]>
Commit | Line | Data |
---|---|---|
45b966db | 1 | /* CPP Library - lexical analysis. |
5d8ebbd8 | 2 | Copyright (C) 2000, 2001, 2002 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 | |
c5a04734 | 7 | Single-pass line tokenization by Neil Booth, April 2000 |
45b966db ZW |
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 | |
11 | Free Software Foundation; either version 2, or (at your option) any | |
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 | |
20 | along with this program; if not, write to the Free Software | |
21 | Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ | |
22 | ||
23 | #include "config.h" | |
24 | #include "system.h" | |
45b966db ZW |
25 | #include "cpplib.h" |
26 | #include "cpphash.h" | |
27 | ||
c8a96070 NB |
28 | /* MULTIBYTE_CHARS support only works for native compilers. |
29 | ??? Ideally what we want is to model widechar support after | |
30 | the current floating point support. */ | |
31 | #ifdef CROSS_COMPILE | |
32 | #undef MULTIBYTE_CHARS | |
33 | #endif | |
34 | ||
35 | #ifdef MULTIBYTE_CHARS | |
36 | #include "mbchar.h" | |
37 | #include <locale.h> | |
38 | #endif | |
39 | ||
93c80368 NB |
40 | /* Tokens with SPELL_STRING store their spelling in the token list, |
41 | and it's length in the token->val.name.len. */ | |
42 | enum spell_type | |
f9a0e96c | 43 | { |
93c80368 NB |
44 | SPELL_OPERATOR = 0, |
45 | SPELL_CHAR, | |
46 | SPELL_IDENT, | |
47ad4138 | 47 | SPELL_NUMBER, |
93c80368 NB |
48 | SPELL_STRING, |
49 | SPELL_NONE | |
f9a0e96c ZW |
50 | }; |
51 | ||
93c80368 | 52 | struct token_spelling |
f9a0e96c | 53 | { |
93c80368 NB |
54 | enum spell_type category; |
55 | const unsigned char *name; | |
f9a0e96c ZW |
56 | }; |
57 | ||
8206c799 ZW |
58 | static const unsigned char *const digraph_spellings[] = |
59 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
93c80368 NB |
60 | |
61 | #define OP(e, s) { SPELL_OPERATOR, U s }, | |
62 | #define TK(e, s) { s, U STRINGX (e) }, | |
8206c799 | 63 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
93c80368 NB |
64 | #undef OP |
65 | #undef TK | |
66 | ||
67 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
68 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
480709cc | 69 | #define BACKUP() do {buffer->cur = buffer->backup_to;} while (0) |
f2d5f0cc | 70 | |
87062813 NB |
71 | static void handle_newline PARAMS ((cpp_reader *)); |
72 | static cppchar_t skip_escaped_newlines PARAMS ((cpp_reader *)); | |
29401c30 | 73 | static cppchar_t get_effective_char PARAMS ((cpp_reader *)); |
0d9f234d | 74 | |
041c3194 | 75 | static int skip_block_comment PARAMS ((cpp_reader *)); |
cbcff6df | 76 | static int skip_line_comment PARAMS ((cpp_reader *)); |
0d9f234d | 77 | static void adjust_column PARAMS ((cpp_reader *)); |
4d6baafa | 78 | static int skip_whitespace PARAMS ((cpp_reader *, cppchar_t)); |
2c3fcba6 | 79 | static cpp_hashnode *parse_identifier PARAMS ((cpp_reader *)); |
10cf9bde NB |
80 | static U_CHAR *parse_slow PARAMS ((cpp_reader *, const U_CHAR *, int, |
81 | unsigned int *)); | |
82 | static void parse_number PARAMS ((cpp_reader *, cpp_string *, int)); | |
93c80368 | 83 | static int unescaped_terminator_p PARAMS ((cpp_reader *, const U_CHAR *)); |
0d9f234d | 84 | static void parse_string PARAMS ((cpp_reader *, cpp_token *, cppchar_t)); |
87062813 | 85 | static bool trigraph_p PARAMS ((cpp_reader *)); |
0d9f234d | 86 | static void save_comment PARAMS ((cpp_reader *, cpp_token *, const U_CHAR *)); |
93c80368 | 87 | static int name_p PARAMS ((cpp_reader *, const cpp_string *)); |
62729350 NB |
88 | static int maybe_read_ucs PARAMS ((cpp_reader *, const unsigned char **, |
89 | const unsigned char *, unsigned int *)); | |
5fddcffc | 90 | static tokenrun *next_tokenrun PARAMS ((tokenrun *)); |
f617b8e2 | 91 | |
c8a96070 | 92 | static unsigned int hex_digit_value PARAMS ((unsigned int)); |
6142088c | 93 | static _cpp_buff *new_buff PARAMS ((size_t)); |
15dad1d9 | 94 | |
041c3194 | 95 | /* Utility routine: |
9e62c811 | 96 | |
bfb9dc7f ZW |
97 | Compares, the token TOKEN to the NUL-terminated string STRING. |
98 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
041c3194 | 99 | int |
bfb9dc7f ZW |
100 | cpp_ideq (token, string) |
101 | const cpp_token *token; | |
041c3194 ZW |
102 | const char *string; |
103 | { | |
bfb9dc7f | 104 | if (token->type != CPP_NAME) |
041c3194 | 105 | return 0; |
bfb9dc7f | 106 | |
a28c5035 | 107 | return !ustrcmp (NODE_NAME (token->val.node), (const U_CHAR *) string); |
15dad1d9 | 108 | } |
1368ee70 | 109 | |
87062813 NB |
110 | /* Call when meeting a newline, assumed to be in buffer->cur[-1]. |
111 | Returns with buffer->cur pointing to the character immediately | |
112 | following the newline (combination). */ | |
113 | static void | |
114 | handle_newline (pfile) | |
1444f2ed | 115 | cpp_reader *pfile; |
0d9f234d | 116 | { |
87062813 | 117 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 118 | |
87062813 | 119 | /* Handle CR-LF and LF-CR. Most other implementations (e.g. java) |
4d6baafa NB |
120 | only accept CR-LF; maybe we should fall back to that behaviour? */ |
121 | if (buffer->cur[-1] + buffer->cur[0] == '\r' + '\n') | |
87062813 | 122 | buffer->cur++; |
0d9f234d | 123 | |
87062813 NB |
124 | buffer->line_base = buffer->cur; |
125 | buffer->col_adjust = 0; | |
126 | pfile->line++; | |
0d9f234d NB |
127 | } |
128 | ||
87062813 NB |
129 | /* Subroutine of skip_escaped_newlines; called when a 3-character |
130 | sequence beginning with "??" is encountered. buffer->cur points to | |
131 | the second '?'. | |
132 | ||
133 | Warn if necessary, and returns true if the sequence forms a | |
134 | trigraph and the trigraph should be honoured. */ | |
135 | static bool | |
136 | trigraph_p (pfile) | |
45b966db | 137 | cpp_reader *pfile; |
45b966db | 138 | { |
87062813 NB |
139 | cpp_buffer *buffer = pfile->buffer; |
140 | cppchar_t from_char = buffer->cur[1]; | |
141 | bool accept; | |
142 | ||
143 | if (!_cpp_trigraph_map[from_char]) | |
144 | return false; | |
145 | ||
146 | accept = CPP_OPTION (pfile, trigraphs); | |
147 | ||
cbcff6df NB |
148 | /* Don't warn about trigraphs in comments. */ |
149 | if (CPP_OPTION (pfile, warn_trigraphs) && !pfile->state.lexing_comment) | |
45b966db | 150 | { |
041c3194 | 151 | if (accept) |
87062813 | 152 | cpp_warning_with_line (pfile, pfile->line, CPP_BUF_COL (buffer) - 1, |
041c3194 | 153 | "trigraph ??%c converted to %c", |
0d9f234d NB |
154 | (int) from_char, |
155 | (int) _cpp_trigraph_map[from_char]); | |
4a5b68a2 NB |
156 | else if (buffer->cur != buffer->last_Wtrigraphs) |
157 | { | |
158 | buffer->last_Wtrigraphs = buffer->cur; | |
67821e3a | 159 | cpp_warning_with_line (pfile, pfile->line, |
87062813 | 160 | CPP_BUF_COL (buffer) - 1, |
4a5b68a2 NB |
161 | "trigraph ??%c ignored", (int) from_char); |
162 | } | |
45b966db | 163 | } |
0d9f234d | 164 | |
041c3194 | 165 | return accept; |
45b966db ZW |
166 | } |
167 | ||
87062813 | 168 | /* Skips any escaped newlines introduced by '?' or a '\\', assumed to |
480709cc NB |
169 | lie in buffer->cur[-1]. Returns the next byte, which will be in |
170 | buffer->cur[-1]. This routine performs preprocessing stages 1 and | |
171 | 2 of the ISO C standard. */ | |
0d9f234d | 172 | static cppchar_t |
87062813 | 173 | skip_escaped_newlines (pfile) |
29401c30 | 174 | cpp_reader *pfile; |
45b966db | 175 | { |
29401c30 | 176 | cpp_buffer *buffer = pfile->buffer; |
87062813 | 177 | cppchar_t next = buffer->cur[-1]; |
29401c30 | 178 | |
a5c3cccd NB |
179 | /* Only do this if we apply stages 1 and 2. */ |
180 | if (!buffer->from_stage3) | |
041c3194 | 181 | { |
a5c3cccd | 182 | const unsigned char *saved_cur; |
87062813 | 183 | cppchar_t next1; |
a5c3cccd NB |
184 | |
185 | do | |
0d9f234d | 186 | { |
a5c3cccd | 187 | if (next == '?') |
0d9f234d | 188 | { |
4d6baafa | 189 | if (buffer->cur[0] != '?' || !trigraph_p (pfile)) |
87062813 | 190 | break; |
a5c3cccd | 191 | |
87062813 NB |
192 | /* Translate the trigraph. */ |
193 | next = _cpp_trigraph_map[buffer->cur[1]]; | |
194 | buffer->cur += 2; | |
4d6baafa | 195 | if (next != '\\') |
a5c3cccd | 196 | break; |
a5c3cccd NB |
197 | } |
198 | ||
4d6baafa NB |
199 | if (buffer->cur == buffer->rlimit) |
200 | break; | |
201 | ||
87062813 NB |
202 | /* We have a backslash, and room for at least one more |
203 | character. Skip horizontal whitespace. */ | |
204 | saved_cur = buffer->cur; | |
a5c3cccd | 205 | do |
87062813 NB |
206 | next1 = *buffer->cur++; |
207 | while (is_nvspace (next1) && buffer->cur < buffer->rlimit); | |
041c3194 | 208 | |
a5c3cccd | 209 | if (!is_vspace (next1)) |
0d9f234d | 210 | { |
87062813 | 211 | buffer->cur = saved_cur; |
0d9f234d NB |
212 | break; |
213 | } | |
45b966db | 214 | |
87062813 NB |
215 | if (saved_cur != buffer->cur - 1 |
216 | && !pfile->state.lexing_comment) | |
29401c30 | 217 | cpp_warning (pfile, "backslash and newline separated by space"); |
0d9f234d | 218 | |
87062813 | 219 | handle_newline (pfile); |
480709cc | 220 | buffer->backup_to = buffer->cur; |
87062813 NB |
221 | if (buffer->cur == buffer->rlimit) |
222 | { | |
223 | cpp_pedwarn (pfile, "backslash-newline at end of file"); | |
224 | next = EOF; | |
225 | } | |
226 | else | |
227 | next = *buffer->cur++; | |
0d9f234d | 228 | } |
a5c3cccd | 229 | while (next == '\\' || next == '?'); |
041c3194 | 230 | } |
45b966db | 231 | |
0d9f234d | 232 | return next; |
45b966db ZW |
233 | } |
234 | ||
0d9f234d | 235 | /* Obtain the next character, after trigraph conversion and skipping |
87062813 NB |
236 | an arbitrarily long string of escaped newlines. The common case of |
237 | no trigraphs or escaped newlines falls through quickly. On return, | |
480709cc NB |
238 | buffer->backup_to points to where to return to if the character is |
239 | not to be processed. */ | |
