]>
Commit | Line | Data |
---|---|---|
45b966db | 1 | /* CPP Library - lexical analysis. |
5d8ebbd8 | 2 | Copyright (C) 2000, 2001, 2002 Free Software Foundation, Inc. |
45b966db ZW |
3 | Contributed by Per Bothner, 1994-95. |
4 | Based on CCCP program by Paul Rubin, June 1986 | |
5 | Adapted to ANSI C, Richard Stallman, Jan 1987 | |
6 | Broken out to separate file, Zack Weinberg, Mar 2000 | |
c5a04734 | 7 | Single-pass line tokenization by Neil Booth, April 2000 |
45b966db ZW |
8 | |
9 | This program is free software; you can redistribute it and/or modify it | |
10 | under the terms of the GNU General Public License as published by the | |
11 | Free Software Foundation; either version 2, or (at your option) any | |
12 | later version. | |
13 | ||
14 | This program is distributed in the hope that it will be useful, | |
15 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
17 | GNU General Public License for more details. | |
18 | ||
19 | You should have received a copy of the GNU General Public License | |
20 | along with this program; if not, write to the Free Software | |
21 | Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ | |
22 | ||
23 | #include "config.h" | |
24 | #include "system.h" | |
45b966db ZW |
25 | #include "cpplib.h" |
26 | #include "cpphash.h" | |
27 | ||
c8a96070 NB |
28 | #ifdef MULTIBYTE_CHARS |
29 | #include "mbchar.h" | |
30 | #include <locale.h> | |
31 | #endif | |
32 | ||
93c80368 NB |
33 | /* Tokens with SPELL_STRING store their spelling in the token list, |
34 | and it's length in the token->val.name.len. */ | |
35 | enum spell_type | |
f9a0e96c | 36 | { |
93c80368 NB |
37 | SPELL_OPERATOR = 0, |
38 | SPELL_CHAR, | |
39 | SPELL_IDENT, | |
47ad4138 | 40 | SPELL_NUMBER, |
93c80368 NB |
41 | SPELL_STRING, |
42 | SPELL_NONE | |
f9a0e96c ZW |
43 | }; |
44 | ||
93c80368 | 45 | struct token_spelling |
f9a0e96c | 46 | { |
93c80368 NB |
47 | enum spell_type category; |
48 | const unsigned char *name; | |
f9a0e96c ZW |
49 | }; |
50 | ||
8206c799 ZW |
51 | static const unsigned char *const digraph_spellings[] = |
52 | { U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" }; | |
93c80368 NB |
53 | |
54 | #define OP(e, s) { SPELL_OPERATOR, U s }, | |
55 | #define TK(e, s) { s, U STRINGX (e) }, | |
8206c799 | 56 | static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE }; |
93c80368 NB |
57 | #undef OP |
58 | #undef TK | |
59 | ||
60 | #define TOKEN_SPELL(token) (token_spellings[(token)->type].category) | |
61 | #define TOKEN_NAME(token) (token_spellings[(token)->type].name) | |
480709cc | 62 | #define BACKUP() do {buffer->cur = buffer->backup_to;} while (0) |
f2d5f0cc | 63 | |
87062813 NB |
64 | static void handle_newline PARAMS ((cpp_reader *)); |
65 | static cppchar_t skip_escaped_newlines PARAMS ((cpp_reader *)); | |
29401c30 | 66 | static cppchar_t get_effective_char PARAMS ((cpp_reader *)); |
0d9f234d | 67 | |
041c3194 | 68 | static int skip_block_comment PARAMS ((cpp_reader *)); |
cbcff6df | 69 | static int skip_line_comment PARAMS ((cpp_reader *)); |
0d9f234d | 70 | static void adjust_column PARAMS ((cpp_reader *)); |
4d6baafa | 71 | static int skip_whitespace PARAMS ((cpp_reader *, cppchar_t)); |
2c3fcba6 | 72 | static cpp_hashnode *parse_identifier PARAMS ((cpp_reader *)); |
562a5c27 | 73 | static uchar *parse_slow PARAMS ((cpp_reader *, const uchar *, int, |
10cf9bde NB |
74 | unsigned int *)); |
75 | static void parse_number PARAMS ((cpp_reader *, cpp_string *, int)); | |
562a5c27 | 76 | static int unescaped_terminator_p PARAMS ((cpp_reader *, const uchar *)); |
0d9f234d | 77 | static void parse_string PARAMS ((cpp_reader *, cpp_token *, cppchar_t)); |
87062813 | 78 | static bool trigraph_p PARAMS ((cpp_reader *)); |
562a5c27 | 79 | static void save_comment PARAMS ((cpp_reader *, cpp_token *, const uchar *, |
477cdac7 | 80 | cppchar_t)); |
93c80368 | 81 | static int name_p PARAMS ((cpp_reader *, const cpp_string *)); |
62729350 | 82 | static int maybe_read_ucs PARAMS ((cpp_reader *, const unsigned char **, |
625458d0 | 83 | const unsigned char *, cppchar_t *)); |
5fddcffc | 84 | static tokenrun *next_tokenrun PARAMS ((tokenrun *)); |
f617b8e2 | 85 | |
c8a96070 | 86 | static unsigned int hex_digit_value PARAMS ((unsigned int)); |
6142088c | 87 | static _cpp_buff *new_buff PARAMS ((size_t)); |
15dad1d9 | 88 | |
041c3194 | 89 | /* Utility routine: |
9e62c811 | 90 | |
bfb9dc7f ZW |
91 | Compares, the token TOKEN to the NUL-terminated string STRING. |
92 | TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */ | |
041c3194 | 93 | int |
bfb9dc7f ZW |
94 | cpp_ideq (token, string) |
95 | const cpp_token *token; | |
041c3194 ZW |
96 | const char *string; |
97 | { | |
bfb9dc7f | 98 | if (token->type != CPP_NAME) |
041c3194 | 99 | return 0; |
bfb9dc7f | 100 | |
562a5c27 | 101 | return !ustrcmp (NODE_NAME (token->val.node), (const uchar *) string); |
15dad1d9 | 102 | } |
1368ee70 | 103 | |
87062813 NB |
104 | /* Call when meeting a newline, assumed to be in buffer->cur[-1]. |
105 | Returns with buffer->cur pointing to the character immediately | |
106 | following the newline (combination). */ | |
107 | static void | |
108 | handle_newline (pfile) | |
1444f2ed | 109 | cpp_reader *pfile; |
0d9f234d | 110 | { |
87062813 | 111 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 112 | |
87062813 | 113 | /* Handle CR-LF and LF-CR. Most other implementations (e.g. java) |
4d6baafa NB |
114 | only accept CR-LF; maybe we should fall back to that behaviour? */ |
115 | if (buffer->cur[-1] + buffer->cur[0] == '\r' + '\n') | |
87062813 | 116 | buffer->cur++; |
0d9f234d | 117 | |
87062813 NB |
118 | buffer->line_base = buffer->cur; |
119 | buffer->col_adjust = 0; | |
120 | pfile->line++; | |
0d9f234d NB |
121 | } |
122 | ||
87062813 NB |
123 | /* Subroutine of skip_escaped_newlines; called when a 3-character |
124 | sequence beginning with "??" is encountered. buffer->cur points to | |
125 | the second '?'. | |
126 | ||
127 | Warn if necessary, and returns true if the sequence forms a | |
128 | trigraph and the trigraph should be honoured. */ | |
129 | static bool | |
130 | trigraph_p (pfile) | |
45b966db | 131 | cpp_reader *pfile; |
45b966db | 132 | { |
87062813 NB |
133 | cpp_buffer *buffer = pfile->buffer; |
134 | cppchar_t from_char = buffer->cur[1]; | |
135 | bool accept; | |
136 | ||
137 | if (!_cpp_trigraph_map[from_char]) | |
138 | return false; | |
139 | ||
140 | accept = CPP_OPTION (pfile, trigraphs); | |
141 | ||
cbcff6df NB |
142 | /* Don't warn about trigraphs in comments. */ |
143 | if (CPP_OPTION (pfile, warn_trigraphs) && !pfile->state.lexing_comment) | |
45b966db | 144 | { |
041c3194 | 145 | if (accept) |
ebef4e8c NB |
146 | cpp_error_with_line (pfile, DL_WARNING, |
147 | pfile->line, CPP_BUF_COL (buffer) - 1, | |
148 | "trigraph ??%c converted to %c", | |
149 | (int) from_char, | |
150 | (int) _cpp_trigraph_map[from_char]); | |
4a5b68a2 NB |
151 | else if (buffer->cur != buffer->last_Wtrigraphs) |
152 | { | |
153 | buffer->last_Wtrigraphs = buffer->cur; | |
ebef4e8c NB |
154 | cpp_error_with_line (pfile, DL_WARNING, |
155 | pfile->line, CPP_BUF_COL (buffer) - 1, | |
156 | "trigraph ??%c ignored", (int) from_char); | |
4a5b68a2 | 157 | } |
45b966db | 158 | } |
0d9f234d | 159 | |
041c3194 | 160 | return accept; |
45b966db ZW |
161 | } |
162 | ||
87062813 | 163 | /* Skips any escaped newlines introduced by '?' or a '\\', assumed to |
480709cc NB |
164 | lie in buffer->cur[-1]. Returns the next byte, which will be in |
165 | buffer->cur[-1]. This routine performs preprocessing stages 1 and | |
166 | 2 of the ISO C standard. */ | |
0d9f234d | 167 | static cppchar_t |
87062813 | 168 | skip_escaped_newlines (pfile) |
29401c30 | 169 | cpp_reader *pfile; |
45b966db | 170 | { |
29401c30 | 171 | cpp_buffer *buffer = pfile->buffer; |
87062813 | 172 | cppchar_t next = buffer->cur[-1]; |
29401c30 | 173 | |
a5c3cccd NB |
174 | /* Only do this if we apply stages 1 and 2. */ |
175 | if (!buffer->from_stage3) | |
041c3194 | 176 | { |
a5c3cccd | 177 | const unsigned char *saved_cur; |
87062813 | 178 | cppchar_t next1; |
a5c3cccd NB |
179 | |
180 | do | |
0d9f234d | 181 | { |
a5c3cccd | 182 | if (next == '?') |
0d9f234d | 183 | { |
4d6baafa | 184 | if (buffer->cur[0] != '?' || !trigraph_p (pfile)) |
87062813 | 185 | break; |
a5c3cccd | 186 | |
87062813 NB |
187 | /* Translate the trigraph. */ |
188 | next = _cpp_trigraph_map[buffer->cur[1]]; | |
189 | buffer->cur += 2; | |
4d6baafa | 190 | if (next != '\\') |
a5c3cccd | 191 | break; |
a5c3cccd NB |
192 | } |
193 | ||
4d6baafa NB |
194 | if (buffer->cur == buffer->rlimit) |
195 | break; | |
196 | ||
87062813 NB |
197 | /* We have a backslash, and room for at least one more |
198 | character. Skip horizontal whitespace. */ | |
199 | saved_cur = buffer->cur; | |
a5c3cccd | 200 | do |
87062813 NB |
201 | next1 = *buffer->cur++; |
202 | while (is_nvspace (next1) && buffer->cur < buffer->rlimit); | |
041c3194 | 203 | |
a5c3cccd | 204 | if (!is_vspace (next1)) |
0d9f234d | 205 | { |
87062813 | 206 | buffer->cur = saved_cur; |
0d9f234d NB |
207 | break; |
208 | } | |
45b966db | 209 | |
87062813 NB |
210 | if (saved_cur != buffer->cur - 1 |
211 | && !pfile->state.lexing_comment) | |
ebef4e8c NB |
212 | cpp_error (pfile, DL_WARNING, |
213 | "backslash and newline separated by space"); | |
0d9f234d | 214 | |
87062813 | 215 | handle_newline (pfile); |
480709cc | 216 | buffer->backup_to = buffer->cur; |
87062813 NB |
217 | if (buffer->cur == buffer->rlimit) |
218 | { | |
ebef4e8c NB |
219 | cpp_error (pfile, DL_PEDWARN, |
220 | "backslash-newline at end of file"); | |
87062813 NB |
221 | next = EOF; |
222 | } | |
223 | else | |
224 | next = *buffer->cur++; | |
0d9f234d | 225 | } |
a5c3cccd | 226 | while (next == '\\' || next == '?'); |
041c3194 | 227 | } |
45b966db | 228 | |
0d9f234d | 229 | return next; |
45b966db ZW |
230 | } |
231 | ||
0d9f234d | 232 | /* Obtain the next character, after trigraph conversion and skipping |
87062813 NB |
233 | an arbitrarily long string of escaped newlines. The common case of |
234 | no trigraphs or escaped newlines falls through quickly. On return, | |
480709cc NB |
235 | buffer->backup_to points to where to return to if the character is |
236 | not to be processed. */ | |
0d9f234d | 237 | static cppchar_t |
29401c30 NB |
238 | get_effective_char (pfile) |
239 | cpp_reader *pfile; | |
64aaf407 | 240 | { |
