]>
Commit | Line | Data |
---|---|---|
45b966db | 1 | /* CPP Library - lexical analysis. |
78b8811a | 2 | Copyright (C) 2000, 2001, 2002, 2003 Free Software Foundation, Inc. |
45b966db ZW |
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 | |
7 | ||
8 | This program is free software; you can redistribute it and/or modify it | |
9 | under the terms of the GNU General Public License as published by the | |
10 | Free Software Foundation; either version 2, or (at your option) any | |
11 | later version. | |
12 | ||
13 | This program is distributed in the hope that it will be useful, | |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
17 | ||
18 | You should have received a copy of the GNU General Public License | |
19 | along with this program; if not, write to the Free Software | |
20 | Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ | |
21 | ||
22 | #include "config.h" | |
23 | #include "system.h" | |
45b966db ZW |
24 | #include "cpplib.h" |
25 | #include "cpphash.h" | |
26 | ||
93c80368 | 27 | enum spell_type |
f9a0e96c | 28 | { |
93c80368 | 29 | SPELL_OPERATOR = 0, |
93c80368 | 30 | SPELL_IDENT, |
6338b358 | 31 | SPELL_LITERAL, |
93c80368 | 32 | SPELL_NONE |
f9a0e96c ZW |
33 | }; |
34 | ||
93c80368 | 35 | struct token_spelling |
f9a0e96c | 36 | { |
93c80368 NB |
37 | enum spell_type category; |
38 | const unsigned char *name; | |
f9a0e96c ZW |
39 | }; |
40 | ||
8206c799 ZW |
41 | static const unsigned char *const digraph_spellings[] = |
42 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
93c80368 NB |
43 | |
44 | #define OP(e, s) { SPELL_OPERATOR, U s }, | |
9a238586 | 45 | #define TK(e, s) { s, U #e }, |
8206c799 | 46 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
93c80368 NB |
47 | #undef OP |
48 | #undef TK | |
49 | ||
50 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
51 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
f2d5f0cc | 52 | |
6cf87ca4 ZW |
53 | static void add_line_note (cpp_buffer *, const uchar *, unsigned int); |
54 | static int skip_line_comment (cpp_reader *); | |
55 | static void skip_whitespace (cpp_reader *, cppchar_t); | |
56 | static cpp_hashnode *lex_identifier (cpp_reader *, const uchar *); | |
57 | static void lex_number (cpp_reader *, cpp_string *); | |
58 | static bool forms_identifier_p (cpp_reader *, int); | |
59 | static void lex_string (cpp_reader *, cpp_token *, const uchar *); | |
60 | static void save_comment (cpp_reader *, cpp_token *, const uchar *, cppchar_t); | |
61 | static void create_literal (cpp_reader *, cpp_token *, const uchar *, | |
62 | unsigned int, enum cpp_ttype); | |
63 | static bool warn_in_comment (cpp_reader *, _cpp_line_note *); | |
64 | static int name_p (cpp_reader *, const cpp_string *); | |
6cf87ca4 ZW |
65 | static tokenrun *next_tokenrun (tokenrun *); |
66 | ||
6cf87ca4 | 67 | static _cpp_buff *new_buff (size_t); |
15dad1d9 | 68 | |
9d10c9a9 | 69 | |
041c3194 | 70 | /* Utility routine: |
9e62c811 | 71 | |
bfb9dc7f ZW |
72 | Compares, the token TOKEN to the NUL-terminated string STRING. |
73 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
041c3194 | 74 | int |
6cf87ca4 | 75 | cpp_ideq (const cpp_token *token, const char *string) |
041c3194 | 76 | { |
bfb9dc7f | 77 | if (token->type != CPP_NAME) |
041c3194 | 78 | return 0; |
bfb9dc7f | 79 | |
562a5c27 | 80 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
15dad1d9 | 81 | } |
1368ee70 | 82 | |
26aea073 NB |
83 | /* Record a note TYPE at byte POS into the current cleaned logical |
84 | line. */ | |
87062813 | 85 | static void |
6cf87ca4 | 86 | add_line_note (cpp_buffer *buffer, const uchar *pos, unsigned int type) |
0d9f234d | 87 | { |
26aea073 NB |
88 | if (buffer->notes_used == buffer->notes_cap) |
89 | { | |
90 | buffer->notes_cap = buffer->notes_cap * 2 + 200; | |
703ad42b KG |
91 | buffer->notes = xrealloc (buffer->notes, |
92 | buffer->notes_cap * sizeof (_cpp_line_note)); | |
26aea073 | 93 | } |
0d9f234d | 94 | |
26aea073 NB |
95 | buffer->notes[buffer->notes_used].pos = pos; |
96 | buffer->notes[buffer->notes_used].type = type; | |
97 | buffer->notes_used++; | |
0d9f234d NB |
98 | } |
99 | ||
26aea073 NB |
100 | /* Returns with a logical line that contains no escaped newlines or |
101 | trigraphs. This is a time-critical inner loop. */ | |
102 | void | |
6cf87ca4 | 103 | _cpp_clean_line (cpp_reader *pfile) |
45b966db | 104 | { |
26aea073 NB |
105 | cpp_buffer *buffer; |
106 | const uchar *s; | |
107 | uchar c, *d, *p; | |
87062813 | 108 | |
26aea073 NB |
109 | buffer = pfile->buffer; |
110 | buffer->cur_note = buffer->notes_used = 0; | |
111 | buffer->cur = buffer->line_base = buffer->next_line; | |
112 | buffer->need_line = false; | |
113 | s = buffer->next_line - 1; | |
87062813 | 114 | |
26aea073 | 115 | if (!buffer->from_stage3) |
45b966db | 116 | { |
26aea073 NB |
117 | d = (uchar *) s; |
118 | ||
119 | for (;;) | |
4a5b68a2 | 120 | { |
26aea073 NB |
121 | c = *++s; |
122 | *++d = c; | |
123 | ||
124 | if (c == '\n' || c == '\r') | |
125 | { | |
126 | /* Handle DOS line endings. */ | |
127 | if (c == '\r' && s != buffer->rlimit && s[1] == '\n') | |
128 | s++; | |
129 | if (s == buffer->rlimit) | |
130 | break; | |
131 | ||
132 | /* Escaped? */ | |
133 | p = d; | |
134 | while (p != buffer->next_line && is_nvspace (p[-1])) | |
135 | p--; | |
136 | if (p == buffer->next_line || p[-1] != '\\') | |
137 | break; | |
138 | ||
41c32c98 | 139 | add_line_note (buffer, p - 1, p != d ? ' ': '\\'); |
26aea073 NB |
140 | d = p - 2; |
141 | buffer->next_line = p - 1; | |
142 | } | |
143 | else if (c == '?' && s[1] == '?' && _cpp_trigraph_map[s[2]]) | |
144 | { | |
145 | /* Add a note regardless, for the benefit of -Wtrigraphs. */ | |
41c32c98 | 146 | add_line_note (buffer, d, s[2]); |
26aea073 NB |
147 | if (CPP_OPTION (pfile, trigraphs)) |
148 | { | |
149 | *d = _cpp_trigraph_map[s[2]]; | |
150 | s += 2; | |
151 | } | |
152 | } | |
4a5b68a2 | 153 | } |
45b966db | 154 | } |
26aea073 NB |
155 | else |
156 | { | |
157 | do | |
158 | s++; | |
159 | while (*s != '\n' && *s != '\r'); | |
160 | d = (uchar *) s; | |
161 | ||
162 | /* Handle DOS line endings. */ | |
163 | if (*s == '\r' && s != buffer->rlimit && s[1] == '\n') | |
164 | s++; | |
165 | } | |
0d9f234d | 166 | |
26aea073 | 167 | *d = '\n'; |
41c32c98 NB |
168 | /* A sentinel note that should never be processed. */ |
169 | add_line_note (buffer, d + 1, '\n'); | |
26aea073 | 170 | buffer->next_line = s + 1; |
45b966db ZW |
171 | } |
172 | ||
a8eb6044 NB |
173 | /* Return true if the trigraph indicated by NOTE should be warned |
174 | about in a comment. */ | |
175 | static bool | |
6cf87ca4 | 176 | warn_in_comment (cpp_reader *pfile, _cpp_line_note *note) |
a8eb6044 NB |
177 | { |
178 | const uchar *p; | |
179 | ||
180 | /* Within comments we don't warn about trigraphs, unless the | |
181 | trigraph forms an escaped newline, as that may change | |
6356f892 | 182 | behavior. */ |
a8eb6044 NB |
183 | if (note->type != '/') |
184 | return false; | |
185 | ||
186 | /* If -trigraphs, then this was an escaped newline iff the next note | |
187 | is coincident. */ | |
188 | if (CPP_OPTION (pfile, trigraphs)) | |
189 | return note[1].pos == note->pos; | |
