?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'js/trace2-fetch-push'
[git.git] / grep.c
blob0bb4cbd3d82ceef5517905a35a7f6820b7fb555a
1 #include "cache.h"
2 #include "config.h"
3 #include "grep.h"
4 #include "object-store.h"
5 #include "userdiff.h"
6 #include "xdiff-interface.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "commit.h"
10 #include "quote.h"
11 #include "help.h"
12
13 static int grep_source_load(struct grep_source *gs);
14 static int grep_source_is_binary(struct grep_source *gs,
15 struct index_state *istate);
16
17 static struct grep_opt grep_defaults;
18
19 static const char *color_grep_slots[] = {
20 [GREP_COLOR_CONTEXT] = "context",
21 [GREP_COLOR_FILENAME] = "filename",
22 [GREP_COLOR_FUNCTION] = "function",
23 [GREP_COLOR_LINENO] = "lineNumber",
24 [GREP_COLOR_COLUMNNO] = "column",
25 [GREP_COLOR_MATCH_CONTEXT] = "matchContext",
26 [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
27 [GREP_COLOR_SELECTED] = "selected",
28 [GREP_COLOR_SEP] = "separator",
29 };
30
31 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
32 {
33 fwrite(buf, size, 1, stdout);
34 }
35
36 static void color_set(char *dst, const char *color_bytes)
37 {
38 xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
39 }
40
41 /*
42 * Initialize the grep_defaults template with hardcoded defaults.
43 * We could let the compiler do this, but without C99 initializers
44 * the code gets unwieldy and unreadable, so...
45 */
46 void init_grep_defaults(struct repository *repo)
47 {
48 struct grep_opt *opt = &grep_defaults;
49 static int run_once;
50
51 if (run_once)
52 return;
53 run_once++;
54
55 memset(opt, 0, sizeof(*opt));
56 opt->repo = repo;
57 opt->relative = 1;
58 opt->pathname = 1;
59 opt->max_depth = -1;
60 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
61 color_set(opt->colors[GREP_COLOR_CONTEXT], "");
62 color_set(opt->colors[GREP_COLOR_FILENAME], "");
63 color_set(opt->colors[GREP_COLOR_FUNCTION], "");
64 color_set(opt->colors[GREP_COLOR_LINENO], "");
65 color_set(opt->colors[GREP_COLOR_COLUMNNO], "");
66 color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED);
67 color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED);
68 color_set(opt->colors[GREP_COLOR_SELECTED], "");
69 color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN);
70 opt->only_matching = 0;
71 opt->color = -1;
72 opt->output = std_output;
73 }
74
75 static int parse_pattern_type_arg(const char *opt, const char *arg)
76 {
77 if (!strcmp(arg, "default"))
78 return GREP_PATTERN_TYPE_UNSPECIFIED;
79 else if (!strcmp(arg, "basic"))
80 return GREP_PATTERN_TYPE_BRE;
81 else if (!strcmp(arg, "extended"))
82 return GREP_PATTERN_TYPE_ERE;
83 else if (!strcmp(arg, "fixed"))
84 return GREP_PATTERN_TYPE_FIXED;
85 else if (!strcmp(arg, "perl"))
86 return GREP_PATTERN_TYPE_PCRE;
87 die("bad %s argument: %s", opt, arg);
88 }
89
90 define_list_config_array_extra(color_grep_slots, {"match"});
91
92 /*
93 * Read the configuration file once and store it in
94 * the grep_defaults template.
95 */
96 int grep_config(const char *var, const char *value, void *cb)
97 {
98 struct grep_opt *opt = &grep_defaults;
99 const char *slot;
100
101 if (userdiff_config(var, value) < 0)
102 return -1;
103
104 if (!strcmp(var, "grep.extendedregexp")) {
105 opt->extended_regexp_option = git_config_bool(var, value);
106 return 0;
107 }
108
109 if (!strcmp(var, "grep.patterntype")) {
110 opt->pattern_type_option = parse_pattern_type_arg(var, value);
111 return 0;
112 }
113
114 if (!strcmp(var, "grep.linenumber")) {
115 opt->linenum = git_config_bool(var, value);
116 return 0;
117 }
118 if (!strcmp(var, "grep.column")) {
119 opt->columnnum = git_config_bool(var, value);
120 return 0;
121 }
122
123 if (!strcmp(var, "grep.fullname")) {
124 opt->relative = !git_config_bool(var, value);
125 return 0;
126 }
127
128 if (!strcmp(var, "color.grep"))
129 opt->color = git_config_colorbool(var, value);
130 if (!strcmp(var, "color.grep.match")) {
131 if (grep_config("color.grep.matchcontext", value, cb) < 0)
132 return -1;
133 if (grep_config("color.grep.matchselected", value, cb) < 0)
134 return -1;
135 } else if (skip_prefix(var, "color.grep.", &slot)) {
136 int i = LOOKUP_CONFIG(color_grep_slots, slot);
137 char *color;
138
139 if (i < 0)
140 return -1;
141 color = opt->colors[i];
142 if (!value)
143 return config_error_nonbool(var);
144 return color_parse(value, color);
145 }
146 return 0;
147 }
148
149 /*
150 * Initialize one instance of grep_opt and copy the
151 * default values from the template we read the configuration
152 * information in an earlier call to git_config(grep_config).
153 */
154 void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
155 {
156 struct grep_opt *def = &grep_defaults;
157 int i;
158
159 memset(opt, 0, sizeof(*opt));
160 opt->repo = repo;
161 opt->prefix = prefix;
162 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
163 opt->pattern_tail = &opt->pattern_list;
164 opt->header_tail = &opt->header_list;
165
166 opt->only_matching = def->only_matching;
167 opt->color = def->color;
168 opt->extended_regexp_option = def->extended_regexp_option;
169 opt->pattern_type_option = def->pattern_type_option;
170 opt->linenum = def->linenum;
171 opt->columnnum = def->columnnum;
172 opt->max_depth = def->max_depth;
173 opt->pathname = def->pathname;
174 opt->relative = def->relative;
175 opt->output = def->output;
176
177 for (i = 0; i < NR_GREP_COLORS; i++)
178 color_set(opt->colors[i], def->colors[i]);
179 }
180
181 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
182 {
183 /*
184 * When committing to the pattern type by setting the relevant
185 * fields in grep_opt it's generally not necessary to zero out
186 * the fields we're not choosing, since they won't have been
187 * set by anything. The extended_regexp_option field is the
188 * only exception to this.
189 *
190 * This is because in the process of parsing grep.patternType
191 * & grep.extendedRegexp we set opt->pattern_type_option and
192 * opt->extended_regexp_option, respectively. We then
193 * internally use opt->extended_regexp_option to see if we're
194 * compiling an ERE. It must be unset if that's not actually
195 * the case.
196 */
197 if (pattern_type != GREP_PATTERN_TYPE_ERE &&
198 opt->extended_regexp_option)
199 opt->extended_regexp_option = 0;
200
201 switch (pattern_type) {
202 case GREP_PATTERN_TYPE_UNSPECIFIED:
203 /* fall through */
204
205 case GREP_PATTERN_TYPE_BRE:
206 break;
207
208 case GREP_PATTERN_TYPE_ERE:
209 opt->extended_regexp_option = 1;
210 break;
211
212 case GREP_PATTERN_TYPE_FIXED:
213 opt->fixed = 1;
214 break;
215
216 case GREP_PATTERN_TYPE_PCRE:
217 #ifdef USE_LIBPCRE2
218 opt->pcre2 = 1;
219 #else
220 /*
221 * It's important that pcre1 always be assigned to
222 * even when there's no USE_LIBPCRE* defined. We still
223 * call the PCRE stub function, it just dies with
224 * "cannot use Perl-compatible regexes[...]".
225 */
226 opt->pcre1 = 1;
227 #endif
228 break;
229 }
230 }
231
232 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
233 {
234 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
235 grep_set_pattern_type_option(pattern_type, opt);
236 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
237 grep_set_pattern_type_option(opt->pattern_type_option, opt);
238 else if (opt->extended_regexp_option)
239 /*
240 * This branch *must* happen after setting from the
241 * opt->pattern_type_option above, we don't want
242 * grep.extendedRegexp to override grep.patternType!
