]>
Commit | Line | Data |
---|---|---|
0578f103 | 1 | /* CPP Library - lexical analysis. |
f0c2775b | 2 | Copyright (C) 2000, 2001, 2002, 2003 Free Software Foundation, Inc. |
0578f103 | 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" | |
805e22b2 | 24 | #include "coretypes.h" |
25 | #include "tm.h" | |
0578f103 | 26 | #include "cpplib.h" |
27 | #include "cpphash.h" | |
28 | ||
79bd622b | 29 | enum spell_type |
241e762e | 30 | { |
79bd622b | 31 | SPELL_OPERATOR = 0, |
79bd622b | 32 | SPELL_IDENT, |
4970d4c2 | 33 | SPELL_LITERAL, |
79bd622b | 34 | SPELL_NONE |
241e762e | 35 | }; |
36 | ||
79bd622b | 37 | struct token_spelling |
241e762e | 38 | { |
79bd622b | 39 | enum spell_type category; |
40 | const unsigned char *name; | |
241e762e | 41 | }; |
42 | ||
0ca849f9 | 43 | static const unsigned char *const digraph_spellings[] = |
44 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
79bd622b | 45 | |
46 | #define OP(e, s) { SPELL_OPERATOR, U s }, | |
18e43155 | 47 | #define TK(e, s) { s, U #e }, |
0ca849f9 | 48 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
79bd622b | 49 | #undef OP |
50 | #undef TK | |
51 | ||
52 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
53 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
e2f9a79f | 54 | |
f7fdd7a1 | 55 | static void add_line_note (cpp_buffer *, const uchar *, unsigned int); |
56 | static int skip_line_comment (cpp_reader *); | |
57 | static void skip_whitespace (cpp_reader *, cppchar_t); | |
58 | static cpp_hashnode *lex_identifier (cpp_reader *, const uchar *); | |
59 | static void lex_number (cpp_reader *, cpp_string *); | |
60 | static bool forms_identifier_p (cpp_reader *, int); | |
61 | static void lex_string (cpp_reader *, cpp_token *, const uchar *); | |
62 | static void save_comment (cpp_reader *, cpp_token *, const uchar *, cppchar_t); | |
63 | static void create_literal (cpp_reader *, cpp_token *, const uchar *, | |
64 | unsigned int, enum cpp_ttype); | |
65 | static bool warn_in_comment (cpp_reader *, _cpp_line_note *); | |
66 | static int name_p (cpp_reader *, const cpp_string *); | |
f7fdd7a1 | 67 | static tokenrun *next_tokenrun (tokenrun *); |
68 | ||
f7fdd7a1 | 69 | static _cpp_buff *new_buff (size_t); |
bce8e0c0 | 70 | |
e920deaf | 71 | |
f80e83a9 | 72 | /* Utility routine: |
2c63d6c8 | 73 | |
76faa4c0 | 74 | Compares, the token TOKEN to the NUL-terminated string STRING. |
75 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
f80e83a9 | 76 | int |
f7fdd7a1 | 77 | cpp_ideq (const cpp_token *token, const char *string) |
f80e83a9 | 78 | { |
76faa4c0 | 79 | if (token->type != CPP_NAME) |
f80e83a9 | 80 | return 0; |
76faa4c0 | 81 | |
b6d18b0a | 82 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
bce8e0c0 | 83 | } |
50fd6b48 | 84 | |
a54e0bf8 | 85 | /* Record a note TYPE at byte POS into the current cleaned logical |
86 | line. */ | |
1e0ef2fd | 87 | static void |
f7fdd7a1 | 88 | add_line_note (cpp_buffer *buffer, const uchar *pos, unsigned int type) |
338fa5f7 | 89 | { |
a54e0bf8 | 90 | if (buffer->notes_used == buffer->notes_cap) |
91 | { | |
92 | buffer->notes_cap = buffer->notes_cap * 2 + 200; | |
93 | buffer->notes = (_cpp_line_note *) | |
94 | xrealloc (buffer->notes, buffer->notes_cap * sizeof (_cpp_line_note)); | |
95 | } | |
338fa5f7 | 96 | |
a54e0bf8 | 97 | buffer->notes[buffer->notes_used].pos = pos; |
98 | buffer->notes[buffer->notes_used].type = type; | |
99 | buffer->notes_used++; | |
338fa5f7 | 100 | } |
101 | ||
a54e0bf8 | 102 | /* Returns with a logical line that contains no escaped newlines or |
103 | trigraphs. This is a time-critical inner loop. */ | |
104 | void | |
f7fdd7a1 | 105 | _cpp_clean_line (cpp_reader *pfile) |
0578f103 | 106 | { |
a54e0bf8 | 107 | cpp_buffer *buffer; |
108 | const uchar *s; | |
109 | uchar c, *d, *p; | |
1e0ef2fd | 110 | |
a54e0bf8 | 111 | buffer = pfile->buffer; |
112 | buffer->cur_note = buffer->notes_used = 0; | |
113 | buffer->cur = buffer->line_base = buffer->next_line; | |
114 | buffer->need_line = false; | |
115 | s = buffer->next_line - 1; | |
1e0ef2fd | 116 | |
a54e0bf8 | 117 | if (!buffer->from_stage3) |
0578f103 | 118 | { |
a54e0bf8 | 119 | d = (uchar *) s; |
120 | ||
121 | for (;;) | |
4b912310 | 122 | { |
a54e0bf8 | 123 | c = *++s; |
124 | *++d = c; | |
125 | ||
126 | if (c == '\n' || c == '\r') | |
127 | { | |
128 | /* Handle DOS line endings. */ | |
129 | if (c == '\r' && s != buffer->rlimit && s[1] == '\n') | |
130 | s++; | |
131 | if (s == buffer->rlimit) | |
132 | break; | |
133 | ||
134 | /* Escaped? */ | |
135 | p = d; | |
136 | while (p != buffer->next_line && is_nvspace (p[-1])) | |
137 | p--; | |
138 | if (p == buffer->next_line || p[-1] != '\\') | |
139 | break; | |
140 | ||
aad4a87f | 141 | add_line_note (buffer, p - 1, p != d ? ' ': '\\'); |
a54e0bf8 | 142 | d = p - 2; |
143 | buffer->next_line = p - 1; | |
144 | } | |
145 | else if (c == '?' && s[1] == '?' && _cpp_trigraph_map[s[2]]) | |
146 | { | |
147 | /* Add a note regardless, for the benefit of -Wtrigraphs. */ | |
aad4a87f | 148 | add_line_note (buffer, d, s[2]); |
a54e0bf8 | 149 | if (CPP_OPTION (pfile, trigraphs)) |
150 | { | |
151 | *d = _cpp_trigraph_map[s[2]]; | |
152 | s += 2; | |
153 | } | |
154 | } | |
4b912310 | 155 | } |
0578f103 | 156 | } |
a54e0bf8 | 157 | else |
158 | { | |
159 | do | |
160 | s++; | |
161 | while (*s != '\n' && *s != '\r'); | |
162 | d = (uchar *) s; | |
163 | ||
164 | /* Handle DOS line endings. */ | |
165 | if (*s == '\r' && s != buffer->rlimit && s[1] == '\n') | |
166 | s++; | |
167 | } | |
338fa5f7 | 168 | |
a54e0bf8 | 169 | *d = '\n'; |
aad4a87f | 170 | /* A sentinel note that should never be processed. */ |
171 | add_line_note (buffer, d + 1, '\n'); | |
a54e0bf8 | 172 | buffer->next_line = s + 1; |
0578f103 | 173 | } |
174 | ||
3078f2b2 | 175 | /* Return true if the trigraph indicated by NOTE should be warned |
176 | about in a comment. */ | |
177 | static bool | |
f7fdd7a1 | 178 | warn_in_comment (cpp_reader *pfile, _cpp_line_note *note) |
3078f2b2 | 179 | { |
180 | const uchar *p; | |
181 | ||
182 | /* Within comments we don't warn about trigraphs, unless the | |
183 | trigraph forms an escaped newline, as that may change | |
7ef5b942 | 184 | behavior. */ |
3078f2b2 | 185 | if (note->type != '/') |
186 | return false; | |
187 | ||
188 | /* If -trigraphs, then this was an escaped newline iff the next note | |
189 | is coincident. */ | |
190 | if (CPP_OPTION (pfile, trigraphs)) | |
191 | return note[1].pos == note->pos; | |
192 | ||
193 | /* Otherwise, see if this forms an escaped newline. */ | |
194 | p = note->pos + 3; | |
195 | while (is_nvspace (*p)) | |
196 | p++; | |
197 | ||
198 | /* There might have been escaped newlines between the trigraph and the | |
199 | newline we found. Hence the position test. */ | |
200 | return (*p == '\n' && p < note[1].pos); | |
201 | } | |
202 | ||
a54e0bf8 | 203 | /* Process the notes created by add_line_note as far as the current |
204 | location. */ | |
205 | void | |
