]>
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; | |
ed40a095 | 10 | p->patternlen = strlen(pat); |
a4d7d2c6 JH |
11 | p->origin = "header"; |
12 | p->no = 0; | |
13 | p->token = GREP_PATTERN_HEAD; | |
14 | p->field = field; | |
80235ba7 JH |
15 | *opt->header_tail = p; |
16 | opt->header_tail = &p->next; | |
a4d7d2c6 JH |
17 | p->next = NULL; |
18 | } | |
19 | ||
83b5d2f5 JH |
20 | void append_grep_pattern(struct grep_opt *opt, const char *pat, |
21 | const char *origin, int no, enum grep_pat_token t) | |
ed40a095 RS |
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) | |
83b5d2f5 JH |
28 | { |
29 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
30 | p->pattern = pat; | |
ed40a095 | 31 | p->patternlen = patlen; |
83b5d2f5 JH |
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 | ||
5b594f45 FK |
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 | |
ed40a095 RS |
55 | append_grep_pat(ret, pat->pattern, pat->patternlen, |
56 | pat->origin, pat->no, pat->token); | |
5b594f45 FK |
57 | } |
58 | ||
59 | return ret; | |
60 | } | |
61 | ||
83b5d2f5 JH |
62 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) |
63 | { | |
c822255c RS |
64 | int err; |
65 | ||
d7eb527d | 66 | p->word_regexp = opt->word_regexp; |
5183bf67 | 67 | p->ignore_case = opt->ignore_case; |
79286102 | 68 | p->fixed = opt->fixed; |
d7eb527d | 69 | |
c822255c RS |
70 | if (p->fixed) |
71 | return; | |
72 | ||
73 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
83b5d2f5 JH |
74 | if (err) { |
75 | char errbuf[1024]; | |
76 | char where[1024]; | |
77 | if (p->no) | |
78 | sprintf(where, "In '%s' at %d, ", | |
79 | p->origin, p->no); | |
80 | else if (p->origin) | |
81 | sprintf(where, "%s, ", p->origin); | |
82 | else | |
83 | where[0] = 0; | |
84 | regerror(err, &p->regexp, errbuf, 1024); | |
85 | regfree(&p->regexp); | |
86 | die("%s'%s': %s", where, p->pattern, errbuf); | |
87 | } | |
88 | } | |
89 | ||
0ab7befa | 90 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
91 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
92 | { | |
93 | struct grep_pat *p; | |
94 | struct grep_expr *x; | |
95 | ||
96 | p = *list; | |
c922b01f LT |
97 | if (!p) |
98 | return NULL; | |
83b5d2f5 JH |
99 | switch (p->token) { |
100 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
101 | case GREP_PATTERN_HEAD: |
102 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
103 | x = xcalloc(1, sizeof (struct grep_expr)); |
104 | x->node = GREP_NODE_ATOM; | |
105 | x->u.atom = p; | |
106 | *list = p->next; | |
107 | return x; | |
108 | case GREP_OPEN_PAREN: | |
109 | *list = p->next; | |
0ab7befa | 110 | x = compile_pattern_or(list); |
83b5d2f5 JH |
111 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
112 | die("unmatched parenthesis"); | |
113 | *list = (*list)->next; | |
114 | return x; | |
115 | default: | |
116 | return NULL; | |
117 | } | |
118 | } | |
119 | ||
120 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
121 | { | |
122 | struct grep_pat *p; | |
123 | struct grep_expr *x; | |
124 | ||
125 | p = *list; | |
c922b01f LT |
126 | if (!p) |
127 | return NULL; | |
83b5d2f5 JH |
128 | switch (p->token) { |
129 | case GREP_NOT: | |
130 | if (!p->next) | |
131 | die("--not not followed by pattern expression"); | |
132 | *list = p->next; | |
133 | x = xcalloc(1, sizeof (struct grep_expr)); | |
134 | x->node = GREP_NODE_NOT; | |
135 | x->u.unary = compile_pattern_not(list); | |
136 | if (!x->u.unary) | |
137 | die("--not followed by non pattern expression"); | |
138 | return x; | |
139 | default: | |
140 | return compile_pattern_atom(list); | |
141 | } | |
142 | } | |
143 | ||
144 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
145 | { | |
146 | struct grep_pat *p; | |
147 | struct grep_expr *x, *y, *z; | |
148 | ||
149 | x = compile_pattern_not(list); | |
150 | p = *list; | |
151 | if (p && p->token == GREP_AND) { | |
152 | if (!p->next) | |
153 | die("--and not followed by pattern expression"); | |
154 | *list = p->next; | |
155 | y = compile_pattern_and(list); | |
156 | if (!y) | |
157 | die("--and not followed by pattern expression"); | |
158 | z = xcalloc(1, sizeof (struct grep_expr)); | |
159 | z->node = GREP_NODE_AND; | |
160 | z->u.binary.left = x; | |
161 | z->u.binary.right = y; | |
162 | return z; | |
