]>
Commit | Line | Data |
---|---|---|
83b5d2f5 | 1 | #include "cache.h" |
83b5d2f5 | 2 | #include "grep.h" |
60ecac98 | 3 | #include "userdiff.h" |
6bfce93e | 4 | #include "xdiff-interface.h" |
83b5d2f5 | 5 | |
07a7d656 JH |
6 | static int grep_source_load(struct grep_source *gs); |
7 | static int grep_source_is_binary(struct grep_source *gs); | |
8 | ||
9 | ||
fc456751 RS |
10 | static struct grep_pat *create_grep_pat(const char *pat, size_t patlen, |
11 | const char *origin, int no, | |
12 | enum grep_pat_token t, | |
13 | enum grep_header_field field) | |
a4d7d2c6 JH |
14 | { |
15 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
526a858a | 16 | p->pattern = xmemdupz(pat, patlen); |
fc456751 RS |
17 | p->patternlen = patlen; |
18 | p->origin = origin; | |
19 | p->no = no; | |
20 | p->token = t; | |
a4d7d2c6 | 21 | p->field = field; |
fc456751 RS |
22 | return p; |
23 | } | |
24 | ||
2b3873ff RS |
25 | static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p) |
26 | { | |
27 | **tail = p; | |
28 | *tail = &p->next; | |
a4d7d2c6 | 29 | p->next = NULL; |
526a858a RS |
30 | |
31 | switch (p->token) { | |
32 | case GREP_PATTERN: /* atom */ | |
33 | case GREP_PATTERN_HEAD: | |
34 | case GREP_PATTERN_BODY: | |
35 | for (;;) { | |
36 | struct grep_pat *new_pat; | |
37 | size_t len = 0; | |
38 | char *cp = p->pattern + p->patternlen, *nl = NULL; | |
39 | while (++len <= p->patternlen) { | |
40 | if (*(--cp) == '\n') { | |
41 | nl = cp; | |
42 | break; | |
43 | } | |
44 | } | |
45 | if (!nl) | |
46 | break; | |
47 | new_pat = create_grep_pat(nl + 1, len - 1, p->origin, | |
48 | p->no, p->token, p->field); | |
49 | new_pat->next = p->next; | |
50 | if (!p->next) | |
51 | *tail = &new_pat->next; | |
52 | p->next = new_pat; | |
53 | *nl = '\0'; | |
54 | p->patternlen -= len; | |
55 | } | |
56 | break; | |
57 | default: | |
58 | break; | |
59 | } | |
2b3873ff RS |
60 | } |
61 | ||
fc456751 RS |
62 | void append_header_grep_pattern(struct grep_opt *opt, |
63 | enum grep_header_field field, const char *pat) | |
64 | { | |
65 | struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0, | |
66 | GREP_PATTERN_HEAD, field); | |
baa6378f JH |
67 | if (field == GREP_HEADER_REFLOG) |
68 | opt->use_reflog_filter = 1; | |
2b3873ff | 69 | do_append_grep_pat(&opt->header_tail, p); |
a4d7d2c6 JH |
70 | } |
71 | ||
83b5d2f5 JH |
72 | void append_grep_pattern(struct grep_opt *opt, const char *pat, |
73 | const char *origin, int no, enum grep_pat_token t) | |
ed40a095 RS |
74 | { |
75 | append_grep_pat(opt, pat, strlen(pat), origin, no, t); | |
76 | } | |
77 | ||
78 | void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen, | |
79 | const char *origin, int no, enum grep_pat_token t) | |
83b5d2f5 | 80 | { |
fc456751 | 81 | struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0); |
2b3873ff | 82 | do_append_grep_pat(&opt->pattern_tail, p); |
83b5d2f5 JH |
83 | } |
84 | ||
5b594f45 FK |
85 | struct grep_opt *grep_opt_dup(const struct grep_opt *opt) |
86 | { | |
87 | struct grep_pat *pat; | |
88 | struct grep_opt *ret = xmalloc(sizeof(struct grep_opt)); | |
89 | *ret = *opt; | |
90 | ||
91 | ret->pattern_list = NULL; | |
92 | ret->pattern_tail = &ret->pattern_list; | |
93 | ||
94 | for(pat = opt->pattern_list; pat != NULL; pat = pat->next) | |
95 | { | |
96 | if(pat->token == GREP_PATTERN_HEAD) | |
97 | append_header_grep_pattern(ret, pat->field, | |
98 | pat->pattern); | |
99 | else | |
ed40a095 RS |
100 | append_grep_pat(ret, pat->pattern, pat->patternlen, |
101 | pat->origin, pat->no, pat->token); | |
5b594f45 FK |
102 | } |
103 | ||
104 | return ret; | |
105 | } | |
106 | ||
a30c148a MK |
107 | static NORETURN void compile_regexp_failed(const struct grep_pat *p, |
108 | const char *error) | |
109 | { | |
110 | char where[1024]; | |
111 | ||
112 | if (p->no) | |
113 | sprintf(where, "In '%s' at %d, ", p->origin, p->no); | |
114 | else if (p->origin) | |
115 | sprintf(where, "%s, ", p->origin); | |
116 | else | |
117 | where[0] = 0; | |
118 | ||
119 | die("%s'%s': %s", where, p->pattern, error); | |
120 | } | |
121 | ||
63e7e9d8 MK |
122 | #ifdef USE_LIBPCRE |
123 | static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt) | |
124 | { | |
125 | const char *error; | |
126 | int erroffset; | |
fba4f125 | 127 | int options = PCRE_MULTILINE; |
63e7e9d8 MK |
128 | |
129 | if (opt->ignore_case) | |
130 | options |= PCRE_CASELESS; | |
131 | ||
132 | p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset, | |
133 | NULL); | |
134 | if (!p->pcre_regexp) | |
135 | compile_regexp_failed(p, error); | |
136 | ||
137 | p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error); | |
138 | if (!p->pcre_extra_info && error) | |
139 | die("%s", error); | |
140 | } | |
141 | ||
142 | static int pcrematch(struct grep_pat *p, const char *line, const char *eol, | |
143 | regmatch_t *match, int eflags) | |
144 | { | |
145 | int ovector[30], ret, flags = 0; | |
146 | ||
147 | if (eflags & REG_NOTBOL) | |
148 | flags |= PCRE_NOTBOL; | |
149 | ||
150 | ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line, | |
151 | 0, flags, ovector, ARRAY_SIZE(ovector)); | |
152 | if (ret < 0 && ret != PCRE_ERROR_NOMATCH) | |
153 | die("pcre_exec failed with error code %d", ret); | |
154 | if (ret > 0) { | |
155 | ret = 0; | |
156 | match->rm_so = ovector[0]; | |
157 | match->rm_eo = ovector[1]; | |
158 | } | |
159 | ||
160 | return ret; | |
161 | } | |
162 | ||
163 | static void free_pcre_regexp(struct grep_pat *p) | |
164 | { | |
165 | pcre_free(p->pcre_regexp); | |
166 | pcre_free(p->pcre_extra_info); | |
167 | } | |
168 | #else /* !USE_LIBPCRE */ | |
169 | static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt) | |
170 | { | |
171 | die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE"); | |
172 | } | |
173 | ||
174 | static int pcrematch(struct grep_pat *p, const char *line, const char *eol, | |
175 | regmatch_t *match, int eflags) | |
176 | { | |
177 | return 1; | |
178 | } | |
179 | ||
180 | static void free_pcre_regexp(struct grep_pat *p) | |
181 | { | |
182 | } | |
183 | #endif /* !USE_LIBPCRE */ | |
184 | ||
9eceddee FK |
185 | static int is_fixed(const char *s, size_t len) |
186 | { | |
187 | size_t i; | |
188 | ||
189 | /* regcomp cannot accept patterns with NULs so we | |
190 | * consider any pattern containing a NUL fixed. | |
191 | */ | |
192 | if (memchr(s, 0, len)) | |
193 | return 1; | |
194 | ||