243 */
244 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
245 }
246
247 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
248 const char *origin, int no,
249 enum grep_pat_token t,
250 enum grep_header_field field)
251 {
252 struct grep_pat *p = xcalloc(1, sizeof(*p));
253 p->pattern = xmemdupz(pat, patlen);
254 p->patternlen = patlen;
255 p->origin = origin;
256 p->no = no;
257 p->token = t;
258 p->field = field;
259 return p;
260 }
261
262 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
263 {
264 **tail = p;
265 *tail = &p->next;
266 p->next = NULL;
267
268 switch (p->token) {
269 case GREP_PATTERN: /* atom */
270 case GREP_PATTERN_HEAD:
271 case GREP_PATTERN_BODY:
272 for (;;) {
273 struct grep_pat *new_pat;
274 size_t len = 0;
275 char *cp = p->pattern + p->patternlen, *nl = NULL;
276 while (++len <= p->patternlen) {
277 if (*(--cp) == '\n') {
278 nl = cp;
279 break;
280 }
281 }
282 if (!nl)
283 break;
284 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
285 p->no, p->token, p->field);
286 new_pat->next = p->next;
287 if (!p->next)
288 *tail = &new_pat->next;
289 p->next = new_pat;
290 *nl = '\0';
291 p->patternlen -= len;
292 }
293 break;
294 default:
295 break;
296 }
297 }
298
299 void append_header_grep_pattern(struct grep_opt *opt,
300 enum grep_header_field field, const char *pat)
301 {
302 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
303 GREP_PATTERN_HEAD, field);
304 if (field == GREP_HEADER_REFLOG)
305 opt->use_reflog_filter = 1;
306 do_append_grep_pat(&opt->header_tail, p);
307 }
308
309 void append_grep_pattern(struct grep_opt *opt, const char *pat,
310 const char *origin, int no, enum grep_pat_token t)
311 {
312 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
313 }
314
315 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
316 const char *origin, int no, enum grep_pat_token t)
317 {
318 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
319 do_append_grep_pat(&opt->pattern_tail, p);
320 }
321
322 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
323 {
324 struct grep_pat *pat;
325 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
326 *ret = *opt;
327
328 ret->pattern_list = NULL;
329 ret->pattern_tail = &ret->pattern_list;
330
331 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
332 {
333 if(pat->token == GREP_PATTERN_HEAD)
334 append_header_grep_pattern(ret, pat->field,
335 pat->pattern);
336 else
337 append_grep_pat(ret, pat->pattern, pat->patternlen,
338 pat->origin, pat->no, pat->token);
339 }
340
341 return ret;
342 }
343
344 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
345 const char *error)
346 {
347 char where[1024];
348
349 if (p->no)
350 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
351 else if (p->origin)
352 xsnprintf(where, sizeof(where), "%s, ", p->origin);
353 else
354 where[0] = 0;
355
356 die("%s'%s': %s", where, p->pattern, error);
357 }
358
359 static int is_fixed(const char *s, size_t len)
360 {
361 size_t i;
362
363 for (i = 0; i < len; i++) {
364 if (is_regex_special(s[i]))
365 return 0;
366 }
367
368 return 1;
369 }
370
371 #ifdef USE_LIBPCRE1
372 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
373 {
374 const char *error;
375 int erroffset;
376 int options = PCRE_MULTILINE;
377 int study_options = 0;
378
379 if (opt->ignore_case) {
380 if (!opt->ignore_locale && has_non_ascii(p->pattern))
381 p->pcre1_tables = pcre_maketables();
382 options |= PCRE_CASELESS;
383 }
384 if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern))
385 options |= PCRE_UTF8;
386
387 p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
388 p->pcre1_tables);
389 if (!p->pcre1_regexp)
390 compile_regexp_failed(p, error);
391
392 #if defined(PCRE_CONFIG_JIT) && !defined(NO_LIBPCRE1_JIT)
393 pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
394 if (opt->debug)
395 fprintf(stderr, "pcre1_jit_on=%d\n", p->pcre1_jit_on);
396
397 if (p->pcre1_jit_on)
398 study_options = PCRE_STUDY_JIT_COMPILE;
399 #endif
400
401 p->pcre1_extra_info = pcre_study(p->pcre1_regexp, study_options, &error);
402 if (!p->pcre1_extra_info && error)
403 die("%s", error);
404 }
405
406 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
407 regmatch_t *match, int eflags)
408 {
409 int ovector[30], ret, flags = PCRE_NO_UTF8_CHECK;
410
411 if (eflags & REG_NOTBOL)
412 flags |= PCRE_NOTBOL;
413
414 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
415 eol - line, 0, flags, ovector,
416 ARRAY_SIZE(ovector));
417
418 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
419 die("pcre_exec failed with error code %d", ret);
420 if (ret > 0) {
421 ret = 0;
422 match->rm_so = ovector[0];
423 match->rm_eo = ovector[1];
424 }
425
426 return ret;
427 }
428
429 static void free_pcre1_regexp(struct grep_pat *p)
430 {
431 pcre_free(p->pcre1_regexp);
432 #ifdef PCRE_CONFIG_JIT
433 if (p->pcre1_jit_on)
434 pcre_free_study(p->pcre1_extra_info);
435 else
436 #endif
437 pcre_free(p->pcre1_extra_info);
438 pcre_free((void *)p->pcre1_tables);
439 }
440 #else /* !USE_LIBPCRE1 */
441 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
442 {
443 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
444 }
445
446 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
447 regmatch_t *match, int eflags)
448 {
449 return 1;
450 }
451
452 static void free_pcre1_regexp(struct grep_pat *p)
453 {
454 }
455 #endif /* !USE_LIBPCRE1 */
456
457 #ifdef USE_LIBPCRE2
458 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
459 {
460 int error;
461 PCRE2_UCHAR errbuf[256];
462 PCRE2_SIZE erroffset;
463 int options = PCRE2_MULTILINE;
464 const uint8_t *character_tables = NULL;
465 int jitret;
466 int patinforet;
467 size_t jitsizearg;
468
469 assert(opt->pcre2);
470
471 p->pcre2_compile_context = NULL;
472
473 if (opt->ignore_case) {
474 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
475 character_tables = pcre2_maketables(NULL);
476 p->pcre2_compile_context = pcre2_compile_context_create(NULL);
477 pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
478 }
479 options |= PCRE2_CASELESS;
480 }
481 if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern) &&
482 !(!opt->ignore_case && (p->fixed || p->is_fixed)))
483 options |= PCRE2_UTF;
484
485 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
486 p->patternlen, options, &error, &erroffset,
487 p->pcre2_compile_context);
488
489 if (p->pcre2_pattern) {
490 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
491 if (!p->pcre2_match_data)
492 die("Couldn't allocate PCRE2 match data");
493 } else {
494 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
495 compile_regexp_failed(p, (const char *)&errbuf);
496 }
497
498 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
499 if (opt->debug)
500 fprintf(stderr, "pcre2_jit_on=%d\n", p->pcre2_jit_on);
501 if (p->pcre2_jit_on) {
502 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
503 if (jitret)
504 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
505
506 /*
507 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
508 * tells us whether the library itself supports JIT,
509 * but to see whether we're going to be actually using
510 * JIT we need to extract PCRE2_INFO_JITSIZE from the
511 * pattern *after* we do pcre2_jit_compile() above.
