]>
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" | |
4977bab6 ZW |
25 | #include "coretypes.h" |
26 | #include "tm.h" | |
45b966db ZW |
27 | #include "cpplib.h" |
28 | #include "cpphash.h" | |
29 | ||
c8a96070 NB |
30 | #ifdef MULTIBYTE_CHARS |
31 | #include "mbchar.h" | |
32 | #include <locale.h> | |
33 | #endif | |
34 | ||
93c80368 NB |
35 | /* Tokens with SPELL_STRING store their spelling in the token list, |
36 | and it's length in the token->val.name.len. */ | |
37 | enum spell_type | |
f9a0e96c | 38 | { |
93c80368 NB |
39 | SPELL_OPERATOR = 0, |
40 | SPELL_CHAR, | |
41 | SPELL_IDENT, | |
47ad4138 | 42 | SPELL_NUMBER, |
93c80368 NB |
43 | SPELL_STRING, |
44 | SPELL_NONE | |
f9a0e96c ZW |
45 | }; |
46 | ||
93c80368 | 47 | struct token_spelling |
f9a0e96c | 48 | { |
93c80368 NB |
49 | enum spell_type category; |
50 | const unsigned char *name; | |
f9a0e96c ZW |
51 | }; |
52 | ||
8206c799 ZW |
53 | static const unsigned char *const digraph_spellings[] = |
54 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
93c80368 NB |
55 | |
56 | #define OP(e, s) { SPELL_OPERATOR, U s }, | |
57 | #define TK(e, s) { s, U STRINGX (e) }, | |
8206c799 | 58 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
93c80368 NB |
59 | #undef OP |
60 | #undef TK | |
61 | ||
62 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
63 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
480709cc | 64 | #define BACKUP() do {buffer->cur = buffer->backup_to;} while (0) |
f2d5f0cc | 65 | |
87062813 NB |
66 | static void handle_newline PARAMS ((cpp_reader *)); |
67 | static cppchar_t skip_escaped_newlines PARAMS ((cpp_reader *)); | |
29401c30 | 68 | static cppchar_t get_effective_char PARAMS ((cpp_reader *)); |
0d9f234d | 69 | |
041c3194 | 70 | static int skip_block_comment PARAMS ((cpp_reader *)); |
cbcff6df | 71 | static int skip_line_comment PARAMS ((cpp_reader *)); |
0d9f234d | 72 | static void adjust_column PARAMS ((cpp_reader *)); |
4d6baafa | 73 | static int skip_whitespace PARAMS ((cpp_reader *, cppchar_t)); |
2c3fcba6 | 74 | static cpp_hashnode *parse_identifier PARAMS ((cpp_reader *)); |
562a5c27 | 75 | static uchar *parse_slow PARAMS ((cpp_reader *, const uchar *, int, |
df383483 | 76 | unsigned int *)); |
10cf9bde | 77 | static void parse_number PARAMS ((cpp_reader *, cpp_string *, int)); |
562a5c27 | 78 | static int unescaped_terminator_p PARAMS ((cpp_reader *, const uchar *)); |
0d9f234d | 79 | static void parse_string PARAMS ((cpp_reader *, cpp_token *, cppchar_t)); |
87062813 | 80 | static bool trigraph_p PARAMS ((cpp_reader *)); |
562a5c27 | 81 | static void save_comment PARAMS ((cpp_reader *, cpp_token *, const uchar *, |
477cdac7 | 82 | cppchar_t)); |
004cb263 | 83 | static bool continue_after_nul PARAMS ((cpp_reader *)); |
93c80368 | 84 | static int name_p PARAMS ((cpp_reader *, const cpp_string *)); |
62729350 | 85 | static int maybe_read_ucs PARAMS ((cpp_reader *, const unsigned char **, |
625458d0 | 86 | const unsigned char *, cppchar_t *)); |
5fddcffc | 87 | static tokenrun *next_tokenrun PARAMS ((tokenrun *)); |
f617b8e2 | 88 | |
c8a96070 | 89 | static unsigned int hex_digit_value PARAMS ((unsigned int)); |
6142088c | 90 | static _cpp_buff *new_buff PARAMS ((size_t)); |
15dad1d9 | 91 | |
9d10c9a9 NB |
92 | /* Change to the native locale for multibyte conversions. */ |
93 | void | |
94 | _cpp_init_mbchar () | |
95 | { | |
96 | #ifdef MULTIBYTE_CHARS | |
97 | setlocale (LC_CTYPE, ""); | |
98 | GET_ENVIRONMENT (literal_codeset, "LANG"); | |
99 | #endif | |
100 | } | |
101 | ||
041c3194 | 102 | /* Utility routine: |
9e62c811 | 103 | |
bfb9dc7f ZW |
104 | Compares, the token TOKEN to the NUL-terminated string STRING. |
105 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
041c3194 | 106 | int |
bfb9dc7f ZW |
107 | cpp_ideq (token, string) |
108 | const cpp_token *token; | |
041c3194 ZW |
109 | const char *string; |
110 | { | |
bfb9dc7f | 111 | if (token->type != CPP_NAME) |
041c3194 | 112 | return 0; |
bfb9dc7f | 113 | |
562a5c27 | 114 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
15dad1d9 | 115 | } |
1368ee70 | 116 | |
87062813 NB |
117 | /* Call when meeting a newline, assumed to be in buffer->cur[-1]. |
118 | Returns with buffer->cur pointing to the character immediately | |
119 | following the newline (combination). */ | |
120 | static void | |
121 | handle_newline (pfile) | |
1444f2ed | 122 | cpp_reader *pfile; |
0d9f234d | 123 | { |
87062813 | 124 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 125 | |
87062813 | 126 | /* Handle CR-LF and LF-CR. Most other implementations (e.g. java) |
8d9afc4e | 127 | only accept CR-LF; maybe we should fall back to that behavior? */ |
4d6baafa | 128 | if (buffer->cur[-1] + buffer->cur[0] == '\r' + '\n') |
87062813 | 129 | buffer->cur++; |
0d9f234d | 130 | |
87062813 NB |
131 | buffer->line_base = buffer->cur; |
132 | buffer->col_adjust = 0; | |
133 | pfile->line++; | |
0d9f234d NB |
134 | } |
135 | ||
87062813 NB |
136 | /* Subroutine of skip_escaped_newlines; called when a 3-character |
137 | sequence beginning with "??" is encountered. buffer->cur points to | |
138 | the second '?'. | |
139 | ||
140 | Warn if necessary, and returns true if the sequence forms a | |
8d9afc4e | 141 | trigraph and the trigraph should be honored. */ |
87062813 NB |
142 | static bool |
143 | trigraph_p (pfile) | |
45b966db | 144 | cpp_reader *pfile; |
45b966db | 145 | { |
87062813 NB |
146 | cpp_buffer *buffer = pfile->buffer; |
147 | cppchar_t from_char = buffer->cur[1]; | |
148 | bool accept; | |
149 | ||
150 | if (!_cpp_trigraph_map[from_char]) | |
151 | return false; | |
152 | ||
153 | accept = CPP_OPTION (pfile, trigraphs); | |
154 | ||
cbcff6df NB |
155 | /* Don't warn about trigraphs in comments. */ |
156 | if (CPP_OPTION (pfile, warn_trigraphs) && !pfile->state.lexing_comment) | |
45b966db | 157 | { |
041c3194 | 158 | if (accept) |
ebef4e8c NB |
159 | cpp_error_with_line (pfile, DL_WARNING, |
160 | pfile->line, CPP_BUF_COL (buffer) - 1, | |
161 | "trigraph ??%c converted to %c", | |
162 | (int) from_char, | |
163 | (int) _cpp_trigraph_map[from_char]); | |
4a5b68a2 NB |
164 | else if (buffer->cur != buffer->last_Wtrigraphs) |
165 | { | |
166 | buffer->last_Wtrigraphs = buffer->cur; | |
ebef4e8c NB |
167 | cpp_error_with_line (pfile, DL_WARNING, |
168 | pfile->line, CPP_BUF_COL (buffer) - 1, | |
169 | "trigraph ??%c ignored", (int) from_char); | |
4a5b68a2 | 170 | } |
45b966db | 171 | } |
0d9f234d | 172 | |
041c3194 | 173 | return accept; |
45b966db ZW |
174 | } |
175 | ||
87062813 | 176 | /* Skips any escaped newlines introduced by '?' or a '\\', assumed to |
480709cc NB |
177 | lie in buffer->cur[-1]. Returns the next byte, which will be in |
178 | buffer->cur[-1]. This routine performs preprocessing stages 1 and | |
179 | 2 of the ISO C standard. */ | |
0d9f234d | 180 | static cppchar_t |
87062813 | 181 | skip_escaped_newlines (pfile) |
29401c30 | 182 | cpp_reader *pfile; |
45b966db | 183 | { |
29401c30 | 184 | cpp_buffer *buffer = pfile->buffer; |
87062813 | 185 | cppchar_t next = buffer->cur[-1]; |
29401c30 | 186 | |
a5c3cccd NB |
187 | /* Only do this if we apply stages 1 and 2. */ |
188 | if (!buffer->from_stage3) | |
041c3194 | 189 | { |
a5c3cccd | 190 | const unsigned char *saved_cur; |
87062813 | 191 | cppchar_t next1; |
a5c3cccd NB |
192 | |
193 | do | |
0d9f234d | 194 | { |
a5c3cccd | 195 | if (next == '?') |
0d9f234d | 196 | { |
4d6baafa | 197 | if (buffer->cur[0] != '?' || !trigraph_p (pfile)) |
87062813 | 198 | break; |
a5c3cccd | 199 | |
87062813 NB |
200 | /* Translate the trigraph. */ |
201 | next = _cpp_trigraph_map[buffer->cur[1]]; | |
202 | buffer->cur += 2; | |
4d6baafa | 203 | if (next != '\\') |
a5c3cccd | 204 | break; |
a5c3cccd NB |
205 | } |
206 | ||
4d6baafa NB |
207 | if (buffer->cur == buffer->rlimit) |
208 | break; | |
209 | ||
87062813 NB |
210 | /* We have a backslash, and room for at least one more |
211 | character. Skip horizontal whitespace. */ | |
212 | saved_cur = buffer->cur; | |
a5c3cccd | 213 | do |
87062813 NB |
214 | next1 = *buffer->cur++; |
215 | while (is_nvspace (next1) && buffer->cur < buffer->rlimit); | |
041c3194 | 216 | |
a5c3cccd | 217 | if (!is_vspace (next1)) |
0d9f234d | 218 | { |
87062813 | 219 | buffer->cur = saved_cur; |
0d9f234d NB |
220 | break; |
221 | } | |
45b966db | 222 | |
87062813 NB |
223 | if (saved_cur != buffer->cur - 1 |
224 | && !pfile->state.lexing_comment) | |
ebef4e8c NB |
225 | cpp_error (pfile, DL_WARNING, |
226 | "backslash and newline separated by space"); | |
0d9f234d | 227 | |
87062813 | 228 | handle_newline (pfile); |
480709cc | 229 | buffer->backup_to = buffer->cur; |
87062813 NB |
230 | if (buffer->cur == buffer->rlimit) |
231 | { | |
ebef4e8c NB |
232 | cpp_error (pfile, DL_PEDWARN, |
233 | "backslash-newline at end of file"); | |
87062813 NB |
234 | next = EOF; |
235 | } | |
236 | else | |
237 | next = *buffer->cur++; | |
0d9f234d | 238 | } |
a5c3cccd | 239 | while (next == '\\' || next == '?'); |
041c3194 | 240 | } |
45b966db | 241 | |
0d9f234d | 242 | return next; |
45b966db ZW |
243 | } |
244 | ||
0d9f234d | 245 | /* Obtain the next character, after trigraph conversion and skipping |
87062813 NB |
246 | an arbitrarily long string of escaped newlines. The common case of |
247 | no trigraphs or escaped newlines falls through quickly. On return, | |
480709cc NB |
248 | buffer->backup_to points to where to return to if the character is |
249 | not to be processed. */ | |
0d9f234d | 250 | static cppchar_t |
29401c30 NB |
251 | get_effective_char (pfile) |
252 | cpp_reader *pfile; | |
64aaf407 | 253 | { |
4d6baafa | 254 | cppchar_t next; |
480709cc | 255 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 256 | |
480709cc | 257 | buffer->backup_to = buffer->cur; |
4d6baafa NB |
258 | next = *buffer->cur++; |
259 | if (__builtin_expect (next == '?' || next == '\\', 0)) | |
260 | next = skip_escaped_newlines (pfile); | |
0d9f234d | 261 | |
df383483 | 262 | return next; |
64aaf407 NB |
263 | } |
264 | ||
0d9f234d NB |
265 | /* Skip a C-style block comment. We find the end of the comment by |
266 | seeing if an asterisk is before every '/' we encounter. Returns | |
da7d8304 | 267 | nonzero if comment terminated by EOF, zero otherwise. */ |
041c3194 ZW |
268 | static int |
269 | skip_block_comment (pfile) | |
45b966db ZW |
270 | cpp_reader *pfile; |
271 | { | |
041c3194 | 272 | cpp_buffer *buffer = pfile->buffer; |
d8090680 | 273 | cppchar_t c = EOF, prevc = EOF; |