4d6baafa | 241 | cppchar_t next; |
480709cc | 242 | cpp_buffer *buffer = pfile->buffer; |
0d9f234d | 243 | |
480709cc | 244 | buffer->backup_to = buffer->cur; |
4d6baafa NB |
245 | next = *buffer->cur++; |
246 | if (__builtin_expect (next == '?' || next == '\\', 0)) | |
247 | next = skip_escaped_newlines (pfile); | |
0d9f234d | 248 | |
480709cc | 249 | return next; |
64aaf407 NB |
250 | } |
251 | ||
0d9f234d NB |
252 | /* Skip a C-style block comment. We find the end of the comment by |
253 | seeing if an asterisk is before every '/' we encounter. Returns | |
254 | non-zero if comment terminated by EOF, zero otherwise. */ | |
041c3194 ZW |
255 | static int |
256 | skip_block_comment (pfile) | |
45b966db ZW |
257 | cpp_reader *pfile; |
258 | { | |
041c3194 | 259 | cpp_buffer *buffer = pfile->buffer; |
d8090680 | 260 | cppchar_t c = EOF, prevc = EOF; |
0d9f234d | 261 | |
cbcff6df | 262 | pfile->state.lexing_comment = 1; |
0d9f234d | 263 | while (buffer->cur != buffer->rlimit) |
45b966db | 264 | { |
0d9f234d NB |
265 | prevc = c, c = *buffer->cur++; |
266 | ||
0d9f234d | 267 | /* FIXME: For speed, create a new character class of characters |
93c80368 | 268 | of interest inside block comments. */ |
0d9f234d | 269 | if (c == '?' || c == '\\') |
87062813 | 270 | c = skip_escaped_newlines (pfile); |
041c3194 | 271 | |
0d9f234d NB |
272 | /* People like decorating comments with '*', so check for '/' |
273 | instead for efficiency. */ | |
041c3194 | 274 | if (c == '/') |
45b966db | 275 | { |
0d9f234d NB |
276 | if (prevc == '*') |
277 | break; | |
041c3194 | 278 | |
0d9f234d | 279 | /* Warn about potential nested comments, but not if the '/' |
a1f300c0 | 280 | comes immediately before the true comment delimiter. |
041c3194 | 281 | Don't bother to get it right across escaped newlines. */ |
0d9f234d | 282 | if (CPP_OPTION (pfile, warn_comments) |
87062813 | 283 | && buffer->cur[0] == '*' && buffer->cur[1] != '/') |
ebef4e8c NB |
284 | cpp_error_with_line (pfile, DL_WARNING, |
285 | pfile->line, CPP_BUF_COL (buffer), | |
286 | "\"/*\" within comment"); | |
45b966db | 287 | } |
91fcd158 | 288 | else if (is_vspace (c)) |
87062813 | 289 | handle_newline (pfile); |
52fadca8 | 290 | else if (c == '\t') |
0d9f234d | 291 | adjust_column (pfile); |
45b966db | 292 | } |
041c3194 | 293 | |
cbcff6df | 294 | pfile->state.lexing_comment = 0; |
0d9f234d | 295 | return c != '/' || prevc != '*'; |
45b966db ZW |
296 | } |
297 | ||
480709cc NB |
298 | /* Skip a C++ line comment, leaving buffer->cur pointing to the |
299 | terminating newline. Handles escaped newlines. Returns non-zero | |
300 | if a multiline comment. */ | |
041c3194 | 301 | static int |
cbcff6df NB |
302 | skip_line_comment (pfile) |
303 | cpp_reader *pfile; | |
45b966db | 304 | { |
cbcff6df | 305 | cpp_buffer *buffer = pfile->buffer; |
67821e3a | 306 | unsigned int orig_line = pfile->line; |
0d9f234d | 307 | cppchar_t c; |
64cdc383 MH |
308 | #ifdef MULTIBYTE_CHARS |
309 | wchar_t wc; | |
310 | int char_len; | |
311 | #endif | |
041c3194 | 312 | |
cbcff6df | 313 | pfile->state.lexing_comment = 1; |
64cdc383 MH |
314 | #ifdef MULTIBYTE_CHARS |
315 | /* Reset multibyte conversion state. */ | |
316 | (void) local_mbtowc (NULL, NULL, 0); | |
317 | #endif | |
0d9f234d | 318 | do |
041c3194 | 319 | { |
0d9f234d | 320 | if (buffer->cur == buffer->rlimit) |
480709cc | 321 | goto at_eof; |
041c3194 | 322 | |
64cdc383 MH |
323 | #ifdef MULTIBYTE_CHARS |
324 | char_len = local_mbtowc (&wc, (const char *) buffer->cur, | |
325 | buffer->rlimit - buffer->cur); | |
326 | if (char_len == -1) | |
327 | { | |
328 | cpp_error (pfile, DL_WARNING, | |
329 | "ignoring invalid multibyte character"); | |
330 | char_len = 1; | |
331 | c = *buffer->cur++; | |
332 | } | |
333 | else | |
334 | { | |
335 | buffer->cur += char_len; | |
336 | c = wc; | |
337 | } | |
338 | #else | |
0d9f234d | 339 | c = *buffer->cur++; |
64cdc383 | 340 | #endif |
0d9f234d | 341 | if (c == '?' || c == '\\') |
87062813 | 342 | c = skip_escaped_newlines (pfile); |
041c3194 | 343 | } |
0d9f234d | 344 | while (!is_vspace (c)); |
45b966db | 345 | |
480709cc NB |
346 | /* Step back over the newline, except at EOF. */ |
347 | buffer->cur--; | |
348 | at_eof: | |
349 | ||
cbcff6df | 350 | pfile->state.lexing_comment = 0; |
67821e3a | 351 | return orig_line != pfile->line; |
041c3194 | 352 | } |
45b966db | 353 | |
0d9f234d NB |
354 | /* pfile->buffer->cur is one beyond the \t character. Update |
355 | col_adjust so we track the column correctly. */ | |
52fadca8 | 356 | static void |
0d9f234d | 357 | adjust_column (pfile) |
52fadca8 | 358 | cpp_reader *pfile; |
52fadca8 | 359 | { |
0d9f234d NB |
360 | cpp_buffer *buffer = pfile->buffer; |
361 | unsigned int col = CPP_BUF_COL (buffer) - 1; /* Zero-based column. */ | |
52fadca8 NB |
362 | |
363 | /* Round it up to multiple of the tabstop, but subtract 1 since the | |
364 | tab itself occupies a character position. */ | |
0d9f234d NB |
365 | buffer->col_adjust += (CPP_OPTION (pfile, tabstop) |
366 | - col % CPP_OPTION (pfile, tabstop)) - 1; | |
52fadca8 NB |
367 | } |
368 | ||
0d9f234d NB |
369 | /* Skips whitespace, saving the next non-whitespace character. |
370 | Adjusts pfile->col_adjust to account for tabs. Without this, | |
371 | tokens might be assigned an incorrect column. */ | |
4d6baafa | 372 | static int |
0d9f234d | 373 | skip_whitespace (pfile, c) |
041c3194 | 374 | cpp_reader *pfile; |
0d9f234d | 375 | cppchar_t c; |
041c3194 ZW |
376 | { |
377 | cpp_buffer *buffer = pfile->buffer; | |
0d9f234d | 378 | unsigned int warned = 0; |
45b966db | 379 | |
0d9f234d | 380 | do |
041c3194 | 381 | { |
91fcd158 NB |
382 | /* Horizontal space always OK. */ |
383 | if (c == ' ') | |
0d9f234d | 384 | ; |
91fcd158 | 385 | else if (c == '\t') |
0d9f234d NB |
386 | adjust_column (pfile); |
387 | /* Just \f \v or \0 left. */ | |
91fcd158 | 388 | else if (c == '\0') |
041c3194 | 389 | { |
4d6baafa NB |
390 | if (buffer->cur - 1 == buffer->rlimit) |
391 | return 0; | |
91fcd158 | 392 | if (!warned) |
0d9f234d | 393 | { |
ebef4e8c | 394 | cpp_error (pfile, DL_WARNING, "null character(s) ignored"); |
0d9f234d NB |
395 | warned = 1; |
396 | } | |
45b966db | 397 | } |
93c80368 | 398 | else if (pfile->state.in_directive && CPP_PEDANTIC (pfile)) |
ebef4e8c NB |
399 | cpp_error_with_line (pfile, DL_PEDWARN, pfile->line, |
400 | CPP_BUF_COL (buffer), | |
401 | "%s in preprocessing directive", | |
402 | c == '\f' ? "form feed" : "vertical tab"); | |
0d9f234d | 403 | |
0d9f234d | 404 | c = *buffer->cur++; |
45b966db | 405 | } |
ec5c56db | 406 | /* We only want non-vertical space, i.e. ' ' \t \f \v \0. */ |
0d9f234d NB |
407 | while (is_nvspace (c)); |
408 | ||
480709cc | 409 | buffer->cur--; |
4d6baafa | 410 | return 1; |
041c3194 | 411 | } |
45b966db | 412 | |
93c80368 NB |
413 | /* See if the characters of a number token are valid in a name (no |
414 | '.', '+' or '-'). */ | |
415 | static int | |
416 | name_p (pfile, string) | |
417 | cpp_reader *pfile; | |
418 | const cpp_string *string; | |
419 | { | |
420 | unsigned int i; | |
421 | ||
422 | for (i = 0; i < string->len; i++) | |
423 | if (!is_idchar (string->text[i])) | |
424 | return 0; | |
425 | ||
426 | return 1; | |
427 | } | |
428 | ||
2c3fcba6 ZW |
429 | /* Parse an identifier, skipping embedded backslash-newlines. This is |
430 | a critical inner loop. The common case is an identifier which has | |
431 | not been split by backslash-newline, does not contain a dollar | |
432 | sign, and has already been scanned (roughly 10:1 ratio of | |
433 | seen:unseen identifiers in normal code; the distribution is | |
434 | Poisson-like). Second most common case is a new identifier, not | |
435 | split and no dollar sign. The other possibilities are rare and | |
10cf9bde | 436 | have been relegated to parse_slow. */ |
0d9f234d | 437 | static cpp_hashnode * |
2c3fcba6 | 438 | parse_identifier (pfile) |
45b966db | 439 | cpp_reader *pfile; |
45b966db | 440 | { |
93c80368 | 441 | cpp_hashnode *result; |
562a5c27 | 442 | const uchar *cur, *base; |
2c3fcba6 ZW |
443 | |
444 | /* Fast-path loop. Skim over a normal identifier. | |
445 | N.B. ISIDNUM does not include $. */ | |
4d6baafa NB |
446 | cur = pfile->buffer->cur; |
447 | while (ISIDNUM (*cur)) | |
2c3fcba6 | 448 | cur++; |
2c3fcba6 ZW |
449 | |
450 | /* Check for slow-path cases. */ | |
4d6baafa | 451 | if (*cur == '?' || *cur == '\\' || *cur == '$') |
10cf9bde NB |
452 | { |
453 | unsigned int len; | |
454 | ||
455 | base = parse_slow (pfile, cur, 0, &len); | |
456 | result = (cpp_hashnode *) | |
457 | ht_lookup (pfile->hash_table, base, len, HT_ALLOCED); | |
458 | } | |
2c3fcba6 ZW |
459 | else |
460 | { | |
10cf9bde NB |
461 | base = pfile->buffer->cur - 1; |
462 | pfile->buffer->cur = cur; | |
2c3fcba6 ZW |
463 | result = (cpp_hashnode *) |
464 | ht_lookup (pfile->hash_table, base, cur - base, HT_ALLOC); | |
2c3fcba6 ZW |
465 | } |
466 | ||
467 | /* Rarely, identifiers require diagnostics when lexed. | |
468 | XXX Has to be forced out of the fast path. */ | |
469 | if (__builtin_expect ((result->flags & NODE_DIAGNOSTIC) | |
470 | && !pfile->state.skipping, 0)) | |
471 | { | |
472 | /* It is allowed to poison the same identifier twice. */ | |
473 | if ((result->flags & NODE_POISONED) && !pfile->state.poisoned_ok) | |
ebef4e8c | 474 | cpp_error (pfile, DL_ERROR, "attempt to use poisoned \"%s\"", |
2c3fcba6 ZW |
475 | NODE_NAME (result)); |
476 | ||
477 | /* Constraint 6.10.3.5: __VA_ARGS__ should only appear in the | |
478 | replacement list of a variadic macro. */ | |
479 | if (result == pfile->spec_nodes.n__VA_ARGS__ | |
480 | && !pfile->state.va_args_ok) | |
ebef4e8c | 481 | cpp_error (pfile, DL_PEDWARN, |
2c3fcba6 ZW |
482 | "__VA_ARGS__ can only appear in the expansion of a C99 variadic macro"); |
483 | } | |
484 | ||
485 | return result; | |
486 | } | |
487 | ||
10cf9bde NB |
488 | /* Slow path. This handles numbers and identifiers which have been |
489 | split, or contain dollar signs. The part of the token from | |
490 | PFILE->buffer->cur-1 to CUR has already been scanned. NUMBER_P is | |
491 | 1 if it's a number, and 2 if it has a leading period. Returns a | |
492 | pointer to the token's NUL-terminated spelling in permanent | |
493 | storage, and sets PLEN to its length. */ | |
562a5c27 | 494 | static uchar * |
10cf9bde | 495 | parse_slow (pfile, cur, number_p, plen) |