190 | ||
191 | /* Otherwise, see if this forms an escaped newline. */ | |
192 | p = note->pos + 3; | |
193 | while (is_nvspace (*p)) | |
194 | p++; | |
195 | ||
196 | /* There might have been escaped newlines between the trigraph and the | |
197 | newline we found. Hence the position test. */ | |
198 | return (*p == '\n' && p < note[1].pos); | |
199 | } | |
200 | ||
26aea073 NB |
201 | /* Process the notes created by add_line_note as far as the current |
202 | location. */ | |
203 | void | |
6cf87ca4 | 204 | _cpp_process_line_notes (cpp_reader *pfile, int in_comment) |
45b966db | 205 | { |
29401c30 NB |
206 | cpp_buffer *buffer = pfile->buffer; |
207 | ||
26aea073 | 208 | for (;;) |
041c3194 | 209 | { |
26aea073 NB |
210 | _cpp_line_note *note = &buffer->notes[buffer->cur_note]; |
211 | unsigned int col; | |
a5c3cccd | 212 | |
26aea073 NB |
213 | if (note->pos > buffer->cur) |
214 | break; | |
a5c3cccd | 215 | |
26aea073 NB |
216 | buffer->cur_note++; |
217 | col = CPP_BUF_COLUMN (buffer, note->pos + 1); | |
4d6baafa | 218 | |
41c32c98 | 219 | if (note->type == '\\' || note->type == ' ') |
26aea073 | 220 | { |
41c32c98 | 221 | if (note->type == ' ' && !in_comment) |
26aea073 NB |
222 | cpp_error_with_line (pfile, DL_WARNING, pfile->line, col, |
223 | "backslash and newline separated by space"); | |
41c32c98 | 224 | |
26aea073 | 225 | if (buffer->next_line > buffer->rlimit) |
87062813 | 226 | { |
26aea073 NB |
227 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line, col, |
228 | "backslash-newline at end of file"); | |
229 | /* Prevent "no newline at end of file" warning. */ | |
230 | buffer->next_line = buffer->rlimit; | |
87062813 | 231 | } |
26aea073 NB |
232 | |
233 | buffer->line_base = note->pos; | |
234 | pfile->line++; | |
0d9f234d | 235 | } |
41c32c98 NB |
236 | else if (_cpp_trigraph_map[note->type]) |
237 | { | |
a8eb6044 NB |
238 | if (CPP_OPTION (pfile, warn_trigraphs) |
239 | && (!in_comment || warn_in_comment (pfile, note))) | |
41c32c98 NB |
240 | { |
241 | if (CPP_OPTION (pfile, trigraphs)) | |
242 | cpp_error_with_line (pfile, DL_WARNING, pfile->line, col, | |
243 | "trigraph ??%c converted to %c", | |
244 | note->type, | |
245 | (int) _cpp_trigraph_map[note->type]); | |
246 | else | |
905bd7b5 GK |
247 | { |
248 | cpp_error_with_line | |
249 | (pfile, DL_WARNING, pfile->line, col, | |
250 | "trigraph ??%c ignored, use -trigraphs to enable", | |
251 | note->type); | |
252 | } | |
41c32c98 NB |
253 | } |
254 | } | |
255 | else | |
256 | abort (); | |
041c3194 | 257 | } |
45b966db ZW |
258 | } |
259 | ||
0d9f234d NB |
260 | /* Skip a C-style block comment. We find the end of the comment by |
261 | seeing if an asterisk is before every '/' we encounter. Returns | |
6f572ac2 NB |
262 | nonzero if comment terminated by EOF, zero otherwise. |
263 | ||
264 | Buffer->cur points to the initial asterisk of the comment. */ | |
26aea073 | 265 | bool |
6cf87ca4 | 266 | _cpp_skip_block_comment (cpp_reader *pfile) |
45b966db | 267 | { |
041c3194 | 268 | cpp_buffer *buffer = pfile->buffer; |
26aea073 | 269 | cppchar_t c; |
0d9f234d | 270 | |
6f572ac2 | 271 | buffer->cur++; |
26aea073 NB |
272 | if (*buffer->cur == '/') |
273 | buffer->cur++; | |
0d9f234d | 274 | |
26aea073 NB |
275 | for (;;) |
276 | { | |
277 | c = *buffer->cur++; | |
041c3194 | 278 | |
0d9f234d NB |
279 | /* People like decorating comments with '*', so check for '/' |
280 | instead for efficiency. */ | |
041c3194 | 281 | if (c == '/') |
45b966db | 282 | { |
26aea073 | 283 | if (buffer->cur[-2] == '*') |
0d9f234d | 284 | break; |
041c3194 | 285 | |
0d9f234d | 286 | /* Warn about potential nested comments, but not if the '/' |
a1f300c0 | 287 | comes immediately before the true comment delimiter. |
041c3194 | 288 | Don't bother to get it right across escaped newlines. */ |
0d9f234d | 289 | if (CPP_OPTION (pfile, warn_comments) |
87062813 | 290 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
ebef4e8c NB |
291 | cpp_error_with_line (pfile, DL_WARNING, |
292 | pfile->line, CPP_BUF_COL (buffer), | |
293 | "\"/*\" within comment"); | |
45b966db | 294 | } |
26aea073 NB |
295 | else if (c == '\n') |
296 | { | |
297 | buffer->cur--; | |
298 | _cpp_process_line_notes (pfile, true); | |
299 | if (buffer->next_line >= buffer->rlimit) | |
300 | return true; | |
301 | _cpp_clean_line (pfile); | |
302 | pfile->line++; | |
303 | } | |
45b966db | 304 | } |
041c3194 | 305 | |
a8eb6044 | 306 | _cpp_process_line_notes (pfile, true); |
26aea073 | 307 | return false; |
45b966db ZW |
308 | } |
309 | ||
480709cc | 310 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
da7d8304 | 311 | terminating newline. Handles escaped newlines. Returns nonzero |
480709cc | 312 | if a multiline comment. */ |
041c3194 | 313 | static int |
6cf87ca4 | 314 | skip_line_comment (cpp_reader *pfile) |
45b966db | 315 | { |
cbcff6df | 316 | cpp_buffer *buffer = pfile->buffer; |
67821e3a | 317 | unsigned int orig_line = pfile->line; |
041c3194 | 318 | |
26aea073 NB |
319 | while (*buffer->cur != '\n') |
320 | buffer->cur++; | |
480709cc | 321 | |
26aea073 | 322 | _cpp_process_line_notes (pfile, true); |
67821e3a | 323 | return orig_line != pfile->line; |
041c3194 | 324 | } |
45b966db | 325 | |
26aea073 | 326 | /* Skips whitespace, saving the next non-whitespace character. */ |
52fadca8 | 327 | static void |
6cf87ca4 | 328 | skip_whitespace (cpp_reader *pfile, cppchar_t c) |
041c3194 ZW |
329 | { |
330 | cpp_buffer *buffer = pfile->buffer; | |
f7d151fb | 331 | bool saw_NUL = false; |
45b966db | 332 | |
0d9f234d | 333 | do |
041c3194 | 334 | { |
91fcd158 | 335 | /* Horizontal space always OK. */ |
26aea073 | 336 | if (c == ' ' || c == '\t') |
0d9f234d | 337 | ; |
0d9f234d | 338 | /* Just \f \v or \0 left. */ |
91fcd158 | 339 | else if (c == '\0') |
f7d151fb | 340 | saw_NUL = true; |
93c80368 | 341 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
ebef4e8c NB |
342 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line, |
343 | CPP_BUF_COL (buffer), | |
344 | "%s in preprocessing directive", | |
345 | c == '\f' ? "form feed" : "vertical tab"); | |
0d9f234d | 346 | |
0d9f234d | 347 | c = *buffer->cur++; |
45b966db | 348 | } |
ec5c56db | 349 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
0d9f234d NB |
350 | while (is_nvspace (c)); |
351 | ||
f7d151fb NB |
352 | if (saw_NUL) |
353 | cpp_error (pfile, DL_WARNING, "null character(s) ignored"); | |
354 | ||
480709cc | 355 | buffer->cur--; |
041c3194 | 356 | } |
45b966db | 357 | |
93c80368 NB |
358 | /* See if the characters of a number token are valid in a name (no |
359 | '.', '+' or '-'). */ | |
360 | static int | |
6cf87ca4 | 361 | name_p (cpp_reader *pfile, const cpp_string *string) |
93c80368 NB |
362 | { |
363 | unsigned int i; | |
364 | ||
365 | for (i = 0; i < string->len; i++) | |
366 | if (!is_idchar (string->text[i])) | |
367 | return 0; | |
368 | ||
df383483 | 369 | return 1; |
93c80368 NB |
370 | } |
371 | ||
bced6edf | 372 | /* Returns TRUE if the sequence starting at buffer->cur is invalid in |
1613e52b | 373 | an identifier. FIRST is TRUE if this starts an identifier. */ |
bced6edf | 374 | static bool |
