]>
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 | ||
ff2b53ef ZW |
29 | #define PEEKBUF(BUFFER, N) \ |
30 | ((BUFFER)->rlimit - (BUFFER)->cur > (N) ? (BUFFER)->cur[N] : EOF) | |
31 | #define GETBUF(BUFFER) \ | |
32 | ((BUFFER)->cur < (BUFFER)->rlimit ? *(BUFFER)->cur++ : EOF) | |
33 | #define FORWARDBUF(BUFFER, N) ((BUFFER)->cur += (N)) | |
34 | ||
35 | #define PEEKN(N) PEEKBUF (CPP_BUFFER (pfile), N) | |
36 | #define FORWARD(N) FORWARDBUF (CPP_BUFFER (pfile), (N)) | |
37 | #define GETC() GETBUF (CPP_BUFFER (pfile)) | |
38 | #define PEEKC() PEEKBUF (CPP_BUFFER (pfile), 0) | |
45b966db ZW |
39 | |
40 | static void skip_block_comment PARAMS ((cpp_reader *)); | |
41 | static void skip_line_comment PARAMS ((cpp_reader *)); | |
42 | static int maybe_macroexpand PARAMS ((cpp_reader *, long)); | |
43 | static int skip_comment PARAMS ((cpp_reader *, int)); | |
44 | static int copy_comment PARAMS ((cpp_reader *, int)); | |
45 | static void skip_string PARAMS ((cpp_reader *, int)); | |
46 | static void parse_string PARAMS ((cpp_reader *, int)); | |
47 | static U_CHAR *find_position PARAMS ((U_CHAR *, U_CHAR *, unsigned long *)); | |
64aaf407 | 48 | static void null_warning PARAMS ((cpp_reader *, unsigned int)); |
45b966db | 49 | |
f2d5f0cc ZW |
50 | static void safe_fwrite PARAMS ((cpp_reader *, const U_CHAR *, |
51 | size_t, FILE *)); | |
1368ee70 ZW |
52 | static void output_line_command PARAMS ((cpp_reader *, cpp_printer *, |
53 | unsigned int)); | |
54 | static void bump_column PARAMS ((cpp_printer *, unsigned int, | |
55 | unsigned int)); | |
c5a04734 | 56 | static void expand_name_space PARAMS ((cpp_toklist *, unsigned int)); |
1368ee70 | 57 | static void expand_token_space PARAMS ((cpp_toklist *)); |
9e62c811 ZW |
58 | static void pedantic_whitespace PARAMS ((cpp_reader *, U_CHAR *, |
59 | unsigned int)); | |
f2d5f0cc | 60 | |
c5a04734 | 61 | #define auto_expand_name_space(list) \ |
f617b8e2 | 62 | expand_name_space ((list), 1 + (list)->name_cap / 2) |
c5a04734 | 63 | |
b8f41010 NB |
64 | #ifdef NEW_LEXER |
65 | ||
66 | static void expand_comment_space PARAMS ((cpp_toklist *)); | |
67 | void init_trigraph_map PARAMS ((void)); | |
68 | static unsigned char* trigraph_replace PARAMS ((cpp_reader *, unsigned char *, | |
69 | unsigned char *)); | |
70 | static const unsigned char *backslash_start PARAMS ((cpp_reader *, | |
71 | const unsigned char *)); | |
72 | static int skip_block_comment2 PARAMS ((cpp_reader *)); | |
73 | static int skip_line_comment2 PARAMS ((cpp_reader *)); | |
74 | static void skip_whitespace PARAMS ((cpp_reader *, int)); | |
75 | static void parse_name PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
76 | static void parse_number PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
77 | static void parse_string2 PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *, | |
78 | unsigned int)); | |
79 | static int trigraph_ok PARAMS ((cpp_reader *, const unsigned char *)); | |
80 | static void save_comment PARAMS ((cpp_toklist *, const unsigned char *, | |
81 | unsigned int, unsigned int, unsigned int)); | |
82 | void _cpp_lex_line PARAMS ((cpp_reader *, cpp_toklist *)); | |
83 | ||
84 | static void _cpp_output_list PARAMS ((cpp_reader *, cpp_toklist *)); | |
85 | ||
3fef5b2b | 86 | static unsigned char * spell_token PARAMS ((cpp_reader *, cpp_token *, |
f617b8e2 | 87 | unsigned char *, int)); |
b8f41010 NB |
88 | |
89 | typedef unsigned int (* speller) PARAMS ((unsigned char *, cpp_toklist *, | |
90 | cpp_token *)); | |
91 | ||
92 | /* Macros on a cpp_name. */ | |
93 | #define INIT_NAME(list, name) \ | |
f617b8e2 NB |
94 | do {(name).len = 0; \ |
95 | (name).text = (list)->namebuf + (list)->name_used;} while (0) | |
b8f41010 NB |
96 | |
97 | #define IS_DIRECTIVE(list) (TOK_TYPE (list, 0) == CPP_HASH) | |
b8f41010 NB |
98 | |
99 | /* Maybe put these in the ISTABLE eventually. */ | |
100 | #define IS_HSPACE(c) ((c) == ' ' || (c) == '\t') | |
101 | #define IS_NEWLINE(c) ((c) == '\n' || (c) == '\r') | |
102 | ||
103 | /* Handle LF, CR, CR-LF and LF-CR style newlines. Assumes next | |
104 | character, if any, is in buffer. */ | |
105 | #define handle_newline(cur, limit, c) \ | |
106 | do {\ | |
107 | if ((cur) < (limit) && *(cur) == '\r' + '\n' - c) \ | |
108 | (cur)++; \ | |
109 | CPP_BUMP_LINE_CUR (pfile, (cur)); \ | |
6ab3e7dd | 110 | pfile->col_adjust = 0; \ |
b8f41010 NB |
111 | } while (0) |
112 | ||
113 | #define IMMED_TOKEN() (!(cur_token->flags & PREV_WHITESPACE)) | |
114 | #define PREV_TOKEN_TYPE (cur_token[-1].type) | |
115 | ||
f617b8e2 NB |
116 | #define PUSH_TOKEN(ttype) cur_token++->type = ttype |
117 | #define REVISE_TOKEN(ttype) cur_token[-1].type = ttype | |
118 | #define BACKUP_TOKEN(ttype) (--cur_token)->type = ttype | |
119 | #define BACKUP_DIGRAPH(ttype) do { \ | |
120 | BACKUP_TOKEN(ttype); cur_token->flags |= DIGRAPH;} while (0) | |
121 | ||
122 | /* An upper bound on the number of bytes needed to spell a token, | |
123 | including preceding whitespace. */ | |
124 | #define TOKEN_LEN(token) (5 + (token_spellings[token->type].type > \ | |
125 | SPELL_NONE ? token->val.name.len: 0)) | |
126 | ||
127 | #endif | |
128 | ||
5d7ee2fa NB |
129 | /* Order here matters. Those beyond SPELL_NONE store their spelling |
130 | in the token list, and it's length in the token->val.name.len. */ | |
131 | #define SPELL_OPERATOR 0 | |
f617b8e2 NB |
132 | #define SPELL_CHAR 2 /* FIXME: revert order after transition. */ |
133 | #define SPELL_NONE 1 | |
5d7ee2fa NB |
134 | #define SPELL_IDENT 3 |
135 | #define SPELL_STRING 4 | |
136 | ||
f617b8e2 | 137 | #define T(e, s) {SPELL_OPERATOR, (const U_CHAR *) s}, |
5d7ee2fa NB |
138 | #define I(e, s) {SPELL_IDENT, s}, |
139 | #define S(e, s) {SPELL_STRING, s}, | |
b8f41010 NB |
140 | #define C(e, s) {SPELL_CHAR, s}, |
141 | #define N(e, s) {SPELL_NONE, s}, | |
b8f41010 NB |
142 | |
143 | static const struct token_spelling | |
144 | { | |
f617b8e2 NB |
145 | U_CHAR type; |
146 | const U_CHAR *spelling; | |
b8f41010 NB |
147 | } token_spellings [N_TTYPES + 1] = {TTYPE_TABLE {0, 0} }; |
148 | ||
149 | #undef T | |
5d7ee2fa NB |
150 | #undef I |
151 | #undef S | |
b8f41010 NB |
152 | #undef C |
153 | #undef N | |
b8f41010 | 154 | |
45b966db ZW |
155 | /* Re-allocates PFILE->token_buffer so it will hold at least N more chars. */ |
156 | ||
157 | void | |
158 | _cpp_grow_token_buffer (pfile, n) | |
159 | cpp_reader *pfile; | |
160 | long n; | |
161 | { | |
162 | long old_written = CPP_WRITTEN (pfile); | |
163 | pfile->token_buffer_size = n + 2 * pfile->token_buffer_size; | |
164 | pfile->token_buffer = (U_CHAR *) | |
165 | xrealloc(pfile->token_buffer, pfile->token_buffer_size); | |
166 | CPP_SET_WRITTEN (pfile, old_written); | |
167 | } | |
168 | ||
45b966db ZW |
169 | /* Allocate a new cpp_buffer for PFILE, and push it on the input buffer stack. |
170 | If BUFFER != NULL, then use the LENGTH characters in BUFFER | |
171 | as the new input buffer. | |
172 | Return the new buffer, or NULL on failure. */ | |
173 | ||
174 | cpp_buffer * | |
175 | cpp_push_buffer (pfile, buffer, length) | |
176 | cpp_reader *pfile; | |
177 | const U_CHAR *buffer; | |
178 | long length; | |
179 | { | |
180 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
181 | cpp_buffer *new; | |
182 | if (++pfile->buffer_stack_depth == CPP_STACK_MAX) | |
183 | { | |
184 | cpp_fatal (pfile, "macro or `#include' recursion too deep"); | |
185 | return NULL; | |
186 | } | |
187 | ||
188 | new = (cpp_buffer *) xcalloc (1, sizeof (cpp_buffer)); | |
189 | ||
190 | new->if_stack = pfile->if_stack; | |
45b966db | 191 | new->buf = new->cur = buffer; |
ff2b53ef | 192 | new->rlimit = buffer + length; |
45b966db | 193 | new->prev = buf; |
ff2b53ef | 194 | new->mark = NULL; |
45b966db ZW |
195 | new->line_base = NULL; |
196 | ||
197 | CPP_BUFFER (pfile) = new; | |
198 | return new; | |
199 | } | |
200 | ||
201 | cpp_buffer * | |
202 | cpp_pop_buffer (pfile) | |
203 | cpp_reader *pfile; | |
204 | { | |
205 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
206 | if (ACTIVE_MARK_P (pfile)) | |
207 | cpp_ice (pfile, "mark active in cpp_pop_buffer"); | |
c56c2073 ZW |
208 | |
209 | if (buf->ihash) | |
210 | { | |
211 | _cpp_unwind_if_stack (pfile, buf); | |
212 | if (buf->buf) | |
213 | free ((PTR) buf->buf); | |
214 | if (pfile->system_include_depth) | |
215 | pfile->system_include_depth--; | |
216 | if (pfile->potential_control_macro) | |
217 | { | |
218 | buf->ihash->control_macro = pfile->potential_control_macro; | |
219 | pfile->potential_control_macro = 0; | |
220 | } | |
221 | pfile->input_stack_listing_current = 0; | |
222 | } | |
223 | else if (buf->macro) | |
224 | { | |
225 | HASHNODE *m = buf->macro; | |
226 | ||
227 | m->disabled = 0; | |
228 | if ((m->type == T_FMACRO && buf->mapped) | |
229 | || m->type == T_SPECLINE || m->type == T_FILE | |
230 | || m->type == T_BASE_FILE || m->type == T_INCLUDE_LEVEL | |
231 | || m->type == T_STDC) | |
232 | free ((PTR) buf->buf); | |
233 | } | |
45b966db ZW |
234 | CPP_BUFFER (pfile) = CPP_PREV_BUFFER (buf); |
235 | free (buf); | |
236 | pfile->buffer_stack_depth--; | |
237 | return CPP_BUFFER (pfile); | |
238 | } | |
239 | ||
f2d5f0cc ZW |
240 | /* Deal with the annoying semantics of fwrite. */ |
241 | static void | |
242 | safe_fwrite (pfile, buf, len, fp) | |
243 | cpp_reader *pfile; | |
244 | const U_CHAR *buf; | |
245 | size_t len; | |
246 | FILE *fp; | |
247 | { | |
248 | size_t count; | |
45b966db | 249 | |
f2d5f0cc ZW |
250 | while (len) |
251 | { | |
252 | count = fwrite (buf, 1, len, fp); | |
253 | if (count == 0) | |
254 | goto error; | |
255 | len -= count; | |
256 | buf += count; | |
257 | } | |
258 | return; | |
259 | ||
260 | error: | |
261 | cpp_notice_from_errno (pfile, CPP_OPTION (pfile, out_fname)); | |
262 | } | |
263 | ||
264 | /* Notify the compiler proper that the current line number has jumped, | |
265 | or the current file name has changed. */ | |
266 | ||
267 | static void | |
1368ee70 | 268 | output_line_command (pfile, print, line) |
45b966db | 269 | cpp_reader *pfile; |
f2d5f0cc | 270 | cpp_printer *print; |
1368ee70 | 271 | unsigned int line; |
45b966db | 272 | { |
1368ee70 | 273 | cpp_buffer *ip = cpp_file_buffer (pfile); |
f2d5f0cc ZW |
274 | enum { same = 0, enter, leave, rname } change; |
275 | static const char * const codes[] = { "", " 1", " 2", "" }; | |
276 | ||
277 | if (CPP_OPTION (pfile, no_line_commands)) | |
278 | return; | |
279 | ||
f2d5f0cc ZW |
280 | /* Determine whether the current filename has changed, and if so, |
281 | how. 'nominal_fname' values are unique, so they can be compared | |
282 | by comparing pointers. */ | |
283 | if (ip->nominal_fname == print->last_fname) | |
284 | change = same; | |
285 | else | |
45b966db | 286 | { |
f2d5f0cc ZW |
287 | if (pfile->buffer_stack_depth == print->last_bsd) |
288 | change = rname; | |
289 | else | |
45b966db | 290 | { |
f2d5f0cc ZW |
291 | if (pfile->buffer_stack_depth > print->last_bsd) |
292 | change = enter; | |
293 | else | |
294 | change = leave; | |
295 | print->last_bsd = pfile->buffer_stack_depth; | |
45b966db | 296 | } |
f2d5f0cc | 297 | print->last_fname = ip->nominal_fname; |
45b966db | 298 | } |
f2d5f0cc ZW |
299 | /* If the current file has not changed, we can output a few newlines |
300 | instead if we want to increase the line number by a small amount. | |
301 | We cannot do this if print->lineno is zero, because that means we | |
302 | haven't output any line commands yet. (The very first line | |
303 | command output is a `same_file' command.) */ | |
304 | if (change == same && print->lineno != 0 | |
305 | && line >= print->lineno && line < print->lineno + 8) | |
45b966db | 306 | { |
f2d5f0cc | 307 | while (line > print->lineno) |
45b966db | 308 | { |
f2d5f0cc ZW |
309 | putc ('\n', print->outf); |
310 | print->lineno++; | |
45b966db | 311 | } |
f2d5f0cc | 312 | return; |
45b966db | 313 | } |
f2d5f0cc ZW |
314 | |
315 | #ifndef NO_IMPLICIT_EXTERN_C | |
316 | if (CPP_OPTION (pfile, cplusplus)) | |
317 | fprintf (print->outf, "# %u \"%s\"%s%s%s\n", line, ip->nominal_fname, | |
318 | codes[change], | |
319 | ip->system_header_p ? " 3" : "", | |
320 | (ip->system_header_p == 2) ? " 4" : ""); | |
321 | else | |
322 | #endif | |
323 | fprintf (print->outf, "# %u \"%s\"%s%s\n", line, ip->nominal_fname, | |
324 | codes[change], | |
325 | ip->system_header_p ? " 3" : ""); | |
326 | print->lineno = line; | |
327 | } | |
328 | ||
329 | /* Write the contents of the token_buffer to the output stream, and | |
330 | clear the token_buffer. Also handles generating line commands and | |
331 | keeping track of file transitions. */ | |
332 | ||
333 | void | |
334 | cpp_output_tokens (pfile, print) | |
335 | cpp_reader *pfile; | |
336 | cpp_printer *print; | |
337 | { | |
1368ee70 ZW |
338 | cpp_buffer *ip; |
339 | ||
f6fab919 ZW |
340 | if (CPP_WRITTEN (pfile) - print->written) |
341 | { | |
342 | if (CPP_PWRITTEN (pfile)[-1] == '\n' && print->lineno) | |
343 | print->lineno++; | |
344 | safe_fwrite (pfile, pfile->token_buffer, | |
345 | CPP_WRITTEN (pfile) - print->written, print->outf); | |
346 | } | |
1368ee70 ZW |
347 | |
348 | ip = cpp_file_buffer (pfile); | |
349 | if (ip) | |
350 | output_line_command (pfile, print, CPP_BUF_LINE (ip)); | |
351 | ||
f2d5f0cc | 352 | CPP_SET_WRITTEN (pfile, print->written); |
45b966db ZW |
353 | } |
354 | ||
1368ee70 ZW |
355 | /* Helper for cpp_output_list - increases the column number to match |
356 | what we expect it to be. */ | |
357 | ||
358 | static void | |
359 | bump_column (print, from, to) | |
360 | cpp_printer *print; | |
361 | unsigned int from, to; | |
362 | { | |
363 | unsigned int tabs, spcs; | |
364 | unsigned int delta = to - from; | |
365 | ||
366 | /* Only if FROM is 0, advance by tabs. */ | |
367 | if (from == 0) | |
368 | tabs = delta / 8, spcs = delta % 8; | |
369 | else | |
370 | tabs = 0, spcs = delta; | |
371 | ||
372 | while (tabs--) putc ('\t', print->outf); | |
373 | while (spcs--) putc (' ', print->outf); | |
374 | } | |
375 | ||
376 | /* Write out the list L onto pfile->token_buffer. This function is | |
377 | incomplete: | |
378 | ||
379 | 1) pfile->token_buffer is not going to continue to exist. | |
380 | 2) At the moment, tokens don't carry the information described | |
381 | in cpplib.h; they are all strings. | |
382 | 3) The list has to be a complete line, and has to be written starting | |
383 | at the beginning of a line. */ | |
384 | ||
385 | void | |
386 | cpp_output_list (pfile, print, list) | |
387 | cpp_reader *pfile; | |
388 | cpp_printer *print; | |
389 | const cpp_toklist *list; | |
390 | { | |
391 | unsigned int i; | |
392 | unsigned int curcol = 1; | |
393 | ||
394 | /* XXX Probably does not do what is intended. */ | |
395 | if (print->lineno != list->line) | |
396 | output_line_command (pfile, print, list->line); | |
397 | ||
398 | for (i = 0; i < list->tokens_used; i++) | |
399 | { | |
1920de47 | 400 | if (TOK_TYPE (list, i) == CPP_VSPACE) |
1368ee70 ZW |
401 | { |
402 | output_line_command (pfile, print, list->tokens[i].aux); | |
403 | continue; | |
404 | } | |
405 | ||
1920de47 | 406 | if (curcol < TOK_COL (list, i)) |
1368ee70 ZW |
407 | { |
408 | /* Insert space to bring the column to what it should be. */ | |
1920de47 ZW |
409 | bump_column (print, curcol - 1, TOK_COL (list, i)); |
410 | curcol = TOK_COL (list, i); | |
1368ee70 ZW |
411 | } |
412 | /* XXX We may have to insert space to prevent an accidental | |
413 | token paste. */ | |
1920de47 ZW |
414 | safe_fwrite (pfile, TOK_NAME (list, i), TOK_LEN (list, i), print->outf); |
415 | curcol += TOK_LEN (list, i); | |
1368ee70 ZW |
416 | } |
417 | } | |
418 | ||
f2d5f0cc ZW |
419 | /* Scan a string (which may have escape marks), perform macro expansion, |