512 *
513 * This is because if the pattern contains the
514 * (*NO_JIT) verb (see pcre2syntax(3))
515 * pcre2_jit_compile() will exit early with 0. If we
516 * then proceed to call pcre2_jit_match() further down
517 * the line instead of pcre2_match() we'll either
518 * segfault (pre PCRE 10.31) or run into a fatal error
519 * (post PCRE2 10.31)
520 */
521 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
522 if (patinforet)
523 BUG("pcre2_pattern_info() failed: %d", patinforet);
524 if (jitsizearg == 0) {
525 p->pcre2_jit_on = 0;
526 if (opt->debug)
527 fprintf(stderr, "pcre2_jit_on=%d: (*NO_JIT) in regex\n",
528 p->pcre2_jit_on);
529 return;
530 }
531 }
532 }
533
534 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
535 regmatch_t *match, int eflags)
536 {
537 int ret, flags = 0;
538 PCRE2_SIZE *ovector;
539 PCRE2_UCHAR errbuf[256];
540
541 if (eflags & REG_NOTBOL)
542 flags |= PCRE2_NOTBOL;
543
544 if (p->pcre2_jit_on)
545 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
546 eol - line, 0, flags, p->pcre2_match_data,
547 NULL);
548 else
549 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
550 eol - line, 0, flags, p->pcre2_match_data,
551 NULL);
552
553 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
554 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
555 die("%s failed with error code %d: %s",
556 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
557 errbuf);
558 }
559 if (ret > 0) {
560 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
561 ret = 0;
562 match->rm_so = (int)ovector[0];
563 match->rm_eo = (int)ovector[1];
564 }
565
566 return ret;
567 }
568
569 static void free_pcre2_pattern(struct grep_pat *p)
570 {
571 pcre2_compile_context_free(p->pcre2_compile_context);
572 pcre2_code_free(p->pcre2_pattern);
573 pcre2_match_data_free(p->pcre2_match_data);
574 }
575 #else /* !USE_LIBPCRE2 */
576 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
577 {
578 /*
579 * Unreachable until USE_LIBPCRE2 becomes synonymous with
580 * USE_LIBPCRE. See the sibling comment in
581 * grep_set_pattern_type_option().
582 */
583 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
584 }
585
586 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
587 regmatch_t *match, int eflags)
588 {
589 return 1;
590 }
591
592 static void free_pcre2_pattern(struct grep_pat *p)
593 {
594 }
595
596 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
597 {
598 struct strbuf sb = STRBUF_INIT;
599 int err;
600 int regflags = 0;
601
602 basic_regex_quote_buf(&sb, p->pattern);
603 if (opt->ignore_case)
604 regflags |= REG_ICASE;
605 err = regcomp(&p->regexp, sb.buf, regflags);
606 if (opt->debug)
607 fprintf(stderr, "fixed %s\n", sb.buf);
608 strbuf_release(&sb);
609 if (err) {
610 char errbuf[1024];
611 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
612 compile_regexp_failed(p, errbuf);
613 }
614 }
615 #endif /* !USE_LIBPCRE2 */
616
617 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
618 {
619 int err;
620 int regflags = REG_NEWLINE;
621
622 p->word_regexp = opt->word_regexp;
623 p->ignore_case = opt->ignore_case;
624 p->fixed = opt->fixed;
625
626 if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2)
627 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
628
629 p->is_fixed = is_fixed(p->pattern, p->patternlen);
630 #ifdef USE_LIBPCRE2
631 if (!p->fixed && !p->is_fixed) {
632 const char *no_jit = "(*NO_JIT)";
633 const int no_jit_len = strlen(no_jit);
634 if (starts_with(p->pattern, no_jit) &&
635 is_fixed(p->pattern + no_jit_len,
636 p->patternlen - no_jit_len))
637 p->is_fixed = 1;
638 }
639 #endif
640 if (p->fixed || p->is_fixed) {
641 #ifdef USE_LIBPCRE2
642 opt->pcre2 = 1;
643 if (p->is_fixed) {
644 compile_pcre2_pattern(p, opt);
645 } else {
646 /*
647 * E.g. t7811-grep-open.sh relies on the
648 * pattern being restored.
649 */
650 char *old_pattern = p->pattern;
651 size_t old_patternlen = p->patternlen;
652 struct strbuf sb = STRBUF_INIT;
653
654 /*
655 * There is the PCRE2_LITERAL flag, but it's
656 * only in PCRE v2 10.30 and later. Needing to
657 * ifdef our way around that and dealing with
658 * it + PCRE2_MULTILINE being an error is more
659 * complex than just quoting this ourselves.
660 */
661 strbuf_add(&sb, "\\Q", 2);
662 strbuf_add(&sb, p->pattern, p->patternlen);
663 strbuf_add(&sb, "\\E", 2);
664
665 p->pattern = sb.buf;
666 p->patternlen = sb.len;
667 compile_pcre2_pattern(p, opt);
668 p->pattern = old_pattern;
669 p->patternlen = old_patternlen;
670 strbuf_release(&sb);
671 }
672 #else /* !USE_LIBPCRE2 */
673 compile_fixed_regexp(p, opt);
674 #endif /* !USE_LIBPCRE2 */
675 return;
676 }
677
678 if (opt->pcre2) {
679 compile_pcre2_pattern(p, opt);
680 return;
681 }
682
683 if (opt->pcre1) {
684 compile_pcre1_regexp(p, opt);
685 return;
686 }
687
688 if (p->ignore_case)
689 regflags |= REG_ICASE;
690 if (opt->extended_regexp_option)
691 regflags |= REG_EXTENDED;
692 err = regcomp(&p->regexp, p->pattern, regflags);
693 if (err) {
694 char errbuf[1024];
695 regerror(err, &p->regexp, errbuf, 1024);
696 compile_regexp_failed(p, errbuf);
697 }
698 }
699
700 static struct grep_expr *compile_pattern_or(struct grep_pat **);
701 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
702 {
703 struct grep_pat *p;
704 struct grep_expr *x;
705
706 p = *list;
707 if (!p)
708 return NULL;
709 switch (p->token) {
710 case GREP_PATTERN: /* atom */
711 case GREP_PATTERN_HEAD:
712 case GREP_PATTERN_BODY:
713 x = xcalloc(1, sizeof (struct grep_expr));
714 x->node = GREP_NODE_ATOM;
715 x->u.atom = p;
716 *list = p->next;
717 return x;
718 case GREP_OPEN_PAREN:
719 *list = p->next;
720 x = compile_pattern_or(list);
721 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
722 die("unmatched parenthesis");
723 *list = (*list)->next;
724 return x;
725 default:
726 return NULL;
727 }
728 }
729
730 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
731 {
732 struct grep_pat *p;
733 struct grep_expr *x;
734
735 p = *list;
736 if (!p)
737 return NULL;
738 switch (p->token) {
739 case GREP_NOT:
740 if (!p->next)
741 die("--not not followed by pattern expression");
742 *list = p->next;
743 x = xcalloc(1, sizeof (struct grep_expr));
744 x->node = GREP_NODE_NOT;
745 x->u.unary = compile_pattern_not(list);
746 if (!x->u.unary)
747 die("--not followed by non pattern expression");
748 return x;
749 default:
750 return compile_pattern_atom(list);
751 }
752 }
753
754 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
755 {
756 struct grep_pat *p;
757 struct grep_expr *x, *y, *z;
758
759 x = compile_pattern_not(list);
760 p = *list;
761 if (p && p->token == GREP_AND) {
762 if (!p->next)
763 die("--and not followed by pattern expression");
764 *list = p->next;
765 y = compile_pattern_and(list);
766 if (!y)
767 die("--and not followed by pattern expression");
768 z = xcalloc(1, sizeof (struct grep_expr));
769 z->node = GREP_NODE_AND;
770 z->u.binary.left = x;
771 z->u.binary.right = y;
772 return z;
773 }
774 return x;
775 }
776
777 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
778 {
779 struct grep_pat *p;