0d9f234d | 274 | |
cbcff6df | 275 | pfile->state.lexing_comment = 1; |
0d9f234d | 276 | while (buffer->cur != buffer->rlimit) |
45b966db | 277 | { |
0d9f234d NB |
278 | prevc = c, c = *buffer->cur++; |
279 | ||
0d9f234d | 280 | /* FIXME: For speed, create a new character class of characters |
93c80368 | 281 | of interest inside block comments. */ |
0d9f234d | 282 | if (c == '?' || c == '\\') |
87062813 | 283 | c = skip_escaped_newlines (pfile); |
041c3194 | 284 | |
0d9f234d NB |
285 | /* People like decorating comments with '*', so check for '/' |
286 | instead for efficiency. */ | |
041c3194 | 287 | if (c == '/') |
45b966db | 288 | { |
0d9f234d NB |
289 | if (prevc == '*') |
290 | break; | |
041c3194 | 291 | |
0d9f234d | 292 | /* Warn about potential nested comments, but not if the '/' |
a1f300c0 | 293 | comes immediately before the true comment delimiter. |
041c3194 | 294 | Don't bother to get it right across escaped newlines. */ |
0d9f234d | 295 | if (CPP_OPTION (pfile, warn_comments) |
87062813 | 296 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
ebef4e8c NB |
297 | cpp_error_with_line (pfile, DL_WARNING, |
298 | pfile->line, CPP_BUF_COL (buffer), | |
299 | "\"/*\" within comment"); | |
45b966db | 300 | } |
91fcd158 | 301 | else if (is_vspace (c)) |
87062813 | 302 | handle_newline (pfile); |
52fadca8 | 303 | else if (c == '\t') |
0d9f234d | 304 | adjust_column (pfile); |
45b966db | 305 | } |
041c3194 | 306 | |
cbcff6df | 307 | pfile->state.lexing_comment = 0; |
0d9f234d | 308 | return c != '/' || prevc != '*'; |
45b966db ZW |
309 | } |
310 | ||
480709cc | 311 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
da7d8304 | 312 | terminating newline. Handles escaped newlines. Returns nonzero |
480709cc | 313 | if a multiline comment. */ |
041c3194 | 314 | static int |
cbcff6df NB |
315 | skip_line_comment (pfile) |
316 | cpp_reader *pfile; | |
45b966db | 317 | { |
cbcff6df | 318 | cpp_buffer *buffer = pfile->buffer; |
67821e3a | 319 | unsigned int orig_line = pfile->line; |
0d9f234d | 320 | cppchar_t c; |
64cdc383 MH |
321 | #ifdef MULTIBYTE_CHARS |
322 | wchar_t wc; | |
323 | int char_len; | |
324 | #endif | |
041c3194 | 325 | |
cbcff6df | 326 | pfile->state.lexing_comment = 1; |
64cdc383 MH |
327 | #ifdef MULTIBYTE_CHARS |
328 | /* Reset multibyte conversion state. */ | |
329 | (void) local_mbtowc (NULL, NULL, 0); | |
330 | #endif | |
0d9f234d | 331 | do |
041c3194 | 332 | { |
0d9f234d | 333 | if (buffer->cur == buffer->rlimit) |
480709cc | 334 | goto at_eof; |
041c3194 | 335 | |
64cdc383 MH |
336 | #ifdef MULTIBYTE_CHARS |
337 | char_len = local_mbtowc (&wc, (const char *) buffer->cur, | |
338 | buffer->rlimit - buffer->cur); | |
339 | if (char_len == -1) | |
340 | { | |
341 | cpp_error (pfile, DL_WARNING, | |
342 | "ignoring invalid multibyte character"); | |
343 | char_len = 1; | |
344 | c = *buffer->cur++; | |
345 | } | |
346 | else | |
347 | { | |
348 | buffer->cur += char_len; | |
349 | c = wc; | |
350 | } | |
351 | #else | |
0d9f234d | 352 | c = *buffer->cur++; |
64cdc383 | 353 | #endif |
0d9f234d | 354 | if (c == '?' || c == '\\') |
87062813 | 355 | c = skip_escaped_newlines (pfile); |
041c3194 | 356 | } |
0d9f234d | 357 | while (!is_vspace (c)); |
45b966db | 358 | |
480709cc NB |
359 | /* Step back over the newline, except at EOF. */ |
360 | buffer->cur--; | |
361 | at_eof: | |
362 | ||
cbcff6df | 363 | pfile->state.lexing_comment = 0; |
67821e3a | 364 | return orig_line != pfile->line; |
041c3194 | 365 | } |
45b966db | 366 | |
0d9f234d NB |
367 | /* pfile->buffer->cur is one beyond the \t character. Update |
368 | col_adjust so we track the column correctly. */ | |
52fadca8 | 369 | static void |
0d9f234d | 370 | adjust_column (pfile) |
52fadca8 | 371 | cpp_reader *pfile; |
52fadca8 | 372 | { |
0d9f234d NB |
373 | cpp_buffer *buffer = pfile->buffer; |
374 | unsigned int col = CPP_BUF_COL (buffer) - 1; /* Zero-based column. */ | |
52fadca8 NB |
375 | |
376 | /* Round it up to multiple of the tabstop, but subtract 1 since the | |
377 | tab itself occupies a character position. */ | |
0d9f234d NB |
378 | buffer->col_adjust += (CPP_OPTION (pfile, tabstop) |
379 | - col % CPP_OPTION (pfile, tabstop)) - 1; | |
52fadca8 NB |
380 | } |
381 | ||
0d9f234d NB |
382 | /* Skips whitespace, saving the next non-whitespace character. |
383 | Adjusts pfile->col_adjust to account for tabs. Without this, | |
384 | tokens might be assigned an incorrect column. */ | |
4d6baafa | 385 | static int |
0d9f234d | 386 | skip_whitespace (pfile, c) |
041c3194 | 387 | cpp_reader *pfile; |
0d9f234d | 388 | cppchar_t c; |
041c3194 ZW |
389 | { |
390 | cpp_buffer *buffer = pfile->buffer; | |
0d9f234d | 391 | unsigned int warned = 0; |
45b966db | 392 | |
0d9f234d | 393 | do |
041c3194 | 394 | { |
91fcd158 NB |
395 | /* Horizontal space always OK. */ |
396 | if (c == ' ') | |
0d9f234d | 397 | ; |
91fcd158 | 398 | else if (c == '\t') |
0d9f234d NB |
399 | adjust_column (pfile); |
400 | /* Just \f \v or \0 left. */ | |
91fcd158 | 401 | else if (c == '\0') |
041c3194 | 402 | { |
4d6baafa NB |
403 | if (buffer->cur - 1 == buffer->rlimit) |
404 | return 0; | |
91fcd158 | 405 | if (!warned) |
0d9f234d | 406 | { |
ebef4e8c | 407 | cpp_error (pfile, DL_WARNING, "null character(s) ignored"); |
0d9f234d NB |
408 | warned = 1; |
409 | } | |
45b966db | 410 | } |
93c80368 | 411 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
ebef4e8c NB |
412 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line, |
413 | CPP_BUF_COL (buffer), | |
414 | "%s in preprocessing directive", | |
415 | c == '\f' ? "form feed" : "vertical tab"); | |
0d9f234d | 416 | |
0d9f234d | 417 | c = *buffer->cur++; |
45b966db | 418 | } |
ec5c56db | 419 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
0d9f234d NB |
420 | while (is_nvspace (c)); |
421 | ||
480709cc | 422 | buffer->cur--; |
4d6baafa | 423 | return 1; |
041c3194 | 424 | } |
45b966db | 425 | |
93c80368 NB |
426 | /* See if the characters of a number token are valid in a name (no |
427 | '.', '+' or '-'). */ | |
428 | static int | |
429 | name_p (pfile, string) | |
430 | cpp_reader *pfile; | |
431 | const cpp_string *string; | |
432 | { | |
433 | unsigned int i; | |
434 | ||
435 | for (i = 0; i < string->len; i++) | |
436 | if (!is_idchar (string->text[i])) | |
437 | return 0; | |
438 | ||
df383483 | 439 | return 1; |
93c80368 NB |
440 | } |
441 | ||
2c3fcba6 ZW |
442 | /* Parse an identifier, skipping embedded backslash-newlines. This is |
443 | a critical inner loop. The common case is an identifier which has | |
444 | not been split by backslash-newline, does not contain a dollar | |
445 | sign, and has already been scanned (roughly 10:1 ratio of | |
446 | seen:unseen identifiers in normal code; the distribution is | |
447 | Poisson-like). Second most common case is a new identifier, not | |
448 | split and no dollar sign. The other possibilities are rare and | |
10cf9bde | 449 | have been relegated to parse_slow. */ |
0d9f234d | 450 | static cpp_hashnode * |
2c3fcba6 | 451 | parse_identifier (pfile) |
45b966db | 452 | cpp_reader *pfile; |
45b966db | 453 | { |
93c80368 | 454 | cpp_hashnode *result; |
562a5c27 | 455 | const uchar *cur, *base; |
2c3fcba6 ZW |
456 | |
457 | /* Fast-path loop. Skim over a normal identifier. | |
458 | N.B. ISIDNUM does not include $. */ | |
4d6baafa NB |
459 | cur = pfile->buffer->cur; |
460 | while (ISIDNUM (*cur)) | |
2c3fcba6 | 461 | cur++; |
2c3fcba6 ZW |
462 | |
463 | /* Check for slow-path cases. */ | |
4d6baafa | 464 | if (*cur == '?' || *cur == '\\' || *cur == '$') |
10cf9bde NB |
465 | { |
466 | unsigned int len; | |
467 | ||
468 | base = parse_slow (pfile, cur, 0, &len); | |
469 | result = (cpp_hashnode *) | |
470 | ht_lookup (pfile->hash_table, base, len, HT_ALLOCED); | |
471 | } | |
2c3fcba6 ZW |
472 | else |
473 | { | |
10cf9bde NB |
474 | base = pfile->buffer->cur - 1; |
475 | pfile->buffer->cur = cur; | |
2c3fcba6 ZW |
476 | result = (cpp_hashnode *) |
477 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
2c3fcba6 ZW |
478 | } |
479 | ||
480 | /* Rarely, identifiers require diagnostics when lexed. | |
481 | XXX Has to be forced out of the fast path. */ | |
482 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) | |
483 | && !pfile->state.skipping, 0)) | |
484 | { | |
485 | /* It is allowed to poison the same identifier twice. */ | |
486 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
ebef4e8c | 487 | cpp_error (pfile, DL_ERROR, "attempt to use poisoned \"%s\"", |
2c3fcba6 ZW |
488 | NODE_NAME (result)); |
489 | ||
490 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
491 | replacement list of a variadic macro. */ | |
492 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
493 | && !pfile->state.va_args_ok) | |
ebef4e8c | 494 | cpp_error (pfile, DL_PEDWARN, |
2c3fcba6 ZW |
495 | "__VA_ARGS__ can only appear in the expansion of a C99 variadic macro"); |
496 | } | |
497 | ||
498 | return result; | |
499 | } | |
500 | ||
10cf9bde NB |
501 | /* Slow path. This handles numbers and identifiers which have been |
502 | split, or contain dollar signs. The part of the token from | |
503 | PFILE->buffer->cur-1 to CUR has already been scanned. NUMBER_P is | |
504 | 1 if it's a number, and 2 if it has a leading period. Returns a | |
505 | pointer to the token's NUL-terminated spelling in permanent | |
506 | storage, and sets PLEN to its length. */ | |
562a5c27 | 507 | static uchar * |
10cf9bde | 508 | parse_slow (pfile, cur, number_p, plen) |
2c3fcba6 | 509 | cpp_reader *pfile; |
562a5c27 | 510 | const uchar *cur; |
10cf9bde NB |
511 | int number_p; |
512 | unsigned int *plen; | |
2c3fcba6 | 513 | { |
0d9f234d | 514 | cpp_buffer *buffer = pfile->buffer; |
562a5c27 | 515 | const uchar *base = buffer->cur - 1; |
2a967f3d | 516 | struct obstack *stack = &pfile->hash_table->stack; |
10cf9bde NB |
517 | unsigned int c, prevc, saw_dollar = 0; |
518 | ||
519 | /* Place any leading period. */ | |
520 | if (number_p == 2) | |
521 | obstack_1grow (stack, '.'); | |
2c3fcba6 ZW |
522 | |
523 | /* Copy the part of the token which is known to be okay. */ | |
524 | obstack_grow (stack, base, cur - base); | |
041c3194 | 525 | |
2c3fcba6 ZW |
526 | /* Now process the part which isn't. We are looking at one of |
527 | '$', '\\', or '?' on entry to this loop. */ | |
10cf9bde | 528 | prevc = cur[-1]; |
2c3fcba6 ZW |
529 | c = *cur++; |
530 | buffer->cur = cur; | |
10cf9bde | 531 | for (;;) |
041c3194 | 532 | { |
10cf9bde NB |
533 | /* Potential escaped newline? */ |