0d9f234d | 240 | static cppchar_t |
29401c30 NB |
241 | get_effective_char (pfile) |
242 | cpp_reader *pfile; | |
64aaf407 | 243 | { |
4d6baafa | 244 | cppchar_t next; |
480709cc | 245 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 246 | |
480709cc | 247 | buffer->backup_to = buffer->cur; |
4d6baafa NB |
248 | next = *buffer->cur++; |
249 | if (__builtin_expect (next == '?' || next == '\\', 0)) | |
250 | next = skip_escaped_newlines (pfile); | |
0d9f234d | 251 | |
480709cc | 252 | return next; |
64aaf407 NB |
253 | } |
254 | ||
0d9f234d NB |
255 | /* Skip a C-style block comment. We find the end of the comment by |
256 | seeing if an asterisk is before every '/' we encounter. Returns | |
257 | non-zero if comment terminated by EOF, zero otherwise. */ | |
041c3194 ZW |
258 | static int |
259 | skip_block_comment (pfile) | |
45b966db ZW |
260 | cpp_reader *pfile; |
261 | { | |
041c3194 | 262 | cpp_buffer *buffer = pfile->buffer; |
d8090680 | 263 | cppchar_t c = EOF, prevc = EOF; |
0d9f234d | 264 | |
cbcff6df | 265 | pfile->state.lexing_comment = 1; |
0d9f234d | 266 | while (buffer->cur != buffer->rlimit) |
45b966db | 267 | { |
0d9f234d NB |
268 | prevc = c, c = *buffer->cur++; |
269 | ||
0d9f234d | 270 | /* FIXME: For speed, create a new character class of characters |
93c80368 | 271 | of interest inside block comments. */ |
0d9f234d | 272 | if (c == '?' || c == '\\') |
87062813 | 273 | c = skip_escaped_newlines (pfile); |
041c3194 | 274 | |
0d9f234d NB |
275 | /* People like decorating comments with '*', so check for '/' |
276 | instead for efficiency. */ | |
041c3194 | 277 | if (c == '/') |
45b966db | 278 | { |
0d9f234d NB |
279 | if (prevc == '*') |
280 | break; | |
041c3194 | 281 | |
0d9f234d | 282 | /* Warn about potential nested comments, but not if the '/' |
a1f300c0 | 283 | comes immediately before the true comment delimiter. |
041c3194 | 284 | Don't bother to get it right across escaped newlines. */ |
0d9f234d | 285 | if (CPP_OPTION (pfile, warn_comments) |
87062813 NB |
286 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
287 | cpp_warning_with_line (pfile, | |
288 | pfile->line, CPP_BUF_COL (buffer), | |
289 | "\"/*\" within comment"); | |
45b966db | 290 | } |
91fcd158 | 291 | else if (is_vspace (c)) |
87062813 | 292 | handle_newline (pfile); |
52fadca8 | 293 | else if (c == '\t') |
0d9f234d | 294 | adjust_column (pfile); |
45b966db | 295 | } |
041c3194 | 296 | |
cbcff6df | 297 | pfile->state.lexing_comment = 0; |
0d9f234d | 298 | return c != '/' || prevc != '*'; |
45b966db ZW |
299 | } |
300 | ||
480709cc NB |
301 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
302 | terminating newline. Handles escaped newlines. Returns non-zero | |
303 | if a multiline comment. */ | |
041c3194 | 304 | static int |
cbcff6df NB |
305 | skip_line_comment (pfile) |
306 | cpp_reader *pfile; | |
45b966db | 307 | { |
cbcff6df | 308 | cpp_buffer *buffer = pfile->buffer; |
67821e3a | 309 | unsigned int orig_line = pfile->line; |
0d9f234d | 310 | cppchar_t c; |
041c3194 | 311 | |
cbcff6df | 312 | pfile->state.lexing_comment = 1; |
0d9f234d | 313 | do |
041c3194 | 314 | { |
0d9f234d | 315 | if (buffer->cur == buffer->rlimit) |
480709cc | 316 | goto at_eof; |
041c3194 | 317 | |
0d9f234d NB |
318 | c = *buffer->cur++; |
319 | if (c == '?' || c == '\\') | |
87062813 | 320 | c = skip_escaped_newlines (pfile); |
041c3194 | 321 | } |
0d9f234d | 322 | while (!is_vspace (c)); |
45b966db | 323 | |
480709cc NB |
324 | /* Step back over the newline, except at EOF. */ |
325 | buffer->cur--; | |
326 | at_eof: | |
327 | ||
cbcff6df | 328 | pfile->state.lexing_comment = 0; |
67821e3a | 329 | return orig_line != pfile->line; |
041c3194 | 330 | } |
45b966db | 331 | |
0d9f234d NB |
332 | /* pfile->buffer->cur is one beyond the \t character. Update |
333 | col_adjust so we track the column correctly. */ | |
52fadca8 | 334 | static void |
0d9f234d | 335 | adjust_column (pfile) |
52fadca8 | 336 | cpp_reader *pfile; |
52fadca8 | 337 | { |
0d9f234d NB |
338 | cpp_buffer *buffer = pfile->buffer; |
339 | unsigned int col = CPP_BUF_COL (buffer) - 1; /* Zero-based column. */ | |
52fadca8 NB |
340 | |
341 | /* Round it up to multiple of the tabstop, but subtract 1 since the | |
342 | tab itself occupies a character position. */ | |
0d9f234d NB |
343 | buffer->col_adjust += (CPP_OPTION (pfile, tabstop) |
344 | - col % CPP_OPTION (pfile, tabstop)) - 1; | |
52fadca8 NB |
345 | } |
346 | ||
0d9f234d NB |
347 | /* Skips whitespace, saving the next non-whitespace character. |
348 | Adjusts pfile->col_adjust to account for tabs. Without this, | |
349 | tokens might be assigned an incorrect column. */ | |
4d6baafa | 350 | static int |
0d9f234d | 351 | skip_whitespace (pfile, c) |
041c3194 | 352 | cpp_reader *pfile; |
0d9f234d | 353 | cppchar_t c; |
041c3194 ZW |
354 | { |
355 | cpp_buffer *buffer = pfile->buffer; | |
0d9f234d | 356 | unsigned int warned = 0; |
45b966db | 357 | |
0d9f234d | 358 | do |
041c3194 | 359 | { |
91fcd158 NB |
360 | /* Horizontal space always OK. */ |
361 | if (c == ' ') | |
0d9f234d | 362 | ; |
91fcd158 | 363 | else if (c == '\t') |
0d9f234d NB |
364 | adjust_column (pfile); |
365 | /* Just \f \v or \0 left. */ | |
91fcd158 | 366 | else if (c == '\0') |
041c3194 | 367 | { |
4d6baafa NB |
368 | if (buffer->cur - 1 == buffer->rlimit) |
369 | return 0; | |
91fcd158 | 370 | if (!warned) |
0d9f234d NB |
371 | { |
372 | cpp_warning (pfile, "null character(s) ignored"); | |
373 | warned = 1; | |
374 | } | |
45b966db | 375 | } |
93c80368 | 376 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
67821e3a | 377 | cpp_pedwarn_with_line (pfile, pfile->line, |
91fcd158 NB |
378 | CPP_BUF_COL (buffer), |
379 | "%s in preprocessing directive", | |
380 | c == '\f' ? "form feed" : "vertical tab"); | |
0d9f234d | 381 | |
0d9f234d | 382 | c = *buffer->cur++; |
45b966db | 383 | } |
ec5c56db | 384 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
0d9f234d NB |
385 | while (is_nvspace (c)); |
386 | ||
480709cc | 387 | buffer->cur--; |
4d6baafa | 388 | return 1; |
041c3194 | 389 | } |
45b966db | 390 | |
93c80368 NB |
391 | /* See if the characters of a number token are valid in a name (no |
392 | '.', '+' or '-'). */ | |
393 | static int | |
394 | name_p (pfile, string) | |
395 | cpp_reader *pfile; | |
396 | const cpp_string *string; | |
397 | { | |
398 | unsigned int i; | |
399 | ||
400 | for (i = 0; i < string->len; i++) | |
401 | if (!is_idchar (string->text[i])) | |
402 | return 0; | |
403 | ||
404 | return 1; | |
405 | } | |
406 | ||
2c3fcba6 ZW |
407 | /* Parse an identifier, skipping embedded backslash-newlines. This is |
408 | a critical inner loop. The common case is an identifier which has | |
409 | not been split by backslash-newline, does not contain a dollar | |
410 | sign, and has already been scanned (roughly 10:1 ratio of | |
411 | seen:unseen identifiers in normal code; the distribution is | |
412 | Poisson-like). Second most common case is a new identifier, not | |
413 | split and no dollar sign. The other possibilities are rare and | |
10cf9bde | 414 | have been relegated to parse_slow. */ |
0d9f234d | 415 | static cpp_hashnode * |
2c3fcba6 | 416 | parse_identifier (pfile) |
45b966db | 417 | cpp_reader *pfile; |
45b966db | 418 | { |
93c80368 | 419 | cpp_hashnode *result; |
10cf9bde | 420 | const U_CHAR *cur, *base; |
2c3fcba6 ZW |
421 | |
422 | /* Fast-path loop. Skim over a normal identifier. | |
423 | N.B. ISIDNUM does not include $. */ | |
4d6baafa NB |
424 | cur = pfile->buffer->cur; |
425 | while (ISIDNUM (*cur)) | |
2c3fcba6 | 426 | cur++; |
2c3fcba6 ZW |
427 | |
428 | /* Check for slow-path cases. */ | |
4d6baafa | 429 | if (*cur == '?' || *cur == '\\' || *cur == '$') |
10cf9bde NB |
430 | { |
431 | unsigned int len; | |
432 | ||
433 | base = parse_slow (pfile, cur, 0, &len); | |
434 | result = (cpp_hashnode *) | |
435 | ht_lookup (pfile->hash_table, base, len, HT_ALLOCED); | |
436 | } | |
2c3fcba6 ZW |
437 | else |
438 | { | |
10cf9bde NB |
439 | base = pfile->buffer->cur - 1; |
440 | pfile->buffer->cur = cur; | |
2c3fcba6 ZW |
441 | result = (cpp_hashnode *) |
442 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
2c3fcba6 ZW |
443 | } |
444 | ||
445 | /* Rarely, identifiers require diagnostics when lexed. | |
446 | XXX Has to be forced out of the fast path. */ | |
447 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) | |
448 | && !pfile->state.skipping, 0)) | |
449 | { | |
450 | /* It is allowed to poison the same identifier twice. */ | |
451 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
452 | cpp_error (pfile, "attempt to use poisoned \"%s\"", | |
453 | NODE_NAME (result)); | |
454 | ||
455 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
456 | replacement list of a variadic macro. */ | |
457 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
458 | && !pfile->state.va_args_ok) | |
459 | cpp_pedwarn (pfile, | |
460 | "__VA_ARGS__ can only appear in the expansion of a C99 variadic macro"); | |
461 | } | |
462 | ||
463 | return result; | |
464 | } | |
465 | ||
10cf9bde NB |
466 | /* Slow path. This handles numbers and identifiers which have been |
467 | split, or contain dollar signs. The part of the token from | |
468 | PFILE->buffer->cur-1 to CUR has already been scanned. NUMBER_P is | |
469 | 1 if it's a number, and 2 if it has a leading period. Returns a | |
470 | pointer to the token's NUL-terminated spelling in permanent | |
471 | storage, and sets PLEN to its length. */ | |
472 | static U_CHAR * | |
473 | parse_slow (pfile, cur, number_p, plen) | |
2c3fcba6 ZW |
474 | cpp_reader *pfile; |
475 | const U_CHAR *cur; | |
10cf9bde NB |
476 | int number_p; |
477 | unsigned int *plen; | |
2c3fcba6 | 478 | { |
0d9f234d | 479 | cpp_buffer *buffer = pfile->buffer; |
