freshlybakedca.ke
Git fork
1#define DISABLE_SIGN_COMPARE_WARNINGS
2
3#include "git-compat-util.h"
4#include "config.h"
5#include "gettext.h"
6#include "grep.h"
7#include "hex.h"
8#include "odb.h"
9#include "pretty.h"
10#include "userdiff.h"
11#include "xdiff-interface.h"
12#include "diff.h"
13#include "diffcore.h"
14#include "quote.h"
15#include "help.h"
16
17static int grep_source_load(struct grep_source *gs);
18static int grep_source_is_binary(struct grep_source *gs,
19 struct index_state *istate);
20
21static void std_output(struct grep_opt *opt UNUSED, const void *buf, size_t size)
22{
23 fwrite(buf, size, 1, stdout);
24}
25
26static const char *color_grep_slots[] = {
27 [GREP_COLOR_CONTEXT] = "context",
28 [GREP_COLOR_FILENAME] = "filename",
29 [GREP_COLOR_FUNCTION] = "function",
30 [GREP_COLOR_LINENO] = "lineNumber",
31 [GREP_COLOR_COLUMNNO] = "column",
32 [GREP_COLOR_MATCH_CONTEXT] = "matchContext",
33 [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
34 [GREP_COLOR_SELECTED] = "selected",
35 [GREP_COLOR_SEP] = "separator",
36};
37
38static int parse_pattern_type_arg(const char *opt, const char *arg)
39{
40 if (!strcmp(arg, "default"))
41 return GREP_PATTERN_TYPE_UNSPECIFIED;
42 else if (!strcmp(arg, "basic"))
43 return GREP_PATTERN_TYPE_BRE;
44 else if (!strcmp(arg, "extended"))
45 return GREP_PATTERN_TYPE_ERE;
46 else if (!strcmp(arg, "fixed"))
47 return GREP_PATTERN_TYPE_FIXED;
48 else if (!strcmp(arg, "perl"))
49 return GREP_PATTERN_TYPE_PCRE;
50 die("bad %s argument: %s", opt, arg);
51}
52
53define_list_config_array_extra(color_grep_slots, {"match"});
54
55/*
56 * Read the configuration file once and store it in
57 * the grep_defaults template.
58 */
59int grep_config(const char *var, const char *value,
60 const struct config_context *ctx, void *cb)
61{
62 struct grep_opt *opt = cb;
63 const char *slot;
64
65 if (userdiff_config(var, value) < 0)
66 return -1;
67
68 if (!strcmp(var, "grep.extendedregexp")) {
69 opt->extended_regexp_option = git_config_bool(var, value);
70 return 0;
71 }
72
73 if (!strcmp(var, "grep.patterntype")) {
74 opt->pattern_type_option = parse_pattern_type_arg(var, value);
75 return 0;
76 }
77
78 if (!strcmp(var, "grep.linenumber")) {
79 opt->linenum = git_config_bool(var, value);
80 return 0;
81 }
82 if (!strcmp(var, "grep.column")) {
83 opt->columnnum = git_config_bool(var, value);
84 return 0;
85 }
86
87 if (!strcmp(var, "grep.fullname")) {
88 opt->relative = !git_config_bool(var, value);
89 return 0;
90 }
91
92 if (!strcmp(var, "color.grep"))
93 opt->color = git_config_colorbool(var, value);
94 if (!strcmp(var, "color.grep.match")) {
95 if (grep_config("color.grep.matchcontext", value, ctx, cb) < 0)
96 return -1;
97 if (grep_config("color.grep.matchselected", value, ctx, cb) < 0)
98 return -1;
99 } else if (skip_prefix(var, "color.grep.", &slot)) {
100 int i = LOOKUP_CONFIG(color_grep_slots, slot);
101 char *color;
102
103 if (i < 0)
104 return -1;
105 color = opt->colors[i];
106 if (!value)
107 return config_error_nonbool(var);
108 return color_parse(value, color);
109 }
110 return 0;
111}
112
113void grep_init(struct grep_opt *opt, struct repository *repo)
114{
115 struct grep_opt blank = GREP_OPT_INIT;
116 memcpy(opt, &blank, sizeof(*opt));
117
118 opt->repo = repo;
119 opt->pattern_tail = &opt->pattern_list;
120 opt->header_tail = &opt->header_list;
121}
122
123static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
124 const char *origin, int no,
125 enum grep_pat_token t,
126 enum grep_header_field field)
127{
128 struct grep_pat *p = xcalloc(1, sizeof(*p));
129 p->pattern = xmemdupz(pat, patlen);
130 p->patternlen = patlen;
131 p->origin = origin;
132 p->no = no;
133 p->token = t;
134 p->field = field;
135 return p;
136}
137
138static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
139{
140 **tail = p;
141 *tail = &p->next;
142 p->next = NULL;
143
144 switch (p->token) {
145 case GREP_PATTERN: /* atom */
146 case GREP_PATTERN_HEAD:
147 case GREP_PATTERN_BODY:
148 for (;;) {
149 struct grep_pat *new_pat;
150 size_t len = 0;
151 char *cp = p->pattern + p->patternlen, *nl = NULL;
152 while (++len <= p->patternlen) {
153 if (*(--cp) == '\n') {
154 nl = cp;
155 break;
156 }
157 }
158 if (!nl)
159 break;
160 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
161 p->no, p->token, p->field);
162 new_pat->next = p->next;
163 if (!p->next)
164 *tail = &new_pat->next;
165 p->next = new_pat;
166 *nl = '\0';
167 p->patternlen -= len;
168 }
169 break;
170 default:
171 break;
172 }
173}
174
175void append_header_grep_pattern(struct grep_opt *opt,
176 enum grep_header_field field, const char *pat)
177{
178 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
179 GREP_PATTERN_HEAD, field);
180 if (field == GREP_HEADER_REFLOG)
181 opt->use_reflog_filter = 1;
182 do_append_grep_pat(&opt->header_tail, p);
183}
184
185void append_grep_pattern(struct grep_opt *opt, const char *pat,
186 const char *origin, int no, enum grep_pat_token t)
187{
188 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
189}
190
191void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
192 const char *origin, int no, enum grep_pat_token t)
193{
194 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
195 do_append_grep_pat(&opt->pattern_tail, p);
196}
197
198struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
199{
200 struct grep_pat *pat;
201 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
202 *ret = *opt;
203
204 ret->pattern_list = NULL;
205 ret->pattern_tail = &ret->pattern_list;
206
207 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
208 {
209 if(pat->token == GREP_PATTERN_HEAD)
210 append_header_grep_pattern(ret, pat->field,
211 pat->pattern);
212 else
213 append_grep_pat(ret, pat->pattern, pat->patternlen,
214 pat->origin, pat->no, pat->token);
215 }
216
217 return ret;
218}
219
220static NORETURN void compile_regexp_failed(const struct grep_pat *p,
221 const char *error)
222{
223 char where[1024];
224
225 if (p->no)
226 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
227 else if (p->origin)
228 xsnprintf(where, sizeof(where), "%s, ", p->origin);
229 else
230 where[0] = 0;
231
232 die("%s'%s': %s", where, p->pattern, error);
233}
234
235static int is_fixed(const char *s, size_t len)
236{
237 size_t i;
238
239 for (i = 0; i < len; i++) {
240 if (is_regex_special(s[i]))
241 return 0;
242 }
243
244 return 1;
245}
246
247#ifdef USE_LIBPCRE2
248#define GREP_PCRE2_DEBUG_MALLOC 0
249
250static void *pcre2_malloc(PCRE2_SIZE size, void *memory_data UNUSED)
251{
252 void *pointer = malloc(size);
253#if GREP_PCRE2_DEBUG_MALLOC
254 static int count = 1;
255 fprintf(stderr, "PCRE2:%p -> #%02d: alloc(%lu)\n", pointer, count++, size);
256#endif
257 return pointer;
258}
259
260static void pcre2_free(void *pointer, void *memory_data UNUSED)
261{
262#if GREP_PCRE2_DEBUG_MALLOC
263 static int count = 1;
264 if (pointer)
265 fprintf(stderr, "PCRE2:%p -> #%02d: free()\n", pointer, count++);
266#endif
267 free(pointer);
268}
269
270static int pcre2_jit_functional(void)
271{
272 static int jit_working = -1;
273 pcre2_code *code;
274 size_t off;
275 int err;
276
277 if (jit_working != -1)
278 return jit_working;
279
280 /*
281 * Try to JIT compile a simple pattern to probe if the JIT is
282 * working in general. It might fail for systems where creating
283 * memory mappings for runtime code generation is restricted.
