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