]>
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" | |
0578f103 | 24 | #include "cpplib.h" |
25 | #include "cpphash.h" | |
26 | ||
79bd622b | 27 | enum spell_type |
241e762e | 28 | { |
79bd622b | 29 | SPELL_OPERATOR = 0, |
79bd622b | 30 | SPELL_IDENT, |
4970d4c2 | 31 | SPELL_LITERAL, |
79bd622b | 32 | SPELL_NONE |
241e762e | 33 | }; |
34 | ||
79bd622b | 35 | struct token_spelling |
241e762e | 36 | { |
79bd622b | 37 | enum spell_type category; |
38 | const unsigned char *name; | |
241e762e | 39 | }; |
40 | ||
0ca849f9 | 41 | static const unsigned char *const digraph_spellings[] = |
42 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
79bd622b | 43 | |
44 | #define OP(e, s) { SPELL_OPERATOR, U s }, | |
18e43155 | 45 | #define TK(e, s) { s, U #e }, |
0ca849f9 | 46 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
79bd622b | 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) | |
e2f9a79f | 52 | |
f7fdd7a1 | 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 *); | |
f7fdd7a1 | 65 | static tokenrun *next_tokenrun (tokenrun *); |
66 | ||
f7fdd7a1 | 67 | static _cpp_buff *new_buff (size_t); |
bce8e0c0 | 68 | |
e920deaf | 69 | |
f80e83a9 | 70 | /* Utility routine: |
2c63d6c8 | 71 | |
76faa4c0 | 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. */ | |
f80e83a9 | 74 | int |
f7fdd7a1 | 75 | cpp_ideq (const cpp_token *token, const char *string) |
f80e83a9 | 76 | { |
76faa4c0 | 77 | if (token->type != CPP_NAME) |
f80e83a9 | 78 | return 0; |
76faa4c0 | 79 | |
b6d18b0a | 80 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
bce8e0c0 | 81 | } |
50fd6b48 | 82 | |
a54e0bf8 | 83 | /* Record a note TYPE at byte POS into the current cleaned logical |
84 | line. */ | |
1e0ef2fd | 85 | static void |
f7fdd7a1 | 86 | add_line_note (cpp_buffer *buffer, const uchar *pos, unsigned int type) |
338fa5f7 | 87 | { |
a54e0bf8 | 88 | if (buffer->notes_used == buffer->notes_cap) |
89 | { | |
90 | buffer->notes_cap = buffer->notes_cap * 2 + 200; | |
f0af5a88 | 91 | buffer->notes = xrealloc (buffer->notes, |
92 | buffer->notes_cap * sizeof (_cpp_line_note)); | |
a54e0bf8 | 93 | } |
338fa5f7 | 94 | |
a54e0bf8 | 95 | buffer->notes[buffer->notes_used].pos = pos; |
96 | buffer->notes[buffer->notes_used].type = type; | |
97 | buffer->notes_used++; | |
338fa5f7 | 98 | } |
99 | ||
a54e0bf8 | 100 | /* Returns with a logical line that contains no escaped newlines or |
101 | trigraphs. This is a time-critical inner loop. */ | |
102 | void | |
f7fdd7a1 | 103 | _cpp_clean_line (cpp_reader *pfile) |
0578f103 | 104 | { |
a54e0bf8 | 105 | cpp_buffer *buffer; |
106 | const uchar *s; | |
107 | uchar c, *d, *p; | |
1e0ef2fd | 108 | |
a54e0bf8 | 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; | |
1e0ef2fd | 114 | |
a54e0bf8 | 115 | if (!buffer->from_stage3) |
0578f103 | 116 | { |
a54e0bf8 | 117 | d = (uchar *) s; |
118 | ||
119 | for (;;) | |
4b912310 | 120 | { |
a54e0bf8 | 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 | ||
aad4a87f | 139 | add_line_note (buffer, p - 1, p != d ? ' ': '\\'); |
a54e0bf8 | 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. */ | |
aad4a87f | 146 | add_line_note (buffer, d, s[2]); |
a54e0bf8 | 147 | if (CPP_OPTION (pfile, trigraphs)) |
148 | { | |
149 | *d = _cpp_trigraph_map[s[2]]; | |
150 | s += 2; | |
151 | } | |
152 | } | |
4b912310 | 153 | } |
0578f103 | 154 | } |
a54e0bf8 | 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 | } | |
338fa5f7 | 166 | |
a54e0bf8 | 167 | *d = '\n'; |
aad4a87f | 168 | /* A sentinel note that should never be processed. */ |
169 | add_line_note (buffer, d + 1, '\n'); | |
a54e0bf8 | 170 | buffer->next_line = s + 1; |
0578f103 | 171 | } |
172 | ||
3078f2b2 | 173 | /* Return true if the trigraph indicated by NOTE should be warned |
174 | about in a comment. */ | |
175 | static bool | |
f7fdd7a1 | 176 | warn_in_comment (cpp_reader *pfile, _cpp_line_note *note) |
3078f2b2 | 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 | |
7ef5b942 | 182 | behavior. */ |
3078f2b2 | 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 | ||
a54e0bf8 | 201 | /* Process the notes created by add_line_note as far as the current |
202 | location. */ | |
203 | void | |
f7fdd7a1 | 204 | _cpp_process_line_notes (cpp_reader *pfile, int in_comment) |
0578f103 | 205 | { |
c808d026 | 206 | cpp_buffer *buffer = pfile->buffer; |
207 | ||
a54e0bf8 | 208 | for (;;) |
f80e83a9 | 209 | { |
a54e0bf8 | 210 | _cpp_line_note *note = &buffer->notes[buffer->cur_note]; |
211 | unsigned int col; | |
396ffa86 | 212 | |
a54e0bf8 | 213 | if (note->pos > buffer->cur) |
214 | break; | |
396ffa86 | 215 | |
a54e0bf8 | 216 | buffer->cur_note++; |
217 | col = CPP_BUF_COLUMN (buffer, note->pos + 1); | |
435fb09b | 218 | |
aad4a87f | 219 | if (note->type == '\\' || note->type == ' ') |
a54e0bf8 | 220 | { |
aad4a87f | 221 | if (note->type == ' ' && !in_comment) |
a54e0bf8 | 222 | cpp_error_with_line (pfile, DL_WARNING, pfile->line, col, |
223 | "backslash and newline separated by space"); | |
aad4a87f | 224 | |
a54e0bf8 | 225 | if (buffer->next_line > buffer->rlimit) |
1e0ef2fd | 226 | { |
a54e0bf8 | 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; | |
1e0ef2fd | 231 | } |
a54e0bf8 | 232 | |
233 | buffer->line_base = note->pos; | |
234 | pfile->line++; | |
338fa5f7 | 235 | } |
aad4a87f | 236 | else if (_cpp_trigraph_map[note->type]) |
237 | { | |
3078f2b2 | 238 | if (CPP_OPTION (pfile, warn_trigraphs) |
239 | && (!in_comment || warn_in_comment (pfile, note))) | |
aad4a87f | 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 | |
1542b1ef | 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 | } | |
aad4a87f | 253 | } |
254 | } | |
255 | else | |
256 | abort (); | |
f80e83a9 | 257 | } |
0578f103 | 258 | } |
259 | ||
338fa5f7 | 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 | |
edaf8cb5 | 262 | nonzero if comment terminated by EOF, zero otherwise. |
263 | ||
264 | Buffer->cur points to the initial asterisk of the comment. */ | |
a54e0bf8 | 265 | bool |
f7fdd7a1 | 266 | _cpp_skip_block_comment (cpp_reader *pfile) |
0578f103 | 267 | { |
f80e83a9 | 268 | cpp_buffer *buffer = pfile->buffer; |
