]>
Commit | Line | Data |
---|---|---|
83b5d2f5 | 1 | #include "cache.h" |
83b5d2f5 | 2 | #include "grep.h" |
6bfce93e | 3 | #include "xdiff-interface.h" |
83b5d2f5 | 4 | |
a4d7d2c6 JH |
5 | void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat) |
6 | { | |
7 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
8 | p->pattern = pat; | |
9 | p->origin = "header"; | |
10 | p->no = 0; | |
11 | p->token = GREP_PATTERN_HEAD; | |
12 | p->field = field; | |
13 | *opt->pattern_tail = p; | |
14 | opt->pattern_tail = &p->next; | |
15 | p->next = NULL; | |
16 | } | |
17 | ||
83b5d2f5 JH |
18 | void append_grep_pattern(struct grep_opt *opt, const char *pat, |
19 | const char *origin, int no, enum grep_pat_token t) | |
20 | { | |
21 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
22 | p->pattern = pat; | |
23 | p->origin = origin; | |
24 | p->no = no; | |
25 | p->token = t; | |
26 | *opt->pattern_tail = p; | |
27 | opt->pattern_tail = &p->next; | |
28 | p->next = NULL; | |
29 | } | |
30 | ||
c822255c RS |
31 | static int is_fixed(const char *s) |
32 | { | |
f9b7cce6 | 33 | while (*s && !is_regex_special(*s)) |
c822255c RS |
34 | s++; |
35 | return !*s; | |
36 | } | |
37 | ||
83b5d2f5 JH |
38 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) |
39 | { | |
c822255c RS |
40 | int err; |
41 | ||
d7eb527d RS |
42 | p->word_regexp = opt->word_regexp; |
43 | ||
c822255c RS |
44 | if (opt->fixed || is_fixed(p->pattern)) |
45 | p->fixed = 1; | |
46 | if (opt->regflags & REG_ICASE) | |
47 | p->fixed = 0; | |
48 | if (p->fixed) | |
49 | return; | |
50 | ||
51 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
83b5d2f5 JH |
52 | if (err) { |
53 | char errbuf[1024]; | |
54 | char where[1024]; | |
55 | if (p->no) | |
56 | sprintf(where, "In '%s' at %d, ", | |
57 | p->origin, p->no); | |
58 | else if (p->origin) | |
59 | sprintf(where, "%s, ", p->origin); | |
60 | else | |
61 | where[0] = 0; | |
62 | regerror(err, &p->regexp, errbuf, 1024); | |
63 | regfree(&p->regexp); | |
64 | die("%s'%s': %s", where, p->pattern, errbuf); | |
65 | } | |
66 | } | |
67 | ||
0ab7befa | 68 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
69 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
70 | { | |
71 | struct grep_pat *p; | |
72 | struct grep_expr *x; | |
73 | ||
74 | p = *list; | |
c922b01f LT |
75 | if (!p) |
76 | return NULL; | |
83b5d2f5 JH |
77 | switch (p->token) { |
78 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
79 | case GREP_PATTERN_HEAD: |
80 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
81 | x = xcalloc(1, sizeof (struct grep_expr)); |
82 | x->node = GREP_NODE_ATOM; | |
83 | x->u.atom = p; | |
84 | *list = p->next; | |
85 | return x; | |
86 | case GREP_OPEN_PAREN: | |
87 | *list = p->next; | |
0ab7befa | 88 | x = compile_pattern_or(list); |
83b5d2f5 JH |
89 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
90 | die("unmatched parenthesis"); | |
91 | *list = (*list)->next; | |
92 | return x; | |
93 | default: | |
94 | return NULL; | |
95 | } | |
96 | } | |
97 | ||
98 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
99 | { | |
100 | struct grep_pat *p; | |
101 | struct grep_expr *x; | |
102 | ||
103 | p = *list; | |
c922b01f LT |
104 | if (!p) |
105 | return NULL; | |
83b5d2f5 JH |
106 | switch (p->token) { |
107 | case GREP_NOT: | |
