminate | |
883 | strings of either variety at end of line. This is a | |
884 | kludge around not knowing where comments are in these | |
885 | languages. */ | |
ae79697b ZW |
886 | if (CPP_OPTION (pfile, lang_fortran) |
887 | || CPP_OPTION (pfile, lang_asm)) | |
45b966db ZW |
888 | { |
889 | FORWARD(-1); | |
64aaf407 | 890 | goto out; |
45b966db ZW |
891 | } |
892 | /* Character constants may not extend over multiple lines. | |
893 | In Standard C, neither may strings. We accept multiline | |
894 | strings as an extension. */ | |
895 | if (c == '\'') | |
896 | { | |
897 | cpp_error_with_line (pfile, start_line, start_column, | |
898 | "unterminated character constant"); | |
899 | FORWARD(-1); | |
64aaf407 | 900 | goto out; |
45b966db ZW |
901 | } |
902 | if (CPP_PEDANTIC (pfile) && pfile->multiline_string_line == 0) | |
903 | cpp_pedwarn_with_line (pfile, start_line, start_column, | |
904 | "string constant runs past end of line"); | |
905 | if (pfile->multiline_string_line == 0) | |
906 | pfile->multiline_string_line = start_line; | |
907 | break; | |
908 | ||
909 | case '\r': | |
910 | if (CPP_BUFFER (pfile)->has_escapes) | |
911 | { | |
912 | cpp_ice (pfile, "\\r escape inside string constant"); | |
913 | FORWARD(1); | |
914 | } | |
915 | else | |
916 | /* Backslash newline is replaced by nothing at all. */ | |
917 | CPP_BUMP_LINE (pfile); | |
918 | break; | |
919 | ||
920 | case '\\': | |
921 | FORWARD(1); | |
922 | break; | |
923 | ||
924 | case '\"': | |
925 | case '\'': | |
926 | if (cc == c) | |
64aaf407 | 927 | goto out; |
45b966db ZW |
928 | break; |
929 | } | |
930 | } | |
64aaf407 NB |
931 | |
932 | out: | |
933 | if (null_count == 1) | |
934 | cpp_warning (pfile, "null character in string or character constant"); | |
935 | else if (null_count > 1) | |
936 | cpp_warning (pfile, "null characters in string or character constant"); | |
45b966db ZW |
937 | } |
938 | ||
939 | /* Parse a string and copy it to the output. */ | |
940 | ||
941 | static void | |
942 | parse_string (pfile, c) | |
943 | cpp_reader *pfile; | |
944 | int c; | |
945 | { | |
946 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
947 | const U_CHAR *limit; | |
948 | ||
949 | skip_string (pfile, c); | |
950 | ||
951 | limit = CPP_BUFFER (pfile)->cur; | |
952 | CPP_RESERVE (pfile, limit - start + 2); | |
953 | CPP_PUTC_Q (pfile, c); | |
954 | for (; start < limit; start++) | |
955 | if (*start != '\r') | |
956 | CPP_PUTC_Q (pfile, *start); | |
957 | } | |
958 | ||
959 | /* Read an assertion into the token buffer, converting to | |
960 | canonical form: `#predicate(a n swe r)' The next non-whitespace | |
961 | character to read should be the first letter of the predicate. | |
962 | Returns 0 for syntax error, 1 for bare predicate, 2 for predicate | |
963 | with answer (see callers for why). In case of 0, an error has been | |
964 | printed. */ | |
965 | int | |
966 | _cpp_parse_assertion (pfile) | |
967 | cpp_reader *pfile; | |
968 | { | |
969 | int c, dropwhite; | |
970 | _cpp_skip_hspace (pfile); | |
971 | c = PEEKC(); | |
bfab56e7 ZW |
972 | if (c == '\n') |
973 | { | |
974 | cpp_error (pfile, "assertion without predicate"); | |
975 | return 0; | |
976 | } | |
977 | else if (! is_idstart(c)) | |
45b966db ZW |
978 | { |
979 | cpp_error (pfile, "assertion predicate is not an identifier"); | |
980 | return 0; | |
981 | } | |
982 | CPP_PUTC(pfile, '#'); | |
983 | FORWARD(1); | |
984 | _cpp_parse_name (pfile, c); | |
985 | ||
986 | c = PEEKC(); | |
987 | if (c != '(') | |
988 | { | |
989 | if (is_hspace(c) || c == '\r') | |
990 | _cpp_skip_hspace (pfile); | |
991 | c = PEEKC(); | |
992 | } | |
993 | if (c != '(') | |
994 | return 1; | |
995 | ||
996 | CPP_PUTC(pfile, '('); | |
997 | FORWARD(1); | |
998 | dropwhite = 1; | |
999 | while ((c = GETC()) != ')') | |
1000 | { | |
1001 | if (is_space(c)) | |
1002 | { | |
1003 | if (! dropwhite) | |
1004 | { | |
1005 | CPP_PUTC(pfile, ' '); | |
1006 | dropwhite = 1; | |
1007 | } | |
1008 | } | |
1009 | else if (c == '\n' || c == EOF) | |
1010 | { | |
1011 | if (c == '\n') FORWARD(-1); | |
1012 | cpp_error (pfile, "un-terminated assertion answer"); | |
1013 | return 0; | |
1014 | } | |
1015 | else if (c == '\r') | |
1016 | /* \r cannot be a macro escape here. */ | |
1017 | CPP_BUMP_LINE (pfile); | |
1018 | else | |
1019 | { | |
1020 | CPP_PUTC (pfile, c); | |
1021 | dropwhite = 0; | |
1022 | } | |
1023 | } | |
1024 | ||
1025 | if (pfile->limit[-1] == ' ') | |
1026 | pfile->limit[-1] = ')'; | |
1027 | else if (pfile->limit[-1] == '(') | |
1028 | { | |
1029 | cpp_error (pfile, "empty token sequence in assertion"); | |
1030 | return 0; | |
1031 | } | |
1032 | else | |
1033 | CPP_PUTC (pfile, ')'); | |
1034 | ||
45b966db ZW |
1035 | return 2; |
1036 | } | |
1037 | ||
1038 | /* Get the next token, and add it to the text in pfile->token_buffer. | |
1039 | Return the kind of token we got. */ | |
1040 | ||
3a2b2c7a | 1041 | enum cpp_ttype |
45b966db ZW |
1042 | _cpp_lex_token (pfile) |
1043 | cpp_reader *pfile; | |
1044 | { | |
5eec0563 | 1045 | register int c, c2; |
3a2b2c7a | 1046 | enum cpp_ttype token; |
45b966db | 1047 | |
f2d5f0cc ZW |
1048 | if (CPP_BUFFER (pfile) == NULL) |
1049 | return CPP_EOF; | |
1050 | ||
45b966db ZW |
1051 | get_next: |
1052 | c = GETC(); | |
1053 | switch (c) | |
1054 | { | |
1055 | case EOF: | |
1056 | return CPP_EOF; | |
1057 | ||
1058 | case '/': | |
1059 | if (PEEKC () == '=') | |
1060 | goto op2; | |
1061 | ||
1062 | comment: | |
ae79697b | 1063 | if (CPP_OPTION (pfile, discard_comments)) |
45b966db ZW |
1064 | c = skip_comment (pfile, c); |
1065 | else | |
1066 | c = copy_comment (pfile, c); | |
1067 | if (c != ' ') | |
1068 | goto randomchar; | |
1069 | ||
1070 | /* Comments are equivalent to spaces. | |
1071 | For -traditional, a comment is equivalent to nothing. */ | |
ff2b53ef | 1072 | if (!CPP_OPTION (pfile, discard_comments)) |
45b966db | 1073 | return CPP_COMMENT; |
9e62c811 | 1074 | else if (CPP_TRADITIONAL (pfile)) |
0f89df67 ZW |
1075 | { |
1076 | if (pfile->parsing_define_directive) | |
1077 | return CPP_COMMENT; | |
1078 | goto get_next; | |
1079 | } | |
45b966db ZW |
1080 | else |
1081 | { | |
1082 | CPP_PUTC (pfile, c); | |
1083 | return CPP_HSPACE; | |
1084 | } | |
1085 | ||
1086 | case '#': | |
5eec0563 JM |
1087 | CPP_PUTC (pfile, c); |
1088 | ||
1089 | hash: | |
45b966db ZW |
1090 | if (pfile->parsing_if_directive) |
1091 | { | |
f2d5f0cc | 1092 | CPP_ADJUST_WRITTEN (pfile, -1); |
bfab56e7 ZW |
1093 | if (_cpp_parse_assertion (pfile)) |
1094 | return CPP_ASSERTION; | |
5eec0563 | 1095 | return CPP_OTHER; |
45b966db ZW |
1096 | } |
1097 | ||
9e62c811 | 1098 | if (pfile->parsing_define_directive) |
45b966db | 1099 | { |
5eec0563 JM |
1100 | c2 = PEEKC (); |
1101 | if (c2 == '#') | |
1102 | { | |
1103 | FORWARD (1); | |
1104 | CPP_PUTC (pfile, c2); | |
1105 | } | |
1106 | else if (c2 == '%' && PEEKN (1) == ':') | |
1107 | { | |
1108 | /* Digraph: "%:" == "#". */ | |
1109 | FORWARD (1); | |
1110 | CPP_RESERVE (pfile, 2); | |
1111 | CPP_PUTC_Q (pfile, c2); | |
1112 | CPP_PUTC_Q (pfile, GETC ()); | |
1113 | } | |
1114 | else | |
1368ee70 | 1115 | return CPP_HASH; |
5eec0563 | 1116 | |
1368ee70 | 1117 | return CPP_PASTE; |
45b966db ZW |
1118 | } |
1119 | ||
1120 | if (!pfile->only_seen_white) | |
5eec0563 JM |
1121 | return CPP_OTHER; |
1122 | ||
1123 | /* Remove the "#" or "%:" from the token buffer. */ | |
1124 | CPP_ADJUST_WRITTEN (pfile, (c == '#' ? -1 : -2)); | |
45b966db ZW |
1125 | return CPP_DIRECTIVE; |
1126 | ||
1127 | case '\"': | |
1128 | case '\'': | |
1129 | parse_string (pfile, c); | |
45b966db ZW |
1130 | return c == '\'' ? CPP_CHAR : CPP_STRING; |
1131 | ||
1132 | case '$': | |
ae79697b | 1133 | if (!CPP_OPTION (pfile, dollars_in_ident)) |
45b966db ZW |
1134 | goto randomchar; |
1135 | goto letter; | |
1136 | ||
1137 | case ':': | |
5eec0563 JM |
1138 | c2 = PEEKC (); |
1139 | /* Digraph: ":>" == "]". */ | |
1140 | if (c2 == '>' | |
1141 | || (c2 == ':' && CPP_OPTION (pfile, cplusplus))) | |
45b966db ZW |
1142 | goto op2; |
1143 | goto randomchar; | |
1144 | ||
1145 | case '&': | |
1146 | case '+': | |
1147 | case '|': | |
1148 | c2 = PEEKC (); | |
1149 | if (c2 == c || c2 == '=') | |
1150 | goto op2; | |
1151 | goto randomchar; | |
1152 | ||
5eec0563 JM |
1153 | case '%': |
1154 | /* Digraphs: "%:" == "#", "%>" == "}". */ | |
1155 | c2 = PEEKC (); | |
1156 | if (c2 == ':') | |
1157 | { | |
1158 | FORWARD (1); | |
1159 | CPP_RESERVE (pfile, 2); | |
1160 | CPP_PUTC_Q (pfile, c); | |
1161 | CPP_PUTC_Q (pfile, c2); | |
1162 | goto hash; | |
1163 | } | |
1164 | else if (c2 == '>') | |
1165 | { | |
1166 | FORWARD (1); | |
1167 | CPP_RESERVE (pfile, 2); | |
1168 | CPP_PUTC_Q (pfile, c); | |
1169 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1170 | return CPP_OPEN_BRACE; |
5eec0563 JM |
1171 | } |
1172 | /* else fall through */ | |
1173 | ||
45b966db ZW |
1174 | case '*': |
1175 | case '!': | |
45b966db ZW |
1176 | case '=': |
1177 | case '^': | |
1178 | if (PEEKC () == '=') | |
1179 | goto op2; | |
1180 | goto randomchar; | |
1181 | ||
1182 | case '-': | |
1183 | c2 = PEEKC (); | |
1184 | if (c2 == '-') | |
1185 | { | |
ae79697b | 1186 | if (CPP_OPTION (pfile, chill)) |
45b966db ZW |
1187 | goto comment; /* Chill style comment */ |
1188 | else | |
1189 | goto op2; | |
1190 | } | |
1191 | else if (c2 == '=') | |
1192 | goto op2; | |
1193 | else if (c2 == '>') | |
1194 | { | |
ae79697b | 1195 | if (CPP_OPTION (pfile, cplusplus) && PEEKN (1) == '*') |
45b966db ZW |
1196 | { |
1197 | /* In C++, there's a ->* operator. */ | |
1198 | token = CPP_OTHER; | |
45b966db ZW |
1199 | CPP_RESERVE (pfile, 4); |
1200 | CPP_PUTC_Q (pfile, c); | |
1201 | CPP_PUTC_Q (pfile, GETC ()); | |
1202 | CPP_PUTC_Q (pfile, GETC ()); | |
45b966db ZW |
1203 | return token; |
1204 | } | |
1205 | goto op2; | |
1206 | } | |
1207 | goto randomchar; | |
1208 | ||
1209 | case '<': | |
1210 | if (pfile->parsing_include_directive) | |
1211 | { | |
1212 | for (;;) | |
1213 | { | |
1214 | CPP_PUTC (pfile, c); | |
1215 | if (c == '>') | |
1216 | break; | |
1217 | c = GETC (); | |
1218 | if (c == '\n' || c == EOF) | |
1219 | { | |
1220 | cpp_error (pfile, | |
1221 | "missing '>' in `#include <FILENAME>'"); | |
1222 | break; | |
1223 | } | |
1224 | else if (c == '\r') | |
1225 | { | |
1226 | if (!CPP_BUFFER (pfile)->has_escapes) | |
1227 | { | |
1228 | /* Backslash newline is replaced by nothing. */ | |
1229 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1230 | CPP_BUMP_LINE (pfile); | |
1231 | } | |
1232 | else | |
1233 | { | |
1234 | /* We might conceivably get \r- or \r<space> in | |
1235 | here. Just delete 'em. */ | |
1236 | int d = GETC(); | |
1237 | if (d != '-' && d != ' ') | |
1238 | cpp_ice (pfile, "unrecognized escape \\r%c", d); | |
1239 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1240 | } | |
1241 | } | |
1242 | } | |
1243 | return CPP_STRING; | |
1244 | } | |
5eec0563 JM |
1245 | /* Digraphs: "<%" == "{", "<:" == "[". */ |
1246 | c2 = PEEKC (); | |
1247 | if (c2 == '%') | |
1248 | { | |
1249 | FORWARD (1); | |
1250 | CPP_RESERVE (pfile, 2); | |
1251 | CPP_PUTC_Q (pfile, c); | |
1252 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1253 | return CPP_CLOSE_BRACE; |
5eec0563 JM |
1254 | } |
1255 | else if (c2 == ':') | |
1256 | goto op2; | |
45b966db ZW |