163 | } | |
164 | return x; | |
165 | } | |
166 | ||
167 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
168 | { | |
169 | struct grep_pat *p; | |
170 | struct grep_expr *x, *y, *z; | |
171 | ||
172 | x = compile_pattern_and(list); | |
173 | p = *list; | |
174 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
175 | y = compile_pattern_or(list); | |
176 | if (!y) | |
177 | die("not a pattern expression %s", p->pattern); | |
178 | z = xcalloc(1, sizeof (struct grep_expr)); | |
179 | z->node = GREP_NODE_OR; | |
180 | z->u.binary.left = x; | |
181 | z->u.binary.right = y; | |
182 | return z; | |
183 | } | |
184 | return x; | |
185 | } | |
186 | ||
187 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
188 | { | |
189 | return compile_pattern_or(list); | |
190 | } | |
191 | ||
5aaeb733 JH |
192 | static struct grep_expr *grep_true_expr(void) |
193 | { | |
194 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
195 | z->node = GREP_NODE_TRUE; | |
196 | return z; | |
197 | } | |
198 | ||
199 | static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right) | |
200 | { | |
201 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
202 | z->node = GREP_NODE_OR; | |
203 | z->u.binary.left = left; | |
204 | z->u.binary.right = right; | |
205 | return z; | |
206 | } | |
207 | ||
95ce9ce2 | 208 | static struct grep_expr *prep_header_patterns(struct grep_opt *opt) |
83b5d2f5 JH |
209 | { |
210 | struct grep_pat *p; | |
95ce9ce2 | 211 | struct grep_expr *header_expr; |
5aaeb733 JH |
212 | struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]); |
213 | enum grep_header_field fld; | |
83b5d2f5 | 214 | |
95ce9ce2 JH |
215 | if (!opt->header_list) |
216 | return NULL; | |
217 | p = opt->header_list; | |
95ce9ce2 JH |
218 | for (p = opt->header_list; p; p = p->next) { |
219 | if (p->token != GREP_PATTERN_HEAD) | |
220 | die("bug: a non-header pattern in grep header list."); | |
221 | if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field) | |
222 | die("bug: unknown header field %d", p->field); | |
223 | compile_regexp(p, opt); | |
80235ba7 | 224 | } |
5aaeb733 JH |
225 | |
226 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) | |
227 | header_group[fld] = NULL; | |
228 | ||
229 | for (p = opt->header_list; p; p = p->next) { | |
230 | struct grep_expr *h; | |
231 | struct grep_pat *pp = p; | |
232 | ||
233 | h = compile_pattern_atom(&pp); | |
234 | if (!h || pp != p->next) | |
235 | die("bug: malformed header expr"); | |
236 | if (!header_group[p->field]) { | |
237 | header_group[p->field] = h; | |
238 | continue; | |
239 | } | |
240 | header_group[p->field] = grep_or_expr(h, header_group[p->field]); | |
241 | } | |
242 | ||
243 | header_expr = NULL; | |
244 | ||
245 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) { | |
246 | if (!header_group[fld]) | |
247 | continue; | |
248 | if (!header_expr) | |
249 | header_expr = grep_true_expr(); | |
250 | header_expr = grep_or_expr(header_group[fld], header_expr); | |
251 | } | |
95ce9ce2 JH |
252 | return header_expr; |
253 | } | |
254 | ||
255 | void compile_grep_patterns(struct grep_opt *opt) | |
256 | { | |
257 | struct grep_pat *p; | |
258 | struct grep_expr *header_expr = prep_header_patterns(opt); | |
0ab7befa | 259 | |
83b5d2f5 | 260 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
261 | switch (p->token) { |
262 | case GREP_PATTERN: /* atom */ | |
263 | case GREP_PATTERN_HEAD: | |
264 | case GREP_PATTERN_BODY: | |
c822255c | 265 | compile_regexp(p, opt); |
480c1ca6 JH |
266 | break; |
267 | default: | |
83b5d2f5 | 268 | opt->extended = 1; |
480c1ca6 JH |
269 | break; |
270 | } | |
83b5d2f5 JH |
271 | } |
272 | ||
80235ba7 JH |
273 | if (opt->all_match || header_expr) |
274 | opt->extended = 1; | |
275 | else if (!opt->extended) | |
83b5d2f5 JH |
276 | return; |
277 | ||
83b5d2f5 | 278 | p = opt->pattern_list; |
ba150a3f MB |
279 | if (p) |
280 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
281 | if (p) |
282 | die("incomplete pattern expression: %s", p->pattern); | |
80235ba7 JH |
283 | |
284 | if (!header_expr) | |
285 | return; | |
286 | ||
5aaeb733 | 287 | if (!opt->pattern_expression) |
80235ba7 | 288 | opt->pattern_expression = header_expr; |
5aaeb733 JH |
289 | else |
290 | opt->pattern_expression = grep_or_expr(opt->pattern_expression, | |
