]>
Commit | Line | Data |
---|---|---|
1 | #include "cache.h" | |
2 | #include "grep.h" | |
3 | #include "userdiff.h" | |
4 | #include "xdiff-interface.h" | |
5 | ||
6 | void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat) | |
7 | { | |
8 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
9 | p->pattern = pat; | |
10 | p->patternlen = strlen(pat); | |
11 | p->origin = "header"; | |
12 | p->no = 0; | |
13 | p->token = GREP_PATTERN_HEAD; | |
14 | p->field = field; | |
15 | *opt->header_tail = p; | |
16 | opt->header_tail = &p->next; | |
17 | p->next = NULL; | |
18 | } | |
19 | ||
20 | void append_grep_pattern(struct grep_opt *opt, const char *pat, | |
21 | const char *origin, int no, enum grep_pat_token t) | |
22 | { | |
23 | append_grep_pat(opt, pat, strlen(pat), origin, no, t); | |
24 | } | |
25 | ||
26 | void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen, | |
27 | const char *origin, int no, enum grep_pat_token t) | |
28 | { | |
29 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
30 | p->pattern = pat; | |
31 | p->patternlen = patlen; | |
32 | p->origin = origin; | |
33 | p->no = no; | |
34 | p->token = t; | |
35 | *opt->pattern_tail = p; | |
36 | opt->pattern_tail = &p->next; | |
37 | p->next = NULL; | |
38 | } | |
39 | ||
40 | struct grep_opt *grep_opt_dup(const struct grep_opt *opt) | |
41 | { | |
42 | struct grep_pat *pat; | |
43 | struct grep_opt *ret = xmalloc(sizeof(struct grep_opt)); | |
44 | *ret = *opt; | |
45 | ||
46 | ret->pattern_list = NULL; | |
47 | ret->pattern_tail = &ret->pattern_list; | |
48 | ||
49 | for(pat = opt->pattern_list; pat != NULL; pat = pat->next) | |
50 | { | |
51 | if(pat->token == GREP_PATTERN_HEAD) | |
52 | append_header_grep_pattern(ret, pat->field, | |
53 | pat->pattern); | |
54 | else | |
55 | append_grep_pat(ret, pat->pattern, pat->patternlen, | |
56 | pat->origin, pat->no, pat->token); | |
57 | } | |
58 | ||
59 | return ret; | |
60 | } | |
61 | ||
62 | static NORETURN void compile_regexp_failed(const struct grep_pat *p, | |
63 | const char *error) | |
64 | { | |
65 | char where[1024]; | |
66 | ||
67 | if (p->no) | |
68 | sprintf(where, "In '%s' at %d, ", p->origin, p->no); | |
69 | else if (p->origin) | |
70 | sprintf(where, "%s, ", p->origin); | |
71 | else | |
72 | where[0] = 0; | |
73 | ||
74 | die("%s'%s': %s", where, p->pattern, error); | |
75 | } | |
76 | ||
77 | #ifdef USE_LIBPCRE | |
78 | static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt) | |
79 | { | |
80 | const char *error; | |
81 | int erroffset; | |
82 | int options = 0; | |
83 | ||
84 | if (opt->ignore_case) | |
85 | options |= PCRE_CASELESS; | |
86 | ||
87 | p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset, | |
88 | NULL); | |
89 | if (!p->pcre_regexp) | |
90 | compile_regexp_failed(p, error); | |
91 | ||
92 | p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error); | |
93 | if (!p->pcre_extra_info && error) | |
94 | die("%s", error); | |
95 | } | |
96 | ||
97 | static int pcrematch(struct grep_pat *p, const char *line, const char *eol, | |
98 | regmatch_t *match, int eflags) | |
99 | { | |
100 | int ovector[30], ret, flags = 0; | |
101 | ||
102 | if (eflags & REG_NOTBOL) | |
103 | flags |= PCRE_NOTBOL; | |
104 | ||
105 | ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line, | |
106 | 0, flags, ovector, ARRAY_SIZE(ovector)); | |
107 | if (ret < 0 && ret != PCRE_ERROR_NOMATCH) | |
108 | die("pcre_exec failed with error code %d", ret); | |
109 | if (ret > 0) { | |
110 | ret = 0; | |
111 | match->rm_so = ovector[0]; | |
112 | match->rm_eo = ovector[1]; | |
113 | } | |
114 | ||
115 | return ret; | |
116 | } | |
117 | ||
118 | static void free_pcre_regexp(struct grep_pat *p) | |
119 | { | |
120 | pcre_free(p->pcre_regexp); | |
121 | pcre_free(p->pcre_extra_info); | |
122 | } | |
123 | #else /* !USE_LIBPCRE */ | |
124 | static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt) | |
125 | { | |
126 | die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE"); | |
127 | } | |
128 | ||
129 | static int pcrematch(struct grep_pat *p, const char *line, const char *eol, | |
130 | regmatch_t *match, int eflags) | |
131 | { | |
132 | return 1; | |
133 | } | |
134 | ||
135 | static void free_pcre_regexp(struct grep_pat *p) | |
136 | { | |
137 | } | |
138 | #endif /* !USE_LIBPCRE */ | |
139 | ||
140 | static int is_fixed(const char *s, size_t len) | |
141 | { | |
142 | size_t i; | |
143 | ||
144 | /* regcomp cannot accept patterns with NULs so we | |
145 | * consider any pattern containing a NUL fixed. | |
146 | */ | |
147 | if (memchr(s, 0, len)) | |
148 | return 1; | |
149 | ||
150 | for (i = 0; i < len; i++) { | |
151 | if (is_regex_special(s[i])) | |
152 | return 0; | |
153 | } | |
154 | ||
155 | return 1; | |
156 | } | |
157 | ||
158 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) | |
159 | { | |
160 | int err; | |
161 | ||
162 | p->word_regexp = opt->word_regexp; | |
163 | p->ignore_case = opt->ignore_case; | |
164 | ||
165 | if (opt->fixed || is_fixed(p->pattern, p->patternlen)) | |
166 | p->fixed = 1; | |
167 | else | |
168 | p->fixed = 0; | |
169 | ||
170 | if (p->fixed) { | |
171 | if (opt->regflags & REG_ICASE || p->ignore_case) { | |