a54e0bf8 | 269 | cppchar_t c; |
338fa5f7 | 270 | |
edaf8cb5 | 271 | buffer->cur++; |
a54e0bf8 | 272 | if (*buffer->cur == '/') |
273 | buffer->cur++; | |
338fa5f7 | 274 | |
a54e0bf8 | 275 | for (;;) |
276 | { | |
277 | c = *buffer->cur++; | |
f80e83a9 | 278 | |
338fa5f7 | 279 | /* People like decorating comments with '*', so check for '/' |
280 | instead for efficiency. */ | |
f80e83a9 | 281 | if (c == '/') |
0578f103 | 282 | { |
a54e0bf8 | 283 | if (buffer->cur[-2] == '*') |
338fa5f7 | 284 | break; |
f80e83a9 | 285 | |
338fa5f7 | 286 | /* Warn about potential nested comments, but not if the '/' |
3fb1e43b | 287 | comes immediately before the true comment delimiter. |
f80e83a9 | 288 | Don't bother to get it right across escaped newlines. */ |
338fa5f7 | 289 | if (CPP_OPTION (pfile, warn_comments) |
1e0ef2fd | 290 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
73328dce | 291 | cpp_error_with_line (pfile, DL_WARNING, |
292 | pfile->line, CPP_BUF_COL (buffer), | |
293 | "\"/*\" within comment"); | |
0578f103 | 294 | } |
a54e0bf8 | 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 | } | |
0578f103 | 304 | } |
f80e83a9 | 305 | |
3078f2b2 | 306 | _cpp_process_line_notes (pfile, true); |
a54e0bf8 | 307 | return false; |
0578f103 | 308 | } |
309 | ||
1c124f85 | 310 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
d10cfa8d | 311 | terminating newline. Handles escaped newlines. Returns nonzero |
1c124f85 | 312 | if a multiline comment. */ |
f80e83a9 | 313 | static int |
f7fdd7a1 | 314 | skip_line_comment (cpp_reader *pfile) |
0578f103 | 315 | { |
f669338a | 316 | cpp_buffer *buffer = pfile->buffer; |
1ea7ed21 | 317 | unsigned int orig_line = pfile->line; |
f80e83a9 | 318 | |
a54e0bf8 | 319 | while (*buffer->cur != '\n') |
320 | buffer->cur++; | |
1c124f85 | 321 | |
a54e0bf8 | 322 | _cpp_process_line_notes (pfile, true); |
1ea7ed21 | 323 | return orig_line != pfile->line; |
f80e83a9 | 324 | } |
0578f103 | 325 | |
a54e0bf8 | 326 | /* Skips whitespace, saving the next non-whitespace character. */ |
b86584f6 | 327 | static void |
f7fdd7a1 | 328 | skip_whitespace (cpp_reader *pfile, cppchar_t c) |
f80e83a9 | 329 | { |
330 | cpp_buffer *buffer = pfile->buffer; | |
fe9eb18b | 331 | bool saw_NUL = false; |
0578f103 | 332 | |
338fa5f7 | 333 | do |
f80e83a9 | 334 | { |
78719282 | 335 | /* Horizontal space always OK. */ |
a54e0bf8 | 336 | if (c == ' ' || c == '\t') |
338fa5f7 | 337 | ; |
338fa5f7 | 338 | /* Just \f \v or \0 left. */ |
78719282 | 339 | else if (c == '\0') |
fe9eb18b | 340 | saw_NUL = true; |
79bd622b | 341 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
73328dce | 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"); | |
338fa5f7 | 346 | |
338fa5f7 | 347 | c = *buffer->cur++; |
0578f103 | 348 | } |
2c0e001b | 349 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
338fa5f7 | 350 | while (is_nvspace (c)); |
351 | ||
fe9eb18b | 352 | if (saw_NUL) |
353 | cpp_error (pfile, DL_WARNING, "null character(s) ignored"); | |
354 | ||
1c124f85 | 355 | buffer->cur--; |
f80e83a9 | 356 | } |
0578f103 | 357 | |
79bd622b | 358 | /* See if the characters of a number token are valid in a name (no |
359 | '.', '+' or '-'). */ | |
360 | static int | |
f7fdd7a1 | 361 | name_p (cpp_reader *pfile, const cpp_string *string) |
79bd622b | 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 | ||
b1a9ff83 | 369 | return 1; |
79bd622b | 370 | } |
371 | ||
5bb46c08 | 372 | /* Returns TRUE if the sequence starting at buffer->cur is invalid in |
2cbf1359 | 373 | an identifier. FIRST is TRUE if this starts an identifier. */ |
5bb46c08 | 374 | static bool |
f7fdd7a1 | 375 | forms_identifier_p (cpp_reader *pfile, int first) |
5bb46c08 | 376 | { |
2cbf1359 | 377 | cpp_buffer *buffer = pfile->buffer; |
378 | ||
379 | if (*buffer->cur == '$') | |
380 | { | |
381 | if (!CPP_OPTION (pfile, dollars_in_ident)) | |
382 | return false; | |
383 | ||
384 | buffer->cur++; | |
f0c2775b | 385 | if (CPP_OPTION (pfile, warn_dollars) && !pfile->state.skipping) |
2cbf1359 | 386 | { |
f0c2775b | 387 | CPP_OPTION (pfile, warn_dollars) = 0; |
2cbf1359 | 388 | cpp_error (pfile, DL_PEDWARN, "'$' in identifier or number"); |
389 | } | |
390 | ||
391 | return true; | |
392 | } | |
5bb46c08 | 393 | |
2cbf1359 | 394 | /* Is this a syntactically valid UCN? */ |
395 | if (0 && *buffer->cur == '\\' | |
396 | && (buffer->cur[1] == 'u' || buffer->cur[1] == 'U')) | |
5bb46c08 | 397 | { |
2cbf1359 | 398 | buffer->cur += 2; |
ebc03810 | 399 | if (_cpp_valid_ucn (pfile, &buffer->cur, buffer->rlimit, 1 + !first)) |
2cbf1359 | 400 | return true; |
401 | buffer->cur -= 2; | |
5bb46c08 | 402 | } |
5bb46c08 | 403 | |
2cbf1359 | 404 | return false; |
5bb46c08 | 405 | } |
406 | ||
407 | /* Lex an identifier starting at BUFFER->CUR - 1. */ | |
338fa5f7 | 408 | static cpp_hashnode * |
f7fdd7a1 | 409 | lex_identifier (cpp_reader *pfile, const uchar *base) |
0578f103 | 410 | { |
79bd622b | 411 | cpp_hashnode *result; |
2cbf1359 | 412 | const uchar *cur; |
66a5287e | 413 | |
5bb46c08 | 414 | do |
78a11351 | 415 | { |
5bb46c08 | 416 | cur = pfile->buffer->cur; |
417 | ||
418 | /* N.B. ISIDNUM does not include $. */ | |
419 | while (ISIDNUM (*cur)) | |
420 | cur++; | |
78a11351 | 421 | |
78a11351 | 422 | pfile->buffer->cur = cur; |
66a5287e | 423 | } |
2cbf1359 | 424 | while (forms_identifier_p (pfile, false)); |
5bb46c08 | 425 | |
426 | result = (cpp_hashnode *) | |
427 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
66a5287e | 428 | |
5bb46c08 | 429 | /* Rarely, identifiers require diagnostics when lexed. */ |
66a5287e | 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) | |
73328dce | 435 | cpp_error (pfile, DL_ERROR, "attempt to use poisoned \"%s\"", |
66a5287e | 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) | |
73328dce | 442 | cpp_error (pfile, DL_PEDWARN, |
f7fdd7a1 | 443 | "__VA_ARGS__ can only appear in the expansion" |
444 | " of a C99 variadic macro"); | |
66a5287e | 445 | } |
446 | ||
447 | return result; | |
448 | } | |
449 | ||
5bb46c08 | 450 | /* Lex a number to NUMBER starting at BUFFER->CUR - 1. */ |