534 | buffer->backup_to = buffer->cur - 1; | |
535 | if (c == '?' || c == '\\') | |
536 | c = skip_escaped_newlines (pfile); | |
537 | ||
538 | if (!is_idchar (c)) | |
539 | { | |
540 | if (!number_p) | |
541 | break; | |
542 | if (c != '.' && !VALID_SIGN (c, prevc)) | |
543 | break; | |
544 | } | |
545 | ||
546 | /* Handle normal identifier characters in this loop. */ | |
547 | do | |
df383483 | 548 | { |
10cf9bde | 549 | prevc = c; |
df383483 | 550 | obstack_1grow (stack, c); |
45b966db | 551 | |
df383483 KH |
552 | if (c == '$') |
553 | saw_dollar++; | |
ba89d661 | 554 | |
df383483 KH |
555 | c = *buffer->cur++; |
556 | } | |
10cf9bde | 557 | while (is_idchar (c)); |
041c3194 | 558 | } |
0d9f234d | 559 | |
4d6baafa | 560 | /* Step back over the unwanted char. */ |
480709cc | 561 | BACKUP (); |
93c80368 | 562 | |
4fe9b91c | 563 | /* $ is not an identifier character in the standard, but is commonly |
0d9f234d NB |
564 | accepted as an extension. Don't warn about it in skipped |
565 | conditional blocks. */ | |
cef0d199 | 566 | if (saw_dollar && CPP_PEDANTIC (pfile) && ! pfile->state.skipping) |
ebef4e8c | 567 | cpp_error (pfile, DL_PEDWARN, "'$' character(s) in identifier or number"); |
0d9f234d | 568 | |
10cf9bde NB |
569 | /* Identifiers and numbers are null-terminated. */ |
570 | *plen = obstack_object_size (stack); | |
2a967f3d | 571 | obstack_1grow (stack, '\0'); |
10cf9bde | 572 | return obstack_finish (stack); |
45b966db ZW |
573 | } |
574 | ||
5d8ebbd8 | 575 | /* Parse a number, beginning with character C, skipping embedded |
da7d8304 | 576 | backslash-newlines. LEADING_PERIOD is nonzero if there was a "." |
5d8ebbd8 | 577 | before C. Place the result in NUMBER. */ |
45b966db | 578 | static void |
10cf9bde | 579 | parse_number (pfile, number, leading_period) |
45b966db | 580 | cpp_reader *pfile; |
0d9f234d | 581 | cpp_string *number; |
93c80368 | 582 | int leading_period; |
45b966db | 583 | { |
562a5c27 | 584 | const uchar *cur; |
45b966db | 585 | |
10cf9bde NB |
586 | /* Fast-path loop. Skim over a normal number. |
587 | N.B. ISIDNUM does not include $. */ | |
588 | cur = pfile->buffer->cur; | |
589 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
590 | cur++; | |
cbcff6df | 591 | |
10cf9bde NB |
592 | /* Check for slow-path cases. */ |
593 | if (*cur == '?' || *cur == '\\' || *cur == '$') | |
594 | number->text = parse_slow (pfile, cur, 1 + leading_period, &number->len); | |
595 | else | |
041c3194 | 596 | { |
562a5c27 NB |
597 | const uchar *base = pfile->buffer->cur - 1; |
598 | uchar *dest; | |
0d9f234d | 599 | |
10cf9bde NB |
600 | number->len = cur - base + leading_period; |
601 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
602 | dest[number->len] = '\0'; | |
603 | number->text = dest; | |
45b966db | 604 | |
10cf9bde NB |
605 | if (leading_period) |
606 | *dest++ = '.'; | |
607 | memcpy (dest, base, cur - base); | |
608 | pfile->buffer->cur = cur; | |
45b966db | 609 | } |
0d9f234d NB |
610 | } |
611 | ||
93c80368 NB |
612 | /* Subroutine of parse_string. */ |
613 | static int | |
614 | unescaped_terminator_p (pfile, dest) | |
615 | cpp_reader *pfile; | |
616 | const unsigned char *dest; | |
617 | { | |
618 | const unsigned char *start, *temp; | |
619 | ||
95bd1dd7 | 620 | /* In #include-style directives, terminators are not escapable. */ |
93c80368 NB |
621 | if (pfile->state.angled_headers) |
622 | return 1; | |
623 | ||
ece54d54 | 624 | start = BUFF_FRONT (pfile->u_buff); |
93c80368 NB |
625 | |
626 | /* An odd number of consecutive backslashes represents an escaped | |
627 | terminator. */ | |
628 | for (temp = dest; temp > start && temp[-1] == '\\'; temp--) | |
629 | ; | |
630 | ||
631 | return ((dest - temp) & 1) == 0; | |
632 | } | |
633 | ||
0d9f234d | 634 | /* Parses a string, character constant, or angle-bracketed header file |
7868b4a2 NB |
635 | name. Handles embedded trigraphs and escaped newlines. The stored |
636 | string is guaranteed NUL-terminated, but it is not guaranteed that | |
637 | this is the first NUL since embedded NULs are preserved. | |
45b966db | 638 | |
87062813 NB |
639 | When this function returns, buffer->cur points to the next |
640 | character to be processed. */ | |
041c3194 | 641 | static void |
0d9f234d | 642 | parse_string (pfile, token, terminator) |
45b966db | 643 | cpp_reader *pfile; |
041c3194 | 644 | cpp_token *token; |
0d9f234d | 645 | cppchar_t terminator; |
45b966db | 646 | { |
041c3194 | 647 | cpp_buffer *buffer = pfile->buffer; |
93c80368 | 648 | unsigned char *dest, *limit; |
0d9f234d | 649 | cppchar_t c; |
d4e6133f | 650 | bool warned_nulls = false; |
64cdc383 MH |
651 | #ifdef MULTIBYTE_CHARS |
652 | wchar_t wc; | |
653 | int char_len; | |
654 | #endif | |
0d9f234d | 655 | |
ece54d54 NB |
656 | dest = BUFF_FRONT (pfile->u_buff); |
657 | limit = BUFF_LIMIT (pfile->u_buff); | |
93c80368 | 658 | |
64cdc383 MH |
659 | #ifdef MULTIBYTE_CHARS |
660 | /* Reset multibyte conversion state. */ | |
661 | (void) local_mbtowc (NULL, NULL, 0); | |
662 | #endif | |
0d9f234d | 663 | for (;;) |
45b966db | 664 | { |
87062813 | 665 | /* We need room for another char, possibly the terminating NUL. */ |
ece54d54 NB |
666 | if ((size_t) (limit - dest) < 1) |
667 | { | |
668 | size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff); | |
8c3b2693 | 669 | _cpp_extend_buff (pfile, &pfile->u_buff, 2); |
ece54d54 NB |
670 | dest = BUFF_FRONT (pfile->u_buff) + len_so_far; |
671 | limit = BUFF_LIMIT (pfile->u_buff); | |
672 | } | |
7868b4a2 | 673 | |
64cdc383 MH |
674 | #ifdef MULTIBYTE_CHARS |
675 | char_len = local_mbtowc (&wc, (const char *) buffer->cur, | |
676 | buffer->rlimit - buffer->cur); | |
677 | if (char_len == -1) | |
678 | { | |
679 | cpp_error (pfile, DL_WARNING, | |
df383483 | 680 | "ignoring invalid multibyte character"); |
64cdc383 MH |
681 | char_len = 1; |
682 | c = *buffer->cur++; | |
683 | } | |
684 | else | |
685 | { | |
686 | buffer->cur += char_len; | |
687 | c = wc; | |
688 | } | |
689 | #else | |
87062813 | 690 | c = *buffer->cur++; |
64cdc383 MH |
691 | #endif |
692 | ||
693 | /* Handle trigraphs, escaped newlines etc. */ | |
0d9f234d | 694 | if (c == '?' || c == '\\') |
87062813 | 695 | c = skip_escaped_newlines (pfile); |
45b966db | 696 | |
87062813 | 697 | if (c == terminator) |
45b966db | 698 | { |
87062813 NB |
699 | if (unescaped_terminator_p (pfile, dest)) |
700 | break; | |
0d9f234d NB |
701 | } |
702 | else if (is_vspace (c)) | |
703 | { | |
d4e6133f NB |
704 | /* No string literal may extend over multiple lines. In |
705 | assembly language, suppress the error except for <> | |
706 | includes. This is a kludge around not knowing where | |
707 | comments are. */ | |
708 | unterminated: | |
709 | if (CPP_OPTION (pfile, lang) != CLK_ASM || terminator == '>') | |
ebef4e8c | 710 | cpp_error (pfile, DL_ERROR, "missing terminating %c character", |
625458d0 | 711 | (int) terminator); |
d4e6133f NB |
712 | buffer->cur--; |
713 | break; | |
0d9f234d | 714 | } |
4d6baafa | 715 | else if (c == '\0') |
0d9f234d | 716 | { |
4d6baafa | 717 | if (buffer->cur - 1 == buffer->rlimit) |
d4e6133f | 718 | goto unterminated; |
4d6baafa NB |
719 | if (!warned_nulls) |
720 | { | |
721 | warned_nulls = true; | |
ebef4e8c NB |
722 | cpp_error (pfile, DL_WARNING, |
723 | "null character(s) preserved in literal"); | |
4d6baafa | 724 | } |
45b966db | 725 | } |
64cdc383 MH |
726 | #ifdef MULTIBYTE_CHARS |
727 | if (char_len > 1) | |
728 | { | |
729 | for ( ; char_len > 0; --char_len) | |
730 | *dest++ = (*buffer->cur - char_len); | |
731 | } | |
732 | else | |
733 | #endif | |
734 | *dest++ = c; | |
45b966db ZW |
735 | } |
736 | ||
7868b4a2 | 737 | *dest = '\0'; |
45b966db | 738 | |
ece54d54 NB |
739 | token->val.str.text = BUFF_FRONT (pfile->u_buff); |
740 | token->val.str.len = dest - BUFF_FRONT (pfile->u_buff); | |
741 | BUFF_FRONT (pfile->u_buff) = dest + 1; | |
0d9f234d | 742 | } |
041c3194 | 743 | |
93c80368 | 744 | /* The stored comment includes the comment start and any terminator. */ |
9e62c811 | 745 | static void |
477cdac7 | 746 | save_comment (pfile, token, from, type) |
0d9f234d | 747 | cpp_reader *pfile; |
041c3194 ZW |
748 | cpp_token *token; |
749 | const unsigned char *from; | |
477cdac7 | 750 | cppchar_t type; |
9e62c811 | 751 | { |
041c3194 | 752 | unsigned char *buffer; |
477cdac7 | 753 | unsigned int len, clen; |
df383483 | 754 | |
1c6d33ef | 755 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
480709cc | 756 | |
3542203b NB |
757 | /* C++ comments probably (not definitely) have moved past a new |
758 | line, which we don't want to save in the comment. */ | |
480709cc | 759 | if (is_vspace (pfile->buffer->cur[-1])) |
3542203b | 760 | len--; |
477cdac7 JT |
761 | |
762 | /* If we are currently in a directive, then we need to store all | |
763 | C++ comments as C comments internally, and so we need to | |
764 | allocate a little extra space in that case. | |
765 | ||
766 | Note that the only time we encounter a directive here is | |
767 | when we are saving comments in a "#define". */ | |
768 | clen = (pfile->state.in_directive && type == '/') ? len + 2 : len; | |
769 | ||
770 | buffer = _cpp_unaligned_alloc (pfile, clen); | |
df383483 | 771 | |
041c3194 | 772 | token->type = CPP_COMMENT; |
477cdac7 | 773 | token->val.str.len = clen; |
0d9f234d | 774 | token->val.str.text = buffer; |
45b966db | 775 | |
1c6d33ef NB |
776 | buffer[0] = '/'; |
777 | memcpy (buffer + 1, from, len - 1); | |
477cdac7 | 778 | |
1eeeb6a4 | 779 | /* Finish conversion to a C comment, if necessary. */ |
477cdac7 JT |
780 | if (pfile->state.in_directive && type == '/') |
781 | { | |
782 | buffer[1] = '*'; | |
783 | buffer[clen - 2] = '*'; | |
784 | buffer[clen - 1] = '/'; | |
785 | } | |
0d9f234d | 786 | } |
45b966db | 787 | |
5fddcffc NB |
788 | /* Allocate COUNT tokens for RUN. */ |
789 | void | |
790 | _cpp_init_tokenrun (run, count) | |
791 | tokenrun *run; | |
792 | unsigned int count; | |
793 | { | |
794 | run->base = xnewvec (cpp_token, count); | |
795 | run->limit = run->base + count; | |
796 | run->next = NULL; | |
797 | } | |
798 | ||
799 | /* Returns the next tokenrun, or creates one if there is none. */ | |
800 | static tokenrun * | |
801 | next_tokenrun (run) | |
802 | tokenrun *run; | |
803 | { | |
804 | if (run->next == NULL) | |
805 | { | |
806 | run->next = xnew (tokenrun); | |
bdcbe496 | 807 | run->next->prev = run; |
5fddcffc NB |
808 | _cpp_init_tokenrun (run->next, 250); |
809 | } | |
810 | ||
811 | return run->next; | |