420 | and write the result to the token_buffer. */ | |
45b966db ZW |
421 | |
422 | void | |
f2d5f0cc | 423 | _cpp_expand_to_buffer (pfile, buf, length) |
45b966db ZW |
424 | cpp_reader *pfile; |
425 | const U_CHAR *buf; | |
426 | int length; | |
427 | { | |
c56c2073 | 428 | cpp_buffer *stop; |
f2d5f0cc | 429 | enum cpp_ttype token; |
f6fab919 | 430 | U_CHAR *buf1; |
45b966db ZW |
431 | |
432 | if (length < 0) | |
433 | { | |
434 | cpp_ice (pfile, "length < 0 in cpp_expand_to_buffer"); | |
435 | return; | |
436 | } | |
437 | ||
f6fab919 ZW |
438 | /* Copy the buffer, because it might be in an unsafe place - for |
439 | example, a sequence on the token_buffer, where the pointers will | |
440 | be invalidated if we enlarge the token_buffer. */ | |
441 | buf1 = alloca (length); | |
442 | memcpy (buf1, buf, length); | |
443 | ||
45b966db | 444 | /* Set up the input on the input stack. */ |
c56c2073 ZW |
445 | stop = CPP_BUFFER (pfile); |
446 | if (cpp_push_buffer (pfile, buf1, length) == NULL) | |
45b966db | 447 | return; |
c56c2073 | 448 | CPP_BUFFER (pfile)->has_escapes = 1; |
45b966db ZW |
449 | |
450 | /* Scan the input, create the output. */ | |
f2d5f0cc ZW |
451 | for (;;) |
452 | { | |
453 | token = cpp_get_token (pfile); | |
c56c2073 | 454 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 455 | break; |
f2d5f0cc | 456 | } |
45b966db ZW |
457 | } |
458 | ||
c56c2073 | 459 | /* Scan until CPP_BUFFER (PFILE) is exhausted, discarding output. */ |
f2d5f0cc ZW |
460 | |
461 | void | |
462 | cpp_scan_buffer_nooutput (pfile) | |
463 | cpp_reader *pfile; | |
464 | { | |
c56c2073 | 465 | cpp_buffer *stop = CPP_PREV_BUFFER (CPP_BUFFER (pfile)); |
f2d5f0cc ZW |
466 | enum cpp_ttype token; |
467 | unsigned int old_written = CPP_WRITTEN (pfile); | |
468 | /* In no-output mode, we can ignore everything but directives. */ | |
469 | for (;;) | |
470 | { | |
471 | if (! pfile->only_seen_white) | |
472 | _cpp_skip_rest_of_line (pfile); | |
473 | token = cpp_get_token (pfile); | |
c56c2073 | 474 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 475 | break; |
f2d5f0cc ZW |
476 | } |
477 | CPP_SET_WRITTEN (pfile, old_written); | |
478 | } | |
479 | ||
c56c2073 | 480 | /* Scan until CPP_BUFFER (pfile) is exhausted, writing output to PRINT. */ |
f2d5f0cc ZW |
481 | |
482 | void | |
483 | cpp_scan_buffer (pfile, print) | |
484 | cpp_reader *pfile; | |
485 | cpp_printer *print; | |
486 | { | |
c56c2073 | 487 | cpp_buffer *stop = CPP_PREV_BUFFER (CPP_BUFFER (pfile)); |
f2d5f0cc ZW |
488 | enum cpp_ttype token; |
489 | ||
490 | for (;;) | |
491 | { | |
492 | token = cpp_get_token (pfile); | |
15dad1d9 | 493 | if (token == CPP_VSPACE || token == CPP_EOF |
f2d5f0cc ZW |
494 | /* XXX Temporary kluge - force flush after #include only */ |
495 | || (token == CPP_DIRECTIVE | |
496 | && CPP_BUFFER (pfile)->nominal_fname != print->last_fname)) | |
497 | { | |
498 | cpp_output_tokens (pfile, print); | |
c56c2073 | 499 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 500 | return; |
f2d5f0cc ZW |
501 | } |
502 | } | |
503 | } | |
504 | ||
45b966db ZW |
505 | /* Return the topmost cpp_buffer that corresponds to a file (not a macro). */ |
506 | ||
507 | cpp_buffer * | |
508 | cpp_file_buffer (pfile) | |
509 | cpp_reader *pfile; | |
510 | { | |
511 | cpp_buffer *ip; | |
512 | ||
513 | for (ip = CPP_BUFFER (pfile); ip; ip = CPP_PREV_BUFFER (ip)) | |
514 | if (ip->ihash != NULL) | |
515 | return ip; | |
516 | return NULL; | |
517 | } | |
518 | ||
1368ee70 ZW |
519 | /* Token-buffer helper functions. */ |
520 | ||
521 | /* Expand a token list's string space. */ | |
522 | static void | |
c5a04734 | 523 | expand_name_space (list, len) |
1368ee70 | 524 | cpp_toklist *list; |
c5a04734 ZW |
525 | unsigned int len; |
526 | { | |
f617b8e2 | 527 | const U_CHAR *old_namebuf; |
f617b8e2 NB |
528 | |
529 | old_namebuf = list->namebuf; | |
c5a04734 ZW |
530 | list->name_cap += len; |
531 | list->namebuf = (unsigned char *) xrealloc (list->namebuf, list->name_cap); | |
f617b8e2 NB |
532 | |
533 | /* Fix up token text pointers. */ | |
79f50f2a | 534 | if (list->namebuf != old_namebuf) |
f617b8e2 NB |
535 | { |
536 | unsigned int i; | |
537 | ||
538 | for (i = 0; i < list->tokens_used; i++) | |
539 | if (token_spellings[list->tokens[i].type].type > SPELL_NONE) | |
79f50f2a | 540 | list->tokens[i].val.name.text += (list->namebuf - old_namebuf); |
f617b8e2 | 541 | } |
1368ee70 ZW |
542 | } |
543 | ||
544 | /* Expand the number of tokens in a list. */ | |
545 | static void | |
546 | expand_token_space (list) | |
547 | cpp_toklist *list; | |
548 | { | |
549 | list->tokens_cap *= 2; | |
15dad1d9 ZW |
550 | if (list->flags & LIST_OFFSET) |
551 | list->tokens--; | |
1368ee70 | 552 | list->tokens = (cpp_token *) |
15dad1d9 ZW |
553 | xrealloc (list->tokens, (list->tokens_cap + 1) * sizeof (cpp_token)); |
554 | if (list->flags & LIST_OFFSET) | |
555 | list->tokens++; /* Skip the dummy. */ | |
1368ee70 ZW |
556 | } |
557 | ||
c5a04734 ZW |
558 | /* Initialize a token list. We allocate an extra token in front of |
559 | the token list, as this allows us to always peek at the previous | |
560 | token without worrying about underflowing the list. */ | |
15dad1d9 ZW |
561 | void |
562 | _cpp_init_toklist (list) | |
1368ee70 | 563 | cpp_toklist *list; |
1368ee70 | 564 | { |
15dad1d9 ZW |
565 | /* Initialize token space. Put a dummy token before the start |
566 | that will fail matches. */ | |
567 | list->tokens_cap = 256; /* 4K's worth. */ | |
568 | list->tokens = (cpp_token *) | |
569 | xmalloc ((list->tokens_cap + 1) * sizeof (cpp_token)); | |
570 | list->tokens[0].type = CPP_EOF; | |
571 | list->tokens++; | |
572 | ||
573 | /* Initialize name space. */ | |
574 | list->name_cap = 1024; | |
575 | list->namebuf = (unsigned char *) xmalloc (list->name_cap); | |
576 | ||
577 | /* Only create a comment space on demand. */ | |
578 | list->comments_cap = 0; | |
579 | list->comments = 0; | |
580 | ||
581 | list->flags = LIST_OFFSET; | |
582 | _cpp_clear_toklist (list); | |
583 | } | |
1368ee70 | 584 | |
15dad1d9 ZW |
585 | /* Clear a token list. */ |
586 | void | |
587 | _cpp_clear_toklist (list) | |
588 | cpp_toklist *list; | |
589 | { | |
c5a04734 ZW |
590 | list->tokens_used = 0; |
591 | list->name_used = 0; | |
592 | list->comments_used = 0; | |
15dad1d9 ZW |
593 | list->dirno = -1; |
594 | list->flags &= LIST_OFFSET; /* clear all but that one */ | |
595 | } | |
596 | ||
597 | /* Free a token list. Does not free the list itself, which may be | |
598 | embedded in a larger structure. */ | |
599 | void | |
600 | _cpp_free_toklist (list) | |
601 | cpp_toklist *list; | |
602 | { | |
603 | if (list->comments) | |
604 | free (list->comments); | |
605 | if (list->flags & LIST_OFFSET) | |
606 | free (list->tokens - 1); /* Backup over dummy token. */ | |
607 | else | |
608 | free (list->tokens); | |
609 | free (list->namebuf); | |
1368ee70 ZW |
610 | } |
611 | ||
15dad1d9 ZW |
612 | /* Slice a token list: copy the sublist [START, FINISH) into COPY. |
613 | COPY is assumed not to be initialized. The comment space is not | |
614 | copied. */ | |
615 | void | |
616 | _cpp_slice_toklist (copy, start, finish) | |
617 | cpp_toklist *copy; | |
618 | const cpp_token *start, *finish; | |
619 | { | |
620 | unsigned int i, n; | |
621 | size_t bytes; | |
622 | ||
623 | n = finish - start; | |
624 | copy->tokens_cap = n; | |
625 | copy->tokens = (cpp_token *) xmalloc (n * sizeof (cpp_token)); | |
626 | memcpy (copy->tokens, start, n * sizeof (cpp_token)); | |
627 | ||
628 | bytes = 0; | |
629 | for (i = 0; i < n; i++) | |
630 | if (token_spellings[start[i].type].type > SPELL_NONE) | |
631 | bytes += start[i].val.name.len; | |
632 | ||
633 | copy->namebuf = xmalloc (bytes); | |
634 | bytes = 0; | |
635 | for (i = 0; i < n; i++) | |
636 | if (token_spellings[start[i].type].type > SPELL_NONE) | |
637 | { | |
638 | memcpy (copy->namebuf + bytes, | |
639 | start[i].val.name.text, start[i].val.name.len); | |
640 | copy->tokens[i].val.name.text = copy->namebuf + bytes; | |
641 | bytes += start[i].val.name.len; | |
642 | } | |
643 | ||
644 | copy->tokens_cap = n; | |
645 | copy->tokens_used = n; | |
646 | copy->name_used = bytes; | |
647 | copy->name_cap = bytes; | |
648 | copy->comments = 0; | |
649 | copy->comments_cap = 0; | |
650 | copy->comments_used = 0; | |
651 | ||
652 | copy->flags = 0; | |
653 | copy->dirno = -1; | |
654 | } | |
1368ee70 | 655 | |
15dad1d9 | 656 | /* Shrink a token list down to the minimum size. */ |
1368ee70 | 657 | void |
15dad1d9 ZW |
658 | _cpp_squeeze_toklist (list) |
659 | cpp_toklist *list; | |
660 | { | |
661 | long delta; | |
662 | const U_CHAR *old_namebuf; | |
663 | ||
664 | if (list->flags & LIST_OFFSET) | |
665 | { | |
666 | list->tokens--; | |
667 | memmove (list->tokens, list->tokens + 1, | |
668 | list->tokens_used * sizeof (cpp_token)); | |
669 | list->tokens = xrealloc (list->tokens, | |
670 | list->tokens_used * sizeof (cpp_token)); | |
671 | list->flags &= ~LIST_OFFSET; | |
672 | } | |
673 | else | |
674 | list->tokens = xrealloc (list->tokens, | |
675 | list->tokens_used * sizeof (cpp_token)); | |
676 | list->tokens_cap = list->tokens_used; | |
677 | ||
678 | old_namebuf = list->namebuf; | |
679 | list->namebuf = xrealloc (list->namebuf, list->name_used); | |
680 | list->name_cap = list->name_used; | |
681 | ||
682 | /* Fix up token text pointers. */ | |
683 | delta = list->namebuf - old_namebuf; | |
684 | if (delta) | |
685 | { | |
686 | unsigned int i; | |
687 | ||
688 | for (i = 0; i < list->tokens_used; i++) | |
689 | if (token_spellings[list->tokens[i].type].type > SPELL_NONE) | |
690 | list->tokens[i].val.name.text += delta; | |
691 | } | |
692 | ||
693 | if (list->comments_cap) | |
694 | { | |
695 | list->comments = xrealloc (list->comments, | |
696 | list->comments_used * sizeof (cpp_token)); | |
697 | list->comments_cap = list->comments_used; | |
698 | } | |
699 | } | |
700 | ||
701 | /* Compare two tokens. */ | |
702 | int | |
703 | _cpp_equiv_tokens (a, b) | |
704 | const cpp_token *a, *b; | |
705 | { | |
706 | if (a->type != b->type | |
707 | || a->flags != b->flags | |
708 | || a->aux != b->aux) | |
709 | return 0; | |
710 | ||
711 | if (token_spellings[a->type].type > SPELL_NONE) | |
712 | { | |
713 | if (a->val.name.len != b->val.name.len | |
714 | || ustrncmp(a->val.name.text, | |
715 | b->val.name.text, | |
716 | a->val.name.len)) | |
717 | return 0; | |
718 | } | |
719 | return 1; | |
720 | } | |
721 | ||
722 | /* Compare two token lists. */ | |
723 | int | |
724 | _cpp_equiv_toklists (a, b) | |
725 | const cpp_toklist *a, *b; | |
726 | { | |
727 | unsigned int i; | |
728 | ||
729 | if (a->tokens_used != b->tokens_used) | |
730 | return 0; | |
731 | ||
732 | for (i = 0; i < a->tokens_used; i++) | |
733 | if (! _cpp_equiv_tokens (&a->tokens[i], &b->tokens[i])) | |
734 | return 0; | |
735 | return 1; | |
736 | } | |
737 | ||
738 | /* Scan until we encounter a token of type STOP or a newline, and | |
739 | create a token list for it. Does not macro-expand or execute | |
740 | directives. The final token is not included in the list or | |
741 | consumed from the input. Returns the type of the token stopped at. */ | |
742 | ||
743 | enum cpp_ttype | |
744 | _cpp_scan_until (pfile, list, stop) | |
1368ee70 ZW |
745 | cpp_reader *pfile; |
746 | cpp_toklist *list; | |
15dad1d9 | 747 | enum cpp_ttype stop; |
1368ee70 ZW |
748 | { |
749 | int i, col; | |
750 | long written, len; | |
751 | enum cpp_ttype type; | |
9e62c811 | 752 | int space_before; |
1368ee70 | 753 | |
15dad1d9 ZW |
754 | _cpp_clear_toklist (list); |
755 | list->line = CPP_BUF_LINE (CPP_BUFFER (pfile)); | |
1368ee70 ZW |
756 | |
757 | written = CPP_WRITTEN (pfile); | |
758 | i = 0; | |
9e62c811 | 759 | space_before = 0; |
1368ee70 ZW |
760 | for (;;) |
761 | { | |
762 | col = CPP_BUFFER (pfile)->cur - CPP_BUFFER (pfile)->line_base; | |
763 | type = _cpp_lex_token (pfile); | |
764 | len = CPP_WRITTEN (pfile) - written; | |
765 | CPP_SET_WRITTEN (pfile, written); | |
766 | if (type == CPP_HSPACE) | |
9e62c811 ZW |
767 | { |
768 | if (CPP_PEDANTIC (pfile)) | |
769 | pedantic_whitespace (pfile, pfile->token_buffer + written, len); | |
770 | space_before = 1; | |
771 | continue; | |
772 | } | |
0f89df67 ZW |
773 | else if (type == CPP_COMMENT) |
774 | /* Only happens when processing -traditional macro definitions. | |
775 | Do not give this a token entry, but do not change space_before | |
776 | either. */ | |
777 | continue; | |
1368ee70 ZW |
778 | |
779 | if (list->tokens_used >= list->tokens_cap) | |
780 | expand_token_space (list); | |
781 | if (list->name_used + len >= list->name_cap) | |
bb1ec1d7 | 782 | expand_name_space (list, list->name_used + len + 1 - list->name_cap); |
1368ee70 | 783 | |
9e62c811 ZW |
784 | if (type == CPP_MACRO) |
785 | type = CPP_NAME; | |
786 | ||
15dad1d9 ZW |
787 | if (type == CPP_VSPACE || type == stop) |
788 | break; | |
789 | ||
1368ee70 | 790 | list->tokens_used++; |
1920de47 ZW |
791 | TOK_TYPE (list, i) = type; |
792 | TOK_COL (list, i) = col; | |
a58f64f5 | 793 | TOK_AUX (list, i) = 0; |
1920de47 | 794 | TOK_FLAGS (list, i) = space_before ? PREV_WHITESPACE : 0; |
9e62c811 | 795 | |
1920de47 | 796 | TOK_LEN (list, i) = len; |
f617b8e2 NB |
797 | if (token_spellings[type].type > SPELL_NONE) |
798 | { | |
799 | memcpy (list->namebuf + list->name_used, CPP_PWRITTEN (pfile), len); | |
800 | TOK_NAME (list, i) = list->namebuf + list->name_used; | |
801 | list->name_used += len; | |
802 | } | |
803 | else | |
804 | TOK_NAME (list, i) = token_spellings[type].spelling; | |
1368ee70 | 805 | i++; |
9e62c811 | 806 | space_before = 0; |
1368ee70 | 807 | } |
9e62c811 | 808 | |
15dad1d9 | 809 | /* XXX Temporary kluge: put back the newline (or whatever). */ |
9e62c811 | 810 | FORWARD(-1); |
1368ee70 | 811 | |
15dad1d9 ZW |
812 | /* Don't consider the first token to have white before. */ |
813 | TOK_FLAGS (list, 0) &= ~PREV_WHITESPACE; | |
814 | return type; | |
815 | } | |
1368ee70 | 816 | |
45b966db ZW |
817 | /* Skip a C-style block comment. We know it's a comment, and point is |
818 | at the second character of the starter. */ | |
819 | static void | |
820 | skip_block_comment (pfile) | |
821 | cpp_reader *pfile; | |
822 | { | |
3a2b2c7a | 823 | unsigned int line, col; |
61474454 | 824 | const U_CHAR *limit, *cur; |
45b966db ZW |
825 | |
826 | FORWARD(1); | |
3a2b2c7a ZW |
827 | line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
828 | col = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
61474454 NB |
829 | limit = CPP_BUFFER (pfile)->rlimit; |
830 | cur = CPP_BUFFER (pfile)->cur; | |
831 | ||
832 | while (cur < limit) | |
45b966db | 833 | { |
61474454 NB |
834 | char c = *cur++; |
835 | if (c == '\n' || c == '\r') | |
45b966db ZW |
836 | { |
837 | /* \r cannot be a macro escape marker here. */ | |
838 | if (!ACTIVE_MARK_P (pfile)) | |
61474454 NB |
839 | CPP_BUMP_LINE_CUR (pfile, cur); |
840 | } | |
841 | else if (c == '*') | |
842 | { | |
843 | /* Check for teminator. */ | |
844 | if (cur < limit && *cur == '/') | |
845 | goto out; | |
846 | ||
847 | /* Warn about comment starter embedded in comment. */ | |
848 | if (cur[-2] == '/' && CPP_OPTION (pfile, warn_comments)) | |
849 | cpp_warning_with_line (pfile, CPP_BUFFER (pfile)->lineno, | |
850 | cur - CPP_BUFFER (pfile)->line_base, | |
851 | "'/*' within comment"); | |
45b966db | 852 | } |
45b966db | 853 | } |
61474454 NB |
854 | |
855 | cpp_error_with_line (pfile, line, col, "unterminated comment"); | |
856 | cur--; | |
857 | out: | |
858 | CPP_BUFFER (pfile)->cur = cur + 1; | |
45b966db ZW |
859 | } |
860 | ||
861 | /* Skip a C++/Chill line comment. We know it's a comment, and point | |