0578f103 | 451 | static void |
f7fdd7a1 | 452 | lex_number (cpp_reader *pfile, cpp_string *number) |
0578f103 | 453 | { |
b6d18b0a | 454 | const uchar *cur; |
5bb46c08 | 455 | const uchar *base; |
456 | uchar *dest; | |
0578f103 | 457 | |
5bb46c08 | 458 | base = pfile->buffer->cur - 1; |
459 | do | |
f80e83a9 | 460 | { |
5bb46c08 | 461 | cur = pfile->buffer->cur; |
338fa5f7 | 462 | |
5bb46c08 | 463 | /* N.B. ISIDNUM does not include $. */ |
464 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
465 | cur++; | |
0578f103 | 466 | |
78a11351 | 467 | pfile->buffer->cur = cur; |
0578f103 | 468 | } |
2cbf1359 | 469 | while (forms_identifier_p (pfile, false)); |
79bd622b | 470 | |
5bb46c08 | 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; | |
79bd622b | 476 | } |
477 | ||
4970d4c2 | 478 | /* Create a token of type TYPE with a literal spelling. */ |
479 | static void | |
f7fdd7a1 | 480 | create_literal (cpp_reader *pfile, cpp_token *token, const uchar *base, |
481 | unsigned int len, enum cpp_ttype type) | |
4970d4c2 | 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 | ||
5bb46c08 | 492 | /* Lexes a string, character constant, or angle-bracketed header file |
4970d4c2 | 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. */ | |
f80e83a9 | 499 | static void |
f7fdd7a1 | 500 | lex_string (cpp_reader *pfile, cpp_token *token, const uchar *base) |
0578f103 | 501 | { |
4970d4c2 | 502 | bool saw_NUL = false; |
503 | const uchar *cur; | |
5bb46c08 | 504 | cppchar_t terminator; |
4970d4c2 | 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; | |
79bd622b | 517 | |
338fa5f7 | 518 | for (;;) |
0578f103 | 519 | { |
4970d4c2 | 520 | cppchar_t c = *cur++; |
4b0c16ee | 521 | |
edaf8cb5 | 522 | /* In #include-style directives, terminators are not escapable. */ |
4970d4c2 | 523 | if (c == '\\' && !pfile->state.angled_headers && *cur != '\n') |
524 | cur++; | |
525 | else if (c == terminator) | |
5bb46c08 | 526 | break; |
4970d4c2 | 527 | else if (c == '\n') |
338fa5f7 | 528 | { |
4970d4c2 | 529 | cur--; |
530 | type = CPP_OTHER; | |
531 | break; | |
0578f103 | 532 | } |
4970d4c2 | 533 | else if (c == '\0') |
534 | saw_NUL = true; | |
0578f103 | 535 | } |
536 | ||
4970d4c2 | 537 | if (saw_NUL && !pfile->state.skipping) |
538 | cpp_error (pfile, DL_WARNING, "null character(s) preserved in literal"); | |
0578f103 | 539 | |
4970d4c2 | 540 | pfile->buffer->cur = cur; |
541 | create_literal (pfile, token, base, cur - base, type); | |
338fa5f7 | 542 | } |
f80e83a9 | 543 | |
79bd622b | 544 | /* The stored comment includes the comment start and any terminator. */ |
2c63d6c8 | 545 | static void |
f7fdd7a1 | 546 | save_comment (cpp_reader *pfile, cpp_token *token, const unsigned char *from, |
547 | cppchar_t type) | |
2c63d6c8 | 548 | { |
f80e83a9 | 549 | unsigned char *buffer; |
d3f7919d | 550 | unsigned int len, clen; |
b1a9ff83 | 551 | |
f0495c2c | 552 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
1c124f85 | 553 | |
a543b315 | 554 | /* C++ comments probably (not definitely) have moved past a new |
555 | line, which we don't want to save in the comment. */ | |
1c124f85 | 556 | if (is_vspace (pfile->buffer->cur[-1])) |
a543b315 | 557 | len--; |
d3f7919d | 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); | |
b1a9ff83 | 568 | |
f80e83a9 | 569 | token->type = CPP_COMMENT; |
d3f7919d | 570 | token->val.str.len = clen; |
338fa5f7 | 571 | token->val.str.text = buffer; |
0578f103 | 572 | |
f0495c2c | 573 | buffer[0] = '/'; |
574 | memcpy (buffer + 1, from, len - 1); | |
d3f7919d | 575 | |
a113df96 | 576 | /* Finish conversion to a C comment, if necessary. */ |
d3f7919d | 577 | if (pfile->state.in_directive && type == '/') |
578 | { | |
579 | buffer[1] = '*'; | |
580 | buffer[clen - 2] = '*'; | |
581 | buffer[clen - 1] = '/'; | |
582 | } | |
338fa5f7 | 583 | } |
0578f103 | 584 | |
83dcbb5c | 585 | /* Allocate COUNT tokens for RUN. */ |
586 | void | |
f7fdd7a1 | 587 | _cpp_init_tokenrun (tokenrun *run, unsigned int count) |
83dcbb5c | 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 * | |
f7fdd7a1 | 596 | next_tokenrun (tokenrun *run) |
83dcbb5c | 597 | { |
598 | if (run->next == NULL) | |
599 | { | |
600 | run->next = xnew (tokenrun); | |
fb5ab82c | 601 | run->next->prev = run; |
83dcbb5c | 602 | _cpp_init_tokenrun (run->next, 250); |
603 | } | |
604 | ||
605 | return run->next; | |
606 | } | |
607 | ||
f9b5f742 | 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 * | |
f7fdd7a1 | 613 | _cpp_temp_token (cpp_reader *pfile) |
f9b5f742 | 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 | ||
10b4496a | 630 | /* Lex a token into RESULT (external interface). Takes care of issues |
631 | like directive handling, token lookahead, multiple include | |
3fb1e43b | 632 | optimization and skipping. */ |
c00e481c | 633 | const cpp_token * |
f7fdd7a1 | 634 | _cpp_lex_token (cpp_reader *pfile) |
83dcbb5c | 635 | { |
fb5ab82c | 636 | cpp_token *result; |
83dcbb5c | 637 | |
fb5ab82c | 638 | for (;;) |
83dcbb5c | 639 | { |
fb5ab82c | 640 | if (pfile->cur_token == pfile->cur_run->limit) |
83dcbb5c | 641 | { |
fb5ab82c | 642 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
643 | pfile->cur_token = pfile->cur_run->base; | |
83dcbb5c | 644 | } |
645 | ||
fb5ab82c | 646 | if (pfile->lookaheads) |
10b4496a | 647 | { |
648 | pfile->lookaheads--; | |
649 | result = pfile->cur_token++; | |
650 | } | |
fb5ab82c | 651 | else |
10b4496a | 652 | result = _cpp_lex_direct (pfile); |
fb5ab82c | 653 | |
654 | if (result->flags & BOL) | |
83dcbb5c | 655 | { |
fb5ab82c | 656 | /* Is this a directive. If _cpp_handle_directive returns |
657 | false, it is an assembler #. */ | |
658 | if (result->type == CPP_HASH | |
d6af0368 | 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 | |
fb5ab82c | 663 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
664 | continue; | |
5621a364 | 665 | if (pfile->cb.line_change && !pfile->state.skipping) |
f7fdd7a1 | 666 | pfile->cb.line_change (pfile, result, pfile->state.parsing_args); |
83dcbb5c | 667 | } |
83dcbb5c | 668 | |
fb5ab82c | 669 | /* We don't skip tokens in directives. */ |