108 | if (!p->next) | |
109 | die("--not not followed by pattern expression"); | |
110 | *list = p->next; | |
111 | x = xcalloc(1, sizeof (struct grep_expr)); | |
112 | x->node = GREP_NODE_NOT; | |
113 | x->u.unary = compile_pattern_not(list); | |
114 | if (!x->u.unary) | |
115 | die("--not followed by non pattern expression"); | |
116 | return x; | |
117 | default: | |
118 | return compile_pattern_atom(list); | |
119 | } | |
120 | } | |
121 | ||
122 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
123 | { | |
124 | struct grep_pat *p; | |
125 | struct grep_expr *x, *y, *z; | |
126 | ||
127 | x = compile_pattern_not(list); | |
128 | p = *list; | |
129 | if (p && p->token == GREP_AND) { | |
130 | if (!p->next) | |
131 | die("--and not followed by pattern expression"); | |
132 | *list = p->next; | |
133 | y = compile_pattern_and(list); | |
134 | if (!y) | |
135 | die("--and not followed by pattern expression"); | |
136 | z = xcalloc(1, sizeof (struct grep_expr)); | |
137 | z->node = GREP_NODE_AND; | |
138 | z->u.binary.left = x; | |
139 | z->u.binary.right = y; | |
140 | return z; | |
141 | } | |
142 | return x; | |
143 | } | |
144 | ||
145 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
146 | { | |
147 | struct grep_pat *p; | |
148 | struct grep_expr *x, *y, *z; | |
149 | ||
150 | x = compile_pattern_and(list); | |
151 | p = *list; | |
152 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
153 | y = compile_pattern_or(list); | |
154 | if (!y) | |
155 | die("not a pattern expression %s", p->pattern); | |
156 | z = xcalloc(1, sizeof (struct grep_expr)); | |
157 | z->node = GREP_NODE_OR; | |
158 | z->u.binary.left = x; | |
159 | z->u.binary.right = y; | |
160 | return z; | |
161 | } | |
162 | return x; | |
163 | } | |
164 | ||
165 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
166 | { | |
167 | return compile_pattern_or(list); | |
168 | } | |
169 | ||
170 | void compile_grep_patterns(struct grep_opt *opt) | |
171 | { | |
172 | struct grep_pat *p; | |
173 | ||
0ab7befa JH |
174 | if (opt->all_match) |
175 | opt->extended = 1; | |
176 | ||
83b5d2f5 | 177 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
178 | switch (p->token) { |
179 | case GREP_PATTERN: /* atom */ | |
180 | case GREP_PATTERN_HEAD: | |
181 | case GREP_PATTERN_BODY: | |
c822255c | 182 | compile_regexp(p, opt); |
480c1ca6 JH |
183 | break; |
184 | default: | |
83b5d2f5 | 185 | opt->extended = 1; |
480c1ca6 JH |
186 | break; |
187 | } | |
83b5d2f5 JH |
188 | } |
189 | ||
190 | if (!opt->extended) | |
191 | return; | |
192 | ||
193 | /* Then bundle them up in an expression. | |
194 | * A classic recursive descent parser would do. | |
195 | */ | |
196 | p = opt->pattern_list; | |
ba150a3f MB |
197 | if (p) |
198 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
199 | if (p) |
200 | die("incomplete pattern expression: %s", p->pattern); | |
201 | } | |
202 | ||
b48fb5b6 JH |
203 | static void free_pattern_expr(struct grep_expr *x) |
204 | { | |
205 | switch (x->node) { | |
206 | case GREP_NODE_ATOM: | |
207 | break; | |
208 | case GREP_NODE_NOT: | |
209 | free_pattern_expr(x->u.unary); | |
210 | break; | |
211 | case GREP_NODE_AND: | |
212 | case GREP_NODE_OR: | |
213 | free_pattern_expr(x->u.binary.left); | |
214 | free_pattern_expr(x->u.binary.right); | |