291 | header_expr); | |
80235ba7 | 292 | opt->all_match = 1; |
83b5d2f5 JH |
293 | } |
294 | ||
b48fb5b6 JH |
295 | static void free_pattern_expr(struct grep_expr *x) |
296 | { | |
297 | switch (x->node) { | |
5aaeb733 | 298 | case GREP_NODE_TRUE: |
b48fb5b6 JH |
299 | case GREP_NODE_ATOM: |
300 | break; | |
301 | case GREP_NODE_NOT: | |
302 | free_pattern_expr(x->u.unary); | |
303 | break; | |
304 | case GREP_NODE_AND: | |
305 | case GREP_NODE_OR: | |
306 | free_pattern_expr(x->u.binary.left); | |
307 | free_pattern_expr(x->u.binary.right); | |
308 | break; | |
309 | } | |
310 | free(x); | |
311 | } | |
312 | ||
313 | void free_grep_patterns(struct grep_opt *opt) | |
314 | { | |
315 | struct grep_pat *p, *n; | |
316 | ||
317 | for (p = opt->pattern_list; p; p = n) { | |
318 | n = p->next; | |
319 | switch (p->token) { | |
320 | case GREP_PATTERN: /* atom */ | |
321 | case GREP_PATTERN_HEAD: | |
322 | case GREP_PATTERN_BODY: | |
323 | regfree(&p->regexp); | |
324 | break; | |
325 | default: | |
326 | break; | |
327 | } | |
328 | free(p); | |
329 | } | |
330 | ||
331 | if (!opt->extended) | |
332 | return; | |
333 | free_pattern_expr(opt->pattern_expression); | |
334 | } | |
335 | ||
83b5d2f5 JH |
336 | static char *end_of_line(char *cp, unsigned long *left) |
337 | { | |
338 | unsigned long l = *left; | |
339 | while (l && *cp != '\n') { | |
340 | l--; | |
341 | cp++; | |
342 | } | |
343 | *left = l; | |
344 | return cp; | |
345 | } | |
346 | ||
347 | static int word_char(char ch) | |
348 | { | |
349 | return isalnum(ch) || ch == '_'; | |
350 | } | |
351 | ||
55f638bd ML |
352 | static void output_color(struct grep_opt *opt, const void *data, size_t size, |
353 | const char *color) | |
354 | { | |
355 | if (opt->color && color && color[0]) { | |
356 | opt->output(opt, color, strlen(color)); | |
357 | opt->output(opt, data, size); | |
358 | opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET)); | |
359 | } else | |
360 | opt->output(opt, data, size); | |
361 | } | |
362 | ||
363 | static void output_sep(struct grep_opt *opt, char sign) | |
364 | { | |
365 | if (opt->null_following_name) | |
366 | opt->output(opt, "\0", 1); | |
367 | else | |
368 | output_color(opt, &sign, 1, opt->color_sep); | |
369 | } | |
370 | ||
83caecca RZ |
371 | static void show_name(struct grep_opt *opt, const char *name) |
372 | { | |
55f638bd | 373 | output_color(opt, name, strlen(name), opt->color_filename); |
5b594f45 | 374 | opt->output(opt, opt->null_following_name ? "\0" : "\n", 1); |
83caecca RZ |
375 | } |
376 | ||
ed40a095 RS |
377 | static int fixmatch(struct grep_pat *p, char *line, char *eol, |
378 | regmatch_t *match) | |
83b5d2f5 | 379 | { |
5183bf67 | 380 | char *hit; |
1baddf4b | 381 | |
ed40a095 | 382 | if (p->ignore_case) { |
52d799a7 RS |
383 | char *s = line; |
384 | do { | |
ed40a095 | 385 | hit = strcasestr(s, p->pattern); |
52d799a7 RS |
386 | if (hit) |
387 | break; | |
388 | s += strlen(s) + 1; | |
389 | } while (s < eol); | |
390 | } else | |
ed40a095 | 391 | hit = memmem(line, eol - line, p->pattern, p->patternlen); |
5183bf67 | 392 | |
83b5d2f5 JH |
393 | if (!hit) { |
394 | match->rm_so = match->rm_eo = -1; | |
395 | return REG_NOMATCH; | |
396 | } | |
397 | else { | |
398 | match->rm_so = hit - line; | |
ed40a095 | 399 | match->rm_eo = match->rm_so + p->patternlen; |
83b5d2f5 JH |
400 | return 0; |
401 | } | |
402 | } | |
403 | ||
f96e5673 RS |
404 | static int regmatch(const regex_t *preg, char *line, char *eol, |
405 | regmatch_t *match, int eflags) | |
406 | { | |
407 | #ifdef REG_STARTEND | |
408 | match->rm_so = 0; | |
409 | match->rm_eo = eol - line; | |
410 | eflags |= REG_STARTEND; | |
411 | #endif | |
412 | return regexec(preg, line, 1, match, eflags); | |
413 | } | |
414 | ||
a4d7d2c6 JH |
415 | static int strip_timestamp(char *bol, char **eol_p) |
416 | { | |
417 | char *eol = *eol_p; | |
418 | int ch; | |
419 | ||
420 | while (bol < --eol) { | |
421 | if (*eol != '>') | |
422 | continue; | |
423 | *eol_p = ++eol; | |
424 | ch = *eol; | |
425 | *eol = '\0'; | |
426 | return ch; | |
427 | } | |
428 | return 0; | |
429 | } | |
430 | ||
431 | static struct { | |
432 | const char *field; | |
433 | size_t len; | |
434 | } header_field[] = { | |
435 | { "author ", 7 }, | |
436 | { "committer ", 10 }, | |