284 */
285 code = pcre2_compile((PCRE2_SPTR)".", 1, 0, &err, &off, NULL);
286 if (!code)
287 return 0;
288
289 jit_working = pcre2_jit_compile(code, PCRE2_JIT_COMPLETE) == 0;
290 pcre2_code_free(code);
291
292 return jit_working;
293}
294
295static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
296{
297 int error;
298 PCRE2_UCHAR errbuf[256];
299 PCRE2_SIZE erroffset;
300 int options = PCRE2_MULTILINE;
301 int jitret;
302 int patinforet;
303 size_t jitsizearg;
304 int literal = !opt->ignore_case && (p->fixed || p->is_fixed);
305
306 /*
307 * Call pcre2_general_context_create() before calling any
308 * other pcre2_*(). It sets up our malloc()/free() functions
309 * with which everything else is allocated.
310 */
311 p->pcre2_general_context = pcre2_general_context_create(
312 pcre2_malloc, pcre2_free, NULL);
313 if (!p->pcre2_general_context)
314 die("Couldn't allocate PCRE2 general context");
315
316 if (opt->ignore_case) {
317 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
318 p->pcre2_tables = pcre2_maketables(p->pcre2_general_context);
319 p->pcre2_compile_context = pcre2_compile_context_create(p->pcre2_general_context);
320 pcre2_set_character_tables(p->pcre2_compile_context,
321 p->pcre2_tables);
322 }
323 options |= PCRE2_CASELESS;
324 }
325 if (!opt->ignore_locale && is_utf8_locale() && !literal)
326 options |= (PCRE2_UTF | PCRE2_UCP | PCRE2_MATCH_INVALID_UTF);
327
328#ifndef GIT_PCRE2_VERSION_10_35_OR_HIGHER
329 /*
330 * Work around a JIT bug related to invalid Unicode character handling
331 * fixed in 10.35:
332 * https://github.com/PCRE2Project/pcre2/commit/c21bd977547d
333 */
334 options &= ~PCRE2_UCP;
335#endif
336
337#ifndef GIT_PCRE2_VERSION_10_36_OR_HIGHER
338 /* Work around https://bugs.exim.org/show_bug.cgi?id=2642 fixed in 10.36 */
339 if (PCRE2_MATCH_INVALID_UTF && options & (PCRE2_UTF | PCRE2_CASELESS))
340 options |= PCRE2_NO_START_OPTIMIZE;
341#endif
342
343 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
344 p->patternlen, options, &error, &erroffset,
345 p->pcre2_compile_context);
346
347 if (p->pcre2_pattern) {
348 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, p->pcre2_general_context);
349 if (!p->pcre2_match_data)
350 die("Couldn't allocate PCRE2 match data");
351 } else {
352 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
353 compile_regexp_failed(p, (const char *)&errbuf);
354 }
355
356 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
357 if (p->pcre2_jit_on) {
358 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
359 if (jitret == PCRE2_ERROR_NOMEMORY && !pcre2_jit_functional()) {
360 /*
361 * Even though pcre2_config(PCRE2_CONFIG_JIT, ...)
362 * indicated JIT support, the library might still
363 * fail to generate JIT code for various reasons,
364 * e.g. when SELinux's 'deny_execmem' or PaX's
365 * MPROTECT prevent creating W|X memory mappings.
366 *
367 * Instead of faling hard, fall back to interpreter
368 * mode, just as if the pattern was prefixed with
369 * '(*NO_JIT)'.
370 */
371 p->pcre2_jit_on = 0;
372 return;
373 } else if (jitret) {
374 int need_clip = p->patternlen > 64;
375 int clip_len = need_clip ? 64 : p->patternlen;
376 die("Couldn't JIT the PCRE2 pattern '%.*s'%s, got '%d'%s",
377 clip_len, p->pattern, need_clip ? "..." : "", jitret,
378 pcre2_jit_functional()
379 ? "\nPerhaps prefix (*NO_JIT) to your pattern?"
380 : "");
381 }
382
383 /*
384 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
385 * tells us whether the library itself supports JIT,
386 * but to see whether we're going to be actually using
387 * JIT we need to extract PCRE2_INFO_JITSIZE from the
388 * pattern *after* we do pcre2_jit_compile() above.
389 *
390 * This is because if the pattern contains the
391 * (*NO_JIT) verb (see pcre2syntax(3))
392 * pcre2_jit_compile() will exit early with 0. If we
393 * then proceed to call pcre2_jit_match() further down
394 * the line instead of pcre2_match() we'll either
395 * segfault (pre PCRE 10.31) or run into a fatal error
396 * (post PCRE2 10.31)
397 */
398 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
399 if (patinforet)
400 BUG("pcre2_pattern_info() failed: %d", patinforet);
401 if (jitsizearg == 0) {
402 p->pcre2_jit_on = 0;
403 return;
404 }
405 }
406}
407
408static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
409 regmatch_t *match, int eflags)
410{
411 int ret, flags = 0;
412 PCRE2_SIZE *ovector;
413 PCRE2_UCHAR errbuf[256];
414
415 if (eflags & REG_NOTBOL)
416 flags |= PCRE2_NOTBOL;
417
418 if (p->pcre2_jit_on)
419 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
420 eol - line, 0, flags, p->pcre2_match_data,
421 NULL);
422 else
423 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
424 eol - line, 0, flags, p->pcre2_match_data,
425 NULL);
426
427 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
428 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
429 die("%s failed with error code %d: %s",
430 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
431 errbuf);
432 }
433 if (ret > 0) {
434 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
435 ret = 0;
436 match->rm_so = (int)ovector[0];
437 match->rm_eo = (int)ovector[1];
438 }
439
440 return ret;
441}
442
443static void free_pcre2_pattern(struct grep_pat *p)
444{
445 pcre2_compile_context_free(p->pcre2_compile_context);
446 pcre2_code_free(p->pcre2_pattern);
447 pcre2_match_data_free(p->pcre2_match_data);
448#ifdef GIT_PCRE2_VERSION_10_34_OR_HIGHER
449 pcre2_maketables_free(p->pcre2_general_context, p->pcre2_tables);
450#else
451 free((void *)p->pcre2_tables);
452#endif
453 pcre2_general_context_free(p->pcre2_general_context);
454}
455#else /* !USE_LIBPCRE2 */
456static void compile_pcre2_pattern(struct grep_pat *p UNUSED,
457 const struct grep_opt *opt UNUSED)
458{
459 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
460}
461
462static int pcre2match(struct grep_pat *p UNUSED, const char *line UNUSED,
463 const char *eol UNUSED, regmatch_t *match UNUSED,
464 int eflags UNUSED)
465{
466 return 1;
467}
468
469static void free_pcre2_pattern(struct grep_pat *p UNUSED)
470{
471}
472
473static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
474{
475 struct strbuf sb = STRBUF_INIT;
476 int err;
477 int regflags = 0;
478
479 basic_regex_quote_buf(&sb, p->pattern);
480 if (opt->ignore_case)
481 regflags |= REG_ICASE;
482 err = regcomp(&p->regexp, sb.buf, regflags);
483 strbuf_release(&sb);
484 if (err) {
485 char errbuf[1024];
486 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
487 compile_regexp_failed(p, errbuf);
488 }
489}
490#endif /* !USE_LIBPCRE2 */
491
492static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
493{
494 int err;
495 int regflags = REG_NEWLINE;
496
497 if (opt->pattern_type_option == GREP_PATTERN_TYPE_UNSPECIFIED)
498 opt->pattern_type_option = (opt->extended_regexp_option
499 ? GREP_PATTERN_TYPE_ERE
500 : GREP_PATTERN_TYPE_BRE);
501
502 p->word_regexp = opt->word_regexp;
503 p->ignore_case = opt->ignore_case;
504 p->fixed = opt->pattern_type_option == GREP_PATTERN_TYPE_FIXED;
505
506 if (opt->pattern_type_option != GREP_PATTERN_TYPE_PCRE &&
507 memchr(p->pattern, 0, p->patternlen))
508 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
509
510 p->is_fixed = is_fixed(p->pattern, p->patternlen);
511#ifdef USE_LIBPCRE2
512 if (!p->fixed && !p->is_fixed) {
513 const char *no_jit = "(*NO_JIT)";
514 const int no_jit_len = strlen(no_jit);
515 if (starts_with(p->pattern, no_jit) &&
516 is_fixed(p->pattern + no_jit_len,
517 p->patternlen - no_jit_len))
518 p->is_fixed = 1;
519 }
520#endif
521 if (p->fixed || p->is_fixed) {
522#ifdef USE_LIBPCRE2
523 if (p->is_fixed) {
524 compile_pcre2_pattern(p, opt);
525 } else {
526 /*
527 * E.g. t7811-grep-open.sh relies on the
528 * pattern being restored.
