]>
Commit | Line | Data |
---|---|---|
45b966db | 1 | /* CPP Library - lexical analysis. |
e1e7d56b | 2 | Copyright (C) 2000, 2001 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 ZW |
79 | static cpp_hashnode *parse_identifier PARAMS ((cpp_reader *)); |
80 | static cpp_hashnode *parse_identifier_slow PARAMS ((cpp_reader *, | |
81 | const U_CHAR *)); | |
93c80368 NB |
82 | static void parse_number PARAMS ((cpp_reader *, cpp_string *, cppchar_t, int)); |
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)); |
93c80368 | 85 | static void unterminated PARAMS ((cpp_reader *, int)); |
87062813 | 86 | static bool trigraph_p PARAMS ((cpp_reader *)); |
0d9f234d | 87 | static void save_comment PARAMS ((cpp_reader *, cpp_token *, const U_CHAR *)); |
93c80368 | 88 | static int name_p PARAMS ((cpp_reader *, const cpp_string *)); |
62729350 NB |
89 | static int maybe_read_ucs PARAMS ((cpp_reader *, const unsigned char **, |
90 | const unsigned char *, unsigned int *)); | |
5fddcffc | 91 | static tokenrun *next_tokenrun PARAMS ((tokenrun *)); |
f617b8e2 | 92 | |
c8a96070 | 93 | static unsigned int hex_digit_value PARAMS ((unsigned int)); |
6142088c | 94 | static _cpp_buff *new_buff PARAMS ((size_t)); |
15dad1d9 | 95 | |
041c3194 | 96 | /* Utility routine: |
9e62c811 | 97 | |
bfb9dc7f ZW |
98 | Compares, the token TOKEN to the NUL-terminated string STRING. |
99 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
15dad1d9 | 100 | |
041c3194 | 101 | int |
bfb9dc7f ZW |
102 | cpp_ideq (token, string) |
103 | const cpp_token *token; | |
041c3194 ZW |
104 | const char *string; |
105 | { | |
bfb9dc7f | 106 | if (token->type != CPP_NAME) |
041c3194 | 107 | return 0; |
bfb9dc7f | 108 | |
a28c5035 | 109 | return !ustrcmp (NODE_NAME (token->val.node), (const U_CHAR *) string); |
15dad1d9 | 110 | } |
1368ee70 | 111 | |
87062813 NB |
112 | /* Call when meeting a newline, assumed to be in buffer->cur[-1]. |
113 | Returns with buffer->cur pointing to the character immediately | |
114 | following the newline (combination). */ | |
115 | static void | |
116 | handle_newline (pfile) | |
1444f2ed | 117 | cpp_reader *pfile; |
0d9f234d | 118 | { |
87062813 | 119 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 120 | |
87062813 | 121 | /* Handle CR-LF and LF-CR. Most other implementations (e.g. java) |
4d6baafa NB |
122 | only accept CR-LF; maybe we should fall back to that behaviour? */ |
123 | if (buffer->cur[-1] + buffer->cur[0] == '\r' + '\n') | |
87062813 | 124 | buffer->cur++; |
0d9f234d | 125 | |
87062813 NB |
126 | buffer->line_base = buffer->cur; |
127 | buffer->col_adjust = 0; | |
128 | pfile->line++; | |
0d9f234d NB |
129 | } |
130 | ||
87062813 NB |
131 | /* Subroutine of skip_escaped_newlines; called when a 3-character |
132 | sequence beginning with "??" is encountered. buffer->cur points to | |
133 | the second '?'. | |
134 | ||
135 | Warn if necessary, and returns true if the sequence forms a | |
136 | trigraph and the trigraph should be honoured. */ | |
137 | static bool | |
138 | trigraph_p (pfile) | |
45b966db | 139 | cpp_reader *pfile; |
45b966db | 140 | { |
87062813 NB |
141 | cpp_buffer *buffer = pfile->buffer; |
142 | cppchar_t from_char = buffer->cur[1]; | |
143 | bool accept; | |
144 | ||
145 | if (!_cpp_trigraph_map[from_char]) | |
146 | return false; | |
147 | ||
148 | accept = CPP_OPTION (pfile, trigraphs); | |
149 | ||
cbcff6df NB |
150 | /* Don't warn about trigraphs in comments. */ |
151 | if (CPP_OPTION (pfile, warn_trigraphs) && !pfile->state.lexing_comment) | |
45b966db | 152 | { |
041c3194 | 153 | if (accept) |
87062813 | 154 | cpp_warning_with_line (pfile, pfile->line, CPP_BUF_COL (buffer) - 1, |
041c3194 | 155 | "trigraph ??%c converted to %c", |
0d9f234d NB |
156 | (int) from_char, |
157 | (int) _cpp_trigraph_map[from_char]); | |
4a5b68a2 NB |
158 | else if (buffer->cur != buffer->last_Wtrigraphs) |
159 | { | |
160 | buffer->last_Wtrigraphs = buffer->cur; | |
67821e3a | 161 | cpp_warning_with_line (pfile, pfile->line, |
87062813 | 162 | CPP_BUF_COL (buffer) - 1, |
4a5b68a2 NB |
163 | "trigraph ??%c ignored", (int) from_char); |
164 | } | |
45b966db | 165 | } |
0d9f234d | 166 | |
041c3194 | 167 | return accept; |
45b966db ZW |
168 | } |
169 | ||
87062813 | 170 | /* Skips any escaped newlines introduced by '?' or a '\\', assumed to |
480709cc NB |
171 | lie in buffer->cur[-1]. Returns the next byte, which will be in |
172 | buffer->cur[-1]. This routine performs preprocessing stages 1 and | |
173 | 2 of the ISO C standard. */ | |
0d9f234d | 174 | static cppchar_t |
87062813 | 175 | skip_escaped_newlines (pfile) |
29401c30 | 176 | cpp_reader *pfile; |
45b966db | 177 | { |
29401c30 | 178 | cpp_buffer *buffer = pfile->buffer; |
87062813 | 179 | cppchar_t next = buffer->cur[-1]; |
29401c30 | 180 | |
a5c3cccd NB |
181 | /* Only do this if we apply stages 1 and 2. */ |
182 | if (!buffer->from_stage3) | |
041c3194 | 183 | { |
a5c3cccd | 184 | const unsigned char *saved_cur; |
87062813 | 185 | cppchar_t next1; |
a5c3cccd NB |
186 | |
187 | do | |
0d9f234d | 188 | { |
a5c3cccd | 189 | if (next == '?') |
0d9f234d | 190 | { |
4d6baafa | 191 | if (buffer->cur[0] != '?' || !trigraph_p (pfile)) |
87062813 | 192 | break; |
a5c3cccd | 193 | |
87062813 NB |
194 | /* Translate the trigraph. */ |
195 | next = _cpp_trigraph_map[buffer->cur[1]]; | |
196 | buffer->cur += 2; | |
4d6baafa | 197 | if (next != '\\') |
a5c3cccd | 198 | break; |
a5c3cccd NB |
199 | } |
200 | ||
4d6baafa NB |
201 | if (buffer->cur == buffer->rlimit) |
202 | break; | |
203 | ||
87062813 NB |
204 | /* We have a backslash, and room for at least one more |
205 | character. Skip horizontal whitespace. */ | |
206 | saved_cur = buffer->cur; | |
a5c3cccd | 207 | do |
87062813 NB |
208 | next1 = *buffer->cur++; |
209 | while (is_nvspace (next1) && buffer->cur < buffer->rlimit); | |
041c3194 | 210 | |
a5c3cccd | 211 | if (!is_vspace (next1)) |
0d9f234d | 212 | { |
87062813 | 213 | buffer->cur = saved_cur; |
0d9f234d NB |
214 | break; |
215 | } | |
45b966db | 216 | |
87062813 NB |
217 | if (saved_cur != buffer->cur - 1 |
218 | && !pfile->state.lexing_comment) | |
29401c30 | 219 | cpp_warning (pfile, "backslash and newline separated by space"); |
0d9f234d | 220 | |
87062813 | 221 | handle_newline (pfile); |
480709cc | 222 | buffer->backup_to = buffer->cur; |
87062813 NB |
223 | if (buffer->cur == buffer->rlimit) |
224 | { | |
225 | cpp_pedwarn (pfile, "backslash-newline at end of file"); | |
226 | next = EOF; | |
227 | } | |
228 | else | |
229 | next = *buffer->cur++; | |
0d9f234d | 230 | } |
a5c3cccd | 231 | while (next == '\\' || next == '?'); |
041c3194 | 232 | } |
45b966db | 233 | |
0d9f234d | 234 | return next; |
45b966db ZW |
235 | } |
236 | ||
0d9f234d | 237 | /* Obtain the next character, after trigraph conversion and skipping |
87062813 NB |
238 | an arbitrarily long string of escaped newlines. The common case of |
239 | no trigraphs or escaped newlines falls through quickly. On return, | |
480709cc NB |
240 | buffer->backup_to points to where to return to if the character is |
241 | not to be processed. */ | |
0d9f234d | 242 | static cppchar_t |
29401c30 NB |
243 | get_effective_char (pfile) |
244 | cpp_reader *pfile; | |
64aaf407 | 245 | { |
4d6baafa | 246 | cppchar_t next; |
480709cc | 247 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 248 | |
480709cc | 249 | buffer->backup_to = buffer->cur; |
4d6baafa NB |
250 | next = *buffer->cur++; |
251 | if (__builtin_expect (next == '?' || next == '\\', 0)) | |
252 | next = skip_escaped_newlines (pfile); | |
0d9f234d | 253 | |
480709cc | 254 | return next; |
64aaf407 NB |
255 | } |
256 | ||
0d9f234d NB |
257 | /* Skip a C-style block comment. We find the end of the comment by |
258 | seeing if an asterisk is before every '/' we encounter. Returns | |
259 | non-zero if comment terminated by EOF, zero otherwise. */ | |
041c3194 ZW |
260 | static int |
261 | skip_block_comment (pfile) | |
45b966db ZW |
262 | cpp_reader *pfile; |
263 | { | |
041c3194 | 264 | cpp_buffer *buffer = pfile->buffer; |
d8090680 | 265 | cppchar_t c = EOF, prevc = EOF; |
0d9f234d | 266 | |
cbcff6df | 267 | pfile->state.lexing_comment = 1; |
0d9f234d | 268 | while (buffer->cur != buffer->rlimit) |
45b966db | 269 | { |
0d9f234d NB |
270 | prevc = c, c = *buffer->cur++; |
271 | ||
0d9f234d | 272 | /* FIXME: For speed, create a new character class of characters |
93c80368 | 273 | of interest inside block comments. */ |
0d9f234d | 274 | if (c == '?' || c == '\\') |
87062813 | 275 | c = skip_escaped_newlines (pfile); |
041c3194 | 276 | |
0d9f234d NB |
277 | /* People like decorating comments with '*', so check for '/' |
278 | instead for efficiency. */ | |
041c3194 | 279 | if (c == '/') |
45b966db | 280 | { |
0d9f234d NB |
281 | if (prevc == '*') |
282 | break; | |
041c3194 | 283 | |
0d9f234d | 284 | /* Warn about potential nested comments, but not if the '/' |
a1f300c0 | 285 | comes immediately before the true comment delimiter. |
041c3194 | 286 | Don't bother to get it right across escaped newlines. */ |
0d9f234d | 287 | if (CPP_OPTION (pfile, warn_comments) |
87062813 NB |
288 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
289 | cpp_warning_with_line (pfile, | |
290 | pfile->line, CPP_BUF_COL (buffer), | |
291 | "\"/*\" within comment"); | |
45b966db | 292 | } |
91fcd158 | 293 | else if (is_vspace (c)) |
87062813 | 294 | handle_newline (pfile); |
52fadca8 | 295 | else if (c == '\t') |
0d9f234d | 296 | adjust_column (pfile); |
45b966db | 297 | } |
041c3194 | 298 | |
cbcff6df | 299 | pfile->state.lexing_comment = 0; |
0d9f234d | 300 | return c != '/' || prevc != '*'; |
45b966db ZW |
301 | } |
302 | ||
