?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
CodingGuidelines: hint why we value clearly written log messages
[git/debian.git] / grep.c
blob7bb0360869a64bca7af1e2d7faecb937a63b7910
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 void std_output(struct grep_opt *opt, const void *buf, size_t size)
18 {
19 fwrite(buf, size, 1, stdout);
20 }
21
22 static struct grep_opt grep_defaults = {
23 .relative = 1,
24 .pathname = 1,
25 .max_depth = -1,
26 .pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED,
27 .colors = {
28 [GREP_COLOR_CONTEXT] = "",
29 [GREP_COLOR_FILENAME] = GIT_COLOR_MAGENTA,
30 [GREP_COLOR_FUNCTION] = "",
31 [GREP_COLOR_LINENO] = GIT_COLOR_GREEN,
32 [GREP_COLOR_COLUMNNO] = GIT_COLOR_GREEN,
33 [GREP_COLOR_MATCH_CONTEXT] = GIT_COLOR_BOLD_RED,
34 [GREP_COLOR_MATCH_SELECTED] = GIT_COLOR_BOLD_RED,
35 [GREP_COLOR_SELECTED] = "",
36 [GREP_COLOR_SEP] = GIT_COLOR_CYAN,
37 },
38 .only_matching = 0,
39 .color = -1,
40 .output = std_output,
41 };
42
43 static const char *color_grep_slots[] = {
44 [GREP_COLOR_CONTEXT] = "context",
45 [GREP_COLOR_FILENAME] = "filename",
46 [GREP_COLOR_FUNCTION] = "function",
47 [GREP_COLOR_LINENO] = "lineNumber",
48 [GREP_COLOR_COLUMNNO] = "column",
49 [GREP_COLOR_MATCH_CONTEXT] = "matchContext",
50 [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
51 [GREP_COLOR_SELECTED] = "selected",
52 [GREP_COLOR_SEP] = "separator",
53 };
54
55 static int parse_pattern_type_arg(const char *opt, const char *arg)
56 {
57 if (!strcmp(arg, "default"))
58 return GREP_PATTERN_TYPE_UNSPECIFIED;
59 else if (!strcmp(arg, "basic"))
60 return GREP_PATTERN_TYPE_BRE;
61 else if (!strcmp(arg, "extended"))
62 return GREP_PATTERN_TYPE_ERE;
63 else if (!strcmp(arg, "fixed"))
64 return GREP_PATTERN_TYPE_FIXED;
65 else if (!strcmp(arg, "perl"))
66 return GREP_PATTERN_TYPE_PCRE;
67 die("bad %s argument: %s", opt, arg);
68 }
69
70 define_list_config_array_extra(color_grep_slots, {"match"});
71
72 /*
73 * Read the configuration file once and store it in
74 * the grep_defaults template.
75 */
76 int grep_config(const char *var, const char *value, void *cb)
77 {
78 struct grep_opt *opt = &grep_defaults;
79 const char *slot;
80
81 if (userdiff_config(var, value) < 0)
82 return -1;
83
84 /*
85 * The instance of grep_opt that we set up here is copied by
86 * grep_init() to be used by each individual invocation.
87 * When populating a new field of this structure here, be
88 * sure to think about ownership -- e.g., you might need to
89 * override the shallow copy in grep_init() with a deep copy.
90 */
91
92 if (!strcmp(var, "grep.extendedregexp")) {
93 opt->extended_regexp_option = git_config_bool(var, value);
94 return 0;
95 }
96
97 if (!strcmp(var, "grep.patterntype")) {
98 opt->pattern_type_option = parse_pattern_type_arg(var, value);
99 return 0;
100 }
101
102 if (!strcmp(var, "grep.linenumber")) {
103 opt->linenum = git_config_bool(var, value);
104 return 0;
105 }
106 if (!strcmp(var, "grep.column")) {
107 opt->columnnum = git_config_bool(var, value);
108 return 0;
109 }
110
111 if (!strcmp(var, "grep.fullname")) {
112 opt->relative = !git_config_bool(var, value);
113 return 0;
114 }
115
116 if (!strcmp(var, "color.grep"))
117 opt->color = git_config_colorbool(var, value);
118 if (!strcmp(var, "color.grep.match")) {
119 if (grep_config("color.grep.matchcontext", value, cb) < 0)
120 return -1;
121 if (grep_config("color.grep.matchselected", value, cb) < 0)
122 return -1;
123 } else if (skip_prefix(var, "color.grep.", &slot)) {
124 int i = LOOKUP_CONFIG(color_grep_slots, slot);
125 char *color;
126
127 if (i < 0)
128 return -1;
129 color = opt->colors[i];
130 if (!value)
131 return config_error_nonbool(var);
132 return color_parse(value, color);
133 }
134 return 0;
135 }
136
137 /*
138 * Initialize one instance of grep_opt and copy the
139 * default values from the template we read the configuration
140 * information in an earlier call to git_config(grep_config).
141 */
142 void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
143 {
144 *opt = grep_defaults;
145
146 opt->repo = repo;
147 opt->prefix = prefix;
148 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
149 opt->pattern_tail = &opt->pattern_list;
150 opt->header_tail = &opt->header_list;
151 }
152
153 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
154 {
155 /*
156 * When committing to the pattern type by setting the relevant
157 * fields in grep_opt it's generally not necessary to zero out
158 * the fields we're not choosing, since they won't have been
159 * set by anything. The extended_regexp_option field is the
160 * only exception to this.
161 *
162 * This is because in the process of parsing grep.patternType
163 * & grep.extendedRegexp we set opt->pattern_type_option and
164 * opt->extended_regexp_option, respectively. We then
165 * internally use opt->extended_regexp_option to see if we're
166 * compiling an ERE. It must be unset if that's not actually
167 * the case.
168 */
169 if (pattern_type != GREP_PATTERN_TYPE_ERE &&
170 opt->extended_regexp_option)
171 opt->extended_regexp_option = 0;
172
173 switch (pattern_type) {
174 case GREP_PATTERN_TYPE_UNSPECIFIED:
175 /* fall through */
176
177 case GREP_PATTERN_TYPE_BRE:
178 break;
179
180 case GREP_PATTERN_TYPE_ERE:
181 opt->extended_regexp_option = 1;
182 break;
183
184 case GREP_PATTERN_TYPE_FIXED:
185 opt->fixed = 1;
186 break;
187
188 case GREP_PATTERN_TYPE_PCRE:
189 opt->pcre2 = 1;
190 break;
191 }
192 }
193
194 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
195 {
196 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
197 grep_set_pattern_type_option(pattern_type, opt);
198 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
199 grep_set_pattern_type_option(opt->pattern_type_option, opt);
200 else if (opt->extended_regexp_option)
201 /*
202 * This branch *must* happen after setting from the
203 * opt->pattern_type_option above, we don't want
204 * grep.extendedRegexp to override grep.patternType!