437 | }; | |
438 | ||
d7eb527d | 439 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
440 | enum grep_context ctx, |
441 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
442 | { |
443 | int hit = 0; | |
a4d7d2c6 | 444 | int saved_ch = 0; |
e701fadb | 445 | const char *start = bol; |
83b5d2f5 | 446 | |
480c1ca6 JH |
447 | if ((p->token != GREP_PATTERN) && |
448 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
449 | return 0; | |
450 | ||
a4d7d2c6 JH |
451 | if (p->token == GREP_PATTERN_HEAD) { |
452 | const char *field; | |
453 | size_t len; | |
454 | assert(p->field < ARRAY_SIZE(header_field)); | |
455 | field = header_field[p->field].field; | |
456 | len = header_field[p->field].len; | |
457 | if (strncmp(bol, field, len)) | |
458 | return 0; | |
459 | bol += len; | |
460 | saved_ch = strip_timestamp(bol, &eol); | |
461 | } | |
462 | ||
83b5d2f5 | 463 | again: |
79212772 | 464 | if (p->fixed) |
ed40a095 | 465 | hit = !fixmatch(p, bol, eol, pmatch); |
79212772 | 466 | else |
f96e5673 | 467 | hit = !regmatch(&p->regexp, bol, eol, pmatch, eflags); |
83b5d2f5 | 468 | |
d7eb527d | 469 | if (hit && p->word_regexp) { |
83b5d2f5 | 470 | if ((pmatch[0].rm_so < 0) || |
84201eae | 471 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
472 | (pmatch[0].rm_eo < 0) || |
473 | (eol - bol) < pmatch[0].rm_eo) | |
474 | die("regexp returned nonsense"); | |
475 | ||
476 | /* Match beginning must be either beginning of the | |
477 | * line, or at word boundary (i.e. the last char must | |
478 | * not be a word char). Similarly, match end must be | |
479 | * either end of the line, or at word boundary | |
480 | * (i.e. the next char must not be a word char). | |
481 | */ | |
fb62eb7f | 482 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
483 | !word_char(bol[pmatch[0].rm_so-1])) && |
484 | ((pmatch[0].rm_eo == (eol-bol)) || | |
485 | !word_char(bol[pmatch[0].rm_eo])) ) | |
486 | ; | |
487 | else | |
488 | hit = 0; | |
489 | ||
84201eae RS |
490 | /* Words consist of at least one character. */ |
491 | if (pmatch->rm_so == pmatch->rm_eo) | |
492 | hit = 0; | |
493 | ||
83b5d2f5 JH |
494 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
495 | /* There could be more than one match on the | |
496 | * line, and the first match might not be | |
497 | * strict word match. But later ones could be! | |
fb62eb7f RS |
498 | * Forward to the next possible start, i.e. the |
499 | * next position following a non-word char. | |
83b5d2f5 JH |
500 | */ |
501 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
502 | while (word_char(bol[-1]) && bol < eol) |
503 | bol++; | |
dbb6a4ad | 504 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
505 | if (bol < eol) |
506 | goto again; | |
83b5d2f5 JH |
507 | } |
508 | } | |
a4d7d2c6 JH |
509 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
510 | *eol = saved_ch; | |
e701fadb RS |
511 | if (hit) { |
512 | pmatch[0].rm_so += bol - start; | |
513 | pmatch[0].rm_eo += bol - start; | |
514 | } | |
83b5d2f5 JH |
515 | return hit; |
516 | } | |
517 | ||
d7eb527d RS |
518 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, |
519 | enum grep_context ctx, int collect_hits) | |
83b5d2f5 | 520 | { |
0ab7befa | 521 | int h = 0; |
79212772 | 522 | regmatch_t match; |
0ab7befa | 523 | |
c922b01f LT |
524 | if (!x) |
525 | die("Not a valid grep expression"); | |
83b5d2f5 | 526 | switch (x->node) { |
5aaeb733 JH |
527 | case GREP_NODE_TRUE: |
528 | h = 1; | |
529 | break; | |
83b5d2f5 | 530 | case GREP_NODE_ATOM: |
79212772 | 531 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); |
83b5d2f5 JH |
532 | break; |
533 | case GREP_NODE_NOT: | |
d7eb527d | 534 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); |
0ab7befa | 535 | break; |
83b5d2f5 | 536 | case GREP_NODE_AND: |
d7eb527d | 537 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) |
252d560d | 538 | return 0; |
d7eb527d | 539 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); |
0ab7befa | 540 | break; |
83b5d2f5 | 541 | case GREP_NODE_OR: |
0ab7befa | 542 | if (!collect_hits) |
d7eb527d | 543 | return (match_expr_eval(x->u.binary.left, |
0ab7befa | 544 | bol, eol, ctx, 0) || |
d7eb527d | 545 | match_expr_eval(x->u.binary.right, |
0ab7befa | 546 | bol, eol, ctx, 0)); |
