]>
Commit | Line | Data |
---|---|---|
0578f103 | 1 | /* CPP Library - lexical analysis. |
6bc9506f | 2 | Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005, 2007, 2008, 2009 |
3 | Free Software Foundation, Inc. | |
0578f103 | 4 | Contributed by Per Bothner, 1994-95. |
5 | Based on CCCP program by Paul Rubin, June 1986 | |
6 | Adapted to ANSI C, Richard Stallman, Jan 1987 | |
7 | Broken out to separate file, Zack Weinberg, Mar 2000 | |
8 | ||
9 | This program is free software; you can redistribute it and/or modify it | |
10 | under the terms of the GNU General Public License as published by the | |
6bc9506f | 11 | Free Software Foundation; either version 3, or (at your option) any |
0578f103 | 12 | later version. |
13 | ||
14 | This program is distributed in the hope that it will be useful, | |
15 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
17 | GNU General Public License for more details. | |
18 | ||
19 | You should have received a copy of the GNU General Public License | |
6bc9506f | 20 | along with this program; see the file COPYING3. If not see |
21 | <http://www.gnu.org/licenses/>. */ | |
0578f103 | 22 | |
23 | #include "config.h" | |
24 | #include "system.h" | |
0578f103 | 25 | #include "cpplib.h" |
d856c8a6 | 26 | #include "internal.h" |
0578f103 | 27 | |
79bd622b | 28 | enum spell_type |
241e762e | 29 | { |
79bd622b | 30 | SPELL_OPERATOR = 0, |
79bd622b | 31 | SPELL_IDENT, |
4970d4c2 | 32 | SPELL_LITERAL, |
79bd622b | 33 | SPELL_NONE |
241e762e | 34 | }; |
35 | ||
79bd622b | 36 | struct token_spelling |
241e762e | 37 | { |
79bd622b | 38 | enum spell_type category; |
39 | const unsigned char *name; | |
241e762e | 40 | }; |
41 | ||
0ca849f9 | 42 | static const unsigned char *const digraph_spellings[] = |
924bbf02 | 43 | { UC"%:", UC"%:%:", UC"<:", UC":>", UC"<%", UC"%>" }; |
79bd622b | 44 | |
924bbf02 | 45 | #define OP(e, s) { SPELL_OPERATOR, UC s }, |
46 | #define TK(e, s) { SPELL_ ## s, UC #e }, | |
0ca849f9 | 47 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
79bd622b | 48 | #undef OP |
49 | #undef TK | |
50 | ||
51 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
52 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
e2f9a79f | 53 | |
f7fdd7a1 | 54 | static void add_line_note (cpp_buffer *, const uchar *, unsigned int); |
55 | static int skip_line_comment (cpp_reader *); | |
56 | static void skip_whitespace (cpp_reader *, cppchar_t); | |
f7fdd7a1 | 57 | static void lex_string (cpp_reader *, cpp_token *, const uchar *); |
58 | static void save_comment (cpp_reader *, cpp_token *, const uchar *, cppchar_t); | |
956c6108 | 59 | static void store_comment (cpp_reader *, cpp_token *); |
f7fdd7a1 | 60 | static void create_literal (cpp_reader *, cpp_token *, const uchar *, |
61 | unsigned int, enum cpp_ttype); | |
62 | static bool warn_in_comment (cpp_reader *, _cpp_line_note *); | |
63 | static int name_p (cpp_reader *, const cpp_string *); | |
f7fdd7a1 | 64 | static tokenrun *next_tokenrun (tokenrun *); |
65 | ||
f7fdd7a1 | 66 | static _cpp_buff *new_buff (size_t); |
bce8e0c0 | 67 | |
e920deaf | 68 | |
f80e83a9 | 69 | /* Utility routine: |
2c63d6c8 | 70 | |
76faa4c0 | 71 | Compares, the token TOKEN to the NUL-terminated string STRING. |
72 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
f80e83a9 | 73 | int |
f7fdd7a1 | 74 | cpp_ideq (const cpp_token *token, const char *string) |
f80e83a9 | 75 | { |
76faa4c0 | 76 | if (token->type != CPP_NAME) |
f80e83a9 | 77 | return 0; |
76faa4c0 | 78 | |
b6d18b0a | 79 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
bce8e0c0 | 80 | } |
50fd6b48 | 81 | |
a54e0bf8 | 82 | /* Record a note TYPE at byte POS into the current cleaned logical |
83 | line. */ | |
1e0ef2fd | 84 | static void |
f7fdd7a1 | 85 | add_line_note (cpp_buffer *buffer, const uchar *pos, unsigned int type) |
338fa5f7 | 86 | { |
a54e0bf8 | 87 | if (buffer->notes_used == buffer->notes_cap) |
88 | { | |
89 | buffer->notes_cap = buffer->notes_cap * 2 + 200; | |
720aca92 | 90 | buffer->notes = XRESIZEVEC (_cpp_line_note, buffer->notes, |
91 | buffer->notes_cap); | |
a54e0bf8 | 92 | } |
338fa5f7 | 93 | |
a54e0bf8 | 94 | buffer->notes[buffer->notes_used].pos = pos; |
95 | buffer->notes[buffer->notes_used].type = type; | |
96 | buffer->notes_used++; | |
338fa5f7 | 97 | } |
98 | ||
a54e0bf8 | 99 | /* Returns with a logical line that contains no escaped newlines or |
100 | trigraphs. This is a time-critical inner loop. */ | |
101 | void | |
f7fdd7a1 | 102 | _cpp_clean_line (cpp_reader *pfile) |
0578f103 | 103 | { |
a54e0bf8 | 104 | cpp_buffer *buffer; |
105 | const uchar *s; | |
106 | uchar c, *d, *p; | |
1e0ef2fd | 107 | |
a54e0bf8 | 108 | buffer = pfile->buffer; |
109 | buffer->cur_note = buffer->notes_used = 0; | |
110 | buffer->cur = buffer->line_base = buffer->next_line; | |
111 | buffer->need_line = false; | |
112 | s = buffer->next_line - 1; | |
1e0ef2fd | 113 | |
a54e0bf8 | 114 | if (!buffer->from_stage3) |
0578f103 | 115 | { |
5008f5c5 | 116 | const uchar *pbackslash = NULL; |
117 | ||
54d3be91 | 118 | /* Short circuit for the common case of an un-escaped line with |
119 | no trigraphs. The primary win here is by not writing any | |
120 | data back to memory until we have to. */ | |
121 | for (;;) | |
122 | { | |
123 | c = *++s; | |
5008f5c5 | 124 | if (__builtin_expect (c == '\n', false) |
125 | || __builtin_expect (c == '\r', false)) | |
54d3be91 | 126 | { |
127 | d = (uchar *) s; | |
128 | ||
5008f5c5 | 129 | if (__builtin_expect (s == buffer->rlimit, false)) |
54d3be91 | 130 | goto done; |
131 | ||
132 | /* DOS line ending? */ | |
5008f5c5 | 133 | if (__builtin_expect (c == '\r', false) |
134 | && s[1] == '\n') | |
135 | { | |
136 | s++; | |
137 | if (s == buffer->rlimit) | |
138 | goto done; | |
139 | } | |
54d3be91 | 140 | |
5008f5c5 | 141 | if (__builtin_expect (pbackslash == NULL, true)) |
54d3be91 | 142 | goto done; |
143 | ||
5008f5c5 | 144 | /* Check for escaped newline. */ |
54d3be91 | 145 | p = d; |
5008f5c5 | 146 | while (is_nvspace (p[-1])) |
54d3be91 | 147 | p--; |
5008f5c5 | 148 | if (p - 1 != pbackslash) |
54d3be91 | 149 | goto done; |
150 | ||
151 | /* Have an escaped newline; process it and proceed to | |
152 | the slow path. */ | |
153 | add_line_note (buffer, p - 1, p != d ? ' ' : '\\'); | |
154 | d = p - 2; | |
155 | buffer->next_line = p - 1; | |
156 | break; | |
157 | } | |
5008f5c5 | 158 | if (__builtin_expect (c == '\\', false)) |
159 | pbackslash = s; | |
160 | else if (__builtin_expect (c == '?', false) | |
161 | && __builtin_expect (s[1] == '?', false) | |
162 | && _cpp_trigraph_map[s[2]]) | |
54d3be91 | 163 | { |
164 | /* Have a trigraph. We may or may not have to convert | |
165 | it. Add a line note regardless, for -Wtrigraphs. */ | |
166 | add_line_note (buffer, s, s[2]); | |
167 | if (CPP_OPTION (pfile, trigraphs)) | |
168 | { | |
169 | /* We do, and that means we have to switch to the | |
170 | slow path. */ | |
171 | d = (uchar *) s; | |
172 | *d = _cpp_trigraph_map[s[2]]; | |
173 | s += 2; | |
174 | break; | |
175 | } | |
176 | } | |
177 | } | |
178 | ||
a54e0bf8 | 179 | |
180 | for (;;) | |
4b912310 | 181 | { |
a54e0bf8 | 182 | c = *++s; |
183 | *++d = c; | |
184 | ||
185 | if (c == '\n' || c == '\r') | |
186 | { | |
187 | /* Handle DOS line endings. */ | |
188 | if (c == '\r' && s != buffer->rlimit && s[1] == '\n') | |
189 | s++; | |
190 | if (s == buffer->rlimit) | |
191 | break; | |
192 | ||
193 | /* Escaped? */ | |
194 | p = d; | |
195 | while (p != buffer->next_line && is_nvspace (p[-1])) | |
196 | p--; | |
197 | if (p == buffer->next_line || p[-1] != '\\') | |
198 | break; | |
199 | ||
aad4a87f | 200 | add_line_note (buffer, p - 1, p != d ? ' ': '\\'); |
a54e0bf8 | 201 | d = p - 2; |
202 | buffer->next_line = p - 1; | |
203 | } | |
204 | else if (c == '?' && s[1] == '?' && _cpp_trigraph_map[s[2]]) | |
205 | { | |
206 | /* Add a note regardless, for the benefit of -Wtrigraphs. */ | |
aad4a87f | 207 | add_line_note (buffer, d, s[2]); |
a54e0bf8 | 208 | if (CPP_OPTION (pfile, trigraphs)) |
209 | { | |
210 | *d = _cpp_trigraph_map[s[2]]; | |
211 | s += 2; | |
212 | } | |
213 | } | |
4b912310 | 214 | } |
0578f103 | 215 | } |
a54e0bf8 | 216 | else |
217 | { | |
218 | do | |
219 | s++; | |
220 | while (*s != '\n' && *s != '\r'); | |
221 | d = (uchar *) s; | |
222 | ||
223 | /* Handle DOS line endings. */ | |
224 | if (*s == '\r' && s != buffer->rlimit && s[1] == '\n') | |
225 | s++; | |
226 | } | |
338fa5f7 | 227 | |
54d3be91 | 228 | done: |
a54e0bf8 | 229 | *d = '\n'; |
aad4a87f | 230 | /* A sentinel note that should never be processed. */ |
231 | add_line_note (buffer, d + 1, '\n'); | |
a54e0bf8 | 232 | buffer->next_line = s + 1; |
0578f103 | 233 | } |
234 | ||
3078f2b2 | 235 | /* Return true if the trigraph indicated by NOTE should be warned |
236 | about in a comment. */ | |
237 | static bool | |
f7fdd7a1 | 238 | warn_in_comment (cpp_reader *pfile, _cpp_line_note *note) |
3078f2b2 | 239 | { |
240 | const uchar *p; | |
241 | ||
242 | /* Within comments we don't warn about trigraphs, unless the | |
243 | trigraph forms an escaped newline, as that may change | |
7ef5b942 | 244 | behavior. */ |
3078f2b2 | 245 | if (note->type != '/') |
246 | return false; | |
247 | ||
248 | /* If -trigraphs, then this was an escaped newline iff the next note | |
