]>
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; | |
14 | *opt->pattern_tail = p; | |
15 | opt->pattern_tail = &p->next; | |
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 | ||
32 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) | |
33 | { | |
c822255c RS |
34 | int err; |
35 | ||
d7eb527d | 36 | p->word_regexp = opt->word_regexp; |
5183bf67 | 37 | p->ignore_case = opt->ignore_case; |
d7eb527d | 38 | |
885d211e | 39 | if (opt->fixed) |
c822255c RS |
40 | p->fixed = 1; |
41 | if (opt->regflags & REG_ICASE) | |
42 | p->fixed = 0; | |
43 | if (p->fixed) | |
44 | return; | |
45 | ||
46 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
83b5d2f5 JH |
47 | if (err) { |
48 | char errbuf[1024]; | |
49 | char where[1024]; | |
50 | if (p->no) | |
51 | sprintf(where, "In '%s' at %d, ", | |
52 | p->origin, p->no); | |
53 | else if (p->origin) | |
54 | sprintf(where, "%s, ", p->origin); | |
55 | else | |
56 | where[0] = 0; | |
57 | regerror(err, &p->regexp, errbuf, 1024); | |
58 | regfree(&p->regexp); | |
59 | die("%s'%s': %s", where, p->pattern, errbuf); | |
60 | } | |
61 | } | |
62 | ||
0ab7befa | 63 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
64 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
65 | { | |
66 | struct grep_pat *p; | |
67 | struct grep_expr *x; | |
68 | ||
69 | p = *list; | |
c922b01f LT |
70 | if (!p) |
71 | return NULL; | |
83b5d2f5 JH |
72 | switch (p->token) { |
73 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
74 | case GREP_PATTERN_HEAD: |
75 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
76 | x = xcalloc(1, sizeof (struct grep_expr)); |
77 | x->node = GREP_NODE_ATOM; | |
78 | x->u.atom = p; | |
79 | *list = p->next; | |
80 | return x; | |
81 | case GREP_OPEN_PAREN: | |
82 | *list = p->next; | |
0ab7befa | 83 | x = compile_pattern_or(list); |
83b5d2f5 JH |
84 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
85 | die("unmatched parenthesis"); | |
86 | *list = (*list)->next; | |
87 | return x; | |
88 | default: | |
89 | return NULL; | |
90 | } | |
91 | } | |
92 | ||
93 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
94 | { | |
95 | struct grep_pat *p; | |
96 | struct grep_expr *x; | |
97 | ||
98 | p = *list; | |
c922b01f LT |
99 | if (!p) |
100 | return NULL; | |
83b5d2f5 JH |
101 | switch (p->token) { |
102 | case GREP_NOT: | |
103 | if (!p->next) | |
104 | die("--not not followed by pattern expression"); | |
105 | *list = p->next; | |
106 | x = xcalloc(1, sizeof (struct grep_expr)); | |
107 | x->node = GREP_NODE_NOT; | |
108 | x->u.unary = compile_pattern_not(list); | |
109 | if (!x->u.unary) | |
110 | die("--not followed by non pattern expression"); | |
111 | return x; | |
112 | default: | |
113 | return compile_pattern_atom(list); | |
114 | } | |
115 | } | |
116 | ||
117 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
118 | { | |
119 | struct grep_pat *p; | |
120 | struct grep_expr *x, *y, *z; | |
121 | ||
122 | x = compile_pattern_not(list); | |
123 | p = *list; | |
124 | if (p && p->token == GREP_AND) { | |
125 | if (!p->next) | |
126 | die("--and not followed by pattern expression"); | |
127 | *list = p->next; | |
128 | y = compile_pattern_and(list); | |
129 | if (!y) | |
130 | die("--and not followed by pattern expression"); | |
131 | z = xcalloc(1, sizeof (struct grep_expr)); | |
132 | z->node = GREP_NODE_AND; | |
133 | z->u.binary.left = x; | |
134 | z->u.binary.right = y; | |