529 */
530 char *old_pattern = p->pattern;
531 size_t old_patternlen = p->patternlen;
532 struct strbuf sb = STRBUF_INIT;
533
534 /*
535 * There is the PCRE2_LITERAL flag, but it's
536 * only in PCRE v2 10.30 and later. Needing to
537 * ifdef our way around that and dealing with
538 * it + PCRE2_MULTILINE being an error is more
539 * complex than just quoting this ourselves.
540 */
541 strbuf_add(&sb, "\\Q", 2);
542 strbuf_add(&sb, p->pattern, p->patternlen);
543 strbuf_add(&sb, "\\E", 2);
544
545 p->pattern = sb.buf;
546 p->patternlen = sb.len;
547 compile_pcre2_pattern(p, opt);
548 p->pattern = old_pattern;
549 p->patternlen = old_patternlen;
550 strbuf_release(&sb);
551 }
552#else /* !USE_LIBPCRE2 */
553 compile_fixed_regexp(p, opt);
554#endif /* !USE_LIBPCRE2 */
555 return;
556 }
557
558 if (opt->pattern_type_option == GREP_PATTERN_TYPE_PCRE) {
559 compile_pcre2_pattern(p, opt);
560 return;
561 }
562
563 if (p->ignore_case)
564 regflags |= REG_ICASE;
565 if (opt->pattern_type_option == GREP_PATTERN_TYPE_ERE)
566 regflags |= REG_EXTENDED;
567 err = regcomp(&p->regexp, p->pattern, regflags);
568 if (err) {
569 char errbuf[1024];
570 regerror(err, &p->regexp, errbuf, 1024);
571 compile_regexp_failed(p, errbuf);
572 }
573}
574
575static struct grep_expr *grep_not_expr(struct grep_expr *expr)
576{
577 struct grep_expr *z = xcalloc(1, sizeof(*z));
578 z->node = GREP_NODE_NOT;
579 z->u.unary = expr;
580 return z;
581}
582
583static struct grep_expr *grep_binexp(enum grep_expr_node kind,
584 struct grep_expr *left,
585 struct grep_expr *right)
586{
587 struct grep_expr *z = xcalloc(1, sizeof(*z));
588 z->node = kind;
589 z->u.binary.left = left;
590 z->u.binary.right = right;
591 return z;
592}
593
594static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
595{
596 return grep_binexp(GREP_NODE_OR, left, right);
597}
598
599static struct grep_expr *grep_and_expr(struct grep_expr *left, struct grep_expr *right)
600{
601 return grep_binexp(GREP_NODE_AND, left, right);
602}
603
604static struct grep_expr *compile_pattern_or(struct grep_pat **);
605static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
606{
607 struct grep_pat *p;
608 struct grep_expr *x;
609
610 p = *list;
611 if (!p)
612 return NULL;
613 switch (p->token) {
614 case GREP_PATTERN: /* atom */
615 case GREP_PATTERN_HEAD:
616 case GREP_PATTERN_BODY:
617 CALLOC_ARRAY(x, 1);
618 x->node = GREP_NODE_ATOM;
619 x->u.atom = p;
620 *list = p->next;
621 return x;
622 case GREP_OPEN_PAREN:
623 *list = p->next;
624 x = compile_pattern_or(list);
625 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
626 die("unmatched ( for expression group");
627 *list = (*list)->next;
628 return x;
629 default:
630 return NULL;
631 }
632}
633
634static struct grep_expr *compile_pattern_not(struct grep_pat **list)
635{
636 struct grep_pat *p;
637 struct grep_expr *x;
638
639 p = *list;
640 if (!p)
641 return NULL;
642 switch (p->token) {
643 case GREP_NOT:
644 if (!p->next)
645 die("--not not followed by pattern expression");
646 *list = p->next;
647 x = compile_pattern_not(list);
648 if (!x)
649 die("--not followed by non pattern expression");
650 return grep_not_expr(x);
651 default:
652 return compile_pattern_atom(list);
653 }
654}
655
656static struct grep_expr *compile_pattern_and(struct grep_pat **list)
657{
658 struct grep_pat *p;
659 struct grep_expr *x, *y;
660
661 x = compile_pattern_not(list);
662 p = *list;
663 if (p && p->token == GREP_AND) {
664 if (!x)
665 die("--and not preceded by pattern expression");
666 if (!p->next)
667 die("--and not followed by pattern expression");
668 *list = p->next;
669 y = compile_pattern_and(list);
670 if (!y)
671 die("--and not followed by pattern expression");
672 return grep_and_expr(x, y);
673 }
674 return x;
675}
676
677static struct grep_expr *compile_pattern_or(struct grep_pat **list)
678{
679 struct grep_pat *p;
680 struct grep_expr *x, *y;
681
682 x = compile_pattern_and(list);
683 p = *list;
684 if (x && p && p->token != GREP_CLOSE_PAREN) {
685 y = compile_pattern_or(list);
686 if (!y)
687 die("not a pattern expression %s", p->pattern);
688 return grep_or_expr(x, y);
689 }
690 return x;
691}
692
693static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
694{
695 return compile_pattern_or(list);
696}
697
698static struct grep_expr *grep_true_expr(void)
699{
700 struct grep_expr *z = xcalloc(1, sizeof(*z));
701 z->node = GREP_NODE_TRUE;
702 return z;
703}
704
705static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
706{
707 struct grep_pat *p;
708 struct grep_expr *header_expr;
709 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
710 enum grep_header_field fld;
711
712 if (!opt->header_list)
713 return NULL;
714
715 for (p = opt->header_list; p; p = p->next) {
716 if (p->token != GREP_PATTERN_HEAD)
717 BUG("a non-header pattern in grep header list.");
718 if (p->field < GREP_HEADER_FIELD_MIN ||
719 GREP_HEADER_FIELD_MAX <= p->field)
720 BUG("unknown header field %d", p->field);
721 compile_regexp(p, opt);
722 }
723
724 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
725 header_group[fld] = NULL;
726
727 for (p = opt->header_list; p; p = p->next) {
728 struct grep_expr *h;
729 struct grep_pat *pp = p;
730
731 h = compile_pattern_atom(&pp);
732 if (!h || pp != p->next)
733 BUG("malformed header expr");
734 if (!header_group[p->field]) {
735 header_group[p->field] = h;
736 continue;
737 }
738 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
739 }
740
741 header_expr = NULL;
742
743 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
744 if (!header_group[fld])
745 continue;
746 if (!header_expr)
747 header_expr = grep_true_expr();
748 header_expr = grep_or_expr(header_group[fld], header_expr);
749 }
750 return header_expr;
751}
752
753static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
754{