249 | is coincident. */ | |
250 | if (CPP_OPTION (pfile, trigraphs)) | |
251 | return note[1].pos == note->pos; | |
252 | ||
253 | /* Otherwise, see if this forms an escaped newline. */ | |
254 | p = note->pos + 3; | |
255 | while (is_nvspace (*p)) | |
256 | p++; | |
257 | ||
258 | /* There might have been escaped newlines between the trigraph and the | |
259 | newline we found. Hence the position test. */ | |
260 | return (*p == '\n' && p < note[1].pos); | |
261 | } | |
262 | ||
a54e0bf8 | 263 | /* Process the notes created by add_line_note as far as the current |
264 | location. */ | |
265 | void | |
f7fdd7a1 | 266 | _cpp_process_line_notes (cpp_reader *pfile, int in_comment) |
0578f103 | 267 | { |
c808d026 | 268 | cpp_buffer *buffer = pfile->buffer; |
269 | ||
a54e0bf8 | 270 | for (;;) |
f80e83a9 | 271 | { |
a54e0bf8 | 272 | _cpp_line_note *note = &buffer->notes[buffer->cur_note]; |
273 | unsigned int col; | |
396ffa86 | 274 | |
a54e0bf8 | 275 | if (note->pos > buffer->cur) |
276 | break; | |
396ffa86 | 277 | |
a54e0bf8 | 278 | buffer->cur_note++; |
279 | col = CPP_BUF_COLUMN (buffer, note->pos + 1); | |
435fb09b | 280 | |
aad4a87f | 281 | if (note->type == '\\' || note->type == ' ') |
a54e0bf8 | 282 | { |
aad4a87f | 283 | if (note->type == ' ' && !in_comment) |
dbddc569 | 284 | cpp_error_with_line (pfile, CPP_DL_WARNING, pfile->line_table->highest_line, col, |
a54e0bf8 | 285 | "backslash and newline separated by space"); |
aad4a87f | 286 | |
a54e0bf8 | 287 | if (buffer->next_line > buffer->rlimit) |
1e0ef2fd | 288 | { |
dbddc569 | 289 | cpp_error_with_line (pfile, CPP_DL_PEDWARN, pfile->line_table->highest_line, col, |
a54e0bf8 | 290 | "backslash-newline at end of file"); |
291 | /* Prevent "no newline at end of file" warning. */ | |
292 | buffer->next_line = buffer->rlimit; | |
1e0ef2fd | 293 | } |
a54e0bf8 | 294 | |
295 | buffer->line_base = note->pos; | |
610625e3 | 296 | CPP_INCREMENT_LINE (pfile, 0); |
338fa5f7 | 297 | } |
aad4a87f | 298 | else if (_cpp_trigraph_map[note->type]) |
299 | { | |
3078f2b2 | 300 | if (CPP_OPTION (pfile, warn_trigraphs) |
301 | && (!in_comment || warn_in_comment (pfile, note))) | |
aad4a87f | 302 | { |
303 | if (CPP_OPTION (pfile, trigraphs)) | |
dbddc569 | 304 | cpp_error_with_line (pfile, CPP_DL_WARNING, pfile->line_table->highest_line, col, |
aad4a87f | 305 | "trigraph ??%c converted to %c", |
306 | note->type, | |
307 | (int) _cpp_trigraph_map[note->type]); | |
308 | else | |
1542b1ef | 309 | { |
310 | cpp_error_with_line | |
dbddc569 | 311 | (pfile, CPP_DL_WARNING, pfile->line_table->highest_line, col, |
1542b1ef | 312 | "trigraph ??%c ignored, use -trigraphs to enable", |
313 | note->type); | |
314 | } | |
aad4a87f | 315 | } |
316 | } | |
317 | else | |
318 | abort (); | |
f80e83a9 | 319 | } |
0578f103 | 320 | } |
321 | ||
338fa5f7 | 322 | /* Skip a C-style block comment. We find the end of the comment by |
323 | seeing if an asterisk is before every '/' we encounter. Returns | |
edaf8cb5 | 324 | nonzero if comment terminated by EOF, zero otherwise. |
325 | ||
326 | Buffer->cur points to the initial asterisk of the comment. */ | |
a54e0bf8 | 327 | bool |
f7fdd7a1 | 328 | _cpp_skip_block_comment (cpp_reader *pfile) |
0578f103 | 329 | { |
f80e83a9 | 330 | cpp_buffer *buffer = pfile->buffer; |
54d3be91 | 331 | const uchar *cur = buffer->cur; |
332 | uchar c; | |
338fa5f7 | 333 | |
54d3be91 | 334 | cur++; |
335 | if (*cur == '/') | |
336 | cur++; | |
338fa5f7 | 337 | |
a54e0bf8 | 338 | for (;;) |
339 | { | |
338fa5f7 | 340 | /* People like decorating comments with '*', so check for '/' |
341 | instead for efficiency. */ | |
54d3be91 | 342 | c = *cur++; |
343 | ||
f80e83a9 | 344 | if (c == '/') |
0578f103 | 345 | { |
54d3be91 | 346 | if (cur[-2] == '*') |
338fa5f7 | 347 | break; |
f80e83a9 | 348 | |
338fa5f7 | 349 | /* Warn about potential nested comments, but not if the '/' |
3fb1e43b | 350 | comes immediately before the true comment delimiter. |
f80e83a9 | 351 | Don't bother to get it right across escaped newlines. */ |
338fa5f7 | 352 | if (CPP_OPTION (pfile, warn_comments) |
54d3be91 | 353 | && cur[0] == '*' && cur[1] != '/') |
354 | { | |
355 | buffer->cur = cur; | |
d80d2074 | 356 | cpp_error_with_line (pfile, CPP_DL_WARNING, |
dbddc569 | 357 | pfile->line_table->highest_line, CPP_BUF_COL (buffer), |
54d3be91 | 358 | "\"/*\" within comment"); |
359 | } | |
0578f103 | 360 | } |
a54e0bf8 | 361 | else if (c == '\n') |
362 | { | |
610625e3 | 363 | unsigned int cols; |
54d3be91 | 364 | buffer->cur = cur - 1; |
a54e0bf8 | 365 | _cpp_process_line_notes (pfile, true); |
366 | if (buffer->next_line >= buffer->rlimit) | |
367 | return true; | |
368 | _cpp_clean_line (pfile); | |
610625e3 | 369 | |
370 | cols = buffer->next_line - buffer->line_base; | |
371 | CPP_INCREMENT_LINE (pfile, cols); | |
372 | ||
54d3be91 | 373 | cur = buffer->cur; |
a54e0bf8 | 374 | } |
0578f103 | 375 | } |
f80e83a9 | 376 | |
54d3be91 | 377 | buffer->cur = cur; |
3078f2b2 | 378 | _cpp_process_line_notes (pfile, true); |
a54e0bf8 | 379 | return false; |
0578f103 | 380 | } |
381 | ||
1c124f85 | 382 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
d10cfa8d | 383 | terminating newline. Handles escaped newlines. Returns nonzero |
1c124f85 | 384 | if a multiline comment. */ |
f80e83a9 | 385 | static int |
f7fdd7a1 | 386 | skip_line_comment (cpp_reader *pfile) |
0578f103 | 387 | { |
f669338a | 388 | cpp_buffer *buffer = pfile->buffer; |
4999c35b | 389 | source_location orig_line = pfile->line_table->highest_line; |
f80e83a9 | 390 | |
a54e0bf8 | 391 | while (*buffer->cur != '\n') |
392 | buffer->cur++; | |
1c124f85 | 393 | |
a54e0bf8 | 394 | _cpp_process_line_notes (pfile, true); |
dbddc569 | 395 | return orig_line != pfile->line_table->highest_line; |
f80e83a9 | 396 | } |
0578f103 | 397 | |
a54e0bf8 | 398 | /* Skips whitespace, saving the next non-whitespace character. */ |
b86584f6 | 399 | static void |
f7fdd7a1 | 400 | skip_whitespace (cpp_reader *pfile, cppchar_t c) |
f80e83a9 | 401 | { |
402 | cpp_buffer *buffer = pfile->buffer; | |
fe9eb18b | 403 | bool saw_NUL = false; |
0578f103 | 404 | |
338fa5f7 | 405 | do |
f80e83a9 | 406 | { |
78719282 | 407 | /* Horizontal space always OK. */ |
a54e0bf8 | 408 | if (c == ' ' || c == '\t') |
338fa5f7 | 409 | ; |
338fa5f7 | 410 | /* Just \f \v or \0 left. */ |
78719282 | 411 | else if (c == '\0') |
fe9eb18b | 412 | saw_NUL = true; |
79bd622b | 413 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
dbddc569 | 414 | cpp_error_with_line (pfile, CPP_DL_PEDWARN, pfile->line_table->highest_line, |
73328dce | 415 | CPP_BUF_COL (buffer), |
416 | "%s in preprocessing directive", | |
417 | c == '\f' ? "form feed" : "vertical tab"); | |
338fa5f7 | 418 | |
338fa5f7 | 419 | c = *buffer->cur++; |
0578f103 | 420 | } |
2c0e001b | 421 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
338fa5f7 | 422 | while (is_nvspace (c)); |
423 | ||
fe9eb18b | 424 | if (saw_NUL) |
d80d2074 | 425 | cpp_error (pfile, CPP_DL_WARNING, "null character(s) ignored"); |
fe9eb18b | 426 | |
1c124f85 | 427 | buffer->cur--; |
f80e83a9 | 428 | } |
0578f103 | 429 | |
79bd622b | 430 | /* See if the characters of a number token are valid in a name (no |
431 | '.', '+' or '-'). */ | |
432 | static int | |
f7fdd7a1 | 433 | name_p (cpp_reader *pfile, const cpp_string *string) |
79bd622b | 434 | { |
435 | unsigned int i; | |
436 | ||
437 | for (i = 0; i < string->len; i++) | |
438 | if (!is_idchar (string->text[i])) | |
439 | return 0; | |
440 | ||
b1a9ff83 | 441 | return 1; |
79bd622b | 442 | } |
443 | ||
bce47149 | 444 | /* After parsing an identifier or other sequence, produce a warning about |
445 | sequences not in NFC/NFKC. */ | |
446 | static void | |
447 | warn_about_normalization (cpp_reader *pfile, | |
448 | const cpp_token *token, | |
449 | const struct normalize_state *s) | |
450 | { | |
451 | if (CPP_OPTION (pfile, warn_normalize) < NORMALIZE_STATE_RESULT (s) | |
452 | && !pfile->state.skipping) | |
453 | { | |
454 | /* Make sure that the token is printed using UCNs, even | |
455 | if we'd otherwise happily print UTF-8. */ | |
720aca92 | 456 | unsigned char *buf = XNEWVEC (unsigned char, cpp_token_len (token)); |
bce47149 | 457 | size_t sz; |
458 | ||
459 | sz = cpp_spell_token (pfile, token, buf, false) - buf; | |
460 | if (NORMALIZE_STATE_RESULT (s) == normalized_C) | |
461 | cpp_error_with_line (pfile, CPP_DL_WARNING, token->src_loc, 0, | |
3827dee5 | 462 | "`%.*s' is not in NFKC", (int) sz, buf); |
bce47149 | 463 | else |
464 | cpp_error_with_line (pfile, CPP_DL_WARNING, token->src_loc, 0, | |
3827dee5 | 465 | "`%.*s' is not in NFC", (int) sz, buf); |
bce47149 | 466 | } |
467 | } | |
468 | ||
5bb46c08 | 469 | /* Returns TRUE if the sequence starting at buffer->cur is invalid in |
2cbf1359 | 470 | an identifier. FIRST is TRUE if this starts an identifier. */ |
5bb46c08 | 471 | static bool |
bce47149 | 472 | forms_identifier_p (cpp_reader *pfile, int first, |
473 | struct normalize_state *state) | |
5bb46c08 | 474 | { |
2cbf1359 | 475 | cpp_buffer *buffer = pfile->buffer; |
476 | ||
477 | if (*buffer->cur == '$') | |
478 | { | |
479 | if (!CPP_OPTION (pfile, dollars_in_ident)) | |
480 | return false; | |
481 | ||
482 | buffer->cur++; | |