6cf87ca4 | 375 | forms_identifier_p (cpp_reader *pfile, int first) |
bced6edf | 376 | { |
1613e52b NB |
377 | cpp_buffer *buffer = pfile->buffer; |
378 | ||
379 | if (*buffer->cur == '$') | |
380 | { | |
381 | if (!CPP_OPTION (pfile, dollars_in_ident)) | |
382 | return false; | |
383 | ||
384 | buffer->cur++; | |
78b8811a | 385 | if (CPP_OPTION (pfile, warn_dollars) && !pfile->state.skipping) |
1613e52b | 386 | { |
78b8811a | 387 | CPP_OPTION (pfile, warn_dollars) = 0; |
1613e52b NB |
388 | cpp_error (pfile, DL_PEDWARN, "'$' in identifier or number"); |
389 | } | |
390 | ||
391 | return true; | |
392 | } | |
bced6edf | 393 | |
1613e52b NB |
394 | /* Is this a syntactically valid UCN? */ |
395 | if (0 && *buffer->cur == '\\' | |
396 | && (buffer->cur[1] == 'u' || buffer->cur[1] == 'U')) | |
bced6edf | 397 | { |
1613e52b | 398 | buffer->cur += 2; |
e6cc3a24 | 399 | if (_cpp_valid_ucn (pfile, &buffer->cur, buffer->rlimit, 1 + !first)) |
1613e52b NB |
400 | return true; |
401 | buffer->cur -= 2; | |
bced6edf | 402 | } |
bced6edf | 403 | |
1613e52b | 404 | return false; |
bced6edf NB |
405 | } |
406 | ||
407 | /* Lex an identifier starting at BUFFER->CUR - 1. */ | |
0d9f234d | 408 | static cpp_hashnode * |
6cf87ca4 | 409 | lex_identifier (cpp_reader *pfile, const uchar *base) |
45b966db | 410 | { |
93c80368 | 411 | cpp_hashnode *result; |
1613e52b | 412 | const uchar *cur; |
2c3fcba6 | 413 | |
bced6edf | 414 | do |
10cf9bde | 415 | { |
bced6edf NB |
416 | cur = pfile->buffer->cur; |
417 | ||
418 | /* N.B. ISIDNUM does not include $. */ | |
419 | while (ISIDNUM (*cur)) | |
420 | cur++; | |
10cf9bde | 421 | |
10cf9bde | 422 | pfile->buffer->cur = cur; |
2c3fcba6 | 423 | } |
1613e52b | 424 | while (forms_identifier_p (pfile, false)); |
bced6edf NB |
425 | |
426 | result = (cpp_hashnode *) | |
427 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
2c3fcba6 | 428 | |
bced6edf | 429 | /* Rarely, identifiers require diagnostics when lexed. */ |
2c3fcba6 ZW |
430 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) |
431 | && !pfile->state.skipping, 0)) | |
432 | { | |
433 | /* It is allowed to poison the same identifier twice. */ | |
434 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
ebef4e8c | 435 | cpp_error (pfile, DL_ERROR, "attempt to use poisoned \"%s\"", |
2c3fcba6 ZW |
436 | NODE_NAME (result)); |
437 | ||
438 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
439 | replacement list of a variadic macro. */ | |
440 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
441 | && !pfile->state.va_args_ok) | |
ebef4e8c | 442 | cpp_error (pfile, DL_PEDWARN, |
6cf87ca4 ZW |
443 | "__VA_ARGS__ can only appear in the expansion" |
444 | " of a C99 variadic macro"); | |
2c3fcba6 ZW |
445 | } |
446 | ||
447 | return result; | |
448 | } | |
449 | ||
bced6edf | 450 | /* Lex a number to NUMBER starting at BUFFER->CUR - 1. */ |
45b966db | 451 | static void |
6cf87ca4 | 452 | lex_number (cpp_reader *pfile, cpp_string *number) |
45b966db | 453 | { |
562a5c27 | 454 | const uchar *cur; |
bced6edf NB |
455 | const uchar *base; |
456 | uchar *dest; | |
45b966db | 457 | |
bced6edf NB |
458 | base = pfile->buffer->cur - 1; |
459 | do | |
041c3194 | 460 | { |
bced6edf | 461 | cur = pfile->buffer->cur; |
0d9f234d | 462 | |
bced6edf NB |
463 | /* N.B. ISIDNUM does not include $. */ |
464 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
465 | cur++; | |
45b966db | 466 | |
10cf9bde | 467 | pfile->buffer->cur = cur; |
45b966db | 468 | } |
1613e52b | 469 | while (forms_identifier_p (pfile, false)); |
93c80368 | 470 | |
bced6edf NB |
471 | number->len = cur - base; |
472 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
473 | memcpy (dest, base, number->len); | |
474 | dest[number->len] = '\0'; | |
475 | number->text = dest; | |
93c80368 NB |
476 | } |
477 | ||
6338b358 NB |
478 | /* Create a token of type TYPE with a literal spelling. */ |
479 | static void | |
6cf87ca4 ZW |
480 | create_literal (cpp_reader *pfile, cpp_token *token, const uchar *base, |
481 | unsigned int len, enum cpp_ttype type) | |
6338b358 NB |
482 | { |
483 | uchar *dest = _cpp_unaligned_alloc (pfile, len + 1); | |
484 | ||
485 | memcpy (dest, base, len); | |
486 | dest[len] = '\0'; | |
487 | token->type = type; | |
488 | token->val.str.len = len; | |
489 | token->val.str.text = dest; | |
490 | } | |
491 | ||
bced6edf | 492 | /* Lexes a string, character constant, or angle-bracketed header file |
6338b358 NB |
493 | name. The stored string contains the spelling, including opening |
494 | quote and leading any leading 'L'. It returns the type of the | |
495 | literal, or CPP_OTHER if it was not properly terminated. | |
496 | ||
497 | The spelling is NUL-terminated, but it is not guaranteed that this | |
498 | is the first NUL since embedded NULs are preserved. */ | |
041c3194 | 499 | static void |
6cf87ca4 | 500 | lex_string (cpp_reader *pfile, cpp_token *token, const uchar *base) |
45b966db | 501 | { |
6338b358 NB |
502 | bool saw_NUL = false; |
503 | const uchar *cur; | |
bced6edf | 504 | cppchar_t terminator; |
6338b358 NB |
505 | enum cpp_ttype type; |
506 | ||
507 | cur = base; | |
508 | terminator = *cur++; | |
509 | if (terminator == 'L') | |
510 | terminator = *cur++; | |
511 | if (terminator == '\"') | |
512 | type = *base == 'L' ? CPP_WSTRING: CPP_STRING; | |
513 | else if (terminator == '\'') | |
514 | type = *base == 'L' ? CPP_WCHAR: CPP_CHAR; | |
515 | else | |
516 | terminator = '>', type = CPP_HEADER_NAME; | |
93c80368 | 517 | |
0d9f234d | 518 | for (;;) |
45b966db | 519 | { |
6338b358 | 520 | cppchar_t c = *cur++; |
7868b4a2 | 521 | |
6f572ac2 | 522 | /* In #include-style directives, terminators are not escapable. */ |
6338b358 NB |
523 | if (c == '\\' && !pfile->state.angled_headers && *cur != '\n') |
524 | cur++; | |
525 | else if (c == terminator) | |
bced6edf | 526 | break; |
6338b358 | 527 | else if (c == '\n') |
0d9f234d | 528 | { |
6338b358 NB |
529 | cur--; |
530 | type = CPP_OTHER; | |
531 | break; | |
45b966db | 532 | } |
6338b358 NB |
533 | else if (c == '\0') |
534 | saw_NUL = true; | |
45b966db ZW |
535 | } |
536 | ||
6338b358 NB |
537 | if (saw_NUL && !pfile->state.skipping) |
538 | cpp_error (pfile, DL_WARNING, "null character(s) preserved in literal"); | |
45b966db | 539 | |
6338b358 NB |
540 | pfile->buffer->cur = cur; |
541 | create_literal (pfile, token, base, cur - base, type); | |
0d9f234d | 542 | } |
041c3194 | 543 | |
93c80368 | 544 | /* The stored comment includes the comment start and any terminator. */ |
9e62c811 | 545 | static void |
6cf87ca4 ZW |
546 | save_comment (cpp_reader *pfile, cpp_token *token, const unsigned char *from, |
547 | cppchar_t type) | |
9e62c811 | 548 | { |
041c3194 | 549 | unsigned char *buffer; |
477cdac7 | 550 | unsigned int len, clen; |
df383483 | 551 | |
1c6d33ef | 552 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
480709cc | 553 | |
3542203b NB |
554 | /* C++ comments probably (not definitely) have moved past a new |
555 | line, which we don't want to save in the comment. */ | |
480709cc | 556 | if (is_vspace (pfile->buffer->cur[-1])) |
3542203b | 557 | len--; |
477cdac7 JT |
558 | |
