?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
pretty: clear signature check
[git/gitster.git] / grep.c
blobe9337f32cbf6e7afd426420e15d8828dc73ffb10
1 #include "git-compat-util.h"
2 #include "config.h"
3 #include "gettext.h"
4 #include "grep.h"
5 #include "hex.h"
6 #include "object-store-ll.h"
7 #include "pretty.h"
8 #include "userdiff.h"
9 #include "xdiff-interface.h"
10 #include "diff.h"
11 #include "diffcore.h"
12 #include "quote.h"
13 #include "help.h"
14
15 static int grep_source_load(struct grep_source *gs);
16 static int grep_source_is_binary(struct grep_source *gs,
17 struct index_state *istate);
18
19 static void std_output(struct grep_opt *opt UNUSED, const void *buf, size_t size)
20 {
21 fwrite(buf, size, 1, stdout);
22 }
23
24 static const char *color_grep_slots[] = {
25 [GREP_COLOR_CONTEXT] = "context",
26 [GREP_COLOR_FILENAME] = "filename",
27 [GREP_COLOR_FUNCTION] = "function",
28 [GREP_COLOR_LINENO] = "lineNumber",
29 [GREP_COLOR_COLUMNNO] = "column",
30 [GREP_COLOR_MATCH_CONTEXT] = "matchContext",
31 [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
32 [GREP_COLOR_SELECTED] = "selected",
33 [GREP_COLOR_SEP] = "separator",
34 };
35
36 static int parse_pattern_type_arg(const char *opt, const char *arg)
37 {
38 if (!strcmp(arg, "default"))
39 return GREP_PATTERN_TYPE_UNSPECIFIED;
40 else if (!strcmp(arg, "basic"))
41 return GREP_PATTERN_TYPE_BRE;
42 else if (!strcmp(arg, "extended"))
43 return GREP_PATTERN_TYPE_ERE;
44 else if (!strcmp(arg, "fixed"))
45 return GREP_PATTERN_TYPE_FIXED;
46 else if (!strcmp(arg, "perl"))
47 return GREP_PATTERN_TYPE_PCRE;
48 die("bad %s argument: %s", opt, arg);
49 }
50
51 define_list_config_array_extra(color_grep_slots, {"match"});
52
53 /*
54 * Read the configuration file once and store it in
55 * the grep_defaults template.
56 */
57 int grep_config(const char *var, const char *value,
58 const struct config_context *ctx, void *cb)
59 {
60 struct grep_opt *opt = cb;
61 const char *slot;
62
63 if (userdiff_config(var, value) < 0)
64 return -1;
65
66 if (!strcmp(var, "grep.extendedregexp")) {
67 opt->extended_regexp_option = git_config_bool(var, value);
68 return 0;
69 }
70
71 if (!strcmp(var, "grep.patterntype")) {
72 opt->pattern_type_option = parse_pattern_type_arg(var, value);
73 return 0;
74 }
75
76 if (!strcmp(var, "grep.linenumber")) {
77 opt->linenum = git_config_bool(var, value);
78 return 0;
79 }
80 if (!strcmp(var, "grep.column")) {
81 opt->columnnum = git_config_bool(var, value);
82 return 0;
83 }
84
85 if (!strcmp(var, "grep.fullname")) {
86 opt->relative = !git_config_bool(var, value);
87 return 0;
88 }
89
90 if (!strcmp(var, "color.grep"))
91 opt->color = git_config_colorbool(var, value);
92 if (!strcmp(var, "color.grep.match")) {
93 if (grep_config("color.grep.matchcontext", value, ctx, cb) < 0)
94 return -1;
95 if (grep_config("color.grep.matchselected", value, ctx, cb) < 0)
96 return -1;
97 } else if (skip_prefix(var, "color.grep.", &slot)) {
98 int i = LOOKUP_CONFIG(color_grep_slots, slot);
99 char *color;
100
101 if (i < 0)
102 return -1;
103 color = opt->colors[i];
104 if (!value)
105 return config_error_nonbool(var);
106 return color_parse(value, color);
107 }
108 return 0;
109 }
110
111 void grep_init(struct grep_opt *opt, struct repository *repo)
112 {
113 struct grep_opt blank = GREP_OPT_INIT;
114 memcpy(opt, &blank, sizeof(*opt));
115
116 opt->repo = repo;
117 opt->pattern_tail = &opt->pattern_list;
118 opt->header_tail = &opt->header_list;
119 }
120
121 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
122 const char *origin, int no,
123 enum grep_pat_token t,
124 enum grep_header_field field)
125 {
126 struct grep_pat *p = xcalloc(1, sizeof(*p));
127 p->pattern = xmemdupz(pat, patlen);
128 p->patternlen = patlen;
129 p->origin = origin;
130 p->no = no;
131 p->token = t;
132 p->field = field;
133 return p;
134 }
135
136 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
137 {
138 **tail = p;
139 *tail = &p->next;
140 p->next = NULL;
141
142 switch (p->token) {
143 case GREP_PATTERN: /* atom */
144 case GREP_PATTERN_HEAD:
145 case GREP_PATTERN_BODY:
146 for (;;) {
147 struct grep_pat *new_pat;
148 size_t len = 0;
149 char *cp = p->pattern + p->patternlen, *nl = NULL;
150 while (++len <= p->patternlen) {
151 if (*(--cp) == '\n') {
152 nl = cp;
153 break;
154 }
155 }
156 if (!nl)
157 break;
158 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
159 p->no, p->token, p->field);
160 new_pat->next = p->next;
161 if (!p->next)
162 *tail = &new_pat->next;
163 p->next = new_pat;
164 *nl = '\0';
165 p->patternlen -= len;
166 }
167 break;
168 default:
169 break;
170 }
171 }
172
173 void append_header_grep_pattern(struct grep_opt *opt,
174 enum grep_header_field field, const char *pat)
175 {
176 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
177 GREP_PATTERN_HEAD, field);
178 if (field == GREP_HEADER_REFLOG)
179 opt->use_reflog_filter = 1;
180 do_append_grep_pat(&opt->header_tail, p);
181 }
182
183 void append_grep_pattern(struct grep_opt *opt, const char *pat,
184 const char *origin, int no, enum grep_pat_token t)
185 {
186 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
187 }
188
189 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
190 const char *origin, int no, enum grep_pat_token t)
191 {
192 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
193 do_append_grep_pat(&opt->pattern_tail, p);
194 }
195
196 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
197 {
198 struct grep_pat *pat;
199 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
200 *ret = *opt;
201
202 ret->pattern_list = NULL;
203 ret->pattern_tail = &ret->pattern_list;
204
205 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
206 {
207 if(pat->token == GREP_PATTERN_HEAD)
208 append_header_grep_pattern(ret, pat->field,
209 pat->pattern);
210 else
211 append_grep_pat(ret, pat->pattern, pat->patternlen,
212 pat->origin, pat->no, pat->token);
213 }
214
215 return ret;
216 }
217
218 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
219 const char *error)
220 {
221 char where[1024];
222
223 if (p->no)
224 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
225 else if (p->origin)
226 xsnprintf(where, sizeof(where), "%s, ", p->origin);
227 else
228 where[0] = 0;
229
230 die("%s'%s': %s", where, p->pattern, error);
231 }
232
233 static int is_fixed(const char *s, size_t len)
234 {
235 size_t i;
236
237 for (i = 0; i < len; i++) {
238 if (is_regex_special(s[i]))
239 return 0;
240 }
241
242 return 1;
243 }
244
245 #ifdef USE_LIBPCRE2
246 #define GREP_PCRE2_DEBUG_MALLOC 0
247
248 static void *pcre2_malloc(PCRE2_SIZE size, void *memory_data UNUSED)
249 {
250 void *pointer = malloc(size);
251 #if GREP_PCRE2_DEBUG_MALLOC
252 static int count = 1;
253 fprintf(stderr, "PCRE2:%p -> #%02d: alloc(%lu)\n", pointer, count++, size);
254 #endif
255 return pointer;
256 }
257
258 static void pcre2_free(void *pointer, void *memory_data UNUSED)
259 {
260 #if GREP_PCRE2_DEBUG_MALLOC
261 static int count = 1;
262 if (pointer)
263 fprintf(stderr, "PCRE2:%p -> #%02d: free()\n", pointer, count++);
264 #endif
265 free(pointer);
266 }
267
268 static int pcre2_jit_functional(void)
269 {
270 static int jit_working = -1;
271 pcre2_code *code;
272 size_t off;
273 int err;
274
275 if (jit_working != -1)
276 return jit_working;
277
278 /*
279 * Try to JIT compile a simple pattern to probe if the JIT is
280 * working in general. It might fail for systems where creating
281 * memory mappings for runtime code generation is restricted.