862 | is at the second character of the initiator. */ | |
863 | static void | |
864 | skip_line_comment (pfile) | |
865 | cpp_reader *pfile; | |
866 | { | |
867 | FORWARD(1); | |
868 | for (;;) | |
869 | { | |
870 | int c = GETC (); | |
871 | ||
872 | /* We don't have to worry about EOF in here. */ | |
873 | if (c == '\n') | |
874 | { | |
875 | /* Don't consider final '\n' to be part of comment. */ | |
876 | FORWARD(-1); | |
877 | return; | |
878 | } | |
879 | else if (c == '\r') | |
880 | { | |
881 | /* \r cannot be a macro escape marker here. */ | |
882 | if (!ACTIVE_MARK_P (pfile)) | |
883 | CPP_BUMP_LINE (pfile); | |
ae79697b | 884 | if (CPP_OPTION (pfile, warn_comments)) |
45b966db ZW |
885 | cpp_warning (pfile, "backslash-newline within line comment"); |
886 | } | |
887 | } | |
888 | } | |
889 | ||
890 | /* Skip a comment - C, C++, or Chill style. M is the first character | |
891 | of the comment marker. If this really is a comment, skip to its | |
892 | end and return ' '. If this is not a comment, return M (which will | |
893 | be '/' or '-'). */ | |
894 | ||
895 | static int | |
896 | skip_comment (pfile, m) | |
897 | cpp_reader *pfile; | |
898 | int m; | |
899 | { | |
900 | if (m == '/' && PEEKC() == '*') | |
901 | { | |
902 | skip_block_comment (pfile); | |
903 | return ' '; | |
904 | } | |
905 | else if (m == '/' && PEEKC() == '/') | |
906 | { | |
907 | if (CPP_BUFFER (pfile)->system_header_p) | |
908 | { | |
909 | /* We silently allow C++ comments in system headers, irrespective | |
910 | of conformance mode, because lots of busted systems do that | |
911 | and trying to clean it up in fixincludes is a nightmare. */ | |
912 | skip_line_comment (pfile); | |
913 | return ' '; | |
914 | } | |
ae79697b | 915 | else if (CPP_OPTION (pfile, cplusplus_comments)) |
45b966db | 916 | { |
0f89df67 | 917 | if (! CPP_BUFFER (pfile)->warned_cplusplus_comments) |
45b966db | 918 | { |
0f89df67 ZW |
919 | if (CPP_WTRADITIONAL (pfile)) |
920 | cpp_pedwarn (pfile, | |
921 | "C++ style comments are not allowed in traditional C"); | |
922 | else if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile)) | |
923 | cpp_pedwarn (pfile, | |
924 | "C++ style comments are not allowed in ISO C89"); | |
925 | if (CPP_WTRADITIONAL (pfile) | |
926 | || (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile))) | |
927 | cpp_pedwarn (pfile, | |
45b966db ZW |
928 | "(this will be reported only once per input file)"); |
929 | CPP_BUFFER (pfile)->warned_cplusplus_comments = 1; | |
930 | } | |
931 | skip_line_comment (pfile); | |
932 | return ' '; | |
933 | } | |
934 | else | |
935 | return m; | |
936 | } | |
937 | else if (m == '-' && PEEKC() == '-' | |
ae79697b | 938 | && CPP_OPTION (pfile, chill)) |
45b966db ZW |
939 | { |
940 | skip_line_comment (pfile); | |
941 | return ' '; | |
942 | } | |
943 | else | |
944 | return m; | |
945 | } | |
946 | ||
947 | /* Identical to skip_comment except that it copies the comment into the | |
948 | token_buffer. This is used if !discard_comments. */ | |
949 | static int | |
950 | copy_comment (pfile, m) | |
951 | cpp_reader *pfile; | |
952 | int m; | |
953 | { | |
954 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
955 | const U_CHAR *limit; | |
956 | ||
957 | if (skip_comment (pfile, m) == m) | |
958 | return m; | |
959 | ||
960 | limit = CPP_BUFFER (pfile)->cur; | |
961 | CPP_RESERVE (pfile, limit - start + 2); | |
962 | CPP_PUTC_Q (pfile, m); | |
963 | for (; start <= limit; start++) | |
964 | if (*start != '\r') | |
965 | CPP_PUTC_Q (pfile, *start); | |
966 | ||
967 | return ' '; | |
968 | } | |
969 | ||
64aaf407 NB |
970 | static void |
971 | null_warning (pfile, count) | |
972 | cpp_reader *pfile; | |
973 | unsigned int count; | |
974 | { | |
975 | if (count == 1) | |
976 | cpp_warning (pfile, "embedded null character ignored"); | |
977 | else | |
978 | cpp_warning (pfile, "embedded null characters ignored"); | |
979 | } | |
980 | ||
45b966db ZW |
981 | /* Skip whitespace \-newline and comments. Does not macro-expand. */ |
982 | ||
983 | void | |
984 | _cpp_skip_hspace (pfile) | |
985 | cpp_reader *pfile; | |
986 | { | |
64aaf407 | 987 | unsigned int null_count = 0; |
45b966db | 988 | int c; |
64aaf407 | 989 | |
45b966db ZW |
990 | while (1) |
991 | { | |
992 | c = GETC(); | |
993 | if (c == EOF) | |
64aaf407 | 994 | goto out; |
45b966db ZW |
995 | else if (is_hspace(c)) |
996 | { | |
997 | if ((c == '\f' || c == '\v') && CPP_PEDANTIC (pfile)) | |
998 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
999 | c == '\f' ? "formfeed" : "vertical tab"); | |
64aaf407 NB |
1000 | else if (c == '\0') |
1001 | null_count++; | |
45b966db ZW |
1002 | } |
1003 | else if (c == '\r') | |
1004 | { | |
1005 | /* \r is a backslash-newline marker if !has_escapes, and | |
1006 | a deletable-whitespace or no-reexpansion marker otherwise. */ | |
1007 | if (CPP_BUFFER (pfile)->has_escapes) | |
1008 | { | |
1009 | if (PEEKC() == ' ') | |
1010 | FORWARD(1); | |
1011 | else | |
1012 | break; | |
1013 | } | |
1014 | else | |
1015 | CPP_BUMP_LINE (pfile); | |
1016 | } | |
1017 | else if (c == '/' || c == '-') | |
1018 | { | |
1019 | c = skip_comment (pfile, c); | |
1020 | if (c != ' ') | |
1021 | break; | |
1022 | } | |
1023 | else | |
1024 | break; | |
1025 | } | |
1026 | FORWARD(-1); | |
64aaf407 NB |
1027 | out: |
1028 | if (null_count) | |
1029 | null_warning (pfile, null_count); | |
45b966db ZW |
1030 | } |
1031 | ||
1032 | /* Read and discard the rest of the current line. */ | |
1033 | ||
1034 | void | |
1035 | _cpp_skip_rest_of_line (pfile) | |
1036 | cpp_reader *pfile; | |
1037 | { | |
1038 | for (;;) | |
1039 | { | |
1040 | int c = GETC(); | |
1041 | switch (c) | |
1042 | { | |
1043 | case '\n': | |
1044 | FORWARD(-1); | |
1045 | case EOF: | |
1046 | return; | |
1047 | ||
1048 | case '\r': | |
1049 | if (! CPP_BUFFER (pfile)->has_escapes) | |
1050 | CPP_BUMP_LINE (pfile); | |
1051 | break; | |
1052 | ||
1053 | case '\'': | |
1054 | case '\"': | |
1055 | skip_string (pfile, c); | |
1056 | break; | |
1057 | ||
1058 | case '/': | |
1059 | case '-': | |
1060 | skip_comment (pfile, c); | |
1061 | break; | |
1062 | ||
1063 | case '\f': | |
1064 | case '\v': | |
1065 | if (CPP_PEDANTIC (pfile)) | |
1066 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
1067 | c == '\f' ? "formfeed" : "vertical tab"); | |
1068 | break; | |
1069 | ||
1070 | } | |
1071 | } | |
1072 | } | |
1073 | ||
1074 | /* Parse an identifier starting with C. */ | |
1075 | ||
1076 | void | |
1077 | _cpp_parse_name (pfile, c) | |
1078 | cpp_reader *pfile; | |
1079 | int c; | |
1080 | { | |
1081 | for (;;) | |
1082 | { | |
1083 | if (! is_idchar(c)) | |
1084 | { | |
1085 | FORWARD (-1); | |
1086 | break; | |
1087 | } | |
1088 | ||
1089 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
1090 | cpp_pedwarn (pfile, "`$' in identifier"); | |
1091 | ||
1092 | CPP_RESERVE(pfile, 2); /* One more for final NUL. */ | |
1093 | CPP_PUTC_Q (pfile, c); | |
1094 | c = GETC(); | |
1095 | if (c == EOF) | |
1096 | break; | |
1097 | } | |
45b966db ZW |
1098 | return; |
1099 | } | |
1100 | ||
1101 | /* Parse and skip over a string starting with C. A single quoted | |
1102 | string is treated like a double -- some programs (e.g., troff) are | |
1103 | perverse this way. (However, a single quoted string is not allowed | |
1104 | to extend over multiple lines.) */ | |
1105 | static void | |
1106 | skip_string (pfile, c) | |
1107 | cpp_reader *pfile; | |
1108 | int c; | |
1109 | { | |
3a2b2c7a | 1110 | unsigned int start_line, start_column; |
64aaf407 | 1111 | unsigned int null_count = 0; |
45b966db | 1112 | |
3a2b2c7a ZW |
1113 | start_line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
1114 | start_column = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
45b966db ZW |
1115 | while (1) |
1116 | { | |
1117 | int cc = GETC(); | |
1118 | switch (cc) | |
1119 | { | |
1120 | case EOF: | |
1121 | cpp_error_with_line (pfile, start_line, start_column, | |
1122 | "unterminated string or character constant"); | |
1123 | if (pfile->multiline_string_line != start_line | |
1124 | && pfile->multiline_string_line != 0) | |
1125 | cpp_error_with_line (pfile, | |
1126 | pfile->multiline_string_line, -1, | |
1127 | "possible real start of unterminated constant"); | |
1128 | pfile->multiline_string_line = 0; | |
64aaf407 | 1129 | goto out; |
45b966db | 1130 | |
64aaf407 NB |
1131 | case '\0': |
1132 | null_count++; | |
1133 | break; | |
1134 | ||
45b966db ZW |
1135 | case '\n': |
1136 | CPP_BUMP_LINE (pfile); | |
1137 | /* In Fortran and assembly language, silently terminate | |
1138 | strings of either variety at end of line. This is a | |
1139 | kludge around not knowing where comments are in these | |
1140 | languages. */ | |
ae79697b ZW |
1141 | if (CPP_OPTION (pfile, lang_fortran) |
1142 | || CPP_OPTION (pfile, lang_asm)) | |
45b966db ZW |
1143 | { |
1144 | FORWARD(-1); | |
64aaf407 | 1145 | goto out; |
45b966db ZW |
1146 | } |
1147 | /* Character constants may not extend over multiple lines. | |
1148 | In Standard C, neither may strings. We accept multiline | |
1149 | strings as an extension. */ | |
1150 | if (c == '\'') | |
1151 | { | |
1152 | cpp_error_with_line (pfile, start_line, start_column, | |
1153 | "unterminated character constant"); | |
1154 | FORWARD(-1); | |
64aaf407 | 1155 | goto out; |
45b966db ZW |
1156 | } |
1157 | if (CPP_PEDANTIC (pfile) && pfile->multiline_string_line == 0) | |
1158 | cpp_pedwarn_with_line (pfile, start_line, start_column, | |
1159 | "string constant runs past end of line"); | |
1160 | if (pfile->multiline_string_line == 0) | |
1161 | pfile->multiline_string_line = start_line; | |
1162 | break; | |
1163 | ||
1164 | case '\r': | |
1165 | if (CPP_BUFFER (pfile)->has_escapes) | |
1166 | { | |
1167 | cpp_ice (pfile, "\\r escape inside string constant"); | |
1168 | FORWARD(1); | |
1169 | } | |
1170 | else | |
1171 | /* Backslash newline is replaced by nothing at all. */ | |
1172 | CPP_BUMP_LINE (pfile); | |
1173 | break; | |
1174 | ||
1175 | case '\\': | |
1176 | FORWARD(1); | |
1177 | break; | |
1178 | ||
1179 | case '\"': | |
1180 | case '\'': | |
1181 | if (cc == c) | |
64aaf407 | 1182 | goto out; |
45b966db ZW |
1183 | break; |
1184 | } | |
1185 | } | |
64aaf407 NB |
1186 | |
1187 | out: | |
1188 | if (null_count == 1) | |
1189 | cpp_warning (pfile, "null character in string or character constant"); | |
1190 | else if (null_count > 1) | |
1191 | cpp_warning (pfile, "null characters in string or character constant"); | |
45b966db ZW |
1192 | } |
1193 | ||
1194 | /* Parse a string and copy it to the output. */ | |
1195 | ||
1196 | static void | |
1197 | parse_string (pfile, c) | |
1198 | cpp_reader *pfile; | |
1199 | int c; | |
1200 | { | |
1201 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
1202 | const U_CHAR *limit; | |
1203 | ||
1204 | skip_string (pfile, c); | |
1205 | ||
1206 | limit = CPP_BUFFER (pfile)->cur; | |
1207 | CPP_RESERVE (pfile, limit - start + 2); | |
1208 | CPP_PUTC_Q (pfile, c); | |
1209 | for (; start < limit; start++) | |
1210 | if (*start != '\r') | |
1211 | CPP_PUTC_Q (pfile, *start); | |
1212 | } | |
1213 | ||
45b966db ZW |
1214 | /* Get the next token, and add it to the text in pfile->token_buffer. |
1215 | Return the kind of token we got. */ | |
1216 | ||
3a2b2c7a | 1217 | enum cpp_ttype |
45b966db ZW |
1218 | _cpp_lex_token (pfile) |
1219 | cpp_reader *pfile; | |
1220 | { | |
5eec0563 | 1221 | register int c, c2; |
3a2b2c7a | 1222 | enum cpp_ttype token; |
45b966db | 1223 | |
f2d5f0cc ZW |
1224 | if (CPP_BUFFER (pfile) == NULL) |
1225 | return CPP_EOF; | |
1226 | ||
45b966db ZW |
1227 | get_next: |
1228 | c = GETC(); | |
1229 | switch (c) | |
1230 | { | |
1231 | case EOF: | |
1232 | return CPP_EOF; | |
1233 | ||
1234 | case '/': | |
1235 | if (PEEKC () == '=') | |
1236 | goto op2; | |
1237 | ||
1238 | comment: | |
ae79697b | 1239 | if (CPP_OPTION (pfile, discard_comments)) |
45b966db ZW |
1240 | c = skip_comment (pfile, c); |
1241 | else | |
1242 | c = copy_comment (pfile, c); | |
1243 | if (c != ' ') | |
1244 | goto randomchar; | |
1245 | ||
1246 | /* Comments are equivalent to spaces. | |
1247 | For -traditional, a comment is equivalent to nothing. */ | |
ff2b53ef | 1248 | if (!CPP_OPTION (pfile, discard_comments)) |
45b966db | 1249 | return CPP_COMMENT; |
9e62c811 | 1250 | else if (CPP_TRADITIONAL (pfile)) |
15dad1d9 | 1251 | goto get_next; |
45b966db ZW |
1252 | else |
1253 | { | |
1254 | CPP_PUTC (pfile, c); | |
1255 | return CPP_HSPACE; | |
1256 | } | |
1257 | ||
1258 | case '#': | |
5eec0563 JM |
1259 | CPP_PUTC (pfile, c); |
1260 | ||
1261 | hash: | |
15dad1d9 ZW |
1262 | c2 = PEEKC (); |
1263 | if (c2 == '#') | |
45b966db | 1264 | { |
15dad1d9 ZW |
1265 | FORWARD (1); |
1266 | CPP_PUTC (pfile, c2); | |
1267 | return CPP_PASTE; | |
45b966db | 1268 | } |
15dad1d9 | 1269 | else if (c2 == '%' && PEEKN (1) == ':') |
45b966db | 1270 | { |
15dad1d9 ZW |
1271 | /* Digraph: "%:" == "#". */ |
1272 | FORWARD (1); | |
1273 | CPP_RESERVE (pfile, 2); | |
1274 | CPP_PUTC_Q (pfile, c2); | |
1275 | CPP_PUTC_Q (pfile, GETC ()); | |
1368ee70 | 1276 | return CPP_PASTE; |
45b966db | 1277 | } |
15dad1d9 ZW |
1278 | else |
1279 | return CPP_HASH; | |
45b966db ZW |
1280 | |
1281 | case '\"': | |
1282 | case '\'': | |
1283 | parse_string (pfile, c); | |
45b966db ZW |
1284 | return c == '\'' ? CPP_CHAR : CPP_STRING; |
1285 | ||
1286 | case '$': | |
ae79697b | 1287 | if (!CPP_OPTION (pfile, dollars_in_ident)) |
45b966db ZW |
1288 | goto randomchar; |
1289 | goto letter; | |
1290 | ||
1291 | case ':': | |
5eec0563 JM |
1292 | c2 = PEEKC (); |
1293 | /* Digraph: ":>" == "]". */ | |
1294 | if (c2 == '>' | |
1295 | || (c2 == ':' && CPP_OPTION (pfile, cplusplus))) | |
45b966db ZW |
1296 | goto op2; |
1297 | goto randomchar; | |
1298 | ||
1299 | case '&': | |
1300 | case '+': | |
1301 | case '|': | |
1302 | c2 = PEEKC (); | |
1303 | if (c2 == c || c2 == '=') | |
1304 | goto op2; | |
1305 | goto randomchar; | |
1306 | ||
5eec0563 JM |
1307 | case '%': |
1308 | /* Digraphs: "%:" == "#", "%>" == "}". */ | |
1309 | c2 = PEEKC (); | |
1310 | if (c2 == ':') | |
1311 | { | |
1312 | FORWARD (1); | |
1313 | CPP_RESERVE (pfile, 2); | |
1314 | CPP_PUTC_Q (pfile, c); | |
1315 | CPP_PUTC_Q (pfile, c2); | |
1316 | goto hash; | |
1317 | } | |
1318 | else if (c2 == '>') | |
1319 | { | |
1320 | FORWARD (1); | |
1321 | CPP_RESERVE (pfile, 2); | |
1322 | CPP_PUTC_Q (pfile, c); | |
1323 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1324 | return CPP_OPEN_BRACE; |
5eec0563 JM |
1325 | } |
1326 | /* else fall through */ | |
1327 | ||
45b966db ZW |
1328 | case '*': |
1329 | case '!': | |
45b966db ZW |
1330 | case '=': |
1331 | case '^': | |
1332 | if (PEEKC () == '=') | |
1333 | goto op2; | |
1334 | goto randomchar; | |
1335 | ||
1336 | case '-': | |
1337 | c2 = PEEKC (); | |
1338 | if (c2 == '-') | |
1339 | { | |
ae79697b | 1340 | if (CPP_OPTION (pfile, chill)) |
45b966db ZW |
1341 | goto comment; /* Chill style comment */ |
1342 | else | |
1343 | goto op2; | |
1344 | } | |
1345 | else if (c2 == '=') | |
1346 | goto op2; | |
1347 | else if (c2 == '>') | |
1348 | { | |
ae79697b | 1349 | if (CPP_OPTION (pfile, cplusplus) && PEEKN (1) == '*') |
45b966db ZW |
1350 | { |
1351 | /* In C++, there's a ->* operator. */ | |
1352 | token = CPP_OTHER; | |
45b966db ZW |
1353 | CPP_RESERVE (pfile, 4); |
1354 | CPP_PUTC_Q (pfile, c); | |
1355 | CPP_PUTC_Q (pfile, GETC ()); | |
1356 | CPP_PUTC_Q (pfile, GETC ()); | |
45b966db ZW |
1357 | return token; |
1358 | } | |
1359 | goto op2; | |
1360 | } | |
1361 | goto randomchar; | |
1362 | ||
1363 | case '<': | |
1364 | if (pfile->parsing_include_directive) | |
1365 | { | |
1366 | for (;;) | |
1367 | { | |