215 | break; | |
216 | } | |
217 | free(x); | |
218 | } | |
219 | ||
220 | void free_grep_patterns(struct grep_opt *opt) | |
221 | { | |
222 | struct grep_pat *p, *n; | |
223 | ||
224 | for (p = opt->pattern_list; p; p = n) { | |
225 | n = p->next; | |
226 | switch (p->token) { | |
227 | case GREP_PATTERN: /* atom */ | |
228 | case GREP_PATTERN_HEAD: | |
229 | case GREP_PATTERN_BODY: | |
230 | regfree(&p->regexp); | |
231 | break; | |
232 | default: | |
233 | break; | |
234 | } | |
235 | free(p); | |
236 | } | |
237 | ||
238 | if (!opt->extended) | |
239 | return; | |
240 | free_pattern_expr(opt->pattern_expression); | |
241 | } | |
242 | ||
83b5d2f5 JH |
243 | static char *end_of_line(char *cp, unsigned long *left) |
244 | { | |
245 | unsigned long l = *left; | |
246 | while (l && *cp != '\n') { | |
247 | l--; | |
248 | cp++; | |
249 | } | |
250 | *left = l; | |
251 | return cp; | |
252 | } | |
253 | ||
254 | static int word_char(char ch) | |
255 | { | |
256 | return isalnum(ch) || ch == '_'; | |
257 | } | |
258 | ||
83caecca RZ |
259 | static void show_name(struct grep_opt *opt, const char *name) |
260 | { | |
261 | printf("%s%c", name, opt->null_following_name ? '\0' : '\n'); | |
262 | } | |
263 | ||
83b5d2f5 JH |
264 | static int fixmatch(const char *pattern, char *line, regmatch_t *match) |
265 | { | |
266 | char *hit = strstr(line, pattern); | |
267 | if (!hit) { | |
268 | match->rm_so = match->rm_eo = -1; | |
269 | return REG_NOMATCH; | |
270 | } | |
271 | else { | |
272 | match->rm_so = hit - line; | |
273 | match->rm_eo = match->rm_so + strlen(pattern); | |
274 | return 0; | |
275 | } | |
276 | } | |
277 | ||
a4d7d2c6 JH |
278 | static int strip_timestamp(char *bol, char **eol_p) |
279 | { | |
280 | char *eol = *eol_p; | |
281 | int ch; | |
282 | ||
283 | while (bol < --eol) { | |
284 | if (*eol != '>') | |
285 | continue; | |
286 | *eol_p = ++eol; | |
287 | ch = *eol; | |
288 | *eol = '\0'; | |
289 | return ch; | |
290 | } | |
291 | return 0; | |
292 | } | |
293 | ||
294 | static struct { | |
295 | const char *field; | |
296 | size_t len; | |
297 | } header_field[] = { | |
298 | { "author ", 7 }, | |
299 | { "committer ", 10 }, | |
300 | }; | |
301 | ||
d7eb527d | 302 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
303 | enum grep_context ctx, |
304 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
305 | { |
306 | int hit = 0; | |
a4d7d2c6 | 307 | int saved_ch = 0; |
e701fadb | 308 | const char *start = bol; |
83b5d2f5 | 309 | |
480c1ca6 JH |
310 | if ((p->token != GREP_PATTERN) && |
311 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
312 | return 0; | |
313 | ||
a4d7d2c6 JH |
314 | if (p->token == GREP_PATTERN_HEAD) { |
315 | const char *field; | |
316 | size_t len; | |
317 | assert(p->field < ARRAY_SIZE(header_field)); | |
318 | field = header_field[p->field].field; | |
319 | len = header_field[p->field].len; | |
320 | if (strncmp(bol, field, len)) | |
321 | return 0; | |
322 | bol += len; | |
323 | saved_ch = strip_timestamp(bol, &eol); | |
324 | } | |
325 | ||
83b5d2f5 | 326 | again: |
79212772 | 327 | if (p->fixed) |
83b5d2f5 | 328 | hit = !fixmatch(p->pattern, bol, pmatch); |
79212772 RS |
329 | else |
330 | hit = !regexec(&p->regexp, bol, 1, pmatch, eflags); | |
83b5d2f5 | 331 | |
d7eb527d | 332 | if (hit && p->word_regexp) { |
