]>
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 ZW |
57 | static void expand_token_space PARAMS ((cpp_toklist *)); |
58 | static void init_token_list PARAMS ((cpp_reader *, cpp_toklist *, int)); | |
9e62c811 ZW |
59 | static void pedantic_whitespace PARAMS ((cpp_reader *, U_CHAR *, |
60 | unsigned int)); | |
f2d5f0cc | 61 | |
c5a04734 ZW |
62 | #define auto_expand_name_space(list) \ |
63 | expand_name_space ((list), (list)->name_cap / 2) | |
64 | ||
b8f41010 NB |
65 | #ifdef NEW_LEXER |
66 | ||
67 | static void expand_comment_space PARAMS ((cpp_toklist *)); | |
68 | void init_trigraph_map PARAMS ((void)); | |
69 | static unsigned char* trigraph_replace PARAMS ((cpp_reader *, unsigned char *, | |
70 | unsigned char *)); | |
71 | static const unsigned char *backslash_start PARAMS ((cpp_reader *, | |
72 | const unsigned char *)); | |
73 | static int skip_block_comment2 PARAMS ((cpp_reader *)); | |
74 | static int skip_line_comment2 PARAMS ((cpp_reader *)); | |
75 | static void skip_whitespace PARAMS ((cpp_reader *, int)); | |
76 | static void parse_name PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
77 | static void parse_number PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *)); | |
78 | static void parse_string2 PARAMS ((cpp_reader *, cpp_toklist *, cpp_name *, | |
79 | unsigned int)); | |
80 | static int trigraph_ok PARAMS ((cpp_reader *, const unsigned char *)); | |
81 | static void save_comment PARAMS ((cpp_toklist *, const unsigned char *, | |
82 | unsigned int, unsigned int, unsigned int)); | |
83 | void _cpp_lex_line PARAMS ((cpp_reader *, cpp_toklist *)); | |
84 | ||
85 | static void _cpp_output_list PARAMS ((cpp_reader *, cpp_toklist *)); | |
86 | ||
87 | unsigned int spell_string PARAMS ((unsigned char *, cpp_toklist *, | |
88 | cpp_token *token)); | |
89 | unsigned int spell_comment PARAMS ((unsigned char *, cpp_toklist *, | |
90 | cpp_token *token)); | |
91 | unsigned int spell_name PARAMS ((unsigned char *, cpp_toklist *, | |
92 | cpp_token *token)); | |
93 | ||
94 | typedef unsigned int (* speller) PARAMS ((unsigned char *, cpp_toklist *, | |
95 | cpp_token *)); | |
96 | ||
97 | /* Macros on a cpp_name. */ | |
98 | #define INIT_NAME(list, name) \ | |
99 | do {(name).len = 0; (name).offset = (list)->name_used;} while (0) | |
100 | ||
101 | #define IS_DIRECTIVE(list) (TOK_TYPE (list, 0) == CPP_HASH) | |
102 | #define COLUMN(cur) ((cur) - buffer->line_base) | |
103 | ||
104 | /* Maybe put these in the ISTABLE eventually. */ | |
105 | #define IS_HSPACE(c) ((c) == ' ' || (c) == '\t') | |
106 | #define IS_NEWLINE(c) ((c) == '\n' || (c) == '\r') | |
107 | ||
108 | /* Handle LF, CR, CR-LF and LF-CR style newlines. Assumes next | |
109 | character, if any, is in buffer. */ | |
110 | #define handle_newline(cur, limit, c) \ | |
111 | do {\ | |
112 | if ((cur) < (limit) && *(cur) == '\r' + '\n' - c) \ | |
113 | (cur)++; \ | |
114 | CPP_BUMP_LINE_CUR (pfile, (cur)); \ | |
115 | } while (0) | |
116 | ||
117 | #define IMMED_TOKEN() (!(cur_token->flags & PREV_WHITESPACE)) | |
118 | #define PREV_TOKEN_TYPE (cur_token[-1].type) | |
119 | ||
120 | #define SPELL_TEXT 0 | |
121 | #define SPELL_HANDLER 1 | |
122 | #define SPELL_CHAR 2 | |
123 | #define SPELL_NONE 3 | |
124 | #define SPELL_EOL 4 | |
125 | ||
126 | #define T(e, s) {SPELL_TEXT, s}, | |
127 | #define H(e, s) {SPELL_HANDLER, (PTR) s}, | |
128 | #define C(e, s) {SPELL_CHAR, s}, | |
129 | #define N(e, s) {SPELL_NONE, s}, | |
130 | #define E(e, s) {SPELL_EOL, s}, | |
131 | ||
132 | static const struct token_spelling | |
133 | { | |
134 | unsigned char type; | |
135 | PTR speller; | |
136 | } token_spellings [N_TTYPES + 1] = {TTYPE_TABLE {0, 0} }; | |
137 | ||
138 | #undef T | |
139 | #undef H | |
140 | #undef C | |
141 | #undef N | |
142 | #undef E | |
143 | ||
144 | #define PUSH_TOKEN(ttype) cur_token++->type = ttype | |
145 | #define REVISE_TOKEN(ttype) cur_token[-1].type = ttype | |
146 | #define BACKUP_TOKEN(ttype) (--cur_token)->type = ttype | |
147 | #define BACKUP_DIGRAPH(ttype) do { \ | |
148 | BACKUP_TOKEN(ttype); cur_token->flags |= DIGRAPH;} while (0) | |
149 | ||
150 | /* If there is this many bytes in a buffer, you have enough room to | |
151 | spell the token, not including preceding whitespace. */ | |
152 | #define TOKEN_LEN(token) (4 + (token_spellings[token->type].type == \ | |
153 | SPELL_HANDLER ? token->val.name.len: 0)) | |
154 | ||
155 | #endif | |
156 | ||
45b966db ZW |
157 | /* Re-allocates PFILE->token_buffer so it will hold at least N more chars. */ |
158 | ||
159 | void | |
160 | _cpp_grow_token_buffer (pfile, n) | |
161 | cpp_reader *pfile; | |
162 | long n; | |
163 | { | |
164 | long old_written = CPP_WRITTEN (pfile); | |
165 | pfile->token_buffer_size = n + 2 * pfile->token_buffer_size; | |
166 | pfile->token_buffer = (U_CHAR *) | |
167 | xrealloc(pfile->token_buffer, pfile->token_buffer_size); | |
168 | CPP_SET_WRITTEN (pfile, old_written); | |
169 | } | |
170 | ||
45b966db ZW |
171 | /* Allocate a new cpp_buffer for PFILE, and push it on the input buffer stack. |
172 | If BUFFER != NULL, then use the LENGTH characters in BUFFER | |
173 | as the new input buffer. | |
174 | Return the new buffer, or NULL on failure. */ | |
175 | ||
176 | cpp_buffer * | |
177 | cpp_push_buffer (pfile, buffer, length) | |
178 | cpp_reader *pfile; | |
179 | const U_CHAR *buffer; | |
180 | long length; | |
181 | { | |
182 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
183 | cpp_buffer *new; | |
184 | if (++pfile->buffer_stack_depth == CPP_STACK_MAX) | |
185 | { | |
186 | cpp_fatal (pfile, "macro or `#include' recursion too deep"); | |
187 | return NULL; | |
188 | } | |
189 | ||
190 | new = (cpp_buffer *) xcalloc (1, sizeof (cpp_buffer)); | |
191 | ||
192 | new->if_stack = pfile->if_stack; | |
45b966db | 193 | new->buf = new->cur = buffer; |
ff2b53ef | 194 | new->rlimit = buffer + length; |
45b966db | 195 | new->prev = buf; |
ff2b53ef | 196 | new->mark = NULL; |
45b966db ZW |
197 | new->line_base = NULL; |
198 | ||
199 | CPP_BUFFER (pfile) = new; | |
200 | return new; | |
201 | } | |
202 | ||
203 | cpp_buffer * | |
204 | cpp_pop_buffer (pfile) | |
205 | cpp_reader *pfile; | |
206 | { | |
207 | cpp_buffer *buf = CPP_BUFFER (pfile); | |
208 | if (ACTIVE_MARK_P (pfile)) | |
209 | cpp_ice (pfile, "mark active in cpp_pop_buffer"); | |
c56c2073 ZW |
210 | |
211 | if (buf->ihash) | |
212 | { | |
213 | _cpp_unwind_if_stack (pfile, buf); | |
214 | if (buf->buf) | |
215 | free ((PTR) buf->buf); | |
216 | if (pfile->system_include_depth) | |
217 | pfile->system_include_depth--; | |
218 | if (pfile->potential_control_macro) | |
219 | { | |
220 | buf->ihash->control_macro = pfile->potential_control_macro; | |
221 | pfile->potential_control_macro = 0; | |
222 | } | |
223 | pfile->input_stack_listing_current = 0; | |
224 | } | |
225 | else if (buf->macro) | |
226 | { | |
227 | HASHNODE *m = buf->macro; | |
228 | ||
229 | m->disabled = 0; | |
230 | if ((m->type == T_FMACRO && buf->mapped) | |
231 | || m->type == T_SPECLINE || m->type == T_FILE | |
232 | || m->type == T_BASE_FILE || m->type == T_INCLUDE_LEVEL | |
233 | || m->type == T_STDC) | |
234 | free ((PTR) buf->buf); | |
235 | } | |
45b966db ZW |
236 | CPP_BUFFER (pfile) = CPP_PREV_BUFFER (buf); |
237 | free (buf); | |
238 | pfile->buffer_stack_depth--; | |
239 | return CPP_BUFFER (pfile); | |
240 | } | |
241 | ||
f2d5f0cc ZW |
242 | /* Deal with the annoying semantics of fwrite. */ |
243 | static void | |
244 | safe_fwrite (pfile, buf, len, fp) | |
245 | cpp_reader *pfile; | |
246 | const U_CHAR *buf; | |
247 | size_t len; | |
248 | FILE *fp; | |
249 | { | |
250 | size_t count; | |
45b966db | 251 | |
f2d5f0cc ZW |
252 | while (len) |
253 | { | |
254 | count = fwrite (buf, 1, len, fp); | |
255 | if (count == 0) | |
256 | goto error; | |
257 | len -= count; | |
258 | buf += count; | |
259 | } | |
260 | return; | |
261 | ||
262 | error: | |
263 | cpp_notice_from_errno (pfile, CPP_OPTION (pfile, out_fname)); | |
264 | } | |
265 | ||
266 | /* Notify the compiler proper that the current line number has jumped, | |
267 | or the current file name has changed. */ | |
268 | ||
269 | static void | |
1368ee70 | 270 | output_line_command (pfile, print, line) |
45b966db | 271 | cpp_reader *pfile; |
f2d5f0cc | 272 | cpp_printer *print; |
1368ee70 | 273 | unsigned int line; |
45b966db | 274 | { |
1368ee70 | 275 | cpp_buffer *ip = cpp_file_buffer (pfile); |
f2d5f0cc ZW |
276 | enum { same = 0, enter, leave, rname } change; |
277 | static const char * const codes[] = { "", " 1", " 2", "" }; | |
278 | ||
279 | if (CPP_OPTION (pfile, no_line_commands)) | |
280 | return; | |
281 | ||
f2d5f0cc ZW |
282 | /* Determine whether the current filename has changed, and if so, |
283 | how. 'nominal_fname' values are unique, so they can be compared | |
284 | by comparing pointers. */ | |
285 | if (ip->nominal_fname == print->last_fname) | |
286 | change = same; | |
287 | else | |
45b966db | 288 | { |
f2d5f0cc ZW |
289 | if (pfile->buffer_stack_depth == print->last_bsd) |
290 | change = rname; | |
291 | else | |
45b966db | 292 | { |
f2d5f0cc ZW |
293 | if (pfile->buffer_stack_depth > print->last_bsd) |
294 | change = enter; | |
295 | else | |
296 | change = leave; | |
297 | print->last_bsd = pfile->buffer_stack_depth; | |
45b966db | 298 | } |
f2d5f0cc | 299 | print->last_fname = ip->nominal_fname; |
45b966db | 300 | } |
f2d5f0cc ZW |
301 | /* If the current file has not changed, we can output a few newlines |
302 | instead if we want to increase the line number by a small amount. | |
303 | We cannot do this if print->lineno is zero, because that means we | |
304 | haven't output any line commands yet. (The very first line | |
305 | command output is a `same_file' command.) */ | |
306 | if (change == same && print->lineno != 0 | |
307 | && line >= print->lineno && line < print->lineno + 8) | |
45b966db | 308 | { |
f2d5f0cc | 309 | while (line > print->lineno) |
45b966db | 310 | { |
f2d5f0cc ZW |
311 | putc ('\n', print->outf); |
312 | print->lineno++; | |
45b966db | 313 | } |
f2d5f0cc | 314 | return; |
45b966db | 315 | } |
f2d5f0cc ZW |
316 | |
317 | #ifndef NO_IMPLICIT_EXTERN_C | |
318 | if (CPP_OPTION (pfile, cplusplus)) | |
319 | fprintf (print->outf, "# %u \"%s\"%s%s%s\n", line, ip->nominal_fname, | |
320 | codes[change], | |
321 | ip->system_header_p ? " 3" : "", | |
322 | (ip->system_header_p == 2) ? " 4" : ""); | |
323 | else | |
324 | #endif | |
325 | fprintf (print->outf, "# %u \"%s\"%s%s\n", line, ip->nominal_fname, | |
326 | codes[change], | |
327 | ip->system_header_p ? " 3" : ""); | |
328 | print->lineno = line; | |
329 | } | |
330 | ||
331 | /* Write the contents of the token_buffer to the output stream, and | |
332 | clear the token_buffer. Also handles generating line commands and | |
333 | keeping track of file transitions. */ | |
334 | ||
335 | void | |
336 | cpp_output_tokens (pfile, print) | |
337 | cpp_reader *pfile; | |
338 | cpp_printer *print; | |
339 | { | |
1368ee70 ZW |
340 | cpp_buffer *ip; |
341 | ||
f6fab919 ZW |
342 | if (CPP_WRITTEN (pfile) - print->written) |
343 | { | |
344 | if (CPP_PWRITTEN (pfile)[-1] == '\n' && print->lineno) | |
345 | print->lineno++; | |
346 | safe_fwrite (pfile, pfile->token_buffer, | |
347 | CPP_WRITTEN (pfile) - print->written, print->outf); | |
348 | } | |
1368ee70 ZW |
349 | |
350 | ip = cpp_file_buffer (pfile); | |
351 | if (ip) | |
352 | output_line_command (pfile, print, CPP_BUF_LINE (ip)); | |
353 | ||
f2d5f0cc | 354 | CPP_SET_WRITTEN (pfile, print->written); |
45b966db ZW |
355 | } |
356 | ||
1368ee70 ZW |
357 | /* Helper for cpp_output_list - increases the column number to match |
358 | what we expect it to be. */ | |
359 | ||
360 | static void | |
361 | bump_column (print, from, to) | |
362 | cpp_printer *print; | |
363 | unsigned int from, to; | |
364 | { | |
365 | unsigned int tabs, spcs; | |
366 | unsigned int delta = to - from; | |
367 | ||
368 | /* Only if FROM is 0, advance by tabs. */ | |
369 | if (from == 0) | |
370 | tabs = delta / 8, spcs = delta % 8; | |
371 | else | |
372 | tabs = 0, spcs = delta; | |
373 | ||
374 | while (tabs--) putc ('\t', print->outf); | |
375 | while (spcs--) putc (' ', print->outf); | |
376 | } | |
377 | ||
378 | /* Write out the list L onto pfile->token_buffer. This function is | |
379 | incomplete: | |
380 | ||
381 | 1) pfile->token_buffer is not going to continue to exist. | |
382 | 2) At the moment, tokens don't carry the information described | |
383 | in cpplib.h; they are all strings. | |
384 | 3) The list has to be a complete line, and has to be written starting | |
385 | at the beginning of a line. */ | |
386 | ||
387 | void | |
388 | cpp_output_list (pfile, print, list) | |
389 | cpp_reader *pfile; | |
390 | cpp_printer *print; | |
391 | const cpp_toklist *list; | |
392 | { | |
393 | unsigned int i; | |
394 | unsigned int curcol = 1; | |
395 | ||
396 | /* XXX Probably does not do what is intended. */ | |
397 | if (print->lineno != list->line) | |
398 | output_line_command (pfile, print, list->line); | |
399 | ||
400 | for (i = 0; i < list->tokens_used; i++) | |
401 | { | |
1920de47 | 402 | if (TOK_TYPE (list, i) == CPP_VSPACE) |
1368ee70 ZW |
403 | { |
404 | output_line_command (pfile, print, list->tokens[i].aux); | |
405 | continue; | |
406 | } | |
407 | ||
1920de47 | 408 | if (curcol < TOK_COL (list, i)) |
1368ee70 ZW |
409 | { |
410 | /* Insert space to bring the column to what it should be. */ | |
1920de47 ZW |
411 | bump_column (print, curcol - 1, TOK_COL (list, i)); |
412 | curcol = TOK_COL (list, i); | |
1368ee70 ZW |
413 | } |
414 | /* XXX We may have to insert space to prevent an accidental | |
415 | token paste. */ | |
1920de47 ZW |
416 | safe_fwrite (pfile, TOK_NAME (list, i), TOK_LEN (list, i), print->outf); |
417 | curcol += TOK_LEN (list, i); | |
1368ee70 ZW |
418 | } |
419 | } | |
420 | ||
f2d5f0cc ZW |
421 | /* Scan a string (which may have escape marks), perform macro expansion, |
422 | and write the result to the token_buffer. */ | |
45b966db ZW |
423 | |
424 | void | |
f2d5f0cc | 425 | _cpp_expand_to_buffer (pfile, buf, length) |
45b966db ZW |
426 | cpp_reader *pfile; |
427 | const U_CHAR *buf; | |
428 | int length; | |
429 | { | |
c56c2073 | 430 | cpp_buffer *stop; |
f2d5f0cc | 431 | enum cpp_ttype token; |
f6fab919 | 432 | U_CHAR *buf1; |
45b966db ZW |
433 | |
434 | if (length < 0) | |
435 | { | |
436 | cpp_ice (pfile, "length < 0 in cpp_expand_to_buffer"); | |
437 | return; | |
438 | } | |
439 | ||
f6fab919 ZW |
440 | /* Copy the buffer, because it might be in an unsafe place - for |
441 | example, a sequence on the token_buffer, where the pointers will | |
442 | be invalidated if we enlarge the token_buffer. */ | |
443 | buf1 = alloca (length); | |
444 | memcpy (buf1, buf, length); | |
445 | ||
45b966db | 446 | /* Set up the input on the input stack. */ |
c56c2073 ZW |
447 | stop = CPP_BUFFER (pfile); |
448 | if (cpp_push_buffer (pfile, buf1, length) == NULL) | |
45b966db | 449 | return; |
c56c2073 | 450 | CPP_BUFFER (pfile)->has_escapes = 1; |
45b966db ZW |