1368 | CPP_PUTC (pfile, c); | |
1369 | if (c == '>') | |
1370 | break; | |
1371 | c = GETC (); | |
1372 | if (c == '\n' || c == EOF) | |
1373 | { | |
1374 | cpp_error (pfile, | |
1375 | "missing '>' in `#include <FILENAME>'"); | |
1376 | break; | |
1377 | } | |
1378 | else if (c == '\r') | |
1379 | { | |
1380 | if (!CPP_BUFFER (pfile)->has_escapes) | |
1381 | { | |
1382 | /* Backslash newline is replaced by nothing. */ | |
1383 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1384 | CPP_BUMP_LINE (pfile); | |
1385 | } | |
1386 | else | |
1387 | { | |
1388 | /* We might conceivably get \r- or \r<space> in | |
1389 | here. Just delete 'em. */ | |
1390 | int d = GETC(); | |
1391 | if (d != '-' && d != ' ') | |
1392 | cpp_ice (pfile, "unrecognized escape \\r%c", d); | |
1393 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1394 | } | |
1395 | } | |
1396 | } | |
1397 | return CPP_STRING; | |
1398 | } | |
5eec0563 JM |
1399 | /* Digraphs: "<%" == "{", "<:" == "[". */ |
1400 | c2 = PEEKC (); | |
1401 | if (c2 == '%') | |
1402 | { | |
1403 | FORWARD (1); | |
1404 | CPP_RESERVE (pfile, 2); | |
1405 | CPP_PUTC_Q (pfile, c); | |
1406 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1407 | return CPP_CLOSE_BRACE; |
5eec0563 JM |
1408 | } |
1409 | else if (c2 == ':') | |
1410 | goto op2; | |
45b966db ZW |
1411 | /* else fall through */ |
1412 | case '>': | |
1413 | c2 = PEEKC (); | |
1414 | if (c2 == '=') | |
1415 | goto op2; | |
1416 | /* GNU C++ supports MIN and MAX operators <? and >?. */ | |
ae79697b | 1417 | if (c2 != c && (!CPP_OPTION (pfile, cplusplus) || c2 != '?')) |
45b966db ZW |
1418 | goto randomchar; |
1419 | FORWARD(1); | |
5eec0563 JM |
1420 | CPP_RESERVE (pfile, 3); |
1421 | CPP_PUTC_Q (pfile, c); | |
1422 | CPP_PUTC_Q (pfile, c2); | |
1423 | if (PEEKC () == '=') | |
45b966db | 1424 | CPP_PUTC_Q (pfile, GETC ()); |
45b966db ZW |
1425 | return CPP_OTHER; |
1426 | ||
1427 | case '.': | |
1428 | c2 = PEEKC (); | |
5eec0563 | 1429 | if (ISDIGIT (c2)) |
45b966db | 1430 | { |
5eec0563 | 1431 | CPP_PUTC (pfile, c); |
45b966db ZW |
1432 | c = GETC (); |
1433 | goto number; | |
1434 | } | |
1435 | ||
1436 | /* In C++ there's a .* operator. */ | |
ae79697b | 1437 | if (CPP_OPTION (pfile, cplusplus) && c2 == '*') |
45b966db ZW |
1438 | goto op2; |
1439 | ||
1440 | if (c2 == '.' && PEEKN(1) == '.') | |
1441 | { | |
5eec0563 | 1442 | CPP_RESERVE (pfile, 3); |
45b966db ZW |
1443 | CPP_PUTC_Q (pfile, '.'); |
1444 | CPP_PUTC_Q (pfile, '.'); | |
1445 | CPP_PUTC_Q (pfile, '.'); | |
1446 | FORWARD (2); | |
1368ee70 | 1447 | return CPP_ELLIPSIS; |
45b966db ZW |
1448 | } |
1449 | goto randomchar; | |
1450 | ||
1451 | op2: | |
5eec0563 | 1452 | CPP_RESERVE (pfile, 2); |
45b966db ZW |
1453 | CPP_PUTC_Q (pfile, c); |
1454 | CPP_PUTC_Q (pfile, GETC ()); | |
5eec0563 | 1455 | return CPP_OTHER; |
45b966db ZW |
1456 | |
1457 | case 'L': | |
1458 | c2 = PEEKC (); | |
1459 | if ((c2 == '\'' || c2 == '\"') && !CPP_TRADITIONAL (pfile)) | |
1460 | { | |
1461 | CPP_PUTC (pfile, c); | |
1462 | c = GETC (); | |
1463 | parse_string (pfile, c); | |
45b966db ZW |
1464 | return c == '\'' ? CPP_WCHAR : CPP_WSTRING; |
1465 | } | |
1466 | goto letter; | |
1467 | ||
1468 | case '0': case '1': case '2': case '3': case '4': | |
1469 | case '5': case '6': case '7': case '8': case '9': | |
1470 | number: | |
1471 | c2 = '.'; | |
1472 | for (;;) | |
1473 | { | |
1474 | CPP_RESERVE (pfile, 2); | |
1475 | CPP_PUTC_Q (pfile, c); | |
1476 | c = PEEKC (); | |
1477 | if (c == EOF) | |
1478 | break; | |
1479 | if (!is_numchar(c) && c != '.' | |
1480 | && ((c2 != 'e' && c2 != 'E' | |
1481 | && ((c2 != 'p' && c2 != 'P') | |
ae79697b | 1482 | || CPP_OPTION (pfile, c89))) |
45b966db ZW |
1483 | || (c != '+' && c != '-'))) |
1484 | break; | |
1485 | FORWARD(1); | |
1486 | c2= c; | |
1487 | } | |
45b966db ZW |
1488 | return CPP_NUMBER; |
1489 | case 'b': case 'c': case 'd': case 'h': case 'o': | |
1490 | case 'B': case 'C': case 'D': case 'H': case 'O': | |
ae79697b | 1491 | if (CPP_OPTION (pfile, chill) && PEEKC () == '\'') |
45b966db | 1492 | { |
45b966db ZW |
1493 | CPP_RESERVE (pfile, 2); |
1494 | CPP_PUTC_Q (pfile, c); | |
1495 | CPP_PUTC_Q (pfile, '\''); | |
1496 | FORWARD(1); | |
1497 | for (;;) | |
1498 | { | |
1499 | c = GETC(); | |
1500 | if (c == EOF) | |
1501 | goto chill_number_eof; | |
1502 | if (!is_numchar(c)) | |
1503 | break; | |
1504 | CPP_PUTC (pfile, c); | |
1505 | } | |
1506 | if (c == '\'') | |
1507 | { | |
1508 | CPP_RESERVE (pfile, 2); | |
1509 | CPP_PUTC_Q (pfile, c); | |
45b966db ZW |
1510 | return CPP_STRING; |
1511 | } | |
1512 | else | |
1513 | { | |
1514 | FORWARD(-1); | |
1515 | chill_number_eof: | |
45b966db ZW |
1516 | return CPP_NUMBER; |
1517 | } | |
1518 | } | |
1519 | else | |
1520 | goto letter; | |
1521 | case '_': | |
1522 | case 'a': case 'e': case 'f': case 'g': case 'i': case 'j': | |
1523 | case 'k': case 'l': case 'm': case 'n': case 'p': case 'q': | |
1524 | case 'r': case 's': case 't': case 'u': case 'v': case 'w': | |
1525 | case 'x': case 'y': case 'z': | |
1526 | case 'A': case 'E': case 'F': case 'G': case 'I': case 'J': | |
1527 | case 'K': case 'M': case 'N': case 'P': case 'Q': case 'R': | |
1528 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1529 | case 'Y': case 'Z': | |
1530 | letter: | |
45b966db ZW |
1531 | _cpp_parse_name (pfile, c); |
1532 | return CPP_MACRO; | |
1533 | ||
64aaf407 NB |
1534 | case ' ': case '\t': case '\v': case '\f': case '\0': |
1535 | { | |
1536 | int null_count = 0; | |
1537 | ||
1538 | for (;;) | |
1539 | { | |
1540 | if (c == '\0') | |
1541 | null_count++; | |
1542 | else | |
1543 | CPP_PUTC (pfile, c); | |
1544 | c = PEEKC (); | |
1545 | if (c == EOF || !is_hspace(c)) | |
1546 | break; | |
1547 | FORWARD(1); | |
1548 | } | |
1549 | if (null_count) | |
1550 | null_warning (pfile, null_count); | |
1551 | return CPP_HSPACE; | |
1552 | } | |
45b966db ZW |
1553 | |
1554 | case '\r': | |
1555 | if (CPP_BUFFER (pfile)->has_escapes) | |
1556 | { | |
1557 | c = GETC (); | |
1558 | if (c == '-') | |
1559 | { | |
1560 | if (pfile->output_escapes) | |
1561 | CPP_PUTS (pfile, "\r-", 2); | |
1562 | _cpp_parse_name (pfile, GETC ()); | |
1563 | return CPP_NAME; | |
1564 | } | |
1565 | else if (c == ' ') | |
1566 | { | |
ff2b53ef ZW |
1567 | /* "\r " means a space, but only if necessary to prevent |
1568 | accidental token concatenation. */ | |
45b966db ZW |
1569 | CPP_RESERVE (pfile, 2); |
1570 | if (pfile->output_escapes) | |
1571 | CPP_PUTC_Q (pfile, '\r'); | |
1572 | CPP_PUTC_Q (pfile, c); | |
1573 | return CPP_HSPACE; | |
1574 | } | |
1575 | else | |
1576 | { | |
1577 | cpp_ice (pfile, "unrecognized escape \\r%c", c); | |
1578 | goto get_next; | |
1579 | } | |
1580 | } | |
1581 | else | |
1582 | { | |
1583 | /* Backslash newline is ignored. */ | |
cbccf5e8 MM |
1584 | if (!ACTIVE_MARK_P (pfile)) |
1585 | CPP_BUMP_LINE (pfile); | |
45b966db ZW |
1586 | goto get_next; |
1587 | } | |
1588 | ||
1589 | case '\n': | |
1590 | CPP_PUTC (pfile, c); | |
45b966db ZW |
1591 | return CPP_VSPACE; |
1592 | ||
1368ee70 ZW |
1593 | case '(': token = CPP_OPEN_PAREN; goto char1; |
1594 | case ')': token = CPP_CLOSE_PAREN; goto char1; | |
1595 | case '{': token = CPP_OPEN_BRACE; goto char1; | |
1596 | case '}': token = CPP_CLOSE_BRACE; goto char1; | |
1597 | case ',': token = CPP_COMMA; goto char1; | |
1598 | case ';': token = CPP_SEMICOLON; goto char1; | |
45b966db ZW |
1599 | |
1600 | randomchar: | |
1601 | default: | |
1602 | token = CPP_OTHER; | |
1603 | char1: | |
45b966db ZW |
1604 | CPP_PUTC (pfile, c); |
1605 | return token; | |
1606 | } | |
1607 | } | |
1608 | ||
1609 | /* Check for and expand a macro, which is from WRITTEN to CPP_WRITTEN (pfile). | |
1610 | Caller is expected to have checked no_macro_expand. */ | |
1611 | static int | |
1612 | maybe_macroexpand (pfile, written) | |
1613 | cpp_reader *pfile; | |
1614 | long written; | |
1615 | { | |
1616 | U_CHAR *macro = pfile->token_buffer + written; | |
1617 | size_t len = CPP_WRITTEN (pfile) - written; | |
1618 | HASHNODE *hp = _cpp_lookup (pfile, macro, len); | |
1619 | ||
a7abcbbf ZW |
1620 | /* _cpp_lookup never returns null. */ |
1621 | if (hp->type == T_VOID) | |
45b966db | 1622 | return 0; |
d9e0bd53 | 1623 | if (hp->disabled || hp->type == T_IDENTITY) |
45b966db ZW |
1624 | { |
1625 | if (pfile->output_escapes) | |
1626 | { | |
1627 | /* Insert a no-reexpand marker before IDENT. */ | |
1628 | CPP_RESERVE (pfile, 2); | |
1629 | CPP_ADJUST_WRITTEN (pfile, 2); | |
1630 | macro = pfile->token_buffer + written; | |
1631 | ||
1632 | memmove (macro + 2, macro, len); | |
1633 | macro[0] = '\r'; | |
1634 | macro[1] = '-'; | |
1635 | } | |
1636 | return 0; | |
1637 | } | |
ff2b53ef ZW |
1638 | if (hp->type == T_EMPTY) |
1639 | { | |
1640 | /* Special case optimization: macro expands to nothing. */ | |
1641 | CPP_SET_WRITTEN (pfile, written); | |
1642 | CPP_PUTC_Q (pfile, ' '); | |
1643 | return 1; | |
1644 | } | |
45b966db ZW |
1645 | |
1646 | /* If macro wants an arglist, verify that a '(' follows. */ | |
d9e0bd53 | 1647 | if (hp->type == T_FMACRO) |
45b966db ZW |
1648 | { |
1649 | int macbuf_whitespace = 0; | |
1650 | int c; | |
1651 | ||
1652 | while (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
1653 | { | |
1654 | const U_CHAR *point = CPP_BUFFER (pfile)->cur; | |
1655 | for (;;) | |
1656 | { | |
1657 | _cpp_skip_hspace (pfile); | |
1658 | c = PEEKC (); | |
1659 | if (c == '\n') | |
1660 | FORWARD(1); | |
1661 | else | |
1662 | break; | |
1663 | } | |
1664 | if (point != CPP_BUFFER (pfile)->cur) | |
1665 | macbuf_whitespace = 1; | |
1666 | if (c == '(') | |
1667 | goto is_macro_call; | |
1668 | else if (c != EOF) | |
1669 | goto not_macro_call; | |
1670 | cpp_pop_buffer (pfile); | |
1671 | } | |
1672 | ||
1673 | CPP_SET_MARK (pfile); | |
1674 | for (;;) | |
1675 | { | |
1676 | _cpp_skip_hspace (pfile); | |
1677 | c = PEEKC (); | |
1678 | if (c == '\n') | |
1679 | FORWARD(1); | |
1680 | else | |
1681 | break; | |
1682 | } | |
1683 | CPP_GOTO_MARK (pfile); | |
1684 | ||
1685 | if (c != '(') | |
1686 | { | |
1687 | not_macro_call: | |
1688 | if (macbuf_whitespace) | |
1689 | CPP_PUTC (pfile, ' '); | |
476f2869 ZW |
1690 | |
1691 | /* K+R treated this as a hard error. */ | |
1692 | if (CPP_OPTION (pfile, warn_traditional)) | |
1693 | cpp_warning (pfile, | |
1694 | "traditional C rejects function macro %s in non-function context", | |
1695 | hp->name); | |
45b966db ZW |
1696 | return 0; |
1697 | } | |
1698 | } | |
1699 | ||
1700 | is_macro_call: | |
1701 | /* This is now known to be a macro call. | |
1702 | Expand the macro, reading arguments as needed, | |
1703 | and push the expansion on the input stack. */ | |
1704 | _cpp_macroexpand (pfile, hp); | |
1705 | CPP_SET_WRITTEN (pfile, written); | |
1706 | return 1; | |
1707 | } | |
1708 | ||
9e62c811 ZW |
1709 | /* Complain about \v or \f in a preprocessing directive (constraint |
1710 | violation, C99 6.10 para 5). Caller has checked CPP_PEDANTIC. */ | |
1711 | static void | |
1712 | pedantic_whitespace (pfile, p, len) | |
1713 | cpp_reader *pfile; | |
1714 | U_CHAR *p; | |
1715 | unsigned int len; | |
1716 | { | |
1717 | while (len) | |
1718 | { | |
1719 | if (*p == '\v') | |
1720 | cpp_pedwarn (pfile, "vertical tab in preprocessing directive"); | |
1721 | else if (*p == '\f') | |
1722 | cpp_pedwarn (pfile, "form feed in preprocessing directive"); | |
1723 | p++; | |
1724 | len--; | |
1725 | } | |
1726 | } | |
1727 | ||
1728 | ||
3a2b2c7a | 1729 | enum cpp_ttype |
45b966db ZW |
1730 | cpp_get_token (pfile) |
1731 | cpp_reader *pfile; | |
1732 | { | |
3a2b2c7a | 1733 | enum cpp_ttype token; |
45b966db ZW |
1734 | long written = CPP_WRITTEN (pfile); |
1735 | ||
1736 | get_next: | |
1737 | token = _cpp_lex_token (pfile); | |
1738 | ||
1739 | switch (token) | |
1740 | { | |
1741 | default: | |
ff2b53ef ZW |
1742 | pfile->potential_control_macro = 0; |
1743 | pfile->only_seen_white = 0; | |
1744 | return token; | |
1745 | ||
1746 | case CPP_VSPACE: | |
1747 | if (pfile->only_seen_white == 0) | |
1748 | pfile->only_seen_white = 1; | |
1749 | CPP_BUMP_LINE (pfile); | |
ff2b53ef ZW |
1750 | return token; |
1751 | ||
1752 | case CPP_HSPACE: | |
1753 | case CPP_COMMENT: | |
45b966db ZW |
1754 | return token; |
1755 | ||
15dad1d9 | 1756 | case CPP_HASH: |
ff2b53ef | 1757 | pfile->potential_control_macro = 0; |
15dad1d9 ZW |
1758 | if (!pfile->only_seen_white) |
1759 | return CPP_HASH; | |
1760 | /* XXX shouldn't have to do this - remove the hash or %: from | |
1761 | the token buffer. */ | |
1762 | if (CPP_PWRITTEN (pfile)[-1] == '#') | |
1763 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1764 | else | |
1765 | CPP_ADJUST_WRITTEN (pfile, -2); | |
1766 | ||
45b966db | 1767 | if (_cpp_handle_directive (pfile)) |
15dad1d9 | 1768 | return CPP_DIRECTIVE; |
45b966db ZW |
1769 | pfile->only_seen_white = 0; |
1770 | CPP_PUTC (pfile, '#'); | |
15dad1d9 | 1771 | return CPP_HASH; |
45b966db ZW |
1772 | |
1773 | case CPP_MACRO: | |
ff2b53ef ZW |
1774 | pfile->potential_control_macro = 0; |
1775 | pfile->only_seen_white = 0; | |
45b966db ZW |
1776 | if (! pfile->no_macro_expand |
1777 | && maybe_macroexpand (pfile, written)) | |
1778 | goto get_next; | |
1779 | return CPP_NAME; | |
1780 | ||
1781 | case CPP_EOF: | |
f2d5f0cc ZW |
1782 | if (CPP_BUFFER (pfile) == NULL) |
1783 | return CPP_EOF; | |
c56c2073 | 1784 | if (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) |
f2d5f0cc | 1785 | { |
45b966db ZW |
1786 | cpp_pop_buffer (pfile); |
1787 | goto get_next; | |
1788 | } | |
c56c2073 ZW |
1789 | cpp_pop_buffer (pfile); |
1790 | return CPP_EOF; | |
45b966db ZW |
1791 | } |
1792 | } | |
1793 | ||
1794 | /* Like cpp_get_token, but skip spaces and comments. */ | |
1795 | ||
3a2b2c7a | 1796 | enum cpp_ttype |
45b966db ZW |
1797 | cpp_get_non_space_token (pfile) |
1798 | cpp_reader *pfile; | |
1799 | { | |
1800 | int old_written = CPP_WRITTEN (pfile); | |
1801 | for (;;) | |
1802 | { | |
3a2b2c7a | 1803 | enum cpp_ttype token = cpp_get_token (pfile); |
ff2b53ef | 1804 | if (token != CPP_COMMENT && token != CPP_HSPACE && token != CPP_VSPACE) |
45b966db ZW |
1805 | return token; |
1806 | CPP_SET_WRITTEN (pfile, old_written); | |
1807 | } | |
1808 | } | |
1809 | ||
ff2b53ef | 1810 | /* Like cpp_get_token, except that it does not execute directives, |
c56c2073 | 1811 | does not consume vertical space, and discards horizontal space. */ |
3a2b2c7a | 1812 | enum cpp_ttype |
9e62c811 | 1813 | _cpp_get_directive_token (pfile) |
45b966db ZW |
1814 | cpp_reader *pfile; |
1815 | { | |
ff2b53ef | 1816 | long old_written; |
3a2b2c7a | 1817 | enum cpp_ttype token; |
57c578a6 | 1818 | int at_bol; |
45b966db | 1819 | |
ff2b53ef | 1820 | get_next: |
57c578a6 | 1821 | at_bol = (CPP_BUFFER (pfile)->cur == CPP_BUFFER (pfile)->line_base); |
ff2b53ef ZW |
1822 | old_written = CPP_WRITTEN (pfile); |
1823 | token = _cpp_lex_token (pfile); | |
1824 | switch (token) | |
45b966db | 1825 | { |
ff2b53ef ZW |
1826 | default: |
1827 | return token; | |
45b966db | 1828 | |
ff2b53ef ZW |
1829 | case CPP_VSPACE: |
1830 | /* Put it back and return VSPACE. */ | |
1831 | FORWARD(-1); | |
1832 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1833 | return CPP_VSPACE; | |
45b966db | 1834 | |
ff2b53ef | 1835 | case CPP_HSPACE: |
57c578a6 ZW |
1836 | /* The purpose of this rather strange check is to prevent pedantic |
1837 | warnings for ^L in an #ifdefed out block. */ | |
1838 | if (CPP_PEDANTIC (pfile) && ! at_bol) | |
9e62c811 ZW |
1839 | pedantic_whitespace (pfile, pfile->token_buffer + old_written, |
1840 | CPP_WRITTEN (pfile) - old_written); | |
1841 | CPP_SET_WRITTEN (pfile, old_written); | |
1842 | goto get_next; | |
ff2b53ef | 1843 | return CPP_HSPACE; |