812 | } | |
813 | ||
4ed5bcfb NB |
814 | /* Allocate a single token that is invalidated at the same time as the |
815 | rest of the tokens on the line. Has its line and col set to the | |
816 | same as the last lexed token, so that diagnostics appear in the | |
817 | right place. */ | |
818 | cpp_token * | |
819 | _cpp_temp_token (pfile) | |
820 | cpp_reader *pfile; | |
821 | { | |
822 | cpp_token *old, *result; | |
823 | ||
824 | old = pfile->cur_token - 1; | |
825 | if (pfile->cur_token == pfile->cur_run->limit) | |
826 | { | |
827 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
828 | pfile->cur_token = pfile->cur_run->base; | |
829 | } | |
830 | ||
831 | result = pfile->cur_token++; | |
832 | result->line = old->line; | |
833 | result->col = old->col; | |
834 | return result; | |
835 | } | |
836 | ||
14baae01 NB |
837 | /* Lex a token into RESULT (external interface). Takes care of issues |
838 | like directive handling, token lookahead, multiple include | |
a1f300c0 | 839 | optimization and skipping. */ |
345894b4 NB |
840 | const cpp_token * |
841 | _cpp_lex_token (pfile) | |
45b966db | 842 | cpp_reader *pfile; |
5fddcffc | 843 | { |
bdcbe496 | 844 | cpp_token *result; |
5fddcffc | 845 | |
bdcbe496 | 846 | for (;;) |
5fddcffc | 847 | { |
bdcbe496 | 848 | if (pfile->cur_token == pfile->cur_run->limit) |
5fddcffc | 849 | { |
bdcbe496 NB |
850 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
851 | pfile->cur_token = pfile->cur_run->base; | |
5fddcffc NB |
852 | } |
853 | ||
bdcbe496 | 854 | if (pfile->lookaheads) |
14baae01 NB |
855 | { |
856 | pfile->lookaheads--; | |
857 | result = pfile->cur_token++; | |
858 | } | |
bdcbe496 | 859 | else |
14baae01 | 860 | result = _cpp_lex_direct (pfile); |
bdcbe496 NB |
861 | |
862 | if (result->flags & BOL) | |
5fddcffc | 863 | { |
bdcbe496 NB |
864 | /* Is this a directive. If _cpp_handle_directive returns |
865 | false, it is an assembler #. */ | |
866 | if (result->type == CPP_HASH | |
e808ec9c NB |
867 | /* 6.10.3 p 11: Directives in a list of macro arguments |
868 | gives undefined behavior. This implementation | |
869 | handles the directive as normal. */ | |
870 | && pfile->state.parsing_args != 1 | |
bdcbe496 NB |
871 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
872 | continue; | |
97293897 NB |
873 | if (pfile->cb.line_change && !pfile->state.skipping) |
874 | (*pfile->cb.line_change)(pfile, result, pfile->state.parsing_args); | |
5fddcffc | 875 | } |
5fddcffc | 876 | |
bdcbe496 NB |
877 | /* We don't skip tokens in directives. */ |
878 | if (pfile->state.in_directive) | |
879 | break; | |
5fddcffc | 880 | |
bdcbe496 | 881 | /* Outside a directive, invalidate controlling macros. At file |
14baae01 | 882 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
bdcbe496 | 883 | get here and MI optimisation works. */ |
5fddcffc | 884 | pfile->mi_valid = false; |
bdcbe496 NB |
885 | |
886 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
887 | break; | |
5fddcffc NB |
888 | } |
889 | ||
345894b4 | 890 | return result; |
5fddcffc NB |
891 | } |
892 | ||
004cb263 NB |
893 | /* A NUL terminates the current buffer. For ISO preprocessing this is |
894 | EOF, but for traditional preprocessing it indicates we need a line | |
895 | refill. Returns TRUE to continue preprocessing a new buffer, FALSE | |
896 | to return a CPP_EOF to the caller. */ | |
897 | static bool | |
898 | continue_after_nul (pfile) | |
899 | cpp_reader *pfile; | |
900 | { | |
901 | cpp_buffer *buffer = pfile->buffer; | |
902 | bool more = false; | |
df383483 | 903 | |
004cb263 NB |
904 | buffer->saved_flags = BOL; |
905 | if (CPP_OPTION (pfile, traditional)) | |
1a76916c NB |
906 | { |
907 | if (pfile->state.in_directive) | |
908 | return false; | |
909 | ||
910 | _cpp_remove_overlay (pfile); | |
911 | more = _cpp_read_logical_line_trad (pfile); | |
912 | _cpp_overlay_buffer (pfile, pfile->out.base, | |
913 | pfile->out.cur - pfile->out.base); | |
914 | pfile->line = pfile->out.first_line; | |
915 | } | |
004cb263 NB |
916 | else |
917 | { | |
918 | /* Stop parsing arguments with a CPP_EOF. When we finally come | |
919 | back here, do the work of popping the buffer. */ | |
920 | if (!pfile->state.parsing_args) | |
921 | { | |
922 | if (buffer->cur != buffer->line_base) | |
923 | { | |
924 | /* Non-empty files should end in a newline. Don't warn | |
925 | for command line and _Pragma buffers. */ | |
926 | if (!buffer->from_stage3) | |
927 | cpp_error (pfile, DL_PEDWARN, "no newline at end of file"); | |
928 | handle_newline (pfile); | |
929 | } | |
930 | ||
931 | /* Similarly, finish an in-progress directive with CPP_EOF | |
932 | before popping the buffer. */ | |
933 | if (!pfile->state.in_directive && buffer->prev) | |
934 | { | |
935 | more = !buffer->return_at_eof; | |
936 | _cpp_pop_buffer (pfile); | |
937 | } | |
938 | } | |
939 | } | |
940 | ||
941 | return more; | |
942 | } | |
943 | ||
480709cc NB |
944 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
945 | do { \ | |
946 | if (get_effective_char (pfile) == CHAR) \ | |
947 | result->type = THEN_TYPE; \ | |
948 | else \ | |
949 | { \ | |
950 | BACKUP (); \ | |
951 | result->type = ELSE_TYPE; \ | |
952 | } \ | |
953 | } while (0) | |
954 | ||
14baae01 NB |
955 | /* Lex a token into pfile->cur_token, which is also incremented, to |
956 | get diagnostics pointing to the correct location. | |
957 | ||
958 | Does not handle issues such as token lookahead, multiple-include | |
959 | optimisation, directives, skipping etc. This function is only | |
960 | suitable for use by _cpp_lex_token, and in special cases like | |
961 | lex_expansion_token which doesn't care for any of these issues. | |
962 | ||
963 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
964 | otherwise returns to the start of the token buffer if permissible. | |
965 | Returns the location of the lexed token. */ | |
966 | cpp_token * | |
967 | _cpp_lex_direct (pfile) | |
5fddcffc | 968 | cpp_reader *pfile; |
45b966db | 969 | { |
0d9f234d | 970 | cppchar_t c; |
adb84b42 | 971 | cpp_buffer *buffer; |
0d9f234d | 972 | const unsigned char *comment_start; |
14baae01 | 973 | cpp_token *result = pfile->cur_token++; |
9ec7291f | 974 | |
5fddcffc | 975 | fresh_line: |
adb84b42 | 976 | buffer = pfile->buffer; |
bd969772 NB |
977 | result->flags = buffer->saved_flags; |
978 | buffer->saved_flags = 0; | |
5fddcffc | 979 | update_tokens_line: |
1444f2ed | 980 | result->line = pfile->line; |
041c3194 | 981 | |
5fddcffc | 982 | skipped_white: |
480709cc | 983 | c = *buffer->cur++; |
5fddcffc | 984 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
5fddcffc NB |
985 | |
986 | trigraph: | |
0d9f234d | 987 | switch (c) |
45b966db | 988 | { |
4d6baafa NB |
989 | case ' ': case '\t': case '\f': case '\v': case '\0': |
990 | result->flags |= PREV_WHITE; | |
991 | if (skip_whitespace (pfile, c)) | |
992 | goto skipped_white; | |
993 | ||
004cb263 | 994 | /* End of buffer. */ |
4d6baafa | 995 | buffer->cur--; |
004cb263 NB |
996 | if (continue_after_nul (pfile)) |
997 | goto fresh_line; | |
0d9f234d | 998 | result->type = CPP_EOF; |
5fddcffc | 999 | break; |
45b966db | 1000 | |
0d9f234d | 1001 | case '\n': case '\r': |
87062813 | 1002 | handle_newline (pfile); |
bdcbe496 NB |
1003 | buffer->saved_flags = BOL; |
1004 | if (! pfile->state.in_directive) | |
45b966db | 1005 | { |
4ed5bcfb NB |
1006 | if (pfile->state.parsing_args == 2) |
1007 | buffer->saved_flags |= PREV_WHITE; | |
bdcbe496 NB |
1008 | if (!pfile->keep_tokens) |
1009 | { | |
1010 | pfile->cur_run = &pfile->base_run; | |
1011 | result = pfile->base_run.base; | |
1012 | pfile->cur_token = result + 1; | |
1013 | } | |
1014 | goto fresh_line; | |
45b966db | 1015 | } |
5fddcffc NB |
1016 | result->type = CPP_EOF; |
1017 | break; | |
46d07497 | 1018 | |
0d9f234d NB |
1019 | case '?': |
1020 | case '\\': | |
1021 | /* These could start an escaped newline, or '?' a trigraph. Let | |
1022 | skip_escaped_newlines do all the work. */ | |
1023 | { | |
67821e3a | 1024 | unsigned int line = pfile->line; |
0d9f234d | 1025 | |
87062813 | 1026 | c = skip_escaped_newlines (pfile); |
67821e3a | 1027 | if (line != pfile->line) |
87062813 | 1028 | { |
480709cc | 1029 | buffer->cur--; |
87062813 NB |
1030 | /* We had at least one escaped newline of some sort. |
1031 | Update the token's line and column. */ | |
5fddcffc | 1032 | goto update_tokens_line; |
87062813 | 1033 | } |
480709cc | 1034 | } |
0d9f234d | 1035 | |
480709cc NB |
1036 | /* We are either the original '?' or '\\', or a trigraph. */ |
1037 | if (c == '?') | |
0d9f234d | 1038 | result->type = CPP_QUERY; |
480709cc NB |
1039 | else if (c == '\\') |
1040 | goto random_char; | |
1041 | else | |
1042 | goto trigraph; | |
0d9f234d | 1043 | break; |
46d07497 | 1044 | |
0d9f234d NB |
1045 | case '0': case '1': case '2': case '3': case '4': |
1046 | case '5': case '6': case '7': case '8': case '9': | |
1047 | result->type = CPP_NUMBER; | |
10cf9bde | 1048 | parse_number (pfile, &result->val.str, 0); |
0d9f234d | 1049 | break; |
46d07497 | 1050 | |
0abc6a6a NB |
1051 | case 'L': |
1052 | /* 'L' may introduce wide characters or strings. */ | |
df383483 KH |
1053 | { |
1054 | const unsigned char *pos = buffer->cur; | |
0d9f234d | 1055 | |
df383483 KH |
1056 | c = get_effective_char (pfile); |
1057 | if (c == '\'' || c == '"') | |
1058 | { | |
1059 | result->type = (c == '"' ? CPP_WSTRING: CPP_WCHAR); | |
1060 | parse_string (pfile, result, c); | |
1061 | break; | |
1062 | } | |
1063 | buffer->cur = pos; | |
1064 | } | |
1065 | /* Fall through. */ | |
0abc6a6a NB |
1066 | |
1067 | start_ident: | |
0d9f234d NB |
1068 | case '_': |
1069 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
1070 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
1071 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
1072 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
1073 | case 'y': case 'z': | |
1074 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
0abc6a6a | 1075 | case 'G': case 'H': case 'I': case 'J': case 'K': |
0d9f234d NB |
1076 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
1077 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1078 | case 'Y': case 'Z': | |
1079 | result->type = CPP_NAME; | |
2c3fcba6 | 1080 | result->val.node = parse_identifier (pfile); |
0d9f234d | 1081 | |
0d9f234d | 1082 | /* Convert named operators to their proper types. */ |
0abc6a6a | 1083 | if (result->val.node->flags & NODE_OPERATOR) |
0d9f234d NB |
1084 | { |