2c3fcba6 | 480 | const U_CHAR *base = buffer->cur - 1; |
2a967f3d | 481 | struct obstack *stack = &pfile->hash_table->stack; |
10cf9bde NB |
482 | unsigned int c, prevc, saw_dollar = 0; |
483 | ||
484 | /* Place any leading period. */ | |
485 | if (number_p == 2) | |
486 | obstack_1grow (stack, '.'); | |
2c3fcba6 ZW |
487 | |
488 | /* Copy the part of the token which is known to be okay. */ | |
489 | obstack_grow (stack, base, cur - base); | |
041c3194 | 490 | |
2c3fcba6 ZW |
491 | /* Now process the part which isn't. We are looking at one of |
492 | '$', '\\', or '?' on entry to this loop. */ | |
10cf9bde | 493 | prevc = cur[-1]; |
2c3fcba6 ZW |
494 | c = *cur++; |
495 | buffer->cur = cur; | |
10cf9bde | 496 | for (;;) |
041c3194 | 497 | { |
10cf9bde NB |
498 | /* Potential escaped newline? */ |
499 | buffer->backup_to = buffer->cur - 1; | |
500 | if (c == '?' || c == '\\') | |
501 | c = skip_escaped_newlines (pfile); | |
502 | ||
503 | if (!is_idchar (c)) | |
504 | { | |
505 | if (!number_p) | |
506 | break; | |
507 | if (c != '.' && !VALID_SIGN (c, prevc)) | |
508 | break; | |
509 | } | |
510 | ||
511 | /* Handle normal identifier characters in this loop. */ | |
512 | do | |
2c3fcba6 | 513 | { |
10cf9bde | 514 | prevc = c; |
2c3fcba6 | 515 | obstack_1grow (stack, c); |
45b966db | 516 | |
2c3fcba6 ZW |
517 | if (c == '$') |
518 | saw_dollar++; | |
ba89d661 | 519 | |
2c3fcba6 ZW |
520 | c = *buffer->cur++; |
521 | } | |
10cf9bde | 522 | while (is_idchar (c)); |
041c3194 | 523 | } |
0d9f234d | 524 | |
4d6baafa | 525 | /* Step back over the unwanted char. */ |
480709cc | 526 | BACKUP (); |
93c80368 | 527 | |
4fe9b91c | 528 | /* $ is not an identifier character in the standard, but is commonly |
0d9f234d NB |
529 | accepted as an extension. Don't warn about it in skipped |
530 | conditional blocks. */ | |
cef0d199 | 531 | if (saw_dollar && CPP_PEDANTIC (pfile) && ! pfile->state.skipping) |
10cf9bde | 532 | cpp_pedwarn (pfile, "'$' character(s) in identifier or number"); |
0d9f234d | 533 | |
10cf9bde NB |
534 | /* Identifiers and numbers are null-terminated. */ |
535 | *plen = obstack_object_size (stack); | |
2a967f3d | 536 | obstack_1grow (stack, '\0'); |
10cf9bde | 537 | return obstack_finish (stack); |
45b966db ZW |
538 | } |
539 | ||
5d8ebbd8 NB |
540 | /* Parse a number, beginning with character C, skipping embedded |
541 | backslash-newlines. LEADING_PERIOD is non-zero if there was a "." | |
542 | before C. Place the result in NUMBER. */ | |
45b966db | 543 | static void |
10cf9bde | 544 | parse_number (pfile, number, leading_period) |
45b966db | 545 | cpp_reader *pfile; |
0d9f234d | 546 | cpp_string *number; |
93c80368 | 547 | int leading_period; |
45b966db | 548 | { |
10cf9bde | 549 | const U_CHAR *cur; |
45b966db | 550 | |
10cf9bde NB |
551 | /* Fast-path loop. Skim over a normal number. |
552 | N.B. ISIDNUM does not include $. */ | |
553 | cur = pfile->buffer->cur; | |
554 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
555 | cur++; | |
cbcff6df | 556 | |
10cf9bde NB |
557 | /* Check for slow-path cases. */ |
558 | if (*cur == '?' || *cur == '\\' || *cur == '$') | |
559 | number->text = parse_slow (pfile, cur, 1 + leading_period, &number->len); | |
560 | else | |
041c3194 | 561 | { |
10cf9bde NB |
562 | const U_CHAR *base = pfile->buffer->cur - 1; |
563 | U_CHAR *dest; | |
0d9f234d | 564 | |
10cf9bde NB |
565 | number->len = cur - base + leading_period; |
566 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
567 | dest[number->len] = '\0'; | |
568 | number->text = dest; | |
45b966db | 569 | |
10cf9bde NB |
570 | if (leading_period) |
571 | *dest++ = '.'; | |
572 | memcpy (dest, base, cur - base); | |
573 | pfile->buffer->cur = cur; | |
45b966db | 574 | } |
0d9f234d NB |
575 | } |
576 | ||
93c80368 NB |
577 | /* Subroutine of parse_string. */ |
578 | static int | |
579 | unescaped_terminator_p (pfile, dest) | |
580 | cpp_reader *pfile; | |
581 | const unsigned char *dest; | |
582 | { | |
583 | const unsigned char *start, *temp; | |
584 | ||
585 | /* In #include-style directives, terminators are not escapeable. */ | |
586 | if (pfile->state.angled_headers) | |
587 | return 1; | |
588 | ||
ece54d54 | 589 | start = BUFF_FRONT (pfile->u_buff); |
93c80368 NB |
590 | |
591 | /* An odd number of consecutive backslashes represents an escaped | |
592 | terminator. */ | |
593 | for (temp = dest; temp > start && temp[-1] == '\\'; temp--) | |
594 | ; | |
595 | ||
596 | return ((dest - temp) & 1) == 0; | |
597 | } | |
598 | ||
0d9f234d | 599 | /* Parses a string, character constant, or angle-bracketed header file |
7868b4a2 NB |
600 | name. Handles embedded trigraphs and escaped newlines. The stored |
601 | string is guaranteed NUL-terminated, but it is not guaranteed that | |
602 | this is the first NUL since embedded NULs are preserved. | |
45b966db | 603 | |
87062813 NB |
604 | When this function returns, buffer->cur points to the next |
605 | character to be processed. */ | |
041c3194 | 606 | static void |
0d9f234d | 607 | parse_string (pfile, token, terminator) |
45b966db | 608 | cpp_reader *pfile; |
041c3194 | 609 | cpp_token *token; |
0d9f234d | 610 | cppchar_t terminator; |
45b966db | 611 | { |
041c3194 | 612 | cpp_buffer *buffer = pfile->buffer; |
93c80368 | 613 | unsigned char *dest, *limit; |
0d9f234d | 614 | cppchar_t c; |
d4e6133f | 615 | bool warned_nulls = false; |
0d9f234d | 616 | |
ece54d54 NB |
617 | dest = BUFF_FRONT (pfile->u_buff); |
618 | limit = BUFF_LIMIT (pfile->u_buff); | |
93c80368 | 619 | |
0d9f234d | 620 | for (;;) |
45b966db | 621 | { |
87062813 | 622 | /* We need room for another char, possibly the terminating NUL. */ |
ece54d54 NB |
623 | if ((size_t) (limit - dest) < 1) |
624 | { | |
625 | size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff); | |
8c3b2693 | 626 | _cpp_extend_buff (pfile, &pfile->u_buff, 2); |
ece54d54 NB |
627 | dest = BUFF_FRONT (pfile->u_buff) + len_so_far; |
628 | limit = BUFF_LIMIT (pfile->u_buff); | |
629 | } | |
7868b4a2 | 630 | |
0d9f234d | 631 | /* Handle trigraphs, escaped newlines etc. */ |
87062813 | 632 | c = *buffer->cur++; |
0d9f234d | 633 | if (c == '?' || c == '\\') |
87062813 | 634 | c = skip_escaped_newlines (pfile); |
45b966db | 635 | |
87062813 | 636 | if (c == terminator) |
45b966db | 637 | { |
87062813 NB |
638 | if (unescaped_terminator_p (pfile, dest)) |
639 | break; | |
0d9f234d NB |
640 | } |
641 | else if (is_vspace (c)) | |
642 | { | |
d4e6133f NB |
643 | /* No string literal may extend over multiple lines. In |
644 | assembly language, suppress the error except for <> | |
645 | includes. This is a kludge around not knowing where | |
646 | comments are. */ | |
647 | unterminated: | |
648 | if (CPP_OPTION (pfile, lang) != CLK_ASM || terminator == '>') | |
649 | cpp_error (pfile, "missing terminating %c character", terminator); | |
650 | buffer->cur--; | |
651 | break; | |
0d9f234d | 652 | } |
4d6baafa | 653 | else if (c == '\0') |
0d9f234d | 654 | { |
4d6baafa | 655 | if (buffer->cur - 1 == buffer->rlimit) |
d4e6133f | 656 | goto unterminated; |
4d6baafa NB |
657 | if (!warned_nulls) |
658 | { | |
659 | warned_nulls = true; | |
660 | cpp_warning (pfile, "null character(s) preserved in literal"); | |
661 | } | |
45b966db | 662 | } |
45b966db | 663 | |
93c80368 | 664 | *dest++ = c; |
45b966db ZW |
665 | } |
666 | ||
7868b4a2 | 667 | *dest = '\0'; |
45b966db | 668 | |
ece54d54 NB |
669 | token->val.str.text = BUFF_FRONT (pfile->u_buff); |
670 | token->val.str.len = dest - BUFF_FRONT (pfile->u_buff); | |
671 | BUFF_FRONT (pfile->u_buff) = dest + 1; | |
0d9f234d | 672 | } |
041c3194 | 673 | |
93c80368 | 674 | /* The stored comment includes the comment start and any terminator. */ |
9e62c811 | 675 | static void |
0d9f234d NB |
676 | save_comment (pfile, token, from) |
677 | cpp_reader *pfile; | |
041c3194 ZW |
678 | cpp_token *token; |
679 | const unsigned char *from; | |
9e62c811 | 680 | { |
041c3194 | 681 | unsigned char *buffer; |
0d9f234d | 682 | unsigned int len; |
0d9f234d | 683 | |
1c6d33ef | 684 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
480709cc | 685 | |
3542203b NB |
686 | /* C++ comments probably (not definitely) have moved past a new |
687 | line, which we don't want to save in the comment. */ | |
480709cc | 688 | if (is_vspace (pfile->buffer->cur[-1])) |
3542203b | 689 | len--; |
ece54d54 | 690 | buffer = _cpp_unaligned_alloc (pfile, len); |
041c3194 | 691 | |
041c3194 | 692 | token->type = CPP_COMMENT; |
bfb9dc7f | 693 | token->val.str.len = len; |
0d9f234d | 694 | token->val.str.text = buffer; |
45b966db | 695 | |
1c6d33ef NB |
696 | buffer[0] = '/'; |
697 | memcpy (buffer + 1, from, len - 1); | |
0d9f234d | 698 | } |
45b966db | 699 | |
5fddcffc NB |
700 | /* Allocate COUNT tokens for RUN. */ |
701 | void | |
702 | _cpp_init_tokenrun (run, count) | |
703 | tokenrun *run; | |
704 | unsigned int count; | |
705 | { | |
706 | run->base = xnewvec (cpp_token, count); | |
707 | run->limit = run->base + count; | |
708 | run->next = NULL; | |
709 | } | |
710 | ||
711 | /* Returns the next tokenrun, or creates one if there is none. */ | |
712 | static tokenrun * | |
713 | next_tokenrun (run) | |
714 | tokenrun *run; | |
715 | { | |
716 | if (run->next == NULL) | |
717 | { | |
718 | run->next = xnew (tokenrun); | |
bdcbe496 | 719 | run->next->prev = run; |
5fddcffc NB |
720 | _cpp_init_tokenrun (run->next, 250); |
721 | } | |
722 | ||
723 | return run->next; | |
724 | } | |
725 | ||
4ed5bcfb NB |
726 | /* Allocate a single token that is invalidated at the same time as the |
727 | rest of the tokens on the line. Has its line and col set to the | |
728 | same as the last lexed token, so that diagnostics appear in the | |
729 | right place. */ | |
730 | cpp_token * | |
731 | _cpp_temp_token (pfile) | |
732 | cpp_reader *pfile; | |