559 | /* If we are currently in a directive, then we need to store all | |
560 | C++ comments as C comments internally, and so we need to | |
561 | allocate a little extra space in that case. | |
562 | ||
563 | Note that the only time we encounter a directive here is | |
564 | when we are saving comments in a "#define". */ | |
565 | clen = (pfile->state.in_directive && type == '/') ? len + 2 : len; | |
566 | ||
567 | buffer = _cpp_unaligned_alloc (pfile, clen); | |
df383483 | 568 | |
041c3194 | 569 | token->type = CPP_COMMENT; |
477cdac7 | 570 | token->val.str.len = clen; |
0d9f234d | 571 | token->val.str.text = buffer; |
45b966db | 572 | |
1c6d33ef NB |
573 | buffer[0] = '/'; |
574 | memcpy (buffer + 1, from, len - 1); | |
477cdac7 | 575 | |
1eeeb6a4 | 576 | /* Finish conversion to a C comment, if necessary. */ |
477cdac7 JT |
577 | if (pfile->state.in_directive && type == '/') |
578 | { | |
579 | buffer[1] = '*'; | |
580 | buffer[clen - 2] = '*'; | |
581 | buffer[clen - 1] = '/'; | |
582 | } | |
0d9f234d | 583 | } |
45b966db | 584 | |
5fddcffc NB |
585 | /* Allocate COUNT tokens for RUN. */ |
586 | void | |
6cf87ca4 | 587 | _cpp_init_tokenrun (tokenrun *run, unsigned int count) |
5fddcffc NB |
588 | { |
589 | run->base = xnewvec (cpp_token, count); | |
590 | run->limit = run->base + count; | |
591 | run->next = NULL; | |
592 | } | |
593 | ||
594 | /* Returns the next tokenrun, or creates one if there is none. */ | |
595 | static tokenrun * | |
6cf87ca4 | 596 | next_tokenrun (tokenrun *run) |
5fddcffc NB |
597 | { |
598 | if (run->next == NULL) | |
599 | { | |
600 | run->next = xnew (tokenrun); | |
bdcbe496 | 601 | run->next->prev = run; |
5fddcffc NB |
602 | _cpp_init_tokenrun (run->next, 250); |
603 | } | |
604 | ||
605 | return run->next; | |
606 | } | |
607 | ||
4ed5bcfb NB |
608 | /* Allocate a single token that is invalidated at the same time as the |
609 | rest of the tokens on the line. Has its line and col set to the | |
610 | same as the last lexed token, so that diagnostics appear in the | |
611 | right place. */ | |
612 | cpp_token * | |
6cf87ca4 | 613 | _cpp_temp_token (cpp_reader *pfile) |
4ed5bcfb NB |
614 | { |
615 | cpp_token *old, *result; | |
616 | ||
617 | old = pfile->cur_token - 1; | |
618 | if (pfile->cur_token == pfile->cur_run->limit) | |
619 | { | |
620 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
621 | pfile->cur_token = pfile->cur_run->base; | |
622 | } | |
623 | ||
624 | result = pfile->cur_token++; | |
625 | result->line = old->line; | |
626 | result->col = old->col; | |
627 | return result; | |
628 | } | |
629 | ||
14baae01 NB |
630 | /* Lex a token into RESULT (external interface). Takes care of issues |
631 | like directive handling, token lookahead, multiple include | |
a1f300c0 | 632 | optimization and skipping. */ |
345894b4 | 633 | const cpp_token * |
6cf87ca4 | 634 | _cpp_lex_token (cpp_reader *pfile) |
5fddcffc | 635 | { |
bdcbe496 | 636 | cpp_token *result; |
5fddcffc | 637 | |
bdcbe496 | 638 | for (;;) |
5fddcffc | 639 | { |
bdcbe496 | 640 | if (pfile->cur_token == pfile->cur_run->limit) |
5fddcffc | 641 | { |
bdcbe496 NB |
642 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
643 | pfile->cur_token = pfile->cur_run->base; | |
5fddcffc NB |
644 | } |
645 | ||
bdcbe496 | 646 | if (pfile->lookaheads) |
14baae01 NB |
647 | { |
648 | pfile->lookaheads--; | |
649 | result = pfile->cur_token++; | |
650 | } | |
bdcbe496 | 651 | else |
14baae01 | 652 | result = _cpp_lex_direct (pfile); |
bdcbe496 NB |
653 | |
654 | if (result->flags & BOL) | |
5fddcffc | 655 | { |
bdcbe496 NB |
656 | /* Is this a directive. If _cpp_handle_directive returns |
657 | false, it is an assembler #. */ | |
658 | if (result->type == CPP_HASH | |
e808ec9c NB |
659 | /* 6.10.3 p 11: Directives in a list of macro arguments |
660 | gives undefined behavior. This implementation | |
661 | handles the directive as normal. */ | |
662 | && pfile->state.parsing_args != 1 | |
bdcbe496 NB |
663 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
664 | continue; | |
97293897 | 665 | if (pfile->cb.line_change && !pfile->state.skipping) |
6cf87ca4 | 666 | pfile->cb.line_change (pfile, result, pfile->state.parsing_args); |
5fddcffc | 667 | } |
5fddcffc | 668 | |
bdcbe496 NB |
669 | /* We don't skip tokens in directives. */ |
670 | if (pfile->state.in_directive) | |
671 | break; | |
5fddcffc | 672 | |
bdcbe496 | 673 | /* Outside a directive, invalidate controlling macros. At file |
14baae01 | 674 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
6356f892 | 675 | get here and MI optimization works. */ |
5fddcffc | 676 | pfile->mi_valid = false; |
bdcbe496 NB |
677 | |
678 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
679 | break; | |
5fddcffc NB |
680 | } |
681 | ||
345894b4 | 682 | return result; |
5fddcffc NB |
683 | } |
684 | ||
26aea073 NB |
685 | /* Returns true if a fresh line has been loaded. */ |
686 | bool | |
6cf87ca4 | 687 | _cpp_get_fresh_line (cpp_reader *pfile) |
004cb263 | 688 | { |
26aea073 NB |
689 | /* We can't get a new line until we leave the current directive. */ |
690 | if (pfile->state.in_directive) | |
691 | return false; | |
df383483 | 692 | |
26aea073 | 693 | for (;;) |
1a76916c | 694 | { |
26aea073 | 695 | cpp_buffer *buffer = pfile->buffer; |
1a76916c | 696 | |
26aea073 NB |
697 | if (!buffer->need_line) |
698 | return true; | |
699 | ||
700 | if (buffer->next_line < buffer->rlimit) | |
004cb263 | 701 | { |
26aea073 NB |
702 | _cpp_clean_line (pfile); |
703 | return true; | |
704 | } | |
004cb263 | 705 | |
26aea073 NB |
706 | /* First, get out of parsing arguments state. */ |
707 | if (pfile->state.parsing_args) | |
708 | return false; | |
709 | ||
710 | /* End of buffer. Non-empty files should end in a newline. */ | |
711 | if (buffer->buf != buffer->rlimit | |
712 | && buffer->next_line > buffer->rlimit | |
713 | && !buffer->from_stage3) | |
714 | { | |
715 | /* Only warn once. */ | |
716 | buffer->next_line = buffer->rlimit; | |
717 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line - 1, | |
718 | CPP_BUF_COLUMN (buffer, buffer->cur), | |
719 | "no newline at end of file"); | |
720 | } | |
721 | ||
722 | if (buffer->return_at_eof) | |
723 | { | |
b78f9414 | 724 | _cpp_pop_buffer (pfile); |
26aea073 | 725 | return false; |
004cb263 | 726 | } |
004cb263 | 727 | |
26aea073 NB |
728 | _cpp_pop_buffer (pfile); |
729 | } | |
004cb263 NB |
730 | } |
731 | ||
6f572ac2 NB |
732 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
733 | do \ | |
734 | { \ | |
735 | result->type = ELSE_TYPE; \ | |
736 | if (*buffer->cur == CHAR) \ | |
737 | buffer->cur++, result->type = THEN_TYPE; \ | |
738 | } \ | |
739 | while (0) | |
480709cc | 740 | |
14baae01 NB |
741 | /* Lex a token into pfile->cur_token, which is also incremented, to |
742 | get diagnostics pointing to the correct location. | |
743 | ||
744 | Does not handle issues such as token lookahead, multiple-include | |
f1ba665b | 745 | optimization, directives, skipping etc. This function is only |
14baae01 NB |
746 | suitable for use by _cpp_lex_token, and in special cases like |
747 | lex_expansion_token which doesn't care for any of these issues. | |
748 | ||
