?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'ma/doc-markup-fix'
[git/debian.git] / grep.c
blobaabfaaa4c32e48569805dbe631db64fcf7b773c6
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] = "",
30 [GREP_COLOR_FUNCTION] = "",
31 [GREP_COLOR_LINENO] = "",
32 [GREP_COLOR_COLUMNNO] = "",
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 #ifdef USE_LIBPCRE2
44 static pcre2_general_context *pcre2_global_context;
45
46 static void *pcre2_malloc(PCRE2_SIZE size, MAYBE_UNUSED void *memory_data)
47 {
48 return malloc(size);
49 }
50
51 static void pcre2_free(void *pointer, MAYBE_UNUSED void *memory_data)
52 {
53 free(pointer);
54 }
55 #endif
56
57 static const char *color_grep_slots[] = {
58 [GREP_COLOR_CONTEXT] = "context",
59 [GREP_COLOR_FILENAME] = "filename",
60 [GREP_COLOR_FUNCTION] = "function",
61 [GREP_COLOR_LINENO] = "lineNumber",
62 [GREP_COLOR_COLUMNNO] = "column",
63 [GREP_COLOR_MATCH_CONTEXT] = "matchContext",
64 [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
65 [GREP_COLOR_SELECTED] = "selected",
66 [GREP_COLOR_SEP] = "separator",
67 };
68
69 static int parse_pattern_type_arg(const char *opt, const char *arg)
70 {
71 if (!strcmp(arg, "default"))
72 return GREP_PATTERN_TYPE_UNSPECIFIED;
73 else if (!strcmp(arg, "basic"))
74 return GREP_PATTERN_TYPE_BRE;
75 else if (!strcmp(arg, "extended"))
76 return GREP_PATTERN_TYPE_ERE;
77 else if (!strcmp(arg, "fixed"))
78 return GREP_PATTERN_TYPE_FIXED;
79 else if (!strcmp(arg, "perl"))
80 return GREP_PATTERN_TYPE_PCRE;
81 die("bad %s argument: %s", opt, arg);
82 }
83
84 define_list_config_array_extra(color_grep_slots, {"match"});
85
86 /*
87 * Read the configuration file once and store it in
88 * the grep_defaults template.
89 */
90 int grep_config(const char *var, const char *value, void *cb)
91 {
92 struct grep_opt *opt = &grep_defaults;
93 const char *slot;
94
95 if (userdiff_config(var, value) < 0)
96 return -1;
97
98 /*
99 * The instance of grep_opt that we set up here is copied by
100 * grep_init() to be used by each individual invocation.
101 * When populating a new field of this structure here, be
102 * sure to think about ownership -- e.g., you might need to
103 * override the shallow copy in grep_init() with a deep copy.
104 */
105
106 if (!strcmp(var, "grep.extendedregexp")) {
107 opt->extended_regexp_option = git_config_bool(var, value);
108 return 0;
109 }
110
111 if (!strcmp(var, "grep.patterntype")) {
112 opt->pattern_type_option = parse_pattern_type_arg(var, value);
113 return 0;
114 }
115
116 if (!strcmp(var, "grep.linenumber")) {
117 opt->linenum = git_config_bool(var, value);
118 return 0;
119 }
120 if (!strcmp(var, "grep.column")) {
121 opt->columnnum = git_config_bool(var, value);
122 return 0;
123 }
124
125 if (!strcmp(var, "grep.fullname")) {
126 opt->relative = !git_config_bool(var, value);
127 return 0;
128 }
129
130 if (!strcmp(var, "color.grep"))
131 opt->color = git_config_colorbool(var, value);
132 if (!strcmp(var, "color.grep.match")) {
133 if (grep_config("color.grep.matchcontext", value, cb) < 0)
134 return -1;
135 if (grep_config("color.grep.matchselected", value, cb) < 0)
136 return -1;
137 } else if (skip_prefix(var, "color.grep.", &slot)) {
138 int i = LOOKUP_CONFIG(color_grep_slots, slot);
139 char *color;
140
141 if (i < 0)
142 return -1;
143 color = opt->colors[i];
144 if (!value)
145 return config_error_nonbool(var);
146 return color_parse(value, color);
147 }
148 return 0;
149 }
150
151 /*
152 * Initialize one instance of grep_opt and copy the
153 * default values from the template we read the configuration
154 * information in an earlier call to git_config(grep_config).
155 *
156 * If using PCRE, make sure that the library is configured
157 * to use the same allocator as Git (e.g. nedmalloc on Windows).
158 *
159 * Any allocated memory needs to be released in grep_destroy().
160 */
161 void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
162 {
163 #if defined(USE_LIBPCRE2)
164 if (!pcre2_global_context)
165 pcre2_global_context = pcre2_general_context_create(
166 pcre2_malloc, pcre2_free, NULL);
167 #endif
168
169 *opt = grep_defaults;
170
171 opt->repo = repo;
172 opt->prefix = prefix;
173 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
174 opt->pattern_tail = &opt->pattern_list;
175 opt->header_tail = &opt->header_list;
176 }
177
178 void grep_destroy(void)
179 {
180 #ifdef USE_LIBPCRE2
181 pcre2_general_context_free(pcre2_global_context);
182 #endif
183 }
184
185 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
186 {
187 /*
188 * When committing to the pattern type by setting the relevant
189 * fields in grep_opt it's generally not necessary to zero out
190 * the fields we're not choosing, since they won't have been
191 * set by anything. The extended_regexp_option field is the
192 * only exception to this.
193 *
194 * This is because in the process of parsing grep.patternType
195 * & grep.extendedRegexp we set opt->pattern_type_option and
196 * opt->extended_regexp_option, respectively. We then
197 * internally use opt->extended_regexp_option to see if we're
198 * compiling an ERE. It must be unset if that's not actually
199 * the case.
200 */
201 if (pattern_type != GREP_PATTERN_TYPE_ERE &&
202 opt->extended_regexp_option)
203 opt->extended_regexp_option = 0;
204
205 switch (pattern_type) {
206 case GREP_PATTERN_TYPE_UNSPECIFIED:
207 /* fall through */
208
209 case GREP_PATTERN_TYPE_BRE:
210 break;
211
212 case GREP_PATTERN_TYPE_ERE:
213 opt->extended_regexp_option = 1;
214 break;
215
216 case GREP_PATTERN_TYPE_FIXED:
217 opt->fixed = 1;
218 break;
219
220 case GREP_PATTERN_TYPE_PCRE:
221 opt->pcre2 = 1;
222 break;
223 }
224 }
225
226 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
227 {
228 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
229 grep_set_pattern_type_option(pattern_type, opt);
230 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
231 grep_set_pattern_type_option(opt->pattern_type_option, opt);
232 else if (opt->extended_regexp_option)
233 /*
234 * This branch *must* happen after setting from the
235 * opt->pattern_type_option above, we don't want
236 * grep.extendedRegexp to override grep.patternType!