205 */
206 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
207 }
208
209 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
210 const char *origin, int no,
211 enum grep_pat_token t,
212 enum grep_header_field field)
213 {
214 struct grep_pat *p = xcalloc(1, sizeof(*p));
215 p->pattern = xmemdupz(pat, patlen);
216 p->patternlen = patlen;
217 p->origin = origin;
218 p->no = no;
219 p->token = t;
220 p->field = field;
221 return p;
222 }
223
224 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
225 {
226 **tail = p;
227 *tail = &p->next;
228 p->next = NULL;
229
230 switch (p->token) {
231 case GREP_PATTERN: /* atom */
232 case GREP_PATTERN_HEAD:
233 case GREP_PATTERN_BODY:
234 for (;;) {
235 struct grep_pat *new_pat;
236 size_t len = 0;
237 char *cp = p->pattern + p->patternlen, *nl = NULL;
238 while (++len <= p->patternlen) {
239 if (*(--cp) == '\n') {
240 nl = cp;
241 break;
242 }
243 }
244 if (!nl)
245 break;
246 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
247 p->no, p->token, p->field);
248 new_pat->next = p->next;
249 if (!p->next)
250 *tail = &new_pat->next;
251 p->next = new_pat;
252 *nl = '\0';
253 p->patternlen -= len;
254 }
255 break;
256 default:
257 break;
258 }
259 }
260
261 void append_header_grep_pattern(struct grep_opt *opt,
262 enum grep_header_field field, const char *pat)
263 {
264 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
265 GREP_PATTERN_HEAD, field);
266 if (field == GREP_HEADER_REFLOG)
267 opt->use_reflog_filter = 1;
268 do_append_grep_pat(&opt->header_tail, p);
269 }
270
271 void append_grep_pattern(struct grep_opt *opt, const char *pat,
272 const char *origin, int no, enum grep_pat_token t)
273 {
274 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
275 }
276
277 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
278 const char *origin, int no, enum grep_pat_token t)
279 {
280 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
281 do_append_grep_pat(&opt->pattern_tail, p);
282 }
283
284 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
285 {
286 struct grep_pat *pat;
287 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
288 *ret = *opt;
289
290 ret->pattern_list = NULL;
291 ret->pattern_tail = &ret->pattern_list;
292
293 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
294 {
295 if(pat->token == GREP_PATTERN_HEAD)
296 append_header_grep_pattern(ret, pat->field,
297 pat->pattern);
298 else
299 append_grep_pat(ret, pat->pattern, pat->patternlen,
300 pat->origin, pat->no, pat->token);
301 }
302
303 return ret;
304 }
305
306 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
307 const char *error)
308 {
309 char where[1024];
310
311 if (p->no)
312 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
313 else if (p->origin)
314 xsnprintf(where, sizeof(where), "%s, ", p->origin);
315 else
316 where[0] = 0;
317
318 die("%s'%s': %s", where, p->pattern, error);
319 }
320
321 static int is_fixed(const char *s, size_t len)
322 {
323 size_t i;
324
325 for (i = 0; i < len; i++) {
326 if (is_regex_special(s[i]))
327 return 0;
328 }
329
330 return 1;
331 }
332
333 #ifdef USE_LIBPCRE2
334 #define GREP_PCRE2_DEBUG_MALLOC 0
335
336 static void *pcre2_malloc(PCRE2_SIZE size, MAYBE_UNUSED void *memory_data)
337 {
338 void *pointer = malloc(size);
339 #if GREP_PCRE2_DEBUG_MALLOC
340 static int count = 1;
341 fprintf(stderr, "PCRE2:%p -> #%02d: alloc(%lu)\n", pointer, count++, size);
342 #endif
343 return pointer;
344 }
345
346 static void pcre2_free(void *pointer, MAYBE_UNUSED void *memory_data)
347 {
348 #if GREP_PCRE2_DEBUG_MALLOC
349 static int count = 1;
350 if (pointer)
351 fprintf(stderr, "PCRE2:%p -> #%02d: free()\n", pointer, count++);
352 #endif
353 free(pointer);
354 }
355
356 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
357 {
358 int error;
359 PCRE2_UCHAR errbuf[256];
360 PCRE2_SIZE erroffset;
361 int options = PCRE2_MULTILINE;
362 int jitret;
363 int patinforet;
364 size_t jitsizearg;
365 int literal = !opt->ignore_case && (p->fixed || p->is_fixed);
366
367 /*
368 * Call pcre2_general_context_create() before calling any
369 * other pcre2_*(). It sets up our malloc()/free() functions
370 * with which everything else is allocated.
371 */
372 p->pcre2_general_context = pcre2_general_context_create(
373 pcre2_malloc, pcre2_free, NULL);
374 if (!p->pcre2_general_context)
375 die("Couldn't allocate PCRE2 general context");
376
377 if (opt->ignore_case) {
378 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
379 p->pcre2_tables = pcre2_maketables(p->pcre2_general_context);
380 p->pcre2_compile_context = pcre2_compile_context_create(p->pcre2_general_context);
381 pcre2_set_character_tables(p->pcre2_compile_context,
382 p->pcre2_tables);
383 }
384 options |= PCRE2_CASELESS;
385 }
386 if (!opt->ignore_locale && is_utf8_locale() && !literal)
387 options |= (PCRE2_UTF | PCRE2_MATCH_INVALID_UTF);
388
389 #ifdef GIT_PCRE2_VERSION_10_36_OR_HIGHER
390 /* Work around https://bugs.exim.org/show_bug.cgi?id=2642 fixed in 10.36 */
391 if (PCRE2_MATCH_INVALID_UTF && options & (PCRE2_UTF | PCRE2_CASELESS))
392 options |= PCRE2_NO_START_OPTIMIZE;
393 #endif
394
395 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
396 p->patternlen, options, &error, &erroffset,
397 p->pcre2_compile_context);
398
399 if (p->pcre2_pattern) {
400 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, p->pcre2_general_context);
401 if (!p->pcre2_match_data)
402 die("Couldn't allocate PCRE2 match data");
403 } else {
404 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
405 compile_regexp_failed(p, (const char *)&errbuf);
406 }
407
408 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
409 if (p->pcre2_jit_on) {
410 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
411 if (jitret)
412 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
413
414 /*
415 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
416 * tells us whether the library itself supports JIT,
417 * but to see whether we're going to be actually using
418 * JIT we need to extract PCRE2_INFO_JITSIZE from the
419 * pattern *after* we do pcre2_jit_compile() above.
420 *
421 * This is because if the pattern contains the
422 * (*NO_JIT) verb (see pcre2syntax(3))
423 * pcre2_jit_compile() will exit early with 0. If we
424 * then proceed to call pcre2_jit_match() further down
425 * the line instead of pcre2_match() we'll either
426 * segfault (pre PCRE 10.31) or run into a fatal error
427 * (post PCRE2 10.31)
428 */
429 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
430 if (patinforet)
431 BUG("pcre2_pattern_info() failed: %d", patinforet);
432 if (jitsizearg == 0) {
433 p->pcre2_jit_on = 0;
434 return;
435 }
436 }
437 }
438
439 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
440 regmatch_t *match, int eflags)
441 {
442 int ret, flags = 0;
443 PCRE2_SIZE *ovector;
444 PCRE2_UCHAR errbuf[256];
445
446 if (eflags & REG_NOTBOL)
447 flags |= PCRE2_NOTBOL;
448
449 if (p->pcre2_jit_on)
450 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
451 eol - line, 0, flags, p->pcre2_match_data,
452 NULL);
453 else
454 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
455 eol - line, 0, flags, p->pcre2_match_data,
456 NULL);
457
458 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
459 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
460 die("%s failed with error code %d: %s",
461 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
462 errbuf);
463 }
464 if (ret > 0) {
465 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
466 ret = 0;
467 match->rm_so = (int)ovector[0];
468 match->rm_eo = (int)ovector[1];
469 }
470
471 return ret;
472 }
473
474 static void free_pcre2_pattern(struct grep_pat *p)
475 {
476 pcre2_compile_context_free(p->pcre2_compile_context);
477 pcre2_code_free(p->pcre2_pattern);
478 pcre2_match_data_free(p->pcre2_match_data);
479 #ifdef GIT_PCRE2_VERSION_10_34_OR_HIGHER
480 pcre2_maketables_free(p->pcre2_general_context, p->pcre2_tables);
481 #else
482 free((void *)p->pcre2_tables);
483 #endif
484 pcre2_general_context_free(p->pcre2_general_context);
485 }
486 #else /* !USE_LIBPCRE2 */
487 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
488 {
489 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
490 }
491
492 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
493 regmatch_t *match, int eflags)
494 {
495 return 1;
496 }
497
498 static void free_pcre2_pattern(struct grep_pat *p)
499 {
500 }
501
502 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
503 {
504 struct strbuf sb = STRBUF_INIT;
505 int err;
506 int regflags = 0;
507
508 basic_regex_quote_buf(&sb, p->pattern);
509 if (opt->ignore_case)
510 regflags |= REG_ICASE;
511 err = regcomp(&p->regexp, sb.buf, regflags);
512 strbuf_release(&sb);
513 if (err) {
514 char errbuf[1024];
515 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
516 compile_regexp_failed(p, errbuf);
517 }
518 }
519 #endif /* !USE_LIBPCRE2 */
520
521 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
522 {
523 int err;
524 int regflags = REG_NEWLINE;
525
526 p->word_regexp = opt->word_regexp;
527 p->ignore_case = opt->ignore_case;
528 p->fixed = opt->fixed;
529
530 if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2)
531 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
532
533 p->is_fixed = is_fixed(p->pattern, p->patternlen);
534 #ifdef USE_LIBPCRE2
535 if (!p->fixed && !p->is_fixed) {
536 const char *no_jit = "(*NO_JIT)";
537 const int no_jit_len = strlen(no_jit);
538 if (starts_with(p->pattern, no_jit) &&
539 is_fixed(p->pattern + no_jit_len,
540 p->patternlen - no_jit_len))
541 p->is_fixed = 1;
542 }
543 #endif
544 if (p->fixed || p->is_fixed) {
545 #ifdef USE_LIBPCRE2
546 if (p->is_fixed) {
547 compile_pcre2_pattern(p, opt);
548 } else {
549 /*
550 * E.g. t7811-grep-open.sh relies on the
551 * pattern being restored.