195 | for (i = 0; i < len; i++) { | |
196 | if (is_regex_special(s[i])) | |
197 | return 0; | |
198 | } | |
199 | ||
200 | return 1; | |
201 | } | |
202 | ||
83b5d2f5 JH |
203 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) |
204 | { | |
c822255c RS |
205 | int err; |
206 | ||
d7eb527d | 207 | p->word_regexp = opt->word_regexp; |
5183bf67 | 208 | p->ignore_case = opt->ignore_case; |
d7eb527d | 209 | |
9eceddee FK |
210 | if (opt->fixed || is_fixed(p->pattern, p->patternlen)) |
211 | p->fixed = 1; | |
212 | else | |
213 | p->fixed = 0; | |
214 | ||
215 | if (p->fixed) { | |
0f871cf5 JH |
216 | if (opt->regflags & REG_ICASE || p->ignore_case) |
217 | p->kws = kwsalloc(tolower_trans_tbl); | |
218 | else | |
9eceddee | 219 | p->kws = kwsalloc(NULL); |
9eceddee FK |
220 | kwsincr(p->kws, p->pattern, p->patternlen); |
221 | kwsprep(p->kws); | |
c822255c | 222 | return; |
9eceddee | 223 | } |
c822255c | 224 | |
63e7e9d8 MK |
225 | if (opt->pcre) { |
226 | compile_pcre_regexp(p, opt); | |
227 | return; | |
228 | } | |
229 | ||
c822255c | 230 | err = regcomp(&p->regexp, p->pattern, opt->regflags); |
83b5d2f5 JH |
231 | if (err) { |
232 | char errbuf[1024]; | |
83b5d2f5 JH |
233 | regerror(err, &p->regexp, errbuf, 1024); |
234 | regfree(&p->regexp); | |
a30c148a | 235 | compile_regexp_failed(p, errbuf); |
83b5d2f5 JH |
236 | } |
237 | } | |
238 | ||
0ab7befa | 239 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
240 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
241 | { | |
242 | struct grep_pat *p; | |
243 | struct grep_expr *x; | |
244 | ||
245 | p = *list; | |
c922b01f LT |
246 | if (!p) |
247 | return NULL; | |
83b5d2f5 JH |
248 | switch (p->token) { |
249 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
250 | case GREP_PATTERN_HEAD: |
251 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
252 | x = xcalloc(1, sizeof (struct grep_expr)); |
253 | x->node = GREP_NODE_ATOM; | |
254 | x->u.atom = p; | |
255 | *list = p->next; | |
256 | return x; | |
257 | case GREP_OPEN_PAREN: | |
258 | *list = p->next; | |
0ab7befa | 259 | x = compile_pattern_or(list); |
83b5d2f5 JH |
260 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
261 | die("unmatched parenthesis"); | |
262 | *list = (*list)->next; | |
263 | return x; | |
264 | default: | |
265 | return NULL; | |
266 | } | |
267 | } | |
268 | ||
269 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
270 | { | |
271 | struct grep_pat *p; | |
272 | struct grep_expr *x; | |
273 | ||
274 | p = *list; | |
c922b01f LT |
275 | if (!p) |
276 | return NULL; | |
83b5d2f5 JH |
277 | switch (p->token) { |
278 | case GREP_NOT: | |
279 | if (!p->next) | |
280 | die("--not not followed by pattern expression"); | |
281 | *list = p->next; | |
282 | x = xcalloc(1, sizeof (struct grep_expr)); | |
283 | x->node = GREP_NODE_NOT; | |
284 | x->u.unary = compile_pattern_not(list); | |
285 | if (!x->u.unary) | |
286 | die("--not followed by non pattern expression"); | |
287 | return x; | |
288 | default: | |
289 | return compile_pattern_atom(list); | |
290 | } | |
291 | } | |
292 | ||
293 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
294 | { | |
295 | struct grep_pat *p; | |
296 | struct grep_expr *x, *y, *z; | |
297 | ||
298 | x = compile_pattern_not(list); | |
299 | p = *list; | |
300 | if (p && p->token == GREP_AND) { | |
301 | if (!p->next) | |
302 | die("--and not followed by pattern expression"); | |
303 | *list = p->next; | |
304 | y = compile_pattern_and(list); | |
305 | if (!y) | |
306 | die("--and not followed by pattern expression"); | |
307 | z = xcalloc(1, sizeof (struct grep_expr)); | |
308 | z->node = GREP_NODE_AND; | |
309 | z->u.binary.left = x; | |
310 | z->u.binary.right = y; | |
311 | return z; | |
312 | } | |
313 | return x; | |
314 | } | |
315 | ||
316 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
317 | { | |
318 | struct grep_pat *p; | |
319 | struct grep_expr *x, *y, *z; | |
320 | ||
321 | x = compile_pattern_and(list); | |
322 | p = *list; | |
323 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
324 | y = compile_pattern_or(list); | |
325 | if (!y) | |
326 | die("not a pattern expression %s", p->pattern); | |
327 | z = xcalloc(1, sizeof (struct grep_expr)); | |
328 | z->node = GREP_NODE_OR; | |
329 | z->u.binary.left = x; | |
330 | z->u.binary.right = y; | |
331 | return z; | |
332 | } | |
333 | return x; | |
334 | } | |
335 | ||
336 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
337 | { | |
338 | return compile_pattern_or(list); | |
339 | } | |
340 | ||
17bf35a3 JH |
341 | static void indent(int in) |
342 | { | |
343 | while (in-- > 0) | |
344 | fputc(' ', stderr); | |
345 | } | |
346 | ||
347 | static void dump_grep_pat(struct grep_pat *p) | |
348 | { | |
349 | switch (p->token) { | |
350 | case GREP_AND: fprintf(stderr, "*and*"); break; | |
351 | case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break; | |
352 | case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break; | |
353 | case GREP_NOT: fprintf(stderr, "*not*"); break; | |
354 | case GREP_OR: fprintf(stderr, "*or*"); break; | |
355 | ||
356 | case GREP_PATTERN: fprintf(stderr, "pattern"); break; | |
357 | case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break; | |
358 | case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break; | |
359 | } | |
360 | ||
361 | switch (p->token) { | |
362 | default: break; | |
363 | case GREP_PATTERN_HEAD: | |
364 | fprintf(stderr, "<head %d>", p->field); break; | |
365 | case GREP_PATTERN_BODY: | |
366 | fprintf(stderr, "<body>"); break; | |
367 | } | |
368 | switch (p->token) { | |
369 | default: break; | |
370 | case GREP_PATTERN_HEAD: | |
371 | case GREP_PATTERN_BODY: | |
372 | case GREP_PATTERN: | |
373 | fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern); | |
374 | break; | |
375 | } | |
376 | fputc('\n', stderr); | |
377 | } | |
378 | ||
379 | static void dump_grep_expression_1(struct grep_expr *x, int in) | |
380 | { | |
381 | indent(in); | |
382 | switch (x->node) { | |
383 | case GREP_NODE_TRUE: | |
384 | fprintf(stderr, "true\n"); | |
385 | break; | |
386 | case GREP_NODE_ATOM: | |
387 | dump_grep_pat(x->u.atom); | |
388 | break; | |
389 | case GREP_NODE_NOT: | |
390 | fprintf(stderr, "(not\n"); | |
391 | dump_grep_expression_1(x->u.unary, in+1); | |
392 | indent(in); | |
393 | fprintf(stderr, ")\n"); | |
394 | break; | |
395 | case GREP_NODE_AND: | |