733 | { | |
734 | cpp_token *old, *result; | |
735 | ||
736 | old = pfile->cur_token - 1; | |
737 | if (pfile->cur_token == pfile->cur_run->limit) | |
738 | { | |
739 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
740 | pfile->cur_token = pfile->cur_run->base; | |
741 | } | |
742 | ||
743 | result = pfile->cur_token++; | |
744 | result->line = old->line; | |
745 | result->col = old->col; | |
746 | return result; | |
747 | } | |
748 | ||
14baae01 NB |
749 | /* Lex a token into RESULT (external interface). Takes care of issues |
750 | like directive handling, token lookahead, multiple include | |
a1f300c0 | 751 | optimization and skipping. */ |
345894b4 NB |
752 | const cpp_token * |
753 | _cpp_lex_token (pfile) | |
45b966db | 754 | cpp_reader *pfile; |
5fddcffc | 755 | { |
bdcbe496 | 756 | cpp_token *result; |
5fddcffc | 757 | |
bdcbe496 | 758 | for (;;) |
5fddcffc | 759 | { |
bdcbe496 | 760 | if (pfile->cur_token == pfile->cur_run->limit) |
5fddcffc | 761 | { |
bdcbe496 NB |
762 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
763 | pfile->cur_token = pfile->cur_run->base; | |
5fddcffc NB |
764 | } |
765 | ||
bdcbe496 | 766 | if (pfile->lookaheads) |
14baae01 NB |
767 | { |
768 | pfile->lookaheads--; | |
769 | result = pfile->cur_token++; | |
770 | } | |
bdcbe496 | 771 | else |
14baae01 | 772 | result = _cpp_lex_direct (pfile); |
bdcbe496 NB |
773 | |
774 | if (result->flags & BOL) | |
5fddcffc | 775 | { |
bdcbe496 NB |
776 | /* Is this a directive. If _cpp_handle_directive returns |
777 | false, it is an assembler #. */ | |
778 | if (result->type == CPP_HASH | |
e808ec9c NB |
779 | /* 6.10.3 p 11: Directives in a list of macro arguments |
780 | gives undefined behavior. This implementation | |
781 | handles the directive as normal. */ | |
782 | && pfile->state.parsing_args != 1 | |
bdcbe496 NB |
783 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
784 | continue; | |
97293897 NB |
785 | if (pfile->cb.line_change && !pfile->state.skipping) |
786 | (*pfile->cb.line_change)(pfile, result, pfile->state.parsing_args); | |
5fddcffc | 787 | } |
5fddcffc | 788 | |
bdcbe496 NB |
789 | /* We don't skip tokens in directives. */ |
790 | if (pfile->state.in_directive) | |
791 | break; | |
5fddcffc | 792 | |
bdcbe496 | 793 | /* Outside a directive, invalidate controlling macros. At file |
14baae01 | 794 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
bdcbe496 | 795 | get here and MI optimisation works. */ |
5fddcffc | 796 | pfile->mi_valid = false; |
bdcbe496 NB |
797 | |
798 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
799 | break; | |
5fddcffc NB |
800 | } |
801 | ||
345894b4 | 802 | return result; |
5fddcffc NB |
803 | } |
804 | ||
480709cc NB |
805 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
806 | do { \ | |
807 | if (get_effective_char (pfile) == CHAR) \ | |
808 | result->type = THEN_TYPE; \ | |
809 | else \ | |
810 | { \ | |
811 | BACKUP (); \ | |
812 | result->type = ELSE_TYPE; \ | |
813 | } \ | |
814 | } while (0) | |
815 | ||
14baae01 NB |
816 | /* Lex a token into pfile->cur_token, which is also incremented, to |
817 | get diagnostics pointing to the correct location. | |
818 | ||
819 | Does not handle issues such as token lookahead, multiple-include | |
820 | optimisation, directives, skipping etc. This function is only | |
821 | suitable for use by _cpp_lex_token, and in special cases like | |
822 | lex_expansion_token which doesn't care for any of these issues. | |
823 | ||
824 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
825 | otherwise returns to the start of the token buffer if permissible. | |
826 | Returns the location of the lexed token. */ | |
827 | cpp_token * | |
828 | _cpp_lex_direct (pfile) | |
5fddcffc | 829 | cpp_reader *pfile; |
45b966db | 830 | { |
0d9f234d | 831 | cppchar_t c; |
adb84b42 | 832 | cpp_buffer *buffer; |
0d9f234d | 833 | const unsigned char *comment_start; |
14baae01 | 834 | cpp_token *result = pfile->cur_token++; |
9ec7291f | 835 | |
5fddcffc | 836 | fresh_line: |
adb84b42 | 837 | buffer = pfile->buffer; |
bd969772 NB |
838 | result->flags = buffer->saved_flags; |
839 | buffer->saved_flags = 0; | |
5fddcffc | 840 | update_tokens_line: |
1444f2ed | 841 | result->line = pfile->line; |
041c3194 | 842 | |
5fddcffc | 843 | skipped_white: |
480709cc | 844 | c = *buffer->cur++; |
5fddcffc | 845 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
5fddcffc NB |
846 | |
847 | trigraph: | |
0d9f234d | 848 | switch (c) |
45b966db | 849 | { |
4d6baafa NB |
850 | case ' ': case '\t': case '\f': case '\v': case '\0': |
851 | result->flags |= PREV_WHITE; | |
852 | if (skip_whitespace (pfile, c)) | |
853 | goto skipped_white; | |
854 | ||
855 | /* EOF. */ | |
856 | buffer->cur--; | |
bdcbe496 | 857 | buffer->saved_flags = BOL; |
5fddcffc | 858 | if (!pfile->state.parsing_args && !pfile->state.in_directive) |
ef6e958a | 859 | { |
bdcbe496 | 860 | if (buffer->cur != buffer->line_base) |
5fddcffc NB |
861 | { |
862 | /* Non-empty files should end in a newline. Don't warn | |
863 | for command line and _Pragma buffers. */ | |
864 | if (!buffer->from_stage3) | |
865 | cpp_pedwarn (pfile, "no newline at end of file"); | |
87062813 | 866 | handle_newline (pfile); |
7364fdd8 | 867 | } |
bdcbe496 NB |
868 | |
869 | /* Don't pop the last buffer. */ | |
870 | if (buffer->prev) | |
871 | { | |
872 | unsigned char stop = buffer->return_at_eof; | |
873 | ||
874 | _cpp_pop_buffer (pfile); | |
875 | if (!stop) | |
876 | goto fresh_line; | |
877 | } | |
ef6e958a | 878 | } |
0d9f234d | 879 | result->type = CPP_EOF; |
5fddcffc | 880 | break; |
45b966db | 881 | |
0d9f234d | 882 | case '\n': case '\r': |
87062813 | 883 | handle_newline (pfile); |
bdcbe496 NB |
884 | buffer->saved_flags = BOL; |
885 | if (! pfile->state.in_directive) | |
45b966db | 886 | { |
4ed5bcfb NB |
887 | if (pfile->state.parsing_args == 2) |
888 | buffer->saved_flags |= PREV_WHITE; | |
bdcbe496 NB |
889 | if (!pfile->keep_tokens) |
890 | { | |
891 | pfile->cur_run = &pfile->base_run; | |
892 | result = pfile->base_run.base; | |
893 | pfile->cur_token = result + 1; | |
894 | } | |
895 | goto fresh_line; | |
45b966db | 896 | } |
5fddcffc NB |
897 | result->type = CPP_EOF; |
898 | break; | |
46d07497 | 899 | |
0d9f234d NB |
900 | case '?': |
901 | case '\\': | |
902 | /* These could start an escaped newline, or '?' a trigraph. Let | |
903 | skip_escaped_newlines do all the work. */ | |
904 | { | |
67821e3a | 905 | unsigned int line = pfile->line; |
0d9f234d | 906 | |
87062813 | 907 | c = skip_escaped_newlines (pfile); |
67821e3a | 908 | if (line != pfile->line) |
87062813 | 909 | { |
480709cc | 910 | buffer->cur--; |
87062813 NB |
911 | /* We had at least one escaped newline of some sort. |
912 | Update the token's line and column. */ | |
5fddcffc | 913 | goto update_tokens_line; |
87062813 | 914 | } |
480709cc | 915 | } |
0d9f234d | 916 | |
480709cc NB |
917 | /* We are either the original '?' or '\\', or a trigraph. */ |
918 | if (c == '?') | |
0d9f234d | 919 | result->type = CPP_QUERY; |
480709cc NB |
920 | else if (c == '\\') |
921 | goto random_char; | |
922 | else | |
923 | goto trigraph; | |
0d9f234d | 924 | break; |
46d07497 | 925 | |
0d9f234d NB |
926 | case '0': case '1': case '2': case '3': case '4': |
927 | case '5': case '6': case '7': case '8': case '9': | |
928 | result->type = CPP_NUMBER; | |
10cf9bde | 929 | parse_number (pfile, &result->val.str, 0); |
0d9f234d | 930 | break; |
46d07497 | 931 | |
0abc6a6a NB |
932 | case 'L': |
933 | /* 'L' may introduce wide characters or strings. */ | |
934 | { | |
935 | const unsigned char *pos = buffer->cur; | |
0d9f234d | 936 | |
0abc6a6a NB |
937 | c = get_effective_char (pfile); |
938 | if (c == '\'' || c == '"') | |
939 | { | |
940 | result->type = (c == '"' ? CPP_WSTRING: CPP_WCHAR); | |
941 | parse_string (pfile, result, c); | |
942 | break; | |
943 | } | |
944 | buffer->cur = pos; | |
945 | } | |
946 | /* Fall through. */ | |
947 | ||
948 | start_ident: | |
0d9f234d NB |
949 | case '_': |
950 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
951 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
952 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
953 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
954 | case 'y': case 'z': | |
955 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
0abc6a6a | 956 | case 'G': case 'H': case 'I': case 'J': case 'K': |
0d9f234d NB |
957 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
958 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
959 | case 'Y': case 'Z': | |
960 | result->type = CPP_NAME; | |
2c3fcba6 | 961 | result->val.node = parse_identifier (pfile); |
0d9f234d | 962 | |
0d9f234d | 963 | /* Convert named operators to their proper types. */ |
0abc6a6a | 964 | if (result->val.node->flags & NODE_OPERATOR) |
0d9f234d NB |
965 | { |
966 | result->flags |= NAMED_OP; | |
93c80368 | 967 | result->type = result->val.node->value.operator; |
0d9f234d NB |
968 | } |
969 | break; | |
970 | ||
971 | case '\'': | |
972 | case '"': | |
973 | result->type = c == '"' ? CPP_STRING: CPP_CHAR; | |
0d9f234d NB |
974 | parse_string (pfile, result, c); |
975 | break; | |
041c3194 | 976 | |
0d9f234d | 977 | case '/': |
1c6d33ef NB |
978 | /* A potential block or line comment. */ |
979 | comment_start = buffer->cur; | |
29401c30 | 980 | c = get_effective_char (pfile); |
480709cc | 981 | |
1c6d33ef NB |
982 | if (c == '*') |
983 | { | |
0d9f234d | 984 | if (skip_block_comment (pfile)) |
67821e3a | 985 | cpp_error (pfile, "unterminated comment"); |
0d9f234d | 986 | } |
480709cc NB |
987 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
988 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
0d9f234d | 989 | { |
bdb05a7b NB |
990 | /* Warn about comments only if pedantically GNUC89, and not |