552 */
553 char *old_pattern = p->pattern;
554 size_t old_patternlen = p->patternlen;
555 struct strbuf sb = STRBUF_INIT;
556
557 /*
558 * There is the PCRE2_LITERAL flag, but it's
559 * only in PCRE v2 10.30 and later. Needing to
560 * ifdef our way around that and dealing with
561 * it + PCRE2_MULTILINE being an error is more
562 * complex than just quoting this ourselves.
563 */
564 strbuf_add(&sb, "\\Q", 2);
565 strbuf_add(&sb, p->pattern, p->patternlen);
566 strbuf_add(&sb, "\\E", 2);
567
568 p->pattern = sb.buf;
569 p->patternlen = sb.len;
570 compile_pcre2_pattern(p, opt);
571 p->pattern = old_pattern;
572 p->patternlen = old_patternlen;
573 strbuf_release(&sb);
574 }
575 #else /* !USE_LIBPCRE2 */
576 compile_fixed_regexp(p, opt);
577 #endif /* !USE_LIBPCRE2 */
578 return;
579 }
580
581 if (opt->pcre2) {
582 compile_pcre2_pattern(p, opt);
583 return;
584 }
585
586 if (p->ignore_case)
587 regflags |= REG_ICASE;
588 if (opt->extended_regexp_option)
589 regflags |= REG_EXTENDED;
590 err = regcomp(&p->regexp, p->pattern, regflags);
591 if (err) {
592 char errbuf[1024];
593 regerror(err, &p->regexp, errbuf, 1024);
594 compile_regexp_failed(p, errbuf);
595 }
596 }
597
598 static struct grep_expr *compile_pattern_or(struct grep_pat **);
599 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
600 {
601 struct grep_pat *p;
602 struct grep_expr *x;
603
604 p = *list;
605 if (!p)
606 return NULL;
607 switch (p->token) {
608 case GREP_PATTERN: /* atom */
609 case GREP_PATTERN_HEAD:
610 case GREP_PATTERN_BODY:
611 CALLOC_ARRAY(x, 1);
612 x->node = GREP_NODE_ATOM;
613 x->u.atom = p;
614 *list = p->next;
615 return x;
616 case GREP_OPEN_PAREN:
617 *list = p->next;
618 x = compile_pattern_or(list);
619 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
620 die("unmatched parenthesis");
621 *list = (*list)->next;
622 return x;
623 default:
624 return NULL;
625 }
626 }
627
628 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
629 {
630 struct grep_pat *p;
631 struct grep_expr *x;
632
633 p = *list;
634 if (!p)
635 return NULL;
636 switch (p->token) {
637 case GREP_NOT:
638 if (!p->next)
639 die("--not not followed by pattern expression");
640 *list = p->next;
641 CALLOC_ARRAY(x, 1);
642 x->node = GREP_NODE_NOT;
643 x->u.unary = compile_pattern_not(list);
644 if (!x->u.unary)
645 die("--not followed by non pattern expression");
646 return x;
647 default:
648 return compile_pattern_atom(list);
649 }
650 }
651
652 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
653 {
654 struct grep_pat *p;
655 struct grep_expr *x, *y, *z;
656
657 x = compile_pattern_not(list);
658 p = *list;
659 if (p && p->token == GREP_AND) {
660 if (!x)
661 die("--and not preceded by pattern expression");
662 if (!p->next)
663 die("--and not followed by pattern expression");
664 *list = p->next;
665 y = compile_pattern_and(list);
666 if (!y)
667 die("--and not followed by pattern expression");
668 CALLOC_ARRAY(z, 1);
669 z->node = GREP_NODE_AND;
670 z->u.binary.left = x;
671 z->u.binary.right = y;
672 return z;
673 }
674 return x;
675 }
676
677 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
678 {
679 struct grep_pat *p;
680 struct grep_expr *x, *y, *z;
681
682 x = compile_pattern_and(list);
683 p = *list;
684 if (x && p && p->token != GREP_CLOSE_PAREN) {
685 y = compile_pattern_or(list);
686 if (!y)
687 die("not a pattern expression %s", p->pattern);
688 CALLOC_ARRAY(z, 1);
689 z->node = GREP_NODE_OR;
690 z->u.binary.left = x;
691 z->u.binary.right = y;
692 return z;
693 }
694 return x;
695 }
696
697 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
698 {
699 return compile_pattern_or(list);
700 }
701
702 static struct grep_expr *grep_not_expr(struct grep_expr *expr)
703 {
704 struct grep_expr *z = xcalloc(1, sizeof(*z));
705 z->node = GREP_NODE_NOT;
706 z->u.unary = expr;
707 return z;
708 }
709
710 static struct grep_expr *grep_true_expr(void)
711 {
712 struct grep_expr *z = xcalloc(1, sizeof(*z));
713 z->node = GREP_NODE_TRUE;
714 return z;
715 }
716
717 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
718 {
719 struct grep_expr *z = xcalloc(1, sizeof(*z));
720 z->node = GREP_NODE_OR;
721 z->u.binary.left = left;
722 z->u.binary.right = right;
723 return z;
724 }
725
726 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
727 {
728 struct grep_pat *p;
729 struct grep_expr *header_expr;
730 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
731 enum grep_header_field fld;
732
733 if (!opt->header_list)
734 return NULL;
735
736 for (p = opt->header_list; p; p = p->next) {
737 if (p->token != GREP_PATTERN_HEAD)
738 BUG("a non-header pattern in grep header list.");
739 if (p->field < GREP_HEADER_FIELD_MIN ||
740 GREP_HEADER_FIELD_MAX <= p->field)
741 BUG("unknown header field %d", p->field);
742 compile_regexp(p, opt);
743 }
744
745 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
746 header_group[fld] = NULL;
747
748 for (p = opt->header_list; p; p = p->next) {
749 struct grep_expr *h;
750 struct grep_pat *pp = p;
751
752 h = compile_pattern_atom(&pp);
753 if (!h || pp != p->next)
754 BUG("malformed header expr");
755 if (!header_group[p->field]) {
756 header_group[p->field] = h;
757 continue;
758 }
759 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
760 }
761
762 header_expr = NULL;
763
764 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
765 if (!header_group[fld])
766 continue;
767 if (!header_expr)
768 header_expr = grep_true_expr();
769 header_expr = grep_or_expr(header_group[fld], header_expr);
770 }
771 return header_expr;
772 }
773
774 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
775 {
