]>
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 | ||
c822255c RS |
32 | static int is_fixed(const char *s) |
33 | { | |
f9b7cce6 | 34 | while (*s && !is_regex_special(*s)) |
c822255c RS |
35 | s++; |
36 | return !*s; | |
37 | } | |
38 | ||
83b5d2f5 JH |
39 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) |
40 | { | |
c822255c RS |
41 | int err; |
42 | ||
d7eb527d | 43 | p->word_regexp = opt->word_regexp; |
5183bf67 | 44 | p->ignore_case = opt->ignore_case; |
d7eb527d | 45 | |
c822255c RS |
46 | if (opt->fixed || is_fixed(p->pattern)) |
47 | p->fixed = 1; | |
48 | if (opt->regflags & REG_ICASE) | |
49 | p->fixed = 0; | |
50 | if (p->fixed) | |
51 | return; | |
52 | ||
53 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
83b5d2f5 JH |
54 | if (err) { |
55 | char errbuf[1024]; | |
56 | char where[1024]; | |
57 | if (p->no) | |
58 | sprintf(where, "In '%s' at %d, ", | |
59 | p->origin, p->no); | |
60 | else if (p->origin) | |
61 | sprintf(where, "%s, ", p->origin); | |
62 | else | |
63 | where[0] = 0; | |
64 | regerror(err, &p->regexp, errbuf, 1024); | |
65 | regfree(&p->regexp); | |
66 | die("%s'%s': %s", where, p->pattern, errbuf); | |
67 | } | |
68 | } | |
69 | ||
0ab7befa | 70 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
71 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
72 | { | |
73 | struct grep_pat *p; | |
74 | struct grep_expr *x; | |
75 | ||
76 | p = *list; | |
c922b01f LT |
77 | if (!p) |
78 | return NULL; | |
83b5d2f5 JH |
79 | switch (p->token) { |
80 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
81 | case GREP_PATTERN_HEAD: |
82 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
83 | x = xcalloc(1, sizeof (struct grep_expr)); |
84 | x->node = GREP_NODE_ATOM; | |
85 | x->u.atom = p; | |
86 | *list = p->next; | |
87 | return x; | |
88 | case GREP_OPEN_PAREN: | |
89 | *list = p->next; | |
0ab7befa | 90 | x = compile_pattern_or(list); |
83b5d2f5 JH |
91 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
92 | die("unmatched parenthesis"); | |
93 | *list = (*list)->next; | |
94 | return x; | |
95 | default: | |
96 | return NULL; | |
97 | } | |
98 | } | |
99 | ||
100 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
101 | { | |
102 | struct grep_pat *p; | |
103 | struct grep_expr *x; | |
104 | ||
105 | p = *list; | |
c922b01f LT |
106 | if (!p) |
107 | return NULL; | |
83b5d2f5 JH |
108 | switch (p->token) { |
109 | case GREP_NOT: | |
110 | if (!p->next) | |
111 | die("--not not followed by pattern expression"); | |
112 | *list = p->next; | |
113 | x = xcalloc(1, sizeof (struct grep_expr)); | |
114 | x->node = GREP_NODE_NOT; | |
115 | x->u.unary = compile_pattern_not(list); | |
116 | if (!x->u.unary) | |
117 | die("--not followed by non pattern expression"); | |
118 | return x; | |
119 | default: | |
120 | return compile_pattern_atom(list); | |
121 | } | |
122 | } | |
123 | ||
124 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
125 | { | |
126 | struct grep_pat *p; | |
127 | struct grep_expr *x, *y, *z; | |
128 | ||
129 | x = compile_pattern_not(list); | |
130 | p = *list; | |
131 | if (p && p->token == GREP_AND) { | |
132 | if (!p->next) | |
133 | die("--and not followed by pattern expression"); | |
134 | *list = p->next; | |
135 | y = compile_pattern_and(list); | |
136 | if (!y) | |
137 | die("--and not followed by pattern expression"); | |
138 | z = xcalloc(1, sizeof (struct grep_expr)); | |
139 | z->node = GREP_NODE_AND; | |
140 | z->u.binary.left = x; | |
141 | z->u.binary.right = y; | |
142 | return z; | |
143 | } | |