1085 | result->flags |= NAMED_OP; | |
4977bab6 | 1086 | result->type = result->val.node->directive_index; |
0d9f234d NB |
1087 | } |
1088 | break; | |
1089 | ||
1090 | case '\'': | |
1091 | case '"': | |
1092 | result->type = c == '"' ? CPP_STRING: CPP_CHAR; | |
0d9f234d NB |
1093 | parse_string (pfile, result, c); |
1094 | break; | |
041c3194 | 1095 | |
0d9f234d | 1096 | case '/': |
1c6d33ef NB |
1097 | /* A potential block or line comment. */ |
1098 | comment_start = buffer->cur; | |
29401c30 | 1099 | c = get_effective_char (pfile); |
480709cc | 1100 | |
1c6d33ef NB |
1101 | if (c == '*') |
1102 | { | |
0d9f234d | 1103 | if (skip_block_comment (pfile)) |
ebef4e8c | 1104 | cpp_error (pfile, DL_ERROR, "unterminated comment"); |
0d9f234d | 1105 | } |
480709cc NB |
1106 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
1107 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
0d9f234d | 1108 | { |
bdb05a7b NB |
1109 | /* Warn about comments only if pedantically GNUC89, and not |
1110 | in system headers. */ | |
1111 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
a94c1199 | 1112 | && ! buffer->warned_cplusplus_comments) |
041c3194 | 1113 | { |
ebef4e8c | 1114 | cpp_error (pfile, DL_PEDWARN, |
56508306 | 1115 | "C++ style comments are not allowed in ISO C90"); |
ebef4e8c NB |
1116 | cpp_error (pfile, DL_PEDWARN, |
1117 | "(this will be reported only once per input file)"); | |
1c6d33ef NB |
1118 | buffer->warned_cplusplus_comments = 1; |
1119 | } | |
0d9f234d | 1120 | |
01ef6563 | 1121 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
ebef4e8c | 1122 | cpp_error (pfile, DL_WARNING, "multi-line comment"); |
1c6d33ef | 1123 | } |
480709cc NB |
1124 | else if (c == '=') |
1125 | { | |
1126 | result->type = CPP_DIV_EQ; | |
1127 | break; | |
1128 | } | |
1129 | else | |
1130 | { | |
1131 | BACKUP (); | |
1132 | result->type = CPP_DIV; | |
1133 | break; | |
1134 | } | |
0d9f234d | 1135 | |
1c6d33ef NB |
1136 | if (!pfile->state.save_comments) |
1137 | { | |
1138 | result->flags |= PREV_WHITE; | |
5fddcffc | 1139 | goto update_tokens_line; |
0d9f234d | 1140 | } |
1c6d33ef NB |
1141 | |
1142 | /* Save the comment as a token in its own right. */ | |
477cdac7 | 1143 | save_comment (pfile, result, comment_start, c); |
bdcbe496 | 1144 | break; |
0d9f234d NB |
1145 | |
1146 | case '<': | |
1147 | if (pfile->state.angled_headers) | |
1148 | { | |
1149 | result->type = CPP_HEADER_NAME; | |
480709cc NB |
1150 | parse_string (pfile, result, '>'); |
1151 | break; | |
0d9f234d | 1152 | } |
45b966db | 1153 | |
29401c30 | 1154 | c = get_effective_char (pfile); |
0d9f234d | 1155 | if (c == '=') |
480709cc | 1156 | result->type = CPP_LESS_EQ; |
0d9f234d | 1157 | else if (c == '<') |
480709cc | 1158 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); |
0d9f234d | 1159 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1160 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); |
0d9f234d NB |
1161 | else if (c == ':' && CPP_OPTION (pfile, digraphs)) |
1162 | { | |
480709cc | 1163 | result->type = CPP_OPEN_SQUARE; |
0d9f234d NB |
1164 | result->flags |= DIGRAPH; |
1165 | } | |
1166 | else if (c == '%' && CPP_OPTION (pfile, digraphs)) | |
1167 | { | |
480709cc | 1168 | result->type = CPP_OPEN_BRACE; |
0d9f234d NB |
1169 | result->flags |= DIGRAPH; |
1170 | } | |
480709cc NB |
1171 | else |
1172 | { | |
1173 | BACKUP (); | |
1174 | result->type = CPP_LESS; | |
1175 | } | |
0d9f234d NB |
1176 | break; |
1177 | ||
1178 | case '>': | |
29401c30 | 1179 | c = get_effective_char (pfile); |
0d9f234d | 1180 | if (c == '=') |
480709cc | 1181 | result->type = CPP_GREATER_EQ; |
0d9f234d | 1182 | else if (c == '>') |
480709cc | 1183 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); |
0d9f234d | 1184 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc NB |
1185 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); |
1186 | else | |
0d9f234d | 1187 | { |
480709cc NB |
1188 | BACKUP (); |
1189 | result->type = CPP_GREATER; | |
0d9f234d NB |
1190 | } |
1191 | break; | |
1192 | ||
cbcff6df | 1193 | case '%': |
480709cc NB |
1194 | c = get_effective_char (pfile); |
1195 | if (c == '=') | |
1196 | result->type = CPP_MOD_EQ; | |
1197 | else if (CPP_OPTION (pfile, digraphs) && c == ':') | |
1198 | { | |
1199 | result->flags |= DIGRAPH; | |
1200 | result->type = CPP_HASH; | |
1201 | if (get_effective_char (pfile) == '%') | |
1202 | { | |
1203 | const unsigned char *pos = buffer->cur; | |
1204 | ||
1205 | if (get_effective_char (pfile) == ':') | |
1206 | result->type = CPP_PASTE; | |
1207 | else | |
1208 | buffer->cur = pos - 1; | |
1209 | } | |
1210 | else | |
1211 | BACKUP (); | |
1212 | } | |
1213 | else if (CPP_OPTION (pfile, digraphs) && c == '>') | |
1214 | { | |
1215 | result->flags |= DIGRAPH; | |
1216 | result->type = CPP_CLOSE_BRACE; | |
1217 | } | |
1218 | else | |
1219 | { | |
1220 | BACKUP (); | |
1221 | result->type = CPP_MOD; | |
1222 | } | |
0d9f234d NB |
1223 | break; |
1224 | ||
cbcff6df | 1225 | case '.': |
480709cc NB |
1226 | result->type = CPP_DOT; |
1227 | c = get_effective_char (pfile); | |
1228 | if (c == '.') | |
1229 | { | |
1230 | const unsigned char *pos = buffer->cur; | |
1231 | ||
1232 | if (get_effective_char (pfile) == '.') | |
1233 | result->type = CPP_ELLIPSIS; | |
1234 | else | |
1235 | buffer->cur = pos - 1; | |
1236 | } | |
1237 | /* All known character sets have 0...9 contiguous. */ | |
0df6c2c7 | 1238 | else if (ISDIGIT (c)) |
480709cc NB |
1239 | { |
1240 | result->type = CPP_NUMBER; | |
10cf9bde | 1241 | parse_number (pfile, &result->val.str, 1); |
480709cc NB |
1242 | } |
1243 | else if (c == '*' && CPP_OPTION (pfile, cplusplus)) | |
1244 | result->type = CPP_DOT_STAR; | |
1245 | else | |
1246 | BACKUP (); | |
0d9f234d | 1247 | break; |
45b966db | 1248 | |
0d9f234d | 1249 | case '+': |
29401c30 | 1250 | c = get_effective_char (pfile); |
480709cc NB |
1251 | if (c == '+') |
1252 | result->type = CPP_PLUS_PLUS; | |
1253 | else if (c == '=') | |
1254 | result->type = CPP_PLUS_EQ; | |
1255 | else | |
1256 | { | |
1257 | BACKUP (); | |
1258 | result->type = CPP_PLUS; | |
1259 | } | |
0d9f234d | 1260 | break; |
04e3ec78 | 1261 | |
0d9f234d | 1262 | case '-': |
29401c30 | 1263 | c = get_effective_char (pfile); |
0d9f234d NB |
1264 | if (c == '>') |
1265 | { | |
480709cc NB |
1266 | result->type = CPP_DEREF; |
1267 | if (CPP_OPTION (pfile, cplusplus)) | |
1268 | { | |
1269 | if (get_effective_char (pfile) == '*') | |
1270 | result->type = CPP_DEREF_STAR; | |
1271 | else | |
1272 | BACKUP (); | |
1273 | } | |
0d9f234d | 1274 | } |
0d9f234d | 1275 | else if (c == '-') |
480709cc NB |
1276 | result->type = CPP_MINUS_MINUS; |
1277 | else if (c == '=') | |
1278 | result->type = CPP_MINUS_EQ; | |
1279 | else | |
1280 | { | |
1281 | BACKUP (); | |
1282 | result->type = CPP_MINUS; | |
1283 | } | |
0d9f234d | 1284 | break; |
45b966db | 1285 | |
0d9f234d | 1286 | case '&': |
29401c30 | 1287 | c = get_effective_char (pfile); |
480709cc NB |
1288 | if (c == '&') |
1289 | result->type = CPP_AND_AND; | |
1290 | else if (c == '=') | |
1291 | result->type = CPP_AND_EQ; | |
1292 | else | |
1293 | { | |
1294 | BACKUP (); | |
1295 | result->type = CPP_AND; | |
1296 | } | |
0d9f234d | 1297 | break; |
df383483 | 1298 | |
0d9f234d | 1299 | case '|': |
29401c30 | 1300 | c = get_effective_char (pfile); |
480709cc NB |
1301 | if (c == '|') |
1302 | result->type = CPP_OR_OR; | |
1303 | else if (c == '=') | |
1304 | result->type = CPP_OR_EQ; | |
1305 | else | |
1306 | { | |
1307 | BACKUP (); | |
1308 | result->type = CPP_OR; | |
1309 | } | |
0d9f234d | 1310 | break; |
45b966db | 1311 | |
0d9f234d | 1312 | case ':': |
29401c30 | 1313 | c = get_effective_char (pfile); |
0d9f234d | 1314 | if (c == ':' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1315 | result->type = CPP_SCOPE; |
0d9f234d NB |
1316 | else if (c == '>' && CPP_OPTION (pfile, digraphs)) |
1317 | { | |
1318 | result->flags |= DIGRAPH; | |
480709cc NB |
1319 | result->type = CPP_CLOSE_SQUARE; |
1320 | } | |
1321 | else | |
1322 | { | |
1323 | BACKUP (); | |
1324 | result->type = CPP_COLON; | |
0d9f234d NB |
1325 | } |
1326 | break; | |
45b966db | 1327 | |
480709cc NB |
1328 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1329 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1330 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1331 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1332 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1333 | ||
0d9f234d NB |
1334 | case '~': result->type = CPP_COMPL; break; |
1335 | case ',': result->type = CPP_COMMA; break; | |
1336 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1337 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1338 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1339 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1340 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1341 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1342 | case ';': result->type = CPP_SEMICOLON; break; | |
1343 | ||
40f03658 | 1344 | /* @ is a punctuator in Objective-C. */ |
cc937581 | 1345 | case '@': result->type = CPP_ATSIGN; break; |
0d9f234d | 1346 | |
0abc6a6a NB |
1347 | case '$': |
1348 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
1349 | goto start_ident; | |
1350 | /* Fall through... */ | |
1351 | ||
0d9f234d NB |
1352 | random_char: |
1353 | default: | |
1354 | result->type = CPP_OTHER; | |
6c53ebff | 1355 | result->val.c = c; |
0d9f234d NB |
1356 | break; |
1357 | } | |
bdcbe496 NB |
1358 | |
1359 | return result; | |
0d9f234d NB |
1360 | } |
1361 | ||
5d8ebbd8 | 1362 | /* An upper bound on the number of bytes needed to spell TOKEN, |
93c80368 NB |
1363 | including preceding whitespace. */ |
1364 | unsigned int | |
1365 | cpp_token_len (token) | |
1366 | const cpp_token *token; | |
0d9f234d | 1367 | { |
93c80368 | 1368 | unsigned int len; |
6d2c2047 | 1369 | |
93c80368 | 1370 | switch (TOKEN_SPELL (token)) |
041c3194 | 1371 | { |
a28c5035 | 1372 | default: len = 0; break; |
47ad4138 | 1373 | case SPELL_NUMBER: |
a28c5035 NB |
1374 | case SPELL_STRING: len = token->val.str.len; break; |
1375 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; | |
041c3194 | 1376 | } |
47ad4138 | 1377 | /* 1 for whitespace, 4 for comment delimiters. */ |
93c80368 | 1378 | return len + 5; |
6d2c2047 ZW |
1379 | } |
1380 | ||
041c3194 | 1381 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
cf00a885 ZW |
1382 | already contain the enough space to hold the token's spelling. |
