]>
Commit | Line | Data |
---|---|---|
45b966db ZW |
1 | /* CPP Library - lexical analysis. |
2 | Copyright (C) 2000 Free Software Foundation, Inc. | |
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" | |
25 | #include "intl.h" | |
26 | #include "cpplib.h" | |
27 | #include "cpphash.h" | |
28 | ||
f8f769ea ZW |
29 | #ifdef HAVE_MMAP_FILE |
30 | # include <sys/mman.h> | |
31 | #endif | |
32 | ||
ff2b53ef ZW |
33 | #define PEEKBUF(BUFFER, N) \ |
34 | ((BUFFER)->rlimit - (BUFFER)->cur > (N) ? (BUFFER)->cur[N] : EOF) | |
35 | #define GETBUF(BUFFER) \ | |
36 | ((BUFFER)->cur < (BUFFER)->rlimit ? *(BUFFER)->cur++ : EOF) | |
37 | #define FORWARDBUF(BUFFER, N) ((BUFFER)->cur += (N)) | |
38 | ||
39 | #define PEEKN(N) PEEKBUF (CPP_BUFFER (pfile), N) | |
40 | #define FORWARD(N) FORWARDBUF (CPP_BUFFER (pfile), (N)) | |
41 | #define GETC() GETBUF (CPP_BUFFER (pfile)) | |
42 | #define PEEKC() PEEKBUF (CPP_BUFFER (pfile), 0) | |
45b966db ZW |
43 | |
44 | static void skip_block_comment PARAMS ((cpp_reader *)); | |
45 | static void skip_line_comment PARAMS ((cpp_reader *)); | |
46 | static int maybe_macroexpand PARAMS ((cpp_reader *, long)); | |
47 | static int skip_comment PARAMS ((cpp_reader *, int)); | |
48 | static int copy_comment PARAMS ((cpp_reader *, int)); | |
49 | static void skip_string PARAMS ((cpp_reader *, int)); | |
50 | static void parse_string PARAMS ((cpp_reader *, int)); | |
51 | static U_CHAR *find_position PARAMS ((U_CHAR *, U_CHAR *, unsigned long *)); | |
64aaf407 | 52 | static void null_warning PARAMS ((cpp_reader *, unsigned int)); |
45b966db | 53 | |
f2d5f0cc ZW |
54 | static void safe_fwrite PARAMS ((cpp_reader *, const U_CHAR *, |
55 | size_t, FILE *)); | |
1368ee70 ZW |
56 | static void output_line_command PARAMS ((cpp_reader *, cpp_printer *, |
57 | unsigned int)); | |
58 | static void bump_column PARAMS ((cpp_printer *, unsigned int, | |
59 | unsigned int)); | |
c5a04734 | 60 | static void expand_name_space PARAMS ((cpp_toklist *, unsigned int)); |
9e62c811 ZW |
61 | static void pedantic_whitespace PARAMS ((cpp_reader *, U_CHAR *, |
62 | unsigned int)); | |
f2d5f0cc | 63 | |
c5a04734 | 64 | #define auto_expand_name_space(list) \ |
f617b8e2 | 65 | expand_name_space ((list), 1 + (list)->name_cap / 2) |
c5a04734 | 66 | |
b8f41010 NB |
67 | #ifdef NEW_LEXER |
68 | ||
b8f41010 NB |
69 | void init_trigraph_map PARAMS ((void)); |
70 | static unsigned char* trigraph_replace PARAMS ((cpp_reader *, unsigned char *, | |
71 | unsigned char *)); | |
72 | static const unsigned char *backslash_start PARAMS ((cpp_reader *, | |
73 | const unsigned char *)); | |
74 | static int skip_block_comment2 PARAMS ((cpp_reader *)); | |
75 | static int skip_line_comment2 PARAMS ((cpp_reader *)); | |
76 | static void skip_whitespace PARAMS ((cpp_reader *, int)); | |
77 | static void parse_name PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
78 | static void parse_number PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
79 | static void parse_string2 PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *, | |
d1d9a6bd | 80 | unsigned int, int)); |
b8f41010 | 81 | static int trigraph_ok PARAMS ((cpp_reader *, const unsigned char *)); |
f624ffa7 NB |
82 | static void save_comment PARAMS ((cpp_toklist *, cpp_token *, |
83 | const unsigned char *, | |
ad265aa4 | 84 | unsigned int, unsigned int)); |
b8f41010 NB |
85 | void _cpp_lex_line PARAMS ((cpp_reader *, cpp_toklist *)); |
86 | ||
87 | static void _cpp_output_list PARAMS ((cpp_reader *, cpp_toklist *)); | |
88 | ||
d1d9a6bd | 89 | static unsigned char * spell_token PARAMS ((cpp_reader *, const cpp_token *, |
f617b8e2 | 90 | unsigned char *, int)); |
b8f41010 NB |
91 | |
92 | typedef unsigned int (* speller) PARAMS ((unsigned char *, cpp_toklist *, | |
93 | cpp_token *)); | |
94 | ||
95 | /* Macros on a cpp_name. */ | |
d1d9a6bd NB |
96 | #define INIT_TOKEN_NAME(list, token) \ |
97 | do {(token)->val.name.len = 0; \ | |
98 | (token)->val.name.text = (list)->namebuf + (list)->name_used; \ | |
99 | (list)->tokens_used = token - (list)->tokens + 1; \ | |
100 | } while (0) | |
b8f41010 NB |
101 | |
102 | /* Maybe put these in the ISTABLE eventually. */ | |
103 | #define IS_HSPACE(c) ((c) == ' ' || (c) == '\t') | |
104 | #define IS_NEWLINE(c) ((c) == '\n' || (c) == '\r') | |
105 | ||
106 | /* Handle LF, CR, CR-LF and LF-CR style newlines. Assumes next | |
107 | character, if any, is in buffer. */ | |
108 | #define handle_newline(cur, limit, c) \ | |
109 | do {\ | |
110 | if ((cur) < (limit) && *(cur) == '\r' + '\n' - c) \ | |
111 | (cur)++; \ | |
112 | CPP_BUMP_LINE_CUR (pfile, (cur)); \ | |
6ab3e7dd | 113 | pfile->col_adjust = 0; \ |
b8f41010 NB |
114 | } while (0) |
115 | ||
116 | #define IMMED_TOKEN() (!(cur_token->flags & PREV_WHITESPACE)) | |
117 | #define PREV_TOKEN_TYPE (cur_token[-1].type) | |
118 | ||
f617b8e2 NB |
119 | #define PUSH_TOKEN(ttype) cur_token++->type = ttype |
120 | #define REVISE_TOKEN(ttype) cur_token[-1].type = ttype | |
121 | #define BACKUP_TOKEN(ttype) (--cur_token)->type = ttype | |
122 | #define BACKUP_DIGRAPH(ttype) do { \ | |
123 | BACKUP_TOKEN(ttype); cur_token->flags |= DIGRAPH;} while (0) | |
124 | ||
125 | /* An upper bound on the number of bytes needed to spell a token, | |
126 | including preceding whitespace. */ | |
f624ffa7 NB |
127 | #define TOKEN_LEN(token) (5 + (token_spellings[(token)->type].type > \ |
128 | SPELL_NONE ? (token)->val.name.len: 0)) | |
f617b8e2 NB |
129 | |
130 | #endif | |
131 | ||
5d7ee2fa NB |
132 | /* Order here matters. Those beyond SPELL_NONE store their spelling |
133 | in the token list, and it's length in the token->val.name.len. */ | |
d1d9a6bd NB |
134 | enum spell_type |
135 | { | |
136 | SPELL_OPERATOR = 0, | |
137 | SPELL_NONE, | |
138 | SPELL_CHAR, /* FIXME: revert order of NONE and CHAR after transition. */ | |
139 | SPELL_IDENT, | |
140 | SPELL_STRING | |
141 | }; | |
5d7ee2fa | 142 | |
f617b8e2 | 143 | #define T(e, s) {SPELL_OPERATOR, (const U_CHAR *) s}, |
5d7ee2fa NB |
144 | #define I(e, s) {SPELL_IDENT, s}, |
145 | #define S(e, s) {SPELL_STRING, s}, | |
b8f41010 NB |
146 | #define C(e, s) {SPELL_CHAR, s}, |
147 | #define N(e, s) {SPELL_NONE, s}, | |
b8f41010 NB |
148 | |
149 | static const struct token_spelling | |
150 | { | |
d1d9a6bd | 151 | ENUM_BITFIELD(spell_type) type : CHAR_BIT; |
f617b8e2 | 152 | const U_CHAR *spelling; |
b8f41010 NB |
153 | } token_spellings [N_TTYPES + 1] = {TTYPE_TABLE {0, 0} }; |
154 | ||
155 | #undef T | |
5d7ee2fa NB |
156 | #undef I |
157 | #undef S | |
b8f41010 NB |
158 | #undef C |
159 | #undef N | |
b8f41010 | 160 | |
45b966db ZW |
161 | /* Re-allocates PFILE->token_buffer so it will hold at least N more chars. */ |
162 | ||
163 | void | |
164 | _cpp_grow_token_buffer (pfile, n) | |
165 | cpp_reader *pfile; | |
166 | long n; | |
167 | { | |
168 | long old_written = CPP_WRITTEN (pfile); | |
169 | pfile->token_buffer_size = n + 2 * pfile->token_buffer_size; | |
170 | pfile->token_buffer = (U_CHAR *) | |
171 | xrealloc(pfile->token_buffer, pfile->token_buffer_size); | |
172 | CPP_SET_WRITTEN (pfile, old_written); | |
173 | } | |
174 | ||
45b966db ZW |
175 | /* Allocate a new cpp_buffer for PFILE, and push it on the input buffer stack. |
176 | If BUFFER != NULL, then use the LENGTH characters in BUFFER | |
177 | as the new input buffer. | |
178 | Return the new buffer, or NULL on failure. */ | |
179 | ||
180 | cpp_buffer * | |
181 | cpp_push_buffer (pfile, buffer, length) | |
182 | cpp_reader *pfile; | |
183 | const U_CHAR *buffer; | |
184 | long length; | |
185 | { | |
186 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
187 | cpp_buffer *new; | |
188 | if (++pfile->buffer_stack_depth == CPP_STACK_MAX) | |
189 | { | |
190 | cpp_fatal (pfile, "macro or `#include' recursion too deep"); | |
191 | return NULL; | |
192 | } | |
193 | ||
194 | new = (cpp_buffer *) xcalloc (1, sizeof (cpp_buffer)); | |
195 | ||
45b966db | 196 | new->buf = new->cur = buffer; |
ff2b53ef | 197 | new->rlimit = buffer + length; |
45b966db | 198 | new->prev = buf; |
ff2b53ef | 199 | new->mark = NULL; |
45b966db ZW |
200 | new->line_base = NULL; |
201 | ||
202 | CPP_BUFFER (pfile) = new; | |
203 | return new; | |
204 | } | |
205 | ||
206 | cpp_buffer * | |
207 | cpp_pop_buffer (pfile) | |
208 | cpp_reader *pfile; | |
209 | { | |
210 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
211 | if (ACTIVE_MARK_P (pfile)) | |
212 | cpp_ice (pfile, "mark active in cpp_pop_buffer"); | |
c56c2073 | 213 | |
c31a6508 | 214 | if (buf->inc) |
c56c2073 ZW |
215 | { |
216 | _cpp_unwind_if_stack (pfile, buf); | |
217 | if (buf->buf) | |
218 | free ((PTR) buf->buf); | |
219 | if (pfile->system_include_depth) | |
220 | pfile->system_include_depth--; | |
221 | if (pfile->potential_control_macro) | |
222 | { | |
c31a6508 ZW |
223 | if (buf->inc->cmacro != NEVER_REREAD) |
224 | buf->inc->cmacro = pfile->potential_control_macro; | |
c56c2073 ZW |
225 | pfile->potential_control_macro = 0; |
226 | } | |
227 | pfile->input_stack_listing_current = 0; | |
c31a6508 ZW |
228 | /* If the file will not be included again, then close it. */ |
229 | if (DO_NOT_REREAD (buf->inc)) | |
230 | { | |
231 | close (buf->inc->fd); | |
232 | buf->inc->fd = -1; | |
233 | } | |
c56c2073 ZW |
234 | } |
235 | else if (buf->macro) | |
236 | { | |
f8f769ea | 237 | cpp_hashnode *m = buf->macro; |
c56c2073 ZW |
238 | |
239 | m->disabled = 0; | |
240 | if ((m->type == T_FMACRO && buf->mapped) | |
241 | || m->type == T_SPECLINE || m->type == T_FILE | |
242 | || m->type == T_BASE_FILE || m->type == T_INCLUDE_LEVEL | |
243 | || m->type == T_STDC) | |
244 | free ((PTR) buf->buf); | |
245 | } | |
45b966db ZW |
246 | CPP_BUFFER (pfile) = CPP_PREV_BUFFER (buf); |
247 | free (buf); | |
248 | pfile->buffer_stack_depth--; | |
249 | return CPP_BUFFER (pfile); | |
250 | } | |
251 | ||
f2d5f0cc ZW |
252 | /* Deal with the annoying semantics of fwrite. */ |
253 | static void | |
254 | safe_fwrite (pfile, buf, len, fp) | |
255 | cpp_reader *pfile; | |
256 | const U_CHAR *buf; | |
257 | size_t len; | |
258 | FILE *fp; | |
259 | { | |
260 | size_t count; | |
45b966db | 261 | |
f2d5f0cc ZW |
262 | while (len) |
263 | { | |
264 | count = fwrite (buf, 1, len, fp); | |
265 | if (count == 0) | |
266 | goto error; | |
267 | len -= count; | |
268 | buf += count; | |
269 | } | |
270 | return; | |
271 | ||
272 | error: | |
273 | cpp_notice_from_errno (pfile, CPP_OPTION (pfile, out_fname)); | |
274 | } | |
275 | ||
276 | /* Notify the compiler proper that the current line number has jumped, | |
277 | or the current file name has changed. */ | |
278 | ||
279 | static void | |
1368ee70 | 280 | output_line_command (pfile, print, line) |
45b966db | 281 | cpp_reader *pfile; |
f2d5f0cc | 282 | cpp_printer *print; |
1368ee70 | 283 | unsigned int line; |
45b966db | 284 | { |
1368ee70 | 285 | cpp_buffer *ip = cpp_file_buffer (pfile); |
f2d5f0cc ZW |
286 | enum { same = 0, enter, leave, rname } change; |
287 | static const char * const codes[] = { "", " 1", " 2", "" }; | |
288 | ||
289 | if (CPP_OPTION (pfile, no_line_commands)) | |
290 | return; | |
291 | ||
0e500c78 JJ |
292 | /* Determine whether the current filename has changed, and if so, |
293 | how. 'nominal_fname' values are unique, so they can be compared | |
294 | by comparing pointers. */ | |
295 | if (ip->nominal_fname == print->last_fname) | |
296 | change = same; | |
54bef41d JJ |
297 | else |
298 | { | |
0e500c78 JJ |
299 | if (pfile->buffer_stack_depth == print->last_bsd) |
300 | change = rname; | |
f2d5f0cc | 301 | else |
0e500c78 JJ |
302 | { |
303 | if (pfile->buffer_stack_depth > print->last_bsd) | |
304 | change = enter; | |
305 | else | |
306 | change = leave; | |
307 | print->last_bsd = pfile->buffer_stack_depth; | |
308 | } | |
309 | print->last_fname = ip->nominal_fname; | |
45b966db | 310 | } |
f2d5f0cc ZW |
311 | /* If the current file has not changed, we can output a few newlines |
312 | instead if we want to increase the line number by a small amount. | |
313 | We cannot do this if print->lineno is zero, because that means we | |
314 | haven't output any line commands yet. (The very first line | |
315 | command output is a `same_file' command.) */ | |
316 | if (change == same && print->lineno != 0 | |
317 | && line >= print->lineno && line < print->lineno + 8) | |
45b966db | 318 | { |
f2d5f0cc | 319 | while (line > print->lineno) |
45b966db | 320 | { |
f2d5f0cc ZW |
321 | putc ('\n', print->outf); |
322 | print->lineno++; | |
45b966db | 323 | } |
f2d5f0cc | 324 | return; |
45b966db | 325 | } |
f2d5f0cc ZW |
326 | |
327 | #ifndef NO_IMPLICIT_EXTERN_C | |
328 | if (CPP_OPTION (pfile, cplusplus)) | |
329 | fprintf (print->outf, "# %u \"%s\"%s%s%s\n", line, ip->nominal_fname, | |
330 | codes[change], | |
c31a6508 ZW |
331 | ip->inc->sysp ? " 3" : "", |
332 | (ip->inc->sysp == 2) ? " 4" : ""); | |
f2d5f0cc ZW |
333 | else |
334 | #endif | |
335 | fprintf (print->outf, "# %u \"%s\"%s%s\n", line, ip->nominal_fname, | |
336 | codes[change], | |
c31a6508 | 337 | ip->inc->sysp ? " 3" : ""); |
f2d5f0cc ZW |
338 | print->lineno = line; |
339 | } | |
340 | ||
341 | /* Write the contents of the token_buffer to the output stream, and | |
342 | clear the token_buffer. Also handles generating line commands and | |
343 | keeping track of file transitions. */ | |
344 | ||
345 | void | |
346 | cpp_output_tokens (pfile, print) | |
347 | cpp_reader *pfile; | |
348 | cpp_printer *print; | |
349 | { | |
1368ee70 ZW |
350 | cpp_buffer *ip; |
351 | ||
f6fab919 ZW |
352 | if (CPP_WRITTEN (pfile) - print->written) |
353 | { | |
354 | if (CPP_PWRITTEN (pfile)[-1] == '\n' && print->lineno) | |
355 | print->lineno++; | |
356 | safe_fwrite (pfile, pfile->token_buffer, | |
357 | CPP_WRITTEN (pfile) - print->written, print->outf); | |
358 | } | |
1368ee70 ZW |
359 | |
360 | ip = cpp_file_buffer (pfile); | |
361 | if (ip) | |
362 | output_line_command (pfile, print, CPP_BUF_LINE (ip)); | |
363 | ||
f2d5f0cc | 364 | CPP_SET_WRITTEN (pfile, print->written); |
45b966db ZW |
365 | } |
366 | ||
1368ee70 ZW |
367 | /* Helper for cpp_output_list - increases the column number to match |
368 | what we expect it to be. */ | |
369 | ||
370 | static void | |
371 | bump_column (print, from, to) | |
372 | cpp_printer *print; | |
373 | unsigned int from, to; | |
374 | { | |
375 | unsigned int tabs, spcs; | |
376 | unsigned int delta = to - from; | |
377 | ||
378 | /* Only if FROM is 0, advance by tabs. */ | |
379 | if (from == 0) | |
380 | tabs = delta / 8, spcs = delta % 8; | |
381 | else | |
382 | tabs = 0, spcs = delta; | |
383 | ||
384 | while (tabs--) putc ('\t', print->outf); | |
385 | while (spcs--) putc (' ', print->outf); | |
386 | } | |
387 | ||
388 | /* Write out the list L onto pfile->token_buffer. This function is | |
389 | incomplete: | |
390 | ||
391 | 1) pfile->token_buffer is not going to continue to exist. | |
392 | 2) At the moment, tokens don't carry the information described | |
393 | in cpplib.h; they are all strings. | |
394 | 3) The list has to be a complete line, and has to be written starting | |
395 | at the beginning of a line. */ | |
396 | ||
397 | void | |
398 | cpp_output_list (pfile, print, list) | |
399 | cpp_reader *pfile; | |
400 | cpp_printer *print; | |
401 | const cpp_toklist *list; | |
402 | { | |
403 | unsigned int i; | |
404 | unsigned int curcol = 1; | |
405 | ||
406 | /* XXX Probably does not do what is intended. */ | |