991 | in system headers. */ | |
992 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
a94c1199 | 993 | && ! buffer->warned_cplusplus_comments) |
041c3194 | 994 | { |
1c6d33ef NB |
995 | cpp_pedwarn (pfile, |
996 | "C++ style comments are not allowed in ISO C89"); | |
997 | cpp_pedwarn (pfile, | |
998 | "(this will be reported only once per input file)"); | |
999 | buffer->warned_cplusplus_comments = 1; | |
1000 | } | |
0d9f234d | 1001 | |
01ef6563 | 1002 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
50410426 | 1003 | cpp_warning (pfile, "multi-line comment"); |
1c6d33ef | 1004 | } |
480709cc NB |
1005 | else if (c == '=') |
1006 | { | |
1007 | result->type = CPP_DIV_EQ; | |
1008 | break; | |
1009 | } | |
1010 | else | |
1011 | { | |
1012 | BACKUP (); | |
1013 | result->type = CPP_DIV; | |
1014 | break; | |
1015 | } | |
0d9f234d | 1016 | |
1c6d33ef NB |
1017 | if (!pfile->state.save_comments) |
1018 | { | |
1019 | result->flags |= PREV_WHITE; | |
5fddcffc | 1020 | goto update_tokens_line; |
0d9f234d | 1021 | } |
1c6d33ef NB |
1022 | |
1023 | /* Save the comment as a token in its own right. */ | |
1024 | save_comment (pfile, result, comment_start); | |
bdcbe496 | 1025 | break; |
0d9f234d NB |
1026 | |
1027 | case '<': | |
1028 | if (pfile->state.angled_headers) | |
1029 | { | |
1030 | result->type = CPP_HEADER_NAME; | |
480709cc NB |
1031 | parse_string (pfile, result, '>'); |
1032 | break; | |
0d9f234d | 1033 | } |
45b966db | 1034 | |
29401c30 | 1035 | c = get_effective_char (pfile); |
0d9f234d | 1036 | if (c == '=') |
480709cc | 1037 | result->type = CPP_LESS_EQ; |
0d9f234d | 1038 | else if (c == '<') |
480709cc | 1039 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); |
0d9f234d | 1040 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1041 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); |
0d9f234d NB |
1042 | else if (c == ':' && CPP_OPTION (pfile, digraphs)) |
1043 | { | |
480709cc | 1044 | result->type = CPP_OPEN_SQUARE; |
0d9f234d NB |
1045 | result->flags |= DIGRAPH; |
1046 | } | |
1047 | else if (c == '%' && CPP_OPTION (pfile, digraphs)) | |
1048 | { | |
480709cc | 1049 | result->type = CPP_OPEN_BRACE; |
0d9f234d NB |
1050 | result->flags |= DIGRAPH; |
1051 | } | |
480709cc NB |
1052 | else |
1053 | { | |
1054 | BACKUP (); | |
1055 | result->type = CPP_LESS; | |
1056 | } | |
0d9f234d NB |
1057 | break; |
1058 | ||
1059 | case '>': | |
29401c30 | 1060 | c = get_effective_char (pfile); |
0d9f234d | 1061 | if (c == '=') |
480709cc | 1062 | result->type = CPP_GREATER_EQ; |
0d9f234d | 1063 | else if (c == '>') |
480709cc | 1064 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); |
0d9f234d | 1065 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc NB |
1066 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); |
1067 | else | |
0d9f234d | 1068 | { |
480709cc NB |
1069 | BACKUP (); |
1070 | result->type = CPP_GREATER; | |
0d9f234d NB |
1071 | } |
1072 | break; | |
1073 | ||
cbcff6df | 1074 | case '%': |
480709cc NB |
1075 | c = get_effective_char (pfile); |
1076 | if (c == '=') | |
1077 | result->type = CPP_MOD_EQ; | |
1078 | else if (CPP_OPTION (pfile, digraphs) && c == ':') | |
1079 | { | |
1080 | result->flags |= DIGRAPH; | |
1081 | result->type = CPP_HASH; | |
1082 | if (get_effective_char (pfile) == '%') | |
1083 | { | |
1084 | const unsigned char *pos = buffer->cur; | |
1085 | ||
1086 | if (get_effective_char (pfile) == ':') | |
1087 | result->type = CPP_PASTE; | |
1088 | else | |
1089 | buffer->cur = pos - 1; | |
1090 | } | |
1091 | else | |
1092 | BACKUP (); | |
1093 | } | |
1094 | else if (CPP_OPTION (pfile, digraphs) && c == '>') | |
1095 | { | |
1096 | result->flags |= DIGRAPH; | |
1097 | result->type = CPP_CLOSE_BRACE; | |
1098 | } | |
1099 | else | |
1100 | { | |
1101 | BACKUP (); | |
1102 | result->type = CPP_MOD; | |
1103 | } | |
0d9f234d NB |
1104 | break; |
1105 | ||
cbcff6df | 1106 | case '.': |
480709cc NB |
1107 | result->type = CPP_DOT; |
1108 | c = get_effective_char (pfile); | |
1109 | if (c == '.') | |
1110 | { | |
1111 | const unsigned char *pos = buffer->cur; | |
1112 | ||
1113 | if (get_effective_char (pfile) == '.') | |
1114 | result->type = CPP_ELLIPSIS; | |
1115 | else | |
1116 | buffer->cur = pos - 1; | |
1117 | } | |
1118 | /* All known character sets have 0...9 contiguous. */ | |
0df6c2c7 | 1119 | else if (ISDIGIT (c)) |
480709cc NB |
1120 | { |
1121 | result->type = CPP_NUMBER; | |
10cf9bde | 1122 | parse_number (pfile, &result->val.str, 1); |
480709cc NB |
1123 | } |
1124 | else if (c == '*' && CPP_OPTION (pfile, cplusplus)) | |
1125 | result->type = CPP_DOT_STAR; | |
1126 | else | |
1127 | BACKUP (); | |
0d9f234d | 1128 | break; |
45b966db | 1129 | |
0d9f234d | 1130 | case '+': |
29401c30 | 1131 | c = get_effective_char (pfile); |
480709cc NB |
1132 | if (c == '+') |
1133 | result->type = CPP_PLUS_PLUS; | |
1134 | else if (c == '=') | |
1135 | result->type = CPP_PLUS_EQ; | |
1136 | else | |
1137 | { | |
1138 | BACKUP (); | |
1139 | result->type = CPP_PLUS; | |
1140 | } | |
0d9f234d | 1141 | break; |
04e3ec78 | 1142 | |
0d9f234d | 1143 | case '-': |
29401c30 | 1144 | c = get_effective_char (pfile); |
0d9f234d NB |
1145 | if (c == '>') |
1146 | { | |
480709cc NB |
1147 | result->type = CPP_DEREF; |
1148 | if (CPP_OPTION (pfile, cplusplus)) | |
1149 | { | |
1150 | if (get_effective_char (pfile) == '*') | |
1151 | result->type = CPP_DEREF_STAR; | |
1152 | else | |
1153 | BACKUP (); | |
1154 | } | |
0d9f234d | 1155 | } |
0d9f234d | 1156 | else if (c == '-') |
480709cc NB |
1157 | result->type = CPP_MINUS_MINUS; |
1158 | else if (c == '=') | |
1159 | result->type = CPP_MINUS_EQ; | |
1160 | else | |
1161 | { | |
1162 | BACKUP (); | |
1163 | result->type = CPP_MINUS; | |
1164 | } | |
0d9f234d | 1165 | break; |
45b966db | 1166 | |
0d9f234d | 1167 | case '&': |
29401c30 | 1168 | c = get_effective_char (pfile); |
480709cc NB |
1169 | if (c == '&') |
1170 | result->type = CPP_AND_AND; | |
1171 | else if (c == '=') | |
1172 | result->type = CPP_AND_EQ; | |
1173 | else | |
1174 | { | |
1175 | BACKUP (); | |
1176 | result->type = CPP_AND; | |
1177 | } | |
0d9f234d NB |
1178 | break; |
1179 | ||
0d9f234d | 1180 | case '|': |
29401c30 | 1181 | c = get_effective_char (pfile); |
480709cc NB |
1182 | if (c == '|') |
1183 | result->type = CPP_OR_OR; | |
1184 | else if (c == '=') | |
1185 | result->type = CPP_OR_EQ; | |
1186 | else | |
1187 | { | |
1188 | BACKUP (); | |
1189 | result->type = CPP_OR; | |
1190 | } | |
0d9f234d | 1191 | break; |
45b966db | 1192 | |
0d9f234d | 1193 | case ':': |
29401c30 | 1194 | c = get_effective_char (pfile); |
0d9f234d | 1195 | if (c == ':' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1196 | result->type = CPP_SCOPE; |
0d9f234d NB |
1197 | else if (c == '>' && CPP_OPTION (pfile, digraphs)) |
1198 | { | |
1199 | result->flags |= DIGRAPH; | |
480709cc NB |
1200 | result->type = CPP_CLOSE_SQUARE; |
1201 | } | |
1202 | else | |
1203 | { | |
1204 | BACKUP (); | |
1205 | result->type = CPP_COLON; | |
0d9f234d NB |
1206 | } |
1207 | break; | |
45b966db | 1208 | |
480709cc NB |
1209 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1210 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1211 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1212 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1213 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1214 | ||
0d9f234d NB |
1215 | case '~': result->type = CPP_COMPL; break; |
1216 | case ',': result->type = CPP_COMMA; break; | |
1217 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1218 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1219 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1220 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1221 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1222 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1223 | case ';': result->type = CPP_SEMICOLON; break; | |
1224 | ||
cc937581 ZW |
1225 | /* @ is a punctuator in Objective C. */ |
1226 | case '@': result->type = CPP_ATSIGN; break; | |
0d9f234d | 1227 | |
0abc6a6a NB |
1228 | case '$': |
1229 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
1230 | goto start_ident; | |
1231 | /* Fall through... */ | |
1232 | ||
0d9f234d NB |
1233 | random_char: |
1234 | default: | |
1235 | result->type = CPP_OTHER; | |
6c53ebff | 1236 | result->val.c = c; |
0d9f234d NB |
1237 | break; |
1238 | } | |
bdcbe496 NB |
1239 | |
1240 | return result; | |
0d9f234d NB |
1241 | } |
1242 | ||
5d8ebbd8 | 1243 | /* An upper bound on the number of bytes needed to spell TOKEN, |
93c80368 NB |
1244 | including preceding whitespace. */ |
1245 | unsigned int | |
1246 | cpp_token_len (token) | |
1247 | const cpp_token *token; | |
0d9f234d | 1248 | { |
93c80368 | 1249 | unsigned int len; |
6d2c2047 | 1250 | |
93c80368 | 1251 | switch (TOKEN_SPELL (token)) |
041c3194 | 1252 | { |
a28c5035 | 1253 | default: len = 0; break; |
47ad4138 | 1254 | case SPELL_NUMBER: |
a28c5035 NB |
1255 | case SPELL_STRING: len = token->val.str.len; break; |
1256 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; | |
041c3194 | 1257 | } |
47ad4138 | 1258 | /* 1 for whitespace, 4 for comment delimiters. */ |
93c80368 | 1259 | return len + 5; |
6d2c2047 ZW |
1260 | } |
1261 | ||
041c3194 | 1262 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
cf00a885 ZW |
1263 | already contain the enough space to hold the token's spelling. |
1264 | Returns a pointer to the character after the last character | |
1265 | written. */ | |
93c80368 NB |
1266 | unsigned char * |
1267 | cpp_spell_token (pfile, token, buffer) | |
041c3194 ZW |
1268 | cpp_reader *pfile; /* Would be nice to be rid of this... */ |
1269 | const cpp_token *token; | |
1270 | unsigned char *buffer; | |
1271 | { | |
