]>
Commit | Line | Data |
---|---|---|
45b966db | 1 | /* CPP Library - lexical analysis. |
cbada204 | 2 | Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005 Free Software Foundation, Inc. |
45b966db ZW |
3 | Contributed by Per Bothner, 1994-95. |
4 | Based on CCCP program by Paul Rubin, June 1986 | |
5 | Adapted to ANSI C, Richard Stallman, Jan 1987 | |
6 | Broken out to separate file, Zack Weinberg, Mar 2000 | |
7 | ||
8 | This program is free software; you can redistribute it and/or modify it | |
9 | under the terms of the GNU General Public License as published by the | |
10 | Free Software Foundation; either version 2, or (at your option) any | |
11 | later version. | |
12 | ||
13 | This program is distributed in the hope that it will be useful, | |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
17 | ||
18 | You should have received a copy of the GNU General Public License | |
19 | along with this program; if not, write to the Free Software | |
20 | Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ | |
21 | ||
22 | #include "config.h" | |
23 | #include "system.h" | |
45b966db | 24 | #include "cpplib.h" |
4f4e53dd | 25 | #include "internal.h" |
45b966db | 26 | |
93c80368 | 27 | enum spell_type |
f9a0e96c | 28 | { |
93c80368 | 29 | SPELL_OPERATOR = 0, |
93c80368 | 30 | SPELL_IDENT, |
6338b358 | 31 | SPELL_LITERAL, |
93c80368 | 32 | SPELL_NONE |
f9a0e96c ZW |
33 | }; |
34 | ||
93c80368 | 35 | struct token_spelling |
f9a0e96c | 36 | { |
93c80368 NB |
37 | enum spell_type category; |
38 | const unsigned char *name; | |
f9a0e96c ZW |
39 | }; |
40 | ||
8206c799 ZW |
41 | static const unsigned char *const digraph_spellings[] = |
42 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
93c80368 | 43 | |
21b11495 ZW |
44 | #define OP(e, s) { SPELL_OPERATOR, U s }, |
45 | #define TK(e, s) { SPELL_ ## s, U #e }, | |
8206c799 | 46 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
93c80368 NB |
47 | #undef OP |
48 | #undef TK | |
49 | ||
50 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
51 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
f2d5f0cc | 52 | |
6cf87ca4 ZW |
53 | static void add_line_note (cpp_buffer *, const uchar *, unsigned int); |
54 | static int skip_line_comment (cpp_reader *); | |
55 | static void skip_whitespace (cpp_reader *, cppchar_t); | |
6cf87ca4 ZW |
56 | static void lex_string (cpp_reader *, cpp_token *, const uchar *); |
57 | static void save_comment (cpp_reader *, cpp_token *, const uchar *, cppchar_t); | |
58 | static void create_literal (cpp_reader *, cpp_token *, const uchar *, | |
59 | unsigned int, enum cpp_ttype); | |
60 | static bool warn_in_comment (cpp_reader *, _cpp_line_note *); | |
61 | static int name_p (cpp_reader *, const cpp_string *); | |
6cf87ca4 ZW |
62 | static tokenrun *next_tokenrun (tokenrun *); |
63 | ||
6cf87ca4 | 64 | static _cpp_buff *new_buff (size_t); |
15dad1d9 | 65 | |
9d10c9a9 | 66 | |
041c3194 | 67 | /* Utility routine: |
9e62c811 | 68 | |
bfb9dc7f ZW |
69 | Compares, the token TOKEN to the NUL-terminated string STRING. |
70 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
041c3194 | 71 | int |
6cf87ca4 | 72 | cpp_ideq (const cpp_token *token, const char *string) |
041c3194 | 73 | { |
bfb9dc7f | 74 | if (token->type != CPP_NAME) |
041c3194 | 75 | return 0; |
bfb9dc7f | 76 | |
562a5c27 | 77 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
15dad1d9 | 78 | } |
1368ee70 | 79 | |
26aea073 NB |
80 | /* Record a note TYPE at byte POS into the current cleaned logical |
81 | line. */ | |
87062813 | 82 | static void |
6cf87ca4 | 83 | add_line_note (cpp_buffer *buffer, const uchar *pos, unsigned int type) |
0d9f234d | 84 | { |
26aea073 NB |
85 | if (buffer->notes_used == buffer->notes_cap) |
86 | { | |
87 | buffer->notes_cap = buffer->notes_cap * 2 + 200; | |
703ad42b KG |
88 | buffer->notes = xrealloc (buffer->notes, |
89 | buffer->notes_cap * sizeof (_cpp_line_note)); | |
26aea073 | 90 | } |
0d9f234d | 91 | |
26aea073 NB |
92 | buffer->notes[buffer->notes_used].pos = pos; |
93 | buffer->notes[buffer->notes_used].type = type; | |
94 | buffer->notes_used++; | |
0d9f234d NB |
95 | } |
96 | ||
26aea073 NB |
97 | /* Returns with a logical line that contains no escaped newlines or |
98 | trigraphs. This is a time-critical inner loop. */ | |
99 | void | |
6cf87ca4 | 100 | _cpp_clean_line (cpp_reader *pfile) |
45b966db | 101 | { |
26aea073 NB |
102 | cpp_buffer *buffer; |
103 | const uchar *s; | |
104 | uchar c, *d, *p; | |
87062813 | 105 | |
26aea073 NB |
106 | buffer = pfile->buffer; |
107 | buffer->cur_note = buffer->notes_used = 0; | |
108 | buffer->cur = buffer->line_base = buffer->next_line; | |
109 | buffer->need_line = false; | |
110 | s = buffer->next_line - 1; | |
87062813 | 111 | |
26aea073 | 112 | if (!buffer->from_stage3) |
45b966db | 113 | { |
d08dcf87 ZW |
114 | /* Short circuit for the common case of an un-escaped line with |
115 | no trigraphs. The primary win here is by not writing any | |
116 | data back to memory until we have to. */ | |
117 | for (;;) | |
118 | { | |
119 | c = *++s; | |
120 | if (c == '\n' || c == '\r') | |
121 | { | |
122 | d = (uchar *) s; | |
123 | ||
124 | if (s == buffer->rlimit) | |
125 | goto done; | |
126 | ||
127 | /* DOS line ending? */ | |
128 | if (c == '\r' && s[1] == '\n') | |
129 | s++; | |
130 | ||
131 | if (s == buffer->rlimit) | |
132 | goto done; | |
133 | ||
134 | /* check for escaped newline */ | |
135 | p = d; | |
136 | while (p != buffer->next_line && is_nvspace (p[-1])) | |
137 | p--; | |
138 | if (p == buffer->next_line || p[-1] != '\\') | |
139 | goto done; | |
140 | ||
141 | /* Have an escaped newline; process it and proceed to | |
142 | the slow path. */ | |
143 | add_line_note (buffer, p - 1, p != d ? ' ' : '\\'); | |
144 | d = p - 2; | |
145 | buffer->next_line = p - 1; | |
146 | break; | |
147 | } | |
148 | if (c == '?' && s[1] == '?' && _cpp_trigraph_map[s[2]]) | |
149 | { | |
150 | /* Have a trigraph. We may or may not have to convert | |
151 | it. Add a line note regardless, for -Wtrigraphs. */ | |
152 | add_line_note (buffer, s, s[2]); | |
153 | if (CPP_OPTION (pfile, trigraphs)) | |
154 | { | |
155 | /* We do, and that means we have to switch to the | |
156 | slow path. */ | |
157 | d = (uchar *) s; | |
158 | *d = _cpp_trigraph_map[s[2]]; | |
159 | s += 2; | |
160 | break; | |
161 | } | |
162 | } | |
163 | } | |
164 | ||
26aea073 NB |
165 | |
166 | for (;;) | |
4a5b68a2 | 167 | { |
26aea073 NB |
168 | c = *++s; |
169 | *++d = c; | |
170 | ||
171 | if (c == '\n' || c == '\r') | |
172 | { | |
173 | /* Handle DOS line endings. */ | |
174 | if (c == '\r' && s != buffer->rlimit && s[1] == '\n') | |
175 | s++; | |
176 | if (s == buffer->rlimit) | |
177 | break; | |
178 | ||
179 | /* Escaped? */ | |
180 | p = d; | |
181 | while (p != buffer->next_line && is_nvspace (p[-1])) | |
182 | p--; | |
183 | if (p == buffer->next_line || p[-1] != '\\') | |
184 | break; | |
185 | ||
41c32c98 | 186 | add_line_note (buffer, p - 1, p != d ? ' ': '\\'); |
26aea073 NB |
187 | d = p - 2; |
188 | buffer->next_line = p - 1; | |
189 | } | |
190 | else if (c == '?' && s[1] == '?' && _cpp_trigraph_map[s[2]]) | |
191 | { | |
192 | /* Add a note regardless, for the benefit of -Wtrigraphs. */ | |
41c32c98 | 193 | add_line_note (buffer, d, s[2]); |
26aea073 NB |
194 | if (CPP_OPTION (pfile, trigraphs)) |
195 | { | |
196 | *d = _cpp_trigraph_map[s[2]]; | |
197 | s += 2; | |
198 | } | |
199 | } | |
4a5b68a2 | 200 | } |
45b966db | 201 | } |
26aea073 NB |
202 | else |
203 | { | |
204 | do | |
205 | s++; | |
206 | while (*s != '\n' && *s != '\r'); | |
207 | d = (uchar *) s; | |
208 | ||
209 | /* Handle DOS line endings. */ | |
210 | if (*s == '\r' && s != buffer->rlimit && s[1] == '\n') | |
211 | s++; | |
212 | } | |
0d9f234d | 213 | |
d08dcf87 | 214 | done: |
26aea073 | 215 | *d = '\n'; |
41c32c98 NB |
216 | /* A sentinel note that should never be processed. */ |
217 | add_line_note (buffer, d + 1, '\n'); | |
26aea073 | 218 | buffer->next_line = s + 1; |
45b966db ZW |
219 | } |
220 | ||
a8eb6044 NB |
221 | /* Return true if the trigraph indicated by NOTE should be warned |
222 | about in a comment. */ | |
223 | static bool | |
6cf87ca4 | 224 | warn_in_comment (cpp_reader *pfile, _cpp_line_note *note) |