407 | if (print->lineno != list->line) | |
408 | output_line_command (pfile, print, list->line); | |
409 | ||
410 | for (i = 0; i < list->tokens_used; i++) | |
411 | { | |
1920de47 | 412 | if (TOK_TYPE (list, i) == CPP_VSPACE) |
1368ee70 ZW |
413 | { |
414 | output_line_command (pfile, print, list->tokens[i].aux); | |
415 | continue; | |
416 | } | |
417 | ||
1920de47 | 418 | if (curcol < TOK_COL (list, i)) |
1368ee70 ZW |
419 | { |
420 | /* Insert space to bring the column to what it should be. */ | |
1920de47 ZW |
421 | bump_column (print, curcol - 1, TOK_COL (list, i)); |
422 | curcol = TOK_COL (list, i); | |
1368ee70 ZW |
423 | } |
424 | /* XXX We may have to insert space to prevent an accidental | |
425 | token paste. */ | |
1920de47 ZW |
426 | safe_fwrite (pfile, TOK_NAME (list, i), TOK_LEN (list, i), print->outf); |
427 | curcol += TOK_LEN (list, i); | |
1368ee70 ZW |
428 | } |
429 | } | |
430 | ||
f2d5f0cc ZW |
431 | /* Scan a string (which may have escape marks), perform macro expansion, |
432 | and write the result to the token_buffer. */ | |
45b966db ZW |
433 | |
434 | void | |
f2d5f0cc | 435 | _cpp_expand_to_buffer (pfile, buf, length) |
45b966db ZW |
436 | cpp_reader *pfile; |
437 | const U_CHAR *buf; | |
438 | int length; | |
439 | { | |
c56c2073 | 440 | cpp_buffer *stop; |
f2d5f0cc | 441 | enum cpp_ttype token; |
f6fab919 | 442 | U_CHAR *buf1; |
45b966db ZW |
443 | |
444 | if (length < 0) | |
445 | { | |
446 | cpp_ice (pfile, "length < 0 in cpp_expand_to_buffer"); | |
447 | return; | |
448 | } | |
449 | ||
f6fab919 ZW |
450 | /* Copy the buffer, because it might be in an unsafe place - for |
451 | example, a sequence on the token_buffer, where the pointers will | |
452 | be invalidated if we enlarge the token_buffer. */ | |
453 | buf1 = alloca (length); | |
454 | memcpy (buf1, buf, length); | |
455 | ||
45b966db | 456 | /* Set up the input on the input stack. */ |
c56c2073 ZW |
457 | stop = CPP_BUFFER (pfile); |
458 | if (cpp_push_buffer (pfile, buf1, length) == NULL) | |
45b966db | 459 | return; |
c56c2073 | 460 | CPP_BUFFER (pfile)->has_escapes = 1; |
45b966db ZW |
461 | |
462 | /* Scan the input, create the output. */ | |
f2d5f0cc ZW |
463 | for (;;) |
464 | { | |
465 | token = cpp_get_token (pfile); | |
c56c2073 | 466 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 467 | break; |
f2d5f0cc | 468 | } |
45b966db ZW |
469 | } |
470 | ||
c56c2073 | 471 | /* Scan until CPP_BUFFER (PFILE) is exhausted, discarding output. */ |
f2d5f0cc ZW |
472 | |
473 | void | |
474 | cpp_scan_buffer_nooutput (pfile) | |
475 | cpp_reader *pfile; | |
476 | { | |
c56c2073 | 477 | cpp_buffer *stop = CPP_PREV_BUFFER (CPP_BUFFER (pfile)); |
f2d5f0cc ZW |
478 | enum cpp_ttype token; |
479 | unsigned int old_written = CPP_WRITTEN (pfile); | |
480 | /* In no-output mode, we can ignore everything but directives. */ | |
481 | for (;;) | |
482 | { | |
483 | if (! pfile->only_seen_white) | |
484 | _cpp_skip_rest_of_line (pfile); | |
485 | token = cpp_get_token (pfile); | |
c56c2073 | 486 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 487 | break; |
f2d5f0cc ZW |
488 | } |
489 | CPP_SET_WRITTEN (pfile, old_written); | |
490 | } | |
491 | ||
c56c2073 | 492 | /* Scan until CPP_BUFFER (pfile) is exhausted, writing output to PRINT. */ |
f2d5f0cc ZW |
493 | |
494 | void | |
495 | cpp_scan_buffer (pfile, print) | |
496 | cpp_reader *pfile; | |
497 | cpp_printer *print; | |
498 | { | |
c56c2073 | 499 | cpp_buffer *stop = CPP_PREV_BUFFER (CPP_BUFFER (pfile)); |
f2d5f0cc ZW |
500 | enum cpp_ttype token; |
501 | ||
502 | for (;;) | |
503 | { | |
504 | token = cpp_get_token (pfile); | |
15dad1d9 | 505 | if (token == CPP_VSPACE || token == CPP_EOF |
f2d5f0cc ZW |
506 | /* XXX Temporary kluge - force flush after #include only */ |
507 | || (token == CPP_DIRECTIVE | |
508 | && CPP_BUFFER (pfile)->nominal_fname != print->last_fname)) | |
509 | { | |
510 | cpp_output_tokens (pfile, print); | |
c56c2073 | 511 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 512 | return; |
f2d5f0cc ZW |
513 | } |
514 | } | |
515 | } | |
516 | ||
45b966db ZW |
517 | /* Return the topmost cpp_buffer that corresponds to a file (not a macro). */ |
518 | ||
519 | cpp_buffer * | |
520 | cpp_file_buffer (pfile) | |
521 | cpp_reader *pfile; | |
522 | { | |
523 | cpp_buffer *ip; | |
524 | ||
525 | for (ip = CPP_BUFFER (pfile); ip; ip = CPP_PREV_BUFFER (ip)) | |
c31a6508 | 526 | if (ip->inc != NULL) |
45b966db ZW |
527 | return ip; |
528 | return NULL; | |
529 | } | |
530 | ||
1368ee70 ZW |
531 | /* Token-buffer helper functions. */ |
532 | ||
d1d9a6bd NB |
533 | /* Expand a token list's string space. It is *vital* that |
534 | list->tokens_used is correct, to get pointer fix-up right. */ | |
1368ee70 | 535 | static void |
c5a04734 | 536 | expand_name_space (list, len) |
1368ee70 | 537 | cpp_toklist *list; |
c5a04734 ZW |
538 | unsigned int len; |
539 | { | |
f617b8e2 | 540 | const U_CHAR *old_namebuf; |
f617b8e2 NB |
541 | |
542 | old_namebuf = list->namebuf; | |
c5a04734 ZW |
543 | list->name_cap += len; |
544 | list->namebuf = (unsigned char *) xrealloc (list->namebuf, list->name_cap); | |
f617b8e2 NB |
545 | |
546 | /* Fix up token text pointers. */ | |
79f50f2a | 547 | if (list->namebuf != old_namebuf) |
f617b8e2 NB |
548 | { |
549 | unsigned int i; | |
550 | ||
551 | for (i = 0; i < list->tokens_used; i++) | |
552 | if (token_spellings[list->tokens[i].type].type > SPELL_NONE) | |
79f50f2a | 553 | list->tokens[i].val.name.text += (list->namebuf - old_namebuf); |
f617b8e2 | 554 | } |
1368ee70 ZW |
555 | } |
556 | ||
557 | /* Expand the number of tokens in a list. */ | |
d1d9a6bd NB |
558 | void |
559 | _cpp_expand_token_space (list, count) | |
1368ee70 | 560 | cpp_toklist *list; |
d1d9a6bd | 561 | unsigned int count; |
1368ee70 | 562 | { |
d1d9a6bd NB |
563 | unsigned int n; |
564 | ||
565 | list->tokens_cap += count; | |
566 | n = list->tokens_cap; | |
15dad1d9 | 567 | if (list->flags & LIST_OFFSET) |
d1d9a6bd | 568 | list->tokens--, n++; |
1368ee70 | 569 | list->tokens = (cpp_token *) |
d1d9a6bd | 570 | xrealloc (list->tokens, n * sizeof (cpp_token)); |
15dad1d9 ZW |
571 | if (list->flags & LIST_OFFSET) |
572 | list->tokens++; /* Skip the dummy. */ | |
1368ee70 ZW |
573 | } |
574 | ||
d1d9a6bd NB |
575 | /* Initialize a token list. If flags is DUMMY_TOKEN, we allocate |
576 | an extra token in front of the token list, as this allows the lexer | |
577 | to always peek at the previous token without worrying about | |
578 | underflowing the list, and some initial space. Otherwise, no | |
579 | token- or name-space is allocated, and there is no dummy token. */ | |
15dad1d9 | 580 | void |
d1d9a6bd | 581 | _cpp_init_toklist (list, flags) |
1368ee70 | 582 | cpp_toklist *list; |
d1d9a6bd | 583 | int flags; |
1368ee70 | 584 | { |
d1d9a6bd NB |
585 | /* We malloc zero bytes because we may want to realloc later, and |
586 | some old implementations don't like realloc-ing a null pointer. */ | |
587 | if (flags == NO_DUMMY_TOKEN) | |
588 | { | |
589 | list->tokens_cap = 0; | |
590 | list->tokens = (cpp_token *) malloc (0); | |
591 | list->name_cap = 0; | |
592 | list->flags = 0; | |
593 | } | |
594 | else | |
595 | { | |
596 | /* Initialize token space. Put a dummy token before the start | |
597 | that will fail matches. */ | |
598 | list->tokens_cap = 256; /* 4K's worth. */ | |
599 | list->tokens = (cpp_token *) | |
600 | xmalloc ((list->tokens_cap + 1) * sizeof (cpp_token)); | |
601 | list->tokens[0].type = CPP_EOF; | |
602 | list->tokens++; | |
603 | ||
604 | /* Initialize name space. */ | |
605 | list->name_cap = 1024; | |
606 | list->flags = LIST_OFFSET; | |
607 | } | |
15dad1d9 | 608 | |
d1d9a6bd | 609 | /* Allocate name space. */ |
15dad1d9 ZW |
610 | list->namebuf = (unsigned char *) xmalloc (list->name_cap); |
611 | ||
15dad1d9 ZW |
612 | _cpp_clear_toklist (list); |
613 | } | |
1368ee70 | 614 | |
15dad1d9 ZW |
615 | /* Clear a token list. */ |
616 | void | |
617 | _cpp_clear_toklist (list) | |
618 | cpp_toklist *list; | |
619 | { | |
c5a04734 ZW |
620 | list->tokens_used = 0; |
621 | list->name_used = 0; | |
15dad1d9 ZW |
622 | list->dirno = -1; |
623 | list->flags &= LIST_OFFSET; /* clear all but that one */ | |
624 | } | |
625 | ||
626 | /* Free a token list. Does not free the list itself, which may be | |
627 | embedded in a larger structure. */ | |
628 | void | |
629 | _cpp_free_toklist (list) | |
630 | cpp_toklist *list; | |
631 | { | |
15dad1d9 ZW |
632 | if (list->flags & LIST_OFFSET) |
633 | free (list->tokens - 1); /* Backup over dummy token. */ | |
634 | else | |
635 | free (list->tokens); | |
636 | free (list->namebuf); | |
1368ee70 ZW |
637 | } |
638 | ||
15dad1d9 ZW |
639 | /* Slice a token list: copy the sublist [START, FINISH) into COPY. |
640 | COPY is assumed not to be initialized. The comment space is not | |
641 | copied. */ | |
642 | void | |
643 | _cpp_slice_toklist (copy, start, finish) | |
644 | cpp_toklist *copy; | |
645 | const cpp_token *start, *finish; | |
646 | { | |
647 | unsigned int i, n; | |
648 | size_t bytes; | |
649 | ||
650 | n = finish - start; | |
651 | copy->tokens_cap = n; | |
652 | copy->tokens = (cpp_token *) xmalloc (n * sizeof (cpp_token)); | |
653 | memcpy (copy->tokens, start, n * sizeof (cpp_token)); | |
654 | ||
655 | bytes = 0; | |
656 | for (i = 0; i < n; i++) | |
657 | if (token_spellings[start[i].type].type > SPELL_NONE) | |
658 | bytes += start[i].val.name.len; | |
659 | ||
660 | copy->namebuf = xmalloc (bytes); | |
661 | bytes = 0; | |
662 | for (i = 0; i < n; i++) | |
663 | if (token_spellings[start[i].type].type > SPELL_NONE) | |
664 | { | |
665 | memcpy (copy->namebuf + bytes, | |
666 | start[i].val.name.text, start[i].val.name.len); | |
667 | copy->tokens[i].val.name.text = copy->namebuf + bytes; | |
668 | bytes += start[i].val.name.len; | |
669 | } | |
670 | ||
671 | copy->tokens_cap = n; | |
672 | copy->tokens_used = n; | |
673 | copy->name_used = bytes; | |
674 | copy->name_cap = bytes; | |
15dad1d9 ZW |
675 | |
676 | copy->flags = 0; | |
677 | copy->dirno = -1; | |
678 | } | |
1368ee70 | 679 | |
15dad1d9 | 680 | /* Shrink a token list down to the minimum size. */ |
1368ee70 | 681 | void |
15dad1d9 ZW |
682 | _cpp_squeeze_toklist (list) |
683 | cpp_toklist *list; | |
684 | { | |
685 | long delta; | |
686 | const U_CHAR *old_namebuf; | |
687 | ||
688 | if (list->flags & LIST_OFFSET) | |
689 | { | |
690 | list->tokens--; | |
691 | memmove (list->tokens, list->tokens + 1, | |
692 | list->tokens_used * sizeof (cpp_token)); | |
693 | list->tokens = xrealloc (list->tokens, | |
694 | list->tokens_used * sizeof (cpp_token)); | |
695 | list->flags &= ~LIST_OFFSET; | |
696 | } | |
697 | else | |
698 | list->tokens = xrealloc (list->tokens, | |
699 | list->tokens_used * sizeof (cpp_token)); | |
700 | list->tokens_cap = list->tokens_used; | |
701 | ||
702 | old_namebuf = list->namebuf; | |
703 | list->namebuf = xrealloc (list->namebuf, list->name_used); | |
704 | list->name_cap = list->name_used; | |
705 | ||
706 | /* Fix up token text pointers. */ | |
707 | delta = list->namebuf - old_namebuf; | |
708 | if (delta) | |
709 | { | |
710 | unsigned int i; | |
711 | ||
712 | for (i = 0; i < list->tokens_used; i++) | |
713 | if (token_spellings[list->tokens[i].type].type > SPELL_NONE) | |
714 | list->tokens[i].val.name.text += delta; | |
715 | } | |
15dad1d9 ZW |
716 | } |
717 | ||
718 | /* Compare two tokens. */ | |
719 | int | |
720 | _cpp_equiv_tokens (a, b) | |
721 | const cpp_token *a, *b; | |
722 | { | |
723 | if (a->type != b->type | |
724 | || a->flags != b->flags | |
725 | || a->aux != b->aux) | |
726 | return 0; | |
727 | ||
728 | if (token_spellings[a->type].type > SPELL_NONE) | |
729 | { | |
730 | if (a->val.name.len != b->val.name.len | |
731 | || ustrncmp(a->val.name.text, | |
732 | b->val.name.text, | |
733 | a->val.name.len)) | |
734 | return 0; | |
735 | } | |
736 | return 1; | |
737 | } | |
738 | ||
739 | /* Compare two token lists. */ | |
740 | int | |
741 | _cpp_equiv_toklists (a, b) | |
742 | const cpp_toklist *a, *b; | |
743 | { | |
744 | unsigned int i; | |
745 | ||
746 | if (a->tokens_used != b->tokens_used) | |
747 | return 0; | |
748 | ||
749 | for (i = 0; i < a->tokens_used; i++) | |
750 | if (! _cpp_equiv_tokens (&a->tokens[i], &b->tokens[i])) | |
751 | return 0; | |
752 | return 1; | |
753 | } | |
754 | ||
755 | /* Scan until we encounter a token of type STOP or a newline, and | |
756 | create a token list for it. Does not macro-expand or execute | |
757 | directives. The final token is not included in the list or | |
758 | consumed from the input. Returns the type of the token stopped at. */ | |
759 | ||
760 | enum cpp_ttype | |
761 | _cpp_scan_until (pfile, list, stop) | |
1368ee70 ZW |
762 | cpp_reader *pfile; |
763 | cpp_toklist *list; | |
15dad1d9 | 764 | enum cpp_ttype stop; |
1368ee70 ZW |
765 | { |
766 | int i, col; | |
767 | long written, len; | |
768 | enum cpp_ttype type; | |
9e62c811 | 769 | int space_before; |
1368ee70 | 770 | |
15dad1d9 ZW |
771 | _cpp_clear_toklist (list); |
772 | list->line = CPP_BUF_LINE (CPP_BUFFER (pfile)); | |
1368ee70 ZW |
773 | |
774 | written = CPP_WRITTEN (pfile); | |
775 | i = 0; | |
9e62c811 | 776 | space_before = 0; |
1368ee70 ZW |
777 | for (;;) |
778 | { | |
779 | col = CPP_BUFFER (pfile)->cur - CPP_BUFFER (pfile)->line_base; | |
780 | type = _cpp_lex_token (pfile); | |
781 | len = CPP_WRITTEN (pfile) - written; | |
782 | CPP_SET_WRITTEN (pfile, written); | |
783 | if (type == CPP_HSPACE) | |
9e62c811 ZW |
784 | { |
785 | if (CPP_PEDANTIC (pfile)) | |
786 | pedantic_whitespace (pfile, pfile->token_buffer + written, len); | |
787 | space_before = 1; | |
788 | continue; | |
789 | } | |
0f89df67 ZW |
790 | else if (type == CPP_COMMENT) |
791 | /* Only happens when processing -traditional macro definitions. | |
792 | Do not give this a token entry, but do not change space_before | |
793 | either. */ | |
794 | continue; | |
1368ee70 ZW |
795 | |
796 | if (list->tokens_used >= list->tokens_cap) | |
d1d9a6bd | 797 | _cpp_expand_token_space (list, 256); |
1368ee70 | 798 | if (list->name_used + len >= list->name_cap) |
bb1ec1d7 | 799 | expand_name_space (list, list->name_used + len + 1 - list->name_cap); |
1368ee70 | 800 | |
9e62c811 ZW |
801 | if (type == CPP_MACRO) |
802 | type = CPP_NAME; | |
803 | ||
15dad1d9 ZW |
804 | if (type == CPP_VSPACE || type == stop) |
805 | break; | |
806 | ||
1368ee70 | 807 | list->tokens_used++; |
1920de47 ZW |
808 | TOK_TYPE (list, i) = type; |
809 | TOK_COL (list, i) = col; | |
a58f64f5 | 810 | TOK_AUX (list, i) = 0; |
1920de47 | 811 | TOK_FLAGS (list, i) = space_before ? PREV_WHITESPACE : 0; |
9e62c811 | 812 | |
1920de47 | 813 | TOK_LEN (list, i) = len; |
f617b8e2 NB |
814 | if (token_spellings[type].type > SPELL_NONE) |
815 | { | |
816 | memcpy (list->namebuf + list->name_used, CPP_PWRITTEN (pfile), len); | |
817 | TOK_NAME (list, i) = list->namebuf + list->name_used; | |
818 | list->name_used += len; | |
819 | } | |
820 | else | |
821 | TOK_NAME (list, i) = token_spellings[type].spelling; | |
1368ee70 | 822 | i++; |
9e62c811 | 823 | space_before = 0; |
1368ee70 | 824 | } |
9e62c811 | 825 | |
15dad1d9 | 826 | /* XXX Temporary kluge: put back the newline (or whatever). */ |
9e62c811 | 827 | FORWARD(-1); |
1368ee70 | 828 | |
15dad1d9 ZW |
829 | /* Don't consider the first token to have white before. */ |
830 | TOK_FLAGS (list, 0) &= ~PREV_WHITESPACE; | |