780 struct grep_expr *x, *y, *z;
781
782 x = compile_pattern_and(list);
783 p = *list;
784 if (x && p && p->token != GREP_CLOSE_PAREN) {
785 y = compile_pattern_or(list);
786 if (!y)
787 die("not a pattern expression %s", p->pattern);
788 z = xcalloc(1, sizeof (struct grep_expr));
789 z->node = GREP_NODE_OR;
790 z->u.binary.left = x;
791 z->u.binary.right = y;
792 return z;
793 }
794 return x;
795 }
796
797 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
798 {
799 return compile_pattern_or(list);
800 }
801
802 static void indent(int in)
803 {
804 while (in-- > 0)
805 fputc(' ', stderr);
806 }
807
808 static void dump_grep_pat(struct grep_pat *p)
809 {
810 switch (p->token) {
811 case GREP_AND: fprintf(stderr, "*and*"); break;
812 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
813 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
814 case GREP_NOT: fprintf(stderr, "*not*"); break;
815 case GREP_OR: fprintf(stderr, "*or*"); break;
816
817 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
818 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
819 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
820 }
821
822 switch (p->token) {
823 default: break;
824 case GREP_PATTERN_HEAD:
825 fprintf(stderr, "<head %d>", p->field); break;
826 case GREP_PATTERN_BODY:
827 fprintf(stderr, "<body>"); break;
828 }
829 switch (p->token) {
830 default: break;
831 case GREP_PATTERN_HEAD:
832 case GREP_PATTERN_BODY:
833 case GREP_PATTERN:
834 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
835 break;
836 }
837 fputc('\n', stderr);
838 }
839
840 static void dump_grep_expression_1(struct grep_expr *x, int in)
841 {
842 indent(in);
843 switch (x->node) {
844 case GREP_NODE_TRUE:
845 fprintf(stderr, "true\n");
846 break;
847 case GREP_NODE_ATOM:
848 dump_grep_pat(x->u.atom);
849 break;
850 case GREP_NODE_NOT:
851 fprintf(stderr, "(not\n");
852 dump_grep_expression_1(x->u.unary, in+1);
853 indent(in);
854 fprintf(stderr, ")\n");
855 break;
856 case GREP_NODE_AND:
857 fprintf(stderr, "(and\n");
858 dump_grep_expression_1(x->u.binary.left, in+1);
859 dump_grep_expression_1(x->u.binary.right, in+1);
860 indent(in);
861 fprintf(stderr, ")\n");
862 break;
863 case GREP_NODE_OR:
864 fprintf(stderr, "(or\n");
865 dump_grep_expression_1(x->u.binary.left, in+1);
866 dump_grep_expression_1(x->u.binary.right, in+1);
867 indent(in);
868 fprintf(stderr, ")\n");
869 break;
870 }
871 }
872
873 static void dump_grep_expression(struct grep_opt *opt)
874 {
875 struct grep_expr *x = opt->pattern_expression;
876
877 if (opt->all_match)
878 fprintf(stderr, "[all-match]\n");
879 dump_grep_expression_1(x, 0);
880 fflush(NULL);
881 }
882
883 static struct grep_expr *grep_true_expr(void)
884 {
885 struct grep_expr *z = xcalloc(1, sizeof(*z));
886 z->node = GREP_NODE_TRUE;
887 return z;
888 }
889
890 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
891 {
892 struct grep_expr *z = xcalloc(1, sizeof(*z));
893 z->node = GREP_NODE_OR;
894 z->u.binary.left = left;
895 z->u.binary.right = right;
896 return z;
897 }
898
899 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
900 {
901 struct grep_pat *p;
902 struct grep_expr *header_expr;
903 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
904 enum grep_header_field fld;
905
906 if (!opt->header_list)
907 return NULL;
908
909 for (p = opt->header_list; p; p = p->next) {
910 if (p->token != GREP_PATTERN_HEAD)
911 BUG("a non-header pattern in grep header list.");
912 if (p->field < GREP_HEADER_FIELD_MIN ||
913 GREP_HEADER_FIELD_MAX <= p->field)
914 BUG("unknown header field %d", p->field);
915 compile_regexp(p, opt);
916 }
917
918 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
919 header_group[fld] = NULL;
920
921 for (p = opt->header_list; p; p = p->next) {
922 struct grep_expr *h;
923 struct grep_pat *pp = p;
924
925 h = compile_pattern_atom(&pp);
926 if (!h || pp != p->next)
927 BUG("malformed header expr");
928 if (!header_group[p->field]) {
929 header_group[p->field] = h;
930 continue;
931 }
932 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
933 }
934
935 header_expr = NULL;
936
937 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
938 if (!header_group[fld])
939 continue;
940 if (!header_expr)
941 header_expr = grep_true_expr();
942 header_expr = grep_or_expr(header_group[fld], header_expr);
943 }
944 return header_expr;
945 }
946
947 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
948 {
949 struct grep_expr *z = x;
950
951 while (x) {
952 assert(x->node == GREP_NODE_OR);
953 if (x->u.binary.right &&
954 x->u.binary.right->node == GREP_NODE_TRUE) {
955 x->u.binary.right = y;
956 break;
957 }
958 x = x->u.binary.right;
959 }
960 return z;
961 }
962
963 static void compile_grep_patterns_real(struct grep_opt *opt)
964 {
965 struct grep_pat *p;
966 struct grep_expr *header_expr = prep_header_patterns(opt);
967
968 for (p = opt->pattern_list; p; p = p->next) {
969 switch (p->token) {
970 case GREP_PATTERN: /* atom */
971 case GREP_PATTERN_HEAD:
972 case GREP_PATTERN_BODY:
973 compile_regexp(p, opt);
974 break;
975 default:
976 opt->extended = 1;
977 break;
978 }
979 }
980
981 if (opt->all_match || header_expr)
982 opt->extended = 1;
983 else if (!opt->extended && !opt->debug)
984 return;
985
986 p = opt->pattern_list;
987 if (p)
988 opt->pattern_expression = compile_pattern_expr(&p);
989 if (p)
990 die("incomplete pattern expression: %s", p->pattern);
991
992 if (!header_expr)
993 return;
994
995 if (!opt->pattern_expression)
996 opt->pattern_expression = header_expr;
997 else if (opt->all_match)
998 opt->pattern_expression = grep_splice_or(header_expr,
999 opt->pattern_expression);
1000 else
1001 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
1002 header_expr);
1003 opt->all_match = 1;
1004 }
1005
1006 void compile_grep_patterns(struct grep_opt *opt)
1007 {
1008 compile_grep_patterns_real(opt);
1009 if (opt->debug)
1010 dump_grep_expression(opt);
1011 }
1012
1013 static void free_pattern_expr(struct grep_expr *x)
1014 {
1015 switch (x->node) {
1016 case GREP_NODE_TRUE:
1017 case GREP_NODE_ATOM:
1018 break;
1019 case GREP_NODE_NOT:
1020 free_pattern_expr(x->u.unary);
1021 break;
1022 case GREP_NODE_AND:
1023 case GREP_NODE_OR:
1024 free_pattern_expr(x->u.binary.left);
1025 free_pattern_expr(x->u.binary.right);
1026 break;
1027 }
1028 free(x);
1029 }
1030
1031 void free_grep_patterns(struct grep_opt *opt)
1032 {
1033 struct grep_pat *p, *n;
1034
1035 for (p = opt->pattern_list; p; p = n) {
1036 n = p->next;
1037 switch (p->token) {
1038 case GREP_PATTERN: /* atom */
1039 case GREP_PATTERN_HEAD:
1040 case GREP_PATTERN_BODY:
1041 if (p->pcre1_regexp)
1042 free_pcre1_regexp(p);
1043 else if (p->pcre2_pattern)
1044 free_pcre2_pattern(p);
1045 else
1046 regfree(&p->regexp);
1047 free(p->pattern);
1048 break;
1049 default:
1050 break;
1051 }
1052 free(p);
1053 }
1054
1055 if (!opt->extended)
1056 return;
1057 free_pattern_expr(opt->pattern_expression);
1058 }
1059
1060 static char *end_of_line(char *cp, unsigned long *left)
1061 {
1062 unsigned long l = *left;
1063 while (l && *cp != '\n') {