2c3fcba6 | 496 | cpp_reader *pfile; |
562a5c27 | 497 | const uchar *cur; |
10cf9bde NB |
498 | int number_p; |
499 | unsigned int *plen; | |
2c3fcba6 | 500 | { |
0d9f234d | 501 | cpp_buffer *buffer = pfile->buffer; |
562a5c27 | 502 | const uchar *base = buffer->cur - 1; |
2a967f3d | 503 | struct obstack *stack = &pfile->hash_table->stack; |
10cf9bde NB |
504 | unsigned int c, prevc, saw_dollar = 0; |
505 | ||
506 | /* Place any leading period. */ | |
507 | if (number_p == 2) | |
508 | obstack_1grow (stack, '.'); | |
2c3fcba6 ZW |
509 | |
510 | /* Copy the part of the token which is known to be okay. */ | |
511 | obstack_grow (stack, base, cur - base); | |
041c3194 | 512 | |
2c3fcba6 ZW |
513 | /* Now process the part which isn't. We are looking at one of |
514 | '$', '\\', or '?' on entry to this loop. */ | |
10cf9bde | 515 | prevc = cur[-1]; |
2c3fcba6 ZW |
516 | c = *cur++; |
517 | buffer->cur = cur; | |
10cf9bde | 518 | for (;;) |
041c3194 | 519 | { |
10cf9bde NB |
520 | /* Potential escaped newline? */ |
521 | buffer->backup_to = buffer->cur - 1; | |
522 | if (c == '?' || c == '\\') | |
523 | c = skip_escaped_newlines (pfile); | |
524 | ||
525 | if (!is_idchar (c)) | |
526 | { | |
527 | if (!number_p) | |
528 | break; | |
529 | if (c != '.' && !VALID_SIGN (c, prevc)) | |
530 | break; | |
531 | } | |
532 | ||
533 | /* Handle normal identifier characters in this loop. */ | |
534 | do | |
2c3fcba6 | 535 | { |
10cf9bde | 536 | prevc = c; |
2c3fcba6 | 537 | obstack_1grow (stack, c); |
45b966db | 538 | |
2c3fcba6 ZW |
539 | if (c == '$') |
540 | saw_dollar++; | |
ba89d661 | 541 | |
2c3fcba6 ZW |
542 | c = *buffer->cur++; |
543 | } | |
10cf9bde | 544 | while (is_idchar (c)); |
041c3194 | 545 | } |
0d9f234d | 546 | |
4d6baafa | 547 | /* Step back over the unwanted char. */ |
480709cc | 548 | BACKUP (); |
93c80368 | 549 | |
4fe9b91c | 550 | /* $ is not an identifier character in the standard, but is commonly |
0d9f234d NB |
551 | accepted as an extension. Don't warn about it in skipped |
552 | conditional blocks. */ | |
cef0d199 | 553 | if (saw_dollar && CPP_PEDANTIC (pfile) && ! pfile->state.skipping) |
ebef4e8c | 554 | cpp_error (pfile, DL_PEDWARN, "'$' character(s) in identifier or number"); |
0d9f234d | 555 | |
10cf9bde NB |
556 | /* Identifiers and numbers are null-terminated. */ |
557 | *plen = obstack_object_size (stack); | |
2a967f3d | 558 | obstack_1grow (stack, '\0'); |
10cf9bde | 559 | return obstack_finish (stack); |
45b966db ZW |
560 | } |
561 | ||
5d8ebbd8 NB |
562 | /* Parse a number, beginning with character C, skipping embedded |
563 | backslash-newlines. LEADING_PERIOD is non-zero if there was a "." | |
564 | before C. Place the result in NUMBER. */ | |
45b966db | 565 | static void |
10cf9bde | 566 | parse_number (pfile, number, leading_period) |
45b966db | 567 | cpp_reader *pfile; |
0d9f234d | 568 | cpp_string *number; |
93c80368 | 569 | int leading_period; |
45b966db | 570 | { |
562a5c27 | 571 | const uchar *cur; |
45b966db | 572 | |
10cf9bde NB |
573 | /* Fast-path loop. Skim over a normal number. |
574 | N.B. ISIDNUM does not include $. */ | |
575 | cur = pfile->buffer->cur; | |
576 | while (ISIDNUM (*cur) || *cur == '.' || VALID_SIGN (*cur, cur[-1])) | |
577 | cur++; | |
cbcff6df | 578 | |
10cf9bde NB |
579 | /* Check for slow-path cases. */ |
580 | if (*cur == '?' || *cur == '\\' || *cur == '$') | |
581 | number->text = parse_slow (pfile, cur, 1 + leading_period, &number->len); | |
582 | else | |
041c3194 | 583 | { |
562a5c27 NB |
584 | const uchar *base = pfile->buffer->cur - 1; |
585 | uchar *dest; | |
0d9f234d | 586 | |
10cf9bde NB |
587 | number->len = cur - base + leading_period; |
588 | dest = _cpp_unaligned_alloc (pfile, number->len + 1); | |
589 | dest[number->len] = '\0'; | |
590 | number->text = dest; | |
45b966db | 591 | |
10cf9bde NB |
592 | if (leading_period) |
593 | *dest++ = '.'; | |
594 | memcpy (dest, base, cur - base); | |
595 | pfile->buffer->cur = cur; | |
45b966db | 596 | } |
0d9f234d NB |
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. | |
45b966db | 625 | |
87062813 NB |
626 | When this function returns, buffer->cur points to the next |
627 | character to be processed. */ | |
041c3194 | 628 | static void |
0d9f234d | 629 | parse_string (pfile, token, terminator) |
45b966db | 630 | cpp_reader *pfile; |
041c3194 | 631 | cpp_token *token; |
0d9f234d | 632 | cppchar_t terminator; |
45b966db | 633 | { |
041c3194 | 634 | cpp_buffer *buffer = pfile->buffer; |
93c80368 | 635 | unsigned char *dest, *limit; |
0d9f234d | 636 | cppchar_t c; |
d4e6133f | 637 | bool warned_nulls = false; |
64cdc383 MH |
638 | #ifdef MULTIBYTE_CHARS |
639 | wchar_t wc; | |
640 | int char_len; | |
641 | #endif | |
0d9f234d | 642 | |
ece54d54 NB |
643 | dest = BUFF_FRONT (pfile->u_buff); |
644 | limit = BUFF_LIMIT (pfile->u_buff); | |
93c80368 | 645 | |
64cdc383 MH |
646 | #ifdef MULTIBYTE_CHARS |
647 | /* Reset multibyte conversion state. */ | |
648 | (void) local_mbtowc (NULL, NULL, 0); | |
649 | #endif | |
0d9f234d | 650 | for (;;) |
45b966db | 651 | { |
87062813 | 652 | /* We need room for another char, possibly the terminating NUL. */ |
ece54d54 NB |
653 | if ((size_t) (limit - dest) < 1) |
654 | { | |
655 | size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff); | |
8c3b2693 | 656 | _cpp_extend_buff (pfile, &pfile->u_buff, 2); |
ece54d54 NB |
657 | dest = BUFF_FRONT (pfile->u_buff) + len_so_far; |
658 | limit = BUFF_LIMIT (pfile->u_buff); | |
659 | } | |
7868b4a2 | 660 | |
64cdc383 MH |
661 | #ifdef MULTIBYTE_CHARS |
662 | char_len = local_mbtowc (&wc, (const char *) buffer->cur, | |
663 | buffer->rlimit - buffer->cur); | |
664 | if (char_len == -1) | |
665 | { | |
666 | cpp_error (pfile, DL_WARNING, | |
667 | "ignoring invalid multibyte character"); | |
668 | char_len = 1; | |
669 | c = *buffer->cur++; | |
670 | } | |
671 | else | |
672 | { | |
673 | buffer->cur += char_len; | |
674 | c = wc; | |
675 | } | |
676 | #else | |
87062813 | 677 | c = *buffer->cur++; |
64cdc383 MH |
678 | #endif |
679 | ||
680 | /* Handle trigraphs, escaped newlines etc. */ | |
0d9f234d | 681 | if (c == '?' || c == '\\') |
87062813 | 682 | c = skip_escaped_newlines (pfile); |
45b966db | 683 | |
87062813 | 684 | if (c == terminator) |
45b966db | 685 | { |
87062813 NB |
686 | if (unescaped_terminator_p (pfile, dest)) |
687 | break; | |
0d9f234d NB |
688 | } |
689 | else if (is_vspace (c)) | |
690 | { | |
d4e6133f NB |
691 | /* No string literal may extend over multiple lines. In |
692 | assembly language, suppress the error except for <> | |
693 | includes. This is a kludge around not knowing where | |
694 | comments are. */ | |
695 | unterminated: | |
696 | if (CPP_OPTION (pfile, lang) != CLK_ASM || terminator == '>') | |
ebef4e8c | 697 | cpp_error (pfile, DL_ERROR, "missing terminating %c character", |
625458d0 | 698 | (int) terminator); |
d4e6133f NB |
699 | buffer->cur--; |
700 | break; | |
0d9f234d | 701 | } |
4d6baafa | 702 | else if (c == '\0') |
0d9f234d | 703 | { |
4d6baafa | 704 | if (buffer->cur - 1 == buffer->rlimit) |
d4e6133f | 705 | goto unterminated; |
4d6baafa NB |
706 | if (!warned_nulls) |
707 | { | |
708 | warned_nulls = true; | |
ebef4e8c NB |
709 | cpp_error (pfile, DL_WARNING, |
710 | "null character(s) preserved in literal"); | |
4d6baafa | 711 | } |
45b966db | 712 | } |
64cdc383 MH |
713 | #ifdef MULTIBYTE_CHARS |
714 | if (char_len > 1) | |
715 | { | |
716 | for ( ; char_len > 0; --char_len) | |
717 | *dest++ = (*buffer->cur - char_len); | |
718 | } | |
719 | else | |
720 | #endif | |
721 | *dest++ = c; | |
45b966db ZW |
722 | } |
723 | ||
7868b4a2 | 724 | *dest = '\0'; |
45b966db | 725 | |
ece54d54 NB |
726 | token->val.str.text = BUFF_FRONT (pfile->u_buff); |
727 | token->val.str.len = dest - BUFF_FRONT (pfile->u_buff); | |
728 | BUFF_FRONT (pfile->u_buff) = dest + 1; | |
0d9f234d | 729 | } |
041c3194 | 730 | |
93c80368 | 731 | /* The stored comment includes the comment start and any terminator. */ |
9e62c811 | 732 | static void |
477cdac7 | 733 | save_comment (pfile, token, from, type) |
0d9f234d | 734 | cpp_reader *pfile; |
041c3194 ZW |
735 | cpp_token *token; |
736 | const unsigned char *from; | |
477cdac7 | 737 | cppchar_t type; |
9e62c811 | 738 | { |
041c3194 | 739 | unsigned char *buffer; |
477cdac7 | 740 | unsigned int len, clen; |
0d9f234d | 741 | |
1c6d33ef | 742 | len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */ |
480709cc | 743 | |
3542203b NB |
744 | /* C++ comments probably (not definitely) have moved past a new |
745 | line, which we don't want to save in the comment. */ | |
480709cc | 746 | if (is_vspace (pfile->buffer->cur[-1])) |
3542203b | 747 | len--; |
477cdac7 JT |
748 | |
749 | /* If we are currently in a directive, then we need to store all | |
750 | C++ comments as C comments internally, and so we need to | |
751 | allocate a little extra space in that case. | |
752 | ||
753 | Note that the only time we encounter a directive here is | |
754 | when we are saving comments in a "#define". */ | |
755 | clen = (pfile->state.in_directive && type == '/') ? len + 2 : len; | |
756 | ||
757 | buffer = _cpp_unaligned_alloc (pfile, clen); | |
041c3194 | 758 | |
041c3194 | 759 | token->type = CPP_COMMENT; |
477cdac7 | 760 | token->val.str.len = clen; |
0d9f234d | 761 | token->val.str.text = buffer; |
45b966db | 762 | |
1c6d33ef NB |
763 | buffer[0] = '/'; |
764 | memcpy (buffer + 1, from, len - 1); | |
477cdac7 | 765 | |
1eeeb6a4 | 766 | /* Finish conversion to a C comment, if necessary. */ |
477cdac7 JT |
767 | if (pfile->state.in_directive && type == '/') |
768 | { | |
769 | buffer[1] = '*'; | |
770 | buffer[clen - 2] = '*'; | |
771 | buffer[clen - 1] = '/'; | |
772 | } | |
0d9f234d | 773 | } |
45b966db | 774 | |
5fddcffc NB |
775 | /* Allocate COUNT tokens for RUN. */ |
776 | void | |
777 | _cpp_init_tokenrun (run, count) | |
778 | tokenrun *run; | |
779 | unsigned int count; | |
780 | { | |
781 | run->base = xnewvec (cpp_token, count); | |
782 | run->limit = run->base + count; | |
783 | run->next = NULL; | |
784 | } | |
785 | ||
786 | /* Returns the next tokenrun, or creates one if there is none. */ | |
787 | static tokenrun * | |
788 | next_tokenrun (run) | |
789 | tokenrun *run; | |
790 | { | |
791 | if (run->next == NULL) | |
792 | { | |
793 | run->next = xnew (tokenrun); | |
bdcbe496 | 794 | run->next->prev = run; |