45b966db | 1844 | |
ff2b53ef ZW |
1845 | case CPP_MACRO: |
1846 | if (! pfile->no_macro_expand | |
1847 | && maybe_macroexpand (pfile, old_written)) | |
1848 | goto get_next; | |
1849 | return CPP_NAME; | |
45b966db | 1850 | |
ff2b53ef ZW |
1851 | case CPP_EOF: |
1852 | if (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
45b966db | 1853 | { |
ff2b53ef ZW |
1854 | cpp_pop_buffer (pfile); |
1855 | goto get_next; | |
45b966db | 1856 | } |
ff2b53ef ZW |
1857 | else |
1858 | /* This can happen for files that don't end with a newline, | |
1859 | and for cpp_define and friends. Pretend they do, so | |
1860 | callers don't have to deal. A warning will be issued by | |
1861 | someone else, if necessary. */ | |
1862 | return CPP_VSPACE; | |
1863 | } | |
1864 | } | |
1865 | ||
45b966db ZW |
1866 | /* Determine the current line and column. Used only by read_and_prescan. */ |
1867 | static U_CHAR * | |
1868 | find_position (start, limit, linep) | |
1869 | U_CHAR *start; | |
1870 | U_CHAR *limit; | |
1871 | unsigned long *linep; | |
1872 | { | |
1873 | unsigned long line = *linep; | |
1874 | U_CHAR *lbase = start; | |
1875 | while (start < limit) | |
1876 | { | |
1877 | U_CHAR ch = *start++; | |
1878 | if (ch == '\n' || ch == '\r') | |
1879 | { | |
1880 | line++; | |
1881 | lbase = start; | |
1882 | } | |
1883 | } | |
1884 | *linep = line; | |
1885 | return lbase; | |
1886 | } | |
1887 | ||
2a87fbe8 ZW |
1888 | /* The following table is used by _cpp_read_and_prescan. If we have |
1889 | designated initializers, it can be constant data; otherwise, it is | |
1890 | set up at runtime by _cpp_init_input_buffer. */ | |
46d07497 ZW |
1891 | |
1892 | #ifndef UCHAR_MAX | |
1893 | #define UCHAR_MAX 255 /* assume 8-bit bytes */ | |
1894 | #endif | |
1895 | ||
12cf91fe | 1896 | #if (GCC_VERSION >= 2007) |
2a87fbe8 | 1897 | #define init_chartab() /* nothing */ |
12cf91fe | 1898 | #define CHARTAB __extension__ static const U_CHAR chartab[UCHAR_MAX + 1] = { |
46d07497 ZW |
1899 | #define END }; |
1900 | #define s(p, v) [p] = v, | |
1901 | #else | |
12cf91fe | 1902 | #define CHARTAB static U_CHAR chartab[UCHAR_MAX + 1] = { 0 }; \ |
2a87fbe8 ZW |
1903 | static void init_chartab PARAMS ((void)) { \ |
1904 | unsigned char *x = chartab; | |
46d07497 ZW |
1905 | #define END } |
1906 | #define s(p, v) x[p] = v; | |
1907 | #endif | |
1908 | ||
1909 | /* Table of characters that can't be handled in the inner loop. | |
2a87fbe8 ZW |
1910 | Also contains the mapping between trigraph third characters and their |
1911 | replacements. */ | |
46d07497 ZW |
1912 | #define SPECCASE_CR 1 |
1913 | #define SPECCASE_BACKSLASH 2 | |
1914 | #define SPECCASE_QUESTION 3 | |
1915 | ||
2a87fbe8 | 1916 | CHARTAB |
46d07497 ZW |
1917 | s('\r', SPECCASE_CR) |
1918 | s('\\', SPECCASE_BACKSLASH) | |
1919 | s('?', SPECCASE_QUESTION) | |
46d07497 | 1920 | |
46d07497 ZW |
1921 | s('=', '#') s(')', ']') s('!', '|') |
1922 | s('(', '[') s('\'', '^') s('>', '}') | |
1923 | s('/', '\\') s('<', '{') s('-', '~') | |
1924 | END | |
1925 | ||
1926 | #undef CHARTAB | |
46d07497 ZW |
1927 | #undef END |
1928 | #undef s | |
1929 | ||
2a87fbe8 ZW |
1930 | #define NORMAL(c) ((chartab[c]) == 0 || (chartab[c]) > SPECCASE_QUESTION) |
1931 | #define NONTRI(c) ((c) <= SPECCASE_QUESTION) | |
1932 | ||
45b966db ZW |
1933 | /* Read the entire contents of file DESC into buffer BUF. LEN is how |
1934 | much memory to allocate initially; more will be allocated if | |
1935 | necessary. Convert end-of-line markers (\n, \r, \r\n, \n\r) to | |
1936 | canonical form (\n). If enabled, convert and/or warn about | |
1937 | trigraphs. Convert backslash-newline to a one-character escape | |
1938 | (\r) and remove it from "embarrassing" places (i.e. the middle of a | |
1939 | token). If there is no newline at the end of the file, add one and | |
1940 | warn. Returns -1 on failure, or the actual length of the data to | |
1941 | be scanned. | |
1942 | ||
1943 | This function does a lot of work, and can be a serious performance | |
1944 | bottleneck. It has been tuned heavily; make sure you understand it | |
1945 | before hacking. The common case - no trigraphs, Unix style line | |
1946 | breaks, backslash-newline set off by whitespace, newline at EOF - | |
1947 | has been optimized at the expense of the others. The performance | |
1948 | penalty for DOS style line breaks (\r\n) is about 15%. | |
1949 | ||
1950 | Warnings lose particularly heavily since we have to determine the | |
1951 | line number, which involves scanning from the beginning of the file | |
1952 | or from the last warning. The penalty for the absence of a newline | |
1953 | at the end of reload1.c is about 60%. (reload1.c is 329k.) | |
1954 | ||
1955 | If your file has more than one kind of end-of-line marker, you | |
04e3ec78 NB |
1956 | will get messed-up line numbering. |
1957 | ||
1958 | So that the cases of the switch statement do not have to concern | |
1959 | themselves with the complications of reading beyond the end of the | |
1960 | buffer, the buffer is guaranteed to have at least 3 characters in | |
1961 | it (or however many are left in the file, if less) on entry to the | |
1962 | switch. This is enough to handle trigraphs and the "\\\n\r" and | |
1963 | "\\\r\n" cases. | |
1964 | ||
1965 | The end of the buffer is marked by a '\\', which, being a special | |
1966 | character, guarantees we will exit the fast-scan loops and perform | |
1967 | a refill. */ | |
46d07497 | 1968 | |
45b966db ZW |
1969 | long |
1970 | _cpp_read_and_prescan (pfile, fp, desc, len) | |
1971 | cpp_reader *pfile; | |
1972 | cpp_buffer *fp; | |
1973 | int desc; | |
1974 | size_t len; | |
1975 | { | |
1976 | U_CHAR *buf = (U_CHAR *) xmalloc (len); | |
1977 | U_CHAR *ip, *op, *line_base; | |
1978 | U_CHAR *ibase; | |
45b966db ZW |
1979 | unsigned long line; |
1980 | unsigned int deferred_newlines; | |
45b966db | 1981 | size_t offset; |
04e3ec78 | 1982 | int count = 0; |
45b966db ZW |
1983 | |
1984 | offset = 0; | |
04e3ec78 | 1985 | deferred_newlines = 0; |
45b966db ZW |
1986 | op = buf; |
1987 | line_base = buf; | |
1988 | line = 1; | |
04e3ec78 NB |
1989 | ibase = pfile->input_buffer + 3; |
1990 | ip = ibase; | |
1991 | ip[-1] = '\0'; /* Guarantee no match with \n for SPECCASE_CR */ | |
45b966db ZW |
1992 | |
1993 | for (;;) | |
1994 | { | |
04e3ec78 NB |
1995 | U_CHAR *near_buff_end; |
1996 | ||
04e3ec78 | 1997 | count = read (desc, ibase, pfile->input_buffer_len); |
45b966db ZW |
1998 | if (count < 0) |
1999 | goto error; | |
04e3ec78 NB |
2000 | |
2001 | ibase[count] = '\\'; /* Marks end of buffer */ | |
2002 | if (count) | |
45b966db | 2003 | { |
04e3ec78 NB |
2004 | near_buff_end = pfile->input_buffer + count; |
2005 | offset += count; | |
45b966db | 2006 | if (offset > len) |
04e3ec78 NB |
2007 | { |
2008 | size_t delta_op; | |
2009 | size_t delta_line_base; | |
1b955cba | 2010 | len = offset * 2; |
04e3ec78 NB |
2011 | if (offset > len) |
2012 | /* len overflowed. | |
2013 | This could happen if the file is larger than half the | |
2014 | maximum address space of the machine. */ | |
2015 | goto too_big; | |
2016 | ||
2017 | delta_op = op - buf; | |
2018 | delta_line_base = line_base - buf; | |
2019 | buf = (U_CHAR *) xrealloc (buf, len); | |
2020 | op = buf + delta_op; | |
2021 | line_base = buf + delta_line_base; | |
2022 | } | |
2023 | } | |
2024 | else | |
2025 | { | |
2026 | if (ip == ibase) | |
2027 | break; | |
2028 | /* Allow normal processing of the (at most 2) remaining | |
2029 | characters. The end-of-buffer marker is still present | |
2030 | and prevents false matches within the switch. */ | |
2031 | near_buff_end = ibase - 1; | |
45b966db ZW |
2032 | } |
2033 | ||
2034 | for (;;) | |
2035 | { | |
04e3ec78 | 2036 | unsigned int span; |
45b966db | 2037 | |
04e3ec78 | 2038 | /* Deal with \-newline, potentially in the middle of a token. */ |
45b966db ZW |
2039 | if (deferred_newlines) |
2040 | { | |
2a87fbe8 | 2041 | if (op != buf && ! is_space (op[-1]) && op[-1] != '\r') |
04e3ec78 NB |
2042 | { |
2043 | /* Previous was not white space. Skip to white | |
2044 | space, if we can, before outputting the \r's */ | |
2045 | span = 0; | |
2046 | while (ip[span] != ' ' | |
2047 | && ip[span] != '\t' | |
2048 | && ip[span] != '\n' | |
2a87fbe8 | 2049 | && NORMAL(ip[span])) |
04e3ec78 NB |
2050 | span++; |
2051 | memcpy (op, ip, span); | |
2052 | op += span; | |
2053 | ip += span; | |
2a87fbe8 | 2054 | if (! NORMAL(ip[0])) |
04e3ec78 NB |
2055 | goto do_speccase; |
2056 | } | |
2057 | while (deferred_newlines) | |
2058 | deferred_newlines--, *op++ = '\r'; | |
45b966db ZW |
2059 | } |
2060 | ||
2061 | /* Copy as much as we can without special treatment. */ | |
04e3ec78 | 2062 | span = 0; |
2a87fbe8 | 2063 | while (NORMAL (ip[span])) span++; |
45b966db ZW |
2064 | memcpy (op, ip, span); |
2065 | op += span; | |
2066 | ip += span; | |
2067 | ||
04e3ec78 NB |
2068 | do_speccase: |
2069 | if (ip > near_buff_end) /* Do we have enough chars? */ | |
2070 | break; | |
2a87fbe8 | 2071 | switch (chartab[*ip++]) |
45b966db | 2072 | { |
45b966db | 2073 | case SPECCASE_CR: /* \r */ |
04e3ec78 | 2074 | if (ip[-2] != '\n') |
45b966db | 2075 | { |
04e3ec78 NB |
2076 | if (*ip == '\n') |
2077 | ip++; | |
2078 | *op++ = '\n'; | |
45b966db | 2079 | } |
45b966db ZW |
2080 | break; |
2081 | ||
2082 | case SPECCASE_BACKSLASH: /* \ */ | |
04e3ec78 | 2083 | if (*ip == '\n') |
45b966db | 2084 | { |
04e3ec78 | 2085 | deferred_newlines++; |
45b966db ZW |
2086 | ip++; |
2087 | if (*ip == '\r') ip++; | |
45b966db ZW |
2088 | } |
2089 | else if (*ip == '\r') | |
2090 | { | |
04e3ec78 | 2091 | deferred_newlines++; |
45b966db ZW |
2092 | ip++; |
2093 | if (*ip == '\n') ip++; | |
45b966db ZW |
2094 | } |
2095 | else | |
2096 | *op++ = '\\'; | |
04e3ec78 | 2097 | break; |
45b966db ZW |
2098 | |
2099 | case SPECCASE_QUESTION: /* ? */ | |
2100 | { | |
2101 | unsigned int d, t; | |
04e3ec78 NB |
2102 | |
2103 | *op++ = '?'; /* Normal non-trigraph case */ | |
2104 | if (ip[0] != '?') | |
2105 | break; | |
2106 | ||
45b966db | 2107 | d = ip[1]; |
2a87fbe8 ZW |
2108 | t = chartab[d]; |
2109 | if (NONTRI (t)) | |
04e3ec78 | 2110 | break; |
45b966db | 2111 | |
ae79697b | 2112 | if (CPP_OPTION (pfile, warn_trigraphs)) |
45b966db ZW |
2113 | { |
2114 | unsigned long col; | |
2115 | line_base = find_position (line_base, op, &line); | |
2116 | col = op - line_base + 1; | |
ae79697b | 2117 | if (CPP_OPTION (pfile, trigraphs)) |
45b966db | 2118 | cpp_warning_with_line (pfile, line, col, |
04e3ec78 | 2119 | "trigraph ??%c converted to %c", d, t); |
45b966db ZW |
2120 | else |
2121 | cpp_warning_with_line (pfile, line, col, | |
04e3ec78 | 2122 | "trigraph ??%c ignored", d); |
45b966db | 2123 | } |
04e3ec78 NB |
2124 | |
2125 | ip += 2; | |
ae79697b | 2126 | if (CPP_OPTION (pfile, trigraphs)) |
45b966db | 2127 | { |
04e3ec78 | 2128 | op[-1] = t; /* Overwrite '?' */ |
45b966db | 2129 | if (t == '\\') |
04e3ec78 NB |
2130 | { |
2131 | op--; | |
2132 | *--ip = '\\'; | |
2133 | goto do_speccase; /* May need buffer refill */ | |
2134 | } | |
45b966db ZW |
2135 | } |
2136 | else | |
2137 | { | |
45b966db ZW |
2138 | *op++ = '?'; |
2139 | *op++ = d; | |
2140 | } | |
2141 | } | |
04e3ec78 | 2142 | break; |
45b966db ZW |
2143 | } |
2144 | } | |
f6fab919 ZW |
2145 | /* Copy previous char plus unprocessed (at most 2) chars |
2146 | to beginning of buffer, refill it with another | |
2147 | read(), and continue processing */ | |
2148 | memmove (ip - count - 1, ip - 1, 4 - (ip - near_buff_end)); | |
2149 | ip -= count; | |
45b966db ZW |
2150 | } |
2151 | ||
2152 | if (offset == 0) | |
2153 | return 0; | |
2154 | ||
45b966db ZW |
2155 | if (op[-1] != '\n') |
2156 | { | |
2157 | unsigned long col; | |
2158 | line_base = find_position (line_base, op, &line); | |
2159 | col = op - line_base + 1; | |
f6fab919 | 2160 | cpp_warning_with_line (pfile, line, col, "no newline at end of file"); |
45b966db ZW |
2161 | if (offset + 1 > len) |
2162 | { | |
2163 | len += 1; | |
2164 | if (offset + 1 > len) | |
2165 | goto too_big; | |
2166 | buf = (U_CHAR *) xrealloc (buf, len); | |
2167 | op = buf + offset; | |
2168 | } | |
2169 | *op++ = '\n'; | |
2170 | } | |
2171 | ||
2172 | fp->buf = ((len - offset < 20) ? buf : (U_CHAR *)xrealloc (buf, op - buf)); | |
2173 | return op - buf; | |
2174 | ||
2175 | too_big: | |
f6fab919 ZW |
2176 | cpp_notice (pfile, "%s is too large (>%lu bytes)", fp->ihash->name, |
2177 | (unsigned long)offset); | |
45b966db ZW |
2178 | free (buf); |
2179 | return -1; | |
2180 | ||
2181 | error: | |
2182 | cpp_error_from_errno (pfile, fp->ihash->name); | |
2183 | free (buf); | |
2184 | return -1; | |
2185 | } | |
2186 | ||
2a87fbe8 ZW |
2187 | /* Allocate pfile->input_buffer, and initialize chartab[] |
2188 | if it hasn't happened already. */ | |
46d07497 | 2189 | |
45b966db ZW |
2190 | void |
2191 | _cpp_init_input_buffer (pfile) | |
2192 | cpp_reader *pfile; | |
2193 | { | |
2194 | U_CHAR *tmp; | |
2195 | ||
2a87fbe8 | 2196 | init_chartab (); |
15dad1d9 | 2197 | _cpp_init_toklist (&pfile->directbuf); |
04e3ec78 | 2198 | |
45b966db ZW |
2199 | /* Determine the appropriate size for the input buffer. Normal C |
2200 | source files are smaller than eight K. */ | |
04e3ec78 NB |
2201 | /* 8Kbytes of buffer proper, 1 to detect running off the end without |
2202 | address arithmetic all the time, and 3 for pushback during buffer | |
2203 | refill, in case there's a potential trigraph or end-of-line | |
2204 | digraph at the end of a block. */ | |
45b966db | 2205 | |
04e3ec78 | 2206 | tmp = (U_CHAR *) xmalloc (8192 + 1 + 3); |
45b966db ZW |
2207 | pfile->input_buffer = tmp; |
2208 | pfile->input_buffer_len = 8192; | |
2209 | } | |
c5a04734 | 2210 | |
6d2c2047 ZW |
2211 | /* Utility routine: |
2212 | Compares, in the manner of strcmp(3), the token beginning at TOKEN | |
2213 | and extending for LEN characters to the NUL-terminated string | |
2214 | STRING. Typical usage: | |
2215 | ||
2216 | if (! cpp_idcmp (pfile->token_buffer + here, CPP_WRITTEN (pfile) - here, | |
2217 | "inline")) | |
2218 | { ... } | |
2219 | */ | |
2220 | ||
2221 | int | |
2222 | cpp_idcmp (token, len, string) | |
2223 | const U_CHAR *token; | |
2224 | size_t len; | |
2225 | const char *string; | |
2226 | { | |
2227 | size_t len2 = strlen (string); | |
2228 | int r; | |
2229 | ||
2230 | if ((r = memcmp (token, string, MIN (len, len2)))) | |
2231 | return r; | |
2232 | ||
2233 | /* The longer of the two strings sorts after the shorter. */ | |
2234 | if (len == len2) | |
2235 | return 0; | |
2236 | else if (len < len2) | |
2237 | return -1; | |
2238 | else | |
2239 | return 1; | |
2240 | } | |
2241 | ||
b8f41010 | 2242 | #ifdef NEW_LEXER |
c5a04734 | 2243 | |
d6d5f795 NB |
2244 | /* Lexing algorithm. |
2245 | ||
2246 | The original lexer in cpplib was made up of two passes: a first pass | |
2247 | that replaced trigraphs and deleted esacped newlines, and a second | |
2248 | pass that tokenized the result of the first pass. Tokenisation was | |
2249 | performed by peeking at the next character in the input stream. For | |
6777db6d | 2250 | example, if the input stream contained "!=", the handler for the ! |
d6d5f795 | 2251 | character would peek at the next character, and if it were a '=' |
6777db6d NB |
2252 | would skip over it, and return a "!=" token, otherwise it would |
2253 | return just the "!" token. | |
d6d5f795 NB |
2254 | |
2255 | To implement a single-pass lexer, this peeking ahead is unworkable. | |
2256 | An arbitrary number of escaped newlines, and trigraphs (in particular | |
6777db6d NB |
2257 | ??/ which translates to the escape \), could separate the '!' and '=' |
2258 | in the input stream, yet the next token is still a "!=". | |
d6d5f795 NB |