396 | fprintf(stderr, "(and\n"); | |
397 | dump_grep_expression_1(x->u.binary.left, in+1); | |
398 | dump_grep_expression_1(x->u.binary.right, in+1); | |
399 | indent(in); | |
400 | fprintf(stderr, ")\n"); | |
401 | break; | |
402 | case GREP_NODE_OR: | |
403 | fprintf(stderr, "(or\n"); | |
404 | dump_grep_expression_1(x->u.binary.left, in+1); | |
405 | dump_grep_expression_1(x->u.binary.right, in+1); | |
406 | indent(in); | |
407 | fprintf(stderr, ")\n"); | |
408 | break; | |
409 | } | |
410 | } | |
411 | ||
07a7d656 | 412 | static void dump_grep_expression(struct grep_opt *opt) |
17bf35a3 JH |
413 | { |
414 | struct grep_expr *x = opt->pattern_expression; | |
415 | ||
416 | if (opt->all_match) | |
417 | fprintf(stderr, "[all-match]\n"); | |
418 | dump_grep_expression_1(x, 0); | |
419 | fflush(NULL); | |
420 | } | |
421 | ||
5aaeb733 JH |
422 | static struct grep_expr *grep_true_expr(void) |
423 | { | |
424 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
425 | z->node = GREP_NODE_TRUE; | |
426 | return z; | |
427 | } | |
428 | ||
429 | static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right) | |
430 | { | |
431 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
432 | z->node = GREP_NODE_OR; | |
433 | z->u.binary.left = left; | |
434 | z->u.binary.right = right; | |
435 | return z; | |
436 | } | |
437 | ||
95ce9ce2 | 438 | static struct grep_expr *prep_header_patterns(struct grep_opt *opt) |
83b5d2f5 JH |
439 | { |
440 | struct grep_pat *p; | |
95ce9ce2 | 441 | struct grep_expr *header_expr; |
5aaeb733 JH |
442 | struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]); |
443 | enum grep_header_field fld; | |
83b5d2f5 | 444 | |
95ce9ce2 JH |
445 | if (!opt->header_list) |
446 | return NULL; | |
2385f246 | 447 | |
95ce9ce2 JH |
448 | for (p = opt->header_list; p; p = p->next) { |
449 | if (p->token != GREP_PATTERN_HEAD) | |
450 | die("bug: a non-header pattern in grep header list."); | |
451 | if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field) | |
452 | die("bug: unknown header field %d", p->field); | |
453 | compile_regexp(p, opt); | |
80235ba7 | 454 | } |
5aaeb733 JH |
455 | |
456 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) | |
457 | header_group[fld] = NULL; | |
458 | ||
459 | for (p = opt->header_list; p; p = p->next) { | |
460 | struct grep_expr *h; | |
461 | struct grep_pat *pp = p; | |
462 | ||
463 | h = compile_pattern_atom(&pp); | |
464 | if (!h || pp != p->next) | |
465 | die("bug: malformed header expr"); | |
466 | if (!header_group[p->field]) { | |
467 | header_group[p->field] = h; | |
468 | continue; | |
469 | } | |
470 | header_group[p->field] = grep_or_expr(h, header_group[p->field]); | |
471 | } | |
472 | ||
473 | header_expr = NULL; | |
474 | ||
475 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) { | |
476 | if (!header_group[fld]) | |
477 | continue; | |
478 | if (!header_expr) | |
479 | header_expr = grep_true_expr(); | |
480 | header_expr = grep_or_expr(header_group[fld], header_expr); | |
481 | } | |
95ce9ce2 JH |
482 | return header_expr; |
483 | } | |
484 | ||
13e4fc7e JH |
485 | static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y) |
486 | { | |
487 | struct grep_expr *z = x; | |
488 | ||
489 | while (x) { | |
490 | assert(x->node == GREP_NODE_OR); | |
491 | if (x->u.binary.right && | |
492 | x->u.binary.right->node == GREP_NODE_TRUE) { | |
493 | x->u.binary.right = y; | |
494 | break; | |
495 | } | |
496 | x = x->u.binary.right; | |
497 | } | |
498 | return z; | |
499 | } | |
500 | ||
17bf35a3 | 501 | static void compile_grep_patterns_real(struct grep_opt *opt) |
95ce9ce2 JH |
502 | { |
503 | struct grep_pat *p; | |
504 | struct grep_expr *header_expr = prep_header_patterns(opt); | |
0ab7befa | 505 | |
83b5d2f5 | 506 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
507 | switch (p->token) { |
508 | case GREP_PATTERN: /* atom */ | |
509 | case GREP_PATTERN_HEAD: | |
510 | case GREP_PATTERN_BODY: | |
c822255c | 511 | compile_regexp(p, opt); |
480c1ca6 JH |
512 | break; |
513 | default: | |
83b5d2f5 | 514 | opt->extended = 1; |
480c1ca6 JH |
515 | break; |
516 | } | |
83b5d2f5 JH |
517 | } |
518 | ||
80235ba7 JH |
519 | if (opt->all_match || header_expr) |
520 | opt->extended = 1; | |
17bf35a3 | 521 | else if (!opt->extended && !opt->debug) |
83b5d2f5 JH |
522 | return; |
523 | ||
83b5d2f5 | 524 | p = opt->pattern_list; |
ba150a3f MB |
525 | if (p) |
526 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
527 | if (p) |
528 | die("incomplete pattern expression: %s", p->pattern); | |
80235ba7 JH |
529 | |
530 | if (!header_expr) | |
531 | return; | |
532 | ||
5aaeb733 | 533 | if (!opt->pattern_expression) |
80235ba7 | 534 | opt->pattern_expression = header_expr; |
13e4fc7e JH |
535 | else if (opt->all_match) |
536 | opt->pattern_expression = grep_splice_or(header_expr, | |
537 | opt->pattern_expression); | |
5aaeb733 JH |
538 | else |
539 | opt->pattern_expression = grep_or_expr(opt->pattern_expression, | |
540 | header_expr); | |
80235ba7 | 541 | opt->all_match = 1; |
83b5d2f5 JH |
542 | } |
543 | ||
17bf35a3 JH |
544 | void compile_grep_patterns(struct grep_opt *opt) |
545 | { | |
546 | compile_grep_patterns_real(opt); | |
547 | if (opt->debug) | |
548 | dump_grep_expression(opt); | |
549 | } | |
550 | ||
b48fb5b6 JH |
551 | static void free_pattern_expr(struct grep_expr *x) |
552 | { | |
553 | switch (x->node) { | |
5aaeb733 | 554 | case GREP_NODE_TRUE: |
b48fb5b6 JH |
555 | case GREP_NODE_ATOM: |
556 | break; | |
557 | case GREP_NODE_NOT: | |
558 | free_pattern_expr(x->u.unary); | |
559 | break; | |
560 | case GREP_NODE_AND: | |
561 | case GREP_NODE_OR: | |
562 | free_pattern_expr(x->u.binary.left); | |
563 | free_pattern_expr(x->u.binary.right); | |
564 | break; | |
565 | } | |
566 | free(x); | |
567 | } | |
568 | ||
569 | void free_grep_patterns(struct grep_opt *opt) | |
570 | { | |
571 | struct grep_pat *p, *n; | |
572 | ||
573 | for (p = opt->pattern_list; p; p = n) { | |
574 | n = p->next; | |
575 | switch (p->token) { | |
576 | case GREP_PATTERN: /* atom */ | |
577 | case GREP_PATTERN_HEAD: | |
578 | case GREP_PATTERN_BODY: | |
9eceddee FK |
579 | if (p->kws) |
580 | kwsfree(p->kws); | |
581 | else if (p->pcre_regexp) | |
63e7e9d8 MK |
582 | free_pcre_regexp(p); |
583 | else | |
584 | regfree(&p->regexp); | |
526a858a | 585 | free(p->pattern); |
b48fb5b6 JH |
586 | break; |
587 | default: | |