144 | return x; | |
145 | } | |
146 | ||
147 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
148 | { | |
149 | struct grep_pat *p; | |
150 | struct grep_expr *x, *y, *z; | |
151 | ||
152 | x = compile_pattern_and(list); | |
153 | p = *list; | |
154 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
155 | y = compile_pattern_or(list); | |
156 | if (!y) | |
157 | die("not a pattern expression %s", p->pattern); | |
158 | z = xcalloc(1, sizeof (struct grep_expr)); | |
159 | z->node = GREP_NODE_OR; | |
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_expr(struct grep_pat **list) | |
168 | { | |
169 | return compile_pattern_or(list); | |
170 | } | |
171 | ||
172 | void compile_grep_patterns(struct grep_opt *opt) | |
173 | { | |
174 | struct grep_pat *p; | |
175 | ||
0ab7befa JH |
176 | if (opt->all_match) |
177 | opt->extended = 1; | |
178 | ||
83b5d2f5 | 179 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
180 | switch (p->token) { |
181 | case GREP_PATTERN: /* atom */ | |
182 | case GREP_PATTERN_HEAD: | |
183 | case GREP_PATTERN_BODY: | |
c822255c | 184 | compile_regexp(p, opt); |
480c1ca6 JH |
185 | break; |
186 | default: | |
83b5d2f5 | 187 | opt->extended = 1; |
480c1ca6 JH |
188 | break; |
189 | } | |
83b5d2f5 JH |
190 | } |
191 | ||
192 | if (!opt->extended) | |
193 | return; | |
194 | ||
195 | /* Then bundle them up in an expression. | |
196 | * A classic recursive descent parser would do. | |
197 | */ | |
198 | p = opt->pattern_list; | |
ba150a3f MB |
199 | if (p) |
200 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
201 | if (p) |
202 | die("incomplete pattern expression: %s", p->pattern); | |
203 | } | |
204 | ||
b48fb5b6 JH |
205 | static void free_pattern_expr(struct grep_expr *x) |
206 | { | |
207 | switch (x->node) { | |
208 | case GREP_NODE_ATOM: | |
209 | break; | |
210 | case GREP_NODE_NOT: | |
211 | free_pattern_expr(x->u.unary); | |
212 | break; | |
213 | case GREP_NODE_AND: | |
214 | case GREP_NODE_OR: | |
215 | free_pattern_expr(x->u.binary.left); | |
216 | free_pattern_expr(x->u.binary.right); | |
217 | break; | |
218 | } | |
219 | free(x); | |
220 | } | |
221 | ||
222 | void free_grep_patterns(struct grep_opt *opt) | |
223 | { | |
224 | struct grep_pat *p, *n; | |
225 | ||
226 | for (p = opt->pattern_list; p; p = n) { | |
227 | n = p->next; | |
228 | switch (p->token) { | |
229 | case GREP_PATTERN: /* atom */ | |
230 | case GREP_PATTERN_HEAD: | |
231 | case GREP_PATTERN_BODY: | |
232 | regfree(&p->regexp); | |
233 | break; | |
234 | default: | |
235 | break; | |
236 | } | |
237 | free(p); | |
238 | } | |
239 | ||
240 | if (!opt->extended) | |
241 | return; | |
242 | free_pattern_expr(opt->pattern_expression); | |
243 | } | |
244 | ||
83b5d2f5 JH |
245 | static char *end_of_line(char *cp, unsigned long *left) |
246 | { | |
247 | unsigned long l = *left; | |
248 | while (l && *cp != '\n') { | |
249 | l--; | |
250 | cp++; | |
251 | } | |
252 | *left = l; | |
253 | return cp; | |
254 | } | |
255 | ||
256 | static int word_char(char ch) | |
257 | { | |
258 | return isalnum(ch) || ch == '_'; | |
259 | } | |
260 | ||
83caecca RZ |
261 | static void show_name(struct grep_opt *opt, const char *name) |
262 | { | |
263 | printf("%s%c", name, opt->null_following_name ? '\0' : '\n'); | |
264 | } | |
265 | ||
5183bf67 BC |
266 | |
267 | static int fixmatch(const char *pattern, char *line, int ignore_case, regmatch_t *match) | |
83b5d2f5 | 268 | { |
5183bf67 BC |
269 | char *hit; |
270 | if (ignore_case) | |
271 | hit = strcasestr(line, pattern); | |
272 | else | |
