]>
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 | |
a4d7d2c6 JH |
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->origin = "header"; | |
11 | p->no = 0; | |
12 | p->token = GREP_PATTERN_HEAD; | |
13 | p->field = field; | |
80235ba7 JH |
14 | *opt->header_tail = p; |
15 | opt->header_tail = &p->next; | |
a4d7d2c6 JH |
16 | p->next = NULL; |
17 | } | |
18 | ||
83b5d2f5 JH |
19 | void append_grep_pattern(struct grep_opt *opt, const char *pat, |
20 | const char *origin, int no, enum grep_pat_token t) | |
21 | { | |
22 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
23 | p->pattern = pat; | |
24 | p->origin = origin; | |
25 | p->no = no; | |
26 | p->token = t; | |
27 | *opt->pattern_tail = p; | |
28 | opt->pattern_tail = &p->next; | |
29 | p->next = NULL; | |
30 | } | |
31 | ||
5b594f45 FK |
32 | struct grep_opt *grep_opt_dup(const struct grep_opt *opt) |
33 | { | |
34 | struct grep_pat *pat; | |
35 | struct grep_opt *ret = xmalloc(sizeof(struct grep_opt)); | |
36 | *ret = *opt; | |
37 | ||
38 | ret->pattern_list = NULL; | |
39 | ret->pattern_tail = &ret->pattern_list; | |
40 | ||
41 | for(pat = opt->pattern_list; pat != NULL; pat = pat->next) | |
42 | { | |
43 | if(pat->token == GREP_PATTERN_HEAD) | |
44 | append_header_grep_pattern(ret, pat->field, | |
45 | pat->pattern); | |
46 | else | |
47 | append_grep_pattern(ret, pat->pattern, pat->origin, | |
48 | pat->no, pat->token); | |
49 | } | |
50 | ||
51 | return ret; | |
52 | } | |
53 | ||
83b5d2f5 JH |
54 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) |
55 | { | |
c822255c RS |
56 | int err; |
57 | ||
d7eb527d | 58 | p->word_regexp = opt->word_regexp; |
5183bf67 | 59 | p->ignore_case = opt->ignore_case; |
79286102 | 60 | p->fixed = opt->fixed; |
d7eb527d | 61 | |
c822255c RS |
62 | if (p->fixed) |
63 | return; | |
64 | ||
65 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
83b5d2f5 JH |
66 | if (err) { |
67 | char errbuf[1024]; | |
68 | char where[1024]; | |
69 | if (p->no) | |
70 | sprintf(where, "In '%s' at %d, ", | |
71 | p->origin, p->no); | |
72 | else if (p->origin) | |
73 | sprintf(where, "%s, ", p->origin); | |
74 | else | |
75 | where[0] = 0; | |
76 | regerror(err, &p->regexp, errbuf, 1024); | |
77 | regfree(&p->regexp); | |
78 | die("%s'%s': %s", where, p->pattern, errbuf); | |
79 | } | |
80 | } | |
81 | ||
0ab7befa | 82 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
83 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
84 | { | |
85 | struct grep_pat *p; | |
86 | struct grep_expr *x; | |
87 | ||
88 | p = *list; | |
c922b01f LT |
89 | if (!p) |
90 | return NULL; | |
83b5d2f5 JH |
91 | switch (p->token) { |
92 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
93 | case GREP_PATTERN_HEAD: |
94 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
95 | x = xcalloc(1, sizeof (struct grep_expr)); |
96 | x->node = GREP_NODE_ATOM; | |
97 | x->u.atom = p; | |
98 | *list = p->next; | |
99 | return x; | |
100 | case GREP_OPEN_PAREN: | |
101 | *list = p->next; | |
0ab7befa | 102 | x = compile_pattern_or(list); |
83b5d2f5 JH |
103 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
104 | die("unmatched parenthesis"); | |
105 | *list = (*list)->next; | |
106 | return x; | |
107 | default: | |
108 | return NULL; | |
109 | } | |
110 | } | |
111 | ||
112 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
113 | { | |
114 | struct grep_pat *p; | |
115 | struct grep_expr *x; | |
116 | ||
117 | p = *list; | |
c922b01f LT |
118 | if (!p) |
119 | return NULL; | |
83b5d2f5 JH |
120 | switch (p->token) { |
121 | case GREP_NOT: | |
122 | if (!p->next) | |
123 | die("--not not followed by pattern expression"); | |
124 | *list = p->next; | |
125 | x = xcalloc(1, sizeof (struct grep_expr)); | |
126 | x->node = GREP_NODE_NOT; | |
127 | x->u.unary = compile_pattern_not(list); | |
128 | if (!x->u.unary) | |
129 | die("--not followed by non pattern expression"); | |
130 | return x; | |
131 | default: | |