1257 | /* else fall through */ |
1258 | case '>': | |
1259 | c2 = PEEKC (); | |
1260 | if (c2 == '=') | |
1261 | goto op2; | |
1262 | /* GNU C++ supports MIN and MAX operators <? and >?. */ | |
ae79697b | 1263 | if (c2 != c && (!CPP_OPTION (pfile, cplusplus) || c2 != '?')) |
45b966db ZW |
1264 | goto randomchar; |
1265 | FORWARD(1); | |
5eec0563 JM |
1266 | CPP_RESERVE (pfile, 3); |
1267 | CPP_PUTC_Q (pfile, c); | |
1268 | CPP_PUTC_Q (pfile, c2); | |
1269 | if (PEEKC () == '=') | |
45b966db | 1270 | CPP_PUTC_Q (pfile, GETC ()); |
45b966db ZW |
1271 | return CPP_OTHER; |
1272 | ||
1273 | case '.': | |
1274 | c2 = PEEKC (); | |
5eec0563 | 1275 | if (ISDIGIT (c2)) |
45b966db | 1276 | { |
5eec0563 | 1277 | CPP_PUTC (pfile, c); |
45b966db ZW |
1278 | c = GETC (); |
1279 | goto number; | |
1280 | } | |
1281 | ||
1282 | /* In C++ there's a .* operator. */ | |
ae79697b | 1283 | if (CPP_OPTION (pfile, cplusplus) && c2 == '*') |
45b966db ZW |
1284 | goto op2; |
1285 | ||
1286 | if (c2 == '.' && PEEKN(1) == '.') | |
1287 | { | |
5eec0563 | 1288 | CPP_RESERVE (pfile, 3); |
45b966db ZW |
1289 | CPP_PUTC_Q (pfile, '.'); |
1290 | CPP_PUTC_Q (pfile, '.'); | |
1291 | CPP_PUTC_Q (pfile, '.'); | |
1292 | FORWARD (2); | |
1368ee70 | 1293 | return CPP_ELLIPSIS; |
45b966db ZW |
1294 | } |
1295 | goto randomchar; | |
1296 | ||
1297 | op2: | |
5eec0563 | 1298 | CPP_RESERVE (pfile, 2); |
45b966db ZW |
1299 | CPP_PUTC_Q (pfile, c); |
1300 | CPP_PUTC_Q (pfile, GETC ()); | |
5eec0563 | 1301 | return CPP_OTHER; |
45b966db ZW |
1302 | |
1303 | case 'L': | |
1304 | c2 = PEEKC (); | |
1305 | if ((c2 == '\'' || c2 == '\"') && !CPP_TRADITIONAL (pfile)) | |
1306 | { | |
1307 | CPP_PUTC (pfile, c); | |
1308 | c = GETC (); | |
1309 | parse_string (pfile, c); | |
45b966db ZW |
1310 | return c == '\'' ? CPP_WCHAR : CPP_WSTRING; |
1311 | } | |
1312 | goto letter; | |
1313 | ||
1314 | case '0': case '1': case '2': case '3': case '4': | |
1315 | case '5': case '6': case '7': case '8': case '9': | |
1316 | number: | |
1317 | c2 = '.'; | |
1318 | for (;;) | |
1319 | { | |
1320 | CPP_RESERVE (pfile, 2); | |
1321 | CPP_PUTC_Q (pfile, c); | |
1322 | c = PEEKC (); | |
1323 | if (c == EOF) | |
1324 | break; | |
1325 | if (!is_numchar(c) && c != '.' | |
1326 | && ((c2 != 'e' && c2 != 'E' | |
1327 | && ((c2 != 'p' && c2 != 'P') | |
ae79697b | 1328 | || CPP_OPTION (pfile, c89))) |
45b966db ZW |
1329 | || (c != '+' && c != '-'))) |
1330 | break; | |
1331 | FORWARD(1); | |
1332 | c2= c; | |
1333 | } | |
45b966db ZW |
1334 | return CPP_NUMBER; |
1335 | case 'b': case 'c': case 'd': case 'h': case 'o': | |
1336 | case 'B': case 'C': case 'D': case 'H': case 'O': | |
ae79697b | 1337 | if (CPP_OPTION (pfile, chill) && PEEKC () == '\'') |
45b966db | 1338 | { |
45b966db ZW |
1339 | CPP_RESERVE (pfile, 2); |
1340 | CPP_PUTC_Q (pfile, c); | |
1341 | CPP_PUTC_Q (pfile, '\''); | |
1342 | FORWARD(1); | |
1343 | for (;;) | |
1344 | { | |
1345 | c = GETC(); | |
1346 | if (c == EOF) | |
1347 | goto chill_number_eof; | |
1348 | if (!is_numchar(c)) | |
1349 | break; | |
1350 | CPP_PUTC (pfile, c); | |
1351 | } | |
1352 | if (c == '\'') | |
1353 | { | |
1354 | CPP_RESERVE (pfile, 2); | |
1355 | CPP_PUTC_Q (pfile, c); | |
45b966db ZW |
1356 | return CPP_STRING; |
1357 | } | |
1358 | else | |
1359 | { | |
1360 | FORWARD(-1); | |
1361 | chill_number_eof: | |
45b966db ZW |
1362 | return CPP_NUMBER; |
1363 | } | |
1364 | } | |
1365 | else | |
1366 | goto letter; | |
1367 | case '_': | |
1368 | case 'a': case 'e': case 'f': case 'g': case 'i': case 'j': | |
1369 | case 'k': case 'l': case 'm': case 'n': case 'p': case 'q': | |
1370 | case 'r': case 's': case 't': case 'u': case 'v': case 'w': | |
1371 | case 'x': case 'y': case 'z': | |
1372 | case 'A': case 'E': case 'F': case 'G': case 'I': case 'J': | |
1373 | case 'K': case 'M': case 'N': case 'P': case 'Q': case 'R': | |
1374 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1375 | case 'Y': case 'Z': | |
1376 | letter: | |
45b966db ZW |
1377 | _cpp_parse_name (pfile, c); |
1378 | return CPP_MACRO; | |
1379 | ||
64aaf407 NB |
1380 | case ' ': case '\t': case '\v': case '\f': case '\0': |
1381 | { | |
1382 | int null_count = 0; | |
1383 | ||
1384 | for (;;) | |
1385 | { | |
1386 | if (c == '\0') | |
1387 | null_count++; | |
1388 | else | |
1389 | CPP_PUTC (pfile, c); | |
1390 | c = PEEKC (); | |
1391 | if (c == EOF || !is_hspace(c)) | |
1392 | break; | |
1393 | FORWARD(1); | |
1394 | } | |
1395 | if (null_count) | |
1396 | null_warning (pfile, null_count); | |
1397 | return CPP_HSPACE; | |
1398 | } | |
45b966db ZW |
1399 | |
1400 | case '\r': | |
1401 | if (CPP_BUFFER (pfile)->has_escapes) | |
1402 | { | |
1403 | c = GETC (); | |
1404 | if (c == '-') | |
1405 | { | |
1406 | if (pfile->output_escapes) | |
1407 | CPP_PUTS (pfile, "\r-", 2); | |
1408 | _cpp_parse_name (pfile, GETC ()); | |
1409 | return CPP_NAME; | |
1410 | } | |
1411 | else if (c == ' ') | |
1412 | { | |
ff2b53ef ZW |
1413 | /* "\r " means a space, but only if necessary to prevent |
1414 | accidental token concatenation. */ | |
45b966db ZW |
1415 | CPP_RESERVE (pfile, 2); |
1416 | if (pfile->output_escapes) | |
1417 | CPP_PUTC_Q (pfile, '\r'); | |
1418 | CPP_PUTC_Q (pfile, c); | |
1419 | return CPP_HSPACE; | |
1420 | } | |
1421 | else | |
1422 | { | |
1423 | cpp_ice (pfile, "unrecognized escape \\r%c", c); | |
1424 | goto get_next; | |
1425 | } | |
1426 | } | |
1427 | else | |
1428 | { | |
1429 | /* Backslash newline is ignored. */ | |
cbccf5e8 MM |
1430 | if (!ACTIVE_MARK_P (pfile)) |
1431 | CPP_BUMP_LINE (pfile); | |
45b966db ZW |
1432 | goto get_next; |
1433 | } | |
1434 | ||
1435 | case '\n': | |
1436 | CPP_PUTC (pfile, c); | |
45b966db ZW |
1437 | return CPP_VSPACE; |
1438 | ||
1368ee70 ZW |
1439 | case '(': token = CPP_OPEN_PAREN; goto char1; |
1440 | case ')': token = CPP_CLOSE_PAREN; goto char1; | |
1441 | case '{': token = CPP_OPEN_BRACE; goto char1; | |
1442 | case '}': token = CPP_CLOSE_BRACE; goto char1; | |
1443 | case ',': token = CPP_COMMA; goto char1; | |
1444 | case ';': token = CPP_SEMICOLON; goto char1; | |
45b966db ZW |
1445 | |
1446 | randomchar: | |
1447 | default: | |
1448 | token = CPP_OTHER; | |
1449 | char1: | |
45b966db ZW |
1450 | CPP_PUTC (pfile, c); |
1451 | return token; | |
1452 | } | |
1453 | } | |
1454 | ||
1455 | /* Check for and expand a macro, which is from WRITTEN to CPP_WRITTEN (pfile). | |
1456 | Caller is expected to have checked no_macro_expand. */ | |
1457 | static int | |
1458 | maybe_macroexpand (pfile, written) | |
1459 | cpp_reader *pfile; | |
1460 | long written; | |
1461 | { | |
1462 | U_CHAR *macro = pfile->token_buffer + written; | |
1463 | size_t len = CPP_WRITTEN (pfile) - written; | |
1464 | HASHNODE *hp = _cpp_lookup (pfile, macro, len); | |
1465 | ||
1466 | if (!hp) | |
1467 | return 0; | |
d9e0bd53 | 1468 | if (hp->disabled || hp->type == T_IDENTITY) |
45b966db ZW |
1469 | { |
1470 | if (pfile->output_escapes) | |
1471 | { | |
1472 | /* Insert a no-reexpand marker before IDENT. */ | |
1473 | CPP_RESERVE (pfile, 2); | |
1474 | CPP_ADJUST_WRITTEN (pfile, 2); | |
1475 | macro = pfile->token_buffer + written; | |
1476 | ||
1477 | memmove (macro + 2, macro, len); | |
1478 | macro[0] = '\r'; | |
1479 | macro[1] = '-'; | |
1480 | } | |
1481 | return 0; | |
1482 | } | |
ff2b53ef ZW |
1483 | if (hp->type == T_EMPTY) |
1484 | { | |
1485 | /* Special case optimization: macro expands to nothing. */ | |
1486 | CPP_SET_WRITTEN (pfile, written); | |
1487 | CPP_PUTC_Q (pfile, ' '); | |
1488 | return 1; | |
1489 | } | |
45b966db ZW |
1490 | |
1491 | /* If macro wants an arglist, verify that a '(' follows. */ | |
d9e0bd53 | 1492 | if (hp->type == T_FMACRO) |
45b966db ZW |
1493 | { |
1494 | int macbuf_whitespace = 0; | |
1495 | int c; | |
1496 | ||
1497 | while (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
1498 | { | |
1499 | const U_CHAR *point = CPP_BUFFER (pfile)->cur; | |
1500 | for (;;) | |
1501 | { | |
1502 | _cpp_skip_hspace (pfile); | |
1503 | c = PEEKC (); | |
1504 | if (c == '\n') | |
1505 | FORWARD(1); | |
1506 | else | |
1507 | break; | |
1508 | } | |
1509 | if (point != CPP_BUFFER (pfile)->cur) | |
1510 | macbuf_whitespace = 1; | |
1511 | if (c == '(') | |
1512 | goto is_macro_call; | |
1513 | else if (c != EOF) | |
1514 | goto not_macro_call; | |
1515 | cpp_pop_buffer (pfile); | |
1516 | } | |
1517 | ||
1518 | CPP_SET_MARK (pfile); | |
1519 | for (;;) | |
1520 | { | |
1521 | _cpp_skip_hspace (pfile); | |
1522 | c = PEEKC (); | |
1523 | if (c == '\n') | |
1524 | FORWARD(1); | |
1525 | else | |
1526 | break; | |
1527 | } | |
1528 | CPP_GOTO_MARK (pfile); | |
1529 | ||
1530 | if (c != '(') | |
1531 | { | |
1532 | not_macro_call: | |
1533 | if (macbuf_whitespace) | |
1534 | CPP_PUTC (pfile, ' '); | |
1535 | return 0; | |
1536 | } | |
1537 | } | |
1538 | ||
1539 | is_macro_call: | |
1540 | /* This is now known to be a macro call. | |
1541 | Expand the macro, reading arguments as needed, | |
1542 | and push the expansion on the input stack. */ | |
1543 | _cpp_macroexpand (pfile, hp); | |
1544 | CPP_SET_WRITTEN (pfile, written); | |
1545 | return 1; | |
1546 | } | |
1547 | ||
9e62c811 ZW |
1548 | /* Complain about \v or \f in a preprocessing directive (constraint |
1549 | violation, C99 6.10 para 5). Caller has checked CPP_PEDANTIC. */ | |
1550 | static void | |
1551 | pedantic_whitespace (pfile, p, len) | |
1552 | cpp_reader *pfile; | |
1553 | U_CHAR *p; | |
1554 | unsigned int len; | |
1555 | { | |
1556 | while (len) | |
1557 | { | |
1558 | if (*p == '\v') | |
1559 | cpp_pedwarn (pfile, "vertical tab in preprocessing directive"); | |
1560 | else if (*p == '\f') | |
1561 | cpp_pedwarn (pfile, "form feed in preprocessing directive"); | |
1562 | p++; | |
1563 | len--; | |
1564 | } | |
1565 | } | |
1566 | ||
1567 | ||
3a2b2c7a | 1568 | enum cpp_ttype |
45b966db ZW |
1569 | cpp_get_token (pfile) |
1570 | cpp_reader *pfile; | |
1571 | { | |
3a2b2c7a | 1572 | enum cpp_ttype token; |
45b966db ZW |
1573 | long written = CPP_WRITTEN (pfile); |
1574 | ||
1575 | get_next: | |
1576 | token = _cpp_lex_token (pfile); | |
1577 | ||
1578 | switch (token) | |
1579 | { | |
1580 | default: | |
ff2b53ef ZW |
1581 | pfile->potential_control_macro = 0; |
1582 | pfile->only_seen_white = 0; | |
1583 | return token; | |
1584 | ||
1585 | case CPP_VSPACE: | |
1586 | if (pfile->only_seen_white == 0) | |
1587 | pfile->only_seen_white = 1; | |
1588 | CPP_BUMP_LINE (pfile); | |
ff2b53ef ZW |
1589 | return token; |
1590 | ||
1591 | case CPP_HSPACE: | |
1592 | case CPP_COMMENT: | |
45b966db ZW |
1593 | return token; |
1594 | ||
1595 | case CPP_DIRECTIVE: | |
ff2b53ef | 1596 | pfile->potential_control_macro = 0; |
45b966db ZW |
1597 | if (_cpp_handle_directive (pfile)) |
1598 | return CPP_DIRECTIVE; | |
1599 | pfile->only_seen_white = 0; | |
1600 | CPP_PUTC (pfile, '#'); | |
1601 | return CPP_OTHER; | |
1602 | ||
1603 | case CPP_MACRO: | |
ff2b53ef ZW |
1604 | pfile->potential_control_macro = 0; |
1605 | pfile->only_seen_white = 0; | |
45b966db ZW |
1606 | if (! pfile->no_macro_expand |
1607 | && maybe_macroexpand (pfile, written)) | |
1608 | goto get_next; | |
1609 | return CPP_NAME; | |
1610 | ||
1611 | case CPP_EOF: | |
f2d5f0cc ZW |
1612 | if (CPP_BUFFER (pfile) == NULL) |
1613 | return CPP_EOF; | |
45b966db ZW |
1614 | if (CPP_BUFFER (pfile)->manual_pop) |