5fddcffc NB |
795 | _cpp_init_tokenrun (run->next, 250); |
796 | } | |
797 | ||
798 | return run->next; | |
799 | } | |
800 | ||
4ed5bcfb NB |
801 | /* Allocate a single token that is invalidated at the same time as the |
802 | rest of the tokens on the line. Has its line and col set to the | |
803 | same as the last lexed token, so that diagnostics appear in the | |
804 | right place. */ | |
805 | cpp_token * | |
806 | _cpp_temp_token (pfile) | |
807 | cpp_reader *pfile; | |
808 | { | |
809 | cpp_token *old, *result; | |
810 | ||
811 | old = pfile->cur_token - 1; | |
812 | if (pfile->cur_token == pfile->cur_run->limit) | |
813 | { | |
814 | pfile->cur_run = next_tokenrun (pfile->cur_run); | |
815 | pfile->cur_token = pfile->cur_run->base; | |
816 | } | |
817 | ||
818 | result = pfile->cur_token++; | |
819 | result->line = old->line; | |
820 | result->col = old->col; | |
821 | return result; | |
822 | } | |
823 | ||
14baae01 NB |
824 | /* Lex a token into RESULT (external interface). Takes care of issues |
825 | like directive handling, token lookahead, multiple include | |
a1f300c0 | 826 | optimization and skipping. */ |
345894b4 NB |
827 | const cpp_token * |
828 | _cpp_lex_token (pfile) | |
45b966db | 829 | cpp_reader *pfile; |
5fddcffc | 830 | { |
bdcbe496 | 831 | cpp_token *result; |
5fddcffc | 832 | |
bdcbe496 | 833 | for (;;) |
5fddcffc | 834 | { |
bdcbe496 | 835 | if (pfile->cur_token == pfile->cur_run->limit) |
5fddcffc | 836 | { |
bdcbe496 NB |
837 | pfile->cur_run = next_tokenrun (pfile->cur_run); |
838 | pfile->cur_token = pfile->cur_run->base; | |
5fddcffc NB |
839 | } |
840 | ||
bdcbe496 | 841 | if (pfile->lookaheads) |
14baae01 NB |
842 | { |
843 | pfile->lookaheads--; | |
844 | result = pfile->cur_token++; | |
845 | } | |
bdcbe496 | 846 | else |
14baae01 | 847 | result = _cpp_lex_direct (pfile); |
bdcbe496 NB |
848 | |
849 | if (result->flags & BOL) | |
5fddcffc | 850 | { |
bdcbe496 NB |
851 | /* Is this a directive. If _cpp_handle_directive returns |
852 | false, it is an assembler #. */ | |
853 | if (result->type == CPP_HASH | |
e808ec9c NB |
854 | /* 6.10.3 p 11: Directives in a list of macro arguments |
855 | gives undefined behavior. This implementation | |
856 | handles the directive as normal. */ | |
857 | && pfile->state.parsing_args != 1 | |
bdcbe496 NB |
858 | && _cpp_handle_directive (pfile, result->flags & PREV_WHITE)) |
859 | continue; | |
97293897 NB |
860 | if (pfile->cb.line_change && !pfile->state.skipping) |
861 | (*pfile->cb.line_change)(pfile, result, pfile->state.parsing_args); | |
5fddcffc | 862 | } |
5fddcffc | 863 | |
bdcbe496 NB |
864 | /* We don't skip tokens in directives. */ |
865 | if (pfile->state.in_directive) | |
866 | break; | |
5fddcffc | 867 | |
bdcbe496 | 868 | /* Outside a directive, invalidate controlling macros. At file |
14baae01 | 869 | EOF, _cpp_lex_direct takes care of popping the buffer, so we never |
bdcbe496 | 870 | get here and MI optimisation works. */ |
5fddcffc | 871 | pfile->mi_valid = false; |
bdcbe496 NB |
872 | |
873 | if (!pfile->state.skipping || result->type == CPP_EOF) | |
874 | break; | |
5fddcffc NB |
875 | } |
876 | ||
345894b4 | 877 | return result; |
5fddcffc NB |
878 | } |
879 | ||
480709cc NB |
880 | #define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \ |
881 | do { \ | |
882 | if (get_effective_char (pfile) == CHAR) \ | |
883 | result->type = THEN_TYPE; \ | |
884 | else \ | |
885 | { \ | |
886 | BACKUP (); \ | |
887 | result->type = ELSE_TYPE; \ | |
888 | } \ | |
889 | } while (0) | |
890 | ||
14baae01 NB |
891 | /* Lex a token into pfile->cur_token, which is also incremented, to |
892 | get diagnostics pointing to the correct location. | |
893 | ||
894 | Does not handle issues such as token lookahead, multiple-include | |
895 | optimisation, directives, skipping etc. This function is only | |
896 | suitable for use by _cpp_lex_token, and in special cases like | |
897 | lex_expansion_token which doesn't care for any of these issues. | |
898 | ||
899 | When meeting a newline, returns CPP_EOF if parsing a directive, | |
900 | otherwise returns to the start of the token buffer if permissible. | |
901 | Returns the location of the lexed token. */ | |
902 | cpp_token * | |
903 | _cpp_lex_direct (pfile) | |
5fddcffc | 904 | cpp_reader *pfile; |
45b966db | 905 | { |
0d9f234d | 906 | cppchar_t c; |
adb84b42 | 907 | cpp_buffer *buffer; |
0d9f234d | 908 | const unsigned char *comment_start; |
14baae01 | 909 | cpp_token *result = pfile->cur_token++; |
9ec7291f | 910 | |
5fddcffc | 911 | fresh_line: |
adb84b42 | 912 | buffer = pfile->buffer; |
bd969772 NB |
913 | result->flags = buffer->saved_flags; |
914 | buffer->saved_flags = 0; | |
5fddcffc | 915 | update_tokens_line: |
1444f2ed | 916 | result->line = pfile->line; |
041c3194 | 917 | |
5fddcffc | 918 | skipped_white: |
480709cc | 919 | c = *buffer->cur++; |
5fddcffc | 920 | result->col = CPP_BUF_COLUMN (buffer, buffer->cur); |
5fddcffc NB |
921 | |
922 | trigraph: | |
0d9f234d | 923 | switch (c) |
45b966db | 924 | { |
4d6baafa NB |
925 | case ' ': case '\t': case '\f': case '\v': case '\0': |
926 | result->flags |= PREV_WHITE; | |
927 | if (skip_whitespace (pfile, c)) | |
928 | goto skipped_white; | |
929 | ||
930 | /* EOF. */ | |
931 | buffer->cur--; | |
bdcbe496 | 932 | buffer->saved_flags = BOL; |
387f9e32 | 933 | if (!pfile->state.parsing_args) |
ef6e958a | 934 | { |
bdcbe496 | 935 | if (buffer->cur != buffer->line_base) |
5fddcffc NB |
936 | { |
937 | /* Non-empty files should end in a newline. Don't warn | |
938 | for command line and _Pragma buffers. */ | |
939 | if (!buffer->from_stage3) | |
ebef4e8c | 940 | cpp_error (pfile, DL_PEDWARN, "no newline at end of file"); |
87062813 | 941 | handle_newline (pfile); |
7364fdd8 | 942 | } |
bdcbe496 NB |
943 | |
944 | /* Don't pop the last buffer. */ | |
387f9e32 | 945 | if (!pfile->state.in_directive && buffer->prev) |
bdcbe496 NB |
946 | { |
947 | unsigned char stop = buffer->return_at_eof; | |
948 | ||
949 | _cpp_pop_buffer (pfile); | |
950 | if (!stop) | |
951 | goto fresh_line; | |
952 | } | |
ef6e958a | 953 | } |
0d9f234d | 954 | result->type = CPP_EOF; |
5fddcffc | 955 | break; |
45b966db | 956 | |
0d9f234d | 957 | case '\n': case '\r': |
87062813 | 958 | handle_newline (pfile); |
bdcbe496 NB |
959 | buffer->saved_flags = BOL; |
960 | if (! pfile->state.in_directive) | |
45b966db | 961 | { |
4ed5bcfb NB |
962 | if (pfile->state.parsing_args == 2) |
963 | buffer->saved_flags |= PREV_WHITE; | |
bdcbe496 NB |
964 | if (!pfile->keep_tokens) |
965 | { | |
966 | pfile->cur_run = &pfile->base_run; | |
967 | result = pfile->base_run.base; | |
968 | pfile->cur_token = result + 1; | |
969 | } | |
970 | goto fresh_line; | |
45b966db | 971 | } |
5fddcffc NB |
972 | result->type = CPP_EOF; |
973 | break; | |
46d07497 | 974 | |
0d9f234d NB |
975 | case '?': |
976 | case '\\': | |
977 | /* These could start an escaped newline, or '?' a trigraph. Let | |
978 | skip_escaped_newlines do all the work. */ | |
979 | { | |
67821e3a | 980 | unsigned int line = pfile->line; |
0d9f234d | 981 | |
87062813 | 982 | c = skip_escaped_newlines (pfile); |
67821e3a | 983 | if (line != pfile->line) |
87062813 | 984 | { |
480709cc | 985 | buffer->cur--; |
87062813 NB |
986 | /* We had at least one escaped newline of some sort. |
987 | Update the token's line and column. */ | |
5fddcffc | 988 | goto update_tokens_line; |
87062813 | 989 | } |
480709cc | 990 | } |
0d9f234d | 991 | |
480709cc NB |
992 | /* We are either the original '?' or '\\', or a trigraph. */ |
993 | if (c == '?') | |
0d9f234d | 994 | result->type = CPP_QUERY; |
480709cc NB |
995 | else if (c == '\\') |
996 | goto random_char; | |
997 | else | |
998 | goto trigraph; | |
0d9f234d | 999 | break; |
46d07497 | 1000 | |
0d9f234d NB |
1001 | case '0': case '1': case '2': case '3': case '4': |
1002 | case '5': case '6': case '7': case '8': case '9': | |
1003 | result->type = CPP_NUMBER; | |
10cf9bde | 1004 | parse_number (pfile, &result->val.str, 0); |
0d9f234d | 1005 | break; |
46d07497 | 1006 | |
0abc6a6a NB |
1007 | case 'L': |
1008 | /* 'L' may introduce wide characters or strings. */ | |
1009 | { | |
1010 | const unsigned char *pos = buffer->cur; | |
0d9f234d | 1011 | |
0abc6a6a NB |
1012 | c = get_effective_char (pfile); |
1013 | if (c == '\'' || c == '"') | |
1014 | { | |
1015 | result->type = (c == '"' ? CPP_WSTRING: CPP_WCHAR); | |
1016 | parse_string (pfile, result, c); | |
1017 | break; | |
1018 | } | |
1019 | buffer->cur = pos; | |
1020 | } | |
1021 | /* Fall through. */ | |
1022 | ||
1023 | start_ident: | |
0d9f234d NB |
1024 | case '_': |
1025 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
1026 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
1027 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
1028 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
1029 | case 'y': case 'z': | |
1030 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
0abc6a6a | 1031 | case 'G': case 'H': case 'I': case 'J': case 'K': |
0d9f234d NB |
1032 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': |
1033 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1034 | case 'Y': case 'Z': | |
1035 | result->type = CPP_NAME; | |
2c3fcba6 | 1036 | result->val.node = parse_identifier (pfile); |
0d9f234d | 1037 | |
0d9f234d | 1038 | /* Convert named operators to their proper types. */ |
0abc6a6a | 1039 | if (result->val.node->flags & NODE_OPERATOR) |
0d9f234d NB |
1040 | { |
1041 | result->flags |= NAMED_OP; | |
93c80368 | 1042 | result->type = result->val.node->value.operator; |
0d9f234d NB |
1043 | } |
1044 | break; | |
1045 | ||
1046 | case '\'': | |
1047 | case '"': | |
1048 | result->type = c == '"' ? CPP_STRING: CPP_CHAR; | |
0d9f234d NB |
1049 | parse_string (pfile, result, c); |
1050 | break; | |
041c3194 | 1051 | |
0d9f234d | 1052 | case '/': |
1c6d33ef NB |
1053 | /* A potential block or line comment. */ |
1054 | comment_start = buffer->cur; | |
29401c30 | 1055 | c = get_effective_char (pfile); |
480709cc | 1056 | |
1c6d33ef NB |
1057 | if (c == '*') |
1058 | { | |
0d9f234d | 1059 | if (skip_block_comment (pfile)) |
ebef4e8c | 1060 | cpp_error (pfile, DL_ERROR, "unterminated comment"); |
0d9f234d | 1061 | } |
480709cc NB |
1062 | else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments) |
1063 | || CPP_IN_SYSTEM_HEADER (pfile))) | |
0d9f234d | 1064 | { |