776 struct grep_expr *z = x;
777
778 while (x) {
779 assert(x->node == GREP_NODE_OR);
780 if (x->u.binary.right &&
781 x->u.binary.right->node == GREP_NODE_TRUE) {
782 x->u.binary.right = y;
783 break;
784 }
785 x = x->u.binary.right;
786 }
787 return z;
788 }
789
790 void compile_grep_patterns(struct grep_opt *opt)
791 {
792 struct grep_pat *p;
793 struct grep_expr *header_expr = prep_header_patterns(opt);
794
795 for (p = opt->pattern_list; p; p = p->next) {
796 switch (p->token) {
797 case GREP_PATTERN: /* atom */
798 case GREP_PATTERN_HEAD:
799 case GREP_PATTERN_BODY:
800 compile_regexp(p, opt);
801 break;
802 default:
803 opt->extended = 1;
804 break;
805 }
806 }
807
808 if (opt->all_match || opt->no_body_match || header_expr)
809 opt->extended = 1;
810 else if (!opt->extended)
811 return;
812
813 p = opt->pattern_list;
814 if (p)
815 opt->pattern_expression = compile_pattern_expr(&p);
816 if (p)
817 die("incomplete pattern expression: %s", p->pattern);
818
819 if (opt->no_body_match && opt->pattern_expression)
820 opt->pattern_expression = grep_not_expr(opt->pattern_expression);
821
822 if (!header_expr)
823 return;
824
825 if (!opt->pattern_expression)
826 opt->pattern_expression = header_expr;
827 else if (opt->all_match)
828 opt->pattern_expression = grep_splice_or(header_expr,
829 opt->pattern_expression);
830 else
831 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
832 header_expr);
833 opt->all_match = 1;
834 }
835
836 static void free_pattern_expr(struct grep_expr *x)
837 {
838 switch (x->node) {
839 case GREP_NODE_TRUE:
840 case GREP_NODE_ATOM:
841 break;
842 case GREP_NODE_NOT:
843 free_pattern_expr(x->u.unary);
844 break;
845 case GREP_NODE_AND:
846 case GREP_NODE_OR:
847 free_pattern_expr(x->u.binary.left);
848 free_pattern_expr(x->u.binary.right);
849 break;
850 }
851 free(x);
852 }
853
854 void free_grep_patterns(struct grep_opt *opt)
855 {
856 struct grep_pat *p, *n;
857
858 for (p = opt->pattern_list; p; p = n) {
859 n = p->next;
860 switch (p->token) {
861 case GREP_PATTERN: /* atom */
862 case GREP_PATTERN_HEAD:
863 case GREP_PATTERN_BODY:
864 if (p->pcre2_pattern)
865 free_pcre2_pattern(p);
866 else
867 regfree(&p->regexp);
868 free(p->pattern);
869 break;
870 default:
871 break;
872 }
873 free(p);
874 }
875
876 if (!opt->extended)
877 return;
878 free_pattern_expr(opt->pattern_expression);
879 }
880
881 static const char *end_of_line(const char *cp, unsigned long *left)
882 {
883 unsigned long l = *left;
884 while (l && *cp != '\n') {
885 l--;
886 cp++;
887 }
888 *left = l;
889 return cp;
890 }
891
892 static int word_char(char ch)
893 {
894 return isalnum(ch) || ch == '_';
895 }
896
897 static void output_color(struct grep_opt *opt, const void *data, size_t size,
898 const char *color)
899 {
900 if (want_color(opt->color) && color && color[0]) {
901 opt->output(opt, color, strlen(color));
902 opt->output(opt, data, size);
903 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
904 } else
905 opt->output(opt, data, size);
906 }
907
908 static void output_sep(struct grep_opt *opt, char sign)
909 {
910 if (opt->null_following_name)
911 opt->output(opt, "\0", 1);
912 else
913 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
914 }
915
916 static void show_name(struct grep_opt *opt, const char *name)
917 {
918 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
919 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
920 }
921
922 static int patmatch(struct grep_pat *p,
923 const char *line, const char *eol,
924 regmatch_t *match, int eflags)
925 {
926 int hit;
927
928 if (p->pcre2_pattern)
929 hit = !pcre2match(p, line, eol, match, eflags);
930 else
931 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
932 eflags);
933
934 return hit;
935 }
936
937 static void strip_timestamp(const char *bol, const char **eol_p)
938 {
939 const char *eol = *eol_p;
940
941 while (bol < --eol) {
942 if (*eol != '>')
943 continue;
944 *eol_p = ++eol;
945 break;
946 }
947 }
948
949 static struct {
950 const char *field;
951 size_t len;
952 } header_field[] = {
953 { "author ", 7 },
954 { "committer ", 10 },
955 { "reflog ", 7 },
956 };
957
958 static int headerless_match_one_pattern(struct grep_pat *p,
959 const char *bol, const char *eol,
960 enum grep_context ctx,
961 regmatch_t *pmatch, int eflags)
962 {
963 int hit = 0;
964 const char *start = bol;
965
966 if ((p->token != GREP_PATTERN) &&
967 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
968 return 0;
969
970 again:
971 hit = patmatch(p, bol, eol, pmatch, eflags);
972
973 if (hit && p->word_regexp) {
974 if ((pmatch[0].rm_so < 0) ||
975 (eol - bol) < pmatch[0].rm_so ||
976 (pmatch[0].rm_eo < 0) ||
977 (eol - bol) < pmatch[0].rm_eo)
978 die("regexp returned nonsense");
979
980 /* Match beginning must be either beginning of the
981 * line, or at word boundary (i.e. the last char must
982 * not be a word char). Similarly, match end must be
983 * either end of the line, or at word boundary
984 * (i.e. the next char must not be a word char).
985 */
986 if ( ((pmatch[0].rm_so == 0) ||
987 !word_char(bol[pmatch[0].rm_so-1])) &&
988 ((pmatch[0].rm_eo == (eol-bol)) ||
989 !word_char(bol[pmatch[0].rm_eo])) )
990 ;
991 else
992 hit = 0;
993
994 /* Words consist of at least one character. */
995 if (pmatch->rm_so == pmatch->rm_eo)
996 hit = 0;
997
998 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
999 /* There could be more than one match on the
1000 * line, and the first match might not be
1001 * strict word match. But later ones could be!
1002 * Forward to the next possible start, i.e. the
1003 * next position following a non-word char.