1383 | Returns a pointer to the character after the last character | |
1384 | written. */ | |
93c80368 NB |
1385 | unsigned char * |
1386 | cpp_spell_token (pfile, token, buffer) | |
041c3194 ZW |
1387 | cpp_reader *pfile; /* Would be nice to be rid of this... */ |
1388 | const cpp_token *token; | |
1389 | unsigned char *buffer; | |
1390 | { | |
96be6998 | 1391 | switch (TOKEN_SPELL (token)) |
041c3194 ZW |
1392 | { |
1393 | case SPELL_OPERATOR: | |
1394 | { | |
1395 | const unsigned char *spelling; | |
1396 | unsigned char c; | |
d6d5f795 | 1397 | |
041c3194 | 1398 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1399 | spelling |
1400 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
92936ecf ZW |
1401 | else if (token->flags & NAMED_OP) |
1402 | goto spell_ident; | |
041c3194 | 1403 | else |
96be6998 | 1404 | spelling = TOKEN_NAME (token); |
df383483 | 1405 | |
041c3194 ZW |
1406 | while ((c = *spelling++) != '\0') |
1407 | *buffer++ = c; | |
1408 | } | |
1409 | break; | |
d6d5f795 | 1410 | |
47ad4138 ZW |
1411 | case SPELL_CHAR: |
1412 | *buffer++ = token->val.c; | |
1413 | break; | |
1414 | ||
1415 | spell_ident: | |
041c3194 | 1416 | case SPELL_IDENT: |
a28c5035 NB |
1417 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1418 | buffer += NODE_LEN (token->val.node); | |
041c3194 | 1419 | break; |
d6d5f795 | 1420 | |
47ad4138 ZW |
1421 | case SPELL_NUMBER: |
1422 | memcpy (buffer, token->val.str.text, token->val.str.len); | |
1423 | buffer += token->val.str.len; | |
1424 | break; | |
1425 | ||
041c3194 ZW |
1426 | case SPELL_STRING: |
1427 | { | |
ba89d661 ZW |
1428 | int left, right, tag; |
1429 | switch (token->type) | |
1430 | { | |
1431 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1432 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
ba89d661 ZW |
1433 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1434 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1435 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 | 1436 | default: |
ebef4e8c NB |
1437 | cpp_error (pfile, DL_ICE, "unknown string token %s\n", |
1438 | TOKEN_NAME (token)); | |
47ad4138 | 1439 | return buffer; |
ba89d661 ZW |
1440 | } |
1441 | if (tag) *buffer++ = tag; | |
47ad4138 | 1442 | *buffer++ = left; |
bfb9dc7f ZW |
1443 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1444 | buffer += token->val.str.len; | |
47ad4138 | 1445 | *buffer++ = right; |
041c3194 ZW |
1446 | } |
1447 | break; | |
d6d5f795 | 1448 | |
041c3194 | 1449 | case SPELL_NONE: |
ebef4e8c | 1450 | cpp_error (pfile, DL_ICE, "unspellable token %s", TOKEN_NAME (token)); |
041c3194 ZW |
1451 | break; |
1452 | } | |
d6d5f795 | 1453 | |
041c3194 ZW |
1454 | return buffer; |
1455 | } | |
d6d5f795 | 1456 | |
5d8ebbd8 NB |
1457 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1458 | freed when the reader is destroyed. Useful for diagnostics. */ | |
93c80368 NB |
1459 | unsigned char * |
1460 | cpp_token_as_text (pfile, token) | |
c5a04734 | 1461 | cpp_reader *pfile; |
041c3194 | 1462 | const cpp_token *token; |
c5a04734 | 1463 | { |
93c80368 | 1464 | unsigned int len = cpp_token_len (token); |
ece54d54 | 1465 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
c5a04734 | 1466 | |
93c80368 NB |
1467 | end = cpp_spell_token (pfile, token, start); |
1468 | end[0] = '\0'; | |
c5a04734 | 1469 | |
93c80368 NB |
1470 | return start; |
1471 | } | |
c5a04734 | 1472 | |
5d8ebbd8 NB |
1473 | /* Used by C front ends, which really should move to using |
1474 | cpp_token_as_text. */ | |
93c80368 NB |
1475 | const char * |
1476 | cpp_type2name (type) | |
1477 | enum cpp_ttype type; | |
1478 | { | |
1479 | return (const char *) token_spellings[type].name; | |
1480 | } | |
c5a04734 | 1481 | |
4ed5bcfb NB |
1482 | /* Writes the spelling of token to FP, without any preceding space. |
1483 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1484 | double-buffering. */ | |
93c80368 NB |
1485 | void |
1486 | cpp_output_token (token, fp) | |
1487 | const cpp_token *token; | |
1488 | FILE *fp; | |
1489 | { | |
93c80368 | 1490 | switch (TOKEN_SPELL (token)) |
c5a04734 | 1491 | { |
93c80368 NB |
1492 | case SPELL_OPERATOR: |
1493 | { | |
1494 | const unsigned char *spelling; | |
3b681e9d | 1495 | int c; |
c5a04734 | 1496 | |
93c80368 | 1497 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1498 | spelling |
1499 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
93c80368 NB |
1500 | else if (token->flags & NAMED_OP) |
1501 | goto spell_ident; | |
1502 | else | |
1503 | spelling = TOKEN_NAME (token); | |
041c3194 | 1504 | |
3b681e9d ZW |
1505 | c = *spelling; |
1506 | do | |
1507 | putc (c, fp); | |
1508 | while ((c = *++spelling) != '\0'); | |
93c80368 NB |
1509 | } |
1510 | break; | |
041c3194 | 1511 | |
47ad4138 ZW |
1512 | case SPELL_CHAR: |
1513 | putc (token->val.c, fp); | |
1514 | break; | |
1515 | ||
93c80368 NB |
1516 | spell_ident: |
1517 | case SPELL_IDENT: | |
3b681e9d | 1518 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
93c80368 | 1519 | break; |
041c3194 | 1520 | |
47ad4138 ZW |
1521 | case SPELL_NUMBER: |
1522 | fwrite (token->val.str.text, 1, token->val.str.len, fp); | |
1523 | break; | |
1524 | ||
93c80368 NB |
1525 | case SPELL_STRING: |
1526 | { | |
1527 | int left, right, tag; | |
1528 | switch (token->type) | |
1529 | { | |
1530 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1531 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
93c80368 NB |
1532 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1533 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1534 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 ZW |
1535 | default: |
1536 | fprintf (stderr, "impossible STRING token %s\n", TOKEN_NAME (token)); | |
1537 | return; | |
93c80368 NB |
1538 | } |
1539 | if (tag) putc (tag, fp); | |
47ad4138 | 1540 | putc (left, fp); |
93c80368 | 1541 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
47ad4138 | 1542 | putc (right, fp); |
93c80368 NB |
1543 | } |
1544 | break; | |
c5a04734 | 1545 | |
93c80368 NB |
1546 | case SPELL_NONE: |
1547 | /* An error, most probably. */ | |
1548 | break; | |
041c3194 | 1549 | } |
c5a04734 ZW |
1550 | } |
1551 | ||
93c80368 NB |
1552 | /* Compare two tokens. */ |
1553 | int | |
1554 | _cpp_equiv_tokens (a, b) | |
1555 | const cpp_token *a, *b; | |
c5a04734 | 1556 | { |
93c80368 NB |
1557 | if (a->type == b->type && a->flags == b->flags) |
1558 | switch (TOKEN_SPELL (a)) | |
1559 | { | |
1560 | default: /* Keep compiler happy. */ | |
1561 | case SPELL_OPERATOR: | |
1562 | return 1; | |
1563 | case SPELL_CHAR: | |
6c53ebff | 1564 | return a->val.c == b->val.c; /* Character. */ |
93c80368 | 1565 | case SPELL_NONE: |
56051c0a | 1566 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
93c80368 NB |
1567 | case SPELL_IDENT: |
1568 | return a->val.node == b->val.node; | |
47ad4138 | 1569 | case SPELL_NUMBER: |
93c80368 NB |
1570 | case SPELL_STRING: |
1571 | return (a->val.str.len == b->val.str.len | |
1572 | && !memcmp (a->val.str.text, b->val.str.text, | |
1573 | a->val.str.len)); | |
1574 | } | |
c5a04734 | 1575 | |
041c3194 ZW |
1576 | return 0; |
1577 | } | |
1578 | ||
93c80368 NB |
1579 | /* Returns nonzero if a space should be inserted to avoid an |
1580 | accidental token paste for output. For simplicity, it is | |
1581 | conservative, and occasionally advises a space where one is not | |
1582 | needed, e.g. "." and ".2". */ | |
93c80368 NB |
1583 | int |
1584 | cpp_avoid_paste (pfile, token1, token2) | |
c5a04734 | 1585 | cpp_reader *pfile; |
93c80368 | 1586 | const cpp_token *token1, *token2; |
c5a04734 | 1587 | { |
93c80368 NB |
1588 | enum cpp_ttype a = token1->type, b = token2->type; |
1589 | cppchar_t c; | |
c5a04734 | 1590 | |
93c80368 NB |
1591 | if (token1->flags & NAMED_OP) |
1592 | a = CPP_NAME; | |
1593 | if (token2->flags & NAMED_OP) | |
1594 | b = CPP_NAME; | |
c5a04734 | 1595 | |
93c80368 NB |
1596 | c = EOF; |
1597 | if (token2->flags & DIGRAPH) | |
37b8524c | 1598 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
93c80368 NB |
1599 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1600 | c = token_spellings[b].name[0]; | |
c5a04734 | 1601 | |
93c80368 | 1602 | /* Quickly get everything that can paste with an '='. */ |
37b8524c | 1603 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
93c80368 | 1604 | return 1; |
c5a04734 | 1605 | |
93c80368 | 1606 | switch (a) |
c5a04734 | 1607 | { |
93c80368 NB |
1608 | case CPP_GREATER: return c == '>' || c == '?'; |
1609 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1610 | case CPP_PLUS: return c == '+'; | |
1611 | case CPP_MINUS: return c == '-' || c == '>'; | |
1612 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1613 | case CPP_MOD: return c == ':' || c == '>'; | |
1614 | case CPP_AND: return c == '&'; | |
1615 | case CPP_OR: return c == '|'; | |
1616 | case CPP_COLON: return c == ':' || c == '>'; | |
1617 | case CPP_DEREF: return c == '*'; | |
26ec42ee | 1618 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
93c80368 NB |
1619 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1620 | case CPP_NAME: return ((b == CPP_NUMBER | |
1621 | && name_p (pfile, &token2->val.str)) | |
1622 | || b == CPP_NAME | |
1623 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1624 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1625 | || c == '.' || c == '+' || c == '-'); | |
1626 | case CPP_OTHER: return (CPP_OPTION (pfile, objc) | |
6c53ebff | 1627 | && token1->val.c == '@' |
93c80368 NB |
1628 | && (b == CPP_NAME || b == CPP_STRING)); |
1629 | default: break; | |
c5a04734 | 1630 | } |
c5a04734 | 1631 | |
417f3e3a | 1632 | return 0; |
c5a04734 ZW |
1633 | } |
1634 | ||
93c80368 | 1635 | /* Output all the remaining tokens on the current line, and a newline |
4ed5bcfb NB |
1636 | character, to FP. Leading whitespace is removed. If there are |
1637 | macros, special token padding is not performed. */ | |
c5a04734 | 1638 | void |
93c80368 | 1639 | cpp_output_line (pfile, fp) |
c5a04734 | 1640 | cpp_reader *pfile; |
93c80368 | 1641 | FILE *fp; |
c5a04734 | 1642 | { |
4ed5bcfb | 1643 | const cpp_token *token; |
96be6998 | 1644 | |
4ed5bcfb NB |
1645 | token = cpp_get_token (pfile); |
1646 | while (token->type != CPP_EOF) | |
96be6998 | 1647 | { |
4ed5bcfb NB |
1648 | cpp_output_token (token, fp); |
1649 | token = cpp_get_token (pfile); | |
1650 | if (token->flags & PREV_WHITE) | |
1651 | putc (' ', fp); | |
96be6998 ZW |
1652 | } |
1653 | ||
93c80368 | 1654 | putc ('\n', fp); |