132 | return compile_pattern_atom(list); | |
133 | } | |
134 | } | |
135 | ||
136 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
137 | { | |
138 | struct grep_pat *p; | |
139 | struct grep_expr *x, *y, *z; | |
140 | ||
141 | x = compile_pattern_not(list); | |
142 | p = *list; | |
143 | if (p && p->token == GREP_AND) { | |
144 | if (!p->next) | |
145 | die("--and not followed by pattern expression"); | |
146 | *list = p->next; | |
147 | y = compile_pattern_and(list); | |
148 | if (!y) | |
149 | die("--and not followed by pattern expression"); | |
150 | z = xcalloc(1, sizeof (struct grep_expr)); | |
151 | z->node = GREP_NODE_AND; | |
152 | z->u.binary.left = x; | |
153 | z->u.binary.right = y; | |
154 | return z; | |
155 | } | |
156 | return x; | |
157 | } | |
158 | ||
159 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
160 | { | |
161 | struct grep_pat *p; | |
162 | struct grep_expr *x, *y, *z; | |
163 | ||
164 | x = compile_pattern_and(list); | |
165 | p = *list; | |
166 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
167 | y = compile_pattern_or(list); | |
168 | if (!y) | |
169 | die("not a pattern expression %s", p->pattern); | |
170 | z = xcalloc(1, sizeof (struct grep_expr)); | |
171 | z->node = GREP_NODE_OR; | |
172 | z->u.binary.left = x; | |
173 | z->u.binary.right = y; | |
174 | return z; | |
175 | } | |
176 | return x; | |
177 | } | |
178 | ||
179 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
180 | { | |
181 | return compile_pattern_or(list); | |
182 | } | |
183 | ||
184 | void compile_grep_patterns(struct grep_opt *opt) | |
185 | { | |
186 | struct grep_pat *p; | |
80235ba7 | 187 | struct grep_expr *header_expr = NULL; |
83b5d2f5 | 188 | |
80235ba7 JH |
189 | if (opt->header_list) { |
190 | p = opt->header_list; | |
191 | header_expr = compile_pattern_expr(&p); | |
192 | if (p) | |
193 | die("incomplete pattern expression: %s", p->pattern); | |
194 | for (p = opt->header_list; p; p = p->next) { | |
195 | switch (p->token) { | |
196 | case GREP_PATTERN: /* atom */ | |
197 | case GREP_PATTERN_HEAD: | |
198 | case GREP_PATTERN_BODY: | |
199 | compile_regexp(p, opt); | |
200 | break; | |
201 | default: | |
202 | opt->extended = 1; | |
203 | break; | |
204 | } | |
205 | } | |
206 | } | |
0ab7befa | 207 | |
83b5d2f5 | 208 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
209 | switch (p->token) { |
210 | case GREP_PATTERN: /* atom */ | |
211 | case GREP_PATTERN_HEAD: | |
212 | case GREP_PATTERN_BODY: | |
c822255c | 213 | compile_regexp(p, opt); |
480c1ca6 JH |
214 | break; |
215 | default: | |
83b5d2f5 | 216 | opt->extended = 1; |
480c1ca6 JH |
217 | break; |
218 | } | |
83b5d2f5 JH |
219 | } |
220 | ||
80235ba7 JH |
221 | if (opt->all_match || header_expr) |
222 | opt->extended = 1; | |
223 | else if (!opt->extended) | |
83b5d2f5 JH |
224 | return; |
225 | ||
226 | /* Then bundle them up in an expression. | |
227 | * A classic recursive descent parser would do. | |
228 | */ | |
229 | p = opt->pattern_list; | |
ba150a3f MB |
230 | if (p) |
231 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
232 | if (p) |
233 | die("incomplete pattern expression: %s", p->pattern); | |
80235ba7 JH |
234 | |
235 | if (!header_expr) | |
236 | return; | |
237 | ||
238 | if (opt->pattern_expression) { | |
239 | struct grep_expr *z; | |
240 | z = xcalloc(1, sizeof(*z)); | |
241 | z->node = GREP_NODE_OR; | |
242 | z->u.binary.left = opt->pattern_expression; | |
243 | z->u.binary.right = header_expr; | |
244 | opt->pattern_expression = z; | |
245 | } else { | |
246 | opt->pattern_expression = header_expr; | |
247 | } | |
248 | opt->all_match = 1; | |
83b5d2f5 JH |
249 | } |
250 | ||
b48fb5b6 JH |
251 | static void free_pattern_expr(struct grep_expr *x) |
252 | { | |
253 | switch (x->node) { | |
254 | case GREP_NODE_ATOM: | |
255 | break; | |
256 | case GREP_NODE_NOT: | |
257 | free_pattern_expr(x->u.unary); | |
258 | break; | |
259 | case GREP_NODE_AND: | |
260 | case GREP_NODE_OR: | |