bdb05a7b NB |
1065 | /* Warn about comments only if pedantically GNUC89, and not |
1066 | in system headers. */ | |
1067 | if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile) | |
a94c1199 | 1068 | && ! buffer->warned_cplusplus_comments) |
041c3194 | 1069 | { |
ebef4e8c | 1070 | cpp_error (pfile, DL_PEDWARN, |
1c6d33ef | 1071 | "C++ style comments are not allowed in ISO C89"); |
ebef4e8c NB |
1072 | cpp_error (pfile, DL_PEDWARN, |
1073 | "(this will be reported only once per input file)"); | |
1c6d33ef NB |
1074 | buffer->warned_cplusplus_comments = 1; |
1075 | } | |
0d9f234d | 1076 | |
01ef6563 | 1077 | if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments)) |
ebef4e8c | 1078 | cpp_error (pfile, DL_WARNING, "multi-line comment"); |
1c6d33ef | 1079 | } |
480709cc NB |
1080 | else if (c == '=') |
1081 | { | |
1082 | result->type = CPP_DIV_EQ; | |
1083 | break; | |
1084 | } | |
1085 | else | |
1086 | { | |
1087 | BACKUP (); | |
1088 | result->type = CPP_DIV; | |
1089 | break; | |
1090 | } | |
0d9f234d | 1091 | |
1c6d33ef NB |
1092 | if (!pfile->state.save_comments) |
1093 | { | |
1094 | result->flags |= PREV_WHITE; | |
5fddcffc | 1095 | goto update_tokens_line; |
0d9f234d | 1096 | } |
1c6d33ef NB |
1097 | |
1098 | /* Save the comment as a token in its own right. */ | |
477cdac7 | 1099 | save_comment (pfile, result, comment_start, c); |
bdcbe496 | 1100 | break; |
0d9f234d NB |
1101 | |
1102 | case '<': | |
1103 | if (pfile->state.angled_headers) | |
1104 | { | |
1105 | result->type = CPP_HEADER_NAME; | |
480709cc NB |
1106 | parse_string (pfile, result, '>'); |
1107 | break; | |
0d9f234d | 1108 | } |
45b966db | 1109 | |
29401c30 | 1110 | c = get_effective_char (pfile); |
0d9f234d | 1111 | if (c == '=') |
480709cc | 1112 | result->type = CPP_LESS_EQ; |
0d9f234d | 1113 | else if (c == '<') |
480709cc | 1114 | IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT); |
0d9f234d | 1115 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1116 | IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN); |
0d9f234d NB |
1117 | else if (c == ':' && CPP_OPTION (pfile, digraphs)) |
1118 | { | |
480709cc | 1119 | result->type = CPP_OPEN_SQUARE; |
0d9f234d NB |
1120 | result->flags |= DIGRAPH; |
1121 | } | |
1122 | else if (c == '%' && CPP_OPTION (pfile, digraphs)) | |
1123 | { | |
480709cc | 1124 | result->type = CPP_OPEN_BRACE; |
0d9f234d NB |
1125 | result->flags |= DIGRAPH; |
1126 | } | |
480709cc NB |
1127 | else |
1128 | { | |
1129 | BACKUP (); | |
1130 | result->type = CPP_LESS; | |
1131 | } | |
0d9f234d NB |
1132 | break; |
1133 | ||
1134 | case '>': | |
29401c30 | 1135 | c = get_effective_char (pfile); |
0d9f234d | 1136 | if (c == '=') |
480709cc | 1137 | result->type = CPP_GREATER_EQ; |
0d9f234d | 1138 | else if (c == '>') |
480709cc | 1139 | IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT); |
0d9f234d | 1140 | else if (c == '?' && CPP_OPTION (pfile, cplusplus)) |
480709cc NB |
1141 | IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX); |
1142 | else | |
0d9f234d | 1143 | { |
480709cc NB |
1144 | BACKUP (); |
1145 | result->type = CPP_GREATER; | |
0d9f234d NB |
1146 | } |
1147 | break; | |
1148 | ||
cbcff6df | 1149 | case '%': |
480709cc NB |
1150 | c = get_effective_char (pfile); |
1151 | if (c == '=') | |
1152 | result->type = CPP_MOD_EQ; | |
1153 | else if (CPP_OPTION (pfile, digraphs) && c == ':') | |
1154 | { | |
1155 | result->flags |= DIGRAPH; | |
1156 | result->type = CPP_HASH; | |
1157 | if (get_effective_char (pfile) == '%') | |
1158 | { | |
1159 | const unsigned char *pos = buffer->cur; | |
1160 | ||
1161 | if (get_effective_char (pfile) == ':') | |
1162 | result->type = CPP_PASTE; | |
1163 | else | |
1164 | buffer->cur = pos - 1; | |
1165 | } | |
1166 | else | |
1167 | BACKUP (); | |
1168 | } | |
1169 | else if (CPP_OPTION (pfile, digraphs) && c == '>') | |
1170 | { | |
1171 | result->flags |= DIGRAPH; | |
1172 | result->type = CPP_CLOSE_BRACE; | |
1173 | } | |
1174 | else | |
1175 | { | |
1176 | BACKUP (); | |
1177 | result->type = CPP_MOD; | |
1178 | } | |
0d9f234d NB |
1179 | break; |
1180 | ||
cbcff6df | 1181 | case '.': |
480709cc NB |
1182 | result->type = CPP_DOT; |
1183 | c = get_effective_char (pfile); | |
1184 | if (c == '.') | |
1185 | { | |
1186 | const unsigned char *pos = buffer->cur; | |
1187 | ||
1188 | if (get_effective_char (pfile) == '.') | |
1189 | result->type = CPP_ELLIPSIS; | |
1190 | else | |
1191 | buffer->cur = pos - 1; | |
1192 | } | |
1193 | /* All known character sets have 0...9 contiguous. */ | |
0df6c2c7 | 1194 | else if (ISDIGIT (c)) |
480709cc NB |
1195 | { |
1196 | result->type = CPP_NUMBER; | |
10cf9bde | 1197 | parse_number (pfile, &result->val.str, 1); |
480709cc NB |
1198 | } |
1199 | else if (c == '*' && CPP_OPTION (pfile, cplusplus)) | |
1200 | result->type = CPP_DOT_STAR; | |
1201 | else | |
1202 | BACKUP (); | |
0d9f234d | 1203 | break; |
45b966db | 1204 | |
0d9f234d | 1205 | case '+': |
29401c30 | 1206 | c = get_effective_char (pfile); |
480709cc NB |
1207 | if (c == '+') |
1208 | result->type = CPP_PLUS_PLUS; | |
1209 | else if (c == '=') | |
1210 | result->type = CPP_PLUS_EQ; | |
1211 | else | |
1212 | { | |
1213 | BACKUP (); | |
1214 | result->type = CPP_PLUS; | |
1215 | } | |
0d9f234d | 1216 | break; |
04e3ec78 | 1217 | |
0d9f234d | 1218 | case '-': |
29401c30 | 1219 | c = get_effective_char (pfile); |
0d9f234d NB |
1220 | if (c == '>') |
1221 | { | |
480709cc NB |
1222 | result->type = CPP_DEREF; |
1223 | if (CPP_OPTION (pfile, cplusplus)) | |
1224 | { | |
1225 | if (get_effective_char (pfile) == '*') | |
1226 | result->type = CPP_DEREF_STAR; | |
1227 | else | |
1228 | BACKUP (); | |
1229 | } | |
0d9f234d | 1230 | } |
0d9f234d | 1231 | else if (c == '-') |
480709cc NB |
1232 | result->type = CPP_MINUS_MINUS; |
1233 | else if (c == '=') | |
1234 | result->type = CPP_MINUS_EQ; | |
1235 | else | |
1236 | { | |
1237 | BACKUP (); | |
1238 | result->type = CPP_MINUS; | |
1239 | } | |
0d9f234d | 1240 | break; |
45b966db | 1241 | |
0d9f234d | 1242 | case '&': |
29401c30 | 1243 | c = get_effective_char (pfile); |
480709cc NB |
1244 | if (c == '&') |
1245 | result->type = CPP_AND_AND; | |
1246 | else if (c == '=') | |
1247 | result->type = CPP_AND_EQ; | |
1248 | else | |
1249 | { | |
1250 | BACKUP (); | |
1251 | result->type = CPP_AND; | |
1252 | } | |
0d9f234d NB |
1253 | break; |
1254 | ||
0d9f234d | 1255 | case '|': |
29401c30 | 1256 | c = get_effective_char (pfile); |
480709cc NB |
1257 | if (c == '|') |
1258 | result->type = CPP_OR_OR; | |
1259 | else if (c == '=') | |
1260 | result->type = CPP_OR_EQ; | |
1261 | else | |
1262 | { | |
1263 | BACKUP (); | |
1264 | result->type = CPP_OR; | |
1265 | } | |
0d9f234d | 1266 | break; |
45b966db | 1267 | |
0d9f234d | 1268 | case ':': |
29401c30 | 1269 | c = get_effective_char (pfile); |
0d9f234d | 1270 | if (c == ':' && CPP_OPTION (pfile, cplusplus)) |
480709cc | 1271 | result->type = CPP_SCOPE; |
0d9f234d NB |
1272 | else if (c == '>' && CPP_OPTION (pfile, digraphs)) |
1273 | { | |
1274 | result->flags |= DIGRAPH; | |
480709cc NB |
1275 | result->type = CPP_CLOSE_SQUARE; |
1276 | } | |
1277 | else | |
1278 | { | |
1279 | BACKUP (); | |
1280 | result->type = CPP_COLON; | |
0d9f234d NB |
1281 | } |
1282 | break; | |
45b966db | 1283 | |
480709cc NB |
1284 | case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break; |
1285 | case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break; | |
1286 | case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break; | |
1287 | case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break; | |
1288 | case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break; | |
1289 | ||
0d9f234d NB |
1290 | case '~': result->type = CPP_COMPL; break; |
1291 | case ',': result->type = CPP_COMMA; break; | |
1292 | case '(': result->type = CPP_OPEN_PAREN; break; | |
1293 | case ')': result->type = CPP_CLOSE_PAREN; break; | |
1294 | case '[': result->type = CPP_OPEN_SQUARE; break; | |
1295 | case ']': result->type = CPP_CLOSE_SQUARE; break; | |
1296 | case '{': result->type = CPP_OPEN_BRACE; break; | |
1297 | case '}': result->type = CPP_CLOSE_BRACE; break; | |
1298 | case ';': result->type = CPP_SEMICOLON; break; | |
1299 | ||
cc937581 ZW |
1300 | /* @ is a punctuator in Objective C. */ |
1301 | case '@': result->type = CPP_ATSIGN; break; | |
0d9f234d | 1302 | |
0abc6a6a NB |
1303 | case '$': |
1304 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
1305 | goto start_ident; | |
1306 | /* Fall through... */ | |
1307 | ||
0d9f234d NB |
1308 | random_char: |
1309 | default: | |
1310 | result->type = CPP_OTHER; | |
6c53ebff | 1311 | result->val.c = c; |
0d9f234d NB |
1312 | break; |
1313 | } | |
bdcbe496 NB |
1314 | |
1315 | return result; | |
0d9f234d NB |
1316 | } |
1317 | ||
5d8ebbd8 | 1318 | /* An upper bound on the number of bytes needed to spell TOKEN, |
93c80368 NB |
1319 | including preceding whitespace. */ |
1320 | unsigned int | |
1321 | cpp_token_len (token) | |
1322 | const cpp_token *token; | |
0d9f234d | 1323 | { |
93c80368 | 1324 | unsigned int len; |
6d2c2047 | 1325 | |
93c80368 | 1326 | switch (TOKEN_SPELL (token)) |
041c3194 | 1327 | { |
a28c5035 | 1328 | default: len = 0; break; |
47ad4138 | 1329 | case SPELL_NUMBER: |
a28c5035 NB |
1330 | case SPELL_STRING: len = token->val.str.len; break; |
1331 | case SPELL_IDENT: len = NODE_LEN (token->val.node); break; | |
041c3194 | 1332 | } |
47ad4138 | 1333 | /* 1 for whitespace, 4 for comment delimiters. */ |
93c80368 | 1334 | return len + 5; |
6d2c2047 ZW |
1335 | } |
1336 | ||
041c3194 | 1337 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
cf00a885 ZW |
1338 | already contain the enough space to hold the token's spelling. |
1339 | Returns a pointer to the character after the last character | |
1340 | written. */ | |
93c80368 NB |
1341 | unsigned char * |
1342 | cpp_spell_token (pfile, token, buffer) | |
041c3194 ZW |
1343 | cpp_reader *pfile; /* Would be nice to be rid of this... */ |
1344 | const cpp_token *token; | |
1345 | unsigned char *buffer; | |
1346 | { | |
96be6998 | 1347 | switch (TOKEN_SPELL (token)) |
041c3194 ZW |
1348 | { |
1349 | case SPELL_OPERATOR: | |
1350 | { | |
1351 | const unsigned char *spelling; | |
1352 | unsigned char c; | |
d6d5f795 | 1353 | |
041c3194 | 1354 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1355 | spelling |