451 | |
452 | /* Scan the input, create the output. */ | |
f2d5f0cc ZW |
453 | for (;;) |
454 | { | |
455 | token = cpp_get_token (pfile); | |
c56c2073 | 456 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 457 | break; |
f2d5f0cc | 458 | } |
45b966db ZW |
459 | } |
460 | ||
c56c2073 | 461 | /* Scan until CPP_BUFFER (PFILE) is exhausted, discarding output. */ |
f2d5f0cc ZW |
462 | |
463 | void | |
464 | cpp_scan_buffer_nooutput (pfile) | |
465 | cpp_reader *pfile; | |
466 | { | |
c56c2073 | 467 | cpp_buffer *stop = CPP_PREV_BUFFER (CPP_BUFFER (pfile)); |
f2d5f0cc ZW |
468 | enum cpp_ttype token; |
469 | unsigned int old_written = CPP_WRITTEN (pfile); | |
470 | /* In no-output mode, we can ignore everything but directives. */ | |
471 | for (;;) | |
472 | { | |
473 | if (! pfile->only_seen_white) | |
474 | _cpp_skip_rest_of_line (pfile); | |
475 | token = cpp_get_token (pfile); | |
c56c2073 | 476 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 477 | break; |
f2d5f0cc ZW |
478 | } |
479 | CPP_SET_WRITTEN (pfile, old_written); | |
480 | } | |
481 | ||
c56c2073 | 482 | /* Scan until CPP_BUFFER (pfile) is exhausted, writing output to PRINT. */ |
f2d5f0cc ZW |
483 | |
484 | void | |
485 | cpp_scan_buffer (pfile, print) | |
486 | cpp_reader *pfile; | |
487 | cpp_printer *print; | |
488 | { | |
c56c2073 | 489 | cpp_buffer *stop = CPP_PREV_BUFFER (CPP_BUFFER (pfile)); |
f2d5f0cc ZW |
490 | enum cpp_ttype token; |
491 | ||
492 | for (;;) | |
493 | { | |
494 | token = cpp_get_token (pfile); | |
c56c2073 | 495 | if (token == CPP_EOF || token == CPP_VSPACE |
f2d5f0cc ZW |
496 | /* XXX Temporary kluge - force flush after #include only */ |
497 | || (token == CPP_DIRECTIVE | |
498 | && CPP_BUFFER (pfile)->nominal_fname != print->last_fname)) | |
499 | { | |
500 | cpp_output_tokens (pfile, print); | |
c56c2073 | 501 | if (token == CPP_EOF && CPP_BUFFER (pfile) == stop) |
f2d5f0cc | 502 | return; |
f2d5f0cc ZW |
503 | } |
504 | } | |
505 | } | |
506 | ||
45b966db ZW |
507 | /* Return the topmost cpp_buffer that corresponds to a file (not a macro). */ |
508 | ||
509 | cpp_buffer * | |
510 | cpp_file_buffer (pfile) | |
511 | cpp_reader *pfile; | |
512 | { | |
513 | cpp_buffer *ip; | |
514 | ||
515 | for (ip = CPP_BUFFER (pfile); ip; ip = CPP_PREV_BUFFER (ip)) | |
516 | if (ip->ihash != NULL) | |
517 | return ip; | |
518 | return NULL; | |
519 | } | |
520 | ||
1368ee70 ZW |
521 | /* Token-buffer helper functions. */ |
522 | ||
523 | /* Expand a token list's string space. */ | |
524 | static void | |
c5a04734 | 525 | expand_name_space (list, len) |
1368ee70 | 526 | cpp_toklist *list; |
c5a04734 ZW |
527 | unsigned int len; |
528 | { | |
529 | list->name_cap += len; | |
530 | list->namebuf = (unsigned char *) xrealloc (list->namebuf, list->name_cap); | |
1368ee70 ZW |
531 | } |
532 | ||
533 | /* Expand the number of tokens in a list. */ | |
534 | static void | |
535 | expand_token_space (list) | |
536 | cpp_toklist *list; | |
537 | { | |
538 | list->tokens_cap *= 2; | |
539 | list->tokens = (cpp_token *) | |
c5a04734 ZW |
540 | xrealloc (list->tokens - 1, (list->tokens_cap + 1) * sizeof (cpp_token)); |
541 | list->tokens++; /* Skip the dummy. */ | |
1368ee70 ZW |
542 | } |
543 | ||
c5a04734 ZW |
544 | /* Initialize a token list. We allocate an extra token in front of |
545 | the token list, as this allows us to always peek at the previous | |
546 | token without worrying about underflowing the list. */ | |
1368ee70 ZW |
547 | static void |
548 | init_token_list (pfile, list, recycle) | |
549 | cpp_reader *pfile; | |
550 | cpp_toklist *list; | |
551 | int recycle; | |
552 | { | |
c5a04734 ZW |
553 | /* Recycling a used list saves 3 free-malloc pairs. */ |
554 | if (!recycle) | |
1368ee70 | 555 | { |
c5a04734 ZW |
556 | /* Initialize token space. Put a dummy token before the start |
557 | that will fail matches. */ | |
558 | list->tokens_cap = 256; /* 4K's worth. */ | |
1368ee70 | 559 | list->tokens = (cpp_token *) |
c5a04734 ZW |
560 | xmalloc ((list->tokens_cap + 1) * sizeof (cpp_token)); |
561 | list->tokens[0].type = CPP_EOF; | |
562 | list->tokens++; | |
1368ee70 | 563 | |
c5a04734 | 564 | /* Initialize name space. */ |
1368ee70 | 565 | list->name_cap = 1024; |
1368ee70 | 566 | list->namebuf = (unsigned char *) xmalloc (list->name_cap); |
c5a04734 ZW |
567 | |
568 | /* Only create a comment space on demand. */ | |
569 | list->comments_cap = 0; | |
570 | list->comments = 0; | |
1368ee70 ZW |
571 | } |
572 | ||
c5a04734 ZW |
573 | list->tokens_used = 0; |
574 | list->name_used = 0; | |
575 | list->comments_used = 0; | |
9e62c811 ZW |
576 | if (pfile->buffer) |
577 | list->line = pfile->buffer->lineno; | |
1368ee70 ZW |
578 | list->dir_handler = 0; |
579 | list->dir_flags = 0; | |
580 | } | |
581 | ||
582 | /* Scan an entire line and create a token list for it. Does not | |
583 | macro-expand or execute directives. */ | |
584 | ||
585 | void | |
586 | _cpp_scan_line (pfile, list) | |
587 | cpp_reader *pfile; | |
588 | cpp_toklist *list; | |
589 | { | |
590 | int i, col; | |
591 | long written, len; | |
592 | enum cpp_ttype type; | |
9e62c811 | 593 | int space_before; |
1368ee70 ZW |
594 | |
595 | init_token_list (pfile, list, 1); | |
596 | ||
597 | written = CPP_WRITTEN (pfile); | |
598 | i = 0; | |
9e62c811 | 599 | space_before = 0; |
1368ee70 ZW |
600 | for (;;) |
601 | { | |
602 | col = CPP_BUFFER (pfile)->cur - CPP_BUFFER (pfile)->line_base; | |
603 | type = _cpp_lex_token (pfile); | |
604 | len = CPP_WRITTEN (pfile) - written; | |
605 | CPP_SET_WRITTEN (pfile, written); | |
606 | if (type == CPP_HSPACE) | |
9e62c811 ZW |
607 | { |
608 | if (CPP_PEDANTIC (pfile)) | |
609 | pedantic_whitespace (pfile, pfile->token_buffer + written, len); | |
610 | space_before = 1; | |
611 | continue; | |
612 | } | |
0f89df67 ZW |
613 | else if (type == CPP_COMMENT) |
614 | /* Only happens when processing -traditional macro definitions. | |
615 | Do not give this a token entry, but do not change space_before | |
616 | either. */ | |
617 | continue; | |
1368ee70 ZW |
618 | |
619 | if (list->tokens_used >= list->tokens_cap) | |
620 | expand_token_space (list); | |
621 | if (list->name_used + len >= list->name_cap) | |
bb1ec1d7 | 622 | expand_name_space (list, list->name_used + len + 1 - list->name_cap); |
1368ee70 | 623 | |
9e62c811 ZW |
624 | if (type == CPP_MACRO) |
625 | type = CPP_NAME; | |
626 | ||
1368ee70 | 627 | list->tokens_used++; |
1920de47 ZW |
628 | TOK_TYPE (list, i) = type; |
629 | TOK_COL (list, i) = col; | |
630 | TOK_FLAGS (list, i) = space_before ? PREV_WHITESPACE : 0; | |
9e62c811 | 631 | |
1368ee70 ZW |
632 | if (type == CPP_VSPACE) |
633 | break; | |
634 | ||
1920de47 ZW |
635 | TOK_LEN (list, i) = len; |
636 | TOK_OFFSET (list, i) = list->name_used; | |
637 | memcpy (TOK_NAME (list, i), CPP_PWRITTEN (pfile), len); | |
1368ee70 ZW |
638 | list->name_used += len; |
639 | i++; | |
9e62c811 | 640 | space_before = 0; |
1368ee70 | 641 | } |
1920de47 | 642 | TOK_AUX (list, i) = CPP_BUFFER (pfile)->lineno + 1; |
9e62c811 ZW |
643 | |
644 | /* XXX Temporary kluge: put back the newline. */ | |
645 | FORWARD(-1); | |
1368ee70 ZW |
646 | } |
647 | ||
648 | ||
45b966db ZW |
649 | /* Skip a C-style block comment. We know it's a comment, and point is |
650 | at the second character of the starter. */ | |
651 | static void | |
652 | skip_block_comment (pfile) | |
653 | cpp_reader *pfile; | |
654 | { | |
3a2b2c7a | 655 | unsigned int line, col; |
61474454 | 656 | const U_CHAR *limit, *cur; |
45b966db ZW |
657 | |
658 | FORWARD(1); | |
3a2b2c7a ZW |
659 | line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
660 | col = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
61474454 NB |
661 | limit = CPP_BUFFER (pfile)->rlimit; |
662 | cur = CPP_BUFFER (pfile)->cur; | |
663 | ||
664 | while (cur < limit) | |
45b966db | 665 | { |
61474454 NB |
666 | char c = *cur++; |
667 | if (c == '\n' || c == '\r') | |
45b966db ZW |
668 | { |
669 | /* \r cannot be a macro escape marker here. */ | |
670 | if (!ACTIVE_MARK_P (pfile)) | |
61474454 NB |
671 | CPP_BUMP_LINE_CUR (pfile, cur); |
672 | } | |
673 | else if (c == '*') | |
674 | { | |
675 | /* Check for teminator. */ | |
676 | if (cur < limit && *cur == '/') | |
677 | goto out; | |
678 | ||
679 | /* Warn about comment starter embedded in comment. */ | |
680 | if (cur[-2] == '/' && CPP_OPTION (pfile, warn_comments)) | |
681 | cpp_warning_with_line (pfile, CPP_BUFFER (pfile)->lineno, | |
682 | cur - CPP_BUFFER (pfile)->line_base, | |
683 | "'/*' within comment"); | |
45b966db | 684 | } |
45b966db | 685 | } |
61474454 NB |
686 | |
687 | cpp_error_with_line (pfile, line, col, "unterminated comment"); | |
688 | cur--; | |
689 | out: | |
690 | CPP_BUFFER (pfile)->cur = cur + 1; | |
45b966db ZW |
691 | } |
692 | ||
693 | /* Skip a C++/Chill line comment. We know it's a comment, and point | |
694 | is at the second character of the initiator. */ | |
695 | static void | |
696 | skip_line_comment (pfile) | |
697 | cpp_reader *pfile; | |
698 | { | |
699 | FORWARD(1); | |
700 | for (;;) | |
701 | { | |
702 | int c = GETC (); | |
703 | ||
704 | /* We don't have to worry about EOF in here. */ | |
705 | if (c == '\n') | |
706 | { | |
707 | /* Don't consider final '\n' to be part of comment. */ | |
708 | FORWARD(-1); | |
709 | return; | |
710 | } | |
711 | else if (c == '\r') | |
712 | { | |
713 | /* \r cannot be a macro escape marker here. */ | |
714 | if (!ACTIVE_MARK_P (pfile)) | |
715 | CPP_BUMP_LINE (pfile); | |
ae79697b | 716 | if (CPP_OPTION (pfile, warn_comments)) |
45b966db ZW |
717 | cpp_warning (pfile, "backslash-newline within line comment"); |
718 | } | |
719 | } | |
720 | } | |
721 | ||
722 | /* Skip a comment - C, C++, or Chill style. M is the first character | |
723 | of the comment marker. If this really is a comment, skip to its | |
724 | end and return ' '. If this is not a comment, return M (which will | |
725 | be '/' or '-'). */ | |
726 | ||
727 | static int | |
728 | skip_comment (pfile, m) | |
729 | cpp_reader *pfile; | |
730 | int m; | |
731 | { | |
732 | if (m == '/' && PEEKC() == '*') | |
733 | { | |
734 | skip_block_comment (pfile); | |
735 | return ' '; | |
736 | } | |
737 | else if (m == '/' && PEEKC() == '/') | |
738 | { | |
739 | if (CPP_BUFFER (pfile)->system_header_p) | |
740 | { | |
741 | /* We silently allow C++ comments in system headers, irrespective | |
742 | of conformance mode, because lots of busted systems do that | |
743 | and trying to clean it up in fixincludes is a nightmare. */ | |
744 | skip_line_comment (pfile); | |
745 | return ' '; | |
746 | } | |
ae79697b | 747 | else if (CPP_OPTION (pfile, cplusplus_comments)) |
45b966db | 748 | { |
0f89df67 | 749 | if (! CPP_BUFFER (pfile)->warned_cplusplus_comments) |
45b966db | 750 | { |
0f89df67 ZW |
751 | if (CPP_WTRADITIONAL (pfile)) |
752 | cpp_pedwarn (pfile, | |
753 | "C++ style comments are not allowed in traditional C"); | |
754 | else if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile)) | |
755 | cpp_pedwarn (pfile, | |
756 | "C++ style comments are not allowed in ISO C89"); | |
757 | if (CPP_WTRADITIONAL (pfile) | |
758 | || (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile))) | |
759 | cpp_pedwarn (pfile, | |
45b966db ZW |
760 | "(this will be reported only once per input file)"); |
761 | CPP_BUFFER (pfile)->warned_cplusplus_comments = 1; | |
762 | } | |
763 | skip_line_comment (pfile); | |
764 | return ' '; | |
765 | } | |
766 | else | |
767 | return m; | |
768 | } | |
769 | else if (m == '-' && PEEKC() == '-' | |
ae79697b | 770 | && CPP_OPTION (pfile, chill)) |
45b966db ZW |
771 | { |
772 | skip_line_comment (pfile); | |
773 | return ' '; | |
774 | } | |
775 | else | |
776 | return m; | |
777 | } | |
778 | ||
779 | /* Identical to skip_comment except that it copies the comment into the | |
780 | token_buffer. This is used if !discard_comments. */ | |
781 | static int | |
782 | copy_comment (pfile, m) | |
783 | cpp_reader *pfile; | |
784 | int m; | |
785 | { | |
786 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
787 | const U_CHAR *limit; | |
788 | ||
789 | if (skip_comment (pfile, m) == m) | |
790 | return m; | |
791 | ||
792 | limit = CPP_BUFFER (pfile)->cur; | |
793 | CPP_RESERVE (pfile, limit - start + 2); | |
794 | CPP_PUTC_Q (pfile, m); | |
795 | for (; start <= limit; start++) | |
796 | if (*start != '\r') | |
797 | CPP_PUTC_Q (pfile, *start); | |
798 | ||
799 | return ' '; | |
800 | } | |
801 | ||
64aaf407 NB |
802 | static void |
803 | null_warning (pfile, count) | |
804 | cpp_reader *pfile; | |
805 | unsigned int count; | |
806 | { | |
807 | if (count == 1) | |
808 | cpp_warning (pfile, "embedded null character ignored"); | |
809 | else | |
810 | cpp_warning (pfile, "embedded null characters ignored"); | |
811 | } | |
812 | ||
45b966db ZW |
813 | /* Skip whitespace \-newline and comments. Does not macro-expand. */ |
814 | ||
815 | void | |
816 | _cpp_skip_hspace (pfile) | |
817 | cpp_reader *pfile; | |
818 | { | |
64aaf407 | 819 | unsigned int null_count = 0; |
45b966db | 820 | int c; |
64aaf407 | 821 | |
45b966db ZW |
822 | while (1) |
823 | { | |
824 | c = GETC(); | |
825 | if (c == EOF) | |
64aaf407 | 826 | goto out; |
45b966db ZW |
827 | else if (is_hspace(c)) |
828 | { | |
829 | if ((c == '\f' || c == '\v') && CPP_PEDANTIC (pfile)) | |
830 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
831 | c == '\f' ? "formfeed" : "vertical tab"); | |
64aaf407 NB |
832 | else if (c == '\0') |
833 | null_count++; | |
45b966db ZW |
834 | } |
835 | else if (c == '\r') | |
836 | { | |
837 | /* \r is a backslash-newline marker if !has_escapes, and | |
838 | a deletable-whitespace or no-reexpansion marker otherwise. */ | |
839 | if (CPP_BUFFER (pfile)->has_escapes) | |
840 | { | |
841 | if (PEEKC() == ' ') | |
842 | FORWARD(1); | |
843 | else | |
844 | break; | |
845 | } | |
846 | else | |
847 | CPP_BUMP_LINE (pfile); | |
848 | } | |
849 | else if (c == '/' || c == '-') | |
850 | { | |
851 | c = skip_comment (pfile, c); | |
852 | if (c != ' ') | |
853 | break; | |
854 | } | |
855 | else | |
856 | break; | |
857 | } | |
858 | FORWARD(-1); | |
64aaf407 NB |
859 | out: |
860 | if (null_count) | |
861 | null_warning (pfile, null_count); | |
45b966db ZW |
862 | } |
863 | ||
864 | /* Read and discard the rest of the current line. */ | |
865 | ||
866 | void | |
867 | _cpp_skip_rest_of_line (pfile) | |
868 | cpp_reader *pfile; | |
869 | { | |
870 | for (;;) | |
871 | { | |
872 | int c = GETC(); | |
873 | switch (c) | |
874 | { | |
875 | case '\n': | |
876 | FORWARD(-1); | |
877 | case EOF: | |
878 | return; | |
879 | ||
880 | case '\r': | |
881 | if (! CPP_BUFFER (pfile)->has_escapes) | |
882 | CPP_BUMP_LINE (pfile); | |
883 | break; | |
884 | ||
885 | case '\'': | |
886 | case '\"': | |
887 | skip_string (pfile, c); | |
888 | break; | |
889 | ||
890 | case '/': | |
891 | case '-': | |
892 | skip_comment (pfile, c); | |
893 | break; | |
894 | ||
895 | case '\f': | |
896 | case '\v': | |
897 | if (CPP_PEDANTIC (pfile)) | |
898 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
