Merge branch 'bw/submodule-helper-usage-fix'
[git/git.git] / grep.c
CommitLineData
83b5d2f5 1#include "cache.h"
b2141fc1 2#include "config.h"
83b5d2f5 3#include "grep.h"
cbd53a21 4#include "object-store.h"
60ecac98 5#include "userdiff.h"
6bfce93e 6#include "xdiff-interface.h"
335ec3bf
JK
7#include "diff.h"
8#include "diffcore.h"
5c1ebcca 9#include "commit.h"
793dc676 10#include "quote.h"
3ac68a93 11#include "help.h"
83b5d2f5 12
07a7d656 13static int grep_source_load(struct grep_source *gs);
acd00ea0
NTND
14static int grep_source_is_binary(struct grep_source *gs,
15 struct index_state *istate);
07a7d656 16
7687a054
JH
17static struct grep_opt grep_defaults;
18
fa151dc5
NTND
19static const char *color_grep_slots[] = {
20 [GREP_COLOR_CONTEXT] = "context",
21 [GREP_COLOR_FILENAME] = "filename",
22 [GREP_COLOR_FUNCTION] = "function",
23 [GREP_COLOR_LINENO] = "lineNumber",
d036d667 24 [GREP_COLOR_COLUMNNO] = "column",
fa151dc5
NTND
25 [GREP_COLOR_MATCH_CONTEXT] = "matchContext",
26 [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
27 [GREP_COLOR_SELECTED] = "selected",
28 [GREP_COLOR_SEP] = "separator",
29};
30
379642bc
BW
31static void std_output(struct grep_opt *opt, const void *buf, size_t size)
32{
33 fwrite(buf, size, 1, stdout);
34}
35
75e5e9c3
SB
36static void color_set(char *dst, const char *color_bytes)
37{
38 xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
39}
40
7687a054
JH
41/*
42 * Initialize the grep_defaults template with hardcoded defaults.
43 * We could let the compiler do this, but without C99 initializers
44 * the code gets unwieldy and unreadable, so...
45 */
38bbc2ea 46void init_grep_defaults(struct repository *repo)
7687a054
JH
47{
48 struct grep_opt *opt = &grep_defaults;
918d4e1c
JH
49 static int run_once;
50
51 if (run_once)
52 return;
53 run_once++;
7687a054
JH
54
55 memset(opt, 0, sizeof(*opt));
38bbc2ea 56 opt->repo = repo;
7687a054
JH
57 opt->relative = 1;
58 opt->pathname = 1;
7687a054
JH
59 opt->max_depth = -1;
60 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
fa151dc5
NTND
61 color_set(opt->colors[GREP_COLOR_CONTEXT], "");
62 color_set(opt->colors[GREP_COLOR_FILENAME], "");
63 color_set(opt->colors[GREP_COLOR_FUNCTION], "");
64 color_set(opt->colors[GREP_COLOR_LINENO], "");
d036d667 65 color_set(opt->colors[GREP_COLOR_COLUMNNO], "");
fa151dc5
NTND
66 color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED);
67 color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED);
68 color_set(opt->colors[GREP_COLOR_SELECTED], "");
69 color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN);
9d8db06e 70 opt->only_matching = 0;
7687a054 71 opt->color = -1;
379642bc 72 opt->output = std_output;
7687a054
JH
73}
74
75static int parse_pattern_type_arg(const char *opt, const char *arg)
76{
77 if (!strcmp(arg, "default"))
78 return GREP_PATTERN_TYPE_UNSPECIFIED;
79 else if (!strcmp(arg, "basic"))
80 return GREP_PATTERN_TYPE_BRE;
81 else if (!strcmp(arg, "extended"))
82 return GREP_PATTERN_TYPE_ERE;
83 else if (!strcmp(arg, "fixed"))
84 return GREP_PATTERN_TYPE_FIXED;
85 else if (!strcmp(arg, "perl"))
86 return GREP_PATTERN_TYPE_PCRE;
87 die("bad %s argument: %s", opt, arg);
88}
89
3ac68a93
NTND
90define_list_config_array_extra(color_grep_slots, {"match"});
91
7687a054
JH
92/*
93 * Read the configuration file once and store it in
94 * the grep_defaults template.
95 */
96int grep_config(const char *var, const char *value, void *cb)
97{
98 struct grep_opt *opt = &grep_defaults;
fa151dc5 99 const char *slot;
7687a054
JH
100
101 if (userdiff_config(var, value) < 0)
102 return -1;
103
104 if (!strcmp(var, "grep.extendedregexp")) {
c7e38551 105 opt->extended_regexp_option = git_config_bool(var, value);
7687a054
JH
106 return 0;
107 }
108
109 if (!strcmp(var, "grep.patterntype")) {
110 opt->pattern_type_option = parse_pattern_type_arg(var, value);
111 return 0;
112 }
113
114 if (!strcmp(var, "grep.linenumber")) {
115 opt->linenum = git_config_bool(var, value);
116 return 0;
117 }
6653fec3
TB
118 if (!strcmp(var, "grep.column")) {
119 opt->columnnum = git_config_bool(var, value);
120 return 0;
121 }
7687a054 122
6453f7b3
AS
123 if (!strcmp(var, "grep.fullname")) {
124 opt->relative = !git_config_bool(var, value);
125 return 0;
126 }
127
7687a054
JH
128 if (!strcmp(var, "color.grep"))
129 opt->color = git_config_colorbool(var, value);
fa151dc5
NTND
130 if (!strcmp(var, "color.grep.match")) {
131 if (grep_config("color.grep.matchcontext", value, cb) < 0)
132 return -1;
133 if (grep_config("color.grep.matchselected", value, cb) < 0)
134 return -1;
135 } else if (skip_prefix(var, "color.grep.", &slot)) {
136 int i = LOOKUP_CONFIG(color_grep_slots, slot);
137 char *color;
138
139 if (i < 0)
140 return -1;
141 color = opt->colors[i];
7687a054
JH
142 if (!value)
143 return config_error_nonbool(var);
f6c5a296 144 return color_parse(value, color);
7687a054
JH
145 }
146 return 0;
147}
148
149/*
150 * Initialize one instance of grep_opt and copy the
151 * default values from the template we read the configuration
152 * information in an earlier call to git_config(grep_config).
153 */
38bbc2ea 154void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
7687a054
JH
155{
156 struct grep_opt *def = &grep_defaults;
fa151dc5 157 int i;
7687a054
JH
158
159 memset(opt, 0, sizeof(*opt));
38bbc2ea 160 opt->repo = repo;
7687a054
JH
161 opt->prefix = prefix;
162 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
163 opt->pattern_tail = &opt->pattern_list;
164 opt->header_tail = &opt->header_list;
165
9d8db06e 166 opt->only_matching = def->only_matching;
7687a054
JH
167 opt->color = def->color;
168 opt->extended_regexp_option = def->extended_regexp_option;
169 opt->pattern_type_option = def->pattern_type_option;
170 opt->linenum = def->linenum;
017c0fcf 171 opt->columnnum = def->columnnum;
7687a054
JH
172 opt->max_depth = def->max_depth;
173 opt->pathname = def->pathname;
7687a054 174 opt->relative = def->relative;
379642bc 175 opt->output = def->output;
7687a054 176
fa151dc5
NTND
177 for (i = 0; i < NR_GREP_COLORS; i++)
178 color_set(opt->colors[i], def->colors[i]);
7687a054 179}
07a7d656 180
8465541e 181static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
c5c31d33 182{
07a3d411
ÆAB
183 /*
184 * When committing to the pattern type by setting the relevant
185 * fields in grep_opt it's generally not necessary to zero out
186 * the fields we're not choosing, since they won't have been
187 * set by anything. The extended_regexp_option field is the
188 * only exception to this.
189 *
190 * This is because in the process of parsing grep.patternType
191 * & grep.extendedRegexp we set opt->pattern_type_option and
192 * opt->extended_regexp_option, respectively. We then
193 * internally use opt->extended_regexp_option to see if we're
194 * compiling an ERE. It must be unset if that's not actually
195 * the case.
196 */
197 if (pattern_type != GREP_PATTERN_TYPE_ERE &&
198 opt->extended_regexp_option)
199 opt->extended_regexp_option = 0;
200
c5c31d33
JH
201 switch (pattern_type) {
202 case GREP_PATTERN_TYPE_UNSPECIFIED:
203 /* fall through */
204
205 case GREP_PATTERN_TYPE_BRE:
c5c31d33
JH
206 break;
207
208 case GREP_PATTERN_TYPE_ERE:
07a3d411 209 opt->extended_regexp_option = 1;
c5c31d33
JH
210 break;
211
212 case GREP_PATTERN_TYPE_FIXED:
213 opt->fixed = 1;
c5c31d33
JH
214 break;
215
216 case GREP_PATTERN_TYPE_PCRE:
94da9193 217#ifdef USE_LIBPCRE2
94da9193
ÆAB
218 opt->pcre2 = 1;
219#else
220 /*
221 * It's important that pcre1 always be assigned to
222 * even when there's no USE_LIBPCRE* defined. We still
223 * call the PCRE stub function, it just dies with
224 * "cannot use Perl-compatible regexes[...]".