172 | static char trans[256]; | |
173 | int i; | |
174 | for (i = 0; i < 256; i++) | |
175 | trans[i] = tolower(i); | |
176 | p->kws = kwsalloc(trans); | |
177 | } else { | |
178 | p->kws = kwsalloc(NULL); | |
179 | } | |
180 | kwsincr(p->kws, p->pattern, p->patternlen); | |
181 | kwsprep(p->kws); | |
182 | return; | |
183 | } | |
184 | ||
185 | if (opt->pcre) { | |
186 | compile_pcre_regexp(p, opt); | |
187 | return; | |
188 | } | |
189 | ||
190 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
191 | if (err) { | |
192 | char errbuf[1024]; | |
193 | regerror(err, &p->regexp, errbuf, 1024); | |
194 | regfree(&p->regexp); | |
195 | compile_regexp_failed(p, errbuf); | |
196 | } | |
197 | } | |
198 | ||
199 | static struct grep_expr *compile_pattern_or(struct grep_pat **); | |
200 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) | |
201 | { | |
202 | struct grep_pat *p; | |
203 | struct grep_expr *x; | |
204 | ||
205 | p = *list; | |
206 | if (!p) | |
207 | return NULL; | |
208 | switch (p->token) { | |
209 | case GREP_PATTERN: /* atom */ | |
210 | case GREP_PATTERN_HEAD: | |
211 | case GREP_PATTERN_BODY: | |
212 | x = xcalloc(1, sizeof (struct grep_expr)); | |
213 | x->node = GREP_NODE_ATOM; | |
214 | x->u.atom = p; | |
215 | *list = p->next; | |
216 | return x; | |
217 | case GREP_OPEN_PAREN: | |
218 | *list = p->next; | |
219 | x = compile_pattern_or(list); | |
220 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) | |
221 | die("unmatched parenthesis"); | |
222 | *list = (*list)->next; | |
223 | return x; | |
224 | default: | |
225 | return NULL; | |
226 | } | |
227 | } | |
228 | ||
229 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
230 | { | |
231 | struct grep_pat *p; | |
232 | struct grep_expr *x; | |
233 | ||
234 | p = *list; | |
235 | if (!p) | |
236 | return NULL; | |
237 | switch (p->token) { | |
238 | case GREP_NOT: | |
239 | if (!p->next) | |
240 | die("--not not followed by pattern expression"); | |
241 | *list = p->next; | |
242 | x = xcalloc(1, sizeof (struct grep_expr)); | |
243 | x->node = GREP_NODE_NOT; | |
244 | x->u.unary = compile_pattern_not(list); | |
245 | if (!x->u.unary) | |
246 | die("--not followed by non pattern expression"); | |
247 | return x; | |
248 | default: | |
249 | return compile_pattern_atom(list); | |
250 | } | |
251 | } | |
252 | ||
253 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
254 | { | |
255 | struct grep_pat *p; | |
256 | struct grep_expr *x, *y, *z; | |
257 | ||
258 | x = compile_pattern_not(list); | |
259 | p = *list; | |
260 | if (p && p->token == GREP_AND) { | |
261 | if (!p->next) | |
262 | die("--and not followed by pattern expression"); | |
263 | *list = p->next; | |
264 | y = compile_pattern_and(list); | |
265 | if (!y) | |
266 | die("--and not followed by pattern expression"); | |
267 | z = xcalloc(1, sizeof (struct grep_expr)); | |
268 | z->node = GREP_NODE_AND; | |
269 | z->u.binary.left = x; | |
270 | z->u.binary.right = y; | |
271 | return z; | |
272 | } | |
273 | return x; | |
274 | } | |
275 | ||
276 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
277 | { | |
278 | struct grep_pat *p; | |
279 | struct grep_expr *x, *y, *z; | |
280 | ||
281 | x = compile_pattern_and(list); | |
282 | p = *list; | |
283 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
284 | y = compile_pattern_or(list); | |
285 | if (!y) | |
286 | die("not a pattern expression %s", p->pattern); | |
287 | z = xcalloc(1, sizeof (struct grep_expr)); | |
288 | z->node = GREP_NODE_OR; | |
289 | z->u.binary.left = x; | |
290 | z->u.binary.right = y; | |
291 | return z; | |
292 | } | |
293 | return x; | |
294 | } | |
295 | ||
296 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
297 | { | |
298 | return compile_pattern_or(list); | |
299 | } | |
300 | ||
301 | static struct grep_expr *grep_true_expr(void) | |
302 | { | |
303 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
304 | z->node = GREP_NODE_TRUE; | |
305 | return z; | |
306 | } | |
307 | ||
308 | static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right) | |
309 | { | |
310 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
311 | z->node = GREP_NODE_OR; | |
312 | z->u.binary.left = left; | |
313 | z->u.binary.right = right; | |
314 | return z; | |
315 | } | |
316 | ||
317 | static struct grep_expr *prep_header_patterns(struct grep_opt *opt) | |
318 | { | |
319 | struct grep_pat *p; | |
320 | struct grep_expr *header_expr; | |
321 | struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]); | |
322 | enum grep_header_field fld; | |
323 | ||
324 | if (!opt->header_list) | |
325 | return NULL; | |
326 | p = opt->header_list; | |
327 | for (p = opt->header_list; p; p = p->next) { | |
328 | if (p->token != GREP_PATTERN_HEAD) | |
329 | die("bug: a non-header pattern in grep header list."); | |
330 | if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field) | |
331 | die("bug: unknown header field %d", p->field); | |
332 | compile_regexp(p, opt); | |
333 | } | |
334 | ||
335 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) | |
336 | header_group[fld] = NULL; | |
337 | ||
338 | for (p = opt->header_list; p; p = p->next) { | |