135 | return z; | |
136 | } | |
137 | return x; | |
138 | } | |
139 | ||
140 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
141 | { | |
142 | struct grep_pat *p; | |
143 | struct grep_expr *x, *y, *z; | |
144 | ||
145 | x = compile_pattern_and(list); | |
146 | p = *list; | |
147 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
148 | y = compile_pattern_or(list); | |
149 | if (!y) | |
150 | die("not a pattern expression %s", p->pattern); | |
151 | z = xcalloc(1, sizeof (struct grep_expr)); | |
152 | z->node = GREP_NODE_OR; | |
153 | z->u.binary.left = x; | |
154 | z->u.binary.right = y; | |
155 | return z; | |
156 | } | |
157 | return x; | |
158 | } | |
159 | ||
160 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
161 | { | |
162 | return compile_pattern_or(list); | |
163 | } | |
164 | ||
165 | void compile_grep_patterns(struct grep_opt *opt) | |
166 | { | |
167 | struct grep_pat *p; | |
168 | ||
0ab7befa JH |
169 | if (opt->all_match) |
170 | opt->extended = 1; | |
171 | ||
83b5d2f5 | 172 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
173 | switch (p->token) { |
174 | case GREP_PATTERN: /* atom */ | |
175 | case GREP_PATTERN_HEAD: | |
176 | case GREP_PATTERN_BODY: | |
c822255c | 177 | compile_regexp(p, opt); |
480c1ca6 JH |
178 | break; |
179 | default: | |
83b5d2f5 | 180 | opt->extended = 1; |
480c1ca6 JH |
181 | break; |
182 | } | |
83b5d2f5 JH |
183 | } |
184 | ||
185 | if (!opt->extended) | |
186 | return; | |
187 | ||
188 | /* Then bundle them up in an expression. | |
189 | * A classic recursive descent parser would do. | |
190 | */ | |
191 | p = opt->pattern_list; | |
ba150a3f MB |
192 | if (p) |
193 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
194 | if (p) |
195 | die("incomplete pattern expression: %s", p->pattern); | |
196 | } | |
197 | ||
b48fb5b6 JH |
198 | static void free_pattern_expr(struct grep_expr *x) |
199 | { | |
200 | switch (x->node) { | |
201 | case GREP_NODE_ATOM: | |
202 | break; | |
203 | case GREP_NODE_NOT: | |
204 | free_pattern_expr(x->u.unary); | |
205 | break; | |
206 | case GREP_NODE_AND: | |
207 | case GREP_NODE_OR: | |
208 | free_pattern_expr(x->u.binary.left); | |
209 | free_pattern_expr(x->u.binary.right); | |
210 | break; | |
211 | } | |
212 | free(x); | |
213 | } | |
214 | ||
215 | void free_grep_patterns(struct grep_opt *opt) | |
216 | { | |
217 | struct grep_pat *p, *n; | |
218 | ||
219 | for (p = opt->pattern_list; p; p = n) { | |
220 | n = p->next; | |
221 | switch (p->token) { | |
222 | case GREP_PATTERN: /* atom */ | |
223 | case GREP_PATTERN_HEAD: | |
224 | case GREP_PATTERN_BODY: | |
225 | regfree(&p->regexp); | |
226 | break; | |
227 | default: | |
228 | break; | |
229 | } | |
230 | free(p); | |
231 | } | |
232 | ||
233 | if (!opt->extended) | |
234 | return; | |
235 | free_pattern_expr(opt->pattern_expression); | |
236 | } | |
237 | ||
83b5d2f5 JH |
238 | static char *end_of_line(char *cp, unsigned long *left) |
239 | { | |
240 | unsigned long l = *left; | |
241 | while (l && *cp != '\n') { | |
242 | l--; | |
243 | cp++; | |
244 | } | |
245 | *left = l; | |
246 | return cp; | |
247 | } | |
248 | ||
249 | static int word_char(char ch) | |
250 | { | |
251 | return isalnum(ch) || ch == '_'; | |
252 | } | |
253 | ||
83caecca RZ |
254 | static void show_name(struct grep_opt *opt, const char *name) |
255 | { | |
256 | printf("%s%c", name, opt->null_following_name ? '\0' : '\n'); | |