670 | if (pfile->state.in_directive) | |
671 | break; | |
83dcbb5c | 672 | |
fb5ab82c | 673 | /* Outside a directive, invalidate controlling macros. At file |
10b4496a | 674 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
7ef5b942 | 675 | get here and MI optimization works. */ |
83dcbb5c | 676 | pfile->mi_valid = false; |
fb5ab82c | 677 | |
678 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
679 | break; | |
83dcbb5c | 680 | } |
681 | ||
c00e481c | 682 | return result; |
83dcbb5c | 683 | } |
684 | ||
a54e0bf8 | 685 | /* Returns true if a fresh line has been loaded. */ |
686 | bool | |
f7fdd7a1 | 687 | _cpp_get_fresh_line (cpp_reader *pfile) |
0bb65704 | 688 | { |
a54e0bf8 | 689 | /* We can't get a new line until we leave the current directive. */ |
690 | if (pfile->state.in_directive) | |
691 | return false; | |
b1a9ff83 | 692 | |
a54e0bf8 | 693 | for (;;) |
fb83e0d6 | 694 | { |
a54e0bf8 | 695 | cpp_buffer *buffer = pfile->buffer; |
fb83e0d6 | 696 | |
a54e0bf8 | 697 | if (!buffer->need_line) |
698 | return true; | |
699 | ||
700 | if (buffer->next_line < buffer->rlimit) | |
0bb65704 | 701 | { |
a54e0bf8 | 702 | _cpp_clean_line (pfile); |
703 | return true; | |
704 | } | |
0bb65704 | 705 | |
a54e0bf8 | 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 | ||
a54e0bf8 | 722 | _cpp_pop_buffer (pfile); |
11b5269c | 723 | if (pfile->buffer == NULL) |
724 | return false; | |
a54e0bf8 | 725 | } |
0bb65704 | 726 | } |
727 | ||
edaf8cb5 | 728 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
729 | do \ | |
730 | { \ | |
731 | result->type = ELSE_TYPE; \ | |
732 | if (*buffer->cur == CHAR) \ | |
733 | buffer->cur++, result->type = THEN_TYPE; \ | |
734 | } \ | |
735 | while (0) | |
1c124f85 | 736 | |
10b4496a | 737 | /* Lex a token into pfile->cur_token, which is also incremented, to |
738 | get diagnostics pointing to the correct location. | |
739 | ||
740 | Does not handle issues such as token lookahead, multiple-include | |
4172d65e | 741 | optimization, directives, skipping etc. This function is only |
10b4496a | 742 | suitable for use by _cpp_lex_token, and in special cases like |
743 | lex_expansion_token which doesn't care for any of these issues. | |
744 | ||
745 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
746 | otherwise returns to the start of the token buffer if permissible. | |
747 | Returns the location of the lexed token. */ | |
748 | cpp_token * | |
f7fdd7a1 | 749 | _cpp_lex_direct (cpp_reader *pfile) |
0578f103 | 750 | { |
338fa5f7 | 751 | cppchar_t c; |
230f0943 | 752 | cpp_buffer *buffer; |
338fa5f7 | 753 | const unsigned char *comment_start; |
10b4496a | 754 | cpp_token *result = pfile->cur_token++; |
0653b94e | 755 | |
83dcbb5c | 756 | fresh_line: |
a54e0bf8 | 757 | result->flags = 0; |
82166c5c | 758 | buffer = pfile->buffer; |
11b5269c | 759 | if (buffer->need_line) |
a54e0bf8 | 760 | { |
761 | if (!_cpp_get_fresh_line (pfile)) | |
762 | { | |
763 | result->type = CPP_EOF; | |
2908f819 | 764 | if (!pfile->state.in_directive) |
765 | { | |
766 | /* Tell the compiler the line number of the EOF token. */ | |
767 | result->line = pfile->line; | |
768 | result->flags = BOL; | |
769 | } | |
a54e0bf8 | 770 | return result; |
771 | } | |
772 | if (!pfile->keep_tokens) | |
773 | { | |
774 | pfile->cur_run = &pfile->base_run; | |
775 | result = pfile->base_run.base; | |
776 | pfile->cur_token = result + 1; | |
777 | } | |
778 | result->flags = BOL; | |
779 | if (pfile->state.parsing_args == 2) | |
780 | result->flags |= PREV_WHITE; | |
781 | } | |
11b5269c | 782 | buffer = pfile->buffer; |
83dcbb5c | 783 | update_tokens_line: |
36a0aa7c | 784 | result->line = pfile->line; |
f80e83a9 | 785 | |
83dcbb5c | 786 | skipped_white: |
a54e0bf8 | 787 | if (buffer->cur >= buffer->notes[buffer->cur_note].pos |
788 | && !pfile->overlaid_buffer) | |
789 | { | |
790 | _cpp_process_line_notes (pfile, false); | |
791 | result->line = pfile->line; | |
792 | } | |
1c124f85 | 793 | c = *buffer->cur++; |
83dcbb5c | 794 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
83dcbb5c | 795 | |
338fa5f7 | 796 | switch (c) |
0578f103 | 797 | { |
435fb09b | 798 | case ' ': case '\t': case '\f': case '\v': case '\0': |
799 | result->flags |= PREV_WHITE; | |
a54e0bf8 | 800 | skip_whitespace (pfile, c); |
801 | goto skipped_white; | |
338fa5f7 | 802 | |
a54e0bf8 | 803 | case '\n': |
804 | pfile->line++; | |
805 | buffer->need_line = true; | |
806 | goto fresh_line; | |
732cb4c9 | 807 | |
338fa5f7 | 808 | case '0': case '1': case '2': case '3': case '4': |
809 | case '5': case '6': case '7': case '8': case '9': | |
810 | result->type = CPP_NUMBER; | |
5bb46c08 | 811 | lex_number (pfile, &result->val.str); |
338fa5f7 | 812 | break; |
732cb4c9 | 813 | |
78c551ad | 814 | case 'L': |
815 | /* 'L' may introduce wide characters or strings. */ | |
5bb46c08 | 816 | if (*buffer->cur == '\'' || *buffer->cur == '"') |
817 | { | |
4970d4c2 | 818 | lex_string (pfile, result, buffer->cur - 1); |
5bb46c08 | 819 | break; |
820 | } | |
b1a9ff83 | 821 | /* Fall through. */ |
78c551ad | 822 | |
338fa5f7 | 823 | case '_': |
824 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
825 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
826 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
827 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
828 | case 'y': case 'z': | |
829 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
78c551ad | 830 | case 'G': case 'H': case 'I': case 'J': case 'K': |
338fa5f7 | 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 | result->type = CPP_NAME; | |
2cbf1359 | 835 | result->val.node = lex_identifier (pfile, buffer->cur - 1); |
338fa5f7 | 836 | |
338fa5f7 | 837 | /* Convert named operators to their proper types. */ |
78c551ad | 838 | if (result->val.node->flags & NODE_OPERATOR) |
338fa5f7 | 839 | { |
840 | result->flags |= NAMED_OP; | |
805e22b2 | 841 | result->type = result->val.node->directive_index; |
338fa5f7 | 842 | } |
843 | break; | |
844 | ||
845 | case '\'': | |
846 | case '"': | |
4970d4c2 | 847 | lex_string (pfile, result, buffer->cur - 1); |