480709cc NB |
303 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
304 | terminating newline. Handles escaped newlines. Returns non-zero | |
305 | if a multiline comment. */ | |
041c3194 | 306 | static int |
cbcff6df NB |
307 | skip_line_comment (pfile) |
308 | cpp_reader *pfile; | |
45b966db | 309 | { |
cbcff6df | 310 | cpp_buffer *buffer = pfile->buffer; |
67821e3a | 311 | unsigned int orig_line = pfile->line; |
0d9f234d | 312 | cppchar_t c; |
041c3194 | 313 | |
cbcff6df | 314 | pfile->state.lexing_comment = 1; |
0d9f234d | 315 | do |
041c3194 | 316 | { |
0d9f234d | 317 | if (buffer->cur == buffer->rlimit) |
480709cc | 318 | goto at_eof; |
041c3194 | 319 | |
0d9f234d NB |
320 | c = *buffer->cur++; |
321 | if (c == '?' || c == '\\') | |
87062813 | 322 | c = skip_escaped_newlines (pfile); |
041c3194 | 323 | } |
0d9f234d | 324 | while (!is_vspace (c)); |
45b966db | 325 | |
480709cc NB |
326 | /* Step back over the newline, except at EOF. */ |
327 | buffer->cur--; | |
328 | at_eof: | |
329 | ||
cbcff6df | 330 | pfile->state.lexing_comment = 0; |
67821e3a | 331 | return orig_line != pfile->line; |
041c3194 | 332 | } |
45b966db | 333 | |
0d9f234d NB |
334 | /* pfile->buffer->cur is one beyond the \t character. Update |
335 | col_adjust so we track the column correctly. */ | |
52fadca8 | 336 | static void |
0d9f234d | 337 | adjust_column (pfile) |
52fadca8 | 338 | cpp_reader *pfile; |
52fadca8 | 339 | { |
0d9f234d NB |
340 | cpp_buffer *buffer = pfile->buffer; |
341 | unsigned int col = CPP_BUF_COL (buffer) - 1; /* Zero-based column. */ | |
52fadca8 NB |
342 | |
343 | /* Round it up to multiple of the tabstop, but subtract 1 since the | |
344 | tab itself occupies a character position. */ | |
0d9f234d NB |
345 | buffer->col_adjust += (CPP_OPTION (pfile, tabstop) |
346 | - col % CPP_OPTION (pfile, tabstop)) - 1; | |
52fadca8 NB |
347 | } |
348 | ||
0d9f234d NB |
349 | /* Skips whitespace, saving the next non-whitespace character. |
350 | Adjusts pfile->col_adjust to account for tabs. Without this, | |
351 | tokens might be assigned an incorrect column. */ | |
4d6baafa | 352 | static int |
0d9f234d | 353 | skip_whitespace (pfile, c) |
041c3194 | 354 | cpp_reader *pfile; |
0d9f234d | 355 | cppchar_t c; |
041c3194 ZW |
356 | { |
357 | cpp_buffer *buffer = pfile->buffer; | |
0d9f234d | 358 | unsigned int warned = 0; |
45b966db | 359 | |
0d9f234d | 360 | do |
041c3194 | 361 | { |
91fcd158 NB |
362 | /* Horizontal space always OK. */ |
363 | if (c == ' ') | |
0d9f234d | 364 | ; |
91fcd158 | 365 | else if (c == '\t') |
0d9f234d NB |
366 | adjust_column (pfile); |
367 | /* Just \f \v or \0 left. */ | |
91fcd158 | 368 | else if (c == '\0') |
041c3194 | 369 | { |
4d6baafa NB |
370 | if (buffer->cur - 1 == buffer->rlimit) |
371 | return 0; | |
91fcd158 | 372 | if (!warned) |
0d9f234d NB |
373 | { |
374 | cpp_warning (pfile, "null character(s) ignored"); | |
375 | warned = 1; | |
376 | } | |
45b966db | 377 | } |
93c80368 | 378 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
67821e3a | 379 | cpp_pedwarn_with_line (pfile, pfile->line, |
91fcd158 NB |
380 | CPP_BUF_COL (buffer), |
381 | "%s in preprocessing directive", | |
382 | c == '\f' ? "form feed" : "vertical tab"); | |
0d9f234d | 383 | |
0d9f234d | 384 | c = *buffer->cur++; |
45b966db | 385 | } |
ec5c56db | 386 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
0d9f234d NB |
387 | while (is_nvspace (c)); |
388 | ||
480709cc | 389 | buffer->cur--; |
4d6baafa | 390 | return 1; |
041c3194 | 391 | } |
45b966db | 392 | |
93c80368 NB |
393 | /* See if the characters of a number token are valid in a name (no |
394 | '.', '+' or '-'). */ | |
395 | static int | |
396 | name_p (pfile, string) | |
397 | cpp_reader *pfile; | |
398 | const cpp_string *string; | |
399 | { | |
400 | unsigned int i; | |
401 | ||
402 | for (i = 0; i < string->len; i++) | |
403 | if (!is_idchar (string->text[i])) | |
404 | return 0; | |
405 | ||
406 | return 1; | |
407 | } | |
408 | ||
2c3fcba6 ZW |
409 | /* Parse an identifier, skipping embedded backslash-newlines. This is |
410 | a critical inner loop. The common case is an identifier which has | |
411 | not been split by backslash-newline, does not contain a dollar | |
412 | sign, and has already been scanned (roughly 10:1 ratio of | |
413 | seen:unseen identifiers in normal code; the distribution is | |
414 | Poisson-like). Second most common case is a new identifier, not | |
415 | split and no dollar sign. The other possibilities are rare and | |
416 | have been relegated to parse_identifier_slow. */ | |
0d9f234d NB |
417 | |
418 | static cpp_hashnode * | |
2c3fcba6 | 419 | parse_identifier (pfile) |
45b966db | 420 | cpp_reader *pfile; |
45b966db | 421 | { |
93c80368 | 422 | cpp_hashnode *result; |
4d6baafa | 423 | const U_CHAR *cur; |
2c3fcba6 ZW |
424 | |
425 | /* Fast-path loop. Skim over a normal identifier. | |
426 | N.B. ISIDNUM does not include $. */ | |
4d6baafa NB |
427 | cur = pfile->buffer->cur; |
428 | while (ISIDNUM (*cur)) | |
2c3fcba6 | 429 | cur++; |
2c3fcba6 ZW |
430 | |
431 | /* Check for slow-path cases. */ | |
4d6baafa | 432 | if (*cur == '?' || *cur == '\\' || *cur == '$') |
2c3fcba6 ZW |
433 | result = parse_identifier_slow (pfile, cur); |
434 | else | |
435 | { | |
436 | const U_CHAR *base = pfile->buffer->cur - 1; | |
437 | result = (cpp_hashnode *) | |
438 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
439 | pfile->buffer->cur = cur; | |
440 | } | |
441 | ||
442 | /* Rarely, identifiers require diagnostics when lexed. | |
443 | XXX Has to be forced out of the fast path. */ | |
444 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) | |
445 | && !pfile->state.skipping, 0)) | |
446 | { | |
447 | /* It is allowed to poison the same identifier twice. */ | |
448 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
449 | cpp_error (pfile, "attempt to use poisoned \"%s\"", | |
450 | NODE_NAME (result)); | |
451 | ||
452 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
453 | replacement list of a variadic macro. */ | |
454 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
455 | && !pfile->state.va_args_ok) | |
456 | cpp_pedwarn (pfile, | |
457 | "__VA_ARGS__ can only appear in the expansion of a C99 variadic macro"); | |
458 | } | |
459 | ||
460 | return result; | |
461 | } | |
462 | ||
463 | /* Slow path. This handles identifiers which have been split, and | |
464 | identifiers which contain dollar signs. The part of the identifier | |
465 | from PFILE->buffer->cur-1 to CUR has already been scanned. */ | |
466 | static cpp_hashnode * | |
467 | parse_identifier_slow (pfile, cur) | |
468 | cpp_reader *pfile; | |
469 | const U_CHAR *cur; | |
470 | { | |
0d9f234d | 471 | cpp_buffer *buffer = pfile->buffer; |
2c3fcba6 | 472 | const U_CHAR *base = buffer->cur - 1; |
2a967f3d | 473 | struct obstack *stack = &pfile->hash_table->stack; |
2c3fcba6 ZW |
474 | unsigned int c, saw_dollar = 0, len; |
475 | ||
476 | /* Copy the part of the token which is known to be okay. */ | |
477 | obstack_grow (stack, base, cur - base); | |
041c3194 | 478 | |
2c3fcba6 ZW |
479 | /* Now process the part which isn't. We are looking at one of |
480 | '$', '\\', or '?' on entry to this loop. */ | |
481 | c = *cur++; | |
482 | buffer->cur = cur; | |
0d9f234d | 483 | do |
041c3194 | 484 | { |
2c3fcba6 ZW |
485 | while (is_idchar (c)) |
486 | { | |
487 | obstack_1grow (stack, c); | |
45b966db | 488 | |
2c3fcba6 ZW |
489 | if (c == '$') |
490 | saw_dollar++; | |
ba89d661 | 491 | |
2c3fcba6 ZW |
492 | c = *buffer->cur++; |
493 | } | |
ba89d661 | 494 | |
0d9f234d | 495 | /* Potential escaped newline? */ |
480709cc | 496 | buffer->backup_to = buffer->cur - 1; |
0d9f234d | 497 | if (c != '?' && c != '\\') |
2c3fcba6 | 498 | break; |
87062813 | 499 | c = skip_escaped_newlines (pfile); |
041c3194 | 500 | } |
0d9f234d NB |
501 | while (is_idchar (c)); |
502 | ||
4d6baafa | 503 | /* Step back over the unwanted char. */ |
480709cc | 504 | BACKUP (); |
93c80368 | 505 | |
4fe9b91c | 506 | /* $ is not an identifier character in the standard, but is commonly |
0d9f234d NB |
507 | accepted as an extension. Don't warn about it in skipped |
508 | conditional blocks. */ | |
cef0d199 | 509 | if (saw_dollar && CPP_PEDANTIC (pfile) && ! pfile->state.skipping) |
0d9f234d NB |
510 | cpp_pedwarn (pfile, "'$' character(s) in identifier"); |
511 | ||
93c80368 | 512 | /* Identifiers are null-terminated. */ |
2a967f3d NB |
513 | len = obstack_object_size (stack); |
514 | obstack_1grow (stack, '\0'); | |
93c80368 | 515 | |
2c3fcba6 | 516 | return (cpp_hashnode *) |
2a967f3d | 517 | ht_lookup (pfile->hash_table, obstack_finish (stack), len, HT_ALLOCED); |
45b966db ZW |
518 | } |
519 | ||
0d9f234d | 520 | /* Parse a number, skipping embedded backslash-newlines. */ |
45b966db | 521 | static void |
93c80368 | 522 | parse_number (pfile, number, c, leading_period) |
45b966db | 523 | cpp_reader *pfile; |
0d9f234d NB |
524 | cpp_string *number; |
525 | cppchar_t c; | |
93c80368 | 526 | int leading_period; |
45b966db | 527 | { |
041c3194 | 528 | cpp_buffer *buffer = pfile->buffer; |
93c80368 | 529 | unsigned char *dest, *limit; |
45b966db | 530 | |
ece54d54 NB |
531 | dest = BUFF_FRONT (pfile->u_buff); |
532 | limit = BUFF_LIMIT (pfile->u_buff); | |
cbcff6df | 533 | |
93c80368 NB |
534 | /* Place a leading period. */ |
535 | if (leading_period) | |
536 | { | |
ece54d54 NB |
537 | if (dest == limit) |
538 | { | |
8c3b2693 | 539 | _cpp_extend_buff (pfile, &pfile->u_buff, 1); |
ece54d54 NB |
540 | dest = BUFF_FRONT (pfile->u_buff); |
541 | limit = BUFF_LIMIT (pfile->u_buff); | |
542 | } | |
93c80368 NB |
543 | *dest++ = '.'; |
544 | } | |
545 | ||