588 | break; | |
589 | } | |
590 | free(p); | |
591 | } | |
592 | ||
593 | if (!opt->extended) | |
594 | return; | |
595 | free_pattern_expr(opt->pattern_expression); | |
596 | } | |
597 | ||
83b5d2f5 JH |
598 | static char *end_of_line(char *cp, unsigned long *left) |
599 | { | |
600 | unsigned long l = *left; | |
601 | while (l && *cp != '\n') { | |
602 | l--; | |
603 | cp++; | |
604 | } | |
605 | *left = l; | |
606 | return cp; | |
607 | } | |
608 | ||
609 | static int word_char(char ch) | |
610 | { | |
611 | return isalnum(ch) || ch == '_'; | |
612 | } | |
613 | ||
55f638bd ML |
614 | static void output_color(struct grep_opt *opt, const void *data, size_t size, |
615 | const char *color) | |
616 | { | |
daa0c3d9 | 617 | if (want_color(opt->color) && color && color[0]) { |
55f638bd ML |
618 | opt->output(opt, color, strlen(color)); |
619 | opt->output(opt, data, size); | |
620 | opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET)); | |
621 | } else | |
622 | opt->output(opt, data, size); | |
623 | } | |
624 | ||
625 | static void output_sep(struct grep_opt *opt, char sign) | |
626 | { | |
627 | if (opt->null_following_name) | |
628 | opt->output(opt, "\0", 1); | |
629 | else | |
630 | output_color(opt, &sign, 1, opt->color_sep); | |
631 | } | |
632 | ||
83caecca RZ |
633 | static void show_name(struct grep_opt *opt, const char *name) |
634 | { | |
55f638bd | 635 | output_color(opt, name, strlen(name), opt->color_filename); |
5b594f45 | 636 | opt->output(opt, opt->null_following_name ? "\0" : "\n", 1); |
83caecca RZ |
637 | } |
638 | ||
ed40a095 RS |
639 | static int fixmatch(struct grep_pat *p, char *line, char *eol, |
640 | regmatch_t *match) | |
83b5d2f5 | 641 | { |
9eceddee FK |
642 | struct kwsmatch kwsm; |
643 | size_t offset = kwsexec(p->kws, line, eol - line, &kwsm); | |
644 | if (offset == -1) { | |
83b5d2f5 JH |
645 | match->rm_so = match->rm_eo = -1; |
646 | return REG_NOMATCH; | |
9eceddee FK |
647 | } else { |
648 | match->rm_so = offset; | |
649 | match->rm_eo = match->rm_so + kwsm.size[0]; | |
83b5d2f5 JH |
650 | return 0; |
651 | } | |
652 | } | |
653 | ||
f96e5673 RS |
654 | static int regmatch(const regex_t *preg, char *line, char *eol, |
655 | regmatch_t *match, int eflags) | |
656 | { | |
657 | #ifdef REG_STARTEND | |
658 | match->rm_so = 0; | |
659 | match->rm_eo = eol - line; | |
660 | eflags |= REG_STARTEND; | |
661 | #endif | |
662 | return regexec(preg, line, 1, match, eflags); | |
663 | } | |
664 | ||
97e77784 MK |
665 | static int patmatch(struct grep_pat *p, char *line, char *eol, |
666 | regmatch_t *match, int eflags) | |
667 | { | |
668 | int hit; | |
669 | ||
670 | if (p->fixed) | |
671 | hit = !fixmatch(p, line, eol, match); | |
63e7e9d8 MK |
672 | else if (p->pcre_regexp) |
673 | hit = !pcrematch(p, line, eol, match, eflags); | |
97e77784 MK |
674 | else |
675 | hit = !regmatch(&p->regexp, line, eol, match, eflags); | |
676 | ||
677 | return hit; | |
678 | } | |
679 | ||
a4d7d2c6 JH |
680 | static int strip_timestamp(char *bol, char **eol_p) |
681 | { | |
682 | char *eol = *eol_p; | |
683 | int ch; | |
684 | ||
685 | while (bol < --eol) { | |
686 | if (*eol != '>') | |
687 | continue; | |
688 | *eol_p = ++eol; | |
689 | ch = *eol; | |
690 | *eol = '\0'; | |
691 | return ch; | |
692 | } | |
693 | return 0; | |
694 | } | |
695 | ||
696 | static struct { | |
697 | const char *field; | |
698 | size_t len; | |
699 | } header_field[] = { | |
700 | { "author ", 7 }, | |
701 | { "committer ", 10 }, | |
72fd13f7 | 702 | { "reflog ", 7 }, |
a4d7d2c6 JH |
703 | }; |
704 | ||
d7eb527d | 705 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
706 | enum grep_context ctx, |
707 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
708 | { |
709 | int hit = 0; | |
a4d7d2c6 | 710 | int saved_ch = 0; |
e701fadb | 711 | const char *start = bol; |
83b5d2f5 | 712 | |
480c1ca6 JH |
713 | if ((p->token != GREP_PATTERN) && |
714 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
715 | return 0; | |
716 | ||
a4d7d2c6 JH |
717 | if (p->token == GREP_PATTERN_HEAD) { |
718 | const char *field; | |
719 | size_t len; | |
720 | assert(p->field < ARRAY_SIZE(header_field)); | |
721 | field = header_field[p->field].field; | |
722 | len = header_field[p->field].len; | |
723 | if (strncmp(bol, field, len)) | |
724 | return 0; | |
725 | bol += len; | |
ad4813b3 NTND |
726 | switch (p->field) { |
727 | case GREP_HEADER_AUTHOR: | |
728 | case GREP_HEADER_COMMITTER: | |
729 | saved_ch = strip_timestamp(bol, &eol); | |
730 | break; | |
731 | default: | |
732 | break; | |
733 | } | |
a4d7d2c6 JH |
734 | } |
735 | ||
83b5d2f5 | 736 | again: |
97e77784 | 737 | hit = patmatch(p, bol, eol, pmatch, eflags); |
83b5d2f5 | 738 | |
d7eb527d | 739 | if (hit && p->word_regexp) { |
83b5d2f5 | 740 | if ((pmatch[0].rm_so < 0) || |
84201eae | 741 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
742 | (pmatch[0].rm_eo < 0) || |
743 | (eol - bol) < pmatch[0].rm_eo) | |
744 | die("regexp returned nonsense"); | |
745 | ||
746 | /* Match beginning must be either beginning of the | |
747 | * line, or at word boundary (i.e. the last char must | |
748 | * not be a word char). Similarly, match end must be | |
749 | * either end of the line, or at word boundary | |
750 | * (i.e. the next char must not be a word char). | |
751 | */ | |
fb62eb7f | 752 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
753 | !word_char(bol[pmatch[0].rm_so-1])) && |
754 | ((pmatch[0].rm_eo == (eol-bol)) || | |
755 | !word_char(bol[pmatch[0].rm_eo])) ) | |
756 | ; | |
757 | else | |
758 | hit = 0; | |
759 | ||
84201eae RS |
760 | /* Words consist of at least one character. */ |
761 | if (pmatch->rm_so == pmatch->rm_eo) | |
762 | hit = 0; | |
763 | ||
83b5d2f5 JH |
764 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
765 | /* There could be more than one match on the | |
766 | * line, and the first match might not be | |
767 | * strict word match. But later ones could be! | |
fb62eb7f RS |
768 | * Forward to the next possible start, i.e. the |
769 | * next position following a non-word char. | |
83b5d2f5 JH |
770 | */ |
771 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
772 | while (word_char(bol[-1]) && bol < eol) |
773 | bol++; | |
dbb6a4ad | 774 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
775 | if (bol < eol) |
776 | goto again; | |
83b5d2f5 JH |
777 | } |
778 | } | |
a4d7d2c6 JH |