f0c2775b | 483 | if (CPP_OPTION (pfile, warn_dollars) && !pfile->state.skipping) |
2cbf1359 | 484 | { |
f0c2775b | 485 | CPP_OPTION (pfile, warn_dollars) = 0; |
d80d2074 | 486 | cpp_error (pfile, CPP_DL_PEDWARN, "'$' in identifier or number"); |
2cbf1359 | 487 | } |
488 | ||
489 | return true; | |
490 | } | |
5bb46c08 | 491 | |
2cbf1359 | 492 | /* Is this a syntactically valid UCN? */ |
865c4e44 | 493 | if (CPP_OPTION (pfile, extended_identifiers) |
4e9d1e6d | 494 | && *buffer->cur == '\\' |
2cbf1359 | 495 | && (buffer->cur[1] == 'u' || buffer->cur[1] == 'U')) |
5bb46c08 | 496 | { |
2cbf1359 | 497 | buffer->cur += 2; |
bce47149 | 498 | if (_cpp_valid_ucn (pfile, &buffer->cur, buffer->rlimit, 1 + !first, |
499 | state)) | |
2cbf1359 | 500 | return true; |
501 | buffer->cur -= 2; | |
5bb46c08 | 502 | } |
5bb46c08 | 503 | |
2cbf1359 | 504 | return false; |
5bb46c08 | 505 | } |
506 | ||
507 | /* Lex an identifier starting at BUFFER->CUR - 1. */ | |
338fa5f7 | 508 | static cpp_hashnode * |
bce47149 | 509 | lex_identifier (cpp_reader *pfile, const uchar *base, bool starts_ucn, |
510 | struct normalize_state *nst) | |
0578f103 | 511 | { |
79bd622b | 512 | cpp_hashnode *result; |
bb1fa6bb | 513 | const uchar *cur; |
3eb3f293 | 514 | unsigned int len; |
515 | unsigned int hash = HT_HASHSTEP (0, *base); | |
66a5287e | 516 | |
3eb3f293 | 517 | cur = pfile->buffer->cur; |
bb1fa6bb | 518 | if (! starts_ucn) |
519 | while (ISIDNUM (*cur)) | |
520 | { | |
521 | hash = HT_HASHSTEP (hash, *cur); | |
522 | cur++; | |
523 | } | |
524 | pfile->buffer->cur = cur; | |
bce47149 | 525 | if (starts_ucn || forms_identifier_p (pfile, false, nst)) |
78a11351 | 526 | { |
bb1fa6bb | 527 | /* Slower version for identifiers containing UCNs (or $). */ |
528 | do { | |
529 | while (ISIDNUM (*pfile->buffer->cur)) | |
bce47149 | 530 | { |
531 | pfile->buffer->cur++; | |
532 | NORMALIZE_STATE_UPDATE_IDNUM (nst); | |
533 | } | |
534 | } while (forms_identifier_p (pfile, false, nst)); | |
bb1fa6bb | 535 | result = _cpp_interpret_identifier (pfile, base, |
536 | pfile->buffer->cur - base); | |
66a5287e | 537 | } |
bb1fa6bb | 538 | else |
539 | { | |
540 | len = cur - base; | |
541 | hash = HT_HASHFINISH (hash, len); | |
5bb46c08 | 542 | |
e297899b | 543 | result = CPP_HASHNODE (ht_lookup_with_hash (pfile->hash_table, |
544 | base, len, hash, HT_ALLOC)); | |
bb1fa6bb | 545 | } |
66a5287e | 546 | |
5bb46c08 | 547 | /* Rarely, identifiers require diagnostics when lexed. */ |
66a5287e | 548 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) |
549 | && !pfile->state.skipping, 0)) | |
550 | { | |
551 | /* It is allowed to poison the same identifier twice. */ | |
552 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
d80d2074 | 553 | cpp_error (pfile, CPP_DL_ERROR, "attempt to use poisoned \"%s\"", |
66a5287e | 554 | NODE_NAME (result)); |
555 | ||
556 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
557 | replacement list of a variadic macro. */ | |
558 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
559 | && !pfile->state.va_args_ok) | |
d80d2074 | 560 | cpp_error (pfile, CPP_DL_PEDWARN, |
f7fdd7a1 | 561 | "__VA_ARGS__ can only appear in the expansion" |
562 | " of a C99 variadic macro"); | |
66a5287e | 563 | } |
564 | ||
565 | return result; | |
566 | } | |
567 | ||
5bb46c08 | 568 | /* Lex a number to NUMBER starting at BUFFER->CUR - 1. */ |
0578f103 | 569 | static void |
bce47149 | 570 | lex_number (cpp_reader *pfile, cpp_string *number, |
571 | struct normalize_state *nst) | |
0578f103 | 572 | { |
b6d18b0a | 573 | const uchar *cur; |
5bb46c08 | 574 | const uchar *base; |
575 | uchar *dest; | |
0578f103 | 576 | |
5bb46c08 | 577 | base = pfile->buffer->cur - 1; |
578 | do | |
f80e83a9 | 579 | { |
5bb46c08 | 580 | cur = pfile->buffer->cur; |
338fa5f7 | 581 | |
5bb46c08 | 582 | /* N.B. ISIDNUM does not include $. */ |
583 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
bce47149 | 584 | { |
585 | cur++; | |
586 | NORMALIZE_STATE_UPDATE_IDNUM (nst); | |
587 | } | |
0578f103 | 588 | |
78a11351 | 589 | pfile->buffer->cur = cur; |
0578f103 | 590 | } |
bce47149 | 591 | while (forms_identifier_p (pfile, false, nst)); |
79bd622b | 592 | |
5bb46c08 | 593 | number->len = cur - base; |
594 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
595 | memcpy (dest, base, number->len); | |
596 | dest[number->len] = '\0'; | |
597 | number->text = dest; | |
79bd622b | 598 | } |
599 | ||
4970d4c2 | 600 | /* Create a token of type TYPE with a literal spelling. */ |
601 | static void | |
f7fdd7a1 | 602 | create_literal (cpp_reader *pfile, cpp_token *token, const uchar *base, |
603 | unsigned int len, enum cpp_ttype type) | |
4970d4c2 | 604 | { |
605 | uchar *dest = _cpp_unaligned_alloc (pfile, len + 1); | |
606 | ||
607 | memcpy (dest, base, len); | |
608 | dest[len] = '\0'; | |
609 | token->type = type; | |
610 | token->val.str.len = len; | |
611 | token->val.str.text = dest; | |
612 | } | |
613 | ||
5bb46c08 | 614 | /* Lexes a string, character constant, or angle-bracketed header file |
4970d4c2 | 615 | name. The stored string contains the spelling, including opening |
924bbf02 | 616 | quote and leading any leading 'L', 'u' or 'U'. It returns the type |
7811eab5 | 617 | of the literal, or CPP_OTHER if it was not properly terminated, or |
618 | CPP_LESS for an unterminated header name which must be relexed as | |
619 | normal tokens. | |
4970d4c2 | 620 | |
621 | The spelling is NUL-terminated, but it is not guaranteed that this | |
622 | is the first NUL since embedded NULs are preserved. */ | |
f80e83a9 | 623 | static void |
f7fdd7a1 | 624 | lex_string (cpp_reader *pfile, cpp_token *token, const uchar *base) |
0578f103 | 625 | { |
4970d4c2 | 626 | bool saw_NUL = false; |
627 | const uchar *cur; | |
5bb46c08 | 628 | cppchar_t terminator; |
4970d4c2 | 629 | enum cpp_ttype type; |
630 | ||
631 | cur = base; | |
632 | terminator = *cur++; | |
924bbf02 | 633 | if (terminator == 'L' || terminator == 'u' || terminator == 'U') |
4970d4c2 | 634 | terminator = *cur++; |
635 | if (terminator == '\"') | |
924bbf02 | 636 | type = (*base == 'L' ? CPP_WSTRING : |
637 | *base == 'U' ? CPP_STRING32 : | |
638 | *base == 'u' ? CPP_STRING16 : CPP_STRING); | |
4970d4c2 | 639 | else if (terminator == '\'') |
924bbf02 | 640 | type = (*base == 'L' ? CPP_WCHAR : |
641 | *base == 'U' ? CPP_CHAR32 : | |
642 | *base == 'u' ? CPP_CHAR16 : CPP_CHAR); | |
4970d4c2 | 643 | else |
644 | terminator = '>', type = CPP_HEADER_NAME; | |
79bd622b | 645 | |
338fa5f7 | 646 | for (;;) |
0578f103 | 647 | { |
4970d4c2 | 648 | cppchar_t c = *cur++; |
4b0c16ee | 649 | |
edaf8cb5 | 650 | /* In #include-style directives, terminators are not escapable. */ |
4970d4c2 | 651 | if (c == '\\' && !pfile->state.angled_headers && *cur != '\n') |
652 | cur++; | |
653 | else if (c == terminator) | |
5bb46c08 | 654 | break; |
4970d4c2 | 655 | else if (c == '\n') |
338fa5f7 | 656 | { |
4970d4c2 | 657 | cur--; |
7811eab5 | 658 | /* Unmatched quotes always yield undefined behavior, but |
659 | greedy lexing means that what appears to be an unterminated | |
660 | header name may actually be a legitimate sequence of tokens. */ | |
661 | if (terminator == '>') | |
662 | { | |
663 | token->type = CPP_LESS; | |
664 | return; | |
665 | } | |
4970d4c2 | 666 | type = CPP_OTHER; |
667 | break; | |
0578f103 | 668 | } |
4970d4c2 | 669 | else if (c == '\0') |
670 | saw_NUL = true; | |
0578f103 | 671 | } |
672 | ||
4970d4c2 | 673 | if (saw_NUL && !pfile->state.skipping) |
d80d2074 | 674 | cpp_error (pfile, CPP_DL_WARNING, |
675 | "null character(s) preserved in literal"); | |
0578f103 | 676 | |
0b67f687 | 677 | if (type == CPP_OTHER && CPP_OPTION (pfile, lang) != CLK_ASM) |
678 | cpp_error (pfile, CPP_DL_PEDWARN, "missing terminating %c character", | |
679 | (int) terminator); | |
680 | ||
4970d4c2 | 681 | pfile->buffer->cur = cur; |
682 | create_literal (pfile, token, base, cur - base, type); | |
338fa5f7 | 683 | } |
f80e83a9 | 684 | |
956c6108 | 685 | /* Return the comment table. The client may not make any assumption |
686 | about the ordering of the table. */ | |
687 | cpp_comment_table * | |
688 | cpp_get_comments (cpp_reader *pfile) | |
689 | { | |
690 | return &pfile->comments; | |
691 | } | |
692 | ||
693 | /* Append a comment to the end of the comment table. */ | |
694 | static void | |
695 | store_comment (cpp_reader *pfile, cpp_token *token) | |
696 | { | |
697 | int len; | |
698 | ||
699 | if (pfile->comments.allocated == 0) | |
700 | { | |
701 | pfile->comments.allocated = 256; | |
702 | pfile->comments.entries = (cpp_comment *) xmalloc | |
703 | (pfile->comments.allocated * sizeof (cpp_comment)); | |
704 | } | |
705 | ||
706 | if (pfile->comments.count == pfile->comments.allocated) | |
707 | { | |
708 | pfile->comments.allocated *= 2; | |
709 | pfile->comments.entries = (cpp_comment *) xrealloc | |
710 | (pfile->comments.entries, | |
711 | pfile->comments.allocated * sizeof (cpp_comment)); | |
712 | } | |
713 | ||
714 | len = token->val.str.len; | |
715 | ||
716 | /* Copy comment. Note, token may not be NULL terminated. */ | |
717 | pfile->comments.entries[pfile->comments.count].comment = | |
718 | (char *) xmalloc (sizeof (char) * (len + 1)); | |
719 | memcpy (pfile->comments.entries[pfile->comments.count].comment, | |