f7fdd7a1 | 206 | _cpp_process_line_notes (cpp_reader *pfile, int in_comment) |
0578f103 | 207 | { |
c808d026 | 208 | cpp_buffer *buffer = pfile->buffer; |
209 | ||
a54e0bf8 | 210 | for (;;) |
f80e83a9 | 211 | { |
a54e0bf8 | 212 | _cpp_line_note *note = &buffer->notes[buffer->cur_note]; |
213 | unsigned int col; | |
396ffa86 | 214 | |
a54e0bf8 | 215 | if (note->pos > buffer->cur) |
216 | break; | |
396ffa86 | 217 | |
a54e0bf8 | 218 | buffer->cur_note++; |
219 | col = CPP_BUF_COLUMN (buffer, note->pos + 1); | |
435fb09b | 220 | |
aad4a87f | 221 | if (note->type == '\\' || note->type == ' ') |
a54e0bf8 | 222 | { |
aad4a87f | 223 | if (note->type == ' ' && !in_comment) |
a54e0bf8 | 224 | cpp_error_with_line (pfile, DL_WARNING, pfile->line, col, |
225 | "backslash and newline separated by space"); | |
aad4a87f | 226 | |
a54e0bf8 | 227 | if (buffer->next_line > buffer->rlimit) |
1e0ef2fd | 228 | { |
a54e0bf8 | 229 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line, col, |
230 | "backslash-newline at end of file"); | |
231 | /* Prevent "no newline at end of file" warning. */ | |
232 | buffer->next_line = buffer->rlimit; | |
1e0ef2fd | 233 | } |
a54e0bf8 | 234 | |
235 | buffer->line_base = note->pos; | |
236 | pfile->line++; | |
338fa5f7 | 237 | } |
aad4a87f | 238 | else if (_cpp_trigraph_map[note->type]) |
239 | { | |
3078f2b2 | 240 | if (CPP_OPTION (pfile, warn_trigraphs) |
241 | && (!in_comment || warn_in_comment (pfile, note))) | |
aad4a87f | 242 | { |
243 | if (CPP_OPTION (pfile, trigraphs)) | |
244 | cpp_error_with_line (pfile, DL_WARNING, pfile->line, col, | |
245 | "trigraph ??%c converted to %c", | |
246 | note->type, | |
247 | (int) _cpp_trigraph_map[note->type]); | |
248 | else | |
249 | cpp_error_with_line (pfile, DL_WARNING, pfile->line, col, | |
250 | "trigraph ??%c ignored", | |
251 | note->type); | |
252 | } | |
253 | } | |
254 | else | |
255 | abort (); | |
f80e83a9 | 256 | } |
0578f103 | 257 | } |
258 | ||
338fa5f7 | 259 | /* Skip a C-style block comment. We find the end of the comment by |
260 | seeing if an asterisk is before every '/' we encounter. Returns | |
edaf8cb5 | 261 | nonzero if comment terminated by EOF, zero otherwise. |
262 | ||
263 | Buffer->cur points to the initial asterisk of the comment. */ | |
a54e0bf8 | 264 | bool |
f7fdd7a1 | 265 | _cpp_skip_block_comment (cpp_reader *pfile) |
0578f103 | 266 | { |
f80e83a9 | 267 | cpp_buffer *buffer = pfile->buffer; |
a54e0bf8 | 268 | cppchar_t c; |
338fa5f7 | 269 | |
edaf8cb5 | 270 | buffer->cur++; |
a54e0bf8 | 271 | if (*buffer->cur == '/') |
272 | buffer->cur++; | |
338fa5f7 | 273 | |
a54e0bf8 | 274 | for (;;) |
275 | { | |
276 | c = *buffer->cur++; | |
f80e83a9 | 277 | |
338fa5f7 | 278 | /* People like decorating comments with '*', so check for '/' |
279 | instead for efficiency. */ | |
f80e83a9 | 280 | if (c == '/') |
0578f103 | 281 | { |
a54e0bf8 | 282 | if (buffer->cur[-2] == '*') |
338fa5f7 | 283 | break; |
f80e83a9 | 284 | |
338fa5f7 | 285 | /* Warn about potential nested comments, but not if the '/' |
3fb1e43b | 286 | comes immediately before the true comment delimiter. |
f80e83a9 | 287 | Don't bother to get it right across escaped newlines. */ |
338fa5f7 | 288 | if (CPP_OPTION (pfile, warn_comments) |
1e0ef2fd | 289 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
73328dce | 290 | cpp_error_with_line (pfile, DL_WARNING, |
291 | pfile->line, CPP_BUF_COL (buffer), | |
292 | "\"/*\" within comment"); | |
0578f103 | 293 | } |
a54e0bf8 | 294 | else if (c == '\n') |
295 | { | |
296 | buffer->cur--; | |
297 | _cpp_process_line_notes (pfile, true); | |
298 | if (buffer->next_line >= buffer->rlimit) | |
299 | return true; | |
300 | _cpp_clean_line (pfile); | |
301 | pfile->line++; | |
302 | } | |
0578f103 | 303 | } |
f80e83a9 | 304 | |
3078f2b2 | 305 | _cpp_process_line_notes (pfile, true); |
a54e0bf8 | 306 | return false; |
0578f103 | 307 | } |
308 | ||
1c124f85 | 309 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
d10cfa8d | 310 | terminating newline. Handles escaped newlines. Returns nonzero |
1c124f85 | 311 | if a multiline comment. */ |
f80e83a9 | 312 | static int |
f7fdd7a1 | 313 | skip_line_comment (cpp_reader *pfile) |
0578f103 | 314 | { |
f669338a | 315 | cpp_buffer *buffer = pfile->buffer; |
1ea7ed21 | 316 | unsigned int orig_line = pfile->line; |
f80e83a9 | 317 | |
a54e0bf8 | 318 | while (*buffer->cur != '\n') |
319 | buffer->cur++; | |
1c124f85 | 320 | |
a54e0bf8 | 321 | _cpp_process_line_notes (pfile, true); |
1ea7ed21 | 322 | return orig_line != pfile->line; |
f80e83a9 | 323 | } |
0578f103 | 324 | |
a54e0bf8 | 325 | /* Skips whitespace, saving the next non-whitespace character. */ |
b86584f6 | 326 | static void |
f7fdd7a1 | 327 | skip_whitespace (cpp_reader *pfile, cppchar_t c) |
f80e83a9 | 328 | { |
329 | cpp_buffer *buffer = pfile->buffer; | |
fe9eb18b | 330 | bool saw_NUL = false; |
0578f103 | 331 | |
338fa5f7 | 332 | do |
f80e83a9 | 333 | { |
78719282 | 334 | /* Horizontal space always OK. */ |
a54e0bf8 | 335 | if (c == ' ' || c == '\t') |
338fa5f7 | 336 | ; |
338fa5f7 | 337 | /* Just \f \v or \0 left. */ |
78719282 | 338 | else if (c == '\0') |
fe9eb18b | 339 | saw_NUL = true; |
79bd622b | 340 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
73328dce | 341 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line, |
342 | CPP_BUF_COL (buffer), | |
343 | "%s in preprocessing directive", | |
344 | c == '\f' ? "form feed" : "vertical tab"); | |
338fa5f7 | 345 | |
338fa5f7 | 346 | c = *buffer->cur++; |
0578f103 | 347 | } |
2c0e001b | 348 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
338fa5f7 | 349 | while (is_nvspace (c)); |
350 | ||
fe9eb18b | 351 | if (saw_NUL) |
352 | cpp_error (pfile, DL_WARNING, "null character(s) ignored"); | |
353 | ||
1c124f85 | 354 | buffer->cur--; |
f80e83a9 | 355 | } |
0578f103 | 356 | |
79bd622b | 357 | /* See if the characters of a number token are valid in a name (no |
358 | '.', '+' or '-'). */ | |
359 | static int | |
f7fdd7a1 | 360 | name_p (cpp_reader *pfile, const cpp_string *string) |
79bd622b | 361 | { |
362 | unsigned int i; | |
363 | ||
364 | for (i = 0; i < string->len; i++) | |
365 | if (!is_idchar (string->text[i])) | |
366 | return 0; | |
367 | ||
b1a9ff83 | 368 | return 1; |
79bd622b | 369 | } |
370 | ||
5bb46c08 | 371 | /* Returns TRUE if the sequence starting at buffer->cur is invalid in |
2cbf1359 | 372 | an identifier. FIRST is TRUE if this starts an identifier. */ |
5bb46c08 | 373 | static bool |
f7fdd7a1 | 374 | forms_identifier_p (cpp_reader *pfile, int first) |
5bb46c08 | 375 | { |
2cbf1359 | 376 | cpp_buffer *buffer = pfile->buffer; |
377 | ||
378 | if (*buffer->cur == '$') | |
379 | { | |
380 | if (!CPP_OPTION (pfile, dollars_in_ident)) | |
381 | return false; | |
382 | ||
383 | buffer->cur++; | |
f0c2775b | 384 | if (CPP_OPTION (pfile, warn_dollars) && !pfile->state.skipping) |
2cbf1359 | 385 | { |
f0c2775b | 386 | CPP_OPTION (pfile, warn_dollars) = 0; |
2cbf1359 | 387 | cpp_error (pfile, DL_PEDWARN, "'$' in identifier or number"); |