261 | free_pattern_expr(x->u.binary.left); | |
262 | free_pattern_expr(x->u.binary.right); | |
263 | break; | |
264 | } | |
265 | free(x); | |
266 | } | |
267 | ||
268 | void free_grep_patterns(struct grep_opt *opt) | |
269 | { | |
270 | struct grep_pat *p, *n; | |
271 | ||
272 | for (p = opt->pattern_list; p; p = n) { | |
273 | n = p->next; | |
274 | switch (p->token) { | |
275 | case GREP_PATTERN: /* atom */ | |
276 | case GREP_PATTERN_HEAD: | |
277 | case GREP_PATTERN_BODY: | |
278 | regfree(&p->regexp); | |
279 | break; | |
280 | default: | |
281 | break; | |
282 | } | |
283 | free(p); | |
284 | } | |
285 | ||
286 | if (!opt->extended) | |
287 | return; | |
288 | free_pattern_expr(opt->pattern_expression); | |
289 | } | |
290 | ||
83b5d2f5 JH |
291 | static char *end_of_line(char *cp, unsigned long *left) |
292 | { | |
293 | unsigned long l = *left; | |
294 | while (l && *cp != '\n') { | |
295 | l--; | |
296 | cp++; | |
297 | } | |
298 | *left = l; | |
299 | return cp; | |
300 | } | |
301 | ||
302 | static int word_char(char ch) | |
303 | { | |
304 | return isalnum(ch) || ch == '_'; | |
305 | } | |
306 | ||
83caecca RZ |
307 | static void show_name(struct grep_opt *opt, const char *name) |
308 | { | |
5b594f45 FK |
309 | opt->output(opt, name, strlen(name)); |
310 | opt->output(opt, opt->null_following_name ? "\0" : "\n", 1); | |
83caecca RZ |
311 | } |
312 | ||
5183bf67 BC |
313 | |
314 | static int fixmatch(const char *pattern, char *line, int ignore_case, regmatch_t *match) | |
83b5d2f5 | 315 | { |
5183bf67 BC |
316 | char *hit; |
317 | if (ignore_case) | |
318 | hit = strcasestr(line, pattern); | |
319 | else | |
320 | hit = strstr(line, pattern); | |
321 | ||
83b5d2f5 JH |
322 | if (!hit) { |
323 | match->rm_so = match->rm_eo = -1; | |
324 | return REG_NOMATCH; | |
325 | } | |
326 | else { | |
327 | match->rm_so = hit - line; | |
328 | match->rm_eo = match->rm_so + strlen(pattern); | |
329 | return 0; | |
330 | } | |
331 | } | |
332 | ||
a4d7d2c6 JH |
333 | static int strip_timestamp(char *bol, char **eol_p) |
334 | { | |
335 | char *eol = *eol_p; | |
336 | int ch; | |
337 | ||
338 | while (bol < --eol) { | |
339 | if (*eol != '>') | |
340 | continue; | |
341 | *eol_p = ++eol; | |
342 | ch = *eol; | |
343 | *eol = '\0'; | |
344 | return ch; | |
345 | } | |
346 | return 0; | |
347 | } | |
348 | ||
349 | static struct { | |
350 | const char *field; | |
351 | size_t len; | |
352 | } header_field[] = { | |
353 | { "author ", 7 }, | |
354 | { "committer ", 10 }, | |
355 | }; | |
356 | ||
d7eb527d | 357 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
358 | enum grep_context ctx, |
359 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
360 | { |
361 | int hit = 0; | |
a4d7d2c6 | 362 | int saved_ch = 0; |
e701fadb | 363 | const char *start = bol; |
83b5d2f5 | 364 | |
480c1ca6 JH |
365 | if ((p->token != GREP_PATTERN) && |
366 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
367 | return 0; | |
368 | ||
a4d7d2c6 JH |
369 | if (p->token == GREP_PATTERN_HEAD) { |
370 | const char *field; | |
371 | size_t len; | |
372 | assert(p->field < ARRAY_SIZE(header_field)); | |
373 | field = header_field[p->field].field; | |
374 | len = header_field[p->field].len; | |
375 | if (strncmp(bol, field, len)) | |
376 | return 0; | |
377 | bol += len; | |
378 | saved_ch = strip_timestamp(bol, &eol); | |
379 | } | |
380 | ||
83b5d2f5 | 381 | again: |
79212772 | 382 | if (p->fixed) |
5183bf67 | 383 | hit = !fixmatch(p->pattern, bol, p->ignore_case, pmatch); |
79212772 RS |
384 | else |
385 | hit = !regexec(&p->regexp, bol, 1, pmatch, eflags); | |
83b5d2f5 | 386 | |
d7eb527d | 387 | if (hit && p->word_regexp) { |
83b5d2f5 | 388 | if ((pmatch[0].rm_so < 0) || |
84201eae | 389 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
390 | (pmatch[0].rm_eo < 0) || |
391 | (eol - bol) < pmatch[0].rm_eo) | |
392 | die("regexp returned nonsense"); | |
393 | ||
394 | /* Match beginning must be either beginning of the | |