225 */
6d4b5747 226 opt->pcre1 = 1;
94da9193 227#endif
c5c31d33
JH
228 break;
229 }
230}
231
8465541e
JH
232void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
233{
234 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
235 grep_set_pattern_type_option(pattern_type, opt);
236 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
237 grep_set_pattern_type_option(opt->pattern_type_option, opt);
238 else if (opt->extended_regexp_option)
07a3d411
ÆAB
239 /*
240 * This branch *must* happen after setting from the
241 * opt->pattern_type_option above, we don't want
242 * grep.extendedRegexp to override grep.patternType!
243 */
8465541e
JH
244 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
245}
246
fc456751
RS
247static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
248 const char *origin, int no,
249 enum grep_pat_token t,
250 enum grep_header_field field)
a4d7d2c6
JH
251{
252 struct grep_pat *p = xcalloc(1, sizeof(*p));
526a858a 253 p->pattern = xmemdupz(pat, patlen);
fc456751
RS
254 p->patternlen = patlen;
255 p->origin = origin;
256 p->no = no;
257 p->token = t;
a4d7d2c6 258 p->field = field;
fc456751
RS
259 return p;
260}
261
2b3873ff
RS
262static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
263{
264 **tail = p;
265 *tail = &p->next;
a4d7d2c6 266 p->next = NULL;
526a858a
RS
267
268 switch (p->token) {
269 case GREP_PATTERN: /* atom */
270 case GREP_PATTERN_HEAD:
271 case GREP_PATTERN_BODY:
272 for (;;) {
273 struct grep_pat *new_pat;
274 size_t len = 0;
275 char *cp = p->pattern + p->patternlen, *nl = NULL;
276 while (++len <= p->patternlen) {
277 if (*(--cp) == '\n') {
278 nl = cp;
279 break;
280 }
281 }
282 if (!nl)
283 break;
284 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
285 p->no, p->token, p->field);
286 new_pat->next = p->next;
287 if (!p->next)
288 *tail = &new_pat->next;
289 p->next = new_pat;
290 *nl = '\0';
291 p->patternlen -= len;
292 }
293 break;
294 default:
295 break;
296 }
2b3873ff
RS
297}
298
fc456751
RS
299void append_header_grep_pattern(struct grep_opt *opt,
300 enum grep_header_field field, const char *pat)
301{
302 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
303 GREP_PATTERN_HEAD, field);
baa6378f
JH
304 if (field == GREP_HEADER_REFLOG)
305 opt->use_reflog_filter = 1;
2b3873ff 306 do_append_grep_pat(&opt->header_tail, p);
a4d7d2c6
JH
307}
308
83b5d2f5
JH
309void append_grep_pattern(struct grep_opt *opt, const char *pat,
310 const char *origin, int no, enum grep_pat_token t)
ed40a095
RS
311{
312 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
313}
314
315void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
316 const char *origin, int no, enum grep_pat_token t)
83b5d2f5 317{
fc456751 318 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
2b3873ff 319 do_append_grep_pat(&opt->pattern_tail, p);
83b5d2f5
JH
320}
321
5b594f45
FK
322struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
323{
324 struct grep_pat *pat;
325 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
326 *ret = *opt;
327
328 ret->pattern_list = NULL;
329 ret->pattern_tail = &ret->pattern_list;
330
331 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
332 {
333 if(pat->token == GREP_PATTERN_HEAD)
334 append_header_grep_pattern(ret, pat->field,
335 pat->pattern);
336 else
ed40a095
RS
337 append_grep_pat(ret, pat->pattern, pat->patternlen,
338 pat->origin, pat->no, pat->token);
5b594f45
FK
339 }
340
341 return ret;
342}
343
a30c148a
MK
344static NORETURN void compile_regexp_failed(const struct grep_pat *p,
345 const char *error)
346{
347 char where[1024];
348
349 if (p->no)
19bdd3e7 350 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
a30c148a 351 else if (p->origin)
19bdd3e7 352 xsnprintf(where, sizeof(where), "%s, ", p->origin);
a30c148a
MK
353 else
354 where[0] = 0;
355
356 die("%s'%s': %s", where, p->pattern, error);
357}
358
543f1c0c
ÆAB
359static int is_fixed(const char *s, size_t len)
360{
361 size_t i;
362
363 for (i = 0; i < len; i++) {
364 if (is_regex_special(s[i]))
365 return 0;
366 }
367
368 return 1;
369}
370
219e65b6
ÆAB
371static int has_null(const char *s, size_t len)
372{
373 /*
374 * regcomp cannot accept patterns with NULs so when using it
375 * we consider any pattern containing a NUL fixed.
376 */
377 if (memchr(s, 0, len))
378 return 1;
379
380 return 0;
381}
382
3485bea1 383#ifdef USE_LIBPCRE1
6d4b5747 384static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
63e7e9d8
MK
385{
386 const char *error;
387 int erroffset;
fba4f125 388 int options = PCRE_MULTILINE;
63e7e9d8 389
9d9babb8
NTND
390 if (opt->ignore_case) {
391 if (has_non_ascii(p->pattern))
6d4b5747 392 p->pcre1_tables = pcre_maketables();
63e7e9d8 393 options |= PCRE_CASELESS;
9d9babb8 394 }
18547aac
NTND
395 if (is_utf8_locale() && has_non_ascii(p->pattern))
396 options |= PCRE_UTF8;
63e7e9d8 397
6d4b5747
ÆAB
398 p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
399 p->pcre1_tables);
400 if (!p->pcre1_regexp)
63e7e9d8
MK
401 compile_regexp_failed(p, error);
402
2fff1e19 403 p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
6d4b5747 404 if (!p->pcre1_extra_info && error)
63e7e9d8 405 die("%s", error);
fbaceaac 406
e87de7ca 407#ifdef GIT_PCRE1_USE_JIT
fbaceaac
ÆAB
408 pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
409 if (p->pcre1_jit_on == 1) {
410 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
411 if (!p->pcre1_jit_stack)
412 die("Couldn't allocate PCRE JIT stack");
413 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
414 } else if (p->pcre1_jit_on != 0) {
033abf97 415 BUG("The pcre1_jit_on variable should be 0 or 1, not %d",
fbaceaac
ÆAB
416 p->pcre1_jit_on);
417 }
418#endif
63e7e9d8
MK
419}
420
6d4b5747 421static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
63e7e9d8
MK
422 regmatch_t *match, int eflags)
423{
ad7c543e 424 int ovector[30], ret, flags = PCRE_NO_UTF8_CHECK;
63e7e9d8
MK
425
426 if (eflags & REG_NOTBOL)
427 flags |= PCRE_NOTBOL;
428
e87de7ca 429#ifdef GIT_PCRE1_USE_JIT
fbaceaac
ÆAB
430 if (p->pcre1_jit_on) {
431 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
432 eol - line, 0, flags, ovector,
433 ARRAY_SIZE(ovector), p->pcre1_jit_stack);
434 } else
435#endif
436 {
437 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
438 eol - line, 0, flags, ovector,
439 ARRAY_SIZE(ovector));
440 }
441
63e7e9d8
MK
442 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
443 die("pcre_exec failed with error code %d", ret);
444 if (ret > 0) {
445 ret = 0;
446 match->rm_so = ovector[0];
447 match->rm_eo = ovector[1];
448 }
449
450 return ret;
451}
452
6d4b5747 453static void free_pcre1_regexp(struct grep_pat *p)
63e7e9d8 454{
6d4b5747 455 pcre_free(p->pcre1_regexp);
e87de7ca 456#ifdef GIT_PCRE1_USE_JIT
fbaceaac
ÆAB
457 if (p->pcre1_jit_on) {
458 pcre_free_study(p->pcre1_extra_info);
459 pcre_jit_stack_free(p->pcre1_jit_stack);
460 } else
461#endif
462 {
463 pcre_free(p->pcre1_extra_info);
464 }
6d4b5747 465 pcre_free((void *)p->pcre1_tables);
63e7e9d8 466}
3485bea1 467#else /* !USE_LIBPCRE1 */
6d4b5747 468static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
63e7e9d8
MK
469{
470 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
471}
472
6d4b5747 473static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
63e7e9d8
MK
474 regmatch_t *match, int eflags)
475{
476 return 1;
477}
478
6d4b5747 479static void free_pcre1_regexp(struct grep_pat *p)
63e7e9d8
MK
480{
481}
3485bea1 482#endif /* !USE_LIBPCRE1 */
63e7e9d8 483
94da9193
ÆAB
484#ifdef USE_LIBPCRE2
485static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
486{
487 int error;
488 PCRE2_UCHAR errbuf[256];
489 PCRE2_SIZE erroffset;
490 int options = PCRE2_MULTILINE;
491 const uint8_t *character_tables = NULL;
492 int jitret;
a25b9085
ÆAB
493 int patinforet;
494 size_t jitsizearg;
94da9193
ÆAB
495
496 assert(opt->pcre2);
497
498 p->pcre2_compile_context = NULL;
499
500 if (opt->ignore_case) {
501 if (has_non_ascii(p->pattern)) {
502 character_tables = pcre2_maketables(NULL);
503 p->pcre2_compile_context = pcre2_compile_context_create(NULL);
504 pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
505 }
506 options |= PCRE2_CASELESS;
507 }
508 if (is_utf8_locale() && has_non_ascii(p->pattern))
509 options |= PCRE2_UTF;
510
511 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
512 p->patternlen, options, &error, &erroffset,
513 p->pcre2_compile_context);
514
515 if (p->pcre2_pattern) {
516 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
517 if (!p->pcre2_match_data)
518 die("Couldn't allocate PCRE2 match data");
519 } else {
520 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
521 compile_regexp_failed(p, (const char *)&errbuf);
522 }
523
524 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
525 if (p->pcre2_jit_on == 1) {
526 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
527 if (jitret)
528 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
a25b9085
ÆAB
529
530 /*
531 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
532 * tells us whether the library itself supports JIT,
533 * but to see whether we're going to be actually using
534 * JIT we need to extract PCRE2_INFO_JITSIZE from the
535 * pattern *after* we do pcre2_jit_compile() above.