96be6998 | 1272 | switch (TOKEN_SPELL (token)) |
041c3194 ZW |
1273 | { |
1274 | case SPELL_OPERATOR: | |
1275 | { | |
1276 | const unsigned char *spelling; | |
1277 | unsigned char c; | |
d6d5f795 | 1278 | |
041c3194 | 1279 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1280 | spelling |
1281 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
92936ecf ZW |
1282 | else if (token->flags & NAMED_OP) |
1283 | goto spell_ident; | |
041c3194 | 1284 | else |
96be6998 | 1285 | spelling = TOKEN_NAME (token); |
041c3194 ZW |
1286 | |
1287 | while ((c = *spelling++) != '\0') | |
1288 | *buffer++ = c; | |
1289 | } | |
1290 | break; | |
d6d5f795 | 1291 | |
47ad4138 ZW |
1292 | case SPELL_CHAR: |
1293 | *buffer++ = token->val.c; | |
1294 | break; | |
1295 | ||
1296 | spell_ident: | |
041c3194 | 1297 | case SPELL_IDENT: |
a28c5035 NB |
1298 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1299 | buffer += NODE_LEN (token->val.node); | |
041c3194 | 1300 | break; |
d6d5f795 | 1301 | |
47ad4138 ZW |
1302 | case SPELL_NUMBER: |
1303 | memcpy (buffer, token->val.str.text, token->val.str.len); | |
1304 | buffer += token->val.str.len; | |
1305 | break; | |
1306 | ||
041c3194 ZW |
1307 | case SPELL_STRING: |
1308 | { | |
ba89d661 ZW |
1309 | int left, right, tag; |
1310 | switch (token->type) | |
1311 | { | |
1312 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1313 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
ba89d661 ZW |
1314 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1315 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1316 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 ZW |
1317 | default: |
1318 | cpp_ice (pfile, "unknown string token %s\n", TOKEN_NAME (token)); | |
1319 | return buffer; | |
ba89d661 ZW |
1320 | } |
1321 | if (tag) *buffer++ = tag; | |
47ad4138 | 1322 | *buffer++ = left; |
bfb9dc7f ZW |
1323 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1324 | buffer += token->val.str.len; | |
47ad4138 | 1325 | *buffer++ = right; |
041c3194 ZW |
1326 | } |
1327 | break; | |
d6d5f795 | 1328 | |
041c3194 | 1329 | case SPELL_NONE: |
1f978f5f | 1330 | cpp_ice (pfile, "unspellable token %s", TOKEN_NAME (token)); |
041c3194 ZW |
1331 | break; |
1332 | } | |
d6d5f795 | 1333 | |
041c3194 ZW |
1334 | return buffer; |
1335 | } | |
d6d5f795 | 1336 | |
5d8ebbd8 NB |
1337 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1338 | freed when the reader is destroyed. Useful for diagnostics. */ | |
93c80368 NB |
1339 | unsigned char * |
1340 | cpp_token_as_text (pfile, token) | |
c5a04734 | 1341 | cpp_reader *pfile; |
041c3194 | 1342 | const cpp_token *token; |
c5a04734 | 1343 | { |
93c80368 | 1344 | unsigned int len = cpp_token_len (token); |
ece54d54 | 1345 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
c5a04734 | 1346 | |
93c80368 NB |
1347 | end = cpp_spell_token (pfile, token, start); |
1348 | end[0] = '\0'; | |
c5a04734 | 1349 | |
93c80368 NB |
1350 | return start; |
1351 | } | |
c5a04734 | 1352 | |
5d8ebbd8 NB |
1353 | /* Used by C front ends, which really should move to using |
1354 | cpp_token_as_text. */ | |
93c80368 NB |
1355 | const char * |
1356 | cpp_type2name (type) | |
1357 | enum cpp_ttype type; | |
1358 | { | |
1359 | return (const char *) token_spellings[type].name; | |
1360 | } | |
c5a04734 | 1361 | |
4ed5bcfb NB |
1362 | /* Writes the spelling of token to FP, without any preceding space. |
1363 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1364 | double-buffering. */ | |
93c80368 NB |
1365 | void |
1366 | cpp_output_token (token, fp) | |
1367 | const cpp_token *token; | |
1368 | FILE *fp; | |
1369 | { | |
93c80368 | 1370 | switch (TOKEN_SPELL (token)) |
c5a04734 | 1371 | { |
93c80368 NB |
1372 | case SPELL_OPERATOR: |
1373 | { | |
1374 | const unsigned char *spelling; | |
3b681e9d | 1375 | int c; |
c5a04734 | 1376 | |
93c80368 | 1377 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1378 | spelling |
1379 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
93c80368 NB |
1380 | else if (token->flags & NAMED_OP) |
1381 | goto spell_ident; | |
1382 | else | |
1383 | spelling = TOKEN_NAME (token); | |
041c3194 | 1384 | |
3b681e9d ZW |
1385 | c = *spelling; |
1386 | do | |
1387 | putc (c, fp); | |
1388 | while ((c = *++spelling) != '\0'); | |
93c80368 NB |
1389 | } |
1390 | break; | |
041c3194 | 1391 | |
47ad4138 ZW |
1392 | case SPELL_CHAR: |
1393 | putc (token->val.c, fp); | |
1394 | break; | |
1395 | ||
93c80368 NB |
1396 | spell_ident: |
1397 | case SPELL_IDENT: | |
3b681e9d | 1398 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
93c80368 | 1399 | break; |
041c3194 | 1400 | |
47ad4138 ZW |
1401 | case SPELL_NUMBER: |
1402 | fwrite (token->val.str.text, 1, token->val.str.len, fp); | |
1403 | break; | |
1404 | ||
93c80368 NB |
1405 | case SPELL_STRING: |
1406 | { | |
1407 | int left, right, tag; | |
1408 | switch (token->type) | |
1409 | { | |
1410 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1411 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
93c80368 NB |
1412 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1413 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1414 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 ZW |
1415 | default: |
1416 | fprintf (stderr, "impossible STRING token %s\n", TOKEN_NAME (token)); | |
1417 | return; | |
93c80368 NB |
1418 | } |
1419 | if (tag) putc (tag, fp); | |
47ad4138 | 1420 | putc (left, fp); |
93c80368 | 1421 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
47ad4138 | 1422 | putc (right, fp); |
93c80368 NB |
1423 | } |
1424 | break; | |
c5a04734 | 1425 | |
93c80368 NB |
1426 | case SPELL_NONE: |
1427 | /* An error, most probably. */ | |
1428 | break; | |
041c3194 | 1429 | } |
c5a04734 ZW |
1430 | } |
1431 | ||
93c80368 NB |
1432 | /* Compare two tokens. */ |
1433 | int | |
1434 | _cpp_equiv_tokens (a, b) | |
1435 | const cpp_token *a, *b; | |
c5a04734 | 1436 | { |
93c80368 NB |
1437 | if (a->type == b->type && a->flags == b->flags) |
1438 | switch (TOKEN_SPELL (a)) | |
1439 | { | |
1440 | default: /* Keep compiler happy. */ | |
1441 | case SPELL_OPERATOR: | |
1442 | return 1; | |
1443 | case SPELL_CHAR: | |
6c53ebff | 1444 | return a->val.c == b->val.c; /* Character. */ |
93c80368 | 1445 | case SPELL_NONE: |
56051c0a | 1446 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
93c80368 NB |
1447 | case SPELL_IDENT: |
1448 | return a->val.node == b->val.node; | |
47ad4138 | 1449 | case SPELL_NUMBER: |
93c80368 NB |
1450 | case SPELL_STRING: |
1451 | return (a->val.str.len == b->val.str.len | |
1452 | && !memcmp (a->val.str.text, b->val.str.text, | |
1453 | a->val.str.len)); | |
1454 | } | |
c5a04734 | 1455 | |
041c3194 ZW |
1456 | return 0; |
1457 | } | |
1458 | ||
93c80368 NB |
1459 | /* Returns nonzero if a space should be inserted to avoid an |
1460 | accidental token paste for output. For simplicity, it is | |
1461 | conservative, and occasionally advises a space where one is not | |
1462 | needed, e.g. "." and ".2". */ | |
93c80368 NB |
1463 | int |
1464 | cpp_avoid_paste (pfile, token1, token2) | |
c5a04734 | 1465 | cpp_reader *pfile; |
93c80368 | 1466 | const cpp_token *token1, *token2; |
c5a04734 | 1467 | { |
93c80368 NB |
1468 | enum cpp_ttype a = token1->type, b = token2->type; |
1469 | cppchar_t c; | |
c5a04734 | 1470 | |
93c80368 NB |
1471 | if (token1->flags & NAMED_OP) |
1472 | a = CPP_NAME; | |
1473 | if (token2->flags & NAMED_OP) | |
1474 | b = CPP_NAME; | |
c5a04734 | 1475 | |
93c80368 NB |
1476 | c = EOF; |
1477 | if (token2->flags & DIGRAPH) | |
37b8524c | 1478 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
93c80368 NB |
1479 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1480 | c = token_spellings[b].name[0]; | |
c5a04734 | 1481 | |
93c80368 | 1482 | /* Quickly get everything that can paste with an '='. */ |
37b8524c | 1483 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
93c80368 | 1484 | return 1; |
c5a04734 | 1485 | |
93c80368 | 1486 | switch (a) |
c5a04734 | 1487 | { |
93c80368 NB |
1488 | case CPP_GREATER: return c == '>' || c == '?'; |
1489 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1490 | case CPP_PLUS: return c == '+'; | |
1491 | case CPP_MINUS: return c == '-' || c == '>'; | |
1492 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1493 | case CPP_MOD: return c == ':' || c == '>'; | |
1494 | case CPP_AND: return c == '&'; | |
1495 | case CPP_OR: return c == '|'; | |
1496 | case CPP_COLON: return c == ':' || c == '>'; | |
1497 | case CPP_DEREF: return c == '*'; | |
26ec42ee | 1498 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
93c80368 NB |
1499 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1500 | case CPP_NAME: return ((b == CPP_NUMBER | |
1501 | && name_p (pfile, &token2->val.str)) | |
1502 | || b == CPP_NAME | |
1503 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1504 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1505 | || c == '.' || c == '+' || c == '-'); | |
1506 | case CPP_OTHER: return (CPP_OPTION (pfile, objc) | |
6c53ebff | 1507 | && token1->val.c == '@' |
93c80368 NB |
1508 | && (b == CPP_NAME || b == CPP_STRING)); |
1509 | default: break; | |
c5a04734 | 1510 | } |
c5a04734 | 1511 | |
417f3e3a | 1512 | return 0; |
c5a04734 ZW |
1513 | } |
1514 | ||
93c80368 | 1515 | /* Output all the remaining tokens on the current line, and a newline |
4ed5bcfb NB |
1516 | character, to FP. Leading whitespace is removed. If there are |
1517 | macros, special token padding is not performed. */ | |
c5a04734 | 1518 | void |
93c80368 | 1519 | cpp_output_line (pfile, fp) |
c5a04734 | 1520 | cpp_reader *pfile; |
93c80368 | 1521 | FILE *fp; |
c5a04734 | 1522 | { |
4ed5bcfb | 1523 | const cpp_token *token; |
96be6998 | 1524 | |
4ed5bcfb NB |
1525 | token = cpp_get_token (pfile); |
1526 | while (token->type != CPP_EOF) | |