755 struct grep_expr *z = x;
756
757 while (x) {
758 assert(x->node == GREP_NODE_OR);
759 if (x->u.binary.right &&
760 x->u.binary.right->node == GREP_NODE_TRUE) {
761 free(x->u.binary.right);
762 x->u.binary.right = y;
763 break;
764 }
765 x = x->u.binary.right;
766 }
767 return z;
768}
769
770void compile_grep_patterns(struct grep_opt *opt)
771{
772 struct grep_pat *p;
773 struct grep_expr *header_expr = prep_header_patterns(opt);
774 int extended = 0;
775
776 for (p = opt->pattern_list; p; p = p->next) {
777 switch (p->token) {
778 case GREP_PATTERN: /* atom */
779 case GREP_PATTERN_HEAD:
780 case GREP_PATTERN_BODY:
781 compile_regexp(p, opt);
782 break;
783 default:
784 extended = 1;
785 break;
786 }
787 }
788
789 if (opt->all_match || opt->no_body_match || header_expr)
790 extended = 1;
791 else if (!extended)
792 return;
793
794 p = opt->pattern_list;
795 if (p)
796 opt->pattern_expression = compile_pattern_expr(&p);
797 if (p)
798 die("incomplete pattern expression group: %s", p->pattern);
799
800 if (opt->no_body_match && opt->pattern_expression)
801 opt->pattern_expression = grep_not_expr(opt->pattern_expression);
802
803 if (!header_expr)
804 return;
805
806 if (!opt->pattern_expression)
807 opt->pattern_expression = header_expr;
808 else if (opt->all_match)
809 opt->pattern_expression = grep_splice_or(header_expr,
810 opt->pattern_expression);
811 else
812 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
813 header_expr);
814 opt->all_match = 1;
815}
816
817static void free_pattern_expr(struct grep_expr *x)
818{
819 switch (x->node) {
820 case GREP_NODE_TRUE:
821 case GREP_NODE_ATOM:
822 break;
823 case GREP_NODE_NOT:
824 free_pattern_expr(x->u.unary);
825 break;
826 case GREP_NODE_AND:
827 case GREP_NODE_OR:
828 free_pattern_expr(x->u.binary.left);
829 free_pattern_expr(x->u.binary.right);
830 break;
831 }
832 free(x);
833}
834
835static void free_grep_pat(struct grep_pat *pattern)
836{
837 struct grep_pat *p, *n;
838
839 for (p = pattern; p; p = n) {
840 n = p->next;
841 switch (p->token) {
842 case GREP_PATTERN: /* atom */
843 case GREP_PATTERN_HEAD:
844 case GREP_PATTERN_BODY:
845 if (p->pcre2_pattern)
846 free_pcre2_pattern(p);
847 else
848 regfree(&p->regexp);
849 break;
850 default:
851 break;
852 }
853 free(p->pattern);
854 free(p);
855 }
856}
857
858void free_grep_patterns(struct grep_opt *opt)
859{
860 free_grep_pat(opt->pattern_list);
861 free_grep_pat(opt->header_list);
862
863 if (opt->pattern_expression)
864 free_pattern_expr(opt->pattern_expression);
865}
866
867static const char *end_of_line(const char *cp, unsigned long *left)
868{
869 unsigned long l = *left;
870 while (l && *cp != '\n') {
871 l--;
872 cp++;
873 }
874 *left = l;
875 return cp;
876}
877
878static int word_char(char ch)
879{
880 return isalnum(ch) || ch == '_';
881}
882
883static void output_color(struct grep_opt *opt, const void *data, size_t size,
884 const char *color)
885{
886 if (want_color(opt->color) && color && color[0]) {
887 opt->output(opt, color, strlen(color));
888 opt->output(opt, data, size);
889 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
890 } else
891 opt->output(opt, data, size);
892}
893
894static void output_sep(struct grep_opt *opt, char sign)
895{
896 if (opt->null_following_name)
897 opt->output(opt, "\0", 1);
898 else
899 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
900}
901
902static void show_name(struct grep_opt *opt, const char *name)
903{
904 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
905 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
906}
907
908static int patmatch(struct grep_pat *p,
909 const char *line, const char *eol,
910 regmatch_t *match, int eflags)
911{
912 if (p->pcre2_pattern)
913 return !pcre2match(p, line, eol, match, eflags);
914
915 switch (regexec_buf(&p->regexp, line, eol - line, 1, match, eflags)) {
916 case 0:
917 return 1;
918 case REG_NOMATCH:
919 return 0;
920 default:
921 return -1;
922 }
923}
924
925static void strip_timestamp(const char *bol, const char **eol_p)
926{
927 const char *eol = *eol_p;
928
929 while (bol < --eol) {
930 if (*eol != '>')
931 continue;
932 *eol_p = ++eol;
933 break;
934 }
935}
936
937static struct {
938 const char *field;
939 size_t len;
940} header_field[] = {
941 { "author ", 7 },
942 { "committer ", 10 },
943 { "reflog ", 7 },
944};
945
946static int headerless_match_one_pattern(struct grep_pat *p,
947 const char *bol, const char *eol,
948 enum grep_context ctx,
949 regmatch_t *pmatch, int eflags)
950{
951 int hit = 0;
952 const char *start = bol;
953
954 if ((p->token != GREP_PATTERN) &&
955 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
956 return 0;
957
958 again:
959 hit = patmatch(p, bol, eol, pmatch, eflags);
960 if (hit < 0)
961 hit = 0;
962
963 if (hit && p->word_regexp) {
964 if ((pmatch[0].rm_so < 0) ||
965 (eol - bol) < pmatch[0].rm_so ||
966 (pmatch[0].rm_eo < 0) ||
967 (eol - bol) < pmatch[0].rm_eo)
968 die("regexp returned nonsense");
969
970 /* Match beginning must be either beginning of the
971 * line, or at word boundary (i.e. the last char must
972 * not be a word char). Similarly, match end must be
973 * either end of the line, or at word boundary
974 * (i.e. the next char must not be a word char).
975 */
976 if ( ((pmatch[0].rm_so == 0) ||
977 !word_char(bol[pmatch[0].rm_so-1])) &&
978 ((pmatch[0].rm_eo == (eol-bol)) ||
979 !word_char(bol[pmatch[0].rm_eo])) )
980 ;
981 else
982 hit = 0;
983
984 /* Words consist of at least one character. */
985 if (pmatch->rm_so == pmatch->rm_eo)
986 hit = 0;
987
988 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
989 /* There could be more than one match on the
990 * line, and the first match might not be
991 * strict word match. But later ones could be!
992 * Forward to the next possible start, i.e. the
993 * next position following a non-word char.