041c3194 | 1655 | } |
c5a04734 | 1656 | |
c8a96070 NB |
1657 | /* Returns the value of a hexadecimal digit. */ |
1658 | static unsigned int | |
1659 | hex_digit_value (c) | |
1660 | unsigned int c; | |
1661 | { | |
9e1ac915 KG |
1662 | if (hex_p (c)) |
1663 | return hex_value (c); | |
1664 | else | |
1665 | abort (); | |
c8a96070 NB |
1666 | } |
1667 | ||
62729350 NB |
1668 | /* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. Returns 1 to indicate |
1669 | failure if cpplib is not parsing C++ or C99. Such failure is | |
1670 | silent, and no variables are updated. Otherwise returns 0, and | |
1671 | warns if -Wtraditional. | |
c8a96070 NB |
1672 | |
1673 | [lex.charset]: The character designated by the universal character | |
1674 | name \UNNNNNNNN is that character whose character short name in | |
1675 | ISO/IEC 10646 is NNNNNNNN; the character designated by the | |
1676 | universal character name \uNNNN is that character whose character | |
1677 | short name in ISO/IEC 10646 is 0000NNNN. If the hexadecimal value | |
1678 | for a universal character name is less than 0x20 or in the range | |
1679 | 0x7F-0x9F (inclusive), or if the universal character name | |
1680 | designates a character in the basic source character set, then the | |
1681 | program is ill-formed. | |
1682 | ||
1683 | We assume that wchar_t is Unicode, so we don't need to do any | |
62729350 | 1684 | mapping. Is this ever wrong? |
c8a96070 | 1685 | |
62729350 NB |
1686 | PC points to the 'u' or 'U', PSTR is points to the byte after PC, |
1687 | LIMIT is the end of the string or charconst. PSTR is updated to | |
1688 | point after the UCS on return, and the UCS is written into PC. */ | |
1689 | ||
1690 | static int | |
1691 | maybe_read_ucs (pfile, pstr, limit, pc) | |
c8a96070 NB |
1692 | cpp_reader *pfile; |
1693 | const unsigned char **pstr; | |
1694 | const unsigned char *limit; | |
625458d0 | 1695 | cppchar_t *pc; |
c8a96070 NB |
1696 | { |
1697 | const unsigned char *p = *pstr; | |
62729350 NB |
1698 | unsigned int code = 0; |
1699 | unsigned int c = *pc, length; | |
1700 | ||
1701 | /* Only attempt to interpret a UCS for C++ and C99. */ | |
1702 | if (! (CPP_OPTION (pfile, cplusplus) || CPP_OPTION (pfile, c99))) | |
1703 | return 1; | |
c8a96070 | 1704 | |
62729350 | 1705 | if (CPP_WTRADITIONAL (pfile)) |
ebef4e8c NB |
1706 | cpp_error (pfile, DL_WARNING, |
1707 | "the meaning of '\\%c' is different in traditional C", c); | |
c8a96070 | 1708 | |
f8710242 NB |
1709 | length = (c == 'u' ? 4: 8); |
1710 | ||
1711 | if ((size_t) (limit - p) < length) | |
1712 | { | |
ebef4e8c | 1713 | cpp_error (pfile, DL_ERROR, "incomplete universal-character-name"); |
f8710242 NB |
1714 | /* Skip to the end to avoid more diagnostics. */ |
1715 | p = limit; | |
1716 | } | |
1717 | else | |
1718 | { | |
1719 | for (; length; length--, p++) | |
c8a96070 | 1720 | { |
f8710242 NB |
1721 | c = *p; |
1722 | if (ISXDIGIT (c)) | |
1723 | code = (code << 4) + hex_digit_value (c); | |
1724 | else | |
1725 | { | |
ebef4e8c | 1726 | cpp_error (pfile, DL_ERROR, |
f8710242 NB |
1727 | "non-hex digit '%c' in universal-character-name", c); |
1728 | /* We shouldn't skip in case there are multibyte chars. */ | |
1729 | break; | |
1730 | } | |
c8a96070 | 1731 | } |
c8a96070 NB |
1732 | } |
1733 | ||
783e2989 NB |
1734 | if (CPP_OPTION (pfile, EBCDIC)) |
1735 | { | |
1736 | cpp_error (pfile, DL_ERROR, "universal-character-name on EBCDIC target"); | |
1737 | code = 0x3f; /* EBCDIC invalid character */ | |
1738 | } | |
1739 | /* True extended characters are OK. */ | |
1740 | else if (code >= 0xa0 | |
1741 | && !(code & 0x80000000) | |
1742 | && !(code >= 0xD800 && code <= 0xDFFF)) | |
f8710242 NB |
1743 | ; |
1744 | /* The standard permits $, @ and ` to be specified as UCNs. We use | |
1745 | hex escapes so that this also works with EBCDIC hosts. */ | |
1746 | else if (code == 0x24 || code == 0x40 || code == 0x60) | |
1747 | ; | |
1748 | /* Don't give another error if one occurred above. */ | |
1749 | else if (length == 0) | |
ebef4e8c | 1750 | cpp_error (pfile, DL_ERROR, "universal-character-name out of range"); |
c8a96070 NB |
1751 | |
1752 | *pstr = p; | |
62729350 NB |
1753 | *pc = code; |
1754 | return 0; | |
c8a96070 NB |
1755 | } |
1756 | ||
4268e8bb NB |
1757 | /* Returns the value of an escape sequence, truncated to the correct |
1758 | target precision. PSTR points to the input pointer, which is just | |
1759 | after the backslash. LIMIT is how much text we have. WIDE is true | |
1760 | if the escape sequence is part of a wide character constant or | |
1761 | string literal. Handles all relevant diagnostics. */ | |
1762 | cppchar_t | |
1763 | cpp_parse_escape (pfile, pstr, limit, wide) | |
c8a96070 NB |
1764 | cpp_reader *pfile; |
1765 | const unsigned char **pstr; | |
1766 | const unsigned char *limit; | |
4268e8bb | 1767 | int wide; |
c8a96070 | 1768 | { |
783e2989 NB |
1769 | /* Values of \a \b \e \f \n \r \t \v respectively. */ |
1770 | static const uchar ascii[] = { 7, 8, 27, 12, 10, 13, 9, 11 }; | |
1771 | static const uchar ebcdic[] = { 47, 22, 39, 12, 21, 13, 5, 11 }; | |
1772 | ||
c8a96070 | 1773 | int unknown = 0; |
783e2989 | 1774 | const unsigned char *str = *pstr, *charconsts; |
4268e8bb NB |
1775 | cppchar_t c, mask; |
1776 | unsigned int width; | |
1777 | ||
783e2989 NB |
1778 | if (CPP_OPTION (pfile, EBCDIC)) |
1779 | charconsts = ebcdic; | |
1780 | else | |
1781 | charconsts = ascii; | |
1782 | ||
4268e8bb NB |
1783 | if (wide) |
1784 | width = CPP_OPTION (pfile, wchar_precision); | |
1785 | else | |
1786 | width = CPP_OPTION (pfile, char_precision); | |
1787 | if (width < BITS_PER_CPPCHAR_T) | |
1788 | mask = ((cppchar_t) 1 << width) - 1; | |
1789 | else | |
1790 | mask = ~0; | |
c8a96070 | 1791 | |
4268e8bb | 1792 | c = *str++; |
c8a96070 NB |
1793 | switch (c) |
1794 | { | |
1795 | case '\\': case '\'': case '"': case '?': break; | |
783e2989 NB |
1796 | case 'b': c = charconsts[1]; break; |
1797 | case 'f': c = charconsts[3]; break; | |
1798 | case 'n': c = charconsts[4]; break; | |
1799 | case 'r': c = charconsts[5]; break; | |
1800 | case 't': c = charconsts[6]; break; | |
1801 | case 'v': c = charconsts[7]; break; | |
c8a96070 NB |
1802 | |
1803 | case '(': case '{': case '[': case '%': | |
1804 | /* '\(', etc, are used at beginning of line to avoid confusing Emacs. | |
1805 | '\%' is used to prevent SCCS from getting confused. */ | |
1806 | unknown = CPP_PEDANTIC (pfile); | |
1807 | break; | |
1808 | ||
1809 | case 'a': | |
1810 | if (CPP_WTRADITIONAL (pfile)) | |
ebef4e8c NB |
1811 | cpp_error (pfile, DL_WARNING, |
1812 | "the meaning of '\\a' is different in traditional C"); | |
783e2989 | 1813 | c = charconsts[0]; |
c8a96070 NB |
1814 | break; |
1815 | ||
1816 | case 'e': case 'E': | |
1817 | if (CPP_PEDANTIC (pfile)) | |
ebef4e8c | 1818 | cpp_error (pfile, DL_PEDWARN, |
625458d0 | 1819 | "non-ISO-standard escape sequence, '\\%c'", (int) c); |
783e2989 | 1820 | c = charconsts[2]; |
c8a96070 | 1821 | break; |
df383483 | 1822 | |
c8a96070 | 1823 | case 'u': case 'U': |
62729350 | 1824 | unknown = maybe_read_ucs (pfile, &str, limit, &c); |
c8a96070 NB |
1825 | break; |
1826 | ||
1827 | case 'x': | |
1828 | if (CPP_WTRADITIONAL (pfile)) | |
ebef4e8c NB |
1829 | cpp_error (pfile, DL_WARNING, |
1830 | "the meaning of '\\x' is different in traditional C"); | |
c8a96070 | 1831 | |
df383483 KH |
1832 | { |
1833 | cppchar_t i = 0, overflow = 0; | |
1834 | int digits_found = 0; | |
c8a96070 | 1835 | |
df383483 KH |
1836 | while (str < limit) |
1837 | { | |
1838 | c = *str; | |
1839 | if (! ISXDIGIT (c)) | |
1840 | break; | |
1841 | str++; | |
1842 | overflow |= i ^ (i << 4 >> 4); | |
1843 | i = (i << 4) + hex_digit_value (c); | |
1844 | digits_found = 1; | |
1845 | } | |
c8a96070 | 1846 | |
df383483 KH |
1847 | if (!digits_found) |
1848 | cpp_error (pfile, DL_ERROR, | |
ebef4e8c | 1849 | "\\x used with no following hex digits"); |
c8a96070 | 1850 | |
df383483 KH |
1851 | if (overflow | (i != (i & mask))) |
1852 | { | |
1853 | cpp_error (pfile, DL_PEDWARN, | |
1854 | "hex escape sequence out of range"); | |
1855 | i &= mask; | |
1856 | } | |
1857 | c = i; | |
1858 | } | |
c8a96070 NB |
1859 | break; |
1860 | ||
1861 | case '0': case '1': case '2': case '3': | |
1862 | case '4': case '5': case '6': case '7': | |
1863 | { | |
4268e8bb NB |
1864 | size_t count = 0; |
1865 | cppchar_t i = c - '0'; | |
c8a96070 NB |
1866 | |
1867 | while (str < limit && ++count < 3) | |
1868 | { | |
1869 | c = *str; | |
1870 | if (c < '0' || c > '7') | |
1871 | break; | |
1872 | str++; | |
1873 | i = (i << 3) + c - '0'; | |
1874 | } | |
1875 | ||
1876 | if (i != (i & mask)) | |
1877 | { | |
ebef4e8c NB |
1878 | cpp_error (pfile, DL_PEDWARN, |
1879 | "octal escape sequence out of range"); | |
c8a96070 NB |
1880 | i &= mask; |
1881 | } | |
1882 | c = i; | |
1883 | } | |
1884 | break; | |
1885 | ||
1886 | default: | |
1887 | unknown = 1; | |
1888 | break; | |
1889 | } | |
1890 | ||
1891 | if (unknown) | |
1892 | { | |
1893 | if (ISGRAPH (c)) | |
625458d0 NB |
1894 | cpp_error (pfile, DL_PEDWARN, |
1895 | "unknown escape sequence '\\%c'", (int) c); | |
c8a96070 | 1896 | else |
625458d0 NB |
1897 | cpp_error (pfile, DL_PEDWARN, |
1898 | "unknown escape sequence: '\\%03o'", (int) c); | |
c8a96070 NB |
1899 | } |
1900 | ||
62729350 | 1901 | if (c > mask) |
4268e8bb | 1902 | { |
639e8b0c | 1903 | cpp_error (pfile, DL_PEDWARN, "escape sequence out of range for its type"); |
4268e8bb NB |
1904 | c &= mask; |
1905 | } | |
62729350 | 1906 | |
c8a96070 NB |
1907 | *pstr = str; |
1908 | return c; | |
1909 | } | |
1910 | ||
c8a96070 | 1911 | /* Interpret a (possibly wide) character constant in TOKEN. |
4268e8bb NB |
1912 | WARN_MULTI warns about multi-character charconsts. PCHARS_SEEN |
1913 | points to a variable that is filled in with the number of | |
1914 | characters seen, and UNSIGNEDP to a variable that indicates whether | |
1915 | the result has signed type. */ | |
1916 | cppchar_t | |
a5a49440 | 1917 | cpp_interpret_charconst (pfile, token, pchars_seen, unsignedp) |
c8a96070 NB |
1918 | cpp_reader *pfile; |
1919 | const cpp_token *token; | |
c8a96070 | 1920 | unsigned int *pchars_seen; |
4268e8bb | 1921 | int *unsignedp; |
c8a96070 NB |
1922 | { |
1923 | const unsigned char *str = token->val.str.text; | |
1924 | const unsigned char *limit = str + token->val.str.len; | |
1925 | unsigned int chars_seen = 0; | |
639e8b0c | 1926 | size_t width, max_chars; |