2259 | |
2260 | Suppose instead that we lex by one logical line at a time, producing | |
6777db6d NB |
2261 | a token list or stack for each logical line, and when seeing the '!' |
2262 | push a CPP_NOT token on the list. Then if the '!' is part of a | |
2263 | longer token ("!=") we know we must see the remainder of the token by | |
2264 | the time we reach the end of the logical line. Thus we can have the | |
2265 | '=' handler look at the previous token (at the end of the list / top | |
2266 | of the stack) and see if it is a "!" token, and if so, instead of | |
2267 | pushing a "=" token revise the existing token to be a "!=" token. | |
d6d5f795 NB |
2268 | |
2269 | This works in the presence of escaped newlines, because the '\' would | |
2270 | have been pushed on the top of the stack as a CPP_BACKSLASH. The | |
2271 | newline ('\n' or '\r') handler looks at the token at the top of the | |
2272 | stack to see if it is a CPP_BACKSLASH, and if so discards both. | |
2273 | Otherwise it pushes the newline (CPP_VSPACE) token as normal. Hence | |
2274 | the '=' handler would never see any intervening escaped newlines. | |
2275 | ||
2276 | To make trigraphs work in this context, as in precedence trigraphs | |
2277 | are highest and converted before anything else, the '?' handler does | |
2278 | lookahead to see if it is a trigraph, and if so skips the trigraph | |
2279 | and pushes the token it represents onto the top of the stack. This | |
2280 | also works in the particular case of a CPP_BACKSLASH trigraph. | |
2281 | ||
2282 | To the preprocessor, whitespace is only significant to the point of | |
2283 | knowing whether whitespace precedes a particular token. For example, | |
2284 | the '=' handler needs to know whether there was whitespace between it | |
6777db6d | 2285 | and a "!" token on the top of the stack, to make the token conversion |
d6d5f795 NB |
2286 | decision correctly. So each token has a PREV_WHITESPACE flag to |
2287 | indicate this - the standard permits consecutive whitespace to be | |
2288 | regarded as a single space. The compiler front ends are not | |
2289 | interested in whitespace at all; they just require a token stream. | |
2290 | Another place where whitespace is significant to the preprocessor is | |
2291 | a #define statment - if there is whitespace between the macro name | |
2292 | and an initial "(" token the macro is "object-like", otherwise it is | |
2293 | a function-like macro that takes arguments. | |
2294 | ||
2295 | However, all is not rosy. Parsing of identifiers, numbers, comments | |
2296 | and strings becomes trickier because of the possibility of raw | |
2297 | trigraphs and escaped newlines in the input stream. | |
2298 | ||
2299 | The trigraphs are three consecutive characters beginning with two | |
c2e25d51 NB |
2300 | question marks. A question mark is not valid as part of a number or |
2301 | identifier, so parsing of a number or identifier terminates normally | |
2302 | upon reaching it, returning to the mainloop which handles the | |
2303 | trigraph just like it would in any other position. Similarly for the | |
2304 | backslash of a backslash-newline combination. So we just need the | |
2305 | escaped-newline dropper in the mainloop to check if the token on the | |
2306 | top of the stack after dropping the escaped newline is a number or | |
2307 | identifier, and if so to continue the processing it as if nothing had | |
2308 | happened. | |
d6d5f795 NB |
2309 | |
2310 | For strings, we replace trigraphs whenever we reach a quote or | |
2311 | newline, because there might be a backslash trigraph escaping them. | |
2312 | We need to be careful that we start trigraph replacing from where we | |
2313 | left off previously, because it is possible for a first scan to leave | |
2314 | "fake" trigraphs that a second scan would pick up as real (e.g. the | |
c2e25d51 | 2315 | sequence "????/\n=" would find a fake ??= trigraph after removing the |
d6d5f795 NB |
2316 | escaped newline.) |
2317 | ||
2318 | For line comments, on reaching a newline we scan the previous | |
2319 | character(s) to see if it escaped, and continue if it is. Block | |
2320 | comments ignore everything and just focus on finding the comment | |
2321 | termination mark. The only difficult thing, and it is surprisingly | |
2322 | tricky, is checking if an asterisk precedes the final slash since | |
2323 | they could be separated by escaped newlines. If the preprocessor is | |
2324 | invoked with the output comments option, we don't bother removing | |
2325 | escaped newlines and replacing trigraphs for output. | |
2326 | ||
2327 | Finally, numbers can begin with a period, which is pushed initially | |
2328 | as a CPP_DOT token in its own right. The digit handler checks if the | |
2329 | previous token was a CPP_DOT not separated by whitespace, and if so | |
2330 | pops it off the stack and pushes a period into the number's buffer | |
2331 | before calling the number parser. | |
2332 | ||
2333 | */ | |
2334 | ||
b8f41010 NB |
2335 | static const unsigned char *digraph_spellings [] = {U"%:", U"%:%:", U"<:", |
2336 | U":>", U"<%", U"%>"}; | |
2337 | static unsigned char trigraph_map[256]; | |
c5a04734 ZW |
2338 | |
2339 | static void | |
2340 | expand_comment_space (list) | |
2341 | cpp_toklist *list; | |
2342 | { | |
2343 | if (list->comments_cap == 0) | |
2344 | { | |
2345 | list->comments_cap = 10; | |
2346 | list->comments = (cpp_token *) | |
2347 | xmalloc (list->comments_cap * sizeof (cpp_token)); | |
2348 | } | |
2349 | else | |
2350 | { | |
2351 | list->comments_cap *= 2; | |
2352 | list->comments = (cpp_token *) | |
2353 | xrealloc (list->comments, list->comments_cap); | |
2354 | } | |
2355 | } | |
2356 | ||
c5a04734 ZW |
2357 | void |
2358 | init_trigraph_map () | |
2359 | { | |
2360 | trigraph_map['='] = '#'; | |
2361 | trigraph_map['('] = '['; | |
2362 | trigraph_map[')'] = ']'; | |
2363 | trigraph_map['/'] = '\\'; | |
2364 | trigraph_map['\''] = '^'; | |
2365 | trigraph_map['<'] = '{'; | |
2366 | trigraph_map['>'] = '}'; | |
2367 | trigraph_map['!'] = '|'; | |
2368 | trigraph_map['-'] = '~'; | |
2369 | } | |
2370 | ||
2371 | /* Call when a trigraph is encountered. It warns if necessary, and | |
2372 | returns true if the trigraph should be honoured. END is the third | |
2373 | character of a trigraph in the input stream. */ | |
2374 | static int | |
2375 | trigraph_ok (pfile, end) | |
2376 | cpp_reader *pfile; | |
2377 | const unsigned char *end; | |
2378 | { | |
2379 | int accept = CPP_OPTION (pfile, trigraphs); | |
2380 | ||
2381 | if (CPP_OPTION (pfile, warn_trigraphs)) | |
2382 | { | |
2383 | unsigned int col = end - 1 - pfile->buffer->line_base; | |
2384 | if (accept) | |
2385 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2386 | "trigraph ??%c converted to %c", | |
2387 | (int) *end, (int) trigraph_map[*end]); | |
2388 | else | |
2389 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2390 | "trigraph ??%c ignored", (int) *end); | |
2391 | } | |
2392 | return accept; | |
2393 | } | |
2394 | ||
2395 | /* Scan a string for trigraphs, warning or replacing them inline as | |
2396 | appropriate. When parsing a string, we must call this routine | |
2397 | before processing a newline character (if trigraphs are enabled), | |
2398 | since the newline might be escaped by a preceding backslash | |
2399 | trigraph sequence. Returns a pointer to the end of the name after | |
2400 | replacement. */ | |
2401 | ||
2402 | static unsigned char* | |
2403 | trigraph_replace (pfile, src, limit) | |
2404 | cpp_reader *pfile; | |
2405 | unsigned char *src; | |
2406 | unsigned char* limit; | |
2407 | { | |
2408 | unsigned char *dest; | |
2409 | ||
2410 | /* Starting with src[1], find two consecutive '?'. The case of no | |
2411 | trigraphs is streamlined. */ | |
2412 | ||
2413 | for (; src + 1 < limit; src += 2) | |
2414 | { | |
2415 | if (src[0] != '?') | |
2416 | continue; | |
2417 | ||
2418 | /* Make src point to the 1st (NOT 2nd) of two consecutive '?'s. */ | |
2419 | if (src[-1] == '?') | |
2420 | src--; | |
2421 | else if (src + 2 == limit || src[1] != '?') | |
2422 | continue; | |
2423 | ||
2424 | /* Check if it really is a trigraph. */ | |
2425 | if (trigraph_map[src[2]] == 0) | |
2426 | continue; | |
2427 | ||
2428 | dest = src; | |
2429 | goto trigraph_found; | |
2430 | } | |
2431 | return limit; | |
2432 | ||
2433 | /* Now we have a trigraph, we need to scan the remaining buffer, and | |
2434 | copy-shifting its contents left if replacement is enabled. */ | |
2435 | for (; src + 2 < limit; dest++, src++) | |
2436 | if ((*dest = *src) == '?' && src[1] == '?' && trigraph_map[src[2]]) | |
2437 | { | |
2438 | trigraph_found: | |
2439 | src += 2; | |
2440 | if (trigraph_ok (pfile, pfile->buffer->cur - (limit - src))) | |
2441 | *dest = trigraph_map[*src]; | |
2442 | } | |
2443 | ||
2444 | /* Copy remaining (at most 2) characters. */ | |
2445 | while (src < limit) | |
2446 | *dest++ = *src++; | |
2447 | return dest; | |
2448 | } | |
2449 | ||
2450 | /* If CUR is a backslash or the end of a trigraphed backslash, return | |
2451 | a pointer to its beginning, otherwise NULL. We don't read beyond | |
2452 | the buffer start, because there is the start of the comment in the | |
2453 | buffer. */ | |
2454 | static const unsigned char * | |
2455 | backslash_start (pfile, cur) | |
2456 | cpp_reader *pfile; | |
2457 | const unsigned char *cur; | |
2458 | { | |
2459 | if (cur[0] == '\\') | |
2460 | return cur; | |
2461 | if (cur[0] == '/' && cur[-1] == '?' && cur[-2] == '?' | |
2462 | && trigraph_ok (pfile, cur)) | |
2463 | return cur - 2; | |
2464 | return 0; | |
2465 | } | |
2466 | ||
2467 | /* Skip a C-style block comment. This is probably the trickiest | |
2468 | handler. We find the end of the comment by seeing if an asterisk | |
2469 | is before every '/' we encounter. The nasty complication is that a | |
2470 | previous asterisk may be separated by one or more escaped newlines. | |
2471 | Returns non-zero if comment terminated by EOF, zero otherwise. */ | |
2472 | static int | |
b8f41010 | 2473 | skip_block_comment2 (pfile) |
c5a04734 ZW |
2474 | cpp_reader *pfile; |
2475 | { | |
2476 | cpp_buffer *buffer = pfile->buffer; | |
2477 | const unsigned char *char_after_star = 0; | |
2478 | register const unsigned char *cur = buffer->cur; | |
2479 | int seen_eof = 0; | |
2480 | ||
2481 | /* Inner loop would think the comment has ended if the first comment | |
2482 | character is a '/'. Avoid this and keep the inner loop clean by | |
2483 | skipping such a character. */ | |
2484 | if (cur < buffer->rlimit && cur[0] == '/') | |
2485 | cur++; | |
2486 | ||
2487 | for (; cur < buffer->rlimit; ) | |
2488 | { | |
2489 | unsigned char c = *cur++; | |
2490 | ||
2491 | /* People like decorating comments with '*', so check for | |
2492 | '/' instead for efficiency. */ | |
2493 | if (c == '/') | |
2494 | { | |
2495 | if (cur[-2] == '*' || cur - 1 == char_after_star) | |
2496 | goto out; | |
2497 | ||
2498 | /* Warn about potential nested comments, but not when | |
2499 | the final character inside the comment is a '/'. | |
2500 | Don't bother to get it right across escaped newlines. */ | |
2501 | if (CPP_OPTION (pfile, warn_comments) && cur + 1 < buffer->rlimit | |
2502 | && cur[0] == '*' && cur[1] != '/') | |
2503 | { | |
2504 | buffer->cur = cur; | |
2505 | cpp_warning (pfile, "'/*' within comment"); | |
2506 | } | |
2507 | } | |
2508 | else if (IS_NEWLINE(c)) | |
2509 | { | |
2510 | const unsigned char* bslash = backslash_start (pfile, cur - 2); | |
2511 | ||
2512 | handle_newline (cur, buffer->rlimit, c); | |
2513 | /* Work correctly if there is an asterisk before an | |
2514 | arbirtrarily long sequence of escaped newlines. */ | |
2515 | if (bslash && (bslash[-1] == '*' || bslash == char_after_star)) | |
2516 | char_after_star = cur; | |
2517 | else | |
2518 | char_after_star = 0; | |
2519 | } | |
2520 | } | |
2521 | seen_eof = 1; | |
2522 | ||
2523 | out: | |
2524 | buffer->cur = cur; | |
2525 | return seen_eof; | |
2526 | } | |
2527 | ||
2528 | /* Skip a C++ or Chill line comment. Handles escaped newlines. | |
2529 | Returns non-zero if a multiline comment. */ | |
2530 | static int | |
b8f41010 | 2531 | skip_line_comment2 (pfile) |
c5a04734 ZW |
2532 | cpp_reader *pfile; |
2533 | { | |
2534 | cpp_buffer *buffer = pfile->buffer; | |
2535 | register const unsigned char *cur = buffer->cur; | |
2536 | int multiline = 0; | |
2537 | ||
2538 | for (; cur < buffer->rlimit; ) | |
2539 | { | |
2540 | unsigned char c = *cur++; | |
2541 | ||
2542 | if (IS_NEWLINE (c)) | |
2543 | { | |
2544 | /* Check for a (trigaph?) backslash escaping the newline. */ | |
2545 | if (!backslash_start (pfile, cur - 2)) | |
2546 | goto out; | |
2547 | multiline = 1; | |
2548 | handle_newline (cur, buffer->rlimit, c); | |
2549 | } | |
2550 | } | |
2551 | cur++; | |
2552 | ||
2553 | out: | |
2554 | buffer->cur = cur - 1; /* Leave newline for caller. */ | |
2555 | return multiline; | |
2556 | } | |
2557 | ||
6ab3e7dd NB |
2558 | /* Skips whitespace, stopping at next non-whitespace character. |
2559 | Adjusts pfile->col_adjust to account for tabs. This enables tokens | |
2560 | to be assigned the correct column. */ | |
c5a04734 ZW |
2561 | static void |
2562 | skip_whitespace (pfile, in_directive) | |
2563 | cpp_reader *pfile; | |
2564 | int in_directive; | |
2565 | { | |
2566 | cpp_buffer *buffer = pfile->buffer; | |
2567 | register const unsigned char *cur = buffer->cur; | |
2568 | unsigned short null_count = 0; | |
2569 | ||
2570 | for (; cur < buffer->rlimit; ) | |
2571 | { | |
2572 | unsigned char c = *cur++; | |
2573 | ||
6ab3e7dd NB |
2574 | if (c == '\t') |
2575 | { | |
2576 | unsigned int col = CPP_BUF_COLUMN (buffer, cur - 1); | |
2577 | pfile->col_adjust += (CPP_OPTION (pfile, tabstop) - 1 | |
2578 | - col % CPP_OPTION(pfile, tabstop)); | |
2579 | } | |
c5a04734 ZW |
2580 | if (IS_HSPACE(c)) /* FIXME: Fix ISTABLE. */ |
2581 | continue; | |
2582 | if (!is_space(c) || IS_NEWLINE (c)) /* Main loop handles newlines. */ | |
2583 | goto out; | |
2584 | if (c == '\0') | |
2585 | null_count++; | |
2586 | /* Mut be '\f' or '\v' */ | |
2587 | else if (in_directive && CPP_PEDANTIC (pfile)) | |
2588 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
2589 | c == '\f' ? "formfeed" : "vertical tab"); | |
2590 | } | |
2591 | cur++; | |
2592 | ||
2593 | out: | |
2594 | buffer->cur = cur - 1; | |
2595 | if (null_count) | |
2596 | cpp_warning (pfile, null_count > 1 ? "embedded null characters ignored" | |
2597 | : "embedded null character ignored"); | |
2598 | } | |
2599 | ||
2600 | /* Parse (append) an identifier. */ | |
2601 | static void | |
2602 | parse_name (pfile, list, name) | |
2603 | cpp_reader *pfile; | |
2604 | cpp_toklist *list; | |
2605 | cpp_name *name; | |
2606 | { | |
2607 | const unsigned char *name_limit; | |
2608 | unsigned char *namebuf; | |
2609 | cpp_buffer *buffer = pfile->buffer; | |
2610 | register const unsigned char *cur = buffer->cur; | |
2611 | ||
2612 | expanded: | |
2613 | name_limit = list->namebuf + list->name_cap; | |
2614 | namebuf = list->namebuf + list->name_used; | |
2615 | ||
2616 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2617 | { | |
2618 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2619 | ||
2620 | if (! is_idchar(c)) | |
2621 | goto out; | |
2622 | namebuf++; | |
2623 | cur++; | |
2624 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
2625 | { | |
2626 | buffer->cur = cur; | |
2627 | cpp_pedwarn (pfile, "'$' character in identifier"); | |
2628 | } | |
2629 | } | |
2630 | ||
2631 | /* Run out of name space? */ | |
2632 | if (cur < buffer->rlimit) | |
2633 | { | |
2634 | list->name_used = namebuf - list->namebuf; | |