339 | struct grep_expr *h; | |
340 | struct grep_pat *pp = p; | |
341 | ||
342 | h = compile_pattern_atom(&pp); | |
343 | if (!h || pp != p->next) | |
344 | die("bug: malformed header expr"); | |
345 | if (!header_group[p->field]) { | |
346 | header_group[p->field] = h; | |
347 | continue; | |
348 | } | |
349 | header_group[p->field] = grep_or_expr(h, header_group[p->field]); | |
350 | } | |
351 | ||
352 | header_expr = NULL; | |
353 | ||
354 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) { | |
355 | if (!header_group[fld]) | |
356 | continue; | |
357 | if (!header_expr) | |
358 | header_expr = grep_true_expr(); | |
359 | header_expr = grep_or_expr(header_group[fld], header_expr); | |
360 | } | |
361 | return header_expr; | |
362 | } | |
363 | ||
364 | void compile_grep_patterns(struct grep_opt *opt) | |
365 | { | |
366 | struct grep_pat *p; | |
367 | struct grep_expr *header_expr = prep_header_patterns(opt); | |
368 | ||
369 | for (p = opt->pattern_list; p; p = p->next) { | |
370 | switch (p->token) { | |
371 | case GREP_PATTERN: /* atom */ | |
372 | case GREP_PATTERN_HEAD: | |
373 | case GREP_PATTERN_BODY: | |
374 | compile_regexp(p, opt); | |
375 | break; | |
376 | default: | |
377 | opt->extended = 1; | |
378 | break; | |
379 | } | |
380 | } | |
381 | ||
382 | if (opt->all_match || header_expr) | |
383 | opt->extended = 1; | |
384 | else if (!opt->extended) | |
385 | return; | |
386 | ||
387 | p = opt->pattern_list; | |
388 | if (p) | |
389 | opt->pattern_expression = compile_pattern_expr(&p); | |
390 | if (p) | |
391 | die("incomplete pattern expression: %s", p->pattern); | |
392 | ||
393 | if (!header_expr) | |
394 | return; | |
395 | ||
396 | if (!opt->pattern_expression) | |
397 | opt->pattern_expression = header_expr; | |
398 | else | |
399 | opt->pattern_expression = grep_or_expr(opt->pattern_expression, | |
400 | header_expr); | |
401 | opt->all_match = 1; | |
402 | } | |
403 | ||
404 | static void free_pattern_expr(struct grep_expr *x) | |
405 | { | |
406 | switch (x->node) { | |
407 | case GREP_NODE_TRUE: | |
408 | case GREP_NODE_ATOM: | |
409 | break; | |
410 | case GREP_NODE_NOT: | |
411 | free_pattern_expr(x->u.unary); | |
412 | break; | |
413 | case GREP_NODE_AND: | |
414 | case GREP_NODE_OR: | |
415 | free_pattern_expr(x->u.binary.left); | |
416 | free_pattern_expr(x->u.binary.right); | |
417 | break; | |
418 | } | |
419 | free(x); | |
420 | } | |
421 | ||
422 | void free_grep_patterns(struct grep_opt *opt) | |
423 | { | |
424 | struct grep_pat *p, *n; | |
425 | ||
426 | for (p = opt->pattern_list; p; p = n) { | |
427 | n = p->next; | |
428 | switch (p->token) { | |
429 | case GREP_PATTERN: /* atom */ | |
430 | case GREP_PATTERN_HEAD: | |
431 | case GREP_PATTERN_BODY: | |
432 | if (p->kws) | |
433 | kwsfree(p->kws); | |
434 | else if (p->pcre_regexp) | |
435 | free_pcre_regexp(p); | |
436 | else | |
437 | regfree(&p->regexp); | |
438 | break; | |
439 | default: | |
440 | break; | |
441 | } | |
442 | free(p); | |
443 | } | |
444 | ||
445 | if (!opt->extended) | |
446 | return; | |
447 | free_pattern_expr(opt->pattern_expression); | |
448 | } | |
449 | ||
450 | static char *end_of_line(char *cp, unsigned long *left) | |
451 | { | |
452 | unsigned long l = *left; | |
453 | while (l && *cp != '\n') { | |
454 | l--; | |
455 | cp++; | |
456 | } | |
457 | *left = l; | |
458 | return cp; | |
459 | } | |
460 | ||
461 | static int word_char(char ch) | |
462 | { | |
463 | return isalnum(ch) || ch == '_'; | |
464 | } | |
465 | ||
466 | static void output_color(struct grep_opt *opt, const void *data, size_t size, | |
467 | const char *color) | |
468 | { | |
469 | if (want_color(opt->color) && color && color[0]) { | |
470 | opt->output(opt, color, strlen(color)); | |
471 | opt->output(opt, data, size); | |
472 | opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET)); | |
473 | } else | |
474 | opt->output(opt, data, size); | |
475 | } | |
476 | ||
477 | static void output_sep(struct grep_opt *opt, char sign) | |
478 | { | |
479 | if (opt->null_following_name) | |
480 | opt->output(opt, "\0", 1); | |
481 | else | |
482 | output_color(opt, &sign, 1, opt->color_sep); | |
483 | } | |
484 | ||
485 | static void show_name(struct grep_opt *opt, const char *name) | |
486 | { | |
487 | output_color(opt, name, strlen(name), opt->color_filename); | |
488 | opt->output(opt, opt->null_following_name ? "\0" : "\n", 1); | |
489 | } | |
490 | ||
491 | static int fixmatch(struct grep_pat *p, char *line, char *eol, | |
492 | regmatch_t *match) | |
493 | { | |
494 | struct kwsmatch kwsm; | |
495 | size_t offset = kwsexec(p->kws, line, eol - line, &kwsm); | |
496 | if (offset == -1) { | |
497 | match->rm_so = match->rm_eo = -1; | |
498 | return REG_NOMATCH; | |
499 | } else { | |
500 | match->rm_so = offset; | |
501 | match->rm_eo = match->rm_so + kwsm.size[0]; | |
502 | return 0; | |
503 | } | |
504 | } | |
505 | ||
506 | static int regmatch(const regex_t *preg, char *line, char *eol, | |
507 | regmatch_t *match, int eflags) | |
508 | { | |
509 | #ifdef REG_STARTEND | |
510 | match->rm_so = 0; | |
511 | match->rm_eo = eol - line; | |
512 | eflags |= REG_STARTEND; | |