831 | return type; | |
832 | } | |
1368ee70 | 833 | |
45b966db ZW |
834 | /* Skip a C-style block comment. We know it's a comment, and point is |
835 | at the second character of the starter. */ | |
836 | static void | |
837 | skip_block_comment (pfile) | |
838 | cpp_reader *pfile; | |
839 | { | |
3a2b2c7a | 840 | unsigned int line, col; |
61474454 | 841 | const U_CHAR *limit, *cur; |
45b966db ZW |
842 | |
843 | FORWARD(1); | |
3a2b2c7a ZW |
844 | line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
845 | col = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
61474454 NB |
846 | limit = CPP_BUFFER (pfile)->rlimit; |
847 | cur = CPP_BUFFER (pfile)->cur; | |
848 | ||
849 | while (cur < limit) | |
45b966db | 850 | { |
61474454 NB |
851 | char c = *cur++; |
852 | if (c == '\n' || c == '\r') | |
45b966db ZW |
853 | { |
854 | /* \r cannot be a macro escape marker here. */ | |
855 | if (!ACTIVE_MARK_P (pfile)) | |
61474454 NB |
856 | CPP_BUMP_LINE_CUR (pfile, cur); |
857 | } | |
858 | else if (c == '*') | |
859 | { | |
860 | /* Check for teminator. */ | |
861 | if (cur < limit && *cur == '/') | |
862 | goto out; | |
863 | ||
864 | /* Warn about comment starter embedded in comment. */ | |
865 | if (cur[-2] == '/' && CPP_OPTION (pfile, warn_comments)) | |
866 | cpp_warning_with_line (pfile, CPP_BUFFER (pfile)->lineno, | |
867 | cur - CPP_BUFFER (pfile)->line_base, | |
868 | "'/*' within comment"); | |
45b966db | 869 | } |
45b966db | 870 | } |
61474454 NB |
871 | |
872 | cpp_error_with_line (pfile, line, col, "unterminated comment"); | |
873 | cur--; | |
874 | out: | |
875 | CPP_BUFFER (pfile)->cur = cur + 1; | |
45b966db ZW |
876 | } |
877 | ||
878 | /* Skip a C++/Chill line comment. We know it's a comment, and point | |
879 | is at the second character of the initiator. */ | |
880 | static void | |
881 | skip_line_comment (pfile) | |
882 | cpp_reader *pfile; | |
883 | { | |
884 | FORWARD(1); | |
885 | for (;;) | |
886 | { | |
887 | int c = GETC (); | |
888 | ||
889 | /* We don't have to worry about EOF in here. */ | |
890 | if (c == '\n') | |
891 | { | |
892 | /* Don't consider final '\n' to be part of comment. */ | |
893 | FORWARD(-1); | |
894 | return; | |
895 | } | |
896 | else if (c == '\r') | |
897 | { | |
898 | /* \r cannot be a macro escape marker here. */ | |
899 | if (!ACTIVE_MARK_P (pfile)) | |
900 | CPP_BUMP_LINE (pfile); | |
ae79697b | 901 | if (CPP_OPTION (pfile, warn_comments)) |
45b966db ZW |
902 | cpp_warning (pfile, "backslash-newline within line comment"); |
903 | } | |
904 | } | |
905 | } | |
906 | ||
907 | /* Skip a comment - C, C++, or Chill style. M is the first character | |
908 | of the comment marker. If this really is a comment, skip to its | |
909 | end and return ' '. If this is not a comment, return M (which will | |
910 | be '/' or '-'). */ | |
911 | ||
912 | static int | |
913 | skip_comment (pfile, m) | |
914 | cpp_reader *pfile; | |
915 | int m; | |
916 | { | |
917 | if (m == '/' && PEEKC() == '*') | |
918 | { | |
919 | skip_block_comment (pfile); | |
920 | return ' '; | |
921 | } | |
922 | else if (m == '/' && PEEKC() == '/') | |
923 | { | |
c31a6508 | 924 | if (CPP_IN_SYSTEM_HEADER (pfile)) |
45b966db ZW |
925 | { |
926 | /* We silently allow C++ comments in system headers, irrespective | |
927 | of conformance mode, because lots of busted systems do that | |
928 | and trying to clean it up in fixincludes is a nightmare. */ | |
929 | skip_line_comment (pfile); | |
930 | return ' '; | |
931 | } | |
ae79697b | 932 | else if (CPP_OPTION (pfile, cplusplus_comments)) |
45b966db | 933 | { |
0f89df67 | 934 | if (! CPP_BUFFER (pfile)->warned_cplusplus_comments) |
45b966db | 935 | { |
0f89df67 ZW |
936 | if (CPP_WTRADITIONAL (pfile)) |
937 | cpp_pedwarn (pfile, | |
938 | "C++ style comments are not allowed in traditional C"); | |
939 | else if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile)) | |
940 | cpp_pedwarn (pfile, | |
941 | "C++ style comments are not allowed in ISO C89"); | |
942 | if (CPP_WTRADITIONAL (pfile) | |
943 | || (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile))) | |
944 | cpp_pedwarn (pfile, | |
45b966db ZW |
945 | "(this will be reported only once per input file)"); |
946 | CPP_BUFFER (pfile)->warned_cplusplus_comments = 1; | |
947 | } | |
948 | skip_line_comment (pfile); | |
949 | return ' '; | |
950 | } | |
951 | else | |
952 | return m; | |
953 | } | |
954 | else if (m == '-' && PEEKC() == '-' | |
ae79697b | 955 | && CPP_OPTION (pfile, chill)) |
45b966db ZW |
956 | { |
957 | skip_line_comment (pfile); | |
958 | return ' '; | |
959 | } | |
960 | else | |
961 | return m; | |
962 | } | |
963 | ||
964 | /* Identical to skip_comment except that it copies the comment into the | |
965 | token_buffer. This is used if !discard_comments. */ | |
966 | static int | |
967 | copy_comment (pfile, m) | |
968 | cpp_reader *pfile; | |
969 | int m; | |
970 | { | |
971 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
972 | const U_CHAR *limit; | |
973 | ||
974 | if (skip_comment (pfile, m) == m) | |
975 | return m; | |
976 | ||
977 | limit = CPP_BUFFER (pfile)->cur; | |
978 | CPP_RESERVE (pfile, limit - start + 2); | |
979 | CPP_PUTC_Q (pfile, m); | |
980 | for (; start <= limit; start++) | |
981 | if (*start != '\r') | |
982 | CPP_PUTC_Q (pfile, *start); | |
983 | ||
984 | return ' '; | |
985 | } | |
986 | ||
64aaf407 NB |
987 | static void |
988 | null_warning (pfile, count) | |
989 | cpp_reader *pfile; | |
990 | unsigned int count; | |
991 | { | |
992 | if (count == 1) | |
993 | cpp_warning (pfile, "embedded null character ignored"); | |
994 | else | |
995 | cpp_warning (pfile, "embedded null characters ignored"); | |
996 | } | |
997 | ||
45b966db ZW |
998 | /* Skip whitespace \-newline and comments. Does not macro-expand. */ |
999 | ||
1000 | void | |
1001 | _cpp_skip_hspace (pfile) | |
1002 | cpp_reader *pfile; | |
1003 | { | |
64aaf407 | 1004 | unsigned int null_count = 0; |
45b966db | 1005 | int c; |
64aaf407 | 1006 | |
45b966db ZW |
1007 | while (1) |
1008 | { | |
1009 | c = GETC(); | |
1010 | if (c == EOF) | |
64aaf407 | 1011 | goto out; |
45b966db ZW |
1012 | else if (is_hspace(c)) |
1013 | { | |
1014 | if ((c == '\f' || c == '\v') && CPP_PEDANTIC (pfile)) | |
1015 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
1016 | c == '\f' ? "formfeed" : "vertical tab"); | |
64aaf407 NB |
1017 | else if (c == '\0') |
1018 | null_count++; | |
45b966db ZW |
1019 | } |
1020 | else if (c == '\r') | |
1021 | { | |
1022 | /* \r is a backslash-newline marker if !has_escapes, and | |
1023 | a deletable-whitespace or no-reexpansion marker otherwise. */ | |
1024 | if (CPP_BUFFER (pfile)->has_escapes) | |
1025 | { | |
1026 | if (PEEKC() == ' ') | |
1027 | FORWARD(1); | |
1028 | else | |
1029 | break; | |
1030 | } | |
1031 | else | |
1032 | CPP_BUMP_LINE (pfile); | |
1033 | } | |
1034 | else if (c == '/' || c == '-') | |
1035 | { | |
1036 | c = skip_comment (pfile, c); | |
1037 | if (c != ' ') | |
1038 | break; | |
1039 | } | |
1040 | else | |
1041 | break; | |
1042 | } | |
1043 | FORWARD(-1); | |
64aaf407 NB |
1044 | out: |
1045 | if (null_count) | |
1046 | null_warning (pfile, null_count); | |
45b966db ZW |
1047 | } |
1048 | ||
1049 | /* Read and discard the rest of the current line. */ | |
1050 | ||
1051 | void | |
1052 | _cpp_skip_rest_of_line (pfile) | |
1053 | cpp_reader *pfile; | |
1054 | { | |
1055 | for (;;) | |
1056 | { | |
1057 | int c = GETC(); | |
1058 | switch (c) | |
1059 | { | |
1060 | case '\n': | |
1061 | FORWARD(-1); | |
1062 | case EOF: | |
1063 | return; | |
1064 | ||
1065 | case '\r': | |
1066 | if (! CPP_BUFFER (pfile)->has_escapes) | |
1067 | CPP_BUMP_LINE (pfile); | |
1068 | break; | |
1069 | ||
1070 | case '\'': | |
1071 | case '\"': | |
1072 | skip_string (pfile, c); | |
1073 | break; | |
1074 | ||
1075 | case '/': | |
1076 | case '-': | |
1077 | skip_comment (pfile, c); | |
1078 | break; | |
1079 | ||
1080 | case '\f': | |
1081 | case '\v': | |
1082 | if (CPP_PEDANTIC (pfile)) | |
1083 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
1084 | c == '\f' ? "formfeed" : "vertical tab"); | |
1085 | break; | |
1086 | ||
1087 | } | |
1088 | } | |
1089 | } | |
1090 | ||
1091 | /* Parse an identifier starting with C. */ | |
1092 | ||
1093 | void | |
1094 | _cpp_parse_name (pfile, c) | |
1095 | cpp_reader *pfile; | |
1096 | int c; | |
1097 | { | |
1098 | for (;;) | |
1099 | { | |
1100 | if (! is_idchar(c)) | |
1101 | { | |
1102 | FORWARD (-1); | |
1103 | break; | |
1104 | } | |
1105 | ||
e5ec2402 ZW |
1106 | /* $ is not a legal identifier character in the standard, but is |
1107 | commonly accepted as an extension. Don't warn about it in | |
1108 | skipped conditional blocks. */ | |
1109 | if (c == '$' && CPP_PEDANTIC (pfile) && ! pfile->skipping) | |
45b966db ZW |
1110 | cpp_pedwarn (pfile, "`$' in identifier"); |
1111 | ||
1112 | CPP_RESERVE(pfile, 2); /* One more for final NUL. */ | |
1113 | CPP_PUTC_Q (pfile, c); | |
1114 | c = GETC(); | |
1115 | if (c == EOF) | |
1116 | break; | |
1117 | } | |
45b966db ZW |
1118 | return; |
1119 | } | |
1120 | ||
1121 | /* Parse and skip over a string starting with C. A single quoted | |
1122 | string is treated like a double -- some programs (e.g., troff) are | |
1123 | perverse this way. (However, a single quoted string is not allowed | |
1124 | to extend over multiple lines.) */ | |
1125 | static void | |
1126 | skip_string (pfile, c) | |
1127 | cpp_reader *pfile; | |
1128 | int c; | |
1129 | { | |
3a2b2c7a | 1130 | unsigned int start_line, start_column; |
64aaf407 | 1131 | unsigned int null_count = 0; |
45b966db | 1132 | |
3a2b2c7a ZW |
1133 | start_line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
1134 | start_column = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
45b966db ZW |
1135 | while (1) |
1136 | { | |
1137 | int cc = GETC(); | |
1138 | switch (cc) | |
1139 | { | |
1140 | case EOF: | |
1141 | cpp_error_with_line (pfile, start_line, start_column, | |
1142 | "unterminated string or character constant"); | |
1143 | if (pfile->multiline_string_line != start_line | |
1144 | && pfile->multiline_string_line != 0) | |
1145 | cpp_error_with_line (pfile, | |
1146 | pfile->multiline_string_line, -1, | |
1147 | "possible real start of unterminated constant"); | |
1148 | pfile->multiline_string_line = 0; | |
64aaf407 | 1149 | goto out; |
45b966db | 1150 | |
64aaf407 NB |
1151 | case '\0': |
1152 | null_count++; | |
1153 | break; | |
1154 | ||
45b966db ZW |
1155 | case '\n': |
1156 | CPP_BUMP_LINE (pfile); | |
1157 | /* In Fortran and assembly language, silently terminate | |
1158 | strings of either variety at end of line. This is a | |
1159 | kludge around not knowing where comments are in these | |
1160 | languages. */ | |
ae79697b ZW |
1161 | if (CPP_OPTION (pfile, lang_fortran) |
1162 | || CPP_OPTION (pfile, lang_asm)) | |
45b966db ZW |
1163 | { |
1164 | FORWARD(-1); | |
64aaf407 | 1165 | goto out; |
45b966db ZW |
1166 | } |
1167 | /* Character constants may not extend over multiple lines. | |
1168 | In Standard C, neither may strings. We accept multiline | |
1169 | strings as an extension. */ | |
1170 | if (c == '\'') | |
1171 | { | |
1172 | cpp_error_with_line (pfile, start_line, start_column, | |
1173 | "unterminated character constant"); | |
1174 | FORWARD(-1); | |
64aaf407 | 1175 | goto out; |
45b966db ZW |
1176 | } |
1177 | if (CPP_PEDANTIC (pfile) && pfile->multiline_string_line == 0) | |
1178 | cpp_pedwarn_with_line (pfile, start_line, start_column, | |
1179 | "string constant runs past end of line"); | |
1180 | if (pfile->multiline_string_line == 0) | |
1181 | pfile->multiline_string_line = start_line; | |
1182 | break; | |
1183 | ||
1184 | case '\r': | |
1185 | if (CPP_BUFFER (pfile)->has_escapes) | |
1186 | { | |
1187 | cpp_ice (pfile, "\\r escape inside string constant"); | |
1188 | FORWARD(1); | |
1189 | } | |
1190 | else | |
1191 | /* Backslash newline is replaced by nothing at all. */ | |
1192 | CPP_BUMP_LINE (pfile); | |
1193 | break; | |
1194 | ||
1195 | case '\\': | |
1196 | FORWARD(1); | |
1197 | break; | |
1198 | ||
1199 | case '\"': | |
1200 | case '\'': | |
1201 | if (cc == c) | |
64aaf407 | 1202 | goto out; |
45b966db ZW |
1203 | break; |
1204 | } | |
1205 | } | |
64aaf407 NB |
1206 | |
1207 | out: | |
1208 | if (null_count == 1) | |
1209 | cpp_warning (pfile, "null character in string or character constant"); | |
1210 | else if (null_count > 1) | |
1211 | cpp_warning (pfile, "null characters in string or character constant"); | |
45b966db ZW |
1212 | } |
1213 | ||
1214 | /* Parse a string and copy it to the output. */ | |
1215 | ||
1216 | static void | |
1217 | parse_string (pfile, c) | |
1218 | cpp_reader *pfile; | |
1219 | int c; | |
1220 | { | |
1221 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
1222 | const U_CHAR *limit; | |
1223 | ||
1224 | skip_string (pfile, c); | |
1225 | ||
1226 | limit = CPP_BUFFER (pfile)->cur; | |
1227 | CPP_RESERVE (pfile, limit - start + 2); | |
1228 | CPP_PUTC_Q (pfile, c); | |
1229 | for (; start < limit; start++) | |
1230 | if (*start != '\r') | |
1231 | CPP_PUTC_Q (pfile, *start); | |
1232 | } | |
1233 | ||
45b966db ZW |
1234 | /* Get the next token, and add it to the text in pfile->token_buffer. |
1235 | Return the kind of token we got. */ | |
1236 | ||
3a2b2c7a | 1237 | enum cpp_ttype |
45b966db ZW |
1238 | _cpp_lex_token (pfile) |
1239 | cpp_reader *pfile; | |
1240 | { | |
5eec0563 | 1241 | register int c, c2; |
3a2b2c7a | 1242 | enum cpp_ttype token; |
45b966db | 1243 | |
f2d5f0cc ZW |
1244 | if (CPP_BUFFER (pfile) == NULL) |
1245 | return CPP_EOF; | |
1246 | ||
45b966db ZW |
1247 | get_next: |
1248 | c = GETC(); | |
1249 | switch (c) | |
1250 | { | |
1251 | case EOF: | |
1252 | return CPP_EOF; | |
1253 | ||
1254 | case '/': | |
1255 | if (PEEKC () == '=') | |
1256 | goto op2; | |
1257 | ||
1258 | comment: | |
ae79697b | 1259 | if (CPP_OPTION (pfile, discard_comments)) |
45b966db ZW |
1260 | c = skip_comment (pfile, c); |
1261 | else | |
1262 | c = copy_comment (pfile, c); | |
1263 | if (c != ' ') | |
1264 | goto randomchar; | |
1265 | ||
1266 | /* Comments are equivalent to spaces. | |
1267 | For -traditional, a comment is equivalent to nothing. */ | |
ff2b53ef | 1268 | if (!CPP_OPTION (pfile, discard_comments)) |
45b966db | 1269 | return CPP_COMMENT; |
9e62c811 | 1270 | else if (CPP_TRADITIONAL (pfile)) |
15dad1d9 | 1271 | goto get_next; |
45b966db ZW |
1272 | else |
1273 | { | |
1274 | CPP_PUTC (pfile, c); | |
1275 | return CPP_HSPACE; | |
1276 | } | |
1277 | ||
1278 | case '#': | |
5eec0563 JM |
1279 | CPP_PUTC (pfile, c); |
1280 | ||
1281 | hash: | |
15dad1d9 ZW |
1282 | c2 = PEEKC (); |
1283 | if (c2 == '#') | |
45b966db | 1284 | { |
15dad1d9 ZW |
1285 | FORWARD (1); |
1286 | CPP_PUTC (pfile, c2); | |
1287 | return CPP_PASTE; | |
45b966db | 1288 | } |
15dad1d9 | 1289 | else if (c2 == '%' && PEEKN (1) == ':') |
45b966db | 1290 | { |
15dad1d9 ZW |
1291 | /* Digraph: "%:" == "#". */ |
1292 | FORWARD (1); | |
1293 | CPP_RESERVE (pfile, 2); | |
1294 | CPP_PUTC_Q (pfile, c2); | |
1295 | CPP_PUTC_Q (pfile, GETC ()); | |
1368ee70 | 1296 | return CPP_PASTE; |
45b966db | 1297 | } |
15dad1d9 ZW |
1298 | else |
1299 | return CPP_HASH; | |
45b966db ZW |
1300 | |
1301 | case '\"': | |
1302 | case '\'': | |
1303 | parse_string (pfile, c); | |
45b966db ZW |
1304 | return c == '\'' ? CPP_CHAR : CPP_STRING; |
1305 | ||
1306 | case '$': | |