749 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
750 | otherwise returns to the start of the token buffer if permissible. | |
751 | Returns the location of the lexed token. */ | |
752 | cpp_token * | |
6cf87ca4 | 753 | _cpp_lex_direct (cpp_reader *pfile) |
45b966db | 754 | { |
0d9f234d | 755 | cppchar_t c; |
adb84b42 | 756 | cpp_buffer *buffer; |
0d9f234d | 757 | const unsigned char *comment_start; |
14baae01 | 758 | cpp_token *result = pfile->cur_token++; |
9ec7291f | 759 | |
5fddcffc | 760 | fresh_line: |
26aea073 NB |
761 | result->flags = 0; |
762 | if (pfile->buffer->need_line) | |
763 | { | |
764 | if (!_cpp_get_fresh_line (pfile)) | |
765 | { | |
766 | result->type = CPP_EOF; | |
9ff7868d NB |
767 | if (!pfile->state.in_directive) |
768 | { | |
769 | /* Tell the compiler the line number of the EOF token. */ | |
770 | result->line = pfile->line; | |
771 | result->flags = BOL; | |
772 | } | |
26aea073 NB |
773 | return result; |
774 | } | |
775 | if (!pfile->keep_tokens) | |
776 | { | |
777 | pfile->cur_run = &pfile->base_run; | |
778 | result = pfile->base_run.base; | |
779 | pfile->cur_token = result + 1; | |
780 | } | |
781 | result->flags = BOL; | |
782 | if (pfile->state.parsing_args == 2) | |
783 | result->flags |= PREV_WHITE; | |
784 | } | |
adb84b42 | 785 | buffer = pfile->buffer; |
5fddcffc | 786 | update_tokens_line: |
1444f2ed | 787 | result->line = pfile->line; |
041c3194 | 788 | |
5fddcffc | 789 | skipped_white: |
26aea073 NB |
790 | if (buffer->cur >= buffer->notes[buffer->cur_note].pos |
791 | && !pfile->overlaid_buffer) | |
792 | { | |
793 | _cpp_process_line_notes (pfile, false); | |
794 | result->line = pfile->line; | |
795 | } | |
480709cc | 796 | c = *buffer->cur++; |
5fddcffc | 797 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
5fddcffc | 798 | |
0d9f234d | 799 | switch (c) |
45b966db | 800 | { |
4d6baafa NB |
801 | case ' ': case '\t': case '\f': case '\v': case '\0': |
802 | result->flags |= PREV_WHITE; | |
26aea073 NB |
803 | skip_whitespace (pfile, c); |
804 | goto skipped_white; | |
0d9f234d | 805 | |
26aea073 NB |
806 | case '\n': |
807 | pfile->line++; | |
808 | buffer->need_line = true; | |
809 | goto fresh_line; | |
46d07497 | 810 | |
0d9f234d NB |
811 | case '0': case '1': case '2': case '3': case '4': |
812 | case '5': case '6': case '7': case '8': case '9': | |
813 | result->type = CPP_NUMBER; | |
bced6edf | 814 | lex_number (pfile, &result->val.str); |
0d9f234d | 815 | break; |
46d07497 | 816 | |
0abc6a6a NB |
817 | case 'L': |
818 | /* 'L' may introduce wide characters or strings. */ | |
bced6edf NB |
819 | if (*buffer->cur == '\'' || *buffer->cur == '"') |
820 | { | |
6338b358 | 821 | lex_string (pfile, result, buffer->cur - 1); |
bced6edf NB |
822 | break; |
823 | } | |
df383483 | 824 | /* Fall through. */ |
0abc6a6a | 825 | |
0d9f234d NB |
826 | case '_': |
827 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
828 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
829 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
830 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
831 | case 'y': case 'z': | |
832 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
0abc6a6a | 833 | case 'G': case 'H': case 'I': case 'J': case 'K': |
0d9f234d NB |
834 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
835 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
836 | case 'Y': case 'Z': | |
837 | result->type = CPP_NAME; | |
1613e52b | 838 | result->val.node = lex_identifier (pfile, buffer->cur - 1); |
0d9f234d | 839 | |
0d9f234d | 840 | /* Convert named operators to their proper types. */ |
0abc6a6a | 841 | if (result->val.node->flags & NODE_OPERATOR) |
0d9f234d NB |
842 | { |
843 | result->flags |= NAMED_OP; | |
4977bab6 | 844 | result->type = result->val.node->directive_index; |
0d9f234d NB |
845 | } |
846 | break; | |
847 | ||
848 | case '\'': | |
849 | case '"': | |
6338b358 | 850 | lex_string (pfile, result, buffer->cur - 1); |
0d9f234d | 851 | break; |
041c3194 | 852 | |
0d9f234d | 853 | case '/': |
1c6d33ef NB |
854 | /* A potential block or line comment. */ |
855 | comment_start = buffer->cur; | |
6f572ac2 NB |
856 | c = *buffer->cur; |
857 | ||
1c6d33ef NB |
858 | if (c == '*') |
859 | { | |
26aea073 | 860 | if (_cpp_skip_block_comment (pfile)) |
ebef4e8c | 861 | cpp_error (pfile, DL_ERROR, "unterminated comment"); |
0d9f234d | 862 | } |
480709cc NB |
863 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
864 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
0d9f234d | 865 | { |
bdb05a7b NB |
866 | /* Warn about comments only if pedantically GNUC89, and not |
867 | in system headers. */ | |
868 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
a94c1199 | 869 | && ! buffer->warned_cplusplus_comments) |
041c3194 | 870 | { |
ebef4e8c | 871 | cpp_error (pfile, DL_PEDWARN, |
56508306 | 872 | "C++ style comments are not allowed in ISO C90"); |
ebef4e8c NB |
873 | cpp_error (pfile, DL_PEDWARN, |
874 | "(this will be reported only once per input file)"); | |
1c6d33ef NB |
875 | buffer->warned_cplusplus_comments = 1; |
876 | } | |
0d9f234d | 877 | |
01ef6563 | 878 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
ebef4e8c | 879 | cpp_error (pfile, DL_WARNING, "multi-line comment"); |
1c6d33ef | 880 | } |
480709cc NB |
881 | else if (c == '=') |
882 | { | |
6f572ac2 | 883 | buffer->cur++; |
480709cc NB |
884 | result->type = CPP_DIV_EQ; |
885 | break; | |
886 | } | |
887 | else | |
888 | { | |
480709cc NB |
889 | result->type = CPP_DIV; |
890 | break; | |
891 | } | |
0d9f234d | 892 | |
1c6d33ef NB |
893 | if (!pfile->state.save_comments) |
894 | { | |
895 | result->flags |= PREV_WHITE; | |
5fddcffc | 896 | goto update_tokens_line; |
0d9f234d | 897 | } |
1c6d33ef NB |
898 | |
899 | /* Save the comment as a token in its own right. */ | |
477cdac7 | 900 | save_comment (pfile, result, comment_start, c); |
bdcbe496 | 901 | break; |
0d9f234d NB |
902 | |
903 | case '<': | |
904 | if (pfile->state.angled_headers) | |
905 | { | |
6338b358 | 906 | lex_string (pfile, result, buffer->cur - 1); |
480709cc | 907 | break; |
0d9f234d | 908 | } |
45b966db | 909 | |
6f572ac2 NB |
910 | result->type = CPP_LESS; |
911 | if (*buffer->cur == '=') | |
912 | buffer->cur++, result->type = CPP_LESS_EQ; | |
913 | else if (*buffer->cur == '<') | |
0d9f234d | 914 | { |
6f572ac2 NB |
915 | buffer->cur++; |
916 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); | |
0d9f234d | 917 | } |
6f572ac2 | 918 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) |
0d9f234d | 919 | { |
6f572ac2 NB |
920 | buffer->cur++; |
921 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); | |
0d9f234d | 922 | } |
6f572ac2 | 923 | else if (CPP_OPTION (pfile, digraphs)) |
480709cc | 924 | { |
6f572ac2 NB |
925 | if (*buffer->cur == ':') |
926 | { | |
927 | buffer->cur++; | |
928 | result->flags |= DIGRAPH; | |
929 | result->type = CPP_OPEN_SQUARE; | |
930 | } | |
931 | else if (*buffer->cur == '%') | |
932 | { | |
933 | buffer->cur++; | |
934 | result->flags |= DIGRAPH; | |
935 | result->type = CPP_OPEN_BRACE; | |
936 | } | |
480709cc | 937 | } |
0d9f234d NB |
938 | break; |
939 | ||
940 | case '>': | |