1064 l--;
1065 cp++;
1066 }
1067 *left = l;
1068 return cp;
1069 }
1070
1071 static int word_char(char ch)
1072 {
1073 return isalnum(ch) || ch == '_';
1074 }
1075
1076 static void output_color(struct grep_opt *opt, const void *data, size_t size,
1077 const char *color)
1078 {
1079 if (want_color(opt->color) && color && color[0]) {
1080 opt->output(opt, color, strlen(color));
1081 opt->output(opt, data, size);
1082 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1083 } else
1084 opt->output(opt, data, size);
1085 }
1086
1087 static void output_sep(struct grep_opt *opt, char sign)
1088 {
1089 if (opt->null_following_name)
1090 opt->output(opt, "\0", 1);
1091 else
1092 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
1093 }
1094
1095 static void show_name(struct grep_opt *opt, const char *name)
1096 {
1097 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1098 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1099 }
1100
1101 static int patmatch(struct grep_pat *p, char *line, char *eol,
1102 regmatch_t *match, int eflags)
1103 {
1104 int hit;
1105
1106 if (p->pcre1_regexp)
1107 hit = !pcre1match(p, line, eol, match, eflags);
1108 else if (p->pcre2_pattern)
1109 hit = !pcre2match(p, line, eol, match, eflags);
1110 else
1111 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1112 eflags);
1113
1114 return hit;
1115 }
1116
1117 static int strip_timestamp(char *bol, char **eol_p)
1118 {
1119 char *eol = *eol_p;
1120 int ch;
1121
1122 while (bol < --eol) {
1123 if (*eol != '>')
1124 continue;
1125 *eol_p = ++eol;
1126 ch = *eol;
1127 *eol = '\0';
1128 return ch;
1129 }
1130 return 0;
1131 }
1132
1133 static struct {
1134 const char *field;
1135 size_t len;
1136 } header_field[] = {
1137 { "author ", 7 },
1138 { "committer ", 10 },
1139 { "reflog ", 7 },
1140 };
1141
1142 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1143 enum grep_context ctx,
1144 regmatch_t *pmatch, int eflags)
1145 {
1146 int hit = 0;
1147 int saved_ch = 0;
1148 const char *start = bol;
1149
1150 if ((p->token != GREP_PATTERN) &&
1151 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1152 return 0;
1153
1154 if (p->token == GREP_PATTERN_HEAD) {
1155 const char *field;
1156 size_t len;
1157 assert(p->field < ARRAY_SIZE(header_field));
1158 field = header_field[p->field].field;
1159 len = header_field[p->field].len;
1160 if (strncmp(bol, field, len))
1161 return 0;
1162 bol += len;
1163 switch (p->field) {
1164 case GREP_HEADER_AUTHOR:
1165 case GREP_HEADER_COMMITTER:
1166 saved_ch = strip_timestamp(bol, &eol);
1167 break;
1168 default:
1169 break;
1170 }
1171 }
1172
1173 again:
1174 hit = patmatch(p, bol, eol, pmatch, eflags);
1175
1176 if (hit && p->word_regexp) {
1177 if ((pmatch[0].rm_so < 0) ||
1178 (eol - bol) < pmatch[0].rm_so ||
1179 (pmatch[0].rm_eo < 0) ||
1180 (eol - bol) < pmatch[0].rm_eo)
1181 die("regexp returned nonsense");
1182
1183 /* Match beginning must be either beginning of the
1184 * line, or at word boundary (i.e. the last char must
1185 * not be a word char). Similarly, match end must be
1186 * either end of the line, or at word boundary
1187 * (i.e. the next char must not be a word char).
1188 */
1189 if ( ((pmatch[0].rm_so == 0) ||
1190 !word_char(bol[pmatch[0].rm_so-1])) &&
1191 ((pmatch[0].rm_eo == (eol-bol)) ||
1192 !word_char(bol[pmatch[0].rm_eo])) )
1193 ;
1194 else
1195 hit = 0;
1196
1197 /* Words consist of at least one character. */
1198 if (pmatch->rm_so == pmatch->rm_eo)
1199 hit = 0;
1200
1201 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1202 /* There could be more than one match on the
1203 * line, and the first match might not be
1204 * strict word match. But later ones could be!
1205 * Forward to the next possible start, i.e. the
1206 * next position following a non-word char.
1207 */
1208 bol = pmatch[0].rm_so + bol + 1;
1209 while (word_char(bol[-1]) && bol < eol)
1210 bol++;
1211 eflags |= REG_NOTBOL;
1212 if (bol < eol)
1213 goto again;
1214 }
1215 }
1216 if (p->token == GREP_PATTERN_HEAD && saved_ch)
1217 *eol = saved_ch;
1218 if (hit) {
1219 pmatch[0].rm_so += bol - start;
1220 pmatch[0].rm_eo += bol - start;
1221 }
1222 return hit;
1223 }
1224
1225 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol,
1226 char *eol, enum grep_context ctx, ssize_t *col,
1227 ssize_t *icol, int collect_hits)
1228 {
1229 int h = 0;
1230
1231 if (!x)
1232 die("Not a valid grep expression");
1233 switch (x->node) {
1234 case GREP_NODE_TRUE:
1235 h = 1;
1236 break;
1237 case GREP_NODE_ATOM:
1238 {
1239 regmatch_t tmp;
1240 h = match_one_pattern(x->u.atom, bol, eol, ctx,
1241 &tmp, 0);
1242 if (h && (*col < 0 || tmp.rm_so < *col))
1243 *col = tmp.rm_so;
1244 }
1245 break;
1246 case GREP_NODE_NOT:
1247 /*
1248 * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1249 */
1250 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1251 0);
1252 break;
1253 case GREP_NODE_AND:
1254 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1255 icol, 0);
1256 if (h || opt->columnnum) {
1257 /*
1258 * Don't short-circuit AND when given --column, since a
1259 * NOT earlier in the tree may turn this into an OR. In
1260 * this case, see the below comment.
1261 */
1262 h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1263 ctx, col, icol, 0);
1264 }
1265 break;
1266 case GREP_NODE_OR:
1267 if (!(collect_hits || opt->columnnum)) {
1268 /*
1269 * Don't short-circuit OR when given --column (or
1270 * collecting hits) to ensure we don't skip a later
1271 * child that would produce an earlier match.
1272 */
1273 return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1274 ctx, col, icol, 0) ||
1275 match_expr_eval(opt, x->u.binary.right, bol,
1276 eol, ctx, col, icol, 0));
1277 }
1278 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1279 icol, 0);
1280 if (collect_hits)
1281 x->u.binary.left->hit |= h;
1282 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1283 icol, collect_hits);
1284 break;
1285 default:
1286 die("Unexpected node type (internal error) %d", x->node);
1287 }
1288 if (collect_hits)
1289 x->hit |= h;
1290 return h;
1291 }
1292
1293 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1294 enum grep_context ctx, ssize_t *col,
1295 ssize_t *icol, int collect_hits)
1296 {
1297 struct grep_expr *x = opt->pattern_expression;
1298 return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1299 }
1300
1301 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1302 ssize_t *col, ssize_t *icol,
1303 enum grep_context ctx, int collect_hits)
1304 {
1305 struct grep_pat *p;
1306 int hit = 0;
1307
1308 if (opt->extended)
1309 return match_expr(opt, bol, eol, ctx, col, icol,
1310 collect_hits);
1311
1312 /* we do not call with collect_hits without being extended */
1313 for (p = opt->pattern_list; p; p = p->next) {
1314 regmatch_t tmp;
1315 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1316 hit |= 1;
1317 if (!opt->columnnum) {
1318 /*
1319 * Without --column, any single match on a line
1320 * is enough to know that it needs to be
1321 * printed. With --column, scan _all_ patterns
1322 * to find the earliest.