83b5d2f5 | 333 | if ((pmatch[0].rm_so < 0) || |
84201eae | 334 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
335 | (pmatch[0].rm_eo < 0) || |
336 | (eol - bol) < pmatch[0].rm_eo) | |
337 | die("regexp returned nonsense"); | |
338 | ||
339 | /* Match beginning must be either beginning of the | |
340 | * line, or at word boundary (i.e. the last char must | |
341 | * not be a word char). Similarly, match end must be | |
342 | * either end of the line, or at word boundary | |
343 | * (i.e. the next char must not be a word char). | |
344 | */ | |
fb62eb7f | 345 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
346 | !word_char(bol[pmatch[0].rm_so-1])) && |
347 | ((pmatch[0].rm_eo == (eol-bol)) || | |
348 | !word_char(bol[pmatch[0].rm_eo])) ) | |
349 | ; | |
350 | else | |
351 | hit = 0; | |
352 | ||
84201eae RS |
353 | /* Words consist of at least one character. */ |
354 | if (pmatch->rm_so == pmatch->rm_eo) | |
355 | hit = 0; | |
356 | ||
83b5d2f5 JH |
357 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
358 | /* There could be more than one match on the | |
359 | * line, and the first match might not be | |
360 | * strict word match. But later ones could be! | |
fb62eb7f RS |
361 | * Forward to the next possible start, i.e. the |
362 | * next position following a non-word char. | |
83b5d2f5 JH |
363 | */ |
364 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
365 | while (word_char(bol[-1]) && bol < eol) |
366 | bol++; | |
dbb6a4ad | 367 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
368 | if (bol < eol) |
369 | goto again; | |
83b5d2f5 JH |
370 | } |
371 | } | |
a4d7d2c6 JH |
372 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
373 | *eol = saved_ch; | |
e701fadb RS |
374 | if (hit) { |
375 | pmatch[0].rm_so += bol - start; | |
376 | pmatch[0].rm_eo += bol - start; | |
377 | } | |
83b5d2f5 JH |
378 | return hit; |
379 | } | |
380 | ||
d7eb527d RS |
381 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, |
382 | enum grep_context ctx, int collect_hits) | |
83b5d2f5 | 383 | { |
0ab7befa | 384 | int h = 0; |
79212772 | 385 | regmatch_t match; |
0ab7befa | 386 | |
c922b01f LT |
387 | if (!x) |
388 | die("Not a valid grep expression"); | |
83b5d2f5 JH |
389 | switch (x->node) { |
390 | case GREP_NODE_ATOM: | |
79212772 | 391 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); |
83b5d2f5 JH |
392 | break; |
393 | case GREP_NODE_NOT: | |
d7eb527d | 394 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); |
0ab7befa | 395 | break; |
83b5d2f5 | 396 | case GREP_NODE_AND: |
d7eb527d | 397 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) |
252d560d | 398 | return 0; |
d7eb527d | 399 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); |
0ab7befa | 400 | break; |
83b5d2f5 | 401 | case GREP_NODE_OR: |
0ab7befa | 402 | if (!collect_hits) |
d7eb527d | 403 | return (match_expr_eval(x->u.binary.left, |
0ab7befa | 404 | bol, eol, ctx, 0) || |
d7eb527d | 405 | match_expr_eval(x->u.binary.right, |
0ab7befa | 406 | bol, eol, ctx, 0)); |
d7eb527d | 407 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); |
0ab7befa | 408 | x->u.binary.left->hit |= h; |
d7eb527d | 409 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); |
0ab7befa JH |
410 | break; |
411 | default: | |