273 | hit = strstr(line, pattern); | |
274 | ||
83b5d2f5 JH |
275 | if (!hit) { |
276 | match->rm_so = match->rm_eo = -1; | |
277 | return REG_NOMATCH; | |
278 | } | |
279 | else { | |
280 | match->rm_so = hit - line; | |
281 | match->rm_eo = match->rm_so + strlen(pattern); | |
282 | return 0; | |
283 | } | |
284 | } | |
285 | ||
a4d7d2c6 JH |
286 | static int strip_timestamp(char *bol, char **eol_p) |
287 | { | |
288 | char *eol = *eol_p; | |
289 | int ch; | |
290 | ||
291 | while (bol < --eol) { | |
292 | if (*eol != '>') | |
293 | continue; | |
294 | *eol_p = ++eol; | |
295 | ch = *eol; | |
296 | *eol = '\0'; | |
297 | return ch; | |
298 | } | |
299 | return 0; | |
300 | } | |
301 | ||
302 | static struct { | |
303 | const char *field; | |
304 | size_t len; | |
305 | } header_field[] = { | |
306 | { "author ", 7 }, | |
307 | { "committer ", 10 }, | |
308 | }; | |
309 | ||
d7eb527d | 310 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
311 | enum grep_context ctx, |
312 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
313 | { |
314 | int hit = 0; | |
a4d7d2c6 | 315 | int saved_ch = 0; |
e701fadb | 316 | const char *start = bol; |
83b5d2f5 | 317 | |
480c1ca6 JH |
318 | if ((p->token != GREP_PATTERN) && |
319 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
320 | return 0; | |
321 | ||
a4d7d2c6 JH |
322 | if (p->token == GREP_PATTERN_HEAD) { |
323 | const char *field; | |
324 | size_t len; | |
325 | assert(p->field < ARRAY_SIZE(header_field)); | |
326 | field = header_field[p->field].field; | |
327 | len = header_field[p->field].len; | |
328 | if (strncmp(bol, field, len)) | |
329 | return 0; | |
330 | bol += len; | |
331 | saved_ch = strip_timestamp(bol, &eol); | |
332 | } | |
333 | ||
83b5d2f5 | 334 | again: |
79212772 | 335 | if (p->fixed) |
5183bf67 | 336 | hit = !fixmatch(p->pattern, bol, p->ignore_case, pmatch); |
79212772 RS |
337 | else |
338 | hit = !regexec(&p->regexp, bol, 1, pmatch, eflags); | |
83b5d2f5 | 339 | |
d7eb527d | 340 | if (hit && p->word_regexp) { |
83b5d2f5 | 341 | if ((pmatch[0].rm_so < 0) || |
84201eae | 342 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
343 | (pmatch[0].rm_eo < 0) || |
344 | (eol - bol) < pmatch[0].rm_eo) | |
345 | die("regexp returned nonsense"); | |
346 | ||
347 | /* Match beginning must be either beginning of the | |
348 | * line, or at word boundary (i.e. the last char must | |
349 | * not be a word char). Similarly, match end must be | |
350 | * either end of the line, or at word boundary | |
351 | * (i.e. the next char must not be a word char). | |
352 | */ | |
fb62eb7f | 353 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
354 | !word_char(bol[pmatch[0].rm_so-1])) && |
355 | ((pmatch[0].rm_eo == (eol-bol)) || | |
356 | !word_char(bol[pmatch[0].rm_eo])) ) | |
357 | ; | |
358 | else | |
359 | hit = 0; | |
360 | ||
84201eae RS |
361 | /* Words consist of at least one character. */ |
362 | if (pmatch->rm_so == pmatch->rm_eo) | |
363 | hit = 0; | |
364 | ||
83b5d2f5 JH |
365 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
366 | /* There could be more than one match on the | |
367 | * line, and the first match might not be | |
368 | * strict word match. But later ones could be! | |
fb62eb7f RS |
369 | * Forward to the next possible start, i.e. the |
370 | * next position following a non-word char. | |
83b5d2f5 JH |
371 | */ |
372 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
373 | while (word_char(bol[-1]) && bol < eol) |
374 | bol++; | |
dbb6a4ad | 375 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