720 | token->val.str.text, len); | |
721 | pfile->comments.entries[pfile->comments.count].comment[len] = '\0'; | |
722 | ||
723 | /* Set source location. */ | |
724 | pfile->comments.entries[pfile->comments.count].sloc = token->src_loc; | |
725 | ||
726 | /* Increment the count of entries in the comment table. */ | |
727 | pfile->comments.count++; | |
728 | } | |
729 | ||
79bd622b | 730 | /* The stored comment includes the comment start and any terminator. */ |
2c63d6c8 | 731 | static void |
f7fdd7a1 | 732 | save_comment (cpp_reader *pfile, cpp_token *token, const unsigned char *from, |
733 | cppchar_t type) | |
2c63d6c8 | 734 | { |
f80e83a9 | 735 | unsigned char *buffer; |
d3f7919d | 736 | unsigned int len, clen; |
b1a9ff83 | 737 | |
f0495c2c | 738 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
1c124f85 | 739 | |
a543b315 | 740 | /* C++ comments probably (not definitely) have moved past a new |
741 | line, which we don't want to save in the comment. */ | |
1c124f85 | 742 | if (is_vspace (pfile->buffer->cur[-1])) |
a543b315 | 743 | len--; |
d3f7919d | 744 | |
745 | /* If we are currently in a directive, then we need to store all | |
746 | C++ comments as C comments internally, and so we need to | |
747 | allocate a little extra space in that case. | |
748 | ||
749 | Note that the only time we encounter a directive here is | |
750 | when we are saving comments in a "#define". */ | |
751 | clen = (pfile->state.in_directive && type == '/') ? len + 2 : len; | |
752 | ||
753 | buffer = _cpp_unaligned_alloc (pfile, clen); | |
b1a9ff83 | 754 | |
f80e83a9 | 755 | token->type = CPP_COMMENT; |
d3f7919d | 756 | token->val.str.len = clen; |
338fa5f7 | 757 | token->val.str.text = buffer; |
0578f103 | 758 | |
f0495c2c | 759 | buffer[0] = '/'; |
760 | memcpy (buffer + 1, from, len - 1); | |
d3f7919d | 761 | |
a113df96 | 762 | /* Finish conversion to a C comment, if necessary. */ |
d3f7919d | 763 | if (pfile->state.in_directive && type == '/') |
764 | { | |
765 | buffer[1] = '*'; | |
766 | buffer[clen - 2] = '*'; | |
767 | buffer[clen - 1] = '/'; | |
768 | } | |
956c6108 | 769 | |
770 | /* Finally store this comment for use by clients of libcpp. */ | |
771 | store_comment (pfile, token); | |
338fa5f7 | 772 | } |
0578f103 | 773 | |
83dcbb5c | 774 | /* Allocate COUNT tokens for RUN. */ |
775 | void | |
f7fdd7a1 | 776 | _cpp_init_tokenrun (tokenrun *run, unsigned int count) |
83dcbb5c | 777 | { |
3b298764 | 778 | run->base = XNEWVEC (cpp_token, count); |
83dcbb5c | 779 | run->limit = run->base + count; |
780 | run->next = NULL; | |
781 | } | |
782 | ||
783 | /* Returns the next tokenrun, or creates one if there is none. */ | |
784 | static tokenrun * | |
f7fdd7a1 | 785 | next_tokenrun (tokenrun *run) |
83dcbb5c | 786 | { |
787 | if (run->next == NULL) | |
788 | { | |
3b298764 | 789 | run->next = XNEW (tokenrun); |
fb5ab82c | 790 | run->next->prev = run; |
83dcbb5c | 791 | _cpp_init_tokenrun (run->next, 250); |
792 | } | |
793 | ||
794 | return run->next; | |
795 | } | |
796 | ||
89768577 | 797 | /* Look ahead in the input stream. */ |
798 | const cpp_token * | |
799 | cpp_peek_token (cpp_reader *pfile, int index) | |
800 | { | |
801 | cpp_context *context = pfile->context; | |
802 | const cpp_token *peektok; | |
803 | int count; | |
804 | ||
805 | /* First, scan through any pending cpp_context objects. */ | |
806 | while (context->prev) | |
807 | { | |
808 | ptrdiff_t sz = (context->direct_p | |
809 | ? LAST (context).token - FIRST (context).token | |
810 | : LAST (context).ptoken - FIRST (context).ptoken); | |
811 | ||
812 | if (index < (int) sz) | |
813 | return (context->direct_p | |
814 | ? FIRST (context).token + index | |
815 | : *(FIRST (context).ptoken + index)); | |
816 | ||
817 | index -= (int) sz; | |
818 | context = context->prev; | |
819 | } | |
820 | ||
821 | /* We will have to read some new tokens after all (and do so | |
822 | without invalidating preceding tokens). */ | |
823 | count = index; | |
824 | pfile->keep_tokens++; | |
825 | ||
826 | do | |
827 | { | |
828 | peektok = _cpp_lex_token (pfile); | |
829 | if (peektok->type == CPP_EOF) | |
830 | return peektok; | |
831 | } | |
832 | while (index--); | |
833 | ||
834 | _cpp_backup_tokens_direct (pfile, count + 1); | |
835 | pfile->keep_tokens--; | |
836 | ||
837 | return peektok; | |
838 | } | |
839 | ||
f9b5f742 | 840 | /* Allocate a single token that is invalidated at the same time as the |
841 | rest of the tokens on the line. Has its line and col set to the | |
842 | same as the last lexed token, so that diagnostics appear in the | |
843 | right place. */ | |
844 | cpp_token * | |
f7fdd7a1 | 845 | _cpp_temp_token (cpp_reader *pfile) |
f9b5f742 | 846 | { |
847 | cpp_token *old, *result; | |
89768577 | 848 | ptrdiff_t sz = pfile->cur_run->limit - pfile->cur_token; |
849 | ptrdiff_t la = (ptrdiff_t) pfile->lookaheads; | |
f9b5f742 | 850 | |
851 | old = pfile->cur_token - 1; | |
89768577 | 852 | /* Any pre-existing lookaheads must not be clobbered. */ |
853 | if (la) | |
854 | { | |
855 | if (sz <= la) | |
856 | { | |
857 | tokenrun *next = next_tokenrun (pfile->cur_run); | |
858 | ||
859 | if (sz < la) | |
860 | memmove (next->base + 1, next->base, | |
861 | (la - sz) * sizeof (cpp_token)); | |
862 | ||
863 | next->base[0] = pfile->cur_run->limit[-1]; | |
864 | } | |
865 | ||
866 | if (sz > 1) | |
867 | memmove (pfile->cur_token + 1, pfile->cur_token, | |
868 | MIN (la, sz - 1) * sizeof (cpp_token)); | |
869 | } | |
870 | ||
871 | if (!sz && pfile->cur_token == pfile->cur_run->limit) | |
f9b5f742 | 872 | { |
873 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
874 | pfile->cur_token = pfile->cur_run->base; | |
875 | } | |
876 | ||
877 | result = pfile->cur_token++; | |
610625e3 | 878 | result->src_loc = old->src_loc; |
f9b5f742 | 879 | return result; |
880 | } | |
881 | ||
10b4496a | 882 | /* Lex a token into RESULT (external interface). Takes care of issues |
883 | like directive handling, token lookahead, multiple include | |
3fb1e43b | 884 | optimization and skipping. */ |
c00e481c | 885 | const cpp_token * |
f7fdd7a1 | 886 | _cpp_lex_token (cpp_reader *pfile) |
83dcbb5c | 887 | { |
fb5ab82c | 888 | cpp_token *result; |
83dcbb5c | 889 | |
fb5ab82c | 890 | for (;;) |
83dcbb5c | 891 | { |
fb5ab82c | 892 | if (pfile->cur_token == pfile->cur_run->limit) |
83dcbb5c | 893 | { |
fb5ab82c | 894 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
895 | pfile->cur_token = pfile->cur_run->base; | |
83dcbb5c | 896 | } |
e0ff7935 | 897 | /* We assume that the current token is somewhere in the current |
898 | run. */ | |
899 | if (pfile->cur_token < pfile->cur_run->base | |
900 | || pfile->cur_token >= pfile->cur_run->limit) | |
901 | abort (); | |
83dcbb5c | 902 | |
fb5ab82c | 903 | if (pfile->lookaheads) |
10b4496a | 904 | { |
905 | pfile->lookaheads--; | |
906 | result = pfile->cur_token++; | |
907 | } | |
fb5ab82c | 908 | else |
10b4496a | 909 | result = _cpp_lex_direct (pfile); |
fb5ab82c | 910 | |
911 | if (result->flags & BOL) | |
83dcbb5c | 912 | { |
fb5ab82c | 913 | /* Is this a directive. If _cpp_handle_directive returns |
914 | false, it is an assembler #. */ | |
915 | if (result->type == CPP_HASH | |
d6af0368 | 916 | /* 6.10.3 p 11: Directives in a list of macro arguments |
917 | gives undefined behavior. This implementation | |
918 | handles the directive as normal. */ | |
b75b98aa | 919 | && pfile->state.parsing_args != 1) |
d6d3c909 | 920 | { |
b75b98aa | 921 | if (_cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
d6d3c909 | 922 | { |
b75b98aa | 923 | if (pfile->directive_result.type == CPP_PADDING) |
924 | continue; | |
d6d3c909 | 925 | result = &pfile->directive_result; |
d6d3c909 | 926 | } |
927 | } | |
b75b98aa | 928 | else if (pfile->state.in_deferred_pragma) |
929 | result = &pfile->directive_result; | |
d6d3c909 | 930 | |
5621a364 | 931 | if (pfile->cb.line_change && !pfile->state.skipping) |
f7fdd7a1 | 932 | pfile->cb.line_change (pfile, result, pfile->state.parsing_args); |
83dcbb5c | 933 | } |
83dcbb5c | 934 | |
fb5ab82c | 935 | /* We don't skip tokens in directives. */ |
b75b98aa | 936 | if (pfile->state.in_directive || pfile->state.in_deferred_pragma) |
fb5ab82c | 937 | break; |
83dcbb5c | 938 | |
fb5ab82c | 939 | /* Outside a directive, invalidate controlling macros. At file |
10b4496a | 940 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
7ef5b942 | 941 | get here and MI optimization works. */ |
83dcbb5c | 942 | pfile->mi_valid = false; |
fb5ab82c | 943 | |
944 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
945 | break; | |
83dcbb5c | 946 | } |
947 | ||
c00e481c | 948 | return result; |
83dcbb5c | 949 | } |
950 | ||
a54e0bf8 | 951 | /* Returns true if a fresh line has been loaded. */ |
952 | bool | |
f7fdd7a1 | 953 | _cpp_get_fresh_line (cpp_reader *pfile) |
0bb65704 | 954 | { |
6e04daf1 | 955 | int return_at_eof; |
956 | ||
a54e0bf8 | 957 | /* We can't get a new line until we leave the current directive. */ |
958 | if (pfile->state.in_directive) | |
959 | return false; | |
b1a9ff83 | 960 | |
a54e0bf8 | 961 | for (;;) |
fb83e0d6 | 962 | { |
a54e0bf8 | 963 | cpp_buffer *buffer = pfile->buffer; |
fb83e0d6 | 964 | |
a54e0bf8 | 965 | if (!buffer->need_line) |
966 | return true; | |
967 | ||