0d9f234d | 546 | do |
041c3194 | 547 | { |
0d9f234d NB |
548 | do |
549 | { | |
93c80368 | 550 | /* Need room for terminating null. */ |
ece54d54 NB |
551 | if ((size_t) (limit - dest) < 2) |
552 | { | |
553 | size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff); | |
8c3b2693 | 554 | _cpp_extend_buff (pfile, &pfile->u_buff, 2); |
ece54d54 NB |
555 | dest = BUFF_FRONT (pfile->u_buff) + len_so_far; |
556 | limit = BUFF_LIMIT (pfile->u_buff); | |
557 | } | |
93c80368 | 558 | *dest++ = c; |
0d9f234d | 559 | |
0d9f234d NB |
560 | c = *buffer->cur++; |
561 | } | |
93c80368 | 562 | while (is_numchar (c) || c == '.' || VALID_SIGN (c, dest[-1])); |
45b966db | 563 | |
0d9f234d | 564 | /* Potential escaped newline? */ |
480709cc | 565 | buffer->backup_to = buffer->cur - 1; |
0d9f234d NB |
566 | if (c != '?' && c != '\\') |
567 | break; | |
87062813 | 568 | c = skip_escaped_newlines (pfile); |
45b966db | 569 | } |
93c80368 | 570 | while (is_numchar (c) || c == '.' || VALID_SIGN (c, dest[-1])); |
cbcff6df | 571 | |
4d6baafa | 572 | /* Step back over the unwanted char. */ |
480709cc | 573 | BACKUP (); |
64aaf407 | 574 | |
93c80368 NB |
575 | /* Null-terminate the number. */ |
576 | *dest = '\0'; | |
577 | ||
ece54d54 | 578 | number->text = BUFF_FRONT (pfile->u_buff); |
93c80368 | 579 | number->len = dest - number->text; |
ece54d54 | 580 | BUFF_FRONT (pfile->u_buff) = dest + 1; |
0d9f234d NB |
581 | } |
582 | ||
583 | /* Subroutine of parse_string. Emits error for unterminated strings. */ | |
584 | static void | |
93c80368 | 585 | unterminated (pfile, term) |
0d9f234d | 586 | cpp_reader *pfile; |
0d9f234d NB |
587 | int term; |
588 | { | |
589 | cpp_error (pfile, "missing terminating %c character", term); | |
590 | ||
50410426 | 591 | if (term == '\"' && pfile->mls_line && pfile->mls_line != pfile->line) |
041c3194 | 592 | { |
50410426 | 593 | cpp_error_with_line (pfile, pfile->mls_line, pfile->mls_col, |
0d9f234d | 594 | "possible start of unterminated string literal"); |
50410426 | 595 | pfile->mls_line = 0; |
041c3194 | 596 | } |
45b966db ZW |
597 | } |
598 | ||
93c80368 NB |
599 | /* Subroutine of parse_string. */ |
600 | static int | |
601 | unescaped_terminator_p (pfile, dest) | |
602 | cpp_reader *pfile; | |
603 | const unsigned char *dest; | |
604 | { | |
605 | const unsigned char *start, *temp; | |
606 | ||
607 | /* In #include-style directives, terminators are not escapeable. */ | |
608 | if (pfile->state.angled_headers) | |
609 | return 1; | |
610 | ||
ece54d54 | 611 | start = BUFF_FRONT (pfile->u_buff); |
93c80368 NB |
612 | |
613 | /* An odd number of consecutive backslashes represents an escaped | |
614 | terminator. */ | |
615 | for (temp = dest; temp > start && temp[-1] == '\\'; temp--) | |
616 | ; | |
617 | ||
618 | return ((dest - temp) & 1) == 0; | |
619 | } | |
620 | ||
0d9f234d | 621 | /* Parses a string, character constant, or angle-bracketed header file |
7868b4a2 NB |
622 | name. Handles embedded trigraphs and escaped newlines. The stored |
623 | string is guaranteed NUL-terminated, but it is not guaranteed that | |
624 | this is the first NUL since embedded NULs are preserved. | |
87062813 | 625 | Multi-line strings are allowed, but they are deprecated. |
45b966db | 626 | |
87062813 NB |
627 | When this function returns, buffer->cur points to the next |
628 | character to be processed. */ | |
041c3194 | 629 | static void |
0d9f234d | 630 | parse_string (pfile, token, terminator) |
45b966db | 631 | cpp_reader *pfile; |
041c3194 | 632 | cpp_token *token; |
0d9f234d | 633 | cppchar_t terminator; |
45b966db | 634 | { |
041c3194 | 635 | cpp_buffer *buffer = pfile->buffer; |
93c80368 | 636 | unsigned char *dest, *limit; |
0d9f234d | 637 | cppchar_t c; |
d82fc108 | 638 | bool warned_nulls = false, warned_multi = false; |
0d9f234d | 639 | |
ece54d54 NB |
640 | dest = BUFF_FRONT (pfile->u_buff); |
641 | limit = BUFF_LIMIT (pfile->u_buff); | |
93c80368 | 642 | |
0d9f234d | 643 | for (;;) |
45b966db | 644 | { |
87062813 | 645 | /* We need room for another char, possibly the terminating NUL. */ |
ece54d54 NB |
646 | if ((size_t) (limit - dest) < 1) |
647 | { | |
648 | size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff); | |
8c3b2693 | 649 | _cpp_extend_buff (pfile, &pfile->u_buff, 2); |
ece54d54 NB |
650 | dest = BUFF_FRONT (pfile->u_buff) + len_so_far; |
651 | limit = BUFF_LIMIT (pfile->u_buff); | |
652 | } | |
7868b4a2 | 653 | |
0d9f234d | 654 | /* Handle trigraphs, escaped newlines etc. */ |
87062813 | 655 | c = *buffer->cur++; |
0d9f234d | 656 | if (c == '?' || c == '\\') |
87062813 | 657 | c = skip_escaped_newlines (pfile); |
45b966db | 658 | |
87062813 | 659 | if (c == terminator) |
45b966db | 660 | { |
87062813 NB |
661 | if (unescaped_terminator_p (pfile, dest)) |
662 | break; | |
0d9f234d NB |
663 | } |
664 | else if (is_vspace (c)) | |
665 | { | |
666 | /* In assembly language, silently terminate string and | |
667 | character literals at end of line. This is a kludge | |
668 | around not knowing where comments are. */ | |
bdb05a7b | 669 | if (CPP_OPTION (pfile, lang) == CLK_ASM && terminator != '>') |
87062813 NB |
670 | { |
671 | buffer->cur--; | |
672 | break; | |
673 | } | |
45b966db | 674 | |
0d9f234d NB |
675 | /* Character constants and header names may not extend over |
676 | multiple lines. In Standard C, neither may strings. | |
677 | Unfortunately, we accept multiline strings as an | |
16eb2788 NB |
678 | extension, except in #include family directives. */ |
679 | if (terminator != '"' || pfile->state.angled_headers) | |
45b966db | 680 | { |
93c80368 | 681 | unterminated (pfile, terminator); |
87062813 | 682 | buffer->cur--; |
0d9f234d | 683 | break; |
45b966db | 684 | } |
45b966db | 685 | |
d82fc108 NB |
686 | if (!warned_multi) |
687 | { | |
688 | warned_multi = true; | |
689 | cpp_pedwarn (pfile, "multi-line string literals are deprecated"); | |
690 | } | |
691 | ||
50410426 NB |
692 | if (pfile->mls_line == 0) |
693 | { | |
694 | pfile->mls_line = token->line; | |
695 | pfile->mls_col = token->col; | |
696 | } | |
0d9f234d | 697 | |
87062813 NB |
698 | handle_newline (pfile); |
699 | c = '\n'; | |
0d9f234d | 700 | } |
4d6baafa | 701 | else if (c == '\0') |
0d9f234d | 702 | { |
4d6baafa NB |
703 | if (buffer->cur - 1 == buffer->rlimit) |
704 | { | |
705 | unterminated (pfile, terminator); | |
706 | buffer->cur--; | |
707 | break; | |
708 | } | |
709 | if (!warned_nulls) | |
710 | { | |
711 | warned_nulls = true; | |
712 | cpp_warning (pfile, "null character(s) preserved in literal"); | |
713 | } | |
45b966db | 714 | } |
45b966db | 715 | |
93c80368 | 716 | *dest++ = c; |
45b966db ZW |
717 | } |
718 | ||
7868b4a2 | 719 | *dest = '\0'; |
45b966db | 720 | |
ece54d54 NB |
721 | token->val.str.text = BUFF_FRONT (pfile->u_buff); |
722 | token->val.str.len = dest - BUFF_FRONT (pfile->u_buff); | |
723 | BUFF_FRONT (pfile->u_buff) = dest + 1; | |
0d9f234d | 724 | } |
041c3194 | 725 | |
93c80368 | 726 | /* The stored comment includes the comment start and any terminator. */ |
9e62c811 | 727 | static void |
0d9f234d NB |
728 | save_comment (pfile, token, from) |
729 | cpp_reader *pfile; | |
041c3194 ZW |
730 | cpp_token *token; |
731 | const unsigned char *from; | |
9e62c811 | 732 | { |
041c3194 | 733 | unsigned char *buffer; |
0d9f234d | 734 | unsigned int len; |
0d9f234d | 735 | |
1c6d33ef | 736 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
480709cc | 737 | |
3542203b NB |
738 | /* C++ comments probably (not definitely) have moved past a new |
739 | line, which we don't want to save in the comment. */ | |
480709cc | 740 | if (is_vspace (pfile->buffer->cur[-1])) |
3542203b | 741 | len--; |
ece54d54 | 742 | buffer = _cpp_unaligned_alloc (pfile, len); |
041c3194 | 743 | |
041c3194 | 744 | token->type = CPP_COMMENT; |
bfb9dc7f | 745 | token->val.str.len = len; |
0d9f234d | 746 | token->val.str.text = buffer; |
45b966db | 747 | |
1c6d33ef NB |
748 | buffer[0] = '/'; |
749 | memcpy (buffer + 1, from, len - 1); | |
0d9f234d | 750 | } |
45b966db | 751 | |
5fddcffc NB |
752 | /* Allocate COUNT tokens for RUN. */ |
753 | void | |
754 | _cpp_init_tokenrun (run, count) | |
755 | tokenrun *run; | |
756 | unsigned int count; | |
757 | { | |
758 | run->base = xnewvec (cpp_token, count); | |
759 | run->limit = run->base + count; | |
760 | run->next = NULL; | |
761 | } | |
762 | ||
763 | /* Returns the next tokenrun, or creates one if there is none. */ | |
764 | static tokenrun * | |
765 | next_tokenrun (run) | |
766 | tokenrun *run; | |
767 | { | |
768 | if (run->next == NULL) | |
769 | { | |
770 | run->next = xnew (tokenrun); | |
bdcbe496 | 771 | run->next->prev = run; |
5fddcffc NB |
772 | _cpp_init_tokenrun (run->next, 250); |
773 | } | |
774 | ||
775 | return run->next; | |
776 | } | |
777 | ||
4ed5bcfb NB |
778 | /* Allocate a single token that is invalidated at the same time as the |
779 | rest of the tokens on the line. Has its line and col set to the | |
780 | same as the last lexed token, so that diagnostics appear in the | |
781 | right place. */ | |
782 | cpp_token * | |
783 | _cpp_temp_token (pfile) | |
784 | cpp_reader *pfile; | |
785 | { | |
786 | cpp_token *old, *result; | |
787 | ||
788 | old = pfile->cur_token - 1; | |
789 | if (pfile->cur_token == pfile->cur_run->limit) | |
790 | { | |
791 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
792 | pfile->cur_token = pfile->cur_run->base; | |
793 | } | |
794 | ||
795 | result = pfile->cur_token++; | |
796 | result->line = old->line; | |
797 | result->col = old->col; | |
798 | return result; | |
799 | } | |
800 | ||
14baae01 NB |
801 | /* Lex a token into RESULT (external interface). Takes care of issues |