96be6998 | 1527 | { |
4ed5bcfb NB |
1528 | cpp_output_token (token, fp); |
1529 | token = cpp_get_token (pfile); | |
1530 | if (token->flags & PREV_WHITE) | |
1531 | putc (' ', fp); | |
96be6998 ZW |
1532 | } |
1533 | ||
93c80368 | 1534 | putc ('\n', fp); |
041c3194 | 1535 | } |
c5a04734 | 1536 | |
c8a96070 NB |
1537 | /* Returns the value of a hexadecimal digit. */ |
1538 | static unsigned int | |
1539 | hex_digit_value (c) | |
1540 | unsigned int c; | |
1541 | { | |
9e1ac915 KG |
1542 | if (hex_p (c)) |
1543 | return hex_value (c); | |
1544 | else | |
1545 | abort (); | |
c8a96070 NB |
1546 | } |
1547 | ||
62729350 NB |
1548 | /* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. Returns 1 to indicate |
1549 | failure if cpplib is not parsing C++ or C99. Such failure is | |
1550 | silent, and no variables are updated. Otherwise returns 0, and | |
1551 | warns if -Wtraditional. | |
c8a96070 NB |
1552 | |
1553 | [lex.charset]: The character designated by the universal character | |
1554 | name \UNNNNNNNN is that character whose character short name in | |
1555 | ISO/IEC 10646 is NNNNNNNN; the character designated by the | |
1556 | universal character name \uNNNN is that character whose character | |
1557 | short name in ISO/IEC 10646 is 0000NNNN. If the hexadecimal value | |
1558 | for a universal character name is less than 0x20 or in the range | |
1559 | 0x7F-0x9F (inclusive), or if the universal character name | |
1560 | designates a character in the basic source character set, then the | |
1561 | program is ill-formed. | |
1562 | ||
1563 | We assume that wchar_t is Unicode, so we don't need to do any | |
62729350 | 1564 | mapping. Is this ever wrong? |
c8a96070 | 1565 | |
62729350 NB |
1566 | PC points to the 'u' or 'U', PSTR is points to the byte after PC, |
1567 | LIMIT is the end of the string or charconst. PSTR is updated to | |
1568 | point after the UCS on return, and the UCS is written into PC. */ | |
1569 | ||
1570 | static int | |
1571 | maybe_read_ucs (pfile, pstr, limit, pc) | |
c8a96070 NB |
1572 | cpp_reader *pfile; |
1573 | const unsigned char **pstr; | |
1574 | const unsigned char *limit; | |
62729350 | 1575 | unsigned int *pc; |
c8a96070 NB |
1576 | { |
1577 | const unsigned char *p = *pstr; | |
62729350 NB |
1578 | unsigned int code = 0; |
1579 | unsigned int c = *pc, length; | |
1580 | ||
1581 | /* Only attempt to interpret a UCS for C++ and C99. */ | |
1582 | if (! (CPP_OPTION (pfile, cplusplus) || CPP_OPTION (pfile, c99))) | |
1583 | return 1; | |
c8a96070 | 1584 | |
62729350 | 1585 | if (CPP_WTRADITIONAL (pfile)) |
f458d1d5 | 1586 | cpp_warning (pfile, "the meaning of '\\%c' is different in traditional C", c); |
c8a96070 | 1587 | |
f8710242 NB |
1588 | length = (c == 'u' ? 4: 8); |
1589 | ||
1590 | if ((size_t) (limit - p) < length) | |
1591 | { | |
1592 | cpp_error (pfile, "incomplete universal-character-name"); | |
1593 | /* Skip to the end to avoid more diagnostics. */ | |
1594 | p = limit; | |
1595 | } | |
1596 | else | |
1597 | { | |
1598 | for (; length; length--, p++) | |
c8a96070 | 1599 | { |
f8710242 NB |
1600 | c = *p; |
1601 | if (ISXDIGIT (c)) | |
1602 | code = (code << 4) + hex_digit_value (c); | |
1603 | else | |
1604 | { | |
1605 | cpp_error (pfile, | |
1606 | "non-hex digit '%c' in universal-character-name", c); | |
1607 | /* We shouldn't skip in case there are multibyte chars. */ | |
1608 | break; | |
1609 | } | |
c8a96070 | 1610 | } |
c8a96070 NB |
1611 | } |
1612 | ||
1613 | #ifdef TARGET_EBCDIC | |
1614 | cpp_error (pfile, "universal-character-name on EBCDIC target"); | |
1615 | code = 0x3f; /* EBCDIC invalid character */ | |
1616 | #else | |
f8710242 NB |
1617 | /* True extended characters are OK. */ |
1618 | if (code >= 0xa0 | |
1619 | && !(code & 0x80000000) | |
1620 | && !(code >= 0xD800 && code <= 0xDFFF)) | |
1621 | ; | |
1622 | /* The standard permits $, @ and ` to be specified as UCNs. We use | |
1623 | hex escapes so that this also works with EBCDIC hosts. */ | |
1624 | else if (code == 0x24 || code == 0x40 || code == 0x60) | |
1625 | ; | |
1626 | /* Don't give another error if one occurred above. */ | |
1627 | else if (length == 0) | |
1628 | cpp_error (pfile, "universal-character-name out of range"); | |
c8a96070 NB |
1629 | #endif |
1630 | ||
1631 | *pstr = p; | |
62729350 NB |
1632 | *pc = code; |
1633 | return 0; | |
c8a96070 NB |
1634 | } |
1635 | ||
1636 | /* Interpret an escape sequence, and return its value. PSTR points to | |
1637 | the input pointer, which is just after the backslash. LIMIT is how | |
62729350 | 1638 | much text we have. MASK is a bitmask for the precision for the |
f458d1d5 | 1639 | destination type (char or wchar_t). |
c8a96070 | 1640 | |
62729350 | 1641 | Handles all relevant diagnostics. */ |
62729350 | 1642 | unsigned int |
f458d1d5 | 1643 | cpp_parse_escape (pfile, pstr, limit, mask) |
c8a96070 NB |
1644 | cpp_reader *pfile; |
1645 | const unsigned char **pstr; | |
1646 | const unsigned char *limit; | |
62729350 | 1647 | unsigned HOST_WIDE_INT mask; |
c8a96070 NB |
1648 | { |
1649 | int unknown = 0; | |
1650 | const unsigned char *str = *pstr; | |
1651 | unsigned int c = *str++; | |
1652 | ||
1653 | switch (c) | |
1654 | { | |
1655 | case '\\': case '\'': case '"': case '?': break; | |
1656 | case 'b': c = TARGET_BS; break; | |
1657 | case 'f': c = TARGET_FF; break; | |
1658 | case 'n': c = TARGET_NEWLINE; break; | |
1659 | case 'r': c = TARGET_CR; break; | |
1660 | case 't': c = TARGET_TAB; break; | |
1661 | case 'v': c = TARGET_VT; break; | |
1662 | ||
1663 | case '(': case '{': case '[': case '%': | |
1664 | /* '\(', etc, are used at beginning of line to avoid confusing Emacs. | |
1665 | '\%' is used to prevent SCCS from getting confused. */ | |
1666 | unknown = CPP_PEDANTIC (pfile); | |
1667 | break; | |
1668 | ||
1669 | case 'a': | |
1670 | if (CPP_WTRADITIONAL (pfile)) | |
f458d1d5 | 1671 | cpp_warning (pfile, "the meaning of '\\a' is different in traditional C"); |
001e3fee | 1672 | c = TARGET_BELL; |
c8a96070 NB |
1673 | break; |
1674 | ||
1675 | case 'e': case 'E': | |
1676 | if (CPP_PEDANTIC (pfile)) | |
1677 | cpp_pedwarn (pfile, "non-ISO-standard escape sequence, '\\%c'", c); | |
1678 | c = TARGET_ESC; | |
1679 | break; | |
1680 | ||
c8a96070 | 1681 | case 'u': case 'U': |
62729350 | 1682 | unknown = maybe_read_ucs (pfile, &str, limit, &c); |
c8a96070 NB |
1683 | break; |
1684 | ||
1685 | case 'x': | |
1686 | if (CPP_WTRADITIONAL (pfile)) | |
f458d1d5 | 1687 | cpp_warning (pfile, "the meaning of '\\x' is different in traditional C"); |
c8a96070 | 1688 | |
c8a96070 NB |
1689 | { |
1690 | unsigned int i = 0, overflow = 0; | |
1691 | int digits_found = 0; | |
1692 | ||
1693 | while (str < limit) | |
1694 | { | |
1695 | c = *str; | |
1696 | if (! ISXDIGIT (c)) | |
1697 | break; | |
1698 | str++; | |
1699 | overflow |= i ^ (i << 4 >> 4); | |
1700 | i = (i << 4) + hex_digit_value (c); | |
1701 | digits_found = 1; | |
1702 | } | |
1703 | ||
1704 | if (!digits_found) | |
1705 | cpp_error (pfile, "\\x used with no following hex digits"); | |
1706 | ||
1707 | if (overflow | (i != (i & mask))) | |
1708 | { | |
1709 | cpp_pedwarn (pfile, "hex escape sequence out of range"); | |
1710 | i &= mask; | |
1711 | } | |
1712 | c = i; | |
1713 | } | |
1714 | break; | |
1715 | ||
1716 | case '0': case '1': case '2': case '3': | |
1717 | case '4': case '5': case '6': case '7': | |
1718 | { | |
1719 | unsigned int i = c - '0'; | |
1720 | int count = 0; | |
1721 | ||
1722 | while (str < limit && ++count < 3) | |
1723 | { | |
1724 | c = *str; | |
1725 | if (c < '0' || c > '7') | |
1726 | break; | |
1727 | str++; | |
1728 | i = (i << 3) + c - '0'; | |
1729 | } | |
1730 | ||
1731 | if (i != (i & mask)) | |
1732 | { | |
1733 | cpp_pedwarn (pfile, "octal escape sequence out of range"); | |
1734 | i &= mask; | |
1735 | } | |
1736 | c = i; | |
1737 | } | |
1738 | break; | |
1739 | ||
1740 | default: | |
1741 | unknown = 1; | |
1742 | break; | |
1743 | } | |
1744 | ||
1745 | if (unknown) | |
1746 | { | |
1747 | if (ISGRAPH (c)) | |
1748 | cpp_pedwarn (pfile, "unknown escape sequence '\\%c'", c); | |
1749 | else | |
1750 | cpp_pedwarn (pfile, "unknown escape sequence: '\\%03o'", c); | |
1751 | } | |
1752 | ||
62729350 NB |
1753 | if (c > mask) |
1754 | cpp_pedwarn (pfile, "escape sequence out of range for character"); | |
1755 | ||
c8a96070 NB |
1756 | *pstr = str; |
1757 | return c; | |
1758 | } | |
1759 | ||
1760 | #ifndef MAX_CHAR_TYPE_SIZE | |
1761 | #define MAX_CHAR_TYPE_SIZE CHAR_TYPE_SIZE | |
1762 | #endif | |
1763 | ||
1764 | #ifndef MAX_WCHAR_TYPE_SIZE | |
1765 | #define MAX_WCHAR_TYPE_SIZE WCHAR_TYPE_SIZE | |
1766 | #endif | |
1767 | ||
1768 | /* Interpret a (possibly wide) character constant in TOKEN. | |
f458d1d5 ZW |
1769 | WARN_MULTI warns about multi-character charconsts. PCHARS_SEEN points |
1770 | to a variable that is filled in with the number of characters seen. */ | |
c8a96070 | 1771 | HOST_WIDE_INT |
f458d1d5 | 1772 | cpp_interpret_charconst (pfile, token, warn_multi, pchars_seen) |
c8a96070 NB |
1773 | cpp_reader *pfile; |
1774 | const cpp_token *token; | |
1775 | int warn_multi; | |
c8a96070 NB |
1776 | unsigned int *pchars_seen; |
1777 | { | |
1778 | const unsigned char *str = token->val.str.text; | |
1779 | const unsigned char *limit = str + token->val.str.len; | |
1780 | unsigned int chars_seen = 0; | |
1781 | unsigned int width, max_chars, c; | |
2a967f3d NB |
1782 | unsigned HOST_WIDE_INT mask; |
1783 | HOST_WIDE_INT result = 0; | |
a47ed310 | 1784 | bool unsigned_p; |
c8a96070 NB |
1785 | |
1786 | #ifdef MULTIBYTE_CHARS | |
1787 | (void) local_mbtowc (NULL, NULL, 0); | |
1788 | #endif | |
1789 | ||
1790 | /* Width in bits. */ | |
1791 | if (token->type == CPP_CHAR) | |
a47ed310 NB |
1792 | { |