257 | } | |
258 | ||
5183bf67 BC |
259 | |
260 | static int fixmatch(const char *pattern, char *line, int ignore_case, regmatch_t *match) | |
83b5d2f5 | 261 | { |
5183bf67 BC |
262 | char *hit; |
263 | if (ignore_case) | |
264 | hit = strcasestr(line, pattern); | |
265 | else | |
266 | hit = strstr(line, pattern); | |
267 | ||
83b5d2f5 JH |
268 | if (!hit) { |
269 | match->rm_so = match->rm_eo = -1; | |
270 | return REG_NOMATCH; | |
271 | } | |
272 | else { | |
273 | match->rm_so = hit - line; | |
274 | match->rm_eo = match->rm_so + strlen(pattern); | |
275 | return 0; | |
276 | } | |
277 | } | |
278 | ||
a4d7d2c6 JH |
279 | static int strip_timestamp(char *bol, char **eol_p) |
280 | { | |
281 | char *eol = *eol_p; | |
282 | int ch; | |
283 | ||
284 | while (bol < --eol) { | |
285 | if (*eol != '>') | |
286 | continue; | |
287 | *eol_p = ++eol; | |
288 | ch = *eol; | |
289 | *eol = '\0'; | |
290 | return ch; | |
291 | } | |
292 | return 0; | |
293 | } | |
294 | ||
295 | static struct { | |
296 | const char *field; | |
297 | size_t len; | |
298 | } header_field[] = { | |
299 | { "author ", 7 }, | |
300 | { "committer ", 10 }, | |
301 | }; | |
302 | ||
d7eb527d | 303 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
304 | enum grep_context ctx, |
305 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
306 | { |
307 | int hit = 0; | |
a4d7d2c6 | 308 | int saved_ch = 0; |
e701fadb | 309 | const char *start = bol; |
83b5d2f5 | 310 | |
480c1ca6 JH |
311 | if ((p->token != GREP_PATTERN) && |
312 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
313 | return 0; | |
314 | ||
a4d7d2c6 JH |
315 | if (p->token == GREP_PATTERN_HEAD) { |
316 | const char *field; | |
317 | size_t len; | |
318 | assert(p->field < ARRAY_SIZE(header_field)); | |
319 | field = header_field[p->field].field; | |
320 | len = header_field[p->field].len; | |
321 | if (strncmp(bol, field, len)) | |
322 | return 0; | |
323 | bol += len; | |
324 | saved_ch = strip_timestamp(bol, &eol); | |
325 | } | |
326 | ||
83b5d2f5 | 327 | again: |
79212772 | 328 | if (p->fixed) |
5183bf67 | 329 | hit = !fixmatch(p->pattern, bol, p->ignore_case, pmatch); |
79212772 RS |
330 | else |
331 | hit = !regexec(&p->regexp, bol, 1, pmatch, eflags); | |
83b5d2f5 | 332 | |
d7eb527d | 333 | if (hit && p->word_regexp) { |
83b5d2f5 | 334 | if ((pmatch[0].rm_so < 0) || |
84201eae | 335 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
336 | (pmatch[0].rm_eo < 0) || |
337 | (eol - bol) < pmatch[0].rm_eo) | |
338 | die("regexp returned nonsense"); | |
339 | ||
340 | /* Match beginning must be either beginning of the | |
341 | * line, or at word boundary (i.e. the last char must | |
342 | * not be a word char). Similarly, match end must be | |
343 | * either end of the line, or at word boundary | |
344 | * (i.e. the next char must not be a word char). | |
345 | */ | |
fb62eb7f | 346 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
347 | !word_char(bol[pmatch[0].rm_so-1])) && |
348 | ((pmatch[0].rm_eo == (eol-bol)) || | |
349 | !word_char(bol[pmatch[0].rm_eo])) ) | |
350 | ; | |
351 | else | |
352 | hit = 0; | |
353 | ||
84201eae RS |
354 | /* Words consist of at least one character. */ |
355 | if (pmatch->rm_so == pmatch->rm_eo) | |
356 | hit = 0; | |
357 | ||
83b5d2f5 JH |
358 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
359 | /* There could be more than one match on the | |