899 | c == '\f' ? "formfeed" : "vertical tab"); | |
900 | break; | |
901 | ||
902 | } | |
903 | } | |
904 | } | |
905 | ||
906 | /* Parse an identifier starting with C. */ | |
907 | ||
908 | void | |
909 | _cpp_parse_name (pfile, c) | |
910 | cpp_reader *pfile; | |
911 | int c; | |
912 | { | |
913 | for (;;) | |
914 | { | |
915 | if (! is_idchar(c)) | |
916 | { | |
917 | FORWARD (-1); | |
918 | break; | |
919 | } | |
920 | ||
921 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
922 | cpp_pedwarn (pfile, "`$' in identifier"); | |
923 | ||
924 | CPP_RESERVE(pfile, 2); /* One more for final NUL. */ | |
925 | CPP_PUTC_Q (pfile, c); | |
926 | c = GETC(); | |
927 | if (c == EOF) | |
928 | break; | |
929 | } | |
45b966db ZW |
930 | return; |
931 | } | |
932 | ||
933 | /* Parse and skip over a string starting with C. A single quoted | |
934 | string is treated like a double -- some programs (e.g., troff) are | |
935 | perverse this way. (However, a single quoted string is not allowed | |
936 | to extend over multiple lines.) */ | |
937 | static void | |
938 | skip_string (pfile, c) | |
939 | cpp_reader *pfile; | |
940 | int c; | |
941 | { | |
3a2b2c7a | 942 | unsigned int start_line, start_column; |
64aaf407 | 943 | unsigned int null_count = 0; |
45b966db | 944 | |
3a2b2c7a ZW |
945 | start_line = CPP_BUF_LINE (CPP_BUFFER (pfile)); |
946 | start_column = CPP_BUF_COL (CPP_BUFFER (pfile)); | |
45b966db ZW |
947 | while (1) |
948 | { | |
949 | int cc = GETC(); | |
950 | switch (cc) | |
951 | { | |
952 | case EOF: | |
953 | cpp_error_with_line (pfile, start_line, start_column, | |
954 | "unterminated string or character constant"); | |
955 | if (pfile->multiline_string_line != start_line | |
956 | && pfile->multiline_string_line != 0) | |
957 | cpp_error_with_line (pfile, | |
958 | pfile->multiline_string_line, -1, | |
959 | "possible real start of unterminated constant"); | |
960 | pfile->multiline_string_line = 0; | |
64aaf407 | 961 | goto out; |
45b966db | 962 | |
64aaf407 NB |
963 | case '\0': |
964 | null_count++; | |
965 | break; | |
966 | ||
45b966db ZW |
967 | case '\n': |
968 | CPP_BUMP_LINE (pfile); | |
969 | /* In Fortran and assembly language, silently terminate | |
970 | strings of either variety at end of line. This is a | |
971 | kludge around not knowing where comments are in these | |
972 | languages. */ | |
ae79697b ZW |
973 | if (CPP_OPTION (pfile, lang_fortran) |
974 | || CPP_OPTION (pfile, lang_asm)) | |
45b966db ZW |
975 | { |
976 | FORWARD(-1); | |
64aaf407 | 977 | goto out; |
45b966db ZW |
978 | } |
979 | /* Character constants may not extend over multiple lines. | |
980 | In Standard C, neither may strings. We accept multiline | |
981 | strings as an extension. */ | |
982 | if (c == '\'') | |
983 | { | |
984 | cpp_error_with_line (pfile, start_line, start_column, | |
985 | "unterminated character constant"); | |
986 | FORWARD(-1); | |
64aaf407 | 987 | goto out; |
45b966db ZW |
988 | } |
989 | if (CPP_PEDANTIC (pfile) && pfile->multiline_string_line == 0) | |
990 | cpp_pedwarn_with_line (pfile, start_line, start_column, | |
991 | "string constant runs past end of line"); | |
992 | if (pfile->multiline_string_line == 0) | |
993 | pfile->multiline_string_line = start_line; | |
994 | break; | |
995 | ||
996 | case '\r': | |
997 | if (CPP_BUFFER (pfile)->has_escapes) | |
998 | { | |
999 | cpp_ice (pfile, "\\r escape inside string constant"); | |
1000 | FORWARD(1); | |
1001 | } | |
1002 | else | |
1003 | /* Backslash newline is replaced by nothing at all. */ | |
1004 | CPP_BUMP_LINE (pfile); | |
1005 | break; | |
1006 | ||
1007 | case '\\': | |
1008 | FORWARD(1); | |
1009 | break; | |
1010 | ||
1011 | case '\"': | |
1012 | case '\'': | |
1013 | if (cc == c) | |
64aaf407 | 1014 | goto out; |
45b966db ZW |
1015 | break; |
1016 | } | |
1017 | } | |
64aaf407 NB |
1018 | |
1019 | out: | |
1020 | if (null_count == 1) | |
1021 | cpp_warning (pfile, "null character in string or character constant"); | |
1022 | else if (null_count > 1) | |
1023 | cpp_warning (pfile, "null characters in string or character constant"); | |
45b966db ZW |
1024 | } |
1025 | ||
1026 | /* Parse a string and copy it to the output. */ | |
1027 | ||
1028 | static void | |
1029 | parse_string (pfile, c) | |
1030 | cpp_reader *pfile; | |
1031 | int c; | |
1032 | { | |
1033 | const U_CHAR *start = CPP_BUFFER (pfile)->cur; /* XXX Layering violation */ | |
1034 | const U_CHAR *limit; | |
1035 | ||
1036 | skip_string (pfile, c); | |
1037 | ||
1038 | limit = CPP_BUFFER (pfile)->cur; | |
1039 | CPP_RESERVE (pfile, limit - start + 2); | |
1040 | CPP_PUTC_Q (pfile, c); | |
1041 | for (; start < limit; start++) | |
1042 | if (*start != '\r') | |
1043 | CPP_PUTC_Q (pfile, *start); | |
1044 | } | |
1045 | ||
1046 | /* Read an assertion into the token buffer, converting to | |
1047 | canonical form: `#predicate(a n swe r)' The next non-whitespace | |
1048 | character to read should be the first letter of the predicate. | |
1049 | Returns 0 for syntax error, 1 for bare predicate, 2 for predicate | |
1050 | with answer (see callers for why). In case of 0, an error has been | |
1051 | printed. */ | |
1052 | int | |
1053 | _cpp_parse_assertion (pfile) | |
1054 | cpp_reader *pfile; | |
1055 | { | |
1056 | int c, dropwhite; | |
1057 | _cpp_skip_hspace (pfile); | |
1058 | c = PEEKC(); | |
bfab56e7 ZW |
1059 | if (c == '\n') |
1060 | { | |
1061 | cpp_error (pfile, "assertion without predicate"); | |
1062 | return 0; | |
1063 | } | |
1064 | else if (! is_idstart(c)) | |
45b966db ZW |
1065 | { |
1066 | cpp_error (pfile, "assertion predicate is not an identifier"); | |
1067 | return 0; | |
1068 | } | |
1069 | CPP_PUTC(pfile, '#'); | |
1070 | FORWARD(1); | |
1071 | _cpp_parse_name (pfile, c); | |
1072 | ||
1073 | c = PEEKC(); | |
1074 | if (c != '(') | |
1075 | { | |
1076 | if (is_hspace(c) || c == '\r') | |
1077 | _cpp_skip_hspace (pfile); | |
1078 | c = PEEKC(); | |
1079 | } | |
1080 | if (c != '(') | |
1081 | return 1; | |
1082 | ||
1083 | CPP_PUTC(pfile, '('); | |
1084 | FORWARD(1); | |
1085 | dropwhite = 1; | |
1086 | while ((c = GETC()) != ')') | |
1087 | { | |
1088 | if (is_space(c)) | |
1089 | { | |
1090 | if (! dropwhite) | |
1091 | { | |
1092 | CPP_PUTC(pfile, ' '); | |
1093 | dropwhite = 1; | |
1094 | } | |
1095 | } | |
1096 | else if (c == '\n' || c == EOF) | |
1097 | { | |
1098 | if (c == '\n') FORWARD(-1); | |
1099 | cpp_error (pfile, "un-terminated assertion answer"); | |
1100 | return 0; | |
1101 | } | |
1102 | else if (c == '\r') | |
1103 | /* \r cannot be a macro escape here. */ | |
1104 | CPP_BUMP_LINE (pfile); | |
1105 | else | |
1106 | { | |
1107 | CPP_PUTC (pfile, c); | |
1108 | dropwhite = 0; | |
1109 | } | |
1110 | } | |
1111 | ||
1112 | if (pfile->limit[-1] == ' ') | |
1113 | pfile->limit[-1] = ')'; | |
1114 | else if (pfile->limit[-1] == '(') | |
1115 | { | |
1116 | cpp_error (pfile, "empty token sequence in assertion"); | |
1117 | return 0; | |
1118 | } | |
1119 | else | |
1120 | CPP_PUTC (pfile, ')'); | |
1121 | ||
45b966db ZW |
1122 | return 2; |
1123 | } | |
1124 | ||
1125 | /* Get the next token, and add it to the text in pfile->token_buffer. | |
1126 | Return the kind of token we got. */ | |
1127 | ||
3a2b2c7a | 1128 | enum cpp_ttype |
45b966db ZW |
1129 | _cpp_lex_token (pfile) |
1130 | cpp_reader *pfile; | |
1131 | { | |
5eec0563 | 1132 | register int c, c2; |
3a2b2c7a | 1133 | enum cpp_ttype token; |
45b966db | 1134 | |
f2d5f0cc ZW |
1135 | if (CPP_BUFFER (pfile) == NULL) |
1136 | return CPP_EOF; | |
1137 | ||
45b966db ZW |
1138 | get_next: |
1139 | c = GETC(); | |
1140 | switch (c) | |
1141 | { | |
1142 | case EOF: | |
1143 | return CPP_EOF; | |
1144 | ||
1145 | case '/': | |
1146 | if (PEEKC () == '=') | |
1147 | goto op2; | |
1148 | ||
1149 | comment: | |
ae79697b | 1150 | if (CPP_OPTION (pfile, discard_comments)) |
45b966db ZW |
1151 | c = skip_comment (pfile, c); |
1152 | else | |
1153 | c = copy_comment (pfile, c); | |
1154 | if (c != ' ') | |
1155 | goto randomchar; | |
1156 | ||
1157 | /* Comments are equivalent to spaces. | |
1158 | For -traditional, a comment is equivalent to nothing. */ | |
ff2b53ef | 1159 | if (!CPP_OPTION (pfile, discard_comments)) |
45b966db | 1160 | return CPP_COMMENT; |
9e62c811 | 1161 | else if (CPP_TRADITIONAL (pfile)) |
0f89df67 ZW |
1162 | { |
1163 | if (pfile->parsing_define_directive) | |
1164 | return CPP_COMMENT; | |
1165 | goto get_next; | |
1166 | } | |
45b966db ZW |
1167 | else |
1168 | { | |
1169 | CPP_PUTC (pfile, c); | |
1170 | return CPP_HSPACE; | |
1171 | } | |
1172 | ||
1173 | case '#': | |
5eec0563 JM |
1174 | CPP_PUTC (pfile, c); |
1175 | ||
1176 | hash: | |
45b966db ZW |
1177 | if (pfile->parsing_if_directive) |
1178 | { | |
f2d5f0cc | 1179 | CPP_ADJUST_WRITTEN (pfile, -1); |
bfab56e7 ZW |
1180 | if (_cpp_parse_assertion (pfile)) |
1181 | return CPP_ASSERTION; | |
5eec0563 | 1182 | return CPP_OTHER; |
45b966db ZW |
1183 | } |
1184 | ||
9e62c811 | 1185 | if (pfile->parsing_define_directive) |
45b966db | 1186 | { |
5eec0563 JM |
1187 | c2 = PEEKC (); |
1188 | if (c2 == '#') | |
1189 | { | |
1190 | FORWARD (1); | |
1191 | CPP_PUTC (pfile, c2); | |
1192 | } | |
1193 | else if (c2 == '%' && PEEKN (1) == ':') | |
1194 | { | |
1195 | /* Digraph: "%:" == "#". */ | |
1196 | FORWARD (1); | |
1197 | CPP_RESERVE (pfile, 2); | |
1198 | CPP_PUTC_Q (pfile, c2); | |
1199 | CPP_PUTC_Q (pfile, GETC ()); | |
1200 | } | |
1201 | else | |
1368ee70 | 1202 | return CPP_HASH; |
5eec0563 | 1203 | |
1368ee70 | 1204 | return CPP_PASTE; |
45b966db ZW |
1205 | } |
1206 | ||
1207 | if (!pfile->only_seen_white) | |
5eec0563 JM |
1208 | return CPP_OTHER; |
1209 | ||
1210 | /* Remove the "#" or "%:" from the token buffer. */ | |
1211 | CPP_ADJUST_WRITTEN (pfile, (c == '#' ? -1 : -2)); | |
45b966db ZW |
1212 | return CPP_DIRECTIVE; |
1213 | ||
1214 | case '\"': | |
1215 | case '\'': | |
1216 | parse_string (pfile, c); | |
45b966db ZW |
1217 | return c == '\'' ? CPP_CHAR : CPP_STRING; |
1218 | ||
1219 | case '$': | |
ae79697b | 1220 | if (!CPP_OPTION (pfile, dollars_in_ident)) |
45b966db ZW |
1221 | goto randomchar; |
1222 | goto letter; | |
1223 | ||
1224 | case ':': | |
5eec0563 JM |
1225 | c2 = PEEKC (); |
1226 | /* Digraph: ":>" == "]". */ | |
1227 | if (c2 == '>' | |
1228 | || (c2 == ':' && CPP_OPTION (pfile, cplusplus))) | |
45b966db ZW |
1229 | goto op2; |
1230 | goto randomchar; | |
1231 | ||
1232 | case '&': | |
1233 | case '+': | |
1234 | case '|': | |
1235 | c2 = PEEKC (); | |
1236 | if (c2 == c || c2 == '=') | |
1237 | goto op2; | |
1238 | goto randomchar; | |
1239 | ||
5eec0563 JM |
1240 | case '%': |
1241 | /* Digraphs: "%:" == "#", "%>" == "}". */ | |
1242 | c2 = PEEKC (); | |
1243 | if (c2 == ':') | |
1244 | { | |
1245 | FORWARD (1); | |
1246 | CPP_RESERVE (pfile, 2); | |
1247 | CPP_PUTC_Q (pfile, c); | |
1248 | CPP_PUTC_Q (pfile, c2); | |
1249 | goto hash; | |
1250 | } | |
1251 | else if (c2 == '>') | |
1252 | { | |
1253 | FORWARD (1); | |
1254 | CPP_RESERVE (pfile, 2); | |
1255 | CPP_PUTC_Q (pfile, c); | |
1256 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1257 | return CPP_OPEN_BRACE; |
5eec0563 JM |
1258 | } |
1259 | /* else fall through */ | |
1260 | ||
45b966db ZW |
1261 | case '*': |
1262 | case '!': | |
45b966db ZW |
1263 | case '=': |
1264 | case '^': | |
1265 | if (PEEKC () == '=') | |
1266 | goto op2; | |
1267 | goto randomchar; | |
1268 | ||
1269 | case '-': | |
1270 | c2 = PEEKC (); | |
1271 | if (c2 == '-') | |
1272 | { | |
ae79697b | 1273 | if (CPP_OPTION (pfile, chill)) |
45b966db ZW |
1274 | goto comment; /* Chill style comment */ |
1275 | else | |
1276 | goto op2; | |
1277 | } | |
1278 | else if (c2 == '=') | |
1279 | goto op2; | |
1280 | else if (c2 == '>') | |
1281 | { | |
ae79697b | 1282 | if (CPP_OPTION (pfile, cplusplus) && PEEKN (1) == '*') |
45b966db ZW |
1283 | { |
1284 | /* In C++, there's a ->* operator. */ | |
1285 | token = CPP_OTHER; | |
45b966db ZW |
1286 | CPP_RESERVE (pfile, 4); |
1287 | CPP_PUTC_Q (pfile, c); | |
1288 | CPP_PUTC_Q (pfile, GETC ()); | |
1289 | CPP_PUTC_Q (pfile, GETC ()); | |
45b966db ZW |
1290 | return token; |
1291 | } | |
1292 | goto op2; | |
1293 | } | |
1294 | goto randomchar; | |
1295 | ||
1296 | case '<': | |
1297 | if (pfile->parsing_include_directive) | |
1298 | { | |
1299 | for (;;) | |
1300 | { | |
1301 | CPP_PUTC (pfile, c); | |
1302 | if (c == '>') | |
1303 | break; | |
1304 | c = GETC (); | |
1305 | if (c == '\n' || c == EOF) | |
1306 | { | |
1307 | cpp_error (pfile, | |
1308 | "missing '>' in `#include <FILENAME>'"); | |
1309 | break; | |
1310 | } | |
1311 | else if (c == '\r') | |
1312 | { | |
1313 | if (!CPP_BUFFER (pfile)->has_escapes) | |
1314 | { | |
1315 | /* Backslash newline is replaced by nothing. */ | |
1316 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1317 | CPP_BUMP_LINE (pfile); | |
1318 | } | |
1319 | else | |
1320 | { | |
1321 | /* We might conceivably get \r- or \r<space> in | |
1322 | here. Just delete 'em. */ | |
1323 | int d = GETC(); | |
1324 | if (d != '-' && d != ' ') | |
1325 | cpp_ice (pfile, "unrecognized escape \\r%c", d); | |
1326 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1327 | } | |
1328 | } | |
1329 | } | |
1330 | return CPP_STRING; | |
1331 | } | |
5eec0563 JM |
1332 | /* Digraphs: "<%" == "{", "<:" == "[". */ |
1333 | c2 = PEEKC (); | |
1334 | if (c2 == '%') | |
1335 | { | |
1336 | FORWARD (1); | |
1337 | CPP_RESERVE (pfile, 2); | |
1338 | CPP_PUTC_Q (pfile, c); | |
1339 | CPP_PUTC_Q (pfile, c2); | |
1368ee70 | 1340 | return CPP_CLOSE_BRACE; |
5eec0563 JM |
1341 | } |
1342 | else if (c2 == ':') | |
1343 | goto op2; | |
45b966db ZW |
1344 | /* else fall through */ |
1345 | case '>': | |
1346 | c2 = PEEKC (); | |
1347 | if (c2 == '=') | |
1348 | goto op2; | |
1349 | /* GNU C++ supports MIN and MAX operators <? and >?. */ | |
ae79697b | 1350 | if (c2 != c && (!CPP_OPTION (pfile, cplusplus) || c2 != '?')) |
45b966db ZW |
1351 | goto randomchar; |
1352 | FORWARD(1); | |
5eec0563 JM |
1353 | CPP_RESERVE (pfile, 3); |
1354 | CPP_PUTC_Q (pfile, c); | |
1355 | CPP_PUTC_Q (pfile, c2); | |
1356 | if (PEEKC () == '=') | |
45b966db | 1357 | CPP_PUTC_Q (pfile, GETC ()); |
45b966db ZW |
1358 | return CPP_OTHER; |
1359 | ||
1360 | case '.': | |
1361 | c2 = PEEKC (); | |
5eec0563 | 1362 | if (ISDIGIT (c2)) |
45b966db | 1363 | { |
5eec0563 | 1364 | CPP_PUTC (pfile, c); |
45b966db ZW |
1365 | c = GETC (); |
1366 | goto number; | |
1367 | } | |
1368 | ||
1369 | /* In C++ there's a .* operator. */ | |
ae79697b | 1370 | if (CPP_OPTION (pfile, cplusplus) && c2 == '*') |
45b966db ZW |
1371 | goto op2; |
1372 | ||
1373 | if (c2 == '.' && PEEKN(1) == '.') | |
1374 | { | |
5eec0563 | 1375 | CPP_RESERVE (pfile, 3); |
45b966db ZW |
1376 | CPP_PUTC_Q (pfile, '.'); |
1377 | CPP_PUTC_Q (pfile, '.'); | |
1378 | CPP_PUTC_Q (pfile, '.'); | |
1379 | FORWARD (2); | |
1368ee70 | 1380 | return CPP_ELLIPSIS; |
45b966db ZW |
1381 | } |
1382 | goto randomchar; | |
1383 | ||
1384 | op2: | |
5eec0563 | 1385 | CPP_RESERVE (pfile, 2); |
45b966db ZW |