a8eb6044 NB |
225 | { |
226 | const uchar *p; | |
227 | ||
228 | /* Within comments we don't warn about trigraphs, unless the | |
229 | trigraph forms an escaped newline, as that may change | |
6356f892 | 230 | behavior. */ |
a8eb6044 NB |
231 | if (note->type != '/') |
232 | return false; | |
233 | ||
234 | /* If -trigraphs, then this was an escaped newline iff the next note | |
235 | is coincident. */ | |
236 | if (CPP_OPTION (pfile, trigraphs)) | |
237 | return note[1].pos == note->pos; | |
238 | ||
239 | /* Otherwise, see if this forms an escaped newline. */ | |
240 | p = note->pos + 3; | |
241 | while (is_nvspace (*p)) | |
242 | p++; | |
243 | ||
244 | /* There might have been escaped newlines between the trigraph and the | |
245 | newline we found. Hence the position test. */ | |
246 | return (*p == '\n' && p < note[1].pos); | |
247 | } | |
248 | ||
26aea073 NB |
249 | /* Process the notes created by add_line_note as far as the current |
250 | location. */ | |
251 | void | |
6cf87ca4 | 252 | _cpp_process_line_notes (cpp_reader *pfile, int in_comment) |
45b966db | 253 | { |
29401c30 NB |
254 | cpp_buffer *buffer = pfile->buffer; |
255 | ||
26aea073 | 256 | for (;;) |
041c3194 | 257 | { |
26aea073 NB |
258 | _cpp_line_note *note = &buffer->notes[buffer->cur_note]; |
259 | unsigned int col; | |
a5c3cccd | 260 | |
26aea073 NB |
261 | if (note->pos > buffer->cur) |
262 | break; | |
a5c3cccd | 263 | |
26aea073 NB |
264 | buffer->cur_note++; |
265 | col = CPP_BUF_COLUMN (buffer, note->pos + 1); | |
4d6baafa | 266 | |
41c32c98 | 267 | if (note->type == '\\' || note->type == ' ') |
26aea073 | 268 | { |
41c32c98 | 269 | if (note->type == ' ' && !in_comment) |
500bee0a | 270 | cpp_error_with_line (pfile, CPP_DL_WARNING, pfile->line_table->highest_line, col, |
26aea073 | 271 | "backslash and newline separated by space"); |
41c32c98 | 272 | |
26aea073 | 273 | if (buffer->next_line > buffer->rlimit) |
87062813 | 274 | { |
500bee0a | 275 | cpp_error_with_line (pfile, CPP_DL_PEDWARN, pfile->line_table->highest_line, col, |
26aea073 NB |
276 | "backslash-newline at end of file"); |
277 | /* Prevent "no newline at end of file" warning. */ | |
278 | buffer->next_line = buffer->rlimit; | |
87062813 | 279 | } |
26aea073 NB |
280 | |
281 | buffer->line_base = note->pos; | |
12f9df4e | 282 | CPP_INCREMENT_LINE (pfile, 0); |
0d9f234d | 283 | } |
41c32c98 NB |
284 | else if (_cpp_trigraph_map[note->type]) |
285 | { | |
a8eb6044 NB |
286 | if (CPP_OPTION (pfile, warn_trigraphs) |
287 | && (!in_comment || warn_in_comment (pfile, note))) | |
41c32c98 NB |
288 | { |
289 | if (CPP_OPTION (pfile, trigraphs)) | |
500bee0a | 290 | cpp_error_with_line (pfile, CPP_DL_WARNING, pfile->line_table->highest_line, col, |
41c32c98 NB |
291 | "trigraph ??%c converted to %c", |
292 | note->type, | |
293 | (int) _cpp_trigraph_map[note->type]); | |
294 | else | |
905bd7b5 GK |
295 | { |
296 | cpp_error_with_line | |
500bee0a | 297 | (pfile, CPP_DL_WARNING, pfile->line_table->highest_line, col, |
905bd7b5 GK |
298 | "trigraph ??%c ignored, use -trigraphs to enable", |
299 | note->type); | |
300 | } | |
41c32c98 NB |
301 | } |
302 | } | |
303 | else | |
304 | abort (); | |
041c3194 | 305 | } |
45b966db ZW |
306 | } |
307 | ||
0d9f234d NB |
308 | /* Skip a C-style block comment. We find the end of the comment by |
309 | seeing if an asterisk is before every '/' we encounter. Returns | |
6f572ac2 NB |
310 | nonzero if comment terminated by EOF, zero otherwise. |
311 | ||
312 | Buffer->cur points to the initial asterisk of the comment. */ | |
26aea073 | 313 | bool |
6cf87ca4 | 314 | _cpp_skip_block_comment (cpp_reader *pfile) |
45b966db | 315 | { |
041c3194 | 316 | cpp_buffer *buffer = pfile->buffer; |
d08dcf87 ZW |
317 | const uchar *cur = buffer->cur; |
318 | uchar c; | |
0d9f234d | 319 | |
d08dcf87 ZW |
320 | cur++; |
321 | if (*cur == '/') | |
322 | cur++; | |
0d9f234d | 323 | |
26aea073 NB |
324 | for (;;) |
325 | { | |
0d9f234d NB |
326 | /* People like decorating comments with '*', so check for '/' |
327 | instead for efficiency. */ | |
d08dcf87 ZW |
328 | c = *cur++; |
329 | ||
041c3194 | 330 | if (c == '/') |
45b966db | 331 | { |
d08dcf87 | 332 | if (cur[-2] == '*') |
0d9f234d | 333 | break; |
041c3194 | 334 | |
0d9f234d | 335 | /* Warn about potential nested comments, but not if the '/' |
a1f300c0 | 336 | comes immediately before the true comment delimiter. |
041c3194 | 337 | Don't bother to get it right across escaped newlines. */ |
0d9f234d | 338 | if (CPP_OPTION (pfile, warn_comments) |
d08dcf87 ZW |
339 | && cur[0] == '*' && cur[1] != '/') |
340 | { | |
341 | buffer->cur = cur; | |
0527bc4e | 342 | cpp_error_with_line (pfile, CPP_DL_WARNING, |
500bee0a | 343 | pfile->line_table->highest_line, CPP_BUF_COL (buffer), |
d08dcf87 ZW |
344 | "\"/*\" within comment"); |
345 | } | |
45b966db | 346 | } |
26aea073 NB |
347 | else if (c == '\n') |
348 | { | |
12f9df4e | 349 | unsigned int cols; |
d08dcf87 | 350 | buffer->cur = cur - 1; |
26aea073 NB |
351 | _cpp_process_line_notes (pfile, true); |
352 | if (buffer->next_line >= buffer->rlimit) | |
353 | return true; | |
354 | _cpp_clean_line (pfile); | |
12f9df4e PB |
355 | |
356 | cols = buffer->next_line - buffer->line_base; | |
357 | CPP_INCREMENT_LINE (pfile, cols); | |
358 | ||
d08dcf87 | 359 | cur = buffer->cur; |
26aea073 | 360 | } |
45b966db | 361 | } |
041c3194 | 362 | |
d08dcf87 | 363 | buffer->cur = cur; |
a8eb6044 | 364 | _cpp_process_line_notes (pfile, true); |
26aea073 | 365 | return false; |
45b966db ZW |
366 | } |
367 | ||
480709cc | 368 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
da7d8304 | 369 | terminating newline. Handles escaped newlines. Returns nonzero |
480709cc | 370 | if a multiline comment. */ |
041c3194 | 371 | static int |
6cf87ca4 | 372 | skip_line_comment (cpp_reader *pfile) |
45b966db | 373 | { |
cbcff6df | 374 | cpp_buffer *buffer = pfile->buffer; |
500bee0a | 375 | unsigned int orig_line = pfile->line_table->highest_line; |
041c3194 | 376 | |
26aea073 NB |
377 | while (*buffer->cur != '\n') |
378 | buffer->cur++; | |
480709cc | 379 | |
26aea073 | 380 | _cpp_process_line_notes (pfile, true); |
500bee0a | 381 | return orig_line != pfile->line_table->highest_line; |
041c3194 | 382 | } |
45b966db | 383 | |
26aea073 | 384 | /* Skips whitespace, saving the next non-whitespace character. */ |
52fadca8 | 385 | static void |
6cf87ca4 | 386 | skip_whitespace (cpp_reader *pfile, cppchar_t c) |
041c3194 ZW |
387 | { |
388 | cpp_buffer *buffer = pfile->buffer; | |
f7d151fb | 389 | bool saw_NUL = false; |
45b966db | 390 | |
0d9f234d | 391 | do |
041c3194 | 392 | { |
91fcd158 | 393 | /* Horizontal space always OK. */ |
26aea073 | 394 | if (c == ' ' || c == '\t') |
0d9f234d | 395 | ; |
0d9f234d | 396 | /* Just \f \v or \0 left. */ |
91fcd158 | 397 | else if (c == '\0') |
f7d151fb | 398 | saw_NUL = true; |
93c80368 | 399 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
500bee0a | 400 | cpp_error_with_line (pfile, CPP_DL_PEDWARN, pfile->line_table->highest_line, |
ebef4e8c NB |
401 | CPP_BUF_COL (buffer), |
402 | "%s in preprocessing directive", | |
403 | c == '\f' ? "form feed" : "vertical tab"); | |
0d9f234d | 404 | |
0d9f234d | 405 | c = *buffer->cur++; |
45b966db | 406 | } |
ec5c56db | 407 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
0d9f234d NB |
408 | while (is_nvspace (c)); |
409 | ||
f7d151fb | 410 | if (saw_NUL) |
0527bc4e | 411 | cpp_error (pfile, CPP_DL_WARNING, "null character(s) ignored"); |
f7d151fb | 412 | |
480709cc | 413 | buffer->cur--; |
041c3194 | 414 | } |
45b966db | 415 | |
93c80368 NB |
416 | /* See if the characters of a number token are valid in a name (no |
417 | '.', '+' or '-'). */ | |
418 | static int | |
6cf87ca4 | 419 | name_p (cpp_reader *pfile, const cpp_string *string) |
93c80368 NB |
420 | { |
421 | unsigned int i; | |
422 | ||
423 | for (i = 0; i < string->len; i++) | |
424 | if (!is_idchar (string->text[i])) | |
425 | return 0; | |
426 | ||
df383483 | 427 | return 1; |
93c80368 NB |
428 | } |
429 | ||
50668cf6 GK |
430 | /* After parsing an identifier or other sequence, produce a warning about |
431 | sequences not in NFC/NFKC. */ | |
432 | static void | |
433 | warn_about_normalization (cpp_reader *pfile, | |