237 */
238 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
239 }
240
241 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
242 const char *origin, int no,
243 enum grep_pat_token t,
244 enum grep_header_field field)
245 {
246 struct grep_pat *p = xcalloc(1, sizeof(*p));
247 p->pattern = xmemdupz(pat, patlen);
248 p->patternlen = patlen;
249 p->origin = origin;
250 p->no = no;
251 p->token = t;
252 p->field = field;
253 return p;
254 }
255
256 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
257 {
258 **tail = p;
259 *tail = &p->next;
260 p->next = NULL;
261
262 switch (p->token) {
263 case GREP_PATTERN: /* atom */
264 case GREP_PATTERN_HEAD:
265 case GREP_PATTERN_BODY:
266 for (;;) {
267 struct grep_pat *new_pat;
268 size_t len = 0;
269 char *cp = p->pattern + p->patternlen, *nl = NULL;
270 while (++len <= p->patternlen) {
271 if (*(--cp) == '\n') {
272 nl = cp;
273 break;
274 }
275 }
276 if (!nl)
277 break;
278 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
279 p->no, p->token, p->field);
280 new_pat->next = p->next;
281 if (!p->next)
282 *tail = &new_pat->next;
283 p->next = new_pat;
284 *nl = '\0';
285 p->patternlen -= len;
286 }
287 break;
288 default:
289 break;
290 }
291 }
292
293 void append_header_grep_pattern(struct grep_opt *opt,
294 enum grep_header_field field, const char *pat)
295 {
296 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
297 GREP_PATTERN_HEAD, field);
298 if (field == GREP_HEADER_REFLOG)
299 opt->use_reflog_filter = 1;
300 do_append_grep_pat(&opt->header_tail, p);
301 }
302
303 void append_grep_pattern(struct grep_opt *opt, const char *pat,
304 const char *origin, int no, enum grep_pat_token t)
305 {
306 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
307 }
308
309 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
310 const char *origin, int no, enum grep_pat_token t)
311 {
312 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
313 do_append_grep_pat(&opt->pattern_tail, p);
314 }
315
316 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
317 {
318 struct grep_pat *pat;
319 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
320 *ret = *opt;
321
322 ret->pattern_list = NULL;
323 ret->pattern_tail = &ret->pattern_list;
324
325 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
326 {
327 if(pat->token == GREP_PATTERN_HEAD)
328 append_header_grep_pattern(ret, pat->field,
329 pat->pattern);
330 else
331 append_grep_pat(ret, pat->pattern, pat->patternlen,
332 pat->origin, pat->no, pat->token);
333 }
334
335 return ret;
336 }
337
338 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
339 const char *error)
340 {
341 char where[1024];
342
343 if (p->no)
344 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
345 else if (p->origin)
346 xsnprintf(where, sizeof(where), "%s, ", p->origin);
347 else
348 where[0] = 0;
349
350 die("%s'%s': %s", where, p->pattern, error);
351 }
352
353 static int is_fixed(const char *s, size_t len)
354 {
355 size_t i;
356
357 for (i = 0; i < len; i++) {
358 if (is_regex_special(s[i]))
359 return 0;
360 }
361
362 return 1;
363 }
364
365 #ifdef USE_LIBPCRE2
366 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
367 {
368 int error;
369 PCRE2_UCHAR errbuf[256];
370 PCRE2_SIZE erroffset;
371 int options = PCRE2_MULTILINE;
372 int jitret;
373 int patinforet;
374 size_t jitsizearg;
375
376 assert(opt->pcre2);
377
378 p->pcre2_compile_context = NULL;
379
380 /* pcre2_global_context is initialized in append_grep_pattern */
381 if (opt->ignore_case) {
382 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
383 if (!pcre2_global_context)
384 BUG("pcre2_global_context uninitialized");
385 p->pcre2_tables = pcre2_maketables(pcre2_global_context);
386 p->pcre2_compile_context = pcre2_compile_context_create(NULL);
387 pcre2_set_character_tables(p->pcre2_compile_context,
388 p->pcre2_tables);
389 }
390 options |= PCRE2_CASELESS;
391 }
392 if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern) &&
393 !(!opt->ignore_case && (p->fixed || p->is_fixed)))
394 options |= (PCRE2_UTF | PCRE2_MATCH_INVALID_UTF);
395
396 /* Work around https://bugs.exim.org/show_bug.cgi?id=2642 fixed in 10.36 */
397 if (PCRE2_MATCH_INVALID_UTF && options & (PCRE2_UTF | PCRE2_CASELESS)) {
398 struct strbuf buf;
399 int len;
400 int err;
401
402 if ((len = pcre2_config(PCRE2_CONFIG_VERSION, NULL)) < 0)
403 BUG("pcre2_config(..., NULL) failed: %d", len);
404 strbuf_init(&buf, len + 1);
405 if ((err = pcre2_config(PCRE2_CONFIG_VERSION, buf.buf)) < 0)
406 BUG("pcre2_config(..., buf.buf) failed: %d", err);
407 if (versioncmp(buf.buf, "10.36") < 0)
408 options |= PCRE2_NO_START_OPTIMIZE;
409 strbuf_release(&buf);
410 }
411
412 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
413 p->patternlen, options, &error, &erroffset,
414 p->pcre2_compile_context);
415
416 if (p->pcre2_pattern) {
417 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
418 if (!p->pcre2_match_data)
419 die("Couldn't allocate PCRE2 match data");
420 } else {
421 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
422 compile_regexp_failed(p, (const char *)&errbuf);
423 }
424
425 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
426 if (p->pcre2_jit_on) {
427 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
428 if (jitret)
429 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
430
431 /*
432 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
433 * tells us whether the library itself supports JIT,
434 * but to see whether we're going to be actually using
435 * JIT we need to extract PCRE2_INFO_JITSIZE from the
436 * pattern *after* we do pcre2_jit_compile() above.