513 | #endif | |
514 | return regexec(preg, line, 1, match, eflags); | |
515 | } | |
516 | ||
517 | static int patmatch(struct grep_pat *p, char *line, char *eol, | |
518 | regmatch_t *match, int eflags) | |
519 | { | |
520 | int hit; | |
521 | ||
522 | if (p->fixed) | |
523 | hit = !fixmatch(p, line, eol, match); | |
524 | else if (p->pcre_regexp) | |
525 | hit = !pcrematch(p, line, eol, match, eflags); | |
526 | else | |
527 | hit = !regmatch(&p->regexp, line, eol, match, eflags); | |
528 | ||
529 | return hit; | |
530 | } | |
531 | ||
532 | static int strip_timestamp(char *bol, char **eol_p) | |
533 | { | |
534 | char *eol = *eol_p; | |
535 | int ch; | |
536 | ||
537 | while (bol < --eol) { | |
538 | if (*eol != '>') | |
539 | continue; | |
540 | *eol_p = ++eol; | |
541 | ch = *eol; | |
542 | *eol = '\0'; | |
543 | return ch; | |
544 | } | |
545 | return 0; | |
546 | } | |
547 | ||
548 | static struct { | |
549 | const char *field; | |
550 | size_t len; | |
551 | } header_field[] = { | |
552 | { "author ", 7 }, | |
553 | { "committer ", 10 }, | |
554 | }; | |
555 | ||
556 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, | |
557 | enum grep_context ctx, | |
558 | regmatch_t *pmatch, int eflags) | |
559 | { | |
560 | int hit = 0; | |
561 | int saved_ch = 0; | |
562 | const char *start = bol; | |
563 | ||
564 | if ((p->token != GREP_PATTERN) && | |
565 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
566 | return 0; | |
567 | ||
568 | if (p->token == GREP_PATTERN_HEAD) { | |
569 | const char *field; | |
570 | size_t len; | |
571 | assert(p->field < ARRAY_SIZE(header_field)); | |
572 | field = header_field[p->field].field; | |
573 | len = header_field[p->field].len; | |
574 | if (strncmp(bol, field, len)) | |
575 | return 0; | |
576 | bol += len; | |
577 | saved_ch = strip_timestamp(bol, &eol); | |
578 | } | |
579 | ||
580 | again: | |
581 | hit = patmatch(p, bol, eol, pmatch, eflags); | |
582 | ||
583 | if (hit && p->word_regexp) { | |
584 | if ((pmatch[0].rm_so < 0) || | |
585 | (eol - bol) < pmatch[0].rm_so || | |
586 | (pmatch[0].rm_eo < 0) || | |
587 | (eol - bol) < pmatch[0].rm_eo) | |
588 | die("regexp returned nonsense"); | |
589 | ||
590 | /* Match beginning must be either beginning of the | |
591 | * line, or at word boundary (i.e. the last char must | |
592 | * not be a word char). Similarly, match end must be | |
593 | * either end of the line, or at word boundary | |
594 | * (i.e. the next char must not be a word char). | |
595 | */ | |
596 | if ( ((pmatch[0].rm_so == 0) || | |
597 | !word_char(bol[pmatch[0].rm_so-1])) && | |
598 | ((pmatch[0].rm_eo == (eol-bol)) || | |
599 | !word_char(bol[pmatch[0].rm_eo])) ) | |
600 | ; | |
601 | else | |
602 | hit = 0; | |
603 | ||
604 | /* Words consist of at least one character. */ | |
605 | if (pmatch->rm_so == pmatch->rm_eo) | |
606 | hit = 0; | |
607 | ||
608 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { | |
609 | /* There could be more than one match on the | |
610 | * line, and the first match might not be | |
611 | * strict word match. But later ones could be! | |
612 | * Forward to the next possible start, i.e. the | |
613 | * next position following a non-word char. | |
614 | */ | |
615 | bol = pmatch[0].rm_so + bol + 1; | |
616 | while (word_char(bol[-1]) && bol < eol) | |
617 | bol++; | |
618 | eflags |= REG_NOTBOL; | |
619 | if (bol < eol) | |
620 | goto again; | |
621 | } | |
622 | } | |
623 | if (p->token == GREP_PATTERN_HEAD && saved_ch) | |
624 | *eol = saved_ch; | |
625 | if (hit) { | |
626 | pmatch[0].rm_so += bol - start; | |
627 | pmatch[0].rm_eo += bol - start; | |
628 | } | |
629 | return hit; | |
630 | } | |
631 | ||
632 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, | |
633 | enum grep_context ctx, int collect_hits) | |
634 | { | |
635 | int h = 0; | |
636 | regmatch_t match; | |
637 | ||
638 | if (!x) | |
639 | die("Not a valid grep expression"); | |
640 | switch (x->node) { | |
641 | case GREP_NODE_TRUE: | |
642 | h = 1; | |
643 | break; | |
644 | case GREP_NODE_ATOM: | |
645 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); | |
646 | break; | |
647 | case GREP_NODE_NOT: | |
648 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); | |
649 | break; | |
650 | case GREP_NODE_AND: | |
651 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) | |
652 | return 0; | |
653 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); | |
654 | break; | |
655 | case GREP_NODE_OR: | |
656 | if (!collect_hits) | |
657 | return (match_expr_eval(x->u.binary.left, | |
658 | bol, eol, ctx, 0) || | |
659 | match_expr_eval(x->u.binary.right, | |
660 | bol, eol, ctx, 0)); | |
661 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); | |
662 | x->u.binary.left->hit |= h; | |
663 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); | |
664 | break; | |
665 | default: | |
666 | die("Unexpected node type (internal error) %d", x->node); | |
667 | } | |
668 | if (collect_hits) | |
669 | x->hit |= h; | |
670 | return h; | |
671 | } | |
672 | ||
673 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, | |
674 | enum grep_context ctx, int collect_hits) | |
675 | { | |