388 | } | |
389 | ||
390 | return true; | |
391 | } | |
5bb46c08 | 392 | |
2cbf1359 | 393 | /* Is this a syntactically valid UCN? */ |
394 | if (0 && *buffer->cur == '\\' | |
395 | && (buffer->cur[1] == 'u' || buffer->cur[1] == 'U')) | |
5bb46c08 | 396 | { |
2cbf1359 | 397 | buffer->cur += 2; |
ebc03810 | 398 | if (_cpp_valid_ucn (pfile, &buffer->cur, buffer->rlimit, 1 + !first)) |
2cbf1359 | 399 | return true; |
400 | buffer->cur -= 2; | |
5bb46c08 | 401 | } |
5bb46c08 | 402 | |
2cbf1359 | 403 | return false; |
5bb46c08 | 404 | } |
405 | ||
406 | /* Lex an identifier starting at BUFFER->CUR - 1. */ | |
338fa5f7 | 407 | static cpp_hashnode * |
f7fdd7a1 | 408 | lex_identifier (cpp_reader *pfile, const uchar *base) |
0578f103 | 409 | { |
79bd622b | 410 | cpp_hashnode *result; |
2cbf1359 | 411 | const uchar *cur; |
66a5287e | 412 | |
5bb46c08 | 413 | do |
78a11351 | 414 | { |
5bb46c08 | 415 | cur = pfile->buffer->cur; |
416 | ||
417 | /* N.B. ISIDNUM does not include $. */ | |
418 | while (ISIDNUM (*cur)) | |
419 | cur++; | |
78a11351 | 420 | |
78a11351 | 421 | pfile->buffer->cur = cur; |
66a5287e | 422 | } |
2cbf1359 | 423 | while (forms_identifier_p (pfile, false)); |
5bb46c08 | 424 | |
425 | result = (cpp_hashnode *) | |
426 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
66a5287e | 427 | |
5bb46c08 | 428 | /* Rarely, identifiers require diagnostics when lexed. */ |
66a5287e | 429 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) |
430 | && !pfile->state.skipping, 0)) | |
431 | { | |
432 | /* It is allowed to poison the same identifier twice. */ | |
433 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
73328dce | 434 | cpp_error (pfile, DL_ERROR, "attempt to use poisoned \"%s\"", |
66a5287e | 435 | NODE_NAME (result)); |
436 | ||
437 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
438 | replacement list of a variadic macro. */ | |
439 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
440 | && !pfile->state.va_args_ok) | |
73328dce | 441 | cpp_error (pfile, DL_PEDWARN, |
f7fdd7a1 | 442 | "__VA_ARGS__ can only appear in the expansion" |
443 | " of a C99 variadic macro"); | |
66a5287e | 444 | } |
445 | ||
446 | return result; | |
447 | } | |
448 | ||
5bb46c08 | 449 | /* Lex a number to NUMBER starting at BUFFER->CUR - 1. */ |
0578f103 | 450 | static void |
f7fdd7a1 | 451 | lex_number (cpp_reader *pfile, cpp_string *number) |
0578f103 | 452 | { |
b6d18b0a | 453 | const uchar *cur; |
5bb46c08 | 454 | const uchar *base; |
455 | uchar *dest; | |
0578f103 | 456 | |
5bb46c08 | 457 | base = pfile->buffer->cur - 1; |
458 | do | |
f80e83a9 | 459 | { |
5bb46c08 | 460 | cur = pfile->buffer->cur; |
338fa5f7 | 461 | |
5bb46c08 | 462 | /* N.B. ISIDNUM does not include $. */ |
463 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
464 | cur++; | |
0578f103 | 465 | |
78a11351 | 466 | pfile->buffer->cur = cur; |
0578f103 | 467 | } |
2cbf1359 | 468 | while (forms_identifier_p (pfile, false)); |
79bd622b | 469 | |
5bb46c08 | 470 | number->len = cur - base; |
471 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
472 | memcpy (dest, base, number->len); | |
473 | dest[number->len] = '\0'; | |
474 | number->text = dest; | |
79bd622b | 475 | } |
476 | ||
4970d4c2 | 477 | /* Create a token of type TYPE with a literal spelling. */ |
478 | static void | |
f7fdd7a1 | 479 | create_literal (cpp_reader *pfile, cpp_token *token, const uchar *base, |
480 | unsigned int len, enum cpp_ttype type) | |
4970d4c2 | 481 | { |
482 | uchar *dest = _cpp_unaligned_alloc (pfile, len + 1); | |
483 | ||
484 | memcpy (dest, base, len); | |
485 | dest[len] = '\0'; | |
486 | token->type = type; | |
487 | token->val.str.len = len; | |
488 | token->val.str.text = dest; | |
489 | } | |
490 | ||
5bb46c08 | 491 | /* Lexes a string, character constant, or angle-bracketed header file |
4970d4c2 | 492 | name. The stored string contains the spelling, including opening |
493 | quote and leading any leading 'L'. It returns the type of the | |
494 | literal, or CPP_OTHER if it was not properly terminated. | |
495 | ||
496 | The spelling is NUL-terminated, but it is not guaranteed that this | |
497 | is the first NUL since embedded NULs are preserved. */ | |
f80e83a9 | 498 | static void |
f7fdd7a1 | 499 | lex_string (cpp_reader *pfile, cpp_token *token, const uchar *base) |
0578f103 | 500 | { |
4970d4c2 | 501 | bool saw_NUL = false; |
502 | const uchar *cur; | |
5bb46c08 | 503 | cppchar_t terminator; |
4970d4c2 | 504 | enum cpp_ttype type; |
505 | ||
506 | cur = base; | |
507 | terminator = *cur++; | |
508 | if (terminator == 'L') | |
509 | terminator = *cur++; | |
510 | if (terminator == '\"') | |
511 | type = *base == 'L' ? CPP_WSTRING: CPP_STRING; | |
512 | else if (terminator == '\'') | |
513 | type = *base == 'L' ? CPP_WCHAR: CPP_CHAR; | |
514 | else | |
515 | terminator = '>', type = CPP_HEADER_NAME; | |
79bd622b | 516 | |
338fa5f7 | 517 | for (;;) |
0578f103 | 518 | { |
4970d4c2 | 519 | cppchar_t c = *cur++; |
4b0c16ee | 520 | |
edaf8cb5 | 521 | /* In #include-style directives, terminators are not escapable. */ |
4970d4c2 | 522 | if (c == '\\' && !pfile->state.angled_headers && *cur != '\n') |
523 | cur++; | |
524 | else if (c == terminator) | |
5bb46c08 | 525 | break; |
4970d4c2 | 526 | else if (c == '\n') |
338fa5f7 | 527 | { |
4970d4c2 | 528 | cur--; |
529 | type = CPP_OTHER; | |
530 | break; | |
0578f103 | 531 | } |
4970d4c2 | 532 | else if (c == '\0') |
533 | saw_NUL = true; | |
0578f103 | 534 | } |
535 | ||
4970d4c2 | 536 | if (saw_NUL && !pfile->state.skipping) |
537 | cpp_error (pfile, DL_WARNING, "null character(s) preserved in literal"); | |
0578f103 | 538 | |
4970d4c2 | 539 | pfile->buffer->cur = cur; |
540 | create_literal (pfile, token, base, cur - base, type); | |
338fa5f7 | 541 | } |
f80e83a9 | 542 | |
79bd622b | 543 | /* The stored comment includes the comment start and any terminator. */ |
2c63d6c8 | 544 | static void |
f7fdd7a1 | 545 | save_comment (cpp_reader *pfile, cpp_token *token, const unsigned char *from, |
546 | cppchar_t type) | |
2c63d6c8 | 547 | { |
f80e83a9 | 548 | unsigned char *buffer; |
d3f7919d | 549 | unsigned int len, clen; |
b1a9ff83 | 550 | |
f0495c2c | 551 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
1c124f85 | 552 | |
a543b315 | 553 | /* C++ comments probably (not definitely) have moved past a new |
554 | line, which we don't want to save in the comment. */ | |
1c124f85 | 555 | if (is_vspace (pfile->buffer->cur[-1])) |
a543b315 | 556 | len--; |
d3f7919d | 557 | |
558 | /* If we are currently in a directive, then we need to store all | |
559 | C++ comments as C comments internally, and so we need to | |
560 | allocate a little extra space in that case. | |
561 | ||
562 | Note that the only time we encounter a directive here is | |
563 | when we are saving comments in a "#define". */ | |
564 | clen = (pfile->state.in_directive && type == '/') ? len + 2 : len; | |
565 | ||
566 | buffer = _cpp_unaligned_alloc (pfile, clen); | |
b1a9ff83 | 567 | |