437 *
438 * This is because if the pattern contains the
439 * (*NO_JIT) verb (see pcre2syntax(3))
440 * pcre2_jit_compile() will exit early with 0. If we
441 * then proceed to call pcre2_jit_match() further down
442 * the line instead of pcre2_match() we'll either
443 * segfault (pre PCRE 10.31) or run into a fatal error
444 * (post PCRE2 10.31)
445 */
446 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
447 if (patinforet)
448 BUG("pcre2_pattern_info() failed: %d", patinforet);
449 if (jitsizearg == 0) {
450 p->pcre2_jit_on = 0;
451 return;
452 }
453 }
454 }
455
456 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
457 regmatch_t *match, int eflags)
458 {
459 int ret, flags = 0;
460 PCRE2_SIZE *ovector;
461 PCRE2_UCHAR errbuf[256];
462
463 if (eflags & REG_NOTBOL)
464 flags |= PCRE2_NOTBOL;
465
466 if (p->pcre2_jit_on)
467 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
468 eol - line, 0, flags, p->pcre2_match_data,
469 NULL);
470 else
471 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
472 eol - line, 0, flags, p->pcre2_match_data,
473 NULL);
474
475 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
476 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
477 die("%s failed with error code %d: %s",
478 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
479 errbuf);
480 }
481 if (ret > 0) {
482 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
483 ret = 0;
484 match->rm_so = (int)ovector[0];
485 match->rm_eo = (int)ovector[1];
486 }
487
488 return ret;
489 }
490
491 static void free_pcre2_pattern(struct grep_pat *p)
492 {
493 pcre2_compile_context_free(p->pcre2_compile_context);
494 pcre2_code_free(p->pcre2_pattern);
495 pcre2_match_data_free(p->pcre2_match_data);
496 free((void *)p->pcre2_tables);
497 }
498 #else /* !USE_LIBPCRE2 */
499 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
500 {
501 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
502 }
503
504 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
505 regmatch_t *match, int eflags)
506 {
507 return 1;
508 }
509
510 static void free_pcre2_pattern(struct grep_pat *p)
511 {
512 }
513
514 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
515 {
516 struct strbuf sb = STRBUF_INIT;
517 int err;
518 int regflags = 0;
519
520 basic_regex_quote_buf(&sb, p->pattern);
521 if (opt->ignore_case)
522 regflags |= REG_ICASE;
523 err = regcomp(&p->regexp, sb.buf, regflags);
524 strbuf_release(&sb);
525 if (err) {
526 char errbuf[1024];
527 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
528 compile_regexp_failed(p, errbuf);
529 }
530 }
531 #endif /* !USE_LIBPCRE2 */
532
533 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
534 {
535 int err;
536 int regflags = REG_NEWLINE;
537
538 p->word_regexp = opt->word_regexp;
539 p->ignore_case = opt->ignore_case;
540 p->fixed = opt->fixed;
541
542 if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2)
543 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
544
545 p->is_fixed = is_fixed(p->pattern, p->patternlen);
546 #ifdef USE_LIBPCRE2
547 if (!p->fixed && !p->is_fixed) {
548 const char *no_jit = "(*NO_JIT)";
549 const int no_jit_len = strlen(no_jit);
550 if (starts_with(p->pattern, no_jit) &&
551 is_fixed(p->pattern + no_jit_len,
552 p->patternlen - no_jit_len))
553 p->is_fixed = 1;
554 }
555 #endif
556 if (p->fixed || p->is_fixed) {
557 #ifdef USE_LIBPCRE2
558 opt->pcre2 = 1;
559 if (p->is_fixed) {
560 compile_pcre2_pattern(p, opt);
561 } else {
562 /*
563 * E.g. t7811-grep-open.sh relies on the
564 * pattern being restored.
565 */
566 char *old_pattern = p->pattern;
567 size_t old_patternlen = p->patternlen;
568 struct strbuf sb = STRBUF_INIT;
569
570 /*
571 * There is the PCRE2_LITERAL flag, but it's
572 * only in PCRE v2 10.30 and later. Needing to
573 * ifdef our way around that and dealing with
574 * it + PCRE2_MULTILINE being an error is more
575 * complex than just quoting this ourselves.
576 */
577 strbuf_add(&sb, "\\Q", 2);
578 strbuf_add(&sb, p->pattern, p->patternlen);
579 strbuf_add(&sb, "\\E", 2);
580
581 p->pattern = sb.buf;
582 p->patternlen = sb.len;
583 compile_pcre2_pattern(p, opt);
584 p->pattern = old_pattern;
585 p->patternlen = old_patternlen;
586 strbuf_release(&sb);
587 }
588 #else /* !USE_LIBPCRE2 */
589 compile_fixed_regexp(p, opt);
590 #endif /* !USE_LIBPCRE2 */
591 return;
592 }
593
594 if (opt->pcre2) {
595 compile_pcre2_pattern(p, opt);
596 return;
597 }
598
599 if (p->ignore_case)
600 regflags |= REG_ICASE;
601 if (opt->extended_regexp_option)
602 regflags |= REG_EXTENDED;
603 err = regcomp(&p->regexp, p->pattern, regflags);
604 if (err) {
605 char errbuf[1024];
606 regerror(err, &p->regexp, errbuf, 1024);
607 compile_regexp_failed(p, errbuf);
608 }
609 }
610
611 static struct grep_expr *compile_pattern_or(struct grep_pat **);
612 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
613 {
614 struct grep_pat *p;
615 struct grep_expr *x;
616
617 p = *list;
618 if (!p)
619 return NULL;
620 switch (p->token) {
621 case GREP_PATTERN: /* atom */
622 case GREP_PATTERN_HEAD:
623 case GREP_PATTERN_BODY:
624 x = xcalloc(1, sizeof (struct grep_expr));
625 x->node = GREP_NODE_ATOM;
626 x->u.atom = p;
627 *list = p->next;
628 return x;
629 case GREP_OPEN_PAREN:
630 *list = p->next;
631 x = compile_pattern_or(list);
632 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
633 die("unmatched parenthesis");
634 *list = (*list)->next;
635 return x;
636 default:
637 return NULL;
638 }
639 }
640
641 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
642 {
643 struct grep_pat *p;
644 struct grep_expr *x;
645
646 p = *list;
647 if (!p)
648 return NULL;
649 switch (p->token) {
650 case GREP_NOT:
651 if (!p->next)
652 die("--not not followed by pattern expression");
653 *list = p->next;
654 x = xcalloc(1, sizeof (struct grep_expr));
655 x->node = GREP_NODE_NOT;
656 x->u.unary = compile_pattern_not(list);
657 if (!x->u.unary)
658 die("--not followed by non pattern expression");
659 return x;
660 default:
661 return compile_pattern_atom(list);
662 }
663 }
664
665 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
666 {
667 struct grep_pat *p;
668 struct grep_expr *x, *y, *z;
669
670 x = compile_pattern_not(list);
671 p = *list;
672 if (p && p->token == GREP_AND) {
673 if (!p->next)
674 die("--and not followed by pattern expression");
675 *list = p->next;
676 y = compile_pattern_and(list);
677 if (!y)
678 die("--and not followed by pattern expression");
679 z = xcalloc(1, sizeof (struct grep_expr));
680 z->node = GREP_NODE_AND;
681 z->u.binary.left = x;
682 z->u.binary.right = y;
683 return z;
684 }
685 return x;
686 }
687
688 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
689 {
690 struct grep_pat *p;
691 struct grep_expr *x, *y, *z;
692
693 x = compile_pattern_and(list);
694 p = *list;
695 if (x && p && p->token != GREP_CLOSE_PAREN) {
696 y = compile_pattern_or(list);
697 if (!y)
698 die("not a pattern expression %s", p->pattern);
699 z = xcalloc(1, sizeof (struct grep_expr));
700 z->node = GREP_NODE_OR;
701 z->u.binary.left = x;