d7eb527d | 547 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); |
0ab7befa | 548 | x->u.binary.left->hit |= h; |
d7eb527d | 549 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); |
0ab7befa JH |
550 | break; |
551 | default: | |
d7530708 | 552 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 553 | } |
0ab7befa JH |
554 | if (collect_hits) |
555 | x->hit |= h; | |
556 | return h; | |
83b5d2f5 JH |
557 | } |
558 | ||
480c1ca6 | 559 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 560 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
561 | { |
562 | struct grep_expr *x = opt->pattern_expression; | |
d7eb527d | 563 | return match_expr_eval(x, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
564 | } |
565 | ||
480c1ca6 | 566 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 567 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
568 | { |
569 | struct grep_pat *p; | |
79212772 RS |
570 | regmatch_t match; |
571 | ||
83b5d2f5 | 572 | if (opt->extended) |
0ab7befa JH |
573 | return match_expr(opt, bol, eol, ctx, collect_hits); |
574 | ||
575 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 576 | for (p = opt->pattern_list; p; p = p->next) { |
79212772 | 577 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) |
83b5d2f5 JH |
578 | return 1; |
579 | } | |
580 | return 0; | |
581 | } | |
582 | ||
7e8f59d5 RS |
583 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
584 | enum grep_context ctx, | |
585 | regmatch_t *pmatch, int eflags) | |
586 | { | |
587 | regmatch_t match; | |
588 | ||
589 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
590 | return 0; | |
591 | if (match.rm_so < 0 || match.rm_eo < 0) | |
592 | return 0; | |
593 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
594 | if (match.rm_so > pmatch->rm_so) | |
595 | return 1; | |
596 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
597 | return 1; | |
598 | } | |
599 | pmatch->rm_so = match.rm_so; | |
600 | pmatch->rm_eo = match.rm_eo; | |
601 | return 1; | |
602 | } | |
603 | ||
604 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
605 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
606 | { | |
607 | struct grep_pat *p; | |
608 | int hit = 0; | |
609 | ||
610 | pmatch->rm_so = pmatch->rm_eo = -1; | |
611 | if (bol < eol) { | |
612 | for (p = opt->pattern_list; p; p = p->next) { | |
613 | switch (p->token) { | |
614 | case GREP_PATTERN: /* atom */ | |
615 | case GREP_PATTERN_HEAD: | |
616 | case GREP_PATTERN_BODY: | |
617 | hit |= match_next_pattern(p, bol, eol, ctx, | |
618 | pmatch, eflags); | |
619 | break; | |
620 | default: | |
621 | break; | |
622 | } | |
623 | } | |
624 | } | |
625 | return hit; | |
626 | } | |
627 | ||
628 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
629 | const char *name, unsigned lno, char sign) | |
630 | { | |
631 | int rest = eol - bol; | |
00588bb5 | 632 | char *line_color = NULL; |
7e8f59d5 | 633 | |
ed24e401 | 634 | if (opt->pre_context || opt->post_context) { |
046802d0 | 635 | if (opt->last_shown == 0) { |
55f638bd ML |
636 | if (opt->show_hunk_mark) { |
637 | output_color(opt, "--", 2, opt->color_sep); | |
638 | opt->output(opt, "\n", 1); | |
07b838f0 | 639 | } |
55f638bd ML |
640 | } else if (lno > opt->last_shown + 1) { |
641 | output_color(opt, "--", 2, opt->color_sep); | |
642 | opt->output(opt, "\n", 1); | |
643 | } | |
5dd06d38 RS |
644 | } |
645 | opt->last_shown = lno; | |
646 | ||
5b594f45 | 647 | if (opt->pathname) { |
55f638bd ML |
648 | output_color(opt, name, strlen(name), opt->color_filename); |
649 | output_sep(opt, sign); | |
5b594f45 FK |
650 | } |
651 | if (opt->linenum) { | |
652 | char buf[32]; | |
653 | snprintf(buf, sizeof(buf), "%d", lno); | |
55f638bd ML |
654 | output_color(opt, buf, strlen(buf), opt->color_lineno); |
655 | output_sep(opt, sign); | |
5b594f45 | 656 | } |
7e8f59d5 RS |
657 | if (opt->color) { |
658 | regmatch_t match; | |
659 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
660 | int ch = *eol; | |
661 | int eflags = 0; | |
662 | ||
00588bb5 ML |
663 | if (sign == ':') |
664 | line_color = opt->color_selected; | |
665 | else if (sign == '-') | |
666 | line_color = opt->color_context; | |
667 | else if (sign == '=') | |
668 | line_color = opt->color_function; | |
7e8f59d5 RS |
669 | *eol = '\0'; |
670 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