6f572ac2 NB |
941 | result->type = CPP_GREATER; |
942 | if (*buffer->cur == '=') | |
943 | buffer->cur++, result->type = CPP_GREATER_EQ; | |
944 | else if (*buffer->cur == '>') | |
0d9f234d | 945 | { |
6f572ac2 NB |
946 | buffer->cur++; |
947 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); | |
948 | } | |
949 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) | |
950 | { | |
951 | buffer->cur++; | |
952 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); | |
0d9f234d NB |
953 | } |
954 | break; | |
955 | ||
cbcff6df | 956 | case '%': |
6f572ac2 NB |
957 | result->type = CPP_MOD; |
958 | if (*buffer->cur == '=') | |
959 | buffer->cur++, result->type = CPP_MOD_EQ; | |
960 | else if (CPP_OPTION (pfile, digraphs)) | |
480709cc | 961 | { |
6f572ac2 | 962 | if (*buffer->cur == ':') |
480709cc | 963 | { |
6f572ac2 NB |
964 | buffer->cur++; |
965 | result->flags |= DIGRAPH; | |
966 | result->type = CPP_HASH; | |
967 | if (*buffer->cur == '%' && buffer->cur[1] == ':') | |
968 | buffer->cur += 2, result->type = CPP_PASTE; | |
969 | } | |
970 | else if (*buffer->cur == '>') | |
971 | { | |
972 | buffer->cur++; | |
973 | result->flags |= DIGRAPH; | |
974 | result->type = CPP_CLOSE_BRACE; | |
480709cc | 975 | } |
480709cc | 976 | } |
0d9f234d NB |
977 | break; |
978 | ||
cbcff6df | 979 | case '.': |
480709cc | 980 | result->type = CPP_DOT; |
6f572ac2 | 981 | if (ISDIGIT (*buffer->cur)) |
480709cc NB |
982 | { |
983 | result->type = CPP_NUMBER; | |
bced6edf | 984 | lex_number (pfile, &result->val.str); |
480709cc | 985 | } |
6f572ac2 NB |
986 | else if (*buffer->cur == '.' && buffer->cur[1] == '.') |
987 | buffer->cur += 2, result->type = CPP_ELLIPSIS; | |
988 | else if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) | |
989 | buffer->cur++, result->type = CPP_DOT_STAR; | |
0d9f234d | 990 | break; |
45b966db | 991 | |
0d9f234d | 992 | case '+': |
6f572ac2 NB |
993 | result->type = CPP_PLUS; |
994 | if (*buffer->cur == '+') | |
995 | buffer->cur++, result->type = CPP_PLUS_PLUS; | |
996 | else if (*buffer->cur == '=') | |
997 | buffer->cur++, result->type = CPP_PLUS_EQ; | |
0d9f234d | 998 | break; |
04e3ec78 | 999 | |
0d9f234d | 1000 | case '-': |
6f572ac2 NB |
1001 | result->type = CPP_MINUS; |
1002 | if (*buffer->cur == '>') | |
0d9f234d | 1003 | { |
6f572ac2 | 1004 | buffer->cur++; |
480709cc | 1005 | result->type = CPP_DEREF; |
6f572ac2 NB |
1006 | if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) |
1007 | buffer->cur++, result->type = CPP_DEREF_STAR; | |
480709cc | 1008 | } |
6f572ac2 NB |
1009 | else if (*buffer->cur == '-') |
1010 | buffer->cur++, result->type = CPP_MINUS_MINUS; | |
1011 | else if (*buffer->cur == '=') | |
1012 | buffer->cur++, result->type = CPP_MINUS_EQ; | |
0d9f234d | 1013 | break; |
45b966db | 1014 | |
0d9f234d | 1015 | case '&': |
6f572ac2 NB |
1016 | result->type = CPP_AND; |
1017 | if (*buffer->cur == '&') | |
1018 | buffer->cur++, result->type = CPP_AND_AND; | |
1019 | else if (*buffer->cur == '=') | |
1020 | buffer->cur++, result->type = CPP_AND_EQ; | |
0d9f234d | 1021 | break; |
df383483 | 1022 | |
0d9f234d | 1023 | case '|': |
6f572ac2 NB |
1024 | result->type = CPP_OR; |
1025 | if (*buffer->cur == '|') | |
1026 | buffer->cur++, result->type = CPP_OR_OR; | |
1027 | else if (*buffer->cur == '=') | |
1028 | buffer->cur++, result->type = CPP_OR_EQ; | |
0d9f234d | 1029 | break; |
45b966db | 1030 | |
0d9f234d | 1031 | case ':': |
6f572ac2 NB |
1032 | result->type = CPP_COLON; |
1033 | if (*buffer->cur == ':' && CPP_OPTION (pfile, cplusplus)) | |
1034 | buffer->cur++, result->type = CPP_SCOPE; | |
1035 | else if (*buffer->cur == '>' && CPP_OPTION (pfile, digraphs)) | |
0d9f234d | 1036 | { |
6f572ac2 | 1037 | buffer->cur++; |
0d9f234d | 1038 | result->flags |= DIGRAPH; |
480709cc NB |
1039 | result->type = CPP_CLOSE_SQUARE; |
1040 | } | |
0d9f234d | 1041 | break; |
45b966db | 1042 | |
480709cc NB |
1043 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1044 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1045 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1046 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1047 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1048 | ||
26aea073 | 1049 | case '?': result->type = CPP_QUERY; break; |
0d9f234d NB |
1050 | case '~': result->type = CPP_COMPL; break; |
1051 | case ',': result->type = CPP_COMMA; break; | |
1052 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1053 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1054 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1055 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1056 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1057 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1058 | case ';': result->type = CPP_SEMICOLON; break; | |
1059 | ||
40f03658 | 1060 | /* @ is a punctuator in Objective-C. */ |
cc937581 | 1061 | case '@': result->type = CPP_ATSIGN; break; |
0d9f234d | 1062 | |
0abc6a6a | 1063 | case '$': |
1613e52b NB |
1064 | case '\\': |
1065 | { | |
1066 | const uchar *base = --buffer->cur; | |
0abc6a6a | 1067 | |
1613e52b NB |
1068 | if (forms_identifier_p (pfile, true)) |
1069 | { | |
1070 | result->type = CPP_NAME; | |
1071 | result->val.node = lex_identifier (pfile, base); | |
1072 | break; | |
1073 | } | |
1074 | buffer->cur++; | |
1067694a | 1075 | } |
1613e52b | 1076 | |
1067694a | 1077 | default: |
6338b358 NB |
1078 | create_literal (pfile, result, buffer->cur - 1, 1, CPP_OTHER); |
1079 | break; | |
0d9f234d | 1080 | } |
bdcbe496 NB |
1081 | |
1082 | return result; | |
0d9f234d NB |
1083 | } |
1084 | ||
59325650 NB |
1085 | /* An upper bound on the number of bytes needed to spell TOKEN. |
1086 | Does not include preceding whitespace. */ | |
93c80368 | 1087 | unsigned int |
6cf87ca4 | 1088 | cpp_token_len (const cpp_token *token) |
0d9f234d | 1089 | { |
93c80368 | 1090 | unsigned int len; |
6d2c2047 | 1091 | |
93c80368 | 1092 | switch (TOKEN_SPELL (token)) |
041c3194 | 1093 | { |
59325650 | 1094 | default: len = 4; break; |
6338b358 | 1095 | case SPELL_LITERAL: len = token->val.str.len; break; |
a28c5035 | 1096 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; |
041c3194 | 1097 | } |
59325650 NB |
1098 | |
1099 | return len; | |
6d2c2047 ZW |
1100 | } |
1101 | ||
041c3194 | 1102 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
cf00a885 | 1103 | already contain the enough space to hold the token's spelling. |
6cf87ca4 ZW |
1104 | Returns a pointer to the character after the last character written. |
1105 | FIXME: Would be nice if we didn't need the PFILE argument. */ | |
93c80368 | 1106 | unsigned char * |
6cf87ca4 ZW |
1107 | cpp_spell_token (cpp_reader *pfile, const cpp_token *token, |
1108 | unsigned char *buffer) | |
041c3194 | 1109 | { |
96be6998 | 1110 | switch (TOKEN_SPELL (token)) |
041c3194 ZW |
1111 | { |
1112 | case SPELL_OPERATOR: | |
1113 | { | |
1114 | const unsigned char *spelling; | |
1115 | unsigned char c; | |
d6d5f795 | 1116 | |
041c3194 | 1117 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1118 | spelling |
1119 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
92936ecf ZW |
1120 | else if (token->flags & NAMED_OP) |