376 | if (bol < eol) |
377 | goto again; | |
83b5d2f5 JH |
378 | } |
379 | } | |
a4d7d2c6 JH |
380 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
381 | *eol = saved_ch; | |
e701fadb RS |
382 | if (hit) { |
383 | pmatch[0].rm_so += bol - start; | |
384 | pmatch[0].rm_eo += bol - start; | |
385 | } | |
83b5d2f5 JH |
386 | return hit; |
387 | } | |
388 | ||
d7eb527d RS |
389 | static int match_expr_eval(struct grep_expr *x, char *bol, char *eol, |
390 | enum grep_context ctx, int collect_hits) | |
83b5d2f5 | 391 | { |
0ab7befa | 392 | int h = 0; |
79212772 | 393 | regmatch_t match; |
0ab7befa | 394 | |
c922b01f LT |
395 | if (!x) |
396 | die("Not a valid grep expression"); | |
83b5d2f5 JH |
397 | switch (x->node) { |
398 | case GREP_NODE_ATOM: | |
79212772 | 399 | h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0); |
83b5d2f5 JH |
400 | break; |
401 | case GREP_NODE_NOT: | |
d7eb527d | 402 | h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0); |
0ab7befa | 403 | break; |
83b5d2f5 | 404 | case GREP_NODE_AND: |
d7eb527d | 405 | if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0)) |
252d560d | 406 | return 0; |
d7eb527d | 407 | h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0); |
0ab7befa | 408 | break; |
83b5d2f5 | 409 | case GREP_NODE_OR: |
0ab7befa | 410 | if (!collect_hits) |
d7eb527d | 411 | return (match_expr_eval(x->u.binary.left, |
0ab7befa | 412 | bol, eol, ctx, 0) || |
d7eb527d | 413 | match_expr_eval(x->u.binary.right, |
0ab7befa | 414 | bol, eol, ctx, 0)); |
d7eb527d | 415 | h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0); |
0ab7befa | 416 | x->u.binary.left->hit |= h; |
d7eb527d | 417 | h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1); |
0ab7befa JH |
418 | break; |
419 | default: | |
d7530708 | 420 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 421 | } |
0ab7befa JH |
422 | if (collect_hits) |
423 | x->hit |= h; | |
424 | return h; | |
83b5d2f5 JH |
425 | } |
426 | ||
480c1ca6 | 427 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 428 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
429 | { |
430 | struct grep_expr *x = opt->pattern_expression; | |
d7eb527d | 431 | return match_expr_eval(x, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
432 | } |
433 | ||
480c1ca6 | 434 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
0ab7befa | 435 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
436 | { |
437 | struct grep_pat *p; | |
79212772 RS |
438 | regmatch_t match; |
439 | ||
83b5d2f5 | 440 | if (opt->extended) |
0ab7befa JH |
441 | return match_expr(opt, bol, eol, ctx, collect_hits); |
442 | ||
443 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 444 | for (p = opt->pattern_list; p; p = p->next) { |
79212772 | 445 | if (match_one_pattern(p, bol, eol, ctx, &match, 0)) |
83b5d2f5 JH |
446 | return 1; |
447 | } | |
448 | return 0; | |
449 | } | |
450 | ||
7e8f59d5 RS |
451 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
452 | enum grep_context ctx, | |
453 | regmatch_t *pmatch, int eflags) | |
454 | { | |
455 | regmatch_t match; | |
456 | ||
457 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
458 | return 0; | |
459 | if (match.rm_so < 0 || match.rm_eo < 0) | |
460 | return 0; | |
461 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
462 | if (match.rm_so > pmatch->rm_so) | |
463 | return 1; | |
464 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
465 | return 1; | |
466 | } | |
467 | pmatch->rm_so = match.rm_so; | |
468 | pmatch->rm_eo = match.rm_eo; | |
469 | return 1; | |