4268e8bb | 1927 | cppchar_t c, mask, result = 0; |
a47ed310 | 1928 | bool unsigned_p; |
c8a96070 NB |
1929 | |
1930 | #ifdef MULTIBYTE_CHARS | |
1931 | (void) local_mbtowc (NULL, NULL, 0); | |
1932 | #endif | |
1933 | ||
1934 | /* Width in bits. */ | |
1935 | if (token->type == CPP_CHAR) | |
a47ed310 | 1936 | { |
4268e8bb | 1937 | width = CPP_OPTION (pfile, char_precision); |
2443d4e1 | 1938 | max_chars = CPP_OPTION (pfile, int_precision) / width; |
44a147ad | 1939 | unsigned_p = CPP_OPTION (pfile, unsigned_char); |
a47ed310 | 1940 | } |
c8a96070 | 1941 | else |
a47ed310 | 1942 | { |
4268e8bb | 1943 | width = CPP_OPTION (pfile, wchar_precision); |
2443d4e1 | 1944 | max_chars = 1; |
44a147ad | 1945 | unsigned_p = CPP_OPTION (pfile, unsigned_wchar); |
a47ed310 | 1946 | } |
c8a96070 | 1947 | |
4268e8bb NB |
1948 | if (width < BITS_PER_CPPCHAR_T) |
1949 | mask = ((cppchar_t) 1 << width) - 1; | |
c8a96070 NB |
1950 | else |
1951 | mask = ~0; | |
c8a96070 NB |
1952 | |
1953 | while (str < limit) | |
1954 | { | |
1955 | #ifdef MULTIBYTE_CHARS | |
1956 | wchar_t wc; | |
1957 | int char_len; | |
1958 | ||
fc69c47c | 1959 | char_len = local_mbtowc (&wc, (const char *)str, limit - str); |
c8a96070 NB |
1960 | if (char_len == -1) |
1961 | { | |
ebef4e8c NB |
1962 | cpp_error (pfile, DL_WARNING, |
1963 | "ignoring invalid multibyte character"); | |
c8a96070 NB |
1964 | c = *str++; |
1965 | } | |
1966 | else | |
1967 | { | |
1968 | str += char_len; | |
1969 | c = wc; | |
1970 | } | |
1971 | #else | |
1972 | c = *str++; | |
1973 | #endif | |
1974 | ||
1975 | if (c == '\\') | |
4268e8bb | 1976 | c = cpp_parse_escape (pfile, &str, limit, token->type == CPP_WCHAR); |
c8a96070 NB |
1977 | |
1978 | #ifdef MAP_CHARACTER | |
1979 | if (ISPRINT (c)) | |
1980 | c = MAP_CHARACTER (c); | |
1981 | #endif | |
df383483 | 1982 | |
639e8b0c NB |
1983 | chars_seen++; |
1984 | ||
a5a49440 NB |
1985 | /* Truncate the character, scale the result and merge the two. */ |
1986 | c &= mask; | |
639e8b0c | 1987 | if (width < BITS_PER_CPPCHAR_T) |
a5a49440 | 1988 | result = (result << width) | c; |
639e8b0c NB |
1989 | else |
1990 | result = c; | |
c8a96070 NB |
1991 | } |
1992 | ||
1993 | if (chars_seen == 0) | |
ebef4e8c | 1994 | cpp_error (pfile, DL_ERROR, "empty character constant"); |
639e8b0c | 1995 | else if (chars_seen > 1) |
c8a96070 | 1996 | { |
639e8b0c NB |
1997 | /* Multichar charconsts are of type int and therefore signed. */ |
1998 | unsigned_p = 0; | |
a5a49440 | 1999 | |
639e8b0c NB |
2000 | if (chars_seen > max_chars) |
2001 | { | |
2002 | chars_seen = max_chars; | |
2003 | cpp_error (pfile, DL_WARNING, | |
2004 | "character constant too long for its type"); | |
2005 | } | |
a5a49440 | 2006 | else if (CPP_OPTION (pfile, warn_multichar)) |
639e8b0c | 2007 | cpp_error (pfile, DL_WARNING, "multi-character character constant"); |
c8a96070 NB |
2008 | } |
2009 | ||
b9e2d17b NB |
2010 | /* Sign-extend or truncate the constant to cppchar_t. The value is |
2011 | in WIDTH bits, but for multi-char charconsts it's value is the | |
2012 | full target type's width. */ | |
2013 | if (chars_seen > 1) | |
2014 | width *= max_chars; | |
2015 | if (width < BITS_PER_CPPCHAR_T) | |
a5a49440 | 2016 | { |
b9e2d17b NB |
2017 | mask = ((cppchar_t) 1 << width) - 1; |
2018 | if (unsigned_p || !(result & (1 << (width - 1)))) | |
2019 | result &= mask; | |
2020 | else | |
2021 | result |= ~mask; | |
a5a49440 NB |
2022 | } |
2023 | ||
c8a96070 | 2024 | *pchars_seen = chars_seen; |
4268e8bb | 2025 | *unsignedp = unsigned_p; |
c8a96070 NB |
2026 | return result; |
2027 | } | |
2028 | ||
1e013d2e NB |
2029 | /* Memory buffers. Changing these three constants can have a dramatic |
2030 | effect on performance. The values here are reasonable defaults, | |
2031 | but might be tuned. If you adjust them, be sure to test across a | |
2032 | range of uses of cpplib, including heavy nested function-like macro | |
2033 | expansion. Also check the change in peak memory usage (NJAMD is a | |
2034 | good tool for this). */ | |
2035 | #define MIN_BUFF_SIZE 8000 | |
87062813 | 2036 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
1e013d2e NB |
2037 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
2038 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
417f3e3a | 2039 | |
87062813 NB |
2040 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
2041 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
2042 | #endif | |
2043 | ||
c9e7a609 NB |
2044 | /* Create a new allocation buffer. Place the control block at the end |
2045 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
b8af0ca5 NB |
2046 | static _cpp_buff * |
2047 | new_buff (len) | |
6142088c | 2048 | size_t len; |
b8af0ca5 NB |
2049 | { |
2050 | _cpp_buff *result; | |
ece54d54 | 2051 | unsigned char *base; |
b8af0ca5 | 2052 | |
1e013d2e NB |
2053 | if (len < MIN_BUFF_SIZE) |
2054 | len = MIN_BUFF_SIZE; | |
c70f6ed3 | 2055 | len = CPP_ALIGN (len); |
b8af0ca5 NB |
2056 | |
2057 | base = xmalloc (len + sizeof (_cpp_buff)); | |
2058 | result = (_cpp_buff *) (base + len); | |
2059 | result->base = base; | |
2060 | result->cur = base; | |
2061 | result->limit = base + len; | |
2062 | result->next = NULL; | |
2063 | return result; | |
2064 | } | |
2065 | ||
2066 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
2067 | void | |
2068 | _cpp_release_buff (pfile, buff) | |
2069 | cpp_reader *pfile; | |
2070 | _cpp_buff *buff; | |
2071 | { | |
2072 | _cpp_buff *end = buff; | |
2073 | ||
2074 | while (end->next) | |
2075 | end = end->next; | |
2076 | end->next = pfile->free_buffs; | |
2077 | pfile->free_buffs = buff; | |
2078 | } | |
2079 | ||
2080 | /* Return a free buffer of size at least MIN_SIZE. */ | |
2081 | _cpp_buff * | |
2082 | _cpp_get_buff (pfile, min_size) | |
2083 | cpp_reader *pfile; | |
6142088c | 2084 | size_t min_size; |
b8af0ca5 NB |
2085 | { |
2086 | _cpp_buff *result, **p; | |
2087 | ||
2088 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
2089 | { | |
6142088c | 2090 | size_t size; |
1e013d2e NB |
2091 | |
2092 | if (*p == NULL) | |
b8af0ca5 | 2093 | return new_buff (min_size); |
1e013d2e NB |
2094 | result = *p; |
2095 | size = result->limit - result->base; | |
2096 | /* Return a buffer that's big enough, but don't waste one that's | |
2097 | way too big. */ | |
34f5271d | 2098 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
b8af0ca5 NB |
2099 | break; |
2100 | } | |
2101 | ||
2102 | *p = result->next; | |
2103 | result->next = NULL; | |
2104 | result->cur = result->base; | |
2105 | return result; | |
2106 | } | |
2107 | ||
4fe9b91c | 2108 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
2109 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
2110 | the excess bytes to the new buffer. Chains the new buffer after | |
2111 | BUFF, and returns the new buffer. */ | |
b8af0ca5 | 2112 | _cpp_buff * |
8c3b2693 | 2113 | _cpp_append_extend_buff (pfile, buff, min_extra) |
b8af0ca5 NB |
2114 | cpp_reader *pfile; |
2115 | _cpp_buff *buff; | |
6142088c | 2116 | size_t min_extra; |
b8af0ca5 | 2117 | { |
6142088c | 2118 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
8c3b2693 | 2119 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
b8af0ca5 | 2120 | |
8c3b2693 NB |
2121 | buff->next = new_buff; |
2122 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
2123 | return new_buff; | |
2124 | } | |
2125 | ||
4fe9b91c | 2126 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
2127 | remaining bytes of the buffer pointed to by BUFF, and at least |
2128 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
2129 | Chains the new buffer before the buffer pointed to by BUFF, and | |
2130 | updates the pointer to point to the new buffer. */ | |
2131 | void | |
2132 | _cpp_extend_buff (pfile, pbuff, min_extra) | |
2133 | cpp_reader *pfile; | |
2134 | _cpp_buff **pbuff; | |
2135 | size_t min_extra; | |
2136 | { | |
2137 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
2138 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
2139 | ||
2140 | new_buff = _cpp_get_buff (pfile, size); | |
2141 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
2142 | new_buff->next = old_buff; | |
2143 | *pbuff = new_buff; | |
b8af0ca5 NB |
2144 | } |
2145 | ||
2146 | /* Free a chain of buffers starting at BUFF. */ | |
2147 | void | |
2148 | _cpp_free_buff (buff) | |
2149 | _cpp_buff *buff; | |
2150 | { | |
2151 | _cpp_buff *next; | |
2152 | ||
2153 | for (; buff; buff = next) | |
2154 | { | |
2155 | next = buff->next; | |
2156 | free (buff->base); | |
2157 | } | |
2158 | } | |
417f3e3a | 2159 | |
ece54d54 NB |
2160 | /* Allocate permanent, unaligned storage of length LEN. */ |
2161 | unsigned char * | |
2162 | _cpp_unaligned_alloc (pfile, len) | |
2163 | cpp_reader *pfile; | |
2164 | size_t len; | |
2165 | { | |
2166 | _cpp_buff *buff = pfile->u_buff; | |
2167 | unsigned char *result = buff->cur; | |
2168 | ||
2169 | if (len > (size_t) (buff->limit - result)) | |
2170 | { | |
2171 | buff = _cpp_get_buff (pfile, len); | |
2172 | buff->next = pfile->u_buff; | |
2173 | pfile->u_buff = buff; | |
2174 | result = buff->cur; | |
2175 | } | |
2176 | ||
2177 | buff->cur = result + len; | |
2178 | return result; | |
2179 | } | |
2180 | ||
87062813 NB |
2181 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
2182 | That buffer is used for growing allocations when saving macro | |
2183 | replacement lists in a #define, and when parsing an answer to an | |
2184 | assertion in #assert, #unassert or #if (and therefore possibly | |
2185 | whilst expanding macros). It therefore must not be used by any | |
2186 | code that they might call: specifically the lexer and the guts of | |
2187 | the macro expander. | |
2188 | ||
2189 | All existing other uses clearly fit this restriction: storing | |
2190 | registered pragmas during initialization. */ | |
93c80368 | 2191 | unsigned char * |
8c3b2693 NB |
2192 | _cpp_aligned_alloc (pfile, len) |
2193 | cpp_reader *pfile; | |
2194 | size_t len; | |
3fef5b2b | 2195 | { |
8c3b2693 NB |
2196 | _cpp_buff *buff = pfile->a_buff; |
2197 | unsigned char *result = buff->cur; | |
3fef5b2b | 2198 | |
8c3b2693 | 2199 | if (len > (size_t) (buff->limit - result)) |
3fef5b2b | 2200 | { |
8c3b2693 NB |
2201 | buff = _cpp_get_buff (pfile, len); |
2202 | buff->next = pfile->a_buff; | |
2203 | pfile->a_buff = buff; | |
2204 | result = buff->cur; | |
3fef5b2b | 2205 | } |
041c3194 | 2206 | |
8c3b2693 | 2207 | buff->cur = result + len; |
93c80368 | 2208 | return result; |
041c3194 | 2209 | } |