1121 | goto spell_ident; | |
041c3194 | 1122 | else |
96be6998 | 1123 | spelling = TOKEN_NAME (token); |
df383483 | 1124 | |
041c3194 ZW |
1125 | while ((c = *spelling++) != '\0') |
1126 | *buffer++ = c; | |
1127 | } | |
1128 | break; | |
d6d5f795 | 1129 | |
47ad4138 | 1130 | spell_ident: |
041c3194 | 1131 | case SPELL_IDENT: |
a28c5035 NB |
1132 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1133 | buffer += NODE_LEN (token->val.node); | |
041c3194 | 1134 | break; |
d6d5f795 | 1135 | |
6338b358 | 1136 | case SPELL_LITERAL: |
47ad4138 ZW |
1137 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1138 | buffer += token->val.str.len; | |
1139 | break; | |
1140 | ||
041c3194 | 1141 | case SPELL_NONE: |
ebef4e8c | 1142 | cpp_error (pfile, DL_ICE, "unspellable token %s", TOKEN_NAME (token)); |
041c3194 ZW |
1143 | break; |
1144 | } | |
d6d5f795 | 1145 | |
041c3194 ZW |
1146 | return buffer; |
1147 | } | |
d6d5f795 | 1148 | |
5d8ebbd8 NB |
1149 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1150 | freed when the reader is destroyed. Useful for diagnostics. */ | |
93c80368 | 1151 | unsigned char * |
6cf87ca4 | 1152 | cpp_token_as_text (cpp_reader *pfile, const cpp_token *token) |
59325650 NB |
1153 | { |
1154 | unsigned int len = cpp_token_len (token) + 1; | |
ece54d54 | 1155 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
c5a04734 | 1156 | |
93c80368 NB |
1157 | end = cpp_spell_token (pfile, token, start); |
1158 | end[0] = '\0'; | |
c5a04734 | 1159 | |
93c80368 NB |
1160 | return start; |
1161 | } | |
c5a04734 | 1162 | |
5d8ebbd8 NB |
1163 | /* Used by C front ends, which really should move to using |
1164 | cpp_token_as_text. */ | |
93c80368 | 1165 | const char * |
6cf87ca4 | 1166 | cpp_type2name (enum cpp_ttype type) |
93c80368 NB |
1167 | { |
1168 | return (const char *) token_spellings[type].name; | |
1169 | } | |
c5a04734 | 1170 | |
4ed5bcfb NB |
1171 | /* Writes the spelling of token to FP, without any preceding space. |
1172 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1173 | double-buffering. */ | |
93c80368 | 1174 | void |
6cf87ca4 | 1175 | cpp_output_token (const cpp_token *token, FILE *fp) |
93c80368 | 1176 | { |
93c80368 | 1177 | switch (TOKEN_SPELL (token)) |
c5a04734 | 1178 | { |
93c80368 NB |
1179 | case SPELL_OPERATOR: |
1180 | { | |
1181 | const unsigned char *spelling; | |
3b681e9d | 1182 | int c; |
c5a04734 | 1183 | |
93c80368 | 1184 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1185 | spelling |
1186 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
93c80368 NB |
1187 | else if (token->flags & NAMED_OP) |
1188 | goto spell_ident; | |
1189 | else | |
1190 | spelling = TOKEN_NAME (token); | |
041c3194 | 1191 | |
3b681e9d ZW |
1192 | c = *spelling; |
1193 | do | |
1194 | putc (c, fp); | |
1195 | while ((c = *++spelling) != '\0'); | |
93c80368 NB |
1196 | } |
1197 | break; | |
041c3194 | 1198 | |
93c80368 NB |
1199 | spell_ident: |
1200 | case SPELL_IDENT: | |
3b681e9d | 1201 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
93c80368 | 1202 | break; |
041c3194 | 1203 | |
6338b358 | 1204 | case SPELL_LITERAL: |
47ad4138 ZW |
1205 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
1206 | break; | |
1207 | ||
93c80368 NB |
1208 | case SPELL_NONE: |
1209 | /* An error, most probably. */ | |
1210 | break; | |
041c3194 | 1211 | } |
c5a04734 ZW |
1212 | } |
1213 | ||
93c80368 NB |
1214 | /* Compare two tokens. */ |
1215 | int | |
6cf87ca4 | 1216 | _cpp_equiv_tokens (const cpp_token *a, const cpp_token *b) |
c5a04734 | 1217 | { |
93c80368 NB |
1218 | if (a->type == b->type && a->flags == b->flags) |
1219 | switch (TOKEN_SPELL (a)) | |
1220 | { | |
1221 | default: /* Keep compiler happy. */ | |
1222 | case SPELL_OPERATOR: | |
1223 | return 1; | |
93c80368 | 1224 | case SPELL_NONE: |
56051c0a | 1225 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
93c80368 NB |
1226 | case SPELL_IDENT: |
1227 | return a->val.node == b->val.node; | |
6338b358 | 1228 | case SPELL_LITERAL: |
93c80368 NB |
1229 | return (a->val.str.len == b->val.str.len |
1230 | && !memcmp (a->val.str.text, b->val.str.text, | |
1231 | a->val.str.len)); | |
1232 | } | |
c5a04734 | 1233 | |
041c3194 ZW |
1234 | return 0; |
1235 | } | |
1236 | ||
93c80368 NB |
1237 | /* Returns nonzero if a space should be inserted to avoid an |
1238 | accidental token paste for output. For simplicity, it is | |
1239 | conservative, and occasionally advises a space where one is not | |
1240 | needed, e.g. "." and ".2". */ | |
93c80368 | 1241 | int |
6cf87ca4 ZW |
1242 | cpp_avoid_paste (cpp_reader *pfile, const cpp_token *token1, |
1243 | const cpp_token *token2) | |
c5a04734 | 1244 | { |
93c80368 NB |
1245 | enum cpp_ttype a = token1->type, b = token2->type; |
1246 | cppchar_t c; | |
c5a04734 | 1247 | |
93c80368 NB |
1248 | if (token1->flags & NAMED_OP) |
1249 | a = CPP_NAME; | |
1250 | if (token2->flags & NAMED_OP) | |
1251 | b = CPP_NAME; | |
c5a04734 | 1252 | |
93c80368 NB |
1253 | c = EOF; |
1254 | if (token2->flags & DIGRAPH) | |
37b8524c | 1255 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
93c80368 NB |
1256 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1257 | c = token_spellings[b].name[0]; | |
c5a04734 | 1258 | |
93c80368 | 1259 | /* Quickly get everything that can paste with an '='. */ |
37b8524c | 1260 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
93c80368 | 1261 | return 1; |
c5a04734 | 1262 | |
93c80368 | 1263 | switch (a) |
c5a04734 | 1264 | { |
93c80368 NB |
1265 | case CPP_GREATER: return c == '>' || c == '?'; |
1266 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1267 | case CPP_PLUS: return c == '+'; | |
1268 | case CPP_MINUS: return c == '-' || c == '>'; | |
1269 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1270 | case CPP_MOD: return c == ':' || c == '>'; | |
1271 | case CPP_AND: return c == '&'; | |
1272 | case CPP_OR: return c == '|'; | |
1273 | case CPP_COLON: return c == ':' || c == '>'; | |
1274 | case CPP_DEREF: return c == '*'; | |
26ec42ee | 1275 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
93c80368 NB |
1276 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1277 | case CPP_NAME: return ((b == CPP_NUMBER | |
1278 | && name_p (pfile, &token2->val.str)) | |
1279 | || b == CPP_NAME | |
1280 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1281 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1282 | || c == '.' || c == '+' || c == '-'); | |
1613e52b | 1283 | /* UCNs */ |
1067694a NB |
1284 | case CPP_OTHER: return ((token1->val.str.text[0] == '\\' |
1285 | && b == CPP_NAME) | |
1613e52b | 1286 | || (CPP_OPTION (pfile, objc) |
1067694a | 1287 | && token1->val.str.text[0] == '@' |
1613e52b | 1288 | && (b == CPP_NAME || b == CPP_STRING))); |
93c80368 | 1289 | default: break; |
c5a04734 | 1290 | } |
c5a04734 | 1291 | |
417f3e3a | 1292 | return 0; |
c5a04734 ZW |
1293 | } |
1294 | ||
93c80368 | 1295 | /* Output all the remaining tokens on the current line, and a newline |
4ed5bcfb NB |
1296 | character, to FP. Leading whitespace is removed. If there are |
1297 | macros, special token padding is not performed. */ | |
c5a04734 | 1298 | void |