ae79697b | 1307 | if (!CPP_OPTION (pfile, dollars_in_ident)) |
45b966db ZW |
1308 | goto randomchar; |
1309 | goto letter; | |
1310 | ||
1311 | case ':': | |
5eec0563 JM |
1312 | c2 = PEEKC (); |
1313 | /* Digraph: ":>" == "]". */ | |
1314 | if (c2 == '>' | |
1315 | || (c2 == ':' && CPP_OPTION (pfile, cplusplus))) | |
45b966db ZW |
1316 | goto op2; |
1317 | goto randomchar; | |
1318 | ||
1319 | case '&': | |
1320 | case '+': | |
1321 | case '|': | |
1322 | c2 = PEEKC (); | |
1323 | if (c2 == c || c2 == '=') | |
1324 | goto op2; | |
1325 | goto randomchar; | |
1326 | ||
5eec0563 JM |
1327 | case '%': |
1328 | /* Digraphs: "%:" == "#", "%>" == "}". */ | |
1329 | c2 = PEEKC (); | |
1330 | if (c2 == ':') | |
1331 | { | |
1332 | FORWARD (1); | |
1333 | CPP_RESERVE (pfile, 2); | |
1334 | CPP_PUTC_Q (pfile, c); | |
1335 | CPP_PUTC_Q (pfile, c2); | |
1336 | goto hash; | |
1337 | } | |
1338 | else if (c2 == '>') | |
1339 | { | |
1340 | FORWARD (1); | |
1341 | CPP_RESERVE (pfile, 2); | |
1342 | CPP_PUTC_Q (pfile, c); | |
1343 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1344 | return CPP_OPEN_BRACE; |
5eec0563 JM |
1345 | } |
1346 | /* else fall through */ | |
1347 | ||
45b966db ZW |
1348 | case '*': |
1349 | case '!': | |
45b966db ZW |
1350 | case '=': |
1351 | case '^': | |
1352 | if (PEEKC () == '=') | |
1353 | goto op2; | |
1354 | goto randomchar; | |
1355 | ||
1356 | case '-': | |
1357 | c2 = PEEKC (); | |
1358 | if (c2 == '-') | |
1359 | { | |
ae79697b | 1360 | if (CPP_OPTION (pfile, chill)) |
45b966db ZW |
1361 | goto comment; /* Chill style comment */ |
1362 | else | |
1363 | goto op2; | |
1364 | } | |
1365 | else if (c2 == '=') | |
1366 | goto op2; | |
1367 | else if (c2 == '>') | |
1368 | { | |
ae79697b | 1369 | if (CPP_OPTION (pfile, cplusplus) && PEEKN (1) == '*') |
45b966db ZW |
1370 | { |
1371 | /* In C++, there's a ->* operator. */ | |
1372 | token = CPP_OTHER; | |
45b966db ZW |
1373 | CPP_RESERVE (pfile, 4); |
1374 | CPP_PUTC_Q (pfile, c); | |
1375 | CPP_PUTC_Q (pfile, GETC ()); | |
1376 | CPP_PUTC_Q (pfile, GETC ()); | |
45b966db ZW |
1377 | return token; |
1378 | } | |
1379 | goto op2; | |
1380 | } | |
1381 | goto randomchar; | |
1382 | ||
1383 | case '<': | |
1384 | if (pfile->parsing_include_directive) | |
1385 | { | |
1386 | for (;;) | |
1387 | { | |
1388 | CPP_PUTC (pfile, c); | |
1389 | if (c == '>') | |
1390 | break; | |
1391 | c = GETC (); | |
1392 | if (c == '\n' || c == EOF) | |
1393 | { | |
1394 | cpp_error (pfile, | |
1395 | "missing '>' in `#include <FILENAME>'"); | |
1396 | break; | |
1397 | } | |
1398 | else if (c == '\r') | |
1399 | { | |
1400 | if (!CPP_BUFFER (pfile)->has_escapes) | |
1401 | { | |
1402 | /* Backslash newline is replaced by nothing. */ | |
1403 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1404 | CPP_BUMP_LINE (pfile); | |
1405 | } | |
1406 | else | |
1407 | { | |
1408 | /* We might conceivably get \r- or \r<space> in | |
1409 | here. Just delete 'em. */ | |
1410 | int d = GETC(); | |
1411 | if (d != '-' && d != ' ') | |
1412 | cpp_ice (pfile, "unrecognized escape \\r%c", d); | |
1413 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1414 | } | |
1415 | } | |
1416 | } | |
1417 | return CPP_STRING; | |
1418 | } | |
5eec0563 JM |
1419 | /* Digraphs: "<%" == "{", "<:" == "[". */ |
1420 | c2 = PEEKC (); | |
1421 | if (c2 == '%') | |
1422 | { | |
1423 | FORWARD (1); | |
1424 | CPP_RESERVE (pfile, 2); | |
1425 | CPP_PUTC_Q (pfile, c); | |
1426 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1427 | return CPP_CLOSE_BRACE; |
5eec0563 JM |
1428 | } |
1429 | else if (c2 == ':') | |
1430 | goto op2; | |
45b966db ZW |
1431 | /* else fall through */ |
1432 | case '>': | |
1433 | c2 = PEEKC (); | |
1434 | if (c2 == '=') | |
1435 | goto op2; | |
1436 | /* GNU C++ supports MIN and MAX operators <? and >?. */ | |
ae79697b | 1437 | if (c2 != c && (!CPP_OPTION (pfile, cplusplus) || c2 != '?')) |
45b966db ZW |
1438 | goto randomchar; |
1439 | FORWARD(1); | |
5eec0563 JM |
1440 | CPP_RESERVE (pfile, 3); |
1441 | CPP_PUTC_Q (pfile, c); | |
1442 | CPP_PUTC_Q (pfile, c2); | |
1443 | if (PEEKC () == '=') | |
45b966db | 1444 | CPP_PUTC_Q (pfile, GETC ()); |
45b966db ZW |
1445 | return CPP_OTHER; |
1446 | ||
1447 | case '.': | |
1448 | c2 = PEEKC (); | |
5eec0563 | 1449 | if (ISDIGIT (c2)) |
45b966db | 1450 | { |
5eec0563 | 1451 | CPP_PUTC (pfile, c); |
45b966db ZW |
1452 | c = GETC (); |
1453 | goto number; | |
1454 | } | |
1455 | ||
1456 | /* In C++ there's a .* operator. */ | |
ae79697b | 1457 | if (CPP_OPTION (pfile, cplusplus) && c2 == '*') |
45b966db ZW |
1458 | goto op2; |
1459 | ||
1460 | if (c2 == '.' && PEEKN(1) == '.') | |
1461 | { | |
5eec0563 | 1462 | CPP_RESERVE (pfile, 3); |
45b966db ZW |
1463 | CPP_PUTC_Q (pfile, '.'); |
1464 | CPP_PUTC_Q (pfile, '.'); | |
1465 | CPP_PUTC_Q (pfile, '.'); | |
1466 | FORWARD (2); | |
1368ee70 | 1467 | return CPP_ELLIPSIS; |
45b966db ZW |
1468 | } |
1469 | goto randomchar; | |
1470 | ||
1471 | op2: | |
5eec0563 | 1472 | CPP_RESERVE (pfile, 2); |
45b966db ZW |
1473 | CPP_PUTC_Q (pfile, c); |
1474 | CPP_PUTC_Q (pfile, GETC ()); | |
5eec0563 | 1475 | return CPP_OTHER; |
45b966db ZW |
1476 | |
1477 | case 'L': | |
1478 | c2 = PEEKC (); | |
1479 | if ((c2 == '\'' || c2 == '\"') && !CPP_TRADITIONAL (pfile)) | |
1480 | { | |
1481 | CPP_PUTC (pfile, c); | |
1482 | c = GETC (); | |
1483 | parse_string (pfile, c); | |
45b966db ZW |
1484 | return c == '\'' ? CPP_WCHAR : CPP_WSTRING; |
1485 | } | |
1486 | goto letter; | |
1487 | ||
1488 | case '0': case '1': case '2': case '3': case '4': | |
1489 | case '5': case '6': case '7': case '8': case '9': | |
1490 | number: | |
1491 | c2 = '.'; | |
1492 | for (;;) | |
1493 | { | |
1494 | CPP_RESERVE (pfile, 2); | |
1495 | CPP_PUTC_Q (pfile, c); | |
1496 | c = PEEKC (); | |
1497 | if (c == EOF) | |
1498 | break; | |
1499 | if (!is_numchar(c) && c != '.' | |
1500 | && ((c2 != 'e' && c2 != 'E' | |
1501 | && ((c2 != 'p' && c2 != 'P') | |
ae79697b | 1502 | || CPP_OPTION (pfile, c89))) |
45b966db ZW |
1503 | || (c != '+' && c != '-'))) |
1504 | break; | |
1505 | FORWARD(1); | |
1506 | c2= c; | |
1507 | } | |
45b966db ZW |
1508 | return CPP_NUMBER; |
1509 | case 'b': case 'c': case 'd': case 'h': case 'o': | |
1510 | case 'B': case 'C': case 'D': case 'H': case 'O': | |
ae79697b | 1511 | if (CPP_OPTION (pfile, chill) && PEEKC () == '\'') |
45b966db | 1512 | { |
45b966db ZW |
1513 | CPP_RESERVE (pfile, 2); |
1514 | CPP_PUTC_Q (pfile, c); | |
1515 | CPP_PUTC_Q (pfile, '\''); | |
1516 | FORWARD(1); | |
1517 | for (;;) | |
1518 | { | |
1519 | c = GETC(); | |
1520 | if (c == EOF) | |
1521 | goto chill_number_eof; | |
1522 | if (!is_numchar(c)) | |
1523 | break; | |
1524 | CPP_PUTC (pfile, c); | |
1525 | } | |
1526 | if (c == '\'') | |
1527 | { | |
1528 | CPP_RESERVE (pfile, 2); | |
1529 | CPP_PUTC_Q (pfile, c); | |
45b966db ZW |
1530 | return CPP_STRING; |
1531 | } | |
1532 | else | |
1533 | { | |
1534 | FORWARD(-1); | |
1535 | chill_number_eof: | |
45b966db ZW |
1536 | return CPP_NUMBER; |
1537 | } | |
1538 | } | |
1539 | else | |
1540 | goto letter; | |
1541 | case '_': | |
1542 | case 'a': case 'e': case 'f': case 'g': case 'i': case 'j': | |
1543 | case 'k': case 'l': case 'm': case 'n': case 'p': case 'q': | |
1544 | case 'r': case 's': case 't': case 'u': case 'v': case 'w': | |
1545 | case 'x': case 'y': case 'z': | |
1546 | case 'A': case 'E': case 'F': case 'G': case 'I': case 'J': | |
1547 | case 'K': case 'M': case 'N': case 'P': case 'Q': case 'R': | |
1548 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1549 | case 'Y': case 'Z': | |
1550 | letter: | |
45b966db ZW |
1551 | _cpp_parse_name (pfile, c); |
1552 | return CPP_MACRO; | |
1553 | ||
64aaf407 NB |
1554 | case ' ': case '\t': case '\v': case '\f': case '\0': |
1555 | { | |
1556 | int null_count = 0; | |
1557 | ||
1558 | for (;;) | |
1559 | { | |
1560 | if (c == '\0') | |
1561 | null_count++; | |
1562 | else | |
1563 | CPP_PUTC (pfile, c); | |
1564 | c = PEEKC (); | |
1565 | if (c == EOF || !is_hspace(c)) | |
1566 | break; | |
1567 | FORWARD(1); | |
1568 | } | |
1569 | if (null_count) | |
1570 | null_warning (pfile, null_count); | |
1571 | return CPP_HSPACE; | |
1572 | } | |
45b966db ZW |
1573 | |
1574 | case '\r': | |
1575 | if (CPP_BUFFER (pfile)->has_escapes) | |
1576 | { | |
1577 | c = GETC (); | |
1578 | if (c == '-') | |
1579 | { | |
1580 | if (pfile->output_escapes) | |
1581 | CPP_PUTS (pfile, "\r-", 2); | |
1582 | _cpp_parse_name (pfile, GETC ()); | |
1583 | return CPP_NAME; | |
1584 | } | |
1585 | else if (c == ' ') | |
1586 | { | |
ff2b53ef ZW |
1587 | /* "\r " means a space, but only if necessary to prevent |
1588 | accidental token concatenation. */ | |
45b966db ZW |
1589 | CPP_RESERVE (pfile, 2); |
1590 | if (pfile->output_escapes) | |
1591 | CPP_PUTC_Q (pfile, '\r'); | |
1592 | CPP_PUTC_Q (pfile, c); | |
1593 | return CPP_HSPACE; | |
1594 | } | |
1595 | else | |
1596 | { | |
1597 | cpp_ice (pfile, "unrecognized escape \\r%c", c); | |
1598 | goto get_next; | |
1599 | } | |
1600 | } | |
1601 | else | |
1602 | { | |
1603 | /* Backslash newline is ignored. */ | |
cbccf5e8 MM |
1604 | if (!ACTIVE_MARK_P (pfile)) |
1605 | CPP_BUMP_LINE (pfile); | |
45b966db ZW |
1606 | goto get_next; |
1607 | } | |
1608 | ||
1609 | case '\n': | |
1610 | CPP_PUTC (pfile, c); | |
45b966db ZW |
1611 | return CPP_VSPACE; |
1612 | ||
1368ee70 ZW |
1613 | case '(': token = CPP_OPEN_PAREN; goto char1; |
1614 | case ')': token = CPP_CLOSE_PAREN; goto char1; | |
1615 | case '{': token = CPP_OPEN_BRACE; goto char1; | |
1616 | case '}': token = CPP_CLOSE_BRACE; goto char1; | |
1617 | case ',': token = CPP_COMMA; goto char1; | |
1618 | case ';': token = CPP_SEMICOLON; goto char1; | |
45b966db ZW |
1619 | |
1620 | randomchar: | |
1621 | default: | |
1622 | token = CPP_OTHER; | |
1623 | char1: | |
45b966db ZW |
1624 | CPP_PUTC (pfile, c); |
1625 | return token; | |
1626 | } | |
1627 | } | |
1628 | ||
1629 | /* Check for and expand a macro, which is from WRITTEN to CPP_WRITTEN (pfile). | |
1630 | Caller is expected to have checked no_macro_expand. */ | |
1631 | static int | |
1632 | maybe_macroexpand (pfile, written) | |
1633 | cpp_reader *pfile; | |
1634 | long written; | |
1635 | { | |
1636 | U_CHAR *macro = pfile->token_buffer + written; | |
1637 | size_t len = CPP_WRITTEN (pfile) - written; | |
f8f769ea | 1638 | cpp_hashnode *hp = cpp_lookup (pfile, macro, len); |
45b966db | 1639 | |
f8f769ea | 1640 | /* cpp_lookup never returns null. */ |
a7abcbbf | 1641 | if (hp->type == T_VOID) |
45b966db | 1642 | return 0; |
d9e0bd53 | 1643 | if (hp->disabled || hp->type == T_IDENTITY) |
45b966db ZW |
1644 | { |
1645 | if (pfile->output_escapes) | |
1646 | { | |
1647 | /* Insert a no-reexpand marker before IDENT. */ | |
1648 | CPP_RESERVE (pfile, 2); | |
1649 | CPP_ADJUST_WRITTEN (pfile, 2); | |
1650 | macro = pfile->token_buffer + written; | |
1651 | ||
1652 | memmove (macro + 2, macro, len); | |
1653 | macro[0] = '\r'; | |
1654 | macro[1] = '-'; | |
1655 | } | |
1656 | return 0; | |
1657 | } | |
ff2b53ef ZW |
1658 | if (hp->type == T_EMPTY) |
1659 | { | |
1660 | /* Special case optimization: macro expands to nothing. */ | |
1661 | CPP_SET_WRITTEN (pfile, written); | |
1662 | CPP_PUTC_Q (pfile, ' '); | |
1663 | return 1; | |
1664 | } | |
45b966db ZW |
1665 | |
1666 | /* If macro wants an arglist, verify that a '(' follows. */ | |
d9e0bd53 | 1667 | if (hp->type == T_FMACRO) |
45b966db ZW |
1668 | { |
1669 | int macbuf_whitespace = 0; | |
1670 | int c; | |
1671 | ||
1672 | while (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
1673 | { | |
1674 | const U_CHAR *point = CPP_BUFFER (pfile)->cur; | |
1675 | for (;;) | |
1676 | { | |
1677 | _cpp_skip_hspace (pfile); | |
1678 | c = PEEKC (); | |
1679 | if (c == '\n') | |
1680 | FORWARD(1); | |
1681 | else | |
1682 | break; | |
1683 | } | |
1684 | if (point != CPP_BUFFER (pfile)->cur) | |
1685 | macbuf_whitespace = 1; | |
1686 | if (c == '(') | |
1687 | goto is_macro_call; | |
1688 | else if (c != EOF) | |
1689 | goto not_macro_call; | |
1690 | cpp_pop_buffer (pfile); | |
1691 | } | |
1692 | ||
1693 | CPP_SET_MARK (pfile); | |
1694 | for (;;) | |
1695 | { | |
1696 | _cpp_skip_hspace (pfile); | |
1697 | c = PEEKC (); | |
1698 | if (c == '\n') | |
1699 | FORWARD(1); | |
1700 | else | |
1701 | break; | |
1702 | } | |
1703 | CPP_GOTO_MARK (pfile); | |
1704 | ||
1705 | if (c != '(') | |
1706 | { | |
1707 | not_macro_call: | |
1708 | if (macbuf_whitespace) | |
1709 | CPP_PUTC (pfile, ' '); | |
476f2869 ZW |
1710 | |
1711 | /* K+R treated this as a hard error. */ | |
d7a2e0f7 | 1712 | if (CPP_WTRADITIONAL (pfile)) |
476f2869 | 1713 | cpp_warning (pfile, |
d7a2e0f7 | 1714 | "function macro %s must be used with arguments in traditional C", |
476f2869 | 1715 | hp->name); |
45b966db ZW |
1716 | return 0; |
1717 | } | |
1718 | } | |
1719 | ||
1720 | is_macro_call: | |
1721 | /* This is now known to be a macro call. | |
1722 | Expand the macro, reading arguments as needed, | |
1723 | and push the expansion on the input stack. */ | |
1724 | _cpp_macroexpand (pfile, hp); | |
1725 | CPP_SET_WRITTEN (pfile, written); | |
1726 | return 1; | |
1727 | } | |
1728 | ||
9e62c811 ZW |
1729 | /* Complain about \v or \f in a preprocessing directive (constraint |
1730 | violation, C99 6.10 para 5). Caller has checked CPP_PEDANTIC. */ | |
1731 | static void | |
1732 | pedantic_whitespace (pfile, p, len) | |
1733 | cpp_reader *pfile; | |
1734 | U_CHAR *p; | |
1735 | unsigned int len; | |
1736 | { | |
1737 | while (len) | |
1738 | { | |
1739 | if (*p == '\v') | |
1740 | cpp_pedwarn (pfile, "vertical tab in preprocessing directive"); | |
1741 | else if (*p == '\f') | |
1742 | cpp_pedwarn (pfile, "form feed in preprocessing directive"); | |
1743 | p++; | |
1744 | len--; | |
1745 | } | |
1746 | } | |
1747 | ||
1748 | ||
3a2b2c7a | 1749 | enum cpp_ttype |
45b966db ZW |
1750 | cpp_get_token (pfile) |
1751 | cpp_reader *pfile; | |
1752 | { | |
3a2b2c7a | 1753 | enum cpp_ttype token; |
45b966db | 1754 | long written = CPP_WRITTEN (pfile); |
ea4a453b | 1755 | int macro_buffer; |
45b966db ZW |
1756 | |
1757 | get_next: | |
1758 | token = _cpp_lex_token (pfile); | |
1759 | ||
1760 | switch (token) | |
1761 | { | |
1762 | default: | |
ea4a453b ZW |
1763 | if (pfile->skipping) |
1764 | break; | |
ff2b53ef ZW |
1765 | pfile->potential_control_macro = 0; |
1766 | pfile->only_seen_white = 0; | |
ea4a453b ZW |
1767 | break; |
1768 | ||
1769 | case CPP_HSPACE: | |
1770 | case CPP_COMMENT: | |
1771 | break; | |
ff2b53ef ZW |
1772 | |
1773 | case CPP_VSPACE: | |
1774 | if (pfile->only_seen_white == 0) | |
1775 | pfile->only_seen_white = 1; | |
1776 | CPP_BUMP_LINE (pfile); | |
ea4a453b | 1777 | break; |
45b966db | 1778 | |
15dad1d9 | 1779 | case CPP_HASH: |
ff2b53ef | 1780 | pfile->potential_control_macro = 0; |
15dad1d9 | 1781 | if (!pfile->only_seen_white) |
ea4a453b | 1782 | break; |
15dad1d9 ZW |
1783 | /* XXX shouldn't have to do this - remove the hash or %: from |
1784 | the token buffer. */ | |
1785 | if (CPP_PWRITTEN (pfile)[-1] == '#') | |
1786 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1787 | else | |
1788 | CPP_ADJUST_WRITTEN (pfile, -2); | |
1789 | ||
45b966db | 1790 | if (_cpp_handle_directive (pfile)) |
ea4a453b ZW |
1791 | { |
1792 | token = CPP_DIRECTIVE; | |
1793 | break; | |
1794 | } | |
45b966db ZW |
1795 | pfile->only_seen_white = 0; |
1796 | CPP_PUTC (pfile, '#'); | |
ea4a453b | 1797 | break; |
45b966db ZW |