1323 */
1324 break;
1325 }
1326 if (*col < 0 || tmp.rm_so < *col)
1327 *col = tmp.rm_so;
1328 }
1329 }
1330 return hit;
1331 }
1332
1333 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1334 enum grep_context ctx,
1335 regmatch_t *pmatch, int eflags)
1336 {
1337 regmatch_t match;
1338
1339 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1340 return 0;
1341 if (match.rm_so < 0 || match.rm_eo < 0)
1342 return 0;
1343 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1344 if (match.rm_so > pmatch->rm_so)
1345 return 1;
1346 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1347 return 1;
1348 }
1349 pmatch->rm_so = match.rm_so;
1350 pmatch->rm_eo = match.rm_eo;
1351 return 1;
1352 }
1353
1354 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1355 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1356 {
1357 struct grep_pat *p;
1358 int hit = 0;
1359
1360 pmatch->rm_so = pmatch->rm_eo = -1;
1361 if (bol < eol) {
1362 for (p = opt->pattern_list; p; p = p->next) {
1363 switch (p->token) {
1364 case GREP_PATTERN: /* atom */
1365 case GREP_PATTERN_HEAD:
1366 case GREP_PATTERN_BODY:
1367 hit |= match_next_pattern(p, bol, eol, ctx,
1368 pmatch, eflags);
1369 break;
1370 default:
1371 break;
1372 }
1373 }
1374 }
1375 return hit;
1376 }
1377
1378 static void show_line_header(struct grep_opt *opt, const char *name,
1379 unsigned lno, ssize_t cno, char sign)
1380 {
1381 if (opt->heading && opt->last_shown == 0) {
1382 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1383 opt->output(opt, "\n", 1);
1384 }
1385 opt->last_shown = lno;
1386
1387 if (!opt->heading && opt->pathname) {
1388 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1389 output_sep(opt, sign);
1390 }
1391 if (opt->linenum) {
1392 char buf[32];
1393 xsnprintf(buf, sizeof(buf), "%d", lno);
1394 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1395 output_sep(opt, sign);
1396 }
1397 /*
1398 * Treat 'cno' as the 1-indexed offset from the start of a non-context
1399 * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1400 * being called with a context line.
1401 */
1402 if (opt->columnnum && cno) {
1403 char buf[32];
1404 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1405 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1406 output_sep(opt, sign);
1407 }
1408 }
1409
1410 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1411 const char *name, unsigned lno, ssize_t cno, char sign)
1412 {
1413 int rest = eol - bol;
1414 const char *match_color = NULL;
1415 const char *line_color = NULL;
1416
1417 if (opt->file_break && opt->last_shown == 0) {
1418 if (opt->show_hunk_mark)
1419 opt->output(opt, "\n", 1);
1420 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1421 if (opt->last_shown == 0) {
1422 if (opt->show_hunk_mark) {
1423 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1424 opt->output(opt, "\n", 1);
1425 }
1426 } else if (lno > opt->last_shown + 1) {
1427 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1428 opt->output(opt, "\n", 1);
1429 }
1430 }
1431 if (!opt->only_matching) {
1432 /*
1433 * In case the line we're being called with contains more than
1434 * one match, leave printing each header to the loop below.
1435 */
1436 show_line_header(opt, name, lno, cno, sign);
1437 }
1438 if (opt->color || opt->only_matching) {
1439 regmatch_t match;
1440 enum grep_context ctx = GREP_CONTEXT_BODY;
1441 int ch = *eol;
1442 int eflags = 0;
1443
1444 if (opt->color) {
1445 if (sign == ':')
1446 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1447 else
1448 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1449 if (sign == ':')
1450 line_color = opt->colors[GREP_COLOR_SELECTED];
1451 else if (sign == '-')
1452 line_color = opt->colors[GREP_COLOR_CONTEXT];
1453 else if (sign == '=')
1454 line_color = opt->colors[GREP_COLOR_FUNCTION];
1455 }
1456 *eol = '\0';
1457 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1458 if (match.rm_so == match.rm_eo)
1459 break;
1460
1461 if (opt->only_matching)
1462 show_line_header(opt, name, lno, cno, sign);
1463 else
1464 output_color(opt, bol, match.rm_so, line_color);
1465 output_color(opt, bol + match.rm_so,
1466 match.rm_eo - match.rm_so, match_color);
1467 if (opt->only_matching)
1468 opt->output(opt, "\n", 1);
1469 bol += match.rm_eo;
1470 cno += match.rm_eo;
1471 rest -= match.rm_eo;
1472 eflags = REG_NOTBOL;
1473 }
1474 *eol = ch;
1475 }
1476 if (!opt->only_matching) {
1477 output_color(opt, bol, rest, line_color);
1478 opt->output(opt, "\n", 1);
1479 }
1480 }
1481
1482 int grep_use_locks;
1483
1484 /*
1485 * This lock protects access to the gitattributes machinery, which is
1486 * not thread-safe.
1487 */
1488 pthread_mutex_t grep_attr_mutex;
1489
1490 static inline void grep_attr_lock(void)
1491 {
1492 if (grep_use_locks)
1493 pthread_mutex_lock(&grep_attr_mutex);
1494 }
1495
1496 static inline void grep_attr_unlock(void)
1497 {
1498 if (grep_use_locks)
1499 pthread_mutex_unlock(&grep_attr_mutex);
1500 }
1501
1502 /*
1503 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1504 */
1505 pthread_mutex_t grep_read_mutex;
1506
1507 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1508 {
1509 xdemitconf_t *xecfg = opt->priv;
1510 if (xecfg && !xecfg->find_func) {
1511 grep_source_load_driver(gs, opt->repo->index);
1512 if (gs->driver->funcname.pattern) {
1513 const struct userdiff_funcname *pe = &gs->driver->funcname;
1514 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1515 } else {
1516 xecfg = opt->priv = NULL;
1517 }
1518 }
1519
1520 if (xecfg) {
1521 char buf[1];
1522 return xecfg->find_func(bol, eol - bol, buf, 1,
1523 xecfg->find_func_priv) >= 0;
1524 }
1525
1526 if (bol == eol)
1527 return 0;
1528 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1529 return 1;
1530 return 0;
1531 }
1532
1533 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1534 char *bol, unsigned lno)
1535 {
1536 while (bol > gs->buf) {
1537 char *eol = --bol;
1538
1539 while (bol > gs->buf && bol[-1] != '\n')
1540 bol--;
1541 lno--;
1542
1543 if (lno <= opt->last_shown)
1544 break;
1545
1546 if (match_funcname(opt, gs, bol, eol)) {
1547 show_line(opt, bol, eol, gs->name, lno, 0, '=');
1548 break;
1549 }
1550 }
1551 }
1552
1553 static int is_empty_line(const char *bol, const char *eol);
1554
1555 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1556 char *bol, char *end, unsigned lno)
1557 {
1558 unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1559 int funcname_needed = !!opt->funcname, comment_needed = 0;
1560
1561 if (opt->pre_context < lno)
1562 from = lno - opt->pre_context;
1563 if (from <= opt->last_shown)
1564 from = opt->last_shown + 1;
1565 orig_from = from;
1566 if (opt->funcbody) {
1567 if (match_funcname(opt, gs, bol, end))
1568 comment_needed = 1;
1569 else
1570 funcname_needed = 1;
1571 from = opt->last_shown + 1;
1572 }
1573
1574 /* Rewind. */
1575 while (bol > gs->buf && cur > from) {
1576 char *next_bol = bol;
1577 char *eol = --bol;
1578
1579 while (bol > gs->buf && bol[-1] != '\n')
1580 bol--;
1581 cur--;
1582 if (comment_needed && (is_empty_line(bol, eol) ||
1583 match_funcname(opt, gs, bol, eol))) {
1584 comment_needed = 0;
1585 from = orig_from;
1586 if (cur < from) {
1587 cur++;
1588 bol = next_bol;
1589 break;
1590 }
1591 }
1592 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1593 funcname_lno = cur;
1594 funcname_needed = 0;
1595 if (opt->funcbody)
1596 comment_needed = 1;
1597 else
1598 from = orig_from;
1599 }
1600 }
1601
1602 /* We need to look even further back to find a function signature. */
1603 if (opt->funcname && funcname_needed)
1604 show_funcname_line(opt, gs, bol, cur);
1605
1606 /* Back forward. */
1607 while (cur < lno) {
1608 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1609
1610 while (*eol != '\n')
1611 eol++;
1612 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1613 bol = eol + 1;
1614 cur++;
1615 }
1616 }
1617
1618 static int should_lookahead(struct grep_opt *opt)
1619 {
1620 struct grep_pat *p;
1621
1622 if (opt->extended)
1623 return 0; /* punt for too complex stuff */
1624 if (opt->invert)
1625 return 0;
1626 for (p = opt->pattern_list; p; p = p->next) {
1627 if (p->token != GREP_PATTERN)
1628 return 0; /* punt for "header only" and stuff */
1629 }
1630 return 1;
1631 }
1632
1633 static int look_ahead(struct grep_opt *opt,
1634 unsigned long *left_p,
1635 unsigned *lno_p,
1636 char **bol_p)
1637 {
1638 unsigned lno = *lno_p;
1639 char *bol = *bol_p;
1640 struct grep_pat *p;
1641 char *sp, *last_bol;
1642 regoff_t earliest = -1;
1643
1644 for (p = opt->pattern_list; p; p = p->next) {
1645 int hit;
1646 regmatch_t m;
1647
1648 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1649 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1650 continue;
1651 if (earliest < 0 || m.rm_so < earliest)
1652 earliest = m.rm_so;
1653 }
1654
1655 if (earliest < 0) {
1656 *bol_p = bol + *left_p;
1657 *left_p = 0;
1658 return 1;
1659 }
1660 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1661 ; /* find the beginning of the line */
1662 last_bol = sp;
1663
1664 for (sp = bol; sp < last_bol; sp++) {
1665 if (*sp == '\n')
1666 lno++;
1667 }
1668 *left_p -= last_bol - bol;
1669 *bol_p = last_bol;
1670 *lno_p = lno;
1671 return 0;
1672 }
1673
1674 static int fill_textconv_grep(struct repository *r,
1675 struct userdiff_driver *driver,
1676 struct grep_source *gs)
1677 {
1678 struct diff_filespec *df;
1679 char *buf;
1680 size_t size;
1681
1682 if (!driver || !driver->textconv)
1683 return grep_source_load(gs);
1684
1685 /*
1686 * The textconv interface is intimately tied to diff_filespecs, so we
1687 * have to pretend to be one. If we could unify the grep_source
1688 * and diff_filespec structs, this mess could just go away.