6cf87ca4 | 1299 | cpp_output_line (cpp_reader *pfile, FILE *fp) |
c5a04734 | 1300 | { |
4ed5bcfb | 1301 | const cpp_token *token; |
96be6998 | 1302 | |
4ed5bcfb NB |
1303 | token = cpp_get_token (pfile); |
1304 | while (token->type != CPP_EOF) | |
96be6998 | 1305 | { |
4ed5bcfb NB |
1306 | cpp_output_token (token, fp); |
1307 | token = cpp_get_token (pfile); | |
1308 | if (token->flags & PREV_WHITE) | |
1309 | putc (' ', fp); | |
96be6998 ZW |
1310 | } |
1311 | ||
93c80368 | 1312 | putc ('\n', fp); |
041c3194 | 1313 | } |
c5a04734 | 1314 | |
1e013d2e NB |
1315 | /* Memory buffers. Changing these three constants can have a dramatic |
1316 | effect on performance. The values here are reasonable defaults, | |
1317 | but might be tuned. If you adjust them, be sure to test across a | |
1318 | range of uses of cpplib, including heavy nested function-like macro | |
1319 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1320 | good tool for this). */ | |
1321 | #define MIN_BUFF_SIZE 8000 | |
87062813 | 1322 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
1e013d2e NB |
1323 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1324 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
417f3e3a | 1325 | |
87062813 NB |
1326 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1327 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1328 | #endif | |
1329 | ||
c9e7a609 NB |
1330 | /* Create a new allocation buffer. Place the control block at the end |
1331 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
b8af0ca5 | 1332 | static _cpp_buff * |
6cf87ca4 | 1333 | new_buff (size_t len) |
b8af0ca5 NB |
1334 | { |
1335 | _cpp_buff *result; | |
ece54d54 | 1336 | unsigned char *base; |
b8af0ca5 | 1337 | |
1e013d2e NB |
1338 | if (len < MIN_BUFF_SIZE) |
1339 | len = MIN_BUFF_SIZE; | |
c70f6ed3 | 1340 | len = CPP_ALIGN (len); |
b8af0ca5 NB |
1341 | |
1342 | base = xmalloc (len + sizeof (_cpp_buff)); | |
1343 | result = (_cpp_buff *) (base + len); | |
1344 | result->base = base; | |
1345 | result->cur = base; | |
1346 | result->limit = base + len; | |
1347 | result->next = NULL; | |
1348 | return result; | |
1349 | } | |
1350 | ||
1351 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1352 | void | |
6cf87ca4 | 1353 | _cpp_release_buff (cpp_reader *pfile, _cpp_buff *buff) |
b8af0ca5 NB |
1354 | { |
1355 | _cpp_buff *end = buff; | |
1356 | ||
1357 | while (end->next) | |
1358 | end = end->next; | |
1359 | end->next = pfile->free_buffs; | |
1360 | pfile->free_buffs = buff; | |
1361 | } | |
1362 | ||
1363 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1364 | _cpp_buff * | |
6cf87ca4 | 1365 | _cpp_get_buff (cpp_reader *pfile, size_t min_size) |
b8af0ca5 NB |
1366 | { |
1367 | _cpp_buff *result, **p; | |
1368 | ||
1369 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1370 | { | |
6142088c | 1371 | size_t size; |
1e013d2e NB |
1372 | |
1373 | if (*p == NULL) | |
b8af0ca5 | 1374 | return new_buff (min_size); |
1e013d2e NB |
1375 | result = *p; |
1376 | size = result->limit - result->base; | |
1377 | /* Return a buffer that's big enough, but don't waste one that's | |
1378 | way too big. */ | |
34f5271d | 1379 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
b8af0ca5 NB |
1380 | break; |
1381 | } | |
1382 | ||
1383 | *p = result->next; | |
1384 | result->next = NULL; | |
1385 | result->cur = result->base; | |
1386 | return result; | |
1387 | } | |
1388 | ||
4fe9b91c | 1389 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
1390 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
1391 | the excess bytes to the new buffer. Chains the new buffer after | |
1392 | BUFF, and returns the new buffer. */ | |
b8af0ca5 | 1393 | _cpp_buff * |
6cf87ca4 | 1394 | _cpp_append_extend_buff (cpp_reader *pfile, _cpp_buff *buff, size_t min_extra) |
b8af0ca5 | 1395 | { |
6142088c | 1396 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
8c3b2693 | 1397 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
b8af0ca5 | 1398 | |
8c3b2693 NB |
1399 | buff->next = new_buff; |
1400 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
1401 | return new_buff; | |
1402 | } | |
1403 | ||
4fe9b91c | 1404 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
1405 | remaining bytes of the buffer pointed to by BUFF, and at least |
1406 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
1407 | Chains the new buffer before the buffer pointed to by BUFF, and | |
1408 | updates the pointer to point to the new buffer. */ | |
1409 | void | |
6cf87ca4 | 1410 | _cpp_extend_buff (cpp_reader *pfile, _cpp_buff **pbuff, size_t min_extra) |
8c3b2693 NB |
1411 | { |
1412 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
1413 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
1414 | ||
1415 | new_buff = _cpp_get_buff (pfile, size); | |
1416 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
1417 | new_buff->next = old_buff; | |
1418 | *pbuff = new_buff; | |
b8af0ca5 NB |
1419 | } |
1420 | ||
1421 | /* Free a chain of buffers starting at BUFF. */ | |
1422 | void | |
5671bf27 | 1423 | _cpp_free_buff (_cpp_buff *buff) |
b8af0ca5 NB |
1424 | { |
1425 | _cpp_buff *next; | |
1426 | ||
1427 | for (; buff; buff = next) | |
1428 | { | |
1429 | next = buff->next; | |
1430 | free (buff->base); | |
1431 | } | |
1432 | } | |
417f3e3a | 1433 | |
ece54d54 NB |
1434 | /* Allocate permanent, unaligned storage of length LEN. */ |
1435 | unsigned char * | |
6cf87ca4 | 1436 | _cpp_unaligned_alloc (cpp_reader *pfile, size_t len) |
ece54d54 NB |
1437 | { |
1438 | _cpp_buff *buff = pfile->u_buff; | |
1439 | unsigned char *result = buff->cur; | |
1440 | ||
1441 | if (len > (size_t) (buff->limit - result)) | |
1442 | { | |
1443 | buff = _cpp_get_buff (pfile, len); | |
1444 | buff->next = pfile->u_buff; | |
1445 | pfile->u_buff = buff; | |
1446 | result = buff->cur; | |
1447 | } | |
1448 | ||
1449 | buff->cur = result + len; | |
1450 | return result; | |
1451 | } | |
1452 | ||
87062813 NB |
1453 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
1454 | That buffer is used for growing allocations when saving macro | |
1455 | replacement lists in a #define, and when parsing an answer to an | |
1456 | assertion in #assert, #unassert or #if (and therefore possibly | |
1457 | whilst expanding macros). It therefore must not be used by any | |
1458 | code that they might call: specifically the lexer and the guts of | |
1459 | the macro expander. | |
1460 | ||
1461 | All existing other uses clearly fit this restriction: storing | |
1462 | registered pragmas during initialization. */ | |
93c80368 | 1463 | unsigned char * |
6cf87ca4 | 1464 | _cpp_aligned_alloc (cpp_reader *pfile, size_t len) |
3fef5b2b | 1465 | { |
8c3b2693 NB |
1466 | _cpp_buff *buff = pfile->a_buff; |
1467 | unsigned char *result = buff->cur; | |
3fef5b2b | 1468 | |
8c3b2693 | 1469 | if (len > (size_t) (buff->limit - result)) |
3fef5b2b | 1470 | { |
8c3b2693 NB |
1471 | buff = _cpp_get_buff (pfile, len); |
1472 | buff->next = pfile->a_buff; | |
1473 | pfile->a_buff = buff; | |
1474 | result = buff->cur; | |
3fef5b2b | 1475 | } |
041c3194 | 1476 | |
8c3b2693 | 1477 | buff->cur = result + len; |
93c80368 | 1478 | return result; |
041c3194 | 1479 | } |