536 *
537 * This is because if the pattern contains the
538 * (*NO_JIT) verb (see pcre2syntax(3))
539 * pcre2_jit_compile() will exit early with 0. If we
540 * then proceed to call pcre2_jit_match() further down
541 * the line instead of pcre2_match() we'll either
542 * segfault (pre PCRE 10.31) or run into a fatal error
543 * (post PCRE2 10.31)
544 */
545 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
546 if (patinforet)
547 BUG("pcre2_pattern_info() failed: %d", patinforet);
548 if (jitsizearg == 0) {
549 p->pcre2_jit_on = 0;
550 return;
551 }
552
94da9193
ÆAB
553 p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
554 if (!p->pcre2_jit_stack)
555 die("Couldn't allocate PCRE2 JIT stack");
556 p->pcre2_match_context = pcre2_match_context_create(NULL);
674ad936 557 if (!p->pcre2_match_context)
94da9193
ÆAB
558 die("Couldn't allocate PCRE2 match context");
559 pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
560 } else if (p->pcre2_jit_on != 0) {
033abf97 561 BUG("The pcre2_jit_on variable should be 0 or 1, not %d",
f0025327 562 p->pcre2_jit_on);
94da9193
ÆAB
563 }
564}
565
566static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
567 regmatch_t *match, int eflags)
568{
569 int ret, flags = 0;
570 PCRE2_SIZE *ovector;
571 PCRE2_UCHAR errbuf[256];
572
573 if (eflags & REG_NOTBOL)
574 flags |= PCRE2_NOTBOL;
575
576 if (p->pcre2_jit_on)
577 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
578 eol - line, 0, flags, p->pcre2_match_data,
579 NULL);
580 else
581 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
582 eol - line, 0, flags, p->pcre2_match_data,
583 NULL);
584
585 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
586 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
587 die("%s failed with error code %d: %s",
588 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
589 errbuf);
590 }
591 if (ret > 0) {
592 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
593 ret = 0;
594 match->rm_so = (int)ovector[0];
595 match->rm_eo = (int)ovector[1];
596 }
597
598 return ret;
599}
600
601static void free_pcre2_pattern(struct grep_pat *p)
602{
603 pcre2_compile_context_free(p->pcre2_compile_context);
604 pcre2_code_free(p->pcre2_pattern);
605 pcre2_match_data_free(p->pcre2_match_data);
606 pcre2_jit_stack_free(p->pcre2_jit_stack);
607 pcre2_match_context_free(p->pcre2_match_context);
608}
609#else /* !USE_LIBPCRE2 */
610static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
611{
612 /*
613 * Unreachable until USE_LIBPCRE2 becomes synonymous with
614 * USE_LIBPCRE. See the sibling comment in
615 * grep_set_pattern_type_option().
616 */
617 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
618}
619
620static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
621 regmatch_t *match, int eflags)
622{
623 return 1;
624}
625
626static void free_pcre2_pattern(struct grep_pat *p)
627{
628}
629#endif /* !USE_LIBPCRE2 */
630
793dc676
NTND
631static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
632{
633 struct strbuf sb = STRBUF_INIT;
634 int err;
1ceababc 635 int regflags = 0;
793dc676
NTND
636
637 basic_regex_quote_buf(&sb, p->pattern);
793dc676
NTND
638 if (opt->ignore_case)
639 regflags |= REG_ICASE;
640 err = regcomp(&p->regexp, sb.buf, regflags);
641 if (opt->debug)
642 fprintf(stderr, "fixed %s\n", sb.buf);
643 strbuf_release(&sb);
644 if (err) {
645 char errbuf[1024];
646 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
793dc676
NTND
647 compile_regexp_failed(p, errbuf);
648 }
649}
650
83b5d2f5
JH
651static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
652{
07a3d411 653 int ascii_only;
c822255c 654 int err;
07a3d411 655 int regflags = REG_NEWLINE;
c822255c 656
d7eb527d 657 p->word_regexp = opt->word_regexp;
5183bf67 658 p->ignore_case = opt->ignore_case;
5c1ebcca 659 ascii_only = !has_non_ascii(p->pattern);
d7eb527d 660
793dc676
NTND
661 /*
662 * Even when -F (fixed) asks us to do a non-regexp search, we
663 * may not be able to correctly case-fold when -i
664 * (ignore-case) is asked (in which case, we'll synthesize a
665 * regexp to match the pattern that matches regexp special
666 * characters literally, while ignoring case differences). On
667 * the other hand, even without -F, if the pattern does not
668 * have any regexp special characters and there is no need for
669 * case-folding search, we can internally turn it into a
670 * simple string match using kws. p->fixed tells us if we
671 * want to use kws.
672 */
219e65b6
ÆAB
673 if (opt->fixed ||
674 has_null(p->pattern, p->patternlen) ||
675 is_fixed(p->pattern, p->patternlen))
07a3d411 676 p->fixed = !p->ignore_case || ascii_only;
9eceddee
FK
677
678 if (p->fixed) {
07a3d411 679 p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
9eceddee
FK
680 kwsincr(p->kws, p->pattern, p->patternlen);
681 kwsprep(p->kws);
c822255c 682 return;
793dc676
NTND
683 } else if (opt->fixed) {
684 /*
685 * We come here when the pattern has the non-ascii
686 * characters we cannot case-fold, and asked to
687 * ignore-case.