1689 */
1690 df = alloc_filespec(gs->path);
1691 switch (gs->type) {
1692 case GREP_SOURCE_OID:
1693 fill_filespec(df, gs->identifier, 1, 0100644);
1694 break;
1695 case GREP_SOURCE_FILE:
1696 fill_filespec(df, &null_oid, 0, 0100644);
1697 break;
1698 default:
1699 BUG("attempt to textconv something without a path?");
1700 }
1701
1702 /*
1703 * fill_textconv is not remotely thread-safe; it may load objects
1704 * behind the scenes, and it modifies the global diff tempfile
1705 * structure.
1706 */
1707 grep_read_lock();
1708 size = fill_textconv(r, driver, df, &buf);
1709 grep_read_unlock();
1710 free_filespec(df);
1711
1712 /*
1713 * The normal fill_textconv usage by the diff machinery would just keep
1714 * the textconv'd buf separate from the diff_filespec. But much of the
1715 * grep code passes around a grep_source and assumes that its "buf"
1716 * pointer is the beginning of the thing we are searching. So let's
1717 * install our textconv'd version into the grep_source, taking care not
1718 * to leak any existing buffer.
1719 */
1720 grep_source_clear_data(gs);
1721 gs->buf = buf;
1722 gs->size = size;
1723
1724 return 0;
1725 }
1726
1727 static int is_empty_line(const char *bol, const char *eol)
1728 {
1729 while (bol < eol && isspace(*bol))
1730 bol++;
1731 return bol == eol;
1732 }
1733
1734 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1735 {
1736 char *bol;
1737 char *peek_bol = NULL;
1738 unsigned long left;
1739 unsigned lno = 1;
1740 unsigned last_hit = 0;
1741 int binary_match_only = 0;
1742 unsigned count = 0;
1743 int try_lookahead = 0;
1744 int show_function = 0;
1745 struct userdiff_driver *textconv = NULL;
1746 enum grep_context ctx = GREP_CONTEXT_HEAD;
1747 xdemitconf_t xecfg;
1748
1749 if (!opt->status_only && gs->name == NULL)
1750 BUG("grep call which could print a name requires "
1751 "grep_source.name be non-NULL");
1752
1753 if (!opt->output)
1754 opt->output = std_output;
1755
1756 if (opt->pre_context || opt->post_context || opt->file_break ||
1757 opt->funcbody) {
1758 /* Show hunk marks, except for the first file. */
1759 if (opt->last_shown)
1760 opt->show_hunk_mark = 1;
1761 /*
1762 * If we're using threads then we can't easily identify
1763 * the first file. Always put hunk marks in that case
1764 * and skip the very first one later in work_done().
1765 */
1766 if (opt->output != std_output)
1767 opt->show_hunk_mark = 1;
1768 }
1769 opt->last_shown = 0;
1770
1771 if (opt->allow_textconv) {
1772 grep_source_load_driver(gs, opt->repo->index);
1773 /*
1774 * We might set up the shared textconv cache data here, which
1775 * is not thread-safe.
1776 */
1777 grep_attr_lock();
1778 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1779 grep_attr_unlock();
1780 }
1781
1782 /*
1783 * We know the result of a textconv is text, so we only have to care
1784 * about binary handling if we are not using it.
1785 */
1786 if (!textconv) {
1787 switch (opt->binary) {
1788 case GREP_BINARY_DEFAULT:
1789 if (grep_source_is_binary(gs, opt->repo->index))
1790 binary_match_only = 1;
1791 break;
1792 case GREP_BINARY_NOMATCH:
1793 if (grep_source_is_binary(gs, opt->repo->index))
1794 return 0; /* Assume unmatch */
1795 break;
1796 case GREP_BINARY_TEXT:
1797 break;
1798 default:
1799 BUG("unknown binary handling mode");
1800 }
1801 }
1802
1803 memset(&xecfg, 0, sizeof(xecfg));
1804 opt->priv = &xecfg;
1805
1806 try_lookahead = should_lookahead(opt);
1807
1808 if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1809 return 0;
1810
1811 bol = gs->buf;
1812 left = gs->size;
1813 while (left) {
1814 char *eol, ch;
1815 int hit;
1816 ssize_t cno;
1817 ssize_t col = -1, icol = -1;
1818
1819 /*
1820 * look_ahead() skips quickly to the line that possibly
1821 * has the next hit; don't call it if we need to do
1822 * something more than just skipping the current line
1823 * in response to an unmatch for the current line. E.g.
1824 * inside a post-context window, we will show the current
1825 * line as a context around the previous hit when it
1826 * doesn't hit.
1827 */
1828 if (try_lookahead
1829 && !(last_hit
1830 && (show_function ||
1831 lno <= last_hit + opt->post_context))
1832 && look_ahead(opt, &left, &lno, &bol))
1833 break;
1834 eol = end_of_line(bol, &left);
1835 ch = *eol;
1836 *eol = 0;
1837
1838 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1839 ctx = GREP_CONTEXT_BODY;
1840
1841 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1842 *eol = ch;
1843
1844 if (collect_hits)
1845 goto next_line;
1846
1847 /* "grep -v -e foo -e bla" should list lines
1848 * that do not have either, so inversion should
1849 * be done outside.
1850 */
1851 if (opt->invert)
1852 hit = !hit;
1853 if (opt->unmatch_name_only) {
1854 if (hit)
1855 return 0;
1856 goto next_line;
1857 }
1858 if (hit) {
1859 count++;
1860 if (opt->status_only)
1861 return 1;
1862 if (opt->name_only) {
1863 show_name(opt, gs->name);
1864 return 1;
1865 }
1866 if (opt->count)
1867 goto next_line;
1868 if (binary_match_only) {
1869 opt->output(opt, "Binary file ", 12);
1870 output_color(opt, gs->name, strlen(gs->name),
1871 opt->colors[GREP_COLOR_FILENAME]);
1872 opt->output(opt, " matches\n", 9);
1873 return 1;
1874 }
1875 /* Hit at this line. If we haven't shown the
1876 * pre-context lines, we would need to show them.
1877 */
1878 if (opt->pre_context || opt->funcbody)
1879 show_pre_context(opt, gs, bol, eol, lno);
1880 else if (opt->funcname)
1881 show_funcname_line(opt, gs, bol, lno);
1882 cno = opt->invert ? icol : col;
1883 if (cno < 0) {
1884 /*
1885 * A negative cno indicates that there was no
1886 * match on the line. We are thus inverted and
1887 * being asked to show all lines that _don't_
1888 * match a given expression. Therefore, set cno
1889 * to 0 to suggest the whole line matches.