802 | like directive handling, token lookahead, multiple include | |
a1f300c0 | 803 | optimization and skipping. */ |
345894b4 NB |
804 | const cpp_token * |
805 | _cpp_lex_token (pfile) | |
45b966db | 806 | cpp_reader *pfile; |
5fddcffc | 807 | { |
bdcbe496 | 808 | cpp_token *result; |
5fddcffc | 809 | |
bdcbe496 | 810 | for (;;) |
5fddcffc | 811 | { |
bdcbe496 | 812 | if (pfile->cur_token == pfile->cur_run->limit) |
5fddcffc | 813 | { |
bdcbe496 NB |
814 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
815 | pfile->cur_token = pfile->cur_run->base; | |
5fddcffc NB |
816 | } |
817 | ||
bdcbe496 | 818 | if (pfile->lookaheads) |
14baae01 NB |
819 | { |
820 | pfile->lookaheads--; | |
821 | result = pfile->cur_token++; | |
822 | } | |
bdcbe496 | 823 | else |
14baae01 | 824 | result = _cpp_lex_direct (pfile); |
bdcbe496 NB |
825 | |
826 | if (result->flags & BOL) | |
5fddcffc | 827 | { |
bdcbe496 NB |
828 | /* Is this a directive. If _cpp_handle_directive returns |
829 | false, it is an assembler #. */ | |
830 | if (result->type == CPP_HASH | |
831 | && !pfile->state.parsing_args | |
832 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) | |
833 | continue; | |
97293897 NB |
834 | if (pfile->cb.line_change && !pfile->state.skipping) |
835 | (*pfile->cb.line_change)(pfile, result, pfile->state.parsing_args); | |
5fddcffc | 836 | } |
5fddcffc | 837 | |
bdcbe496 NB |
838 | /* We don't skip tokens in directives. */ |
839 | if (pfile->state.in_directive) | |
840 | break; | |
5fddcffc | 841 | |
bdcbe496 | 842 | /* Outside a directive, invalidate controlling macros. At file |
14baae01 | 843 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
bdcbe496 | 844 | get here and MI optimisation works. */ |
5fddcffc | 845 | pfile->mi_valid = false; |
bdcbe496 NB |
846 | |
847 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
848 | break; | |
5fddcffc NB |
849 | } |
850 | ||
345894b4 | 851 | return result; |
5fddcffc NB |
852 | } |
853 | ||
480709cc NB |
854 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
855 | do { \ | |
856 | if (get_effective_char (pfile) == CHAR) \ | |
857 | result->type = THEN_TYPE; \ | |
858 | else \ | |
859 | { \ | |
860 | BACKUP (); \ | |
861 | result->type = ELSE_TYPE; \ | |
862 | } \ | |
863 | } while (0) | |
864 | ||
14baae01 NB |
865 | /* Lex a token into pfile->cur_token, which is also incremented, to |
866 | get diagnostics pointing to the correct location. | |
867 | ||
868 | Does not handle issues such as token lookahead, multiple-include | |
869 | optimisation, directives, skipping etc. This function is only | |
870 | suitable for use by _cpp_lex_token, and in special cases like | |
871 | lex_expansion_token which doesn't care for any of these issues. | |
872 | ||
873 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
874 | otherwise returns to the start of the token buffer if permissible. | |
875 | Returns the location of the lexed token. */ | |
876 | cpp_token * | |
877 | _cpp_lex_direct (pfile) | |
5fddcffc | 878 | cpp_reader *pfile; |
45b966db | 879 | { |
0d9f234d | 880 | cppchar_t c; |
adb84b42 | 881 | cpp_buffer *buffer; |
0d9f234d | 882 | const unsigned char *comment_start; |
14baae01 | 883 | cpp_token *result = pfile->cur_token++; |
9ec7291f | 884 | |
5fddcffc | 885 | fresh_line: |
adb84b42 | 886 | buffer = pfile->buffer; |
bd969772 NB |
887 | result->flags = buffer->saved_flags; |
888 | buffer->saved_flags = 0; | |
5fddcffc | 889 | update_tokens_line: |
1444f2ed | 890 | result->line = pfile->line; |
041c3194 | 891 | |
5fddcffc | 892 | skipped_white: |
480709cc | 893 | c = *buffer->cur++; |
5fddcffc | 894 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
5fddcffc NB |
895 | |
896 | trigraph: | |
0d9f234d | 897 | switch (c) |
45b966db | 898 | { |
4d6baafa NB |
899 | case ' ': case '\t': case '\f': case '\v': case '\0': |
900 | result->flags |= PREV_WHITE; | |
901 | if (skip_whitespace (pfile, c)) | |
902 | goto skipped_white; | |
903 | ||
904 | /* EOF. */ | |
905 | buffer->cur--; | |
bdcbe496 | 906 | buffer->saved_flags = BOL; |
5fddcffc | 907 | if (!pfile->state.parsing_args && !pfile->state.in_directive) |
ef6e958a | 908 | { |
bdcbe496 | 909 | if (buffer->cur != buffer->line_base) |
5fddcffc NB |
910 | { |
911 | /* Non-empty files should end in a newline. Don't warn | |
912 | for command line and _Pragma buffers. */ | |
913 | if (!buffer->from_stage3) | |
914 | cpp_pedwarn (pfile, "no newline at end of file"); | |
87062813 | 915 | handle_newline (pfile); |
7364fdd8 | 916 | } |
bdcbe496 NB |
917 | |
918 | /* Don't pop the last buffer. */ | |
919 | if (buffer->prev) | |
920 | { | |
921 | unsigned char stop = buffer->return_at_eof; | |
922 | ||
923 | _cpp_pop_buffer (pfile); | |
924 | if (!stop) | |
925 | goto fresh_line; | |
926 | } | |
ef6e958a | 927 | } |
0d9f234d | 928 | result->type = CPP_EOF; |
5fddcffc | 929 | break; |
45b966db | 930 | |
0d9f234d | 931 | case '\n': case '\r': |
87062813 | 932 | handle_newline (pfile); |
bdcbe496 NB |
933 | buffer->saved_flags = BOL; |
934 | if (! pfile->state.in_directive) | |
45b966db | 935 | { |
4ed5bcfb NB |
936 | if (pfile->state.parsing_args == 2) |
937 | buffer->saved_flags |= PREV_WHITE; | |
bdcbe496 NB |
938 | if (!pfile->keep_tokens) |
939 | { | |
940 | pfile->cur_run = &pfile->base_run; | |
941 | result = pfile->base_run.base; | |
942 | pfile->cur_token = result + 1; | |
943 | } | |
944 | goto fresh_line; | |
45b966db | 945 | } |
5fddcffc NB |
946 | result->type = CPP_EOF; |
947 | break; | |
46d07497 | 948 | |
0d9f234d NB |
949 | case '?': |
950 | case '\\': | |
951 | /* These could start an escaped newline, or '?' a trigraph. Let | |
952 | skip_escaped_newlines do all the work. */ | |
953 | { | |
67821e3a | 954 | unsigned int line = pfile->line; |
0d9f234d | 955 | |
87062813 | 956 | c = skip_escaped_newlines (pfile); |
67821e3a | 957 | if (line != pfile->line) |
87062813 | 958 | { |
480709cc | 959 | buffer->cur--; |
87062813 NB |
960 | /* We had at least one escaped newline of some sort. |
961 | Update the token's line and column. */ | |
5fddcffc | 962 | goto update_tokens_line; |
87062813 | 963 | } |
480709cc | 964 | } |
0d9f234d | 965 | |
480709cc NB |
966 | /* We are either the original '?' or '\\', or a trigraph. */ |
967 | if (c == '?') | |
0d9f234d | 968 | result->type = CPP_QUERY; |
480709cc NB |
969 | else if (c == '\\') |
970 | goto random_char; | |
971 | else | |
972 | goto trigraph; | |
0d9f234d | 973 | break; |
46d07497 | 974 | |
0d9f234d NB |
975 | case '0': case '1': case '2': case '3': case '4': |
976 | case '5': case '6': case '7': case '8': case '9': | |
977 | result->type = CPP_NUMBER; | |
93c80368 | 978 | parse_number (pfile, &result->val.str, c, 0); |
0d9f234d | 979 | break; |
46d07497 | 980 | |
0abc6a6a NB |
981 | case 'L': |
982 | /* 'L' may introduce wide characters or strings. */ | |
983 | { | |
984 | const unsigned char *pos = buffer->cur; | |
0d9f234d | 985 | |
0abc6a6a NB |
986 | c = get_effective_char (pfile); |
987 | if (c == '\'' || c == '"') | |
988 | { | |
989 | result->type = (c == '"' ? CPP_WSTRING: CPP_WCHAR); | |
990 | parse_string (pfile, result, c); | |
991 | break; | |
992 | } | |
993 | buffer->cur = pos; | |
994 | } | |
995 | /* Fall through. */ | |
996 | ||
997 | start_ident: | |
0d9f234d NB |
998 | case '_': |
999 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
1000 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
1001 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
1002 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
1003 | case 'y': case 'z': | |
1004 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
0abc6a6a | 1005 | case 'G': case 'H': case 'I': case 'J': case 'K': |
0d9f234d NB |
1006 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
1007 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1008 | case 'Y': case 'Z': | |
1009 | result->type = CPP_NAME; | |
2c3fcba6 | 1010 | result->val.node = parse_identifier (pfile); |
0d9f234d | 1011 | |
0d9f234d | 1012 | /* Convert named operators to their proper types. */ |
0abc6a6a | 1013 | if (result->val.node->flags & NODE_OPERATOR) |
0d9f234d NB |
1014 | { |
1015 | result->flags |= NAMED_OP; | |
93c80368 | 1016 | result->type = result->val.node->value.operator; |
0d9f234d NB |
1017 | } |
1018 | break; | |
1019 | ||
1020 | case '\'': | |
1021 | case '"': | |
1022 | result->type = c == '"' ? CPP_STRING: CPP_CHAR; | |
0d9f234d NB |
1023 | parse_string (pfile, result, c); |
1024 | break; | |
041c3194 | 1025 | |
0d9f234d | 1026 | case '/': |
1c6d33ef NB |
1027 | /* A potential block or line comment. */ |
1028 | comment_start = buffer->cur; | |
29401c30 | 1029 | c = get_effective_char (pfile); |
480709cc | 1030 | |
1c6d33ef NB |
1031 | if (c == '*') |
1032 | { | |
0d9f234d | 1033 | if (skip_block_comment (pfile)) |
67821e3a | 1034 | cpp_error (pfile, "unterminated comment"); |
0d9f234d | 1035 | } |
480709cc NB |
1036 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
1037 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
0d9f234d | 1038 | { |
bdb05a7b NB |
1039 | /* Warn about comments only if pedantically GNUC89, and not |
1040 | in system headers. */ | |
1041 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
a94c1199 | 1042 | && ! buffer->warned_cplusplus_comments) |
041c3194 | 1043 | { |
1c6d33ef NB |
1044 | cpp_pedwarn (pfile, |
1045 | "C++ style comments are not allowed in ISO C89"); | |
1046 | cpp_pedwarn (pfile, | |
1047 | "(this will be reported only once per input file)"); | |
1048 | buffer->warned_cplusplus_comments = 1; | |
1049 | } | |
0d9f234d | 1050 | |
01ef6563 | 1051 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
50410426 | 1052 | cpp_warning (pfile, "multi-line comment"); |