1004 */
1005 bol = pmatch[0].rm_so + bol + 1;
1006 while (word_char(bol[-1]) && bol < eol)
1007 bol++;
1008 eflags |= REG_NOTBOL;
1009 if (bol < eol)
1010 goto again;
1011 }
1012 }
1013 if (hit) {
1014 pmatch[0].rm_so += bol - start;
1015 pmatch[0].rm_eo += bol - start;
1016 }
1017 return hit;
1018 }
1019
1020 static int match_one_pattern(struct grep_pat *p,
1021 const char *bol, const char *eol,
1022 enum grep_context ctx, regmatch_t *pmatch,
1023 int eflags)
1024 {
1025 const char *field;
1026 size_t len;
1027
1028 if (p->token == GREP_PATTERN_HEAD) {
1029 assert(p->field < ARRAY_SIZE(header_field));
1030 field = header_field[p->field].field;
1031 len = header_field[p->field].len;
1032 if (strncmp(bol, field, len))
1033 return 0;
1034 bol += len;
1035
1036 switch (p->field) {
1037 case GREP_HEADER_AUTHOR:
1038 case GREP_HEADER_COMMITTER:
1039 strip_timestamp(bol, &eol);
1040 break;
1041 default:
1042 break;
1043 }
1044 }
1045
1046 return headerless_match_one_pattern(p, bol, eol, ctx, pmatch, eflags);
1047 }
1048
1049
1050 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x,
1051 const char *bol, const char *eol,
1052 enum grep_context ctx, ssize_t *col,
1053 ssize_t *icol, int collect_hits)
1054 {
1055 int h = 0;
1056
1057 if (!x)
1058 die("Not a valid grep expression");
1059 switch (x->node) {
1060 case GREP_NODE_TRUE:
1061 h = 1;
1062 break;
1063 case GREP_NODE_ATOM:
1064 {
1065 regmatch_t tmp;
1066 h = match_one_pattern(x->u.atom, bol, eol, ctx,
1067 &tmp, 0);
1068 if (h && (*col < 0 || tmp.rm_so < *col))
1069 *col = tmp.rm_so;
1070 }
1071 if (x->u.atom->token == GREP_PATTERN_BODY)
1072 opt->body_hit |= h;
1073 break;
1074 case GREP_NODE_NOT:
1075 /*
1076 * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1077 */
1078 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1079 0);
1080 break;
1081 case GREP_NODE_AND:
1082 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1083 icol, 0);
1084 if (h || opt->columnnum) {
1085 /*
1086 * Don't short-circuit AND when given --column, since a
1087 * NOT earlier in the tree may turn this into an OR. In
1088 * this case, see the below comment.
1089 */
1090 h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1091 ctx, col, icol, 0);
1092 }
1093 break;
1094 case GREP_NODE_OR:
1095 if (!(collect_hits || opt->columnnum)) {
1096 /*
1097 * Don't short-circuit OR when given --column (or
1098 * collecting hits) to ensure we don't skip a later
1099 * child that would produce an earlier match.
1100 */
1101 return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1102 ctx, col, icol, 0) ||
1103 match_expr_eval(opt, x->u.binary.right, bol,
1104 eol, ctx, col, icol, 0));
1105 }
1106 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1107 icol, 0);
1108 if (collect_hits)
1109 x->u.binary.left->hit |= h;
1110 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1111 icol, collect_hits);
1112 break;
1113 default:
1114 die("Unexpected node type (internal error) %d", x->node);
1115 }
1116 if (collect_hits)
1117 x->hit |= h;
1118 return h;
1119 }
1120
1121 static int match_expr(struct grep_opt *opt,
1122 const char *bol, const char *eol,
1123 enum grep_context ctx, ssize_t *col,
1124 ssize_t *icol, int collect_hits)
1125 {
1126 struct grep_expr *x = opt->pattern_expression;
1127 return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1128 }
1129
1130 static int match_line(struct grep_opt *opt,
1131 const char *bol, const char *eol,
1132 ssize_t *col, ssize_t *icol,
1133 enum grep_context ctx, int collect_hits)
1134 {
1135 struct grep_pat *p;
1136 int hit = 0;
1137
1138 if (opt->extended)
1139 return match_expr(opt, bol, eol, ctx, col, icol,
1140 collect_hits);
1141
1142 /* we do not call with collect_hits without being extended */
1143 for (p = opt->pattern_list; p; p = p->next) {
1144 regmatch_t tmp;
1145 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1146 hit |= 1;
1147 if (!opt->columnnum) {
1148 /*
1149 * Without --column, any single match on a line
1150 * is enough to know that it needs to be
1151 * printed. With --column, scan _all_ patterns
1152 * to find the earliest.
1153 */
1154 break;
1155 }
1156 if (*col < 0 || tmp.rm_so < *col)
1157 *col = tmp.rm_so;
1158 }
1159 }
1160 return hit;
1161 }
1162
1163 static int match_next_pattern(struct grep_pat *p,
1164 const char *bol, const char *eol,
1165 enum grep_context ctx,
1166 regmatch_t *pmatch, int eflags)
1167 {
1168 regmatch_t match;
1169
1170 if (!headerless_match_one_pattern(p, bol, eol, ctx, &match, eflags))
1171 return 0;
1172 if (match.rm_so < 0 || match.rm_eo < 0)
1173 return 0;
1174 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1175 if (match.rm_so > pmatch->rm_so)
1176 return 1;
1177 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1178 return 1;
1179 }
1180 pmatch->rm_so = match.rm_so;
1181 pmatch->rm_eo = match.rm_eo;
1182 return 1;
1183 }
1184
1185 int grep_next_match(struct grep_opt *opt,
1186 const char *bol, const char *eol,
1187 enum grep_context ctx, regmatch_t *pmatch,
1188 enum grep_header_field field, int eflags)
1189 {
1190 struct grep_pat *p;
1191 int hit = 0;
1192
1193 pmatch->rm_so = pmatch->rm_eo = -1;
1194 if (bol < eol) {
1195 for (p = ((ctx == GREP_CONTEXT_HEAD)
1196 ? opt->header_list : opt->pattern_list);
1197 p; p = p->next) {
1198 switch (p->token) {
1199 case GREP_PATTERN_HEAD:
1200 if ((field != GREP_HEADER_FIELD_MAX) &&
1201 (p->field != field))
1202 continue;
1203 /* fall thru */
1204 case GREP_PATTERN: /* atom */
1205 case GREP_PATTERN_BODY:
1206 hit |= match_next_pattern(p, bol, eol, ctx,
1207 pmatch, eflags);
1208 break;
1209 default:
1210 break;
1211 }
1212 }
1213 }
1214 return hit;
1215 }
1216
1217 static void show_line_header(struct grep_opt *opt, const char *name,
1218 unsigned lno, ssize_t cno, char sign)
1219 {
1220 if (opt->heading && opt->last_shown == 0) {
1221 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1222 opt->output(opt, "\n", 1);
1223 }
1224 opt->last_shown = lno;
1225
1226 if (!opt->heading && opt->pathname) {
1227 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1228 output_sep(opt, sign);
1229 }
1230 if (opt->linenum) {
1231 char buf[32];
1232 xsnprintf(buf, sizeof(buf), "%d", lno);
1233 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1234 output_sep(opt, sign);
1235 }
1236 /*
1237 * Treat 'cno' as the 1-indexed offset from the start of a non-context
1238 * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1239 * being called with a context line.
1240 */
1241 if (opt->columnnum && cno) {
1242 char buf[32];
1243 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1244 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1245 output_sep(opt, sign);
1246 }
1247 }
1248
1249 static void show_line(struct grep_opt *opt,
1250 const char *bol, const char *eol,
1251 const char *name, unsigned lno, ssize_t cno, char sign)
1252 {
1253 int rest = eol - bol;
1254 const char *match_color = NULL;
1255 const char *line_color = NULL;
1256
1257 if (opt->file_break && opt->last_shown == 0) {
1258 if (opt->show_hunk_mark)
1259 opt->output(opt, "\n", 1);
1260 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1261 if (opt->last_shown == 0) {
1262 if (opt->show_hunk_mark) {
1263 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1264 opt->output(opt, "\n", 1);
1265 }
1266 } else if (lno > opt->last_shown + 1) {
1267 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1268 opt->output(opt, "\n", 1);
1269 }
1270 }
1271 if (!opt->only_matching) {
1272 /*
1273 * In case the line we're being called with contains more than
1274 * one match, leave printing each header to the loop below.