1615 | /* If we've been reading from redirected input, the | |
1616 | frontend will pop the buffer. */ | |
1617 | return CPP_EOF; | |
45b966db | 1618 | |
f2d5f0cc ZW |
1619 | if (CPP_BUFFER (pfile)->seen_eof) |
1620 | { | |
45b966db ZW |
1621 | cpp_pop_buffer (pfile); |
1622 | goto get_next; | |
1623 | } | |
1624 | else | |
1625 | { | |
1626 | _cpp_handle_eof (pfile); | |
1627 | return CPP_POP; | |
1628 | } | |
1629 | } | |
1630 | } | |
1631 | ||
1632 | /* Like cpp_get_token, but skip spaces and comments. */ | |
1633 | ||
3a2b2c7a | 1634 | enum cpp_ttype |
45b966db ZW |
1635 | cpp_get_non_space_token (pfile) |
1636 | cpp_reader *pfile; | |
1637 | { | |
1638 | int old_written = CPP_WRITTEN (pfile); | |
1639 | for (;;) | |
1640 | { | |
3a2b2c7a | 1641 | enum cpp_ttype token = cpp_get_token (pfile); |
ff2b53ef | 1642 | if (token != CPP_COMMENT && token != CPP_HSPACE && token != CPP_VSPACE) |
45b966db ZW |
1643 | return token; |
1644 | CPP_SET_WRITTEN (pfile, old_written); | |
1645 | } | |
1646 | } | |
1647 | ||
ff2b53ef | 1648 | /* Like cpp_get_token, except that it does not execute directives, |
9e62c811 ZW |
1649 | does not consume vertical space, discards horizontal space, and |
1650 | automatically pops off macro buffers. */ | |
3a2b2c7a | 1651 | enum cpp_ttype |
9e62c811 | 1652 | _cpp_get_directive_token (pfile) |
45b966db ZW |
1653 | cpp_reader *pfile; |
1654 | { | |
ff2b53ef | 1655 | long old_written; |
3a2b2c7a | 1656 | enum cpp_ttype token; |
45b966db | 1657 | |
ff2b53ef ZW |
1658 | get_next: |
1659 | old_written = CPP_WRITTEN (pfile); | |
1660 | token = _cpp_lex_token (pfile); | |
1661 | switch (token) | |
45b966db | 1662 | { |
ff2b53ef ZW |
1663 | default: |
1664 | return token; | |
45b966db | 1665 | |
ff2b53ef ZW |
1666 | case CPP_VSPACE: |
1667 | /* Put it back and return VSPACE. */ | |
1668 | FORWARD(-1); | |
1669 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1670 | return CPP_VSPACE; | |
45b966db | 1671 | |
ff2b53ef ZW |
1672 | case CPP_HSPACE: |
1673 | if (CPP_PEDANTIC (pfile)) | |
9e62c811 ZW |
1674 | pedantic_whitespace (pfile, pfile->token_buffer + old_written, |
1675 | CPP_WRITTEN (pfile) - old_written); | |
1676 | CPP_SET_WRITTEN (pfile, old_written); | |
1677 | goto get_next; | |
ff2b53ef | 1678 | return CPP_HSPACE; |
45b966db | 1679 | |
ff2b53ef ZW |
1680 | case CPP_DIRECTIVE: |
1681 | /* Don't execute the directive, but don't smash it to OTHER either. */ | |
1682 | CPP_PUTC (pfile, '#'); | |
1683 | return CPP_DIRECTIVE; | |
1684 | ||
1685 | case CPP_MACRO: | |
1686 | if (! pfile->no_macro_expand | |
1687 | && maybe_macroexpand (pfile, old_written)) | |
1688 | goto get_next; | |
1689 | return CPP_NAME; | |
45b966db | 1690 | |
ff2b53ef ZW |
1691 | case CPP_EOF: |
1692 | if (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
45b966db | 1693 | { |
ff2b53ef ZW |
1694 | cpp_pop_buffer (pfile); |
1695 | goto get_next; | |
45b966db | 1696 | } |
ff2b53ef ZW |
1697 | else |
1698 | /* This can happen for files that don't end with a newline, | |
1699 | and for cpp_define and friends. Pretend they do, so | |
1700 | callers don't have to deal. A warning will be issued by | |
1701 | someone else, if necessary. */ | |
1702 | return CPP_VSPACE; | |
1703 | } | |
1704 | } | |
1705 | ||
45b966db ZW |
1706 | /* Determine the current line and column. Used only by read_and_prescan. */ |
1707 | static U_CHAR * | |
1708 | find_position (start, limit, linep) | |
1709 | U_CHAR *start; | |
1710 | U_CHAR *limit; | |
1711 | unsigned long *linep; | |
1712 | { | |
1713 | unsigned long line = *linep; | |
1714 | U_CHAR *lbase = start; | |
1715 | while (start < limit) | |
1716 | { | |
1717 | U_CHAR ch = *start++; | |
1718 | if (ch == '\n' || ch == '\r') | |
1719 | { | |
1720 | line++; | |
1721 | lbase = start; | |
1722 | } | |
1723 | } | |
1724 | *linep = line; | |
1725 | return lbase; | |
1726 | } | |
1727 | ||
2a87fbe8 ZW |
1728 | /* The following table is used by _cpp_read_and_prescan. If we have |
1729 | designated initializers, it can be constant data; otherwise, it is | |
1730 | set up at runtime by _cpp_init_input_buffer. */ | |
46d07497 ZW |
1731 | |
1732 | #ifndef UCHAR_MAX | |
1733 | #define UCHAR_MAX 255 /* assume 8-bit bytes */ | |
1734 | #endif | |
1735 | ||
1736 | #if (GCC_VERSION >= 2007) || (__STDC_VERSION__ >= 199901L) | |
2a87fbe8 ZW |
1737 | #define init_chartab() /* nothing */ |
1738 | #define CHARTAB static const unsigned char chartab[UCHAR_MAX + 1] = { | |
46d07497 ZW |
1739 | #define END }; |
1740 | #define s(p, v) [p] = v, | |
1741 | #else | |
2a87fbe8 ZW |
1742 | #define CHARTAB static unsigned char chartab[UCHAR_MAX + 1] = { 0 }; \ |
1743 | static void init_chartab PARAMS ((void)) { \ | |
1744 | unsigned char *x = chartab; | |
46d07497 ZW |
1745 | #define END } |
1746 | #define s(p, v) x[p] = v; | |
1747 | #endif | |
1748 | ||
1749 | /* Table of characters that can't be handled in the inner loop. | |
2a87fbe8 ZW |
1750 | Also contains the mapping between trigraph third characters and their |
1751 | replacements. */ | |
46d07497 ZW |
1752 | #define SPECCASE_CR 1 |
1753 | #define SPECCASE_BACKSLASH 2 | |
1754 | #define SPECCASE_QUESTION 3 | |
1755 | ||
2a87fbe8 | 1756 | CHARTAB |
46d07497 ZW |
1757 | s('\r', SPECCASE_CR) |
1758 | s('\\', SPECCASE_BACKSLASH) | |
1759 | s('?', SPECCASE_QUESTION) | |
46d07497 | 1760 | |
46d07497 ZW |
1761 | s('=', '#') s(')', ']') s('!', '|') |
1762 | s('(', '[') s('\'', '^') s('>', '}') | |
1763 | s('/', '\\') s('<', '{') s('-', '~') | |
1764 | END | |
1765 | ||
1766 | #undef CHARTAB | |
46d07497 ZW |
1767 | #undef END |
1768 | #undef s | |
1769 | ||
2a87fbe8 ZW |
1770 | #define NORMAL(c) ((chartab[c]) == 0 || (chartab[c]) > SPECCASE_QUESTION) |
1771 | #define NONTRI(c) ((c) <= SPECCASE_QUESTION) | |
1772 | ||
45b966db ZW |
1773 | /* Read the entire contents of file DESC into buffer BUF. LEN is how |
1774 | much memory to allocate initially; more will be allocated if | |
1775 | necessary. Convert end-of-line markers (\n, \r, \r\n, \n\r) to | |
1776 | canonical form (\n). If enabled, convert and/or warn about | |
1777 | trigraphs. Convert backslash-newline to a one-character escape | |
1778 | (\r) and remove it from "embarrassing" places (i.e. the middle of a | |
1779 | token). If there is no newline at the end of the file, add one and | |
1780 | warn. Returns -1 on failure, or the actual length of the data to | |
1781 | be scanned. | |
1782 | ||
1783 | This function does a lot of work, and can be a serious performance | |
1784 | bottleneck. It has been tuned heavily; make sure you understand it | |
1785 | before hacking. The common case - no trigraphs, Unix style line | |
1786 | breaks, backslash-newline set off by whitespace, newline at EOF - | |
1787 | has been optimized at the expense of the others. The performance | |
1788 | penalty for DOS style line breaks (\r\n) is about 15%. | |
1789 | ||
1790 | Warnings lose particularly heavily since we have to determine the | |
1791 | line number, which involves scanning from the beginning of the file | |
1792 | or from the last warning. The penalty for the absence of a newline | |
1793 | at the end of reload1.c is about 60%. (reload1.c is 329k.) | |
1794 | ||
1795 | If your file has more than one kind of end-of-line marker, you | |
04e3ec78 NB |
1796 | will get messed-up line numbering. |
1797 | ||
1798 | So that the cases of the switch statement do not have to concern | |
1799 | themselves with the complications of reading beyond the end of the | |
1800 | buffer, the buffer is guaranteed to have at least 3 characters in | |
1801 | it (or however many are left in the file, if less) on entry to the | |
1802 | switch. This is enough to handle trigraphs and the "\\\n\r" and | |
1803 | "\\\r\n" cases. | |
1804 | ||
1805 | The end of the buffer is marked by a '\\', which, being a special | |
1806 | character, guarantees we will exit the fast-scan loops and perform | |
1807 | a refill. */ | |
46d07497 | 1808 | |
45b966db ZW |
1809 | long |
1810 | _cpp_read_and_prescan (pfile, fp, desc, len) | |
1811 | cpp_reader *pfile; | |
1812 | cpp_buffer *fp; | |
1813 | int desc; | |
1814 | size_t len; | |
1815 | { | |
1816 | U_CHAR *buf = (U_CHAR *) xmalloc (len); | |
1817 | U_CHAR *ip, *op, *line_base; | |
1818 | U_CHAR *ibase; | |
45b966db ZW |
1819 | unsigned long line; |
1820 | unsigned int deferred_newlines; | |
45b966db | 1821 | size_t offset; |
04e3ec78 | 1822 | int count = 0; |
45b966db ZW |
1823 | |
1824 | offset = 0; | |
04e3ec78 | 1825 | deferred_newlines = 0; |
45b966db ZW |
1826 | op = buf; |
1827 | line_base = buf; | |
1828 | line = 1; | |
04e3ec78 NB |
1829 | ibase = pfile->input_buffer + 3; |
1830 | ip = ibase; | |
1831 | ip[-1] = '\0'; /* Guarantee no match with \n for SPECCASE_CR */ | |
45b966db ZW |
1832 | |
1833 | for (;;) | |
1834 | { | |
04e3ec78 NB |
1835 | U_CHAR *near_buff_end; |
1836 | ||
04e3ec78 | 1837 | count = read (desc, ibase, pfile->input_buffer_len); |
45b966db ZW |
1838 | if (count < 0) |
1839 | goto error; | |
04e3ec78 NB |
1840 | |
1841 | ibase[count] = '\\'; /* Marks end of buffer */ | |
1842 | if (count) | |
45b966db | 1843 | { |
04e3ec78 NB |
1844 | near_buff_end = pfile->input_buffer + count; |
1845 | offset += count; | |
45b966db | 1846 | if (offset > len) |
04e3ec78 NB |
1847 | { |
1848 | size_t delta_op; | |
1849 | size_t delta_line_base; | |
1b955cba | 1850 | len = offset * 2; |
04e3ec78 NB |
1851 | if (offset > len) |
1852 | /* len overflowed. | |
1853 | This could happen if the file is larger than half the | |
1854 | maximum address space of the machine. */ | |
1855 | goto too_big; | |
1856 | ||
1857 | delta_op = op - buf; | |
1858 | delta_line_base = line_base - buf; | |
1859 | buf = (U_CHAR *) xrealloc (buf, len); | |
1860 | op = buf + delta_op; | |
1861 | line_base = buf + delta_line_base; | |
1862 | } | |
1863 | } | |
1864 | else | |
1865 | { | |
1866 | if (ip == ibase) | |
1867 | break; | |
1868 | /* Allow normal processing of the (at most 2) remaining | |
1869 | characters. The end-of-buffer marker is still present | |
1870 | and prevents false matches within the switch. */ | |
1871 | near_buff_end = ibase - 1; | |
45b966db ZW |
1872 | } |
1873 | ||
1874 | for (;;) | |
1875 | { | |
04e3ec78 | 1876 | unsigned int span; |
45b966db | 1877 | |
04e3ec78 | 1878 | /* Deal with \-newline, potentially in the middle of a token. */ |
45b966db ZW |
1879 | if (deferred_newlines) |
1880 | { | |
2a87fbe8 | 1881 | if (op != buf && ! is_space (op[-1]) && op[-1] != '\r') |
04e3ec78 NB |
1882 | { |
1883 | /* Previous was not white space. Skip to white | |
1884 | space, if we can, before outputting the \r's */ | |
1885 | span = 0; | |
1886 | while (ip[span] != ' ' | |
1887 | && ip[span] != '\t' | |
1888 | && ip[span] != '\n' | |
2a87fbe8 | 1889 | && NORMAL(ip[span])) |
04e3ec78 NB |
1890 | span++; |
1891 | memcpy (op, ip, span); | |
1892 | op += span; | |
1893 | ip += span; | |
2a87fbe8 | 1894 | if (! NORMAL(ip[0])) |
04e3ec78 NB |
1895 | goto do_speccase; |
1896 | } | |
1897 | while (deferred_newlines) | |
1898 | deferred_newlines--, *op++ = '\r'; | |
45b966db ZW |
1899 | } |
1900 | ||
1901 | /* Copy as much as we can without special treatment. */ | |
04e3ec78 | 1902 | span = 0; |
2a87fbe8 | 1903 | while (NORMAL (ip[span])) span++; |
45b966db ZW |
1904 | memcpy (op, ip, span); |
1905 | op += span; | |
1906 | ip += span; | |