360 | * line, and the first match might not be | |
361 | * strict word match. But later ones could be! | |
fb62eb7f RS |
362 | * Forward to the next possible start, i.e. the |
363 | * next position following a non-word char. | |
83b5d2f5 JH |
364 | */ |
365 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
366 | while (word_char(bol[-1]) && bol < eol) |
367 | bol++; | |
dbb6a4ad | 368 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
369 | if (bol < eol) |
370 | goto again; | |
83b5d2f5 JH |
371 | } |
372 | } | |
a4d7d2c6 JH |
373 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
374 | *eol = saved_ch; | |
e701fadb RS |
375 | if (hit) { |
376 | pmatch[0].rm_so += bol - start; | |
377 | pmatch[0].rm_eo += bol - start; | |
378 | } | |
83b5d2f5 JH |
379 | return hit; |
380 | } | |
381 | ||
d7eb527d RS |
382 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, |
383 | enum grep_context ctx, int collect_hits) | |
83b5d2f5 | 384 | { |
0ab7befa | 385 | int h = 0; |
79212772 | 386 | regmatch_t match; |
0ab7befa | 387 | |
c922b01f LT |
388 | if (!x) |
389 | die("Not a valid grep expression"); | |
83b5d2f5 JH |
390 | switch (x->node) { |
391 | case GREP_NODE_ATOM: | |
79212772 | 392 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); |
83b5d2f5 JH |
393 | break; |
394 | case GREP_NODE_NOT: | |
d7eb527d | 395 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); |
0ab7befa | 396 | break; |
83b5d2f5 | 397 | case GREP_NODE_AND: |
d7eb527d | 398 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) |
252d560d | 399 | return 0; |
d7eb527d | 400 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); |
0ab7befa | 401 | break; |
83b5d2f5 | 402 | case GREP_NODE_OR: |
0ab7befa | 403 | if (!collect_hits) |
d7eb527d | 404 | return (match_expr_eval(x->u.binary.left, |
0ab7befa | 405 | bol, eol, ctx, 0) || |
d7eb527d | 406 | match_expr_eval(x->u.binary.right, |
0ab7befa | 407 | bol, eol, ctx, 0)); |
d7eb527d | 408 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); |
0ab7befa | 409 | x->u.binary.left->hit |= h; |
d7eb527d | 410 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); |
0ab7befa JH |
411 | break; |
412 | default: | |
d7530708 | 413 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 414 | } |
0ab7befa JH |
415 | if (collect_hits) |
416 | x->hit |= h; | |
417 | return h; | |
83b5d2f5 JH |
418 | } |
419 | ||
480c1ca6 | 420 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 421 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
422 | { |
423 | struct grep_expr *x = opt->pattern_expression; | |
d7eb527d | 424 | return match_expr_eval(x, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
425 | } |
426 | ||
480c1ca6 | 427 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 428 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
429 | { |
430 | struct grep_pat *p; | |
79212772 RS |
431 | regmatch_t match; |
432 | ||
83b5d2f5 | 433 | if (opt->extended) |
0ab7befa JH |
434 | return match_expr(opt, bol, eol, ctx, collect_hits); |
435 | ||
436 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 437 | for (p = opt->pattern_list; p; p = p->next) { |
79212772 | 438 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) |
83b5d2f5 JH |
439 | return 1; |
440 | } | |
441 | return 0; | |
442 | } | |
443 | ||
7e8f59d5 RS |
444 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
445 | enum grep_context ctx, | |
446 | regmatch_t *pmatch, int eflags) | |