676 | struct grep_expr *x = opt->pattern_expression; | |
677 | return match_expr_eval(x, bol, eol, ctx, collect_hits); | |
678 | } | |
679 | ||
680 | static int match_line(struct grep_opt *opt, char *bol, char *eol, | |
681 | enum grep_context ctx, int collect_hits) | |
682 | { | |
683 | struct grep_pat *p; | |
684 | regmatch_t match; | |
685 | ||
686 | if (opt->extended) | |
687 | return match_expr(opt, bol, eol, ctx, collect_hits); | |
688 | ||
689 | /* we do not call with collect_hits without being extended */ | |
690 | for (p = opt->pattern_list; p; p = p->next) { | |
691 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) | |
692 | return 1; | |
693 | } | |
694 | return 0; | |
695 | } | |
696 | ||
697 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, | |
698 | enum grep_context ctx, | |
699 | regmatch_t *pmatch, int eflags) | |
700 | { | |
701 | regmatch_t match; | |
702 | ||
703 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
704 | return 0; | |
705 | if (match.rm_so < 0 || match.rm_eo < 0) | |
706 | return 0; | |
707 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
708 | if (match.rm_so > pmatch->rm_so) | |
709 | return 1; | |
710 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
711 | return 1; | |
712 | } | |
713 | pmatch->rm_so = match.rm_so; | |
714 | pmatch->rm_eo = match.rm_eo; | |
715 | return 1; | |
716 | } | |
717 | ||
718 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
719 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
720 | { | |
721 | struct grep_pat *p; | |
722 | int hit = 0; | |
723 | ||
724 | pmatch->rm_so = pmatch->rm_eo = -1; | |
725 | if (bol < eol) { | |
726 | for (p = opt->pattern_list; p; p = p->next) { | |
727 | switch (p->token) { | |
728 | case GREP_PATTERN: /* atom */ | |
729 | case GREP_PATTERN_HEAD: | |
730 | case GREP_PATTERN_BODY: | |
731 | hit |= match_next_pattern(p, bol, eol, ctx, | |
732 | pmatch, eflags); | |
733 | break; | |
734 | default: | |
735 | break; | |
736 | } | |
737 | } | |
738 | } | |
739 | return hit; | |
740 | } | |
741 | ||
742 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
743 | const char *name, unsigned lno, char sign) | |
744 | { | |
745 | int rest = eol - bol; | |
746 | char *line_color = NULL; | |
747 | ||
748 | if (opt->file_break && opt->last_shown == 0) { | |
749 | if (opt->show_hunk_mark) | |
750 | opt->output(opt, "\n", 1); | |
751 | } else if (opt->pre_context || opt->post_context || opt->funcbody) { | |
752 | if (opt->last_shown == 0) { | |
753 | if (opt->show_hunk_mark) { | |
754 | output_color(opt, "--", 2, opt->color_sep); | |
755 | opt->output(opt, "\n", 1); | |
756 | } | |
757 | } else if (lno > opt->last_shown + 1) { | |
758 | output_color(opt, "--", 2, opt->color_sep); | |
759 | opt->output(opt, "\n", 1); | |
760 | } | |
761 | } | |
762 | if (opt->heading && opt->last_shown == 0) { | |
763 | output_color(opt, name, strlen(name), opt->color_filename); | |
764 | opt->output(opt, "\n", 1); | |
765 | } | |
766 | opt->last_shown = lno; | |
767 | ||
768 | if (!opt->heading && opt->pathname) { | |
769 | output_color(opt, name, strlen(name), opt->color_filename); | |
770 | output_sep(opt, sign); | |
771 | } | |
772 | if (opt->linenum) { | |
773 | char buf[32]; | |
774 | snprintf(buf, sizeof(buf), "%d", lno); | |
775 | output_color(opt, buf, strlen(buf), opt->color_lineno); | |
776 | output_sep(opt, sign); | |
777 | } | |
778 | if (opt->color) { | |
779 | regmatch_t match; | |
780 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
781 | int ch = *eol; | |
782 | int eflags = 0; | |
783 | ||
784 | if (sign == ':') | |
785 | line_color = opt->color_selected; | |
786 | else if (sign == '-') | |
787 | line_color = opt->color_context; | |
788 | else if (sign == '=') | |
789 | line_color = opt->color_function; | |
790 | *eol = '\0'; | |
791 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
792 | if (match.rm_so == match.rm_eo) | |
793 | break; | |
794 | ||
795 | output_color(opt, bol, match.rm_so, line_color); | |
796 | output_color(opt, bol + match.rm_so, | |
797 | match.rm_eo - match.rm_so, | |
798 | opt->color_match); | |
799 | bol += match.rm_eo; | |
800 | rest -= match.rm_eo; | |
801 | eflags = REG_NOTBOL; | |
802 | } | |
803 | *eol = ch; | |
804 | } | |
805 | output_color(opt, bol, rest, line_color); | |
806 | opt->output(opt, "\n", 1); | |
807 | } | |
808 | ||
809 | #ifndef NO_PTHREADS | |
810 | int grep_use_locks; | |
811 | ||
812 | /* | |
813 | * This lock protects access to the gitattributes machinery, which is | |
814 | * not thread-safe. | |
815 | */ | |
816 | pthread_mutex_t grep_attr_mutex; | |
817 | ||
818 | static inline void grep_attr_lock(void) | |
819 | { | |
820 | if (grep_use_locks) | |
821 | pthread_mutex_lock(&grep_attr_mutex); | |
822 | } | |
823 | ||
824 | static inline void grep_attr_unlock(void) | |
825 | { | |
826 | if (grep_use_locks) | |
827 | pthread_mutex_unlock(&grep_attr_mutex); | |
828 | } | |
829 | ||
830 | /* | |
831 | * Same as git_attr_mutex, but protecting the thread-unsafe object db access. | |
832 | */ | |
833 | pthread_mutex_t grep_read_mutex; | |
834 | ||
835 | #else | |
836 | #define grep_attr_lock() | |
837 | #define grep_attr_unlock() | |