671 | if (match.rm_so == match.rm_eo) |
672 | break; | |
5b594f45 | 673 | |
00588bb5 | 674 | output_color(opt, bol, match.rm_so, line_color); |
55f638bd ML |
675 | output_color(opt, bol + match.rm_so, |
676 | match.rm_eo - match.rm_so, | |
677 | opt->color_match); | |
7e8f59d5 RS |
678 | bol += match.rm_eo; |
679 | rest -= match.rm_eo; | |
680 | eflags = REG_NOTBOL; | |
681 | } | |
682 | *eol = ch; | |
683 | } | |
00588bb5 | 684 | output_color(opt, bol, rest, line_color); |
5b594f45 | 685 | opt->output(opt, "\n", 1); |
7e8f59d5 RS |
686 | } |
687 | ||
60ecac98 | 688 | static int match_funcname(struct grep_opt *opt, char *bol, char *eol) |
2944e4e6 | 689 | { |
60ecac98 RS |
690 | xdemitconf_t *xecfg = opt->priv; |
691 | if (xecfg && xecfg->find_func) { | |
692 | char buf[1]; | |
693 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
694 | xecfg->find_func_priv) >= 0; | |
695 | } | |
696 | ||
2944e4e6 RS |
697 | if (bol == eol) |
698 | return 0; | |
699 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
700 | return 1; | |
701 | return 0; | |
702 | } | |
703 | ||
704 | static void show_funcname_line(struct grep_opt *opt, const char *name, | |
705 | char *buf, char *bol, unsigned lno) | |
706 | { | |
707 | while (bol > buf) { | |
708 | char *eol = --bol; | |
709 | ||
710 | while (bol > buf && bol[-1] != '\n') | |
711 | bol--; | |
712 | lno--; | |
713 | ||
714 | if (lno <= opt->last_shown) | |
715 | break; | |
716 | ||
60ecac98 | 717 | if (match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
718 | show_line(opt, bol, eol, name, lno, '='); |
719 | break; | |
720 | } | |
721 | } | |
722 | } | |
723 | ||
49de3216 RS |
724 | static void show_pre_context(struct grep_opt *opt, const char *name, char *buf, |
725 | char *bol, unsigned lno) | |
726 | { | |
2944e4e6 RS |
727 | unsigned cur = lno, from = 1, funcname_lno = 0; |
728 | int funcname_needed = opt->funcname; | |
49de3216 RS |
729 | |
730 | if (opt->pre_context < lno) | |
731 | from = lno - opt->pre_context; | |
732 | if (from <= opt->last_shown) | |
733 | from = opt->last_shown + 1; | |
734 | ||
735 | /* Rewind. */ | |
736 | while (bol > buf && cur > from) { | |
2944e4e6 RS |
737 | char *eol = --bol; |
738 | ||
49de3216 RS |
739 | while (bol > buf && bol[-1] != '\n') |
740 | bol--; | |
741 | cur--; | |
60ecac98 | 742 | if (funcname_needed && match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
743 | funcname_lno = cur; |
744 | funcname_needed = 0; | |
745 | } | |
49de3216 RS |
746 | } |
747 | ||
2944e4e6 RS |
748 | /* We need to look even further back to find a function signature. */ |
749 | if (opt->funcname && funcname_needed) | |
750 | show_funcname_line(opt, name, buf, bol, cur); | |
751 | ||
49de3216 RS |
752 | /* Back forward. */ |
753 | while (cur < lno) { | |
2944e4e6 | 754 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; |
49de3216 RS |
755 | |
756 | while (*eol != '\n') | |
757 | eol++; | |
2944e4e6 | 758 | show_line(opt, bol, eol, name, cur, sign); |
49de3216 RS |
759 | bol = eol + 1; |
760 | cur++; | |
761 | } | |
762 | } | |
763 | ||
a26345b6 JH |
764 | static int should_lookahead(struct grep_opt *opt) |
765 | { | |
766 | struct grep_pat *p; | |
767 | ||
768 | if (opt->extended) | |
769 | return 0; /* punt for too complex stuff */ | |
770 | if (opt->invert) | |
771 | return 0; | |
772 | for (p = opt->pattern_list; p; p = p->next) { | |
773 | if (p->token != GREP_PATTERN) | |
774 | return 0; /* punt for "header only" and stuff */ | |
775 | } | |
776 | return 1; | |
777 | } | |
778 | ||
779 | static int look_ahead(struct grep_opt *opt, | |
780 | unsigned long *left_p, | |
781 | unsigned *lno_p, | |
782 | char **bol_p) | |
783 | { | |
784 | unsigned lno = *lno_p; | |
785 | char *bol = *bol_p; | |
786 | struct grep_pat *p; | |
787 | char *sp, *last_bol; | |
788 | regoff_t earliest = -1; | |
789 | ||
790 | for (p = opt->pattern_list; p; p = p->next) { | |
791 | int hit; | |
792 | regmatch_t m; | |
793 | ||
ed40a095 RS |
794 | if (p->fixed) |
795 | hit = !fixmatch(p, bol, bol + *left_p, &m); | |
796 | else | |
f96e5673 | 797 | hit = !regmatch(&p->regexp, bol, bol + *left_p, &m, 0); |
a26345b6 JH |
798 | if (!hit || m.rm_so < 0 || m.rm_eo < 0) |
799 | continue; | |
800 | if (earliest < 0 || m.rm_so < earliest) | |