968 | if (buffer->next_line < buffer->rlimit) | |
0bb65704 | 969 | { |
a54e0bf8 | 970 | _cpp_clean_line (pfile); |
971 | return true; | |
972 | } | |
0bb65704 | 973 | |
a54e0bf8 | 974 | /* First, get out of parsing arguments state. */ |
975 | if (pfile->state.parsing_args) | |
976 | return false; | |
977 | ||
978 | /* End of buffer. Non-empty files should end in a newline. */ | |
979 | if (buffer->buf != buffer->rlimit | |
980 | && buffer->next_line > buffer->rlimit | |
981 | && !buffer->from_stage3) | |
982 | { | |
0448520c | 983 | /* Clip to buffer size. */ |
a54e0bf8 | 984 | buffer->next_line = buffer->rlimit; |
a54e0bf8 | 985 | } |
6e04daf1 | 986 | |
987 | return_at_eof = buffer->return_at_eof; | |
a54e0bf8 | 988 | _cpp_pop_buffer (pfile); |
6e04daf1 | 989 | if (pfile->buffer == NULL || return_at_eof) |
11b5269c | 990 | return false; |
a54e0bf8 | 991 | } |
0bb65704 | 992 | } |
993 | ||
edaf8cb5 | 994 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
995 | do \ | |
996 | { \ | |
997 | result->type = ELSE_TYPE; \ | |
998 | if (*buffer->cur == CHAR) \ | |
999 | buffer->cur++, result->type = THEN_TYPE; \ | |
1000 | } \ | |
1001 | while (0) | |
1c124f85 | 1002 | |
10b4496a | 1003 | /* Lex a token into pfile->cur_token, which is also incremented, to |
1004 | get diagnostics pointing to the correct location. | |
1005 | ||
1006 | Does not handle issues such as token lookahead, multiple-include | |
4172d65e | 1007 | optimization, directives, skipping etc. This function is only |
10b4496a | 1008 | suitable for use by _cpp_lex_token, and in special cases like |
1009 | lex_expansion_token which doesn't care for any of these issues. | |
1010 | ||
1011 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
1012 | otherwise returns to the start of the token buffer if permissible. | |
1013 | Returns the location of the lexed token. */ | |
1014 | cpp_token * | |
f7fdd7a1 | 1015 | _cpp_lex_direct (cpp_reader *pfile) |
0578f103 | 1016 | { |
338fa5f7 | 1017 | cppchar_t c; |
230f0943 | 1018 | cpp_buffer *buffer; |
338fa5f7 | 1019 | const unsigned char *comment_start; |
10b4496a | 1020 | cpp_token *result = pfile->cur_token++; |
0653b94e | 1021 | |
83dcbb5c | 1022 | fresh_line: |
a54e0bf8 | 1023 | result->flags = 0; |
82166c5c | 1024 | buffer = pfile->buffer; |
11b5269c | 1025 | if (buffer->need_line) |
a54e0bf8 | 1026 | { |
b75b98aa | 1027 | if (pfile->state.in_deferred_pragma) |
1028 | { | |
1029 | result->type = CPP_PRAGMA_EOL; | |
1030 | pfile->state.in_deferred_pragma = false; | |
1031 | if (!pfile->state.pragma_allow_expansion) | |
1032 | pfile->state.prevent_expansion--; | |
1033 | return result; | |
1034 | } | |
a54e0bf8 | 1035 | if (!_cpp_get_fresh_line (pfile)) |
1036 | { | |
1037 | result->type = CPP_EOF; | |
2908f819 | 1038 | if (!pfile->state.in_directive) |
1039 | { | |
1040 | /* Tell the compiler the line number of the EOF token. */ | |
dbddc569 | 1041 | result->src_loc = pfile->line_table->highest_line; |
2908f819 | 1042 | result->flags = BOL; |
1043 | } | |
a54e0bf8 | 1044 | return result; |
1045 | } | |
1046 | if (!pfile->keep_tokens) | |
1047 | { | |
1048 | pfile->cur_run = &pfile->base_run; | |
1049 | result = pfile->base_run.base; | |
1050 | pfile->cur_token = result + 1; | |
1051 | } | |
1052 | result->flags = BOL; | |
1053 | if (pfile->state.parsing_args == 2) | |
1054 | result->flags |= PREV_WHITE; | |
1055 | } | |
11b5269c | 1056 | buffer = pfile->buffer; |
83dcbb5c | 1057 | update_tokens_line: |
dbddc569 | 1058 | result->src_loc = pfile->line_table->highest_line; |
f80e83a9 | 1059 | |
83dcbb5c | 1060 | skipped_white: |
a54e0bf8 | 1061 | if (buffer->cur >= buffer->notes[buffer->cur_note].pos |
1062 | && !pfile->overlaid_buffer) | |
1063 | { | |
1064 | _cpp_process_line_notes (pfile, false); | |
dbddc569 | 1065 | result->src_loc = pfile->line_table->highest_line; |
a54e0bf8 | 1066 | } |
1c124f85 | 1067 | c = *buffer->cur++; |
610625e3 | 1068 | |
dbddc569 | 1069 | LINEMAP_POSITION_FOR_COLUMN (result->src_loc, pfile->line_table, |
1070 | CPP_BUF_COLUMN (buffer, buffer->cur)); | |
83dcbb5c | 1071 | |
338fa5f7 | 1072 | switch (c) |
0578f103 | 1073 | { |
435fb09b | 1074 | case ' ': case '\t': case '\f': case '\v': case '\0': |
1075 | result->flags |= PREV_WHITE; | |
a54e0bf8 | 1076 | skip_whitespace (pfile, c); |
1077 | goto skipped_white; | |
338fa5f7 | 1078 | |
a54e0bf8 | 1079 | case '\n': |
610625e3 | 1080 | if (buffer->cur < buffer->rlimit) |
1081 | CPP_INCREMENT_LINE (pfile, 0); | |
a54e0bf8 | 1082 | buffer->need_line = true; |
1083 | goto fresh_line; | |
732cb4c9 | 1084 | |
338fa5f7 | 1085 | case '0': case '1': case '2': case '3': case '4': |
1086 | case '5': case '6': case '7': case '8': case '9': | |
bce47149 | 1087 | { |
1088 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; | |
1089 | result->type = CPP_NUMBER; | |
1090 | lex_number (pfile, &result->val.str, &nst); | |
1091 | warn_about_normalization (pfile, result, &nst); | |
1092 | break; | |
1093 | } | |
732cb4c9 | 1094 | |
78c551ad | 1095 | case 'L': |
924bbf02 | 1096 | case 'u': |
1097 | case 'U': | |
1098 | /* 'L', 'u' or 'U' may introduce wide characters or strings. */ | |
1099 | if (c == 'L' || CPP_OPTION (pfile, uliterals)) | |
5bb46c08 | 1100 | { |
924bbf02 | 1101 | if (*buffer->cur == '\'' || *buffer->cur == '"') |
1102 | { | |
1103 | lex_string (pfile, result, buffer->cur - 1); | |
1104 | break; | |
1105 | } | |
5bb46c08 | 1106 | } |
b1a9ff83 | 1107 | /* Fall through. */ |
78c551ad | 1108 | |
338fa5f7 | 1109 | case '_': |
1110 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
1111 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
1112 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
924bbf02 | 1113 | case 's': case 't': case 'v': case 'w': case 'x': |
338fa5f7 | 1114 | case 'y': case 'z': |
1115 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
78c551ad | 1116 | case 'G': case 'H': case 'I': case 'J': case 'K': |
338fa5f7 | 1117 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
924bbf02 | 1118 | case 'S': case 'T': case 'V': case 'W': case 'X': |
338fa5f7 | 1119 | case 'Y': case 'Z': |
1120 | result->type = CPP_NAME; | |
bce47149 | 1121 | { |
1122 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; | |
1123 | result->val.node = lex_identifier (pfile, buffer->cur - 1, false, | |
1124 | &nst); | |
1125 | warn_about_normalization (pfile, result, &nst); | |
1126 | } | |
338fa5f7 | 1127 | |
338fa5f7 | 1128 | /* Convert named operators to their proper types. */ |
78c551ad | 1129 | if (result->val.node->flags & NODE_OPERATOR) |
338fa5f7 | 1130 | { |
1131 | result->flags |= NAMED_OP; | |
720aca92 | 1132 | result->type = (enum cpp_ttype) result->val.node->directive_index; |
338fa5f7 | 1133 | } |
1134 | break; | |
1135 | ||
1136 | case '\'': | |
1137 | case '"': | |
4970d4c2 | 1138 | lex_string (pfile, result, buffer->cur - 1); |
338fa5f7 | 1139 | break; |
f80e83a9 | 1140 | |
338fa5f7 | 1141 | case '/': |
f0495c2c | 1142 | /* A potential block or line comment. */ |
1143 | comment_start = buffer->cur; | |
edaf8cb5 | 1144 | c = *buffer->cur; |
1145 | ||
f0495c2c | 1146 | if (c == '*') |
1147 | { | |
a54e0bf8 | 1148 | if (_cpp_skip_block_comment (pfile)) |
d80d2074 | 1149 | cpp_error (pfile, CPP_DL_ERROR, "unterminated comment"); |
338fa5f7 | 1150 | } |
1c124f85 | 1151 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
610625e3 | 1152 | || cpp_in_system_header (pfile))) |
338fa5f7 | 1153 | { |
5db5d057 | 1154 | /* Warn about comments only if pedantically GNUC89, and not |
1155 | in system headers. */ | |
1156 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
66914e49 | 1157 | && ! buffer->warned_cplusplus_comments) |
f80e83a9 | 1158 | { |
d80d2074 | 1159 | cpp_error (pfile, CPP_DL_PEDWARN, |
ba059ac0 | 1160 | "C++ style comments are not allowed in ISO C90"); |
d80d2074 | 1161 | cpp_error (pfile, CPP_DL_PEDWARN, |
73328dce | 1162 | "(this will be reported only once per input file)"); |
f0495c2c | 1163 | buffer->warned_cplusplus_comments = 1; |
1164 | } | |
338fa5f7 | 1165 | |
e1caf668 | 1166 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
d80d2074 | 1167 | cpp_error (pfile, CPP_DL_WARNING, "multi-line comment"); |
f0495c2c | 1168 | } |
1c124f85 | 1169 | else if (c == '=') |
1170 | { | |
edaf8cb5 | 1171 | buffer->cur++; |
1c124f85 | 1172 | result->type = CPP_DIV_EQ; |
1173 | break; | |
1174 | } | |
1175 | else | |
1176 | { | |
1c124f85 | 1177 | result->type = CPP_DIV; |
1178 | break; | |
1179 | } | |
338fa5f7 | 1180 | |
f0495c2c | 1181 | if (!pfile->state.save_comments) |
1182 | { | |
1183 | result->flags |= PREV_WHITE; | |
83dcbb5c | 1184 | goto update_tokens_line; |
338fa5f7 | 1185 | } |
f0495c2c | 1186 | |
1187 | /* Save the comment as a token in its own right. */ | |
d3f7919d | 1188 | save_comment (pfile, result, comment_start, c); |
fb5ab82c | 1189 | break; |
338fa5f7 | 1190 | |
1191 | case '<': | |
1192 | if (pfile->state.angled_headers) | |
1193 | { | |
4970d4c2 | 1194 | lex_string (pfile, result, buffer->cur - 1); |
7811eab5 | 1195 | if (result->type != CPP_LESS) |
1196 | break; | |
338fa5f7 | 1197 | } |
0578f103 | 1198 | |
edaf8cb5 | 1199 | result->type = CPP_LESS; |
1200 | if (*buffer->cur == '=') | |
1201 | buffer->cur++, result->type = CPP_LESS_EQ; | |
1202 | else if (*buffer->cur == '<') | |