838 | #endif | |
839 | ||
840 | static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol) | |
841 | { | |
842 | xdemitconf_t *xecfg = opt->priv; | |
843 | if (xecfg && !xecfg->find_func) { | |
844 | struct userdiff_driver *drv; | |
845 | grep_attr_lock(); | |
846 | drv = userdiff_find_by_path(gs->name); | |
847 | grep_attr_unlock(); | |
848 | if (drv && drv->funcname.pattern) { | |
849 | const struct userdiff_funcname *pe = &drv->funcname; | |
850 | xdiff_set_find_func(xecfg, pe->pattern, pe->cflags); | |
851 | } else { | |
852 | xecfg = opt->priv = NULL; | |
853 | } | |
854 | } | |
855 | ||
856 | if (xecfg) { | |
857 | char buf[1]; | |
858 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
859 | xecfg->find_func_priv) >= 0; | |
860 | } | |
861 | ||
862 | if (bol == eol) | |
863 | return 0; | |
864 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
865 | return 1; | |
866 | return 0; | |
867 | } | |
868 | ||
869 | static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs, | |
870 | char *bol, unsigned lno) | |
871 | { | |
872 | while (bol > gs->buf) { | |
873 | char *eol = --bol; | |
874 | ||
875 | while (bol > gs->buf && bol[-1] != '\n') | |
876 | bol--; | |
877 | lno--; | |
878 | ||
879 | if (lno <= opt->last_shown) | |
880 | break; | |
881 | ||
882 | if (match_funcname(opt, gs, bol, eol)) { | |
883 | show_line(opt, bol, eol, gs->name, lno, '='); | |
884 | break; | |
885 | } | |
886 | } | |
887 | } | |
888 | ||
889 | static void show_pre_context(struct grep_opt *opt, struct grep_source *gs, | |
890 | char *bol, char *end, unsigned lno) | |
891 | { | |
892 | unsigned cur = lno, from = 1, funcname_lno = 0; | |
893 | int funcname_needed = !!opt->funcname; | |
894 | ||
895 | if (opt->funcbody && !match_funcname(opt, gs, bol, end)) | |
896 | funcname_needed = 2; | |
897 | ||
898 | if (opt->pre_context < lno) | |
899 | from = lno - opt->pre_context; | |
900 | if (from <= opt->last_shown) | |
901 | from = opt->last_shown + 1; | |
902 | ||
903 | /* Rewind. */ | |
904 | while (bol > gs->buf && | |
905 | cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) { | |
906 | char *eol = --bol; | |
907 | ||
908 | while (bol > gs->buf && bol[-1] != '\n') | |
909 | bol--; | |
910 | cur--; | |
911 | if (funcname_needed && match_funcname(opt, gs, bol, eol)) { | |
912 | funcname_lno = cur; | |
913 | funcname_needed = 0; | |
914 | } | |
915 | } | |
916 | ||
917 | /* We need to look even further back to find a function signature. */ | |
918 | if (opt->funcname && funcname_needed) | |
919 | show_funcname_line(opt, gs, bol, cur); | |
920 | ||
921 | /* Back forward. */ | |
922 | while (cur < lno) { | |
923 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; | |
924 | ||
925 | while (*eol != '\n') | |
926 | eol++; | |
927 | show_line(opt, bol, eol, gs->name, cur, sign); | |
928 | bol = eol + 1; | |
929 | cur++; | |
930 | } | |
931 | } | |
932 | ||
933 | static int should_lookahead(struct grep_opt *opt) | |
934 | { | |
935 | struct grep_pat *p; | |
936 | ||
937 | if (opt->extended) | |
938 | return 0; /* punt for too complex stuff */ | |
939 | if (opt->invert) | |
940 | return 0; | |
941 | for (p = opt->pattern_list; p; p = p->next) { | |
942 | if (p->token != GREP_PATTERN) | |
943 | return 0; /* punt for "header only" and stuff */ | |
944 | } | |
945 | return 1; | |
946 | } | |
947 | ||
948 | static int look_ahead(struct grep_opt *opt, | |
949 | unsigned long *left_p, | |
950 | unsigned *lno_p, | |
951 | char **bol_p) | |
952 | { | |
953 | unsigned lno = *lno_p; | |
954 | char *bol = *bol_p; | |
955 | struct grep_pat *p; | |
956 | char *sp, *last_bol; | |
957 | regoff_t earliest = -1; | |
958 | ||
959 | for (p = opt->pattern_list; p; p = p->next) { | |
960 | int hit; | |
961 | regmatch_t m; | |
962 | ||
963 | hit = patmatch(p, bol, bol + *left_p, &m, 0); | |
964 | if (!hit || m.rm_so < 0 || m.rm_eo < 0) | |
965 | continue; | |
966 | if (earliest < 0 || m.rm_so < earliest) | |
967 | earliest = m.rm_so; | |
968 | } | |
969 | ||
970 | if (earliest < 0) { | |
971 | *bol_p = bol + *left_p; | |
972 | *left_p = 0; | |
973 | return 1; | |
974 | } | |
975 | for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--) | |
976 | ; /* find the beginning of the line */ | |
977 | last_bol = sp; | |
978 | ||
979 | for (sp = bol; sp < last_bol; sp++) { | |
980 | if (*sp == '\n') | |
981 | lno++; | |
982 | } | |
983 | *left_p -= last_bol - bol; | |
984 | *bol_p = last_bol; | |
985 | *lno_p = lno; | |
986 | return 0; | |
987 | } | |
988 | ||
989 | static void std_output(struct grep_opt *opt, const void *buf, size_t size) | |
990 | { | |
991 | fwrite(buf, size, 1, stdout); | |
992 | } | |
993 | ||
994 | static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits) | |
995 | { | |
996 | char *bol; | |
997 | unsigned long left; | |
998 | unsigned lno = 1; | |
999 | unsigned last_hit = 0; | |
1000 | int binary_match_only = 0; | |
1001 | unsigned count = 0; | |
1002 | int try_lookahead = 0; | |
1003 | int show_function = 0; | |
1004 | enum grep_context ctx = GREP_CONTEXT_HEAD; | |
1005 | xdemitconf_t xecfg; | |
1006 | ||
1007 | if (!opt->output) | |
1008 | opt->output = std_output; | |
1009 | ||