1c6d33ef | 1053 | } |
480709cc NB |
1054 | else if (c == '=') |
1055 | { | |
1056 | result->type = CPP_DIV_EQ; | |
1057 | break; | |
1058 | } | |
1059 | else | |
1060 | { | |
1061 | BACKUP (); | |
1062 | result->type = CPP_DIV; | |
1063 | break; | |
1064 | } | |
0d9f234d | 1065 | |
1c6d33ef NB |
1066 | if (!pfile->state.save_comments) |
1067 | { | |
1068 | result->flags |= PREV_WHITE; | |
5fddcffc | 1069 | goto update_tokens_line; |
0d9f234d | 1070 | } |
1c6d33ef NB |
1071 | |
1072 | /* Save the comment as a token in its own right. */ | |
1073 | save_comment (pfile, result, comment_start); | |
bdcbe496 | 1074 | break; |
0d9f234d NB |
1075 | |
1076 | case '<': | |
1077 | if (pfile->state.angled_headers) | |
1078 | { | |
1079 | result->type = CPP_HEADER_NAME; | |
480709cc NB |
1080 | parse_string (pfile, result, '>'); |
1081 | break; | |
0d9f234d | 1082 | } |
45b966db | 1083 | |
29401c30 | 1084 | c = get_effective_char (pfile); |
0d9f234d | 1085 | if (c == '=') |
480709cc | 1086 | result->type = CPP_LESS_EQ; |
0d9f234d | 1087 | else if (c == '<') |
480709cc | 1088 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); |
0d9f234d | 1089 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1090 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); |
0d9f234d NB |
1091 | else if (c == ':' && CPP_OPTION (pfile, digraphs)) |
1092 | { | |
480709cc | 1093 | result->type = CPP_OPEN_SQUARE; |
0d9f234d NB |
1094 | result->flags |= DIGRAPH; |
1095 | } | |
1096 | else if (c == '%' && CPP_OPTION (pfile, digraphs)) | |
1097 | { | |
480709cc | 1098 | result->type = CPP_OPEN_BRACE; |
0d9f234d NB |
1099 | result->flags |= DIGRAPH; |
1100 | } | |
480709cc NB |
1101 | else |
1102 | { | |
1103 | BACKUP (); | |
1104 | result->type = CPP_LESS; | |
1105 | } | |
0d9f234d NB |
1106 | break; |
1107 | ||
1108 | case '>': | |
29401c30 | 1109 | c = get_effective_char (pfile); |
0d9f234d | 1110 | if (c == '=') |
480709cc | 1111 | result->type = CPP_GREATER_EQ; |
0d9f234d | 1112 | else if (c == '>') |
480709cc | 1113 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); |
0d9f234d | 1114 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc NB |
1115 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); |
1116 | else | |
0d9f234d | 1117 | { |
480709cc NB |
1118 | BACKUP (); |
1119 | result->type = CPP_GREATER; | |
0d9f234d NB |
1120 | } |
1121 | break; | |
1122 | ||
cbcff6df | 1123 | case '%': |
480709cc NB |
1124 | c = get_effective_char (pfile); |
1125 | if (c == '=') | |
1126 | result->type = CPP_MOD_EQ; | |
1127 | else if (CPP_OPTION (pfile, digraphs) && c == ':') | |
1128 | { | |
1129 | result->flags |= DIGRAPH; | |
1130 | result->type = CPP_HASH; | |
1131 | if (get_effective_char (pfile) == '%') | |
1132 | { | |
1133 | const unsigned char *pos = buffer->cur; | |
1134 | ||
1135 | if (get_effective_char (pfile) == ':') | |
1136 | result->type = CPP_PASTE; | |
1137 | else | |
1138 | buffer->cur = pos - 1; | |
1139 | } | |
1140 | else | |
1141 | BACKUP (); | |
1142 | } | |
1143 | else if (CPP_OPTION (pfile, digraphs) && c == '>') | |
1144 | { | |
1145 | result->flags |= DIGRAPH; | |
1146 | result->type = CPP_CLOSE_BRACE; | |
1147 | } | |
1148 | else | |
1149 | { | |
1150 | BACKUP (); | |
1151 | result->type = CPP_MOD; | |
1152 | } | |
0d9f234d NB |
1153 | break; |
1154 | ||
cbcff6df | 1155 | case '.': |
480709cc NB |
1156 | result->type = CPP_DOT; |
1157 | c = get_effective_char (pfile); | |
1158 | if (c == '.') | |
1159 | { | |
1160 | const unsigned char *pos = buffer->cur; | |
1161 | ||
1162 | if (get_effective_char (pfile) == '.') | |
1163 | result->type = CPP_ELLIPSIS; | |
1164 | else | |
1165 | buffer->cur = pos - 1; | |
1166 | } | |
1167 | /* All known character sets have 0...9 contiguous. */ | |
0df6c2c7 | 1168 | else if (ISDIGIT (c)) |
480709cc NB |
1169 | { |
1170 | result->type = CPP_NUMBER; | |
1171 | parse_number (pfile, &result->val.str, c, 1); | |
1172 | } | |
1173 | else if (c == '*' && CPP_OPTION (pfile, cplusplus)) | |
1174 | result->type = CPP_DOT_STAR; | |
1175 | else | |
1176 | BACKUP (); | |
0d9f234d | 1177 | break; |
45b966db | 1178 | |
0d9f234d | 1179 | case '+': |
29401c30 | 1180 | c = get_effective_char (pfile); |
480709cc NB |
1181 | if (c == '+') |
1182 | result->type = CPP_PLUS_PLUS; | |
1183 | else if (c == '=') | |
1184 | result->type = CPP_PLUS_EQ; | |
1185 | else | |
1186 | { | |
1187 | BACKUP (); | |
1188 | result->type = CPP_PLUS; | |
1189 | } | |
0d9f234d | 1190 | break; |
04e3ec78 | 1191 | |
0d9f234d | 1192 | case '-': |
29401c30 | 1193 | c = get_effective_char (pfile); |
0d9f234d NB |
1194 | if (c == '>') |
1195 | { | |
480709cc NB |
1196 | result->type = CPP_DEREF; |
1197 | if (CPP_OPTION (pfile, cplusplus)) | |
1198 | { | |
1199 | if (get_effective_char (pfile) == '*') | |
1200 | result->type = CPP_DEREF_STAR; | |
1201 | else | |
1202 | BACKUP (); | |
1203 | } | |
0d9f234d | 1204 | } |
0d9f234d | 1205 | else if (c == '-') |
480709cc NB |
1206 | result->type = CPP_MINUS_MINUS; |
1207 | else if (c == '=') | |
1208 | result->type = CPP_MINUS_EQ; | |
1209 | else | |
1210 | { | |
1211 | BACKUP (); | |
1212 | result->type = CPP_MINUS; | |
1213 | } | |
0d9f234d | 1214 | break; |
45b966db | 1215 | |
0d9f234d | 1216 | case '&': |
29401c30 | 1217 | c = get_effective_char (pfile); |
480709cc NB |
1218 | if (c == '&') |
1219 | result->type = CPP_AND_AND; | |
1220 | else if (c == '=') | |
1221 | result->type = CPP_AND_EQ; | |
1222 | else | |
1223 | { | |
1224 | BACKUP (); | |
1225 | result->type = CPP_AND; | |
1226 | } | |
0d9f234d NB |
1227 | break; |
1228 | ||
0d9f234d | 1229 | case '|': |
29401c30 | 1230 | c = get_effective_char (pfile); |
480709cc NB |
1231 | if (c == '|') |
1232 | result->type = CPP_OR_OR; | |
1233 | else if (c == '=') | |
1234 | result->type = CPP_OR_EQ; | |
1235 | else | |
1236 | { | |
1237 | BACKUP (); | |
1238 | result->type = CPP_OR; | |
1239 | } | |
0d9f234d | 1240 | break; |
45b966db | 1241 | |
0d9f234d | 1242 | case ':': |
29401c30 | 1243 | c = get_effective_char (pfile); |
0d9f234d | 1244 | if (c == ':' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1245 | result->type = CPP_SCOPE; |
0d9f234d NB |
1246 | else if (c == '>' && CPP_OPTION (pfile, digraphs)) |
1247 | { | |
1248 | result->flags |= DIGRAPH; | |
480709cc NB |
1249 | result->type = CPP_CLOSE_SQUARE; |
1250 | } | |
1251 | else | |
1252 | { | |
1253 | BACKUP (); | |
1254 | result->type = CPP_COLON; | |
0d9f234d NB |
1255 | } |
1256 | break; | |
45b966db | 1257 | |
480709cc NB |
1258 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1259 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1260 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1261 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1262 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1263 | ||
0d9f234d NB |
1264 | case '~': result->type = CPP_COMPL; break; |
1265 | case ',': result->type = CPP_COMMA; break; | |
1266 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1267 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1268 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1269 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1270 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1271 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1272 | case ';': result->type = CPP_SEMICOLON; break; | |
1273 | ||
cc937581 ZW |
1274 | /* @ is a punctuator in Objective C. */ |
1275 | case '@': result->type = CPP_ATSIGN; break; | |
0d9f234d | 1276 | |
0abc6a6a NB |
1277 | case '$': |
1278 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
1279 | goto start_ident; | |
1280 | /* Fall through... */ | |
1281 | ||
0d9f234d NB |
1282 | random_char: |
1283 | default: | |
1284 | result->type = CPP_OTHER; | |
6c53ebff | 1285 | result->val.c = c; |
0d9f234d NB |
1286 | break; |
1287 | } | |
bdcbe496 NB |
1288 | |
1289 | return result; | |
0d9f234d NB |
1290 | } |
1291 | ||
93c80368 NB |
1292 | /* An upper bound on the number of bytes needed to spell a token, |
1293 | including preceding whitespace. */ | |
1294 | unsigned int | |
1295 | cpp_token_len (token) | |
1296 | const cpp_token *token; | |
0d9f234d | 1297 | { |
93c80368 | 1298 | unsigned int len; |
6d2c2047 | 1299 | |
93c80368 | 1300 | switch (TOKEN_SPELL (token)) |
041c3194 | 1301 | { |
a28c5035 | 1302 | default: len = 0; break; |
47ad4138 | 1303 | case SPELL_NUMBER: |
a28c5035 NB |
1304 | case SPELL_STRING: len = token->val.str.len; break; |
1305 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; | |
041c3194 | 1306 | } |
47ad4138 | 1307 | /* 1 for whitespace, 4 for comment delimiters. */ |
93c80368 | 1308 | return len + 5; |
6d2c2047 ZW |
1309 | } |
1310 | ||
041c3194 | 1311 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
cf00a885 ZW |
1312 | already contain the enough space to hold the token's spelling. |
1313 | Returns a pointer to the character after the last character | |
1314 | written. */ | |
93c80368 NB |
1315 | unsigned char * |
1316 | cpp_spell_token (pfile, token, buffer) | |
041c3194 ZW |
1317 | cpp_reader *pfile; /* Would be nice to be rid of this... */ |
1318 | const cpp_token *token; | |
1319 | unsigned char *buffer; | |
1320 | { | |
96be6998 | 1321 | switch (TOKEN_SPELL (token)) |
041c3194 ZW |
1322 | { |
1323 | case SPELL_OPERATOR: | |
1324 | { | |
1325 | const unsigned char *spelling; | |
1326 | unsigned char c; | |
d6d5f795 | 1327 | |
041c3194 | 1328 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1329 | spelling |
1330 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
92936ecf ZW |
1331 | else if (token->flags & NAMED_OP) |
1332 | goto spell_ident; | |
041c3194 | 1333 | else |
96be6998 | 1334 | spelling = TOKEN_NAME (token); |