1907 | ||
04e3ec78 NB |
1908 | do_speccase: |
1909 | if (ip > near_buff_end) /* Do we have enough chars? */ | |
1910 | break; | |
2a87fbe8 | 1911 | switch (chartab[*ip++]) |
45b966db | 1912 | { |
45b966db | 1913 | case SPECCASE_CR: /* \r */ |
04e3ec78 | 1914 | if (ip[-2] != '\n') |
45b966db | 1915 | { |
04e3ec78 NB |
1916 | if (*ip == '\n') |
1917 | ip++; | |
1918 | *op++ = '\n'; | |
45b966db | 1919 | } |
45b966db ZW |
1920 | break; |
1921 | ||
1922 | case SPECCASE_BACKSLASH: /* \ */ | |
04e3ec78 | 1923 | if (*ip == '\n') |
45b966db | 1924 | { |
04e3ec78 | 1925 | deferred_newlines++; |
45b966db ZW |
1926 | ip++; |
1927 | if (*ip == '\r') ip++; | |
45b966db ZW |
1928 | } |
1929 | else if (*ip == '\r') | |
1930 | { | |
04e3ec78 | 1931 | deferred_newlines++; |
45b966db ZW |
1932 | ip++; |
1933 | if (*ip == '\n') ip++; | |
45b966db ZW |
1934 | } |
1935 | else | |
1936 | *op++ = '\\'; | |
04e3ec78 | 1937 | break; |
45b966db ZW |
1938 | |
1939 | case SPECCASE_QUESTION: /* ? */ | |
1940 | { | |
1941 | unsigned int d, t; | |
04e3ec78 NB |
1942 | |
1943 | *op++ = '?'; /* Normal non-trigraph case */ | |
1944 | if (ip[0] != '?') | |
1945 | break; | |
1946 | ||
45b966db | 1947 | d = ip[1]; |
2a87fbe8 ZW |
1948 | t = chartab[d]; |
1949 | if (NONTRI (t)) | |
04e3ec78 | 1950 | break; |
45b966db | 1951 | |
ae79697b | 1952 | if (CPP_OPTION (pfile, warn_trigraphs)) |
45b966db ZW |
1953 | { |
1954 | unsigned long col; | |
1955 | line_base = find_position (line_base, op, &line); | |
1956 | col = op - line_base + 1; | |
ae79697b | 1957 | if (CPP_OPTION (pfile, trigraphs)) |
45b966db | 1958 | cpp_warning_with_line (pfile, line, col, |
04e3ec78 | 1959 | "trigraph ??%c converted to %c", d, t); |
45b966db ZW |
1960 | else |
1961 | cpp_warning_with_line (pfile, line, col, | |
04e3ec78 | 1962 | "trigraph ??%c ignored", d); |
45b966db | 1963 | } |
04e3ec78 NB |
1964 | |
1965 | ip += 2; | |
ae79697b | 1966 | if (CPP_OPTION (pfile, trigraphs)) |
45b966db | 1967 | { |
04e3ec78 | 1968 | op[-1] = t; /* Overwrite '?' */ |
45b966db | 1969 | if (t == '\\') |
04e3ec78 NB |
1970 | { |
1971 | op--; | |
1972 | *--ip = '\\'; | |
1973 | goto do_speccase; /* May need buffer refill */ | |
1974 | } | |
45b966db ZW |
1975 | } |
1976 | else | |
1977 | { | |
45b966db ZW |
1978 | *op++ = '?'; |
1979 | *op++ = d; | |
1980 | } | |
1981 | } | |
04e3ec78 | 1982 | break; |
45b966db ZW |
1983 | } |
1984 | } | |
f6fab919 ZW |
1985 | /* Copy previous char plus unprocessed (at most 2) chars |
1986 | to beginning of buffer, refill it with another | |
1987 | read(), and continue processing */ | |
1988 | memmove (ip - count - 1, ip - 1, 4 - (ip - near_buff_end)); | |
1989 | ip -= count; | |
45b966db ZW |
1990 | } |
1991 | ||
1992 | if (offset == 0) | |
1993 | return 0; | |
1994 | ||
45b966db ZW |
1995 | if (op[-1] != '\n') |
1996 | { | |
1997 | unsigned long col; | |
1998 | line_base = find_position (line_base, op, &line); | |
1999 | col = op - line_base + 1; | |
f6fab919 | 2000 | cpp_warning_with_line (pfile, line, col, "no newline at end of file"); |
45b966db ZW |
2001 | if (offset + 1 > len) |
2002 | { | |
2003 | len += 1; | |
2004 | if (offset + 1 > len) | |
2005 | goto too_big; | |
2006 | buf = (U_CHAR *) xrealloc (buf, len); | |
2007 | op = buf + offset; | |
2008 | } | |
2009 | *op++ = '\n'; | |
2010 | } | |
2011 | ||
2012 | fp->buf = ((len - offset < 20) ? buf : (U_CHAR *)xrealloc (buf, op - buf)); | |
2013 | return op - buf; | |
2014 | ||
2015 | too_big: | |
f6fab919 ZW |
2016 | cpp_notice (pfile, "%s is too large (>%lu bytes)", fp->ihash->name, |
2017 | (unsigned long)offset); | |
45b966db ZW |
2018 | free (buf); |
2019 | return -1; | |
2020 | ||
2021 | error: | |
2022 | cpp_error_from_errno (pfile, fp->ihash->name); | |
2023 | free (buf); | |
2024 | return -1; | |
2025 | } | |
2026 | ||
2a87fbe8 ZW |
2027 | /* Allocate pfile->input_buffer, and initialize chartab[] |
2028 | if it hasn't happened already. */ | |
46d07497 | 2029 | |
45b966db ZW |
2030 | void |
2031 | _cpp_init_input_buffer (pfile) | |
2032 | cpp_reader *pfile; | |
2033 | { | |
2034 | U_CHAR *tmp; | |
2035 | ||
2a87fbe8 | 2036 | init_chartab (); |
9e62c811 | 2037 | init_token_list (pfile, &pfile->directbuf, 0); |
04e3ec78 | 2038 | |
45b966db ZW |
2039 | /* Determine the appropriate size for the input buffer. Normal C |
2040 | source files are smaller than eight K. */ | |
04e3ec78 NB |
2041 | /* 8Kbytes of buffer proper, 1 to detect running off the end without |
2042 | address arithmetic all the time, and 3 for pushback during buffer | |
2043 | refill, in case there's a potential trigraph or end-of-line | |
2044 | digraph at the end of a block. */ | |
45b966db | 2045 | |
04e3ec78 | 2046 | tmp = (U_CHAR *) xmalloc (8192 + 1 + 3); |
45b966db ZW |
2047 | pfile->input_buffer = tmp; |
2048 | pfile->input_buffer_len = 8192; | |
2049 | } | |
c5a04734 ZW |
2050 | |
2051 | #if 0 | |
2052 | ||
d6d5f795 NB |
2053 | /* Lexing algorithm. |
2054 | ||
2055 | The original lexer in cpplib was made up of two passes: a first pass | |
2056 | that replaced trigraphs and deleted esacped newlines, and a second | |
2057 | pass that tokenized the result of the first pass. Tokenisation was | |
2058 | performed by peeking at the next character in the input stream. For | |
6777db6d | 2059 | example, if the input stream contained "!=", the handler for the ! |
d6d5f795 | 2060 | character would peek at the next character, and if it were a '=' |
6777db6d NB |
2061 | would skip over it, and return a "!=" token, otherwise it would |
2062 | return just the "!" token. | |
d6d5f795 NB |
2063 | |
2064 | To implement a single-pass lexer, this peeking ahead is unworkable. | |
2065 | An arbitrary number of escaped newlines, and trigraphs (in particular | |
6777db6d NB |
2066 | ??/ which translates to the escape \), could separate the '!' and '=' |
2067 | in the input stream, yet the next token is still a "!=". | |
d6d5f795 NB |
2068 | |
2069 | Suppose instead that we lex by one logical line at a time, producing | |
6777db6d NB |
2070 | a token list or stack for each logical line, and when seeing the '!' |
2071 | push a CPP_NOT token on the list. Then if the '!' is part of a | |
2072 | longer token ("!=") we know we must see the remainder of the token by | |
2073 | the time we reach the end of the logical line. Thus we can have the | |
2074 | '=' handler look at the previous token (at the end of the list / top | |
2075 | of the stack) and see if it is a "!" token, and if so, instead of | |
2076 | pushing a "=" token revise the existing token to be a "!=" token. | |
d6d5f795 NB |
2077 | |
2078 | This works in the presence of escaped newlines, because the '\' would | |
2079 | have been pushed on the top of the stack as a CPP_BACKSLASH. The | |
2080 | newline ('\n' or '\r') handler looks at the token at the top of the | |
2081 | stack to see if it is a CPP_BACKSLASH, and if so discards both. | |
2082 | Otherwise it pushes the newline (CPP_VSPACE) token as normal. Hence | |
2083 | the '=' handler would never see any intervening escaped newlines. | |
2084 | ||
2085 | To make trigraphs work in this context, as in precedence trigraphs | |
2086 | are highest and converted before anything else, the '?' handler does | |
2087 | lookahead to see if it is a trigraph, and if so skips the trigraph | |
2088 | and pushes the token it represents onto the top of the stack. This | |
2089 | also works in the particular case of a CPP_BACKSLASH trigraph. | |
2090 | ||
2091 | To the preprocessor, whitespace is only significant to the point of | |
2092 | knowing whether whitespace precedes a particular token. For example, | |
2093 | the '=' handler needs to know whether there was whitespace between it | |
6777db6d | 2094 | and a "!" token on the top of the stack, to make the token conversion |
d6d5f795 NB |
2095 | decision correctly. So each token has a PREV_WHITESPACE flag to |
2096 | indicate this - the standard permits consecutive whitespace to be | |
2097 | regarded as a single space. The compiler front ends are not | |
2098 | interested in whitespace at all; they just require a token stream. | |
2099 | Another place where whitespace is significant to the preprocessor is | |
2100 | a #define statment - if there is whitespace between the macro name | |
2101 | and an initial "(" token the macro is "object-like", otherwise it is | |
2102 | a function-like macro that takes arguments. | |
2103 | ||
2104 | However, all is not rosy. Parsing of identifiers, numbers, comments | |
2105 | and strings becomes trickier because of the possibility of raw | |
2106 | trigraphs and escaped newlines in the input stream. | |
2107 | ||
2108 | The trigraphs are three consecutive characters beginning with two | |
c2e25d51 NB |
2109 | question marks. A question mark is not valid as part of a number or |
2110 | identifier, so parsing of a number or identifier terminates normally | |
2111 | upon reaching it, returning to the mainloop which handles the | |
2112 | trigraph just like it would in any other position. Similarly for the | |
2113 | backslash of a backslash-newline combination. So we just need the | |
2114 | escaped-newline dropper in the mainloop to check if the token on the | |
2115 | top of the stack after dropping the escaped newline is a number or | |
2116 | identifier, and if so to continue the processing it as if nothing had | |
2117 | happened. | |
d6d5f795 NB |
2118 | |
2119 | For strings, we replace trigraphs whenever we reach a quote or | |
2120 | newline, because there might be a backslash trigraph escaping them. | |
2121 | We need to be careful that we start trigraph replacing from where we | |
2122 | left off previously, because it is possible for a first scan to leave | |
2123 | "fake" trigraphs that a second scan would pick up as real (e.g. the | |
c2e25d51 | 2124 | sequence "????/\n=" would find a fake ??= trigraph after removing the |
d6d5f795 NB |
2125 | escaped newline.) |
2126 | ||
2127 | For line comments, on reaching a newline we scan the previous | |
2128 | character(s) to see if it escaped, and continue if it is. Block | |
2129 | comments ignore everything and just focus on finding the comment | |
2130 | termination mark. The only difficult thing, and it is surprisingly | |
2131 | tricky, is checking if an asterisk precedes the final slash since | |
2132 | they could be separated by escaped newlines. If the preprocessor is | |
2133 | invoked with the output comments option, we don't bother removing | |
2134 | escaped newlines and replacing trigraphs for output. | |
2135 | ||
2136 | Finally, numbers can begin with a period, which is pushed initially | |
2137 | as a CPP_DOT token in its own right. The digit handler checks if the | |
2138 | previous token was a CPP_DOT not separated by whitespace, and if so | |
2139 | pops it off the stack and pushes a period into the number's buffer | |
2140 | before calling the number parser. | |
2141 | ||
2142 | */ | |
2143 | ||
c5a04734 ZW |
2144 | static void expand_comment_space PARAMS ((cpp_toklist *)); |
2145 | void init_trigraph_map PARAMS ((void)); | |
2146 | static unsigned char* trigraph_replace PARAMS ((cpp_reader *, unsigned char *, | |
2147 | unsigned char *)); | |
2148 | static const unsigned char *backslash_start PARAMS ((cpp_reader *, | |
2149 | const unsigned char *)); | |
2150 | static int skip_block_comment PARAMS ((cpp_reader *)); | |