1356 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
92936ecf ZW |
1357 | else if (token->flags & NAMED_OP) |
1358 | goto spell_ident; | |
041c3194 | 1359 | else |
96be6998 | 1360 | spelling = TOKEN_NAME (token); |
041c3194 ZW |
1361 | |
1362 | while ((c = *spelling++) != '\0') | |
1363 | *buffer++ = c; | |
1364 | } | |
1365 | break; | |
d6d5f795 | 1366 | |
47ad4138 ZW |
1367 | case SPELL_CHAR: |
1368 | *buffer++ = token->val.c; | |
1369 | break; | |
1370 | ||
1371 | spell_ident: | |
041c3194 | 1372 | case SPELL_IDENT: |
a28c5035 NB |
1373 | memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node)); |
1374 | buffer += NODE_LEN (token->val.node); | |
041c3194 | 1375 | break; |
d6d5f795 | 1376 | |
47ad4138 ZW |
1377 | case SPELL_NUMBER: |
1378 | memcpy (buffer, token->val.str.text, token->val.str.len); | |
1379 | buffer += token->val.str.len; | |
1380 | break; | |
1381 | ||
041c3194 ZW |
1382 | case SPELL_STRING: |
1383 | { | |
ba89d661 ZW |
1384 | int left, right, tag; |
1385 | switch (token->type) | |
1386 | { | |
1387 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1388 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
ba89d661 ZW |
1389 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1390 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1391 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 | 1392 | default: |
ebef4e8c NB |
1393 | cpp_error (pfile, DL_ICE, "unknown string token %s\n", |
1394 | TOKEN_NAME (token)); | |
47ad4138 | 1395 | return buffer; |
ba89d661 ZW |
1396 | } |
1397 | if (tag) *buffer++ = tag; | |
47ad4138 | 1398 | *buffer++ = left; |
bfb9dc7f ZW |
1399 | memcpy (buffer, token->val.str.text, token->val.str.len); |
1400 | buffer += token->val.str.len; | |
47ad4138 | 1401 | *buffer++ = right; |
041c3194 ZW |
1402 | } |
1403 | break; | |
d6d5f795 | 1404 | |
041c3194 | 1405 | case SPELL_NONE: |
ebef4e8c | 1406 | cpp_error (pfile, DL_ICE, "unspellable token %s", TOKEN_NAME (token)); |
041c3194 ZW |
1407 | break; |
1408 | } | |
d6d5f795 | 1409 | |
041c3194 ZW |
1410 | return buffer; |
1411 | } | |
d6d5f795 | 1412 | |
5d8ebbd8 NB |
1413 | /* Returns TOKEN spelt as a null-terminated string. The string is |
1414 | freed when the reader is destroyed. Useful for diagnostics. */ | |
93c80368 NB |
1415 | unsigned char * |
1416 | cpp_token_as_text (pfile, token) | |
c5a04734 | 1417 | cpp_reader *pfile; |
041c3194 | 1418 | const cpp_token *token; |
c5a04734 | 1419 | { |
93c80368 | 1420 | unsigned int len = cpp_token_len (token); |
ece54d54 | 1421 | unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end; |
c5a04734 | 1422 | |
93c80368 NB |
1423 | end = cpp_spell_token (pfile, token, start); |
1424 | end[0] = '\0'; | |
c5a04734 | 1425 | |
93c80368 NB |
1426 | return start; |
1427 | } | |
c5a04734 | 1428 | |
5d8ebbd8 NB |
1429 | /* Used by C front ends, which really should move to using |
1430 | cpp_token_as_text. */ | |
93c80368 NB |
1431 | const char * |
1432 | cpp_type2name (type) | |
1433 | enum cpp_ttype type; | |
1434 | { | |
1435 | return (const char *) token_spellings[type].name; | |
1436 | } | |
c5a04734 | 1437 | |
4ed5bcfb NB |
1438 | /* Writes the spelling of token to FP, without any preceding space. |
1439 | Separated from cpp_spell_token for efficiency - to avoid stdio | |
1440 | double-buffering. */ | |
93c80368 NB |
1441 | void |
1442 | cpp_output_token (token, fp) | |
1443 | const cpp_token *token; | |
1444 | FILE *fp; | |
1445 | { | |
93c80368 | 1446 | switch (TOKEN_SPELL (token)) |
c5a04734 | 1447 | { |
93c80368 NB |
1448 | case SPELL_OPERATOR: |
1449 | { | |
1450 | const unsigned char *spelling; | |
3b681e9d | 1451 | int c; |
c5a04734 | 1452 | |
93c80368 | 1453 | if (token->flags & DIGRAPH) |
37b8524c JDA |
1454 | spelling |
1455 | = digraph_spellings[(int) token->type - (int) CPP_FIRST_DIGRAPH]; | |
93c80368 NB |
1456 | else if (token->flags & NAMED_OP) |
1457 | goto spell_ident; | |
1458 | else | |
1459 | spelling = TOKEN_NAME (token); | |
041c3194 | 1460 | |
3b681e9d ZW |
1461 | c = *spelling; |
1462 | do | |
1463 | putc (c, fp); | |
1464 | while ((c = *++spelling) != '\0'); | |
93c80368 NB |
1465 | } |
1466 | break; | |
041c3194 | 1467 | |
47ad4138 ZW |
1468 | case SPELL_CHAR: |
1469 | putc (token->val.c, fp); | |
1470 | break; | |
1471 | ||
93c80368 NB |
1472 | spell_ident: |
1473 | case SPELL_IDENT: | |
3b681e9d | 1474 | fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp); |
93c80368 | 1475 | break; |
041c3194 | 1476 | |
47ad4138 ZW |
1477 | case SPELL_NUMBER: |
1478 | fwrite (token->val.str.text, 1, token->val.str.len, fp); | |
1479 | break; | |
1480 | ||
93c80368 NB |
1481 | case SPELL_STRING: |
1482 | { | |
1483 | int left, right, tag; | |
1484 | switch (token->type) | |
1485 | { | |
1486 | case CPP_STRING: left = '"'; right = '"'; tag = '\0'; break; | |
1487 | case CPP_WSTRING: left = '"'; right = '"'; tag = 'L'; break; | |
93c80368 NB |
1488 | case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break; |
1489 | case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break; | |
1490 | case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break; | |
47ad4138 ZW |
1491 | default: |
1492 | fprintf (stderr, "impossible STRING token %s\n", TOKEN_NAME (token)); | |
1493 | return; | |
93c80368 NB |
1494 | } |
1495 | if (tag) putc (tag, fp); | |
47ad4138 | 1496 | putc (left, fp); |
93c80368 | 1497 | fwrite (token->val.str.text, 1, token->val.str.len, fp); |
47ad4138 | 1498 | putc (right, fp); |
93c80368 NB |
1499 | } |
1500 | break; | |
c5a04734 | 1501 | |
93c80368 NB |
1502 | case SPELL_NONE: |
1503 | /* An error, most probably. */ | |
1504 | break; | |
041c3194 | 1505 | } |
c5a04734 ZW |
1506 | } |
1507 | ||
93c80368 NB |
1508 | /* Compare two tokens. */ |
1509 | int | |
1510 | _cpp_equiv_tokens (a, b) | |
1511 | const cpp_token *a, *b; | |
c5a04734 | 1512 | { |
93c80368 NB |
1513 | if (a->type == b->type && a->flags == b->flags) |
1514 | switch (TOKEN_SPELL (a)) | |
1515 | { | |
1516 | default: /* Keep compiler happy. */ | |
1517 | case SPELL_OPERATOR: | |
1518 | return 1; | |
1519 | case SPELL_CHAR: | |
6c53ebff | 1520 | return a->val.c == b->val.c; /* Character. */ |
93c80368 | 1521 | case SPELL_NONE: |
56051c0a | 1522 | return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no); |
93c80368 NB |
1523 | case SPELL_IDENT: |
1524 | return a->val.node == b->val.node; | |
47ad4138 | 1525 | case SPELL_NUMBER: |
93c80368 NB |
1526 | case SPELL_STRING: |
1527 | return (a->val.str.len == b->val.str.len | |
1528 | && !memcmp (a->val.str.text, b->val.str.text, | |
1529 | a->val.str.len)); | |
1530 | } | |
c5a04734 | 1531 | |
041c3194 ZW |
1532 | return 0; |
1533 | } | |
1534 | ||
93c80368 NB |
1535 | /* Returns nonzero if a space should be inserted to avoid an |
1536 | accidental token paste for output. For simplicity, it is | |
1537 | conservative, and occasionally advises a space where one is not | |
1538 | needed, e.g. "." and ".2". */ | |
93c80368 NB |
1539 | int |
1540 | cpp_avoid_paste (pfile, token1, token2) | |
c5a04734 | 1541 | cpp_reader *pfile; |
93c80368 | 1542 | const cpp_token *token1, *token2; |
c5a04734 | 1543 | { |
93c80368 NB |
1544 | enum cpp_ttype a = token1->type, b = token2->type; |
1545 | cppchar_t c; | |
c5a04734 | 1546 | |
93c80368 NB |
1547 | if (token1->flags & NAMED_OP) |
1548 | a = CPP_NAME; | |
1549 | if (token2->flags & NAMED_OP) | |
1550 | b = CPP_NAME; | |
c5a04734 | 1551 | |
93c80368 NB |
1552 | c = EOF; |
1553 | if (token2->flags & DIGRAPH) | |
37b8524c | 1554 | c = digraph_spellings[(int) b - (int) CPP_FIRST_DIGRAPH][0]; |
93c80368 NB |
1555 | else if (token_spellings[b].category == SPELL_OPERATOR) |
1556 | c = token_spellings[b].name[0]; | |
c5a04734 | 1557 | |
93c80368 | 1558 | /* Quickly get everything that can paste with an '='. */ |
37b8524c | 1559 | if ((int) a <= (int) CPP_LAST_EQ && c == '=') |
93c80368 | 1560 | return 1; |
c5a04734 | 1561 | |
93c80368 | 1562 | switch (a) |
c5a04734 | 1563 | { |
93c80368 NB |
1564 | case CPP_GREATER: return c == '>' || c == '?'; |
1565 | case CPP_LESS: return c == '<' || c == '?' || c == '%' || c == ':'; | |
1566 | case CPP_PLUS: return c == '+'; | |
1567 | case CPP_MINUS: return c == '-' || c == '>'; | |
1568 | case CPP_DIV: return c == '/' || c == '*'; /* Comments. */ | |
1569 | case CPP_MOD: return c == ':' || c == '>'; | |
1570 | case CPP_AND: return c == '&'; | |
1571 | case CPP_OR: return c == '|'; | |
1572 | case CPP_COLON: return c == ':' || c == '>'; | |
1573 | case CPP_DEREF: return c == '*'; | |
26ec42ee | 1574 | case CPP_DOT: return c == '.' || c == '%' || b == CPP_NUMBER; |
93c80368 NB |
1575 | case CPP_HASH: return c == '#' || c == '%'; /* Digraph form. */ |
1576 | case CPP_NAME: return ((b == CPP_NUMBER | |
1577 | && name_p (pfile, &token2->val.str)) | |
1578 | || b == CPP_NAME | |
1579 | || b == CPP_CHAR || b == CPP_STRING); /* L */ | |
1580 | case CPP_NUMBER: return (b == CPP_NUMBER || b == CPP_NAME | |
1581 | || c == '.' || c == '+' || c == '-'); | |
1582 | case CPP_OTHER: return (CPP_OPTION (pfile, objc) | |
6c53ebff | 1583 | && token1->val.c == '@' |
93c80368 NB |
1584 | && (b == CPP_NAME || b == CPP_STRING)); |
1585 | default: break; | |
c5a04734 | 1586 | } |
c5a04734 | 1587 | |
417f3e3a | 1588 | return 0; |
c5a04734 ZW |
1589 | } |
1590 | ||
93c80368 | 1591 | /* Output all the remaining tokens on the current line, and a newline |
4ed5bcfb NB |
1592 | character, to FP. Leading whitespace is removed. If there are |
1593 | macros, special token padding is not performed. */ | |
c5a04734 | 1594 | void |
93c80368 | 1595 | cpp_output_line (pfile, fp) |
c5a04734 | 1596 | cpp_reader *pfile; |
93c80368 | 1597 | FILE *fp; |
c5a04734 | 1598 | { |
4ed5bcfb | 1599 | const cpp_token *token; |
96be6998 | 1600 | |
4ed5bcfb NB |
1601 | token = cpp_get_token (pfile); |
1602 | while (token->type != CPP_EOF) | |
96be6998 | 1603 | { |
4ed5bcfb NB |
1604 | cpp_output_token (token, fp); |
1605 | token = cpp_get_token (pfile); | |
1606 | if (token->flags & PREV_WHITE) | |
1607 | putc (' ', fp); | |
96be6998 ZW |
1608 | } |
1609 | ||
93c80368 | 1610 | putc ('\n', fp); |
041c3194 | 1611 | } |
c5a04734 | 1612 | |
c8a96070 NB |
1613 | /* Returns the value of a hexadecimal digit. */ |
1614 | static unsigned int | |
1615 | hex_digit_value (c) | |
1616 | unsigned int c; | |
1617 | { | |
9e1ac915 KG |
1618 | if (hex_p (c)) |
1619 | return hex_value (c); | |
1620 | else | |