282 */
283 code = pcre2_compile((PCRE2_SPTR)".", 1, 0, &err, &off, NULL);
284 if (!code)
285 return 0;
286
287 jit_working = pcre2_jit_compile(code, PCRE2_JIT_COMPLETE) == 0;
288 pcre2_code_free(code);
289
290 return jit_working;
291 }
292
293 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
294 {
295 int error;
296 PCRE2_UCHAR errbuf[256];
297 PCRE2_SIZE erroffset;
298 int options = PCRE2_MULTILINE;
299 int jitret;
300 int patinforet;
301 size_t jitsizearg;
302 int literal = !opt->ignore_case && (p->fixed || p->is_fixed);
303
304 /*
305 * Call pcre2_general_context_create() before calling any
306 * other pcre2_*(). It sets up our malloc()/free() functions
307 * with which everything else is allocated.
308 */
309 p->pcre2_general_context = pcre2_general_context_create(
310 pcre2_malloc, pcre2_free, NULL);
311 if (!p->pcre2_general_context)
312 die("Couldn't allocate PCRE2 general context");
313
314 if (opt->ignore_case) {
315 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
316 p->pcre2_tables = pcre2_maketables(p->pcre2_general_context);
317 p->pcre2_compile_context = pcre2_compile_context_create(p->pcre2_general_context);
318 pcre2_set_character_tables(p->pcre2_compile_context,
319 p->pcre2_tables);
320 }
321 options |= PCRE2_CASELESS;
322 }
323 if (!opt->ignore_locale && is_utf8_locale() && !literal)
324 options |= (PCRE2_UTF | PCRE2_UCP | PCRE2_MATCH_INVALID_UTF);
325
326 #ifndef GIT_PCRE2_VERSION_10_35_OR_HIGHER
327 /*
328 * Work around a JIT bug related to invalid Unicode character handling
329 * fixed in 10.35:
330 * https://github.com/PCRE2Project/pcre2/commit/c21bd977547d
331 */
332 options &= ~PCRE2_UCP;
333 #endif
334
335 #ifndef GIT_PCRE2_VERSION_10_36_OR_HIGHER
336 /* Work around https://bugs.exim.org/show_bug.cgi?id=2642 fixed in 10.36 */
337 if (PCRE2_MATCH_INVALID_UTF && options & (PCRE2_UTF | PCRE2_CASELESS))
338 options |= PCRE2_NO_START_OPTIMIZE;
339 #endif
340
341 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
342 p->patternlen, options, &error, &erroffset,
343 p->pcre2_compile_context);
344
345 if (p->pcre2_pattern) {
346 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, p->pcre2_general_context);
347 if (!p->pcre2_match_data)
348 die("Couldn't allocate PCRE2 match data");
349 } else {
350 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
351 compile_regexp_failed(p, (const char *)&errbuf);
352 }
353
354 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
355 if (p->pcre2_jit_on) {
356 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
357 if (jitret == PCRE2_ERROR_NOMEMORY && !pcre2_jit_functional()) {
358 /*
359 * Even though pcre2_config(PCRE2_CONFIG_JIT, ...)
360 * indicated JIT support, the library might still
361 * fail to generate JIT code for various reasons,
362 * e.g. when SELinux's 'deny_execmem' or PaX's
363 * MPROTECT prevent creating W|X memory mappings.
364 *
365 * Instead of faling hard, fall back to interpreter
366 * mode, just as if the pattern was prefixed with
367 * '(*NO_JIT)'.
368 */
369 p->pcre2_jit_on = 0;
370 return;
371 } else if (jitret) {
372 int need_clip = p->patternlen > 64;
373 int clip_len = need_clip ? 64 : p->patternlen;
374 die("Couldn't JIT the PCRE2 pattern '%.*s'%s, got '%d'%s",
375 clip_len, p->pattern, need_clip ? "..." : "", jitret,
376 pcre2_jit_functional()
377 ? "\nPerhaps prefix (*NO_JIT) to your pattern?"
378 : "");
379 }
380
381 /*
382 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
383 * tells us whether the library itself supports JIT,
384 * but to see whether we're going to be actually using
385 * JIT we need to extract PCRE2_INFO_JITSIZE from the
386 * pattern *after* we do pcre2_jit_compile() above.