395 | * line, or at word boundary (i.e. the last char must | |
396 | * not be a word char). Similarly, match end must be | |
397 | * either end of the line, or at word boundary | |
398 | * (i.e. the next char must not be a word char). | |
399 | */ | |
fb62eb7f | 400 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
401 | !word_char(bol[pmatch[0].rm_so-1])) && |
402 | ((pmatch[0].rm_eo == (eol-bol)) || | |
403 | !word_char(bol[pmatch[0].rm_eo])) ) | |
404 | ; | |
405 | else | |
406 | hit = 0; | |
407 | ||
84201eae RS |
408 | /* Words consist of at least one character. */ |
409 | if (pmatch->rm_so == pmatch->rm_eo) | |
410 | hit = 0; | |
411 | ||
83b5d2f5 JH |
412 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
413 | /* There could be more than one match on the | |
414 | * line, and the first match might not be | |
415 | * strict word match. But later ones could be! | |
fb62eb7f RS |
416 | * Forward to the next possible start, i.e. the |
417 | * next position following a non-word char. | |
83b5d2f5 JH |
418 | */ |
419 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
420 | while (word_char(bol[-1]) && bol < eol) |
421 | bol++; | |
dbb6a4ad | 422 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
423 | if (bol < eol) |
424 | goto again; | |
83b5d2f5 JH |
425 | } |
426 | } | |
a4d7d2c6 JH |
427 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
428 | *eol = saved_ch; | |
e701fadb RS |
429 | if (hit) { |
430 | pmatch[0].rm_so += bol - start; | |
431 | pmatch[0].rm_eo += bol - start; | |
432 | } | |
83b5d2f5 JH |
433 | return hit; |
434 | } | |
435 | ||
d7eb527d RS |
436 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, |
437 | enum grep_context ctx, int collect_hits) | |
83b5d2f5 | 438 | { |
0ab7befa | 439 | int h = 0; |
79212772 | 440 | regmatch_t match; |
0ab7befa | 441 | |
c922b01f LT |
442 | if (!x) |
443 | die("Not a valid grep expression"); | |
83b5d2f5 JH |
444 | switch (x->node) { |
445 | case GREP_NODE_ATOM: | |
79212772 | 446 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); |
83b5d2f5 JH |
447 | break; |
448 | case GREP_NODE_NOT: | |
d7eb527d | 449 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); |
0ab7befa | 450 | break; |
83b5d2f5 | 451 | case GREP_NODE_AND: |
d7eb527d | 452 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) |
252d560d | 453 | return 0; |
d7eb527d | 454 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); |
0ab7befa | 455 | break; |
83b5d2f5 | 456 | case GREP_NODE_OR: |
0ab7befa | 457 | if (!collect_hits) |
d7eb527d | 458 | return (match_expr_eval(x->u.binary.left, |
0ab7befa | 459 | bol, eol, ctx, 0) || |
d7eb527d | 460 | match_expr_eval(x->u.binary.right, |
0ab7befa | 461 | bol, eol, ctx, 0)); |
d7eb527d | 462 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); |
0ab7befa | 463 | x->u.binary.left->hit |= h; |
d7eb527d | 464 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); |
0ab7befa JH |
465 | break; |
466 | default: | |
d7530708 | 467 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 468 | } |
0ab7befa JH |
469 | if (collect_hits) |
470 | x->hit |= h; | |
471 | return h; | |
83b5d2f5 JH |
472 | } |
473 | ||
480c1ca6 | 474 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 475 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
476 | { |
477 | struct grep_expr *x = opt->pattern_expression; | |
d7eb527d | 478 | return match_expr_eval(x, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
479 | } |
480 | ||
480c1ca6 | 481 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 482 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
483 | { |
484 | struct grep_pat *p; | |
79212772 RS |
485 | regmatch_t match; |
486 | ||
83b5d2f5 | 487 | if (opt->extended) |
0ab7befa JH |
488 | return match_expr(opt, bol, eol, ctx, collect_hits); |
489 | ||
490 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 491 | for (p = opt->pattern_list; p; p = p->next) { |
79212772 | 492 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) |