702 z->u.binary.right = y;
703 return z;
704 }
705 return x;
706 }
707
708 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
709 {
710 return compile_pattern_or(list);
711 }
712
713 static struct grep_expr *grep_true_expr(void)
714 {
715 struct grep_expr *z = xcalloc(1, sizeof(*z));
716 z->node = GREP_NODE_TRUE;
717 return z;
718 }
719
720 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
721 {
722 struct grep_expr *z = xcalloc(1, sizeof(*z));
723 z->node = GREP_NODE_OR;
724 z->u.binary.left = left;
725 z->u.binary.right = right;
726 return z;
727 }
728
729 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
730 {
731 struct grep_pat *p;
732 struct grep_expr *header_expr;
733 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
734 enum grep_header_field fld;
735
736 if (!opt->header_list)
737 return NULL;
738
739 for (p = opt->header_list; p; p = p->next) {
740 if (p->token != GREP_PATTERN_HEAD)
741 BUG("a non-header pattern in grep header list.");
742 if (p->field < GREP_HEADER_FIELD_MIN ||
743 GREP_HEADER_FIELD_MAX <= p->field)
744 BUG("unknown header field %d", p->field);
745 compile_regexp(p, opt);
746 }
747
748 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
749 header_group[fld] = NULL;
750
751 for (p = opt->header_list; p; p = p->next) {
752 struct grep_expr *h;
753 struct grep_pat *pp = p;
754
755 h = compile_pattern_atom(&pp);
756 if (!h || pp != p->next)
757 BUG("malformed header expr");
758 if (!header_group[p->field]) {
759 header_group[p->field] = h;
760 continue;
761 }
762 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
763 }
764
765 header_expr = NULL;
766
767 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
768 if (!header_group[fld])
769 continue;
770 if (!header_expr)
771 header_expr = grep_true_expr();
772 header_expr = grep_or_expr(header_group[fld], header_expr);
773 }
774 return header_expr;
775 }
776
777 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
778 {
779 struct grep_expr *z = x;
780
781 while (x) {
782 assert(x->node == GREP_NODE_OR);
783 if (x->u.binary.right &&
784 x->u.binary.right->node == GREP_NODE_TRUE) {
785 x->u.binary.right = y;
786 break;
787 }
788 x = x->u.binary.right;
789 }
790 return z;
791 }
792
793 void compile_grep_patterns(struct grep_opt *opt)
794 {
795 struct grep_pat *p;
796 struct grep_expr *header_expr = prep_header_patterns(opt);
797
798 for (p = opt->pattern_list; p; p = p->next) {
799 switch (p->token) {
800 case GREP_PATTERN: /* atom */
801 case GREP_PATTERN_HEAD:
802 case GREP_PATTERN_BODY:
803 compile_regexp(p, opt);
804 break;
805 default:
806 opt->extended = 1;
807 break;
808 }
809 }
810
811 if (opt->all_match || header_expr)
812 opt->extended = 1;
813 else if (!opt->extended)
814 return;
815
816 p = opt->pattern_list;
817 if (p)
818 opt->pattern_expression = compile_pattern_expr(&p);
819 if (p)
820 die("incomplete pattern expression: %s", p->pattern);
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 char *end_of_line(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, char *line, char *eol,
923 regmatch_t *match, int eflags)
924 {
925 int hit;
926
927 if (p->pcre2_pattern)
928 hit = !pcre2match(p, line, eol, match, eflags);
929 else
930 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
931 eflags);
932
933 return hit;
934 }
935
936 static int strip_timestamp(char *bol, char **eol_p)
937 {
938 char *eol = *eol_p;
939 int ch;
940
941 while (bol < --eol) {
942 if (*eol != '>')
943 continue;
944 *eol_p = ++eol;
945 ch = *eol;
946 *eol = '\0';
947 return ch;
948 }
949 return 0;
950 }
951
952 static struct {
953 const char *field;
954 size_t len;
955 } header_field[] = {
956 { "author ", 7 },
957 { "committer ", 10 },
958 { "reflog ", 7 },
959 };
960
961 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
962 enum grep_context ctx,
963 regmatch_t *pmatch, int eflags)
964 {
965 int hit = 0;
966 int saved_ch = 0;
967 const char *start = bol;
968
969 if ((p->token != GREP_PATTERN) &&
970 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
971 return 0;
972
973 if (p->token == GREP_PATTERN_HEAD) {
974 const char *field;
975 size_t len;
976 assert(p->field < ARRAY_SIZE(header_field));
977 field = header_field[p->field].field;
978 len = header_field[p->field].len;
979 if (strncmp(bol, field, len))
980 return 0;
981 bol += len;
982 switch (p->field) {
983 case GREP_HEADER_AUTHOR:
984 case GREP_HEADER_COMMITTER:
985 saved_ch = strip_timestamp(bol, &eol);
986 break;
987 default:
988 break;
989 }
990 }
991
992 again:
993 hit = patmatch(p, bol, eol, pmatch, eflags);
994
995 if (hit && p->word_regexp) {
996 if ((pmatch[0].rm_so < 0) ||
997 (eol - bol) < pmatch[0].rm_so ||
998 (pmatch[0].rm_eo < 0) ||
999 (eol - bol) < pmatch[0].rm_eo)
1000 die("regexp returned nonsense");
1001
1002 /* Match beginning must be either beginning of the
1003 * line, or at word boundary (i.e. the last char must
1004 * not be a word char). Similarly, match end must be
1005 * either end of the line, or at word boundary
1006 * (i.e. the next char must not be a word char).
1007 */
1008 if ( ((pmatch[0].rm_so == 0) ||
1009 !word_char(bol[pmatch[0].rm_so-1])) &&
1010 ((pmatch[0].rm_eo == (eol-bol)) ||
1011 !word_char(bol[pmatch[0].rm_eo])) )
1012 ;
1013 else
1014 hit = 0;
1015
1016 /* Words consist of at least one character. */
1017 if (pmatch->rm_so == pmatch->rm_eo)
1018 hit = 0;
1019
1020 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1021 /* There could be more than one match on the
1022 * line, and the first match might not be
1023 * strict word match. But later ones could be!
1024 * Forward to the next possible start, i.e. the
1025 * next position following a non-word char.
1026 */
1027 bol = pmatch[0].rm_so + bol + 1;
1028 while (word_char(bol[-1]) && bol < eol)
1029 bol++;
1030 eflags |= REG_NOTBOL;
1031 if (bol < eol)
1032 goto again;
1033 }
1034 }
1035 if (p->token == GREP_PATTERN_HEAD && saved_ch)
1036 *eol = saved_ch;
1037 if (hit) {
1038 pmatch[0].rm_so += bol - start;
1039 pmatch[0].rm_eo += bol - start;
1040 }
1041 return hit;
1042 }
1043
1044 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol,
1045 char *eol, enum grep_context ctx, ssize_t *col,
1046 ssize_t *icol, int collect_hits)
1047 {
1048 int h = 0;
1049
1050 if (!x)
1051 die("Not a valid grep expression");
1052 switch (x->node) {
1053 case GREP_NODE_TRUE:
1054 h = 1;
1055 break;
1056 case GREP_NODE_ATOM:
1057 {
1058 regmatch_t tmp;
1059 h = match_one_pattern(x->u.atom, bol, eol, ctx,
1060 &tmp, 0);
1061 if (h && (*col < 0 || tmp.rm_so < *col))
1062 *col = tmp.rm_so;
1063 }
1064 break;
1065 case GREP_NODE_NOT:
1066 /*
1067 * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1068 */
1069 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1070 0);
1071 break;
1072 case GREP_NODE_AND:
1073 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1074 icol, 0);
1075 if (h || opt->columnnum) {
1076 /*
1077 * Don't short-circuit AND when given --column, since a
1078 * NOT earlier in the tree may turn this into an OR. In
1079 * this case, see the below comment.