1386 | CPP_PUTC_Q (pfile, c); |
1387 | CPP_PUTC_Q (pfile, GETC ()); | |
5eec0563 | 1388 | return CPP_OTHER; |
45b966db ZW |
1389 | |
1390 | case 'L': | |
1391 | c2 = PEEKC (); | |
1392 | if ((c2 == '\'' || c2 == '\"') && !CPP_TRADITIONAL (pfile)) | |
1393 | { | |
1394 | CPP_PUTC (pfile, c); | |
1395 | c = GETC (); | |
1396 | parse_string (pfile, c); | |
45b966db ZW |
1397 | return c == '\'' ? CPP_WCHAR : CPP_WSTRING; |
1398 | } | |
1399 | goto letter; | |
1400 | ||
1401 | case '0': case '1': case '2': case '3': case '4': | |
1402 | case '5': case '6': case '7': case '8': case '9': | |
1403 | number: | |
1404 | c2 = '.'; | |
1405 | for (;;) | |
1406 | { | |
1407 | CPP_RESERVE (pfile, 2); | |
1408 | CPP_PUTC_Q (pfile, c); | |
1409 | c = PEEKC (); | |
1410 | if (c == EOF) | |
1411 | break; | |
1412 | if (!is_numchar(c) && c != '.' | |
1413 | && ((c2 != 'e' && c2 != 'E' | |
1414 | && ((c2 != 'p' && c2 != 'P') | |
ae79697b | 1415 | || CPP_OPTION (pfile, c89))) |
45b966db ZW |
1416 | || (c != '+' && c != '-'))) |
1417 | break; | |
1418 | FORWARD(1); | |
1419 | c2= c; | |
1420 | } | |
45b966db ZW |
1421 | return CPP_NUMBER; |
1422 | case 'b': case 'c': case 'd': case 'h': case 'o': | |
1423 | case 'B': case 'C': case 'D': case 'H': case 'O': | |
ae79697b | 1424 | if (CPP_OPTION (pfile, chill) && PEEKC () == '\'') |
45b966db | 1425 | { |
45b966db ZW |
1426 | CPP_RESERVE (pfile, 2); |
1427 | CPP_PUTC_Q (pfile, c); | |
1428 | CPP_PUTC_Q (pfile, '\''); | |
1429 | FORWARD(1); | |
1430 | for (;;) | |
1431 | { | |
1432 | c = GETC(); | |
1433 | if (c == EOF) | |
1434 | goto chill_number_eof; | |
1435 | if (!is_numchar(c)) | |
1436 | break; | |
1437 | CPP_PUTC (pfile, c); | |
1438 | } | |
1439 | if (c == '\'') | |
1440 | { | |
1441 | CPP_RESERVE (pfile, 2); | |
1442 | CPP_PUTC_Q (pfile, c); | |
45b966db ZW |
1443 | return CPP_STRING; |
1444 | } | |
1445 | else | |
1446 | { | |
1447 | FORWARD(-1); | |
1448 | chill_number_eof: | |
45b966db ZW |
1449 | return CPP_NUMBER; |
1450 | } | |
1451 | } | |
1452 | else | |
1453 | goto letter; | |
1454 | case '_': | |
1455 | case 'a': case 'e': case 'f': case 'g': case 'i': case 'j': | |
1456 | case 'k': case 'l': case 'm': case 'n': case 'p': case 'q': | |
1457 | case 'r': case 's': case 't': case 'u': case 'v': case 'w': | |
1458 | case 'x': case 'y': case 'z': | |
1459 | case 'A': case 'E': case 'F': case 'G': case 'I': case 'J': | |
1460 | case 'K': case 'M': case 'N': case 'P': case 'Q': case 'R': | |
1461 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
1462 | case 'Y': case 'Z': | |
1463 | letter: | |
45b966db ZW |
1464 | _cpp_parse_name (pfile, c); |
1465 | return CPP_MACRO; | |
1466 | ||
64aaf407 NB |
1467 | case ' ': case '\t': case '\v': case '\f': case '\0': |
1468 | { | |
1469 | int null_count = 0; | |
1470 | ||
1471 | for (;;) | |
1472 | { | |
1473 | if (c == '\0') | |
1474 | null_count++; | |
1475 | else | |
1476 | CPP_PUTC (pfile, c); | |
1477 | c = PEEKC (); | |
1478 | if (c == EOF || !is_hspace(c)) | |
1479 | break; | |
1480 | FORWARD(1); | |
1481 | } | |
1482 | if (null_count) | |
1483 | null_warning (pfile, null_count); | |
1484 | return CPP_HSPACE; | |
1485 | } | |
45b966db ZW |
1486 | |
1487 | case '\r': | |
1488 | if (CPP_BUFFER (pfile)->has_escapes) | |
1489 | { | |
1490 | c = GETC (); | |
1491 | if (c == '-') | |
1492 | { | |
1493 | if (pfile->output_escapes) | |
1494 | CPP_PUTS (pfile, "\r-", 2); | |
1495 | _cpp_parse_name (pfile, GETC ()); | |
1496 | return CPP_NAME; | |
1497 | } | |
1498 | else if (c == ' ') | |
1499 | { | |
ff2b53ef ZW |
1500 | /* "\r " means a space, but only if necessary to prevent |
1501 | accidental token concatenation. */ | |
45b966db ZW |
1502 | CPP_RESERVE (pfile, 2); |
1503 | if (pfile->output_escapes) | |
1504 | CPP_PUTC_Q (pfile, '\r'); | |
1505 | CPP_PUTC_Q (pfile, c); | |
1506 | return CPP_HSPACE; | |
1507 | } | |
1508 | else | |
1509 | { | |
1510 | cpp_ice (pfile, "unrecognized escape \\r%c", c); | |
1511 | goto get_next; | |
1512 | } | |
1513 | } | |
1514 | else | |
1515 | { | |
1516 | /* Backslash newline is ignored. */ | |
cbccf5e8 MM |
1517 | if (!ACTIVE_MARK_P (pfile)) |
1518 | CPP_BUMP_LINE (pfile); | |
45b966db ZW |
1519 | goto get_next; |
1520 | } | |
1521 | ||
1522 | case '\n': | |
1523 | CPP_PUTC (pfile, c); | |
45b966db ZW |
1524 | return CPP_VSPACE; |
1525 | ||
1368ee70 ZW |
1526 | case '(': token = CPP_OPEN_PAREN; goto char1; |
1527 | case ')': token = CPP_CLOSE_PAREN; goto char1; | |
1528 | case '{': token = CPP_OPEN_BRACE; goto char1; | |
1529 | case '}': token = CPP_CLOSE_BRACE; goto char1; | |
1530 | case ',': token = CPP_COMMA; goto char1; | |
1531 | case ';': token = CPP_SEMICOLON; goto char1; | |
45b966db ZW |
1532 | |
1533 | randomchar: | |
1534 | default: | |
1535 | token = CPP_OTHER; | |
1536 | char1: | |
45b966db ZW |
1537 | CPP_PUTC (pfile, c); |
1538 | return token; | |
1539 | } | |
1540 | } | |
1541 | ||
1542 | /* Check for and expand a macro, which is from WRITTEN to CPP_WRITTEN (pfile). | |
1543 | Caller is expected to have checked no_macro_expand. */ | |
1544 | static int | |
1545 | maybe_macroexpand (pfile, written) | |
1546 | cpp_reader *pfile; | |
1547 | long written; | |
1548 | { | |
1549 | U_CHAR *macro = pfile->token_buffer + written; | |
1550 | size_t len = CPP_WRITTEN (pfile) - written; | |
1551 | HASHNODE *hp = _cpp_lookup (pfile, macro, len); | |
1552 | ||
a7abcbbf ZW |
1553 | /* _cpp_lookup never returns null. */ |
1554 | if (hp->type == T_VOID) | |
45b966db | 1555 | return 0; |
d9e0bd53 | 1556 | if (hp->disabled || hp->type == T_IDENTITY) |
45b966db ZW |
1557 | { |
1558 | if (pfile->output_escapes) | |
1559 | { | |
1560 | /* Insert a no-reexpand marker before IDENT. */ | |
1561 | CPP_RESERVE (pfile, 2); | |
1562 | CPP_ADJUST_WRITTEN (pfile, 2); | |
1563 | macro = pfile->token_buffer + written; | |
1564 | ||
1565 | memmove (macro + 2, macro, len); | |
1566 | macro[0] = '\r'; | |
1567 | macro[1] = '-'; | |
1568 | } | |
1569 | return 0; | |
1570 | } | |
ff2b53ef ZW |
1571 | if (hp->type == T_EMPTY) |
1572 | { | |
1573 | /* Special case optimization: macro expands to nothing. */ | |
1574 | CPP_SET_WRITTEN (pfile, written); | |
1575 | CPP_PUTC_Q (pfile, ' '); | |
1576 | return 1; | |
1577 | } | |
45b966db ZW |
1578 | |
1579 | /* If macro wants an arglist, verify that a '(' follows. */ | |
d9e0bd53 | 1580 | if (hp->type == T_FMACRO) |
45b966db ZW |
1581 | { |
1582 | int macbuf_whitespace = 0; | |
1583 | int c; | |
1584 | ||
1585 | while (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
1586 | { | |
1587 | const U_CHAR *point = CPP_BUFFER (pfile)->cur; | |
1588 | for (;;) | |
1589 | { | |
1590 | _cpp_skip_hspace (pfile); | |
1591 | c = PEEKC (); | |
1592 | if (c == '\n') | |
1593 | FORWARD(1); | |
1594 | else | |
1595 | break; | |
1596 | } | |
1597 | if (point != CPP_BUFFER (pfile)->cur) | |
1598 | macbuf_whitespace = 1; | |
1599 | if (c == '(') | |
1600 | goto is_macro_call; | |
1601 | else if (c != EOF) | |
1602 | goto not_macro_call; | |
1603 | cpp_pop_buffer (pfile); | |
1604 | } | |
1605 | ||
1606 | CPP_SET_MARK (pfile); | |
1607 | for (;;) | |
1608 | { | |
1609 | _cpp_skip_hspace (pfile); | |
1610 | c = PEEKC (); | |
1611 | if (c == '\n') | |
1612 | FORWARD(1); | |
1613 | else | |
1614 | break; | |
1615 | } | |
1616 | CPP_GOTO_MARK (pfile); | |
1617 | ||
1618 | if (c != '(') | |
1619 | { | |
1620 | not_macro_call: | |
1621 | if (macbuf_whitespace) | |
1622 | CPP_PUTC (pfile, ' '); | |
1623 | return 0; | |
1624 | } | |
1625 | } | |
1626 | ||
1627 | is_macro_call: | |
1628 | /* This is now known to be a macro call. | |
1629 | Expand the macro, reading arguments as needed, | |
1630 | and push the expansion on the input stack. */ | |
1631 | _cpp_macroexpand (pfile, hp); | |
1632 | CPP_SET_WRITTEN (pfile, written); | |
1633 | return 1; | |
1634 | } | |
1635 | ||
9e62c811 ZW |
1636 | /* Complain about \v or \f in a preprocessing directive (constraint |
1637 | violation, C99 6.10 para 5). Caller has checked CPP_PEDANTIC. */ | |
1638 | static void | |
1639 | pedantic_whitespace (pfile, p, len) | |
1640 | cpp_reader *pfile; | |
1641 | U_CHAR *p; | |
1642 | unsigned int len; | |
1643 | { | |
1644 | while (len) | |
1645 | { | |
1646 | if (*p == '\v') | |
1647 | cpp_pedwarn (pfile, "vertical tab in preprocessing directive"); | |
1648 | else if (*p == '\f') | |
1649 | cpp_pedwarn (pfile, "form feed in preprocessing directive"); | |
1650 | p++; | |
1651 | len--; | |
1652 | } | |
1653 | } | |
1654 | ||
1655 | ||
3a2b2c7a | 1656 | enum cpp_ttype |
45b966db ZW |
1657 | cpp_get_token (pfile) |
1658 | cpp_reader *pfile; | |
1659 | { | |
3a2b2c7a | 1660 | enum cpp_ttype token; |
45b966db ZW |
1661 | long written = CPP_WRITTEN (pfile); |
1662 | ||
1663 | get_next: | |
1664 | token = _cpp_lex_token (pfile); | |
1665 | ||
1666 | switch (token) | |
1667 | { | |
1668 | default: | |
ff2b53ef ZW |
1669 | pfile->potential_control_macro = 0; |
1670 | pfile->only_seen_white = 0; | |
1671 | return token; | |
1672 | ||
1673 | case CPP_VSPACE: | |
1674 | if (pfile->only_seen_white == 0) | |
1675 | pfile->only_seen_white = 1; | |
1676 | CPP_BUMP_LINE (pfile); | |
ff2b53ef ZW |
1677 | return token; |
1678 | ||
1679 | case CPP_HSPACE: | |
1680 | case CPP_COMMENT: | |
45b966db ZW |
1681 | return token; |
1682 | ||
1683 | case CPP_DIRECTIVE: | |
ff2b53ef | 1684 | pfile->potential_control_macro = 0; |
45b966db ZW |
1685 | if (_cpp_handle_directive (pfile)) |
1686 | return CPP_DIRECTIVE; | |
1687 | pfile->only_seen_white = 0; | |
1688 | CPP_PUTC (pfile, '#'); | |
1689 | return CPP_OTHER; | |
1690 | ||
1691 | case CPP_MACRO: | |
ff2b53ef ZW |
1692 | pfile->potential_control_macro = 0; |
1693 | pfile->only_seen_white = 0; | |
45b966db ZW |
1694 | if (! pfile->no_macro_expand |
1695 | && maybe_macroexpand (pfile, written)) | |
1696 | goto get_next; | |
1697 | return CPP_NAME; | |
1698 | ||
1699 | case CPP_EOF: | |
f2d5f0cc ZW |
1700 | if (CPP_BUFFER (pfile) == NULL) |
1701 | return CPP_EOF; | |
c56c2073 | 1702 | if (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) |
f2d5f0cc | 1703 | { |
45b966db ZW |
1704 | cpp_pop_buffer (pfile); |
1705 | goto get_next; | |
1706 | } | |
c56c2073 ZW |
1707 | cpp_pop_buffer (pfile); |
1708 | return CPP_EOF; | |
45b966db ZW |
1709 | } |
1710 | } | |
1711 | ||
1712 | /* Like cpp_get_token, but skip spaces and comments. */ | |
1713 | ||
3a2b2c7a | 1714 | enum cpp_ttype |
45b966db ZW |
1715 | cpp_get_non_space_token (pfile) |
1716 | cpp_reader *pfile; | |
1717 | { | |
1718 | int old_written = CPP_WRITTEN (pfile); | |
1719 | for (;;) | |
1720 | { | |
3a2b2c7a | 1721 | enum cpp_ttype token = cpp_get_token (pfile); |
ff2b53ef | 1722 | if (token != CPP_COMMENT && token != CPP_HSPACE && token != CPP_VSPACE) |
45b966db ZW |
1723 | return token; |
1724 | CPP_SET_WRITTEN (pfile, old_written); | |
1725 | } | |
1726 | } | |
1727 | ||
ff2b53ef | 1728 | /* Like cpp_get_token, except that it does not execute directives, |
c56c2073 | 1729 | does not consume vertical space, and discards horizontal space. */ |
3a2b2c7a | 1730 | enum cpp_ttype |
9e62c811 | 1731 | _cpp_get_directive_token (pfile) |
45b966db ZW |
1732 | cpp_reader *pfile; |
1733 | { | |
ff2b53ef | 1734 | long old_written; |
3a2b2c7a | 1735 | enum cpp_ttype token; |
57c578a6 | 1736 | int at_bol; |
45b966db | 1737 | |
ff2b53ef | 1738 | get_next: |
57c578a6 | 1739 | at_bol = (CPP_BUFFER (pfile)->cur == CPP_BUFFER (pfile)->line_base); |
ff2b53ef ZW |
1740 | old_written = CPP_WRITTEN (pfile); |
1741 | token = _cpp_lex_token (pfile); | |
1742 | switch (token) | |
45b966db | 1743 | { |
ff2b53ef ZW |
1744 | default: |
1745 | return token; | |
45b966db | 1746 | |
ff2b53ef ZW |
1747 | case CPP_VSPACE: |
1748 | /* Put it back and return VSPACE. */ | |
1749 | FORWARD(-1); | |
1750 | CPP_ADJUST_WRITTEN (pfile, -1); | |
1751 | return CPP_VSPACE; | |
45b966db | 1752 | |
ff2b53ef | 1753 | case CPP_HSPACE: |
57c578a6 ZW |
1754 | /* The purpose of this rather strange check is to prevent pedantic |
1755 | warnings for ^L in an #ifdefed out block. */ | |
1756 | if (CPP_PEDANTIC (pfile) && ! at_bol) | |
9e62c811 ZW |
1757 | pedantic_whitespace (pfile, pfile->token_buffer + old_written, |
1758 | CPP_WRITTEN (pfile) - old_written); | |
1759 | CPP_SET_WRITTEN (pfile, old_written); | |
1760 | goto get_next; | |
ff2b53ef | 1761 | return CPP_HSPACE; |
45b966db | 1762 | |
ff2b53ef ZW |
1763 | case CPP_DIRECTIVE: |
1764 | /* Don't execute the directive, but don't smash it to OTHER either. */ | |
1765 | CPP_PUTC (pfile, '#'); | |
1766 | return CPP_DIRECTIVE; | |
1767 | ||
1768 | case CPP_MACRO: | |
1769 | if (! pfile->no_macro_expand | |
1770 | && maybe_macroexpand (pfile, old_written)) | |
1771 | goto get_next; | |
1772 | return CPP_NAME; | |
45b966db | 1773 | |
ff2b53ef ZW |
1774 | case CPP_EOF: |
1775 | if (CPP_IS_MACRO_BUFFER (CPP_BUFFER (pfile))) | |
45b966db | 1776 | { |
ff2b53ef ZW |
1777 | cpp_pop_buffer (pfile); |
1778 | goto get_next; | |
45b966db | 1779 | } |
ff2b53ef ZW |
1780 | else |
1781 | /* This can happen for files that don't end with a newline, | |
1782 | and for cpp_define and friends. Pretend they do, so | |
1783 | callers don't have to deal. A warning will be issued by | |
1784 | someone else, if necessary. */ | |
1785 | return CPP_VSPACE; | |
1786 | } | |
1787 | } | |
1788 | ||
45b966db ZW |
1789 | /* Determine the current line and column. Used only by read_and_prescan. */ |
1790 | static U_CHAR * | |
1791 | find_position (start, limit, linep) | |
1792 | U_CHAR *start; | |
1793 | U_CHAR *limit; | |
1794 | unsigned long *linep; | |
1795 | { | |
1796 | unsigned long line = *linep; | |
1797 | U_CHAR *lbase = start; | |
1798 | while (start < limit) | |
1799 | { | |
1800 | U_CHAR ch = *start++; | |
1801 | if (ch == '\n' || ch == '\r') | |
1802 | { | |
1803 | line++; | |
1804 | lbase = start; | |
1805 | } | |
1806 | } | |
1807 | *linep = line; | |
1808 | return lbase; | |
1809 | } | |
1810 | ||
2a87fbe8 ZW |
1811 | /* The following table is used by _cpp_read_and_prescan. If we have |
1812 | designated initializers, it can be constant data; otherwise, it is | |
1813 | set up at runtime by _cpp_init_input_buffer. */ | |
46d07497 ZW |
1814 | |
1815 | #ifndef UCHAR_MAX | |
1816 | #define UCHAR_MAX 255 /* assume 8-bit bytes */ | |
1817 | #endif | |
1818 | ||
12cf91fe | 1819 | #if (GCC_VERSION >= 2007) |
2a87fbe8 | 1820 | #define init_chartab() /* nothing */ |
12cf91fe | 1821 | #define CHARTAB __extension__ static const U_CHAR chartab[UCHAR_MAX + 1] = { |
46d07497 ZW |
1822 | #define END }; |
1823 | #define s(p, v) [p] = v, | |
1824 | #else | |
12cf91fe | 1825 | #define CHARTAB static U_CHAR chartab[UCHAR_MAX + 1] = { 0 }; \ |
2a87fbe8 ZW |
1826 | static void init_chartab PARAMS ((void)) { \ |
1827 | unsigned char *x = chartab; | |
46d07497 ZW |
1828 | #define END } |
1829 | #define s(p, v) x[p] = v; | |
1830 | #endif | |
1831 | ||
1832 | /* Table of characters that can't be handled in the inner loop. | |
2a87fbe8 ZW |
1833 | Also contains the mapping between trigraph third characters and their |
1834 | replacements. */ | |
46d07497 ZW |