d7530708 | 412 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 413 | } |
0ab7befa JH |
414 | if (collect_hits) |
415 | x->hit |= h; | |
416 | return h; | |
83b5d2f5 JH |
417 | } |
418 | ||
480c1ca6 | 419 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 420 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
421 | { |
422 | struct grep_expr *x = opt->pattern_expression; | |
d7eb527d | 423 | return match_expr_eval(x, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
424 | } |
425 | ||
480c1ca6 | 426 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 427 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
428 | { |
429 | struct grep_pat *p; | |
79212772 RS |
430 | regmatch_t match; |
431 | ||
83b5d2f5 | 432 | if (opt->extended) |
0ab7befa JH |
433 | return match_expr(opt, bol, eol, ctx, collect_hits); |
434 | ||
435 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 436 | for (p = opt->pattern_list; p; p = p->next) { |
79212772 | 437 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) |
83b5d2f5 JH |
438 | return 1; |
439 | } | |
440 | return 0; | |
441 | } | |
442 | ||
7e8f59d5 RS |
443 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
444 | enum grep_context ctx, | |
445 | regmatch_t *pmatch, int eflags) | |
446 | { | |
447 | regmatch_t match; | |
448 | ||
449 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
450 | return 0; | |
451 | if (match.rm_so < 0 || match.rm_eo < 0) | |
452 | return 0; | |
453 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
454 | if (match.rm_so > pmatch->rm_so) | |
455 | return 1; | |
456 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
457 | return 1; | |
458 | } | |
459 | pmatch->rm_so = match.rm_so; | |
460 | pmatch->rm_eo = match.rm_eo; | |
461 | return 1; | |
462 | } | |
463 | ||
464 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
465 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
466 | { | |
467 | struct grep_pat *p; | |
468 | int hit = 0; | |
469 | ||
470 | pmatch->rm_so = pmatch->rm_eo = -1; | |
471 | if (bol < eol) { | |
472 | for (p = opt->pattern_list; p; p = p->next) { | |
473 | switch (p->token) { | |
474 | case GREP_PATTERN: /* atom */ | |
475 | case GREP_PATTERN_HEAD: | |
476 | case GREP_PATTERN_BODY: | |
477 | hit |= match_next_pattern(p, bol, eol, ctx, | |
478 | pmatch, eflags); | |
479 | break; | |
480 | default: | |
481 | break; | |
482 | } | |
483 | } | |
484 | } | |
485 | return hit; | |
486 | } | |
487 | ||
488 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
489 | const char *name, unsigned lno, char sign) | |
490 | { | |
491 | int rest = eol - bol; | |
492 | ||
5dd06d38 | 493 | if (opt->pre_context || opt->post_context) { |
046802d0 RS |
494 | if (opt->last_shown == 0) { |
495 | if (opt->show_hunk_mark) | |
496 | fputs("--\n", stdout); | |
497 | else | |
498 | opt->show_hunk_mark = 1; | |
499 | } else if (lno > opt->last_shown + 1) | |
5dd06d38 RS |
500 | fputs("--\n", stdout); |
501 | } | |
502 | opt->last_shown = lno; | |
503 | ||
7e8f59d5 RS |
504 | if (opt->null_following_name) |
505 | sign = '\0'; | |
506 | if (opt->pathname) | |
507 | printf("%s%c", name, sign); | |
508 | if (opt->linenum) | |
509 | printf("%d%c", lno, sign); | |
510 | if (opt->color) { | |
511 | regmatch_t match; | |
512 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
513 | int ch = *eol; | |