1621 | abort (); | |
c8a96070 NB |
1622 | } |
1623 | ||
62729350 NB |
1624 | /* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. Returns 1 to indicate |
1625 | failure if cpplib is not parsing C++ or C99. Such failure is | |
1626 | silent, and no variables are updated. Otherwise returns 0, and | |
1627 | warns if -Wtraditional. | |
c8a96070 NB |
1628 | |
1629 | [lex.charset]: The character designated by the universal character | |
1630 | name \UNNNNNNNN is that character whose character short name in | |
1631 | ISO/IEC 10646 is NNNNNNNN; the character designated by the | |
1632 | universal character name \uNNNN is that character whose character | |
1633 | short name in ISO/IEC 10646 is 0000NNNN. If the hexadecimal value | |
1634 | for a universal character name is less than 0x20 or in the range | |
1635 | 0x7F-0x9F (inclusive), or if the universal character name | |
1636 | designates a character in the basic source character set, then the | |
1637 | program is ill-formed. | |
1638 | ||
1639 | We assume that wchar_t is Unicode, so we don't need to do any | |
62729350 | 1640 | mapping. Is this ever wrong? |
c8a96070 | 1641 | |
62729350 NB |
1642 | PC points to the 'u' or 'U', PSTR is points to the byte after PC, |
1643 | LIMIT is the end of the string or charconst. PSTR is updated to | |
1644 | point after the UCS on return, and the UCS is written into PC. */ | |
1645 | ||
1646 | static int | |
1647 | maybe_read_ucs (pfile, pstr, limit, pc) | |
c8a96070 NB |
1648 | cpp_reader *pfile; |
1649 | const unsigned char **pstr; | |
1650 | const unsigned char *limit; | |
625458d0 | 1651 | cppchar_t *pc; |
c8a96070 NB |
1652 | { |
1653 | const unsigned char *p = *pstr; | |
62729350 NB |
1654 | unsigned int code = 0; |
1655 | unsigned int c = *pc, length; | |
1656 | ||
1657 | /* Only attempt to interpret a UCS for C++ and C99. */ | |
1658 | if (! (CPP_OPTION (pfile, cplusplus) || CPP_OPTION (pfile, c99))) | |
1659 | return 1; | |
c8a96070 | 1660 | |
62729350 | 1661 | if (CPP_WTRADITIONAL (pfile)) |
ebef4e8c NB |
1662 | cpp_error (pfile, DL_WARNING, |
1663 | "the meaning of '\\%c' is different in traditional C", c); | |
c8a96070 | 1664 | |
f8710242 NB |
1665 | length = (c == 'u' ? 4: 8); |
1666 | ||
1667 | if ((size_t) (limit - p) < length) | |
1668 | { | |
ebef4e8c | 1669 | cpp_error (pfile, DL_ERROR, "incomplete universal-character-name"); |
f8710242 NB |
1670 | /* Skip to the end to avoid more diagnostics. */ |
1671 | p = limit; | |
1672 | } | |
1673 | else | |
1674 | { | |
1675 | for (; length; length--, p++) | |
c8a96070 | 1676 | { |
f8710242 NB |
1677 | c = *p; |
1678 | if (ISXDIGIT (c)) | |
1679 | code = (code << 4) + hex_digit_value (c); | |
1680 | else | |
1681 | { | |
ebef4e8c | 1682 | cpp_error (pfile, DL_ERROR, |
f8710242 NB |
1683 | "non-hex digit '%c' in universal-character-name", c); |
1684 | /* We shouldn't skip in case there are multibyte chars. */ | |
1685 | break; | |
1686 | } | |
c8a96070 | 1687 | } |
c8a96070 NB |
1688 | } |
1689 | ||
1690 | #ifdef TARGET_EBCDIC | |
ebef4e8c | 1691 | cpp_error (pfile, DL_ERROR, "universal-character-name on EBCDIC target"); |
c8a96070 NB |
1692 | code = 0x3f; /* EBCDIC invalid character */ |
1693 | #else | |
f8710242 NB |
1694 | /* True extended characters are OK. */ |
1695 | if (code >= 0xa0 | |
1696 | && !(code & 0x80000000) | |
1697 | && !(code >= 0xD800 && code <= 0xDFFF)) | |
1698 | ; | |
1699 | /* The standard permits $, @ and ` to be specified as UCNs. We use | |
1700 | hex escapes so that this also works with EBCDIC hosts. */ | |
1701 | else if (code == 0x24 || code == 0x40 || code == 0x60) | |
1702 | ; | |
1703 | /* Don't give another error if one occurred above. */ | |
1704 | else if (length == 0) | |
ebef4e8c | 1705 | cpp_error (pfile, DL_ERROR, "universal-character-name out of range"); |
c8a96070 NB |
1706 | #endif |
1707 | ||
1708 | *pstr = p; | |
62729350 NB |
1709 | *pc = code; |
1710 | return 0; | |
c8a96070 NB |
1711 | } |
1712 | ||
4268e8bb NB |
1713 | /* Returns the value of an escape sequence, truncated to the correct |
1714 | target precision. PSTR points to the input pointer, which is just | |
1715 | after the backslash. LIMIT is how much text we have. WIDE is true | |
1716 | if the escape sequence is part of a wide character constant or | |
1717 | string literal. Handles all relevant diagnostics. */ | |
1718 | cppchar_t | |
1719 | cpp_parse_escape (pfile, pstr, limit, wide) | |
c8a96070 NB |
1720 | cpp_reader *pfile; |
1721 | const unsigned char **pstr; | |
1722 | const unsigned char *limit; | |
4268e8bb | 1723 | int wide; |
c8a96070 NB |
1724 | { |
1725 | int unknown = 0; | |
1726 | const unsigned char *str = *pstr; | |
4268e8bb NB |
1727 | cppchar_t c, mask; |
1728 | unsigned int width; | |
1729 | ||
1730 | if (wide) | |
1731 | width = CPP_OPTION (pfile, wchar_precision); | |
1732 | else | |
1733 | width = CPP_OPTION (pfile, char_precision); | |
1734 | if (width < BITS_PER_CPPCHAR_T) | |
1735 | mask = ((cppchar_t) 1 << width) - 1; | |
1736 | else | |
1737 | mask = ~0; | |
c8a96070 | 1738 | |
4268e8bb | 1739 | c = *str++; |
c8a96070 NB |
1740 | switch (c) |
1741 | { | |
1742 | case '\\': case '\'': case '"': case '?': break; | |
1743 | case 'b': c = TARGET_BS; break; | |
1744 | case 'f': c = TARGET_FF; break; | |
1745 | case 'n': c = TARGET_NEWLINE; break; | |
1746 | case 'r': c = TARGET_CR; break; | |
1747 | case 't': c = TARGET_TAB; break; | |
1748 | case 'v': c = TARGET_VT; break; | |
1749 | ||
1750 | case '(': case '{': case '[': case '%': | |
1751 | /* '\(', etc, are used at beginning of line to avoid confusing Emacs. | |
1752 | '\%' is used to prevent SCCS from getting confused. */ | |
1753 | unknown = CPP_PEDANTIC (pfile); | |
1754 | break; | |
1755 | ||
1756 | case 'a': | |
1757 | if (CPP_WTRADITIONAL (pfile)) | |
ebef4e8c NB |
1758 | cpp_error (pfile, DL_WARNING, |
1759 | "the meaning of '\\a' is different in traditional C"); | |
001e3fee | 1760 | c = TARGET_BELL; |
c8a96070 NB |
1761 | break; |
1762 | ||
1763 | case 'e': case 'E': | |
1764 | if (CPP_PEDANTIC (pfile)) | |
ebef4e8c | 1765 | cpp_error (pfile, DL_PEDWARN, |
625458d0 | 1766 | "non-ISO-standard escape sequence, '\\%c'", (int) c); |
c8a96070 NB |
1767 | c = TARGET_ESC; |
1768 | break; | |
1769 | ||
c8a96070 | 1770 | case 'u': case 'U': |
62729350 | 1771 | unknown = maybe_read_ucs (pfile, &str, limit, &c); |
c8a96070 NB |
1772 | break; |
1773 | ||
1774 | case 'x': | |
1775 | if (CPP_WTRADITIONAL (pfile)) | |
ebef4e8c NB |
1776 | cpp_error (pfile, DL_WARNING, |
1777 | "the meaning of '\\x' is different in traditional C"); | |
c8a96070 | 1778 | |
c8a96070 | 1779 | { |
4268e8bb | 1780 | cppchar_t i = 0, overflow = 0; |
c8a96070 NB |
1781 | int digits_found = 0; |
1782 | ||
1783 | while (str < limit) | |
1784 | { | |
1785 | c = *str; | |
1786 | if (! ISXDIGIT (c)) | |
1787 | break; | |
1788 | str++; | |
1789 | overflow |= i ^ (i << 4 >> 4); | |
1790 | i = (i << 4) + hex_digit_value (c); | |
1791 | digits_found = 1; | |
1792 | } | |
1793 | ||
1794 | if (!digits_found) | |
ebef4e8c NB |
1795 | cpp_error (pfile, DL_ERROR, |
1796 | "\\x used with no following hex digits"); | |
c8a96070 NB |
1797 | |
1798 | if (overflow | (i != (i & mask))) | |
1799 | { | |
ebef4e8c NB |
1800 | cpp_error (pfile, DL_PEDWARN, |
1801 | "hex escape sequence out of range"); | |
c8a96070 NB |
1802 | i &= mask; |
1803 | } | |
1804 | c = i; | |
1805 | } | |
1806 | break; | |
1807 | ||
1808 | case '0': case '1': case '2': case '3': | |
1809 | case '4': case '5': case '6': case '7': | |
1810 | { | |
4268e8bb NB |
1811 | size_t count = 0; |
1812 | cppchar_t i = c - '0'; | |
c8a96070 NB |
1813 | |
1814 | while (str < limit && ++count < 3) | |
1815 | { | |
1816 | c = *str; | |
1817 | if (c < '0' || c > '7') | |
1818 | break; | |
1819 | str++; | |
1820 | i = (i << 3) + c - '0'; | |
1821 | } | |
1822 | ||
1823 | if (i != (i & mask)) | |
1824 | { | |
ebef4e8c NB |
1825 | cpp_error (pfile, DL_PEDWARN, |
1826 | "octal escape sequence out of range"); | |
c8a96070 NB |
1827 | i &= mask; |
1828 | } | |
1829 | c = i; | |
1830 | } | |
1831 | break; | |
1832 | ||
1833 | default: | |
1834 | unknown = 1; | |
1835 | break; | |
1836 | } | |
1837 | ||
1838 | if (unknown) | |
1839 | { | |
1840 | if (ISGRAPH (c)) | |
625458d0 NB |
1841 | cpp_error (pfile, DL_PEDWARN, |
1842 | "unknown escape sequence '\\%c'", (int) c); | |
c8a96070 | 1843 | else |
625458d0 NB |
1844 | cpp_error (pfile, DL_PEDWARN, |
1845 | "unknown escape sequence: '\\%03o'", (int) c); | |
c8a96070 NB |
1846 | } |
1847 | ||
62729350 | 1848 | if (c > mask) |
4268e8bb | 1849 | { |
639e8b0c | 1850 | cpp_error (pfile, DL_PEDWARN, "escape sequence out of range for its type"); |
4268e8bb NB |
1851 | c &= mask; |
1852 | } | |
62729350 | 1853 | |
c8a96070 NB |
1854 | *pstr = str; |
1855 | return c; | |
1856 | } | |
1857 | ||
c8a96070 | 1858 | /* Interpret a (possibly wide) character constant in TOKEN. |
4268e8bb NB |
1859 | WARN_MULTI warns about multi-character charconsts. PCHARS_SEEN |
1860 | points to a variable that is filled in with the number of | |
1861 | characters seen, and UNSIGNEDP to a variable that indicates whether | |
1862 | the result has signed type. */ | |
1863 | cppchar_t | |
a5a49440 | 1864 | cpp_interpret_charconst (pfile, token, pchars_seen, unsignedp) |
c8a96070 NB |
1865 | cpp_reader *pfile; |
1866 | const cpp_token *token; | |
c8a96070 | 1867 | unsigned int *pchars_seen; |
4268e8bb | 1868 | int *unsignedp; |
c8a96070 NB |
1869 | { |
1870 | const unsigned char *str = token->val.str.text; | |
1871 | const unsigned char *limit = str + token->val.str.len; | |
1872 | unsigned int chars_seen = 0; | |
639e8b0c | 1873 | size_t width, max_chars; |
4268e8bb | 1874 | cppchar_t c, mask, result = 0; |
a47ed310 | 1875 | bool unsigned_p; |
c8a96070 NB |
1876 | |
1877 | #ifdef MULTIBYTE_CHARS | |
1878 | (void) local_mbtowc (NULL, NULL, 0); | |
1879 | #endif | |
1880 | ||
1881 | /* Width in bits. */ | |
1882 | if (token->type == CPP_CHAR) | |
a47ed310 | 1883 | { |
4268e8bb | 1884 | width = CPP_OPTION (pfile, char_precision); |
2443d4e1 | 1885 | max_chars = CPP_OPTION (pfile, int_precision) / width; |
44a147ad | 1886 | unsigned_p = CPP_OPTION (pfile, unsigned_char); |
a47ed310 | 1887 | } |
c8a96070 | 1888 | else |
a47ed310 | 1889 | { |
4268e8bb | 1890 | width = CPP_OPTION (pfile, wchar_precision); |
2443d4e1 | 1891 | max_chars = 1; |