688 */
689 compile_fixed_regexp(p, opt);
690 return;
9eceddee 691 }
c822255c 692
94da9193
ÆAB
693 if (opt->pcre2) {
694 compile_pcre2_pattern(p, opt);
695 return;
696 }
697
6d4b5747
ÆAB
698 if (opt->pcre1) {
699 compile_pcre1_regexp(p, opt);
63e7e9d8
MK
700 return;
701 }
702
07a3d411
ÆAB
703 if (p->ignore_case)
704 regflags |= REG_ICASE;
705 if (opt->extended_regexp_option)
706 regflags |= REG_EXTENDED;
707 err = regcomp(&p->regexp, p->pattern, regflags);
83b5d2f5
JH
708 if (err) {
709 char errbuf[1024];
83b5d2f5 710 regerror(err, &p->regexp, errbuf, 1024);
a30c148a 711 compile_regexp_failed(p, errbuf);
83b5d2f5
JH
712 }
713}
714
0ab7befa 715static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
716static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
717{
718 struct grep_pat *p;
719 struct grep_expr *x;
720
721 p = *list;
c922b01f
LT
722 if (!p)
723 return NULL;
83b5d2f5
JH
724 switch (p->token) {
725 case GREP_PATTERN: /* atom */
480c1ca6
JH
726 case GREP_PATTERN_HEAD:
727 case GREP_PATTERN_BODY:
83b5d2f5
JH
728 x = xcalloc(1, sizeof (struct grep_expr));
729 x->node = GREP_NODE_ATOM;
730 x->u.atom = p;
731 *list = p->next;
732 return x;
733 case GREP_OPEN_PAREN:
734 *list = p->next;
0ab7befa 735 x = compile_pattern_or(list);
83b5d2f5
JH
736 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
737 die("unmatched parenthesis");
738 *list = (*list)->next;
739 return x;
740 default:
741 return NULL;
742 }
743}
744
745static struct grep_expr *compile_pattern_not(struct grep_pat **list)
746{
747 struct grep_pat *p;
748 struct grep_expr *x;
749
750 p = *list;
c922b01f
LT
751 if (!p)
752 return NULL;
83b5d2f5
JH
753 switch (p->token) {
754 case GREP_NOT:
755 if (!p->next)
756 die("--not not followed by pattern expression");
757 *list = p->next;
758 x = xcalloc(1, sizeof (struct grep_expr));
759 x->node = GREP_NODE_NOT;
760 x->u.unary = compile_pattern_not(list);
761 if (!x->u.unary)
762 die("--not followed by non pattern expression");
763 return x;
764 default:
765 return compile_pattern_atom(list);
766 }
767}
768
769static struct grep_expr *compile_pattern_and(struct grep_pat **list)
770{
771 struct grep_pat *p;
772 struct grep_expr *x, *y, *z;
773
774 x = compile_pattern_not(list);
775 p = *list;
776 if (p && p->token == GREP_AND) {
777 if (!p->next)
778 die("--and not followed by pattern expression");
779 *list = p->next;
780 y = compile_pattern_and(list);
781 if (!y)
782 die("--and not followed by pattern expression");
783 z = xcalloc(1, sizeof (struct grep_expr));
784 z->node = GREP_NODE_AND;
785 z->u.binary.left = x;
786 z->u.binary.right = y;
787 return z;
788 }
789 return x;
790}
791
792static struct grep_expr *compile_pattern_or(struct grep_pat **list)
793{
794 struct grep_pat *p;
795 struct grep_expr *x, *y, *z;
796
797 x = compile_pattern_and(list);
798 p = *list;
799 if (x && p && p->token != GREP_CLOSE_PAREN) {
800 y = compile_pattern_or(list);
801 if (!y)
802 die("not a pattern expression %s", p->pattern);
803 z = xcalloc(1, sizeof (struct grep_expr));
804 z->node = GREP_NODE_OR;
805 z->u.binary.left = x;
806 z->u.binary.right = y;
807 return z;
808 }
809 return x;
810}
811
812static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
813{
814 return compile_pattern_or(list);
815}
816
17bf35a3
JH
817static void indent(int in)
818{
819 while (in-- > 0)
820 fputc(' ', stderr);
821}
822
823static void dump_grep_pat(struct grep_pat *p)
824{
825 switch (p->token) {
826 case GREP_AND: fprintf(stderr, "*and*"); break;
827 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
828 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
829 case GREP_NOT: fprintf(stderr, "*not*"); break;
830 case GREP_OR: fprintf(stderr, "*or*"); break;
831
832 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
833 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
834 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
835 }
836
837 switch (p->token) {
838 default: break;
839 case GREP_PATTERN_HEAD:
840 fprintf(stderr, "<head %d>", p->field); break;
841 case GREP_PATTERN_BODY:
842 fprintf(stderr, "<body>"); break;
843 }
844 switch (p->token) {
845 default: break;
846 case GREP_PATTERN_HEAD:
847 case GREP_PATTERN_BODY:
848 case GREP_PATTERN:
849 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
850 break;
851 }
852 fputc('\n', stderr);
853}
854
855static void dump_grep_expression_1(struct grep_expr *x, int in)
856{
857 indent(in);
858 switch (x->node) {
859 case GREP_NODE_TRUE:
860 fprintf(stderr, "true\n");
861 break;
862 case GREP_NODE_ATOM:
863 dump_grep_pat(x->u.atom);
864 break;
865 case GREP_NODE_NOT:
866 fprintf(stderr, "(not\n");
867 dump_grep_expression_1(x->u.unary, in+1);
868 indent(in);
869 fprintf(stderr, ")\n");
870 break;
871 case GREP_NODE_AND:
872 fprintf(stderr, "(and\n");
873 dump_grep_expression_1(x->u.binary.left, in+1);
874 dump_grep_expression_1(x->u.binary.right, in+1);
875 indent(in);
876 fprintf(stderr, ")\n");
877 break;
878 case GREP_NODE_OR:
879 fprintf(stderr, "(or\n");
880 dump_grep_expression_1(x->u.binary.left, in+1);
881 dump_grep_expression_1(x->u.binary.right, in+1);
882 indent(in);
883 fprintf(stderr, ")\n");
884 break;
885 }
886}
887
07a7d656 888static void dump_grep_expression(struct grep_opt *opt)
17bf35a3
JH
889{
890 struct grep_expr *x = opt->pattern_expression;
891
892 if (opt->all_match)
893 fprintf(stderr, "[all-match]\n");
894 dump_grep_expression_1(x, 0);
895 fflush(NULL);
896}
897
5aaeb733
JH
898static struct grep_expr *grep_true_expr(void)
899{
900 struct grep_expr *z = xcalloc(1, sizeof(*z));
901 z->node = GREP_NODE_TRUE;
902 return z;
903}
904
905static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
906{
907 struct grep_expr *z = xcalloc(1, sizeof(*z));
908 z->node = GREP_NODE_OR;
909 z->u.binary.left = left;
910 z->u.binary.right = right;
911 return z;
912}
913
95ce9ce2 914static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
83b5d2f5
JH
915{
916 struct grep_pat *p;
95ce9ce2 917 struct grep_expr *header_expr;
5aaeb733
JH
918 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
919 enum grep_header_field fld;
83b5d2f5 920
95ce9ce2
JH
921 if (!opt->header_list)
922 return NULL;
2385f246 923
95ce9ce2
JH
924 for (p = opt->header_list; p; p = p->next) {
925 if (p->token != GREP_PATTERN_HEAD)
033abf97 926 BUG("a non-header pattern in grep header list.");
3ce3ffb8
AP
927 if (p->field < GREP_HEADER_FIELD_MIN ||
928 GREP_HEADER_FIELD_MAX <= p->field)
033abf97 929 BUG("unknown header field %d", p->field);
95ce9ce2 930 compile_regexp(p, opt);
80235ba7 931 }
5aaeb733
JH
932
933 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
934 header_group[fld] = NULL;
935
936 for (p = opt->header_list; p; p = p->next) {
937 struct grep_expr *h;
938 struct grep_pat *pp = p;
939
940 h = compile_pattern_atom(&pp);
941 if (!h || pp != p->next)
033abf97 942 BUG("malformed header expr");
5aaeb733
JH
943 if (!header_group[p->field]) {
944 header_group[p->field] = h;
945 continue;
946 }
947 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
948 }
949
950 header_expr = NULL;
951
952 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
953 if (!header_group[fld])
954 continue;
955 if (!header_expr)
956 header_expr = grep_true_expr();
957 header_expr = grep_or_expr(header_group[fld], header_expr);
958 }
95ce9ce2
JH
959 return header_expr;
960}
961
13e4fc7e
JH
962static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
963{
964 struct grep_expr *z = x;
965
966 while (x) {
967 assert(x->node == GREP_NODE_OR);
968 if (x->u.binary.right &&
969 x->u.binary.right->node == GREP_NODE_TRUE) {
970 x->u.binary.right = y;
971 break;
972 }
973 x = x->u.binary.right;
974 }
975 return z;
976}
977
17bf35a3 978static void compile_grep_patterns_real(struct grep_opt *opt)
95ce9ce2
JH
979{
980 struct grep_pat *p;
981 struct grep_expr *header_expr = prep_header_patterns(opt);
0ab7befa 982
83b5d2f5 983 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
984 switch (p->token) {
985 case GREP_PATTERN: /* atom */
986 case GREP_PATTERN_HEAD:
987 case GREP_PATTERN_BODY:
c822255c 988 compile_regexp(p, opt);
480c1ca6
JH
989 break;
990 default:
83b5d2f5 991 opt->extended = 1;
480c1ca6
JH
992 break;
993 }
83b5d2f5
JH
994 }
995
80235ba7
JH
996 if (opt->all_match || header_expr)
997 opt->extended = 1;
17bf35a3 998 else if (!opt->extended && !opt->debug)
83b5d2f5
JH
999 return;
1000
83b5d2f5 1001 p = opt->pattern_list;
ba150a3f
MB
1002 if (p)