779 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
780 | *eol = saved_ch; | |
e701fadb RS |
781 | if (hit) { |
782 | pmatch[0].rm_so += bol - start; | |
783 | pmatch[0].rm_eo += bol - start; | |
784 | } | |
83b5d2f5 JH |
785 | return hit; |
786 | } | |
787 | ||
d7eb527d RS |
788 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, |
789 | enum grep_context ctx, int collect_hits) | |
83b5d2f5 | 790 | { |
0ab7befa | 791 | int h = 0; |
79212772 | 792 | regmatch_t match; |
0ab7befa | 793 | |
c922b01f LT |
794 | if (!x) |
795 | die("Not a valid grep expression"); | |
83b5d2f5 | 796 | switch (x->node) { |
5aaeb733 JH |
797 | case GREP_NODE_TRUE: |
798 | h = 1; | |
799 | break; | |
83b5d2f5 | 800 | case GREP_NODE_ATOM: |
79212772 | 801 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); |
83b5d2f5 JH |
802 | break; |
803 | case GREP_NODE_NOT: | |
d7eb527d | 804 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); |
0ab7befa | 805 | break; |
83b5d2f5 | 806 | case GREP_NODE_AND: |
d7eb527d | 807 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) |
252d560d | 808 | return 0; |
d7eb527d | 809 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); |
0ab7befa | 810 | break; |
83b5d2f5 | 811 | case GREP_NODE_OR: |
0ab7befa | 812 | if (!collect_hits) |
d7eb527d | 813 | return (match_expr_eval(x->u.binary.left, |
0ab7befa | 814 | bol, eol, ctx, 0) || |
d7eb527d | 815 | match_expr_eval(x->u.binary.right, |
0ab7befa | 816 | bol, eol, ctx, 0)); |
d7eb527d | 817 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); |
0ab7befa | 818 | x->u.binary.left->hit |= h; |
d7eb527d | 819 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); |
0ab7befa JH |
820 | break; |
821 | default: | |
d7530708 | 822 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 823 | } |
0ab7befa JH |
824 | if (collect_hits) |
825 | x->hit |= h; | |
826 | return h; | |
83b5d2f5 JH |
827 | } |
828 | ||
480c1ca6 | 829 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 830 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
831 | { |
832 | struct grep_expr *x = opt->pattern_expression; | |
d7eb527d | 833 | return match_expr_eval(x, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
834 | } |
835 | ||
480c1ca6 | 836 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 837 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
838 | { |
839 | struct grep_pat *p; | |
79212772 RS |
840 | regmatch_t match; |
841 | ||
83b5d2f5 | 842 | if (opt->extended) |
0ab7befa JH |
843 | return match_expr(opt, bol, eol, ctx, collect_hits); |
844 | ||
845 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 846 | for (p = opt->pattern_list; p; p = p->next) { |
79212772 | 847 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) |
83b5d2f5 JH |
848 | return 1; |
849 | } | |
850 | return 0; | |
851 | } | |
852 | ||
7e8f59d5 RS |
853 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
854 | enum grep_context ctx, | |
855 | regmatch_t *pmatch, int eflags) | |
856 | { | |
857 | regmatch_t match; | |
858 | ||
859 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
860 | return 0; | |
861 | if (match.rm_so < 0 || match.rm_eo < 0) | |
862 | return 0; | |
863 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
864 | if (match.rm_so > pmatch->rm_so) | |
865 | return 1; | |
866 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
867 | return 1; | |
868 | } | |
869 | pmatch->rm_so = match.rm_so; | |
870 | pmatch->rm_eo = match.rm_eo; | |
871 | return 1; | |
872 | } | |
873 | ||
874 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
875 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
876 | { | |
877 | struct grep_pat *p; | |
878 | int hit = 0; | |
879 | ||
880 | pmatch->rm_so = pmatch->rm_eo = -1; | |
881 | if (bol < eol) { | |
882 | for (p = opt->pattern_list; p; p = p->next) { | |
883 | switch (p->token) { | |
884 | case GREP_PATTERN: /* atom */ | |
885 | case GREP_PATTERN_HEAD: | |
886 | case GREP_PATTERN_BODY: | |
887 | hit |= match_next_pattern(p, bol, eol, ctx, | |
888 | pmatch, eflags); | |
889 | break; | |
890 | default: | |
891 | break; | |
892 | } | |
893 | } | |
894 | } | |
895 | return hit; | |
896 | } | |
897 | ||
898 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
899 | const char *name, unsigned lno, char sign) | |
900 | { | |
901 | int rest = eol - bol; | |
00588bb5 | 902 | char *line_color = NULL; |
7e8f59d5 | 903 | |
a8f0e764 RS |
904 | if (opt->file_break && opt->last_shown == 0) { |
905 | if (opt->show_hunk_mark) | |
906 | opt->output(opt, "\n", 1); | |
ba8ea749 | 907 | } else if (opt->pre_context || opt->post_context || opt->funcbody) { |
046802d0 | 908 | if (opt->last_shown == 0) { |
55f638bd ML |
909 | if (opt->show_hunk_mark) { |
910 | output_color(opt, "--", 2, opt->color_sep); | |
911 | opt->output(opt, "\n", 1); | |
07b838f0 | 912 | } |
55f638bd ML |
913 | } else if (lno > opt->last_shown + 1) { |
914 | output_color(opt, "--", 2, opt->color_sep); | |
915 | opt->output(opt, "\n", 1); | |
916 | } | |
5dd06d38 | 917 | } |
1d84f72e RS |
918 | if (opt->heading && opt->last_shown == 0) { |
919 | output_color(opt, name, strlen(name), opt->color_filename); | |
920 | opt->output(opt, "\n", 1); | |
921 | } | |
5dd06d38 RS |
922 | opt->last_shown = lno; |
923 | ||
1d84f72e | 924 | if (!opt->heading && opt->pathname) { |
55f638bd ML |
925 | output_color(opt, name, strlen(name), opt->color_filename); |
926 | output_sep(opt, sign); | |
5b594f45 FK |
927 | } |
928 | if (opt->linenum) { | |
929 | char buf[32]; | |
930 | snprintf(buf, sizeof(buf), "%d", lno); | |
55f638bd ML |
931 | output_color(opt, buf, strlen(buf), opt->color_lineno); |
932 | output_sep(opt, sign); | |
5b594f45 | 933 | } |
7e8f59d5 RS |
934 | if (opt->color) { |
935 | regmatch_t match; | |
936 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
937 | int ch = *eol; | |
938 | int eflags = 0; | |
939 | ||
00588bb5 ML |
940 | if (sign == ':') |
941 | line_color = opt->color_selected; | |
942 | else if (sign == '-') | |
943 | line_color = opt->color_context; | |
944 | else if (sign == '=') | |
945 | line_color = opt->color_function; | |
7e8f59d5 RS |
946 | *eol = '\0'; |
947 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
948 | if (match.rm_so == match.rm_eo) |
949 | break; | |
5b594f45 | 950 | |
00588bb5 | 951 | output_color(opt, bol, match.rm_so, line_color); |
55f638bd ML |
952 | output_color(opt, bol + match.rm_so, |
953 | match.rm_eo - match.rm_so, | |