f80e83a9 | 568 | token->type = CPP_COMMENT; |
d3f7919d | 569 | token->val.str.len = clen; |
338fa5f7 | 570 | token->val.str.text = buffer; |
0578f103 | 571 | |
f0495c2c | 572 | buffer[0] = '/'; |
573 | memcpy (buffer + 1, from, len - 1); | |
d3f7919d | 574 | |
a113df96 | 575 | /* Finish conversion to a C comment, if necessary. */ |
d3f7919d | 576 | if (pfile->state.in_directive && type == '/') |
577 | { | |
578 | buffer[1] = '*'; | |
579 | buffer[clen - 2] = '*'; | |
580 | buffer[clen - 1] = '/'; | |
581 | } | |
338fa5f7 | 582 | } |
0578f103 | 583 | |
83dcbb5c | 584 | /* Allocate COUNT tokens for RUN. */ |
585 | void | |
f7fdd7a1 | 586 | _cpp_init_tokenrun (tokenrun *run, unsigned int count) |
83dcbb5c | 587 | { |
588 | run->base = xnewvec (cpp_token, count); | |
589 | run->limit = run->base + count; | |
590 | run->next = NULL; | |
591 | } | |
592 | ||
593 | /* Returns the next tokenrun, or creates one if there is none. */ | |
594 | static tokenrun * | |
f7fdd7a1 | 595 | next_tokenrun (tokenrun *run) |
83dcbb5c | 596 | { |
597 | if (run->next == NULL) | |
598 | { | |
599 | run->next = xnew (tokenrun); | |
fb5ab82c | 600 | run->next->prev = run; |
83dcbb5c | 601 | _cpp_init_tokenrun (run->next, 250); |
602 | } | |
603 | ||
604 | return run->next; | |
605 | } | |
606 | ||
f9b5f742 | 607 | /* Allocate a single token that is invalidated at the same time as the |
608 | rest of the tokens on the line. Has its line and col set to the | |
609 | same as the last lexed token, so that diagnostics appear in the | |
610 | right place. */ | |
611 | cpp_token * | |
f7fdd7a1 | 612 | _cpp_temp_token (cpp_reader *pfile) |
f9b5f742 | 613 | { |
614 | cpp_token *old, *result; | |
615 | ||
616 | old = pfile->cur_token - 1; | |
617 | if (pfile->cur_token == pfile->cur_run->limit) | |
618 | { | |
619 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
620 | pfile->cur_token = pfile->cur_run->base; | |
621 | } | |
622 | ||
623 | result = pfile->cur_token++; | |
624 | result->line = old->line; | |
625 | result->col = old->col; | |
626 | return result; | |
627 | } | |
628 | ||
10b4496a | 629 | /* Lex a token into RESULT (external interface). Takes care of issues |
630 | like directive handling, token lookahead, multiple include | |
3fb1e43b | 631 | optimization and skipping. */ |
c00e481c | 632 | const cpp_token * |
f7fdd7a1 | 633 | _cpp_lex_token (cpp_reader *pfile) |
83dcbb5c | 634 | { |
fb5ab82c | 635 | cpp_token *result; |
83dcbb5c | 636 | |
fb5ab82c | 637 | for (;;) |
83dcbb5c | 638 | { |
fb5ab82c | 639 | if (pfile->cur_token == pfile->cur_run->limit) |
83dcbb5c | 640 | { |
fb5ab82c | 641 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
642 | pfile->cur_token = pfile->cur_run->base; | |
83dcbb5c | 643 | } |
644 | ||
fb5ab82c | 645 | if (pfile->lookaheads) |
10b4496a | 646 | { |
647 | pfile->lookaheads--; | |
648 | result = pfile->cur_token++; | |
649 | } | |
fb5ab82c | 650 | else |
10b4496a | 651 | result = _cpp_lex_direct (pfile); |
fb5ab82c | 652 | |
653 | if (result->flags & BOL) | |
83dcbb5c | 654 | { |
fb5ab82c | 655 | /* Is this a directive. If _cpp_handle_directive returns |
656 | false, it is an assembler #. */ | |
657 | if (result->type == CPP_HASH | |
d6af0368 | 658 | /* 6.10.3 p 11: Directives in a list of macro arguments |
659 | gives undefined behavior. This implementation | |
660 | handles the directive as normal. */ | |
661 | && pfile->state.parsing_args != 1 | |
fb5ab82c | 662 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
663 | continue; | |
5621a364 | 664 | if (pfile->cb.line_change && !pfile->state.skipping) |
f7fdd7a1 | 665 | pfile->cb.line_change (pfile, result, pfile->state.parsing_args); |
83dcbb5c | 666 | } |
83dcbb5c | 667 | |
fb5ab82c | 668 | /* We don't skip tokens in directives. */ |
669 | if (pfile->state.in_directive) | |
670 | break; | |
83dcbb5c | 671 | |
fb5ab82c | 672 | /* Outside a directive, invalidate controlling macros. At file |
10b4496a | 673 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
7ef5b942 | 674 | get here and MI optimization works. */ |
83dcbb5c | 675 | pfile->mi_valid = false; |
fb5ab82c | 676 | |
677 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
678 | break; | |
83dcbb5c | 679 | } |
680 | ||
c00e481c | 681 | return result; |
83dcbb5c | 682 | } |
683 | ||
a54e0bf8 | 684 | /* Returns true if a fresh line has been loaded. */ |
685 | bool | |
f7fdd7a1 | 686 | _cpp_get_fresh_line (cpp_reader *pfile) |
0bb65704 | 687 | { |
a54e0bf8 | 688 | /* We can't get a new line until we leave the current directive. */ |
689 | if (pfile->state.in_directive) | |
690 | return false; | |
b1a9ff83 | 691 | |
a54e0bf8 | 692 | for (;;) |
fb83e0d6 | 693 | { |
a54e0bf8 | 694 | cpp_buffer *buffer = pfile->buffer; |
fb83e0d6 | 695 | |
a54e0bf8 | 696 | if (!buffer->need_line) |
697 | return true; | |
698 | ||
699 | if (buffer->next_line < buffer->rlimit) | |
0bb65704 | 700 | { |
a54e0bf8 | 701 | _cpp_clean_line (pfile); |
702 | return true; | |
703 | } | |
0bb65704 | 704 | |
a54e0bf8 | 705 | /* First, get out of parsing arguments state. */ |
706 | if (pfile->state.parsing_args) | |
707 | return false; | |
708 | ||
709 | /* End of buffer. Non-empty files should end in a newline. */ | |
710 | if (buffer->buf != buffer->rlimit | |
711 | && buffer->next_line > buffer->rlimit | |
712 | && !buffer->from_stage3) | |
713 | { | |
714 | /* Only warn once. */ | |
715 | buffer->next_line = buffer->rlimit; | |
716 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line - 1, | |
717 | CPP_BUF_COLUMN (buffer, buffer->cur), | |
718 | "no newline at end of file"); | |
719 | } | |
720 | ||
cc6448c7 | 721 | if (!buffer->prev) |
722 | return false; | |
723 | ||
a54e0bf8 | 724 | if (buffer->return_at_eof) |
725 | { | |
cc6448c7 | 726 | _cpp_pop_buffer (pfile); |
a54e0bf8 | 727 | return false; |
0bb65704 | 728 | } |
0bb65704 | 729 | |
a54e0bf8 | 730 | _cpp_pop_buffer (pfile); |
731 | } | |
0bb65704 | 732 | } |
733 | ||
edaf8cb5 | 734 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
735 | do \ | |
736 | { \ | |
737 | result->type = ELSE_TYPE; \ | |
738 | if (*buffer->cur == CHAR) \ | |
739 | buffer->cur++, result->type = THEN_TYPE; \ | |
740 | } \ | |
741 | while (0) | |
1c124f85 | 742 | |
10b4496a | 743 | /* Lex a token into pfile->cur_token, which is also incremented, to |
744 | get diagnostics pointing to the correct location. | |
745 | ||
746 | Does not handle issues such as token lookahead, multiple-include | |
4172d65e | 747 | optimization, directives, skipping etc. This function is only |
10b4496a | 748 | suitable for use by _cpp_lex_token, and in special cases like |
749 | lex_expansion_token which doesn't care for any of these issues. | |
750 | ||
751 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
752 | otherwise returns to the start of the token buffer if permissible. | |
753 | Returns the location of the lexed token. */ | |
754 | cpp_token * | |
f7fdd7a1 | 755 | _cpp_lex_direct (cpp_reader *pfile) |
0578f103 | 756 | { |
338fa5f7 | 757 | cppchar_t c; |