470 | } | |
471 | ||
472 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
473 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
474 | { | |
475 | struct grep_pat *p; | |
476 | int hit = 0; | |
477 | ||
478 | pmatch->rm_so = pmatch->rm_eo = -1; | |
479 | if (bol < eol) { | |
480 | for (p = opt->pattern_list; p; p = p->next) { | |
481 | switch (p->token) { | |
482 | case GREP_PATTERN: /* atom */ | |
483 | case GREP_PATTERN_HEAD: | |
484 | case GREP_PATTERN_BODY: | |
485 | hit |= match_next_pattern(p, bol, eol, ctx, | |
486 | pmatch, eflags); | |
487 | break; | |
488 | default: | |
489 | break; | |
490 | } | |
491 | } | |
492 | } | |
493 | return hit; | |
494 | } | |
495 | ||
496 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
497 | const char *name, unsigned lno, char sign) | |
498 | { | |
499 | int rest = eol - bol; | |
500 | ||
ed24e401 | 501 | if (opt->pre_context || opt->post_context) { |
046802d0 RS |
502 | if (opt->last_shown == 0) { |
503 | if (opt->show_hunk_mark) | |
ed24e401 | 504 | fputs("--\n", stdout); |
046802d0 RS |
505 | else |
506 | opt->show_hunk_mark = 1; | |
ed24e401 RS |
507 | } else if (lno > opt->last_shown + 1) |
508 | fputs("--\n", stdout); | |
5dd06d38 RS |
509 | } |
510 | opt->last_shown = lno; | |
511 | ||
7e8f59d5 RS |
512 | if (opt->null_following_name) |
513 | sign = '\0'; | |
514 | if (opt->pathname) | |
515 | printf("%s%c", name, sign); | |
516 | if (opt->linenum) | |
517 | printf("%d%c", lno, sign); | |
518 | if (opt->color) { | |
519 | regmatch_t match; | |
520 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
521 | int ch = *eol; | |
522 | int eflags = 0; | |
523 | ||
524 | *eol = '\0'; | |
525 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
526 | if (match.rm_so == match.rm_eo) |
527 | break; | |
7e8f59d5 | 528 | printf("%.*s%s%.*s%s", |
747a322b | 529 | (int)match.rm_so, bol, |
7e8f59d5 | 530 | opt->color_match, |
747a322b | 531 | (int)(match.rm_eo - match.rm_so), bol + match.rm_so, |
7e8f59d5 RS |
532 | GIT_COLOR_RESET); |
533 | bol += match.rm_eo; | |
534 | rest -= match.rm_eo; | |
535 | eflags = REG_NOTBOL; | |
536 | } | |
537 | *eol = ch; | |
538 | } | |
539 | printf("%.*s\n", rest, bol); | |
540 | } | |
541 | ||
60ecac98 | 542 | static int match_funcname(struct grep_opt *opt, char *bol, char *eol) |
2944e4e6 | 543 | { |
60ecac98 RS |
544 | xdemitconf_t *xecfg = opt->priv; |
545 | if (xecfg && xecfg->find_func) { | |
546 | char buf[1]; | |
547 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
548 | xecfg->find_func_priv) >= 0; | |
549 | } | |
550 | ||
2944e4e6 RS |
551 | if (bol == eol) |
552 | return 0; | |
553 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
554 | return 1; | |
555 | return 0; | |
556 | } | |
557 | ||
558 | static void show_funcname_line(struct grep_opt *opt, const char *name, | |
559 | char *buf, char *bol, unsigned lno) | |
560 | { | |
561 | while (bol > buf) { | |
562 | char *eol = --bol; | |
563 | ||
564 | while (bol > buf && bol[-1] != '\n') | |
565 | bol--; | |
566 | lno--; | |
567 | ||
568 | if (lno <= opt->last_shown) | |
569 | break; | |
570 | ||
60ecac98 | 571 | if (match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
572 | show_line(opt, bol, eol, name, lno, '='); |
573 | break; | |
574 | } | |
575 | } | |
576 | } | |
577 | ||
49de3216 RS |
578 | static void show_pre_context(struct grep_opt *opt, const char *name, char *buf, |
579 | char *bol, unsigned lno) | |
580 | { | |
2944e4e6 RS |
581 | unsigned cur = lno, from = 1, funcname_lno = 0; |
582 | int funcname_needed = opt->funcname; | |