041c3194 ZW |
1335 | |
1336 | while ((c = *spelling++) != '\0') | |
1337 | *buffer++ = c; | |
1338 | } | |
1339 | break; | |
d6d5f795 | 1340 | |
47ad4138 ZW |
1341 | case SPELL_CHAR: |
1342 | *buffer++ = token->val.c; | |
1343 | break; | |
1344 | ||
1345 | spell_ident: | |
041c3194 | 1346 | case SPELL_IDENT: |
a28c5035 NB |
1347 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1348 | buffer += NODE_LEN (token->val.node); | |
041c3194 | 1349 | break; |
d6d5f795 | 1350 | |
47ad4138 ZW |
1351 | case SPELL_NUMBER: |
1352 | memcpy (buffer, token->val.str.text, token->val.str.len); | |
1353 | buffer += token->val.str.len; | |
1354 | break; | |
1355 | ||
041c3194 ZW |
1356 | case SPELL_STRING: |
1357 | { | |
ba89d661 ZW |
1358 | int left, right, tag; |
1359 | switch (token->type) | |
1360 | { | |
1361 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1362 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
ba89d661 ZW |
1363 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1364 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1365 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 ZW |
1366 | default: |
1367 | cpp_ice (pfile, "unknown string token %s\n", TOKEN_NAME (token)); | |
1368 | return buffer; | |
ba89d661 ZW |
1369 | } |
1370 | if (tag) *buffer++ = tag; | |
47ad4138 | 1371 | *buffer++ = left; |
bfb9dc7f ZW |
1372 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1373 | buffer += token->val.str.len; | |
47ad4138 | 1374 | *buffer++ = right; |
041c3194 ZW |
1375 | } |
1376 | break; | |
d6d5f795 | 1377 | |
041c3194 | 1378 | case SPELL_NONE: |
1f978f5f | 1379 | cpp_ice (pfile, "unspellable token %s", TOKEN_NAME (token)); |
041c3194 ZW |
1380 | break; |
1381 | } | |
d6d5f795 | 1382 | |
041c3194 ZW |
1383 | return buffer; |
1384 | } | |
d6d5f795 | 1385 | |
93c80368 NB |
1386 | /* Returns a token as a null-terminated string. The string is |
1387 | temporary, and automatically freed later. Useful for diagnostics. */ | |
1388 | unsigned char * | |
1389 | cpp_token_as_text (pfile, token) | |
c5a04734 | 1390 | cpp_reader *pfile; |
041c3194 | 1391 | const cpp_token *token; |
c5a04734 | 1392 | { |
93c80368 | 1393 | unsigned int len = cpp_token_len (token); |
ece54d54 | 1394 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
c5a04734 | 1395 | |
93c80368 NB |
1396 | end = cpp_spell_token (pfile, token, start); |
1397 | end[0] = '\0'; | |
c5a04734 | 1398 | |
93c80368 NB |
1399 | return start; |
1400 | } | |
c5a04734 | 1401 | |
93c80368 NB |
1402 | /* Used by C front ends. Should really move to using cpp_token_as_text. */ |
1403 | const char * | |
1404 | cpp_type2name (type) | |
1405 | enum cpp_ttype type; | |
1406 | { | |
1407 | return (const char *) token_spellings[type].name; | |
1408 | } | |
c5a04734 | 1409 | |
4ed5bcfb NB |
1410 | /* Writes the spelling of token to FP, without any preceding space. |
1411 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1412 | double-buffering. */ | |
93c80368 NB |
1413 | void |
1414 | cpp_output_token (token, fp) | |
1415 | const cpp_token *token; | |
1416 | FILE *fp; | |
1417 | { | |
93c80368 | 1418 | switch (TOKEN_SPELL (token)) |
c5a04734 | 1419 | { |
93c80368 NB |
1420 | case SPELL_OPERATOR: |
1421 | { | |
1422 | const unsigned char *spelling; | |
3b681e9d | 1423 | int c; |
c5a04734 | 1424 | |
93c80368 | 1425 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1426 | spelling |
1427 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
93c80368 NB |
1428 | else if (token->flags & NAMED_OP) |
1429 | goto spell_ident; | |
1430 | else | |
1431 | spelling = TOKEN_NAME (token); | |
041c3194 | 1432 | |
3b681e9d ZW |
1433 | c = *spelling; |
1434 | do | |
1435 | putc (c, fp); | |
1436 | while ((c = *++spelling) != '\0'); | |
93c80368 NB |
1437 | } |
1438 | break; | |
041c3194 | 1439 | |
47ad4138 ZW |
1440 | case SPELL_CHAR: |
1441 | putc (token->val.c, fp); | |
1442 | break; | |
1443 | ||
93c80368 NB |
1444 | spell_ident: |
1445 | case SPELL_IDENT: | |
3b681e9d | 1446 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
93c80368 | 1447 | break; |
041c3194 | 1448 | |
47ad4138 ZW |
1449 | case SPELL_NUMBER: |
1450 | fwrite (token->val.str.text, 1, token->val.str.len, fp); | |
1451 | break; | |
1452 | ||
93c80368 NB |
1453 | case SPELL_STRING: |
1454 | { | |
1455 | int left, right, tag; | |
1456 | switch (token->type) | |
1457 | { | |
1458 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1459 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
93c80368 NB |
1460 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1461 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1462 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 ZW |
1463 | default: |
1464 | fprintf (stderr, "impossible STRING token %s\n", TOKEN_NAME (token)); | |
1465 | return; | |
93c80368 NB |
1466 | } |
1467 | if (tag) putc (tag, fp); | |
47ad4138 | 1468 | putc (left, fp); |
93c80368 | 1469 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
47ad4138 | 1470 | putc (right, fp); |
93c80368 NB |
1471 | } |
1472 | break; | |
c5a04734 | 1473 | |
93c80368 NB |
1474 | case SPELL_NONE: |
1475 | /* An error, most probably. */ | |
1476 | break; | |
041c3194 | 1477 | } |
c5a04734 ZW |
1478 | } |
1479 | ||
93c80368 NB |
1480 | /* Compare two tokens. */ |
1481 | int | |
1482 | _cpp_equiv_tokens (a, b) | |
1483 | const cpp_token *a, *b; | |
c5a04734 | 1484 | { |
93c80368 NB |
1485 | if (a->type == b->type && a->flags == b->flags) |
1486 | switch (TOKEN_SPELL (a)) | |
1487 | { | |
1488 | default: /* Keep compiler happy. */ | |
1489 | case SPELL_OPERATOR: | |
1490 | return 1; | |
1491 | case SPELL_CHAR: | |
6c53ebff | 1492 | return a->val.c == b->val.c; /* Character. */ |
93c80368 | 1493 | case SPELL_NONE: |
56051c0a | 1494 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
93c80368 NB |
1495 | case SPELL_IDENT: |
1496 | return a->val.node == b->val.node; | |
47ad4138 | 1497 | case SPELL_NUMBER: |
93c80368 NB |
1498 | case SPELL_STRING: |
1499 | return (a->val.str.len == b->val.str.len | |
1500 | && !memcmp (a->val.str.text, b->val.str.text, | |
1501 | a->val.str.len)); | |
1502 | } | |
c5a04734 | 1503 | |
041c3194 ZW |
1504 | return 0; |
1505 | } | |
1506 | ||
93c80368 NB |
1507 | /* Returns nonzero if a space should be inserted to avoid an |
1508 | accidental token paste for output. For simplicity, it is | |
1509 | conservative, and occasionally advises a space where one is not | |
1510 | needed, e.g. "." and ".2". */ | |
041c3194 | 1511 | |
93c80368 NB |
1512 | int |
1513 | cpp_avoid_paste (pfile, token1, token2) | |
c5a04734 | 1514 | cpp_reader *pfile; |
93c80368 | 1515 | const cpp_token *token1, *token2; |
c5a04734 | 1516 | { |
93c80368 NB |
1517 | enum cpp_ttype a = token1->type, b = token2->type; |
1518 | cppchar_t c; | |
c5a04734 | 1519 | |
93c80368 NB |
1520 | if (token1->flags & NAMED_OP) |
1521 | a = CPP_NAME; | |
1522 | if (token2->flags & NAMED_OP) | |
1523 | b = CPP_NAME; | |
c5a04734 | 1524 | |
93c80368 NB |
1525 | c = EOF; |
1526 | if (token2->flags & DIGRAPH) | |
37b8524c | 1527 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
93c80368 NB |
1528 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1529 | c = token_spellings[b].name[0]; | |
c5a04734 | 1530 | |
93c80368 | 1531 | /* Quickly get everything that can paste with an '='. */ |
37b8524c | 1532 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
93c80368 | 1533 | return 1; |
c5a04734 | 1534 | |
93c80368 | 1535 | switch (a) |
c5a04734 | 1536 | { |
93c80368 NB |
1537 | case CPP_GREATER: return c == '>' || c == '?'; |
1538 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1539 | case CPP_PLUS: return c == '+'; | |
1540 | case CPP_MINUS: return c == '-' || c == '>'; | |
1541 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1542 | case CPP_MOD: return c == ':' || c == '>'; | |
1543 | case CPP_AND: return c == '&'; | |
1544 | case CPP_OR: return c == '|'; | |
1545 | case CPP_COLON: return c == ':' || c == '>'; | |
1546 | case CPP_DEREF: return c == '*'; | |
26ec42ee | 1547 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
93c80368 NB |
1548 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1549 | case CPP_NAME: return ((b == CPP_NUMBER | |
1550 | && name_p (pfile, &token2->val.str)) | |
1551 | || b == CPP_NAME | |
1552 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1553 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1554 | || c == '.' || c == '+' || c == '-'); | |
1555 | case CPP_OTHER: return (CPP_OPTION (pfile, objc) | |
6c53ebff | 1556 | && token1->val.c == '@' |
93c80368 NB |
1557 | && (b == CPP_NAME || b == CPP_STRING)); |
1558 | default: break; | |
c5a04734 | 1559 | } |
c5a04734 | 1560 | |
417f3e3a | 1561 | return 0; |
c5a04734 ZW |
1562 | } |
1563 | ||
93c80368 | 1564 | /* Output all the remaining tokens on the current line, and a newline |
4ed5bcfb NB |
1565 | character, to FP. Leading whitespace is removed. If there are |
1566 | macros, special token padding is not performed. */ | |
c5a04734 | 1567 | void |
93c80368 | 1568 | cpp_output_line (pfile, fp) |
c5a04734 | 1569 | cpp_reader *pfile; |
93c80368 | 1570 | FILE *fp; |
c5a04734 | 1571 | { |
4ed5bcfb | 1572 | const cpp_token *token; |
96be6998 | 1573 | |
4ed5bcfb NB |
1574 | token = cpp_get_token (pfile); |
1575 | while (token->type != CPP_EOF) | |
96be6998 | 1576 | { |
4ed5bcfb NB |
1577 | cpp_output_token (token, fp); |
1578 | token = cpp_get_token (pfile); | |
1579 | if (token->flags & PREV_WHITE) | |
1580 | putc (' ', fp); | |
96be6998 ZW |
1581 | } |
1582 | ||
93c80368 | 1583 | putc ('\n', fp); |
041c3194 | 1584 | } |
c5a04734 | 1585 | |
c8a96070 NB |
1586 | /* Returns the value of a hexadecimal digit. */ |