1890 */
1891 cno = 0;
1892 }
1893 show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1894 last_hit = lno;
1895 if (opt->funcbody)
1896 show_function = 1;
1897 goto next_line;
1898 }
1899 if (show_function && (!peek_bol || peek_bol < bol)) {
1900 unsigned long peek_left = left;
1901 char *peek_eol = eol;
1902
1903 /*
1904 * Trailing empty lines are not interesting.
1905 * Peek past them to see if they belong to the
1906 * body of the current function.
1907 */
1908 peek_bol = bol;
1909 while (is_empty_line(peek_bol, peek_eol)) {
1910 peek_bol = peek_eol + 1;
1911 peek_eol = end_of_line(peek_bol, &peek_left);
1912 }
1913
1914 if (match_funcname(opt, gs, peek_bol, peek_eol))
1915 show_function = 0;
1916 }
1917 if (show_function ||
1918 (last_hit && lno <= last_hit + opt->post_context)) {
1919 /* If the last hit is within the post context,
1920 * we need to show this line.
1921 */
1922 show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1923 }
1924
1925 next_line:
1926 bol = eol + 1;
1927 if (!left)
1928 break;
1929 left--;
1930 lno++;
1931 }
1932
1933 if (collect_hits)
1934 return 0;
1935
1936 if (opt->status_only)
1937 return opt->unmatch_name_only;
1938 if (opt->unmatch_name_only) {
1939 /* We did not see any hit, so we want to show this */
1940 show_name(opt, gs->name);
1941 return 1;
1942 }
1943
1944 xdiff_clear_find_func(&xecfg);
1945 opt->priv = NULL;
1946
1947 /* NEEDSWORK:
1948 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1949 * which feels mostly useless but sometimes useful. Maybe
1950 * make it another option? For now suppress them.
1951 */
1952 if (opt->count && count) {
1953 char buf[32];
1954 if (opt->pathname) {
1955 output_color(opt, gs->name, strlen(gs->name),
1956 opt->colors[GREP_COLOR_FILENAME]);
1957 output_sep(opt, ':');
1958 }
1959 xsnprintf(buf, sizeof(buf), "%u\n", count);
1960 opt->output(opt, buf, strlen(buf));
1961 return 1;
1962 }
1963 return !!last_hit;
1964 }
1965
1966 static void clr_hit_marker(struct grep_expr *x)
1967 {
1968 /* All-hit markers are meaningful only at the very top level
1969 * OR node.
1970 */
1971 while (1) {
1972 x->hit = 0;
1973 if (x->node != GREP_NODE_OR)
1974 return;
1975 x->u.binary.left->hit = 0;
1976 x = x->u.binary.right;
1977 }
1978 }
1979
1980 static int chk_hit_marker(struct grep_expr *x)
1981 {
1982 /* Top level nodes have hit markers. See if they all are hits */
1983 while (1) {
1984 if (x->node != GREP_NODE_OR)
1985 return x->hit;
1986 if (!x->u.binary.left->hit)
1987 return 0;
1988 x = x->u.binary.right;
1989 }
1990 }
1991
1992 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1993 {
1994 /*
1995 * we do not have to do the two-pass grep when we do not check
1996 * buffer-wide "all-match".
1997 */
1998 if (!opt->all_match)
1999 return grep_source_1(opt, gs, 0);
2000
2001 /* Otherwise the toplevel "or" terms hit a bit differently.
2002 * We first clear hit markers from them.
2003 */
2004 clr_hit_marker(opt->pattern_expression);
2005 grep_source_1(opt, gs, 1);
2006
2007 if (!chk_hit_marker(opt->pattern_expression))
2008 return 0;
2009
2010 return grep_source_1(opt, gs, 0);
2011 }
2012
2013 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
2014 {
2015 struct grep_source gs;
2016 int r;
2017
2018 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
2019 gs.buf = buf;
2020 gs.size = size;
2021
2022 r = grep_source(opt, &gs);
2023
2024 grep_source_clear(&gs);
2025 return r;
2026 }
2027
2028 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
2029 const char *name, const char *path,
2030 const void *identifier)
2031 {
2032 gs->type = type;
2033 gs->name = xstrdup_or_null(name);
2034 gs->path = xstrdup_or_null(path);
2035 gs->buf = NULL;
2036 gs->size = 0;
2037 gs->driver = NULL;
2038
2039 switch (type) {
2040 case GREP_SOURCE_FILE:
2041 gs->identifier = xstrdup(identifier);
2042 break;
2043 case GREP_SOURCE_OID:
2044 gs->identifier = oiddup(identifier);
2045 break;
2046 case GREP_SOURCE_BUF:
2047 gs->identifier = NULL;
2048 break;
2049 }
2050 }
2051
2052 void grep_source_clear(struct grep_source *gs)
2053 {
2054 FREE_AND_NULL(gs->name);
2055 FREE_AND_NULL(gs->path);
2056 FREE_AND_NULL(gs->identifier);
2057 grep_source_clear_data(gs);
2058 }
2059
2060 void grep_source_clear_data(struct grep_source *gs)
2061 {
2062 switch (gs->type) {
2063 case GREP_SOURCE_FILE:
2064 case GREP_SOURCE_OID:
2065 FREE_AND_NULL(gs->buf);
2066 gs->size = 0;
2067 break;
2068 case GREP_SOURCE_BUF:
2069 /* leave user-provided buf intact */
2070 break;
2071 }
2072 }
2073
2074 static int grep_source_load_oid(struct grep_source *gs)
2075 {
2076 enum object_type type;
2077
2078 grep_read_lock();
2079 gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2080 grep_read_unlock();
2081
2082 if (!gs->buf)
2083 return error(_("'%s': unable to read %s"),
2084 gs->name,
2085 oid_to_hex(gs->identifier));
2086 return 0;
2087 }
2088
2089 static int grep_source_load_file(struct grep_source *gs)
2090 {
2091 const char *filename = gs->identifier;
2092 struct stat st;
2093 char *data;
2094 size_t size;
2095 int i;
2096
2097 if (lstat(filename, &st) < 0) {
2098 err_ret:
2099 if (errno != ENOENT)
2100 error_errno(_("failed to stat '%s'"), filename);
2101 return -1;
2102 }
2103 if (!S_ISREG(st.st_mode))
2104 return -1;
2105 size = xsize_t(st.st_size);
2106 i = open(filename, O_RDONLY);
2107 if (i < 0)
2108 goto err_ret;
2109 data = xmallocz(size);
2110 if (st.st_size != read_in_full(i, data, size)) {
2111 error_errno(_("'%s': short read"), filename);
2112 close(i);
2113 free(data);
2114 return -1;
2115 }
2116 close(i);
2117
2118 gs->buf = data;
2119 gs->size = size;
2120 return 0;
2121 }
2122
2123 static int grep_source_load(struct grep_source *gs)
2124 {
2125 if (gs->buf)
2126 return 0;
2127
2128 switch (gs->type) {
2129 case GREP_SOURCE_FILE:
2130 return grep_source_load_file(gs);
2131 case GREP_SOURCE_OID:
2132 return grep_source_load_oid(gs);
2133 case GREP_SOURCE_BUF:
2134 return gs->buf ? 0 : -1;
2135 }
2136 BUG("invalid grep_source type to load");
2137 }
2138
2139 void grep_source_load_driver(struct grep_source *gs,
2140 struct index_state *istate)
2141 {
2142 if (gs->driver)
2143 return;
2144
2145 grep_attr_lock();
2146 if (gs->path)
2147 gs->driver = userdiff_find_by_path(istate, gs->path);
2148 if (!gs->driver)
2149 gs->driver = userdiff_find_by_name("default");
2150 grep_attr_unlock();
2151 }
2152
2153 static int grep_source_is_binary(struct grep_source *gs,
2154 struct index_state *istate)
2155 {
2156 grep_source_load_driver(gs, istate);
2157 if (gs->driver->binary != -1)
2158 return gs->driver->binary;
2159
2160 if (!grep_source_load(gs))
2161 return buffer_is_binary(gs->buf, gs->size);
2162
2163 return 0;
2164 }
The Best VCS Around