338fa5f7 | 1203 | { |
edaf8cb5 | 1204 | buffer->cur++; |
1205 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); | |
338fa5f7 | 1206 | } |
edaf8cb5 | 1207 | else if (CPP_OPTION (pfile, digraphs)) |
1c124f85 | 1208 | { |
edaf8cb5 | 1209 | if (*buffer->cur == ':') |
1210 | { | |
1211 | buffer->cur++; | |
1212 | result->flags |= DIGRAPH; | |
1213 | result->type = CPP_OPEN_SQUARE; | |
1214 | } | |
1215 | else if (*buffer->cur == '%') | |
1216 | { | |
1217 | buffer->cur++; | |
1218 | result->flags |= DIGRAPH; | |
1219 | result->type = CPP_OPEN_BRACE; | |
1220 | } | |
1c124f85 | 1221 | } |
338fa5f7 | 1222 | break; |
1223 | ||
1224 | case '>': | |
edaf8cb5 | 1225 | result->type = CPP_GREATER; |
1226 | if (*buffer->cur == '=') | |
1227 | buffer->cur++, result->type = CPP_GREATER_EQ; | |
1228 | else if (*buffer->cur == '>') | |
338fa5f7 | 1229 | { |
edaf8cb5 | 1230 | buffer->cur++; |
1231 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); | |
1232 | } | |
338fa5f7 | 1233 | break; |
1234 | ||
f669338a | 1235 | case '%': |
edaf8cb5 | 1236 | result->type = CPP_MOD; |
1237 | if (*buffer->cur == '=') | |
1238 | buffer->cur++, result->type = CPP_MOD_EQ; | |
1239 | else if (CPP_OPTION (pfile, digraphs)) | |
1c124f85 | 1240 | { |
edaf8cb5 | 1241 | if (*buffer->cur == ':') |
1c124f85 | 1242 | { |
edaf8cb5 | 1243 | buffer->cur++; |
1244 | result->flags |= DIGRAPH; | |
1245 | result->type = CPP_HASH; | |
1246 | if (*buffer->cur == '%' && buffer->cur[1] == ':') | |
941f2388 | 1247 | buffer->cur += 2, result->type = CPP_PASTE, result->val.arg_no = 0; |
edaf8cb5 | 1248 | } |
1249 | else if (*buffer->cur == '>') | |
1250 | { | |
1251 | buffer->cur++; | |
1252 | result->flags |= DIGRAPH; | |
1253 | result->type = CPP_CLOSE_BRACE; | |
1c124f85 | 1254 | } |
1c124f85 | 1255 | } |
338fa5f7 | 1256 | break; |
1257 | ||
f669338a | 1258 | case '.': |
1c124f85 | 1259 | result->type = CPP_DOT; |
edaf8cb5 | 1260 | if (ISDIGIT (*buffer->cur)) |
1c124f85 | 1261 | { |
bce47149 | 1262 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; |
1c124f85 | 1263 | result->type = CPP_NUMBER; |
bce47149 | 1264 | lex_number (pfile, &result->val.str, &nst); |
1265 | warn_about_normalization (pfile, result, &nst); | |
1c124f85 | 1266 | } |
edaf8cb5 | 1267 | else if (*buffer->cur == '.' && buffer->cur[1] == '.') |
1268 | buffer->cur += 2, result->type = CPP_ELLIPSIS; | |
1269 | else if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) | |
1270 | buffer->cur++, result->type = CPP_DOT_STAR; | |
338fa5f7 | 1271 | break; |
0578f103 | 1272 | |
338fa5f7 | 1273 | case '+': |
edaf8cb5 | 1274 | result->type = CPP_PLUS; |
1275 | if (*buffer->cur == '+') | |
1276 | buffer->cur++, result->type = CPP_PLUS_PLUS; | |
1277 | else if (*buffer->cur == '=') | |
1278 | buffer->cur++, result->type = CPP_PLUS_EQ; | |
338fa5f7 | 1279 | break; |
ac0749c7 | 1280 | |
338fa5f7 | 1281 | case '-': |
edaf8cb5 | 1282 | result->type = CPP_MINUS; |
1283 | if (*buffer->cur == '>') | |
338fa5f7 | 1284 | { |
edaf8cb5 | 1285 | buffer->cur++; |
1c124f85 | 1286 | result->type = CPP_DEREF; |
edaf8cb5 | 1287 | if (*buffer->cur == '*' && CPP_OPTION (pfile, cplusplus)) |
1288 | buffer->cur++, result->type = CPP_DEREF_STAR; | |
1c124f85 | 1289 | } |
edaf8cb5 | 1290 | else if (*buffer->cur == '-') |
1291 | buffer->cur++, result->type = CPP_MINUS_MINUS; | |
1292 | else if (*buffer->cur == '=') | |
1293 | buffer->cur++, result->type = CPP_MINUS_EQ; | |
338fa5f7 | 1294 | break; |
0578f103 | 1295 | |
338fa5f7 | 1296 | case '&': |
edaf8cb5 | 1297 | result->type = CPP_AND; |
1298 | if (*buffer->cur == '&') | |
1299 | buffer->cur++, result->type = CPP_AND_AND; | |
1300 | else if (*buffer->cur == '=') | |
1301 | buffer->cur++, result->type = CPP_AND_EQ; | |
338fa5f7 | 1302 | break; |
b1a9ff83 | 1303 | |
338fa5f7 | 1304 | case '|': |
edaf8cb5 | 1305 | result->type = CPP_OR; |
1306 | if (*buffer->cur == '|') | |
1307 | buffer->cur++, result->type = CPP_OR_OR; | |
1308 | else if (*buffer->cur == '=') | |
1309 | buffer->cur++, result->type = CPP_OR_EQ; | |
338fa5f7 | 1310 | break; |
0578f103 | 1311 | |
338fa5f7 | 1312 | case ':': |
edaf8cb5 | 1313 | result->type = CPP_COLON; |
1314 | if (*buffer->cur == ':' && CPP_OPTION (pfile, cplusplus)) | |
1315 | buffer->cur++, result->type = CPP_SCOPE; | |
1316 | else if (*buffer->cur == '>' && CPP_OPTION (pfile, digraphs)) | |
338fa5f7 | 1317 | { |
edaf8cb5 | 1318 | buffer->cur++; |
338fa5f7 | 1319 | result->flags |= DIGRAPH; |
1c124f85 | 1320 | result->type = CPP_CLOSE_SQUARE; |
1321 | } | |
338fa5f7 | 1322 | break; |
0578f103 | 1323 | |
1c124f85 | 1324 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1325 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1326 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1327 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
941f2388 | 1328 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); result->val.arg_no = 0; break; |
1c124f85 | 1329 | |
a54e0bf8 | 1330 | case '?': result->type = CPP_QUERY; break; |
338fa5f7 | 1331 | case '~': result->type = CPP_COMPL; break; |
1332 | case ',': result->type = CPP_COMMA; break; | |
1333 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1334 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1335 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1336 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1337 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1338 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1339 | case ';': result->type = CPP_SEMICOLON; break; | |
1340 | ||
7fd957fe | 1341 | /* @ is a punctuator in Objective-C. */ |
9ee99ac6 | 1342 | case '@': result->type = CPP_ATSIGN; break; |
338fa5f7 | 1343 | |
78c551ad | 1344 | case '$': |
2cbf1359 | 1345 | case '\\': |
1346 | { | |
1347 | const uchar *base = --buffer->cur; | |
bce47149 | 1348 | struct normalize_state nst = INITIAL_NORMALIZE_STATE; |
78c551ad | 1349 | |
bce47149 | 1350 | if (forms_identifier_p (pfile, true, &nst)) |
2cbf1359 | 1351 | { |
1352 | result->type = CPP_NAME; | |
bce47149 | 1353 | result->val.node = lex_identifier (pfile, base, true, &nst); |
1354 | warn_about_normalization (pfile, result, &nst); | |
2cbf1359 | 1355 | break; |
1356 | } | |
1357 | buffer->cur++; | |
bc205914 | 1358 | } |
2cbf1359 | 1359 | |
bc205914 | 1360 | default: |
4970d4c2 | 1361 | create_literal (pfile, result, buffer->cur - 1, 1, CPP_OTHER); |
1362 | break; | |
338fa5f7 | 1363 | } |
fb5ab82c | 1364 | |
1365 | return result; | |
338fa5f7 | 1366 | } |
1367 | ||
b1280514 | 1368 | /* An upper bound on the number of bytes needed to spell TOKEN. |
1369 | Does not include preceding whitespace. */ | |
79bd622b | 1370 | unsigned int |
f7fdd7a1 | 1371 | cpp_token_len (const cpp_token *token) |
338fa5f7 | 1372 | { |
79bd622b | 1373 | unsigned int len; |
cfad5579 | 1374 | |
79bd622b | 1375 | switch (TOKEN_SPELL (token)) |
f80e83a9 | 1376 | { |
cd740bd5 | 1377 | default: len = 6; break; |
4970d4c2 | 1378 | case SPELL_LITERAL: len = token->val.str.len; break; |
bb1fa6bb | 1379 | case SPELL_IDENT: len = NODE_LEN (token->val.node) * 10; break; |
f80e83a9 | 1380 | } |
b1280514 | 1381 | |
1382 | return len; | |
cfad5579 | 1383 | } |
1384 | ||
bb1fa6bb | 1385 | /* Parse UTF-8 out of NAMEP and place a \U escape in BUFFER. |
1386 | Return the number of bytes read out of NAME. (There are always | |
1387 | 10 bytes written to BUFFER.) */ | |
1388 | ||
1389 | static size_t | |
1390 | utf8_to_ucn (unsigned char *buffer, const unsigned char *name) | |
1391 | { | |
1392 | int j; | |
1393 | int ucn_len = 0; | |
1394 | int ucn_len_c; | |
1395 | unsigned t; | |
1396 | unsigned long utf32; | |
1397 | ||
1398 | /* Compute the length of the UTF-8 sequence. */ | |
1399 | for (t = *name; t & 0x80; t <<= 1) | |
1400 | ucn_len++; | |
1401 | ||
1402 | utf32 = *name & (0x7F >> ucn_len); | |
1403 | for (ucn_len_c = 1; ucn_len_c < ucn_len; ucn_len_c++) | |
1404 | { | |
1405 | utf32 = (utf32 << 6) | (*++name & 0x3F); | |
1406 | ||
1407 | /* Ill-formed UTF-8. */ | |
1408 | if ((*name & ~0x3F) != 0x80) | |
1409 | abort (); | |
1410 | } | |
1411 | ||
1412 | *buffer++ = '\\'; | |
1413 | *buffer++ = 'U'; | |
1414 | for (j = 7; j >= 0; j--) | |
1415 | *buffer++ = "0123456789abcdef"[(utf32 >> (4 * j)) & 0xF]; | |
1416 | return ucn_len; | |
1417 | } | |
1418 | ||
ba99525e | 1419 | /* Given a token TYPE corresponding to a digraph, return a pointer to |
1420 | the spelling of the digraph. */ | |
1421 | static const unsigned char * | |
1422 | cpp_digraph2name (enum cpp_ttype type) | |
1423 | { | |
1424 | return digraph_spellings[(int) type - (int) CPP_FIRST_DIGRAPH]; | |
1425 | } | |
bb1fa6bb | 1426 | |
f80e83a9 | 1427 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
c5ea33a8 | 1428 | already contain the enough space to hold the token's spelling. |
f7fdd7a1 | 1429 | Returns a pointer to the character after the last character written. |
bb1fa6bb | 1430 | FORSTRING is true if this is to be the spelling after translation |
1431 | phase 1 (this is different for UCNs). | |
f7fdd7a1 | 1432 | FIXME: Would be nice if we didn't need the PFILE argument. */ |
79bd622b | 1433 | unsigned char * |
f7fdd7a1 | 1434 | cpp_spell_token (cpp_reader *pfile, const cpp_token *token, |
bb1fa6bb | 1435 | unsigned char *buffer, bool forstring) |
f80e83a9 | 1436 | { |