1798 | |
1799 | case CPP_MACRO: | |
ea4a453b ZW |
1800 | if (pfile->skipping) |
1801 | break; | |
ff2b53ef ZW |
1802 | pfile->potential_control_macro = 0; |
1803 | pfile->only_seen_white = 0; | |
45b966db ZW |
1804 | if (! pfile->no_macro_expand |
1805 | && maybe_macroexpand (pfile, written)) | |
1806 | goto get_next; | |
ea4a453b ZW |
1807 | token = CPP_NAME; |
1808 | break; | |
45b966db | 1809 | |
ea4a453b | 1810 | /* Do not run this case through the 'skipping' logic. */ |
45b966db | 1811 | case CPP_EOF: |
f2d5f0cc ZW |
1812 | if (CPP_BUFFER (pfile) == NULL) |
1813 | return CPP_EOF; | |
ea4a453b ZW |
1814 | macro_buffer = CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile)); |
1815 | ||
c56c2073 | 1816 | cpp_pop_buffer (pfile); |
ea4a453b ZW |
1817 | if (macro_buffer) |
1818 | goto get_next; | |
c56c2073 | 1819 | return CPP_EOF; |
45b966db | 1820 | } |
ea4a453b ZW |
1821 | |
1822 | if (pfile->skipping) | |
1823 | { | |
1824 | CPP_SET_WRITTEN (pfile, written); | |
1825 | goto get_next; | |
1826 | } | |
1827 | return token; | |
45b966db ZW |
1828 | } |
1829 | ||
1830 | /* Like cpp_get_token, but skip spaces and comments. */ | |
1831 | ||
3a2b2c7a | 1832 | enum cpp_ttype |
45b966db ZW |
1833 | cpp_get_non_space_token (pfile) |
1834 | cpp_reader *pfile; | |
1835 | { | |
1836 | int old_written = CPP_WRITTEN (pfile); | |
1837 | for (;;) | |
1838 | { | |
3a2b2c7a | 1839 | enum cpp_ttype token = cpp_get_token (pfile); |
ff2b53ef | 1840 | if (token != CPP_COMMENT && token != CPP_HSPACE && token != CPP_VSPACE) |
45b966db ZW |
1841 | return token; |
1842 | CPP_SET_WRITTEN (pfile, old_written); | |
1843 | } | |
1844 | } | |
1845 | ||
ff2b53ef | 1846 | /* Like cpp_get_token, except that it does not execute directives, |
c56c2073 | 1847 | does not consume vertical space, and discards horizontal space. */ |
3a2b2c7a | 1848 | enum cpp_ttype |
9e62c811 | 1849 | _cpp_get_directive_token (pfile) |
45b966db ZW |
1850 | cpp_reader *pfile; |
1851 | { | |
ff2b53ef | 1852 | long old_written; |
3a2b2c7a | 1853 | enum cpp_ttype token; |
57c578a6 | 1854 | int at_bol; |
45b966db | 1855 | |
ff2b53ef | 1856 | get_next: |
57c578a6 | 1857 | at_bol = (CPP_BUFFER (pfile)->cur == CPP_BUFFER (pfile)->line_base); |
ff2b53ef ZW |
1858 | old_written = CPP_WRITTEN (pfile); |
1859 | token = _cpp_lex_token (pfile); | |
1860 | switch (token) | |
45b966db | 1861 | { |
ff2b53ef ZW |
1862 | default: |
1863 | return token; | |
45b966db | 1864 | |
ff2b53ef ZW |
1865 | case CPP_VSPACE: |
1866 | /* Put it back and return VSPACE. */ | |
1867 | FORWARD(-1); | |
1868 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1869 | return CPP_VSPACE; | |
45b966db | 1870 | |
ff2b53ef | 1871 | case CPP_HSPACE: |
57c578a6 ZW |
1872 | /* The purpose of this rather strange check is to prevent pedantic |
1873 | warnings for ^L in an #ifdefed out block. */ | |
1874 | if (CPP_PEDANTIC (pfile) && ! at_bol) | |
9e62c811 ZW |
1875 | pedantic_whitespace (pfile, pfile->token_buffer + old_written, |
1876 | CPP_WRITTEN (pfile) - old_written); | |
1877 | CPP_SET_WRITTEN (pfile, old_written); | |
1878 | goto get_next; | |
ff2b53ef | 1879 | return CPP_HSPACE; |
45b966db | 1880 | |
ff2b53ef ZW |
1881 | case CPP_MACRO: |
1882 | if (! pfile->no_macro_expand | |
1883 | && maybe_macroexpand (pfile, old_written)) | |
1884 | goto get_next; | |
1885 | return CPP_NAME; | |
45b966db | 1886 | |
ff2b53ef ZW |
1887 | case CPP_EOF: |
1888 | if (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
45b966db | 1889 | { |
ff2b53ef ZW |
1890 | cpp_pop_buffer (pfile); |
1891 | goto get_next; | |
45b966db | 1892 | } |
ff2b53ef ZW |
1893 | else |
1894 | /* This can happen for files that don't end with a newline, | |
1895 | and for cpp_define and friends. Pretend they do, so | |
1896 | callers don't have to deal. A warning will be issued by | |
1897 | someone else, if necessary. */ | |
1898 | return CPP_VSPACE; | |
1899 | } | |
1900 | } | |
1901 | ||
45b966db ZW |
1902 | /* Determine the current line and column. Used only by read_and_prescan. */ |
1903 | static U_CHAR * | |
1904 | find_position (start, limit, linep) | |
1905 | U_CHAR *start; | |
1906 | U_CHAR *limit; | |
1907 | unsigned long *linep; | |
1908 | { | |
1909 | unsigned long line = *linep; | |
1910 | U_CHAR *lbase = start; | |
1911 | while (start < limit) | |
1912 | { | |
1913 | U_CHAR ch = *start++; | |
1914 | if (ch == '\n' || ch == '\r') | |
1915 | { | |
1916 | line++; | |
1917 | lbase = start; | |
1918 | } | |
1919 | } | |
1920 | *linep = line; | |
1921 | return lbase; | |
1922 | } | |
1923 | ||
f8f769ea | 1924 | /* The following table is used by _cpp_prescan. If we have |
2a87fbe8 ZW |
1925 | designated initializers, it can be constant data; otherwise, it is |
1926 | set up at runtime by _cpp_init_input_buffer. */ | |
46d07497 | 1927 | |
12cf91fe | 1928 | #if (GCC_VERSION >= 2007) |
2a87fbe8 | 1929 | #define init_chartab() /* nothing */ |
12cf91fe | 1930 | #define CHARTAB __extension__ static const U_CHAR chartab[UCHAR_MAX + 1] = { |
46d07497 ZW |
1931 | #define END }; |
1932 | #define s(p, v) [p] = v, | |
1933 | #else | |
12cf91fe | 1934 | #define CHARTAB static U_CHAR chartab[UCHAR_MAX + 1] = { 0 }; \ |
2a87fbe8 ZW |
1935 | static void init_chartab PARAMS ((void)) { \ |
1936 | unsigned char *x = chartab; | |
46d07497 ZW |
1937 | #define END } |
1938 | #define s(p, v) x[p] = v; | |
1939 | #endif | |
1940 | ||
1941 | /* Table of characters that can't be handled in the inner loop. | |
2a87fbe8 ZW |
1942 | Also contains the mapping between trigraph third characters and their |
1943 | replacements. */ | |
46d07497 ZW |
1944 | #define SPECCASE_CR 1 |
1945 | #define SPECCASE_BACKSLASH 2 | |
1946 | #define SPECCASE_QUESTION 3 | |
1947 | ||
2a87fbe8 | 1948 | CHARTAB |
46d07497 ZW |
1949 | s('\r', SPECCASE_CR) |
1950 | s('\\', SPECCASE_BACKSLASH) | |
1951 | s('?', SPECCASE_QUESTION) | |
46d07497 | 1952 | |
46d07497 ZW |
1953 | s('=', '#') s(')', ']') s('!', '|') |
1954 | s('(', '[') s('\'', '^') s('>', '}') | |
1955 | s('/', '\\') s('<', '{') s('-', '~') | |
1956 | END | |
1957 | ||
1958 | #undef CHARTAB | |
46d07497 ZW |
1959 | #undef END |
1960 | #undef s | |
1961 | ||
2a87fbe8 ZW |
1962 | #define NORMAL(c) ((chartab[c]) == 0 || (chartab[c]) > SPECCASE_QUESTION) |
1963 | #define NONTRI(c) ((c) <= SPECCASE_QUESTION) | |
1964 | ||
f8f769ea ZW |
1965 | /* Prescan pass over a file already loaded into BUF. This is |
1966 | translation phases 1 and 2 (C99 5.1.1.2). | |
1967 | ||
1968 | Convert end-of-line markers (\n, \r, \r\n, \n\r) to | |
45b966db ZW |
1969 | canonical form (\n). If enabled, convert and/or warn about |
1970 | trigraphs. Convert backslash-newline to a one-character escape | |
1971 | (\r) and remove it from "embarrassing" places (i.e. the middle of a | |
1972 | token). If there is no newline at the end of the file, add one and | |
1973 | warn. Returns -1 on failure, or the actual length of the data to | |
1974 | be scanned. | |
1975 | ||
1976 | This function does a lot of work, and can be a serious performance | |
1977 | bottleneck. It has been tuned heavily; make sure you understand it | |
1978 | before hacking. The common case - no trigraphs, Unix style line | |
1979 | breaks, backslash-newline set off by whitespace, newline at EOF - | |
1980 | has been optimized at the expense of the others. The performance | |
1981 | penalty for DOS style line breaks (\r\n) is about 15%. | |
1982 | ||
1983 | Warnings lose particularly heavily since we have to determine the | |
1984 | line number, which involves scanning from the beginning of the file | |
1985 | or from the last warning. The penalty for the absence of a newline | |
1986 | at the end of reload1.c is about 60%. (reload1.c is 329k.) | |
1987 | ||
1988 | If your file has more than one kind of end-of-line marker, you | |
f8f769ea ZW |
1989 | will get messed-up line numbering. */ |
1990 | ||
1991 | ssize_t | |
1992 | _cpp_prescan (pfile, fp, len) | |
45b966db ZW |
1993 | cpp_reader *pfile; |
1994 | cpp_buffer *fp; | |
f8f769ea | 1995 | ssize_t len; |
45b966db | 1996 | { |
f8f769ea ZW |
1997 | U_CHAR *buf, *op; |
1998 | const U_CHAR *ibase, *ip, *ilimit; | |
1999 | U_CHAR *line_base; | |
45b966db ZW |
2000 | unsigned long line; |
2001 | unsigned int deferred_newlines; | |
45b966db | 2002 | |
f8f769ea ZW |
2003 | /* Allocate an extra byte in case we must add a trailing \n. */ |
2004 | buf = (U_CHAR *) xmalloc (len + 1); | |
2005 | line_base = op = buf; | |
2006 | ip = ibase = fp->buf; | |
2007 | ilimit = ibase + len; | |
45b966db | 2008 | line = 1; |
f8f769ea | 2009 | deferred_newlines = 0; |
45b966db ZW |
2010 | |
2011 | for (;;) | |
2012 | { | |
f8f769ea | 2013 | const U_CHAR *iq; |
04e3ec78 | 2014 | |
f8f769ea ZW |
2015 | /* Deal with \-newline, potentially in the middle of a token. */ |
2016 | if (deferred_newlines) | |
45b966db | 2017 | { |
f8f769ea | 2018 | if (op != buf && ! is_space (op[-1]) && op[-1] != '\r') |
04e3ec78 | 2019 | { |
f8f769ea ZW |
2020 | /* Previous was not white space. Skip to white |
2021 | space, if we can, before outputting the \r's */ | |
2022 | iq = ip; | |
2023 | while (iq < ilimit | |
2024 | && *iq != ' ' | |
2025 | && *iq != '\t' | |
2026 | && *iq != '\n' | |
2027 | && NORMAL(*iq)) | |
2028 | iq++; | |
2029 | memcpy (op, ip, iq - ip); | |
2030 | op += iq - ip; | |
2031 | ip += iq - ip; | |
2032 | if (! NORMAL(*ip)) | |
2033 | goto do_speccase; | |
04e3ec78 | 2034 | } |
f8f769ea ZW |
2035 | while (deferred_newlines) |
2036 | deferred_newlines--, *op++ = '\r'; | |
45b966db ZW |
2037 | } |
2038 | ||
f8f769ea ZW |
2039 | /* Copy as much as we can without special treatment. */ |
2040 | iq = ip; | |
2041 | while (iq < ilimit && NORMAL (*iq)) iq++; | |
2042 | memcpy (op, ip, iq - ip); | |
2043 | op += iq - ip; | |
2044 | ip += iq - ip; | |
45b966db | 2045 | |
f8f769ea ZW |
2046 | do_speccase: |
2047 | if (ip >= ilimit) | |
2048 | break; | |
2049 | ||
2050 | switch (chartab[*ip++]) | |
2051 | { | |
2052 | case SPECCASE_CR: /* \r */ | |
2053 | if (ip[-2] != '\n') | |
45b966db | 2054 | { |
f8f769ea ZW |
2055 | if (ip < ilimit && *ip == '\n') |
2056 | ip++; | |
2057 | *op++ = '\n'; | |
45b966db | 2058 | } |
f8f769ea | 2059 | break; |
45b966db | 2060 | |
f8f769ea ZW |
2061 | case SPECCASE_BACKSLASH: /* \ */ |
2062 | backslash: | |
2063 | if (ip < ilimit) | |
45b966db | 2064 | { |
04e3ec78 | 2065 | if (*ip == '\n') |
45b966db | 2066 | { |
04e3ec78 | 2067 | deferred_newlines++; |
45b966db ZW |
2068 | ip++; |
2069 | if (*ip == '\r') ip++; | |
f8f769ea | 2070 | break; |
45b966db ZW |
2071 | } |
2072 | else if (*ip == '\r') | |
2073 | { | |
04e3ec78 | 2074 | deferred_newlines++; |
45b966db ZW |
2075 | ip++; |
2076 | if (*ip == '\n') ip++; | |
f8f769ea | 2077 | break; |
45b966db | 2078 | } |
f8f769ea | 2079 | } |
45b966db | 2080 | |
f8f769ea ZW |
2081 | *op++ = '\\'; |
2082 | break; | |
04e3ec78 | 2083 | |
f8f769ea ZW |
2084 | case SPECCASE_QUESTION: /* ? */ |
2085 | { | |
2086 | unsigned int d, t; | |
45b966db | 2087 | |
f8f769ea ZW |
2088 | *op++ = '?'; /* Normal non-trigraph case */ |
2089 | if (ip > ilimit - 2 || ip[0] != '?') | |
2090 | break; | |
2091 | ||
2092 | d = ip[1]; | |
2093 | t = chartab[d]; | |
2094 | if (NONTRI (t)) | |
2095 | break; | |
04e3ec78 | 2096 | |
f8f769ea ZW |
2097 | if (CPP_OPTION (pfile, warn_trigraphs)) |
2098 | { | |
2099 | unsigned long col; | |
2100 | line_base = find_position (line_base, op, &line); | |
2101 | col = op - line_base + 1; | |
ae79697b | 2102 | if (CPP_OPTION (pfile, trigraphs)) |
f8f769ea ZW |
2103 | cpp_warning_with_line (pfile, line, col, |
2104 | "trigraph ??%c converted to %c", d, t); | |
45b966db | 2105 | else |
f8f769ea ZW |
2106 | cpp_warning_with_line (pfile, line, col, |
2107 | "trigraph ??%c ignored", d); | |
2108 | } | |
2109 | ||
2110 | ip += 2; | |
2111 | if (CPP_OPTION (pfile, trigraphs)) | |
2112 | { | |
2113 | op[-1] = t; /* Overwrite '?' */ | |
2114 | if (t == '\\') | |
45b966db | 2115 | { |
f8f769ea ZW |
2116 | op--; |
2117 | goto backslash; | |
45b966db ZW |
2118 | } |
2119 | } | |
f8f769ea ZW |
2120 | else |
2121 | { | |
2122 | *op++ = '?'; | |
2123 | *op++ = d; | |
2124 | } | |
2125 | } | |
2126 | break; | |
45b966db ZW |
2127 | } |
2128 | } | |
2129 | ||
f8f769ea ZW |
2130 | #ifdef HAVE_MMAP_FILE |
2131 | if (fp->mapped) | |
2132 | munmap ((caddr_t) fp->buf, len); | |
2133 | else | |
2134 | #endif | |
2135 | free ((PTR) fp->buf); | |
45b966db | 2136 | |
45b966db ZW |
2137 | if (op[-1] != '\n') |
2138 | { | |
2139 | unsigned long col; | |
2140 | line_base = find_position (line_base, op, &line); | |
2141 | col = op - line_base + 1; | |
f6fab919 | 2142 | cpp_warning_with_line (pfile, line, col, "no newline at end of file"); |
45b966db ZW |
2143 | *op++ = '\n'; |
2144 | } | |
2145 | ||
f8f769ea | 2146 | fp->buf = buf; |
45b966db | 2147 | return op - buf; |
45b966db ZW |
2148 | } |
2149 | ||
2a87fbe8 ZW |
2150 | /* Allocate pfile->input_buffer, and initialize chartab[] |
2151 | if it hasn't happened already. */ | |
46d07497 | 2152 | |
45b966db ZW |
2153 | void |
2154 | _cpp_init_input_buffer (pfile) | |
2155 | cpp_reader *pfile; | |
2156 | { | |
2157 | U_CHAR *tmp; | |
2158 | ||
2a87fbe8 | 2159 | init_chartab (); |
d1d9a6bd | 2160 | _cpp_init_toklist (&pfile->directbuf, NO_DUMMY_TOKEN); |
04e3ec78 | 2161 | |
45b966db ZW |
2162 | /* Determine the appropriate size for the input buffer. Normal C |
2163 | source files are smaller than eight K. */ | |
04e3ec78 NB |
2164 | /* 8Kbytes of buffer proper, 1 to detect running off the end without |
2165 | address arithmetic all the time, and 3 for pushback during buffer | |
2166 | refill, in case there's a potential trigraph or end-of-line | |
2167 | digraph at the end of a block. */ | |
45b966db | 2168 | |
04e3ec78 | 2169 | tmp = (U_CHAR *) xmalloc (8192 + 1 + 3); |
45b966db ZW |
2170 | pfile->input_buffer = tmp; |
2171 | pfile->input_buffer_len = 8192; | |
2172 | } | |
c5a04734 | 2173 | |
6d2c2047 ZW |
2174 | /* Utility routine: |
2175 | Compares, in the manner of strcmp(3), the token beginning at TOKEN | |
2176 | and extending for LEN characters to the NUL-terminated string | |
2177 | STRING. Typical usage: | |
2178 | ||
2179 | if (! cpp_idcmp (pfile->token_buffer + here, CPP_WRITTEN (pfile) - here, | |
2180 | "inline")) | |
2181 | { ... } | |
2182 | */ | |
2183 | ||
2184 | int | |
2185 | cpp_idcmp (token, len, string) | |
2186 | const U_CHAR *token; | |
2187 | size_t len; | |
2188 | const char *string; | |
2189 | { | |
2190 | size_t len2 = strlen (string); | |
2191 | int r; | |
2192 | ||
2193 | if ((r = memcmp (token, string, MIN (len, len2)))) | |
2194 | return r; | |
2195 | ||
2196 | /* The longer of the two strings sorts after the shorter. */ | |
2197 | if (len == len2) | |
2198 | return 0; | |
2199 | else if (len < len2) | |
2200 | return -1; | |
2201 | else | |
2202 | return 1; | |
2203 | } | |
2204 | ||
b8f41010 | 2205 | #ifdef NEW_LEXER |
c5a04734 | 2206 | |
d6d5f795 NB |
2207 | /* Lexing algorithm. |
2208 | ||
2209 | The original lexer in cpplib was made up of two passes: a first pass | |
2210 | that replaced trigraphs and deleted esacped newlines, and a second | |
2211 | pass that tokenized the result of the first pass. Tokenisation was | |
2212 | performed by peeking at the next character in the input stream. For | |
6777db6d | 2213 | example, if the input stream contained "!=", the handler for the ! |
d6d5f795 | 2214 | character would peek at the next character, and if it were a '=' |