338fa5f7 | 848 | break; |
f80e83a9 | 849 | |
338fa5f7 | 850 | case '/': |
f0495c2c | 851 | /* A potential block or line comment. */ |
852 | comment_start = buffer->cur; | |
edaf8cb5 | 853 | c = *buffer->cur; |
854 | ||
f0495c2c | 855 | if (c == '*') |
856 | { | |
a54e0bf8 | 857 | if (_cpp_skip_block_comment (pfile)) |
73328dce | 858 | cpp_error (pfile, DL_ERROR, "unterminated comment"); |
338fa5f7 | 859 | } |
1c124f85 | 860 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
861 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
338fa5f7 | 862 | { |
5db5d057 | 863 | /* Warn about comments only if pedantically GNUC89, and not |
864 | in system headers. */ | |
865 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
66914e49 | 866 | && ! buffer->warned_cplusplus_comments) |
f80e83a9 | 867 | { |
73328dce | 868 | cpp_error (pfile, DL_PEDWARN, |
ba059ac0 | 869 | "C++ style comments are not allowed in ISO C90"); |
73328dce | 870 | cpp_error (pfile, DL_PEDWARN, |
871 | "(this will be reported only once per input file)"); | |
f0495c2c | 872 | buffer->warned_cplusplus_comments = 1; |
873 | } | |
338fa5f7 | 874 | |
e1caf668 | 875 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
73328dce | 876 | cpp_error (pfile, DL_WARNING, "multi-line comment"); |
f0495c2c | 877 | } |
1c124f85 | 878 | else if (c == '=') |
879 | { | |
edaf8cb5 | 880 | buffer->cur++; |
1c124f85 | 881 | result->type = CPP_DIV_EQ; |
882 | break; | |
883 | } | |
884 | else | |
885 | { | |
1c124f85 | 886 | result->type = CPP_DIV; |
887 | break; | |
888 | } | |
338fa5f7 | 889 | |
f0495c2c | 890 | if (!pfile->state.save_comments) |
891 | { | |
892 | result->flags |= PREV_WHITE; | |
83dcbb5c | 893 | goto update_tokens_line; |
338fa5f7 | 894 | } |
f0495c2c | 895 | |
896 | /* Save the comment as a token in its own right. */ | |
d3f7919d | 897 | save_comment (pfile, result, comment_start, c); |
fb5ab82c | 898 | break; |
338fa5f7 | 899 | |
900 | case '<': | |
901 | if (pfile->state.angled_headers) | |
902 | { | |
4970d4c2 | 903 | lex_string (pfile, result, buffer->cur - 1); |
1c124f85 | 904 | break; |
338fa5f7 | 905 | } |
0578f103 | 906 | |
edaf8cb5 | 907 | result->type = CPP_LESS; |
908 | if (*buffer->cur == '=') | |
909 | buffer->cur++, result->type = CPP_LESS_EQ; | |
910 | else if (*buffer->cur == '<') | |
338fa5f7 | 911 | { |
edaf8cb5 | 912 | buffer->cur++; |
913 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); | |
338fa5f7 | 914 | } |
edaf8cb5 | 915 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) |
338fa5f7 | 916 | { |
edaf8cb5 | 917 | buffer->cur++; |
918 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); | |
338fa5f7 | 919 | } |
edaf8cb5 | 920 | else if (CPP_OPTION (pfile, digraphs)) |
1c124f85 | 921 | { |
edaf8cb5 | 922 | if (*buffer->cur == ':') |
923 | { | |
924 | buffer->cur++; | |
925 | result->flags |= DIGRAPH; | |
926 | result->type = CPP_OPEN_SQUARE; | |
927 | } | |
928 | else if (*buffer->cur == '%') | |
929 | { | |
930 | buffer->cur++; | |
931 | result->flags |= DIGRAPH; | |
932 | result->type = CPP_OPEN_BRACE; | |
933 | } | |
1c124f85 | 934 | } |
338fa5f7 | 935 | break; |
936 | ||
937 | case '>': | |
edaf8cb5 | 938 | result->type = CPP_GREATER; |
939 | if (*buffer->cur == '=') | |
940 | buffer->cur++, result->type = CPP_GREATER_EQ; | |
941 | else if (*buffer->cur == '>') | |
338fa5f7 | 942 | { |
edaf8cb5 | 943 | buffer->cur++; |
944 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); | |
945 | } | |
946 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) | |
947 | { | |
948 | buffer->cur++; | |
949 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); | |
338fa5f7 | 950 | } |
951 | break; | |
952 | ||
f669338a | 953 | case '%': |
edaf8cb5 | 954 | result->type = CPP_MOD; |
955 | if (*buffer->cur == '=') | |
956 | buffer->cur++, result->type = CPP_MOD_EQ; | |
957 | else if (CPP_OPTION (pfile, digraphs)) | |
1c124f85 | 958 | { |
edaf8cb5 | 959 | if (*buffer->cur == ':') |
1c124f85 | 960 | { |
edaf8cb5 | 961 | buffer->cur++; |
962 | result->flags |= DIGRAPH; | |
963 | result->type = CPP_HASH; | |
964 | if (*buffer->cur == '%' && buffer->cur[1] == ':') | |
965 | buffer->cur += 2, result->type = CPP_PASTE; | |
966 | } | |
967 | else if (*buffer->cur == '>') | |
968 | { | |
969 | buffer->cur++; | |
970 | result->flags |= DIGRAPH; | |
971 | result->type = CPP_CLOSE_BRACE; | |
1c124f85 | 972 | } |
1c124f85 | 973 | } |
338fa5f7 | 974 | break; |
975 | ||
f669338a | 976 | case '.': |
1c124f85 | 977 | result->type = CPP_DOT; |
edaf8cb5 | 978 | if (ISDIGIT (*buffer->cur)) |
1c124f85 | 979 | { |
980 | result->type = CPP_NUMBER; | |
5bb46c08 | 981 | lex_number (pfile, &result->val.str); |
1c124f85 | 982 | } |
edaf8cb5 | 983 | else if (*buffer->cur == '.' && buffer->cur[1] == '.') |
984 | buffer->cur += 2, result->type = CPP_ELLIPSIS; | |
985 | else if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) | |
986 | buffer->cur++, result->type = CPP_DOT_STAR; | |
338fa5f7 | 987 | break; |
0578f103 | 988 | |
338fa5f7 | 989 | case '+': |
edaf8cb5 | 990 | result->type = CPP_PLUS; |
991 | if (*buffer->cur == '+') | |
992 | buffer->cur++, result->type = CPP_PLUS_PLUS; | |
993 | else if (*buffer->cur == '=') | |
994 | buffer->cur++, result->type = CPP_PLUS_EQ; | |
338fa5f7 | 995 | break; |
ac0749c7 | 996 | |
338fa5f7 | 997 | case '-': |
edaf8cb5 | 998 | result->type = CPP_MINUS; |
999 | if (*buffer->cur == '>') | |
338fa5f7 | 1000 | { |
edaf8cb5 | 1001 | buffer->cur++; |
1c124f85 | 1002 | result->type = CPP_DEREF; |
edaf8cb5 | 1003 | if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) |
1004 | buffer->cur++, result->type = CPP_DEREF_STAR; | |
1c124f85 | 1005 | } |
edaf8cb5 | 1006 | else if (*buffer->cur == '-') |
1007 | buffer->cur++, result->type = CPP_MINUS_MINUS; | |
1008 | else if (*buffer->cur == '=') | |
1009 | buffer->cur++, result->type = CPP_MINUS_EQ; | |
338fa5f7 | 1010 | break; |
0578f103 | 1011 | |
338fa5f7 | 1012 | case '&': |
edaf8cb5 | 1013 | result->type = CPP_AND; |
1014 | if (*buffer->cur == '&') | |
1015 | buffer->cur++, result->type = CPP_AND_AND; | |
1016 | else if (*buffer->cur == '=') | |