434 | const cpp_token *token, | |
435 | const struct normalize_state *s) | |
436 | { | |
437 | if (CPP_OPTION (pfile, warn_normalize) < NORMALIZE_STATE_RESULT (s) | |
438 | && !pfile->state.skipping) | |
439 | { | |
440 | /* Make sure that the token is printed using UCNs, even | |
441 | if we'd otherwise happily print UTF-8. */ | |
442 | unsigned char *buf = xmalloc (cpp_token_len (token)); | |
443 | size_t sz; | |
444 | ||
445 | sz = cpp_spell_token (pfile, token, buf, false) - buf; | |
446 | if (NORMALIZE_STATE_RESULT (s) == normalized_C) | |
447 | cpp_error_with_line (pfile, CPP_DL_WARNING, token->src_loc, 0, | |
cbada204 | 448 | "`%.*s' is not in NFKC", (int) sz, buf); |
50668cf6 GK |
449 | else |
450 | cpp_error_with_line (pfile, CPP_DL_WARNING, token->src_loc, 0, | |
cbada204 | 451 | "`%.*s' is not in NFC", (int) sz, buf); |
50668cf6 GK |
452 | } |
453 | } | |
454 | ||
bced6edf | 455 | /* Returns TRUE if the sequence starting at buffer->cur is invalid in |
1613e52b | 456 | an identifier. FIRST is TRUE if this starts an identifier. */ |
bced6edf | 457 | static bool |
50668cf6 GK |
458 | forms_identifier_p (cpp_reader *pfile, int first, |
459 | struct normalize_state *state) | |
bced6edf | 460 | { |
1613e52b NB |
461 | cpp_buffer *buffer = pfile->buffer; |
462 | ||
463 | if (*buffer->cur == '$') | |
464 | { | |
465 | if (!CPP_OPTION (pfile, dollars_in_ident)) | |
466 | return false; | |
467 | ||
468 | buffer->cur++; | |
78b8811a | 469 | if (CPP_OPTION (pfile, warn_dollars) && !pfile->state.skipping) |
1613e52b | 470 | { |
78b8811a | 471 | CPP_OPTION (pfile, warn_dollars) = 0; |
0527bc4e | 472 | cpp_error (pfile, CPP_DL_PEDWARN, "'$' in identifier or number"); |
1613e52b NB |
473 | } |
474 | ||
475 | return true; | |
476 | } | |
bced6edf | 477 | |
1613e52b | 478 | /* Is this a syntactically valid UCN? */ |
6baba9bb GK |
479 | if ((CPP_OPTION (pfile, cplusplus) || CPP_OPTION (pfile, c99)) |
480 | && *buffer->cur == '\\' | |
1613e52b | 481 | && (buffer->cur[1] == 'u' || buffer->cur[1] == 'U')) |
bced6edf | 482 | { |
1613e52b | 483 | buffer->cur += 2; |
50668cf6 GK |
484 | if (_cpp_valid_ucn (pfile, &buffer->cur, buffer->rlimit, 1 + !first, |
485 | state)) | |
1613e52b NB |
486 | return true; |
487 | buffer->cur -= 2; | |
bced6edf | 488 | } |
bced6edf | 489 | |
1613e52b | 490 | return false; |
bced6edf NB |
491 | } |
492 | ||
493 | /* Lex an identifier starting at BUFFER->CUR - 1. */ | |
0d9f234d | 494 | static cpp_hashnode * |
50668cf6 GK |
495 | lex_identifier (cpp_reader *pfile, const uchar *base, bool starts_ucn, |
496 | struct normalize_state *nst) | |
45b966db | 497 | { |
93c80368 | 498 | cpp_hashnode *result; |
47e20491 | 499 | const uchar *cur; |
c6e83800 ZW |
500 | unsigned int len; |
501 | unsigned int hash = HT_HASHSTEP (0, *base); | |
2c3fcba6 | 502 | |
c6e83800 | 503 | cur = pfile->buffer->cur; |
47e20491 GK |
504 | if (! starts_ucn) |
505 | while (ISIDNUM (*cur)) | |
506 | { | |
507 | hash = HT_HASHSTEP (hash, *cur); | |
508 | cur++; | |
509 | } | |
510 | pfile->buffer->cur = cur; | |
50668cf6 | 511 | if (starts_ucn || forms_identifier_p (pfile, false, nst)) |
10cf9bde | 512 | { |
47e20491 GK |
513 | /* Slower version for identifiers containing UCNs (or $). */ |
514 | do { | |
515 | while (ISIDNUM (*pfile->buffer->cur)) | |
50668cf6 GK |
516 | { |
517 | pfile->buffer->cur++; | |
518 | NORMALIZE_STATE_UPDATE_IDNUM (nst); | |
519 | } | |
520 | } while (forms_identifier_p (pfile, false, nst)); | |
47e20491 GK |
521 | result = _cpp_interpret_identifier (pfile, base, |
522 | pfile->buffer->cur - base); | |
2c3fcba6 | 523 | } |
47e20491 GK |
524 | else |
525 | { | |
526 | len = cur - base; | |
527 | hash = HT_HASHFINISH (hash, len); | |
bced6edf | 528 | |
47e20491 GK |
529 | result = (cpp_hashnode *) |
530 | ht_lookup_with_hash (pfile->hash_table, base, len, hash, HT_ALLOC); | |
531 | } | |
2c3fcba6 | 532 | |
bced6edf | 533 | /* Rarely, identifiers require diagnostics when lexed. */ |
2c3fcba6 ZW |
534 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) |
535 | && !pfile->state.skipping, 0)) | |
536 | { | |
537 | /* It is allowed to poison the same identifier twice. */ | |
538 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
0527bc4e | 539 | cpp_error (pfile, CPP_DL_ERROR, "attempt to use poisoned \"%s\"", |
2c3fcba6 ZW |
540 | NODE_NAME (result)); |
541 | ||
542 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
543 | replacement list of a variadic macro. */ | |
544 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
545 | && !pfile->state.va_args_ok) | |
0527bc4e | 546 | cpp_error (pfile, CPP_DL_PEDWARN, |
6cf87ca4 ZW |
547 | "__VA_ARGS__ can only appear in the expansion" |
548 | " of a C99 variadic macro"); | |
2c3fcba6 ZW |
549 | } |
550 | ||
551 | return result; | |
552 | } | |
553 | ||
bced6edf | 554 | /* Lex a number to NUMBER starting at BUFFER->CUR - 1. */ |
45b966db | 555 | static void |
50668cf6 GK |
556 | lex_number (cpp_reader *pfile, cpp_string *number, |
557 | struct normalize_state *nst) | |
45b966db | 558 | { |
562a5c27 | 559 | const uchar *cur; |
bced6edf NB |
560 | const uchar *base; |
561 | uchar *dest; | |
45b966db | 562 | |
bced6edf NB |
563 | base = pfile->buffer->cur - 1; |
564 | do | |
041c3194 | 565 | { |
bced6edf | 566 | cur = pfile->buffer->cur; |
0d9f234d | 567 | |
bced6edf NB |
568 | /* N.B. ISIDNUM does not include $. */ |
569 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
50668cf6 GK |
570 | { |
571 | cur++; | |
572 | NORMALIZE_STATE_UPDATE_IDNUM (nst); | |
573 | } | |
45b966db | 574 | |
10cf9bde | 575 | pfile->buffer->cur = cur; |
45b966db | 576 | } |
50668cf6 | 577 | while (forms_identifier_p (pfile, false, nst)); |
93c80368 | 578 | |
bced6edf NB |
579 | number->len = cur - base; |
580 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
581 | memcpy (dest, base, number->len); | |
582 | dest[number->len] = '\0'; | |
583 | number->text = dest; | |
93c80368 NB |
584 | } |
585 | ||
6338b358 NB |
586 | /* Create a token of type TYPE with a literal spelling. */ |
587 | static void | |
6cf87ca4 ZW |
588 | create_literal (cpp_reader *pfile, cpp_token *token, const uchar *base, |
589 | unsigned int len, enum cpp_ttype type) | |
6338b358 NB |
590 | { |
591 | uchar *dest = _cpp_unaligned_alloc (pfile, len + 1); | |
592 | ||
593 | memcpy (dest, base, len); | |
594 | dest[len] = '\0'; | |
595 | token->type = type; | |
596 | token->val.str.len = len; | |
597 | token->val.str.text = dest; | |
598 | } | |
599 | ||
bced6edf | 600 | /* Lexes a string, character constant, or angle-bracketed header file |
6338b358 NB |
601 | name. The stored string contains the spelling, including opening |
602 | quote and leading any leading 'L'. It returns the type of the | |
603 | literal, or CPP_OTHER if it was not properly terminated. | |
604 | ||
605 | The spelling is NUL-terminated, but it is not guaranteed that this | |
606 | is the first NUL since embedded NULs are preserved. */ | |
041c3194 | 607 | static void |
6cf87ca4 | 608 | lex_string (cpp_reader *pfile, cpp_token *token, const uchar *base) |
45b966db | 609 | { |
6338b358 NB |
610 | bool saw_NUL = false; |
611 | const uchar *cur; | |
bced6edf | 612 | cppchar_t terminator; |
6338b358 NB |
613 | enum cpp_ttype type; |
614 | ||
615 | cur = base; | |
616 | terminator = *cur++; | |
617 | if (terminator == 'L') | |
618 | terminator = *cur++; | |
619 | if (terminator == '\"') | |
620 | type = *base == 'L' ? CPP_WSTRING: CPP_STRING; | |
621 | else if (terminator == '\'') | |
622 | type = *base == 'L' ? CPP_WCHAR: CPP_CHAR; | |
623 | else | |
624 | terminator = '>', type = CPP_HEADER_NAME; | |
93c80368 | 625 | |
0d9f234d | 626 | for (;;) |
45b966db | 627 | { |
6338b358 | 628 | cppchar_t c = *cur++; |
7868b4a2 | 629 | |
6f572ac2 | 630 | /* In #include-style directives, terminators are not escapable. */ |
6338b358 NB |
631 | if (c == '\\' && !pfile->state.angled_headers && *cur != '\n') |
632 | cur++; | |
633 | else if (c == terminator) | |
bced6edf | 634 | break; |
6338b358 | 635 | else if (c == '\n') |
0d9f234d | 636 | { |
6338b358 NB |
637 | cur--; |
638 | type = CPP_OTHER; | |
639 | break; | |
45b966db | 640 | } |
6338b358 NB |
641 | else if (c == '\0') |
642 | saw_NUL = true; | |
45b966db ZW |
643 | } |
644 | ||
6338b358 | 645 | if (saw_NUL && !pfile->state.skipping) |