447 | { | |
448 | regmatch_t match; | |
449 | ||
450 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
451 | return 0; | |
452 | if (match.rm_so < 0 || match.rm_eo < 0) | |
453 | return 0; | |
454 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
455 | if (match.rm_so > pmatch->rm_so) | |
456 | return 1; | |
457 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
458 | return 1; | |
459 | } | |
460 | pmatch->rm_so = match.rm_so; | |
461 | pmatch->rm_eo = match.rm_eo; | |
462 | return 1; | |
463 | } | |
464 | ||
465 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
466 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
467 | { | |
468 | struct grep_pat *p; | |
469 | int hit = 0; | |
470 | ||
471 | pmatch->rm_so = pmatch->rm_eo = -1; | |
472 | if (bol < eol) { | |
473 | for (p = opt->pattern_list; p; p = p->next) { | |
474 | switch (p->token) { | |
475 | case GREP_PATTERN: /* atom */ | |
476 | case GREP_PATTERN_HEAD: | |
477 | case GREP_PATTERN_BODY: | |
478 | hit |= match_next_pattern(p, bol, eol, ctx, | |
479 | pmatch, eflags); | |
480 | break; | |
481 | default: | |
482 | break; | |
483 | } | |
484 | } | |
485 | } | |
486 | return hit; | |
487 | } | |
488 | ||
489 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
490 | const char *name, unsigned lno, char sign) | |
491 | { | |
492 | int rest = eol - bol; | |
493 | ||
ed24e401 | 494 | if (opt->pre_context || opt->post_context) { |
046802d0 RS |
495 | if (opt->last_shown == 0) { |
496 | if (opt->show_hunk_mark) | |
ed24e401 | 497 | fputs("--\n", stdout); |
046802d0 RS |
498 | else |
499 | opt->show_hunk_mark = 1; | |
ed24e401 RS |
500 | } else if (lno > opt->last_shown + 1) |
501 | fputs("--\n", stdout); | |
5dd06d38 RS |
502 | } |
503 | opt->last_shown = lno; | |
504 | ||
7e8f59d5 RS |
505 | if (opt->null_following_name) |
506 | sign = '\0'; | |
507 | if (opt->pathname) | |
508 | printf("%s%c", name, sign); | |
509 | if (opt->linenum) | |
510 | printf("%d%c", lno, sign); | |
511 | if (opt->color) { | |
512 | regmatch_t match; | |
513 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
514 | int ch = *eol; | |
515 | int eflags = 0; | |
516 | ||
517 | *eol = '\0'; | |
518 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
519 | if (match.rm_so == match.rm_eo) |
520 | break; | |
7e8f59d5 | 521 | printf("%.*s%s%.*s%s", |
747a322b | 522 | (int)match.rm_so, bol, |
7e8f59d5 | 523 | opt->color_match, |
747a322b | 524 | (int)(match.rm_eo - match.rm_so), bol + match.rm_so, |
7e8f59d5 RS |
525 | GIT_COLOR_RESET); |
526 | bol += match.rm_eo; | |
527 | rest -= match.rm_eo; | |
528 | eflags = REG_NOTBOL; | |
529 | } | |
530 | *eol = ch; | |
531 | } | |
532 | printf("%.*s\n", rest, bol); | |
533 | } | |
534 | ||
60ecac98 | 535 | static int match_funcname(struct grep_opt *opt, char *bol, char *eol) |
2944e4e6 | 536 | { |
60ecac98 RS |
537 | xdemitconf_t *xecfg = opt->priv; |
538 | if (xecfg && xecfg->find_func) { | |
539 | char buf[1]; | |
540 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
541 | xecfg->find_func_priv) >= 0; | |
542 | } | |
543 | ||
2944e4e6 RS |
544 | if (bol == eol) |
545 | return 0; | |
546 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
547 | return 1; | |
548 | return 0; | |
549 | } | |
550 | ||
551 | static void show_funcname_line(struct grep_opt *opt, const char *name, | |
552 | char *buf, char *bol, unsigned lno) | |
553 | { | |
554 | while (bol > buf) { | |