1587 | static unsigned int | |
1588 | hex_digit_value (c) | |
1589 | unsigned int c; | |
1590 | { | |
9e1ac915 KG |
1591 | if (hex_p (c)) |
1592 | return hex_value (c); | |
1593 | else | |
1594 | abort (); | |
c8a96070 NB |
1595 | } |
1596 | ||
62729350 NB |
1597 | /* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. Returns 1 to indicate |
1598 | failure if cpplib is not parsing C++ or C99. Such failure is | |
1599 | silent, and no variables are updated. Otherwise returns 0, and | |
1600 | warns if -Wtraditional. | |
c8a96070 NB |
1601 | |
1602 | [lex.charset]: The character designated by the universal character | |
1603 | name \UNNNNNNNN is that character whose character short name in | |
1604 | ISO/IEC 10646 is NNNNNNNN; the character designated by the | |
1605 | universal character name \uNNNN is that character whose character | |
1606 | short name in ISO/IEC 10646 is 0000NNNN. If the hexadecimal value | |
1607 | for a universal character name is less than 0x20 or in the range | |
1608 | 0x7F-0x9F (inclusive), or if the universal character name | |
1609 | designates a character in the basic source character set, then the | |
1610 | program is ill-formed. | |
1611 | ||
1612 | We assume that wchar_t is Unicode, so we don't need to do any | |
62729350 | 1613 | mapping. Is this ever wrong? |
c8a96070 | 1614 | |
62729350 NB |
1615 | PC points to the 'u' or 'U', PSTR is points to the byte after PC, |
1616 | LIMIT is the end of the string or charconst. PSTR is updated to | |
1617 | point after the UCS on return, and the UCS is written into PC. */ | |
1618 | ||
1619 | static int | |
1620 | maybe_read_ucs (pfile, pstr, limit, pc) | |
c8a96070 NB |
1621 | cpp_reader *pfile; |
1622 | const unsigned char **pstr; | |
1623 | const unsigned char *limit; | |
62729350 | 1624 | unsigned int *pc; |
c8a96070 NB |
1625 | { |
1626 | const unsigned char *p = *pstr; | |
62729350 NB |
1627 | unsigned int code = 0; |
1628 | unsigned int c = *pc, length; | |
1629 | ||
1630 | /* Only attempt to interpret a UCS for C++ and C99. */ | |
1631 | if (! (CPP_OPTION (pfile, cplusplus) || CPP_OPTION (pfile, c99))) | |
1632 | return 1; | |
c8a96070 | 1633 | |
62729350 NB |
1634 | if (CPP_WTRADITIONAL (pfile)) |
1635 | cpp_warning (pfile, "the meaning of '\\%c' varies with -traditional", c); | |
c8a96070 | 1636 | |
f8710242 NB |
1637 | length = (c == 'u' ? 4: 8); |
1638 | ||
1639 | if ((size_t) (limit - p) < length) | |
1640 | { | |
1641 | cpp_error (pfile, "incomplete universal-character-name"); | |
1642 | /* Skip to the end to avoid more diagnostics. */ | |
1643 | p = limit; | |
1644 | } | |
1645 | else | |
1646 | { | |
1647 | for (; length; length--, p++) | |
c8a96070 | 1648 | { |
f8710242 NB |
1649 | c = *p; |
1650 | if (ISXDIGIT (c)) | |
1651 | code = (code << 4) + hex_digit_value (c); | |
1652 | else | |
1653 | { | |
1654 | cpp_error (pfile, | |
1655 | "non-hex digit '%c' in universal-character-name", c); | |
1656 | /* We shouldn't skip in case there are multibyte chars. */ | |
1657 | break; | |
1658 | } | |
c8a96070 | 1659 | } |
c8a96070 NB |
1660 | } |
1661 | ||
1662 | #ifdef TARGET_EBCDIC | |
1663 | cpp_error (pfile, "universal-character-name on EBCDIC target"); | |
1664 | code = 0x3f; /* EBCDIC invalid character */ | |
1665 | #else | |
f8710242 NB |
1666 | /* True extended characters are OK. */ |
1667 | if (code >= 0xa0 | |
1668 | && !(code & 0x80000000) | |
1669 | && !(code >= 0xD800 && code <= 0xDFFF)) | |
1670 | ; | |
1671 | /* The standard permits $, @ and ` to be specified as UCNs. We use | |
1672 | hex escapes so that this also works with EBCDIC hosts. */ | |
1673 | else if (code == 0x24 || code == 0x40 || code == 0x60) | |
1674 | ; | |
1675 | /* Don't give another error if one occurred above. */ | |
1676 | else if (length == 0) | |
1677 | cpp_error (pfile, "universal-character-name out of range"); | |
c8a96070 NB |
1678 | #endif |
1679 | ||
1680 | *pstr = p; | |
62729350 NB |
1681 | *pc = code; |
1682 | return 0; | |
c8a96070 NB |
1683 | } |
1684 | ||
1685 | /* Interpret an escape sequence, and return its value. PSTR points to | |
1686 | the input pointer, which is just after the backslash. LIMIT is how | |
62729350 NB |
1687 | much text we have. MASK is a bitmask for the precision for the |
1688 | destination type (char or wchar_t). TRADITIONAL, if true, does not | |
1689 | interpret escapes that did not exist in traditional C. | |
c8a96070 | 1690 | |
62729350 NB |
1691 | Handles all relevant diagnostics. */ |
1692 | ||
1693 | unsigned int | |
1694 | cpp_parse_escape (pfile, pstr, limit, mask, traditional) | |
c8a96070 NB |
1695 | cpp_reader *pfile; |
1696 | const unsigned char **pstr; | |
1697 | const unsigned char *limit; | |
62729350 | 1698 | unsigned HOST_WIDE_INT mask; |
c8a96070 NB |
1699 | int traditional; |
1700 | { | |
1701 | int unknown = 0; | |
1702 | const unsigned char *str = *pstr; | |
1703 | unsigned int c = *str++; | |
1704 | ||
1705 | switch (c) | |
1706 | { | |
1707 | case '\\': case '\'': case '"': case '?': break; | |
1708 | case 'b': c = TARGET_BS; break; | |
1709 | case 'f': c = TARGET_FF; break; | |
1710 | case 'n': c = TARGET_NEWLINE; break; | |
1711 | case 'r': c = TARGET_CR; break; | |
1712 | case 't': c = TARGET_TAB; break; | |
1713 | case 'v': c = TARGET_VT; break; | |
1714 | ||
1715 | case '(': case '{': case '[': case '%': | |
1716 | /* '\(', etc, are used at beginning of line to avoid confusing Emacs. | |
1717 | '\%' is used to prevent SCCS from getting confused. */ | |
1718 | unknown = CPP_PEDANTIC (pfile); | |
1719 | break; | |
1720 | ||
1721 | case 'a': | |
1722 | if (CPP_WTRADITIONAL (pfile)) | |
1723 | cpp_warning (pfile, "the meaning of '\\a' varies with -traditional"); | |
1724 | if (!traditional) | |
1725 | c = TARGET_BELL; | |
1726 | break; | |
1727 | ||
1728 | case 'e': case 'E': | |
1729 | if (CPP_PEDANTIC (pfile)) | |
1730 | cpp_pedwarn (pfile, "non-ISO-standard escape sequence, '\\%c'", c); | |
1731 | c = TARGET_ESC; | |
1732 | break; | |
1733 | ||
c8a96070 | 1734 | case 'u': case 'U': |
62729350 | 1735 | unknown = maybe_read_ucs (pfile, &str, limit, &c); |
c8a96070 NB |
1736 | break; |
1737 | ||
1738 | case 'x': | |
1739 | if (CPP_WTRADITIONAL (pfile)) | |
1740 | cpp_warning (pfile, "the meaning of '\\x' varies with -traditional"); | |
1741 | ||
1742 | if (!traditional) | |
1743 | { | |
1744 | unsigned int i = 0, overflow = 0; | |
1745 | int digits_found = 0; | |
1746 | ||
1747 | while (str < limit) | |
1748 | { | |
1749 | c = *str; | |
1750 | if (! ISXDIGIT (c)) | |
1751 | break; | |
1752 | str++; | |
1753 | overflow |= i ^ (i << 4 >> 4); | |
1754 | i = (i << 4) + hex_digit_value (c); | |
1755 | digits_found = 1; | |
1756 | } | |
1757 | ||
1758 | if (!digits_found) | |
1759 | cpp_error (pfile, "\\x used with no following hex digits"); | |
1760 | ||
1761 | if (overflow | (i != (i & mask))) | |
1762 | { | |
1763 | cpp_pedwarn (pfile, "hex escape sequence out of range"); | |
1764 | i &= mask; | |
1765 | } | |
1766 | c = i; | |
1767 | } | |
1768 | break; | |
1769 | ||
1770 | case '0': case '1': case '2': case '3': | |
1771 | case '4': case '5': case '6': case '7': | |
1772 | { | |
1773 | unsigned int i = c - '0'; | |
1774 | int count = 0; | |
1775 | ||
1776 | while (str < limit && ++count < 3) | |
1777 | { | |
1778 | c = *str; | |
1779 | if (c < '0' || c > '7') | |
1780 | break; | |
1781 | str++; | |
1782 | i = (i << 3) + c - '0'; | |
1783 | } | |
1784 | ||
1785 | if (i != (i & mask)) | |
1786 | { | |
1787 | cpp_pedwarn (pfile, "octal escape sequence out of range"); | |
1788 | i &= mask; | |
1789 | } | |
1790 | c = i; | |
1791 | } | |
1792 | break; | |
1793 | ||
1794 | default: | |
1795 | unknown = 1; | |
1796 | break; | |
1797 | } | |
1798 | ||
1799 | if (unknown) | |
1800 | { | |
1801 | if (ISGRAPH (c)) | |
1802 | cpp_pedwarn (pfile, "unknown escape sequence '\\%c'", c); | |
1803 | else | |
1804 | cpp_pedwarn (pfile, "unknown escape sequence: '\\%03o'", c); | |
1805 | } | |
1806 | ||
62729350 NB |
1807 | if (c > mask) |
1808 | cpp_pedwarn (pfile, "escape sequence out of range for character"); | |
1809 | ||
c8a96070 NB |
1810 | *pstr = str; |
1811 | return c; | |
1812 | } | |
1813 | ||
1814 | #ifndef MAX_CHAR_TYPE_SIZE | |
1815 | #define MAX_CHAR_TYPE_SIZE CHAR_TYPE_SIZE | |
1816 | #endif | |
1817 | ||
1818 | #ifndef MAX_WCHAR_TYPE_SIZE | |
1819 | #define MAX_WCHAR_TYPE_SIZE WCHAR_TYPE_SIZE | |
1820 | #endif | |
1821 | ||
1822 | /* Interpret a (possibly wide) character constant in TOKEN. | |
1823 | WARN_MULTI warns about multi-character charconsts, if not | |
1824 | TRADITIONAL. TRADITIONAL also indicates not to interpret escapes | |
1825 | that did not exist in traditional C. PCHARS_SEEN points to a | |
1826 | variable that is filled in with the number of characters seen. */ | |
1827 | HOST_WIDE_INT | |
1828 | cpp_interpret_charconst (pfile, token, warn_multi, traditional, pchars_seen) | |
1829 | cpp_reader *pfile; | |
1830 | const cpp_token *token; | |
1831 | int warn_multi; | |
1832 | int traditional; | |
1833 | unsigned int *pchars_seen; | |
1834 | { | |
1835 | const unsigned char *str = token->val.str.text; | |
1836 | const unsigned char *limit = str + token->val.str.len; | |
1837 | unsigned int chars_seen = 0; | |
1838 | unsigned int width, max_chars, c; | |
2a967f3d NB |
1839 | unsigned HOST_WIDE_INT mask; |
1840 | HOST_WIDE_INT result = 0; | |
c8a96070 NB |
1841 | |
1842 | #ifdef MULTIBYTE_CHARS | |
1843 | (void) local_mbtowc (NULL, NULL, 0); | |
1844 | #endif | |
1845 | ||
1846 | /* Width in bits. */ | |
1847 | if (token->type == CPP_CHAR) | |
1848 | width = MAX_CHAR_TYPE_SIZE; | |
1849 | else | |
1850 | width = MAX_WCHAR_TYPE_SIZE; | |