514 | int eflags = 0; | |
515 | ||
516 | *eol = '\0'; | |
517 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
518 | if (match.rm_so == match.rm_eo) |
519 | break; | |
7e8f59d5 | 520 | printf("%.*s%s%.*s%s", |
747a322b | 521 | (int)match.rm_so, bol, |
7e8f59d5 | 522 | opt->color_match, |
747a322b | 523 | (int)(match.rm_eo - match.rm_so), bol + match.rm_so, |
7e8f59d5 RS |
524 | GIT_COLOR_RESET); |
525 | bol += match.rm_eo; | |
526 | rest -= match.rm_eo; | |
527 | eflags = REG_NOTBOL; | |
528 | } | |
529 | *eol = ch; | |
530 | } | |
531 | printf("%.*s\n", rest, bol); | |
532 | } | |
533 | ||
49de3216 RS |
534 | static void show_pre_context(struct grep_opt *opt, const char *name, char *buf, |
535 | char *bol, unsigned lno) | |
536 | { | |
537 | unsigned cur = lno, from = 1; | |
538 | ||
539 | if (opt->pre_context < lno) | |
540 | from = lno - opt->pre_context; | |
541 | if (from <= opt->last_shown) | |
542 | from = opt->last_shown + 1; | |
543 | ||
544 | /* Rewind. */ | |
545 | while (bol > buf && cur > from) { | |
546 | bol--; | |
547 | while (bol > buf && bol[-1] != '\n') | |
548 | bol--; | |
549 | cur--; | |
550 | } | |
551 | ||
552 | /* Back forward. */ | |
553 | while (cur < lno) { | |
554 | char *eol = bol; | |
555 | ||
556 | while (*eol != '\n') | |
557 | eol++; | |
558 | show_line(opt, bol, eol, name, cur, '-'); | |
559 | bol = eol + 1; | |
560 | cur++; | |
561 | } | |
562 | } | |
563 | ||
0ab7befa JH |
564 | static int grep_buffer_1(struct grep_opt *opt, const char *name, |
565 | char *buf, unsigned long size, int collect_hits) | |
83b5d2f5 JH |
566 | { |
567 | char *bol = buf; | |
568 | unsigned long left = size; | |
569 | unsigned lno = 1; | |
83b5d2f5 | 570 | unsigned last_hit = 0; |
83b5d2f5 | 571 | int binary_match_only = 0; |
83b5d2f5 | 572 | unsigned count = 0; |
480c1ca6 | 573 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
83b5d2f5 | 574 | |
5dd06d38 RS |
575 | opt->last_shown = 0; |
576 | ||
83b5d2f5 JH |
577 | if (buffer_is_binary(buf, size)) { |
578 | switch (opt->binary) { | |
579 | case GREP_BINARY_DEFAULT: | |
580 | binary_match_only = 1; | |
581 | break; | |
582 | case GREP_BINARY_NOMATCH: | |
583 | return 0; /* Assume unmatch */ | |
584 | break; | |
585 | default: | |
586 | break; | |
587 | } | |
588 | } | |
589 | ||
83b5d2f5 JH |
590 | while (left) { |
591 | char *eol, ch; | |
0ab7befa | 592 | int hit; |
83b5d2f5 JH |
593 | |
594 | eol = end_of_line(bol, &left); | |
595 | ch = *eol; | |
596 | *eol = 0; | |
597 | ||
480c1ca6 JH |
598 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
599 | ctx = GREP_CONTEXT_BODY; | |
600 | ||
0ab7befa | 601 | hit = match_line(opt, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
602 | *eol = ch; |
603 | ||
0ab7befa JH |
604 | if (collect_hits) |
605 | goto next_line; | |
606 | ||
83b5d2f5 JH |
607 | /* "grep -v -e foo -e bla" should list lines |
608 | * that do not have either, so inversion should | |
609 | * be done outside. | |
610 | */ | |
611 | if (opt->invert) | |
612 | hit = !hit; | |
613 | if (opt->unmatch_name_only) { | |
614 | if (hit) | |
615 | return 0; | |
616 | goto next_line; | |
617 | } | |
618 | if (hit) { | |
619 | count++; | |
620 | if (opt->status_only) | |
621 | return 1; | |
622 | if (binary_match_only) { | |
623 | printf("Binary file %s matches\n", name); | |