994 */
995 bol = pmatch[0].rm_so + bol + 1;
996 while (word_char(bol[-1]) && bol < eol)
997 bol++;
998 eflags |= REG_NOTBOL;
999 if (bol < eol)
1000 goto again;
1001 }
1002 }
1003 if (hit) {
1004 pmatch[0].rm_so += bol - start;
1005 pmatch[0].rm_eo += bol - start;
1006 }
1007 return hit;
1008}
1009
1010static int match_one_pattern(struct grep_pat *p,
1011 const char *bol, const char *eol,
1012 enum grep_context ctx, regmatch_t *pmatch,
1013 int eflags)
1014{
1015 const char *field;
1016 size_t len;
1017
1018 if (p->token == GREP_PATTERN_HEAD) {
1019 assert(p->field < ARRAY_SIZE(header_field));
1020 field = header_field[p->field].field;
1021 len = header_field[p->field].len;
1022 if (strncmp(bol, field, len))
1023 return 0;
1024 bol += len;
1025
1026 switch (p->field) {
1027 case GREP_HEADER_AUTHOR:
1028 case GREP_HEADER_COMMITTER:
1029 strip_timestamp(bol, &eol);
1030 break;
1031 default:
1032 break;
1033 }
1034 }
1035
1036 return headerless_match_one_pattern(p, bol, eol, ctx, pmatch, eflags);
1037}
1038
1039
1040static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x,
1041 const char *bol, const char *eol,
1042 enum grep_context ctx, ssize_t *col,
1043 ssize_t *icol, int collect_hits)
1044{
1045 int h = 0;
1046
1047 switch (x->node) {
1048 case GREP_NODE_TRUE:
1049 h = 1;
1050 break;
1051 case GREP_NODE_ATOM:
1052 {
1053 regmatch_t tmp;
1054 h = match_one_pattern(x->u.atom, bol, eol, ctx,
1055 &tmp, 0);
1056 if (h && (*col < 0 || tmp.rm_so < *col))
1057 *col = tmp.rm_so;
1058 }
1059 if (x->u.atom->token == GREP_PATTERN_BODY)
1060 opt->body_hit |= h;
1061 break;
1062 case GREP_NODE_NOT:
1063 /*
1064 * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1065 */
1066 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1067 0);
1068 break;
1069 case GREP_NODE_AND:
1070 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1071 icol, 0);
1072 if (h || opt->columnnum) {
1073 /*
1074 * Don't short-circuit AND when given --column, since a
1075 * NOT earlier in the tree may turn this into an OR. In
1076 * this case, see the below comment.
1077 */
1078 h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1079 ctx, col, icol, 0);
1080 }
1081 break;
1082 case GREP_NODE_OR:
1083 if (!(collect_hits || opt->columnnum)) {
1084 /*
1085 * Don't short-circuit OR when given --column (or
1086 * collecting hits) to ensure we don't skip a later
1087 * child that would produce an earlier match.
1088 */
1089 return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1090 ctx, col, icol, 0) ||
1091 match_expr_eval(opt, x->u.binary.right, bol,
1092 eol, ctx, col, icol, 0));
1093 }
1094 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1095 icol, 0);
1096 if (collect_hits)
1097 x->u.binary.left->hit |= h;
1098 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1099 icol, collect_hits);
1100 break;
1101 default:
1102 die("Unexpected node type (internal error) %d", x->node);
1103 }
1104 if (collect_hits)
1105 x->hit |= h;
1106 return h;
1107}
1108
1109static int match_expr(struct grep_opt *opt,
1110 const char *bol, const char *eol,
1111 enum grep_context ctx, ssize_t *col,
1112 ssize_t *icol, int collect_hits)
1113{
1114 struct grep_expr *x = opt->pattern_expression;
1115 return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1116}
1117
1118static int match_line(struct grep_opt *opt,
1119 const char *bol, const char *eol,
1120 ssize_t *col, ssize_t *icol,
1121 enum grep_context ctx, int collect_hits)
1122{
1123 struct grep_pat *p;
1124 int hit = 0;
1125
1126 if (opt->pattern_expression)
1127 return match_expr(opt, bol, eol, ctx, col, icol,
1128 collect_hits);
1129
1130 /* we do not call with collect_hits without being extended */
1131 for (p = opt->pattern_list; p; p = p->next) {
1132 regmatch_t tmp;
1133 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1134 hit |= 1;
1135 if (!opt->columnnum) {
1136 /*
1137 * Without --column, any single match on a line
1138 * is enough to know that it needs to be
1139 * printed. With --column, scan _all_ patterns
1140 * to find the earliest.
1141 */
1142 break;
1143 }
1144 if (*col < 0 || tmp.rm_so < *col)
1145 *col = tmp.rm_so;
1146 }
1147 }
1148 return hit;
1149}
1150
1151static int match_next_pattern(struct grep_pat *p,
1152 const char *bol, const char *eol,
1153 enum grep_context ctx,
1154 regmatch_t *pmatch, int eflags)
1155{
1156 regmatch_t match;
1157
1158 if (!headerless_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
1173int grep_next_match(struct grep_opt *opt,
1174 const char *bol, const char *eol,
1175 enum grep_context ctx, regmatch_t *pmatch,
1176 enum grep_header_field field, int eflags)
1177{
1178 struct grep_pat *p;
1179 int hit = 0;
1180
1181 pmatch->rm_so = pmatch->rm_eo = -1;
1182 if (bol < eol) {
1183 for (p = ((ctx == GREP_CONTEXT_HEAD)
1184 ? opt->header_list : opt->pattern_list);
1185 p; p = p->next) {
1186 switch (p->token) {
1187 case GREP_PATTERN_HEAD:
1188 if ((field != GREP_HEADER_FIELD_MAX) &&
1189 (p->field != field))
1190 continue;
1191 /* fall thru */
1192 case GREP_PATTERN: /* atom */
1193 case GREP_PATTERN_BODY:
1194 hit |= match_next_pattern(p, bol, eol, ctx,
1195 pmatch, eflags);
1196 break;
1197 default:
1198 break;
1199 }
1200 }
1201 }
1202 return hit;
1203}
1204
1205static void show_line_header(struct grep_opt *opt, const char *name,
1206 unsigned lno, ssize_t cno, char sign)
1207{
1208 if (opt->heading && opt->last_shown == 0) {
1209 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1210 opt->output(opt, "\n", 1);
1211 }
1212 opt->last_shown = lno;
1213
1214 if (!opt->heading && opt->pathname) {
1215 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1216 output_sep(opt, sign);
1217 }
1218 if (opt->linenum) {
1219 char buf[32];
1220 xsnprintf(buf, sizeof(buf), "%d", lno);
1221 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1222 output_sep(opt, sign);
1223 }
1224 /*
1225 * Treat 'cno' as the 1-indexed offset from the start of a non-context
1226 * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1227 * being called with a context line.
1228 */
1229 if (opt->columnnum && cno) {
1230 char buf[32];
1231 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1232 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1233 output_sep(opt, sign);
1234 }
1235}
1236
1237static void show_line(struct grep_opt *opt,
1238 const char *bol, const char *eol,
1239 const char *name, unsigned lno, ssize_t cno, char sign)
1240{
1241 int rest = eol - bol;
1242 const char *match_color = NULL;
1243 const char *line_color = NULL;
1244
1245 if (opt->file_break && opt->last_shown == 0) {
1246 if (opt->show_hunk_mark)
1247 opt->output(opt, "\n", 1);
1248 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1249 if (opt->last_shown == 0) {
1250 if (opt->show_hunk_mark) {
1251 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1252 opt->output(opt, "\n", 1);
1253 }
1254 } else if (lno > opt->last_shown + 1) {
1255 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1256 opt->output(opt, "\n", 1);
1257 }
1258 }
1259 if (!opt->only_matching) {
1260 /*
1261 * In case the line we're being called with contains more than
1262 * one match, leave printing each header to the loop below.