387 *
388 * This is because if the pattern contains the
389 * (*NO_JIT) verb (see pcre2syntax(3))
390 * pcre2_jit_compile() will exit early with 0. If we
391 * then proceed to call pcre2_jit_match() further down
392 * the line instead of pcre2_match() we'll either
393 * segfault (pre PCRE 10.31) or run into a fatal error
394 * (post PCRE2 10.31)
395 */
396 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
397 if (patinforet)
398 BUG("pcre2_pattern_info() failed: %d", patinforet);
399 if (jitsizearg == 0) {
400 p->pcre2_jit_on = 0;
401 return;
402 }
403 }
404 }
405
406 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
407 regmatch_t *match, int eflags)
408 {
409 int ret, flags = 0;
410 PCRE2_SIZE *ovector;
411 PCRE2_UCHAR errbuf[256];
412
413 if (eflags & REG_NOTBOL)
414 flags |= PCRE2_NOTBOL;
415
416 if (p->pcre2_jit_on)
417 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
418 eol - line, 0, flags, p->pcre2_match_data,
419 NULL);
420 else
421 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
422 eol - line, 0, flags, p->pcre2_match_data,
423 NULL);
424
425 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
426 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
427 die("%s failed with error code %d: %s",
428 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
429 errbuf);
430 }
431 if (ret > 0) {
432 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
433 ret = 0;
434 match->rm_so = (int)ovector[0];
435 match->rm_eo = (int)ovector[1];
436 }
437
438 return ret;
439 }
440
441 static void free_pcre2_pattern(struct grep_pat *p)
442 {
443 pcre2_compile_context_free(p->pcre2_compile_context);
444 pcre2_code_free(p->pcre2_pattern);
445 pcre2_match_data_free(p->pcre2_match_data);
446 #ifdef GIT_PCRE2_VERSION_10_34_OR_HIGHER
447 pcre2_maketables_free(p->pcre2_general_context, p->pcre2_tables);
448 #else
449 free((void *)p->pcre2_tables);
450 #endif
451 pcre2_general_context_free(p->pcre2_general_context);
452 }
453 #else /* !USE_LIBPCRE2 */
454 static void compile_pcre2_pattern(struct grep_pat *p UNUSED,
455 const struct grep_opt *opt UNUSED)
456 {
457 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
458 }
459
460 static int pcre2match(struct grep_pat *p UNUSED, const char *line UNUSED,
461 const char *eol UNUSED, regmatch_t *match UNUSED,
462 int eflags UNUSED)
463 {
464 return 1;
465 }
466
467 static void free_pcre2_pattern(struct grep_pat *p UNUSED)
468 {
469 }
470
471 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
472 {
473 struct strbuf sb = STRBUF_INIT;
474 int err;
475 int regflags = 0;
476
477 basic_regex_quote_buf(&sb, p->pattern);
478 if (opt->ignore_case)
479 regflags |= REG_ICASE;
480 err = regcomp(&p->regexp, sb.buf, regflags);
481 strbuf_release(&sb);
482 if (err) {
483 char errbuf[1024];
484 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
485 compile_regexp_failed(p, errbuf);
486 }
487 }
488 #endif /* !USE_LIBPCRE2 */
489
490 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
491 {
492 int err;
493 int regflags = REG_NEWLINE;
494
495 if (opt->pattern_type_option == GREP_PATTERN_TYPE_UNSPECIFIED)
496 opt->pattern_type_option = (opt->extended_regexp_option
497 ? GREP_PATTERN_TYPE_ERE
498 : GREP_PATTERN_TYPE_BRE);
499
500 p->word_regexp = opt->word_regexp;
501 p->ignore_case = opt->ignore_case;
502 p->fixed = opt->pattern_type_option == GREP_PATTERN_TYPE_FIXED;
503
504 if (opt->pattern_type_option != GREP_PATTERN_TYPE_PCRE &&
505 memchr(p->pattern, 0, p->patternlen))
506 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
507
508 p->is_fixed = is_fixed(p->pattern, p->patternlen);
509 #ifdef USE_LIBPCRE2
510 if (!p->fixed && !p->is_fixed) {
511 const char *no_jit = "(*NO_JIT)";
512 const int no_jit_len = strlen(no_jit);
513 if (starts_with(p->pattern, no_jit) &&
514 is_fixed(p->pattern + no_jit_len,
515 p->patternlen - no_jit_len))
516 p->is_fixed = 1;
517 }
518 #endif
519 if (p->fixed || p->is_fixed) {
520 #ifdef USE_LIBPCRE2
521 if (p->is_fixed) {
522 compile_pcre2_pattern(p, opt);
523 } else {
524 /*
525 * E.g. t7811-grep-open.sh relies on the
526 * pattern being restored.
527 */
528 char *old_pattern = p->pattern;
529 size_t old_patternlen = p->patternlen;
530 struct strbuf sb = STRBUF_INIT;
531
532 /*
533 * There is the PCRE2_LITERAL flag, but it's
534 * only in PCRE v2 10.30 and later. Needing to
535 * ifdef our way around that and dealing with
536 * it + PCRE2_MULTILINE being an error is more
537 * complex than just quoting this ourselves.
538 */
539 strbuf_add(&sb, "\\Q", 2);
540 strbuf_add(&sb, p->pattern, p->patternlen);
541 strbuf_add(&sb, "\\E", 2);
542
543 p->pattern = sb.buf;
544 p->patternlen = sb.len;
545 compile_pcre2_pattern(p, opt);
546 p->pattern = old_pattern;
547 p->patternlen = old_patternlen;
548 strbuf_release(&sb);
549 }
550 #else /* !USE_LIBPCRE2 */
551 compile_fixed_regexp(p, opt);
552 #endif /* !USE_LIBPCRE2 */
553 return;
554 }
555
556 if (opt->pattern_type_option == GREP_PATTERN_TYPE_PCRE) {
557 compile_pcre2_pattern(p, opt);
558 return;
559 }
560
561 if (p->ignore_case)
562 regflags |= REG_ICASE;
563 if (opt->pattern_type_option == GREP_PATTERN_TYPE_ERE)
564 regflags |= REG_EXTENDED;
565 err = regcomp(&p->regexp, p->pattern, regflags);
566 if (err) {
567 char errbuf[1024];
568 regerror(err, &p->regexp, errbuf, 1024);
569 compile_regexp_failed(p, errbuf);
570 }
571 }
572
573 static struct grep_expr *grep_not_expr(struct grep_expr *expr)
574 {
575 struct grep_expr *z = xcalloc(1, sizeof(*z));
576 z->node = GREP_NODE_NOT;
577 z->u.unary = expr;
578 return z;
579 }
580
581 static struct grep_expr *grep_binexp(enum grep_expr_node kind,
582 struct grep_expr *left,
583 struct grep_expr *right)
584 {
585 struct grep_expr *z = xcalloc(1, sizeof(*z));
586 z->node = kind;
587 z->u.binary.left = left;
588 z->u.binary.right = right;
589 return z;
590 }
591
592 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
593 {
594 return grep_binexp(GREP_NODE_OR, left, right);
595 }
596
597 static struct grep_expr *grep_and_expr(struct grep_expr *left, struct grep_expr *right)
598 {
599 return grep_binexp(GREP_NODE_AND, left, right);
600 }
601
602 static struct grep_expr *compile_pattern_or(struct grep_pat **);
603 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
604 {
605 struct grep_pat *p;
606 struct grep_expr *x;
607
608 p = *list;
609 if (!p)
610 return NULL;
611 switch (p->token) {
612 case GREP_PATTERN: /* atom */
613 case GREP_PATTERN_HEAD:
614 case GREP_PATTERN_BODY:
615 CALLOC_ARRAY(x, 1);
616 x->node = GREP_NODE_ATOM;
617 x->u.atom = p;
618 *list = p->next;
619 return x;
620 case GREP_OPEN_PAREN:
621 *list = p->next;
622 x = compile_pattern_or(list);
623 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
624 die("unmatched ( for expression group");
625 *list = (*list)->next;
626 return x;
627 default:
628 return NULL;
629 }
630 }
631
632 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
633 {
634 struct grep_pat *p;
635 struct grep_expr *x;
636
637 p = *list;
638 if (!p)
639 return NULL;
640 switch (p->token) {
641 case GREP_NOT:
642 if (!p->next)
643 die("--not not followed by pattern expression");
644 *list = p->next;
645 x = compile_pattern_not(list);
646 if (!x)
647 die("--not followed by non pattern expression");
648 return grep_not_expr(x);
649 default:
650 return compile_pattern_atom(list);
651 }
652 }
653
654 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
655 {
656 struct grep_pat *p;
657 struct grep_expr *x, *y;
658
659 x = compile_pattern_not(list);
660 p = *list;
661 if (p && p->token == GREP_AND) {