83b5d2f5 JH |
493 | return 1; |
494 | } | |
495 | return 0; | |
496 | } | |
497 | ||
7e8f59d5 RS |
498 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
499 | enum grep_context ctx, | |
500 | regmatch_t *pmatch, int eflags) | |
501 | { | |
502 | regmatch_t match; | |
503 | ||
504 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
505 | return 0; | |
506 | if (match.rm_so < 0 || match.rm_eo < 0) | |
507 | return 0; | |
508 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
509 | if (match.rm_so > pmatch->rm_so) | |
510 | return 1; | |
511 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
512 | return 1; | |
513 | } | |
514 | pmatch->rm_so = match.rm_so; | |
515 | pmatch->rm_eo = match.rm_eo; | |
516 | return 1; | |
517 | } | |
518 | ||
519 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
520 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
521 | { | |
522 | struct grep_pat *p; | |
523 | int hit = 0; | |
524 | ||
525 | pmatch->rm_so = pmatch->rm_eo = -1; | |
526 | if (bol < eol) { | |
527 | for (p = opt->pattern_list; p; p = p->next) { | |
528 | switch (p->token) { | |
529 | case GREP_PATTERN: /* atom */ | |
530 | case GREP_PATTERN_HEAD: | |
531 | case GREP_PATTERN_BODY: | |
532 | hit |= match_next_pattern(p, bol, eol, ctx, | |
533 | pmatch, eflags); | |
534 | break; | |
535 | default: | |
536 | break; | |
537 | } | |
538 | } | |
539 | } | |
540 | return hit; | |
541 | } | |
542 | ||
543 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
544 | const char *name, unsigned lno, char sign) | |
545 | { | |
546 | int rest = eol - bol; | |
5b594f45 | 547 | char sign_str[1]; |
7e8f59d5 | 548 | |
5b594f45 | 549 | sign_str[0] = sign; |
ed24e401 | 550 | if (opt->pre_context || opt->post_context) { |
046802d0 RS |
551 | if (opt->last_shown == 0) { |
552 | if (opt->show_hunk_mark) | |
5b594f45 | 553 | opt->output(opt, "--\n", 3); |
046802d0 RS |
554 | else |
555 | opt->show_hunk_mark = 1; | |
ed24e401 | 556 | } else if (lno > opt->last_shown + 1) |
5b594f45 | 557 | opt->output(opt, "--\n", 3); |
5dd06d38 RS |
558 | } |
559 | opt->last_shown = lno; | |
560 | ||
7e8f59d5 | 561 | if (opt->null_following_name) |
5b594f45 FK |
562 | sign_str[0] = '\0'; |
563 | if (opt->pathname) { | |
564 | opt->output(opt, name, strlen(name)); | |
565 | opt->output(opt, sign_str, 1); | |
566 | } | |
567 | if (opt->linenum) { | |
568 | char buf[32]; | |
569 | snprintf(buf, sizeof(buf), "%d", lno); | |
570 | opt->output(opt, buf, strlen(buf)); | |
571 | opt->output(opt, sign_str, 1); | |
572 | } | |
7e8f59d5 RS |
573 | if (opt->color) { |
574 | regmatch_t match; | |
575 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
576 | int ch = *eol; | |
577 | int eflags = 0; | |
578 | ||
579 | *eol = '\0'; | |
580 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
581 | if (match.rm_so == match.rm_eo) |
582 | break; | |
5b594f45 FK |
583 | |
584 | opt->output(opt, bol, match.rm_so); | |
585 | opt->output(opt, opt->color_match, | |
586 | strlen(opt->color_match)); | |
587 | opt->output(opt, bol + match.rm_so, | |
588 | (int)(match.rm_eo - match.rm_so)); | |
589 | opt->output(opt, GIT_COLOR_RESET, | |
590 | strlen(GIT_COLOR_RESET)); | |
7e8f59d5 RS |
591 | bol += match.rm_eo; |
592 | rest -= match.rm_eo; | |
593 | eflags = REG_NOTBOL; | |
594 | } | |
595 | *eol = ch; | |
596 | } | |
5b594f45 FK |
597 | opt->output(opt, bol, rest); |
598 | opt->output(opt, "\n", 1); | |
7e8f59d5 RS |
599 | } |
600 | ||
60ecac98 | 601 | static int match_funcname(struct grep_opt *opt, char *bol, char *eol) |
2944e4e6 | 602 | { |
60ecac98 RS |
603 | xdemitconf_t *xecfg = opt->priv; |
604 | if (xecfg && xecfg->find_func) { | |
605 | char buf[1]; | |
606 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
607 | xecfg->find_func_priv) >= 0; | |
608 | } | |
609 | ||
2944e4e6 RS |
610 | if (bol == eol) |
611 | return 0; | |
612 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
613 | return 1; | |
614 | return 0; | |
615 | } | |
616 | ||