1835 | #define SPECCASE_CR 1 |
1836 | #define SPECCASE_BACKSLASH 2 | |
1837 | #define SPECCASE_QUESTION 3 | |
1838 | ||
2a87fbe8 | 1839 | CHARTAB |
46d07497 ZW |
1840 | s('\r', SPECCASE_CR) |
1841 | s('\\', SPECCASE_BACKSLASH) | |
1842 | s('?', SPECCASE_QUESTION) | |
46d07497 | 1843 | |
46d07497 ZW |
1844 | s('=', '#') s(')', ']') s('!', '|') |
1845 | s('(', '[') s('\'', '^') s('>', '}') | |
1846 | s('/', '\\') s('<', '{') s('-', '~') | |
1847 | END | |
1848 | ||
1849 | #undef CHARTAB | |
46d07497 ZW |
1850 | #undef END |
1851 | #undef s | |
1852 | ||
2a87fbe8 ZW |
1853 | #define NORMAL(c) ((chartab[c]) == 0 || (chartab[c]) > SPECCASE_QUESTION) |
1854 | #define NONTRI(c) ((c) <= SPECCASE_QUESTION) | |
1855 | ||
45b966db ZW |
1856 | /* Read the entire contents of file DESC into buffer BUF. LEN is how |
1857 | much memory to allocate initially; more will be allocated if | |
1858 | necessary. Convert end-of-line markers (\n, \r, \r\n, \n\r) to | |
1859 | canonical form (\n). If enabled, convert and/or warn about | |
1860 | trigraphs. Convert backslash-newline to a one-character escape | |
1861 | (\r) and remove it from "embarrassing" places (i.e. the middle of a | |
1862 | token). If there is no newline at the end of the file, add one and | |
1863 | warn. Returns -1 on failure, or the actual length of the data to | |
1864 | be scanned. | |
1865 | ||
1866 | This function does a lot of work, and can be a serious performance | |
1867 | bottleneck. It has been tuned heavily; make sure you understand it | |
1868 | before hacking. The common case - no trigraphs, Unix style line | |
1869 | breaks, backslash-newline set off by whitespace, newline at EOF - | |
1870 | has been optimized at the expense of the others. The performance | |
1871 | penalty for DOS style line breaks (\r\n) is about 15%. | |
1872 | ||
1873 | Warnings lose particularly heavily since we have to determine the | |
1874 | line number, which involves scanning from the beginning of the file | |
1875 | or from the last warning. The penalty for the absence of a newline | |
1876 | at the end of reload1.c is about 60%. (reload1.c is 329k.) | |
1877 | ||
1878 | If your file has more than one kind of end-of-line marker, you | |
04e3ec78 NB |
1879 | will get messed-up line numbering. |
1880 | ||
1881 | So that the cases of the switch statement do not have to concern | |
1882 | themselves with the complications of reading beyond the end of the | |
1883 | buffer, the buffer is guaranteed to have at least 3 characters in | |
1884 | it (or however many are left in the file, if less) on entry to the | |
1885 | switch. This is enough to handle trigraphs and the "\\\n\r" and | |
1886 | "\\\r\n" cases. | |
1887 | ||
1888 | The end of the buffer is marked by a '\\', which, being a special | |
1889 | character, guarantees we will exit the fast-scan loops and perform | |
1890 | a refill. */ | |
46d07497 | 1891 | |
45b966db ZW |
1892 | long |
1893 | _cpp_read_and_prescan (pfile, fp, desc, len) | |
1894 | cpp_reader *pfile; | |
1895 | cpp_buffer *fp; | |
1896 | int desc; | |
1897 | size_t len; | |
1898 | { | |
1899 | U_CHAR *buf = (U_CHAR *) xmalloc (len); | |
1900 | U_CHAR *ip, *op, *line_base; | |
1901 | U_CHAR *ibase; | |
45b966db ZW |
1902 | unsigned long line; |
1903 | unsigned int deferred_newlines; | |
45b966db | 1904 | size_t offset; |
04e3ec78 | 1905 | int count = 0; |
45b966db ZW |
1906 | |
1907 | offset = 0; | |
04e3ec78 | 1908 | deferred_newlines = 0; |
45b966db ZW |
1909 | op = buf; |
1910 | line_base = buf; | |
1911 | line = 1; | |
04e3ec78 NB |
1912 | ibase = pfile->input_buffer + 3; |
1913 | ip = ibase; | |
1914 | ip[-1] = '\0'; /* Guarantee no match with \n for SPECCASE_CR */ | |
45b966db ZW |
1915 | |
1916 | for (;;) | |
1917 | { | |
04e3ec78 NB |
1918 | U_CHAR *near_buff_end; |
1919 | ||
04e3ec78 | 1920 | count = read (desc, ibase, pfile->input_buffer_len); |
45b966db ZW |
1921 | if (count < 0) |
1922 | goto error; | |
04e3ec78 NB |
1923 | |
1924 | ibase[count] = '\\'; /* Marks end of buffer */ | |
1925 | if (count) | |
45b966db | 1926 | { |
04e3ec78 NB |
1927 | near_buff_end = pfile->input_buffer + count; |
1928 | offset += count; | |
45b966db | 1929 | if (offset > len) |
04e3ec78 NB |
1930 | { |
1931 | size_t delta_op; | |
1932 | size_t delta_line_base; | |
1b955cba | 1933 | len = offset * 2; |
04e3ec78 NB |
1934 | if (offset > len) |
1935 | /* len overflowed. | |
1936 | This could happen if the file is larger than half the | |
1937 | maximum address space of the machine. */ | |
1938 | goto too_big; | |
1939 | ||
1940 | delta_op = op - buf; | |
1941 | delta_line_base = line_base - buf; | |
1942 | buf = (U_CHAR *) xrealloc (buf, len); | |
1943 | op = buf + delta_op; | |
1944 | line_base = buf + delta_line_base; | |
1945 | } | |
1946 | } | |
1947 | else | |
1948 | { | |
1949 | if (ip == ibase) | |
1950 | break; | |
1951 | /* Allow normal processing of the (at most 2) remaining | |
1952 | characters. The end-of-buffer marker is still present | |
1953 | and prevents false matches within the switch. */ | |
1954 | near_buff_end = ibase - 1; | |
45b966db ZW |
1955 | } |
1956 | ||
1957 | for (;;) | |
1958 | { | |
04e3ec78 | 1959 | unsigned int span; |
45b966db | 1960 | |
04e3ec78 | 1961 | /* Deal with \-newline, potentially in the middle of a token. */ |
45b966db ZW |
1962 | if (deferred_newlines) |
1963 | { | |
2a87fbe8 | 1964 | if (op != buf && ! is_space (op[-1]) && op[-1] != '\r') |
04e3ec78 NB |
1965 | { |
1966 | /* Previous was not white space. Skip to white | |
1967 | space, if we can, before outputting the \r's */ | |
1968 | span = 0; | |
1969 | while (ip[span] != ' ' | |
1970 | && ip[span] != '\t' | |
1971 | && ip[span] != '\n' | |
2a87fbe8 | 1972 | && NORMAL(ip[span])) |
04e3ec78 NB |
1973 | span++; |
1974 | memcpy (op, ip, span); | |
1975 | op += span; | |
1976 | ip += span; | |
2a87fbe8 | 1977 | if (! NORMAL(ip[0])) |
04e3ec78 NB |
1978 | goto do_speccase; |
1979 | } | |
1980 | while (deferred_newlines) | |
1981 | deferred_newlines--, *op++ = '\r'; | |
45b966db ZW |
1982 | } |
1983 | ||
1984 | /* Copy as much as we can without special treatment. */ | |
04e3ec78 | 1985 | span = 0; |
2a87fbe8 | 1986 | while (NORMAL (ip[span])) span++; |
45b966db ZW |
1987 | memcpy (op, ip, span); |
1988 | op += span; | |
1989 | ip += span; | |
1990 | ||
04e3ec78 NB |
1991 | do_speccase: |
1992 | if (ip > near_buff_end) /* Do we have enough chars? */ | |
1993 | break; | |
2a87fbe8 | 1994 | switch (chartab[*ip++]) |
45b966db | 1995 | { |
45b966db | 1996 | case SPECCASE_CR: /* \r */ |
04e3ec78 | 1997 | if (ip[-2] != '\n') |
45b966db | 1998 | { |
04e3ec78 NB |
1999 | if (*ip == '\n') |
2000 | ip++; | |
2001 | *op++ = '\n'; | |
45b966db | 2002 | } |
45b966db ZW |
2003 | break; |
2004 | ||
2005 | case SPECCASE_BACKSLASH: /* \ */ | |
04e3ec78 | 2006 | if (*ip == '\n') |
45b966db | 2007 | { |
04e3ec78 | 2008 | deferred_newlines++; |
45b966db ZW |
2009 | ip++; |
2010 | if (*ip == '\r') ip++; | |
45b966db ZW |
2011 | } |
2012 | else if (*ip == '\r') | |
2013 | { | |
04e3ec78 | 2014 | deferred_newlines++; |
45b966db ZW |
2015 | ip++; |
2016 | if (*ip == '\n') ip++; | |
45b966db ZW |
2017 | } |
2018 | else | |
2019 | *op++ = '\\'; | |
04e3ec78 | 2020 | break; |
45b966db ZW |
2021 | |
2022 | case SPECCASE_QUESTION: /* ? */ | |
2023 | { | |
2024 | unsigned int d, t; | |
04e3ec78 NB |
2025 | |
2026 | *op++ = '?'; /* Normal non-trigraph case */ | |
2027 | if (ip[0] != '?') | |
2028 | break; | |
2029 | ||
45b966db | 2030 | d = ip[1]; |
2a87fbe8 ZW |
2031 | t = chartab[d]; |
2032 | if (NONTRI (t)) | |
04e3ec78 | 2033 | break; |
45b966db | 2034 | |
ae79697b | 2035 | if (CPP_OPTION (pfile, warn_trigraphs)) |
45b966db ZW |
2036 | { |
2037 | unsigned long col; | |
2038 | line_base = find_position (line_base, op, &line); | |
2039 | col = op - line_base + 1; | |
ae79697b | 2040 | if (CPP_OPTION (pfile, trigraphs)) |
45b966db | 2041 | cpp_warning_with_line (pfile, line, col, |
04e3ec78 | 2042 | "trigraph ??%c converted to %c", d, t); |
45b966db ZW |
2043 | else |
2044 | cpp_warning_with_line (pfile, line, col, | |
04e3ec78 | 2045 | "trigraph ??%c ignored", d); |
45b966db | 2046 | } |
04e3ec78 NB |
2047 | |
2048 | ip += 2; | |
ae79697b | 2049 | if (CPP_OPTION (pfile, trigraphs)) |
45b966db | 2050 | { |
04e3ec78 | 2051 | op[-1] = t; /* Overwrite '?' */ |
45b966db | 2052 | if (t == '\\') |
04e3ec78 NB |
2053 | { |
2054 | op--; | |
2055 | *--ip = '\\'; | |
2056 | goto do_speccase; /* May need buffer refill */ | |
2057 | } | |
45b966db ZW |
2058 | } |
2059 | else | |
2060 | { | |
45b966db ZW |
2061 | *op++ = '?'; |
2062 | *op++ = d; | |
2063 | } | |
2064 | } | |
04e3ec78 | 2065 | break; |
45b966db ZW |
2066 | } |
2067 | } | |
f6fab919 ZW |
2068 | /* Copy previous char plus unprocessed (at most 2) chars |
2069 | to beginning of buffer, refill it with another | |
2070 | read(), and continue processing */ | |
2071 | memmove (ip - count - 1, ip - 1, 4 - (ip - near_buff_end)); | |
2072 | ip -= count; | |
45b966db ZW |
2073 | } |
2074 | ||
2075 | if (offset == 0) | |
2076 | return 0; | |
2077 | ||
45b966db ZW |
2078 | if (op[-1] != '\n') |
2079 | { | |
2080 | unsigned long col; | |
2081 | line_base = find_position (line_base, op, &line); | |
2082 | col = op - line_base + 1; | |
f6fab919 | 2083 | cpp_warning_with_line (pfile, line, col, "no newline at end of file"); |
45b966db ZW |
2084 | if (offset + 1 > len) |
2085 | { | |
2086 | len += 1; | |
2087 | if (offset + 1 > len) | |
2088 | goto too_big; | |
2089 | buf = (U_CHAR *) xrealloc (buf, len); | |
2090 | op = buf + offset; | |
2091 | } | |
2092 | *op++ = '\n'; | |
2093 | } | |
2094 | ||
2095 | fp->buf = ((len - offset < 20) ? buf : (U_CHAR *)xrealloc (buf, op - buf)); | |
2096 | return op - buf; | |
2097 | ||
2098 | too_big: | |
f6fab919 ZW |
2099 | cpp_notice (pfile, "%s is too large (>%lu bytes)", fp->ihash->name, |
2100 | (unsigned long)offset); | |
45b966db ZW |
2101 | free (buf); |
2102 | return -1; | |
2103 | ||
2104 | error: | |
2105 | cpp_error_from_errno (pfile, fp->ihash->name); | |
2106 | free (buf); | |
2107 | return -1; | |
2108 | } | |
2109 | ||
2a87fbe8 ZW |
2110 | /* Allocate pfile->input_buffer, and initialize chartab[] |
2111 | if it hasn't happened already. */ | |
46d07497 | 2112 | |
45b966db ZW |
2113 | void |
2114 | _cpp_init_input_buffer (pfile) | |
2115 | cpp_reader *pfile; | |
2116 | { | |
2117 | U_CHAR *tmp; | |
2118 | ||
2a87fbe8 | 2119 | init_chartab (); |
9e62c811 | 2120 | init_token_list (pfile, &pfile->directbuf, 0); |
04e3ec78 | 2121 | |
45b966db ZW |
2122 | /* Determine the appropriate size for the input buffer. Normal C |
2123 | source files are smaller than eight K. */ | |
04e3ec78 NB |
2124 | /* 8Kbytes of buffer proper, 1 to detect running off the end without |
2125 | address arithmetic all the time, and 3 for pushback during buffer | |
2126 | refill, in case there's a potential trigraph or end-of-line | |
2127 | digraph at the end of a block. */ | |
45b966db | 2128 | |
04e3ec78 | 2129 | tmp = (U_CHAR *) xmalloc (8192 + 1 + 3); |
45b966db ZW |
2130 | pfile->input_buffer = tmp; |
2131 | pfile->input_buffer_len = 8192; | |
2132 | } | |
c5a04734 | 2133 | |
6d2c2047 ZW |
2134 | /* Utility routine: |
2135 | Compares, in the manner of strcmp(3), the token beginning at TOKEN | |
2136 | and extending for LEN characters to the NUL-terminated string | |
2137 | STRING. Typical usage: | |
2138 | ||
2139 | if (! cpp_idcmp (pfile->token_buffer + here, CPP_WRITTEN (pfile) - here, | |
2140 | "inline")) | |
2141 | { ... } | |
2142 | */ | |
2143 | ||
2144 | int | |
2145 | cpp_idcmp (token, len, string) | |
2146 | const U_CHAR *token; | |
2147 | size_t len; | |
2148 | const char *string; | |
2149 | { | |
2150 | size_t len2 = strlen (string); | |
2151 | int r; | |
2152 | ||
2153 | if ((r = memcmp (token, string, MIN (len, len2)))) | |
2154 | return r; | |
2155 | ||
2156 | /* The longer of the two strings sorts after the shorter. */ | |
2157 | if (len == len2) | |
2158 | return 0; | |
2159 | else if (len < len2) | |
2160 | return -1; | |
2161 | else | |
2162 | return 1; | |
2163 | } | |
2164 | ||
b8f41010 | 2165 | #ifdef NEW_LEXER |
c5a04734 | 2166 | |
d6d5f795 NB |
2167 | /* Lexing algorithm. |
2168 | ||
2169 | The original lexer in cpplib was made up of two passes: a first pass | |
2170 | that replaced trigraphs and deleted esacped newlines, and a second | |
2171 | pass that tokenized the result of the first pass. Tokenisation was | |
2172 | performed by peeking at the next character in the input stream. For | |
6777db6d | 2173 | example, if the input stream contained "!=", the handler for the ! |
d6d5f795 | 2174 | character would peek at the next character, and if it were a '=' |
6777db6d NB |
2175 | would skip over it, and return a "!=" token, otherwise it would |
2176 | return just the "!" token. | |
d6d5f795 NB |
2177 | |
2178 | To implement a single-pass lexer, this peeking ahead is unworkable. | |
2179 | An arbitrary number of escaped newlines, and trigraphs (in particular | |
6777db6d NB |
2180 | ??/ which translates to the escape \), could separate the '!' and '=' |
2181 | in the input stream, yet the next token is still a "!=". | |
d6d5f795 NB |
2182 | |
2183 | Suppose instead that we lex by one logical line at a time, producing | |
6777db6d NB |
2184 | a token list or stack for each logical line, and when seeing the '!' |
2185 | push a CPP_NOT token on the list. Then if the '!' is part of a | |
2186 | longer token ("!=") we know we must see the remainder of the token by | |
2187 | the time we reach the end of the logical line. Thus we can have the | |
2188 | '=' handler look at the previous token (at the end of the list / top | |
2189 | of the stack) and see if it is a "!" token, and if so, instead of | |
2190 | pushing a "=" token revise the existing token to be a "!=" token. | |
d6d5f795 NB |
2191 | |
2192 | This works in the presence of escaped newlines, because the '\' would | |
2193 | have been pushed on the top of the stack as a CPP_BACKSLASH. The | |
2194 | newline ('\n' or '\r') handler looks at the token at the top of the | |
2195 | stack to see if it is a CPP_BACKSLASH, and if so discards both. | |
2196 | Otherwise it pushes the newline (CPP_VSPACE) token as normal. Hence | |
2197 | the '=' handler would never see any intervening escaped newlines. | |
2198 | ||
2199 | To make trigraphs work in this context, as in precedence trigraphs | |
2200 | are highest and converted before anything else, the '?' handler does | |
2201 | lookahead to see if it is a trigraph, and if so skips the trigraph | |
2202 | and pushes the token it represents onto the top of the stack. This | |
2203 | also works in the particular case of a CPP_BACKSLASH trigraph. | |
2204 | ||
2205 | To the preprocessor, whitespace is only significant to the point of | |
2206 | knowing whether whitespace precedes a particular token. For example, | |
2207 | the '=' handler needs to know whether there was whitespace between it | |
6777db6d | 2208 | and a "!" token on the top of the stack, to make the token conversion |
d6d5f795 NB |