1263 */
1264 show_line_header(opt, name, lno, cno, sign);
1265 }
1266 if (want_color(opt->color) || opt->only_matching) {
1267 regmatch_t match;
1268 enum grep_context ctx = GREP_CONTEXT_BODY;
1269 int eflags = 0;
1270
1271 if (want_color(opt->color)) {
1272 if (sign == ':')
1273 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1274 else
1275 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1276 if (sign == ':')
1277 line_color = opt->colors[GREP_COLOR_SELECTED];
1278 else if (sign == '-')
1279 line_color = opt->colors[GREP_COLOR_CONTEXT];
1280 else if (sign == '=')
1281 line_color = opt->colors[GREP_COLOR_FUNCTION];
1282 }
1283 while (grep_next_match(opt, bol, eol, ctx, &match,
1284 GREP_HEADER_FIELD_MAX, eflags)) {
1285 if (match.rm_so == match.rm_eo)
1286 break;
1287
1288 if (opt->only_matching)
1289 show_line_header(opt, name, lno, cno, sign);
1290 else
1291 output_color(opt, bol, match.rm_so, line_color);
1292 output_color(opt, bol + match.rm_so,
1293 match.rm_eo - match.rm_so, match_color);
1294 if (opt->only_matching)
1295 opt->output(opt, "\n", 1);
1296 bol += match.rm_eo;
1297 cno += match.rm_eo;
1298 rest -= match.rm_eo;
1299 eflags = REG_NOTBOL;
1300 }
1301 }
1302 if (!opt->only_matching) {
1303 output_color(opt, bol, rest, line_color);
1304 opt->output(opt, "\n", 1);
1305 }
1306}
1307
1308int grep_use_locks;
1309
1310/*
1311 * This lock protects access to the gitattributes machinery, which is
1312 * not thread-safe.
1313 */
1314pthread_mutex_t grep_attr_mutex;
1315
1316static inline void grep_attr_lock(void)
1317{
1318 if (grep_use_locks)
1319 pthread_mutex_lock(&grep_attr_mutex);
1320}
1321
1322static inline void grep_attr_unlock(void)
1323{
1324 if (grep_use_locks)
1325 pthread_mutex_unlock(&grep_attr_mutex);
1326}
1327
1328static int match_funcname(struct grep_opt *opt, struct grep_source *gs,
1329 const char *bol, const char *eol)
1330{
1331 xdemitconf_t *xecfg = opt->priv;
1332 if (xecfg && !xecfg->find_func) {
1333 grep_source_load_driver(gs, opt->repo->index);
1334 if (gs->driver->funcname.pattern) {
1335 const struct userdiff_funcname *pe = &gs->driver->funcname;
1336 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1337 } else {
1338 xecfg = opt->priv = NULL;
1339 }
1340 }
1341
1342 if (xecfg) {
1343 char buf[1];
1344 return xecfg->find_func(bol, eol - bol, buf, 1,
1345 xecfg->find_func_priv) >= 0;
1346 }
1347
1348 if (bol == eol)
1349 return 0;
1350 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1351 return 1;
1352 return 0;
1353}
1354
1355static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1356 const char *bol, unsigned lno)
1357{
1358 while (bol > gs->buf) {
1359 const char *eol = --bol;
1360
1361 while (bol > gs->buf && bol[-1] != '\n')
1362 bol--;
1363 lno--;
1364
1365 if (lno <= opt->last_shown)
1366 break;
1367
1368 if (match_funcname(opt, gs, bol, eol)) {
1369 show_line(opt, bol, eol, gs->name, lno, 0, '=');
1370 break;
1371 }
1372 }
1373}
1374
1375static int is_empty_line(const char *bol, const char *eol);
1376
1377static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1378 const char *bol, const char *end, unsigned lno)
1379{
1380 unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1381 int funcname_needed = !!opt->funcname, comment_needed = 0;
1382
1383 if (opt->pre_context < lno)
1384 from = lno - opt->pre_context;
1385 if (from <= opt->last_shown)
1386 from = opt->last_shown + 1;
1387 orig_from = from;
1388 if (opt->funcbody) {
1389 if (match_funcname(opt, gs, bol, end))
1390 comment_needed = 1;
1391 else
1392 funcname_needed = 1;
1393 from = opt->last_shown + 1;
1394 }
1395
1396 /* Rewind. */
1397 while (bol > gs->buf && cur > from) {
1398 const char *next_bol = bol;
1399 const char *eol = --bol;
1400
1401 while (bol > gs->buf && bol[-1] != '\n')
1402 bol--;
1403 cur--;
1404 if (comment_needed && (is_empty_line(bol, eol) ||
1405 match_funcname(opt, gs, bol, eol))) {
1406 comment_needed = 0;
1407 from = orig_from;
1408 if (cur < from) {
1409 cur++;
1410 bol = next_bol;
1411 break;
1412 }
1413 }
1414 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1415 funcname_lno = cur;
1416 funcname_needed = 0;
1417 if (opt->funcbody)
1418 comment_needed = 1;
1419 else
1420 from = orig_from;
1421 }
1422 }
1423
1424 /* We need to look even further back to find a function signature. */
1425 if (opt->funcname && funcname_needed)
1426 show_funcname_line(opt, gs, bol, cur);
1427
1428 /* Back forward. */
1429 while (cur < lno) {
1430 const char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1431
1432 while (*eol != '\n')
1433 eol++;
1434 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1435 bol = eol + 1;
1436 cur++;
1437 }
1438}
1439
1440static int should_lookahead(struct grep_opt *opt)
1441{
1442 struct grep_pat *p;
1443
1444 if (opt->pattern_expression)
1445 return 0; /* punt for too complex stuff */
1446 if (opt->invert)
1447 return 0;
1448 for (p = opt->pattern_list; p; p = p->next) {
1449 if (p->token != GREP_PATTERN)
1450 return 0; /* punt for "header only" and stuff */
1451 }
1452 return 1;
1453}
1454
1455static int look_ahead(struct grep_opt *opt,
1456 unsigned long *left_p,
1457 unsigned *lno_p,
1458 const char **bol_p)
1459{
1460 unsigned lno = *lno_p;
1461 const char *bol = *bol_p;
1462 struct grep_pat *p;
1463 const char *sp, *last_bol;
1464 regoff_t earliest = -1;
1465
1466 for (p = opt->pattern_list; p; p = p->next) {
1467 int hit;
1468 regmatch_t m;
1469
1470 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1471 if (hit < 0)
1472 return -1;
1473 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1474 continue;
1475 if (earliest < 0 || m.rm_so < earliest)
1476 earliest = m.rm_so;
1477 }
1478
1479 if (earliest < 0) {
1480 *bol_p = bol + *left_p;
1481 *left_p = 0;
1482 return 1;
1483 }
1484 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1485 ; /* find the beginning of the line */
1486 last_bol = sp;
1487
1488 for (sp = bol; sp < last_bol; sp++) {
1489 if (*sp == '\n')
1490 lno++;
1491 }
1492 *left_p -= last_bol - bol;
1493 *bol_p = last_bol;
1494 *lno_p = lno;
1495 return 0;
1496}
1497
1498static int fill_textconv_grep(struct repository *r,
1499 struct userdiff_driver *driver,
1500 struct grep_source *gs)
1501{
1502 struct diff_filespec *df;
1503 char *buf;
1504 size_t size;
1505
1506 if (!driver || !driver->textconv)
1507 return grep_source_load(gs);
1508
1509 /*
1510 * The textconv interface is intimately tied to diff_filespecs, so we
1511 * have to pretend to be one. If we could unify the grep_source
1512 * and diff_filespec structs, this mess could just go away.
1513 */
1514 df = alloc_filespec(gs->path);
1515 switch (gs->type) {
1516 case GREP_SOURCE_OID:
1517 fill_filespec(df, gs->identifier, 1, 0100644);
1518 break;
1519 case GREP_SOURCE_FILE:
1520 fill_filespec(df, null_oid(r->hash_algo), 0, 0100644);
1521 break;
1522 default:
1523 BUG("attempt to textconv something without a path?");
1524 }
1525
1526 /*
1527 * fill_textconv is not remotely thread-safe; it modifies the global
1528 * diff tempfile structure, writes to the_repo's odb and might
1529 * internally call thread-unsafe functions such as the
1530 * prepare_packed_git() lazy-initializator. Because of the last two, we
1531 * must ensure mutual exclusion between this call and the object reading
1532 * API, thus we use obj_read_lock() here.