617 | static void show_funcname_line(struct grep_opt *opt, const char *name, | |
618 | char *buf, char *bol, unsigned lno) | |
619 | { | |
620 | while (bol > buf) { | |
621 | char *eol = --bol; | |
622 | ||
623 | while (bol > buf && bol[-1] != '\n') | |
624 | bol--; | |
625 | lno--; | |
626 | ||
627 | if (lno <= opt->last_shown) | |
628 | break; | |
629 | ||
60ecac98 | 630 | if (match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
631 | show_line(opt, bol, eol, name, lno, '='); |
632 | break; | |
633 | } | |
634 | } | |
635 | } | |
636 | ||
49de3216 RS |
637 | static void show_pre_context(struct grep_opt *opt, const char *name, char *buf, |
638 | char *bol, unsigned lno) | |
639 | { | |
2944e4e6 RS |
640 | unsigned cur = lno, from = 1, funcname_lno = 0; |
641 | int funcname_needed = opt->funcname; | |
49de3216 RS |
642 | |
643 | if (opt->pre_context < lno) | |
644 | from = lno - opt->pre_context; | |
645 | if (from <= opt->last_shown) | |
646 | from = opt->last_shown + 1; | |
647 | ||
648 | /* Rewind. */ | |
649 | while (bol > buf && cur > from) { | |
2944e4e6 RS |
650 | char *eol = --bol; |
651 | ||
49de3216 RS |
652 | while (bol > buf && bol[-1] != '\n') |
653 | bol--; | |
654 | cur--; | |
60ecac98 | 655 | if (funcname_needed && match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
656 | funcname_lno = cur; |
657 | funcname_needed = 0; | |
658 | } | |
49de3216 RS |
659 | } |
660 | ||
2944e4e6 RS |
661 | /* We need to look even further back to find a function signature. */ |
662 | if (opt->funcname && funcname_needed) | |
663 | show_funcname_line(opt, name, buf, bol, cur); | |
664 | ||
49de3216 RS |
665 | /* Back forward. */ |
666 | while (cur < lno) { | |
2944e4e6 | 667 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; |
49de3216 RS |
668 | |
669 | while (*eol != '\n') | |
670 | eol++; | |
2944e4e6 | 671 | show_line(opt, bol, eol, name, cur, sign); |
49de3216 RS |
672 | bol = eol + 1; |
673 | cur++; | |
674 | } | |
675 | } | |
676 | ||
a26345b6 JH |
677 | static int should_lookahead(struct grep_opt *opt) |
678 | { | |
679 | struct grep_pat *p; | |
680 | ||
681 | if (opt->extended) | |
682 | return 0; /* punt for too complex stuff */ | |
683 | if (opt->invert) | |
684 | return 0; | |
685 | for (p = opt->pattern_list; p; p = p->next) { | |
686 | if (p->token != GREP_PATTERN) | |
687 | return 0; /* punt for "header only" and stuff */ | |
688 | } | |
689 | return 1; | |
690 | } | |
691 | ||
692 | static int look_ahead(struct grep_opt *opt, | |
693 | unsigned long *left_p, | |
694 | unsigned *lno_p, | |
695 | char **bol_p) | |
696 | { | |
697 | unsigned lno = *lno_p; | |
698 | char *bol = *bol_p; | |
699 | struct grep_pat *p; | |
700 | char *sp, *last_bol; | |
701 | regoff_t earliest = -1; | |
702 | ||
703 | for (p = opt->pattern_list; p; p = p->next) { | |
704 | int hit; | |
705 | regmatch_t m; | |
706 | ||
707 | if (p->fixed) | |
e2d2e383 | 708 | hit = !fixmatch(p->pattern, bol, p->ignore_case, &m); |
24072c02 BK |
709 | else { |
710 | #ifdef REG_STARTEND | |
711 | m.rm_so = 0; | |
712 | m.rm_eo = *left_p; | |
713 | hit = !regexec(&p->regexp, bol, 1, &m, REG_STARTEND); | |
714 | #else | |
a26345b6 | 715 | hit = !regexec(&p->regexp, bol, 1, &m, 0); |
24072c02 BK |
716 | #endif |
717 | } | |
a26345b6 JH |
718 | if (!hit || m.rm_so < 0 || m.rm_eo < 0) |
719 | continue; | |
720 | if (earliest < 0 || m.rm_so < earliest) | |
721 | earliest = m.rm_so; | |
722 | } | |
723 | ||
724 | if (earliest < 0) { | |
725 | *bol_p = bol + *left_p; | |
726 | *left_p = 0; | |
727 | return 1; | |
728 | } | |
729 | for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--) | |
730 | ; /* find the beginning of the line */ | |
731 | last_bol = sp; | |
732 | ||
733 | for (sp = bol; sp < last_bol; sp++) { | |
734 | if (*sp == '\n') | |
735 | lno++; | |
736 | } | |
737 | *left_p -= last_bol - bol; | |
738 | *bol_p = last_bol; | |
739 | *lno_p = lno; | |
740 | return 0; | |
741 | } | |
742 | ||
5b594f45 FK |