2209 | decision correctly. So each token has a PREV_WHITESPACE flag to |
2210 | indicate this - the standard permits consecutive whitespace to be | |
2211 | regarded as a single space. The compiler front ends are not | |
2212 | interested in whitespace at all; they just require a token stream. | |
2213 | Another place where whitespace is significant to the preprocessor is | |
2214 | a #define statment - if there is whitespace between the macro name | |
2215 | and an initial "(" token the macro is "object-like", otherwise it is | |
2216 | a function-like macro that takes arguments. | |
2217 | ||
2218 | However, all is not rosy. Parsing of identifiers, numbers, comments | |
2219 | and strings becomes trickier because of the possibility of raw | |
2220 | trigraphs and escaped newlines in the input stream. | |
2221 | ||
2222 | The trigraphs are three consecutive characters beginning with two | |
c2e25d51 NB |
2223 | question marks. A question mark is not valid as part of a number or |
2224 | identifier, so parsing of a number or identifier terminates normally | |
2225 | upon reaching it, returning to the mainloop which handles the | |
2226 | trigraph just like it would in any other position. Similarly for the | |
2227 | backslash of a backslash-newline combination. So we just need the | |
2228 | escaped-newline dropper in the mainloop to check if the token on the | |
2229 | top of the stack after dropping the escaped newline is a number or | |
2230 | identifier, and if so to continue the processing it as if nothing had | |
2231 | happened. | |
d6d5f795 NB |
2232 | |
2233 | For strings, we replace trigraphs whenever we reach a quote or | |
2234 | newline, because there might be a backslash trigraph escaping them. | |
2235 | We need to be careful that we start trigraph replacing from where we | |
2236 | left off previously, because it is possible for a first scan to leave | |
2237 | "fake" trigraphs that a second scan would pick up as real (e.g. the | |
c2e25d51 | 2238 | sequence "????/\n=" would find a fake ??= trigraph after removing the |
d6d5f795 NB |
2239 | escaped newline.) |
2240 | ||
2241 | For line comments, on reaching a newline we scan the previous | |
2242 | character(s) to see if it escaped, and continue if it is. Block | |
2243 | comments ignore everything and just focus on finding the comment | |
2244 | termination mark. The only difficult thing, and it is surprisingly | |
2245 | tricky, is checking if an asterisk precedes the final slash since | |
2246 | they could be separated by escaped newlines. If the preprocessor is | |
2247 | invoked with the output comments option, we don't bother removing | |
2248 | escaped newlines and replacing trigraphs for output. | |
2249 | ||
2250 | Finally, numbers can begin with a period, which is pushed initially | |
2251 | as a CPP_DOT token in its own right. The digit handler checks if the | |
2252 | previous token was a CPP_DOT not separated by whitespace, and if so | |
2253 | pops it off the stack and pushes a period into the number's buffer | |
2254 | before calling the number parser. | |
2255 | ||
2256 | */ | |
2257 | ||
b8f41010 NB |
2258 | static const unsigned char *digraph_spellings [] = {U"%:", U"%:%:", U"<:", |
2259 | U":>", U"<%", U"%>"}; | |
2260 | static unsigned char trigraph_map[256]; | |
c5a04734 ZW |
2261 | |
2262 | static void | |
2263 | expand_comment_space (list) | |
2264 | cpp_toklist *list; | |
2265 | { | |
2266 | if (list->comments_cap == 0) | |
2267 | { | |
2268 | list->comments_cap = 10; | |
2269 | list->comments = (cpp_token *) | |
2270 | xmalloc (list->comments_cap * sizeof (cpp_token)); | |
2271 | } | |
2272 | else | |
2273 | { | |
2274 | list->comments_cap *= 2; | |
2275 | list->comments = (cpp_token *) | |
2276 | xrealloc (list->comments, list->comments_cap); | |
2277 | } | |
2278 | } | |
2279 | ||
2280 | void | |
2281 | cpp_free_token_list (list) | |
2282 | cpp_toklist *list; | |
2283 | { | |
2284 | if (list->comments) | |
2285 | free (list->comments); | |
cfd5b8b8 | 2286 | free (list->tokens - 1); /* Backup over dummy token. */ |
c5a04734 ZW |
2287 | free (list->namebuf); |
2288 | free (list); | |
2289 | } | |
2290 | ||
c5a04734 ZW |
2291 | void |
2292 | init_trigraph_map () | |
2293 | { | |
2294 | trigraph_map['='] = '#'; | |
2295 | trigraph_map['('] = '['; | |
2296 | trigraph_map[')'] = ']'; | |
2297 | trigraph_map['/'] = '\\'; | |
2298 | trigraph_map['\''] = '^'; | |
2299 | trigraph_map['<'] = '{'; | |
2300 | trigraph_map['>'] = '}'; | |
2301 | trigraph_map['!'] = '|'; | |
2302 | trigraph_map['-'] = '~'; | |
2303 | } | |
2304 | ||
2305 | /* Call when a trigraph is encountered. It warns if necessary, and | |
2306 | returns true if the trigraph should be honoured. END is the third | |
2307 | character of a trigraph in the input stream. */ | |
2308 | static int | |
2309 | trigraph_ok (pfile, end) | |
2310 | cpp_reader *pfile; | |
2311 | const unsigned char *end; | |
2312 | { | |
2313 | int accept = CPP_OPTION (pfile, trigraphs); | |
2314 | ||
2315 | if (CPP_OPTION (pfile, warn_trigraphs)) | |
2316 | { | |
2317 | unsigned int col = end - 1 - pfile->buffer->line_base; | |
2318 | if (accept) | |
2319 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2320 | "trigraph ??%c converted to %c", | |
2321 | (int) *end, (int) trigraph_map[*end]); | |
2322 | else | |
2323 | cpp_warning_with_line (pfile, pfile->buffer->lineno, col, | |
2324 | "trigraph ??%c ignored", (int) *end); | |
2325 | } | |
2326 | return accept; | |
2327 | } | |
2328 | ||
2329 | /* Scan a string for trigraphs, warning or replacing them inline as | |
2330 | appropriate. When parsing a string, we must call this routine | |
2331 | before processing a newline character (if trigraphs are enabled), | |
2332 | since the newline might be escaped by a preceding backslash | |
2333 | trigraph sequence. Returns a pointer to the end of the name after | |
2334 | replacement. */ | |
2335 | ||
2336 | static unsigned char* | |
2337 | trigraph_replace (pfile, src, limit) | |
2338 | cpp_reader *pfile; | |
2339 | unsigned char *src; | |
2340 | unsigned char* limit; | |
2341 | { | |
2342 | unsigned char *dest; | |
2343 | ||
2344 | /* Starting with src[1], find two consecutive '?'. The case of no | |
2345 | trigraphs is streamlined. */ | |
2346 | ||
2347 | for (; src + 1 < limit; src += 2) | |
2348 | { | |
2349 | if (src[0] != '?') | |
2350 | continue; | |
2351 | ||
2352 | /* Make src point to the 1st (NOT 2nd) of two consecutive '?'s. */ | |
2353 | if (src[-1] == '?') | |
2354 | src--; | |
2355 | else if (src + 2 == limit || src[1] != '?') | |
2356 | continue; | |
2357 | ||
2358 | /* Check if it really is a trigraph. */ | |
2359 | if (trigraph_map[src[2]] == 0) | |
2360 | continue; | |
2361 | ||
2362 | dest = src; | |
2363 | goto trigraph_found; | |
2364 | } | |
2365 | return limit; | |
2366 | ||
2367 | /* Now we have a trigraph, we need to scan the remaining buffer, and | |
2368 | copy-shifting its contents left if replacement is enabled. */ | |
2369 | for (; src + 2 < limit; dest++, src++) | |
2370 | if ((*dest = *src) == '?' && src[1] == '?' && trigraph_map[src[2]]) | |
2371 | { | |
2372 | trigraph_found: | |
2373 | src += 2; | |
2374 | if (trigraph_ok (pfile, pfile->buffer->cur - (limit - src))) | |
2375 | *dest = trigraph_map[*src]; | |
2376 | } | |
2377 | ||
2378 | /* Copy remaining (at most 2) characters. */ | |
2379 | while (src < limit) | |
2380 | *dest++ = *src++; | |
2381 | return dest; | |
2382 | } | |
2383 | ||
2384 | /* If CUR is a backslash or the end of a trigraphed backslash, return | |
2385 | a pointer to its beginning, otherwise NULL. We don't read beyond | |
2386 | the buffer start, because there is the start of the comment in the | |
2387 | buffer. */ | |
2388 | static const unsigned char * | |
2389 | backslash_start (pfile, cur) | |
2390 | cpp_reader *pfile; | |
2391 | const unsigned char *cur; | |
2392 | { | |
2393 | if (cur[0] == '\\') | |
2394 | return cur; | |
2395 | if (cur[0] == '/' && cur[-1] == '?' && cur[-2] == '?' | |
2396 | && trigraph_ok (pfile, cur)) | |
2397 | return cur - 2; | |
2398 | return 0; | |
2399 | } | |
2400 | ||
2401 | /* Skip a C-style block comment. This is probably the trickiest | |
2402 | handler. We find the end of the comment by seeing if an asterisk | |
2403 | is before every '/' we encounter. The nasty complication is that a | |
2404 | previous asterisk may be separated by one or more escaped newlines. | |
2405 | Returns non-zero if comment terminated by EOF, zero otherwise. */ | |
2406 | static int | |
b8f41010 | 2407 | skip_block_comment2 (pfile) |
c5a04734 ZW |
2408 | cpp_reader *pfile; |
2409 | { | |
2410 | cpp_buffer *buffer = pfile->buffer; | |
2411 | const unsigned char *char_after_star = 0; | |
2412 | register const unsigned char *cur = buffer->cur; | |
2413 | int seen_eof = 0; | |
2414 | ||
2415 | /* Inner loop would think the comment has ended if the first comment | |
2416 | character is a '/'. Avoid this and keep the inner loop clean by | |
2417 | skipping such a character. */ | |
2418 | if (cur < buffer->rlimit && cur[0] == '/') | |
2419 | cur++; | |
2420 | ||
2421 | for (; cur < buffer->rlimit; ) | |
2422 | { | |
2423 | unsigned char c = *cur++; | |
2424 | ||
2425 | /* People like decorating comments with '*', so check for | |
2426 | '/' instead for efficiency. */ | |
2427 | if (c == '/') | |
2428 | { | |
2429 | if (cur[-2] == '*' || cur - 1 == char_after_star) | |
2430 | goto out; | |
2431 | ||
2432 | /* Warn about potential nested comments, but not when | |
2433 | the final character inside the comment is a '/'. | |
2434 | Don't bother to get it right across escaped newlines. */ | |
2435 | if (CPP_OPTION (pfile, warn_comments) && cur + 1 < buffer->rlimit | |
2436 | && cur[0] == '*' && cur[1] != '/') | |
2437 | { | |
2438 | buffer->cur = cur; | |
2439 | cpp_warning (pfile, "'/*' within comment"); | |
2440 | } | |
2441 | } | |
2442 | else if (IS_NEWLINE(c)) | |
2443 | { | |
2444 | const unsigned char* bslash = backslash_start (pfile, cur - 2); | |
2445 | ||
2446 | handle_newline (cur, buffer->rlimit, c); | |
2447 | /* Work correctly if there is an asterisk before an | |
2448 | arbirtrarily long sequence of escaped newlines. */ | |
2449 | if (bslash && (bslash[-1] == '*' || bslash == char_after_star)) | |
2450 | char_after_star = cur; | |
2451 | else | |
2452 | char_after_star = 0; | |
2453 | } | |
2454 | } | |
2455 | seen_eof = 1; | |
2456 | ||
2457 | out: | |
2458 | buffer->cur = cur; | |
2459 | return seen_eof; | |
2460 | } | |
2461 | ||
2462 | /* Skip a C++ or Chill line comment. Handles escaped newlines. | |
2463 | Returns non-zero if a multiline comment. */ | |
2464 | static int | |
b8f41010 | 2465 | skip_line_comment2 (pfile) |
c5a04734 ZW |
2466 | cpp_reader *pfile; |
2467 | { | |
2468 | cpp_buffer *buffer = pfile->buffer; | |
2469 | register const unsigned char *cur = buffer->cur; | |
2470 | int multiline = 0; | |
2471 | ||
2472 | for (; cur < buffer->rlimit; ) | |
2473 | { | |
2474 | unsigned char c = *cur++; | |
2475 | ||
2476 | if (IS_NEWLINE (c)) | |
2477 | { | |
2478 | /* Check for a (trigaph?) backslash escaping the newline. */ | |
2479 | if (!backslash_start (pfile, cur - 2)) | |
2480 | goto out; | |
2481 | multiline = 1; | |
2482 | handle_newline (cur, buffer->rlimit, c); | |
2483 | } | |
2484 | } | |
2485 | cur++; | |
2486 | ||
2487 | out: | |
2488 | buffer->cur = cur - 1; /* Leave newline for caller. */ | |
2489 | return multiline; | |
2490 | } | |
2491 | ||
2492 | /* Skips whitespace, stopping at next non-whitespace character. */ | |
2493 | static void | |
2494 | skip_whitespace (pfile, in_directive) | |
2495 | cpp_reader *pfile; | |
2496 | int in_directive; | |
2497 | { | |
2498 | cpp_buffer *buffer = pfile->buffer; | |
2499 | register const unsigned char *cur = buffer->cur; | |
2500 | unsigned short null_count = 0; | |
2501 | ||
2502 | for (; cur < buffer->rlimit; ) | |
2503 | { | |
2504 | unsigned char c = *cur++; | |
2505 | ||
2506 | if (IS_HSPACE(c)) /* FIXME: Fix ISTABLE. */ | |
2507 | continue; | |
2508 | if (!is_space(c) || IS_NEWLINE (c)) /* Main loop handles newlines. */ | |
2509 | goto out; | |
2510 | if (c == '\0') | |
2511 | null_count++; | |
2512 | /* Mut be '\f' or '\v' */ | |
2513 | else if (in_directive && CPP_PEDANTIC (pfile)) | |
2514 | cpp_pedwarn (pfile, "%s in preprocessing directive", | |
2515 | c == '\f' ? "formfeed" : "vertical tab"); | |
2516 | } | |
2517 | cur++; | |
2518 | ||
2519 | out: | |
2520 | buffer->cur = cur - 1; | |
2521 | if (null_count) | |
2522 | cpp_warning (pfile, null_count > 1 ? "embedded null characters ignored" | |
2523 | : "embedded null character ignored"); | |
2524 | } | |
2525 | ||
2526 | /* Parse (append) an identifier. */ | |
2527 | static void | |
2528 | parse_name (pfile, list, name) | |
2529 | cpp_reader *pfile; | |
2530 | cpp_toklist *list; | |
2531 | cpp_name *name; | |
2532 | { | |
2533 | const unsigned char *name_limit; | |
2534 | unsigned char *namebuf; | |
2535 | cpp_buffer *buffer = pfile->buffer; | |
2536 | register const unsigned char *cur = buffer->cur; | |
2537 | ||
2538 | expanded: | |
2539 | name_limit = list->namebuf + list->name_cap; | |
2540 | namebuf = list->namebuf + list->name_used; | |
2541 | ||
2542 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2543 | { | |
2544 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2545 | ||
2546 | if (! is_idchar(c)) | |
2547 | goto out; | |
2548 | namebuf++; | |
2549 | cur++; | |
2550 | if (c == '$' && CPP_PEDANTIC (pfile)) | |
2551 | { | |
2552 | buffer->cur = cur; | |
2553 | cpp_pedwarn (pfile, "'$' character in identifier"); | |
2554 | } | |
2555 | } | |
2556 | ||
2557 | /* Run out of name space? */ | |
2558 | if (cur < buffer->rlimit) | |
2559 | { | |
2560 | list->name_used = namebuf - list->namebuf; | |
2561 | auto_expand_name_space (list); | |
2562 | goto expanded; | |
2563 | } | |
2564 | ||
2565 | out: | |
2566 | buffer->cur = cur; | |
2567 | name->len = namebuf - (list->namebuf + name->offset); | |
2568 | list->name_used = namebuf - list->namebuf; | |
2569 | } | |
2570 | ||
2571 | /* Parse (append) a number. */ | |
2572 | ||
2573 | #define VALID_SIGN(c, prevc) \ | |
2574 | (((c) == '+' || (c) == '-') && \ | |
2575 | ((prevc) == 'e' || (prevc) == 'E' \ | |
2576 | || (((prevc) == 'p' || (prevc) == 'P') && !CPP_OPTION (pfile, c89)))) | |
2577 | ||
2578 | static void | |
2579 | parse_number (pfile, list, name) | |
2580 | cpp_reader *pfile; | |
2581 | cpp_toklist *list; | |
2582 | cpp_name *name; | |
2583 | { | |
2584 | const unsigned char *name_limit; | |
2585 | unsigned char *namebuf; | |
2586 | cpp_buffer *buffer = pfile->buffer; | |
2587 | register const unsigned char *cur = buffer->cur; | |
2588 | ||
2589 | expanded: | |
2590 | name_limit = list->namebuf + list->name_cap; | |
2591 | namebuf = list->namebuf + list->name_used; | |
2592 | ||
2593 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2594 | { | |
2595 | unsigned char c = *namebuf = *cur; /* Copy a single char. */ | |
2596 | ||
2597 | /* Perhaps we should accept '$' here if we accept it for | |