1533 *
1534 * TODO: allowing text conversion to run in parallel with object
1535 * reading operations might increase performance in the multithreaded
1536 * non-worktreee git-grep with --textconv.
1537 */
1538 obj_read_lock();
1539 size = fill_textconv(r, driver, df, &buf);
1540 obj_read_unlock();
1541 free_filespec(df);
1542
1543 /*
1544 * The normal fill_textconv usage by the diff machinery would just keep
1545 * the textconv'd buf separate from the diff_filespec. But much of the
1546 * grep code passes around a grep_source and assumes that its "buf"
1547 * pointer is the beginning of the thing we are searching. So let's
1548 * install our textconv'd version into the grep_source, taking care not
1549 * to leak any existing buffer.
1550 */
1551 grep_source_clear_data(gs);
1552 gs->buf = buf;
1553 gs->size = size;
1554
1555 return 0;
1556}
1557
1558static int is_empty_line(const char *bol, const char *eol)
1559{
1560 while (bol < eol && isspace(*bol))
1561 bol++;
1562 return bol == eol;
1563}
1564
1565static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1566{
1567 const char *bol;
1568 const char *peek_bol = NULL;
1569 unsigned long left;
1570 unsigned lno = 1;
1571 unsigned last_hit = 0;
1572 int binary_match_only = 0;
1573 unsigned count = 0;
1574 int try_lookahead = 0;
1575 int show_function = 0;
1576 struct userdiff_driver *textconv = NULL;
1577 enum grep_context ctx = GREP_CONTEXT_HEAD;
1578 xdemitconf_t xecfg;
1579
1580 if (!opt->status_only && gs->name == NULL)
1581 BUG("grep call which could print a name requires "
1582 "grep_source.name be non-NULL");
1583
1584 if (!opt->output)
1585 opt->output = std_output;
1586
1587 if (opt->pre_context || opt->post_context || opt->file_break ||
1588 opt->funcbody) {
1589 /* Show hunk marks, except for the first file. */
1590 if (opt->last_shown)
1591 opt->show_hunk_mark = 1;
1592 /*
1593 * If we're using threads then we can't easily identify
1594 * the first file. Always put hunk marks in that case
1595 * and skip the very first one later in work_done().
1596 */
1597 if (opt->output != std_output)
1598 opt->show_hunk_mark = 1;
1599 }
1600 opt->last_shown = 0;
1601
1602 if (opt->allow_textconv) {
1603 grep_source_load_driver(gs, opt->repo->index);
1604 /*
1605 * We might set up the shared textconv cache data here, which
1606 * is not thread-safe. Also, get_oid_with_context() and
1607 * parse_object() might be internally called. As they are not
1608 * currently thread-safe and might be racy with object reading,
1609 * obj_read_lock() must be called.
1610 */
1611 grep_attr_lock();
1612 obj_read_lock();
1613 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1614 obj_read_unlock();
1615 grep_attr_unlock();
1616 }
1617
1618 /*
1619 * We know the result of a textconv is text, so we only have to care
1620 * about binary handling if we are not using it.
1621 */
1622 if (!textconv) {
1623 switch (opt->binary) {
1624 case GREP_BINARY_DEFAULT:
1625 if (grep_source_is_binary(gs, opt->repo->index))
1626 binary_match_only = 1;
1627 break;
1628 case GREP_BINARY_NOMATCH:
1629 if (grep_source_is_binary(gs, opt->repo->index))
1630 return 0; /* Assume unmatch */
1631 break;
1632 case GREP_BINARY_TEXT:
1633 break;
1634 default:
1635 BUG("unknown binary handling mode");
1636 }
1637 }
1638
1639 memset(&xecfg, 0, sizeof(xecfg));
1640 opt->priv = &xecfg;
1641
1642 try_lookahead = should_lookahead(opt);
1643
1644 if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1645 return 0;
1646
1647 bol = gs->buf;
1648 left = gs->size;
1649 while (left) {
1650 const char *eol;
1651 int hit;
1652 ssize_t cno;
1653 ssize_t col = -1, icol = -1;
1654
1655 /*
1656 * look_ahead() skips quickly to the line that possibly
1657 * has the next hit; don't call it if we need to do
1658 * something more than just skipping the current line
1659 * in response to an unmatch for the current line. E.g.
1660 * inside a post-context window, we will show the current
1661 * line as a context around the previous hit when it
1662 * doesn't hit.
1663 */
1664 if (try_lookahead
1665 && !(last_hit
1666 && (show_function ||
1667 lno <= last_hit + opt->post_context))) {
1668 hit = look_ahead(opt, &left, &lno, &bol);
1669 if (hit < 0)
1670 try_lookahead = 0;
1671 else if (hit)
1672 break;
1673 }
1674 eol = end_of_line(bol, &left);
1675
1676 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1677 ctx = GREP_CONTEXT_BODY;
1678
1679 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1680
1681 if (collect_hits)
1682 goto next_line;
1683
1684 /* "grep -v -e foo -e bla" should list lines
1685 * that do not have either, so inversion should
1686 * be done outside.
1687 */
1688 if (opt->invert)
1689 hit = !hit;
1690 if (opt->unmatch_name_only) {
1691 if (hit)
1692 return 0;
1693 goto next_line;
1694 }
1695 if (hit && (opt->max_count < 0 || count < opt->max_count)) {
1696 count++;
1697 if (opt->status_only)
1698 return 1;
1699 if (opt->name_only) {
1700 show_name(opt, gs->name);
1701 return 1;
1702 }
1703 if (opt->count)
1704 goto next_line;
1705 if (binary_match_only) {
1706 opt->output(opt, "Binary file ", 12);
1707 output_color(opt, gs->name, strlen(gs->name),
1708 opt->colors[GREP_COLOR_FILENAME]);
1709 opt->output(opt, " matches\n", 9);
1710 return 1;
1711 }
1712 /* Hit at this line. If we haven't shown the
1713 * pre-context lines, we would need to show them.
1714 */
1715 if (opt->pre_context || opt->funcbody)
1716 show_pre_context(opt, gs, bol, eol, lno);
1717 else if (opt->funcname)
1718 show_funcname_line(opt, gs, bol, lno);
1719 cno = opt->invert ? icol : col;
1720 if (cno < 0) {
1721 /*
1722 * A negative cno indicates that there was no
1723 * match on the line. We are thus inverted and
1724 * being asked to show all lines that _don't_
1725 * match a given expression. Therefore, set cno
1726 * to 0 to suggest the whole line matches.
1727 */
1728 cno = 0;
1729 }
1730 show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1731 last_hit = lno;
1732 if (opt->funcbody)
1733 show_function = 1;
1734 goto next_line;
1735 }
1736 if (show_function && (!peek_bol || peek_bol < bol)) {
1737 unsigned long peek_left = left;
1738 const char *peek_eol = eol;
1739
1740 /*
1741 * Trailing empty lines are not interesting.
1742 * Peek past them to see if they belong to the
1743 * body of the current function.
1744 */
1745 peek_bol = bol;
1746 while (is_empty_line(peek_bol, peek_eol)) {
1747 peek_bol = peek_eol + 1;
1748 peek_eol = end_of_line(peek_bol, &peek_left);
1749 }
1750
1751 if (peek_bol >= gs->buf + gs->size ||
1752 match_funcname(opt, gs, peek_bol, peek_eol))
1753 show_function = 0;
1754 }
1755 if (show_function ||
1756 (last_hit && lno <= last_hit + opt->post_context)) {
1757 /* If the last hit is within the post context,
1758 * we need to show this line.