1003 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
1004 if (p)
1005 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
1006
1007 if (!header_expr)
1008 return;
1009
5aaeb733 1010 if (!opt->pattern_expression)
80235ba7 1011 opt->pattern_expression = header_expr;
13e4fc7e
JH
1012 else if (opt->all_match)
1013 opt->pattern_expression = grep_splice_or(header_expr,
1014 opt->pattern_expression);
5aaeb733
JH
1015 else
1016 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
1017 header_expr);
80235ba7 1018 opt->all_match = 1;
83b5d2f5
JH
1019}
1020
17bf35a3
JH
1021void compile_grep_patterns(struct grep_opt *opt)
1022{
1023 compile_grep_patterns_real(opt);
1024 if (opt->debug)
1025 dump_grep_expression(opt);
1026}
1027
b48fb5b6
JH
1028static void free_pattern_expr(struct grep_expr *x)
1029{
1030 switch (x->node) {
5aaeb733 1031 case GREP_NODE_TRUE:
b48fb5b6
JH
1032 case GREP_NODE_ATOM:
1033 break;
1034 case GREP_NODE_NOT:
1035 free_pattern_expr(x->u.unary);
1036 break;
1037 case GREP_NODE_AND:
1038 case GREP_NODE_OR:
1039 free_pattern_expr(x->u.binary.left);
1040 free_pattern_expr(x->u.binary.right);
1041 break;
1042 }
1043 free(x);
1044}
1045
1046void free_grep_patterns(struct grep_opt *opt)
1047{
1048 struct grep_pat *p, *n;
1049
1050 for (p = opt->pattern_list; p; p = n) {
1051 n = p->next;
1052 switch (p->token) {
1053 case GREP_PATTERN: /* atom */
1054 case GREP_PATTERN_HEAD:
1055 case GREP_PATTERN_BODY:
9eceddee
FK
1056 if (p->kws)
1057 kwsfree(p->kws);
6d4b5747
ÆAB
1058 else if (p->pcre1_regexp)
1059 free_pcre1_regexp(p);
94da9193
ÆAB
1060 else if (p->pcre2_pattern)
1061 free_pcre2_pattern(p);
63e7e9d8
MK
1062 else
1063 regfree(&p->regexp);
526a858a 1064 free(p->pattern);
b48fb5b6
JH
1065 break;
1066 default:
1067 break;
1068 }
1069 free(p);
1070 }
1071
1072 if (!opt->extended)
1073 return;
1074 free_pattern_expr(opt->pattern_expression);
1075}
1076
83b5d2f5
JH
1077static char *end_of_line(char *cp, unsigned long *left)
1078{
1079 unsigned long l = *left;
1080 while (l && *cp != '\n') {
1081 l--;
1082 cp++;
1083 }
1084 *left = l;
1085 return cp;
1086}
1087
1088static int word_char(char ch)
1089{
1090 return isalnum(ch) || ch == '_';
1091}
1092
55f638bd
ML
1093static void output_color(struct grep_opt *opt, const void *data, size_t size,
1094 const char *color)
1095{
daa0c3d9 1096 if (want_color(opt->color) && color && color[0]) {
55f638bd
ML
1097 opt->output(opt, color, strlen(color));
1098 opt->output(opt, data, size);
1099 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1100 } else
1101 opt->output(opt, data, size);
1102}
1103
1104static void output_sep(struct grep_opt *opt, char sign)
1105{
1106 if (opt->null_following_name)
1107 opt->output(opt, "\0", 1);
1108 else
fa151dc5 1109 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
55f638bd
ML
1110}
1111
83caecca
RZ
1112static void show_name(struct grep_opt *opt, const char *name)
1113{
fa151dc5 1114 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
5b594f45 1115 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
1116}
1117
ed40a095
RS
1118static int fixmatch(struct grep_pat *p, char *line, char *eol,
1119 regmatch_t *match)
83b5d2f5 1120{
9eceddee
FK
1121 struct kwsmatch kwsm;
1122 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1123 if (offset == -1) {
83b5d2f5
JH
1124 match->rm_so = match->rm_eo = -1;
1125 return REG_NOMATCH;
9eceddee
FK
1126 } else {
1127 match->rm_so = offset;
1128 match->rm_eo = match->rm_so + kwsm.size[0];
83b5d2f5
JH
1129 return 0;
1130 }
1131}
1132
97e77784
MK
1133static int patmatch(struct grep_pat *p, char *line, char *eol,
1134 regmatch_t *match, int eflags)
1135{
1136 int hit;
1137
1138 if (p->fixed)
1139 hit = !fixmatch(p, line, eol, match);
6d4b5747
ÆAB
1140 else if (p->pcre1_regexp)
1141 hit = !pcre1match(p, line, eol, match, eflags);
94da9193
ÆAB
1142 else if (p->pcre2_pattern)
1143 hit = !pcre2match(p, line, eol, match, eflags);
97e77784 1144 else
b7d36ffc
JS
1145 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1146 eflags);
97e77784
MK
1147
1148 return hit;
1149}
1150
a4d7d2c6
JH
1151static int strip_timestamp(char *bol, char **eol_p)
1152{
1153 char *eol = *eol_p;
1154 int ch;
1155
1156 while (bol < --eol) {
1157 if (*eol != '>')
1158 continue;
1159 *eol_p = ++eol;
1160 ch = *eol;
1161 *eol = '\0';
1162 return ch;
1163 }
1164 return 0;
1165}
1166
1167static struct {
1168 const char *field;
1169 size_t len;
1170} header_field[] = {
1171 { "author ", 7 },
1172 { "committer ", 10 },
72fd13f7 1173 { "reflog ", 7 },
a4d7d2c6
JH
1174};
1175
d7eb527d 1176static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
1177 enum grep_context ctx,
1178 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
1179{
1180 int hit = 0;
a4d7d2c6 1181 int saved_ch = 0;
e701fadb 1182 const char *start = bol;
83b5d2f5 1183
480c1ca6
JH
1184 if ((p->token != GREP_PATTERN) &&
1185 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1186 return 0;
1187
a4d7d2c6
JH
1188 if (p->token == GREP_PATTERN_HEAD) {
1189 const char *field;
1190 size_t len;
1191 assert(p->field < ARRAY_SIZE(header_field));
1192 field = header_field[p->field].field;
1193 len = header_field[p->field].len;
1194 if (strncmp(bol, field, len))
1195 return 0;
1196 bol += len;
ad4813b3
NTND
1197 switch (p->field) {
1198 case GREP_HEADER_AUTHOR:
1199 case GREP_HEADER_COMMITTER:
1200 saved_ch = strip_timestamp(bol, &eol);
1201 break;
1202 default:
1203 break;
1204 }
a4d7d2c6
JH
1205 }
1206
83b5d2f5 1207 again:
97e77784 1208 hit = patmatch(p, bol, eol, pmatch, eflags);
83b5d2f5 1209
d7eb527d 1210 if (hit && p->word_regexp) {
83b5d2f5 1211 if ((pmatch[0].rm_so < 0) ||
84201eae 1212 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
1213 (pmatch[0].rm_eo < 0) ||
1214 (eol - bol) < pmatch[0].rm_eo)
1215 die("regexp returned nonsense");
1216
1217 /* Match beginning must be either beginning of the
1218 * line, or at word boundary (i.e. the last char must
1219 * not be a word char). Similarly, match end must be
1220 * either end of the line, or at word boundary
1221 * (i.e. the next char must not be a word char).
1222 */
fb62eb7f 1223 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
1224 !word_char(bol[pmatch[0].rm_so-1])) &&
1225 ((pmatch[0].rm_eo == (eol-bol)) ||
1226 !word_char(bol[pmatch[0].rm_eo])) )
1227 ;
1228 else
1229 hit = 0;
1230
84201eae
RS
1231 /* Words consist of at least one character. */
1232 if (pmatch->rm_so == pmatch->rm_eo)
1233 hit = 0;
1234
83b5d2f5
JH
1235 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1236 /* There could be more than one match on the
1237 * line, and the first match might not be
1238 * strict word match. But later ones could be!
fb62eb7f
RS
1239 * Forward to the next possible start, i.e. the
1240 * next position following a non-word char.
83b5d2f5
JH
1241 */
1242 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
1243 while (word_char(bol[-1]) && bol < eol)
1244 bol++;
dbb6a4ad 1245 eflags |= REG_NOTBOL;
fb62eb7f
RS
1246 if (bol < eol)
1247 goto again;
83b5d2f5
JH
1248 }
1249 }
a4d7d2c6
JH
1250 if (p->token == GREP_PATTERN_HEAD && saved_ch)
1251 *eol = saved_ch;
e701fadb
RS
1252 if (hit) {
1253 pmatch[0].rm_so += bol - start;
1254 pmatch[0].rm_eo += bol - start;
1255 }
83b5d2f5
JH
1256 return hit;
1257}
1258
68d686e6
TB
1259static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol,
1260 char *eol, enum grep_context ctx, ssize_t *col,
1261 ssize_t *icol, int collect_hits)
83b5d2f5 1262{
0ab7befa
JH
1263 int h = 0;
1264
c922b01f
LT
1265 if (!x)
1266 die("Not a valid grep expression");
83b5d2f5 1267 switch (x->node) {
5aaeb733
JH
1268 case GREP_NODE_TRUE:
1269 h = 1;
1270 break;
83b5d2f5 1271 case GREP_NODE_ATOM:
68d686e6
TB
1272 {
1273 regmatch_t tmp;
1274 h = match_one_pattern(x->u.atom, bol, eol, ctx,
1275 &tmp, 0);
1276 if (h && (*col < 0 || tmp.rm_so < *col))
1277 *col = tmp.rm_so;
1278 }
83b5d2f5
JH
1279 break;
1280 case GREP_NODE_NOT:
68d686e6
TB
1281 /*
1282 * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1283 */
1284 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1285 0);
0ab7befa 1286 break;
83b5d2f5 1287 case GREP_NODE_AND:
017c0fcf 1288 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
68d686e6 1289 icol, 0);
017c0fcf
TB
1290 if (h || opt->columnnum) {
1291 /*
1292 * Don't short-circuit AND when given --column, since a
1293 * NOT earlier in the tree may turn this into an OR. In
1294 * this case, see the below comment.