49de3216 RS |
583 | |
584 | if (opt->pre_context < lno) | |
585 | from = lno - opt->pre_context; | |
586 | if (from <= opt->last_shown) | |
587 | from = opt->last_shown + 1; | |
588 | ||
589 | /* Rewind. */ | |
590 | while (bol > buf && cur > from) { | |
2944e4e6 RS |
591 | char *eol = --bol; |
592 | ||
49de3216 RS |
593 | while (bol > buf && bol[-1] != '\n') |
594 | bol--; | |
595 | cur--; | |
60ecac98 | 596 | if (funcname_needed && match_funcname(opt, bol, eol)) { |
2944e4e6 RS |
597 | funcname_lno = cur; |
598 | funcname_needed = 0; | |
599 | } | |
49de3216 RS |
600 | } |
601 | ||
2944e4e6 RS |
602 | /* We need to look even further back to find a function signature. */ |
603 | if (opt->funcname && funcname_needed) | |
604 | show_funcname_line(opt, name, buf, bol, cur); | |
605 | ||
49de3216 RS |
606 | /* Back forward. */ |
607 | while (cur < lno) { | |
2944e4e6 | 608 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; |
49de3216 RS |
609 | |
610 | while (*eol != '\n') | |
611 | eol++; | |
2944e4e6 | 612 | show_line(opt, bol, eol, name, cur, sign); |
49de3216 RS |
613 | bol = eol + 1; |
614 | cur++; | |
615 | } | |
616 | } | |
617 | ||
0ab7befa JH |
618 | static int grep_buffer_1(struct grep_opt *opt, const char *name, |
619 | char *buf, unsigned long size, int collect_hits) | |
83b5d2f5 JH |
620 | { |
621 | char *bol = buf; | |
622 | unsigned long left = size; | |
623 | unsigned lno = 1; | |
83b5d2f5 | 624 | unsigned last_hit = 0; |
83b5d2f5 | 625 | int binary_match_only = 0; |
83b5d2f5 | 626 | unsigned count = 0; |
480c1ca6 | 627 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
60ecac98 | 628 | xdemitconf_t xecfg; |
83b5d2f5 | 629 | |
5dd06d38 RS |
630 | opt->last_shown = 0; |
631 | ||
83b5d2f5 JH |
632 | if (buffer_is_binary(buf, size)) { |
633 | switch (opt->binary) { | |
634 | case GREP_BINARY_DEFAULT: | |
635 | binary_match_only = 1; | |
636 | break; | |
637 | case GREP_BINARY_NOMATCH: | |
638 | return 0; /* Assume unmatch */ | |
639 | break; | |
640 | default: | |
641 | break; | |
642 | } | |
643 | } | |
644 | ||
60ecac98 RS |
645 | memset(&xecfg, 0, sizeof(xecfg)); |
646 | if (opt->funcname && !opt->unmatch_name_only && !opt->status_only && | |
647 | !opt->name_only && !binary_match_only && !collect_hits) { | |
648 | struct userdiff_driver *drv = userdiff_find_by_path(name); | |
649 | if (drv && drv->funcname.pattern) { | |
650 | const struct userdiff_funcname *pe = &drv->funcname; | |
651 | xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags); | |
652 | opt->priv = &xecfg; | |
653 | } | |
654 | } | |
655 | ||
83b5d2f5 JH |
656 | while (left) { |
657 | char *eol, ch; | |
0ab7befa | 658 | int hit; |
83b5d2f5 JH |
659 | |
660 | eol = end_of_line(bol, &left); | |
661 | ch = *eol; | |
662 | *eol = 0; | |
663 | ||
480c1ca6 JH |
664 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
665 | ctx = GREP_CONTEXT_BODY; | |
666 | ||
0ab7befa | 667 | hit = match_line(opt, bol, eol, ctx, collect_hits); |
83b5d2f5 JH |
668 | *eol = ch; |
669 | ||
0ab7befa JH |
670 | if (collect_hits) |
671 | goto next_line; | |
672 | ||
83b5d2f5 JH |
673 | /* "grep -v -e foo -e bla" should list lines |
674 | * that do not have either, so inversion should | |
675 | * be done outside. | |
676 | */ | |
677 | if (opt->invert) | |
678 | hit = !hit; | |
679 | if (opt->unmatch_name_only) { | |
680 | if (hit) | |
681 | return 0; | |
682 | goto next_line; | |
683 | } | |
684 | if (hit) { | |
685 | count++; | |
686 | if (opt->status_only) | |
687 | return 1; | |
688 | if (binary_match_only) { | |