662 if (!x)
663 die("--and not preceded by pattern expression");
664 if (!p->next)
665 die("--and not followed by pattern expression");
666 *list = p->next;
667 y = compile_pattern_and(list);
668 if (!y)
669 die("--and not followed by pattern expression");
670 return grep_and_expr(x, y);
671 }
672 return x;
673 }
674
675 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
676 {
677 struct grep_pat *p;
678 struct grep_expr *x, *y;
679
680 x = compile_pattern_and(list);
681 p = *list;
682 if (x && p && p->token != GREP_CLOSE_PAREN) {
683 y = compile_pattern_or(list);
684 if (!y)
685 die("not a pattern expression %s", p->pattern);
686 return grep_or_expr(x, y);
687 }
688 return x;
689 }
690
691 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
692 {
693 return compile_pattern_or(list);
694 }
695
696 static struct grep_expr *grep_true_expr(void)
697 {
698 struct grep_expr *z = xcalloc(1, sizeof(*z));
699 z->node = GREP_NODE_TRUE;
700 return z;
701 }
702
703 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
704 {
705 struct grep_pat *p;
706 struct grep_expr *header_expr;
707 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
708 enum grep_header_field fld;
709
710 if (!opt->header_list)
711 return NULL;
712
713 for (p = opt->header_list; p; p = p->next) {
714 if (p->token != GREP_PATTERN_HEAD)
715 BUG("a non-header pattern in grep header list.");
716 if (p->field < GREP_HEADER_FIELD_MIN ||
717 GREP_HEADER_FIELD_MAX <= p->field)
718 BUG("unknown header field %d", p->field);
719 compile_regexp(p, opt);
720 }
721
722 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
723 header_group[fld] = NULL;
724
725 for (p = opt->header_list; p; p = p->next) {
726 struct grep_expr *h;
727 struct grep_pat *pp = p;
728
729 h = compile_pattern_atom(&pp);
730 if (!h || pp != p->next)
731 BUG("malformed header expr");
732 if (!header_group[p->field]) {
733 header_group[p->field] = h;
734 continue;
735 }
736 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
737 }
738
739 header_expr = NULL;
740
741 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
742 if (!header_group[fld])
743 continue;
744 if (!header_expr)
745 header_expr = grep_true_expr();
746 header_expr = grep_or_expr(header_group[fld], header_expr);
747 }
748 return header_expr;
749 }
750
751 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
752 {
753 struct grep_expr *z = x;
754
755 while (x) {
756 assert(x->node == GREP_NODE_OR);
757 if (x->u.binary.right &&
758 x->u.binary.right->node == GREP_NODE_TRUE) {
759 free(x->u.binary.right);
760 x->u.binary.right = y;
761 break;
762 }
763 x = x->u.binary.right;
764 }
765 return z;
766 }
767
768 void compile_grep_patterns(struct grep_opt *opt)
769 {
770 struct grep_pat *p;
771 struct grep_expr *header_expr = prep_header_patterns(opt);
772 int extended = 0;
773
774 for (p = opt->pattern_list; p; p = p->next) {
775 switch (p->token) {
776 case GREP_PATTERN: /* atom */
777 case GREP_PATTERN_HEAD:
778 case GREP_PATTERN_BODY:
779 compile_regexp(p, opt);
780 break;
781 default:
782 extended = 1;
783 break;
784 }
785 }
786
787 if (opt->all_match || opt->no_body_match || header_expr)
788 extended = 1;
789 else if (!extended)
790 return;
791
792 p = opt->pattern_list;
793 if (p)
794 opt->pattern_expression = compile_pattern_expr(&p);
795 if (p)
796 die("incomplete pattern expression group: %s", p->pattern);
797
798 if (opt->no_body_match && opt->pattern_expression)
799 opt->pattern_expression = grep_not_expr(opt->pattern_expression);
800
801 if (!header_expr)
802 return;
803
804 if (!opt->pattern_expression)
805 opt->pattern_expression = header_expr;
806 else if (opt->all_match)
807 opt->pattern_expression = grep_splice_or(header_expr,
808 opt->pattern_expression);
809 else
810 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
811 header_expr);
812 opt->all_match = 1;
813 }
814
815 static void free_pattern_expr(struct grep_expr *x)
816 {
817 switch (x->node) {
818 case GREP_NODE_TRUE:
819 case GREP_NODE_ATOM:
820 break;
821 case GREP_NODE_NOT:
822 free_pattern_expr(x->u.unary);
823 break;
824 case GREP_NODE_AND:
825 case GREP_NODE_OR:
826 free_pattern_expr(x->u.binary.left);
827 free_pattern_expr(x->u.binary.right);
828 break;
829 }
830 free(x);
831 }
832
833 static void free_grep_pat(struct grep_pat *pattern)
834 {
835 struct grep_pat *p, *n;
836
837 for (p = pattern; p; p = n) {
838 n = p->next;
839 switch (p->token) {
840 case GREP_PATTERN: /* atom */
841 case GREP_PATTERN_HEAD:
842 case GREP_PATTERN_BODY:
843 if (p->pcre2_pattern)
844 free_pcre2_pattern(p);
845 else
846 regfree(&p->regexp);
847 break;
848 default:
849 break;
850 }
851 free(p->pattern);
852 free(p);
853 }
854 }
855
856 void free_grep_patterns(struct grep_opt *opt)
857 {
858 free_grep_pat(opt->pattern_list);
859 free_grep_pat(opt->header_list);
860
861 if (opt->pattern_expression)
862 free_pattern_expr(opt->pattern_expression);
863 }
864
865 static const char *end_of_line(const char *cp, unsigned long *left)
866 {
867 unsigned long l = *left;
868 while (l && *cp != '\n') {
869 l--;
870 cp++;
871 }
872 *left = l;
873 return cp;
874 }
875
876 static int word_char(char ch)
877 {
878 return isalnum(ch) || ch == '_';
879 }
880
881 static void output_color(struct grep_opt *opt, const void *data, size_t size,
882 const char *color)
883 {
884 if (want_color(opt->color) && color && color[0]) {
885 opt->output(opt, color, strlen(color));
886 opt->output(opt, data, size);
887 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
888 } else
889 opt->output(opt, data, size);
890 }
891
892 static void output_sep(struct grep_opt *opt, char sign)
893 {
894 if (opt->null_following_name)
895 opt->output(opt, "\0", 1);
896 else
897 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
898 }
899
900 static void show_name(struct grep_opt *opt, const char *name)
901 {
902 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
903 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
904 }
905
906 static int patmatch(struct grep_pat *p,
907 const char *line, const char *eol,
908 regmatch_t *match, int eflags)
909 {
910 int hit;
911
912 if (p->pcre2_pattern)
913 hit = !pcre2match(p, line, eol, match, eflags);
914 else
915 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
916 eflags);
917
918 return hit;
919 }
920
921 static void strip_timestamp(const char *bol, const char **eol_p)
922 {
923 const char *eol = *eol_p;
924
925 while (bol < --eol) {
926 if (*eol != '>')
927 continue;
928 *eol_p = ++eol;
929 break;
930 }
931 }
932
933 static struct {
934 const char *field;
935 size_t len;
936 } header_field[] = {
937 { "author ", 7 },
938 { "committer ", 10 },
939 { "reflog ", 7 },
940 };
941
942 static int headerless_match_one_pattern(struct grep_pat *p,
943 const char *bol, const char *eol,
944 enum grep_context ctx,
945 regmatch_t *pmatch, int eflags)
946 {
947 int hit = 0;
948 const char *start = bol;
949
950 if ((p->token != GREP_PATTERN) &&
951 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
952 return 0;
953
954 again:
955 hit = patmatch(p, bol, eol, pmatch, eflags);
956
957 if (hit && p->word_regexp) {
958 if ((pmatch[0].rm_so < 0) ||
959 (eol - bol) < pmatch[0].rm_so ||
960 (pmatch[0].rm_eo < 0) ||
961 (eol - bol) < pmatch[0].rm_eo)
962 die("regexp returned nonsense");
963
964 /* Match beginning must be either beginning of the
965 * line, or at word boundary (i.e. the last char must
966 * not be a word char). Similarly, match end must be
967 * either end of the line, or at word boundary
968 * (i.e. the next char must not be a word char).