954 | opt->color_match); | |
7e8f59d5 RS |
955 | bol += match.rm_eo; |
956 | rest -= match.rm_eo; | |
957 | eflags = REG_NOTBOL; | |
958 | } | |
959 | *eol = ch; | |
960 | } | |
00588bb5 | 961 | output_color(opt, bol, rest, line_color); |
5b594f45 | 962 | opt->output(opt, "\n", 1); |
7e8f59d5 RS |
963 | } |
964 | ||
0579f91d | 965 | #ifndef NO_PTHREADS |
78db6ea9 JK |
966 | int grep_use_locks; |
967 | ||
0579f91d TR |
968 | /* |
969 | * This lock protects access to the gitattributes machinery, which is | |
970 | * not thread-safe. | |
971 | */ | |
972 | pthread_mutex_t grep_attr_mutex; | |
973 | ||
78db6ea9 | 974 | static inline void grep_attr_lock(void) |
0579f91d | 975 | { |
78db6ea9 | 976 | if (grep_use_locks) |
0579f91d TR |
977 | pthread_mutex_lock(&grep_attr_mutex); |
978 | } | |
979 | ||
78db6ea9 | 980 | static inline void grep_attr_unlock(void) |
0579f91d | 981 | { |
78db6ea9 | 982 | if (grep_use_locks) |
0579f91d TR |
983 | pthread_mutex_unlock(&grep_attr_mutex); |
984 | } | |
b3aeb285 JK |
985 | |
986 | /* | |
987 | * Same as git_attr_mutex, but protecting the thread-unsafe object db access. | |
988 | */ | |
989 | pthread_mutex_t grep_read_mutex; | |
990 | ||
0579f91d | 991 | #else |
78db6ea9 JK |
992 | #define grep_attr_lock() |
993 | #define grep_attr_unlock() | |
0579f91d TR |
994 | #endif |
995 | ||
e1327023 | 996 | static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol) |
2944e4e6 | 997 | { |
60ecac98 | 998 | xdemitconf_t *xecfg = opt->priv; |
0579f91d | 999 | if (xecfg && !xecfg->find_func) { |
94ad9d9e JK |
1000 | grep_source_load_driver(gs); |
1001 | if (gs->driver->funcname.pattern) { | |
1002 | const struct userdiff_funcname *pe = &gs->driver->funcname; | |
0579f91d TR |
1003 | xdiff_set_find_func(xecfg, pe->pattern, pe->cflags); |
1004 | } else { | |
1005 | xecfg = opt->priv = NULL; | |
1006 | } | |
1007 | } | |
1008 | ||
1009 | if (xecfg) { | |
60ecac98 RS |
1010 | char buf[1]; |
1011 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
1012 | xecfg->find_func_priv) >= 0; | |
1013 | } | |
1014 | ||
2944e4e6 RS |
1015 | if (bol == eol) |
1016 | return 0; | |
1017 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
1018 | return 1; | |
1019 | return 0; | |
1020 | } | |
1021 | ||
e1327023 JK |
1022 | static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs, |
1023 | char *bol, unsigned lno) | |
2944e4e6 | 1024 | { |
e1327023 | 1025 | while (bol > gs->buf) { |
2944e4e6 RS |
1026 | char *eol = --bol; |
1027 | ||
e1327023 | 1028 | while (bol > gs->buf && bol[-1] != '\n') |
2944e4e6 RS |
1029 | bol--; |
1030 | lno--; | |
1031 | ||
1032 | if (lno <= opt->last_shown) | |
1033 | break; | |
1034 | ||
e1327023 JK |
1035 | if (match_funcname(opt, gs, bol, eol)) { |
1036 | show_line(opt, bol, eol, gs->name, lno, '='); | |
2944e4e6 RS |
1037 | break; |
1038 | } | |
1039 | } | |
1040 | } | |
1041 | ||
e1327023 | 1042 | static void show_pre_context(struct grep_opt *opt, struct grep_source *gs, |
ba8ea749 | 1043 | char *bol, char *end, unsigned lno) |
49de3216 | 1044 | { |
2944e4e6 | 1045 | unsigned cur = lno, from = 1, funcname_lno = 0; |
ba8ea749 RS |
1046 | int funcname_needed = !!opt->funcname; |
1047 | ||
e1327023 | 1048 | if (opt->funcbody && !match_funcname(opt, gs, bol, end)) |
ba8ea749 | 1049 | funcname_needed = 2; |
49de3216 RS |
1050 | |
1051 | if (opt->pre_context < lno) | |
1052 | from = lno - opt->pre_context; | |
1053 | if (from <= opt->last_shown) | |
1054 | from = opt->last_shown + 1; | |
1055 | ||
1056 | /* Rewind. */ | |
e1327023 | 1057 | while (bol > gs->buf && |
ba8ea749 | 1058 | cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) { |
2944e4e6 RS |
1059 | char *eol = --bol; |
1060 | ||
e1327023 | 1061 | while (bol > gs->buf && bol[-1] != '\n') |
49de3216 RS |
1062 | bol--; |
1063 | cur--; | |
e1327023 | 1064 | if (funcname_needed && match_funcname(opt, gs, bol, eol)) { |
2944e4e6 RS |
1065 | funcname_lno = cur; |
1066 | funcname_needed = 0; | |
1067 | } | |
49de3216 RS |
1068 | } |
1069 | ||
2944e4e6 RS |
1070 | /* We need to look even further back to find a function signature. */ |
1071 | if (opt->funcname && funcname_needed) | |
e1327023 | 1072 | show_funcname_line(opt, gs, bol, cur); |
2944e4e6 | 1073 | |
49de3216 RS |
1074 | /* Back forward. */ |
1075 | while (cur < lno) { | |
2944e4e6 | 1076 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; |
49de3216 RS |
1077 | |
1078 | while (*eol != '\n') | |
1079 | eol++; | |
e1327023 | 1080 | show_line(opt, bol, eol, gs->name, cur, sign); |
49de3216 RS |
1081 | bol = eol + 1; |
1082 | cur++; | |
1083 | } | |
1084 | } | |
1085 | ||
a26345b6 JH |
1086 | static int should_lookahead(struct grep_opt *opt) |
1087 | { | |
1088 | struct grep_pat *p; | |
1089 | ||
1090 | if (opt->extended) | |
1091 | return 0; /* punt for too complex stuff */ | |
1092 | if (opt->invert) | |
1093 | return 0; | |
1094 | for (p = opt->pattern_list; p; p = p->next) { | |
1095 | if (p->token != GREP_PATTERN) | |
1096 | return 0; /* punt for "header only" and stuff */ | |
1097 | } | |
1098 | return 1; | |
1099 | } | |
1100 | ||
1101 | static int look_ahead(struct grep_opt *opt, | |
1102 | unsigned long *left_p, | |
1103 | unsigned *lno_p, | |
1104 | char **bol_p) | |
1105 | { | |
1106 | unsigned lno = *lno_p; | |
1107 | char *bol = *bol_p; | |
1108 | struct grep_pat *p; | |
1109 | char *sp, *last_bol; | |
1110 | regoff_t earliest = -1; | |
1111 | ||
1112 | for (p = opt->pattern_list; p; p = p->next) { | |
1113 | int hit; | |
1114 | regmatch_t m; | |
1115 | ||
97e77784 | 1116 | hit = patmatch(p, bol, bol + *left_p, &m, 0); |
a26345b6 JH |
1117 | if (!hit || m.rm_so < 0 || m.rm_eo < 0) |
1118 | continue; | |
1119 | if (earliest < 0 || m.rm_so < earliest) | |
1120 | earliest = m.rm_so; | |
1121 | } | |
1122 | ||
1123 | if (earliest < 0) { | |
1124 | *bol_p = bol + *left_p; | |
1125 | *left_p = 0; | |
1126 | return 1; | |
1127 | } | |
1128 | for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--) | |
1129 | ; /* find the beginning of the line */ | |
1130 | last_bol = sp; | |
1131 | ||
1132 | for (sp = bol; sp < last_bol; sp++) { | |
1133 | if (*sp == '\n') | |
1134 | lno++; | |
1135 | } | |
1136 | *left_p -= last_bol - bol; | |
1137 | *bol_p = last_bol; | |
1138 | *lno_p = lno; | |
1139 | return 0; | |
1140 | } | |
1141 | ||
5b594f45 FK |
1142 | static void std_output(struct grep_opt *opt, const void *buf, size_t size) |