1010 | if (opt->pre_context || opt->post_context || opt->file_break || | |
1011 | opt->funcbody) { | |
1012 | /* Show hunk marks, except for the first file. */ | |
1013 | if (opt->last_shown) | |
1014 | opt->show_hunk_mark = 1; | |
1015 | /* | |
1016 | * If we're using threads then we can't easily identify | |
1017 | * the first file. Always put hunk marks in that case | |
1018 | * and skip the very first one later in work_done(). | |
1019 | */ | |
1020 | if (opt->output != std_output) | |
1021 | opt->show_hunk_mark = 1; | |
1022 | } | |
1023 | opt->last_shown = 0; | |
1024 | ||
1025 | if (grep_source_load(gs) < 0) | |
1026 | return 0; | |
1027 | ||
1028 | switch (opt->binary) { | |
1029 | case GREP_BINARY_DEFAULT: | |
1030 | if (buffer_is_binary(gs->buf, gs->size)) | |
1031 | binary_match_only = 1; | |
1032 | break; | |
1033 | case GREP_BINARY_NOMATCH: | |
1034 | if (buffer_is_binary(gs->buf, gs->size)) | |
1035 | return 0; /* Assume unmatch */ | |
1036 | break; | |
1037 | case GREP_BINARY_TEXT: | |
1038 | break; | |
1039 | default: | |
1040 | die("bug: unknown binary handling mode"); | |
1041 | } | |
1042 | ||
1043 | memset(&xecfg, 0, sizeof(xecfg)); | |
1044 | opt->priv = &xecfg; | |
1045 | ||
1046 | try_lookahead = should_lookahead(opt); | |
1047 | ||
1048 | bol = gs->buf; | |
1049 | left = gs->size; | |
1050 | while (left) { | |
1051 | char *eol, ch; | |
1052 | int hit; | |
1053 | ||
1054 | /* | |
1055 | * look_ahead() skips quickly to the line that possibly | |
1056 | * has the next hit; don't call it if we need to do | |
1057 | * something more than just skipping the current line | |
1058 | * in response to an unmatch for the current line. E.g. | |
1059 | * inside a post-context window, we will show the current | |
1060 | * line as a context around the previous hit when it | |
1061 | * doesn't hit. | |
1062 | */ | |
1063 | if (try_lookahead | |
1064 | && !(last_hit | |
1065 | && (show_function || | |
1066 | lno <= last_hit + opt->post_context)) | |
1067 | && look_ahead(opt, &left, &lno, &bol)) | |
1068 | break; | |
1069 | eol = end_of_line(bol, &left); | |
1070 | ch = *eol; | |
1071 | *eol = 0; | |
1072 | ||
1073 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) | |
1074 | ctx = GREP_CONTEXT_BODY; | |
1075 | ||
1076 | hit = match_line(opt, bol, eol, ctx, collect_hits); | |
1077 | *eol = ch; | |
1078 | ||
1079 | if (collect_hits) | |
1080 | goto next_line; | |
1081 | ||
1082 | /* "grep -v -e foo -e bla" should list lines | |
1083 | * that do not have either, so inversion should | |
1084 | * be done outside. | |
1085 | */ | |
1086 | if (opt->invert) | |
1087 | hit = !hit; | |
1088 | if (opt->unmatch_name_only) { | |
1089 | if (hit) | |
1090 | return 0; | |
1091 | goto next_line; | |
1092 | } | |
1093 | if (hit) { | |
1094 | count++; | |
1095 | if (opt->status_only) | |
1096 | return 1; | |
1097 | if (opt->name_only) { | |
1098 | show_name(opt, gs->name); | |
1099 | return 1; | |
1100 | } | |
1101 | if (opt->count) | |
1102 | goto next_line; | |
1103 | if (binary_match_only) { | |
1104 | opt->output(opt, "Binary file ", 12); | |
1105 | output_color(opt, gs->name, strlen(gs->name), | |
1106 | opt->color_filename); | |
1107 | opt->output(opt, " matches\n", 9); | |
1108 | return 1; | |
1109 | } | |
1110 | /* Hit at this line. If we haven't shown the | |
1111 | * pre-context lines, we would need to show them. | |
1112 | */ | |
1113 | if (opt->pre_context || opt->funcbody) | |
1114 | show_pre_context(opt, gs, bol, eol, lno); | |
1115 | else if (opt->funcname) | |
1116 | show_funcname_line(opt, gs, bol, lno); | |
1117 | show_line(opt, bol, eol, gs->name, lno, ':'); | |
1118 | last_hit = lno; | |
1119 | if (opt->funcbody) | |
1120 | show_function = 1; | |
1121 | goto next_line; | |
1122 | } | |
1123 | if (show_function && match_funcname(opt, gs, bol, eol)) | |
1124 | show_function = 0; | |
1125 | if (show_function || | |
1126 | (last_hit && lno <= last_hit + opt->post_context)) { | |
1127 | /* If the last hit is within the post context, | |
1128 | * we need to show this line. | |
1129 | */ | |
1130 | show_line(opt, bol, eol, gs->name, lno, '-'); | |
1131 | } | |
1132 | ||
1133 | next_line: | |
1134 | bol = eol + 1; | |
1135 | if (!left) | |
1136 | break; | |
1137 | left--; | |
1138 | lno++; | |
1139 | } | |
1140 | ||
1141 | if (collect_hits) | |
1142 | return 0; | |
1143 | ||
1144 | if (opt->status_only) | |
1145 | return 0; | |
1146 | if (opt->unmatch_name_only) { | |
1147 | /* We did not see any hit, so we want to show this */ | |
1148 | show_name(opt, gs->name); | |
1149 | return 1; | |
1150 | } | |
1151 | ||
1152 | xdiff_clear_find_func(&xecfg); | |
1153 | opt->priv = NULL; | |
1154 | ||
1155 | /* NEEDSWORK: | |
1156 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
1157 | * which feels mostly useless but sometimes useful. Maybe | |
1158 | * make it another option? For now suppress them. | |
1159 | */ | |
1160 | if (opt->count && count) { | |
1161 | char buf[32]; | |
1162 | output_color(opt, gs->name, strlen(gs->name), opt->color_filename); | |
1163 | output_sep(opt, ':'); | |
1164 | snprintf(buf, sizeof(buf), "%u\n", count); | |
1165 | opt->output(opt, buf, strlen(buf)); | |