1080 */
1081 h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1082 ctx, col, icol, 0);
1083 }
1084 break;
1085 case GREP_NODE_OR:
1086 if (!(collect_hits || opt->columnnum)) {
1087 /*
1088 * Don't short-circuit OR when given --column (or
1089 * collecting hits) to ensure we don't skip a later
1090 * child that would produce an earlier match.
1091 */
1092 return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1093 ctx, col, icol, 0) ||
1094 match_expr_eval(opt, x->u.binary.right, bol,
1095 eol, ctx, col, icol, 0));
1096 }
1097 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1098 icol, 0);
1099 if (collect_hits)
1100 x->u.binary.left->hit |= h;
1101 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1102 icol, collect_hits);
1103 break;
1104 default:
1105 die("Unexpected node type (internal error) %d", x->node);
1106 }
1107 if (collect_hits)
1108 x->hit |= h;
1109 return h;
1110 }
1111
1112 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1113 enum grep_context ctx, ssize_t *col,
1114 ssize_t *icol, int collect_hits)
1115 {
1116 struct grep_expr *x = opt->pattern_expression;
1117 return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1118 }
1119
1120 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1121 ssize_t *col, ssize_t *icol,
1122 enum grep_context ctx, int collect_hits)
1123 {
1124 struct grep_pat *p;
1125 int hit = 0;
1126
1127 if (opt->extended)
1128 return match_expr(opt, bol, eol, ctx, col, icol,
1129 collect_hits);
1130
1131 /* we do not call with collect_hits without being extended */
1132 for (p = opt->pattern_list; p; p = p->next) {
1133 regmatch_t tmp;
1134 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1135 hit |= 1;
1136 if (!opt->columnnum) {
1137 /*
1138 * Without --column, any single match on a line
1139 * is enough to know that it needs to be
1140 * printed. With --column, scan _all_ patterns
1141 * to find the earliest.
1142 */
1143 break;
1144 }
1145 if (*col < 0 || tmp.rm_so < *col)
1146 *col = tmp.rm_so;
1147 }
1148 }
1149 return hit;
1150 }
1151
1152 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1153 enum grep_context ctx,
1154 regmatch_t *pmatch, int eflags)
1155 {
1156 regmatch_t match;
1157
1158 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1159 return 0;
1160 if (match.rm_so < 0 || match.rm_eo < 0)
1161 return 0;
1162 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1163 if (match.rm_so > pmatch->rm_so)
1164 return 1;
1165 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1166 return 1;
1167 }
1168 pmatch->rm_so = match.rm_so;
1169 pmatch->rm_eo = match.rm_eo;
1170 return 1;
1171 }
1172
1173 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1174 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1175 {
1176 struct grep_pat *p;
1177 int hit = 0;
1178
1179 pmatch->rm_so = pmatch->rm_eo = -1;
1180 if (bol < eol) {
1181 for (p = opt->pattern_list; p; p = p->next) {
1182 switch (p->token) {
1183 case GREP_PATTERN: /* atom */
1184 case GREP_PATTERN_HEAD:
1185 case GREP_PATTERN_BODY:
1186 hit |= match_next_pattern(p, bol, eol, ctx,
1187 pmatch, eflags);
1188 break;
1189 default:
1190 break;
1191 }
1192 }
1193 }
1194 return hit;
1195 }
1196
1197 static void show_line_header(struct grep_opt *opt, const char *name,
1198 unsigned lno, ssize_t cno, char sign)
1199 {
1200 if (opt->heading && opt->last_shown == 0) {
1201 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1202 opt->output(opt, "\n", 1);
1203 }
1204 opt->last_shown = lno;
1205
1206 if (!opt->heading && opt->pathname) {
1207 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1208 output_sep(opt, sign);
1209 }
1210 if (opt->linenum) {
1211 char buf[32];
1212 xsnprintf(buf, sizeof(buf), "%d", lno);
1213 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1214 output_sep(opt, sign);
1215 }
1216 /*
1217 * Treat 'cno' as the 1-indexed offset from the start of a non-context
1218 * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1219 * being called with a context line.
1220 */
1221 if (opt->columnnum && cno) {
1222 char buf[32];
1223 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1224 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1225 output_sep(opt, sign);
1226 }
1227 }
1228
1229 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1230 const char *name, unsigned lno, ssize_t cno, char sign)
1231 {
1232 int rest = eol - bol;
1233 const char *match_color = NULL;
1234 const char *line_color = NULL;
1235
1236 if (opt->file_break && opt->last_shown == 0) {
1237 if (opt->show_hunk_mark)
1238 opt->output(opt, "\n", 1);
1239 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1240 if (opt->last_shown == 0) {
1241 if (opt->show_hunk_mark) {
1242 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1243 opt->output(opt, "\n", 1);
1244 }
1245 } else if (lno > opt->last_shown + 1) {
1246 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1247 opt->output(opt, "\n", 1);
1248 }
1249 }
1250 if (!opt->only_matching) {
1251 /*
1252 * In case the line we're being called with contains more than
1253 * one match, leave printing each header to the loop below.
1254 */
1255 show_line_header(opt, name, lno, cno, sign);
1256 }
1257 if (opt->color || opt->only_matching) {
1258 regmatch_t match;
1259 enum grep_context ctx = GREP_CONTEXT_BODY;
1260 int ch = *eol;
1261 int eflags = 0;
1262
1263 if (opt->color) {
1264 if (sign == ':')
1265 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1266 else
1267 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1268 if (sign == ':')
1269 line_color = opt->colors[GREP_COLOR_SELECTED];
1270 else if (sign == '-')
1271 line_color = opt->colors[GREP_COLOR_CONTEXT];
1272 else if (sign == '=')
1273 line_color = opt->colors[GREP_COLOR_FUNCTION];
1274 }
1275 *eol = '\0';
1276 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1277 if (match.rm_so == match.rm_eo)
1278 break;
1279
1280 if (opt->only_matching)
1281 show_line_header(opt, name, lno, cno, sign);
1282 else
1283 output_color(opt, bol, match.rm_so, line_color);
1284 output_color(opt, bol + match.rm_so,
1285 match.rm_eo - match.rm_so, match_color);
1286 if (opt->only_matching)
1287 opt->output(opt, "\n", 1);
1288 bol += match.rm_eo;
1289 cno += match.rm_eo;
1290 rest -= match.rm_eo;
1291 eflags = REG_NOTBOL;
1292 }
1293 *eol = ch;
1294 }
1295 if (!opt->only_matching) {
1296 output_color(opt, bol, rest, line_color);
1297 opt->output(opt, "\n", 1);
1298 }
1299 }
1300
1301 int grep_use_locks;
1302
1303 /*
1304 * This lock protects access to the gitattributes machinery, which is
1305 * not thread-safe.