1275 */
1276 show_line_header(opt, name, lno, cno, sign);
1277 }
1278 if (opt->color || opt->only_matching) {
1279 regmatch_t match;
1280 enum grep_context ctx = GREP_CONTEXT_BODY;
1281 int eflags = 0;
1282
1283 if (opt->color) {
1284 if (sign == ':')
1285 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1286 else
1287 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1288 if (sign == ':')
1289 line_color = opt->colors[GREP_COLOR_SELECTED];
1290 else if (sign == '-')
1291 line_color = opt->colors[GREP_COLOR_CONTEXT];
1292 else if (sign == '=')
1293 line_color = opt->colors[GREP_COLOR_FUNCTION];
1294 }
1295 while (grep_next_match(opt, bol, eol, ctx, &match,
1296 GREP_HEADER_FIELD_MAX, eflags)) {
1297 if (match.rm_so == match.rm_eo)
1298 break;
1299
1300 if (opt->only_matching)
1301 show_line_header(opt, name, lno, cno, sign);
1302 else
1303 output_color(opt, bol, match.rm_so, line_color);
1304 output_color(opt, bol + match.rm_so,
1305 match.rm_eo - match.rm_so, match_color);
1306 if (opt->only_matching)
1307 opt->output(opt, "\n", 1);
1308 bol += match.rm_eo;
1309 cno += match.rm_eo;
1310 rest -= match.rm_eo;
1311 eflags = REG_NOTBOL;
1312 }
1313 }
1314 if (!opt->only_matching) {
1315 output_color(opt, bol, rest, line_color);
1316 opt->output(opt, "\n", 1);
1317 }
1318 }
1319
1320 int grep_use_locks;
1321
1322 /*
1323 * This lock protects access to the gitattributes machinery, which is
1324 * not thread-safe.
1325 */
1326 pthread_mutex_t grep_attr_mutex;
1327
1328 static inline void grep_attr_lock(void)
1329 {
1330 if (grep_use_locks)
1331 pthread_mutex_lock(&grep_attr_mutex);
1332 }
1333
1334 static inline void grep_attr_unlock(void)
1335 {
1336 if (grep_use_locks)
1337 pthread_mutex_unlock(&grep_attr_mutex);
1338 }
1339
1340 static int match_funcname(struct grep_opt *opt, struct grep_source *gs,
1341 const char *bol, const char *eol)
1342 {
1343 xdemitconf_t *xecfg = opt->priv;
1344 if (xecfg && !xecfg->find_func) {
1345 grep_source_load_driver(gs, opt->repo->index);
1346 if (gs->driver->funcname.pattern) {
1347 const struct userdiff_funcname *pe = &gs->driver->funcname;
1348 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1349 } else {
1350 xecfg = opt->priv = NULL;
1351 }
1352 }
1353
1354 if (xecfg) {
1355 char buf[1];
1356 return xecfg->find_func(bol, eol - bol, buf, 1,
1357 xecfg->find_func_priv) >= 0;
1358 }
1359
1360 if (bol == eol)
1361 return 0;
1362 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1363 return 1;
1364 return 0;
1365 }
1366
1367 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1368 const char *bol, unsigned lno)
1369 {
1370 while (bol > gs->buf) {
1371 const char *eol = --bol;
1372
1373 while (bol > gs->buf && bol[-1] != '\n')
1374 bol--;
1375 lno--;
1376
1377 if (lno <= opt->last_shown)
1378 break;
1379
1380 if (match_funcname(opt, gs, bol, eol)) {
1381 show_line(opt, bol, eol, gs->name, lno, 0, '=');
1382 break;
1383 }
1384 }
1385 }
1386
1387 static int is_empty_line(const char *bol, const char *eol);
1388
1389 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1390 const char *bol, const char *end, unsigned lno)
1391 {
1392 unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1393 int funcname_needed = !!opt->funcname, comment_needed = 0;
1394
1395 if (opt->pre_context < lno)
1396 from = lno - opt->pre_context;
1397 if (from <= opt->last_shown)
1398 from = opt->last_shown + 1;
1399 orig_from = from;
1400 if (opt->funcbody) {
1401 if (match_funcname(opt, gs, bol, end))
1402 comment_needed = 1;
1403 else
1404 funcname_needed = 1;
1405 from = opt->last_shown + 1;
1406 }
1407
1408 /* Rewind. */
1409 while (bol > gs->buf && cur > from) {
1410 const char *next_bol = bol;
1411 const char *eol = --bol;
1412
1413 while (bol > gs->buf && bol[-1] != '\n')
1414 bol--;
1415 cur--;
1416 if (comment_needed && (is_empty_line(bol, eol) ||
1417 match_funcname(opt, gs, bol, eol))) {
1418 comment_needed = 0;
1419 from = orig_from;
1420 if (cur < from) {
1421 cur++;
1422 bol = next_bol;
1423 break;
1424 }
1425 }
1426 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1427 funcname_lno = cur;
1428 funcname_needed = 0;
1429 if (opt->funcbody)
1430 comment_needed = 1;
1431 else
1432 from = orig_from;
1433 }
1434 }
1435
1436 /* We need to look even further back to find a function signature. */
1437 if (opt->funcname && funcname_needed)
1438 show_funcname_line(opt, gs, bol, cur);
1439
1440 /* Back forward. */
1441 while (cur < lno) {
1442 const char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1443
1444 while (*eol != '\n')
1445 eol++;
1446 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1447 bol = eol + 1;
1448 cur++;
1449 }
1450 }
1451
1452 static int should_lookahead(struct grep_opt *opt)
1453 {
1454 struct grep_pat *p;
1455
1456 if (opt->extended)
1457 return 0; /* punt for too complex stuff */
1458 if (opt->invert)
1459 return 0;
1460 for (p = opt->pattern_list; p; p = p->next) {
1461 if (p->token != GREP_PATTERN)
1462 return 0; /* punt for "header only" and stuff */
1463 }
1464 return 1;
1465 }
1466
1467 static int look_ahead(struct grep_opt *opt,
1468 unsigned long *left_p,
1469 unsigned *lno_p,
1470 const char **bol_p)
1471 {
1472 unsigned lno = *lno_p;
1473 const char *bol = *bol_p;
1474 struct grep_pat *p;
1475 const char *sp, *last_bol;
1476 regoff_t earliest = -1;
1477
1478 for (p = opt->pattern_list; p; p = p->next) {
1479 int hit;
1480 regmatch_t m;
1481
1482 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1483 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1484 continue;
1485 if (earliest < 0 || m.rm_so < earliest)
1486 earliest = m.rm_so;
1487 }
1488
1489 if (earliest < 0) {
1490 *bol_p = bol + *left_p;
1491 *left_p = 0;
1492 return 1;
1493 }
1494 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1495 ; /* find the beginning of the line */
1496 last_bol = sp;
1497
1498 for (sp = bol; sp < last_bol; sp++) {
1499 if (*sp == '\n')
1500 lno++;
1501 }
1502 *left_p -= last_bol - bol;
1503 *bol_p = last_bol;
1504 *lno_p = lno;
1505 return 0;
1506 }
1507
1508 static int fill_textconv_grep(struct repository *r,
1509 struct userdiff_driver *driver,
1510 struct grep_source *gs)
1511 {
1512 struct diff_filespec *df;
1513 char *buf;
1514 size_t size;
1515
1516 if (!driver || !driver->textconv)
1517 return grep_source_load(gs);
1518
1519 /*
1520 * The textconv interface is intimately tied to diff_filespecs, so we
1521 * have to pretend to be one. If we could unify the grep_source
1522 * and diff_filespec structs, this mess could just go away.
1523 */
1524 df = alloc_filespec(gs->path);
1525 switch (gs->type) {
1526 case GREP_SOURCE_OID:
1527 fill_filespec(df, gs->identifier, 1, 0100644);
1528 break;
1529 case GREP_SOURCE_FILE:
1530 fill_filespec(df, null_oid(), 0, 0100644);
1531 break;
1532 default:
1533 BUG("attempt to textconv something without a path?");
1534 }
1535
1536 /*
1537 * fill_textconv is not remotely thread-safe; it modifies the global
1538 * diff tempfile structure, writes to the_repo's odb and might
1539 * internally call thread-unsafe functions such as the
1540 * prepare_packed_git() lazy-initializator. Because of the last two, we
1541 * must ensure mutual exclusion between this call and the object reading
1542 * API, thus we use obj_read_lock() here.