6777db6d NB |
2215 | would skip over it, and return a "!=" token, otherwise it would |
2216 | return just the "!" token. | |
d6d5f795 NB |
2217 | |
2218 | To implement a single-pass lexer, this peeking ahead is unworkable. | |
2219 | An arbitrary number of escaped newlines, and trigraphs (in particular | |
6777db6d NB |
2220 | ??/ which translates to the escape \), could separate the '!' and '=' |
2221 | in the input stream, yet the next token is still a "!=". | |
d6d5f795 NB |
2222 | |
2223 | Suppose instead that we lex by one logical line at a time, producing | |
6777db6d NB |
2224 | a token list or stack for each logical line, and when seeing the '!' |
2225 | push a CPP_NOT token on the list. Then if the '!' is part of a | |
2226 | longer token ("!=") we know we must see the remainder of the token by | |
2227 | the time we reach the end of the logical line. Thus we can have the | |
2228 | '=' handler look at the previous token (at the end of the list / top | |
2229 | of the stack) and see if it is a "!" token, and if so, instead of | |
2230 | pushing a "=" token revise the existing token to be a "!=" token. | |
d6d5f795 NB |
2231 | |
2232 | This works in the presence of escaped newlines, because the '\' would | |
2233 | have been pushed on the top of the stack as a CPP_BACKSLASH. The | |
2234 | newline ('\n' or '\r') handler looks at the token at the top of the | |
2235 | stack to see if it is a CPP_BACKSLASH, and if so discards both. | |
2236 | Otherwise it pushes the newline (CPP_VSPACE) token as normal. Hence | |
2237 | the '=' handler would never see any intervening escaped newlines. | |
2238 | ||
2239 | To make trigraphs work in this context, as in precedence trigraphs | |
2240 | are highest and converted before anything else, the '?' handler does | |
2241 | lookahead to see if it is a trigraph, and if so skips the trigraph | |
2242 | and pushes the token it represents onto the top of the stack. This | |
2243 | also works in the particular case of a CPP_BACKSLASH trigraph. | |
2244 | ||
2245 | To the preprocessor, whitespace is only significant to the point of | |
2246 | knowing whether whitespace precedes a particular token. For example, | |
2247 | the '=' handler needs to know whether there was whitespace between it | |
6777db6d | 2248 | and a "!" token on the top of the stack, to make the token conversion |
d6d5f795 NB |
2249 | decision correctly. So each token has a PREV_WHITESPACE flag to |
2250 | indicate this - the standard permits consecutive whitespace to be | |
2251 | regarded as a single space. The compiler front ends are not | |
2252 | interested in whitespace at all; they just require a token stream. | |
2253 | Another place where whitespace is significant to the preprocessor is | |
2254 | a #define statment - if there is whitespace between the macro name | |
2255 | and an initial "(" token the macro is "object-like", otherwise it is | |
2256 | a function-like macro that takes arguments. | |
2257 | ||
2258 | However, all is not rosy. Parsing of identifiers, numbers, comments | |
2259 | and strings becomes trickier because of the possibility of raw | |
2260 | trigraphs and escaped newlines in the input stream. | |
2261 | ||
2262 | The trigraphs are three consecutive characters beginning with two | |
c2e25d51 NB |
2263 | question marks. A question mark is not valid as part of a number or |
2264 | identifier, so parsing of a number or identifier terminates normally | |
2265 | upon reaching it, returning to the mainloop which handles the | |
2266 | trigraph just like it would in any other position. Similarly for the | |
2267 | backslash of a backslash-newline combination. So we just need the | |
2268 | escaped-newline dropper in the mainloop to check if the token on the | |
2269 | top of the stack after dropping the escaped newline is a number or | |
2270 | identifier, and if so to continue the processing it as if nothing had | |
2271 | happened. | |
d6d5f795 NB |
2272 | |
2273 | For strings, we replace trigraphs whenever we reach a quote or | |
2274 | newline, because there might be a backslash trigraph escaping them. | |
2275 | We need to be careful that we start trigraph replacing from where we | |
2276 | left off previously, because it is possible for a first scan to leave | |
2277 | "fake" trigraphs that a second scan would pick up as real (e.g. the | |
c2e25d51 | 2278 | sequence "????/\n=" would find a fake ??= trigraph after removing the |
d6d5f795 NB |
2279 | escaped newline.) |
2280 | ||
2281 | For line comments, on reaching a newline we scan the previous | |
2282 | character(s) to see if it escaped, and continue if it is. Block | |
2283 | comments ignore everything and just focus on finding the comment | |
2284 | termination mark. The only difficult thing, and it is surprisingly | |
2285 | tricky, is checking if an asterisk precedes the final slash since | |
2286 | they could be separated by escaped newlines. If the preprocessor is | |
2287 | invoked with the output comments option, we don't bother removing | |
2288 | escaped newlines and replacing trigraphs for output. | |
2289 | ||
2290 | Finally, numbers can begin with a period, which is pushed initially | |
2291 | as a CPP_DOT token in its own right. The digit handler checks if the | |
2292 | previous token was a CPP_DOT not separated by whitespace, and if so | |
2293 | pops it off the stack and pushes a period into the number's buffer | |
2294 | before calling the number parser. | |
2295 | ||
2296 | */ | |
2297 | ||
b8f41010 NB |
2298 | static const unsigned char *digraph_spellings [] = {U"%:", U"%:%:", U"<:", |
2299 | U":>", U"<%", U"%>"}; | |
2300 | static unsigned char trigraph_map[256]; | |
c5a04734 | 2301 | |
c5a04734 ZW |
2302 | void |
2303 | init_trigraph_map () | |
2304 | { | |
2305 | trigraph_map['='] = '#'; | |
2306 | trigraph_map['('] = '['; | |
2307 | trigraph_map[')'] = ']'; | |
2308 | trigraph_map['/'] = '\\'; | |
2309 | trigraph_map['\''] = '^'; | |
2310 | trigraph_map['<'] = '{'; | |
2311 | trigraph_map['>'] = '}'; | |
2312 | trigraph_map['!'] = '|'; | |
2313 | trigraph_map['-'] = '~'; | |
2314 | } | |
2315 | ||
2316 | /* Call when a trigraph is encountered. It warns if necessary, and | |
2317 | returns true if the trigraph should be honoured. END is the third | |
2318 | character of a trigraph in the input stream. */ | |
2319 | static int | |
2320 | trigraph_ok (pfile, end) | |
2321 | cpp_reader *pfile; | |
2322 | const unsigned char *end; | |
2323 | { | |
2324 | int accept = CPP_OPTION (pfile, trigraphs); | |
2325 | ||
2326 | if (CPP_OPTION (pfile, warn_trigraphs)) | |
2327 | { | |
2328 | unsigned int col = end - 1 - pfile->buffer->line_base; | |
2329 | if (accept) | |
2330 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2331 | "trigraph ??%c converted to %c", | |
2332 | (int) *end, (int) trigraph_map[*end]); | |
2333 | else | |
2334 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2335 | "trigraph ??%c ignored", (int) *end); | |
2336 | } | |
2337 | return accept; | |
2338 | } | |
2339 | ||
2340 | /* Scan a string for trigraphs, warning or replacing them inline as | |
2341 | appropriate. When parsing a string, we must call this routine | |
2342 | before processing a newline character (if trigraphs are enabled), | |
2343 | since the newline might be escaped by a preceding backslash | |
2344 | trigraph sequence. Returns a pointer to the end of the name after | |
2345 | replacement. */ | |
2346 | ||
2347 | static unsigned char* | |
2348 | trigraph_replace (pfile, src, limit) | |
2349 | cpp_reader *pfile; | |
2350 | unsigned char *src; | |
2351 | unsigned char* limit; | |
2352 | { | |
2353 | unsigned char *dest; | |
2354 | ||
2355 | /* Starting with src[1], find two consecutive '?'. The case of no | |
2356 | trigraphs is streamlined. */ | |
2357 | ||
2358 | for (; src + 1 < limit; src += 2) | |
2359 | { | |
2360 | if (src[0] != '?') | |
2361 | continue; | |
2362 | ||
2363 | /* Make src point to the 1st (NOT 2nd) of two consecutive '?'s. */ | |
2364 | if (src[-1] == '?') | |
2365 | src--; | |
2366 | else if (src + 2 == limit || src[1] != '?') | |
2367 | continue; | |
2368 | ||
2369 | /* Check if it really is a trigraph. */ | |
2370 | if (trigraph_map[src[2]] == 0) | |
2371 | continue; | |
2372 | ||
2373 | dest = src; | |
2374 | goto trigraph_found; | |
2375 | } | |
2376 | return limit; | |
2377 | ||
2378 | /* Now we have a trigraph, we need to scan the remaining buffer, and | |
2379 | copy-shifting its contents left if replacement is enabled. */ | |
2380 | for (; src + 2 < limit; dest++, src++) | |
2381 | if ((*dest = *src) == '?' && src[1] == '?' && trigraph_map[src[2]]) | |
2382 | { | |
2383 | trigraph_found: | |
2384 | src += 2; | |
2385 | if (trigraph_ok (pfile, pfile->buffer->cur - (limit - src))) | |
2386 | *dest = trigraph_map[*src]; | |
2387 | } | |
2388 | ||
2389 | /* Copy remaining (at most 2) characters. */ | |
2390 | while (src < limit) | |
2391 | *dest++ = *src++; | |
2392 | return dest; | |
2393 | } | |
2394 | ||
2395 | /* If CUR is a backslash or the end of a trigraphed backslash, return | |
2396 | a pointer to its beginning, otherwise NULL. We don't read beyond | |
2397 | the buffer start, because there is the start of the comment in the | |
2398 | buffer. */ | |
2399 | static const unsigned char * | |
2400 | backslash_start (pfile, cur) | |
2401 | cpp_reader *pfile; | |
2402 | const unsigned char *cur; | |
2403 | { | |
2404 | if (cur[0] == '\\') | |
2405 | return cur; | |
2406 | if (cur[0] == '/' && cur[-1] == '?' && cur[-2] == '?' | |
2407 | && trigraph_ok (pfile, cur)) | |
2408 | return cur - 2; | |
2409 | return 0; | |
2410 | } | |
2411 | ||
2412 | /* Skip a C-style block comment. This is probably the trickiest | |
2413 | handler. We find the end of the comment by seeing if an asterisk | |
2414 | is before every '/' we encounter. The nasty complication is that a | |
2415 | previous asterisk may be separated by one or more escaped newlines. | |
2416 | Returns non-zero if comment terminated by EOF, zero otherwise. */ | |
2417 | static int | |
b8f41010 | 2418 | skip_block_comment2 (pfile) |
c5a04734 ZW |
2419 | cpp_reader *pfile; |
2420 | { | |
2421 | cpp_buffer *buffer = pfile->buffer; | |
2422 | const unsigned char *char_after_star = 0; | |
2423 | register const unsigned char *cur = buffer->cur; | |
2424 | int seen_eof = 0; | |
2425 | ||
2426 | /* Inner loop would think the comment has ended if the first comment | |
2427 | character is a '/'. Avoid this and keep the inner loop clean by | |
2428 | skipping such a character. */ | |
2429 | if (cur < buffer->rlimit && cur[0] == '/') | |
2430 | cur++; | |
2431 | ||
2432 | for (; cur < buffer->rlimit; ) | |
2433 | { | |
2434 | unsigned char c = *cur++; | |
2435 | ||
2436 | /* People like decorating comments with '*', so check for | |
2437 | '/' instead for efficiency. */ | |
2438 | if (c == '/') | |
2439 | { | |
2440 | if (cur[-2] == '*' || cur - 1 == char_after_star) | |
2441 | goto out; | |
2442 | ||
2443 | /* Warn about potential nested comments, but not when | |
2444 | the final character inside the comment is a '/'. | |
2445 | Don't bother to get it right across escaped newlines. */ | |
2446 | if (CPP_OPTION (pfile, warn_comments) && cur + 1 < buffer->rlimit | |
2447 | && cur[0] == '*' && cur[1] != '/') | |
2448 | { | |
2449 | buffer->cur = cur; | |
2450 | cpp_warning (pfile, "'/*' within comment"); | |
2451 | } | |
2452 | } | |
2453 | else if (IS_NEWLINE(c)) | |
2454 | { | |
2455 | const unsigned char* bslash = backslash_start (pfile, cur - 2); | |
2456 | ||
2457 | handle_newline (cur, buffer->rlimit, c); | |
2458 | /* Work correctly if there is an asterisk before an | |
2459 | arbirtrarily long sequence of escaped newlines. */ | |
2460 | if (bslash && (bslash[-1] == '*' || bslash == char_after_star)) | |
2461 | char_after_star = cur; | |
2462 | else | |
2463 | char_after_star = 0; | |
2464 | } | |
2465 | } | |
2466 | seen_eof = 1; | |
2467 | ||
2468 | out: | |
2469 | buffer->cur = cur; | |
2470 | return seen_eof; | |
2471 | } | |
2472 | ||
2473 | /* Skip a C++ or Chill line comment. Handles escaped newlines. | |
2474 | Returns non-zero if a multiline comment. */ | |
2475 | static int | |
b8f41010 | 2476 | skip_line_comment2 (pfile) |
c5a04734 ZW |
2477 | cpp_reader *pfile; |
2478 | { | |
2479 | cpp_buffer *buffer = pfile->buffer; | |
2480 | register const unsigned char *cur = buffer->cur; | |
2481 | int multiline = 0; | |
2482 | ||
2483 | for (; cur < buffer->rlimit; ) | |
2484 | { | |
2485 | unsigned char c = *cur++; | |
2486 | ||
2487 | if (IS_NEWLINE (c)) | |
2488 | { | |
2489 | /* Check for a (trigaph?) backslash escaping the newline. */ | |
2490 | if (!backslash_start (pfile, cur - 2)) | |
2491 | goto out; | |
2492 | multiline = 1; | |
2493 | handle_newline (cur, buffer->rlimit, c); | |
2494 | } | |
2495 | } | |
2496 | cur++; | |
2497 | ||
2498 | out: | |
2499 | buffer->cur = cur - 1; /* Leave newline for caller. */ | |
2500 | return multiline; | |
2501 | } | |
2502 | ||
6ab3e7dd NB |
2503 | /* Skips whitespace, stopping at next non-whitespace character. |
2504 | Adjusts pfile->col_adjust to account for tabs. This enables tokens | |
2505 | to be assigned the correct column. */ | |
c5a04734 ZW |
2506 | static void |
2507 | skip_whitespace (pfile, in_directive) | |
2508 | cpp_reader *pfile; | |
2509 | int in_directive; | |
2510 | { | |
2511 | cpp_buffer *buffer = pfile->buffer; | |
2512 | register const unsigned char *cur = buffer->cur; | |
2513 | unsigned short null_count = 0; | |
2514 | ||
2515 | for (; cur < buffer->rlimit; ) | |
2516 | { | |
2517 | unsigned char c = *cur++; | |
2518 | ||
6ab3e7dd NB |
2519 | if (c == '\t') |
2520 | { | |
2521 | unsigned int col = CPP_BUF_COLUMN (buffer, cur - 1); | |
2522 | pfile->col_adjust += (CPP_OPTION (pfile, tabstop) - 1 | |
2523 | - col % CPP_OPTION(pfile, tabstop)); | |
2524 | } | |
c5a04734 ZW |
2525 | if (IS_HSPACE(c)) /* FIXME: Fix ISTABLE. */ |
2526 | continue; | |
2527 | if (!is_space(c) || IS_NEWLINE (c)) /* Main loop handles newlines. */ | |
2528 | goto out; | |
2529 | if (c == '\0') | |
2530 | null_count++; | |
2531 | /* Mut be '\f' or '\v' */ | |
2532 | else if (in_directive && CPP_PEDANTIC (pfile)) | |
2533 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
2534 | c == '\f' ? "formfeed" : "vertical tab"); | |
2535 | } | |
2536 | cur++; | |
2537 | ||
2538 | out: | |
2539 | buffer->cur = cur - 1; | |
2540 | if (null_count) | |
2541 | cpp_warning (pfile, null_count > 1 ? "embedded null characters ignored" | |
2542 | : "embedded null character ignored"); | |
2543 | } | |
2544 | ||
2545 | /* Parse (append) an identifier. */ | |
2546 | static void | |
2547 | parse_name (pfile, list, name) | |
2548 | cpp_reader *pfile; | |
2549 | cpp_toklist *list; | |
2550 | cpp_name *name; | |
2551 | { | |
2552 | const unsigned char *name_limit; | |
2553 | unsigned char *namebuf; | |
2554 | cpp_buffer *buffer = pfile->buffer; | |
2555 | register const unsigned char *cur = buffer->cur; | |
2556 | ||
2557 | expanded: | |
2558 | name_limit = list->namebuf + list->name_cap; | |
2559 | namebuf = list->namebuf + list->name_used; | |
2560 | ||
2561 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2562 | { | |
2563 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2564 | ||
2565 | if (! is_idchar(c)) | |
2566 | goto out; | |
2567 | namebuf++; | |
2568 | cur++; | |
2569 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
2570 | { | |