1306 */
1307 pthread_mutex_t grep_attr_mutex;
1308
1309 static inline void grep_attr_lock(void)
1310 {
1311 if (grep_use_locks)
1312 pthread_mutex_lock(&grep_attr_mutex);
1313 }
1314
1315 static inline void grep_attr_unlock(void)
1316 {
1317 if (grep_use_locks)
1318 pthread_mutex_unlock(&grep_attr_mutex);
1319 }
1320
1321 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1322 {
1323 xdemitconf_t *xecfg = opt->priv;
1324 if (xecfg && !xecfg->find_func) {
1325 grep_source_load_driver(gs, opt->repo->index);
1326 if (gs->driver->funcname.pattern) {
1327 const struct userdiff_funcname *pe = &gs->driver->funcname;
1328 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1329 } else {
1330 xecfg = opt->priv = NULL;
1331 }
1332 }
1333
1334 if (xecfg) {
1335 char buf[1];
1336 return xecfg->find_func(bol, eol - bol, buf, 1,
1337 xecfg->find_func_priv) >= 0;
1338 }
1339
1340 if (bol == eol)
1341 return 0;
1342 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1343 return 1;
1344 return 0;
1345 }
1346
1347 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1348 char *bol, unsigned lno)
1349 {
1350 while (bol > gs->buf) {
1351 char *eol = --bol;
1352
1353 while (bol > gs->buf && bol[-1] != '\n')
1354 bol--;
1355 lno--;
1356
1357 if (lno <= opt->last_shown)
1358 break;
1359
1360 if (match_funcname(opt, gs, bol, eol)) {
1361 show_line(opt, bol, eol, gs->name, lno, 0, '=');
1362 break;
1363 }
1364 }
1365 }
1366
1367 static int is_empty_line(const char *bol, const char *eol);
1368
1369 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1370 char *bol, char *end, unsigned lno)
1371 {
1372 unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1373 int funcname_needed = !!opt->funcname, comment_needed = 0;
1374
1375 if (opt->pre_context < lno)
1376 from = lno - opt->pre_context;
1377 if (from <= opt->last_shown)
1378 from = opt->last_shown + 1;
1379 orig_from = from;
1380 if (opt->funcbody) {
1381 if (match_funcname(opt, gs, bol, end))
1382 comment_needed = 1;
1383 else
1384 funcname_needed = 1;
1385 from = opt->last_shown + 1;
1386 }
1387
1388 /* Rewind. */
1389 while (bol > gs->buf && cur > from) {
1390 char *next_bol = bol;
1391 char *eol = --bol;
1392
1393 while (bol > gs->buf && bol[-1] != '\n')
1394 bol--;
1395 cur--;
1396 if (comment_needed && (is_empty_line(bol, eol) ||
1397 match_funcname(opt, gs, bol, eol))) {
1398 comment_needed = 0;
1399 from = orig_from;
1400 if (cur < from) {
1401 cur++;
1402 bol = next_bol;
1403 break;
1404 }
1405 }
1406 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1407 funcname_lno = cur;
1408 funcname_needed = 0;
1409 if (opt->funcbody)
1410 comment_needed = 1;
1411 else
1412 from = orig_from;
1413 }
1414 }
1415
1416 /* We need to look even further back to find a function signature. */
1417 if (opt->funcname && funcname_needed)
1418 show_funcname_line(opt, gs, bol, cur);
1419
1420 /* Back forward. */
1421 while (cur < lno) {
1422 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1423
1424 while (*eol != '\n')
1425 eol++;
1426 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1427 bol = eol + 1;
1428 cur++;
1429 }
1430 }
1431
1432 static int should_lookahead(struct grep_opt *opt)
1433 {
1434 struct grep_pat *p;
1435
1436 if (opt->extended)
1437 return 0; /* punt for too complex stuff */
1438 if (opt->invert)
1439 return 0;
1440 for (p = opt->pattern_list; p; p = p->next) {
1441 if (p->token != GREP_PATTERN)
1442 return 0; /* punt for "header only" and stuff */
1443 }
1444 return 1;
1445 }
1446
1447 static int look_ahead(struct grep_opt *opt,
1448 unsigned long *left_p,
1449 unsigned *lno_p,
1450 char **bol_p)
1451 {
1452 unsigned lno = *lno_p;
1453 char *bol = *bol_p;
1454 struct grep_pat *p;
1455 char *sp, *last_bol;
1456 regoff_t earliest = -1;
1457
1458 for (p = opt->pattern_list; p; p = p->next) {
1459 int hit;
1460 regmatch_t m;
1461
1462 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1463 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1464 continue;
1465 if (earliest < 0 || m.rm_so < earliest)
1466 earliest = m.rm_so;
1467 }
1468
1469 if (earliest < 0) {
1470 *bol_p = bol + *left_p;
1471 *left_p = 0;
1472 return 1;
1473 }
1474 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1475 ; /* find the beginning of the line */
1476 last_bol = sp;
1477
1478 for (sp = bol; sp < last_bol; sp++) {
1479 if (*sp == '\n')
1480 lno++;
1481 }
1482 *left_p -= last_bol - bol;
1483 *bol_p = last_bol;
1484 *lno_p = lno;
1485 return 0;
1486 }
1487
1488 static int fill_textconv_grep(struct repository *r,
1489 struct userdiff_driver *driver,
1490 struct grep_source *gs)
1491 {
1492 struct diff_filespec *df;
1493 char *buf;
1494 size_t size;
1495
1496 if (!driver || !driver->textconv)
1497 return grep_source_load(gs);
1498
1499 /*
1500 * The textconv interface is intimately tied to diff_filespecs, so we
1501 * have to pretend to be one. If we could unify the grep_source
1502 * and diff_filespec structs, this mess could just go away.
1503 */
1504 df = alloc_filespec(gs->path);
1505 switch (gs->type) {
1506 case GREP_SOURCE_OID:
1507 fill_filespec(df, gs->identifier, 1, 0100644);
1508 break;
1509 case GREP_SOURCE_FILE:
1510 fill_filespec(df, &null_oid, 0, 0100644);
1511 break;
1512 default:
1513 BUG("attempt to textconv something without a path?");
1514 }
1515
1516 /*
1517 * fill_textconv is not remotely thread-safe; it modifies the global
1518 * diff tempfile structure, writes to the_repo's odb and might
1519 * internally call thread-unsafe functions such as the
1520 * prepare_packed_git() lazy-initializator. Because of the last two, we
1521 * must ensure mutual exclusion between this call and the object reading
1522 * API, thus we use obj_read_lock() here.
1523 *
1524 * TODO: allowing text conversion to run in parallel with object
1525 * reading operations might increase performance in the multithreaded
1526 * non-worktreee git-grep with --textconv.