230f0943 | 758 | cpp_buffer *buffer; |
338fa5f7 | 759 | const unsigned char *comment_start; |
10b4496a | 760 | cpp_token *result = pfile->cur_token++; |
0653b94e | 761 | |
83dcbb5c | 762 | fresh_line: |
a54e0bf8 | 763 | result->flags = 0; |
764 | if (pfile->buffer->need_line) | |
765 | { | |
766 | if (!_cpp_get_fresh_line (pfile)) | |
767 | { | |
768 | result->type = CPP_EOF; | |
2908f819 | 769 | if (!pfile->state.in_directive) |
770 | { | |
771 | /* Tell the compiler the line number of the EOF token. */ | |
772 | result->line = pfile->line; | |
773 | result->flags = BOL; | |
774 | } | |
a54e0bf8 | 775 | return result; |
776 | } | |
777 | if (!pfile->keep_tokens) | |
778 | { | |
779 | pfile->cur_run = &pfile->base_run; | |
780 | result = pfile->base_run.base; | |
781 | pfile->cur_token = result + 1; | |
782 | } | |
783 | result->flags = BOL; | |
784 | if (pfile->state.parsing_args == 2) | |
785 | result->flags |= PREV_WHITE; | |
786 | } | |
230f0943 | 787 | buffer = pfile->buffer; |
83dcbb5c | 788 | update_tokens_line: |
36a0aa7c | 789 | result->line = pfile->line; |
f80e83a9 | 790 | |
83dcbb5c | 791 | skipped_white: |
a54e0bf8 | 792 | if (buffer->cur >= buffer->notes[buffer->cur_note].pos |
793 | && !pfile->overlaid_buffer) | |
794 | { | |
795 | _cpp_process_line_notes (pfile, false); | |
796 | result->line = pfile->line; | |
797 | } | |
1c124f85 | 798 | c = *buffer->cur++; |
83dcbb5c | 799 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
83dcbb5c | 800 | |
338fa5f7 | 801 | switch (c) |
0578f103 | 802 | { |
435fb09b | 803 | case ' ': case '\t': case '\f': case '\v': case '\0': |
804 | result->flags |= PREV_WHITE; | |
a54e0bf8 | 805 | skip_whitespace (pfile, c); |
806 | goto skipped_white; | |
338fa5f7 | 807 | |
a54e0bf8 | 808 | case '\n': |
809 | pfile->line++; | |
810 | buffer->need_line = true; | |
811 | goto fresh_line; | |
732cb4c9 | 812 | |
338fa5f7 | 813 | case '0': case '1': case '2': case '3': case '4': |
814 | case '5': case '6': case '7': case '8': case '9': | |
815 | result->type = CPP_NUMBER; | |
5bb46c08 | 816 | lex_number (pfile, &result->val.str); |
338fa5f7 | 817 | break; |
732cb4c9 | 818 | |
78c551ad | 819 | case 'L': |
820 | /* 'L' may introduce wide characters or strings. */ | |
5bb46c08 | 821 | if (*buffer->cur == '\'' || *buffer->cur == '"') |
822 | { | |
4970d4c2 | 823 | lex_string (pfile, result, buffer->cur - 1); |
5bb46c08 | 824 | break; |
825 | } | |
b1a9ff83 | 826 | /* Fall through. */ |
78c551ad | 827 | |
338fa5f7 | 828 | case '_': |
829 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
830 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
831 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
832 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
833 | case 'y': case 'z': | |
834 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
78c551ad | 835 | case 'G': case 'H': case 'I': case 'J': case 'K': |
338fa5f7 | 836 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
837 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
838 | case 'Y': case 'Z': | |
839 | result->type = CPP_NAME; | |
2cbf1359 | 840 | result->val.node = lex_identifier (pfile, buffer->cur - 1); |
338fa5f7 | 841 | |
338fa5f7 | 842 | /* Convert named operators to their proper types. */ |
78c551ad | 843 | if (result->val.node->flags & NODE_OPERATOR) |
338fa5f7 | 844 | { |
845 | result->flags |= NAMED_OP; | |
805e22b2 | 846 | result->type = result->val.node->directive_index; |
338fa5f7 | 847 | } |
848 | break; | |
849 | ||
850 | case '\'': | |
851 | case '"': | |
4970d4c2 | 852 | lex_string (pfile, result, buffer->cur - 1); |
338fa5f7 | 853 | break; |
f80e83a9 | 854 | |
338fa5f7 | 855 | case '/': |
f0495c2c | 856 | /* A potential block or line comment. */ |
857 | comment_start = buffer->cur; | |
edaf8cb5 | 858 | c = *buffer->cur; |
859 | ||
f0495c2c | 860 | if (c == '*') |
861 | { | |
a54e0bf8 | 862 | if (_cpp_skip_block_comment (pfile)) |
73328dce | 863 | cpp_error (pfile, DL_ERROR, "unterminated comment"); |
338fa5f7 | 864 | } |
1c124f85 | 865 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
866 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
338fa5f7 | 867 | { |
5db5d057 | 868 | /* Warn about comments only if pedantically GNUC89, and not |
869 | in system headers. */ | |
870 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
66914e49 | 871 | && ! buffer->warned_cplusplus_comments) |
f80e83a9 | 872 | { |
73328dce | 873 | cpp_error (pfile, DL_PEDWARN, |
ba059ac0 | 874 | "C++ style comments are not allowed in ISO C90"); |
73328dce | 875 | cpp_error (pfile, DL_PEDWARN, |
876 | "(this will be reported only once per input file)"); | |
f0495c2c | 877 | buffer->warned_cplusplus_comments = 1; |
878 | } | |
338fa5f7 | 879 | |
e1caf668 | 880 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
73328dce | 881 | cpp_error (pfile, DL_WARNING, "multi-line comment"); |
f0495c2c | 882 | } |
1c124f85 | 883 | else if (c == '=') |
884 | { | |
edaf8cb5 | 885 | buffer->cur++; |
1c124f85 | 886 | result->type = CPP_DIV_EQ; |
887 | break; | |
888 | } | |
889 | else | |
890 | { | |
1c124f85 | 891 | result->type = CPP_DIV; |
892 | break; | |
893 | } | |
338fa5f7 | 894 | |
f0495c2c | 895 | if (!pfile->state.save_comments) |
896 | { | |
897 | result->flags |= PREV_WHITE; | |
83dcbb5c | 898 | goto update_tokens_line; |
338fa5f7 | 899 | } |
f0495c2c | 900 | |
901 | /* Save the comment as a token in its own right. */ | |
d3f7919d | 902 | save_comment (pfile, result, comment_start, c); |
fb5ab82c | 903 | break; |
338fa5f7 | 904 | |
905 | case '<': | |
906 | if (pfile->state.angled_headers) | |
907 | { | |
4970d4c2 | 908 | lex_string (pfile, result, buffer->cur - 1); |
1c124f85 | 909 | break; |
338fa5f7 | 910 | } |
0578f103 | 911 | |
edaf8cb5 | 912 | result->type = CPP_LESS; |
913 | if (*buffer->cur == '=') | |
914 | buffer->cur++, result->type = CPP_LESS_EQ; | |
915 | else if (*buffer->cur == '<') | |
338fa5f7 | 916 | { |
edaf8cb5 | 917 | buffer->cur++; |
918 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); | |
338fa5f7 | 919 | } |
edaf8cb5 | 920 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) |
338fa5f7 | 921 | { |
edaf8cb5 | 922 | buffer->cur++; |
923 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); | |
338fa5f7 | 924 | } |
edaf8cb5 | 925 | else if (CPP_OPTION (pfile, digraphs)) |
1c124f85 | 926 | { |
edaf8cb5 | 927 | if (*buffer->cur == ':') |
928 | { | |
929 | buffer->cur++; | |
930 | result->flags |= DIGRAPH; | |
931 | result->type = CPP_OPEN_SQUARE; | |
932 | } | |
933 | else if (*buffer->cur == '%') | |
934 | { | |
935 | buffer->cur++; | |
936 | result->flags |= DIGRAPH; | |
937 | result->type = CPP_OPEN_BRACE; | |
938 | } | |
1c124f85 | 939 | } |
338fa5f7 | 940 | break; |
941 | ||
942 | case '>': | |
edaf8cb5 | 943 | result->type = CPP_GREATER; |
944 | if (*buffer->cur == '=') | |
945 | buffer->cur++, result->type = CPP_GREATER_EQ; | |
946 | else if (*buffer->cur == '>') | |
338fa5f7 | 947 | { |
edaf8cb5 | 948 | buffer->cur++; |