2598 | identifiers. We know namebuf[-1] is safe, because for c to | |
2599 | be a sign we must have pushed at least one character. */ | |
2600 | if (!is_numchar (c) && c != '.' && ! VALID_SIGN (c, namebuf[-1])) | |
2601 | goto out; | |
2602 | ||
2603 | namebuf++; | |
2604 | cur++; | |
2605 | } | |
2606 | ||
2607 | /* Run out of name space? */ | |
2608 | if (cur < buffer->rlimit) | |
2609 | { | |
2610 | list->name_used = namebuf - list->namebuf; | |
2611 | auto_expand_name_space (list); | |
2612 | goto expanded; | |
2613 | } | |
2614 | ||
2615 | out: | |
2616 | buffer->cur = cur; | |
2617 | name->len = namebuf - (list->namebuf + name->offset); | |
2618 | list->name_used = namebuf - list->namebuf; | |
2619 | } | |
2620 | ||
2621 | /* Places a string terminated by an unescaped TERMINATOR into a | |
2622 | cpp_name, which should be expandable and thus at the top of the | |
2623 | list's stack. Handles embedded trigraphs, if necessary, and | |
2624 | escaped newlines. | |
2625 | ||
2626 | Can be used for character constants (terminator = '\''), string | |
2627 | constants ('"'), angled headers ('>') and assertions (')'). */ | |
2628 | ||
2629 | static void | |
b8f41010 | 2630 | parse_string2 (pfile, list, name, terminator) |
c5a04734 ZW |
2631 | cpp_reader *pfile; |
2632 | cpp_toklist *list; | |
2633 | cpp_name *name; | |
2634 | unsigned int terminator; | |
2635 | { | |
2636 | cpp_buffer *buffer = pfile->buffer; | |
2637 | register const unsigned char *cur = buffer->cur; | |
2638 | const unsigned char *name_limit; | |
2639 | unsigned char *namebuf; | |
2640 | unsigned int null_count = 0; | |
2641 | int trigraphed_len = 0; | |
2642 | ||
2643 | expanded: | |
2644 | name_limit = list->namebuf + list->name_cap; | |
2645 | namebuf = list->namebuf + list->name_used; | |
2646 | ||
2647 | for (; cur < buffer->rlimit && namebuf < name_limit; ) | |
2648 | { | |
2649 | unsigned int c = *namebuf++ = *cur++; /* Copy a single char. */ | |
2650 | ||
2651 | if (c == '\0') | |
2652 | null_count++; | |
2653 | else if (c == terminator || IS_NEWLINE (c)) | |
2654 | { | |
2655 | unsigned char* name_start = list->namebuf + name->offset; | |
2656 | ||
2657 | /* Needed for trigraph_replace and multiline string warning. */ | |
2658 | buffer->cur = cur; | |
2659 | ||
2660 | /* Scan for trigraphs before checking if backslash-escaped. */ | |
2661 | if (CPP_OPTION (pfile, trigraphs) | |
2662 | || CPP_OPTION (pfile, warn_trigraphs)) | |
2663 | { | |
2664 | namebuf = trigraph_replace (pfile, name_start + trigraphed_len, | |
2665 | namebuf); | |
2666 | trigraphed_len = namebuf - 2 - (name_start + trigraphed_len); | |
2667 | if (trigraphed_len < 0) | |
2668 | trigraphed_len = 0; | |
2669 | } | |
2670 | ||
2671 | namebuf--; /* Drop the newline / terminator from the name. */ | |
2672 | if (IS_NEWLINE (c)) | |
2673 | { | |
2674 | /* Drop a backslash newline, and continue. */ | |
2675 | if (namebuf[-1] == '\\') | |
2676 | { | |
2677 | handle_newline (cur, buffer->rlimit, c); | |
2678 | namebuf--; | |
2679 | continue; | |
2680 | } | |
2681 | ||
2682 | cur--; | |
2683 | ||
2684 | /* In Fortran and assembly language, silently terminate | |
2685 | strings of either variety at end of line. This is a | |
2686 | kludge around not knowing where comments are in these | |
2687 | languages. */ | |
2688 | if (CPP_OPTION (pfile, lang_fortran) | |
2689 | || CPP_OPTION (pfile, lang_asm)) | |
2690 | goto out; | |
2691 | ||
2692 | /* Character constants, headers and asserts may not | |
2693 | extend over multiple lines. In Standard C, neither | |
2694 | may strings. We accept multiline strings as an | |
2695 | extension, but not in directives. */ | |
2696 | if (terminator != '"' || IS_DIRECTIVE (list)) | |
2697 | goto unterminated; | |
2698 | ||
2699 | cur++; /* Move forwards again. */ | |
2700 | ||
2701 | if (pfile->multiline_string_line == 0) | |
2702 | { | |
2703 | pfile->multiline_string_line = list->line; | |
2704 | if (CPP_PEDANTIC (pfile)) | |
2705 | cpp_pedwarn (pfile, "multi-line string constant"); | |
2706 | } | |
2707 | ||
2708 | *namebuf++ = '\n'; | |
2709 | handle_newline (cur, buffer->rlimit, c); | |
2710 | } | |
2711 | else | |
2712 | { | |
2713 | unsigned char *temp; | |
2714 | ||
2715 | /* An odd number of consecutive backslashes represents | |
2716 | an escaped terminator. */ | |
2717 | temp = namebuf - 1; | |
2718 | while (temp >= name_start && *temp == '\\') | |
2719 | temp--; | |
2720 | ||
2721 | if ((namebuf - temp) & 1) | |
2722 | goto out; | |
2723 | namebuf++; | |
2724 | } | |
2725 | } | |
2726 | } | |
2727 | ||
2728 | /* Run out of name space? */ | |
2729 | if (cur < buffer->rlimit) | |
2730 | { | |
2731 | list->name_used = namebuf - list->namebuf; | |
2732 | auto_expand_name_space (list); | |
2733 | goto expanded; | |
2734 | } | |
2735 | ||
2736 | /* We may not have trigraph-replaced the input for this code path, | |
2737 | but as the input is in error by being unterminated we don't | |
2738 | bother. Prevent warnings about no newlines at EOF. */ | |
2739 | if (IS_NEWLINE(cur[-1])) | |
2740 | cur--; | |
2741 | ||
2742 | unterminated: | |
2743 | cpp_error (pfile, "missing terminating %c character", (int) terminator); | |
2744 | ||
2745 | if (terminator == '\"' && pfile->multiline_string_line != list->line | |
2746 | && pfile->multiline_string_line != 0) | |
2747 | { | |
2748 | cpp_error_with_line (pfile, pfile->multiline_string_line, -1, | |
2749 | "possible start of unterminated string literal"); | |
2750 | pfile->multiline_string_line = 0; | |
2751 | } | |
2752 | ||
2753 | out: | |
2754 | buffer->cur = cur; | |
2755 | name->len = namebuf - (list->namebuf + name->offset); | |
2756 | list->name_used = namebuf - list->namebuf; | |
2757 | ||
2758 | if (null_count > 0) | |
2759 | cpp_warning (pfile, (null_count > 1 ? "null characters preserved" | |
2760 | : "null character preserved")); | |
2761 | } | |
2762 | ||
2763 | /* The character C helps us distinguish comment types: '*' = C style, | |
2764 | '-' = Chill-style and '/' = C++ style. For code simplicity, the | |
2765 | stored comment includes any C-style comment terminator. */ | |
2766 | static void | |
b8f41010 | 2767 | save_comment (list, from, len, tok_no, type) |
c5a04734 ZW |
2768 | cpp_toklist *list; |
2769 | const unsigned char *from; | |
2770 | unsigned int len; | |
2771 | unsigned int tok_no; | |
2772 | unsigned int type; | |
2773 | { | |
2774 | cpp_token *comment; | |
2775 | ||
2776 | if (list->comments_used == list->comments_cap) | |
2777 | expand_comment_space (list); | |
2778 | ||
2779 | if (list->name_used + len > list->name_cap) | |
2780 | expand_name_space (list, len); | |
2781 | ||
2782 | comment = &list->comments[list->comments_used++]; | |
2783 | comment->type = type; | |
2784 | comment->aux = tok_no; | |
2785 | comment->val.name.len = len; | |
2786 | comment->val.name.offset = list->name_used; | |
2787 | ||
2788 | memcpy (list->namebuf + list->name_used, from, len); | |
2789 | list->name_used += len; | |
2790 | } | |
2791 | ||
2792 | /* | |
2793 | * The tokenizer's main loop. Returns a token list, representing a | |
2794 | * logical line in the input file, terminated with a CPP_VSPACE | |
2795 | * token. On EOF, a token list containing the single CPP_EOF token | |
2796 | * is returned. | |
2797 | * | |
2798 | * Implementation relies almost entirely on lookback, rather than | |
2799 | * looking forwards. This means that tokenization requires just | |
2800 | * a single pass of the file, even in the presence of trigraphs and | |
2801 | * escaped newlines, providing significant performance benefits. | |
2802 | * Trigraph overhead is negligible if they are disabled, and low | |
2803 | * even when enabled. | |
2804 | */ | |
2805 | ||
c5a04734 ZW |
2806 | void |
2807 | _cpp_lex_line (pfile, list) | |
2808 | cpp_reader *pfile; | |
2809 | cpp_toklist *list; | |
2810 | { | |
2811 | cpp_token *cur_token, *token_limit; | |
2812 | cpp_buffer *buffer = pfile->buffer; | |
2813 | register const unsigned char *cur = buffer->cur; | |
2814 | unsigned char flags = 0; | |
2815 | ||
2816 | expanded: | |
2817 | token_limit = list->tokens + list->tokens_cap; | |
2818 | cur_token = list->tokens + list->tokens_used; | |
2819 | ||
2820 | for (; cur < buffer->rlimit && cur_token < token_limit;) | |
2821 | { | |
2822 | unsigned char c = *cur++; | |
2823 | ||
2824 | /* Optimize whitespace skipping, in particular the case of a | |
2825 | single whitespace character, as every other token is probably | |
2826 | whitespace. (' ' '\t' '\v' '\f' '\0'). */ | |
2827 | if (is_hspace ((unsigned int) c)) | |
2828 | { | |
2829 | if (c == '\0' || (cur < buffer->rlimit && is_hspace (*cur))) | |
2830 | { | |
2831 | buffer->cur = cur - (c == '\0'); /* Get the null warning. */ | |
2832 | skip_whitespace (pfile, IS_DIRECTIVE (list)); | |
2833 | cur = buffer->cur; | |
2834 | } | |
2835 | flags = PREV_WHITESPACE; | |
2836 | if (cur == buffer->rlimit) | |
2837 | break; | |
2838 | c = *cur++; | |
2839 | } | |
2840 | ||
2841 | /* Initialize current token. Its type is set in the switch. */ | |
2842 | cur_token->col = COLUMN (cur); | |
2843 | cur_token->flags = flags; | |
2844 | flags = 0; | |
2845 | ||
2846 | switch (c) | |
2847 | { | |
2848 | case '0': case '1': case '2': case '3': case '4': | |
2849 | case '5': case '6': case '7': case '8': case '9': | |
2850 | /* Prepend an immediately previous CPP_DOT token. */ | |
2851 | if (PREV_TOKEN_TYPE == CPP_DOT && IMMED_TOKEN ()) | |
2852 | { | |
2853 | cur_token--; | |
2854 | if (list->name_cap == list->name_used) | |
2855 | auto_expand_name_space (list); | |
2856 | ||
2857 | cur_token->val.name.len = 1; | |
2858 | cur_token->val.name.offset = list->name_used; | |
2859 | list->namebuf[list->name_used++] = '.'; | |
2860 | } | |
2861 | else | |
2862 | INIT_NAME (list, cur_token->val.name); | |
2863 | cur--; /* Backup character. */ | |
2864 | ||
2865 | continue_number: | |
2866 | buffer->cur = cur; | |
2867 | parse_number (pfile, list, &cur_token->val.name); | |
2868 | cur = buffer->cur; | |
2869 | ||
2870 | PUSH_TOKEN (CPP_NUMBER); /* Number not yet interpreted. */ | |
2871 | break; | |
2872 | ||
2873 | letter: | |
2874 | case '_': | |
2875 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
2876 | case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': | |
2877 | case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': | |
2878 | case 's': case 't': case 'u': case 'v': case 'w': case 'x': | |
2879 | case 'y': case 'z': | |
2880 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
2881 | case 'G': case 'H': case 'I': case 'J': case 'K': case 'L': | |
2882 | case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': | |
2883 | case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': | |
2884 | case 'Y': case 'Z': | |
2885 | INIT_NAME (list, cur_token->val.name); | |
2886 | cur--; /* Backup character. */ | |
2887 | cur_token->type = CPP_NAME; /* Identifier, macro etc. */ | |
2888 | ||
2889 | continue_name: | |
2890 | buffer->cur = cur; | |
2891 | parse_name (pfile, list, &cur_token->val.name); | |
2892 | cur = buffer->cur; | |
2893 | ||
2894 | /* Find handler for newly created / extended directive. */ | |
2895 | if (IS_DIRECTIVE (list) && cur_token == &list->tokens[1]) | |
2896 | _cpp_check_directive (list, cur_token); | |
2897 | cur_token++; | |
2898 | break; | |
2899 | ||
2900 | case '\'': | |
2901 | /* Fall through. */ | |
2902 | case '\"': | |
2903 | cur_token->type = c == '\'' ? CPP_CHAR : CPP_STRING; | |
2904 | /* Do we have a wide string? */ | |
2905 | if (cur_token[-1].type == CPP_NAME && IMMED_TOKEN () | |
2906 | && cur_token[-1].val.name.len == 1 | |
1920de47 | 2907 | && *(list->namebuf + cur_token[-1].val.name.offset) == 'L' |
c5a04734 ZW |
2908 | && !CPP_TRADITIONAL (pfile)) |
2909 | { | |
2910 | /* No need for 'L' any more. */ | |
2911 | list->name_used--; | |
2912 | (--cur_token)->type = (c == '\'' ? CPP_WCHAR : CPP_WSTRING); | |
2913 | } | |
2914 | ||
2915 | do_parse_string: | |
2916 | /* Here c is one of ' " > or ). */ | |
2917 | INIT_NAME (list, cur_token->val.name); | |
2918 | buffer->cur = cur; | |
b8f41010 | 2919 | parse_string2 (pfile, list, &cur_token->val.name, c); |
c5a04734 ZW |
2920 | cur = buffer->cur; |
2921 | cur_token++; | |
2922 | break; | |
2923 | ||
2924 | case '/': | |
2925 | cur_token->type = CPP_DIV; | |
2926 | if (IMMED_TOKEN ()) | |
2927 | { | |
2928 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
2929 | { | |
2930 | /* We silently allow C++ comments in system headers, | |
2931 | irrespective of conformance mode, because lots of | |
2932 | broken systems do that and trying to clean it up | |
2933 | in fixincludes is a nightmare. */ | |
2934 | if (buffer->system_header_p) | |
2935 | goto do_line_comment; | |
2936 | else if (CPP_OPTION (pfile, cplusplus_comments)) | |
2937 | { | |
2938 | if (CPP_OPTION (pfile, c89) && CPP_PEDANTIC (pfile) | |
2939 | && ! buffer->warned_cplusplus_comments) | |
2940 | { | |
2941 | buffer->cur = cur; | |
2942 | cpp_pedwarn (pfile, | |
2943 | "C++ style comments are not allowed in ISO C89"); | |
2944 | cpp_pedwarn (pfile, | |
2945 | "(this will be reported only once per input file)"); | |
2946 | buffer->warned_cplusplus_comments = 1; | |
2947 | } | |
2948 | do_line_comment: | |
2949 | buffer->cur = cur; | |
2950 | if (cur[-2] != c) | |
2951 | cpp_warning (pfile, | |
2952 | "comment start split across lines"); | |
b8f41010 | 2953 | if (skip_line_comment2 (pfile)) |
c5a04734 ZW |
2954 | cpp_error_with_line (pfile, list->line, |
2955 | cur_token[-1].col, | |
2956 | "multi-line comment"); | |
2957 | if (!CPP_OPTION (pfile, discard_comments)) | |
b8f41010 | 2958 | save_comment (list, cur, buffer->cur - cur, |
c5a04734 ZW |
2959 | cur_token - 1 - list->tokens, c == '/' |
2960 | ? CPP_CPP_COMMENT: CPP_CHILL_COMMENT); | |
2961 | cur = buffer->cur; | |
2962 | ||
2963 | /* Back-up to first '-' or '/'. */ | |
2964 | cur_token -= 2; | |
2965 | if (!CPP_OPTION (pfile, traditional)) | |
2966 | flags = PREV_WHITESPACE; | |
2967 | } | |
2968 | } | |
2969 | } | |
2970 | cur_token++; | |
2971 | break; | |
2972 | ||
2973 | case '*': | |
2974 | cur_token->type = CPP_MULT; | |
2975 | if (IMMED_TOKEN ()) | |
2976 | { | |
2977 | if (PREV_TOKEN_TYPE == CPP_DIV) | |
2978 | { | |
2979 | buffer->cur = cur; | |
2980 | if (cur[-2] != '/') | |
2981 | cpp_warning (pfile, | |
2982 | "comment start '/*' split across lines"); | |
b8f41010 | 2983 | if (skip_block_comment2 (pfile)) |
c5a04734 ZW |
2984 | cpp_error_with_line (pfile, list->line, cur_token[-1].col, |
2985 | "unterminated comment"); | |
2986 | else if (buffer->cur[-2] != '*') | |
2987 | cpp_warning (pfile, | |
2988 | "comment end '*/' split across lines"); | |
2989 | if (!CPP_OPTION (pfile, discard_comments)) | |
b8f41010 | 2990 | save_comment (list, cur, buffer->cur - cur, |
c5a04734 ZW |
2991 | cur_token - 1 - list->tokens, CPP_C_COMMENT); |
2992 | cur = buffer->cur; | |
2993 | ||
2994 | cur_token -= 2; | |
2995 | if (!CPP_OPTION (pfile, traditional)) | |
2996 | flags = PREV_WHITESPACE; | |
2997 | } | |
2998 | else if (CPP_OPTION (pfile, cplusplus)) | |