1527 */
1528 obj_read_lock();
1529 size = fill_textconv(r, driver, df, &buf);
1530 obj_read_unlock();
1531 free_filespec(df);
1532
1533 /*
1534 * The normal fill_textconv usage by the diff machinery would just keep
1535 * the textconv'd buf separate from the diff_filespec. But much of the
1536 * grep code passes around a grep_source and assumes that its "buf"
1537 * pointer is the beginning of the thing we are searching. So let's
1538 * install our textconv'd version into the grep_source, taking care not
1539 * to leak any existing buffer.
1540 */
1541 grep_source_clear_data(gs);
1542 gs->buf = buf;
1543 gs->size = size;
1544
1545 return 0;
1546 }
1547
1548 static int is_empty_line(const char *bol, const char *eol)
1549 {
1550 while (bol < eol && isspace(*bol))
1551 bol++;
1552 return bol == eol;
1553 }
1554
1555 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1556 {
1557 char *bol;
1558 char *peek_bol = NULL;
1559 unsigned long left;
1560 unsigned lno = 1;
1561 unsigned last_hit = 0;
1562 int binary_match_only = 0;
1563 unsigned count = 0;
1564 int try_lookahead = 0;
1565 int show_function = 0;
1566 struct userdiff_driver *textconv = NULL;
1567 enum grep_context ctx = GREP_CONTEXT_HEAD;
1568 xdemitconf_t xecfg;
1569
1570 if (!opt->status_only && gs->name == NULL)
1571 BUG("grep call which could print a name requires "
1572 "grep_source.name be non-NULL");
1573
1574 if (!opt->output)
1575 opt->output = std_output;
1576
1577 if (opt->pre_context || opt->post_context || opt->file_break ||
1578 opt->funcbody) {
1579 /* Show hunk marks, except for the first file. */
1580 if (opt->last_shown)
1581 opt->show_hunk_mark = 1;
1582 /*
1583 * If we're using threads then we can't easily identify
1584 * the first file. Always put hunk marks in that case
1585 * and skip the very first one later in work_done().
1586 */
1587 if (opt->output != std_output)
1588 opt->show_hunk_mark = 1;
1589 }
1590 opt->last_shown = 0;
1591
1592 if (opt->allow_textconv) {
1593 grep_source_load_driver(gs, opt->repo->index);
1594 /*
1595 * We might set up the shared textconv cache data here, which
1596 * is not thread-safe. Also, get_oid_with_context() and
1597 * parse_object() might be internally called. As they are not
1598 * currently thread-safe and might be racy with object reading,
1599 * obj_read_lock() must be called.
1600 */
1601 grep_attr_lock();
1602 obj_read_lock();
1603 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1604 obj_read_unlock();
1605 grep_attr_unlock();
1606 }
1607
1608 /*
1609 * We know the result of a textconv is text, so we only have to care
1610 * about binary handling if we are not using it.
1611 */
1612 if (!textconv) {
1613 switch (opt->binary) {
1614 case GREP_BINARY_DEFAULT:
1615 if (grep_source_is_binary(gs, opt->repo->index))
1616 binary_match_only = 1;
1617 break;
1618 case GREP_BINARY_NOMATCH:
1619 if (grep_source_is_binary(gs, opt->repo->index))
1620 return 0; /* Assume unmatch */
1621 break;
1622 case GREP_BINARY_TEXT:
1623 break;
1624 default:
1625 BUG("unknown binary handling mode");
1626 }
1627 }
1628
1629 memset(&xecfg, 0, sizeof(xecfg));
1630 opt->priv = &xecfg;
1631
1632 try_lookahead = should_lookahead(opt);
1633
1634 if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1635 return 0;
1636
1637 bol = gs->buf;
1638 left = gs->size;
1639 while (left) {
1640 char *eol, ch;
1641 int hit;
1642 ssize_t cno;
1643 ssize_t col = -1, icol = -1;
1644
1645 /*
1646 * look_ahead() skips quickly to the line that possibly
1647 * has the next hit; don't call it if we need to do
1648 * something more than just skipping the current line
1649 * in response to an unmatch for the current line. E.g.
1650 * inside a post-context window, we will show the current
1651 * line as a context around the previous hit when it
1652 * doesn't hit.
1653 */
1654 if (try_lookahead
1655 && !(last_hit
1656 && (show_function ||
1657 lno <= last_hit + opt->post_context))
1658 && look_ahead(opt, &left, &lno, &bol))
1659 break;
1660 eol = end_of_line(bol, &left);
1661 ch = *eol;
1662 *eol = 0;
1663
1664 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1665 ctx = GREP_CONTEXT_BODY;
1666
1667 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1668 *eol = ch;
1669
1670 if (collect_hits)
1671 goto next_line;
1672
1673 /* "grep -v -e foo -e bla" should list lines
1674 * that do not have either, so inversion should
1675 * be done outside.
1676 */
1677 if (opt->invert)
1678 hit = !hit;
1679 if (opt->unmatch_name_only) {
1680 if (hit)
1681 return 0;
1682 goto next_line;
1683 }
1684 if (hit) {
1685 count++;
1686 if (opt->status_only)
1687 return 1;
1688 if (opt->name_only) {
1689 show_name(opt, gs->name);
1690 return 1;
1691 }
1692 if (opt->count)
1693 goto next_line;
1694 if (binary_match_only) {
1695 opt->output(opt, "Binary file ", 12);
1696 output_color(opt, gs->name, strlen(gs->name),
1697 opt->colors[GREP_COLOR_FILENAME]);
1698 opt->output(opt, " matches\n", 9);
1699 return 1;
1700 }
1701 /* Hit at this line. If we haven't shown the
1702 * pre-context lines, we would need to show them.
1703 */
1704 if (opt->pre_context || opt->funcbody)
1705 show_pre_context(opt, gs, bol, eol, lno);
1706 else if (opt->funcname)
1707 show_funcname_line(opt, gs, bol, lno);
1708 cno = opt->invert ? icol : col;
1709 if (cno < 0) {
1710 /*
1711 * A negative cno indicates that there was no
1712 * match on the line. We are thus inverted and
1713 * being asked to show all lines that _don't_
1714 * match a given expression. Therefore, set cno
1715 * to 0 to suggest the whole line matches.
1716 */
1717 cno = 0;
1718 }
1719 show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1720 last_hit = lno;
1721 if (opt->funcbody)
1722 show_function = 1;
1723 goto next_line;
1724 }
1725 if (show_function && (!peek_bol || peek_bol < bol)) {
1726 unsigned long peek_left = left;
1727 char *peek_eol = eol;
1728
1729 /*
1730 * Trailing empty lines are not interesting.
1731 * Peek past them to see if they belong to the
1732 * body of the current function.
1733 */
1734 peek_bol = bol;
1735 while (is_empty_line(peek_bol, peek_eol)) {
1736 peek_bol = peek_eol + 1;
1737 peek_eol = end_of_line(peek_bol, &peek_left);
1738 }
1739
1740 if (match_funcname(opt, gs, peek_bol, peek_eol))
1741 show_function = 0;
1742 }
1743 if (show_function ||
1744 (last_hit && lno <= last_hit + opt->post_context)) {
1745 /* If the last hit is within the post context,
1746 * we need to show this line.