2151 | static int skip_line_comment PARAMS ((cpp_reader *)); | |
2152 | static void skip_whitespace PARAMS ((cpp_reader *, int)); | |
2153 | static void parse_name PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
2154 | static void parse_number PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
2155 | static void parse_string PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *, | |
2156 | unsigned int)); | |
2157 | static int trigraph_ok PARAMS ((cpp_reader *, const unsigned char *)); | |
2158 | static void copy_comment PARAMS ((cpp_toklist *, const unsigned char *, | |
2159 | unsigned int, unsigned int, unsigned int)); | |
2160 | void _cpp_lex_line PARAMS ((cpp_reader *, cpp_toklist *)); | |
2161 | ||
2162 | static void _cpp_output_list PARAMS ((cpp_reader *, cpp_toklist *)); | |
2163 | ||
c5a04734 ZW |
2164 | unsigned int spell_string PARAMS ((unsigned char *, cpp_toklist *, |
2165 | cpp_token *token)); | |
2166 | unsigned int spell_comment PARAMS ((unsigned char *, cpp_toklist *, | |
2167 | cpp_token *token)); | |
2168 | unsigned int spell_name PARAMS ((unsigned char *, cpp_toklist *, | |
2169 | cpp_token *token)); | |
c5a04734 ZW |
2170 | |
2171 | typedef unsigned int (* speller) PARAMS ((unsigned char *, cpp_toklist *, | |
2172 | cpp_token *)); | |
2173 | ||
2174 | /* Macros on a cpp_name. */ | |
2175 | #define INIT_NAME(list, name) \ | |
2176 | do {(name).len = 0; (name).offset = (list)->name_used;} while (0) | |
2177 | ||
2178 | #define IS_DIRECTIVE(list) (list->tokens[0].type == CPP_HASH) | |
2179 | #define COLUMN(cur) ((cur) - buffer->line_base) | |
2180 | ||
2181 | /* Maybe put these in the ISTABLE eventually. */ | |
2182 | #define IS_HSPACE(c) ((c) == ' ' || (c) == '\t') | |
2183 | #define IS_NEWLINE(c) ((c) == '\n' || (c) == '\r') | |
2184 | ||
2185 | /* Handle LF, CR, CR-LF and LF-CR style newlines. Assumes next | |
2186 | character, if any, is in buffer. */ | |
2187 | #define handle_newline(cur, limit, c) \ | |
2188 | do {\ | |
2189 | if ((cur) < (limit) && *(cur) == '\r' + '\n' - c) \ | |
2190 | (cur)++; \ | |
2191 | CPP_BUMP_LINE_CUR (pfile, (cur)); \ | |
2192 | } while (0) | |
2193 | ||
2194 | #define IMMED_TOKEN() (!(cur_token->flags & PREV_WHITESPACE)) | |
2195 | #define PREV_TOKEN_TYPE (cur_token[-1].type) | |
2196 | ||
2197 | #define SPELL_TEXT 0 | |
2198 | #define SPELL_HANDLER 1 | |
cfd5b8b8 NB |
2199 | #define SPELL_CHAR 2 |
2200 | #define SPELL_NONE 3 | |
2201 | #define SPELL_EOL 4 | |
c5a04734 ZW |
2202 | |
2203 | #define T(e, s) {SPELL_TEXT, s}, | |
2204 | #define H(e, s) {SPELL_HANDLER, s}, | |
cfd5b8b8 | 2205 | #define C(e, s) {SPELL_CHAR, s}, |
c5a04734 ZW |
2206 | #define N(e, s) {SPELL_NONE, s}, |
2207 | #define E(e, s) {SPELL_EOL, s}, | |
2208 | ||
2209 | static const struct token_spelling | |
2210 | { | |
cfd5b8b8 | 2211 | unsigned char type; |
c5a04734 ZW |
2212 | PTR speller; |
2213 | } token_spellings [N_TTYPES + 1] = {TTYPE_TABLE {0, 0} }; | |
2214 | ||
2215 | #undef T | |
2216 | #undef H | |
cfd5b8b8 | 2217 | #undef C |
c5a04734 ZW |
2218 | #undef N |
2219 | #undef E | |
2220 | ||
2221 | static const unsigned char *digraph_spellings [] = {"%:", "%:%:", "<:", | |
2222 | ":>", "<%", "%>"}; | |
2223 | ||
2224 | static void | |
2225 | expand_comment_space (list) | |
2226 | cpp_toklist *list; | |
2227 | { | |
2228 | if (list->comments_cap == 0) | |
2229 | { | |
2230 | list->comments_cap = 10; | |
2231 | list->comments = (cpp_token *) | |
2232 | xmalloc (list->comments_cap * sizeof (cpp_token)); | |
2233 | } | |
2234 | else | |
2235 | { | |
2236 | list->comments_cap *= 2; | |
2237 | list->comments = (cpp_token *) | |
2238 | xrealloc (list->comments, list->comments_cap); | |
2239 | } | |
2240 | } | |
2241 | ||
2242 | void | |
2243 | cpp_free_token_list (list) | |
2244 | cpp_toklist *list; | |
2245 | { | |
2246 | if (list->comments) | |
2247 | free (list->comments); | |
cfd5b8b8 | 2248 | free (list->tokens - 1); /* Backup over dummy token. */ |
c5a04734 ZW |
2249 | free (list->namebuf); |
2250 | free (list); | |
2251 | } | |
2252 | ||
cfd5b8b8 | 2253 | static unsigned char trigraph_map[256]; |
c5a04734 ZW |
2254 | |
2255 | void | |
2256 | init_trigraph_map () | |
2257 | { | |
2258 | trigraph_map['='] = '#'; | |
2259 | trigraph_map['('] = '['; | |
2260 | trigraph_map[')'] = ']'; | |
2261 | trigraph_map['/'] = '\\'; | |
2262 | trigraph_map['\''] = '^'; | |
2263 | trigraph_map['<'] = '{'; | |
2264 | trigraph_map['>'] = '}'; | |
2265 | trigraph_map['!'] = '|'; | |
2266 | trigraph_map['-'] = '~'; | |
2267 | } | |
2268 | ||
2269 | /* Call when a trigraph is encountered. It warns if necessary, and | |
2270 | returns true if the trigraph should be honoured. END is the third | |
2271 | character of a trigraph in the input stream. */ | |
2272 | static int | |
2273 | trigraph_ok (pfile, end) | |
2274 | cpp_reader *pfile; | |
2275 | const unsigned char *end; | |
2276 | { | |
2277 | int accept = CPP_OPTION (pfile, trigraphs); | |
2278 | ||
2279 | if (CPP_OPTION (pfile, warn_trigraphs)) | |
2280 | { | |
2281 | unsigned int col = end - 1 - pfile->buffer->line_base; | |
2282 | if (accept) | |
2283 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2284 | "trigraph ??%c converted to %c", | |
2285 | (int) *end, (int) trigraph_map[*end]); | |
2286 | else | |
2287 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2288 | "trigraph ??%c ignored", (int) *end); | |
2289 | } | |
2290 | return accept; | |
2291 | } | |
2292 | ||
2293 | /* Scan a string for trigraphs, warning or replacing them inline as | |
2294 | appropriate. When parsing a string, we must call this routine | |
2295 | before processing a newline character (if trigraphs are enabled), | |
2296 | since the newline might be escaped by a preceding backslash | |
2297 | trigraph sequence. Returns a pointer to the end of the name after | |
2298 | replacement. */ | |
2299 | ||
2300 | static unsigned char* | |
2301 | trigraph_replace (pfile, src, limit) | |
2302 | cpp_reader *pfile; | |
2303 | unsigned char *src; | |
2304 | unsigned char* limit; | |
2305 | { | |
2306 | unsigned char *dest; | |
2307 | ||
2308 | /* Starting with src[1], find two consecutive '?'. The case of no | |
2309 | trigraphs is streamlined. */ | |
2310 | ||
2311 | for (; src + 1 < limit; src += 2) | |
2312 | { | |
2313 | if (src[0] != '?') | |
2314 | continue; | |
2315 | ||
2316 | /* Make src point to the 1st (NOT 2nd) of two consecutive '?'s. */ | |
2317 | if (src[-1] == '?') | |
2318 | src--; | |
2319 | else if (src + 2 == limit || src[1] != '?') | |
2320 | continue; | |
2321 | ||
2322 | /* Check if it really is a trigraph. */ | |
2323 | if (trigraph_map[src[2]] == 0) | |
2324 | continue; | |
2325 | ||
2326 | dest = src; | |
2327 | goto trigraph_found; | |
2328 | } | |
2329 | return limit; | |
2330 | ||
2331 | /* Now we have a trigraph, we need to scan the remaining buffer, and | |
2332 | copy-shifting its contents left if replacement is enabled. */ | |
2333 | for (; src + 2 < limit; dest++, src++) | |
2334 | if ((*dest = *src) == '?' && src[1] == '?' && trigraph_map[src[2]]) | |
2335 | { | |
2336 | trigraph_found: | |
2337 | src += 2; | |
2338 | if (trigraph_ok (pfile, pfile->buffer->cur - (limit - src))) | |
2339 | *dest = trigraph_map[*src]; | |
2340 | } | |
2341 | ||
2342 | /* Copy remaining (at most 2) characters. */ | |
2343 | while (src < limit) | |
2344 | *dest++ = *src++; | |
2345 | return dest; | |
2346 | } | |
2347 | ||
2348 | /* If CUR is a backslash or the end of a trigraphed backslash, return | |
2349 | a pointer to its beginning, otherwise NULL. We don't read beyond | |
2350 | the buffer start, because there is the start of the comment in the | |
2351 | buffer. */ | |
2352 | static const unsigned char * | |
2353 | backslash_start (pfile, cur) | |
2354 | cpp_reader *pfile; | |
2355 | const unsigned char *cur; | |
2356 | { | |
2357 | if (cur[0] == '\\') | |
2358 | return cur; | |
2359 | if (cur[0] == '/' && cur[-1] == '?' && cur[-2] == '?' | |
2360 | && trigraph_ok (pfile, cur)) | |
2361 | return cur - 2; | |
2362 | return 0; | |
2363 | } | |
2364 | ||
2365 | /* Skip a C-style block comment. This is probably the trickiest | |
2366 | handler. We find the end of the comment by seeing if an asterisk | |
2367 | is before every '/' we encounter. The nasty complication is that a | |
2368 | previous asterisk may be separated by one or more escaped newlines. | |
2369 | Returns non-zero if comment terminated by EOF, zero otherwise. */ | |
2370 | static int | |
2371 | skip_block_comment (pfile) | |
2372 | cpp_reader *pfile; | |
2373 | { | |
2374 | cpp_buffer *buffer = pfile->buffer; | |
2375 | const unsigned char *char_after_star = 0; | |
2376 | register const unsigned char *cur = buffer->cur; | |
2377 | int seen_eof = 0; | |
2378 | ||
2379 | /* Inner loop would think the comment has ended if the first comment | |
2380 | character is a '/'. Avoid this and keep the inner loop clean by | |
2381 | skipping such a character. */ | |
2382 | if (cur < buffer->rlimit && cur[0] == '/') | |
2383 | cur++; | |
2384 | ||
2385 | for (; cur < buffer->rlimit; ) | |
2386 | { | |
2387 | unsigned char c = *cur++; | |
2388 | ||
2389 | /* People like decorating comments with '*', so check for | |
2390 | '/' instead for efficiency. */ | |
2391 | if (c == '/') | |
2392 | { | |
2393 | if (cur[-2] == '*' || cur - 1 == char_after_star) | |
2394 | goto out; | |
2395 | ||
2396 | /* Warn about potential nested comments, but not when | |
2397 | the final character inside the comment is a '/'. | |
2398 | Don't bother to get it right across escaped newlines. */ | |
2399 | if (CPP_OPTION (pfile, warn_comments) && cur + 1 < buffer->rlimit | |
2400 | && cur[0] == '*' && cur[1] != '/') | |
2401 | { | |
2402 | buffer->cur = cur; | |
2403 | cpp_warning (pfile, "'/*' within comment"); | |
2404 | } | |
2405 | } | |
2406 | else if (IS_NEWLINE(c)) | |
2407 | { | |
2408 | const unsigned char* bslash = backslash_start (pfile, cur - 2); | |
2409 | ||
2410 | handle_newline (cur, buffer->rlimit, c); | |
2411 | /* Work correctly if there is an asterisk before an | |
2412 | arbirtrarily long sequence of escaped newlines. */ | |
2413 | if (bslash && (bslash[-1] == '*' || bslash == char_after_star)) | |
2414 | char_after_star = cur; | |
2415 | else | |
2416 | char_after_star = 0; | |
2417 | } | |
2418 | } | |
2419 | seen_eof = 1; | |
2420 | ||
2421 | out: | |
2422 | buffer->cur = cur; | |
2423 | return seen_eof; | |
2424 | } | |
2425 | ||
2426 | /* Skip a C++ or Chill line comment. Handles escaped newlines. | |
2427 | Returns non-zero if a multiline comment. */ | |
2428 | static int | |
2429 | skip_line_comment (pfile) | |
2430 | cpp_reader *pfile; | |
2431 | { | |
2432 | cpp_buffer *buffer = pfile->buffer; | |
2433 | register const unsigned char *cur = buffer->cur; | |
2434 | int multiline = 0; | |
2435 | ||
2436 | for (; cur < buffer->rlimit; ) | |
2437 | { | |
2438 | unsigned char c = *cur++; | |
2439 | ||
2440 | if (IS_NEWLINE (c)) | |
2441 | { | |
2442 | /* Check for a (trigaph?) backslash escaping the newline. */ | |
2443 | if (!backslash_start (pfile, cur - 2)) | |
2444 | goto out; | |
2445 | multiline = 1; | |
2446 | handle_newline (cur, buffer->rlimit, c); | |
2447 | } | |
2448 | } | |
2449 | cur++; | |
2450 | ||
2451 | out: | |
2452 | buffer->cur = cur - 1; /* Leave newline for caller. */ | |