0527bc4e JDA |
646 | cpp_error (pfile, CPP_DL_WARNING, |
647 | "null character(s) preserved in literal"); | |
45b966db | 648 | |
6338b358 NB |
649 | pfile->buffer->cur = cur; |
650 | create_literal (pfile, token, base, cur - base, type); | |
0d9f234d | 651 | } |
041c3194 | 652 | |
93c80368 | 653 | /* The stored comment includes the comment start and any terminator. */ |
9e62c811 | 654 | static void |
6cf87ca4 ZW |
655 | save_comment (cpp_reader *pfile, cpp_token *token, const unsigned char *from, |
656 | cppchar_t type) | |
9e62c811 | 657 | { |
041c3194 | 658 | unsigned char *buffer; |
477cdac7 | 659 | unsigned int len, clen; |
df383483 | 660 | |
1c6d33ef | 661 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
480709cc | 662 | |
3542203b NB |
663 | /* C++ comments probably (not definitely) have moved past a new |
664 | line, which we don't want to save in the comment. */ | |
480709cc | 665 | if (is_vspace (pfile->buffer->cur[-1])) |
3542203b | 666 | len--; |
477cdac7 JT |
667 | |
668 | /* If we are currently in a directive, then we need to store all | |
669 | C++ comments as C comments internally, and so we need to | |
670 | allocate a little extra space in that case. | |
671 | ||
672 | Note that the only time we encounter a directive here is | |
673 | when we are saving comments in a "#define". */ | |
674 | clen = (pfile->state.in_directive && type == '/') ? len + 2 : len; | |
675 | ||
676 | buffer = _cpp_unaligned_alloc (pfile, clen); | |
df383483 | 677 | |
041c3194 | 678 | token->type = CPP_COMMENT; |
477cdac7 | 679 | token->val.str.len = clen; |
0d9f234d | 680 | token->val.str.text = buffer; |
45b966db | 681 | |
1c6d33ef NB |
682 | buffer[0] = '/'; |
683 | memcpy (buffer + 1, from, len - 1); | |
477cdac7 | 684 | |
1eeeb6a4 | 685 | /* Finish conversion to a C comment, if necessary. */ |
477cdac7 JT |
686 | if (pfile->state.in_directive && type == '/') |
687 | { | |
688 | buffer[1] = '*'; | |
689 | buffer[clen - 2] = '*'; | |
690 | buffer[clen - 1] = '/'; | |
691 | } | |
0d9f234d | 692 | } |
45b966db | 693 | |
5fddcffc NB |
694 | /* Allocate COUNT tokens for RUN. */ |
695 | void | |
6cf87ca4 | 696 | _cpp_init_tokenrun (tokenrun *run, unsigned int count) |
5fddcffc | 697 | { |
72bb2c39 | 698 | run->base = XNEWVEC (cpp_token, count); |
5fddcffc NB |
699 | run->limit = run->base + count; |
700 | run->next = NULL; | |
701 | } | |
702 | ||
703 | /* Returns the next tokenrun, or creates one if there is none. */ | |
704 | static tokenrun * | |
6cf87ca4 | 705 | next_tokenrun (tokenrun *run) |
5fddcffc NB |
706 | { |
707 | if (run->next == NULL) | |
708 | { | |
72bb2c39 | 709 | run->next = XNEW (tokenrun); |
bdcbe496 | 710 | run->next->prev = run; |
5fddcffc NB |
711 | _cpp_init_tokenrun (run->next, 250); |
712 | } | |
713 | ||
714 | return run->next; | |
715 | } | |
716 | ||
4ed5bcfb NB |
717 | /* Allocate a single token that is invalidated at the same time as the |
718 | rest of the tokens on the line. Has its line and col set to the | |
719 | same as the last lexed token, so that diagnostics appear in the | |
720 | right place. */ | |
721 | cpp_token * | |
6cf87ca4 | 722 | _cpp_temp_token (cpp_reader *pfile) |
4ed5bcfb NB |
723 | { |
724 | cpp_token *old, *result; | |
725 | ||
726 | old = pfile->cur_token - 1; | |
727 | if (pfile->cur_token == pfile->cur_run->limit) | |
728 | { | |
729 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
730 | pfile->cur_token = pfile->cur_run->base; | |
731 | } | |
732 | ||
733 | result = pfile->cur_token++; | |
12f9df4e | 734 | result->src_loc = old->src_loc; |
4ed5bcfb NB |
735 | return result; |
736 | } | |
737 | ||
14baae01 NB |
738 | /* Lex a token into RESULT (external interface). Takes care of issues |
739 | like directive handling, token lookahead, multiple include | |
a1f300c0 | 740 | optimization and skipping. */ |
345894b4 | 741 | const cpp_token * |
6cf87ca4 | 742 | _cpp_lex_token (cpp_reader *pfile) |
5fddcffc | 743 | { |
bdcbe496 | 744 | cpp_token *result; |
5fddcffc | 745 | |
bdcbe496 | 746 | for (;;) |
5fddcffc | 747 | { |
bdcbe496 | 748 | if (pfile->cur_token == pfile->cur_run->limit) |
5fddcffc | 749 | { |
bdcbe496 NB |
750 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
751 | pfile->cur_token = pfile->cur_run->base; | |
5fddcffc NB |
752 | } |
753 | ||
bdcbe496 | 754 | if (pfile->lookaheads) |
14baae01 NB |
755 | { |
756 | pfile->lookaheads--; | |
757 | result = pfile->cur_token++; | |
758 | } | |
bdcbe496 | 759 | else |
14baae01 | 760 | result = _cpp_lex_direct (pfile); |
bdcbe496 NB |
761 | |
762 | if (result->flags & BOL) | |
5fddcffc | 763 | { |
bdcbe496 NB |
764 | /* Is this a directive. If _cpp_handle_directive returns |
765 | false, it is an assembler #. */ | |
766 | if (result->type == CPP_HASH | |
e808ec9c NB |
767 | /* 6.10.3 p 11: Directives in a list of macro arguments |
768 | gives undefined behavior. This implementation | |
769 | handles the directive as normal. */ | |
770 | && pfile->state.parsing_args != 1 | |
bdcbe496 | 771 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
21b11495 ZW |
772 | { |
773 | if (pfile->directive_result.type == CPP_PADDING) | |
774 | continue; | |
775 | else | |
776 | { | |
777 | result = &pfile->directive_result; | |
778 | break; | |
779 | } | |
780 | } | |
781 | ||
97293897 | 782 | if (pfile->cb.line_change && !pfile->state.skipping) |
6cf87ca4 | 783 | pfile->cb.line_change (pfile, result, pfile->state.parsing_args); |
5fddcffc | 784 | } |
5fddcffc | 785 | |
bdcbe496 NB |
786 | /* We don't skip tokens in directives. */ |
787 | if (pfile->state.in_directive) | |
788 | break; | |
5fddcffc | 789 | |
bdcbe496 | 790 | /* Outside a directive, invalidate controlling macros. At file |
14baae01 | 791 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
6356f892 | 792 | get here and MI optimization works. */ |
5fddcffc | 793 | pfile->mi_valid = false; |
bdcbe496 NB |
794 | |
795 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
796 | break; | |
5fddcffc NB |
797 | } |
798 | ||
345894b4 | 799 | return result; |
5fddcffc NB |
800 | } |
801 | ||
26aea073 NB |
802 | /* Returns true if a fresh line has been loaded. */ |
803 | bool | |
6cf87ca4 | 804 | _cpp_get_fresh_line (cpp_reader *pfile) |
004cb263 | 805 | { |
22234f56 PB |
806 | int return_at_eof; |
807 | ||
26aea073 NB |
808 | /* We can't get a new line until we leave the current directive. */ |
809 | if (pfile->state.in_directive) | |
810 | return false; | |
df383483 | 811 | |
26aea073 | 812 | for (;;) |
1a76916c | 813 | { |
26aea073 | 814 | cpp_buffer *buffer = pfile->buffer; |
1a76916c | 815 | |
26aea073 NB |
816 | if (!buffer->need_line) |
817 | return true; | |
818 | ||
819 | if (buffer->next_line < buffer->rlimit) | |
004cb263 | 820 | { |
26aea073 NB |
821 | _cpp_clean_line (pfile); |
822 | return true; | |
823 | } | |
004cb263 | 824 | |
26aea073 NB |
825 | /* First, get out of parsing arguments state. */ |
826 | if (pfile->state.parsing_args) | |
827 | return false; | |
828 | ||
829 | /* End of buffer. Non-empty files should end in a newline. */ | |
830 | if (buffer->buf != buffer->rlimit | |
831 | && buffer->next_line > buffer->rlimit | |
832 | && !buffer->from_stage3) | |
833 | { | |
834 | /* Only warn once. */ | |
835 | buffer->next_line = buffer->rlimit; | |
500bee0a | 836 | cpp_error_with_line (pfile, CPP_DL_PEDWARN, pfile->line_table->highest_line, |
26aea073 NB |
837 | CPP_BUF_COLUMN (buffer, buffer->cur), |
838 | "no newline at end of file"); | |
839 | } | |
22234f56 PB |
840 | |
841 | return_at_eof = buffer->return_at_eof; | |
26aea073 | 842 | _cpp_pop_buffer (pfile); |
22234f56 | 843 | if (pfile->buffer == NULL || return_at_eof) |
a506c55c | 844 | return false; |
26aea073 | 845 | } |
004cb263 NB |
846 | } |
847 | ||
6f572ac2 NB |
848 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
849 | do \ | |
850 | { \ | |
851 | result->type = ELSE_TYPE; \ | |
852 | if (*buffer->cur == CHAR) \ | |
853 | buffer->cur++, result->type = THEN_TYPE; \ | |
854 | } \ | |
855 | while (0) | |
480709cc | 856 | |
14baae01 NB |
857 | /* Lex a token into pfile->cur_token, which is also incremented, to |
858 | get diagnostics pointing to the correct location. | |
859 | ||
860 | Does not handle issues such as token lookahead, multiple-include | |