949 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); | |
950 | } | |
951 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) | |
952 | { | |
953 | buffer->cur++; | |
954 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); | |
338fa5f7 | 955 | } |
956 | break; | |
957 | ||
f669338a | 958 | case '%': |
edaf8cb5 | 959 | result->type = CPP_MOD; |
960 | if (*buffer->cur == '=') | |
961 | buffer->cur++, result->type = CPP_MOD_EQ; | |
962 | else if (CPP_OPTION (pfile, digraphs)) | |
1c124f85 | 963 | { |
edaf8cb5 | 964 | if (*buffer->cur == ':') |
1c124f85 | 965 | { |
edaf8cb5 | 966 | buffer->cur++; |
967 | result->flags |= DIGRAPH; | |
968 | result->type = CPP_HASH; | |
969 | if (*buffer->cur == '%' && buffer->cur[1] == ':') | |
970 | buffer->cur += 2, result->type = CPP_PASTE; | |
971 | } | |
972 | else if (*buffer->cur == '>') | |
973 | { | |
974 | buffer->cur++; | |
975 | result->flags |= DIGRAPH; | |
976 | result->type = CPP_CLOSE_BRACE; | |
1c124f85 | 977 | } |
1c124f85 | 978 | } |
338fa5f7 | 979 | break; |
980 | ||
f669338a | 981 | case '.': |
1c124f85 | 982 | result->type = CPP_DOT; |
edaf8cb5 | 983 | if (ISDIGIT (*buffer->cur)) |
1c124f85 | 984 | { |
985 | result->type = CPP_NUMBER; | |
5bb46c08 | 986 | lex_number (pfile, &result->val.str); |
1c124f85 | 987 | } |
edaf8cb5 | 988 | else if (*buffer->cur == '.' && buffer->cur[1] == '.') |
989 | buffer->cur += 2, result->type = CPP_ELLIPSIS; | |
990 | else if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) | |
991 | buffer->cur++, result->type = CPP_DOT_STAR; | |
338fa5f7 | 992 | break; |
0578f103 | 993 | |
338fa5f7 | 994 | case '+': |
edaf8cb5 | 995 | result->type = CPP_PLUS; |
996 | if (*buffer->cur == '+') | |
997 | buffer->cur++, result->type = CPP_PLUS_PLUS; | |
998 | else if (*buffer->cur == '=') | |
999 | buffer->cur++, result->type = CPP_PLUS_EQ; | |
338fa5f7 | 1000 | break; |
ac0749c7 | 1001 | |
338fa5f7 | 1002 | case '-': |
edaf8cb5 | 1003 | result->type = CPP_MINUS; |
1004 | if (*buffer->cur == '>') | |
338fa5f7 | 1005 | { |
edaf8cb5 | 1006 | buffer->cur++; |
1c124f85 | 1007 | result->type = CPP_DEREF; |
edaf8cb5 | 1008 | if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) |
1009 | buffer->cur++, result->type = CPP_DEREF_STAR; | |
1c124f85 | 1010 | } |
edaf8cb5 | 1011 | else if (*buffer->cur == '-') |
1012 | buffer->cur++, result->type = CPP_MINUS_MINUS; | |
1013 | else if (*buffer->cur == '=') | |
1014 | buffer->cur++, result->type = CPP_MINUS_EQ; | |
338fa5f7 | 1015 | break; |
0578f103 | 1016 | |
338fa5f7 | 1017 | case '&': |
edaf8cb5 | 1018 | result->type = CPP_AND; |
1019 | if (*buffer->cur == '&') | |
1020 | buffer->cur++, result->type = CPP_AND_AND; | |
1021 | else if (*buffer->cur == '=') | |
1022 | buffer->cur++, result->type = CPP_AND_EQ; | |
338fa5f7 | 1023 | break; |
b1a9ff83 | 1024 | |
338fa5f7 | 1025 | case '|': |
edaf8cb5 | 1026 | result->type = CPP_OR; |
1027 | if (*buffer->cur == '|') | |
1028 | buffer->cur++, result->type = CPP_OR_OR; | |
1029 | else if (*buffer->cur == '=') | |
1030 | buffer->cur++, result->type = CPP_OR_EQ; | |
338fa5f7 | 1031 | break; |
0578f103 | 1032 | |
338fa5f7 | 1033 | case ':': |
edaf8cb5 | 1034 | result->type = CPP_COLON; |
1035 | if (*buffer->cur == ':' && CPP_OPTION (pfile, cplusplus)) | |
1036 | buffer->cur++, result->type = CPP_SCOPE; | |
1037 | else if (*buffer->cur == '>' && CPP_OPTION (pfile, digraphs)) | |
338fa5f7 | 1038 | { |
edaf8cb5 | 1039 | buffer->cur++; |
338fa5f7 | 1040 | result->flags |= DIGRAPH; |
1c124f85 | 1041 | result->type = CPP_CLOSE_SQUARE; |
1042 | } | |
338fa5f7 | 1043 | break; |
0578f103 | 1044 | |
1c124f85 | 1045 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1046 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1047 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1048 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1049 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1050 | ||
a54e0bf8 | 1051 | case '?': result->type = CPP_QUERY; break; |
338fa5f7 | 1052 | case '~': result->type = CPP_COMPL; break; |
1053 | case ',': result->type = CPP_COMMA; break; | |
1054 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1055 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1056 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1057 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1058 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1059 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1060 | case ';': result->type = CPP_SEMICOLON; break; | |
1061 | ||
7fd957fe | 1062 | /* @ is a punctuator in Objective-C. */ |
9ee99ac6 | 1063 | case '@': result->type = CPP_ATSIGN; break; |
338fa5f7 | 1064 | |
78c551ad | 1065 | case '$': |
2cbf1359 | 1066 | case '\\': |
1067 | { | |
1068 | const uchar *base = --buffer->cur; | |
78c551ad | 1069 | |
2cbf1359 | 1070 | if (forms_identifier_p (pfile, true)) |
1071 | { | |
1072 | result->type = CPP_NAME; | |
1073 | result->val.node = lex_identifier (pfile, base); | |
1074 | break; | |
1075 | } | |
1076 | buffer->cur++; | |
bc205914 | 1077 | } |
2cbf1359 | 1078 | |
bc205914 | 1079 | default: |
4970d4c2 | 1080 | create_literal (pfile, result, buffer->cur - 1, 1, CPP_OTHER); |
1081 | break; | |
338fa5f7 | 1082 | } |
fb5ab82c | 1083 | |
1084 | return result; | |
338fa5f7 | 1085 | } |
1086 | ||
b1280514 | 1087 | /* An upper bound on the number of bytes needed to spell TOKEN. |
1088 | Does not include preceding whitespace. */ | |
79bd622b | 1089 | unsigned int |
f7fdd7a1 | 1090 | cpp_token_len (const cpp_token *token) |
338fa5f7 | 1091 | { |
79bd622b | 1092 | unsigned int len; |
cfad5579 | 1093 | |
79bd622b | 1094 | switch (TOKEN_SPELL (token)) |
f80e83a9 | 1095 | { |
b1280514 | 1096 | default: len = 4; break; |
4970d4c2 | 1097 | case SPELL_LITERAL: len = token->val.str.len; break; |
c86dbc5b | 1098 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; |
f80e83a9 | 1099 | } |
b1280514 | 1100 | |
1101 | return len; | |
cfad5579 | 1102 | } |
1103 | ||
f80e83a9 | 1104 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
c5ea33a8 | 1105 | already contain the enough space to hold the token's spelling. |
f7fdd7a1 | 1106 | Returns a pointer to the character after the last character written. |
1107 | FIXME: Would be nice if we didn't need the PFILE argument. */ | |
79bd622b | 1108 | unsigned char * |
f7fdd7a1 | 1109 | cpp_spell_token (cpp_reader *pfile, const cpp_token *token, |
1110 | unsigned char *buffer) | |
f80e83a9 | 1111 | { |
7e842f95 | 1112 | switch (TOKEN_SPELL (token)) |
f80e83a9 | 1113 | { |
1114 | case SPELL_OPERATOR: | |
1115 | { | |
1116 | const unsigned char *spelling; | |
1117 | unsigned char c; | |
ab12a39c | 1118 | |
f80e83a9 | 1119 | if (token->flags & DIGRAPH) |
ee6c4e4b | 1120 | spelling |
1121 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
31674461 | 1122 | else if (token->flags & NAMED_OP) |
1123 | goto spell_ident; | |
f80e83a9 | 1124 | else |
7e842f95 | 1125 | spelling = TOKEN_NAME (token); |
b1a9ff83 | 1126 | |
f80e83a9 | 1127 | while ((c = *spelling++) != '\0') |