44a147ad | 1892 | unsigned_p = CPP_OPTION (pfile, unsigned_wchar); |
a47ed310 | 1893 | } |
c8a96070 | 1894 | |
4268e8bb NB |
1895 | if (width < BITS_PER_CPPCHAR_T) |
1896 | mask = ((cppchar_t) 1 << width) - 1; | |
c8a96070 NB |
1897 | else |
1898 | mask = ~0; | |
c8a96070 NB |
1899 | |
1900 | while (str < limit) | |
1901 | { | |
1902 | #ifdef MULTIBYTE_CHARS | |
1903 | wchar_t wc; | |
1904 | int char_len; | |
1905 | ||
1906 | char_len = local_mbtowc (&wc, str, limit - str); | |
1907 | if (char_len == -1) | |
1908 | { | |
ebef4e8c NB |
1909 | cpp_error (pfile, DL_WARNING, |
1910 | "ignoring invalid multibyte character"); | |
c8a96070 NB |
1911 | c = *str++; |
1912 | } | |
1913 | else | |
1914 | { | |
1915 | str += char_len; | |
1916 | c = wc; | |
1917 | } | |
1918 | #else | |
1919 | c = *str++; | |
1920 | #endif | |
1921 | ||
1922 | if (c == '\\') | |
4268e8bb | 1923 | c = cpp_parse_escape (pfile, &str, limit, token->type == CPP_WCHAR); |
c8a96070 NB |
1924 | |
1925 | #ifdef MAP_CHARACTER | |
1926 | if (ISPRINT (c)) | |
1927 | c = MAP_CHARACTER (c); | |
1928 | #endif | |
1929 | ||
639e8b0c NB |
1930 | chars_seen++; |
1931 | ||
a5a49440 NB |
1932 | /* Truncate the character, scale the result and merge the two. */ |
1933 | c &= mask; | |
639e8b0c | 1934 | if (width < BITS_PER_CPPCHAR_T) |
a5a49440 | 1935 | result = (result << width) | c; |
639e8b0c NB |
1936 | else |
1937 | result = c; | |
c8a96070 NB |
1938 | } |
1939 | ||
1940 | if (chars_seen == 0) | |
ebef4e8c | 1941 | cpp_error (pfile, DL_ERROR, "empty character constant"); |
639e8b0c | 1942 | else if (chars_seen > 1) |
c8a96070 | 1943 | { |
639e8b0c NB |
1944 | /* Multichar charconsts are of type int and therefore signed. */ |
1945 | unsigned_p = 0; | |
a5a49440 | 1946 | |
639e8b0c NB |
1947 | if (chars_seen > max_chars) |
1948 | { | |
1949 | chars_seen = max_chars; | |
1950 | cpp_error (pfile, DL_WARNING, | |
1951 | "character constant too long for its type"); | |
1952 | } | |
a5a49440 | 1953 | else if (CPP_OPTION (pfile, warn_multichar)) |
639e8b0c | 1954 | cpp_error (pfile, DL_WARNING, "multi-character character constant"); |
c8a96070 NB |
1955 | } |
1956 | ||
b9e2d17b NB |
1957 | /* Sign-extend or truncate the constant to cppchar_t. The value is |
1958 | in WIDTH bits, but for multi-char charconsts it's value is the | |
1959 | full target type's width. */ | |
1960 | if (chars_seen > 1) | |
1961 | width *= max_chars; | |
1962 | if (width < BITS_PER_CPPCHAR_T) | |
a5a49440 | 1963 | { |
b9e2d17b NB |
1964 | mask = ((cppchar_t) 1 << width) - 1; |
1965 | if (unsigned_p || !(result & (1 << (width - 1)))) | |
1966 | result &= mask; | |
1967 | else | |
1968 | result |= ~mask; | |
a5a49440 NB |
1969 | } |
1970 | ||
c8a96070 | 1971 | *pchars_seen = chars_seen; |
4268e8bb | 1972 | *unsignedp = unsigned_p; |
c8a96070 NB |
1973 | return result; |
1974 | } | |
1975 | ||
1e013d2e NB |
1976 | /* Memory buffers. Changing these three constants can have a dramatic |
1977 | effect on performance. The values here are reasonable defaults, | |
1978 | but might be tuned. If you adjust them, be sure to test across a | |
1979 | range of uses of cpplib, including heavy nested function-like macro | |
1980 | expansion. Also check the change in peak memory usage (NJAMD is a | |
1981 | good tool for this). */ | |
1982 | #define MIN_BUFF_SIZE 8000 | |
87062813 | 1983 | #define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2) |
1e013d2e NB |
1984 | #define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \ |
1985 | (MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2) | |
417f3e3a | 1986 | |
87062813 NB |
1987 | #if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0) |
1988 | #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE! | |
1989 | #endif | |
1990 | ||
93c80368 | 1991 | struct dummy |
417f3e3a | 1992 | { |
93c80368 NB |
1993 | char c; |
1994 | union | |
1995 | { | |
1996 | double d; | |
1997 | int *p; | |
1998 | } u; | |
1999 | }; | |
417f3e3a | 2000 | |
93c80368 | 2001 | #define DEFAULT_ALIGNMENT (offsetof (struct dummy, u)) |
b8af0ca5 NB |
2002 | #define CPP_ALIGN(size, align) (((size) + ((align) - 1)) & ~((align) - 1)) |
2003 | ||
c9e7a609 NB |
2004 | /* Create a new allocation buffer. Place the control block at the end |
2005 | of the buffer, so that buffer overflows will cause immediate chaos. */ | |
b8af0ca5 NB |
2006 | static _cpp_buff * |
2007 | new_buff (len) | |
6142088c | 2008 | size_t len; |
b8af0ca5 NB |
2009 | { |
2010 | _cpp_buff *result; | |
ece54d54 | 2011 | unsigned char *base; |
b8af0ca5 | 2012 | |
1e013d2e NB |
2013 | if (len < MIN_BUFF_SIZE) |
2014 | len = MIN_BUFF_SIZE; | |
b8af0ca5 NB |
2015 | len = CPP_ALIGN (len, DEFAULT_ALIGNMENT); |
2016 | ||
2017 | base = xmalloc (len + sizeof (_cpp_buff)); | |
2018 | result = (_cpp_buff *) (base + len); | |
2019 | result->base = base; | |
2020 | result->cur = base; | |
2021 | result->limit = base + len; | |
2022 | result->next = NULL; | |
2023 | return result; | |
2024 | } | |
2025 | ||
2026 | /* Place a chain of unwanted allocation buffers on the free list. */ | |
2027 | void | |
2028 | _cpp_release_buff (pfile, buff) | |
2029 | cpp_reader *pfile; | |
2030 | _cpp_buff *buff; | |
2031 | { | |
2032 | _cpp_buff *end = buff; | |
2033 | ||
2034 | while (end->next) | |
2035 | end = end->next; | |
2036 | end->next = pfile->free_buffs; | |
2037 | pfile->free_buffs = buff; | |
2038 | } | |
2039 | ||
2040 | /* Return a free buffer of size at least MIN_SIZE. */ | |
2041 | _cpp_buff * | |
2042 | _cpp_get_buff (pfile, min_size) | |
2043 | cpp_reader *pfile; | |
6142088c | 2044 | size_t min_size; |
b8af0ca5 NB |
2045 | { |
2046 | _cpp_buff *result, **p; | |
2047 | ||
2048 | for (p = &pfile->free_buffs;; p = &(*p)->next) | |
2049 | { | |
6142088c | 2050 | size_t size; |
1e013d2e NB |
2051 | |
2052 | if (*p == NULL) | |
b8af0ca5 | 2053 | return new_buff (min_size); |
1e013d2e NB |
2054 | result = *p; |
2055 | size = result->limit - result->base; | |
2056 | /* Return a buffer that's big enough, but don't waste one that's | |
2057 | way too big. */ | |
34f5271d | 2058 | if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size)) |
b8af0ca5 NB |
2059 | break; |
2060 | } | |
2061 | ||
2062 | *p = result->next; | |
2063 | result->next = NULL; | |
2064 | result->cur = result->base; | |
2065 | return result; | |
2066 | } | |
2067 | ||
4fe9b91c | 2068 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
2069 | remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies |
2070 | the excess bytes to the new buffer. Chains the new buffer after | |
2071 | BUFF, and returns the new buffer. */ | |
b8af0ca5 | 2072 | _cpp_buff * |
8c3b2693 | 2073 | _cpp_append_extend_buff (pfile, buff, min_extra) |
b8af0ca5 NB |
2074 | cpp_reader *pfile; |
2075 | _cpp_buff *buff; | |
6142088c | 2076 | size_t min_extra; |
b8af0ca5 | 2077 | { |
6142088c | 2078 | size_t size = EXTENDED_BUFF_SIZE (buff, min_extra); |
8c3b2693 | 2079 | _cpp_buff *new_buff = _cpp_get_buff (pfile, size); |
b8af0ca5 | 2080 | |
8c3b2693 NB |
2081 | buff->next = new_buff; |
2082 | memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff)); | |
2083 | return new_buff; | |
2084 | } | |
2085 | ||
4fe9b91c | 2086 | /* Creates a new buffer with enough space to hold the uncommitted |
8c3b2693 NB |
2087 | remaining bytes of the buffer pointed to by BUFF, and at least |
2088 | MIN_EXTRA more bytes. Copies the excess bytes to the new buffer. | |
2089 | Chains the new buffer before the buffer pointed to by BUFF, and | |
2090 | updates the pointer to point to the new buffer. */ | |
2091 | void | |
2092 | _cpp_extend_buff (pfile, pbuff, min_extra) | |
2093 | cpp_reader *pfile; | |
2094 | _cpp_buff **pbuff; | |
2095 | size_t min_extra; | |
2096 | { | |
2097 | _cpp_buff *new_buff, *old_buff = *pbuff; | |
2098 | size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra); | |
2099 | ||
2100 | new_buff = _cpp_get_buff (pfile, size); | |
2101 | memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff)); | |
2102 | new_buff->next = old_buff; | |
2103 | *pbuff = new_buff; | |
b8af0ca5 NB |
2104 | } |
2105 | ||
2106 | /* Free a chain of buffers starting at BUFF. */ | |
2107 | void | |
2108 | _cpp_free_buff (buff) | |
2109 | _cpp_buff *buff; | |
2110 | { | |
2111 | _cpp_buff *next; | |
2112 | ||
2113 | for (; buff; buff = next) | |
2114 | { | |
2115 | next = buff->next; | |
2116 | free (buff->base); | |
2117 | } | |
2118 | } | |
417f3e3a | 2119 | |
ece54d54 NB |
2120 | /* Allocate permanent, unaligned storage of length LEN. */ |
2121 | unsigned char * | |
2122 | _cpp_unaligned_alloc (pfile, len) | |
2123 | cpp_reader *pfile; | |
2124 | size_t len; | |
2125 | { | |
2126 | _cpp_buff *buff = pfile->u_buff; | |
2127 | unsigned char *result = buff->cur; | |
2128 | ||
2129 | if (len > (size_t) (buff->limit - result)) | |
2130 | { | |
2131 | buff = _cpp_get_buff (pfile, len); | |
2132 | buff->next = pfile->u_buff; | |
2133 | pfile->u_buff = buff; | |
2134 | result = buff->cur; | |
2135 | } | |
2136 | ||
2137 | buff->cur = result + len; | |
2138 | return result; | |
2139 | } | |
2140 | ||
87062813 NB |
2141 | /* Allocate permanent, unaligned storage of length LEN from a_buff. |
2142 | That buffer is used for growing allocations when saving macro | |
2143 | replacement lists in a #define, and when parsing an answer to an | |
2144 | assertion in #assert, #unassert or #if (and therefore possibly | |
2145 | whilst expanding macros). It therefore must not be used by any | |
2146 | code that they might call: specifically the lexer and the guts of | |
2147 | the macro expander. | |
2148 | ||
2149 | All existing other uses clearly fit this restriction: storing | |
2150 | registered pragmas during initialization. */ | |
93c80368 | 2151 | unsigned char * |
8c3b2693 NB |
2152 | _cpp_aligned_alloc (pfile, len) |
2153 | cpp_reader *pfile; | |
2154 | size_t len; | |
3fef5b2b | 2155 | { |
8c3b2693 NB |
2156 | _cpp_buff *buff = pfile->a_buff; |
2157 | unsigned char *result = buff->cur; | |
3fef5b2b | 2158 | |
8c3b2693 | 2159 | if (len > (size_t) (buff->limit - result)) |
3fef5b2b | 2160 | { |
8c3b2693 NB |
2161 | buff = _cpp_get_buff (pfile, len); |
2162 | buff->next = pfile->a_buff; | |
2163 | pfile->a_buff = buff; | |
2164 | result = buff->cur; | |
3fef5b2b | 2165 | } |
041c3194 | 2166 | |
8c3b2693 | 2167 | buff->cur = result + len; |
93c80368 | 2168 | return result; |
041c3194 | 2169 | } |