689 | printf("Binary file %s matches\n", name); | |
690 | return 1; | |
691 | } | |
692 | if (opt->name_only) { | |
83caecca | 693 | show_name(opt, name); |
83b5d2f5 JH |
694 | return 1; |
695 | } | |
696 | /* Hit at this line. If we haven't shown the | |
697 | * pre-context lines, we would need to show them. | |
698 | * When asked to do "count", this still show | |
699 | * the context which is nonsense, but the user | |
700 | * deserves to get that ;-). | |
701 | */ | |
49de3216 RS |
702 | if (opt->pre_context) |
703 | show_pre_context(opt, name, buf, bol, lno); | |
2944e4e6 RS |
704 | else if (opt->funcname) |
705 | show_funcname_line(opt, name, buf, bol, lno); | |
83b5d2f5 JH |
706 | if (!opt->count) |
707 | show_line(opt, bol, eol, name, lno, ':'); | |
5dd06d38 | 708 | last_hit = lno; |
83b5d2f5 JH |
709 | } |
710 | else if (last_hit && | |
711 | lno <= last_hit + opt->post_context) { | |
712 | /* If the last hit is within the post context, | |
713 | * we need to show this line. | |
714 | */ | |
83b5d2f5 | 715 | show_line(opt, bol, eol, name, lno, '-'); |
83b5d2f5 | 716 | } |
83b5d2f5 JH |
717 | |
718 | next_line: | |
719 | bol = eol + 1; | |
720 | if (!left) | |
721 | break; | |
722 | left--; | |
723 | lno++; | |
724 | } | |
725 | ||
0ab7befa JH |
726 | if (collect_hits) |
727 | return 0; | |
b48fb5b6 | 728 | |
83b5d2f5 JH |
729 | if (opt->status_only) |
730 | return 0; | |
731 | if (opt->unmatch_name_only) { | |
732 | /* We did not see any hit, so we want to show this */ | |
83caecca | 733 | show_name(opt, name); |
83b5d2f5 JH |
734 | return 1; |
735 | } | |
736 | ||
60ecac98 RS |
737 | xdiff_clear_find_func(&xecfg); |
738 | opt->priv = NULL; | |
739 | ||
83b5d2f5 JH |
740 | /* NEEDSWORK: |
741 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
742 | * which feels mostly useless but sometimes useful. Maybe | |
743 | * make it another option? For now suppress them. | |
744 | */ | |
745 | if (opt->count && count) | |
83caecca RZ |
746 | printf("%s%c%u\n", name, |
747 | opt->null_following_name ? '\0' : ':', count); | |
83b5d2f5 JH |
748 | return !!last_hit; |
749 | } | |
750 | ||
0ab7befa JH |
751 | static void clr_hit_marker(struct grep_expr *x) |
752 | { | |
753 | /* All-hit markers are meaningful only at the very top level | |
754 | * OR node. | |
755 | */ | |
756 | while (1) { | |
757 | x->hit = 0; | |
758 | if (x->node != GREP_NODE_OR) | |
759 | return; | |
760 | x->u.binary.left->hit = 0; | |
761 | x = x->u.binary.right; | |
762 | } | |
763 | } | |
764 | ||
765 | static int chk_hit_marker(struct grep_expr *x) | |
766 | { | |
767 | /* Top level nodes have hit markers. See if they all are hits */ | |
768 | while (1) { | |
769 | if (x->node != GREP_NODE_OR) | |
770 | return x->hit; | |
771 | if (!x->u.binary.left->hit) | |
772 | return 0; | |
773 | x = x->u.binary.right; | |
774 | } | |
775 | } | |
776 | ||
777 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
778 | { | |
779 | /* | |
780 | * we do not have to do the two-pass grep when we do not check | |
781 | * buffer-wide "all-match". | |
782 | */ | |
783 | if (!opt->all_match) | |
784 | return grep_buffer_1(opt, name, buf, size, 0); | |
785 | ||
786 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
787 | * We first clear hit markers from them. | |
788 | */ | |
789 | clr_hit_marker(opt->pattern_expression); | |
790 | grep_buffer_1(opt, name, buf, size, 1); | |
791 | ||
792 | if (!chk_hit_marker(opt->pattern_expression)) | |
793 | return 0; | |
794 | ||
795 | return grep_buffer_1(opt, name, buf, size, 0); | |
796 | } |