7e842f95 | 1437 | switch (TOKEN_SPELL (token)) |
f80e83a9 | 1438 | { |
1439 | case SPELL_OPERATOR: | |
1440 | { | |
1441 | const unsigned char *spelling; | |
1442 | unsigned char c; | |
ab12a39c | 1443 | |
f80e83a9 | 1444 | if (token->flags & DIGRAPH) |
ba99525e | 1445 | spelling = cpp_digraph2name (token->type); |
31674461 | 1446 | else if (token->flags & NAMED_OP) |
1447 | goto spell_ident; | |
f80e83a9 | 1448 | else |
7e842f95 | 1449 | spelling = TOKEN_NAME (token); |
b1a9ff83 | 1450 | |
f80e83a9 | 1451 | while ((c = *spelling++) != '\0') |
1452 | *buffer++ = c; | |
1453 | } | |
1454 | break; | |
ab12a39c | 1455 | |
8d27e472 | 1456 | spell_ident: |
f80e83a9 | 1457 | case SPELL_IDENT: |
bb1fa6bb | 1458 | if (forstring) |
1459 | { | |
1460 | memcpy (buffer, NODE_NAME (token->val.node), | |
1461 | NODE_LEN (token->val.node)); | |
1462 | buffer += NODE_LEN (token->val.node); | |
1463 | } | |
1464 | else | |
1465 | { | |
1466 | size_t i; | |
1467 | const unsigned char * name = NODE_NAME (token->val.node); | |
1468 | ||
1469 | for (i = 0; i < NODE_LEN (token->val.node); i++) | |
1470 | if (name[i] & ~0x7F) | |
1471 | { | |
1472 | i += utf8_to_ucn (buffer, name + i) - 1; | |
1473 | buffer += 10; | |
1474 | } | |
1475 | else | |
1476 | *buffer++ = NODE_NAME (token->val.node)[i]; | |
1477 | } | |
f80e83a9 | 1478 | break; |
ab12a39c | 1479 | |
4970d4c2 | 1480 | case SPELL_LITERAL: |
8d27e472 | 1481 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1482 | buffer += token->val.str.len; | |
1483 | break; | |
1484 | ||
f80e83a9 | 1485 | case SPELL_NONE: |
d80d2074 | 1486 | cpp_error (pfile, CPP_DL_ICE, |
1487 | "unspellable token %s", TOKEN_NAME (token)); | |
f80e83a9 | 1488 | break; |
1489 | } | |
ab12a39c | 1490 | |
f80e83a9 | 1491 | return buffer; |
1492 | } | |
ab12a39c | 1493 | |
e484a1cc | 1494 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1495 | freed when the reader is destroyed. Useful for diagnostics. */ | |
79bd622b | 1496 | unsigned char * |
f7fdd7a1 | 1497 | cpp_token_as_text (cpp_reader *pfile, const cpp_token *token) |
b1280514 | 1498 | { |
1499 | unsigned int len = cpp_token_len (token) + 1; | |
1fdf6039 | 1500 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
6060326b | 1501 | |
bb1fa6bb | 1502 | end = cpp_spell_token (pfile, token, start, false); |
79bd622b | 1503 | end[0] = '\0'; |
6060326b | 1504 | |
79bd622b | 1505 | return start; |
1506 | } | |
6060326b | 1507 | |
ba99525e | 1508 | /* Returns a pointer to a string which spells the token defined by |
1509 | TYPE and FLAGS. Used by C front ends, which really should move to | |
1510 | using cpp_token_as_text. */ | |
79bd622b | 1511 | const char * |
ba99525e | 1512 | cpp_type2name (enum cpp_ttype type, unsigned char flags) |
79bd622b | 1513 | { |
ba99525e | 1514 | if (flags & DIGRAPH) |
1515 | return (const char *) cpp_digraph2name (type); | |
1516 | else if (flags & NAMED_OP) | |
1517 | return cpp_named_operator2name (type); | |
1518 | ||
79bd622b | 1519 | return (const char *) token_spellings[type].name; |
1520 | } | |
6060326b | 1521 | |
f9b5f742 | 1522 | /* Writes the spelling of token to FP, without any preceding space. |
1523 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1524 | double-buffering. */ | |
79bd622b | 1525 | void |
f7fdd7a1 | 1526 | cpp_output_token (const cpp_token *token, FILE *fp) |
79bd622b | 1527 | { |
79bd622b | 1528 | switch (TOKEN_SPELL (token)) |
6060326b | 1529 | { |
79bd622b | 1530 | case SPELL_OPERATOR: |
1531 | { | |
1532 | const unsigned char *spelling; | |
28874558 | 1533 | int c; |
6060326b | 1534 | |
79bd622b | 1535 | if (token->flags & DIGRAPH) |
ba99525e | 1536 | spelling = cpp_digraph2name (token->type); |
79bd622b | 1537 | else if (token->flags & NAMED_OP) |
1538 | goto spell_ident; | |
1539 | else | |
1540 | spelling = TOKEN_NAME (token); | |
f80e83a9 | 1541 | |
28874558 | 1542 | c = *spelling; |
1543 | do | |
1544 | putc (c, fp); | |
1545 | while ((c = *++spelling) != '\0'); | |
79bd622b | 1546 | } |
1547 | break; | |
f80e83a9 | 1548 | |
79bd622b | 1549 | spell_ident: |
1550 | case SPELL_IDENT: | |
bb1fa6bb | 1551 | { |
1552 | size_t i; | |
1553 | const unsigned char * name = NODE_NAME (token->val.node); | |
1554 | ||
1555 | for (i = 0; i < NODE_LEN (token->val.node); i++) | |
1556 | if (name[i] & ~0x7F) | |
1557 | { | |
1558 | unsigned char buffer[10]; | |
1559 | i += utf8_to_ucn (buffer, name + i) - 1; | |
1560 | fwrite (buffer, 1, 10, fp); | |
1561 | } | |
1562 | else | |
1563 | fputc (NODE_NAME (token->val.node)[i], fp); | |
1564 | } | |
1565 | break; | |
f80e83a9 | 1566 | |
4970d4c2 | 1567 | case SPELL_LITERAL: |
8d27e472 | 1568 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
1569 | break; | |
1570 | ||
79bd622b | 1571 | case SPELL_NONE: |
1572 | /* An error, most probably. */ | |
1573 | break; | |
f80e83a9 | 1574 | } |
6060326b | 1575 | } |
1576 | ||
79bd622b | 1577 | /* Compare two tokens. */ |
1578 | int | |
f7fdd7a1 | 1579 | _cpp_equiv_tokens (const cpp_token *a, const cpp_token *b) |
6060326b | 1580 | { |
79bd622b | 1581 | if (a->type == b->type && a->flags == b->flags) |
1582 | switch (TOKEN_SPELL (a)) | |
1583 | { | |
1584 | default: /* Keep compiler happy. */ | |
1585 | case SPELL_OPERATOR: | |
941f2388 | 1586 | /* arg_no is used to track where multiple consecutive ## |
1587 | tokens were originally located. */ | |
1588 | return (a->type != CPP_PASTE || a->val.arg_no == b->val.arg_no); | |
79bd622b | 1589 | case SPELL_NONE: |
588d632b | 1590 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
79bd622b | 1591 | case SPELL_IDENT: |
1592 | return a->val.node == b->val.node; | |
4970d4c2 | 1593 | case SPELL_LITERAL: |
79bd622b | 1594 | return (a->val.str.len == b->val.str.len |
1595 | && !memcmp (a->val.str.text, b->val.str.text, | |
1596 | a->val.str.len)); | |
1597 | } | |
6060326b | 1598 | |
f80e83a9 | 1599 | return 0; |
1600 | } | |
1601 | ||
79bd622b | 1602 | /* Returns nonzero if a space should be inserted to avoid an |
1603 | accidental token paste for output. For simplicity, it is | |
1604 | conservative, and occasionally advises a space where one is not | |
1605 | needed, e.g. "." and ".2". */ | |
79bd622b | 1606 | int |
f7fdd7a1 | 1607 | cpp_avoid_paste (cpp_reader *pfile, const cpp_token *token1, |
1608 | const cpp_token *token2) | |
6060326b | 1609 | { |
79bd622b | 1610 | enum cpp_ttype a = token1->type, b = token2->type; |
1611 | cppchar_t c; | |
6060326b | 1612 | |
79bd622b | 1613 | if (token1->flags & NAMED_OP) |
1614 | a = CPP_NAME; | |
1615 | if (token2->flags & NAMED_OP) | |
1616 | b = CPP_NAME; | |
6060326b | 1617 | |
79bd622b | 1618 | c = EOF; |
1619 | if (token2->flags & DIGRAPH) | |
ee6c4e4b | 1620 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
79bd622b | 1621 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1622 | c = token_spellings[b].name[0]; | |
6060326b | 1623 | |
79bd622b | 1624 | /* Quickly get everything that can paste with an '='. */ |
ee6c4e4b | 1625 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
79bd622b | 1626 | return 1; |
6060326b | 1627 | |
79bd622b | 1628 | switch (a) |
6060326b | 1629 | { |
e58c07f7 | 1630 | case CPP_GREATER: return c == '>'; |
1631 | case CPP_LESS: return c == '<' || c == '%' || c == ':'; | |
79bd622b | 1632 | case CPP_PLUS: return c == '+'; |
1633 | case CPP_MINUS: return c == '-' || c == '>'; | |
1634 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1635 | case CPP_MOD: return c == ':' || c == '>'; | |
1636 | case CPP_AND: return c == '&'; | |
1637 | case CPP_OR: return c == '|'; | |
1638 | case CPP_COLON: return c == ':' || c == '>'; | |
1639 | case CPP_DEREF: return c == '*'; | |
efdcc728 | 1640 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
79bd622b | 1641 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1642 | case CPP_NAME: return ((b == CPP_NUMBER | |
1643 | && name_p (pfile, &token2->val.str)) | |
1644 | || b == CPP_NAME | |
1645 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1646 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1647 | || c == '.' || c == '+' || c == '-'); | |
2cbf1359 | 1648 | /* UCNs */ |
bc205914 | 1649 | case CPP_OTHER: return ((token1->val.str.text[0] == '\\' |
1650 | && b == CPP_NAME) | |
2cbf1359 | 1651 | || (CPP_OPTION (pfile, objc) |
bc205914 | 1652 | && token1->val.str.text[0] == '@' |
2cbf1359 | 1653 | && (b == CPP_NAME || b == CPP_STRING))); |
79bd622b | 1654 | default: break; |
6060326b | 1655 | } |
6060326b | 1656 | |
deb356cf | 1657 | return 0; |
6060326b | 1658 | } |
1659 | ||
79bd622b | 1660 | /* Output all the remaining tokens on the current line, and a newline |
f9b5f742 | 1661 | character, to FP. Leading whitespace is removed. If there are |
1662 | macros, special token padding is not performed. */ | |
6060326b | 1663 | void |
f7fdd7a1 | 1664 | cpp_output_line (cpp_reader *pfile, FILE *fp) |
6060326b | 1665 | { |
f9b5f742 | 1666 | const cpp_token *token; |
7e842f95 | 1667 | |
f9b5f742 | 1668 | token = cpp_get_token (pfile); |
1669 | while (token->type != CPP_EOF) | |
7e842f95 | 1670 | { |
f9b5f742 | 1671 | cpp_output_token (token, fp); |
1672 | token = cpp_get_token (pfile); | |
1673 | if (token->flags & PREV_WHITE) | |
1674 | putc (' ', fp); | |
7e842f95 | 1675 | } |
1676 | ||
79bd622b | 1677 | putc ('\n', fp); |