2453 | return multiline; | |
2454 | } | |
2455 | ||
2456 | /* Skips whitespace, stopping at next non-whitespace character. */ | |
2457 | static void | |
2458 | skip_whitespace (pfile, in_directive) | |
2459 | cpp_reader *pfile; | |
2460 | int in_directive; | |
2461 | { | |
2462 | cpp_buffer *buffer = pfile->buffer; | |
2463 | register const unsigned char *cur = buffer->cur; | |
2464 | unsigned short null_count = 0; | |
2465 | ||
2466 | for (; cur < buffer->rlimit; ) | |
2467 | { | |
2468 | unsigned char c = *cur++; | |
2469 | ||
2470 | if (IS_HSPACE(c)) /* FIXME: Fix ISTABLE. */ | |
2471 | continue; | |
2472 | if (!is_space(c) || IS_NEWLINE (c)) /* Main loop handles newlines. */ | |
2473 | goto out; | |
2474 | if (c == '\0') | |
2475 | null_count++; | |
2476 | /* Mut be '\f' or '\v' */ | |
2477 | else if (in_directive && CPP_PEDANTIC (pfile)) | |
2478 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
2479 | c == '\f' ? "formfeed" : "vertical tab"); | |
2480 | } | |
2481 | cur++; | |
2482 | ||
2483 | out: | |
2484 | buffer->cur = cur - 1; | |
2485 | if (null_count) | |
2486 | cpp_warning (pfile, null_count > 1 ? "embedded null characters ignored" | |
2487 | : "embedded null character ignored"); | |
2488 | } | |
2489 | ||
2490 | /* Parse (append) an identifier. */ | |
2491 | static void | |
2492 | parse_name (pfile, list, name) | |
2493 | cpp_reader *pfile; | |
2494 | cpp_toklist *list; | |
2495 | cpp_name *name; | |
2496 | { | |
2497 | const unsigned char *name_limit; | |
2498 | unsigned char *namebuf; | |
2499 | cpp_buffer *buffer = pfile->buffer; | |
2500 | register const unsigned char *cur = buffer->cur; | |
2501 | ||
2502 | expanded: | |
2503 | name_limit = list->namebuf + list->name_cap; | |
2504 | namebuf = list->namebuf + list->name_used; | |
2505 | ||
2506 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2507 | { | |
2508 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2509 | ||
2510 | if (! is_idchar(c)) | |
2511 | goto out; | |
2512 | namebuf++; | |
2513 | cur++; | |
2514 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
2515 | { | |
2516 | buffer->cur = cur; | |
2517 | cpp_pedwarn (pfile, "'$' character in identifier"); | |
2518 | } | |
2519 | } | |
2520 | ||
2521 | /* Run out of name space? */ | |
2522 | if (cur < buffer->rlimit) | |
2523 | { | |
2524 | list->name_used = namebuf - list->namebuf; | |
2525 | auto_expand_name_space (list); | |
2526 | goto expanded; | |
2527 | } | |
2528 | ||
2529 | out: | |
2530 | buffer->cur = cur; | |
2531 | name->len = namebuf - (list->namebuf + name->offset); | |
2532 | list->name_used = namebuf - list->namebuf; | |
2533 | } | |
2534 | ||
2535 | /* Parse (append) a number. */ | |
2536 | ||
2537 | #define VALID_SIGN(c, prevc) \ | |
2538 | (((c) == '+' || (c) == '-') && \ | |
2539 | ((prevc) == 'e' || (prevc) == 'E' \ | |
2540 | || (((prevc) == 'p' || (prevc) == 'P') && !CPP_OPTION (pfile, c89)))) | |
2541 | ||
2542 | static void | |
2543 | parse_number (pfile, list, name) | |
2544 | cpp_reader *pfile; | |
2545 | cpp_toklist *list; | |
2546 | cpp_name *name; | |
2547 | { | |
2548 | const unsigned char *name_limit; | |
2549 | unsigned char *namebuf; | |
2550 | cpp_buffer *buffer = pfile->buffer; | |
2551 | register const unsigned char *cur = buffer->cur; | |
2552 | ||
2553 | expanded: | |
2554 | name_limit = list->namebuf + list->name_cap; | |
2555 | namebuf = list->namebuf + list->name_used; | |
2556 | ||
2557 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2558 | { | |
2559 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2560 | ||
2561 | /* Perhaps we should accept '$' here if we accept it for | |
2562 | identifiers. We know namebuf[-1] is safe, because for c to | |
2563 | be a sign we must have pushed at least one character. */ | |
2564 | if (!is_numchar (c) && c != '.' && ! VALID_SIGN (c, namebuf[-1])) | |
2565 | goto out; | |
2566 | ||
2567 | namebuf++; | |
2568 | cur++; | |
2569 | } | |
2570 | ||
2571 | /* Run out of name space? */ | |
2572 | if (cur < buffer->rlimit) | |
2573 | { | |
2574 | list->name_used = namebuf - list->namebuf; | |
2575 | auto_expand_name_space (list); | |
2576 | goto expanded; | |
2577 | } | |
2578 | ||
2579 | out: | |
2580 | buffer->cur = cur; | |
2581 | name->len = namebuf - (list->namebuf + name->offset); | |
2582 | list->name_used = namebuf - list->namebuf; | |
2583 | } | |
2584 | ||
2585 | /* Places a string terminated by an unescaped TERMINATOR into a | |
2586 | cpp_name, which should be expandable and thus at the top of the | |
2587 | list's stack. Handles embedded trigraphs, if necessary, and | |
2588 | escaped newlines. | |
2589 | ||
2590 | Can be used for character constants (terminator = '\''), string | |
2591 | constants ('"'), angled headers ('>') and assertions (')'). */ | |
2592 | ||
2593 | static void | |
2594 | parse_string (pfile, list, name, terminator) | |
2595 | cpp_reader *pfile; | |
2596 | cpp_toklist *list; | |
2597 | cpp_name *name; | |
2598 | unsigned int terminator; | |
2599 | { | |
2600 | cpp_buffer *buffer = pfile->buffer; | |
2601 | register const unsigned char *cur = buffer->cur; | |
2602 | const unsigned char *name_limit; | |
2603 | unsigned char *namebuf; | |
2604 | unsigned int null_count = 0; | |
2605 | int trigraphed_len = 0; | |
2606 | ||
2607 | expanded: | |
2608 | name_limit = list->namebuf + list->name_cap; | |
2609 | namebuf = list->namebuf + list->name_used; | |
2610 | ||
2611 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2612 | { | |
2613 | unsigned int c = *namebuf++ = *cur++; /* Copy a single char. */ | |
2614 | ||
2615 | if (c == '\0') | |
2616 | null_count++; | |
2617 | else if (c == terminator || IS_NEWLINE (c)) | |
2618 | { | |
2619 | unsigned char* name_start = list->namebuf + name->offset; | |
2620 | ||
2621 | /* Needed for trigraph_replace and multiline string warning. */ | |
2622 | buffer->cur = cur; | |
2623 | ||
2624 | /* Scan for trigraphs before checking if backslash-escaped. */ | |
2625 | if (CPP_OPTION (pfile, trigraphs) | |
2626 | || CPP_OPTION (pfile, warn_trigraphs)) | |
2627 | { | |
2628 | namebuf = trigraph_replace (pfile, name_start + trigraphed_len, | |
2629 | namebuf); | |
2630 | trigraphed_len = namebuf - 2 - (name_start + trigraphed_len); | |
2631 | if (trigraphed_len < 0) | |
2632 | trigraphed_len = 0; | |
2633 | } | |
2634 | ||
2635 | namebuf--; /* Drop the newline / terminator from the name. */ | |
2636 | if (IS_NEWLINE (c)) | |
2637 | { | |
2638 | /* Drop a backslash newline, and continue. */ | |
2639 | if (namebuf[-1] == '\\') | |
2640 | { | |
2641 | handle_newline (cur, buffer->rlimit, c); | |
2642 | namebuf--; | |
2643 | continue; | |
2644 | } | |
2645 | ||
2646 | cur--; | |
2647 | ||
2648 | /* In Fortran and assembly language, silently terminate | |
2649 | strings of either variety at end of line. This is a | |
2650 | kludge around not knowing where comments are in these | |
2651 | languages. */ | |
2652 | if (CPP_OPTION (pfile, lang_fortran) | |
2653 | || CPP_OPTION (pfile, lang_asm)) | |
2654 | goto out; | |
2655 | ||
2656 | /* Character constants, headers and asserts may not | |
2657 | extend over multiple lines. In Standard C, neither | |
2658 | may strings. We accept multiline strings as an | |
2659 | extension, but not in directives. */ | |
2660 | if (terminator != '"' || IS_DIRECTIVE (list)) | |
2661 | goto unterminated; | |
2662 | ||
2663 | cur++; /* Move forwards again. */ | |
2664 | ||
2665 | if (pfile->multiline_string_line == 0) | |
2666 | { | |
2667 | pfile->multiline_string_line = list->line; | |
2668 | if (CPP_PEDANTIC (pfile)) | |
2669 | cpp_pedwarn (pfile, "multi-line string constant"); | |
2670 | } | |
2671 | ||
2672 | *namebuf++ = '\n'; | |
2673 | handle_newline (cur, buffer->rlimit, c); | |
2674 | } | |
2675 | else | |
2676 | { | |
2677 | unsigned char *temp; | |
2678 | ||
2679 | /* An odd number of consecutive backslashes represents | |
2680 | an escaped terminator. */ | |
2681 | temp = namebuf - 1; | |
2682 | while (temp >= name_start && *temp == '\\') | |
2683 | temp--; | |
2684 | ||
2685 | if ((namebuf - temp) & 1) | |
2686 | goto out; | |
2687 | namebuf++; | |
2688 | } | |
2689 | } | |
2690 | } | |
2691 | ||
2692 | /* Run out of name space? */ | |
2693 | if (cur < buffer->rlimit) | |
2694 | { | |
2695 | list->name_used = namebuf - list->namebuf; | |
2696 | auto_expand_name_space (list); | |
2697 | goto expanded; | |
2698 | } | |
2699 | ||
2700 | /* We may not have trigraph-replaced the input for this code path, | |
2701 | but as the input is in error by being unterminated we don't | |
2702 | bother. Prevent warnings about no newlines at EOF. */ | |
2703 | if (IS_NEWLINE(cur[-1])) | |
2704 | cur--; | |
2705 | ||
2706 | unterminated: | |
2707 | cpp_error (pfile, "missing terminating %c character", (int) terminator); | |
2708 | ||
2709 | if (terminator == '\"' && pfile->multiline_string_line != list->line | |
2710 | && pfile->multiline_string_line != 0) | |
2711 | { | |
2712 | cpp_error_with_line (pfile, pfile->multiline_string_line, -1, | |
2713 | "possible start of unterminated string literal"); | |
2714 | pfile->multiline_string_line = 0; | |
2715 | } | |
2716 | ||
2717 | out: | |
2718 | buffer->cur = cur; | |
2719 | name->len = namebuf - (list->namebuf + name->offset); | |
2720 | list->name_used = namebuf - list->namebuf; | |
2721 | ||
2722 | if (null_count > 0) | |
2723 | cpp_warning (pfile, (null_count > 1 ? "null characters preserved" | |
2724 | : "null character preserved")); | |
2725 | } | |
2726 | ||
2727 | /* The character C helps us distinguish comment types: '*' = C style, | |
2728 | '-' = Chill-style and '/' = C++ style. For code simplicity, the | |
2729 | stored comment includes any C-style comment terminator. */ | |
2730 | static void | |
2731 | copy_comment (list, from, len, tok_no, type) | |
2732 | cpp_toklist *list; | |
2733 | const unsigned char *from; | |
2734 | unsigned int len; | |
2735 | unsigned int tok_no; | |
2736 | unsigned int type; | |
2737 | { | |
2738 | cpp_token *comment; | |
2739 | ||
2740 | if (list->comments_used == list->comments_cap) | |
2741 | expand_comment_space (list); | |
2742 | ||
2743 | if (list->name_used + len > list->name_cap) | |
2744 | expand_name_space (list, len); | |
2745 | ||
2746 | comment = &list->comments[list->comments_used++]; | |
2747 | comment->type = type; | |
2748 | comment->aux = tok_no; | |
2749 | comment->val.name.len = len; | |
2750 | comment->val.name.offset = list->name_used; | |
2751 | ||
2752 | memcpy (list->namebuf + list->name_used, from, len); | |
2753 | list->name_used += len; | |
2754 | } | |
2755 | ||
2756 | /* | |
2757 | * The tokenizer's main loop. Returns a token list, representing a | |
2758 | * logical line in the input file, terminated with a CPP_VSPACE | |
2759 | * token. On EOF, a token list containing the single CPP_EOF token | |
2760 | * is returned. | |
2761 | * | |
2762 | * Implementation relies almost entirely on lookback, rather than | |
2763 | * looking forwards. This means that tokenization requires just | |
2764 | * a single pass of the file, even in the presence of trigraphs and | |
2765 | * escaped newlines, providing significant performance benefits. | |
2766 | * Trigraph overhead is negligible if they are disabled, and low | |
2767 | * even when enabled. | |
2768 | */ | |
2769 | ||