1295 */
1296 h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1297 ctx, col, icol, 0);
1298 }
0ab7befa 1299 break;
83b5d2f5 1300 case GREP_NODE_OR:
017c0fcf
TB
1301 if (!(collect_hits || opt->columnnum)) {
1302 /*
1303 * Don't short-circuit OR when given --column (or
1304 * collecting hits) to ensure we don't skip a later
1305 * child that would produce an earlier match.
1306 */
68d686e6
TB
1307 return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1308 ctx, col, icol, 0) ||
1309 match_expr_eval(opt, x->u.binary.right, bol,
1310 eol, ctx, col, icol, 0));
017c0fcf 1311 }
68d686e6
TB
1312 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1313 icol, 0);
017c0fcf
TB
1314 if (collect_hits)
1315 x->u.binary.left->hit |= h;
68d686e6 1316 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
017c0fcf 1317 icol, collect_hits);
0ab7befa
JH
1318 break;
1319 default:
d7530708 1320 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 1321 }
0ab7befa
JH
1322 if (collect_hits)
1323 x->hit |= h;
1324 return h;
83b5d2f5
JH
1325}
1326
480c1ca6 1327static int match_expr(struct grep_opt *opt, char *bol, char *eol,
68d686e6
TB
1328 enum grep_context ctx, ssize_t *col,
1329 ssize_t *icol, int collect_hits)
83b5d2f5
JH
1330{
1331 struct grep_expr *x = opt->pattern_expression;
68d686e6 1332 return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
83b5d2f5
JH
1333}
1334
480c1ca6 1335static int match_line(struct grep_opt *opt, char *bol, char *eol,
68d686e6 1336 ssize_t *col, ssize_t *icol,
0ab7befa 1337 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1338{
1339 struct grep_pat *p;
017c0fcf 1340 int hit = 0;
79212772 1341
83b5d2f5 1342 if (opt->extended)
68d686e6
TB
1343 return match_expr(opt, bol, eol, ctx, col, icol,
1344 collect_hits);
0ab7befa
JH
1345
1346 /* we do not call with collect_hits without being extended */
83b5d2f5 1347 for (p = opt->pattern_list; p; p = p->next) {
68d686e6
TB
1348 regmatch_t tmp;
1349 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
017c0fcf
TB
1350 hit |= 1;
1351 if (!opt->columnnum) {
1352 /*
1353 * Without --column, any single match on a line
1354 * is enough to know that it needs to be
1355 * printed. With --column, scan _all_ patterns
1356 * to find the earliest.
1357 */
1358 break;
1359 }
1360 if (*col < 0 || tmp.rm_so < *col)
1361 *col = tmp.rm_so;
68d686e6 1362 }
83b5d2f5 1363 }
017c0fcf 1364 return hit;
83b5d2f5
JH
1365}
1366
7e8f59d5
RS
1367static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1368 enum grep_context ctx,
1369 regmatch_t *pmatch, int eflags)
1370{
1371 regmatch_t match;
1372
1373 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1374 return 0;
1375 if (match.rm_so < 0 || match.rm_eo < 0)
1376 return 0;
1377 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1378 if (match.rm_so > pmatch->rm_so)
1379 return 1;
1380 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1381 return 1;
1382 }
1383 pmatch->rm_so = match.rm_so;
1384 pmatch->rm_eo = match.rm_eo;
1385 return 1;
1386}
1387
1388static int next_match(struct grep_opt *opt, char *bol, char *eol,
1389 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1390{
1391 struct grep_pat *p;
1392 int hit = 0;
1393
1394 pmatch->rm_so = pmatch->rm_eo = -1;
1395 if (bol < eol) {
1396 for (p = opt->pattern_list; p; p = p->next) {
1397 switch (p->token) {
1398 case GREP_PATTERN: /* atom */
1399 case GREP_PATTERN_HEAD:
1400 case GREP_PATTERN_BODY:
1401 hit |= match_next_pattern(p, bol, eol, ctx,
1402 pmatch, eflags);
1403 break;
1404 default:
1405 break;
1406 }
1407 }
1408 }
1409 return hit;
1410}
1411
c707ded3
TB
1412static void show_line_header(struct grep_opt *opt, const char *name,
1413 unsigned lno, ssize_t cno, char sign)
7e8f59d5 1414{
1d84f72e 1415 if (opt->heading && opt->last_shown == 0) {
fa151dc5 1416 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1d84f72e
RS
1417 opt->output(opt, "\n", 1);
1418 }
5dd06d38
RS
1419 opt->last_shown = lno;
1420
1d84f72e 1421 if (!opt->heading && opt->pathname) {
fa151dc5 1422 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
55f638bd 1423 output_sep(opt, sign);
5b594f45
FK
1424 }
1425 if (opt->linenum) {
1426 char buf[32];
1a168e5c 1427 xsnprintf(buf, sizeof(buf), "%d", lno);
fa151dc5 1428 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
55f638bd 1429 output_sep(opt, sign);
5b594f45 1430 }
89252cd0
TB
1431 /*
1432 * Treat 'cno' as the 1-indexed offset from the start of a non-context
1433 * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1434 * being called with a context line.
1435 */
1436 if (opt->columnnum && cno) {
1437 char buf[32];
1438 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
d036d667 1439 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
89252cd0
TB
1440 output_sep(opt, sign);
1441 }
c707ded3
TB
1442}
1443
1444static void show_line(struct grep_opt *opt, char *bol, char *eol,
1445 const char *name, unsigned lno, ssize_t cno, char sign)
1446{
1447 int rest = eol - bol;
9d8db06e
TB
1448 const char *match_color = NULL;
1449 const char *line_color = NULL;
c707ded3
TB
1450
1451 if (opt->file_break && opt->last_shown == 0) {
1452 if (opt->show_hunk_mark)
1453 opt->output(opt, "\n", 1);
1454 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1455 if (opt->last_shown == 0) {
1456 if (opt->show_hunk_mark) {
87ece7ce 1457 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
c707ded3
TB
1458 opt->output(opt, "\n", 1);
1459 }
1460 } else if (lno > opt->last_shown + 1) {
87ece7ce 1461 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
c707ded3
TB
1462 opt->output(opt, "\n", 1);
1463 }
1464 }
9d8db06e
TB
1465 if (!opt->only_matching) {
1466 /*
1467 * In case the line we're being called with contains more than
1468 * one match, leave printing each header to the loop below.
1469 */
1470 show_line_header(opt, name, lno, cno, sign);
1471 }
1472 if (opt->color || opt->only_matching) {
7e8f59d5
RS
1473 regmatch_t match;
1474 enum grep_context ctx = GREP_CONTEXT_BODY;
1475 int ch = *eol;
1476 int eflags = 0;
1477
9d8db06e
TB
1478 if (opt->color) {
1479 if (sign == ':')
87ece7ce 1480 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
9d8db06e 1481 else
87ece7ce 1482 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
9d8db06e 1483 if (sign == ':')
87ece7ce 1484 line_color = opt->colors[GREP_COLOR_SELECTED];
9d8db06e 1485 else if (sign == '-')
87ece7ce 1486 line_color = opt->colors[GREP_COLOR_CONTEXT];
9d8db06e 1487 else if (sign == '=')
87ece7ce 1488 line_color = opt->colors[GREP_COLOR_FUNCTION];
9d8db06e 1489 }
7e8f59d5
RS
1490 *eol = '\0';
1491 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
1492 if (match.rm_so == match.rm_eo)
1493 break;
5b594f45 1494
9d8db06e
TB
1495 if (opt->only_matching)
1496 show_line_header(opt, name, lno, cno, sign);
1497 else
1498 output_color(opt, bol, match.rm_so, line_color);
55f638bd 1499 output_color(opt, bol + match.rm_so,
79a77109 1500 match.rm_eo - match.rm_so, match_color);
9d8db06e
TB
1501 if (opt->only_matching)
1502 opt->output(opt, "\n", 1);
7e8f59d5 1503 bol += match.rm_eo;
9d8db06e 1504 cno += match.rm_eo;
7e8f59d5
RS
1505 rest -= match.rm_eo;
1506 eflags = REG_NOTBOL;
1507 }
1508 *eol = ch;
1509 }
9d8db06e
TB
1510 if (!opt->only_matching) {
1511 output_color(opt, bol, rest, line_color);
1512 opt->output(opt, "\n", 1);
1513 }
7e8f59d5
RS
1514}
1515
78db6ea9
JK
1516int grep_use_locks;
1517
0579f91d
TR
1518/*
1519 * This lock protects access to the gitattributes machinery, which is
1520 * not thread-safe.