1017 | buffer->cur++, result->type = CPP_AND_EQ; | |
338fa5f7 | 1018 | break; |
b1a9ff83 | 1019 | |
338fa5f7 | 1020 | case '|': |
edaf8cb5 | 1021 | result->type = CPP_OR; |
1022 | if (*buffer->cur == '|') | |
1023 | buffer->cur++, result->type = CPP_OR_OR; | |
1024 | else if (*buffer->cur == '=') | |
1025 | buffer->cur++, result->type = CPP_OR_EQ; | |
338fa5f7 | 1026 | break; |
0578f103 | 1027 | |
338fa5f7 | 1028 | case ':': |
edaf8cb5 | 1029 | result->type = CPP_COLON; |
1030 | if (*buffer->cur == ':' && CPP_OPTION (pfile, cplusplus)) | |
1031 | buffer->cur++, result->type = CPP_SCOPE; | |
1032 | else if (*buffer->cur == '>' && CPP_OPTION (pfile, digraphs)) | |
338fa5f7 | 1033 | { |
edaf8cb5 | 1034 | buffer->cur++; |
338fa5f7 | 1035 | result->flags |= DIGRAPH; |
1c124f85 | 1036 | result->type = CPP_CLOSE_SQUARE; |
1037 | } | |
338fa5f7 | 1038 | break; |
0578f103 | 1039 | |
1c124f85 | 1040 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1041 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1042 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1043 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1044 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1045 | ||
a54e0bf8 | 1046 | case '?': result->type = CPP_QUERY; break; |
338fa5f7 | 1047 | case '~': result->type = CPP_COMPL; break; |
1048 | case ',': result->type = CPP_COMMA; break; | |
1049 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1050 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1051 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1052 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1053 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1054 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1055 | case ';': result->type = CPP_SEMICOLON; break; | |
1056 | ||
7fd957fe | 1057 | /* @ is a punctuator in Objective-C. */ |
9ee99ac6 | 1058 | case '@': result->type = CPP_ATSIGN; break; |
338fa5f7 | 1059 | |
78c551ad | 1060 | case '$': |
2cbf1359 | 1061 | case '\\': |
1062 | { | |
1063 | const uchar *base = --buffer->cur; | |
78c551ad | 1064 | |
2cbf1359 | 1065 | if (forms_identifier_p (pfile, true)) |
1066 | { | |
1067 | result->type = CPP_NAME; | |
1068 | result->val.node = lex_identifier (pfile, base); | |
1069 | break; | |
1070 | } | |
1071 | buffer->cur++; | |
bc205914 | 1072 | } |
2cbf1359 | 1073 | |
bc205914 | 1074 | default: |
4970d4c2 | 1075 | create_literal (pfile, result, buffer->cur - 1, 1, CPP_OTHER); |
1076 | break; | |
338fa5f7 | 1077 | } |
fb5ab82c | 1078 | |
1079 | return result; | |
338fa5f7 | 1080 | } |
1081 | ||
b1280514 | 1082 | /* An upper bound on the number of bytes needed to spell TOKEN. |
1083 | Does not include preceding whitespace. */ | |
79bd622b | 1084 | unsigned int |
f7fdd7a1 | 1085 | cpp_token_len (const cpp_token *token) |
338fa5f7 | 1086 | { |
79bd622b | 1087 | unsigned int len; |
cfad5579 | 1088 | |
79bd622b | 1089 | switch (TOKEN_SPELL (token)) |
f80e83a9 | 1090 | { |
b1280514 | 1091 | default: len = 4; break; |
4970d4c2 | 1092 | case SPELL_LITERAL: len = token->val.str.len; break; |
c86dbc5b | 1093 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; |
f80e83a9 | 1094 | } |
b1280514 | 1095 | |
1096 | return len; | |
cfad5579 | 1097 | } |
1098 | ||
f80e83a9 | 1099 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
c5ea33a8 | 1100 | already contain the enough space to hold the token's spelling. |
f7fdd7a1 | 1101 | Returns a pointer to the character after the last character written. |
1102 | FIXME: Would be nice if we didn't need the PFILE argument. */ | |
79bd622b | 1103 | unsigned char * |
f7fdd7a1 | 1104 | cpp_spell_token (cpp_reader *pfile, const cpp_token *token, |
1105 | unsigned char *buffer) | |
f80e83a9 | 1106 | { |
7e842f95 | 1107 | switch (TOKEN_SPELL (token)) |
f80e83a9 | 1108 | { |
1109 | case SPELL_OPERATOR: | |
1110 | { | |
1111 | const unsigned char *spelling; | |
1112 | unsigned char c; | |
ab12a39c | 1113 | |
f80e83a9 | 1114 | if (token->flags & DIGRAPH) |
ee6c4e4b | 1115 | spelling |
1116 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
31674461 | 1117 | else if (token->flags & NAMED_OP) |
1118 | goto spell_ident; | |
f80e83a9 | 1119 | else |
7e842f95 | 1120 | spelling = TOKEN_NAME (token); |
b1a9ff83 | 1121 | |
f80e83a9 | 1122 | while ((c = *spelling++) != '\0') |
1123 | *buffer++ = c; | |
1124 | } | |
1125 | break; | |
ab12a39c | 1126 | |
8d27e472 | 1127 | spell_ident: |
f80e83a9 | 1128 | case SPELL_IDENT: |
c86dbc5b | 1129 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1130 | buffer += NODE_LEN (token->val.node); | |
f80e83a9 | 1131 | break; |
ab12a39c | 1132 | |
4970d4c2 | 1133 | case SPELL_LITERAL: |
8d27e472 | 1134 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1135 | buffer += token->val.str.len; | |
1136 | break; | |
1137 | ||
f80e83a9 | 1138 | case SPELL_NONE: |
73328dce | 1139 | cpp_error (pfile, DL_ICE, "unspellable token %s", TOKEN_NAME (token)); |
f80e83a9 | 1140 | break; |
1141 | } | |
ab12a39c | 1142 | |
f80e83a9 | 1143 | return buffer; |
1144 | } | |
ab12a39c | 1145 | |
e484a1cc | 1146 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1147 | freed when the reader is destroyed. Useful for diagnostics. */ | |
79bd622b | 1148 | unsigned char * |
f7fdd7a1 | 1149 | cpp_token_as_text (cpp_reader *pfile, const cpp_token *token) |
b1280514 | 1150 | { |
1151 | unsigned int len = cpp_token_len (token) + 1; | |
1fdf6039 | 1152 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
6060326b | 1153 | |
79bd622b | 1154 | end = cpp_spell_token (pfile, token, start); |
1155 | end[0] = '\0'; | |
6060326b | 1156 | |
79bd622b | 1157 | return start; |
1158 | } | |
6060326b | 1159 | |
e484a1cc | 1160 | /* Used by C front ends, which really should move to using |
1161 | cpp_token_as_text. */ | |
79bd622b | 1162 | const char * |
f7fdd7a1 | 1163 | cpp_type2name (enum cpp_ttype type) |
79bd622b | 1164 | { |
1165 | return (const char *) token_spellings[type].name; | |
1166 | } | |
6060326b | 1167 | |
f9b5f742 | 1168 | /* Writes the spelling of token to FP, without any preceding space. |
1169 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1170 | double-buffering. */ | |