1543 *
1544 * TODO: allowing text conversion to run in parallel with object
1545 * reading operations might increase performance in the multithreaded
1546 * non-worktreee git-grep with --textconv.
1547 */
1548 obj_read_lock();
1549 size = fill_textconv(r, driver, df, &buf);
1550 obj_read_unlock();
1551 free_filespec(df);
1552
1553 /*
1554 * The normal fill_textconv usage by the diff machinery would just keep
1555 * the textconv'd buf separate from the diff_filespec. But much of the
1556 * grep code passes around a grep_source and assumes that its "buf"
1557 * pointer is the beginning of the thing we are searching. So let's
1558 * install our textconv'd version into the grep_source, taking care not
1559 * to leak any existing buffer.
1560 */
1561 grep_source_clear_data(gs);
1562 gs->buf = buf;
1563 gs->size = size;
1564
1565 return 0;
1566 }
1567
1568 static int is_empty_line(const char *bol, const char *eol)
1569 {
1570 while (bol < eol && isspace(*bol))
1571 bol++;
1572 return bol == eol;
1573 }
1574
1575 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1576 {
1577 const char *bol;
1578 const char *peek_bol = NULL;
1579 unsigned long left;
1580 unsigned lno = 1;
1581 unsigned last_hit = 0;
1582 int binary_match_only = 0;
1583 unsigned count = 0;
1584 int try_lookahead = 0;
1585 int show_function = 0;
1586 struct userdiff_driver *textconv = NULL;
1587 enum grep_context ctx = GREP_CONTEXT_HEAD;
1588 xdemitconf_t xecfg;
1589
1590 if (!opt->status_only && gs->name == NULL)
1591 BUG("grep call which could print a name requires "
1592 "grep_source.name be non-NULL");
1593
1594 if (!opt->output)
1595 opt->output = std_output;
1596
1597 if (opt->pre_context || opt->post_context || opt->file_break ||
1598 opt->funcbody) {
1599 /* Show hunk marks, except for the first file. */
1600 if (opt->last_shown)
1601 opt->show_hunk_mark = 1;
1602 /*
1603 * If we're using threads then we can't easily identify
1604 * the first file. Always put hunk marks in that case
1605 * and skip the very first one later in work_done().
1606 */
1607 if (opt->output != std_output)
1608 opt->show_hunk_mark = 1;
1609 }
1610 opt->last_shown = 0;
1611
1612 if (opt->allow_textconv) {
1613 grep_source_load_driver(gs, opt->repo->index);
1614 /*
1615 * We might set up the shared textconv cache data here, which
1616 * is not thread-safe. Also, get_oid_with_context() and
1617 * parse_object() might be internally called. As they are not
1618 * currently thread-safe and might be racy with object reading,
1619 * obj_read_lock() must be called.
1620 */
1621 grep_attr_lock();
1622 obj_read_lock();
1623 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1624 obj_read_unlock();
1625 grep_attr_unlock();
1626 }
1627
1628 /*
1629 * We know the result of a textconv is text, so we only have to care
1630 * about binary handling if we are not using it.
1631 */
1632 if (!textconv) {
1633 switch (opt->binary) {
1634 case GREP_BINARY_DEFAULT:
1635 if (grep_source_is_binary(gs, opt->repo->index))
1636 binary_match_only = 1;
1637 break;
1638 case GREP_BINARY_NOMATCH:
1639 if (grep_source_is_binary(gs, opt->repo->index))
1640 return 0; /* Assume unmatch */
1641 break;
1642 case GREP_BINARY_TEXT:
1643 break;
1644 default:
1645 BUG("unknown binary handling mode");
1646 }
1647 }
1648
1649 memset(&xecfg, 0, sizeof(xecfg));
1650 opt->priv = &xecfg;
1651
1652 try_lookahead = should_lookahead(opt);
1653
1654 if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1655 return 0;
1656
1657 bol = gs->buf;
1658 left = gs->size;
1659 while (left) {
1660 const char *eol;
1661 int hit;
1662 ssize_t cno;
1663 ssize_t col = -1, icol = -1;
1664
1665 /*
1666 * look_ahead() skips quickly to the line that possibly
1667 * has the next hit; don't call it if we need to do
1668 * something more than just skipping the current line
1669 * in response to an unmatch for the current line. E.g.
1670 * inside a post-context window, we will show the current
1671 * line as a context around the previous hit when it
1672 * doesn't hit.
1673 */
1674 if (try_lookahead
1675 && !(last_hit
1676 && (show_function ||
1677 lno <= last_hit + opt->post_context))
1678 && look_ahead(opt, &left, &lno, &bol))
1679 break;
1680 eol = end_of_line(bol, &left);
1681
1682 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1683 ctx = GREP_CONTEXT_BODY;
1684
1685 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1686
1687 if (collect_hits)
1688 goto next_line;
1689
1690 /* "grep -v -e foo -e bla" should list lines
1691 * that do not have either, so inversion should
1692 * be done outside.
1693 */
1694 if (opt->invert)
1695 hit = !hit;
1696 if (opt->unmatch_name_only) {
1697 if (hit)
1698 return 0;
1699 goto next_line;
1700 }
1701 if (hit) {
1702 count++;
1703 if (opt->status_only)
1704 return 1;
1705 if (opt->name_only) {
1706 show_name(opt, gs->name);
1707 return 1;
1708 }
1709 if (opt->count)
1710 goto next_line;
1711 if (binary_match_only) {
1712 opt->output(opt, "Binary file ", 12);
1713 output_color(opt, gs->name, strlen(gs->name),
1714 opt->colors[GREP_COLOR_FILENAME]);
1715 opt->output(opt, " matches\n", 9);
1716 return 1;
1717 }
1718 /* Hit at this line. If we haven't shown the
1719 * pre-context lines, we would need to show them.
1720 */
1721 if (opt->pre_context || opt->funcbody)
1722 show_pre_context(opt, gs, bol, eol, lno);
1723 else if (opt->funcname)
1724 show_funcname_line(opt, gs, bol, lno);
1725 cno = opt->invert ? icol : col;
1726 if (cno < 0) {
1727 /*
1728 * A negative cno indicates that there was no
1729 * match on the line. We are thus inverted and
1730 * being asked to show all lines that _don't_
1731 * match a given expression. Therefore, set cno
1732 * to 0 to suggest the whole line matches.
1733 */
1734 cno = 0;
1735 }
1736 show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1737 last_hit = lno;
1738 if (opt->funcbody)
1739 show_function = 1;
1740 goto next_line;
1741 }
1742 if (show_function && (!peek_bol || peek_bol < bol)) {
1743 unsigned long peek_left = left;
1744 const char *peek_eol = eol;
1745
1746 /*
1747 * Trailing empty lines are not interesting.
1748 * Peek past them to see if they belong to the
1749 * body of the current function.
1750 */
1751 peek_bol = bol;
1752 while (is_empty_line(peek_bol, peek_eol)) {
1753 peek_bol = peek_eol + 1;
1754 peek_eol = end_of_line(peek_bol, &peek_left);
1755 }
1756
1757 if (match_funcname(opt, gs, peek_bol, peek_eol))
1758 show_function = 0;
1759 }
1760 if (show_function ||
1761 (last_hit && lno <= last_hit + opt->post_context)) {
1762 /* If the last hit is within the post context,
1763 * we need to show this line.