1143 | { | |
1144 | fwrite(buf, size, 1, stdout); | |
1145 | } | |
1146 | ||
e1327023 | 1147 | static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits) |
83b5d2f5 | 1148 | { |
e1327023 JK |
1149 | char *bol; |
1150 | unsigned long left; | |
83b5d2f5 | 1151 | unsigned lno = 1; |
83b5d2f5 | 1152 | unsigned last_hit = 0; |
83b5d2f5 | 1153 | int binary_match_only = 0; |
83b5d2f5 | 1154 | unsigned count = 0; |
a26345b6 | 1155 | int try_lookahead = 0; |
ba8ea749 | 1156 | int show_function = 0; |
480c1ca6 | 1157 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
60ecac98 | 1158 | xdemitconf_t xecfg; |
83b5d2f5 | 1159 | |
5b594f45 FK |
1160 | if (!opt->output) |
1161 | opt->output = std_output; | |
1162 | ||
ba8ea749 RS |
1163 | if (opt->pre_context || opt->post_context || opt->file_break || |
1164 | opt->funcbody) { | |
08303c36 RS |
1165 | /* Show hunk marks, except for the first file. */ |
1166 | if (opt->last_shown) | |
1167 | opt->show_hunk_mark = 1; | |
1168 | /* | |
1169 | * If we're using threads then we can't easily identify | |
1170 | * the first file. Always put hunk marks in that case | |
1171 | * and skip the very first one later in work_done(). | |
1172 | */ | |
1173 | if (opt->output != std_output) | |
1174 | opt->show_hunk_mark = 1; | |
1175 | } | |
431d6e7b RS |
1176 | opt->last_shown = 0; |
1177 | ||
64fcec78 RS |
1178 | switch (opt->binary) { |
1179 | case GREP_BINARY_DEFAULT: | |
41b59bfc | 1180 | if (grep_source_is_binary(gs)) |
83b5d2f5 | 1181 | binary_match_only = 1; |
64fcec78 RS |
1182 | break; |
1183 | case GREP_BINARY_NOMATCH: | |
41b59bfc | 1184 | if (grep_source_is_binary(gs)) |
83b5d2f5 | 1185 | return 0; /* Assume unmatch */ |
64fcec78 RS |
1186 | break; |
1187 | case GREP_BINARY_TEXT: | |
1188 | break; | |
1189 | default: | |
1190 | die("bug: unknown binary handling mode"); | |
83b5d2f5 JH |
1191 | } |
1192 | ||
60ecac98 | 1193 | memset(&xecfg, 0, sizeof(xecfg)); |
0579f91d TR |
1194 | opt->priv = &xecfg; |
1195 | ||
a26345b6 | 1196 | try_lookahead = should_lookahead(opt); |
60ecac98 | 1197 | |
08265798 JK |
1198 | if (grep_source_load(gs) < 0) |
1199 | return 0; | |
1200 | ||
e1327023 JK |
1201 | bol = gs->buf; |
1202 | left = gs->size; | |
83b5d2f5 JH |
1203 | while (left) { |
1204 | char *eol, ch; | |
0ab7befa | 1205 | int hit; |
83b5d2f5 | 1206 | |
a26345b6 | 1207 | /* |
8997da38 | 1208 | * look_ahead() skips quickly to the line that possibly |
a26345b6 JH |
1209 | * has the next hit; don't call it if we need to do |
1210 | * something more than just skipping the current line | |
1211 | * in response to an unmatch for the current line. E.g. | |
1212 | * inside a post-context window, we will show the current | |
1213 | * line as a context around the previous hit when it | |
1214 | * doesn't hit. | |
1215 | */ | |
1216 | if (try_lookahead | |
1217 | && !(last_hit | |
ba8ea749 RS |
1218 | && (show_function || |
1219 | lno <= last_hit + opt->post_context)) | |
a26345b6 JH |
1220 | && look_ahead(opt, &left, &lno, &bol)) |
1221 | break; | |
83b5d2f5 JH |
1222 | eol = end_of_line(bol, &left); |
1223 | ch = *eol; | |
1224 | *eol = 0; | |
1225 | ||
480c1ca6 JH |
1226 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
1227 | ctx = GREP_CONTEXT_BODY; | |
1228 | ||
0ab7befa | 1229 | hit = match_line(opt, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
1230 | *eol = ch; |
1231 | ||
0ab7befa JH |
1232 | if (collect_hits) |
1233 | goto next_line; | |
1234 | ||
83b5d2f5 JH |
1235 | /* "grep -v -e foo -e bla" should list lines |
1236 | * that do not have either, so inversion should | |
1237 | * be done outside. | |
1238 | */ | |
1239 | if (opt->invert) | |
1240 | hit = !hit; | |
1241 | if (opt->unmatch_name_only) { | |
1242 | if (hit) | |
1243 | return 0; | |
1244 | goto next_line; | |
1245 | } | |
1246 | if (hit) { | |
1247 | count++; | |
1248 | if (opt->status_only) | |
1249 | return 1; | |
321ffcc0 | 1250 | if (opt->name_only) { |
e1327023 | 1251 | show_name(opt, gs->name); |
321ffcc0 RS |
1252 | return 1; |
1253 | } | |
c30c10cf RS |
1254 | if (opt->count) |
1255 | goto next_line; | |
83b5d2f5 | 1256 | if (binary_match_only) { |
5b594f45 | 1257 | opt->output(opt, "Binary file ", 12); |
e1327023 | 1258 | output_color(opt, gs->name, strlen(gs->name), |
55f638bd | 1259 | opt->color_filename); |
5b594f45 | 1260 | opt->output(opt, " matches\n", 9); |
83b5d2f5 JH |
1261 | return 1; |
1262 | } | |
83b5d2f5 JH |
1263 | /* Hit at this line. If we haven't shown the |
1264 | * pre-context lines, we would need to show them. | |
83b5d2f5 | 1265 | */ |
ba8ea749 | 1266 | if (opt->pre_context || opt->funcbody) |
e1327023 | 1267 | show_pre_context(opt, gs, bol, eol, lno); |
2944e4e6 | 1268 | else if (opt->funcname) |
e1327023 JK |
1269 | show_funcname_line(opt, gs, bol, lno); |
1270 | show_line(opt, bol, eol, gs->name, lno, ':'); | |
5dd06d38 | 1271 | last_hit = lno; |
ba8ea749 RS |
1272 | if (opt->funcbody) |
1273 | show_function = 1; | |
1274 | goto next_line; | |
83b5d2f5 | 1275 | } |
e1327023 | 1276 | if (show_function && match_funcname(opt, gs, bol, eol)) |
ba8ea749 RS |
1277 | show_function = 0; |
1278 | if (show_function || | |
1279 | (last_hit && lno <= last_hit + opt->post_context)) { | |
83b5d2f5 JH |
1280 | /* If the last hit is within the post context, |
1281 | * we need to show this line. | |
1282 | */ | |
e1327023 | 1283 | show_line(opt, bol, eol, gs->name, lno, '-'); |
83b5d2f5 | 1284 | } |
83b5d2f5 JH |
1285 | |
1286 | next_line: | |
1287 | bol = eol + 1; | |
1288 | if (!left) | |
1289 | break; | |
1290 | left--; | |
1291 | lno++; | |
1292 | } | |
1293 | ||
0ab7befa JH |
1294 | if (collect_hits) |
1295 | return 0; | |
b48fb5b6 | 1296 | |
83b5d2f5 JH |
1297 | if (opt->status_only) |
1298 | return 0; | |
1299 | if (opt->unmatch_name_only) { | |
1300 | /* We did not see any hit, so we want to show this */ | |
e1327023 | 1301 | show_name(opt, gs->name); |
83b5d2f5 JH |
1302 | return 1; |
1303 | } | |
1304 | ||
60ecac98 RS |
1305 | xdiff_clear_find_func(&xecfg); |
1306 | opt->priv = NULL; | |
1307 | ||
83b5d2f5 JH |
1308 | /* NEEDSWORK: |
1309 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
1310 | * which feels mostly useless but sometimes useful. Maybe | |
1311 | * make it another option? For now suppress them. | |
1312 | */ | |
5b594f45 FK |
1313 | if (opt->count && count) { |
1314 | char buf[32]; | |