79bd622b | 1171 | void |
f7fdd7a1 | 1172 | cpp_output_token (const cpp_token *token, FILE *fp) |
79bd622b | 1173 | { |
79bd622b | 1174 | switch (TOKEN_SPELL (token)) |
6060326b | 1175 | { |
79bd622b | 1176 | case SPELL_OPERATOR: |
1177 | { | |
1178 | const unsigned char *spelling; | |
28874558 | 1179 | int c; |
6060326b | 1180 | |
79bd622b | 1181 | if (token->flags & DIGRAPH) |
ee6c4e4b | 1182 | spelling |
1183 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
79bd622b | 1184 | else if (token->flags & NAMED_OP) |
1185 | goto spell_ident; | |
1186 | else | |
1187 | spelling = TOKEN_NAME (token); | |
f80e83a9 | 1188 | |
28874558 | 1189 | c = *spelling; |
1190 | do | |
1191 | putc (c, fp); | |
1192 | while ((c = *++spelling) != '\0'); | |
79bd622b | 1193 | } |
1194 | break; | |
f80e83a9 | 1195 | |
79bd622b | 1196 | spell_ident: |
1197 | case SPELL_IDENT: | |
28874558 | 1198 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
79bd622b | 1199 | break; |
f80e83a9 | 1200 | |
4970d4c2 | 1201 | case SPELL_LITERAL: |
8d27e472 | 1202 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
1203 | break; | |
1204 | ||
79bd622b | 1205 | case SPELL_NONE: |
1206 | /* An error, most probably. */ | |
1207 | break; | |
f80e83a9 | 1208 | } |
6060326b | 1209 | } |
1210 | ||
79bd622b | 1211 | /* Compare two tokens. */ |
1212 | int | |
f7fdd7a1 | 1213 | _cpp_equiv_tokens (const cpp_token *a, const cpp_token *b) |
6060326b | 1214 | { |
79bd622b | 1215 | if (a->type == b->type && a->flags == b->flags) |
1216 | switch (TOKEN_SPELL (a)) | |
1217 | { | |
1218 | default: /* Keep compiler happy. */ | |
1219 | case SPELL_OPERATOR: | |
1220 | return 1; | |
79bd622b | 1221 | case SPELL_NONE: |
588d632b | 1222 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
79bd622b | 1223 | case SPELL_IDENT: |
1224 | return a->val.node == b->val.node; | |
4970d4c2 | 1225 | case SPELL_LITERAL: |
79bd622b | 1226 | return (a->val.str.len == b->val.str.len |
1227 | && !memcmp (a->val.str.text, b->val.str.text, | |
1228 | a->val.str.len)); | |
1229 | } | |
6060326b | 1230 | |
f80e83a9 | 1231 | return 0; |
1232 | } | |
1233 | ||
79bd622b | 1234 | /* Returns nonzero if a space should be inserted to avoid an |
1235 | accidental token paste for output. For simplicity, it is | |
1236 | conservative, and occasionally advises a space where one is not | |
1237 | needed, e.g. "." and ".2". */ | |
79bd622b | 1238 | int |
f7fdd7a1 | 1239 | cpp_avoid_paste (cpp_reader *pfile, const cpp_token *token1, |
1240 | const cpp_token *token2) | |
6060326b | 1241 | { |
79bd622b | 1242 | enum cpp_ttype a = token1->type, b = token2->type; |
1243 | cppchar_t c; | |
6060326b | 1244 | |
79bd622b | 1245 | if (token1->flags & NAMED_OP) |
1246 | a = CPP_NAME; | |
1247 | if (token2->flags & NAMED_OP) | |
1248 | b = CPP_NAME; | |
6060326b | 1249 | |
79bd622b | 1250 | c = EOF; |
1251 | if (token2->flags & DIGRAPH) | |
ee6c4e4b | 1252 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
79bd622b | 1253 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1254 | c = token_spellings[b].name[0]; | |
6060326b | 1255 | |
79bd622b | 1256 | /* Quickly get everything that can paste with an '='. */ |
ee6c4e4b | 1257 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
79bd622b | 1258 | return 1; |
6060326b | 1259 | |
79bd622b | 1260 | switch (a) |
6060326b | 1261 | { |
79bd622b | 1262 | case CPP_GREATER: return c == '>' || c == '?'; |
1263 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1264 | case CPP_PLUS: return c == '+'; | |
1265 | case CPP_MINUS: return c == '-' || c == '>'; | |
1266 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1267 | case CPP_MOD: return c == ':' || c == '>'; | |
1268 | case CPP_AND: return c == '&'; | |
1269 | case CPP_OR: return c == '|'; | |
1270 | case CPP_COLON: return c == ':' || c == '>'; | |
1271 | case CPP_DEREF: return c == '*'; | |
efdcc728 | 1272 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
79bd622b | 1273 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1274 | case CPP_NAME: return ((b == CPP_NUMBER | |
1275 | && name_p (pfile, &token2->val.str)) | |
1276 | || b == CPP_NAME | |
1277 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1278 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1279 | || c == '.' || c == '+' || c == '-'); | |
2cbf1359 | 1280 | /* UCNs */ |
bc205914 | 1281 | case CPP_OTHER: return ((token1->val.str.text[0] == '\\' |
1282 | && b == CPP_NAME) | |
2cbf1359 | 1283 | || (CPP_OPTION (pfile, objc) |
bc205914 | 1284 | && token1->val.str.text[0] == '@' |
2cbf1359 | 1285 | && (b == CPP_NAME || b == CPP_STRING))); |
79bd622b | 1286 | default: break; |
6060326b | 1287 | } |
6060326b | 1288 | |
deb356cf | 1289 | return 0; |
6060326b | 1290 | } |
1291 | ||
79bd622b | 1292 | /* Output all the remaining tokens on the current line, and a newline |
f9b5f742 | 1293 | character, to FP. Leading whitespace is removed. If there are |
1294 | macros, special token padding is not performed. */ | |
6060326b | 1295 | void |
f7fdd7a1 | 1296 | cpp_output_line (cpp_reader *pfile, FILE *fp) |
6060326b | 1297 | { |
f9b5f742 | 1298 | const cpp_token *token; |
7e842f95 | 1299 | |
f9b5f742 | 1300 | token = cpp_get_token (pfile); |
1301 | while (token->type != CPP_EOF) | |
7e842f95 | 1302 | { |
f9b5f742 | 1303 | cpp_output_token (token, fp); |
1304 | token = cpp_get_token (pfile); | |
1305 | if (token->flags & PREV_WHITE) | |
1306 | putc (' ', fp); | |
7e842f95 | 1307 | } |
1308 | ||
79bd622b | 1309 | putc ('\n', fp); |
f80e83a9 | 1310 | } |
6060326b | 1311 | |
084163dc | 1312 | /* Memory buffers. Changing these three constants can have a dramatic |
1313 | effect on performance. The values here are reasonable defaults, | |
1314 | but might be tuned. If you adjust them, be sure to test across a | |
1315 | range of uses of cpplib, including heavy nested function-like macro | |
1316 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1317 | good tool for this). */ | |
1318 | #define MIN_BUFF_SIZE 8000 | |
1e0ef2fd | 1319 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