1851 | ||
1852 | if (width < HOST_BITS_PER_WIDE_INT) | |
1853 | mask = ((unsigned HOST_WIDE_INT) 1 << width) - 1; | |
1854 | else | |
1855 | mask = ~0; | |
1856 | max_chars = HOST_BITS_PER_WIDE_INT / width; | |
1857 | ||
1858 | while (str < limit) | |
1859 | { | |
1860 | #ifdef MULTIBYTE_CHARS | |
1861 | wchar_t wc; | |
1862 | int char_len; | |
1863 | ||
1864 | char_len = local_mbtowc (&wc, str, limit - str); | |
1865 | if (char_len == -1) | |
1866 | { | |
1867 | cpp_warning (pfile, "ignoring invalid multibyte character"); | |
1868 | c = *str++; | |
1869 | } | |
1870 | else | |
1871 | { | |
1872 | str += char_len; | |
1873 | c = wc; | |
1874 | } | |
1875 | #else | |
1876 | c = *str++; | |
1877 | #endif | |
1878 | ||
1879 | if (c == '\\') | |
62729350 | 1880 | c = cpp_parse_escape (pfile, &str, limit, mask, traditional); |
c8a96070 NB |
1881 | |
1882 | #ifdef MAP_CHARACTER | |
1883 | if (ISPRINT (c)) | |
1884 | c = MAP_CHARACTER (c); | |
1885 | #endif | |
1886 | ||
1887 | /* Merge character into result; ignore excess chars. */ | |
1888 | if (++chars_seen <= max_chars) | |
1889 | { | |
1890 | if (width < HOST_BITS_PER_WIDE_INT) | |
1891 | result = (result << width) | (c & mask); | |
1892 | else | |
1893 | result = c; | |
1894 | } | |
1895 | } | |
1896 | ||
1897 | if (chars_seen == 0) | |
1898 | cpp_error (pfile, "empty character constant"); | |
1899 | else if (chars_seen > max_chars) | |
1900 | { | |
1901 | chars_seen = max_chars; | |
f8710242 | 1902 | cpp_warning (pfile, "character constant too long"); |
c8a96070 NB |
1903 | } |
1904 | else if (chars_seen > 1 && !traditional && warn_multi) | |
1905 | cpp_warning (pfile, "multi-character character constant"); | |
1906 | ||
1907 | /* If char type is signed, sign-extend the constant. The | |
1908 | __CHAR_UNSIGNED__ macro is set by the driver if appropriate. */ | |
1909 | if (token->type == CPP_CHAR && chars_seen) | |
1910 | { | |
1911 | unsigned int nbits = chars_seen * width; | |
c8a96070 | 1912 | |
e1e7d56b | 1913 | mask = (unsigned HOST_WIDE_INT) ~0 >> (HOST_BITS_PER_WIDE_INT - nbits); |
c8a96070 NB |
1914 | if (pfile->spec_nodes.n__CHAR_UNSIGNED__->type == NT_MACRO |
1915 | || ((result >> (nbits - 1)) & 1) == 0) | |
1916 | result &= mask; | |
1917 | else | |
1918 | result |= ~mask; | |
1919 | } | |
1920 | ||
1921 | *pchars_seen = chars_seen; | |
1922 | return result; | |
1923 | } | |
1924 | ||
1e013d2e NB |
1925 | /* Memory buffers. Changing these three constants can have a dramatic |
1926 | effect on performance. The values here are reasonable defaults, | |
1927 | but might be tuned. If you adjust them, be sure to test across a | |
1928 | range of uses of cpplib, including heavy nested function-like macro | |
1929 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1930 | good tool for this). */ | |
1931 | #define MIN_BUFF_SIZE 8000 | |
87062813 | 1932 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
1e013d2e NB |
1933 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1934 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
417f3e3a | 1935 | |
87062813 NB |
1936 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1937 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1938 | #endif | |
1939 | ||
93c80368 | 1940 | struct dummy |
417f3e3a | 1941 | { |
93c80368 NB |
1942 | char c; |
1943 | union | |
1944 | { | |
1945 | double d; | |
1946 | int *p; | |
1947 | } u; | |
1948 | }; | |
417f3e3a | 1949 | |
93c80368 | 1950 | #define DEFAULT_ALIGNMENT (offsetof (struct dummy, u)) |
b8af0ca5 NB |
1951 | #define CPP_ALIGN(size, align) (((size) + ((align) - 1)) & ~((align) - 1)) |
1952 | ||
c9e7a609 NB |
1953 | /* Create a new allocation buffer. Place the control block at the end |
1954 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
b8af0ca5 NB |
1955 | static _cpp_buff * |
1956 | new_buff (len) | |
6142088c | 1957 | size_t len; |
b8af0ca5 NB |
1958 | { |
1959 | _cpp_buff *result; | |
ece54d54 | 1960 | unsigned char *base; |
b8af0ca5 | 1961 | |
1e013d2e NB |
1962 | if (len < MIN_BUFF_SIZE) |
1963 | len = MIN_BUFF_SIZE; | |
b8af0ca5 NB |
1964 | len = CPP_ALIGN (len, DEFAULT_ALIGNMENT); |
1965 | ||
1966 | base = xmalloc (len + sizeof (_cpp_buff)); | |
1967 | result = (_cpp_buff *) (base + len); | |
1968 | result->base = base; | |
1969 | result->cur = base; | |
1970 | result->limit = base + len; | |
1971 | result->next = NULL; | |
1972 | return result; | |
1973 | } | |
1974 | ||
1975 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
1976 | void | |
1977 | _cpp_release_buff (pfile, buff) | |
1978 | cpp_reader *pfile; | |
1979 | _cpp_buff *buff; | |
1980 | { | |
1981 | _cpp_buff *end = buff; | |
1982 | ||
1983 | while (end->next) | |
1984 | end = end->next; | |
1985 | end->next = pfile->free_buffs; | |
1986 | pfile->free_buffs = buff; | |
1987 | } | |
1988 | ||
1989 | /* Return a free buffer of size at least MIN_SIZE. */ | |
1990 | _cpp_buff * | |
1991 | _cpp_get_buff (pfile, min_size) | |
1992 | cpp_reader *pfile; | |
6142088c | 1993 | size_t min_size; |
b8af0ca5 NB |
1994 | { |
1995 | _cpp_buff *result, **p; | |
1996 | ||
1997 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
1998 | { | |
6142088c | 1999 | size_t size; |
1e013d2e NB |
2000 | |
2001 | if (*p == NULL) | |
b8af0ca5 | 2002 | return new_buff (min_size); |
1e013d2e NB |
2003 | result = *p; |
2004 | size = result->limit - result->base; | |
2005 | /* Return a buffer that's big enough, but don't waste one that's | |
2006 | way too big. */ | |
34f5271d | 2007 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
b8af0ca5 NB |
2008 | break; |
2009 | } | |
2010 | ||
2011 | *p = result->next; | |
2012 | result->next = NULL; | |
2013 | result->cur = result->base; | |
2014 | return result; | |
2015 | } | |
2016 | ||
4fe9b91c | 2017 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
2018 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
2019 | the excess bytes to the new buffer. Chains the new buffer after | |
2020 | BUFF, and returns the new buffer. */ | |
b8af0ca5 | 2021 | _cpp_buff * |
8c3b2693 | 2022 | _cpp_append_extend_buff (pfile, buff, min_extra) |
b8af0ca5 NB |
2023 | cpp_reader *pfile; |
2024 | _cpp_buff *buff; | |
6142088c | 2025 | size_t min_extra; |
b8af0ca5 | 2026 | { |
6142088c | 2027 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
8c3b2693 | 2028 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
b8af0ca5 | 2029 | |
8c3b2693 NB |
2030 | buff->next = new_buff; |
2031 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
2032 | return new_buff; | |
2033 | } | |
2034 | ||
4fe9b91c | 2035 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
2036 | remaining bytes of the buffer pointed to by BUFF, and at least |
2037 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
2038 | Chains the new buffer before the buffer pointed to by BUFF, and | |
2039 | updates the pointer to point to the new buffer. */ | |
2040 | void | |
2041 | _cpp_extend_buff (pfile, pbuff, min_extra) | |
2042 | cpp_reader *pfile; | |
2043 | _cpp_buff **pbuff; | |
2044 | size_t min_extra; | |
2045 | { | |
2046 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
2047 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
2048 | ||
2049 | new_buff = _cpp_get_buff (pfile, size); | |
2050 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
2051 | new_buff->next = old_buff; | |
2052 | *pbuff = new_buff; | |
b8af0ca5 NB |
2053 | } |
2054 | ||
2055 | /* Free a chain of buffers starting at BUFF. */ | |
2056 | void | |
2057 | _cpp_free_buff (buff) | |
2058 | _cpp_buff *buff; | |
2059 | { | |
2060 | _cpp_buff *next; | |
2061 | ||
2062 | for (; buff; buff = next) | |
2063 | { | |
2064 | next = buff->next; | |
2065 | free (buff->base); | |
2066 | } | |
2067 | } | |
417f3e3a | 2068 | |
ece54d54 NB |
2069 | /* Allocate permanent, unaligned storage of length LEN. */ |
2070 | unsigned char * | |
2071 | _cpp_unaligned_alloc (pfile, len) | |
2072 | cpp_reader *pfile; | |
2073 | size_t len; | |
2074 | { | |
2075 | _cpp_buff *buff = pfile->u_buff; | |
2076 | unsigned char *result = buff->cur; | |
2077 | ||
2078 | if (len > (size_t) (buff->limit - result)) | |
2079 | { | |
2080 | buff = _cpp_get_buff (pfile, len); | |
2081 | buff->next = pfile->u_buff; | |
2082 | pfile->u_buff = buff; | |
2083 | result = buff->cur; | |
2084 | } | |
2085 | ||
2086 | buff->cur = result + len; | |
2087 | return result; | |
2088 | } | |
2089 | ||
87062813 NB |
2090 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
2091 | That buffer is used for growing allocations when saving macro | |
2092 | replacement lists in a #define, and when parsing an answer to an | |
2093 | assertion in #assert, #unassert or #if (and therefore possibly | |
2094 | whilst expanding macros). It therefore must not be used by any | |
2095 | code that they might call: specifically the lexer and the guts of | |
2096 | the macro expander. | |
2097 | ||
2098 | All existing other uses clearly fit this restriction: storing | |
2099 | registered pragmas during initialization. */ | |
93c80368 | 2100 | unsigned char * |
8c3b2693 NB |
2101 | _cpp_aligned_alloc (pfile, len) |
2102 | cpp_reader *pfile; | |
2103 | size_t len; | |
3fef5b2b | 2104 | { |
8c3b2693 NB |
2105 | _cpp_buff *buff = pfile->a_buff; |
2106 | unsigned char *result = buff->cur; | |
3fef5b2b | 2107 | |
8c3b2693 | 2108 | if (len > (size_t) (buff->limit - result)) |
3fef5b2b | 2109 | { |
8c3b2693 NB |
2110 | buff = _cpp_get_buff (pfile, len); |
2111 | buff->next = pfile->a_buff; | |
2112 | pfile->a_buff = buff; | |
2113 | result = buff->cur; | |
3fef5b2b | 2114 | } |
041c3194 | 2115 | |
8c3b2693 | 2116 | buff->cur = result + len; |
93c80368 | 2117 | return result; |
041c3194 | 2118 | } |