2571 | buffer->cur = cur; | |
2572 | cpp_pedwarn (pfile, "'$' character in identifier"); | |
2573 | } | |
2574 | } | |
2575 | ||
2576 | /* Run out of name space? */ | |
2577 | if (cur < buffer->rlimit) | |
2578 | { | |
2579 | list->name_used = namebuf - list->namebuf; | |
2580 | auto_expand_name_space (list); | |
2581 | goto expanded; | |
2582 | } | |
2583 | ||
2584 | out: | |
2585 | buffer->cur = cur; | |
f617b8e2 | 2586 | name->len = namebuf - name->text; |
c5a04734 ZW |
2587 | list->name_used = namebuf - list->namebuf; |
2588 | } | |
2589 | ||
2590 | /* Parse (append) a number. */ | |
2591 | ||
2592 | #define VALID_SIGN(c, prevc) \ | |
2593 | (((c) == '+' || (c) == '-') && \ | |
2594 | ((prevc) == 'e' || (prevc) == 'E' \ | |
2595 | || (((prevc) == 'p' || (prevc) == 'P') && !CPP_OPTION (pfile, c89)))) | |
2596 | ||
2597 | static void | |
2598 | parse_number (pfile, list, name) | |
2599 | cpp_reader *pfile; | |
2600 | cpp_toklist *list; | |
2601 | cpp_name *name; | |
2602 | { | |
2603 | const unsigned char *name_limit; | |
2604 | unsigned char *namebuf; | |
2605 | cpp_buffer *buffer = pfile->buffer; | |
2606 | register const unsigned char *cur = buffer->cur; | |
2607 | ||
2608 | expanded: | |
2609 | name_limit = list->namebuf + list->name_cap; | |
2610 | namebuf = list->namebuf + list->name_used; | |
2611 | ||
2612 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2613 | { | |
2614 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2615 | ||
2616 | /* Perhaps we should accept '$' here if we accept it for | |
2617 | identifiers. We know namebuf[-1] is safe, because for c to | |
2618 | be a sign we must have pushed at least one character. */ | |
2619 | if (!is_numchar (c) && c != '.' && ! VALID_SIGN (c, namebuf[-1])) | |
2620 | goto out; | |
2621 | ||
2622 | namebuf++; | |
2623 | cur++; | |
2624 | } | |
2625 | ||
2626 | /* Run out of name space? */ | |
2627 | if (cur < buffer->rlimit) | |
2628 | { | |
2629 | list->name_used = namebuf - list->namebuf; | |
2630 | auto_expand_name_space (list); | |
2631 | goto expanded; | |
2632 | } | |
2633 | ||
2634 | out: | |
2635 | buffer->cur = cur; | |
f617b8e2 | 2636 | name->len = namebuf - name->text; |
c5a04734 ZW |
2637 | list->name_used = namebuf - list->namebuf; |
2638 | } | |
2639 | ||
2640 | /* Places a string terminated by an unescaped TERMINATOR into a | |
2641 | cpp_name, which should be expandable and thus at the top of the | |
2642 | list's stack. Handles embedded trigraphs, if necessary, and | |
2643 | escaped newlines. | |
2644 | ||
2645 | Can be used for character constants (terminator = '\''), string | |
41e8b1d7 NB |
2646 | constants ('"') and angled headers ('>'). Multi-line strings are |
2647 | allowed, except for within directives. */ | |
c5a04734 ZW |
2648 | |
2649 | static void | |
d1d9a6bd | 2650 | parse_string2 (pfile, list, name, terminator, multiline_ok) |
c5a04734 ZW |
2651 | cpp_reader *pfile; |
2652 | cpp_toklist *list; | |
2653 | cpp_name *name; | |
2654 | unsigned int terminator; | |
d1d9a6bd | 2655 | int multiline_ok; |
c5a04734 ZW |
2656 | { |
2657 | cpp_buffer *buffer = pfile->buffer; | |
2658 | register const unsigned char *cur = buffer->cur; | |
2659 | const unsigned char *name_limit; | |
2660 | unsigned char *namebuf; | |
2661 | unsigned int null_count = 0; | |
2662 | int trigraphed_len = 0; | |
2663 | ||
2664 | expanded: | |
2665 | name_limit = list->namebuf + list->name_cap; | |
2666 | namebuf = list->namebuf + list->name_used; | |
2667 | ||
2668 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2669 | { | |
2670 | unsigned int c = *namebuf++ = *cur++; /* Copy a single char. */ | |
2671 | ||
2672 | if (c == '\0') | |
2673 | null_count++; | |
2674 | else if (c == terminator || IS_NEWLINE (c)) | |
2675 | { | |
c5a04734 ZW |
2676 | /* Needed for trigraph_replace and multiline string warning. */ |
2677 | buffer->cur = cur; | |
2678 | ||
2679 | /* Scan for trigraphs before checking if backslash-escaped. */ | |
2680 | if (CPP_OPTION (pfile, trigraphs) | |
2681 | || CPP_OPTION (pfile, warn_trigraphs)) | |
2682 | { | |
f617b8e2 | 2683 | namebuf = trigraph_replace (pfile, name->text + trigraphed_len, |
c5a04734 | 2684 | namebuf); |
f617b8e2 | 2685 | trigraphed_len = namebuf - 2 - (name->text + trigraphed_len); |
c5a04734 ZW |
2686 | if (trigraphed_len < 0) |
2687 | trigraphed_len = 0; | |
2688 | } | |
2689 | ||
2690 | namebuf--; /* Drop the newline / terminator from the name. */ | |
2691 | if (IS_NEWLINE (c)) | |
2692 | { | |
2693 | /* Drop a backslash newline, and continue. */ | |
2694 | if (namebuf[-1] == '\\') | |
2695 | { | |
2696 | handle_newline (cur, buffer->rlimit, c); | |
2697 | namebuf--; | |
2698 | continue; | |
2699 | } | |
2700 | ||
2701 | cur--; | |
2702 | ||
2703 | /* In Fortran and assembly language, silently terminate | |
2704 | strings of either variety at end of line. This is a | |
2705 | kludge around not knowing where comments are in these | |
2706 | languages. */ | |
2707 | if (CPP_OPTION (pfile, lang_fortran) | |
2708 | || CPP_OPTION (pfile, lang_asm)) | |
2709 | goto out; | |
2710 | ||
2711 | /* Character constants, headers and asserts may not | |
2712 | extend over multiple lines. In Standard C, neither | |
2713 | may strings. We accept multiline strings as an | |
2714 | extension, but not in directives. */ | |
d1d9a6bd | 2715 | if (!multiline_ok) |
c5a04734 ZW |
2716 | goto unterminated; |
2717 | ||
2718 | cur++; /* Move forwards again. */ | |
2719 | ||
2720 | if (pfile->multiline_string_line == 0) | |
2721 | { | |
2722 | pfile->multiline_string_line = list->line; | |
2723 | if (CPP_PEDANTIC (pfile)) | |
2724 | cpp_pedwarn (pfile, "multi-line string constant"); | |
2725 | } | |
2726 | ||
2727 | *namebuf++ = '\n'; | |
2728 | handle_newline (cur, buffer->rlimit, c); | |
2729 | } | |
2730 | else | |
2731 | { | |
2732 | unsigned char *temp; | |
2733 | ||
2734 | /* An odd number of consecutive backslashes represents | |
2735 | an escaped terminator. */ | |
2736 | temp = namebuf - 1; | |
f617b8e2 | 2737 | while (temp >= name->text && *temp == '\\') |
c5a04734 ZW |
2738 | temp--; |
2739 | ||
2740 | if ((namebuf - temp) & 1) | |
2741 | goto out; | |
2742 | namebuf++; | |
2743 | } | |
2744 | } | |
2745 | } | |
2746 | ||
2747 | /* Run out of name space? */ | |
2748 | if (cur < buffer->rlimit) | |
2749 | { | |
2750 | list->name_used = namebuf - list->namebuf; | |
2751 | auto_expand_name_space (list); | |
2752 | goto expanded; | |
2753 | } | |
2754 | ||
2755 | /* We may not have trigraph-replaced the input for this code path, | |
2756 | but as the input is in error by being unterminated we don't | |
2757 | bother. Prevent warnings about no newlines at EOF. */ | |
2758 | if (IS_NEWLINE(cur[-1])) | |
2759 | cur--; | |
2760 | ||
2761 | unterminated: | |
2762 | cpp_error (pfile, "missing terminating %c character", (int) terminator); | |
2763 | ||
2764 | if (terminator == '\"' && pfile->multiline_string_line != list->line | |
2765 | && pfile->multiline_string_line != 0) | |
2766 | { | |
2767 | cpp_error_with_line (pfile, pfile->multiline_string_line, -1, | |
2768 | "possible start of unterminated string literal"); | |
2769 | pfile->multiline_string_line = 0; | |
2770 | } | |
2771 | ||
2772 | out: | |
2773 | buffer->cur = cur; | |
f617b8e2 | 2774 | name->len = namebuf - name->text; |
c5a04734 ZW |
2775 | list->name_used = namebuf - list->namebuf; |
2776 | ||
2777 | if (null_count > 0) | |
2778 | cpp_warning (pfile, (null_count > 1 ? "null characters preserved" | |
2779 | : "null character preserved")); | |
2780 | } | |
2781 | ||
5d7ee2fa NB |
2782 | /* The character TYPE helps us distinguish comment types: '*' = C |
2783 | style, '-' = Chill-style and '/' = C++ style. For code simplicity, | |
2784 | the stored comment includes the comment start and any terminator. */ | |
2785 | ||
2786 | #define COMMENT_START_LEN 2 | |
c5a04734 | 2787 | static void |
ad265aa4 | 2788 | save_comment (list, token, from, len, type) |
c5a04734 | 2789 | cpp_toklist *list; |
ad265aa4 | 2790 | cpp_token *token; |
c5a04734 ZW |
2791 | const unsigned char *from; |
2792 | unsigned int len; | |
c5a04734 ZW |
2793 | unsigned int type; |
2794 | { | |
5d7ee2fa NB |
2795 | unsigned char *buffer; |
2796 | ||
2797 | len += COMMENT_START_LEN; | |
c5a04734 | 2798 | |
c5a04734 ZW |
2799 | if (list->name_used + len > list->name_cap) |
2800 | expand_name_space (list, len); | |
2801 | ||
ad265aa4 NB |
2802 | INIT_TOKEN_NAME (list, token); |
2803 | token->type = CPP_COMMENT; | |
2804 | token->val.name.len = len; | |
c5a04734 | 2805 | |
d1d9a6bd NB |
2806 | buffer = list->namebuf + list->name_used; |
2807 | list->name_used += len; | |
2808 | ||
2809 | /* Copy the comment. */ | |
5d7ee2fa NB |
2810 | if (type == '*') |
2811 | { | |
2812 | *buffer++ = '/'; | |
2813 | *buffer++ = '*'; | |
2814 | } | |
2815 | else | |
2816 | { | |
2817 | *buffer++ = type; | |
2818 | *buffer++ = type; | |
2819 | } | |
5d7ee2fa | 2820 | memcpy (buffer, from, len - COMMENT_START_LEN); |
c5a04734 ZW |
2821 | } |
2822 | ||
2823 | /* | |
2824 | * The tokenizer's main loop. Returns a token list, representing a | |
f624ffa7 NB |
2825 | * logical line in the input file. On EOF after some tokens have |
2826 | * been processed, we return immediately. Then in next call, or if | |
2827 | * EOF occurred at the beginning of a logical line, a single CPP_EOF | |
2828 | * token is placed in the list. | |
c5a04734 ZW |
2829 | * |
2830 | * Implementation relies almost entirely on lookback, rather than | |
2831 | * looking forwards. This means that tokenization requires just | |
2832 | * a single pass of the file, even in the presence of trigraphs and | |
2833 | * escaped newlines, providing significant performance benefits. | |
2834 | * Trigraph overhead is negligible if they are disabled, and low | |
2835 | * even when enabled. | |
2836 | */ | |
2837 | ||
d1d9a6bd NB |
2838 | #define IS_DIRECTIVE() (list->tokens[first_token].type == CPP_HASH) |
2839 | ||
c5a04734 ZW |
2840 | void |
2841 | _cpp_lex_line (pfile, list) | |
2842 | cpp_reader *pfile; | |
2843 | cpp_toklist *list; | |
2844 | { | |
2845 | cpp_token *cur_token, *token_limit; | |
2846 | cpp_buffer *buffer = pfile->buffer; | |
2847 | register const unsigned char *cur = buffer->cur; | |
2848 | unsigned char flags = 0; | |
d1d9a6bd | 2849 | unsigned int first_token = list->tokens_used; |
c5a04734 | 2850 | |
f624ffa7 | 2851 | list->line = CPP_BUF_LINE (buffer); |
6ab3e7dd | 2852 | pfile->col_adjust = 0; |
c5a04734 ZW |
2853 | expanded: |
2854 | token_limit = list->tokens + list->tokens_cap; | |
2855 | cur_token = list->tokens + list->tokens_used; | |
2856 | ||
2857 | for (; cur < buffer->rlimit && cur_token < token_limit;) | |
2858 | { | |
2859 | unsigned char c = *cur++; | |
2860 | ||
6ab3e7dd NB |
2861 | /* Optimize whitespace skipping, as most tokens are probably |
2862 | separated by whitespace. (' ' '\t' '\v' '\f' '\0'). */ | |
2863 | ||
c5a04734 ZW |
2864 | if (is_hspace ((unsigned int) c)) |
2865 | { | |
6ab3e7dd NB |
2866 | /* Step back to get the null warning and tab correction. */ |
2867 | buffer->cur = cur - 1; | |
d1d9a6bd | 2868 | skip_whitespace (pfile, IS_DIRECTIVE ()); |
6ab3e7dd NB |
2869 | cur = buffer->cur; |
2870 | ||
c5a04734 ZW |
2871 | flags = PREV_WHITESPACE; |
2872 | if (cur == buffer->rlimit) | |
2873 | break; | |
2874 | c = *cur++; | |
2875 | } | |
2876 | ||
2877 | /* Initialize current token. Its type is set in the switch. */ | |
6ab3e7dd | 2878 | cur_token->col = CPP_BUF_COLUMN (buffer, cur); |
c5a04734 ZW |
2879 | cur_token->flags = flags; |
2880 | flags = 0; | |
2881 | ||
2882 | switch (c) | |
2883 | { | |
2884 | case '0': case '1': case '2': case '3': case '4': | |
2885 | case '5': case '6': case '7': case '8': case '9': | |
d1d9a6bd NB |
2886 | { |
2887 | int prev_dot; | |
c5a04734 | 2888 | |
d1d9a6bd NB |
2889 | cur--; /* Backup character. */ |
2890 | prev_dot = PREV_TOKEN_TYPE == CPP_DOT && IMMED_TOKEN (); | |
2891 | if (prev_dot) | |
2892 | cur_token--; | |
2893 | INIT_TOKEN_NAME (list, cur_token); | |
2894 | /* Prepend an immediately previous CPP_DOT token. */ | |
2895 | if (prev_dot) | |
2896 | { | |
2897 | if (list->name_cap == list->name_used) | |
2898 | auto_expand_name_space (list); | |
c5a04734 | 2899 | |
d1d9a6bd NB |
2900 | cur_token->val.name.len = 1; |
2901 | list->namebuf[list->name_used++] = '.'; | |
2902 | } | |
c5a04734 | 2903 | |
d1d9a6bd NB |
2904 | continue_number: |
2905 | cur_token->type = CPP_NUMBER; /* Before parse_number. */ | |
2906 | buffer->cur = cur; | |
2907 | parse_number (pfile, list, &cur_token->val.name); | |
2908 | cur = buffer->cur; | |
2909 | cur_token++; | |
2910 | } | |
c5a04734 ZW |
2911 | break; |
2912 | ||
2913 | letter: | |
2914 | case '_': | |
2915 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
2916 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
2917 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
2918 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
2919 | case 'y': case 'z': | |
2920 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
2921 | case 'G': case 'H': case 'I': case 'J': case 'K': case 'L': | |
2922 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': | |
2923 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
2924 | case 'Y': case 'Z': | |
c5a04734 | 2925 | cur--; /* Backup character. */ |
d1d9a6bd | 2926 | INIT_TOKEN_NAME (list, cur_token); |
c5a04734 ZW |
2927 | cur_token->type = CPP_NAME; /* Identifier, macro etc. */ |
2928 | ||
2929 | continue_name: | |
2930 | buffer->cur = cur; | |
2931 | parse_name (pfile, list, &cur_token->val.name); | |
2932 | cur = buffer->cur; | |
2933 | ||
2934 | /* Find handler for newly created / extended directive. */ | |
d1d9a6bd | 2935 | if (IS_DIRECTIVE () && cur_token == &list->tokens[first_token + 1]) |
c5a04734 ZW |
2936 | _cpp_check_directive (list, cur_token); |
2937 | cur_token++; | |
2938 | break; | |
2939 | ||
2940 | case '\'': | |
2941 | /* Fall through. */ | |
2942 | case '\"': | |
2943 | cur_token->type = c == '\'' ? CPP_CHAR : CPP_STRING; | |
2944 | /* Do we have a wide string? */ | |
2945 | if (cur_token[-1].type == CPP_NAME && IMMED_TOKEN () | |
2946 | && cur_token[-1].val.name.len == 1 | |
f617b8e2 | 2947 | && cur_token[-1].val.name.text[0] == 'L' |
c5a04734 ZW |
2948 | && !CPP_TRADITIONAL (pfile)) |
2949 | { | |
2950 | /* No need for 'L' any more. */ | |
2951 | list->name_used--; | |
2952 | (--cur_token)->type = (c == '\'' ? CPP_WCHAR : CPP_WSTRING); | |
2953 | } | |
2954 | ||
2955 | do_parse_string: | |
6ab3e7dd | 2956 | /* Here c is one of ' " or >. */ |
d1d9a6bd | 2957 | INIT_TOKEN_NAME (list, cur_token); |
c5a04734 | 2958 | buffer->cur = cur; |
d1d9a6bd NB |
2959 | parse_string2 (pfile, list, &cur_token->val.name, c, |
2960 | c == '"' && !IS_DIRECTIVE()); | |
c5a04734 ZW |
2961 | cur = buffer->cur; |
2962 | cur_token++; | |
2963 | break; | |
2964 | ||
2965 | case '/': | |
2966 | cur_token->type = CPP_DIV; | |
2967 | if (IMMED_TOKEN ()) | |
2968 | { | |
2969 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
2970 | { | |
2971 | /* We silently allow C++ comments in system headers, | |
2972 | irrespective of conformance mode, because lots of | |
2973 | broken systems do that and trying to clean it up | |
2974 | in fixincludes is a nightmare. */ | |
c31a6508 | 2975 | if (CPP_IN_SYSTEM_HEADER (pfile)) |
c5a04734 ZW |
2976 | goto do_line_comment; |
2977 | else if (CPP_OPTION (pfile, cplusplus_comments)) | |
2978 | { | |