555 | char *eol = --bol; | |
556 | ||
557 | while (bol > buf && bol[-1] != '\n') | |
558 | bol--; | |
559 | lno--; | |
560 | ||
561 | if (lno <= opt->last_shown) | |
562 | break; | |
563 | ||
60ecac98 | 564 | if (match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
565 | show_line(opt, bol, eol, name, lno, '='); |
566 | break; | |
567 | } | |
568 | } | |
569 | } | |
570 | ||
49de3216 RS |
571 | static void show_pre_context(struct grep_opt *opt, const char *name, char *buf, |
572 | char *bol, unsigned lno) | |
573 | { | |
2944e4e6 RS |
574 | unsigned cur = lno, from = 1, funcname_lno = 0; |
575 | int funcname_needed = opt->funcname; | |
49de3216 RS |
576 | |
577 | if (opt->pre_context < lno) | |
578 | from = lno - opt->pre_context; | |
579 | if (from <= opt->last_shown) | |
580 | from = opt->last_shown + 1; | |
581 | ||
582 | /* Rewind. */ | |
583 | while (bol > buf && cur > from) { | |
2944e4e6 RS |
584 | char *eol = --bol; |
585 | ||
49de3216 RS |
586 | while (bol > buf && bol[-1] != '\n') |
587 | bol--; | |
588 | cur--; | |
60ecac98 | 589 | if (funcname_needed && match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
590 | funcname_lno = cur; |
591 | funcname_needed = 0; | |
592 | } | |
49de3216 RS |
593 | } |
594 | ||
2944e4e6 RS |
595 | /* We need to look even further back to find a function signature. */ |
596 | if (opt->funcname && funcname_needed) | |
597 | show_funcname_line(opt, name, buf, bol, cur); | |
598 | ||
49de3216 RS |
599 | /* Back forward. */ |
600 | while (cur < lno) { | |
2944e4e6 | 601 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; |
49de3216 RS |
602 | |
603 | while (*eol != '\n') | |
604 | eol++; | |
2944e4e6 | 605 | show_line(opt, bol, eol, name, cur, sign); |
49de3216 RS |
606 | bol = eol + 1; |
607 | cur++; | |
608 | } | |
609 | } | |
610 | ||
a26345b6 JH |
611 | static int should_lookahead(struct grep_opt *opt) |
612 | { | |
613 | struct grep_pat *p; | |
614 | ||
615 | if (opt->extended) | |
616 | return 0; /* punt for too complex stuff */ | |
617 | if (opt->invert) | |
618 | return 0; | |
619 | for (p = opt->pattern_list; p; p = p->next) { | |
620 | if (p->token != GREP_PATTERN) | |
621 | return 0; /* punt for "header only" and stuff */ | |
622 | } | |
623 | return 1; | |
624 | } | |
625 | ||
626 | static int look_ahead(struct grep_opt *opt, | |
627 | unsigned long *left_p, | |
628 | unsigned *lno_p, | |
629 | char **bol_p) | |
630 | { | |
631 | unsigned lno = *lno_p; | |
632 | char *bol = *bol_p; | |
633 | struct grep_pat *p; | |
634 | char *sp, *last_bol; | |
635 | regoff_t earliest = -1; | |
636 | ||
637 | for (p = opt->pattern_list; p; p = p->next) { | |
638 | int hit; | |
639 | regmatch_t m; | |
640 | ||
641 | if (p->fixed) | |
e2d2e383 | 642 | hit = !fixmatch(p->pattern, bol, p->ignore_case, &m); |
a26345b6 JH |
643 | else |
644 | hit = !regexec(&p->regexp, bol, 1, &m, 0); | |
645 | if (!hit || m.rm_so < 0 || m.rm_eo < 0) | |
646 | continue; | |
647 | if (earliest < 0 || m.rm_so < earliest) | |
648 | earliest = m.rm_so; | |
649 | } | |
650 | ||
651 | if (earliest < 0) { | |
652 | *bol_p = bol + *left_p; | |
653 | *left_p = 0; | |
654 | return 1; | |
655 | } | |
656 | for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--) | |
657 | ; /* find the beginning of the line */ | |
658 | last_bol = sp; | |
659 | ||
660 | for (sp = bol; sp < last_bol; sp++) { | |
661 | if (*sp == '\n') | |