1759 */
1760 show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1761 }
1762
1763 next_line:
1764 bol = eol + 1;
1765 if (!left)
1766 break;
1767 left--;
1768 lno++;
1769 }
1770
1771 if (collect_hits)
1772 return 0;
1773
1774 if (opt->status_only)
1775 return opt->unmatch_name_only;
1776 if (opt->unmatch_name_only) {
1777 /* We did not see any hit, so we want to show this */
1778 show_name(opt, gs->name);
1779 return 1;
1780 }
1781
1782 xdiff_clear_find_func(&xecfg);
1783 opt->priv = NULL;
1784
1785 /* NEEDSWORK:
1786 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1787 * which feels mostly useless but sometimes useful. Maybe
1788 * make it another option? For now suppress them.
1789 */
1790 if (opt->count && count) {
1791 char buf[32];
1792 if (opt->pathname) {
1793 output_color(opt, gs->name, strlen(gs->name),
1794 opt->colors[GREP_COLOR_FILENAME]);
1795 output_sep(opt, ':');
1796 }
1797 xsnprintf(buf, sizeof(buf), "%u\n", count);
1798 opt->output(opt, buf, strlen(buf));
1799 return 1;
1800 }
1801 return !!last_hit;
1802}
1803
1804static void clr_hit_marker(struct grep_expr *x)
1805{
1806 /* All-hit markers are meaningful only at the very top level
1807 * OR node.
1808 */
1809 while (1) {
1810 x->hit = 0;
1811 if (x->node != GREP_NODE_OR)
1812 return;
1813 x->u.binary.left->hit = 0;
1814 x = x->u.binary.right;
1815 }
1816}
1817
1818static int chk_hit_marker(struct grep_expr *x)
1819{
1820 /* Top level nodes have hit markers. See if they all are hits */
1821 while (1) {
1822 if (x->node != GREP_NODE_OR)
1823 return x->hit;
1824 if (!x->u.binary.left->hit)
1825 return 0;
1826 x = x->u.binary.right;
1827 }
1828}
1829
1830int grep_source(struct grep_opt *opt, struct grep_source *gs)
1831{
1832 /*
1833 * we do not have to do the two-pass grep when we do not check
1834 * buffer-wide "all-match".
1835 */
1836 if (!opt->all_match && !opt->no_body_match)
1837 return grep_source_1(opt, gs, 0);
1838
1839 /* Otherwise the toplevel "or" terms hit a bit differently.
1840 * We first clear hit markers from them.
1841 */
1842 clr_hit_marker(opt->pattern_expression);
1843 opt->body_hit = 0;
1844 grep_source_1(opt, gs, 1);
1845
1846 if (opt->all_match && !chk_hit_marker(opt->pattern_expression))
1847 return 0;
1848 if (opt->no_body_match && opt->body_hit)
1849 return 0;
1850
1851 return grep_source_1(opt, gs, 0);
1852}
1853
1854static void grep_source_init_buf(struct grep_source *gs,
1855 const char *buf,
1856 unsigned long size)
1857{
1858 gs->type = GREP_SOURCE_BUF;
1859 gs->name = NULL;
1860 gs->path = NULL;
1861 gs->buf = buf;
1862 gs->size = size;
1863 gs->driver = NULL;
1864 gs->identifier = NULL;
1865}
1866
1867int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size)
1868{
1869 struct grep_source gs;
1870 int r;
1871
1872 grep_source_init_buf(&gs, buf, size);
1873
1874 r = grep_source(opt, &gs);
1875
1876 grep_source_clear(&gs);
1877 return r;
1878}
1879
1880void grep_source_init_file(struct grep_source *gs, const char *name,
1881 const char *path)
1882{
1883 gs->type = GREP_SOURCE_FILE;
1884 gs->name = xstrdup_or_null(name);
1885 gs->path = xstrdup_or_null(path);
1886 gs->buf = NULL;
1887 gs->size = 0;
1888 gs->driver = NULL;
1889 gs->identifier = xstrdup(path);
1890}
1891
1892void grep_source_init_oid(struct grep_source *gs, const char *name,
1893 const char *path, const struct object_id *oid,
1894 struct repository *repo)
1895{
1896 gs->type = GREP_SOURCE_OID;
1897 gs->name = xstrdup_or_null(name);
1898 gs->path = xstrdup_or_null(path);
1899 gs->buf = NULL;
1900 gs->size = 0;
1901 gs->driver = NULL;
1902 gs->identifier = oiddup(oid);
1903 gs->repo = repo;
1904}
1905
1906void grep_source_clear(struct grep_source *gs)
1907{
1908 FREE_AND_NULL(gs->name);
1909 FREE_AND_NULL(gs->path);
1910 FREE_AND_NULL(gs->identifier);
1911 grep_source_clear_data(gs);
1912}
1913
1914void grep_source_clear_data(struct grep_source *gs)
1915{
1916 switch (gs->type) {
1917 case GREP_SOURCE_FILE:
1918 case GREP_SOURCE_OID:
1919 /* these types own the buffer */
1920 free((char *)gs->buf);
1921 gs->buf = NULL;
1922 gs->size = 0;
1923 break;
1924 case GREP_SOURCE_BUF:
1925 /* leave user-provided buf intact */
1926 break;
1927 }
1928}
1929
1930static int grep_source_load_oid(struct grep_source *gs)
1931{
1932 enum object_type type;
1933
1934 gs->buf = odb_read_object(gs->repo->objects, gs->identifier,
1935 &type, &gs->size);
1936 if (!gs->buf)
1937 return error(_("'%s': unable to read %s"),
1938 gs->name,
1939 oid_to_hex(gs->identifier));
1940 return 0;
1941}
1942
1943static int grep_source_load_file(struct grep_source *gs)
1944{
1945 const char *filename = gs->identifier;
1946 struct stat st;
1947 char *data;
1948 size_t size;
1949 int i;
1950
1951 if (lstat(filename, &st) < 0) {
1952 err_ret:
1953 if (errno != ENOENT)
1954 error_errno(_("failed to stat '%s'"), filename);
1955 return -1;
1956 }
1957 if (!S_ISREG(st.st_mode))
1958 return -1;
1959 size = xsize_t(st.st_size);
1960 i = open(filename, O_RDONLY);
1961 if (i < 0)
1962 goto err_ret;
1963 data = xmallocz(size);
1964 if (st.st_size != read_in_full(i, data, size)) {
1965 error_errno(_("'%s': short read"), filename);
1966 close(i);
1967 free(data);
1968 return -1;
1969 }
1970 close(i);
1971
1972 gs->buf = data;
1973 gs->size = size;
1974 return 0;
1975}
1976
1977static int grep_source_load(struct grep_source *gs)
1978{
1979 if (gs->buf)
1980 return 0;
1981
1982 switch (gs->type) {
1983 case GREP_SOURCE_FILE:
1984 return grep_source_load_file(gs);
1985 case GREP_SOURCE_OID:
1986 return grep_source_load_oid(gs);
1987 case GREP_SOURCE_BUF:
1988 return gs->buf ? 0 : -1;
1989 }
1990 BUG("invalid grep_source type to load");
1991}
1992
1993void grep_source_load_driver(struct grep_source *gs,
1994 struct index_state *istate)
1995{
1996 if (gs->driver)
1997 return;
1998
1999 grep_attr_lock();
2000 if (gs->path)
2001 gs->driver = userdiff_find_by_path(istate, gs->path);
2002 if (!gs->driver)
2003 gs->driver = userdiff_find_by_name("default");
2004 grep_attr_unlock();
2005}
2006
2007static int grep_source_is_binary(struct grep_source *gs,
2008 struct index_state *istate)
2009{
2010 grep_source_load_driver(gs, istate);
2011 if (gs->driver->binary != -1)
2012 return gs->driver->binary;
2013
2014 if (!grep_source_load(gs))
2015 return buffer_is_binary(gs->buf, gs->size);
2016
2017 return 0;
2018}