2770 | #define PUSH_TOKEN(ttype) cur_token++->type = ttype | |
2771 | #define REVISE_TOKEN(ttype) cur_token[-1].type = ttype | |
2772 | #define BACKUP_TOKEN(ttype) (--cur_token)->type = ttype | |
2773 | #define BACKUP_DIGRAPH(ttype) do { \ | |
2774 | BACKUP_TOKEN(ttype); cur_token->flags |= DIGRAPH;} while (0) | |
2775 | ||
2776 | void | |
2777 | _cpp_lex_line (pfile, list) | |
2778 | cpp_reader *pfile; | |
2779 | cpp_toklist *list; | |
2780 | { | |
2781 | cpp_token *cur_token, *token_limit; | |
2782 | cpp_buffer *buffer = pfile->buffer; | |
2783 | register const unsigned char *cur = buffer->cur; | |
2784 | unsigned char flags = 0; | |
2785 | ||
2786 | expanded: | |
2787 | token_limit = list->tokens + list->tokens_cap; | |
2788 | cur_token = list->tokens + list->tokens_used; | |
2789 | ||
2790 | for (; cur < buffer->rlimit && cur_token < token_limit;) | |
2791 | { | |
2792 | unsigned char c = *cur++; | |
2793 | ||
2794 | /* Optimize whitespace skipping, in particular the case of a | |
2795 | single whitespace character, as every other token is probably | |
2796 | whitespace. (' ' '\t' '\v' '\f' '\0'). */ | |
2797 | if (is_hspace ((unsigned int) c)) | |
2798 | { | |
2799 | if (c == '\0' || (cur < buffer->rlimit && is_hspace (*cur))) | |
2800 | { | |
2801 | buffer->cur = cur - (c == '\0'); /* Get the null warning. */ | |
2802 | skip_whitespace (pfile, IS_DIRECTIVE (list)); | |
2803 | cur = buffer->cur; | |
2804 | } | |
2805 | flags = PREV_WHITESPACE; | |
2806 | if (cur == buffer->rlimit) | |
2807 | break; | |
2808 | c = *cur++; | |
2809 | } | |
2810 | ||
2811 | /* Initialize current token. Its type is set in the switch. */ | |
2812 | cur_token->col = COLUMN (cur); | |
2813 | cur_token->flags = flags; | |
2814 | flags = 0; | |
2815 | ||
2816 | switch (c) | |
2817 | { | |
2818 | case '0': case '1': case '2': case '3': case '4': | |
2819 | case '5': case '6': case '7': case '8': case '9': | |
2820 | /* Prepend an immediately previous CPP_DOT token. */ | |
2821 | if (PREV_TOKEN_TYPE == CPP_DOT && IMMED_TOKEN ()) | |
2822 | { | |
2823 | cur_token--; | |
2824 | if (list->name_cap == list->name_used) | |
2825 | auto_expand_name_space (list); | |
2826 | ||
2827 | cur_token->val.name.len = 1; | |
2828 | cur_token->val.name.offset = list->name_used; | |
2829 | list->namebuf[list->name_used++] = '.'; | |
2830 | } | |
2831 | else | |
2832 | INIT_NAME (list, cur_token->val.name); | |
2833 | cur--; /* Backup character. */ | |
2834 | ||
2835 | continue_number: | |
2836 | buffer->cur = cur; | |
2837 | parse_number (pfile, list, &cur_token->val.name); | |
2838 | cur = buffer->cur; | |
2839 | ||
2840 | PUSH_TOKEN (CPP_NUMBER); /* Number not yet interpreted. */ | |
2841 | break; | |
2842 | ||
2843 | letter: | |
2844 | case '_': | |
2845 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
2846 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
2847 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
2848 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
2849 | case 'y': case 'z': | |
2850 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
2851 | case 'G': case 'H': case 'I': case 'J': case 'K': case 'L': | |
2852 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': | |
2853 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
2854 | case 'Y': case 'Z': | |
2855 | INIT_NAME (list, cur_token->val.name); | |
2856 | cur--; /* Backup character. */ | |
2857 | cur_token->type = CPP_NAME; /* Identifier, macro etc. */ | |
2858 | ||
2859 | continue_name: | |
2860 | buffer->cur = cur; | |
2861 | parse_name (pfile, list, &cur_token->val.name); | |
2862 | cur = buffer->cur; | |
2863 | ||
2864 | /* Find handler for newly created / extended directive. */ | |
2865 | if (IS_DIRECTIVE (list) && cur_token == &list->tokens[1]) | |
2866 | _cpp_check_directive (list, cur_token); | |
2867 | cur_token++; | |
2868 | break; | |
2869 | ||
2870 | case '\'': | |
2871 | /* Fall through. */ | |
2872 | case '\"': | |
2873 | cur_token->type = c == '\'' ? CPP_CHAR : CPP_STRING; | |
2874 | /* Do we have a wide string? */ | |
2875 | if (cur_token[-1].type == CPP_NAME && IMMED_TOKEN () | |
2876 | && cur_token[-1].val.name.len == 1 | |
2877 | && TOK_NAME (list, cur_token - 1)[0] == 'L' | |
2878 | && !CPP_TRADITIONAL (pfile)) | |
2879 | { | |
2880 | /* No need for 'L' any more. */ | |
2881 | list->name_used--; | |
2882 | (--cur_token)->type = (c == '\'' ? CPP_WCHAR : CPP_WSTRING); | |
2883 | } | |
2884 | ||
2885 | do_parse_string: | |
2886 | /* Here c is one of ' " > or ). */ | |
2887 | INIT_NAME (list, cur_token->val.name); | |
2888 | buffer->cur = cur; | |
2889 | parse_string (pfile, list, &cur_token->val.name, c); | |
2890 | cur = buffer->cur; | |
2891 | cur_token++; | |
2892 | break; | |
2893 | ||
2894 | case '/': | |
2895 | cur_token->type = CPP_DIV; | |
2896 | if (IMMED_TOKEN ()) | |
2897 | { | |
2898 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
2899 | { | |
2900 | /* We silently allow C++ comments in system headers, | |
2901 | irrespective of conformance mode, because lots of | |
2902 | broken systems do that and trying to clean it up | |
2903 | in fixincludes is a nightmare. */ | |
2904 | if (buffer->system_header_p) | |
2905 | goto do_line_comment; | |
2906 | else if (CPP_OPTION (pfile, cplusplus_comments)) | |
2907 | { | |
2908 | if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile) | |
2909 | && ! buffer->warned_cplusplus_comments) | |
2910 | { | |
2911 | buffer->cur = cur; | |
2912 | cpp_pedwarn (pfile, | |
2913 | "C++ style comments are not allowed in ISO C89"); | |
2914 | cpp_pedwarn (pfile, | |
2915 | "(this will be reported only once per input file)"); | |
2916 | buffer->warned_cplusplus_comments = 1; | |
2917 | } | |
2918 | do_line_comment: | |
2919 | buffer->cur = cur; | |
2920 | if (cur[-2] != c) | |
2921 | cpp_warning (pfile, | |
2922 | "comment start split across lines"); | |
2923 | if (skip_line_comment (pfile)) | |
2924 | cpp_error_with_line (pfile, list->line, | |
2925 | cur_token[-1].col, | |
2926 | "multi-line comment"); | |
2927 | if (!CPP_OPTION (pfile, discard_comments)) | |
2928 | copy_comment (list, cur, buffer->cur - cur, | |
2929 | cur_token - 1 - list->tokens, c == '/' | |
2930 | ? CPP_CPP_COMMENT: CPP_CHILL_COMMENT); | |
2931 | cur = buffer->cur; | |
2932 | ||
2933 | /* Back-up to first '-' or '/'. */ | |
2934 | cur_token -= 2; | |
2935 | if (!CPP_OPTION (pfile, traditional)) | |
2936 | flags = PREV_WHITESPACE; | |
2937 | } | |
2938 | } | |
2939 | } | |
2940 | cur_token++; | |
2941 | break; | |
2942 | ||
2943 | case '*': | |
2944 | cur_token->type = CPP_MULT; | |
2945 | if (IMMED_TOKEN ()) | |
2946 | { | |
2947 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
2948 | { | |
2949 | buffer->cur = cur; | |
2950 | if (cur[-2] != '/') | |
2951 | cpp_warning (pfile, | |
2952 | "comment start '/*' split across lines"); | |
2953 | if (skip_block_comment (pfile)) | |
2954 | cpp_error_with_line (pfile, list->line, cur_token[-1].col, | |
2955 | "unterminated comment"); | |
2956 | else if (buffer->cur[-2] != '*') | |
2957 | cpp_warning (pfile, | |
2958 | "comment end '*/' split across lines"); | |
2959 | if (!CPP_OPTION (pfile, discard_comments)) | |
2960 | copy_comment (list, cur, buffer->cur - cur, | |
2961 | cur_token - 1 - list->tokens, CPP_C_COMMENT); | |
2962 | cur = buffer->cur; | |
2963 | ||
2964 | cur_token -= 2; | |
2965 | if (!CPP_OPTION (pfile, traditional)) | |
2966 | flags = PREV_WHITESPACE; | |
2967 | } | |
2968 | else if (CPP_OPTION (pfile, cplusplus)) | |
2969 | { | |
2970 | /* In C++, there are .* and ->* operators. */ | |
2971 | if (PREV_TOKEN_TYPE == CPP_DEREF) | |
2972 | BACKUP_TOKEN (CPP_DEREF_STAR); | |
2973 | else if (PREV_TOKEN_TYPE == CPP_DOT) | |
2974 | BACKUP_TOKEN (CPP_DOT_STAR); | |
2975 | } | |
2976 | } | |
2977 | cur_token++; | |
2978 | break; | |
2979 | ||
2980 | case '\n': | |
2981 | case '\r': | |
2982 | handle_newline (cur, buffer->rlimit, c); | |
2983 | if (PREV_TOKEN_TYPE != CPP_BACKSLASH || !IMMED_TOKEN ()) | |
2984 | { | |
2985 | if (PREV_TOKEN_TYPE == CPP_BACKSLASH) | |
2986 | { | |
2987 | buffer->cur = cur; | |
2988 | cpp_warning (pfile, | |
2989 | "backslash and newline separated by space"); | |
2990 | } | |
2991 | PUSH_TOKEN (CPP_VSPACE); | |
2992 | goto out; | |
2993 | } | |
2994 | /* Remove the escaped newline. Then continue to process | |
2995 | any interrupted name or number. */ | |
2996 | cur_token--; | |
2997 | if (IMMED_TOKEN ()) | |
2998 | { | |
2999 | cur_token--; | |
3000 | if (cur_token->type == CPP_NAME) | |
3001 | goto continue_name; | |
3002 | else if (cur_token->type == CPP_NUMBER) | |
3003 | goto continue_number; | |
3004 | cur_token++; | |
3005 | } | |
3006 | break; | |
3007 | ||
3008 | case '-': | |
3009 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_MINUS) | |
3010 | { | |
3011 | if (CPP_OPTION (pfile, chill)) | |
3012 | goto do_line_comment; | |
3013 | REVISE_TOKEN (CPP_MINUS_MINUS); | |
3014 | } | |
3015 | else | |
3016 | PUSH_TOKEN (CPP_MINUS); | |
3017 | break; | |
3018 | ||
3019 | /* The digraph flag checking ensures that ## and %:%: | |
3020 | are interpreted as CPP_PASTE, but #%: and %:# are not. */ | |
3021 | make_hash: | |
3022 | case '#': | |
3023 | if (PREV_TOKEN_TYPE == CPP_HASH && IMMED_TOKEN () | |
3024 | && ((cur_token->flags ^ cur_token[-1].flags) & DIGRAPH) == 0) | |
3025 | REVISE_TOKEN (CPP_PASTE); | |
3026 | else | |
3027 | PUSH_TOKEN (CPP_HASH); | |
3028 | break; | |
3029 | ||
3030 | case ':': | |
3031 | cur_token->type = CPP_COLON; | |
3032 | if (IMMED_TOKEN ()) | |
3033 | { | |
3034 | if (PREV_TOKEN_TYPE == CPP_COLON | |
3035 | && CPP_OPTION (pfile, cplusplus)) | |
3036 | BACKUP_TOKEN (CPP_SCOPE); | |
3037 | /* Digraph: "<:" is a '[' */ | |
3038 | else if (PREV_TOKEN_TYPE == CPP_LESS) | |
3039 | BACKUP_DIGRAPH (CPP_OPEN_SQUARE); | |
3040 | /* Digraph: "%:" is a '#' */ | |
3041 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3042 | { | |
3043 | (--cur_token)->flags |= DIGRAPH; | |
3044 | goto make_hash; | |
3045 | } | |
3046 | } | |
3047 | cur_token++; | |
3048 | break; | |
3049 | ||
3050 | case '&': | |
3051 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_AND) | |
3052 | REVISE_TOKEN (CPP_AND_AND); | |
3053 | else | |
3054 | PUSH_TOKEN (CPP_AND); | |
3055 | break; | |
3056 | ||
3057 | make_or: | |
3058 | case '|': | |
3059 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_OR) | |
3060 | REVISE_TOKEN (CPP_OR_OR); | |
3061 | else | |
3062 | PUSH_TOKEN (CPP_OR); | |
3063 | break; | |
3064 | ||
3065 | case '+': | |
3066 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_PLUS) | |
3067 | REVISE_TOKEN (CPP_PLUS_PLUS); | |
3068 | else | |
3069 | PUSH_TOKEN (CPP_PLUS); | |
3070 | break; | |
3071 | ||
3072 | case '=': | |
3073 | /* This relies on equidistance of "?=" and "?" tokens. */ | |
3074 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE <= CPP_LAST_EQ) | |
3075 | REVISE_TOKEN (PREV_TOKEN_TYPE + (CPP_EQ_EQ - CPP_EQ)); | |
3076 | else | |
3077 | PUSH_TOKEN (CPP_EQ); | |
3078 | break; | |
3079 | ||
3080 | case '>': | |
3081 | cur_token->type = CPP_GREATER; | |
3082 | if (IMMED_TOKEN ()) | |
3083 | { | |
3084 | if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3085 | BACKUP_TOKEN (CPP_RSHIFT); | |
3086 | else if (PREV_TOKEN_TYPE == CPP_MINUS) | |
3087 | BACKUP_TOKEN (CPP_DEREF); | |
3088 | /* Digraph: ":>" is a ']' */ | |