1128 | *buffer++ = c; | |
1129 | } | |
1130 | break; | |
ab12a39c | 1131 | |
8d27e472 | 1132 | spell_ident: |
f80e83a9 | 1133 | case SPELL_IDENT: |
c86dbc5b | 1134 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1135 | buffer += NODE_LEN (token->val.node); | |
f80e83a9 | 1136 | break; |
ab12a39c | 1137 | |
4970d4c2 | 1138 | case SPELL_LITERAL: |
8d27e472 | 1139 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1140 | buffer += token->val.str.len; | |
1141 | break; | |
1142 | ||
f80e83a9 | 1143 | case SPELL_NONE: |
73328dce | 1144 | cpp_error (pfile, DL_ICE, "unspellable token %s", TOKEN_NAME (token)); |
f80e83a9 | 1145 | break; |
1146 | } | |
ab12a39c | 1147 | |
f80e83a9 | 1148 | return buffer; |
1149 | } | |
ab12a39c | 1150 | |
e484a1cc | 1151 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1152 | freed when the reader is destroyed. Useful for diagnostics. */ | |
79bd622b | 1153 | unsigned char * |
f7fdd7a1 | 1154 | cpp_token_as_text (cpp_reader *pfile, const cpp_token *token) |
b1280514 | 1155 | { |
1156 | unsigned int len = cpp_token_len (token) + 1; | |
1fdf6039 | 1157 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
6060326b | 1158 | |
79bd622b | 1159 | end = cpp_spell_token (pfile, token, start); |
1160 | end[0] = '\0'; | |
6060326b | 1161 | |
79bd622b | 1162 | return start; |
1163 | } | |
6060326b | 1164 | |
e484a1cc | 1165 | /* Used by C front ends, which really should move to using |
1166 | cpp_token_as_text. */ | |
79bd622b | 1167 | const char * |
f7fdd7a1 | 1168 | cpp_type2name (enum cpp_ttype type) |
79bd622b | 1169 | { |
1170 | return (const char *) token_spellings[type].name; | |
1171 | } | |
6060326b | 1172 | |
f9b5f742 | 1173 | /* Writes the spelling of token to FP, without any preceding space. |
1174 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1175 | double-buffering. */ | |
79bd622b | 1176 | void |
f7fdd7a1 | 1177 | cpp_output_token (const cpp_token *token, FILE *fp) |
79bd622b | 1178 | { |
79bd622b | 1179 | switch (TOKEN_SPELL (token)) |
6060326b | 1180 | { |
79bd622b | 1181 | case SPELL_OPERATOR: |
1182 | { | |
1183 | const unsigned char *spelling; | |
28874558 | 1184 | int c; |
6060326b | 1185 | |
79bd622b | 1186 | if (token->flags & DIGRAPH) |
ee6c4e4b | 1187 | spelling |
1188 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
79bd622b | 1189 | else if (token->flags & NAMED_OP) |
1190 | goto spell_ident; | |
1191 | else | |
1192 | spelling = TOKEN_NAME (token); | |
f80e83a9 | 1193 | |
28874558 | 1194 | c = *spelling; |
1195 | do | |
1196 | putc (c, fp); | |
1197 | while ((c = *++spelling) != '\0'); | |
79bd622b | 1198 | } |
1199 | break; | |
f80e83a9 | 1200 | |
79bd622b | 1201 | spell_ident: |
1202 | case SPELL_IDENT: | |
28874558 | 1203 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
79bd622b | 1204 | break; |
f80e83a9 | 1205 | |
4970d4c2 | 1206 | case SPELL_LITERAL: |
8d27e472 | 1207 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
1208 | break; | |
1209 | ||
79bd622b | 1210 | case SPELL_NONE: |
1211 | /* An error, most probably. */ | |
1212 | break; | |
f80e83a9 | 1213 | } |
6060326b | 1214 | } |
1215 | ||
79bd622b | 1216 | /* Compare two tokens. */ |
1217 | int | |
f7fdd7a1 | 1218 | _cpp_equiv_tokens (const cpp_token *a, const cpp_token *b) |
6060326b | 1219 | { |
79bd622b | 1220 | if (a->type == b->type && a->flags == b->flags) |
1221 | switch (TOKEN_SPELL (a)) | |
1222 | { | |
1223 | default: /* Keep compiler happy. */ | |
1224 | case SPELL_OPERATOR: | |
1225 | return 1; | |
79bd622b | 1226 | case SPELL_NONE: |
588d632b | 1227 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
79bd622b | 1228 | case SPELL_IDENT: |
1229 | return a->val.node == b->val.node; | |
4970d4c2 | 1230 | case SPELL_LITERAL: |
79bd622b | 1231 | return (a->val.str.len == b->val.str.len |
1232 | && !memcmp (a->val.str.text, b->val.str.text, | |
1233 | a->val.str.len)); | |
1234 | } | |
6060326b | 1235 | |
f80e83a9 | 1236 | return 0; |
1237 | } | |
1238 | ||
79bd622b | 1239 | /* Returns nonzero if a space should be inserted to avoid an |
1240 | accidental token paste for output. For simplicity, it is | |
1241 | conservative, and occasionally advises a space where one is not | |
1242 | needed, e.g. "." and ".2". */ | |
79bd622b | 1243 | int |
f7fdd7a1 | 1244 | cpp_avoid_paste (cpp_reader *pfile, const cpp_token *token1, |
1245 | const cpp_token *token2) | |
6060326b | 1246 | { |
79bd622b | 1247 | enum cpp_ttype a = token1->type, b = token2->type; |
1248 | cppchar_t c; | |
6060326b | 1249 | |
79bd622b | 1250 | if (token1->flags & NAMED_OP) |
1251 | a = CPP_NAME; | |
1252 | if (token2->flags & NAMED_OP) | |
1253 | b = CPP_NAME; | |
6060326b | 1254 | |
79bd622b | 1255 | c = EOF; |
1256 | if (token2->flags & DIGRAPH) | |
ee6c4e4b | 1257 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
79bd622b | 1258 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1259 | c = token_spellings[b].name[0]; | |
6060326b | 1260 | |
79bd622b | 1261 | /* Quickly get everything that can paste with an '='. */ |
ee6c4e4b | 1262 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
79bd622b | 1263 | return 1; |
6060326b | 1264 | |
79bd622b | 1265 | switch (a) |
6060326b | 1266 | { |
79bd622b | 1267 | case CPP_GREATER: return c == '>' || c == '?'; |
1268 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1269 | case CPP_PLUS: return c == '+'; | |
1270 | case CPP_MINUS: return c == '-' || c == '>'; | |
1271 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1272 | case CPP_MOD: return c == ':' || c == '>'; | |
1273 | case CPP_AND: return c == '&'; | |
1274 | case CPP_OR: return c == '|'; | |
1275 | case CPP_COLON: return c == ':' || c == '>'; | |
1276 | case CPP_DEREF: return c == '*'; | |
efdcc728 | 1277 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
79bd622b | 1278 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1279 | case CPP_NAME: return ((b == CPP_NUMBER | |
1280 | && name_p (pfile, &token2->val.str)) | |
1281 | || b == CPP_NAME | |
1282 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1283 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1284 | || c == '.' || c == '+' || c == '-'); | |
2cbf1359 | 1285 | /* UCNs */ |
bc205914 | 1286 | case CPP_OTHER: return ((token1->val.str.text[0] == '\\' |
1287 | && b == CPP_NAME) | |
2cbf1359 | 1288 | || (CPP_OPTION (pfile, objc) |
bc205914 | 1289 | && token1->val.str.text[0] == '@' |
2cbf1359 | 1290 | && (b == CPP_NAME || b == CPP_STRING))); |
79bd622b | 1291 | default: break; |
6060326b | 1292 | } |
6060326b | 1293 | |
deb356cf | 1294 | return 0; |
6060326b | 1295 | } |
1296 | ||
79bd622b | 1297 | /* Output all the remaining tokens on the current line, and a newline |
f9b5f742 | 1298 | character, to FP. Leading whitespace is removed. If there are |
1299 | macros, special token padding is not performed. */ | |
6060326b | 1300 | void |
f7fdd7a1 | 1301 | cpp_output_line (cpp_reader *pfile, FILE *fp) |
6060326b | 1302 | { |
f9b5f742 | 1303 | const cpp_token *token; |
7e842f95 | 1304 | |