f80e83a9 | 1678 | } |
6060326b | 1679 | |
c0770282 | 1680 | /* Return a string representation of all the remaining tokens on the |
1681 | current line. The result is allocated using xmalloc and must be | |
1682 | freed by the caller. */ | |
1683 | unsigned char * | |
1684 | cpp_output_line_to_string (cpp_reader *pfile, const unsigned char *dir_name) | |
1685 | { | |
1686 | const cpp_token *token; | |
1687 | unsigned int out = dir_name ? ustrlen (dir_name) : 0; | |
1688 | unsigned int alloced = 120 + out; | |
1689 | unsigned char *result = (unsigned char *) xmalloc (alloced); | |
1690 | ||
1691 | /* If DIR_NAME is empty, there are no initial contents. */ | |
1692 | if (dir_name) | |
1693 | { | |
1694 | sprintf ((char *) result, "#%s ", dir_name); | |
1695 | out += 2; | |
1696 | } | |
1697 | ||
1698 | token = cpp_get_token (pfile); | |
1699 | while (token->type != CPP_EOF) | |
1700 | { | |
1701 | unsigned char *last; | |
1702 | /* Include room for a possible space and the terminating nul. */ | |
1703 | unsigned int len = cpp_token_len (token) + 2; | |
1704 | ||
1705 | if (out + len > alloced) | |
1706 | { | |
1707 | alloced *= 2; | |
1708 | if (out + len > alloced) | |
1709 | alloced = out + len; | |
1710 | result = (unsigned char *) xrealloc (result, alloced); | |
1711 | } | |
1712 | ||
1713 | last = cpp_spell_token (pfile, token, &result[out], 0); | |
1714 | out = last - result; | |
1715 | ||
1716 | token = cpp_get_token (pfile); | |
1717 | if (token->flags & PREV_WHITE) | |
1718 | result[out++] = ' '; | |
1719 | } | |
1720 | ||
1721 | result[out] = '\0'; | |
1722 | return result; | |
1723 | } | |
1724 | ||
084163dc | 1725 | /* Memory buffers. Changing these three constants can have a dramatic |
1726 | effect on performance. The values here are reasonable defaults, | |
1727 | but might be tuned. If you adjust them, be sure to test across a | |
1728 | range of uses of cpplib, including heavy nested function-like macro | |
1729 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1730 | good tool for this). */ | |
1731 | #define MIN_BUFF_SIZE 8000 | |
1e0ef2fd | 1732 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
084163dc | 1733 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1734 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
deb356cf | 1735 | |
1e0ef2fd | 1736 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1737 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1738 | #endif | |
1739 | ||
1785b647 | 1740 | /* Create a new allocation buffer. Place the control block at the end |
1741 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
06c92cbc | 1742 | static _cpp_buff * |
f7fdd7a1 | 1743 | new_buff (size_t len) |
06c92cbc | 1744 | { |
1745 | _cpp_buff *result; | |
1fdf6039 | 1746 | unsigned char *base; |
06c92cbc | 1747 | |
084163dc | 1748 | if (len < MIN_BUFF_SIZE) |
1749 | len = MIN_BUFF_SIZE; | |
198b48a0 | 1750 | len = CPP_ALIGN (len); |
06c92cbc | 1751 | |
720aca92 | 1752 | base = XNEWVEC (unsigned char, len + sizeof (_cpp_buff)); |
06c92cbc | 1753 | result = (_cpp_buff *) (base + len); |
1754 | result->base = base; | |
1755 | result->cur = base; | |
1756 | result->limit = base + len; | |
1757 | result->next = NULL; | |
1758 | return result; | |
1759 | } | |
1760 | ||
1761 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1762 | void | |
f7fdd7a1 | 1763 | _cpp_release_buff (cpp_reader *pfile, _cpp_buff *buff) |
06c92cbc | 1764 | { |
1765 | _cpp_buff *end = buff; | |
1766 | ||
1767 | while (end->next) | |
1768 | end = end->next; | |
1769 | end->next = pfile->free_buffs; | |
1770 | pfile->free_buffs = buff; | |
1771 | } | |
1772 | ||
1773 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1774 | _cpp_buff * | |
f7fdd7a1 | 1775 | _cpp_get_buff (cpp_reader *pfile, size_t min_size) |
06c92cbc | 1776 | { |
1777 | _cpp_buff *result, **p; | |
1778 | ||
1779 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1780 | { | |
4b31a107 | 1781 | size_t size; |
084163dc | 1782 | |
1783 | if (*p == NULL) | |
06c92cbc | 1784 | return new_buff (min_size); |
084163dc | 1785 | result = *p; |
1786 | size = result->limit - result->base; | |
1787 | /* Return a buffer that's big enough, but don't waste one that's | |
1788 | way too big. */ | |
4085c149 | 1789 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
06c92cbc | 1790 | break; |
1791 | } | |
1792 | ||
1793 | *p = result->next; | |
1794 | result->next = NULL; | |
1795 | result->cur = result->base; | |
1796 | return result; | |
1797 | } | |
1798 | ||
20dd417a | 1799 | /* Creates a new buffer with enough space to hold the uncommitted |
e6a5f963 | 1800 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
1801 | the excess bytes to the new buffer. Chains the new buffer after | |
1802 | BUFF, and returns the new buffer. */ | |
06c92cbc | 1803 | _cpp_buff * |
f7fdd7a1 | 1804 | _cpp_append_extend_buff (cpp_reader *pfile, _cpp_buff *buff, size_t min_extra) |
06c92cbc | 1805 | { |
4b31a107 | 1806 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
e6a5f963 | 1807 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
06c92cbc | 1808 | |
e6a5f963 | 1809 | buff->next = new_buff; |
1810 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
1811 | return new_buff; | |
1812 | } | |
1813 | ||
20dd417a | 1814 | /* Creates a new buffer with enough space to hold the uncommitted |
e6a5f963 | 1815 | remaining bytes of the buffer pointed to by BUFF, and at least |
1816 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
1817 | Chains the new buffer before the buffer pointed to by BUFF, and | |
1818 | updates the pointer to point to the new buffer. */ | |
1819 | void | |
f7fdd7a1 | 1820 | _cpp_extend_buff (cpp_reader *pfile, _cpp_buff **pbuff, size_t min_extra) |
e6a5f963 | 1821 | { |
1822 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
1823 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
1824 | ||
1825 | new_buff = _cpp_get_buff (pfile, size); | |
1826 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
1827 | new_buff->next = old_buff; | |
1828 | *pbuff = new_buff; | |
06c92cbc | 1829 | } |
1830 | ||
1831 | /* Free a chain of buffers starting at BUFF. */ | |
1832 | void | |
f82b06e0 | 1833 | _cpp_free_buff (_cpp_buff *buff) |
06c92cbc | 1834 | { |
1835 | _cpp_buff *next; | |
1836 | ||
1837 | for (; buff; buff = next) | |
1838 | { | |
1839 | next = buff->next; | |
1840 | free (buff->base); | |
1841 | } | |
1842 | } | |
deb356cf | 1843 | |
1fdf6039 | 1844 | /* Allocate permanent, unaligned storage of length LEN. */ |
1845 | unsigned char * | |
f7fdd7a1 | 1846 | _cpp_unaligned_alloc (cpp_reader *pfile, size_t len) |
1fdf6039 | 1847 | { |
1848 | _cpp_buff *buff = pfile->u_buff; | |
1849 | unsigned char *result = buff->cur; | |
1850 | ||
1851 | if (len > (size_t) (buff->limit - result)) | |
1852 | { | |
1853 | buff = _cpp_get_buff (pfile, len); | |
1854 | buff->next = pfile->u_buff; | |
1855 | pfile->u_buff = buff; | |
1856 | result = buff->cur; | |
1857 | } | |
1858 | ||
1859 | buff->cur = result + len; | |
1860 | return result; | |
1861 | } | |
1862 | ||
1e0ef2fd | 1863 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
1864 | That buffer is used for growing allocations when saving macro | |
1865 | replacement lists in a #define, and when parsing an answer to an | |
1866 | assertion in #assert, #unassert or #if (and therefore possibly | |
1867 | whilst expanding macros). It therefore must not be used by any | |
1868 | code that they might call: specifically the lexer and the guts of | |
1869 | the macro expander. | |
1870 | ||
1871 | All existing other uses clearly fit this restriction: storing | |
1872 | registered pragmas during initialization. */ | |
79bd622b | 1873 | unsigned char * |
f7fdd7a1 | 1874 | _cpp_aligned_alloc (cpp_reader *pfile, size_t len) |
89b05ef6 | 1875 | { |
e6a5f963 | 1876 | _cpp_buff *buff = pfile->a_buff; |
1877 | unsigned char *result = buff->cur; | |
89b05ef6 | 1878 | |
e6a5f963 | 1879 | if (len > (size_t) (buff->limit - result)) |
89b05ef6 | 1880 | { |
e6a5f963 | 1881 | buff = _cpp_get_buff (pfile, len); |
1882 | buff->next = pfile->a_buff; | |
1883 | pfile->a_buff = buff; | |
1884 | result = buff->cur; | |
89b05ef6 | 1885 | } |
f80e83a9 | 1886 | |
e6a5f963 | 1887 | buff->cur = result + len; |
79bd622b | 1888 | return result; |
f80e83a9 | 1889 | } |
c39ed964 | 1890 | |
1891 | /* Say which field of TOK is in use. */ | |
1892 | ||
1893 | enum cpp_token_fld_kind | |
1894 | cpp_token_val_index (cpp_token *tok) | |
1895 | { | |
1896 | switch (TOKEN_SPELL (tok)) | |
1897 | { | |
1898 | case SPELL_IDENT: | |
1899 | return CPP_TOKEN_FLD_NODE; | |
1900 | case SPELL_LITERAL: | |
1901 | return CPP_TOKEN_FLD_STR; | |
941f2388 | 1902 | case SPELL_OPERATOR: |
1903 | if (tok->type == CPP_PASTE) | |
1904 | return CPP_TOKEN_FLD_ARG_NO; | |
1905 | else | |
1906 | return CPP_TOKEN_FLD_NONE; | |
c39ed964 | 1907 | case SPELL_NONE: |
1908 | if (tok->type == CPP_MACRO_ARG) | |
1909 | return CPP_TOKEN_FLD_ARG_NO; | |
1910 | else if (tok->type == CPP_PADDING) | |
1911 | return CPP_TOKEN_FLD_SOURCE; | |
d6d3c909 | 1912 | else if (tok->type == CPP_PRAGMA) |
b75b98aa | 1913 | return CPP_TOKEN_FLD_PRAGMA; |
c39ed964 | 1914 | /* else fall through */ |
1915 | default: | |
1916 | return CPP_TOKEN_FLD_NONE; | |
1917 | } | |
1918 | } |