f1ba665b | 861 | optimization, directives, skipping etc. This function is only |
14baae01 NB |
862 | suitable for use by _cpp_lex_token, and in special cases like |
863 | lex_expansion_token which doesn't care for any of these issues. | |
864 | ||
865 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
866 | otherwise returns to the start of the token buffer if permissible. | |
867 | Returns the location of the lexed token. */ | |
868 | cpp_token * | |
6cf87ca4 | 869 | _cpp_lex_direct (cpp_reader *pfile) |
45b966db | 870 | { |
0d9f234d | 871 | cppchar_t c; |
adb84b42 | 872 | cpp_buffer *buffer; |
0d9f234d | 873 | const unsigned char *comment_start; |
14baae01 | 874 | cpp_token *result = pfile->cur_token++; |
9ec7291f | 875 | |
5fddcffc | 876 | fresh_line: |
26aea073 | 877 | result->flags = 0; |
2be570f9 | 878 | buffer = pfile->buffer; |
a506c55c | 879 | if (buffer->need_line) |
26aea073 NB |
880 | { |
881 | if (!_cpp_get_fresh_line (pfile)) | |
882 | { | |
883 | result->type = CPP_EOF; | |
9ff7868d NB |
884 | if (!pfile->state.in_directive) |
885 | { | |
886 | /* Tell the compiler the line number of the EOF token. */ | |
500bee0a | 887 | result->src_loc = pfile->line_table->highest_line; |
9ff7868d NB |
888 | result->flags = BOL; |
889 | } | |
26aea073 NB |
890 | return result; |
891 | } | |
892 | if (!pfile->keep_tokens) | |
893 | { | |
894 | pfile->cur_run = &pfile->base_run; | |
895 | result = pfile->base_run.base; | |
896 | pfile->cur_token = result + 1; | |
897 | } | |
898 | result->flags = BOL; | |
899 | if (pfile->state.parsing_args == 2) | |
900 | result->flags |= PREV_WHITE; | |
901 | } | |
a506c55c | 902 | buffer = pfile->buffer; |
5fddcffc | 903 | update_tokens_line: |
500bee0a | 904 | result->src_loc = pfile->line_table->highest_line; |
041c3194 | 905 | |
5fddcffc | 906 | skipped_white: |
26aea073 NB |
907 | if (buffer->cur >= buffer->notes[buffer->cur_note].pos |
908 | && !pfile->overlaid_buffer) | |
909 | { | |
910 | _cpp_process_line_notes (pfile, false); | |
500bee0a | 911 | result->src_loc = pfile->line_table->highest_line; |
26aea073 | 912 | } |
480709cc | 913 | c = *buffer->cur++; |
12f9df4e | 914 | |
500bee0a PB |
915 | LINEMAP_POSITION_FOR_COLUMN (result->src_loc, pfile->line_table, |
916 | CPP_BUF_COLUMN (buffer, buffer->cur)); | |
5fddcffc | 917 | |
0d9f234d | 918 | switch (c) |
45b966db | 919 | { |
4d6baafa NB |
920 | case ' ': case '\t': case '\f': case '\v': case '\0': |
921 | result->flags |= PREV_WHITE; | |
26aea073 NB |
922 | skip_whitespace (pfile, c); |
923 | goto skipped_white; | |
0d9f234d | 924 | |
26aea073 | 925 | case '\n': |
12f9df4e PB |
926 | if (buffer->cur < buffer->rlimit) |
927 | CPP_INCREMENT_LINE (pfile, 0); | |
26aea073 NB |
928 | buffer->need_line = true; |
929 | goto fresh_line; | |
46d07497 | 930 | |
0d9f234d NB |
931 | case '0': case '1': case '2': case '3': case '4': |
932 | case '5': case '6': case '7': case '8': case '9': | |
50668cf6 GK |
933 | { |
934 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; | |
935 | result->type = CPP_NUMBER; | |
936 | lex_number (pfile, &result->val.str, &nst); | |
937 | warn_about_normalization (pfile, result, &nst); | |
938 | break; | |
939 | } | |
46d07497 | 940 | |
0abc6a6a NB |
941 | case 'L': |
942 | /* 'L' may introduce wide characters or strings. */ | |
bced6edf NB |
943 | if (*buffer->cur == '\'' || *buffer->cur == '"') |
944 | { | |
6338b358 | 945 | lex_string (pfile, result, buffer->cur - 1); |
bced6edf NB |
946 | break; |
947 | } | |
df383483 | 948 | /* Fall through. */ |
0abc6a6a | 949 | |
0d9f234d NB |
950 | case '_': |
951 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
952 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
953 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
954 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
955 | case 'y': case 'z': | |
956 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
0abc6a6a | 957 | case 'G': case 'H': case 'I': case 'J': case 'K': |
0d9f234d NB |
958 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
959 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
960 | case 'Y': case 'Z': | |
961 | result->type = CPP_NAME; | |
50668cf6 GK |
962 | { |
963 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; | |
964 | result->val.node = lex_identifier (pfile, buffer->cur - 1, false, | |
965 | &nst); | |
966 | warn_about_normalization (pfile, result, &nst); | |
967 | } | |
0d9f234d | 968 | |
0d9f234d | 969 | /* Convert named operators to their proper types. */ |
0abc6a6a | 970 | if (result->val.node->flags & NODE_OPERATOR) |
0d9f234d NB |
971 | { |
972 | result->flags |= NAMED_OP; | |
4977bab6 | 973 | result->type = result->val.node->directive_index; |
0d9f234d NB |
974 | } |
975 | break; | |
976 | ||
977 | case '\'': | |
978 | case '"': | |
6338b358 | 979 | lex_string (pfile, result, buffer->cur - 1); |
0d9f234d | 980 | break; |
041c3194 | 981 | |
0d9f234d | 982 | case '/': |
1c6d33ef NB |
983 | /* A potential block or line comment. */ |
984 | comment_start = buffer->cur; | |
6f572ac2 NB |
985 | c = *buffer->cur; |
986 | ||
1c6d33ef NB |
987 | if (c == '*') |
988 | { | |
26aea073 | 989 | if (_cpp_skip_block_comment (pfile)) |
0527bc4e | 990 | cpp_error (pfile, CPP_DL_ERROR, "unterminated comment"); |
0d9f234d | 991 | } |
480709cc | 992 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
12f9df4e | 993 | || cpp_in_system_header (pfile))) |
0d9f234d | 994 | { |
bdb05a7b NB |
995 | /* Warn about comments only if pedantically GNUC89, and not |
996 | in system headers. */ | |
997 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
a94c1199 | 998 | && ! buffer->warned_cplusplus_comments) |
041c3194 | 999 | { |
0527bc4e | 1000 | cpp_error (pfile, CPP_DL_PEDWARN, |
56508306 | 1001 | "C++ style comments are not allowed in ISO C90"); |
0527bc4e | 1002 | cpp_error (pfile, CPP_DL_PEDWARN, |
ebef4e8c | 1003 | "(this will be reported only once per input file)"); |
1c6d33ef NB |
1004 | buffer->warned_cplusplus_comments = 1; |
1005 | } | |
0d9f234d | 1006 | |
01ef6563 | 1007 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
0527bc4e | 1008 | cpp_error (pfile, CPP_DL_WARNING, "multi-line comment"); |
1c6d33ef | 1009 | } |
480709cc NB |
1010 | else if (c == '=') |
1011 | { | |
6f572ac2 | 1012 | buffer->cur++; |
480709cc NB |
1013 | result->type = CPP_DIV_EQ; |
1014 | break; | |
1015 | } | |
1016 | else | |
1017 | { | |
480709cc NB |
1018 | result->type = CPP_DIV; |
1019 | break; | |
1020 | } | |
0d9f234d | 1021 | |
1c6d33ef NB |
1022 | if (!pfile->state.save_comments) |
1023 | { | |
1024 | result->flags |= PREV_WHITE; | |
5fddcffc | 1025 | goto update_tokens_line; |
0d9f234d | 1026 | } |
1c6d33ef NB |
1027 | |
1028 | /* Save the comment as a token in its own right. */ | |
477cdac7 | 1029 | save_comment (pfile, result, comment_start, c); |
bdcbe496 | 1030 | break; |
0d9f234d NB |
1031 | |
1032 | case '<': | |
1033 | if (pfile->state.angled_headers) | |
1034 | { | |
6338b358 | 1035 | lex_string (pfile, result, buffer->cur - 1); |
480709cc | 1036 | break; |
0d9f234d | 1037 | } |
45b966db | 1038 | |
6f572ac2 NB |
1039 | result->type = CPP_LESS; |
1040 | if (*buffer->cur == '=') | |
1041 | buffer->cur++, result->type = CPP_LESS_EQ; | |
1042 | else if (*buffer->cur == '<') | |
0d9f234d | 1043 | { |
6f572ac2 NB |
1044 | buffer->cur++; |
1045 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); | |
0d9f234d | 1046 | } |
6f572ac2 | 1047 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) |
0d9f234d | 1048 | { |
6f572ac2 NB |
1049 | buffer->cur++; |
1050 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); | |
0d9f234d | 1051 | } |
6f572ac2 | 1052 | else if (CPP_OPTION (pfile, digraphs)) |
480709cc | 1053 | { |
6f572ac2 NB |
1054 | if (*buffer->cur == ':') |
1055 | { | |
1056 | buffer->cur++; | |
1057 | result->flags |= DIGRAPH; | |
1058 | result->type = CPP_OPEN_SQUARE; | |
1059 | } | |
1060 | else if (*buffer->cur == '%') | |
1061 | { | |
1062 | buffer->cur++; | |
1063 | result->flags |= DIGRAPH; | |
1064 | result->type = CPP_OPEN_BRACE; | |
1065 | } | |
480709cc | 1066 | } |
0d9f234d NB |
1067 | break; |
1068 | ||
1069 | case '>': | |
6f572ac2 NB |
1070 | result->type = CPP_GREATER; |
1071 | if (*buffer->cur == '=') | |