662 | lno++; | |
663 | } | |
664 | *left_p -= last_bol - bol; | |
665 | *bol_p = last_bol; | |
666 | *lno_p = lno; | |
667 | return 0; | |
668 | } | |
669 | ||
0ab7befa JH |
670 | static int grep_buffer_1(struct grep_opt *opt, const char *name, |
671 | char *buf, unsigned long size, int collect_hits) | |
83b5d2f5 JH |
672 | { |
673 | char *bol = buf; | |
674 | unsigned long left = size; | |
675 | unsigned lno = 1; | |
83b5d2f5 | 676 | unsigned last_hit = 0; |
83b5d2f5 | 677 | int binary_match_only = 0; |
83b5d2f5 | 678 | unsigned count = 0; |
a26345b6 | 679 | int try_lookahead = 0; |
480c1ca6 | 680 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
60ecac98 | 681 | xdemitconf_t xecfg; |
83b5d2f5 | 682 | |
5dd06d38 RS |
683 | opt->last_shown = 0; |
684 | ||
83b5d2f5 JH |
685 | if (buffer_is_binary(buf, size)) { |
686 | switch (opt->binary) { | |
687 | case GREP_BINARY_DEFAULT: | |
688 | binary_match_only = 1; | |
689 | break; | |
690 | case GREP_BINARY_NOMATCH: | |
691 | return 0; /* Assume unmatch */ | |
692 | break; | |
693 | default: | |
694 | break; | |
695 | } | |
696 | } | |
697 | ||
60ecac98 RS |
698 | memset(&xecfg, 0, sizeof(xecfg)); |
699 | if (opt->funcname && !opt->unmatch_name_only && !opt->status_only && | |
700 | !opt->name_only && !binary_match_only && !collect_hits) { | |
701 | struct userdiff_driver *drv = userdiff_find_by_path(name); | |
702 | if (drv && drv->funcname.pattern) { | |
703 | const struct userdiff_funcname *pe = &drv->funcname; | |
704 | xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags); | |
705 | opt->priv = &xecfg; | |
706 | } | |
707 | } | |
a26345b6 | 708 | try_lookahead = should_lookahead(opt); |
60ecac98 | 709 | |
83b5d2f5 JH |
710 | while (left) { |
711 | char *eol, ch; | |
0ab7befa | 712 | int hit; |
83b5d2f5 | 713 | |
a26345b6 JH |
714 | /* |
715 | * look_ahead() skips quicly to the line that possibly | |
716 | * has the next hit; don't call it if we need to do | |
717 | * something more than just skipping the current line | |
718 | * in response to an unmatch for the current line. E.g. | |
719 | * inside a post-context window, we will show the current | |
720 | * line as a context around the previous hit when it | |
721 | * doesn't hit. | |
722 | */ | |
723 | if (try_lookahead | |
724 | && !(last_hit | |
725 | && lno <= last_hit + opt->post_context) | |
726 | && look_ahead(opt, &left, &lno, &bol)) | |
727 | break; | |
83b5d2f5 JH |
728 | eol = end_of_line(bol, &left); |
729 | ch = *eol; | |
730 | *eol = 0; | |
731 | ||
480c1ca6 JH |
732 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
733 | ctx = GREP_CONTEXT_BODY; | |
734 | ||
0ab7befa | 735 | hit = match_line(opt, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
736 | *eol = ch; |
737 | ||
0ab7befa JH |
738 | if (collect_hits) |
739 | goto next_line; | |
740 | ||
83b5d2f5 JH |
741 | /* "grep -v -e foo -e bla" should list lines |
742 | * that do not have either, so inversion should | |
743 | * be done outside. | |
744 | */ | |
745 | if (opt->invert) | |
746 | hit = !hit; | |
747 | if (opt->unmatch_name_only) { | |
748 | if (hit) | |
749 | return 0; | |
750 | goto next_line; | |
751 | } | |
752 | if (hit) { | |
753 | count++; | |
754 | if (opt->status_only) | |
755 | return 1; | |
756 | if (binary_match_only) { | |
757 | printf("Binary file %s matches\n", name); | |
758 | return 1; | |
759 | } | |
760 | if (opt->name_only) { | |
83caecca | 761 | show_name(opt, name); |