1764 */
1765 show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1766 }
1767
1768 next_line:
1769 bol = eol + 1;
1770 if (!left)
1771 break;
1772 left--;
1773 lno++;
1774 }
1775
1776 if (collect_hits)
1777 return 0;
1778
1779 if (opt->status_only)
1780 return opt->unmatch_name_only;
1781 if (opt->unmatch_name_only) {
1782 /* We did not see any hit, so we want to show this */
1783 show_name(opt, gs->name);
1784 return 1;
1785 }
1786
1787 xdiff_clear_find_func(&xecfg);
1788 opt->priv = NULL;
1789
1790 /* NEEDSWORK:
1791 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1792 * which feels mostly useless but sometimes useful. Maybe
1793 * make it another option? For now suppress them.
1794 */
1795 if (opt->count && count) {
1796 char buf[32];
1797 if (opt->pathname) {
1798 output_color(opt, gs->name, strlen(gs->name),
1799 opt->colors[GREP_COLOR_FILENAME]);
1800 output_sep(opt, ':');
1801 }
1802 xsnprintf(buf, sizeof(buf), "%u\n", count);
1803 opt->output(opt, buf, strlen(buf));
1804 return 1;
1805 }
1806 return !!last_hit;
1807 }
1808
1809 static void clr_hit_marker(struct grep_expr *x)
1810 {
1811 /* All-hit markers are meaningful only at the very top level
1812 * OR node.
1813 */
1814 while (1) {
1815 x->hit = 0;
1816 if (x->node != GREP_NODE_OR)
1817 return;
1818 x->u.binary.left->hit = 0;
1819 x = x->u.binary.right;
1820 }
1821 }
1822
1823 static int chk_hit_marker(struct grep_expr *x)
1824 {
1825 /* Top level nodes have hit markers. See if they all are hits */
1826 while (1) {
1827 if (x->node != GREP_NODE_OR)
1828 return x->hit;
1829 if (!x->u.binary.left->hit)
1830 return 0;
1831 x = x->u.binary.right;
1832 }
1833 }
1834
1835 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1836 {
1837 /*
1838 * we do not have to do the two-pass grep when we do not check
1839 * buffer-wide "all-match".
1840 */
1841 if (!opt->all_match && !opt->no_body_match)
1842 return grep_source_1(opt, gs, 0);
1843
1844 /* Otherwise the toplevel "or" terms hit a bit differently.
1845 * We first clear hit markers from them.
1846 */
1847 clr_hit_marker(opt->pattern_expression);
1848 opt->body_hit = 0;
1849 grep_source_1(opt, gs, 1);
1850
1851 if (opt->all_match && !chk_hit_marker(opt->pattern_expression))
1852 return 0;
1853 if (opt->no_body_match && opt->body_hit)
1854 return 0;
1855
1856 return grep_source_1(opt, gs, 0);
1857 }
1858
1859 static void grep_source_init_buf(struct grep_source *gs,
1860 const char *buf,
1861 unsigned long size)
1862 {
1863 gs->type = GREP_SOURCE_BUF;
1864 gs->name = NULL;
1865 gs->path = NULL;
1866 gs->buf = buf;
1867 gs->size = size;
1868 gs->driver = NULL;
1869 gs->identifier = NULL;
1870 }
1871
1872 int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size)
1873 {
1874 struct grep_source gs;
1875 int r;
1876
1877 grep_source_init_buf(&gs, buf, size);
1878
1879 r = grep_source(opt, &gs);
1880
1881 grep_source_clear(&gs);
1882 return r;
1883 }
1884
1885 void grep_source_init_file(struct grep_source *gs, const char *name,
1886 const char *path)
1887 {
1888 gs->type = GREP_SOURCE_FILE;
1889 gs->name = xstrdup_or_null(name);
1890 gs->path = xstrdup_or_null(path);
1891 gs->buf = NULL;
1892 gs->size = 0;
1893 gs->driver = NULL;
1894 gs->identifier = xstrdup(path);
1895 }
1896
1897 void grep_source_init_oid(struct grep_source *gs, const char *name,
1898 const char *path, const struct object_id *oid,
1899 struct repository *repo)
1900 {
1901 gs->type = GREP_SOURCE_OID;
1902 gs->name = xstrdup_or_null(name);
1903 gs->path = xstrdup_or_null(path);
1904 gs->buf = NULL;
1905 gs->size = 0;
1906 gs->driver = NULL;
1907 gs->identifier = oiddup(oid);
1908 gs->repo = repo;
1909 }
1910
1911 void grep_source_clear(struct grep_source *gs)
1912 {
1913 FREE_AND_NULL(gs->name);
1914 FREE_AND_NULL(gs->path);
1915 FREE_AND_NULL(gs->identifier);
1916 grep_source_clear_data(gs);
1917 }
1918
1919 void grep_source_clear_data(struct grep_source *gs)
1920 {
1921 switch (gs->type) {
1922 case GREP_SOURCE_FILE:
1923 case GREP_SOURCE_OID:
1924 /* these types own the buffer */
1925 free((char *)gs->buf);
1926 gs->buf = NULL;
1927 gs->size = 0;
1928 break;
1929 case GREP_SOURCE_BUF:
1930 /* leave user-provided buf intact */
1931 break;
1932 }
1933 }
1934
1935 static int grep_source_load_oid(struct grep_source *gs)
1936 {
1937 enum object_type type;
1938
1939 gs->buf = repo_read_object_file(gs->repo, gs->identifier, &type,
1940 &gs->size);
1941 if (!gs->buf)
1942 return error(_("'%s': unable to read %s"),
1943 gs->name,
1944 oid_to_hex(gs->identifier));
1945 return 0;
1946 }
1947
1948 static int grep_source_load_file(struct grep_source *gs)
1949 {
1950 const char *filename = gs->identifier;
1951 struct stat st;
1952 char *data;
1953 size_t size;
1954 int i;
1955
1956 if (lstat(filename, &st) < 0) {
1957 err_ret:
1958 if (errno != ENOENT)
1959 error_errno(_("failed to stat '%s'"), filename);
1960 return -1;
1961 }
1962 if (!S_ISREG(st.st_mode))
1963 return -1;
1964 size = xsize_t(st.st_size);
1965 i = open(filename, O_RDONLY);
1966 if (i < 0)
1967 goto err_ret;
1968 data = xmallocz(size);
1969 if (st.st_size != read_in_full(i, data, size)) {
1970 error_errno(_("'%s': short read"), filename);
1971 close(i);
1972 free(data);
1973 return -1;
1974 }
1975 close(i);
1976
1977 gs->buf = data;
1978 gs->size = size;
1979 return 0;
1980 }
1981
1982 static int grep_source_load(struct grep_source *gs)
1983 {
1984 if (gs->buf)
1985 return 0;
1986
1987 switch (gs->type) {
1988 case GREP_SOURCE_FILE:
1989 return grep_source_load_file(gs);
1990 case GREP_SOURCE_OID:
1991 return grep_source_load_oid(gs);
1992 case GREP_SOURCE_BUF:
1993 return gs->buf ? 0 : -1;
1994 }
1995 BUG("invalid grep_source type to load");
1996 }
1997
1998 void grep_source_load_driver(struct grep_source *gs,
1999 struct index_state *istate)
2000 {
2001 if (gs->driver)
2002 return;
2003
2004 grep_attr_lock();
2005 if (gs->path)
2006 gs->driver = userdiff_find_by_path(istate, gs->path);
2007 if (!gs->driver)
2008 gs->driver = userdiff_find_by_name("default");
2009 grep_attr_unlock();
2010 }
2011
2012 static int grep_source_is_binary(struct grep_source *gs,
2013 struct index_state *istate)
2014 {
2015 grep_source_load_driver(gs, istate);
2016 if (gs->driver->binary != -1)
2017 return gs->driver->binary;
2018
2019 if (!grep_source_load(gs))
2020 return buffer_is_binary(gs->buf, gs->size);
2021
2022 return 0;
2023 }
git for Debian