2979 | if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile) | |
2980 | && ! buffer->warned_cplusplus_comments) | |
2981 | { | |
2982 | buffer->cur = cur; | |
2983 | cpp_pedwarn (pfile, | |
2984 | "C++ style comments are not allowed in ISO C89"); | |
2985 | cpp_pedwarn (pfile, | |
2986 | "(this will be reported only once per input file)"); | |
2987 | buffer->warned_cplusplus_comments = 1; | |
2988 | } | |
2989 | do_line_comment: | |
2990 | buffer->cur = cur; | |
2991 | if (cur[-2] != c) | |
2992 | cpp_warning (pfile, | |
2993 | "comment start split across lines"); | |
b8f41010 | 2994 | if (skip_line_comment2 (pfile)) |
c5a04734 ZW |
2995 | cpp_error_with_line (pfile, list->line, |
2996 | cur_token[-1].col, | |
2997 | "multi-line comment"); | |
c5a04734 ZW |
2998 | |
2999 | /* Back-up to first '-' or '/'. */ | |
ad265aa4 NB |
3000 | cur_token--; |
3001 | if (!CPP_OPTION (pfile, discard_comments) | |
3002 | && (!IS_DIRECTIVE() || list->dirno == 0)) | |
3003 | save_comment (list, cur_token++, cur, | |
3004 | buffer->cur - cur, c); | |
3005 | cur = buffer->cur; | |
3006 | ||
c5a04734 ZW |
3007 | if (!CPP_OPTION (pfile, traditional)) |
3008 | flags = PREV_WHITESPACE; | |
ad265aa4 | 3009 | break; |
c5a04734 ZW |
3010 | } |
3011 | } | |
3012 | } | |
3013 | cur_token++; | |
3014 | break; | |
3015 | ||
3016 | case '*': | |
3017 | cur_token->type = CPP_MULT; | |
3018 | if (IMMED_TOKEN ()) | |
3019 | { | |
3020 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
3021 | { | |
3022 | buffer->cur = cur; | |
3023 | if (cur[-2] != '/') | |
3024 | cpp_warning (pfile, | |
3025 | "comment start '/*' split across lines"); | |
b8f41010 | 3026 | if (skip_block_comment2 (pfile)) |
c5a04734 ZW |
3027 | cpp_error_with_line (pfile, list->line, cur_token[-1].col, |
3028 | "unterminated comment"); | |
3029 | else if (buffer->cur[-2] != '*') | |
3030 | cpp_warning (pfile, | |
3031 | "comment end '*/' split across lines"); | |
c5a04734 | 3032 | |
ad265aa4 | 3033 | /* Back up to opening '/'. */ |
f617b8e2 | 3034 | cur_token--; |
ad265aa4 NB |
3035 | if (!CPP_OPTION (pfile, discard_comments) |
3036 | && (!IS_DIRECTIVE() || list->dirno == 0)) | |
3037 | save_comment (list, cur_token++, cur, | |
3038 | buffer->cur - cur, c); | |
3039 | cur = buffer->cur; | |
3040 | ||
c5a04734 ZW |
3041 | if (!CPP_OPTION (pfile, traditional)) |
3042 | flags = PREV_WHITESPACE; | |
f617b8e2 | 3043 | break; |
c5a04734 ZW |
3044 | } |
3045 | else if (CPP_OPTION (pfile, cplusplus)) | |
3046 | { | |
3047 | /* In C++, there are .* and ->* operators. */ | |
3048 | if (PREV_TOKEN_TYPE == CPP_DEREF) | |
3049 | BACKUP_TOKEN (CPP_DEREF_STAR); | |
3050 | else if (PREV_TOKEN_TYPE == CPP_DOT) | |
3051 | BACKUP_TOKEN (CPP_DOT_STAR); | |
3052 | } | |
3053 | } | |
3054 | cur_token++; | |
3055 | break; | |
3056 | ||
3057 | case '\n': | |
3058 | case '\r': | |
3059 | handle_newline (cur, buffer->rlimit, c); | |
fb4527c3 | 3060 | if (PREV_TOKEN_TYPE == CPP_BACKSLASH && IMMED_TOKEN ()) |
c5a04734 | 3061 | { |
fb4527c3 NB |
3062 | /* Remove the escaped newline. Then continue to process |
3063 | any interrupted name or number. */ | |
3064 | cur_token--; | |
3065 | if (IMMED_TOKEN ()) | |
c5a04734 | 3066 | { |
fb4527c3 NB |
3067 | cur_token--; |
3068 | if (cur_token->type == CPP_NAME) | |
3069 | goto continue_name; | |
3070 | else if (cur_token->type == CPP_NUMBER) | |
3071 | goto continue_number; | |
3072 | cur_token++; | |
c5a04734 | 3073 | } |
fb4527c3 NB |
3074 | /* Remember whitespace setting. */ |
3075 | flags = cur_token->flags; | |
3076 | break; | |
c5a04734 | 3077 | } |
fb4527c3 | 3078 | if (PREV_TOKEN_TYPE == CPP_BACKSLASH) |
c5a04734 | 3079 | { |
fb4527c3 NB |
3080 | buffer->cur = cur; |
3081 | cpp_warning (pfile, "backslash and newline separated by space"); | |
c5a04734 | 3082 | } |
f624ffa7 NB |
3083 | /* Skip vertical space until we have at least one token to |
3084 | return. */ | |
3085 | if (cur_token != &list->tokens[first_token]) | |
3086 | goto out; | |
3087 | list->line = CPP_BUF_LINE (buffer); | |
3088 | break; | |
c5a04734 ZW |
3089 | |
3090 | case '-': | |
3091 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_MINUS) | |
3092 | { | |
3093 | if (CPP_OPTION (pfile, chill)) | |
3094 | goto do_line_comment; | |
3095 | REVISE_TOKEN (CPP_MINUS_MINUS); | |
3096 | } | |
3097 | else | |
3098 | PUSH_TOKEN (CPP_MINUS); | |
3099 | break; | |
3100 | ||
3101 | /* The digraph flag checking ensures that ## and %:%: | |
3102 | are interpreted as CPP_PASTE, but #%: and %:# are not. */ | |
3103 | make_hash: | |
3104 | case '#': | |
3105 | if (PREV_TOKEN_TYPE == CPP_HASH && IMMED_TOKEN () | |
3106 | && ((cur_token->flags ^ cur_token[-1].flags) & DIGRAPH) == 0) | |
3107 | REVISE_TOKEN (CPP_PASTE); | |
3108 | else | |
3109 | PUSH_TOKEN (CPP_HASH); | |
3110 | break; | |
3111 | ||
3112 | case ':': | |
3113 | cur_token->type = CPP_COLON; | |
3114 | if (IMMED_TOKEN ()) | |
3115 | { | |
3116 | if (PREV_TOKEN_TYPE == CPP_COLON | |
3117 | && CPP_OPTION (pfile, cplusplus)) | |
3118 | BACKUP_TOKEN (CPP_SCOPE); | |
3119 | /* Digraph: "<:" is a '[' */ | |
3120 | else if (PREV_TOKEN_TYPE == CPP_LESS) | |
3121 | BACKUP_DIGRAPH (CPP_OPEN_SQUARE); | |
3122 | /* Digraph: "%:" is a '#' */ | |
3123 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3124 | { | |
3125 | (--cur_token)->flags |= DIGRAPH; | |
3126 | goto make_hash; | |
3127 | } | |
3128 | } | |
3129 | cur_token++; | |
3130 | break; | |
3131 | ||
3132 | case '&': | |
3133 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_AND) | |
3134 | REVISE_TOKEN (CPP_AND_AND); | |
3135 | else | |
3136 | PUSH_TOKEN (CPP_AND); | |
3137 | break; | |
3138 | ||
3139 | make_or: | |
3140 | case '|': | |
3141 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_OR) | |
3142 | REVISE_TOKEN (CPP_OR_OR); | |
3143 | else | |
3144 | PUSH_TOKEN (CPP_OR); | |
3145 | break; | |
3146 | ||
3147 | case '+': | |
3148 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_PLUS) | |
3149 | REVISE_TOKEN (CPP_PLUS_PLUS); | |
3150 | else | |
3151 | PUSH_TOKEN (CPP_PLUS); | |
3152 | break; | |
3153 | ||
3154 | case '=': | |
3155 | /* This relies on equidistance of "?=" and "?" tokens. */ | |
3156 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE <= CPP_LAST_EQ) | |
3157 | REVISE_TOKEN (PREV_TOKEN_TYPE + (CPP_EQ_EQ - CPP_EQ)); | |
3158 | else | |
3159 | PUSH_TOKEN (CPP_EQ); | |
3160 | break; | |
3161 | ||
3162 | case '>': | |
3163 | cur_token->type = CPP_GREATER; | |
3164 | if (IMMED_TOKEN ()) | |
3165 | { | |
3166 | if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3167 | BACKUP_TOKEN (CPP_RSHIFT); | |
3168 | else if (PREV_TOKEN_TYPE == CPP_MINUS) | |
3169 | BACKUP_TOKEN (CPP_DEREF); | |
3170 | /* Digraph: ":>" is a ']' */ | |
3171 | else if (PREV_TOKEN_TYPE == CPP_COLON) | |
3172 | BACKUP_DIGRAPH (CPP_CLOSE_SQUARE); | |
3173 | /* Digraph: "%>" is a '}' */ | |
3174 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3175 | BACKUP_DIGRAPH (CPP_CLOSE_BRACE); | |
3176 | } | |
3177 | cur_token++; | |
3178 | break; | |
3179 | ||
3180 | case '<': | |
3181 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3182 | { | |
3183 | REVISE_TOKEN (CPP_LSHIFT); | |
3184 | break; | |
3185 | } | |
3186 | /* Is this the beginning of a header name? */ | |
15dad1d9 | 3187 | if (list->flags & SYNTAX_INCLUDE) |
c5a04734 ZW |
3188 | { |
3189 | c = '>'; /* Terminator. */ | |
3190 | cur_token->type = CPP_HEADER_NAME; | |
3191 | goto do_parse_string; | |
3192 | } | |
3193 | PUSH_TOKEN (CPP_LESS); | |
3194 | break; | |
3195 | ||
3196 | case '%': | |
3197 | /* Digraph: "<%" is a '{' */ | |
3198 | cur_token->type = CPP_MOD; | |
3199 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3200 | BACKUP_DIGRAPH (CPP_OPEN_BRACE); | |
3201 | cur_token++; | |
3202 | break; | |
3203 | ||
c5a04734 ZW |
3204 | case '?': |
3205 | if (cur + 1 < buffer->rlimit && *cur == '?' | |
3206 | && trigraph_map[cur[1]] && trigraph_ok (pfile, cur + 1)) | |
3207 | { | |
3208 | /* Handle trigraph. */ | |
3209 | cur++; | |
3210 | switch (*cur++) | |
3211 | { | |
3212 | case '(': goto make_open_square; | |
3213 | case ')': goto make_close_square; | |
3214 | case '<': goto make_open_brace; | |
3215 | case '>': goto make_close_brace; | |
3216 | case '=': goto make_hash; | |
3217 | case '!': goto make_or; | |
3218 | case '-': goto make_complement; | |
3219 | case '/': goto make_backslash; | |
3220 | case '\'': goto make_xor; | |
3221 | } | |
3222 | } | |
3223 | if (IMMED_TOKEN () && CPP_OPTION (pfile, cplusplus)) | |
3224 | { | |
3225 | /* GNU C++ defines <? and >? operators. */ | |
3226 | if (PREV_TOKEN_TYPE == CPP_LESS) | |
3227 | { | |
3228 | REVISE_TOKEN (CPP_MIN); | |
3229 | break; | |
3230 | } | |
3231 | else if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3232 | { | |
3233 | REVISE_TOKEN (CPP_MAX); | |
3234 | break; | |
3235 | } | |
3236 | } | |
3237 | PUSH_TOKEN (CPP_QUERY); | |
3238 | break; | |
3239 | ||
3240 | case '.': | |
3241 | if (PREV_TOKEN_TYPE == CPP_DOT && cur_token[-2].type == CPP_DOT | |
3242 | && IMMED_TOKEN () | |
3243 | && !(cur_token[-1].flags & PREV_WHITESPACE)) | |
3244 | { | |
3245 | cur_token -= 2; | |
3246 | PUSH_TOKEN (CPP_ELLIPSIS); | |
3247 | } | |
3248 | else | |
3249 | PUSH_TOKEN (CPP_DOT); | |
3250 | break; | |
3251 | ||
cfd5b8b8 NB |
3252 | make_complement: |
3253 | case '~': PUSH_TOKEN (CPP_COMPL); break; | |
c5a04734 ZW |
3254 | make_xor: |
3255 | case '^': PUSH_TOKEN (CPP_XOR); break; | |
3256 | make_open_brace: | |
3257 | case '{': PUSH_TOKEN (CPP_OPEN_BRACE); break; | |
3258 | make_close_brace: | |
3259 | case '}': PUSH_TOKEN (CPP_CLOSE_BRACE); break; | |
3260 | make_open_square: | |
3261 | case '[': PUSH_TOKEN (CPP_OPEN_SQUARE); break; | |
3262 | make_close_square: | |
3263 | case ']': PUSH_TOKEN (CPP_CLOSE_SQUARE); break; | |
3264 | make_backslash: | |
3265 | case '\\': PUSH_TOKEN (CPP_BACKSLASH); break; | |
3266 | case '!': PUSH_TOKEN (CPP_NOT); break; | |
3267 | case ',': PUSH_TOKEN (CPP_COMMA); break; | |
3268 | case ';': PUSH_TOKEN (CPP_SEMICOLON); break; | |
41e8b1d7 | 3269 | case '(': PUSH_TOKEN (CPP_OPEN_PAREN); break; |
cfd5b8b8 | 3270 | case ')': PUSH_TOKEN (CPP_CLOSE_PAREN); break; |
c5a04734 ZW |
3271 | |
3272 | case '$': | |
3273 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
3274 | goto letter; | |
3275 | /* Fall through */ | |
3276 | default: | |
3277 | cur_token->aux = c; | |
f617b8e2 | 3278 | cur_token->val.name.len = 0; /* FIXME: needed for transition only */ |
c5a04734 ZW |
3279 | PUSH_TOKEN (CPP_OTHER); |
3280 | break; | |
3281 | } | |
3282 | } | |
3283 | ||
3284 | /* Run out of token space? */ | |
3285 | if (cur_token == token_limit) | |
3286 | { | |
3287 | list->tokens_used = cur_token - list->tokens; | |
d1d9a6bd | 3288 | _cpp_expand_token_space (list, 256); |
c5a04734 ZW |
3289 | goto expanded; |
3290 | } | |
3291 | ||
c5a04734 | 3292 | cur_token->flags = flags; |
f624ffa7 | 3293 | if (cur_token == &list->tokens[first_token]) |
c5a04734 | 3294 | { |
f624ffa7 NB |
3295 | /* FIXME: move this warning to callers who care. */ |
3296 | if (cur > buffer->buf && !IS_NEWLINE (cur[-1])) | |
3297 | cpp_warning (pfile, "no newline at end of file"); | |
3298 | cur_token++->type = CPP_EOF; | |
c5a04734 ZW |
3299 | } |
3300 | ||
3301 | out: | |
f624ffa7 | 3302 | list->tokens[first_token].flags |= BOL; |
c5a04734 | 3303 | buffer->cur = cur; |
c5a04734 | 3304 | list->tokens_used = cur_token - list->tokens; |
c5a04734 ZW |
3305 | } |
3306 | ||
3fef5b2b NB |
3307 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
3308 | already contain the enough space to hold the token's spelling. If | |
3309 | WHITESPACE is true, and the token was preceded by whitespace, | |
3310 | output a single space before the token proper. Returns a pointer | |
3311 | to the character after the last character written. */ | |
3312 | ||
3313 | static unsigned char * | |
f617b8e2 | 3314 | spell_token (pfile, token, buffer, whitespace) |
3fef5b2b | 3315 | cpp_reader *pfile; /* Would be nice to be rid of this... */ |
d1d9a6bd | 3316 | const cpp_token *token; |
3fef5b2b NB |
3317 | unsigned char *buffer; |
3318 | int whitespace; | |
3319 | { | |
3320 | /* Whitespace will not be wanted by handlers of the # and ## | |
3321 | operators calling this function, but will be wanted by the | |
3322 | function that writes out the preprocessed file. */ | |
3323 | if (whitespace && token->flags & PREV_WHITESPACE) | |
3324 | *buffer++ = ' '; | |
3325 | ||
3326 | switch (token_spellings[token->type].type) | |
3327 | { | |
5d7ee2fa | 3328 | case SPELL_OPERATOR: |
3fef5b2b NB |
3329 | { |
3330 | const unsigned char *spelling; | |
3331 | unsigned char c; | |
3332 | ||
3333 | if (token->flags & DIGRAPH) | |
3334 | spelling = digraph_spellings[token->type - CPP_FIRST_DIGRAPH]; | |
3335 | else | |
f617b8e2 | 3336 | spelling = token_spellings[token->type].spelling; |
3fef5b2b NB |
3337 | |
3338 | while ((c = *spelling++) != '\0') | |
3339 | *buffer++ = c; | |
3340 | } | |
3341 | break; | |
3342 | ||
5d7ee2fa | 3343 | case SPELL_IDENT: |
f617b8e2 | 3344 | memcpy (buffer, token->val.name.text, token->val.name.len); |
5d7ee2fa NB |
3345 | buffer += token->val.name.len; |
3346 | break; | |
3347 | ||
3348 | case SPELL_STRING: | |
3fef5b2b | 3349 | { |
5d7ee2fa | 3350 | unsigned char c; |
3fef5b2b | 3351 | |
5d7ee2fa NB |
3352 | if (token->type == CPP_WSTRING || token->type == CPP_WCHAR) |
3353 | *buffer++ = 'L'; | |
3354 | c = '\''; | |
3355 | if (token->type == CPP_STRING || token->type == CPP_WSTRING) | |
3356 | c = '"'; | |
3357 | *buffer++ = c; | |
f617b8e2 | 3358 | memcpy (buffer, token->val.name.text, token->val.name.len); |
5d7ee2fa NB |
3359 | buffer += token->val.name.len; |
3360 | *buffer++ = c; | |
3fef5b2b NB |
3361 | } |
3362 | break; | |
3363 | ||
3364 | case SPELL_CHAR: | |
3365 | *buffer++ = token->aux; | |
3366 | break; | |
3367 | ||
3368 | case SPELL_NONE: | |
3369 | cpp_ice (pfile, "Unspellable token"); | |
3370 | break; | |
3371 | } | |
3372 | ||
3373 | return buffer; | |
3374 | } | |
3375 | ||
3376 | /* Temporary function for illustrative purposes. */ | |
c5a04734 ZW |
3377 | void |
3378 | _cpp_lex_file (pfile) | |
3379 | cpp_reader* pfile; | |
3380 | { | |
c5a04734 ZW |
3381 | cpp_toklist* list; |
3382 | ||
3383 | init_trigraph_map (); | |
3384 | list = (cpp_toklist *) xmalloc (sizeof (cpp_toklist)); | |
d1d9a6bd | 3385 | _cpp_init_toklist (list, DUMMY_TOKEN); |
c5a04734 | 3386 | |
15dad1d9 | 3387 | for (;;) |
c5a04734 | 3388 | { |
c5a04734 ZW |
3389 | _cpp_lex_line (pfile, list); |
3390 | if (list->tokens[0].type == CPP_EOF) | |
3391 | break; | |
3392 | ||
15dad1d9 ZW |
3393 | #if 0 |
3394 | if (list->dirno) | |
3395 | _cpp_handle_directive (pfile, list); | |
c5a04734 | 3396 | else |
15dad1d9 | 3397 | #endif |
c5a04734 | 3398 | _cpp_output_list (pfile, list); |
15dad1d9 | 3399 | _cpp_clear_toklist (list); |
c5a04734 ZW |
3400 | } |
3401 | } | |
3402 | ||
b8f41010 | 3403 | /* Temporary function for illustrative purposes. */ |
c5a04734 ZW |
3404 | static void |
3405 | _cpp_output_list (pfile, list) | |
3406 | cpp_reader *pfile; | |
3407 | cpp_toklist *list; | |
3408 | { | |
f624ffa7 | 3409 | unsigned int i; |
c5a04734 | 3410 | |
f624ffa7 | 3411 | for (i = 0; i < list->tokens_used; i++) |
c5a04734 | 3412 | { |
f624ffa7 NB |
3413 | CPP_RESERVE (pfile, TOKEN_LEN (&list->tokens[i])); |
3414 | pfile->limit = spell_token (pfile, &list->tokens[i], pfile->limit, 1); | |
c5a04734 ZW |
3415 | } |
3416 | } | |
3417 | ||
3418 | #endif |