743 | int grep_threads_ok(const struct grep_opt *opt) |
744 | { | |
745 | /* If this condition is true, then we may use the attribute | |
746 | * machinery in grep_buffer_1. The attribute code is not | |
747 | * thread safe, so we disable the use of threads. | |
748 | */ | |
749 | if (opt->funcname && !opt->unmatch_name_only && !opt->status_only && | |
750 | !opt->name_only) | |
751 | return 0; | |
752 | ||
753 | /* If we are showing hunk marks, we should not do it for the | |
754 | * first match. The synchronization problem we get for this | |
755 | * constraint is not yet solved, so we disable threading in | |
756 | * this case. | |
757 | */ | |
758 | if (opt->pre_context || opt->post_context) | |
759 | return 0; | |
760 | ||
761 | return 1; | |
762 | } | |
763 | ||
764 | static void std_output(struct grep_opt *opt, const void *buf, size_t size) | |
765 | { | |
766 | fwrite(buf, size, 1, stdout); | |
767 | } | |
768 | ||
0ab7befa JH |
769 | static int grep_buffer_1(struct grep_opt *opt, const char *name, |
770 | char *buf, unsigned long size, int collect_hits) | |
83b5d2f5 JH |
771 | { |
772 | char *bol = buf; | |
773 | unsigned long left = size; | |
774 | unsigned lno = 1; | |
83b5d2f5 | 775 | unsigned last_hit = 0; |
83b5d2f5 | 776 | int binary_match_only = 0; |
83b5d2f5 | 777 | unsigned count = 0; |
a26345b6 | 778 | int try_lookahead = 0; |
480c1ca6 | 779 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
60ecac98 | 780 | xdemitconf_t xecfg; |
83b5d2f5 | 781 | |
5dd06d38 RS |
782 | opt->last_shown = 0; |
783 | ||
5b594f45 FK |
784 | if (!opt->output) |
785 | opt->output = std_output; | |
786 | ||
83b5d2f5 JH |
787 | if (buffer_is_binary(buf, size)) { |
788 | switch (opt->binary) { | |
789 | case GREP_BINARY_DEFAULT: | |
790 | binary_match_only = 1; | |
791 | break; | |
792 | case GREP_BINARY_NOMATCH: | |
793 | return 0; /* Assume unmatch */ | |
794 | break; | |
795 | default: | |
796 | break; | |
797 | } | |
798 | } | |
799 | ||
60ecac98 RS |
800 | memset(&xecfg, 0, sizeof(xecfg)); |
801 | if (opt->funcname && !opt->unmatch_name_only && !opt->status_only && | |
802 | !opt->name_only && !binary_match_only && !collect_hits) { | |
803 | struct userdiff_driver *drv = userdiff_find_by_path(name); | |
804 | if (drv && drv->funcname.pattern) { | |
805 | const struct userdiff_funcname *pe = &drv->funcname; | |
806 | xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags); | |
807 | opt->priv = &xecfg; | |
808 | } | |
809 | } | |
a26345b6 | 810 | try_lookahead = should_lookahead(opt); |
60ecac98 | 811 | |
83b5d2f5 JH |
812 | while (left) { |
813 | char *eol, ch; | |
0ab7befa | 814 | int hit; |
83b5d2f5 | 815 | |
a26345b6 JH |
816 | /* |
817 | * look_ahead() skips quicly to the line that possibly | |
818 | * has the next hit; don't call it if we need to do | |
819 | * something more than just skipping the current line | |
820 | * in response to an unmatch for the current line. E.g. | |
821 | * inside a post-context window, we will show the current | |
822 | * line as a context around the previous hit when it | |
823 | * doesn't hit. | |
824 | */ | |
825 | if (try_lookahead | |
826 | && !(last_hit | |
827 | && lno <= last_hit + opt->post_context) | |
828 | && look_ahead(opt, &left, &lno, &bol)) | |
829 | break; | |
83b5d2f5 JH |
830 | eol = end_of_line(bol, &left); |
831 | ch = *eol; | |
832 | *eol = 0; | |
833 | ||
480c1ca6 JH |
834 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
835 | ctx = GREP_CONTEXT_BODY; | |
836 | ||
0ab7befa | 837 | hit = match_line(opt, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
838 | *eol = ch; |
839 | ||
0ab7befa JH |
840 | if (collect_hits) |
841 | goto next_line; | |
842 | ||
83b5d2f5 JH |
843 | /* "grep -v -e foo -e bla" should list lines |
844 | * that do not have either, so inversion should | |
845 | * be done outside. | |
846 | */ | |
847 | if (opt->invert) | |
848 | hit = !hit; | |
849 | if (opt->unmatch_name_only) { | |
850 | if (hit) | |
851 | return 0; | |