3089 | else if (PREV_TOKEN_TYPE == CPP_COLON) | |
3090 | BACKUP_DIGRAPH (CPP_CLOSE_SQUARE); | |
3091 | /* Digraph: "%>" is a '}' */ | |
3092 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3093 | BACKUP_DIGRAPH (CPP_CLOSE_BRACE); | |
3094 | } | |
3095 | cur_token++; | |
3096 | break; | |
3097 | ||
3098 | case '<': | |
3099 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3100 | { | |
3101 | REVISE_TOKEN (CPP_LSHIFT); | |
3102 | break; | |
3103 | } | |
3104 | /* Is this the beginning of a header name? */ | |
3105 | if (list->dir_flags & SYNTAX_INCLUDE) | |
3106 | { | |
3107 | c = '>'; /* Terminator. */ | |
3108 | cur_token->type = CPP_HEADER_NAME; | |
3109 | goto do_parse_string; | |
3110 | } | |
3111 | PUSH_TOKEN (CPP_LESS); | |
3112 | break; | |
3113 | ||
3114 | case '%': | |
3115 | /* Digraph: "<%" is a '{' */ | |
3116 | cur_token->type = CPP_MOD; | |
3117 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3118 | BACKUP_DIGRAPH (CPP_OPEN_BRACE); | |
3119 | cur_token++; | |
3120 | break; | |
3121 | ||
c5a04734 ZW |
3122 | case '(': |
3123 | /* Is this the beginning of an assertion string? */ | |
3124 | if (list->dir_flags & SYNTAX_ASSERT) | |
3125 | { | |
3126 | c = ')'; /* Terminator. */ | |
3127 | cur_token->type = CPP_ASSERTION; | |
3128 | goto do_parse_string; | |
3129 | } | |
3130 | PUSH_TOKEN (CPP_OPEN_PAREN); | |
3131 | break; | |
3132 | ||
c5a04734 ZW |
3133 | case '?': |
3134 | if (cur + 1 < buffer->rlimit && *cur == '?' | |
3135 | && trigraph_map[cur[1]] && trigraph_ok (pfile, cur + 1)) | |
3136 | { | |
3137 | /* Handle trigraph. */ | |
3138 | cur++; | |
3139 | switch (*cur++) | |
3140 | { | |
3141 | case '(': goto make_open_square; | |
3142 | case ')': goto make_close_square; | |
3143 | case '<': goto make_open_brace; | |
3144 | case '>': goto make_close_brace; | |
3145 | case '=': goto make_hash; | |
3146 | case '!': goto make_or; | |
3147 | case '-': goto make_complement; | |
3148 | case '/': goto make_backslash; | |
3149 | case '\'': goto make_xor; | |
3150 | } | |
3151 | } | |
3152 | if (IMMED_TOKEN () && CPP_OPTION (pfile, cplusplus)) | |
3153 | { | |
3154 | /* GNU C++ defines <? and >? operators. */ | |
3155 | if (PREV_TOKEN_TYPE == CPP_LESS) | |
3156 | { | |
3157 | REVISE_TOKEN (CPP_MIN); | |
3158 | break; | |
3159 | } | |
3160 | else if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3161 | { | |
3162 | REVISE_TOKEN (CPP_MAX); | |
3163 | break; | |
3164 | } | |
3165 | } | |
3166 | PUSH_TOKEN (CPP_QUERY); | |
3167 | break; | |
3168 | ||
3169 | case '.': | |
3170 | if (PREV_TOKEN_TYPE == CPP_DOT && cur_token[-2].type == CPP_DOT | |
3171 | && IMMED_TOKEN () | |
3172 | && !(cur_token[-1].flags & PREV_WHITESPACE)) | |
3173 | { | |
3174 | cur_token -= 2; | |
3175 | PUSH_TOKEN (CPP_ELLIPSIS); | |
3176 | } | |
3177 | else | |
3178 | PUSH_TOKEN (CPP_DOT); | |
3179 | break; | |
3180 | ||
cfd5b8b8 NB |
3181 | make_complement: |
3182 | case '~': PUSH_TOKEN (CPP_COMPL); break; | |
c5a04734 ZW |
3183 | make_xor: |
3184 | case '^': PUSH_TOKEN (CPP_XOR); break; | |
3185 | make_open_brace: | |
3186 | case '{': PUSH_TOKEN (CPP_OPEN_BRACE); break; | |
3187 | make_close_brace: | |
3188 | case '}': PUSH_TOKEN (CPP_CLOSE_BRACE); break; | |
3189 | make_open_square: | |
3190 | case '[': PUSH_TOKEN (CPP_OPEN_SQUARE); break; | |
3191 | make_close_square: | |
3192 | case ']': PUSH_TOKEN (CPP_CLOSE_SQUARE); break; | |
3193 | make_backslash: | |
3194 | case '\\': PUSH_TOKEN (CPP_BACKSLASH); break; | |
3195 | case '!': PUSH_TOKEN (CPP_NOT); break; | |
3196 | case ',': PUSH_TOKEN (CPP_COMMA); break; | |
3197 | case ';': PUSH_TOKEN (CPP_SEMICOLON); break; | |
cfd5b8b8 | 3198 | case ')': PUSH_TOKEN (CPP_CLOSE_PAREN); break; |
c5a04734 ZW |
3199 | |
3200 | case '$': | |
3201 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
3202 | goto letter; | |
3203 | /* Fall through */ | |
3204 | default: | |
3205 | cur_token->aux = c; | |
3206 | PUSH_TOKEN (CPP_OTHER); | |
3207 | break; | |
3208 | } | |
3209 | } | |
3210 | ||
3211 | /* Run out of token space? */ | |
3212 | if (cur_token == token_limit) | |
3213 | { | |
3214 | list->tokens_used = cur_token - list->tokens; | |
3215 | expand_token_space (list); | |
3216 | goto expanded; | |
3217 | } | |
3218 | ||
3219 | cur_token->type = CPP_EOF; | |
3220 | cur_token->flags = flags; | |
3221 | ||
3222 | if (cur_token != &list->tokens[0]) | |
3223 | { | |
3224 | /* Next call back will get just a CPP_EOF. */ | |
3225 | buffer->cur = cur; | |
3226 | cpp_warning (pfile, "no newline at end of file"); | |
3227 | PUSH_TOKEN (CPP_VSPACE); | |
3228 | } | |
3229 | ||
3230 | out: | |
3231 | buffer->cur = cur; | |
3232 | ||
3233 | list->tokens_used = cur_token - list->tokens; | |
3234 | ||
3235 | /* FIXME: take this check out and put it in the caller. | |
3236 | list->directive == 0 indicates an unknown directive (but null | |
3237 | directive is OK). This is the first time we can be sure the | |
3238 | directive is invalid, and thus warn about it, because it might | |
3239 | have been split by escaped newlines. Also, don't complain about | |
3240 | invalid directives in assembly source, we don't know where the | |
3241 | comments are, and # may introduce assembler pseudo-ops. */ | |
3242 | ||
3243 | if (IS_DIRECTIVE (list) && list->dir_handler == 0 | |
3244 | && list->tokens[1].type != CPP_VSPACE | |
3245 | && !CPP_OPTION (pfile, lang_asm)) | |
3246 | cpp_error_with_line (pfile, list->line, list->tokens[1].col, | |
3247 | "invalid preprocessing directive"); | |
3248 | } | |
3249 | ||
3250 | /* Token spelling functions. Used for output of a preprocessed file, | |
3251 | stringizing and token pasting. They all assume sufficient buffer | |
3252 | is allocated, and return exactly how much they used. */ | |
3253 | ||
c5a04734 ZW |
3254 | /* Needs buffer of 3 + len. */ |
3255 | unsigned int | |
3256 | spell_string (buffer, list, token) | |
3257 | unsigned char *buffer; | |
3258 | cpp_toklist *list; | |
3259 | cpp_token *token; | |
3260 | { | |
cfd5b8b8 | 3261 | unsigned char c, *orig_buff = buffer; |
c5a04734 ZW |
3262 | size_t len; |
3263 | ||
cfd5b8b8 | 3264 | if (token->type == CPP_WSTRING || token->type == CPP_WCHAR) |
c5a04734 | 3265 | *buffer++ = 'L'; |
cfd5b8b8 NB |
3266 | c = token->type == CPP_STRING || token->type == CPP_WSTRING ? '"': '\''; |
3267 | *buffer++ = c; | |
c5a04734 ZW |
3268 | |
3269 | len = token->val.name.len; | |
3270 | memcpy (buffer, TOK_NAME (list, token), len); | |
3271 | buffer += len; | |
cfd5b8b8 | 3272 | *buffer++ = c; |
c5a04734 ZW |
3273 | return buffer - orig_buff; |
3274 | } | |
3275 | ||
3276 | /* Needs buffer of len + 2. */ | |
3277 | unsigned int | |
3278 | spell_comment (buffer, list, token) | |
3279 | unsigned char *buffer; | |
3280 | cpp_toklist *list; | |
3281 | cpp_token *token; | |
3282 | { | |
3283 | size_t len; | |
3284 | ||
3285 | if (token->type == CPP_C_COMMENT) | |
3286 | { | |
3287 | *buffer++ = '/'; | |
3288 | *buffer++ = '*'; | |
3289 | } | |
3290 | else if (token->type == CPP_CPP_COMMENT) | |
3291 | { | |
3292 | *buffer++ = '/'; | |
3293 | *buffer++ = '/'; | |
3294 | } | |
3295 | else | |
3296 | { | |
3297 | *buffer++ = '-'; | |
3298 | *buffer++ = '-'; | |
3299 | } | |
3300 | ||
3301 | len = token->val.name.len; | |
3302 | memcpy (buffer, TOK_NAME (list, token), len); | |
3303 | ||
3304 | return len + 2; | |
3305 | } | |
3306 | ||
3307 | /* Needs buffer of len. */ | |
3308 | unsigned int | |
3309 | spell_name (buffer, list, token) | |
3310 | unsigned char *buffer; | |
3311 | cpp_toklist *list; | |
3312 | cpp_token *token; | |
3313 | { | |
3314 | size_t len; | |
3315 | ||
3316 | len = token->val.name.len; | |
3317 | memcpy (buffer, TOK_NAME (list, token), len); | |
3318 | buffer += len; | |
3319 | ||
3320 | return len; | |
3321 | } | |
3322 | ||
c5a04734 ZW |
3323 | void |
3324 | _cpp_lex_file (pfile) | |
3325 | cpp_reader* pfile; | |
3326 | { | |
3327 | int recycle; | |
3328 | cpp_toklist* list; | |
3329 | ||
3330 | init_trigraph_map (); | |
3331 | list = (cpp_toklist *) xmalloc (sizeof (cpp_toklist)); | |
3332 | ||
3333 | for (recycle = 0; ;) | |
3334 | { | |
3335 | init_token_list (pfile, list, recycle); | |
3336 | recycle = 1; | |
3337 | ||
3338 | _cpp_lex_line (pfile, list); | |
3339 | if (list->tokens[0].type == CPP_EOF) | |
3340 | break; | |
3341 | ||
3342 | if (list->dir_handler) | |
3343 | { | |
3344 | if (list->dir_handler (pfile)) | |
3345 | { | |
3346 | list = (cpp_toklist *) xmalloc (sizeof (cpp_toklist)); | |
3347 | recycle = 0; | |
3348 | } | |
3349 | } | |
3350 | else | |
3351 | _cpp_output_list (pfile, list); | |
3352 | } | |
3353 | } | |
3354 | ||
cfd5b8b8 NB |
3355 | /* This could be useful to other routines. If you allocate this many |
3356 | bytes, you have enough room to spell the token. */ | |
3357 | #define TOKEN_LEN(token) (4 + (token_spellings[token->type].type == \ | |
3358 | SPELL_HANDLER ? token->val.name.len: 0)) | |
3359 | ||
c5a04734 ZW |
3360 | static void |
3361 | _cpp_output_list (pfile, list) | |
3362 | cpp_reader *pfile; | |
3363 | cpp_toklist *list; | |
3364 | { | |
3365 | unsigned int comment_no = 0; | |
3366 | cpp_token *token, *comment_token = 0; | |
3367 | ||
3368 | if (list->comments_used > 0) | |
3369 | comment_token = list->tokens + list->comments[0].aux; | |
3370 | ||
3371 | CPP_RESERVE (pfile, 2); /* Always have room for " \n". */ | |
3372 | for (token = &list->tokens[0];; token++) | |
3373 | { | |
3374 | if (token->flags & PREV_WHITESPACE) | |
3375 | { | |
3376 | /* Output comments if -C. Otherwise a space will do. */ | |
3377 | if (token == comment_token) | |
3378 | { | |
3379 | cpp_token *comment = &list->comments[comment_no]; | |
3380 | do | |
3381 | { | |
cfd5b8b8 | 3382 | CPP_RESERVE (pfile, 2 + TOKEN_LEN (comment)); |
c5a04734 ZW |
3383 | pfile->limit += spell_comment (pfile->limit, list, comment); |
3384 | comment_no++, comment++; | |
3385 | if (comment_no == list->comments_used) | |
3386 | break; | |
3387 | comment_token = comment->aux + list->tokens; | |
3388 | } | |
3389 | while (comment_token == token); | |
3390 | } | |
3391 | else | |
3392 | CPP_PUTC_Q (pfile, ' '); | |
3393 | } | |
3394 | ||
cfd5b8b8 | 3395 | CPP_RESERVE (pfile, 2 + TOKEN_LEN (token)); |
c5a04734 ZW |
3396 | switch (token_spellings[token->type].type) |
3397 | { | |
3398 | case SPELL_TEXT: | |
3399 | { | |
3400 | const unsigned char *spelling; | |
3401 | unsigned char c; | |
3402 | ||
c5a04734 | 3403 | if (token->flags & DIGRAPH) |
cfd5b8b8 | 3404 | spelling = digraph_spellings[token->type - CPP_FIRST_DIGRAPH]; |
c5a04734 ZW |
3405 | else |
3406 | spelling = token_spellings[token->type].speller; | |
3407 | ||
3408 | while ((c = *spelling++) != '\0') | |
3409 | CPP_PUTC_Q (pfile, c); | |
3410 | } | |
3411 | break; | |
3412 | ||
3413 | case SPELL_HANDLER: | |
3414 | { | |
3415 | speller s; | |
3416 | ||
3417 | s = (speller) token_spellings[token->type].speller; | |
c5a04734 ZW |
3418 | pfile->limit += s (pfile->limit, list, token); |
3419 | } | |
3420 | break; | |
3421 | ||
cfd5b8b8 NB |
3422 | case SPELL_CHAR: |
3423 | *pfile->limit++ = token->aux; | |
3424 | break; | |
3425 | ||
c5a04734 ZW |
3426 | case SPELL_EOL: |
3427 | CPP_PUTC_Q (pfile, '\n'); | |
3428 | return; | |
3429 | ||
3430 | case SPELL_NONE: | |
3431 | cpp_error (pfile, "Unwriteable token"); | |
3432 | break; | |
3433 | } | |
3434 | } | |
3435 | } | |
3436 | ||
3437 | #endif |