969 */
970 if ( ((pmatch[0].rm_so == 0) ||
971 !word_char(bol[pmatch[0].rm_so-1])) &&
972 ((pmatch[0].rm_eo == (eol-bol)) ||
973 !word_char(bol[pmatch[0].rm_eo])) )
974 ;
975 else
976 hit = 0;
977
978 /* Words consist of at least one character. */
979 if (pmatch->rm_so == pmatch->rm_eo)
980 hit = 0;
981
982 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
983 /* There could be more than one match on the
984 * line, and the first match might not be
985 * strict word match. But later ones could be!
986 * Forward to the next possible start, i.e. the
987 * next position following a non-word char.
988 */
989 bol = pmatch[0].rm_so + bol + 1;
990 while (word_char(bol[-1]) && bol < eol)
991 bol++;
992 eflags |= REG_NOTBOL;
993 if (bol < eol)
994 goto again;
995 }
996 }
997 if (hit) {
998 pmatch[0].rm_so += bol - start;
999 pmatch[0].rm_eo += bol - start;
1000 }
1001 return hit;
1002 }
1003
1004 static int match_one_pattern(struct grep_pat *p,
1005 const char *bol, const char *eol,
1006 enum grep_context ctx, regmatch_t *pmatch,
1007 int eflags)
1008 {
1009 const char *field;
1010 size_t len;
1011
1012 if (p->token == GREP_PATTERN_HEAD) {
1013 assert(p->field < ARRAY_SIZE(header_field));
1014 field = header_field[p->field].field;
1015 len = header_field[p->field].len;
1016 if (strncmp(bol, field, len))
1017 return 0;
1018 bol += len;
1019
1020 switch (p->field) {
1021 case GREP_HEADER_AUTHOR:
1022 case GREP_HEADER_COMMITTER:
1023 strip_timestamp(bol, &eol);
1024 break;
1025 default:
1026 break;
1027 }
1028 }
1029
1030 return headerless_match_one_pattern(p, bol, eol, ctx, pmatch, eflags);
1031 }
1032
1033
1034 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x,
1035 const char *bol, const char *eol,
1036 enum grep_context ctx, ssize_t *col,
1037 ssize_t *icol, int collect_hits)
1038 {
1039 int h = 0;
1040
1041 switch (x->node) {
1042 case GREP_NODE_TRUE:
1043 h = 1;
1044 break;
1045 case GREP_NODE_ATOM:
1046 {
1047 regmatch_t tmp;
1048 h = match_one_pattern(x->u.atom, bol, eol, ctx,
1049 &tmp, 0);
1050 if (h && (*col < 0 || tmp.rm_so < *col))
1051 *col = tmp.rm_so;
1052 }
1053 if (x->u.atom->token == GREP_PATTERN_BODY)
1054 opt->body_hit |= h;
1055 break;
1056 case GREP_NODE_NOT:
1057 /*
1058 * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1059 */
1060 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1061 0);
1062 break;
1063 case GREP_NODE_AND:
1064 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1065 icol, 0);
1066 if (h || opt->columnnum) {
1067 /*
1068 * Don't short-circuit AND when given --column, since a
1069 * NOT earlier in the tree may turn this into an OR. In
1070 * this case, see the below comment.
1071 */
1072 h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1073 ctx, col, icol, 0);
1074 }
1075 break;
1076 case GREP_NODE_OR:
1077 if (!(collect_hits || opt->columnnum)) {
1078 /*
1079 * Don't short-circuit OR when given --column (or
1080 * collecting hits) to ensure we don't skip a later
1081 * child that would produce an earlier match.
1082 */
1083 return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1084 ctx, col, icol, 0) ||
1085 match_expr_eval(opt, x->u.binary.right, bol,
1086 eol, ctx, col, icol, 0));
1087 }
1088 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1089 icol, 0);
1090 if (collect_hits)
1091 x->u.binary.left->hit |= h;
1092 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1093 icol, collect_hits);
1094 break;
1095 default:
1096 die("Unexpected node type (internal error) %d", x->node);
1097 }
1098 if (collect_hits)
1099 x->hit |= h;
1100 return h;
1101 }
1102
1103 static int match_expr(struct grep_opt *opt,
1104 const char *bol, const char *eol,
1105 enum grep_context ctx, ssize_t *col,
1106 ssize_t *icol, int collect_hits)
1107 {
1108 struct grep_expr *x = opt->pattern_expression;
1109 return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1110 }
1111
1112 static int match_line(struct grep_opt *opt,
1113 const char *bol, const char *eol,
1114 ssize_t *col, ssize_t *icol,
1115 enum grep_context ctx, int collect_hits)
1116 {
1117 struct grep_pat *p;
1118 int hit = 0;
1119
1120 if (opt->pattern_expression)
1121 return match_expr(opt, bol, eol, ctx, col, icol,
1122 collect_hits);
1123
1124 /* we do not call with collect_hits without being extended */
1125 for (p = opt->pattern_list; p; p = p->next) {
1126 regmatch_t tmp;
1127 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1128 hit |= 1;
1129 if (!opt->columnnum) {
1130 /*
1131 * Without --column, any single match on a line
1132 * is enough to know that it needs to be
1133 * printed. With --column, scan _all_ patterns
1134 * to find the earliest.
1135 */
1136 break;
1137 }
1138 if (*col < 0 || tmp.rm_so < *col)
1139 *col = tmp.rm_so;
1140 }
1141 }
1142 return hit;
1143 }
1144
1145 static int match_next_pattern(struct grep_pat *p,
1146 const char *bol, const char *eol,
1147 enum grep_context ctx,
1148 regmatch_t *pmatch, int eflags)
1149 {
1150 regmatch_t match;
1151
1152 if (!headerless_match_one_pattern(p, bol, eol, ctx, &match, eflags))
1153 return 0;
1154 if (match.rm_so < 0 || match.rm_eo < 0)
1155 return 0;
1156 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1157 if (match.rm_so > pmatch->rm_so)
1158 return 1;
1159 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1160 return 1;
1161 }
1162 pmatch->rm_so = match.rm_so;
1163 pmatch->rm_eo = match.rm_eo;
1164 return 1;
1165 }
1166
1167 int grep_next_match(struct grep_opt *opt,
1168 const char *bol, const char *eol,
1169 enum grep_context ctx, regmatch_t *pmatch,
1170 enum grep_header_field field, int eflags)
1171 {
1172 struct grep_pat *p;
1173 int hit = 0;
1174
1175 pmatch->rm_so = pmatch->rm_eo = -1;
1176 if (bol < eol) {
1177 for (p = ((ctx == GREP_CONTEXT_HEAD)
1178 ? opt->header_list : opt->pattern_list);
1179 p; p = p->next) {
1180 switch (p->token) {
1181 case GREP_PATTERN_HEAD:
1182 if ((field != GREP_HEADER_FIELD_MAX) &&
1183 (p->field != field))
1184 continue;
1185 /* fall thru */
1186 case GREP_PATTERN: /* atom */
1187 case GREP_PATTERN_BODY:
1188 hit |= match_next_pattern(p, bol, eol, ctx,
1189 pmatch, eflags);
1190 break;
1191 default:
1192 break;
1193 }
1194 }
1195 }
1196 return hit;
1197 }
1198
1199 static void show_line_header(struct grep_opt *opt, const char *name,
1200 unsigned lno, ssize_t cno, char sign)
1201 {
1202 if (opt->heading && opt->last_shown == 0) {
1203 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1204 opt->output(opt, "\n", 1);
1205 }
1206 opt->last_shown = lno;
1207
1208 if (!opt->heading && opt->pathname) {
1209 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1210 output_sep(opt, sign);
1211 }
1212 if (opt->linenum) {
1213 char buf[32];
1214 xsnprintf(buf, sizeof(buf), "%d", lno);
1215 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1216 output_sep(opt, sign);
1217 }
1218 /*
1219 * Treat 'cno' as the 1-indexed offset from the start of a non-context
1220 * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1221 * being called with a context line.