2635 | auto_expand_name_space (list); | |
2636 | goto expanded; | |
2637 | } | |
2638 | ||
2639 | out: | |
2640 | buffer->cur = cur; | |
f617b8e2 | 2641 | name->len = namebuf - name->text; |
c5a04734 ZW |
2642 | list->name_used = namebuf - list->namebuf; |
2643 | } | |
2644 | ||
2645 | /* Parse (append) a number. */ | |
2646 | ||
2647 | #define VALID_SIGN(c, prevc) \ | |
2648 | (((c) == '+' || (c) == '-') && \ | |
2649 | ((prevc) == 'e' || (prevc) == 'E' \ | |
2650 | || (((prevc) == 'p' || (prevc) == 'P') && !CPP_OPTION (pfile, c89)))) | |
2651 | ||
2652 | static void | |
2653 | parse_number (pfile, list, name) | |
2654 | cpp_reader *pfile; | |
2655 | cpp_toklist *list; | |
2656 | cpp_name *name; | |
2657 | { | |
2658 | const unsigned char *name_limit; | |
2659 | unsigned char *namebuf; | |
2660 | cpp_buffer *buffer = pfile->buffer; | |
2661 | register const unsigned char *cur = buffer->cur; | |
2662 | ||
2663 | expanded: | |
2664 | name_limit = list->namebuf + list->name_cap; | |
2665 | namebuf = list->namebuf + list->name_used; | |
2666 | ||
2667 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2668 | { | |
2669 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2670 | ||
2671 | /* Perhaps we should accept '$' here if we accept it for | |
2672 | identifiers. We know namebuf[-1] is safe, because for c to | |
2673 | be a sign we must have pushed at least one character. */ | |
2674 | if (!is_numchar (c) && c != '.' && ! VALID_SIGN (c, namebuf[-1])) | |
2675 | goto out; | |
2676 | ||
2677 | namebuf++; | |
2678 | cur++; | |
2679 | } | |
2680 | ||
2681 | /* Run out of name space? */ | |
2682 | if (cur < buffer->rlimit) | |
2683 | { | |
2684 | list->name_used = namebuf - list->namebuf; | |
2685 | auto_expand_name_space (list); | |
2686 | goto expanded; | |
2687 | } | |
2688 | ||
2689 | out: | |
2690 | buffer->cur = cur; | |
f617b8e2 | 2691 | name->len = namebuf - name->text; |
c5a04734 ZW |
2692 | list->name_used = namebuf - list->namebuf; |
2693 | } | |
2694 | ||
2695 | /* Places a string terminated by an unescaped TERMINATOR into a | |
2696 | cpp_name, which should be expandable and thus at the top of the | |
2697 | list's stack. Handles embedded trigraphs, if necessary, and | |
2698 | escaped newlines. | |
2699 | ||
2700 | Can be used for character constants (terminator = '\''), string | |
41e8b1d7 NB |
2701 | constants ('"') and angled headers ('>'). Multi-line strings are |
2702 | allowed, except for within directives. */ | |
c5a04734 ZW |
2703 | |
2704 | static void | |
b8f41010 | 2705 | parse_string2 (pfile, list, name, terminator) |
c5a04734 ZW |
2706 | cpp_reader *pfile; |
2707 | cpp_toklist *list; | |
2708 | cpp_name *name; | |
2709 | unsigned int terminator; | |
2710 | { | |
2711 | cpp_buffer *buffer = pfile->buffer; | |
2712 | register const unsigned char *cur = buffer->cur; | |
2713 | const unsigned char *name_limit; | |
2714 | unsigned char *namebuf; | |
2715 | unsigned int null_count = 0; | |
2716 | int trigraphed_len = 0; | |
2717 | ||
2718 | expanded: | |
2719 | name_limit = list->namebuf + list->name_cap; | |
2720 | namebuf = list->namebuf + list->name_used; | |
2721 | ||
2722 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2723 | { | |
2724 | unsigned int c = *namebuf++ = *cur++; /* Copy a single char. */ | |
2725 | ||
2726 | if (c == '\0') | |
2727 | null_count++; | |
2728 | else if (c == terminator || IS_NEWLINE (c)) | |
2729 | { | |
c5a04734 ZW |
2730 | /* Needed for trigraph_replace and multiline string warning. */ |
2731 | buffer->cur = cur; | |
2732 | ||
2733 | /* Scan for trigraphs before checking if backslash-escaped. */ | |
2734 | if (CPP_OPTION (pfile, trigraphs) | |
2735 | || CPP_OPTION (pfile, warn_trigraphs)) | |
2736 | { | |
f617b8e2 | 2737 | namebuf = trigraph_replace (pfile, name->text + trigraphed_len, |
c5a04734 | 2738 | namebuf); |
f617b8e2 | 2739 | trigraphed_len = namebuf - 2 - (name->text + trigraphed_len); |
c5a04734 ZW |
2740 | if (trigraphed_len < 0) |
2741 | trigraphed_len = 0; | |
2742 | } | |
2743 | ||
2744 | namebuf--; /* Drop the newline / terminator from the name. */ | |
2745 | if (IS_NEWLINE (c)) | |
2746 | { | |
2747 | /* Drop a backslash newline, and continue. */ | |
2748 | if (namebuf[-1] == '\\') | |
2749 | { | |
2750 | handle_newline (cur, buffer->rlimit, c); | |
2751 | namebuf--; | |
2752 | continue; | |
2753 | } | |
2754 | ||
2755 | cur--; | |
2756 | ||
2757 | /* In Fortran and assembly language, silently terminate | |
2758 | strings of either variety at end of line. This is a | |
2759 | kludge around not knowing where comments are in these | |
2760 | languages. */ | |
2761 | if (CPP_OPTION (pfile, lang_fortran) | |
2762 | || CPP_OPTION (pfile, lang_asm)) | |
2763 | goto out; | |
2764 | ||
2765 | /* Character constants, headers and asserts may not | |
2766 | extend over multiple lines. In Standard C, neither | |
2767 | may strings. We accept multiline strings as an | |
2768 | extension, but not in directives. */ | |
2769 | if (terminator != '"' || IS_DIRECTIVE (list)) | |
2770 | goto unterminated; | |
2771 | ||
2772 | cur++; /* Move forwards again. */ | |
2773 | ||
2774 | if (pfile->multiline_string_line == 0) | |
2775 | { | |
2776 | pfile->multiline_string_line = list->line; | |
2777 | if (CPP_PEDANTIC (pfile)) | |
2778 | cpp_pedwarn (pfile, "multi-line string constant"); | |
2779 | } | |
2780 | ||
2781 | *namebuf++ = '\n'; | |
2782 | handle_newline (cur, buffer->rlimit, c); | |
2783 | } | |
2784 | else | |
2785 | { | |
2786 | unsigned char *temp; | |
2787 | ||
2788 | /* An odd number of consecutive backslashes represents | |
2789 | an escaped terminator. */ | |
2790 | temp = namebuf - 1; | |
f617b8e2 | 2791 | while (temp >= name->text && *temp == '\\') |
c5a04734 ZW |
2792 | temp--; |
2793 | ||
2794 | if ((namebuf - temp) & 1) | |
2795 | goto out; | |
2796 | namebuf++; | |
2797 | } | |
2798 | } | |
2799 | } | |
2800 | ||
2801 | /* Run out of name space? */ | |
2802 | if (cur < buffer->rlimit) | |
2803 | { | |
2804 | list->name_used = namebuf - list->namebuf; | |
2805 | auto_expand_name_space (list); | |
2806 | goto expanded; | |
2807 | } | |
2808 | ||
2809 | /* We may not have trigraph-replaced the input for this code path, | |
2810 | but as the input is in error by being unterminated we don't | |
2811 | bother. Prevent warnings about no newlines at EOF. */ | |
2812 | if (IS_NEWLINE(cur[-1])) | |
2813 | cur--; | |
2814 | ||
2815 | unterminated: | |
2816 | cpp_error (pfile, "missing terminating %c character", (int) terminator); | |
2817 | ||
2818 | if (terminator == '\"' && pfile->multiline_string_line != list->line | |
2819 | && pfile->multiline_string_line != 0) | |
2820 | { | |
2821 | cpp_error_with_line (pfile, pfile->multiline_string_line, -1, | |
2822 | "possible start of unterminated string literal"); | |
2823 | pfile->multiline_string_line = 0; | |
2824 | } | |
2825 | ||
2826 | out: | |
2827 | buffer->cur = cur; | |
f617b8e2 | 2828 | name->len = namebuf - name->text; |
c5a04734 ZW |
2829 | list->name_used = namebuf - list->namebuf; |
2830 | ||
2831 | if (null_count > 0) | |
2832 | cpp_warning (pfile, (null_count > 1 ? "null characters preserved" | |
2833 | : "null character preserved")); | |
2834 | } | |
2835 | ||
5d7ee2fa NB |
2836 | /* The character TYPE helps us distinguish comment types: '*' = C |
2837 | style, '-' = Chill-style and '/' = C++ style. For code simplicity, | |
2838 | the stored comment includes the comment start and any terminator. */ | |
2839 | ||
2840 | #define COMMENT_START_LEN 2 | |
c5a04734 | 2841 | static void |
b8f41010 | 2842 | save_comment (list, from, len, tok_no, type) |
c5a04734 ZW |
2843 | cpp_toklist *list; |
2844 | const unsigned char *from; | |
2845 | unsigned int len; | |
2846 | unsigned int tok_no; | |
2847 | unsigned int type; | |
2848 | { | |
2849 | cpp_token *comment; | |
5d7ee2fa NB |
2850 | unsigned char *buffer; |
2851 | ||
2852 | len += COMMENT_START_LEN; | |
c5a04734 ZW |
2853 | |
2854 | if (list->comments_used == list->comments_cap) | |
2855 | expand_comment_space (list); | |
2856 | ||
2857 | if (list->name_used + len > list->name_cap) | |
2858 | expand_name_space (list, len); | |
2859 | ||
f617b8e2 NB |
2860 | buffer = list->namebuf + list->name_used; |
2861 | ||
c5a04734 | 2862 | comment = &list->comments[list->comments_used++]; |
5d7ee2fa | 2863 | comment->type = CPP_COMMENT; |
c5a04734 ZW |
2864 | comment->aux = tok_no; |
2865 | comment->val.name.len = len; | |
f617b8e2 | 2866 | comment->val.name.text = buffer; |
c5a04734 | 2867 | |
5d7ee2fa NB |
2868 | if (type == '*') |
2869 | { | |
2870 | *buffer++ = '/'; | |
2871 | *buffer++ = '*'; | |
2872 | } | |
2873 | else | |
2874 | { | |
2875 | *buffer++ = type; | |
2876 | *buffer++ = type; | |
2877 | } | |
2878 | ||
2879 | memcpy (buffer, from, len - COMMENT_START_LEN); | |
c5a04734 ZW |
2880 | list->name_used += len; |
2881 | } | |
2882 | ||
2883 | /* | |
2884 | * The tokenizer's main loop. Returns a token list, representing a | |
2885 | * logical line in the input file, terminated with a CPP_VSPACE | |
2886 | * token. On EOF, a token list containing the single CPP_EOF token | |
2887 | * is returned. | |
2888 | * | |
2889 | * Implementation relies almost entirely on lookback, rather than | |
2890 | * looking forwards. This means that tokenization requires just | |
2891 | * a single pass of the file, even in the presence of trigraphs and | |
2892 | * escaped newlines, providing significant performance benefits. | |
2893 | * Trigraph overhead is negligible if they are disabled, and low | |
2894 | * even when enabled. | |
2895 | */ | |
2896 | ||
c5a04734 ZW |
2897 | void |
2898 | _cpp_lex_line (pfile, list) | |
2899 | cpp_reader *pfile; | |
2900 | cpp_toklist *list; | |
2901 | { | |
2902 | cpp_token *cur_token, *token_limit; | |
2903 | cpp_buffer *buffer = pfile->buffer; | |
2904 | register const unsigned char *cur = buffer->cur; | |
2905 | unsigned char flags = 0; | |
2906 | ||
6ab3e7dd | 2907 | pfile->col_adjust = 0; |
c5a04734 ZW |
2908 | expanded: |
2909 | token_limit = list->tokens + list->tokens_cap; | |
2910 | cur_token = list->tokens + list->tokens_used; | |
2911 | ||
2912 | for (; cur < buffer->rlimit && cur_token < token_limit;) | |
2913 | { | |
2914 | unsigned char c = *cur++; | |
2915 | ||
6ab3e7dd NB |
2916 | /* Optimize whitespace skipping, as most tokens are probably |
2917 | separated by whitespace. (' ' '\t' '\v' '\f' '\0'). */ | |
2918 | ||
c5a04734 ZW |
2919 | if (is_hspace ((unsigned int) c)) |
2920 | { | |
6ab3e7dd NB |
2921 | /* Step back to get the null warning and tab correction. */ |
2922 | buffer->cur = cur - 1; | |
2923 | skip_whitespace (pfile, IS_DIRECTIVE (list)); | |
2924 | cur = buffer->cur; | |
2925 | ||
c5a04734 ZW |
2926 | flags = PREV_WHITESPACE; |
2927 | if (cur == buffer->rlimit) | |
2928 | break; | |
2929 | c = *cur++; | |
2930 | } | |
2931 | ||
2932 | /* Initialize current token. Its type is set in the switch. */ | |
6ab3e7dd | 2933 | cur_token->col = CPP_BUF_COLUMN (buffer, cur); |
c5a04734 ZW |
2934 | cur_token->flags = flags; |
2935 | flags = 0; | |
2936 | ||
2937 | switch (c) | |
2938 | { | |
2939 | case '0': case '1': case '2': case '3': case '4': | |
2940 | case '5': case '6': case '7': case '8': case '9': | |
f617b8e2 | 2941 | cur--; /* Backup character. */ |
c5a04734 ZW |
2942 | if (PREV_TOKEN_TYPE == CPP_DOT && IMMED_TOKEN ()) |
2943 | { | |
f617b8e2 | 2944 | /* Prepend an immediately previous CPP_DOT token. */ |
c5a04734 ZW |
2945 | cur_token--; |
2946 | if (list->name_cap == list->name_used) | |
2947 | auto_expand_name_space (list); | |
2948 | ||
2949 | cur_token->val.name.len = 1; | |
f617b8e2 | 2950 | cur_token->val.name.text = list->namebuf + list->name_used; |
c5a04734 ZW |
2951 | list->namebuf[list->name_used++] = '.'; |
2952 | } | |
2953 | else | |
2954 | INIT_NAME (list, cur_token->val.name); | |
c5a04734 ZW |
2955 | |
2956 | continue_number: | |
2957 | buffer->cur = cur; | |
2958 | parse_number (pfile, list, &cur_token->val.name); | |
2959 | cur = buffer->cur; | |
2960 | ||
2961 | PUSH_TOKEN (CPP_NUMBER); /* Number not yet interpreted. */ | |
2962 | break; | |
2963 | ||
2964 | letter: | |
2965 | case '_': | |
2966 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
2967 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
2968 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
2969 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
2970 | case 'y': case 'z': | |
2971 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
2972 | case 'G': case 'H': case 'I': case 'J': case 'K': case 'L': | |
2973 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': | |
2974 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
2975 | case 'Y': case 'Z': | |
c5a04734 | 2976 | cur--; /* Backup character. */ |
f617b8e2 | 2977 | INIT_NAME (list, cur_token->val.name); |
c5a04734 ZW |
2978 | cur_token->type = CPP_NAME; /* Identifier, macro etc. */ |
2979 | ||
2980 | continue_name: | |
2981 | buffer->cur = cur; | |
2982 | parse_name (pfile, list, &cur_token->val.name); | |
2983 | cur = buffer->cur; | |
2984 | ||
2985 | /* Find handler for newly created / extended directive. */ | |
2986 | if (IS_DIRECTIVE (list) && cur_token == &list->tokens[1]) | |
2987 | _cpp_check_directive (list, cur_token); | |
2988 | cur_token++; | |
2989 | break; | |
2990 | ||
2991 | case '\'': | |
2992 | /* Fall through. */ | |
2993 | case '\"': | |
2994 | cur_token->type = c == '\'' ? CPP_CHAR : CPP_STRING; | |
2995 | /* Do we have a wide string? */ | |
2996 | if (cur_token[-1].type == CPP_NAME && IMMED_TOKEN () | |
2997 | && cur_token[-1].val.name.len == 1 | |
f617b8e2 | 2998 | && cur_token[-1].val.name.text[0] == 'L' |
c5a04734 ZW |
2999 | && !CPP_TRADITIONAL (pfile)) |
3000 | { | |
3001 | /* No need for 'L' any more. */ | |
3002 | list->name_used--; | |
3003 | (--cur_token)->type = (c == '\'' ? CPP_WCHAR : CPP_WSTRING); | |
3004 | } | |
3005 | ||
3006 | do_parse_string: | |
6ab3e7dd | 3007 | /* Here c is one of ' " or >. */ |
c5a04734 ZW |
3008 | INIT_NAME (list, cur_token->val.name); |
3009 | buffer->cur = cur; | |
b8f41010 | 3010 | parse_string2 (pfile, list, &cur_token->val.name, c); |
c5a04734 ZW |
3011 | cur = buffer->cur; |
3012 | cur_token++; | |
3013 | break; | |
3014 | ||
3015 | case '/': | |
3016 | cur_token->type = CPP_DIV; | |
3017 | if (IMMED_TOKEN ()) | |
3018 | { | |
3019 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
3020 | { | |
3021 | /* We silently allow C++ comments in system headers, | |
3022 | irrespective of conformance mode, because lots of | |
3023 | broken systems do that and trying to clean it up | |
3024 | in fixincludes is a nightmare. */ | |
3025 | if (buffer->system_header_p) | |
3026 | goto do_line_comment; | |
3027 | else if (CPP_OPTION (pfile, cplusplus_comments)) | |
3028 | { | |
3029 | if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile) | |
3030 | && ! buffer->warned_cplusplus_comments) | |
3031 | { | |
3032 | buffer->cur = cur; | |
3033 | cpp_pedwarn (pfile, | |
3034 | "C++ style comments are not allowed in ISO C89"); | |
3035 | cpp_pedwarn (pfile, | |
3036 | "(this will be reported only once per input file)"); | |
3037 | buffer->warned_cplusplus_comments = 1; | |
3038 | } | |
3039 | do_line_comment: | |
3040 | buffer->cur = cur; | |
3041 | if (cur[-2] != c) | |
3042 | cpp_warning (pfile, | |
3043 | "comment start split across lines"); | |
b8f41010 | 3044 | if (skip_line_comment2 (pfile)) |
c5a04734 ZW |
3045 | cpp_error_with_line (pfile, list->line, |
3046 | cur_token[-1].col, | |
3047 | "multi-line comment"); | |
3048 | if (!CPP_OPTION (pfile, discard_comments)) | |
b8f41010 | 3049 | save_comment (list, cur, buffer->cur - cur, |
5d7ee2fa | 3050 | cur_token - 1 - list->tokens, c); |
c5a04734 ZW |