1521 */
1522pthread_mutex_t grep_attr_mutex;
1523
78db6ea9 1524static inline void grep_attr_lock(void)
0579f91d 1525{
78db6ea9 1526 if (grep_use_locks)
0579f91d
TR
1527 pthread_mutex_lock(&grep_attr_mutex);
1528}
1529
78db6ea9 1530static inline void grep_attr_unlock(void)
0579f91d 1531{
78db6ea9 1532 if (grep_use_locks)
0579f91d
TR
1533 pthread_mutex_unlock(&grep_attr_mutex);
1534}
b3aeb285
JK
1535
1536/*
1537 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1538 */
1539pthread_mutex_t grep_read_mutex;
1540
e1327023 1541static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
2944e4e6 1542{
60ecac98 1543 xdemitconf_t *xecfg = opt->priv;
0579f91d 1544 if (xecfg && !xecfg->find_func) {
acd00ea0 1545 grep_source_load_driver(gs, opt->repo->index);
94ad9d9e
JK
1546 if (gs->driver->funcname.pattern) {
1547 const struct userdiff_funcname *pe = &gs->driver->funcname;
0579f91d
TR
1548 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1549 } else {
1550 xecfg = opt->priv = NULL;
1551 }
1552 }
1553
1554 if (xecfg) {
60ecac98
RS
1555 char buf[1];
1556 return xecfg->find_func(bol, eol - bol, buf, 1,
1557 xecfg->find_func_priv) >= 0;
1558 }
1559
2944e4e6
RS
1560 if (bol == eol)
1561 return 0;
1562 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1563 return 1;
1564 return 0;
1565}
1566
e1327023
JK
1567static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1568 char *bol, unsigned lno)
2944e4e6 1569{
e1327023 1570 while (bol > gs->buf) {
2944e4e6
RS
1571 char *eol = --bol;
1572
e1327023 1573 while (bol > gs->buf && bol[-1] != '\n')
2944e4e6
RS
1574 bol--;
1575 lno--;
1576
1577 if (lno <= opt->last_shown)
1578 break;
1579
e1327023 1580 if (match_funcname(opt, gs, bol, eol)) {
89252cd0 1581 show_line(opt, bol, eol, gs->name, lno, 0, '=');
2944e4e6
RS
1582 break;
1583 }
1584 }
1585}
1586
a5dc20b0
RS
1587static int is_empty_line(const char *bol, const char *eol);
1588
e1327023 1589static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
ba8ea749 1590 char *bol, char *end, unsigned lno)
49de3216 1591{
6653a01b 1592 unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
a5dc20b0 1593 int funcname_needed = !!opt->funcname, comment_needed = 0;
ba8ea749 1594
49de3216
RS
1595 if (opt->pre_context < lno)
1596 from = lno - opt->pre_context;
1597 if (from <= opt->last_shown)
1598 from = opt->last_shown + 1;
6653a01b 1599 orig_from = from;
a5dc20b0
RS
1600 if (opt->funcbody) {
1601 if (match_funcname(opt, gs, bol, end))
1602 comment_needed = 1;
1603 else
1604 funcname_needed = 1;
6653a01b
RS
1605 from = opt->last_shown + 1;
1606 }
49de3216
RS
1607
1608 /* Rewind. */
6653a01b 1609 while (bol > gs->buf && cur > from) {
a5dc20b0 1610 char *next_bol = bol;
2944e4e6
RS
1611 char *eol = --bol;
1612
e1327023 1613 while (bol > gs->buf && bol[-1] != '\n')
49de3216
RS
1614 bol--;
1615 cur--;
a5dc20b0
RS
1616 if (comment_needed && (is_empty_line(bol, eol) ||
1617 match_funcname(opt, gs, bol, eol))) {
1618 comment_needed = 0;
1619 from = orig_from;
1620 if (cur < from) {
1621 cur++;
1622 bol = next_bol;
1623 break;
1624 }
1625 }
e1327023 1626 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
2944e4e6
RS
1627 funcname_lno = cur;
1628 funcname_needed = 0;
a5dc20b0
RS
1629 if (opt->funcbody)
1630 comment_needed = 1;
1631 else
1632 from = orig_from;
2944e4e6 1633 }
49de3216
RS
1634 }
1635
2944e4e6
RS
1636 /* We need to look even further back to find a function signature. */
1637 if (opt->funcname && funcname_needed)
e1327023 1638 show_funcname_line(opt, gs, bol, cur);
2944e4e6 1639
49de3216
RS
1640 /* Back forward. */
1641 while (cur < lno) {
2944e4e6 1642 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
1643
1644 while (*eol != '\n')
1645 eol++;
89252cd0 1646 show_line(opt, bol, eol, gs->name, cur, 0, sign);
49de3216
RS
1647 bol = eol + 1;
1648 cur++;
1649 }
1650}
1651
a26345b6
JH
1652static int should_lookahead(struct grep_opt *opt)
1653{
1654 struct grep_pat *p;
1655
1656 if (opt->extended)
1657 return 0; /* punt for too complex stuff */
1658 if (opt->invert)
1659 return 0;
1660 for (p = opt->pattern_list; p; p = p->next) {
1661 if (p->token != GREP_PATTERN)
1662 return 0; /* punt for "header only" and stuff */
1663 }
1664 return 1;
1665}
1666
1667static int look_ahead(struct grep_opt *opt,
1668 unsigned long *left_p,
1669 unsigned *lno_p,
1670 char **bol_p)
1671{
1672 unsigned lno = *lno_p;
1673 char *bol = *bol_p;
1674 struct grep_pat *p;
1675 char *sp, *last_bol;
1676 regoff_t earliest = -1;
1677
1678 for (p = opt->pattern_list; p; p = p->next) {
1679 int hit;
1680 regmatch_t m;
1681
97e77784 1682 hit = patmatch(p, bol, bol + *left_p, &m, 0);
a26345b6
JH
1683 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1684 continue;
1685 if (earliest < 0 || m.rm_so < earliest)
1686 earliest = m.rm_so;
1687 }
1688
1689 if (earliest < 0) {
1690 *bol_p = bol + *left_p;
1691 *left_p = 0;
1692 return 1;
1693 }
1694 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1695 ; /* find the beginning of the line */
1696 last_bol = sp;
1697
1698 for (sp = bol; sp < last_bol; sp++) {
1699 if (*sp == '\n')
1700 lno++;
1701 }
1702 *left_p -= last_bol - bol;
1703 *bol_p = last_bol;
1704 *lno_p = lno;
1705 return 0;
1706}
1707
38bbc2ea
NTND
1708static int fill_textconv_grep(struct repository *r,
1709 struct userdiff_driver *driver,
335ec3bf
JK
1710 struct grep_source *gs)
1711{
1712 struct diff_filespec *df;
1713 char *buf;
1714 size_t size;
1715
1716 if (!driver || !driver->textconv)
1717 return grep_source_load(gs);
1718
1719 /*
1720 * The textconv interface is intimately tied to diff_filespecs, so we
1721 * have to pretend to be one. If we could unify the grep_source
1722 * and diff_filespec structs, this mess could just go away.
1723 */
1724 df = alloc_filespec(gs->path);
1725 switch (gs->type) {
1c41c82b 1726 case GREP_SOURCE_OID:
335ec3bf
JK
1727 fill_filespec(df, gs->identifier, 1, 0100644);
1728 break;
1729 case GREP_SOURCE_FILE:
f9704c2d 1730 fill_filespec(df, &null_oid, 0, 0100644);
335ec3bf
JK
1731 break;
1732 default:
033abf97 1733 BUG("attempt to textconv something without a path?");
335ec3bf
JK
1734 }
1735
1736 /*
1737 * fill_textconv is not remotely thread-safe; it may load objects
1738 * behind the scenes, and it modifies the global diff tempfile
1739 * structure.
1740 */
1741 grep_read_lock();
38bbc2ea 1742 size = fill_textconv(r, driver, df, &buf);
335ec3bf
JK
1743 grep_read_unlock();
1744 free_filespec(df);
1745
1746 /*
1747 * The normal fill_textconv usage by the diff machinery would just keep
1748 * the textconv'd buf separate from the diff_filespec. But much of the
1749 * grep code passes around a grep_source and assumes that its "buf"
1750 * pointer is the beginning of the thing we are searching. So let's
1751 * install our textconv'd version into the grep_source, taking care not
1752 * to leak any existing buffer.
1753 */
1754 grep_source_clear_data(gs);
1755 gs->buf = buf;
1756 gs->size = size;
1757
1758 return 0;
1759}
1760
4aa2c475
RS
1761static int is_empty_line(const char *bol, const char *eol)
1762{
1763 while (bol < eol && isspace(*bol))
1764 bol++;
1765 return bol == eol;
1766}
1767
e1327023 1768static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
83b5d2f5 1769{
e1327023 1770 char *bol;
4aa2c475 1771 char *peek_bol = NULL;
e1327023 1772 unsigned long left;
83b5d2f5 1773 unsigned lno = 1;
83b5d2f5 1774 unsigned last_hit = 0;
83b5d2f5 1775 int binary_match_only = 0;
83b5d2f5 1776 unsigned count = 0;
a26345b6 1777 int try_lookahead = 0;
ba8ea749 1778 int show_function = 0;
335ec3bf 1779 struct userdiff_driver *textconv = NULL;
480c1ca6 1780 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 1781 xdemitconf_t xecfg;
83b5d2f5 1782
de99eb0c
ES
1783 if (!opt->status_only && gs->name == NULL)
1784 BUG("grep call which could print a name requires "
1785 "grep_source.name be non-NULL");
1786
5b594f45
FK
1787 if (!opt->output)
1788 opt->output = std_output;
1789
ba8ea749
RS
1790 if (opt->pre_context || opt->post_context || opt->file_break ||
1791 opt->funcbody) {
08303c36
RS
1792 /* Show hunk marks, except for the first file. */
1793 if (opt->last_shown)
1794 opt->show_hunk_mark = 1;
1795 /*
1796 * If we're using threads then we can't easily identify
1797 * the first file. Always put hunk marks in that case
1798 * and skip the very first one later in work_done().