2999 | { | |
3000 | /* In C++, there are .* and ->* operators. */ | |
3001 | if (PREV_TOKEN_TYPE == CPP_DEREF) | |
3002 | BACKUP_TOKEN (CPP_DEREF_STAR); | |
3003 | else if (PREV_TOKEN_TYPE == CPP_DOT) | |
3004 | BACKUP_TOKEN (CPP_DOT_STAR); | |
3005 | } | |
3006 | } | |
3007 | cur_token++; | |
3008 | break; | |
3009 | ||
3010 | case '\n': | |
3011 | case '\r': | |
3012 | handle_newline (cur, buffer->rlimit, c); | |
3013 | if (PREV_TOKEN_TYPE != CPP_BACKSLASH || !IMMED_TOKEN ()) | |
3014 | { | |
3015 | if (PREV_TOKEN_TYPE == CPP_BACKSLASH) | |
3016 | { | |
3017 | buffer->cur = cur; | |
3018 | cpp_warning (pfile, | |
3019 | "backslash and newline separated by space"); | |
3020 | } | |
3021 | PUSH_TOKEN (CPP_VSPACE); | |
3022 | goto out; | |
3023 | } | |
3024 | /* Remove the escaped newline. Then continue to process | |
3025 | any interrupted name or number. */ | |
3026 | cur_token--; | |
3027 | if (IMMED_TOKEN ()) | |
3028 | { | |
3029 | cur_token--; | |
3030 | if (cur_token->type == CPP_NAME) | |
3031 | goto continue_name; | |
3032 | else if (cur_token->type == CPP_NUMBER) | |
3033 | goto continue_number; | |
3034 | cur_token++; | |
3035 | } | |
3036 | break; | |
3037 | ||
3038 | case '-': | |
3039 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_MINUS) | |
3040 | { | |
3041 | if (CPP_OPTION (pfile, chill)) | |
3042 | goto do_line_comment; | |
3043 | REVISE_TOKEN (CPP_MINUS_MINUS); | |
3044 | } | |
3045 | else | |
3046 | PUSH_TOKEN (CPP_MINUS); | |
3047 | break; | |
3048 | ||
3049 | /* The digraph flag checking ensures that ## and %:%: | |
3050 | are interpreted as CPP_PASTE, but #%: and %:# are not. */ | |
3051 | make_hash: | |
3052 | case '#': | |
3053 | if (PREV_TOKEN_TYPE == CPP_HASH && IMMED_TOKEN () | |
3054 | && ((cur_token->flags ^ cur_token[-1].flags) & DIGRAPH) == 0) | |
3055 | REVISE_TOKEN (CPP_PASTE); | |
3056 | else | |
3057 | PUSH_TOKEN (CPP_HASH); | |
3058 | break; | |
3059 | ||
3060 | case ':': | |
3061 | cur_token->type = CPP_COLON; | |
3062 | if (IMMED_TOKEN ()) | |
3063 | { | |
3064 | if (PREV_TOKEN_TYPE == CPP_COLON | |
3065 | && CPP_OPTION (pfile, cplusplus)) | |
3066 | BACKUP_TOKEN (CPP_SCOPE); | |
3067 | /* Digraph: "<:" is a '[' */ | |
3068 | else if (PREV_TOKEN_TYPE == CPP_LESS) | |
3069 | BACKUP_DIGRAPH (CPP_OPEN_SQUARE); | |
3070 | /* Digraph: "%:" is a '#' */ | |
3071 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3072 | { | |
3073 | (--cur_token)->flags |= DIGRAPH; | |
3074 | goto make_hash; | |
3075 | } | |
3076 | } | |
3077 | cur_token++; | |
3078 | break; | |
3079 | ||
3080 | case '&': | |
3081 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_AND) | |
3082 | REVISE_TOKEN (CPP_AND_AND); | |
3083 | else | |
3084 | PUSH_TOKEN (CPP_AND); | |
3085 | break; | |
3086 | ||
3087 | make_or: | |
3088 | case '|': | |
3089 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_OR) | |
3090 | REVISE_TOKEN (CPP_OR_OR); | |
3091 | else | |
3092 | PUSH_TOKEN (CPP_OR); | |
3093 | break; | |
3094 | ||
3095 | case '+': | |
3096 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_PLUS) | |
3097 | REVISE_TOKEN (CPP_PLUS_PLUS); | |
3098 | else | |
3099 | PUSH_TOKEN (CPP_PLUS); | |
3100 | break; | |
3101 | ||
3102 | case '=': | |
3103 | /* This relies on equidistance of "?=" and "?" tokens. */ | |
3104 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE <= CPP_LAST_EQ) | |
3105 | REVISE_TOKEN (PREV_TOKEN_TYPE + (CPP_EQ_EQ - CPP_EQ)); | |
3106 | else | |
3107 | PUSH_TOKEN (CPP_EQ); | |
3108 | break; | |
3109 | ||
3110 | case '>': | |
3111 | cur_token->type = CPP_GREATER; | |
3112 | if (IMMED_TOKEN ()) | |
3113 | { | |
3114 | if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3115 | BACKUP_TOKEN (CPP_RSHIFT); | |
3116 | else if (PREV_TOKEN_TYPE == CPP_MINUS) | |
3117 | BACKUP_TOKEN (CPP_DEREF); | |
3118 | /* Digraph: ":>" is a ']' */ | |
3119 | else if (PREV_TOKEN_TYPE == CPP_COLON) | |
3120 | BACKUP_DIGRAPH (CPP_CLOSE_SQUARE); | |
3121 | /* Digraph: "%>" is a '}' */ | |
3122 | else if (PREV_TOKEN_TYPE == CPP_MOD) | |
3123 | BACKUP_DIGRAPH (CPP_CLOSE_BRACE); | |
3124 | } | |
3125 | cur_token++; | |
3126 | break; | |
3127 | ||
3128 | case '<': | |
3129 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3130 | { | |
3131 | REVISE_TOKEN (CPP_LSHIFT); | |
3132 | break; | |
3133 | } | |
3134 | /* Is this the beginning of a header name? */ | |
3135 | if (list->dir_flags & SYNTAX_INCLUDE) | |
3136 | { | |
3137 | c = '>'; /* Terminator. */ | |
3138 | cur_token->type = CPP_HEADER_NAME; | |
3139 | goto do_parse_string; | |
3140 | } | |
3141 | PUSH_TOKEN (CPP_LESS); | |
3142 | break; | |
3143 | ||
3144 | case '%': | |
3145 | /* Digraph: "<%" is a '{' */ | |
3146 | cur_token->type = CPP_MOD; | |
3147 | if (IMMED_TOKEN () && PREV_TOKEN_TYPE == CPP_LESS) | |
3148 | BACKUP_DIGRAPH (CPP_OPEN_BRACE); | |
3149 | cur_token++; | |
3150 | break; | |
3151 | ||
c5a04734 ZW |
3152 | case '(': |
3153 | /* Is this the beginning of an assertion string? */ | |
3154 | if (list->dir_flags & SYNTAX_ASSERT) | |
3155 | { | |
3156 | c = ')'; /* Terminator. */ | |
3157 | cur_token->type = CPP_ASSERTION; | |
3158 | goto do_parse_string; | |
3159 | } | |
3160 | PUSH_TOKEN (CPP_OPEN_PAREN); | |
3161 | break; | |
3162 | ||
c5a04734 ZW |
3163 | case '?': |
3164 | if (cur + 1 < buffer->rlimit && *cur == '?' | |
3165 | && trigraph_map[cur[1]] && trigraph_ok (pfile, cur + 1)) | |
3166 | { | |
3167 | /* Handle trigraph. */ | |
3168 | cur++; | |
3169 | switch (*cur++) | |
3170 | { | |
3171 | case '(': goto make_open_square; | |
3172 | case ')': goto make_close_square; | |
3173 | case '<': goto make_open_brace; | |
3174 | case '>': goto make_close_brace; | |
3175 | case '=': goto make_hash; | |
3176 | case '!': goto make_or; | |
3177 | case '-': goto make_complement; | |
3178 | case '/': goto make_backslash; | |
3179 | case '\'': goto make_xor; | |
3180 | } | |
3181 | } | |
3182 | if (IMMED_TOKEN () && CPP_OPTION (pfile, cplusplus)) | |
3183 | { | |
3184 | /* GNU C++ defines <? and >? operators. */ | |
3185 | if (PREV_TOKEN_TYPE == CPP_LESS) | |
3186 | { | |
3187 | REVISE_TOKEN (CPP_MIN); | |
3188 | break; | |
3189 | } | |
3190 | else if (PREV_TOKEN_TYPE == CPP_GREATER) | |
3191 | { | |
3192 | REVISE_TOKEN (CPP_MAX); | |
3193 | break; | |
3194 | } | |
3195 | } | |
3196 | PUSH_TOKEN (CPP_QUERY); | |
3197 | break; | |
3198 | ||
3199 | case '.': | |
3200 | if (PREV_TOKEN_TYPE == CPP_DOT && cur_token[-2].type == CPP_DOT | |
3201 | && IMMED_TOKEN () | |
3202 | && !(cur_token[-1].flags & PREV_WHITESPACE)) | |
3203 | { | |
3204 | cur_token -= 2; | |
3205 | PUSH_TOKEN (CPP_ELLIPSIS); | |
3206 | } | |
3207 | else | |
3208 | PUSH_TOKEN (CPP_DOT); | |
3209 | break; | |
3210 | ||
cfd5b8b8 NB |
3211 | make_complement: |
3212 | case '~': PUSH_TOKEN (CPP_COMPL); break; | |
c5a04734 ZW |
3213 | make_xor: |
3214 | case '^': PUSH_TOKEN (CPP_XOR); break; | |
3215 | make_open_brace: | |
3216 | case '{': PUSH_TOKEN (CPP_OPEN_BRACE); break; | |
3217 | make_close_brace: | |
3218 | case '}': PUSH_TOKEN (CPP_CLOSE_BRACE); break; | |
3219 | make_open_square: | |
3220 | case '[': PUSH_TOKEN (CPP_OPEN_SQUARE); break; | |
3221 | make_close_square: | |
3222 | case ']': PUSH_TOKEN (CPP_CLOSE_SQUARE); break; | |
3223 | make_backslash: | |
3224 | case '\\': PUSH_TOKEN (CPP_BACKSLASH); break; | |
3225 | case '!': PUSH_TOKEN (CPP_NOT); break; | |
3226 | case ',': PUSH_TOKEN (CPP_COMMA); break; | |
3227 | case ';': PUSH_TOKEN (CPP_SEMICOLON); break; | |
cfd5b8b8 | 3228 | case ')': PUSH_TOKEN (CPP_CLOSE_PAREN); break; |
c5a04734 ZW |
3229 | |
3230 | case '$': | |
3231 | if (CPP_OPTION (pfile, dollars_in_ident)) | |
3232 | goto letter; | |
3233 | /* Fall through */ | |
3234 | default: | |
3235 | cur_token->aux = c; | |
3236 | PUSH_TOKEN (CPP_OTHER); | |
3237 | break; | |
3238 | } | |
3239 | } | |
3240 | ||
3241 | /* Run out of token space? */ | |
3242 | if (cur_token == token_limit) | |
3243 | { | |
3244 | list->tokens_used = cur_token - list->tokens; | |
3245 | expand_token_space (list); | |
3246 | goto expanded; | |
3247 | } | |
3248 | ||
3249 | cur_token->type = CPP_EOF; | |
3250 | cur_token->flags = flags; | |
3251 | ||
3252 | if (cur_token != &list->tokens[0]) | |
3253 | { | |
3254 | /* Next call back will get just a CPP_EOF. */ | |
3255 | buffer->cur = cur; | |
3256 | cpp_warning (pfile, "no newline at end of file"); | |
3257 | PUSH_TOKEN (CPP_VSPACE); | |
3258 | } | |
3259 | ||
3260 | out: | |
3261 | buffer->cur = cur; | |
3262 | ||
3263 | list->tokens_used = cur_token - list->tokens; | |
3264 | ||
3265 | /* FIXME: take this check out and put it in the caller. | |
3266 | list->directive == 0 indicates an unknown directive (but null | |
3267 | directive is OK). This is the first time we can be sure the | |
3268 | directive is invalid, and thus warn about it, because it might | |
3269 | have been split by escaped newlines. Also, don't complain about | |
3270 | invalid directives in assembly source, we don't know where the | |
3271 | comments are, and # may introduce assembler pseudo-ops. */ | |
3272 | ||
3273 | if (IS_DIRECTIVE (list) && list->dir_handler == 0 | |
3274 | && list->tokens[1].type != CPP_VSPACE | |
3275 | && !CPP_OPTION (pfile, lang_asm)) | |
3276 | cpp_error_with_line (pfile, list->line, list->tokens[1].col, | |
3277 | "invalid preprocessing directive"); | |
3278 | } | |
3279 | ||
3280 | /* Token spelling functions. Used for output of a preprocessed file, | |
3281 | stringizing and token pasting. They all assume sufficient buffer | |
3282 | is allocated, and return exactly how much they used. */ | |
3283 | ||
c5a04734 ZW |
3284 | /* Needs buffer of 3 + len. */ |
3285 | unsigned int | |
3286 | spell_string (buffer, list, token) | |
3287 | unsigned char *buffer; | |
3288 | cpp_toklist *list; | |
3289 | cpp_token *token; | |
3290 | { | |
cfd5b8b8 | 3291 | unsigned char c, *orig_buff = buffer; |
c5a04734 ZW |
3292 | size_t len; |
3293 | ||
cfd5b8b8 | 3294 | if (token->type == CPP_WSTRING || token->type == CPP_WCHAR) |
c5a04734 | 3295 | *buffer++ = 'L'; |
cfd5b8b8 NB |
3296 | c = token->type == CPP_STRING || token->type == CPP_WSTRING ? '"': '\''; |
3297 | *buffer++ = c; | |
c5a04734 ZW |
3298 | |
3299 | len = token->val.name.len; | |
1920de47 | 3300 | memcpy (buffer, list->namebuf + token->val.name.offset, len); |
c5a04734 | 3301 | buffer += len; |
cfd5b8b8 | 3302 | *buffer++ = c; |
c5a04734 ZW |
3303 | return buffer - orig_buff; |
3304 | } | |
3305 | ||
3306 | /* Needs buffer of len + 2. */ | |
3307 | unsigned int | |
3308 | spell_comment (buffer, list, token) | |
3309 | unsigned char *buffer; | |
3310 | cpp_toklist *list; | |
3311 | cpp_token *token; | |
3312 | { | |
3313 | size_t len; | |
3314 | ||
3315 | if (token->type == CPP_C_COMMENT) | |
3316 | { | |
3317 | *buffer++ = '/'; | |
3318 | *buffer++ = '*'; | |
3319 | } | |
3320 | else if (token->type == CPP_CPP_COMMENT) | |
3321 | { | |
3322 | *buffer++ = '/'; | |
3323 | *buffer++ = '/'; | |
3324 | } | |
3325 | else | |
3326 | { | |
3327 | *buffer++ = '-'; | |
3328 | *buffer++ = '-'; | |
3329 | } | |
3330 | ||
3331 | len = token->val.name.len; | |
1920de47 | 3332 | memcpy (buffer, list->namebuf + token->val.name.offset, len); |
c5a04734 ZW |
3333 | |
3334 | return len + 2; | |
3335 | } | |
3336 | ||
3337 | /* Needs buffer of len. */ | |
3338 | unsigned int | |
3339 | spell_name (buffer, list, token) | |
3340 | unsigned char *buffer; | |
3341 | cpp_toklist *list; | |
3342 | cpp_token *token; | |
3343 | { | |
3344 | size_t len; | |
3345 | ||
3346 | len = token->val.name.len; | |
1920de47 | 3347 | memcpy (buffer, list->namebuf + token->val.name.offset, len); |
c5a04734 ZW |
3348 | buffer += len; |
3349 | ||
3350 | return len; | |
3351 | } | |
3352 | ||
c5a04734 ZW |
3353 | void |
3354 | _cpp_lex_file (pfile) | |
3355 | cpp_reader* pfile; | |
3356 | { | |
3357 | int recycle; | |
3358 | cpp_toklist* list; | |
3359 | ||
3360 | init_trigraph_map (); | |
3361 | list = (cpp_toklist *) xmalloc (sizeof (cpp_toklist)); | |
3362 | ||
3363 | for (recycle = 0; ;) | |
3364 | { | |
3365 | init_token_list (pfile, list, recycle); | |
3366 | recycle = 1; | |
3367 | ||
3368 | _cpp_lex_line (pfile, list); | |
3369 | if (list->tokens[0].type == CPP_EOF) | |
3370 | break; | |
3371 | ||
3372 | if (list->dir_handler) | |
3373 | { | |
3374 | if (list->dir_handler (pfile)) | |
3375 | { | |
3376 | list = (cpp_toklist *) xmalloc (sizeof (cpp_toklist)); | |
3377 | recycle = 0; | |
3378 | } | |
3379 | } | |
3380 | else | |
3381 | _cpp_output_list (pfile, list); | |
3382 | } | |
3383 | } | |
3384 | ||
b8f41010 | 3385 | /* Temporary function for illustrative purposes. */ |
c5a04734 ZW |
3386 | static void |
3387 | _cpp_output_list (pfile, list) | |
3388 | cpp_reader *pfile; | |
3389 | cpp_toklist *list; | |
3390 | { | |
3391 | unsigned int comment_no = 0; | |
3392 | cpp_token *token, *comment_token = 0; | |
3393 | ||
3394 | if (list->comments_used > 0) | |
3395 | comment_token = list->tokens + list->comments[0].aux; | |
3396 | ||
3397 | CPP_RESERVE (pfile, 2); /* Always have room for " \n". */ | |
3398 | for (token = &list->tokens[0];; token++) | |
3399 | { | |
3400 | if (token->flags & PREV_WHITESPACE) | |
3401 | { | |
3402 | /* Output comments if -C. Otherwise a space will do. */ | |
3403 | if (token == comment_token) | |
3404 | { | |
3405 | cpp_token *comment = &list->comments[comment_no]; | |
3406 | do | |
3407 | { | |
cfd5b8b8 | 3408 | CPP_RESERVE (pfile, 2 + TOKEN_LEN (comment)); |
c5a04734 ZW |
3409 | pfile->limit += spell_comment (pfile->limit, list, comment); |
3410 | comment_no++, comment++; | |
3411 | if (comment_no == list->comments_used) | |
3412 | break; | |
3413 | comment_token = comment->aux + list->tokens; | |
3414 | } | |
3415 | while (comment_token == token); | |
3416 | } | |
3417 | else | |
3418 | CPP_PUTC_Q (pfile, ' '); | |
3419 | } | |
3420 | ||
cfd5b8b8 | 3421 | CPP_RESERVE (pfile, 2 + TOKEN_LEN (token)); |
c5a04734 ZW |
3422 | switch (token_spellings[token->type].type) |
3423 | { | |
3424 | case SPELL_TEXT: | |
3425 | { | |
3426 | const unsigned char *spelling; | |
3427 | unsigned char c; | |
3428 | ||
c5a04734 | 3429 | if (token->flags & DIGRAPH) |
cfd5b8b8 | 3430 | spelling = digraph_spellings[token->type - CPP_FIRST_DIGRAPH]; |
c5a04734 ZW |
3431 | else |
3432 | spelling = token_spellings[token->type].speller; | |
3433 | ||
3434 | while ((c = *spelling++) != '\0') | |
3435 | CPP_PUTC_Q (pfile, c); | |
3436 | } | |
3437 | break; | |
3438 | ||
3439 | case SPELL_HANDLER: | |
3440 | { | |
3441 | speller s; | |
3442 | ||
3443 | s = (speller) token_spellings[token->type].speller; | |
c5a04734 ZW |
3444 | pfile->limit += s (pfile->limit, list, token); |
3445 | } | |
3446 | break; | |
3447 | ||
cfd5b8b8 NB |
3448 | case SPELL_CHAR: |
3449 | *pfile->limit++ = token->aux; | |
3450 | break; | |
3451 | ||
c5a04734 ZW |
3452 | case SPELL_EOL: |
3453 | CPP_PUTC_Q (pfile, '\n'); | |
3454 | return; | |
3455 | ||
3456 | case SPELL_NONE: | |
3457 | cpp_error (pfile, "Unwriteable token"); | |
3458 | break; | |
3459 | } | |
3460 | } | |
3461 | } | |
3462 | ||
3463 | #endif |