3051 | cur = buffer->cur; |
3052 | ||
3053 | /* Back-up to first '-' or '/'. */ | |
3054 | cur_token -= 2; | |
3055 | if (!CPP_OPTION (pfile, traditional)) | |
3056 | flags = PREV_WHITESPACE; | |
3057 | } | |
3058 | } | |
3059 | } | |
3060 | cur_token++; | |
3061 | break; | |
3062 | ||
3063 | case '*': | |
3064 | cur_token->type = CPP_MULT; | |
3065 | if (IMMED_TOKEN ()) | |
3066 | { | |
3067 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
3068 | { | |
3069 | buffer->cur = cur; | |
3070 | if (cur[-2] != '/') | |
3071 | cpp_warning (pfile, | |
3072 | "comment start '/*' split across lines"); | |
b8f41010 | 3073 | if (skip_block_comment2 (pfile)) |
c5a04734 ZW |
3074 | cpp_error_with_line (pfile, list->line, cur_token[-1].col, |
3075 | "unterminated comment"); | |
3076 | else if (buffer->cur[-2] != '*') | |
3077 | cpp_warning (pfile, | |
3078 | "comment end '*/' split across lines"); | |
3079 | if (!CPP_OPTION (pfile, discard_comments)) | |
b8f41010 | 3080 | save_comment (list, cur, buffer->cur - cur, |
5d7ee2fa | 3081 | cur_token - 1 - list->tokens, c); |
c5a04734 ZW |
3082 | cur = buffer->cur; |
3083 | ||
f617b8e2 | 3084 | cur_token--; |
c5a04734 ZW |
3085 | if (!CPP_OPTION (pfile, traditional)) |
3086 | flags = PREV_WHITESPACE; | |
f617b8e2 | 3087 | break; |
c5a04734 ZW |
3088 | } |
3089 | else if (CPP_OPTION (pfile, cplusplus)) | |
3090 | { | |
3091 | /* In C++, there are .* and ->* operators. */ | |
3092 | if (PREV_TOKEN_TYPE == CPP_DEREF) | |
3093 | BACKUP_TOKEN (CPP_DEREF_STAR); | |
3094 | else if (PREV_TOKEN_TYPE == CPP_DOT) | |
3095 | BACKUP_TOKEN (CPP_DOT_STAR); | |
3096 | } | |
3097 | } | |
3098 | cur_token++; | |
3099 | break; | |
3100 | ||
3101 | case '\n': | |
3102 | case '\r': | |
3103 | handle_newline (cur, buffer->rlimit, c); | |
fb4527c3 | 3104 | if (PREV_TOKEN_TYPE == CPP_BACKSLASH && IMMED_TOKEN ()) |
c5a04734 | 3105 | { |
fb4527c3 NB |
3106 | /* Remove the escaped newline. Then continue to process |
3107 | any interrupted name or number. */ | |
3108 | cur_token--; | |
3109 | if (IMMED_TOKEN ()) | |
c5a04734 | 3110 | { |
fb4527c3 NB |
3111 | cur_token--; |
3112 | if (cur_token->type == CPP_NAME) | |
3113 | goto continue_name; | |
3114 | else if (cur_token->type == CPP_NUMBER) | |
3115 | goto continue_number; | |
3116 | cur_token++; | |
c5a04734 | 3117 | } |
fb4527c3 NB |
3118 | /* Remember whitespace setting. */ |
3119 | flags = cur_token->flags; | |
3120 | break; | |
c5a04734 | 3121 | } |
fb4527c3 | 3122 | if (PREV_TOKEN_TYPE == CPP_BACKSLASH) |
c5a04734 | 3123 | { |
fb4527c3 NB |
3124 | buffer->cur = cur; |
3125 | cpp_warning (pfile, "backslash and newline separated by space"); | |
c5a04734 | 3126 | } |
fb4527c3 NB |
3127 | PUSH_TOKEN (CPP_VSPACE); |
3128 | goto out; | |
c5a04734 ZW |
3129 | |
3130 | case '-': | |
3131 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_MINUS) | |
3132 | { | |
3133 | if (CPP_OPTION (pfile, chill)) | |
3134 | goto do_line_comment; | |
3135 | REVISE_TOKEN (CPP_MINUS_MINUS); | |
3136 | } | |
3137 | else | |
3138 | PUSH_TOKEN (CPP_MINUS); | |
3139 | break; | |
3140 | ||
3141 | /* The digraph flag checking ensures that ## and %:%: | |
3142 | are interpreted as CPP_PASTE, but #%: and %:# are not. */ | |
3143 | make_hash: | |
3144 | case '#': | |
3145 | if (PREV_TOKEN_TYPE == CPP_HASH && IMMED_TOKEN () | |
3146 | && ((cur_token->flags ^ cur_token[-1].flags) & DIGRAPH) == 0) | |
3147 | REVISE_TOKEN (CPP_PASTE); | |
3148 | else | |
3149 | PUSH_TOKEN (CPP_HASH); | |
3150 | break; | |
3151 | ||
3152 | case ':': | |
3153 | cur_token->type = CPP_COLON; | |
3154 | if (IMMED_TOKEN ()) | |
3155 | { | |
3156 | if (PREV_TOKEN_TYPE == CPP_COLON | |
3157 | && CPP_OPTION (pfile, cplusplus)) | |
3158 | BACKUP_TOKEN (CPP_SCOPE); | |
3159 | /* Digraph: "<:" is a '[' */ | |
3160 | else if (PREV_TOKEN_TYPE == CPP_LESS) | |
3161 | BACKUP_DIGRAPH (CPP_OPEN_SQUARE); | |
3162 | /* Digraph: "%:" is a '#' */ | |
3163 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3164 | { | |
3165 | (--cur_token)->flags |= DIGRAPH; | |
3166 | goto make_hash; | |
3167 | } | |
3168 | } | |
3169 | cur_token++; | |
3170 | break; | |
3171 | ||
3172 | case '&': | |
3173 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_AND) | |
3174 | REVISE_TOKEN (CPP_AND_AND); | |
3175 | else | |
3176 | PUSH_TOKEN (CPP_AND); | |
3177 | break; | |
3178 | ||
3179 | make_or: | |
3180 | case '|': | |
3181 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_OR) | |
3182 | REVISE_TOKEN (CPP_OR_OR); | |
3183 | else | |
3184 | PUSH_TOKEN (CPP_OR); | |
3185 | break; | |
3186 | ||
3187 | case '+': | |
3188 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_PLUS) | |
3189 | REVISE_TOKEN (CPP_PLUS_PLUS); | |
3190 | else | |
3191 | PUSH_TOKEN (CPP_PLUS); | |
3192 | break; | |
3193 | ||
3194 | case '=': | |
3195 | /* This relies on equidistance of "?=" and "?" tokens. */ | |
3196 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE <= CPP_LAST_EQ) | |
3197 | REVISE_TOKEN (PREV_TOKEN_TYPE + (CPP_EQ_EQ - CPP_EQ)); | |
3198 | else | |
3199 | PUSH_TOKEN (CPP_EQ); | |
3200 | break; | |
3201 | ||
3202 | case '>': | |
3203 | cur_token->type = CPP_GREATER; | |
3204 | if (IMMED_TOKEN ()) | |
3205 | { | |
3206 | if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3207 | BACKUP_TOKEN (CPP_RSHIFT); | |
3208 | else if (PREV_TOKEN_TYPE == CPP_MINUS) | |
3209 | BACKUP_TOKEN (CPP_DEREF); | |
3210 | /* Digraph: ":>" is a ']' */ | |
3211 | else if (PREV_TOKEN_TYPE == CPP_COLON) | |
3212 | BACKUP_DIGRAPH (CPP_CLOSE_SQUARE); | |
3213 | /* Digraph: "%>" is a '}' */ | |
3214 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3215 | BACKUP_DIGRAPH (CPP_CLOSE_BRACE); | |
3216 | } | |
3217 | cur_token++; | |
3218 | break; | |
3219 | ||
3220 | case '<': | |
3221 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3222 | { | |
3223 | REVISE_TOKEN (CPP_LSHIFT); | |
3224 | break; | |
3225 | } | |
3226 | /* Is this the beginning of a header name? */ | |
15dad1d9 | 3227 | if (list->flags & SYNTAX_INCLUDE) |
c5a04734 ZW |
3228 | { |
3229 | c = '>'; /* Terminator. */ | |
3230 | cur_token->type = CPP_HEADER_NAME; | |
3231 | goto do_parse_string; | |
3232 | } | |
3233 | PUSH_TOKEN (CPP_LESS); | |
3234 | break; | |
3235 | ||
3236 | case '%': | |
3237 | /* Digraph: "<%" is a '{' */ | |
3238 | cur_token->type = CPP_MOD; | |
3239 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3240 | BACKUP_DIGRAPH (CPP_OPEN_BRACE); | |
3241 | cur_token++; | |
3242 | break; | |
3243 | ||
c5a04734 ZW |
3244 | case '?': |
3245 | if (cur + 1 < buffer->rlimit && *cur == '?' | |
3246 | && trigraph_map[cur[1]] && trigraph_ok (pfile, cur + 1)) | |
3247 | { | |
3248 | /* Handle trigraph. */ | |
3249 | cur++; | |
3250 | switch (*cur++) | |
3251 | { | |
3252 | case '(': goto make_open_square; | |
3253 | case ')': goto make_close_square; | |
3254 | case '<': goto make_open_brace; | |
3255 | case '>': goto make_close_brace; | |
3256 | case '=': goto make_hash; | |
3257 | case '!': goto make_or; | |
3258 | case '-': goto make_complement; | |
3259 | case '/': goto make_backslash; | |
3260 | case '\'': goto make_xor; | |
3261 | } | |
3262 | } | |
3263 | if (IMMED_TOKEN () && CPP_OPTION (pfile, cplusplus)) | |
3264 | { | |
3265 | /* GNU C++ defines <? and >? operators. */ | |
3266 | if (PREV_TOKEN_TYPE == CPP_LESS) | |
3267 | { | |
3268 | REVISE_TOKEN (CPP_MIN); | |
3269 | break; | |
3270 | } | |
3271 | else if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3272 | { | |
3273 | REVISE_TOKEN (CPP_MAX); | |
3274 | break; | |
3275 | } | |
3276 | } | |
3277 | PUSH_TOKEN (CPP_QUERY); | |
3278 | break; | |
3279 | ||
3280 | case '.': | |
3281 | if (PREV_TOKEN_TYPE == CPP_DOT && cur_token[-2].type == CPP_DOT | |
3282 | && IMMED_TOKEN () | |
3283 | && !(cur_token[-1].flags & PREV_WHITESPACE)) | |
3284 | { | |
3285 | cur_token -= 2; | |
3286 | PUSH_TOKEN (CPP_ELLIPSIS); | |
3287 | } | |
3288 | else | |
3289 | PUSH_TOKEN (CPP_DOT); | |
3290 | break; | |
3291 | ||
cfd5b8b8 NB |
3292 | make_complement: |
3293 | case '~': PUSH_TOKEN (CPP_COMPL); break; | |
c5a04734 ZW |
3294 | make_xor: |
3295 | case '^': PUSH_TOKEN (CPP_XOR); break; | |
3296 | make_open_brace: | |
3297 | case '{': PUSH_TOKEN (CPP_OPEN_BRACE); break; | |
3298 | make_close_brace: | |
3299 | case '}': PUSH_TOKEN (CPP_CLOSE_BRACE); break; | |
3300 | make_open_square: | |
3301 | case '[': PUSH_TOKEN (CPP_OPEN_SQUARE); break; | |
3302 | make_close_square: | |
3303 | case ']': PUSH_TOKEN (CPP_CLOSE_SQUARE); break; | |
3304 | make_backslash: | |
3305 | case '\\': PUSH_TOKEN (CPP_BACKSLASH); break; | |
3306 | case '!': PUSH_TOKEN (CPP_NOT); break; | |
3307 | case ',': PUSH_TOKEN (CPP_COMMA); break; | |
3308 | case ';': PUSH_TOKEN (CPP_SEMICOLON); break; | |
41e8b1d7 | 3309 | case '(': PUSH_TOKEN (CPP_OPEN_PAREN); break; |
cfd5b8b8 | 3310 | case ')': PUSH_TOKEN (CPP_CLOSE_PAREN); break; |
c5a04734 ZW |
3311 | |
3312 | case '$': | |
3313 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
3314 | goto letter; | |
3315 | /* Fall through */ | |
3316 | default: | |
3317 | cur_token->aux = c; | |
f617b8e2 | 3318 | cur_token->val.name.len = 0; /* FIXME: needed for transition only */ |
c5a04734 ZW |
3319 | PUSH_TOKEN (CPP_OTHER); |
3320 | break; | |
3321 | } | |
3322 | } | |
3323 | ||
3324 | /* Run out of token space? */ | |
3325 | if (cur_token == token_limit) | |
3326 | { | |
3327 | list->tokens_used = cur_token - list->tokens; | |
3328 | expand_token_space (list); | |
3329 | goto expanded; | |
3330 | } | |
3331 | ||
3332 | cur_token->type = CPP_EOF; | |
3333 | cur_token->flags = flags; | |
3334 | ||
3335 | if (cur_token != &list->tokens[0]) | |
3336 | { | |
3337 | /* Next call back will get just a CPP_EOF. */ | |
3338 | buffer->cur = cur; | |
3339 | cpp_warning (pfile, "no newline at end of file"); | |
3340 | PUSH_TOKEN (CPP_VSPACE); | |
3341 | } | |
3342 | ||
3343 | out: | |
3344 | buffer->cur = cur; | |
3345 | ||
3346 | list->tokens_used = cur_token - list->tokens; | |
3347 | ||
3348 | /* FIXME: take this check out and put it in the caller. | |
3349 | list->directive == 0 indicates an unknown directive (but null | |
3350 | directive is OK). This is the first time we can be sure the | |
3351 | directive is invalid, and thus warn about it, because it might | |
3352 | have been split by escaped newlines. Also, don't complain about | |
3353 | invalid directives in assembly source, we don't know where the | |
3354 | comments are, and # may introduce assembler pseudo-ops. */ | |
3355 | ||
15dad1d9 | 3356 | if (IS_DIRECTIVE (list) && list->dirno == -1 |
c5a04734 ZW |
3357 | && list->tokens[1].type != CPP_VSPACE |
3358 | && !CPP_OPTION (pfile, lang_asm)) | |
3359 | cpp_error_with_line (pfile, list->line, list->tokens[1].col, | |
3360 | "invalid preprocessing directive"); | |
3361 | } | |
3362 | ||
3fef5b2b NB |
3363 | /* Write the spelling of a token TOKEN to BUFFER. The buffer must |
3364 | already contain the enough space to hold the token's spelling. If | |
3365 | WHITESPACE is true, and the token was preceded by whitespace, | |
3366 | output a single space before the token proper. Returns a pointer | |
3367 | to the character after the last character written. */ | |
3368 | ||
3369 | static unsigned char * | |
f617b8e2 | 3370 | spell_token (pfile, token, buffer, whitespace) |
3fef5b2b NB |
3371 | cpp_reader *pfile; /* Would be nice to be rid of this... */ |
3372 | cpp_token *token; | |
3fef5b2b NB |
3373 | unsigned char *buffer; |
3374 | int whitespace; | |
3375 | { | |
3376 | /* Whitespace will not be wanted by handlers of the # and ## | |
3377 | operators calling this function, but will be wanted by the | |
3378 | function that writes out the preprocessed file. */ | |
3379 | if (whitespace && token->flags & PREV_WHITESPACE) | |
3380 | *buffer++ = ' '; | |
3381 | ||
3382 | switch (token_spellings[token->type].type) | |
3383 | { | |
5d7ee2fa | 3384 | case SPELL_OPERATOR: |
3fef5b2b NB |
3385 | { |
3386 | const unsigned char *spelling; | |
3387 | unsigned char c; | |
3388 | ||
3389 | if (token->flags & DIGRAPH) | |
3390 | spelling = digraph_spellings[token->type - CPP_FIRST_DIGRAPH]; | |
3391 | else | |
f617b8e2 | 3392 | spelling = token_spellings[token->type].spelling; |
3fef5b2b NB |
3393 | |
3394 | while ((c = *spelling++) != '\0') | |
3395 | *buffer++ = c; | |
3396 | } | |
3397 | break; | |
3398 | ||
5d7ee2fa | 3399 | case SPELL_IDENT: |
f617b8e2 | 3400 | memcpy (buffer, token->val.name.text, token->val.name.len); |
5d7ee2fa NB |
3401 | buffer += token->val.name.len; |
3402 | break; | |
3403 | ||
3404 | case SPELL_STRING: | |
3fef5b2b | 3405 | { |
5d7ee2fa | 3406 | unsigned char c; |
3fef5b2b | 3407 | |
5d7ee2fa NB |
3408 | if (token->type == CPP_WSTRING || token->type == CPP_WCHAR) |
3409 | *buffer++ = 'L'; | |
3410 | c = '\''; | |
3411 | if (token->type == CPP_STRING || token->type == CPP_WSTRING) | |
3412 | c = '"'; | |
3413 | *buffer++ = c; | |
f617b8e2 | 3414 | memcpy (buffer, token->val.name.text, token->val.name.len); |
5d7ee2fa NB |
3415 | buffer += token->val.name.len; |
3416 | *buffer++ = c; | |
3fef5b2b NB |
3417 | } |
3418 | break; | |
3419 | ||
3420 | case SPELL_CHAR: | |
3421 | *buffer++ = token->aux; | |
3422 | break; | |
3423 | ||
3424 | case SPELL_NONE: | |
3425 | cpp_ice (pfile, "Unspellable token"); | |
3426 | break; | |
3427 | } | |
3428 | ||
3429 | return buffer; | |
3430 | } | |
3431 | ||
3432 | /* Temporary function for illustrative purposes. */ | |
c5a04734 ZW |
3433 | void |
3434 | _cpp_lex_file (pfile) | |
3435 | cpp_reader* pfile; | |
3436 | { | |
c5a04734 ZW |
3437 | cpp_toklist* list; |
3438 | ||
3439 | init_trigraph_map (); | |
3440 | list = (cpp_toklist *) xmalloc (sizeof (cpp_toklist)); | |
15dad1d9 | 3441 | _cpp_init_toklist (list); |
c5a04734 | 3442 | |
15dad1d9 | 3443 | for (;;) |
c5a04734 | 3444 | { |
c5a04734 ZW |
3445 | _cpp_lex_line (pfile, list); |
3446 | if (list->tokens[0].type == CPP_EOF) | |
3447 | break; | |
3448 | ||
15dad1d9 ZW |
3449 | #if 0 |
3450 | if (list->dirno) | |
3451 | _cpp_handle_directive (pfile, list); | |
c5a04734 | 3452 | else |
15dad1d9 | 3453 | #endif |
c5a04734 | 3454 | _cpp_output_list (pfile, list); |
15dad1d9 | 3455 | _cpp_clear_toklist (list); |
c5a04734 ZW |
3456 | } |
3457 | } | |
3458 | ||
b8f41010 | 3459 | /* Temporary function for illustrative purposes. */ |
c5a04734 ZW |
3460 | static void |
3461 | _cpp_output_list (pfile, list) | |
3462 | cpp_reader *pfile; | |
3463 | cpp_toklist *list; | |
3464 | { | |
5d7ee2fa | 3465 | cpp_token *token, *comment, *comment_before = 0; |
c5a04734 ZW |
3466 | |
3467 | if (list->comments_used > 0) | |
5d7ee2fa NB |
3468 | { |
3469 | comment = &list->comments[0]; | |
3470 | comment_before = &list->tokens[comment->aux]; | |
3471 | } | |
c5a04734 | 3472 | |
3fef5b2b NB |
3473 | token = &list->tokens[0]; |
3474 | do | |
c5a04734 | 3475 | { |
3fef5b2b | 3476 | /* Output comments if -C. */ |
5d7ee2fa | 3477 | while (token == comment_before) |
c5a04734 | 3478 | { |
5d7ee2fa NB |
3479 | /* Make space for the comment, and copy it out. */ |
3480 | CPP_RESERVE (pfile, TOKEN_LEN (comment)); | |
f617b8e2 | 3481 | pfile->limit = spell_token (pfile, comment, pfile->limit, 0); |
5d7ee2fa NB |
3482 | |
3483 | /* Stop if no comments left, or no more comments appear | |
3484 | before the current token. */ | |
3485 | comment++; | |
3486 | if (comment == list->comments + list->comments_used) | |
3487 | break; | |
3488 | comment_before = &list->tokens[comment->aux]; | |
c5a04734 ZW |
3489 | } |
3490 | ||
3fef5b2b | 3491 | CPP_RESERVE (pfile, TOKEN_LEN (token)); |
f617b8e2 | 3492 | pfile->limit = spell_token (pfile, token, pfile->limit, 1); |
c5a04734 | 3493 | } |
3fef5b2b | 3494 | while (token++->type != CPP_VSPACE); |
c5a04734 ZW |
3495 | } |
3496 | ||
3497 | #endif |