084163dc | 1320 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1321 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
deb356cf | 1322 | |
1e0ef2fd | 1323 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1324 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1325 | #endif | |
1326 | ||
1785b647 | 1327 | /* Create a new allocation buffer. Place the control block at the end |
1328 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
06c92cbc | 1329 | static _cpp_buff * |
f7fdd7a1 | 1330 | new_buff (size_t len) |
06c92cbc | 1331 | { |
1332 | _cpp_buff *result; | |
1fdf6039 | 1333 | unsigned char *base; |
06c92cbc | 1334 | |
084163dc | 1335 | if (len < MIN_BUFF_SIZE) |
1336 | len = MIN_BUFF_SIZE; | |
198b48a0 | 1337 | len = CPP_ALIGN (len); |
06c92cbc | 1338 | |
1339 | base = xmalloc (len + sizeof (_cpp_buff)); | |
1340 | result = (_cpp_buff *) (base + len); | |
1341 | result->base = base; | |
1342 | result->cur = base; | |
1343 | result->limit = base + len; | |
1344 | result->next = NULL; | |
1345 | return result; | |
1346 | } | |
1347 | ||
1348 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1349 | void | |
f7fdd7a1 | 1350 | _cpp_release_buff (cpp_reader *pfile, _cpp_buff *buff) |
06c92cbc | 1351 | { |
1352 | _cpp_buff *end = buff; | |
1353 | ||
1354 | while (end->next) | |
1355 | end = end->next; | |
1356 | end->next = pfile->free_buffs; | |
1357 | pfile->free_buffs = buff; | |
1358 | } | |
1359 | ||
1360 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1361 | _cpp_buff * | |
f7fdd7a1 | 1362 | _cpp_get_buff (cpp_reader *pfile, size_t min_size) |
06c92cbc | 1363 | { |
1364 | _cpp_buff *result, **p; | |
1365 | ||
1366 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1367 | { | |
4b31a107 | 1368 | size_t size; |
084163dc | 1369 | |
1370 | if (*p == NULL) | |
06c92cbc | 1371 | return new_buff (min_size); |
084163dc | 1372 | result = *p; |
1373 | size = result->limit - result->base; | |
1374 | /* Return a buffer that's big enough, but don't waste one that's | |
1375 | way too big. */ | |
4085c149 | 1376 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
06c92cbc | 1377 | break; |
1378 | } | |
1379 | ||
1380 | *p = result->next; | |
1381 | result->next = NULL; | |
1382 | result->cur = result->base; | |
1383 | return result; | |
1384 | } | |
1385 | ||
20dd417a | 1386 | /* Creates a new buffer with enough space to hold the uncommitted |
e6a5f963 | 1387 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
1388 | the excess bytes to the new buffer. Chains the new buffer after | |
1389 | BUFF, and returns the new buffer. */ | |
06c92cbc | 1390 | _cpp_buff * |
f7fdd7a1 | 1391 | _cpp_append_extend_buff (cpp_reader *pfile, _cpp_buff *buff, size_t min_extra) |
06c92cbc | 1392 | { |
4b31a107 | 1393 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
e6a5f963 | 1394 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
06c92cbc | 1395 | |
e6a5f963 | 1396 | buff->next = new_buff; |
1397 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
1398 | return new_buff; | |
1399 | } | |
1400 | ||
20dd417a | 1401 | /* Creates a new buffer with enough space to hold the uncommitted |
e6a5f963 | 1402 | remaining bytes of the buffer pointed to by BUFF, and at least |
1403 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
1404 | Chains the new buffer before the buffer pointed to by BUFF, and | |
1405 | updates the pointer to point to the new buffer. */ | |
1406 | void | |
f7fdd7a1 | 1407 | _cpp_extend_buff (cpp_reader *pfile, _cpp_buff **pbuff, size_t min_extra) |
e6a5f963 | 1408 | { |
1409 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
1410 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
1411 | ||
1412 | new_buff = _cpp_get_buff (pfile, size); | |
1413 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
1414 | new_buff->next = old_buff; | |
1415 | *pbuff = new_buff; | |
06c92cbc | 1416 | } |
1417 | ||
1418 | /* Free a chain of buffers starting at BUFF. */ | |
1419 | void | |
f82b06e0 | 1420 | _cpp_free_buff (_cpp_buff *buff) |
06c92cbc | 1421 | { |
1422 | _cpp_buff *next; | |
1423 | ||
1424 | for (; buff; buff = next) | |
1425 | { | |
1426 | next = buff->next; | |
1427 | free (buff->base); | |
1428 | } | |
1429 | } | |
deb356cf | 1430 | |
1fdf6039 | 1431 | /* Allocate permanent, unaligned storage of length LEN. */ |
1432 | unsigned char * | |
f7fdd7a1 | 1433 | _cpp_unaligned_alloc (cpp_reader *pfile, size_t len) |
1fdf6039 | 1434 | { |
1435 | _cpp_buff *buff = pfile->u_buff; | |
1436 | unsigned char *result = buff->cur; | |
1437 | ||
1438 | if (len > (size_t) (buff->limit - result)) | |
1439 | { | |
1440 | buff = _cpp_get_buff (pfile, len); | |
1441 | buff->next = pfile->u_buff; | |
1442 | pfile->u_buff = buff; | |
1443 | result = buff->cur; | |
1444 | } | |
1445 | ||
1446 | buff->cur = result + len; | |
1447 | return result; | |
1448 | } | |
1449 | ||
1e0ef2fd | 1450 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
1451 | That buffer is used for growing allocations when saving macro | |
1452 | replacement lists in a #define, and when parsing an answer to an | |
1453 | assertion in #assert, #unassert or #if (and therefore possibly | |
1454 | whilst expanding macros). It therefore must not be used by any | |
1455 | code that they might call: specifically the lexer and the guts of | |
1456 | the macro expander. | |
1457 | ||
1458 | All existing other uses clearly fit this restriction: storing | |
1459 | registered pragmas during initialization. */ | |
79bd622b | 1460 | unsigned char * |
f7fdd7a1 | 1461 | _cpp_aligned_alloc (cpp_reader *pfile, size_t len) |
89b05ef6 | 1462 | { |
e6a5f963 | 1463 | _cpp_buff *buff = pfile->a_buff; |
1464 | unsigned char *result = buff->cur; | |
89b05ef6 | 1465 | |
e6a5f963 | 1466 | if (len > (size_t) (buff->limit - result)) |
89b05ef6 | 1467 | { |
e6a5f963 | 1468 | buff = _cpp_get_buff (pfile, len); |
1469 | buff->next = pfile->a_buff; | |
1470 | pfile->a_buff = buff; | |
1471 | result = buff->cur; | |
89b05ef6 | 1472 | } |
f80e83a9 | 1473 | |
e6a5f963 | 1474 | buff->cur = result + len; |
79bd622b | 1475 | return result; |
f80e83a9 | 1476 | } |