1166 | return 1; | |
1167 | } | |
1168 | return !!last_hit; | |
1169 | } | |
1170 | ||
1171 | static void clr_hit_marker(struct grep_expr *x) | |
1172 | { | |
1173 | /* All-hit markers are meaningful only at the very top level | |
1174 | * OR node. | |
1175 | */ | |
1176 | while (1) { | |
1177 | x->hit = 0; | |
1178 | if (x->node != GREP_NODE_OR) | |
1179 | return; | |
1180 | x->u.binary.left->hit = 0; | |
1181 | x = x->u.binary.right; | |
1182 | } | |
1183 | } | |
1184 | ||
1185 | static int chk_hit_marker(struct grep_expr *x) | |
1186 | { | |
1187 | /* Top level nodes have hit markers. See if they all are hits */ | |
1188 | while (1) { | |
1189 | if (x->node != GREP_NODE_OR) | |
1190 | return x->hit; | |
1191 | if (!x->u.binary.left->hit) | |
1192 | return 0; | |
1193 | x = x->u.binary.right; | |
1194 | } | |
1195 | } | |
1196 | ||
1197 | int grep_source(struct grep_opt *opt, struct grep_source *gs) | |
1198 | { | |
1199 | /* | |
1200 | * we do not have to do the two-pass grep when we do not check | |
1201 | * buffer-wide "all-match". | |
1202 | */ | |
1203 | if (!opt->all_match) | |
1204 | return grep_source_1(opt, gs, 0); | |
1205 | ||
1206 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
1207 | * We first clear hit markers from them. | |
1208 | */ | |
1209 | clr_hit_marker(opt->pattern_expression); | |
1210 | grep_source_1(opt, gs, 1); | |
1211 | ||
1212 | if (!chk_hit_marker(opt->pattern_expression)) | |
1213 | return 0; | |
1214 | ||
1215 | return grep_source_1(opt, gs, 0); | |
1216 | } | |
1217 | ||
1218 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
1219 | { | |
1220 | struct grep_source gs; | |
1221 | int r; | |
1222 | ||
1223 | grep_source_init(&gs, GREP_SOURCE_BUF, name, NULL); | |
1224 | gs.buf = buf; | |
1225 | gs.size = size; | |
1226 | ||
1227 | r = grep_source(opt, &gs); | |
1228 | ||
1229 | grep_source_clear(&gs); | |
1230 | return r; | |
1231 | } | |
1232 | ||
1233 | void grep_source_init(struct grep_source *gs, enum grep_source_type type, | |
1234 | const char *name, const void *identifier) | |
1235 | { | |
1236 | gs->type = type; | |
1237 | gs->name = name ? xstrdup(name) : NULL; | |
1238 | gs->buf = NULL; | |
1239 | gs->size = 0; | |
1240 | ||
1241 | switch (type) { | |
1242 | case GREP_SOURCE_FILE: | |
1243 | gs->identifier = xstrdup(identifier); | |
1244 | break; | |
1245 | case GREP_SOURCE_SHA1: | |
1246 | gs->identifier = xmalloc(20); | |
1247 | memcpy(gs->identifier, identifier, 20); | |
1248 | break; | |
1249 | case GREP_SOURCE_BUF: | |
1250 | gs->identifier = NULL; | |
1251 | } | |
1252 | } | |
1253 | ||
1254 | void grep_source_clear(struct grep_source *gs) | |
1255 | { | |
1256 | free(gs->name); | |
1257 | gs->name = NULL; | |
1258 | free(gs->identifier); | |
1259 | gs->identifier = NULL; | |
1260 | grep_source_clear_data(gs); | |
1261 | } | |
1262 | ||
1263 | void grep_source_clear_data(struct grep_source *gs) | |
1264 | { | |
1265 | switch (gs->type) { | |
1266 | case GREP_SOURCE_FILE: | |
1267 | case GREP_SOURCE_SHA1: | |
1268 | free(gs->buf); | |
1269 | gs->buf = NULL; | |
1270 | gs->size = 0; | |
1271 | break; | |
1272 | case GREP_SOURCE_BUF: | |
1273 | /* leave user-provided buf intact */ | |
1274 | break; | |
1275 | } | |
1276 | } | |
1277 | ||
1278 | static int grep_source_load_sha1(struct grep_source *gs) | |
1279 | { | |
1280 | enum object_type type; | |
1281 | ||
1282 | grep_read_lock(); | |
1283 | gs->buf = read_sha1_file(gs->identifier, &type, &gs->size); | |
1284 | grep_read_unlock(); | |
1285 | ||
1286 | if (!gs->buf) | |
1287 | return error(_("'%s': unable to read %s"), | |
1288 | gs->name, | |
1289 | sha1_to_hex(gs->identifier)); | |
1290 | return 0; | |
1291 | } | |
1292 | ||
1293 | static int grep_source_load_file(struct grep_source *gs) | |
1294 | { | |
1295 | const char *filename = gs->identifier; | |
1296 | struct stat st; | |
1297 | char *data; | |
1298 | size_t size; | |
1299 | int i; | |
1300 | ||
1301 | if (lstat(filename, &st) < 0) { | |
1302 | err_ret: | |
1303 | if (errno != ENOENT) | |
1304 | error(_("'%s': %s"), filename, strerror(errno)); | |
1305 | return -1; | |
1306 | } | |
1307 | if (!S_ISREG(st.st_mode)) | |
1308 | return -1; | |
1309 | size = xsize_t(st.st_size); | |
1310 | i = open(filename, O_RDONLY); | |
1311 | if (i < 0) | |
1312 | goto err_ret; | |
1313 | data = xmalloc(size + 1); | |
1314 | if (st.st_size != read_in_full(i, data, size)) { | |
1315 | error(_("'%s': short read %s"), filename, strerror(errno)); | |
1316 | close(i); | |
1317 | free(data); | |
1318 | return -1; | |
1319 | } | |
1320 | close(i); | |
1321 | data[size] = 0; | |
1322 | ||
1323 | gs->buf = data; | |
1324 | gs->size = size; | |
1325 | return 0; | |
1326 | } | |
1327 | ||
1328 | int grep_source_load(struct grep_source *gs) | |
1329 | { | |
1330 | if (gs->buf) | |
1331 | return 0; | |
1332 | ||
1333 | switch (gs->type) { | |
1334 | case GREP_SOURCE_FILE: | |
1335 | return grep_source_load_file(gs); | |
1336 | case GREP_SOURCE_SHA1: | |
1337 | return grep_source_load_sha1(gs); | |
1338 | case GREP_SOURCE_BUF: | |
1339 | return gs->buf ? 0 : -1; | |
1340 | } | |
1341 | die("BUG: invalid grep_source type"); | |
1342 | } |