1072 | buffer->cur++, result->type = CPP_GREATER_EQ; | |
1073 | else if (*buffer->cur == '>') | |
0d9f234d | 1074 | { |
6f572ac2 NB |
1075 | buffer->cur++; |
1076 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); | |
1077 | } | |
1078 | else if (*buffer->cur == '?' && CPP_OPTION (pfile, cplusplus)) | |
1079 | { | |
1080 | buffer->cur++; | |
1081 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); | |
0d9f234d NB |
1082 | } |
1083 | break; | |
1084 | ||
cbcff6df | 1085 | case '%': |
6f572ac2 NB |
1086 | result->type = CPP_MOD; |
1087 | if (*buffer->cur == '=') | |
1088 | buffer->cur++, result->type = CPP_MOD_EQ; | |
1089 | else if (CPP_OPTION (pfile, digraphs)) | |
480709cc | 1090 | { |
6f572ac2 | 1091 | if (*buffer->cur == ':') |
480709cc | 1092 | { |
6f572ac2 NB |
1093 | buffer->cur++; |
1094 | result->flags |= DIGRAPH; | |
1095 | result->type = CPP_HASH; | |
1096 | if (*buffer->cur == '%' && buffer->cur[1] == ':') | |
1097 | buffer->cur += 2, result->type = CPP_PASTE; | |
1098 | } | |
1099 | else if (*buffer->cur == '>') | |
1100 | { | |
1101 | buffer->cur++; | |
1102 | result->flags |= DIGRAPH; | |
1103 | result->type = CPP_CLOSE_BRACE; | |
480709cc | 1104 | } |
480709cc | 1105 | } |
0d9f234d NB |
1106 | break; |
1107 | ||
cbcff6df | 1108 | case '.': |
480709cc | 1109 | result->type = CPP_DOT; |
6f572ac2 | 1110 | if (ISDIGIT (*buffer->cur)) |
480709cc | 1111 | { |
50668cf6 | 1112 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; |
480709cc | 1113 | result->type = CPP_NUMBER; |
50668cf6 GK |
1114 | lex_number (pfile, &result->val.str, &nst); |
1115 | warn_about_normalization (pfile, result, &nst); | |
480709cc | 1116 | } |
6f572ac2 NB |
1117 | else if (*buffer->cur == '.' && buffer->cur[1] == '.') |
1118 | buffer->cur += 2, result->type = CPP_ELLIPSIS; | |
1119 | else if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) | |
1120 | buffer->cur++, result->type = CPP_DOT_STAR; | |
0d9f234d | 1121 | break; |
45b966db | 1122 | |
0d9f234d | 1123 | case '+': |
6f572ac2 NB |
1124 | result->type = CPP_PLUS; |
1125 | if (*buffer->cur == '+') | |
1126 | buffer->cur++, result->type = CPP_PLUS_PLUS; | |
1127 | else if (*buffer->cur == '=') | |
1128 | buffer->cur++, result->type = CPP_PLUS_EQ; | |
0d9f234d | 1129 | break; |
04e3ec78 | 1130 | |
0d9f234d | 1131 | case '-': |
6f572ac2 NB |
1132 | result->type = CPP_MINUS; |
1133 | if (*buffer->cur == '>') | |
0d9f234d | 1134 | { |
6f572ac2 | 1135 | buffer->cur++; |
480709cc | 1136 | result->type = CPP_DEREF; |
6f572ac2 NB |
1137 | if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) |
1138 | buffer->cur++, result->type = CPP_DEREF_STAR; | |
480709cc | 1139 | } |
6f572ac2 NB |
1140 | else if (*buffer->cur == '-') |
1141 | buffer->cur++, result->type = CPP_MINUS_MINUS; | |
1142 | else if (*buffer->cur == '=') | |
1143 | buffer->cur++, result->type = CPP_MINUS_EQ; | |
0d9f234d | 1144 | break; |
45b966db | 1145 | |
0d9f234d | 1146 | case '&': |
6f572ac2 NB |
1147 | result->type = CPP_AND; |
1148 | if (*buffer->cur == '&') | |
1149 | buffer->cur++, result->type = CPP_AND_AND; | |
1150 | else if (*buffer->cur == '=') | |
1151 | buffer->cur++, result->type = CPP_AND_EQ; | |
0d9f234d | 1152 | break; |
df383483 | 1153 | |
0d9f234d | 1154 | case '|': |
6f572ac2 NB |
1155 | result->type = CPP_OR; |
1156 | if (*buffer->cur == '|') | |
1157 | buffer->cur++, result->type = CPP_OR_OR; | |
1158 | else if (*buffer->cur == '=') | |
1159 | buffer->cur++, result->type = CPP_OR_EQ; | |
0d9f234d | 1160 | break; |
45b966db | 1161 | |
0d9f234d | 1162 | case ':': |
6f572ac2 NB |
1163 | result->type = CPP_COLON; |
1164 | if (*buffer->cur == ':' && CPP_OPTION (pfile, cplusplus)) | |
1165 | buffer->cur++, result->type = CPP_SCOPE; | |
1166 | else if (*buffer->cur == '>' && CPP_OPTION (pfile, digraphs)) | |
0d9f234d | 1167 | { |
6f572ac2 | 1168 | buffer->cur++; |
0d9f234d | 1169 | result->flags |= DIGRAPH; |
480709cc NB |
1170 | result->type = CPP_CLOSE_SQUARE; |
1171 | } | |
0d9f234d | 1172 | break; |
45b966db | 1173 | |
480709cc NB |
1174 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1175 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1176 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1177 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1178 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1179 | ||
26aea073 | 1180 | case '?': result->type = CPP_QUERY; break; |
0d9f234d NB |
1181 | case '~': result->type = CPP_COMPL; break; |
1182 | case ',': result->type = CPP_COMMA; break; | |
1183 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1184 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1185 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1186 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1187 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1188 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1189 | case ';': result->type = CPP_SEMICOLON; break; | |
1190 | ||
40f03658 | 1191 | /* @ is a punctuator in Objective-C. */ |
cc937581 | 1192 | case '@': result->type = CPP_ATSIGN; break; |
0d9f234d | 1193 | |
0abc6a6a | 1194 | case '$': |
1613e52b NB |
1195 | case '\\': |
1196 | { | |
1197 | const uchar *base = --buffer->cur; | |
50668cf6 | 1198 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; |
0abc6a6a | 1199 | |
50668cf6 | 1200 | if (forms_identifier_p (pfile, true, &nst)) |
1613e52b NB |
1201 | { |
1202 | result->type = CPP_NAME; | |
50668cf6 GK |
1203 | result->val.node = lex_identifier (pfile, base, true, &nst); |
1204 | warn_about_normalization (pfile, result, &nst); | |
1613e52b NB |
1205 | break; |
1206 | } | |
1207 | buffer->cur++; | |
1067694a | 1208 | } |
1613e52b | 1209 | |
1067694a | 1210 | default: |
6338b358 NB |
1211 | create_literal (pfile, result, buffer->cur - 1, 1, CPP_OTHER); |
1212 | break; | |
0d9f234d | 1213 | } |
bdcbe496 NB |
1214 | |
1215 | return result; | |
0d9f234d NB |
1216 | } |
1217 | ||
59325650 NB |
1218 | /* An upper bound on the number of bytes needed to spell TOKEN. |
1219 | Does not include preceding whitespace. */ | |
93c80368 | 1220 | unsigned int |
6cf87ca4 | 1221 | cpp_token_len (const cpp_token *token) |
0d9f234d | 1222 | { |
93c80368 | 1223 | unsigned int len; |
6d2c2047 | 1224 | |
93c80368 | 1225 | switch (TOKEN_SPELL (token)) |
041c3194 | 1226 | { |
59325650 | 1227 | default: len = 4; break; |
6338b358 | 1228 | case SPELL_LITERAL: len = token->val.str.len; break; |
47e20491 | 1229 | case SPELL_IDENT: len = NODE_LEN (token->val.node) * 10; break; |
041c3194 | 1230 | } |
59325650 NB |
1231 | |
1232 | return len; | |
6d2c2047 ZW |
1233 | } |
1234 | ||
47e20491 GK |
1235 | /* Parse UTF-8 out of NAMEP and place a \U escape in BUFFER. |
1236 | Return the number of bytes read out of NAME. (There are always | |
1237 | 10 bytes written to BUFFER.) */ | |
1238 | ||
1239 | static size_t | |
1240 | utf8_to_ucn (unsigned char *buffer, const unsigned char *name) | |
1241 | { | |
1242 | int j; | |
1243 | int ucn_len = 0; | |
1244 | int ucn_len_c; | |
1245 | unsigned t; | |
1246 | unsigned long utf32; | |
1247 | ||
1248 | /* Compute the length of the UTF-8 sequence. */ | |
1249 | for (t = *name; t & 0x80; t <<= 1) | |
1250 | ucn_len++; | |
1251 | ||
1252 | utf32 = *name & (0x7F >> ucn_len); | |
1253 | for (ucn_len_c = 1; ucn_len_c < ucn_len; ucn_len_c++) | |
1254 | { | |
1255 | utf32 = (utf32 << 6) | (*++name & 0x3F); | |
1256 | ||
1257 | /* Ill-formed UTF-8. */ | |
1258 | if ((*name & ~0x3F) != 0x80) | |
1259 | abort (); | |
1260 | } | |
1261 | ||
1262 | *buffer++ = '\\'; | |
1263 | *buffer++ = 'U'; | |
1264 | for (j = 7; j >= 0; j--) | |
1265 | *buffer++ = "0123456789abcdef"[(utf32 >> (4 * j)) & 0xF]; | |
1266 | return ucn_len; | |
1267 | } | |
1268 | ||
1269 | ||
041c3194 | 1270 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
cf00a885 | 1271 | already contain the enough space to hold the token's spelling. |
6cf87ca4 | 1272 | Returns a pointer to the character after the last character written. |
47e20491 GK |
1273 | FORSTRING is true if this is to be the spelling after translation |
1274 | phase 1 (this is different for UCNs). | |
6cf87ca4 | 1275 | FIXME: Would be nice if we didn't need the PFILE argument. */ |
93c80368 | 1276 | unsigned char * |
6cf87ca4 | 1277 | cpp_spell_token (cpp_reader *pfile, const cpp_token *token, |