852 | goto next_line; | |
853 | } | |
854 | if (hit) { | |
855 | count++; | |
856 | if (opt->status_only) | |
857 | return 1; | |
858 | if (binary_match_only) { | |
5b594f45 FK |
859 | opt->output(opt, "Binary file ", 12); |
860 | opt->output(opt, name, strlen(name)); | |
861 | opt->output(opt, " matches\n", 9); | |
83b5d2f5 JH |
862 | return 1; |
863 | } | |
864 | if (opt->name_only) { | |
83caecca | 865 | show_name(opt, name); |
83b5d2f5 JH |
866 | return 1; |
867 | } | |
868 | /* Hit at this line. If we haven't shown the | |
869 | * pre-context lines, we would need to show them. | |
870 | * When asked to do "count", this still show | |
871 | * the context which is nonsense, but the user | |
872 | * deserves to get that ;-). | |
873 | */ | |
49de3216 RS |
874 | if (opt->pre_context) |
875 | show_pre_context(opt, name, buf, bol, lno); | |
2944e4e6 RS |
876 | else if (opt->funcname) |
877 | show_funcname_line(opt, name, buf, bol, lno); | |
83b5d2f5 JH |
878 | if (!opt->count) |
879 | show_line(opt, bol, eol, name, lno, ':'); | |
5dd06d38 | 880 | last_hit = lno; |
83b5d2f5 JH |
881 | } |
882 | else if (last_hit && | |
883 | lno <= last_hit + opt->post_context) { | |
884 | /* If the last hit is within the post context, | |
885 | * we need to show this line. | |
886 | */ | |
83b5d2f5 | 887 | show_line(opt, bol, eol, name, lno, '-'); |
83b5d2f5 | 888 | } |
83b5d2f5 JH |
889 | |
890 | next_line: | |
891 | bol = eol + 1; | |
892 | if (!left) | |
893 | break; | |
894 | left--; | |
895 | lno++; | |
896 | } | |
897 | ||
0ab7befa JH |
898 | if (collect_hits) |
899 | return 0; | |
b48fb5b6 | 900 | |
83b5d2f5 JH |
901 | if (opt->status_only) |
902 | return 0; | |
903 | if (opt->unmatch_name_only) { | |
904 | /* We did not see any hit, so we want to show this */ | |
83caecca | 905 | show_name(opt, name); |
83b5d2f5 JH |
906 | return 1; |
907 | } | |
908 | ||
60ecac98 RS |
909 | xdiff_clear_find_func(&xecfg); |
910 | opt->priv = NULL; | |
911 | ||
83b5d2f5 JH |
912 | /* NEEDSWORK: |
913 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
914 | * which feels mostly useless but sometimes useful. Maybe | |
915 | * make it another option? For now suppress them. | |
916 | */ | |
5b594f45 FK |
917 | if (opt->count && count) { |
918 | char buf[32]; | |
919 | opt->output(opt, name, strlen(name)); | |
920 | snprintf(buf, sizeof(buf), "%c%u\n", | |
921 | opt->null_following_name ? '\0' : ':', count); | |
922 | opt->output(opt, buf, strlen(buf)); | |
923 | } | |
83b5d2f5 JH |
924 | return !!last_hit; |
925 | } | |
926 | ||
0ab7befa JH |
927 | static void clr_hit_marker(struct grep_expr *x) |
928 | { | |
929 | /* All-hit markers are meaningful only at the very top level | |
930 | * OR node. | |
931 | */ | |
932 | while (1) { | |
933 | x->hit = 0; | |
934 | if (x->node != GREP_NODE_OR) | |
935 | return; | |
936 | x->u.binary.left->hit = 0; | |
937 | x = x->u.binary.right; | |
938 | } | |
939 | } | |
940 | ||
941 | static int chk_hit_marker(struct grep_expr *x) | |
942 | { | |
943 | /* Top level nodes have hit markers. See if they all are hits */ | |
944 | while (1) { | |
945 | if (x->node != GREP_NODE_OR) | |
946 | return x->hit; | |
947 | if (!x->u.binary.left->hit) | |
948 | return 0; | |
949 | x = x->u.binary.right; | |
950 | } | |
951 | } | |
952 | ||
953 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
954 | { | |
955 | /* | |
956 | * we do not have to do the two-pass grep when we do not check | |
957 | * buffer-wide "all-match". | |
958 | */ | |
959 | if (!opt->all_match) | |
960 | return grep_buffer_1(opt, name, buf, size, 0); | |
961 | ||
962 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
963 | * We first clear hit markers from them. | |
964 | */ | |
965 | clr_hit_marker(opt->pattern_expression); | |
966 | grep_buffer_1(opt, name, buf, size, 1); | |
967 | ||
968 | if (!chk_hit_marker(opt->pattern_expression)) | |
969 | return 0; | |
970 | ||
971 | return grep_buffer_1(opt, name, buf, size, 0); | |
972 | } |