83b5d2f5 JH |
762 | return 1; |
763 | } | |
764 | /* Hit at this line. If we haven't shown the | |
765 | * pre-context lines, we would need to show them. | |
766 | * When asked to do "count", this still show | |
767 | * the context which is nonsense, but the user | |
768 | * deserves to get that ;-). | |
769 | */ | |
49de3216 RS |
770 | if (opt->pre_context) |
771 | show_pre_context(opt, name, buf, bol, lno); | |
2944e4e6 RS |
772 | else if (opt->funcname) |
773 | show_funcname_line(opt, name, buf, bol, lno); | |
83b5d2f5 JH |
774 | if (!opt->count) |
775 | show_line(opt, bol, eol, name, lno, ':'); | |
5dd06d38 | 776 | last_hit = lno; |
83b5d2f5 JH |
777 | } |
778 | else if (last_hit && | |
779 | lno <= last_hit + opt->post_context) { | |
780 | /* If the last hit is within the post context, | |
781 | * we need to show this line. | |
782 | */ | |
83b5d2f5 | 783 | show_line(opt, bol, eol, name, lno, '-'); |
83b5d2f5 | 784 | } |
83b5d2f5 JH |
785 | |
786 | next_line: | |
787 | bol = eol + 1; | |
788 | if (!left) | |
789 | break; | |
790 | left--; | |
791 | lno++; | |
792 | } | |
793 | ||
0ab7befa JH |
794 | if (collect_hits) |
795 | return 0; | |
b48fb5b6 | 796 | |
83b5d2f5 JH |
797 | if (opt->status_only) |
798 | return 0; | |
799 | if (opt->unmatch_name_only) { | |
800 | /* We did not see any hit, so we want to show this */ | |
83caecca | 801 | show_name(opt, name); |
83b5d2f5 JH |
802 | return 1; |
803 | } | |
804 | ||
60ecac98 RS |
805 | xdiff_clear_find_func(&xecfg); |
806 | opt->priv = NULL; | |
807 | ||
83b5d2f5 JH |
808 | /* NEEDSWORK: |
809 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
810 | * which feels mostly useless but sometimes useful. Maybe | |
811 | * make it another option? For now suppress them. | |
812 | */ | |
813 | if (opt->count && count) | |
83caecca RZ |
814 | printf("%s%c%u\n", name, |
815 | opt->null_following_name ? '\0' : ':', count); | |
83b5d2f5 JH |
816 | return !!last_hit; |
817 | } | |
818 | ||
0ab7befa JH |
819 | static void clr_hit_marker(struct grep_expr *x) |
820 | { | |
821 | /* All-hit markers are meaningful only at the very top level | |
822 | * OR node. | |
823 | */ | |
824 | while (1) { | |
825 | x->hit = 0; | |
826 | if (x->node != GREP_NODE_OR) | |
827 | return; | |
828 | x->u.binary.left->hit = 0; | |
829 | x = x->u.binary.right; | |
830 | } | |
831 | } | |
832 | ||
833 | static int chk_hit_marker(struct grep_expr *x) | |
834 | { | |
835 | /* Top level nodes have hit markers. See if they all are hits */ | |
836 | while (1) { | |
837 | if (x->node != GREP_NODE_OR) | |
838 | return x->hit; | |
839 | if (!x->u.binary.left->hit) | |
840 | return 0; | |
841 | x = x->u.binary.right; | |
842 | } | |
843 | } | |
844 | ||
845 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
846 | { | |
847 | /* | |
848 | * we do not have to do the two-pass grep when we do not check | |
849 | * buffer-wide "all-match". | |
850 | */ | |
851 | if (!opt->all_match) | |
852 | return grep_buffer_1(opt, name, buf, size, 0); | |
853 | ||
854 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
855 | * We first clear hit markers from them. | |
856 | */ | |
857 | clr_hit_marker(opt->pattern_expression); | |
858 | grep_buffer_1(opt, name, buf, size, 1); | |
859 | ||
860 | if (!chk_hit_marker(opt->pattern_expression)) | |
861 | return 0; | |
862 | ||
863 | return grep_buffer_1(opt, name, buf, size, 0); | |
864 | } |