1747 */
1748 show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1749 }
1750
1751 next_line:
1752 bol = eol + 1;
1753 if (!left)
1754 break;
1755 left--;
1756 lno++;
1757 }
1758
1759 if (collect_hits)
1760 return 0;
1761
1762 if (opt->status_only)
1763 return opt->unmatch_name_only;
1764 if (opt->unmatch_name_only) {
1765 /* We did not see any hit, so we want to show this */
1766 show_name(opt, gs->name);
1767 return 1;
1768 }
1769
1770 xdiff_clear_find_func(&xecfg);
1771 opt->priv = NULL;
1772
1773 /* NEEDSWORK:
1774 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1775 * which feels mostly useless but sometimes useful. Maybe
1776 * make it another option? For now suppress them.
1777 */
1778 if (opt->count && count) {
1779 char buf[32];
1780 if (opt->pathname) {
1781 output_color(opt, gs->name, strlen(gs->name),
1782 opt->colors[GREP_COLOR_FILENAME]);
1783 output_sep(opt, ':');
1784 }
1785 xsnprintf(buf, sizeof(buf), "%u\n", count);
1786 opt->output(opt, buf, strlen(buf));
1787 return 1;
1788 }
1789 return !!last_hit;
1790 }
1791
1792 static void clr_hit_marker(struct grep_expr *x)
1793 {
1794 /* All-hit markers are meaningful only at the very top level
1795 * OR node.
1796 */
1797 while (1) {
1798 x->hit = 0;
1799 if (x->node != GREP_NODE_OR)
1800 return;
1801 x->u.binary.left->hit = 0;
1802 x = x->u.binary.right;
1803 }
1804 }
1805
1806 static int chk_hit_marker(struct grep_expr *x)
1807 {
1808 /* Top level nodes have hit markers. See if they all are hits */
1809 while (1) {
1810 if (x->node != GREP_NODE_OR)
1811 return x->hit;
1812 if (!x->u.binary.left->hit)
1813 return 0;
1814 x = x->u.binary.right;
1815 }
1816 }
1817
1818 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1819 {
1820 /*
1821 * we do not have to do the two-pass grep when we do not check
1822 * buffer-wide "all-match".
1823 */
1824 if (!opt->all_match)
1825 return grep_source_1(opt, gs, 0);
1826
1827 /* Otherwise the toplevel "or" terms hit a bit differently.
1828 * We first clear hit markers from them.
1829 */
1830 clr_hit_marker(opt->pattern_expression);
1831 grep_source_1(opt, gs, 1);
1832
1833 if (!chk_hit_marker(opt->pattern_expression))
1834 return 0;
1835
1836 return grep_source_1(opt, gs, 0);
1837 }
1838
1839 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1840 {
1841 struct grep_source gs;
1842 int r;
1843
1844 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1845 gs.buf = buf;
1846 gs.size = size;
1847
1848 r = grep_source(opt, &gs);
1849
1850 grep_source_clear(&gs);
1851 return r;
1852 }
1853
1854 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1855 const char *name, const char *path,
1856 const void *identifier)
1857 {
1858 gs->type = type;
1859 gs->name = xstrdup_or_null(name);
1860 gs->path = xstrdup_or_null(path);
1861 gs->buf = NULL;
1862 gs->size = 0;
1863 gs->driver = NULL;
1864
1865 switch (type) {
1866 case GREP_SOURCE_FILE:
1867 gs->identifier = xstrdup(identifier);
1868 break;
1869 case GREP_SOURCE_OID:
1870 gs->identifier = oiddup(identifier);
1871 break;
1872 case GREP_SOURCE_BUF:
1873 gs->identifier = NULL;
1874 break;
1875 }
1876 }
1877
1878 void grep_source_clear(struct grep_source *gs)
1879 {
1880 FREE_AND_NULL(gs->name);
1881 FREE_AND_NULL(gs->path);
1882 FREE_AND_NULL(gs->identifier);
1883 grep_source_clear_data(gs);
1884 }
1885
1886 void grep_source_clear_data(struct grep_source *gs)
1887 {
1888 switch (gs->type) {
1889 case GREP_SOURCE_FILE:
1890 case GREP_SOURCE_OID:
1891 FREE_AND_NULL(gs->buf);
1892 gs->size = 0;
1893 break;
1894 case GREP_SOURCE_BUF:
1895 /* leave user-provided buf intact */
1896 break;
1897 }
1898 }
1899
1900 static int grep_source_load_oid(struct grep_source *gs)
1901 {
1902 enum object_type type;
1903
1904 gs->buf = read_object_file(gs->identifier, &type, &gs->size);
1905 if (!gs->buf)
1906 return error(_("'%s': unable to read %s"),
1907 gs->name,
1908 oid_to_hex(gs->identifier));
1909 return 0;
1910 }
1911
1912 static int grep_source_load_file(struct grep_source *gs)
1913 {
1914 const char *filename = gs->identifier;
1915 struct stat st;
1916 char *data;
1917 size_t size;
1918 int i;
1919
1920 if (lstat(filename, &st) < 0) {
1921 err_ret:
1922 if (errno != ENOENT)
1923 error_errno(_("failed to stat '%s'"), filename);
1924 return -1;
1925 }
1926 if (!S_ISREG(st.st_mode))
1927 return -1;
1928 size = xsize_t(st.st_size);
1929 i = open(filename, O_RDONLY);
1930 if (i < 0)
1931 goto err_ret;
1932 data = xmallocz(size);
1933 if (st.st_size != read_in_full(i, data, size)) {
1934 error_errno(_("'%s': short read"), filename);
1935 close(i);
1936 free(data);
1937 return -1;
1938 }
1939 close(i);
1940
1941 gs->buf = data;
1942 gs->size = size;
1943 return 0;
1944 }
1945
1946 static int grep_source_load(struct grep_source *gs)
1947 {
1948 if (gs->buf)
1949 return 0;
1950
1951 switch (gs->type) {
1952 case GREP_SOURCE_FILE:
1953 return grep_source_load_file(gs);
1954 case GREP_SOURCE_OID:
1955 return grep_source_load_oid(gs);
1956 case GREP_SOURCE_BUF:
1957 return gs->buf ? 0 : -1;
1958 }
1959 BUG("invalid grep_source type to load");
1960 }
1961
1962 void grep_source_load_driver(struct grep_source *gs,
1963 struct index_state *istate)
1964 {
1965 if (gs->driver)
1966 return;
1967
1968 grep_attr_lock();
1969 if (gs->path)
1970 gs->driver = userdiff_find_by_path(istate, gs->path);
1971 if (!gs->driver)
1972 gs->driver = userdiff_find_by_name("default");
1973 grep_attr_unlock();
1974 }
1975
1976 static int grep_source_is_binary(struct grep_source *gs,
1977 struct index_state *istate)
1978 {
1979 grep_source_load_driver(gs, istate);
1980 if (gs->driver->binary != -1)
1981 return gs->driver->binary;
1982
1983 if (!grep_source_load(gs))
1984 return buffer_is_binary(gs->buf, gs->size);
1985
1986 return 0;
1987 }
git for Debian