1222 */
1223 if (opt->columnnum && cno) {
1224 char buf[32];
1225 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1226 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1227 output_sep(opt, sign);
1228 }
1229 }
1230
1231 static void show_line(struct grep_opt *opt,
1232 const char *bol, const char *eol,
1233 const char *name, unsigned lno, ssize_t cno, char sign)
1234 {
1235 int rest = eol - bol;
1236 const char *match_color = NULL;
1237 const char *line_color = NULL;
1238
1239 if (opt->file_break && opt->last_shown == 0) {
1240 if (opt->show_hunk_mark)
1241 opt->output(opt, "\n", 1);
1242 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1243 if (opt->last_shown == 0) {
1244 if (opt->show_hunk_mark) {
1245 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1246 opt->output(opt, "\n", 1);
1247 }
1248 } else if (lno > opt->last_shown + 1) {
1249 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1250 opt->output(opt, "\n", 1);
1251 }
1252 }
1253 if (!opt->only_matching) {
1254 /*
1255 * In case the line we're being called with contains more than
1256 * one match, leave printing each header to the loop below.
1257 */
1258 show_line_header(opt, name, lno, cno, sign);
1259 }
1260 if (opt->color || opt->only_matching) {
1261 regmatch_t match;
1262 enum grep_context ctx = GREP_CONTEXT_BODY;
1263 int eflags = 0;
1264
1265 if (opt->color) {
1266 if (sign == ':')
1267 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1268 else
1269 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1270 if (sign == ':')
1271 line_color = opt->colors[GREP_COLOR_SELECTED];
1272 else if (sign == '-')
1273 line_color = opt->colors[GREP_COLOR_CONTEXT];
1274 else if (sign == '=')
1275 line_color = opt->colors[GREP_COLOR_FUNCTION];
1276 }
1277 while (grep_next_match(opt, bol, eol, ctx, &match,
1278 GREP_HEADER_FIELD_MAX, eflags)) {
1279 if (match.rm_so == match.rm_eo)
1280 break;
1281
1282 if (opt->only_matching)
1283 show_line_header(opt, name, lno, cno, sign);
1284 else
1285 output_color(opt, bol, match.rm_so, line_color);
1286 output_color(opt, bol + match.rm_so,
1287 match.rm_eo - match.rm_so, match_color);
1288 if (opt->only_matching)
1289 opt->output(opt, "\n", 1);
1290 bol += match.rm_eo;
1291 cno += match.rm_eo;
1292 rest -= match.rm_eo;
1293 eflags = REG_NOTBOL;
1294 }
1295 }
1296 if (!opt->only_matching) {
1297 output_color(opt, bol, rest, line_color);
1298 opt->output(opt, "\n", 1);
1299 }
1300 }
1301
1302 int grep_use_locks;
1303
1304 /*
1305 * This lock protects access to the gitattributes machinery, which is
1306 * not thread-safe.
1307 */
1308 pthread_mutex_t grep_attr_mutex;
1309
1310 static inline void grep_attr_lock(void)
1311 {
1312 if (grep_use_locks)
1313 pthread_mutex_lock(&grep_attr_mutex);
1314 }
1315
1316 static inline void grep_attr_unlock(void)
1317 {
1318 if (grep_use_locks)
1319 pthread_mutex_unlock(&grep_attr_mutex);
1320 }
1321
1322 static int match_funcname(struct grep_opt *opt, struct grep_source *gs,
1323 const char *bol, const char *eol)
1324 {
1325 xdemitconf_t *xecfg = opt->priv;
1326 if (xecfg && !xecfg->find_func) {
1327 grep_source_load_driver(gs, opt->repo->index);
1328 if (gs->driver->funcname.pattern) {
1329 const struct userdiff_funcname *pe = &gs->driver->funcname;
1330 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1331 } else {
1332 xecfg = opt->priv = NULL;
1333 }
1334 }
1335
1336 if (xecfg) {
1337 char buf[1];
1338 return xecfg->find_func(bol, eol - bol, buf, 1,
1339 xecfg->find_func_priv) >= 0;
1340 }
1341
1342 if (bol == eol)
1343 return 0;
1344 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1345 return 1;
1346 return 0;
1347 }
1348
1349 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1350 const char *bol, unsigned lno)
1351 {
1352 while (bol > gs->buf) {
1353 const char *eol = --bol;
1354
1355 while (bol > gs->buf && bol[-1] != '\n')
1356 bol--;
1357 lno--;
1358
1359 if (lno <= opt->last_shown)
1360 break;
1361
1362 if (match_funcname(opt, gs, bol, eol)) {
1363 show_line(opt, bol, eol, gs->name, lno, 0, '=');
1364 break;
1365 }
1366 }
1367 }
1368
1369 static int is_empty_line(const char *bol, const char *eol);
1370
1371 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1372 const char *bol, const char *end, unsigned lno)
1373 {
1374 unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1375 int funcname_needed = !!opt->funcname, comment_needed = 0;
1376
1377 if (opt->pre_context < lno)
1378 from = lno - opt->pre_context;
1379 if (from <= opt->last_shown)
1380 from = opt->last_shown + 1;
1381 orig_from = from;
1382 if (opt->funcbody) {
1383 if (match_funcname(opt, gs, bol, end))
1384 comment_needed = 1;
1385 else
1386 funcname_needed = 1;
1387 from = opt->last_shown + 1;
1388 }
1389
1390 /* Rewind. */
1391 while (bol > gs->buf && cur > from) {
1392 const char *next_bol = bol;
1393 const char *eol = --bol;
1394
1395 while (bol > gs->buf && bol[-1] != '\n')
1396 bol--;
1397 cur--;
1398 if (comment_needed && (is_empty_line(bol, eol) ||
1399 match_funcname(opt, gs, bol, eol))) {
1400 comment_needed = 0;
1401 from = orig_from;
1402 if (cur < from) {
1403 cur++;
1404 bol = next_bol;
1405 break;
1406 }
1407 }
1408 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1409 funcname_lno = cur;
1410 funcname_needed = 0;
1411 if (opt->funcbody)
1412 comment_needed = 1;
1413 else
1414 from = orig_from;
1415 }
1416 }
1417
1418 /* We need to look even further back to find a function signature. */
1419 if (opt->funcname && funcname_needed)
1420 show_funcname_line(opt, gs, bol, cur);
1421
1422 /* Back forward. */
1423 while (cur < lno) {
1424 const char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1425
1426 while (*eol != '\n')
1427 eol++;
1428 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1429 bol = eol + 1;
1430 cur++;
1431 }
1432 }
1433
1434 static int should_lookahead(struct grep_opt *opt)
1435 {
1436 struct grep_pat *p;
1437
1438 if (opt->pattern_expression)
1439 return 0; /* punt for too complex stuff */
1440 if (opt->invert)
1441 return 0;
1442 for (p = opt->pattern_list; p; p = p->next) {
1443 if (p->token != GREP_PATTERN)
1444 return 0; /* punt for "header only" and stuff */
1445 }
1446 return 1;
1447 }
1448
1449 static int look_ahead(struct grep_opt *opt,
1450 unsigned long *left_p,
1451 unsigned *lno_p,
1452 const char **bol_p)
1453 {
1454 unsigned lno = *lno_p;
1455 const char *bol = *bol_p;
1456 struct grep_pat *p;
1457 const char *sp, *last_bol;
1458 regoff_t earliest = -1;
1459
1460 for (p = opt->pattern_list; p; p = p->next) {
1461 int hit;
1462 regmatch_t m;
1463
1464 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1465 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1466 continue;
1467 if (earliest < 0 || m.rm_so < earliest)
1468 earliest = m.rm_so;
1469 }
1470
1471 if (earliest < 0) {
1472 *bol_p = bol + *left_p;
1473 *left_p = 0;
1474 return 1;
1475 }
1476 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1477 ; /* find the beginning of the line */
1478 last_bol = sp;
1479
1480 for (sp = bol; sp < last_bol; sp++) {
1481 if (*sp == '\n')
1482 lno++;
1483 }
1484 *left_p -= last_bol - bol;
1485 *bol_p = last_bol;
1486 *lno_p = lno;
1487 return 0;
1488 }
1489
1490 static int fill_textconv_grep(struct repository *r,
1491 struct userdiff_driver *driver,
1492 struct grep_source *gs)
1493 {
1494 struct diff_filespec *df;
1495 char *buf;
1496 size_t size;
1497
1498 if (!driver || !driver->textconv)
1499 return grep_source_load(gs);
1500
1501 /*
1502 * The textconv interface is intimately tied to diff_filespecs, so we
1503 * have to pretend to be one. If we could unify the grep_source
1504 * and diff_filespec structs, this mess could just go away.