e1327023 | 1315 | output_color(opt, gs->name, strlen(gs->name), opt->color_filename); |
55f638bd ML |
1316 | output_sep(opt, ':'); |
1317 | snprintf(buf, sizeof(buf), "%u\n", count); | |
5b594f45 | 1318 | opt->output(opt, buf, strlen(buf)); |
c30c10cf | 1319 | return 1; |
5b594f45 | 1320 | } |
83b5d2f5 JH |
1321 | return !!last_hit; |
1322 | } | |
1323 | ||
0ab7befa JH |
1324 | static void clr_hit_marker(struct grep_expr *x) |
1325 | { | |
1326 | /* All-hit markers are meaningful only at the very top level | |
1327 | * OR node. | |
1328 | */ | |
1329 | while (1) { | |
1330 | x->hit = 0; | |
1331 | if (x->node != GREP_NODE_OR) | |
1332 | return; | |
1333 | x->u.binary.left->hit = 0; | |
1334 | x = x->u.binary.right; | |
1335 | } | |
1336 | } | |
1337 | ||
1338 | static int chk_hit_marker(struct grep_expr *x) | |
1339 | { | |
1340 | /* Top level nodes have hit markers. See if they all are hits */ | |
1341 | while (1) { | |
1342 | if (x->node != GREP_NODE_OR) | |
1343 | return x->hit; | |
1344 | if (!x->u.binary.left->hit) | |
1345 | return 0; | |
1346 | x = x->u.binary.right; | |
1347 | } | |
1348 | } | |
1349 | ||
e1327023 | 1350 | int grep_source(struct grep_opt *opt, struct grep_source *gs) |
0ab7befa JH |
1351 | { |
1352 | /* | |
1353 | * we do not have to do the two-pass grep when we do not check | |
1354 | * buffer-wide "all-match". | |
1355 | */ | |
1356 | if (!opt->all_match) | |
e1327023 | 1357 | return grep_source_1(opt, gs, 0); |
0ab7befa JH |
1358 | |
1359 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
1360 | * We first clear hit markers from them. | |
1361 | */ | |
1362 | clr_hit_marker(opt->pattern_expression); | |
e1327023 | 1363 | grep_source_1(opt, gs, 1); |
0ab7befa JH |
1364 | |
1365 | if (!chk_hit_marker(opt->pattern_expression)) | |
1366 | return 0; | |
1367 | ||
e1327023 JK |
1368 | return grep_source_1(opt, gs, 0); |
1369 | } | |
1370 | ||
c876d6da | 1371 | int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size) |
e1327023 JK |
1372 | { |
1373 | struct grep_source gs; | |
1374 | int r; | |
1375 | ||
c876d6da | 1376 | grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL); |
e1327023 JK |
1377 | gs.buf = buf; |
1378 | gs.size = size; | |
1379 | ||
1380 | r = grep_source(opt, &gs); | |
1381 | ||
1382 | grep_source_clear(&gs); | |
1383 | return r; | |
1384 | } | |
1385 | ||
1386 | void grep_source_init(struct grep_source *gs, enum grep_source_type type, | |
1387 | const char *name, const void *identifier) | |
1388 | { | |
1389 | gs->type = type; | |
1390 | gs->name = name ? xstrdup(name) : NULL; | |
1391 | gs->buf = NULL; | |
1392 | gs->size = 0; | |
94ad9d9e | 1393 | gs->driver = NULL; |
e1327023 JK |
1394 | |
1395 | switch (type) { | |
1396 | case GREP_SOURCE_FILE: | |
1397 | gs->identifier = xstrdup(identifier); | |
1398 | break; | |
1399 | case GREP_SOURCE_SHA1: | |
1400 | gs->identifier = xmalloc(20); | |
1401 | memcpy(gs->identifier, identifier, 20); | |
1402 | break; | |
1403 | case GREP_SOURCE_BUF: | |
1404 | gs->identifier = NULL; | |
1405 | } | |
1406 | } | |
1407 | ||
1408 | void grep_source_clear(struct grep_source *gs) | |
1409 | { | |
1410 | free(gs->name); | |
1411 | gs->name = NULL; | |
1412 | free(gs->identifier); | |
1413 | gs->identifier = NULL; | |
1414 | grep_source_clear_data(gs); | |
1415 | } | |
1416 | ||
1417 | void grep_source_clear_data(struct grep_source *gs) | |
1418 | { | |
1419 | switch (gs->type) { | |
1420 | case GREP_SOURCE_FILE: | |
1421 | case GREP_SOURCE_SHA1: | |
1422 | free(gs->buf); | |
1423 | gs->buf = NULL; | |
1424 | gs->size = 0; | |
1425 | break; | |
1426 | case GREP_SOURCE_BUF: | |
1427 | /* leave user-provided buf intact */ | |
1428 | break; | |
1429 | } | |
1430 | } | |
1431 | ||
1432 | static int grep_source_load_sha1(struct grep_source *gs) | |
1433 | { | |
1434 | enum object_type type; | |
1435 | ||
1436 | grep_read_lock(); | |
1437 | gs->buf = read_sha1_file(gs->identifier, &type, &gs->size); | |
1438 | grep_read_unlock(); | |
1439 | ||
1440 | if (!gs->buf) | |
1441 | return error(_("'%s': unable to read %s"), | |
1442 | gs->name, | |
1443 | sha1_to_hex(gs->identifier)); | |
1444 | return 0; | |
1445 | } | |
1446 | ||
1447 | static int grep_source_load_file(struct grep_source *gs) | |
1448 | { | |
1449 | const char *filename = gs->identifier; | |
1450 | struct stat st; | |
1451 | char *data; | |
1452 | size_t size; | |
1453 | int i; | |
1454 | ||
1455 | if (lstat(filename, &st) < 0) { | |
1456 | err_ret: | |
1457 | if (errno != ENOENT) | |
1458 | error(_("'%s': %s"), filename, strerror(errno)); | |
1459 | return -1; | |
1460 | } | |
1461 | if (!S_ISREG(st.st_mode)) | |
1462 | return -1; | |
1463 | size = xsize_t(st.st_size); | |
1464 | i = open(filename, O_RDONLY); | |
1465 | if (i < 0) | |
1466 | goto err_ret; | |
1467 | data = xmalloc(size + 1); | |
1468 | if (st.st_size != read_in_full(i, data, size)) { | |
1469 | error(_("'%s': short read %s"), filename, strerror(errno)); | |
1470 | close(i); | |
1471 | free(data); | |
1472 | return -1; | |
1473 | } | |
1474 | close(i); | |
1475 | data[size] = 0; | |
1476 | ||
1477 | gs->buf = data; | |
1478 | gs->size = size; | |
1479 | return 0; | |
1480 | } | |
1481 | ||
3083301e | 1482 | static int grep_source_load(struct grep_source *gs) |
e1327023 JK |
1483 | { |
1484 | if (gs->buf) | |
1485 | return 0; | |
1486 | ||
1487 | switch (gs->type) { | |
1488 | case GREP_SOURCE_FILE: | |
1489 | return grep_source_load_file(gs); | |
1490 | case GREP_SOURCE_SHA1: | |
1491 | return grep_source_load_sha1(gs); | |
1492 | case GREP_SOURCE_BUF: | |
1493 | return gs->buf ? 0 : -1; | |
1494 | } | |
1495 | die("BUG: invalid grep_source type"); | |
0ab7befa | 1496 | } |
94ad9d9e JK |
1497 | |
1498 | void grep_source_load_driver(struct grep_source *gs) | |
1499 | { | |
1500 | if (gs->driver) | |
1501 | return; | |
1502 | ||
1503 | grep_attr_lock(); | |
1504 | gs->driver = userdiff_find_by_path(gs->name); | |
1505 | if (!gs->driver) | |
1506 | gs->driver = userdiff_find_by_name("default"); | |
1507 | grep_attr_unlock(); | |
1508 | } | |
41b59bfc | 1509 | |
3083301e | 1510 | static int grep_source_is_binary(struct grep_source *gs) |
41b59bfc JK |
1511 | { |
1512 | grep_source_load_driver(gs); | |
1513 | if (gs->driver->binary != -1) | |
1514 | return gs->driver->binary; | |
1515 | ||
1516 | if (!grep_source_load(gs)) | |
1517 | return buffer_is_binary(gs->buf, gs->size); | |
1518 | ||
1519 | return 0; | |
1520 | } |