801 | earliest = m.rm_so; | |
802 | } | |
803 | ||
804 | if (earliest < 0) { | |
805 | *bol_p = bol + *left_p; | |
806 | *left_p = 0; | |
807 | return 1; | |
808 | } | |
809 | for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--) | |
810 | ; /* find the beginning of the line */ | |
811 | last_bol = sp; | |
812 | ||
813 | for (sp = bol; sp < last_bol; sp++) { | |
814 | if (*sp == '\n') | |
815 | lno++; | |
816 | } | |
817 | *left_p -= last_bol - bol; | |
818 | *bol_p = last_bol; | |
819 | *lno_p = lno; | |
820 | return 0; | |
821 | } | |
822 | ||
5b594f45 FK |
823 | int grep_threads_ok(const struct grep_opt *opt) |
824 | { | |
825 | /* If this condition is true, then we may use the attribute | |
826 | * machinery in grep_buffer_1. The attribute code is not | |
827 | * thread safe, so we disable the use of threads. | |
828 | */ | |
829 | if (opt->funcname && !opt->unmatch_name_only && !opt->status_only && | |
830 | !opt->name_only) | |
831 | return 0; | |
832 | ||
5b594f45 FK |
833 | return 1; |
834 | } | |
835 | ||
836 | static void std_output(struct grep_opt *opt, const void *buf, size_t size) | |
837 | { | |
838 | fwrite(buf, size, 1, stdout); | |
839 | } | |
840 | ||
0ab7befa JH |
841 | static int grep_buffer_1(struct grep_opt *opt, const char *name, |
842 | char *buf, unsigned long size, int collect_hits) | |
83b5d2f5 JH |
843 | { |
844 | char *bol = buf; | |
845 | unsigned long left = size; | |
846 | unsigned lno = 1; | |
83b5d2f5 | 847 | unsigned last_hit = 0; |
83b5d2f5 | 848 | int binary_match_only = 0; |
83b5d2f5 | 849 | unsigned count = 0; |
a26345b6 | 850 | int try_lookahead = 0; |
480c1ca6 | 851 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
60ecac98 | 852 | xdemitconf_t xecfg; |
83b5d2f5 | 853 | |
5b594f45 FK |
854 | if (!opt->output) |
855 | opt->output = std_output; | |
856 | ||
431d6e7b RS |
857 | if (opt->last_shown && (opt->pre_context || opt->post_context) && |
858 | opt->output == std_output) | |
859 | opt->show_hunk_mark = 1; | |
860 | opt->last_shown = 0; | |
861 | ||
64fcec78 RS |
862 | switch (opt->binary) { |
863 | case GREP_BINARY_DEFAULT: | |
864 | if (buffer_is_binary(buf, size)) | |
83b5d2f5 | 865 | binary_match_only = 1; |
64fcec78 RS |
866 | break; |
867 | case GREP_BINARY_NOMATCH: | |
868 | if (buffer_is_binary(buf, size)) | |
83b5d2f5 | 869 | return 0; /* Assume unmatch */ |
64fcec78 RS |
870 | break; |
871 | case GREP_BINARY_TEXT: | |
872 | break; | |
873 | default: | |
874 | die("bug: unknown binary handling mode"); | |
83b5d2f5 JH |
875 | } |
876 | ||
60ecac98 RS |
877 | memset(&xecfg, 0, sizeof(xecfg)); |
878 | if (opt->funcname && !opt->unmatch_name_only && !opt->status_only && | |
879 | !opt->name_only && !binary_match_only && !collect_hits) { | |
880 | struct userdiff_driver *drv = userdiff_find_by_path(name); | |
881 | if (drv && drv->funcname.pattern) { | |
882 | const struct userdiff_funcname *pe = &drv->funcname; | |
883 | xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags); | |
884 | opt->priv = &xecfg; | |
885 | } | |
886 | } | |
a26345b6 | 887 | try_lookahead = should_lookahead(opt); |
60ecac98 | 888 | |
83b5d2f5 JH |
889 | while (left) { |
890 | char *eol, ch; | |
0ab7befa | 891 | int hit; |
83b5d2f5 | 892 | |
a26345b6 JH |
893 | /* |
894 | * look_ahead() skips quicly to the line that possibly | |
895 | * has the next hit; don't call it if we need to do | |
896 | * something more than just skipping the current line | |
897 | * in response to an unmatch for the current line. E.g. | |
898 | * inside a post-context window, we will show the current | |
899 | * line as a context around the previous hit when it | |
900 | * doesn't hit. | |
901 | */ | |
902 | if (try_lookahead | |
903 | && !(last_hit | |
904 | && lno <= last_hit + opt->post_context) | |
905 | && look_ahead(opt, &left, &lno, &bol)) | |
906 | break; | |
83b5d2f5 JH |
907 | eol = end_of_line(bol, &left); |
908 | ch = *eol; | |
909 | *eol = 0; | |
910 | ||
480c1ca6 JH |
911 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
912 | ctx = GREP_CONTEXT_BODY; | |
913 | ||
0ab7befa | 914 | hit = match_line(opt, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
915 | *eol = ch; |
916 | ||
0ab7befa JH |
917 | if (collect_hits) |
918 | goto next_line; | |
919 | ||
83b5d2f5 JH |
920 | /* "grep -v -e foo -e bla" should list lines |