1505 */
1506 df = alloc_filespec(gs->path);
1507 switch (gs->type) {
1508 case GREP_SOURCE_OID:
1509 fill_filespec(df, gs->identifier, 1, 0100644);
1510 break;
1511 case GREP_SOURCE_FILE:
1512 fill_filespec(df, null_oid(), 0, 0100644);
1513 break;
1514 default:
1515 BUG("attempt to textconv something without a path?");
1516 }
1517
1518 /*
1519 * fill_textconv is not remotely thread-safe; it modifies the global
1520 * diff tempfile structure, writes to the_repo's odb and might
1521 * internally call thread-unsafe functions such as the
1522 * prepare_packed_git() lazy-initializator. Because of the last two, we
1523 * must ensure mutual exclusion between this call and the object reading
1524 * API, thus we use obj_read_lock() here.
1525 *
1526 * TODO: allowing text conversion to run in parallel with object
1527 * reading operations might increase performance in the multithreaded
1528 * non-worktreee git-grep with --textconv.
1529 */
1530 obj_read_lock();
1531 size = fill_textconv(r, driver, df, &buf);
1532 obj_read_unlock();
1533 free_filespec(df);
1534
1535 /*
1536 * The normal fill_textconv usage by the diff machinery would just keep
1537 * the textconv'd buf separate from the diff_filespec. But much of the
1538 * grep code passes around a grep_source and assumes that its "buf"
1539 * pointer is the beginning of the thing we are searching. So let's
1540 * install our textconv'd version into the grep_source, taking care not
1541 * to leak any existing buffer.
1542 */
1543 grep_source_clear_data(gs);
1544 gs->buf = buf;
1545 gs->size = size;
1546
1547 return 0;
1548 }
1549
1550 static int is_empty_line(const char *bol, const char *eol)
1551 {
1552 while (bol < eol && isspace(*bol))
1553 bol++;
1554 return bol == eol;
1555 }
1556
1557 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1558 {
1559 const char *bol;
1560 const char *peek_bol = NULL;
1561 unsigned long left;
1562 unsigned lno = 1;
1563 unsigned last_hit = 0;
1564 int binary_match_only = 0;
1565 unsigned count = 0;
1566 int try_lookahead = 0;
1567 int show_function = 0;
1568 struct userdiff_driver *textconv = NULL;
1569 enum grep_context ctx = GREP_CONTEXT_HEAD;
1570 xdemitconf_t xecfg;
1571
1572 if (!opt->status_only && gs->name == NULL)
1573 BUG("grep call which could print a name requires "
1574 "grep_source.name be non-NULL");
1575
1576 if (!opt->output)
1577 opt->output = std_output;
1578
1579 if (opt->pre_context || opt->post_context || opt->file_break ||
1580 opt->funcbody) {
1581 /* Show hunk marks, except for the first file. */
1582 if (opt->last_shown)
1583 opt->show_hunk_mark = 1;
1584 /*
1585 * If we're using threads then we can't easily identify
1586 * the first file. Always put hunk marks in that case
1587 * and skip the very first one later in work_done().
1588 */
1589 if (opt->output != std_output)
1590 opt->show_hunk_mark = 1;
1591 }
1592 opt->last_shown = 0;
1593
1594 if (opt->allow_textconv) {
1595 grep_source_load_driver(gs, opt->repo->index);
1596 /*
1597 * We might set up the shared textconv cache data here, which
1598 * is not thread-safe. Also, get_oid_with_context() and
1599 * parse_object() might be internally called. As they are not
1600 * currently thread-safe and might be racy with object reading,
1601 * obj_read_lock() must be called.
1602 */
1603 grep_attr_lock();
1604 obj_read_lock();
1605 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1606 obj_read_unlock();
1607 grep_attr_unlock();
1608 }
1609
1610 /*
1611 * We know the result of a textconv is text, so we only have to care
1612 * about binary handling if we are not using it.
1613 */
1614 if (!textconv) {
1615 switch (opt->binary) {
1616 case GREP_BINARY_DEFAULT:
1617 if (grep_source_is_binary(gs, opt->repo->index))
1618 binary_match_only = 1;
1619 break;
1620 case GREP_BINARY_NOMATCH:
1621 if (grep_source_is_binary(gs, opt->repo->index))
1622 return 0; /* Assume unmatch */
1623 break;
1624 case GREP_BINARY_TEXT:
1625 break;
1626 default:
1627 BUG("unknown binary handling mode");
1628 }
1629 }
1630
1631 memset(&xecfg, 0, sizeof(xecfg));
1632 opt->priv = &xecfg;
1633
1634 try_lookahead = should_lookahead(opt);
1635
1636 if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1637 return 0;
1638
1639 bol = gs->buf;
1640 left = gs->size;
1641 while (left) {
1642 const char *eol;
1643 int hit;
1644 ssize_t cno;
1645 ssize_t col = -1, icol = -1;
1646
1647 /*
1648 * look_ahead() skips quickly to the line that possibly
1649 * has the next hit; don't call it if we need to do
1650 * something more than just skipping the current line
1651 * in response to an unmatch for the current line. E.g.
1652 * inside a post-context window, we will show the current
1653 * line as a context around the previous hit when it
1654 * doesn't hit.
1655 */
1656 if (try_lookahead
1657 && !(last_hit
1658 && (show_function ||
1659 lno <= last_hit + opt->post_context))
1660 && look_ahead(opt, &left, &lno, &bol))
1661 break;
1662 eol = end_of_line(bol, &left);
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
1669 if (collect_hits)
1670 goto next_line;
1671
1672 /* "grep -v -e foo -e bla" should list lines
1673 * that do not have either, so inversion should
1674 * be done outside.