624 | return 1; | |
625 | } | |
626 | if (opt->name_only) { | |
83caecca | 627 | show_name(opt, name); |
83b5d2f5 JH |
628 | return 1; |
629 | } | |
630 | /* Hit at this line. If we haven't shown the | |
631 | * pre-context lines, we would need to show them. | |
632 | * When asked to do "count", this still show | |
633 | * the context which is nonsense, but the user | |
634 | * deserves to get that ;-). | |
635 | */ | |
49de3216 RS |
636 | if (opt->pre_context) |
637 | show_pre_context(opt, name, buf, bol, lno); | |
83b5d2f5 JH |
638 | if (!opt->count) |
639 | show_line(opt, bol, eol, name, lno, ':'); | |
5dd06d38 | 640 | last_hit = lno; |
83b5d2f5 JH |
641 | } |
642 | else if (last_hit && | |
643 | lno <= last_hit + opt->post_context) { | |
644 | /* If the last hit is within the post context, | |
645 | * we need to show this line. | |
646 | */ | |
83b5d2f5 | 647 | show_line(opt, bol, eol, name, lno, '-'); |
83b5d2f5 | 648 | } |
83b5d2f5 JH |
649 | |
650 | next_line: | |
651 | bol = eol + 1; | |
652 | if (!left) | |
653 | break; | |
654 | left--; | |
655 | lno++; | |
656 | } | |
657 | ||
0ab7befa JH |
658 | if (collect_hits) |
659 | return 0; | |
b48fb5b6 | 660 | |
83b5d2f5 JH |
661 | if (opt->status_only) |
662 | return 0; | |
663 | if (opt->unmatch_name_only) { | |
664 | /* We did not see any hit, so we want to show this */ | |
83caecca | 665 | show_name(opt, name); |
83b5d2f5 JH |
666 | return 1; |
667 | } | |
668 | ||
669 | /* NEEDSWORK: | |
670 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
671 | * which feels mostly useless but sometimes useful. Maybe | |
672 | * make it another option? For now suppress them. | |
673 | */ | |
674 | if (opt->count && count) | |
83caecca RZ |
675 | printf("%s%c%u\n", name, |
676 | opt->null_following_name ? '\0' : ':', count); | |
83b5d2f5 JH |
677 | return !!last_hit; |
678 | } | |
679 | ||
0ab7befa JH |
680 | static void clr_hit_marker(struct grep_expr *x) |
681 | { | |
682 | /* All-hit markers are meaningful only at the very top level | |
683 | * OR node. | |
684 | */ | |
685 | while (1) { | |
686 | x->hit = 0; | |
687 | if (x->node != GREP_NODE_OR) | |
688 | return; | |
689 | x->u.binary.left->hit = 0; | |
690 | x = x->u.binary.right; | |
691 | } | |
692 | } | |
693 | ||
694 | static int chk_hit_marker(struct grep_expr *x) | |
695 | { | |
696 | /* Top level nodes have hit markers. See if they all are hits */ | |
697 | while (1) { | |
698 | if (x->node != GREP_NODE_OR) | |
699 | return x->hit; | |
700 | if (!x->u.binary.left->hit) | |
701 | return 0; | |
702 | x = x->u.binary.right; | |
703 | } | |
704 | } | |
705 | ||
706 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
707 | { | |
708 | /* | |
709 | * we do not have to do the two-pass grep when we do not check | |
710 | * buffer-wide "all-match". | |
711 | */ | |
712 | if (!opt->all_match) | |
713 | return grep_buffer_1(opt, name, buf, size, 0); | |
714 | ||
715 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
716 | * We first clear hit markers from them. | |
717 | */ | |
718 | clr_hit_marker(opt->pattern_expression); | |
719 | grep_buffer_1(opt, name, buf, size, 1); | |
720 | ||
721 | if (!chk_hit_marker(opt->pattern_expression)) | |
722 | return 0; | |
723 | ||
724 | return grep_buffer_1(opt, name, buf, size, 0); | |
725 | } |