47e20491 | 1278 | unsigned char *buffer, bool forstring) |
041c3194 | 1279 | { |
96be6998 | 1280 | switch (TOKEN_SPELL (token)) |
041c3194 ZW |
1281 | { |
1282 | case SPELL_OPERATOR: | |
1283 | { | |
1284 | const unsigned char *spelling; | |
1285 | unsigned char c; | |
d6d5f795 | 1286 | |
041c3194 | 1287 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1288 | spelling |
1289 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
92936ecf ZW |
1290 | else if (token->flags & NAMED_OP) |
1291 | goto spell_ident; | |
041c3194 | 1292 | else |
96be6998 | 1293 | spelling = TOKEN_NAME (token); |
df383483 | 1294 | |
041c3194 ZW |
1295 | while ((c = *spelling++) != '\0') |
1296 | *buffer++ = c; | |
1297 | } | |
1298 | break; | |
d6d5f795 | 1299 | |
47ad4138 | 1300 | spell_ident: |
041c3194 | 1301 | case SPELL_IDENT: |
47e20491 GK |
1302 | if (forstring) |
1303 | { | |
1304 | memcpy (buffer, NODE_NAME (token->val.node), | |
1305 | NODE_LEN (token->val.node)); | |
1306 | buffer += NODE_LEN (token->val.node); | |
1307 | } | |
1308 | else | |
1309 | { | |
1310 | size_t i; | |
1311 | const unsigned char * name = NODE_NAME (token->val.node); | |
1312 | ||
1313 | for (i = 0; i < NODE_LEN (token->val.node); i++) | |
1314 | if (name[i] & ~0x7F) | |
1315 | { | |
1316 | i += utf8_to_ucn (buffer, name + i) - 1; | |
1317 | buffer += 10; | |
1318 | } | |
1319 | else | |
1320 | *buffer++ = NODE_NAME (token->val.node)[i]; | |
1321 | } | |
041c3194 | 1322 | break; |
d6d5f795 | 1323 | |
6338b358 | 1324 | case SPELL_LITERAL: |
47ad4138 ZW |
1325 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1326 | buffer += token->val.str.len; | |
1327 | break; | |
1328 | ||
041c3194 | 1329 | case SPELL_NONE: |
0527bc4e JDA |
1330 | cpp_error (pfile, CPP_DL_ICE, |
1331 | "unspellable token %s", TOKEN_NAME (token)); | |
041c3194 ZW |
1332 | break; |
1333 | } | |
d6d5f795 | 1334 | |
041c3194 ZW |
1335 | return buffer; |
1336 | } | |
d6d5f795 | 1337 | |
5d8ebbd8 NB |
1338 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1339 | freed when the reader is destroyed. Useful for diagnostics. */ | |
93c80368 | 1340 | unsigned char * |
6cf87ca4 | 1341 | cpp_token_as_text (cpp_reader *pfile, const cpp_token *token) |
59325650 NB |
1342 | { |
1343 | unsigned int len = cpp_token_len (token) + 1; | |
ece54d54 | 1344 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
c5a04734 | 1345 | |
47e20491 | 1346 | end = cpp_spell_token (pfile, token, start, false); |
93c80368 | 1347 | end[0] = '\0'; |
c5a04734 | 1348 | |
93c80368 NB |
1349 | return start; |
1350 | } | |
c5a04734 | 1351 | |
5d8ebbd8 NB |
1352 | /* Used by C front ends, which really should move to using |
1353 | cpp_token_as_text. */ | |
93c80368 | 1354 | const char * |
6cf87ca4 | 1355 | cpp_type2name (enum cpp_ttype type) |
93c80368 NB |
1356 | { |
1357 | return (const char *) token_spellings[type].name; | |
1358 | } | |
c5a04734 | 1359 | |
4ed5bcfb NB |
1360 | /* Writes the spelling of token to FP, without any preceding space. |
1361 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1362 | double-buffering. */ | |
93c80368 | 1363 | void |
6cf87ca4 | 1364 | cpp_output_token (const cpp_token *token, FILE *fp) |
93c80368 | 1365 | { |
93c80368 | 1366 | switch (TOKEN_SPELL (token)) |
c5a04734 | 1367 | { |
93c80368 NB |
1368 | case SPELL_OPERATOR: |
1369 | { | |
1370 | const unsigned char *spelling; | |
3b681e9d | 1371 | int c; |
c5a04734 | 1372 | |
93c80368 | 1373 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1374 | spelling |
1375 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
93c80368 NB |
1376 | else if (token->flags & NAMED_OP) |
1377 | goto spell_ident; | |
1378 | else | |
1379 | spelling = TOKEN_NAME (token); | |
041c3194 | 1380 | |
3b681e9d ZW |
1381 | c = *spelling; |
1382 | do | |
1383 | putc (c, fp); | |
1384 | while ((c = *++spelling) != '\0'); | |
93c80368 NB |
1385 | } |
1386 | break; | |
041c3194 | 1387 | |
93c80368 NB |
1388 | spell_ident: |
1389 | case SPELL_IDENT: | |
47e20491 GK |
1390 | { |
1391 | size_t i; | |
1392 | const unsigned char * name = NODE_NAME (token->val.node); | |
1393 | ||
1394 | for (i = 0; i < NODE_LEN (token->val.node); i++) | |
1395 | if (name[i] & ~0x7F) | |
1396 | { | |
1397 | unsigned char buffer[10]; | |
1398 | i += utf8_to_ucn (buffer, name + i) - 1; | |
1399 | fwrite (buffer, 1, 10, fp); | |
1400 | } | |
1401 | else | |
1402 | fputc (NODE_NAME (token->val.node)[i], fp); | |
1403 | } | |
1404 | break; | |
041c3194 | 1405 | |
6338b358 | 1406 | case SPELL_LITERAL: |
47ad4138 ZW |
1407 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
1408 | break; | |
1409 | ||
93c80368 NB |
1410 | case SPELL_NONE: |
1411 | /* An error, most probably. */ | |
1412 | break; | |
041c3194 | 1413 | } |
c5a04734 ZW |
1414 | } |
1415 | ||
93c80368 NB |
1416 | /* Compare two tokens. */ |
1417 | int | |
6cf87ca4 | 1418 | _cpp_equiv_tokens (const cpp_token *a, const cpp_token *b) |
c5a04734 | 1419 | { |
93c80368 NB |
1420 | if (a->type == b->type && a->flags == b->flags) |
1421 | switch (TOKEN_SPELL (a)) | |
1422 | { | |
1423 | default: /* Keep compiler happy. */ | |
1424 | case SPELL_OPERATOR: | |
1425 | return 1; | |
93c80368 | 1426 | case SPELL_NONE: |
56051c0a | 1427 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
93c80368 NB |
1428 | case SPELL_IDENT: |
1429 | return a->val.node == b->val.node; | |
6338b358 | 1430 | case SPELL_LITERAL: |
93c80368 NB |
1431 | return (a->val.str.len == b->val.str.len |
1432 | && !memcmp (a->val.str.text, b->val.str.text, | |
1433 | a->val.str.len)); | |
1434 | } | |
c5a04734 | 1435 | |
041c3194 ZW |
1436 | return 0; |
1437 | } | |
1438 | ||
93c80368 NB |
1439 | /* Returns nonzero if a space should be inserted to avoid an |
1440 | accidental token paste for output. For simplicity, it is | |
1441 | conservative, and occasionally advises a space where one is not | |
1442 | needed, e.g. "." and ".2". */ | |
93c80368 | 1443 | int |
6cf87ca4 ZW |
1444 | cpp_avoid_paste (cpp_reader *pfile, const cpp_token *token1, |
1445 | const cpp_token *token2) | |
c5a04734 | 1446 | { |
93c80368 NB |
1447 | enum cpp_ttype a = token1->type, b = token2->type; |
1448 | cppchar_t c; | |
c5a04734 | 1449 | |
93c80368 NB |
1450 | if (token1->flags & NAMED_OP) |
1451 | a = CPP_NAME; | |
1452 | if (token2->flags & NAMED_OP) | |
1453 | b = CPP_NAME; | |
c5a04734 | 1454 | |
93c80368 NB |
1455 | c = EOF; |
1456 | if (token2->flags & DIGRAPH) | |
37b8524c | 1457 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
93c80368 NB |
1458 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1459 | c = token_spellings[b].name[0]; | |
c5a04734 | 1460 | |
93c80368 | 1461 | /* Quickly get everything that can paste with an '='. */ |
37b8524c | 1462 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
93c80368 | 1463 | return 1; |
c5a04734 | 1464 | |
93c80368 | 1465 | switch (a) |
c5a04734 | 1466 | { |
93c80368 NB |
1467 | case CPP_GREATER: return c == '>' || c == '?'; |
1468 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1469 | case CPP_PLUS: return c == '+'; | |
1470 | case CPP_MINUS: return c == '-' || c == '>'; | |
1471 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1472 | case CPP_MOD: return c == ':' || c == '>'; | |
1473 | case CPP_AND: return c == '&'; | |
1474 | case CPP_OR: return c == '|'; | |
1475 | case CPP_COLON: return c == ':' || c == '>'; | |
1476 | case CPP_DEREF: return c == '*'; | |
26ec42ee | 1477 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
93c80368 NB |
1478 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1479 | case CPP_NAME: return ((b == CPP_NUMBER | |
1480 | && name_p (pfile, &token2->val.str)) | |
1481 | || b == CPP_NAME | |
1482 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1483 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1484 | || c == '.' || c == '+' || c == '-'); | |
1613e52b | 1485 | /* UCNs */ |
1067694a NB |
1486 | case CPP_OTHER: return ((token1->val.str.text[0] == '\\' |
1487 | && b == CPP_NAME) | |
1613e52b | 1488 | || (CPP_OPTION (pfile, objc) |
1067694a | 1489 | && token1->val.str.text[0] == '@' |
1613e52b | 1490 | && (b == CPP_NAME || b == CPP_STRING))); |
93c80368 | 1491 | default: break; |
c5a04734 | 1492 | } |
c5a04734 | 1493 | |
417f3e3a | 1494 | return 0; |
c5a04734 ZW |