921 | * that do not have either, so inversion should | |
922 | * be done outside. | |
923 | */ | |
924 | if (opt->invert) | |
925 | hit = !hit; | |
926 | if (opt->unmatch_name_only) { | |
927 | if (hit) | |
928 | return 0; | |
929 | goto next_line; | |
930 | } | |
931 | if (hit) { | |
932 | count++; | |
933 | if (opt->status_only) | |
934 | return 1; | |
321ffcc0 RS |
935 | if (opt->name_only) { |
936 | show_name(opt, name); | |
937 | return 1; | |
938 | } | |
c30c10cf RS |
939 | if (opt->count) |
940 | goto next_line; | |
83b5d2f5 | 941 | if (binary_match_only) { |
5b594f45 | 942 | opt->output(opt, "Binary file ", 12); |
55f638bd ML |
943 | output_color(opt, name, strlen(name), |
944 | opt->color_filename); | |
5b594f45 | 945 | opt->output(opt, " matches\n", 9); |
83b5d2f5 JH |
946 | return 1; |
947 | } | |
83b5d2f5 JH |
948 | /* Hit at this line. If we haven't shown the |
949 | * pre-context lines, we would need to show them. | |
83b5d2f5 | 950 | */ |
49de3216 RS |
951 | if (opt->pre_context) |
952 | show_pre_context(opt, name, buf, bol, lno); | |
2944e4e6 RS |
953 | else if (opt->funcname) |
954 | show_funcname_line(opt, name, buf, bol, lno); | |
c30c10cf | 955 | show_line(opt, bol, eol, name, lno, ':'); |
5dd06d38 | 956 | last_hit = lno; |
83b5d2f5 JH |
957 | } |
958 | else if (last_hit && | |
959 | lno <= last_hit + opt->post_context) { | |
960 | /* If the last hit is within the post context, | |
961 | * we need to show this line. | |
962 | */ | |
83b5d2f5 | 963 | show_line(opt, bol, eol, name, lno, '-'); |
83b5d2f5 | 964 | } |
83b5d2f5 JH |
965 | |
966 | next_line: | |
967 | bol = eol + 1; | |
968 | if (!left) | |
969 | break; | |
970 | left--; | |
971 | lno++; | |
972 | } | |
973 | ||
0ab7befa JH |
974 | if (collect_hits) |
975 | return 0; | |
b48fb5b6 | 976 | |
83b5d2f5 JH |
977 | if (opt->status_only) |
978 | return 0; | |
979 | if (opt->unmatch_name_only) { | |
980 | /* We did not see any hit, so we want to show this */ | |
83caecca | 981 | show_name(opt, name); |
83b5d2f5 JH |
982 | return 1; |
983 | } | |
984 | ||
60ecac98 RS |
985 | xdiff_clear_find_func(&xecfg); |
986 | opt->priv = NULL; | |
987 | ||
83b5d2f5 JH |
988 | /* NEEDSWORK: |
989 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
990 | * which feels mostly useless but sometimes useful. Maybe | |
991 | * make it another option? For now suppress them. | |
992 | */ | |
5b594f45 FK |
993 | if (opt->count && count) { |
994 | char buf[32]; | |
55f638bd ML |
995 | output_color(opt, name, strlen(name), opt->color_filename); |
996 | output_sep(opt, ':'); | |
997 | snprintf(buf, sizeof(buf), "%u\n", count); | |
5b594f45 | 998 | opt->output(opt, buf, strlen(buf)); |
c30c10cf | 999 | return 1; |
5b594f45 | 1000 | } |
83b5d2f5 JH |
1001 | return !!last_hit; |
1002 | } | |
1003 | ||
0ab7befa JH |
1004 | static void clr_hit_marker(struct grep_expr *x) |
1005 | { | |
1006 | /* All-hit markers are meaningful only at the very top level | |
1007 | * OR node. | |
1008 | */ | |
1009 | while (1) { | |
1010 | x->hit = 0; | |
1011 | if (x->node != GREP_NODE_OR) | |
1012 | return; | |
1013 | x->u.binary.left->hit = 0; | |
1014 | x = x->u.binary.right; | |
1015 | } | |
1016 | } | |
1017 | ||
1018 | static int chk_hit_marker(struct grep_expr *x) | |
1019 | { | |
1020 | /* Top level nodes have hit markers. See if they all are hits */ | |
1021 | while (1) { | |
1022 | if (x->node != GREP_NODE_OR) | |
1023 | return x->hit; | |
1024 | if (!x->u.binary.left->hit) | |
1025 | return 0; | |
1026 | x = x->u.binary.right; | |
1027 | } | |
1028 | } | |
1029 | ||
1030 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
1031 | { | |
1032 | /* | |
1033 | * we do not have to do the two-pass grep when we do not check | |
1034 | * buffer-wide "all-match". | |
1035 | */ | |
1036 | if (!opt->all_match) | |
1037 | return grep_buffer_1(opt, name, buf, size, 0); | |
1038 | ||
1039 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
1040 | * We first clear hit markers from them. | |
1041 | */ | |
1042 | clr_hit_marker(opt->pattern_expression); | |
1043 | grep_buffer_1(opt, name, buf, size, 1); | |
1044 | ||
1045 | if (!chk_hit_marker(opt->pattern_expression)) | |
1046 | return 0; | |
1047 | ||
1048 | return grep_buffer_1(opt, name, buf, size, 0); | |
1049 | } |