f9b5f742 | 1305 | token = cpp_get_token (pfile); |
1306 | while (token->type != CPP_EOF) | |
7e842f95 | 1307 | { |
f9b5f742 | 1308 | cpp_output_token (token, fp); |
1309 | token = cpp_get_token (pfile); | |
1310 | if (token->flags & PREV_WHITE) | |
1311 | putc (' ', fp); | |
7e842f95 | 1312 | } |
1313 | ||
79bd622b | 1314 | putc ('\n', fp); |
f80e83a9 | 1315 | } |
6060326b | 1316 | |
084163dc | 1317 | /* Memory buffers. Changing these three constants can have a dramatic |
1318 | effect on performance. The values here are reasonable defaults, | |
1319 | but might be tuned. If you adjust them, be sure to test across a | |
1320 | range of uses of cpplib, including heavy nested function-like macro | |
1321 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1322 | good tool for this). */ | |
1323 | #define MIN_BUFF_SIZE 8000 | |
1e0ef2fd | 1324 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
084163dc | 1325 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1326 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
deb356cf | 1327 | |
1e0ef2fd | 1328 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1329 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1330 | #endif | |
1331 | ||
1785b647 | 1332 | /* Create a new allocation buffer. Place the control block at the end |
1333 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
06c92cbc | 1334 | static _cpp_buff * |
f7fdd7a1 | 1335 | new_buff (size_t len) |
06c92cbc | 1336 | { |
1337 | _cpp_buff *result; | |
1fdf6039 | 1338 | unsigned char *base; |
06c92cbc | 1339 | |
084163dc | 1340 | if (len < MIN_BUFF_SIZE) |
1341 | len = MIN_BUFF_SIZE; | |
198b48a0 | 1342 | len = CPP_ALIGN (len); |
06c92cbc | 1343 | |
1344 | base = xmalloc (len + sizeof (_cpp_buff)); | |
1345 | result = (_cpp_buff *) (base + len); | |
1346 | result->base = base; | |
1347 | result->cur = base; | |
1348 | result->limit = base + len; | |
1349 | result->next = NULL; | |
1350 | return result; | |
1351 | } | |
1352 | ||
1353 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1354 | void | |
f7fdd7a1 | 1355 | _cpp_release_buff (cpp_reader *pfile, _cpp_buff *buff) |
06c92cbc | 1356 | { |
1357 | _cpp_buff *end = buff; | |
1358 | ||
1359 | while (end->next) | |
1360 | end = end->next; | |
1361 | end->next = pfile->free_buffs; | |
1362 | pfile->free_buffs = buff; | |
1363 | } | |
1364 | ||
1365 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1366 | _cpp_buff * | |
f7fdd7a1 | 1367 | _cpp_get_buff (cpp_reader *pfile, size_t min_size) |
06c92cbc | 1368 | { |
1369 | _cpp_buff *result, **p; | |
1370 | ||
1371 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1372 | { | |
4b31a107 | 1373 | size_t size; |
084163dc | 1374 | |
1375 | if (*p == NULL) | |
06c92cbc | 1376 | return new_buff (min_size); |
084163dc | 1377 | result = *p; |
1378 | size = result->limit - result->base; | |
1379 | /* Return a buffer that's big enough, but don't waste one that's | |
1380 | way too big. */ | |
4085c149 | 1381 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
06c92cbc | 1382 | break; |
1383 | } | |
1384 | ||
1385 | *p = result->next; | |
1386 | result->next = NULL; | |
1387 | result->cur = result->base; | |
1388 | return result; | |
1389 | } | |
1390 | ||
20dd417a | 1391 | /* Creates a new buffer with enough space to hold the uncommitted |
e6a5f963 | 1392 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
1393 | the excess bytes to the new buffer. Chains the new buffer after | |
1394 | BUFF, and returns the new buffer. */ | |
06c92cbc | 1395 | _cpp_buff * |
f7fdd7a1 | 1396 | _cpp_append_extend_buff (cpp_reader *pfile, _cpp_buff *buff, size_t min_extra) |
06c92cbc | 1397 | { |
4b31a107 | 1398 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
e6a5f963 | 1399 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
06c92cbc | 1400 | |
e6a5f963 | 1401 | buff->next = new_buff; |
1402 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
1403 | return new_buff; | |
1404 | } | |
1405 | ||
20dd417a | 1406 | /* Creates a new buffer with enough space to hold the uncommitted |
e6a5f963 | 1407 | remaining bytes of the buffer pointed to by BUFF, and at least |
1408 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
1409 | Chains the new buffer before the buffer pointed to by BUFF, and | |
1410 | updates the pointer to point to the new buffer. */ | |
1411 | void | |
f7fdd7a1 | 1412 | _cpp_extend_buff (cpp_reader *pfile, _cpp_buff **pbuff, size_t min_extra) |
e6a5f963 | 1413 | { |
1414 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
1415 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
1416 | ||
1417 | new_buff = _cpp_get_buff (pfile, size); | |
1418 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
1419 | new_buff->next = old_buff; | |
1420 | *pbuff = new_buff; | |
06c92cbc | 1421 | } |
1422 | ||
1423 | /* Free a chain of buffers starting at BUFF. */ | |
1424 | void | |
1425 | _cpp_free_buff (buff) | |
1426 | _cpp_buff *buff; | |
1427 | { | |
1428 | _cpp_buff *next; | |
1429 | ||
1430 | for (; buff; buff = next) | |
1431 | { | |
1432 | next = buff->next; | |
1433 | free (buff->base); | |
1434 | } | |
1435 | } | |
deb356cf | 1436 | |
1fdf6039 | 1437 | /* Allocate permanent, unaligned storage of length LEN. */ |
1438 | unsigned char * | |
f7fdd7a1 | 1439 | _cpp_unaligned_alloc (cpp_reader *pfile, size_t len) |
1fdf6039 | 1440 | { |
1441 | _cpp_buff *buff = pfile->u_buff; | |
1442 | unsigned char *result = buff->cur; | |
1443 | ||
1444 | if (len > (size_t) (buff->limit - result)) | |
1445 | { | |
1446 | buff = _cpp_get_buff (pfile, len); | |
1447 | buff->next = pfile->u_buff; | |
1448 | pfile->u_buff = buff; | |
1449 | result = buff->cur; | |
1450 | } | |
1451 | ||
1452 | buff->cur = result + len; | |
1453 | return result; | |
1454 | } | |
1455 | ||
1e0ef2fd | 1456 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
1457 | That buffer is used for growing allocations when saving macro | |
1458 | replacement lists in a #define, and when parsing an answer to an | |
1459 | assertion in #assert, #unassert or #if (and therefore possibly | |
1460 | whilst expanding macros). It therefore must not be used by any | |
1461 | code that they might call: specifically the lexer and the guts of | |
1462 | the macro expander. | |
1463 | ||
1464 | All existing other uses clearly fit this restriction: storing | |
1465 | registered pragmas during initialization. */ | |
79bd622b | 1466 | unsigned char * |
f7fdd7a1 | 1467 | _cpp_aligned_alloc (cpp_reader *pfile, size_t len) |
89b05ef6 | 1468 | { |
e6a5f963 | 1469 | _cpp_buff *buff = pfile->a_buff; |
1470 | unsigned char *result = buff->cur; | |
89b05ef6 | 1471 | |
e6a5f963 | 1472 | if (len > (size_t) (buff->limit - result)) |
89b05ef6 | 1473 | { |
e6a5f963 | 1474 | buff = _cpp_get_buff (pfile, len); |
1475 | buff->next = pfile->a_buff; | |
1476 | pfile->a_buff = buff; | |
1477 | result = buff->cur; | |
89b05ef6 | 1478 | } |
f80e83a9 | 1479 | |
e6a5f963 | 1480 | buff->cur = result + len; |
79bd622b | 1481 | return result; |
f80e83a9 | 1482 | } |