1495 | } |
1496 | ||
93c80368 | 1497 | /* Output all the remaining tokens on the current line, and a newline |
4ed5bcfb NB |
1498 | character, to FP. Leading whitespace is removed. If there are |
1499 | macros, special token padding is not performed. */ | |
c5a04734 | 1500 | void |
6cf87ca4 | 1501 | cpp_output_line (cpp_reader *pfile, FILE *fp) |
c5a04734 | 1502 | { |
4ed5bcfb | 1503 | const cpp_token *token; |
96be6998 | 1504 | |
4ed5bcfb NB |
1505 | token = cpp_get_token (pfile); |
1506 | while (token->type != CPP_EOF) | |
96be6998 | 1507 | { |
4ed5bcfb NB |
1508 | cpp_output_token (token, fp); |
1509 | token = cpp_get_token (pfile); | |
1510 | if (token->flags & PREV_WHITE) | |
1511 | putc (' ', fp); | |
96be6998 ZW |
1512 | } |
1513 | ||
93c80368 | 1514 | putc ('\n', fp); |
041c3194 | 1515 | } |
c5a04734 | 1516 | |
1e013d2e NB |
1517 | /* Memory buffers. Changing these three constants can have a dramatic |
1518 | effect on performance. The values here are reasonable defaults, | |
1519 | but might be tuned. If you adjust them, be sure to test across a | |
1520 | range of uses of cpplib, including heavy nested function-like macro | |
1521 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1522 | good tool for this). */ | |
1523 | #define MIN_BUFF_SIZE 8000 | |
87062813 | 1524 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
1e013d2e NB |
1525 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1526 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
417f3e3a | 1527 | |
87062813 NB |
1528 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1529 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1530 | #endif | |
1531 | ||
c9e7a609 NB |
1532 | /* Create a new allocation buffer. Place the control block at the end |
1533 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
b8af0ca5 | 1534 | static _cpp_buff * |
6cf87ca4 | 1535 | new_buff (size_t len) |
b8af0ca5 NB |
1536 | { |
1537 | _cpp_buff *result; | |
ece54d54 | 1538 | unsigned char *base; |
b8af0ca5 | 1539 | |
1e013d2e NB |
1540 | if (len < MIN_BUFF_SIZE) |
1541 | len = MIN_BUFF_SIZE; | |
c70f6ed3 | 1542 | len = CPP_ALIGN (len); |
b8af0ca5 NB |
1543 | |
1544 | base = xmalloc (len + sizeof (_cpp_buff)); | |
1545 | result = (_cpp_buff *) (base + len); | |
1546 | result->base = base; | |
1547 | result->cur = base; | |
1548 | result->limit = base + len; | |
1549 | result->next = NULL; | |
1550 | return result; | |
1551 | } | |
1552 | ||
1553 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1554 | void | |
6cf87ca4 | 1555 | _cpp_release_buff (cpp_reader *pfile, _cpp_buff *buff) |
b8af0ca5 NB |
1556 | { |
1557 | _cpp_buff *end = buff; | |
1558 | ||
1559 | while (end->next) | |
1560 | end = end->next; | |
1561 | end->next = pfile->free_buffs; | |
1562 | pfile->free_buffs = buff; | |
1563 | } | |
1564 | ||
1565 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1566 | _cpp_buff * | |
6cf87ca4 | 1567 | _cpp_get_buff (cpp_reader *pfile, size_t min_size) |
b8af0ca5 NB |
1568 | { |
1569 | _cpp_buff *result, **p; | |
1570 | ||
1571 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1572 | { | |
6142088c | 1573 | size_t size; |
1e013d2e NB |
1574 | |
1575 | if (*p == NULL) | |
b8af0ca5 | 1576 | return new_buff (min_size); |
1e013d2e NB |
1577 | result = *p; |
1578 | size = result->limit - result->base; | |
1579 | /* Return a buffer that's big enough, but don't waste one that's | |
1580 | way too big. */ | |
34f5271d | 1581 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
b8af0ca5 NB |
1582 | break; |
1583 | } | |
1584 | ||
1585 | *p = result->next; | |
1586 | result->next = NULL; | |
1587 | result->cur = result->base; | |
1588 | return result; | |
1589 | } | |
1590 | ||
4fe9b91c | 1591 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
1592 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
1593 | the excess bytes to the new buffer. Chains the new buffer after | |
1594 | BUFF, and returns the new buffer. */ | |
b8af0ca5 | 1595 | _cpp_buff * |
6cf87ca4 | 1596 | _cpp_append_extend_buff (cpp_reader *pfile, _cpp_buff *buff, size_t min_extra) |
b8af0ca5 | 1597 | { |
6142088c | 1598 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
8c3b2693 | 1599 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
b8af0ca5 | 1600 | |
8c3b2693 NB |
1601 | buff->next = new_buff; |
1602 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
1603 | return new_buff; | |
1604 | } | |
1605 | ||
4fe9b91c | 1606 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
1607 | remaining bytes of the buffer pointed to by BUFF, and at least |
1608 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
1609 | Chains the new buffer before the buffer pointed to by BUFF, and | |
1610 | updates the pointer to point to the new buffer. */ | |
1611 | void | |
6cf87ca4 | 1612 | _cpp_extend_buff (cpp_reader *pfile, _cpp_buff **pbuff, size_t min_extra) |
8c3b2693 NB |
1613 | { |
1614 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
1615 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
1616 | ||
1617 | new_buff = _cpp_get_buff (pfile, size); | |
1618 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
1619 | new_buff->next = old_buff; | |
1620 | *pbuff = new_buff; | |
b8af0ca5 NB |
1621 | } |
1622 | ||
1623 | /* Free a chain of buffers starting at BUFF. */ | |
1624 | void | |
5671bf27 | 1625 | _cpp_free_buff (_cpp_buff *buff) |
b8af0ca5 NB |
1626 | { |
1627 | _cpp_buff *next; | |
1628 | ||
1629 | for (; buff; buff = next) | |
1630 | { | |
1631 | next = buff->next; | |
1632 | free (buff->base); | |
1633 | } | |
1634 | } | |
417f3e3a | 1635 | |
ece54d54 NB |
1636 | /* Allocate permanent, unaligned storage of length LEN. */ |
1637 | unsigned char * | |
6cf87ca4 | 1638 | _cpp_unaligned_alloc (cpp_reader *pfile, size_t len) |
ece54d54 NB |
1639 | { |
1640 | _cpp_buff *buff = pfile->u_buff; | |
1641 | unsigned char *result = buff->cur; | |
1642 | ||
1643 | if (len > (size_t) (buff->limit - result)) | |
1644 | { | |
1645 | buff = _cpp_get_buff (pfile, len); | |
1646 | buff->next = pfile->u_buff; | |
1647 | pfile->u_buff = buff; | |
1648 | result = buff->cur; | |
1649 | } | |
1650 | ||
1651 | buff->cur = result + len; | |
1652 | return result; | |
1653 | } | |
1654 | ||
87062813 NB |
1655 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
1656 | That buffer is used for growing allocations when saving macro | |
1657 | replacement lists in a #define, and when parsing an answer to an | |
1658 | assertion in #assert, #unassert or #if (and therefore possibly | |
1659 | whilst expanding macros). It therefore must not be used by any | |
1660 | code that they might call: specifically the lexer and the guts of | |
1661 | the macro expander. | |
1662 | ||
1663 | All existing other uses clearly fit this restriction: storing | |
1664 | registered pragmas during initialization. */ | |
93c80368 | 1665 | unsigned char * |
6cf87ca4 | 1666 | _cpp_aligned_alloc (cpp_reader *pfile, size_t len) |
3fef5b2b | 1667 | { |
8c3b2693 NB |
1668 | _cpp_buff *buff = pfile->a_buff; |
1669 | unsigned char *result = buff->cur; | |
3fef5b2b | 1670 | |
8c3b2693 | 1671 | if (len > (size_t) (buff->limit - result)) |
3fef5b2b | 1672 | { |
8c3b2693 NB |
1673 | buff = _cpp_get_buff (pfile, len); |
1674 | buff->next = pfile->a_buff; | |
1675 | pfile->a_buff = buff; | |
1676 | result = buff->cur; | |
3fef5b2b | 1677 | } |
041c3194 | 1678 | |
8c3b2693 | 1679 | buff->cur = result + len; |
93c80368 | 1680 | return result; |
041c3194 | 1681 | } |
d8044160 GK |
1682 | |
1683 | /* Say which field of TOK is in use. */ | |
1684 | ||
1685 | enum cpp_token_fld_kind | |
1686 | cpp_token_val_index (cpp_token *tok) | |
1687 | { | |
1688 | switch (TOKEN_SPELL (tok)) | |
1689 | { | |
1690 | case SPELL_IDENT: | |
1691 | return CPP_TOKEN_FLD_NODE; | |
1692 | case SPELL_LITERAL: | |
1693 | return CPP_TOKEN_FLD_STR; | |
1694 | case SPELL_NONE: | |
1695 | if (tok->type == CPP_MACRO_ARG) | |
1696 | return CPP_TOKEN_FLD_ARG_NO; | |
1697 | else if (tok->type == CPP_PADDING) | |
1698 | return CPP_TOKEN_FLD_SOURCE; | |
21b11495 ZW |
1699 | else if (tok->type == CPP_PRAGMA) |
1700 | return CPP_TOKEN_FLD_STR; | |
d8044160 GK |
1701 | /* else fall through */ |
1702 | default: | |
1703 | return CPP_TOKEN_FLD_NONE; | |
1704 | } | |
1705 | } |