1793 | width = MAX_CHAR_TYPE_SIZE; | |
1794 | unsigned_p = CPP_OPTION (pfile, signed_char) == 0; | |
1795 | } | |
c8a96070 | 1796 | else |
a47ed310 NB |
1797 | { |
1798 | width = MAX_WCHAR_TYPE_SIZE; | |
1799 | unsigned_p = WCHAR_UNSIGNED; | |
1800 | } | |
c8a96070 NB |
1801 | |
1802 | if (width < HOST_BITS_PER_WIDE_INT) | |
1803 | mask = ((unsigned HOST_WIDE_INT) 1 << width) - 1; | |
1804 | else | |
1805 | mask = ~0; | |
1806 | max_chars = HOST_BITS_PER_WIDE_INT / width; | |
1807 | ||
1808 | while (str < limit) | |
1809 | { | |
1810 | #ifdef MULTIBYTE_CHARS | |
1811 | wchar_t wc; | |
1812 | int char_len; | |
1813 | ||
1814 | char_len = local_mbtowc (&wc, str, limit - str); | |
1815 | if (char_len == -1) | |
1816 | { | |
1817 | cpp_warning (pfile, "ignoring invalid multibyte character"); | |
1818 | c = *str++; | |
1819 | } | |
1820 | else | |
1821 | { | |
1822 | str += char_len; | |
1823 | c = wc; | |
1824 | } | |
1825 | #else | |
1826 | c = *str++; | |
1827 | #endif | |
1828 | ||
1829 | if (c == '\\') | |
f458d1d5 | 1830 | c = cpp_parse_escape (pfile, &str, limit, mask); |
c8a96070 NB |
1831 | |
1832 | #ifdef MAP_CHARACTER | |
1833 | if (ISPRINT (c)) | |
1834 | c = MAP_CHARACTER (c); | |
1835 | #endif | |
1836 | ||
1837 | /* Merge character into result; ignore excess chars. */ | |
1838 | if (++chars_seen <= max_chars) | |
1839 | { | |
1840 | if (width < HOST_BITS_PER_WIDE_INT) | |
1841 | result = (result << width) | (c & mask); | |
1842 | else | |
1843 | result = c; | |
1844 | } | |
1845 | } | |
1846 | ||
1847 | if (chars_seen == 0) | |
1848 | cpp_error (pfile, "empty character constant"); | |
1849 | else if (chars_seen > max_chars) | |
1850 | { | |
1851 | chars_seen = max_chars; | |
f8710242 | 1852 | cpp_warning (pfile, "character constant too long"); |
c8a96070 | 1853 | } |
f458d1d5 | 1854 | else if (chars_seen > 1 && warn_multi) |
c8a96070 NB |
1855 | cpp_warning (pfile, "multi-character character constant"); |
1856 | ||
a47ed310 NB |
1857 | /* If relevant type is signed, sign-extend the constant. */ |
1858 | if (chars_seen) | |
c8a96070 NB |
1859 | { |
1860 | unsigned int nbits = chars_seen * width; | |
c8a96070 | 1861 | |
e1e7d56b | 1862 | mask = (unsigned HOST_WIDE_INT) ~0 >> (HOST_BITS_PER_WIDE_INT - nbits); |
a47ed310 | 1863 | if (unsigned_p || ((result >> (nbits - 1)) & 1) == 0) |
c8a96070 NB |
1864 | result &= mask; |
1865 | else | |
1866 | result |= ~mask; | |
1867 | } | |
1868 | ||
1869 | *pchars_seen = chars_seen; | |
1870 | return result; | |
1871 | } | |
1872 | ||
1e013d2e NB |
1873 | /* Memory buffers. Changing these three constants can have a dramatic |
1874 | effect on performance. The values here are reasonable defaults, | |
1875 | but might be tuned. If you adjust them, be sure to test across a | |
1876 | range of uses of cpplib, including heavy nested function-like macro | |
1877 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1878 | good tool for this). */ | |
1879 | #define MIN_BUFF_SIZE 8000 | |
87062813 | 1880 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
1e013d2e NB |
1881 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1882 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
417f3e3a | 1883 | |
87062813 NB |
1884 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1885 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1886 | #endif | |
1887 | ||
93c80368 | 1888 | struct dummy |
417f3e3a | 1889 | { |
93c80368 NB |
1890 | char c; |
1891 | union | |
1892 | { | |
1893 | double d; | |
1894 | int *p; | |
1895 | } u; | |
1896 | }; | |
417f3e3a | 1897 | |
93c80368 | 1898 | #define DEFAULT_ALIGNMENT (offsetof (struct dummy, u)) |
b8af0ca5 NB |
1899 | #define CPP_ALIGN(size, align) (((size) + ((align) - 1)) & ~((align) - 1)) |
1900 | ||
c9e7a609 NB |
1901 | /* Create a new allocation buffer. Place the control block at the end |
1902 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
b8af0ca5 NB |
1903 | static _cpp_buff * |
1904 | new_buff (len) | |
6142088c | 1905 | size_t len; |
b8af0ca5 NB |
1906 | { |
1907 | _cpp_buff *result; | |
ece54d54 | 1908 | unsigned char *base; |
b8af0ca5 | 1909 | |
1e013d2e NB |
1910 | if (len < MIN_BUFF_SIZE) |
1911 | len = MIN_BUFF_SIZE; | |
b8af0ca5 NB |
1912 | len = CPP_ALIGN (len, DEFAULT_ALIGNMENT); |
1913 | ||
1914 | base = xmalloc (len + sizeof (_cpp_buff)); | |
1915 | result = (_cpp_buff *) (base + len); | |
1916 | result->base = base; | |
1917 | result->cur = base; | |
1918 | result->limit = base + len; | |
1919 | result->next = NULL; | |
1920 | return result; | |
1921 | } | |
1922 | ||
1923 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1924 | void | |
1925 | _cpp_release_buff (pfile, buff) | |
1926 | cpp_reader *pfile; | |
1927 | _cpp_buff *buff; | |
1928 | { | |
1929 | _cpp_buff *end = buff; | |
1930 | ||
1931 | while (end->next) | |
1932 | end = end->next; | |
1933 | end->next = pfile->free_buffs; | |
1934 | pfile->free_buffs = buff; | |
1935 | } | |
1936 | ||
1937 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1938 | _cpp_buff * | |
1939 | _cpp_get_buff (pfile, min_size) | |
1940 | cpp_reader *pfile; | |
6142088c | 1941 | size_t min_size; |
b8af0ca5 NB |
1942 | { |
1943 | _cpp_buff *result, **p; | |
1944 | ||
1945 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1946 | { | |
6142088c | 1947 | size_t size; |
1e013d2e NB |
1948 | |
1949 | if (*p == NULL) | |
b8af0ca5 | 1950 | return new_buff (min_size); |
1e013d2e NB |
1951 | result = *p; |
1952 | size = result->limit - result->base; | |
1953 | /* Return a buffer that's big enough, but don't waste one that's | |
1954 | way too big. */ | |
34f5271d | 1955 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
b8af0ca5 NB |
1956 | break; |
1957 | } | |
1958 | ||
1959 | *p = result->next; | |
1960 | result->next = NULL; | |
1961 | result->cur = result->base; | |
1962 | return result; | |
1963 | } | |
1964 | ||
4fe9b91c | 1965 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
1966 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
1967 | the excess bytes to the new buffer. Chains the new buffer after | |
1968 | BUFF, and returns the new buffer. */ | |
b8af0ca5 | 1969 | _cpp_buff * |
8c3b2693 | 1970 | _cpp_append_extend_buff (pfile, buff, min_extra) |
b8af0ca5 NB |
1971 | cpp_reader *pfile; |
1972 | _cpp_buff *buff; | |
6142088c | 1973 | size_t min_extra; |
b8af0ca5 | 1974 | { |
6142088c | 1975 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
8c3b2693 | 1976 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
b8af0ca5 | 1977 | |
8c3b2693 NB |
1978 | buff->next = new_buff; |
1979 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
1980 | return new_buff; | |
1981 | } | |
1982 | ||
4fe9b91c | 1983 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
1984 | remaining bytes of the buffer pointed to by BUFF, and at least |
1985 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
1986 | Chains the new buffer before the buffer pointed to by BUFF, and | |
1987 | updates the pointer to point to the new buffer. */ | |
1988 | void | |
1989 | _cpp_extend_buff (pfile, pbuff, min_extra) | |
1990 | cpp_reader *pfile; | |
1991 | _cpp_buff **pbuff; | |
1992 | size_t min_extra; | |
1993 | { | |
1994 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
1995 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
1996 | ||
1997 | new_buff = _cpp_get_buff (pfile, size); | |
1998 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
1999 | new_buff->next = old_buff; | |
2000 | *pbuff = new_buff; | |
b8af0ca5 NB |
2001 | } |
2002 | ||
2003 | /* Free a chain of buffers starting at BUFF. */ | |
2004 | void | |
2005 | _cpp_free_buff (buff) | |
2006 | _cpp_buff *buff; | |
2007 | { | |
2008 | _cpp_buff *next; | |
2009 | ||
2010 | for (; buff; buff = next) | |
2011 | { | |
2012 | next = buff->next; | |
2013 | free (buff->base); | |
2014 | } | |
2015 | } | |
417f3e3a | 2016 | |
ece54d54 NB |
2017 | /* Allocate permanent, unaligned storage of length LEN. */ |
2018 | unsigned char * | |
2019 | _cpp_unaligned_alloc (pfile, len) | |
2020 | cpp_reader *pfile; | |
2021 | size_t len; | |
2022 | { | |
2023 | _cpp_buff *buff = pfile->u_buff; | |
2024 | unsigned char *result = buff->cur; | |
2025 | ||
2026 | if (len > (size_t) (buff->limit - result)) | |
2027 | { | |
2028 | buff = _cpp_get_buff (pfile, len); | |
2029 | buff->next = pfile->u_buff; | |
2030 | pfile->u_buff = buff; | |
2031 | result = buff->cur; | |
2032 | } | |
2033 | ||
2034 | buff->cur = result + len; | |
2035 | return result; | |
2036 | } | |
2037 | ||
87062813 NB |
2038 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
2039 | That buffer is used for growing allocations when saving macro | |
2040 | replacement lists in a #define, and when parsing an answer to an | |
2041 | assertion in #assert, #unassert or #if (and therefore possibly | |
2042 | whilst expanding macros). It therefore must not be used by any | |
2043 | code that they might call: specifically the lexer and the guts of | |
2044 | the macro expander. | |
2045 | ||
2046 | All existing other uses clearly fit this restriction: storing | |
2047 | registered pragmas during initialization. */ | |
93c80368 | 2048 | unsigned char * |
8c3b2693 NB |
2049 | _cpp_aligned_alloc (pfile, len) |
2050 | cpp_reader *pfile; | |
2051 | size_t len; | |
3fef5b2b | 2052 | { |
8c3b2693 NB |
2053 | _cpp_buff *buff = pfile->a_buff; |
2054 | unsigned char *result = buff->cur; | |
3fef5b2b | 2055 | |
8c3b2693 | 2056 | if (len > (size_t) (buff->limit - result)) |
3fef5b2b | 2057 | { |
8c3b2693 NB |
2058 | buff = _cpp_get_buff (pfile, len); |
2059 | buff->next = pfile->a_buff; | |
2060 | pfile->a_buff = buff; | |
2061 | result = buff->cur; | |
3fef5b2b | 2062 | } |
041c3194 | 2063 | |
8c3b2693 | 2064 | buff->cur = result + len; |
93c80368 | 2065 | return result; |
041c3194 | 2066 | } |