1675 */
1676 if (opt->invert)
1677 hit = !hit;
1678 if (opt->unmatch_name_only) {
1679 if (hit)
1680 return 0;
1681 goto next_line;
1682 }
1683 if (hit && (opt->max_count < 0 || count < opt->max_count)) {
1684 count++;
1685 if (opt->status_only)
1686 return 1;
1687 if (opt->name_only) {
1688 show_name(opt, gs->name);
1689 return 1;
1690 }
1691 if (opt->count)
1692 goto next_line;
1693 if (binary_match_only) {
1694 opt->output(opt, "Binary file ", 12);
1695 output_color(opt, gs->name, strlen(gs->name),
1696 opt->colors[GREP_COLOR_FILENAME]);
1697 opt->output(opt, " matches\n", 9);
1698 return 1;
1699 }
1700 /* Hit at this line. If we haven't shown the
1701 * pre-context lines, we would need to show them.
1702 */
1703 if (opt->pre_context || opt->funcbody)
1704 show_pre_context(opt, gs, bol, eol, lno);
1705 else if (opt->funcname)
1706 show_funcname_line(opt, gs, bol, lno);
1707 cno = opt->invert ? icol : col;
1708 if (cno < 0) {
1709 /*
1710 * A negative cno indicates that there was no
1711 * match on the line. We are thus inverted and
1712 * being asked to show all lines that _don't_
1713 * match a given expression. Therefore, set cno
1714 * to 0 to suggest the whole line matches.
1715 */
1716 cno = 0;
1717 }
1718 show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1719 last_hit = lno;
1720 if (opt->funcbody)
1721 show_function = 1;
1722 goto next_line;
1723 }
1724 if (show_function && (!peek_bol || peek_bol < bol)) {
1725 unsigned long peek_left = left;
1726 const char *peek_eol = eol;
1727
1728 /*
1729 * Trailing empty lines are not interesting.
1730 * Peek past them to see if they belong to the
1731 * body of the current function.
1732 */
1733 peek_bol = bol;
1734 while (is_empty_line(peek_bol, peek_eol)) {
1735 peek_bol = peek_eol + 1;
1736 peek_eol = end_of_line(peek_bol, &peek_left);
1737 }
1738
1739 if (peek_bol >= gs->buf + gs->size ||
1740 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 && !opt->no_body_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 opt->body_hit = 0;
1832 grep_source_1(opt, gs, 1);
1833
1834 if (opt->all_match && !chk_hit_marker(opt->pattern_expression))
1835 return 0;
1836 if (opt->no_body_match && opt->body_hit)
1837 return 0;
1838
1839 return grep_source_1(opt, gs, 0);
1840 }
1841
1842 static void grep_source_init_buf(struct grep_source *gs,
1843 const char *buf,
1844 unsigned long size)
1845 {
1846 gs->type = GREP_SOURCE_BUF;
1847 gs->name = NULL;
1848 gs->path = NULL;
1849 gs->buf = buf;
1850 gs->size = size;
1851 gs->driver = NULL;
1852 gs->identifier = NULL;
1853 }
1854
1855 int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size)
1856 {
1857 struct grep_source gs;
1858 int r;
1859
1860 grep_source_init_buf(&gs, buf, size);
1861
1862 r = grep_source(opt, &gs);
1863
1864 grep_source_clear(&gs);
1865 return r;
1866 }
1867
1868 void grep_source_init_file(struct grep_source *gs, const char *name,
1869 const char *path)
1870 {
1871 gs->type = GREP_SOURCE_FILE;
1872 gs->name = xstrdup_or_null(name);
1873 gs->path = xstrdup_or_null(path);
1874 gs->buf = NULL;
1875 gs->size = 0;
1876 gs->driver = NULL;
1877 gs->identifier = xstrdup(path);
1878 }
1879
1880 void grep_source_init_oid(struct grep_source *gs, const char *name,
1881 const char *path, const struct object_id *oid,
1882 struct repository *repo)
1883 {
1884 gs->type = GREP_SOURCE_OID;
1885 gs->name = xstrdup_or_null(name);
1886 gs->path = xstrdup_or_null(path);
1887 gs->buf = NULL;
1888 gs->size = 0;
1889 gs->driver = NULL;
1890 gs->identifier = oiddup(oid);
1891 gs->repo = repo;
1892 }
1893
1894 void grep_source_clear(struct grep_source *gs)
1895 {
1896 FREE_AND_NULL(gs->name);
1897 FREE_AND_NULL(gs->path);
1898 FREE_AND_NULL(gs->identifier);
1899 grep_source_clear_data(gs);
1900 }
1901
1902 void grep_source_clear_data(struct grep_source *gs)
1903 {
1904 switch (gs->type) {
1905 case GREP_SOURCE_FILE:
1906 case GREP_SOURCE_OID:
1907 /* these types own the buffer */
1908 free((char *)gs->buf);
1909 gs->buf = NULL;
1910 gs->size = 0;
1911 break;
1912 case GREP_SOURCE_BUF:
1913 /* leave user-provided buf intact */
1914 break;
1915 }
1916 }
1917
1918 static int grep_source_load_oid(struct grep_source *gs)
1919 {
1920 enum object_type type;
1921
1922 gs->buf = repo_read_object_file(gs->repo, gs->identifier, &type,
1923 &gs->size);
1924 if (!gs->buf)
1925 return error(_("'%s': unable to read %s"),
1926 gs->name,
1927 oid_to_hex(gs->identifier));
1928 return 0;
1929 }
1930
1931 static int grep_source_load_file(struct grep_source *gs)
1932 {
1933 const char *filename = gs->identifier;
1934 struct stat st;
1935 char *data;
1936 size_t size;
1937 int i;
1938
1939 if (lstat(filename, &st) < 0) {
1940 err_ret:
1941 if (errno != ENOENT)
1942 error_errno(_("failed to stat '%s'"), filename);
1943 return -1;
1944 }
1945 if (!S_ISREG(st.st_mode))
1946 return -1;
1947 size = xsize_t(st.st_size);
1948 i = open(filename, O_RDONLY);
1949 if (i < 0)
1950 goto err_ret;
1951 data = xmallocz(size);
1952 if (st.st_size != read_in_full(i, data, size)) {
1953 error_errno(_("'%s': short read"), filename);
1954 close(i);
1955 free(data);
1956 return -1;
1957 }
1958 close(i);
1959
1960 gs->buf = data;
1961 gs->size = size;
1962 return 0;
1963 }
1964
1965 static int grep_source_load(struct grep_source *gs)
1966 {
1967 if (gs->buf)
1968 return 0;
1969
1970 switch (gs->type) {
1971 case GREP_SOURCE_FILE:
1972 return grep_source_load_file(gs);
1973 case GREP_SOURCE_OID:
1974 return grep_source_load_oid(gs);
1975 case GREP_SOURCE_BUF:
1976 return gs->buf ? 0 : -1;
1977 }
1978 BUG("invalid grep_source type to load");
1979 }
1980
1981 void grep_source_load_driver(struct grep_source *gs,
1982 struct index_state *istate)
1983 {
1984 if (gs->driver)
1985 return;
1986
1987 grep_attr_lock();
1988 if (gs->path)
1989 gs->driver = userdiff_find_by_path(istate, gs->path);
1990 if (!gs->driver)
1991 gs->driver = userdiff_find_by_name("default");
1992 grep_attr_unlock();
1993 }
1994
1995 static int grep_source_is_binary(struct grep_source *gs,
1996 struct index_state *istate)
1997 {
1998 grep_source_load_driver(gs, istate);
1999 if (gs->driver->binary != -1)
2000 return gs->driver->binary;
2001
2002 if (!grep_source_load(gs))
2003 return buffer_is_binary(gs->buf, gs->size);
2004
2005 return 0;
2006 }
Mirror of Junio's Git repository with all the branches