1799 */
1800 if (opt->output != std_output)
1801 opt->show_hunk_mark = 1;
1802 }
431d6e7b
RS
1803 opt->last_shown = 0;
1804
335ec3bf 1805 if (opt->allow_textconv) {
acd00ea0 1806 grep_source_load_driver(gs, opt->repo->index);
335ec3bf
JK
1807 /*
1808 * We might set up the shared textconv cache data here, which
1809 * is not thread-safe.
1810 */
1811 grep_attr_lock();
bd7ad45b 1812 textconv = userdiff_get_textconv(opt->repo, gs->driver);
335ec3bf
JK
1813 grep_attr_unlock();
1814 }
1815
1816 /*
1817 * We know the result of a textconv is text, so we only have to care
1818 * about binary handling if we are not using it.
1819 */
1820 if (!textconv) {
1821 switch (opt->binary) {
1822 case GREP_BINARY_DEFAULT:
acd00ea0 1823 if (grep_source_is_binary(gs, opt->repo->index))
335ec3bf
JK
1824 binary_match_only = 1;
1825 break;
1826 case GREP_BINARY_NOMATCH:
acd00ea0 1827 if (grep_source_is_binary(gs, opt->repo->index))
335ec3bf
JK
1828 return 0; /* Assume unmatch */
1829 break;
1830 case GREP_BINARY_TEXT:
1831 break;
1832 default:
033abf97 1833 BUG("unknown binary handling mode");
335ec3bf 1834 }
83b5d2f5
JH
1835 }
1836
60ecac98 1837 memset(&xecfg, 0, sizeof(xecfg));
0579f91d
TR
1838 opt->priv = &xecfg;
1839
a26345b6 1840 try_lookahead = should_lookahead(opt);
60ecac98 1841
38bbc2ea 1842 if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
08265798
JK
1843 return 0;
1844
e1327023
JK
1845 bol = gs->buf;
1846 left = gs->size;
83b5d2f5
JH
1847 while (left) {
1848 char *eol, ch;
0ab7befa 1849 int hit;
89252cd0 1850 ssize_t cno;
68d686e6 1851 ssize_t col = -1, icol = -1;
83b5d2f5 1852
a26345b6 1853 /*
8997da38 1854 * look_ahead() skips quickly to the line that possibly
a26345b6
JH
1855 * has the next hit; don't call it if we need to do
1856 * something more than just skipping the current line
1857 * in response to an unmatch for the current line. E.g.
1858 * inside a post-context window, we will show the current
1859 * line as a context around the previous hit when it
1860 * doesn't hit.
1861 */
1862 if (try_lookahead
1863 && !(last_hit
ba8ea749
RS
1864 && (show_function ||
1865 lno <= last_hit + opt->post_context))
a26345b6
JH
1866 && look_ahead(opt, &left, &lno, &bol))
1867 break;
83b5d2f5
JH
1868 eol = end_of_line(bol, &left);
1869 ch = *eol;
1870 *eol = 0;
1871
480c1ca6
JH
1872 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1873 ctx = GREP_CONTEXT_BODY;
1874
68d686e6 1875 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
83b5d2f5
JH
1876 *eol = ch;
1877
0ab7befa
JH
1878 if (collect_hits)
1879 goto next_line;
1880
83b5d2f5
JH
1881 /* "grep -v -e foo -e bla" should list lines
1882 * that do not have either, so inversion should
1883 * be done outside.
1884 */
1885 if (opt->invert)
1886 hit = !hit;
1887 if (opt->unmatch_name_only) {
1888 if (hit)
1889 return 0;
1890 goto next_line;
1891 }
1892 if (hit) {
1893 count++;
1894 if (opt->status_only)
1895 return 1;
321ffcc0 1896 if (opt->name_only) {
e1327023 1897 show_name(opt, gs->name);
321ffcc0
RS
1898 return 1;
1899 }
c30c10cf
RS
1900 if (opt->count)
1901 goto next_line;
83b5d2f5 1902 if (binary_match_only) {
5b594f45 1903 opt->output(opt, "Binary file ", 12);
e1327023 1904 output_color(opt, gs->name, strlen(gs->name),
fa151dc5 1905 opt->colors[GREP_COLOR_FILENAME]);
5b594f45 1906 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
1907 return 1;
1908 }
83b5d2f5
JH
1909 /* Hit at this line. If we haven't shown the
1910 * pre-context lines, we would need to show them.
83b5d2f5 1911 */
ba8ea749 1912 if (opt->pre_context || opt->funcbody)
e1327023 1913 show_pre_context(opt, gs, bol, eol, lno);
2944e4e6 1914 else if (opt->funcname)
e1327023 1915 show_funcname_line(opt, gs, bol, lno);
89252cd0
TB
1916 cno = opt->invert ? icol : col;
1917 if (cno < 0) {
1918 /*
1919 * A negative cno indicates that there was no
1920 * match on the line. We are thus inverted and
1921 * being asked to show all lines that _don't_
1922 * match a given expression. Therefore, set cno
1923 * to 0 to suggest the whole line matches.
1924 */
1925 cno = 0;
1926 }
1927 show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
5dd06d38 1928 last_hit = lno;
ba8ea749
RS
1929 if (opt->funcbody)
1930 show_function = 1;
1931 goto next_line;
83b5d2f5 1932 }
4aa2c475
RS
1933 if (show_function && (!peek_bol || peek_bol < bol)) {
1934 unsigned long peek_left = left;
1935 char *peek_eol = eol;
1936
1937 /*
1938 * Trailing empty lines are not interesting.
1939 * Peek past them to see if they belong to the
1940 * body of the current function.
1941 */
1942 peek_bol = bol;
1943 while (is_empty_line(peek_bol, peek_eol)) {
1944 peek_bol = peek_eol + 1;
1945 peek_eol = end_of_line(peek_bol, &peek_left);
1946 }
1947
1948 if (match_funcname(opt, gs, peek_bol, peek_eol))
1949 show_function = 0;
1950 }
ba8ea749
RS
1951 if (show_function ||
1952 (last_hit && lno <= last_hit + opt->post_context)) {
83b5d2f5
JH
1953 /* If the last hit is within the post context,
1954 * we need to show this line.
1955 */
89252cd0 1956 show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
83b5d2f5 1957 }
83b5d2f5
JH
1958
1959 next_line:
1960 bol = eol + 1;
1961 if (!left)
1962 break;
1963 left--;
1964 lno++;
1965 }
1966
0ab7befa
JH
1967 if (collect_hits)
1968 return 0;
b48fb5b6 1969
83b5d2f5 1970 if (opt->status_only)
e1f68c66 1971 return opt->unmatch_name_only;
83b5d2f5
JH
1972 if (opt->unmatch_name_only) {
1973 /* We did not see any hit, so we want to show this */
e1327023 1974 show_name(opt, gs->name);
83b5d2f5
JH
1975 return 1;
1976 }
1977
60ecac98
RS
1978 xdiff_clear_find_func(&xecfg);
1979 opt->priv = NULL;
1980
83b5d2f5
JH
1981 /* NEEDSWORK:
1982 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1983 * which feels mostly useless but sometimes useful. Maybe
1984 * make it another option? For now suppress them.
1985 */
5b594f45
FK
1986 if (opt->count && count) {
1987 char buf[32];
f76d947a
RS
1988 if (opt->pathname) {
1989 output_color(opt, gs->name, strlen(gs->name),
fa151dc5 1990 opt->colors[GREP_COLOR_FILENAME]);
f76d947a
RS
1991 output_sep(opt, ':');
1992 }
1a168e5c 1993 xsnprintf(buf, sizeof(buf), "%u\n", count);
5b594f45 1994 opt->output(opt, buf, strlen(buf));
c30c10cf 1995 return 1;
5b594f45 1996 }
83b5d2f5
JH
1997 return !!last_hit;
1998}
1999
0ab7befa
JH
2000static void clr_hit_marker(struct grep_expr *x)
2001{
2002 /* All-hit markers are meaningful only at the very top level
2003 * OR node.
2004 */
2005 while (1) {
2006 x->hit = 0;
2007 if (x->node != GREP_NODE_OR)
2008 return;
2009 x->u.binary.left->hit = 0;
2010 x = x->u.binary.right;
2011 }
2012}
2013
2014static int chk_hit_marker(struct grep_expr *x)
2015{
2016 /* Top level nodes have hit markers. See if they all are hits */
2017 while (1) {
2018 if (x->node != GREP_NODE_OR)
2019 return x->hit;
2020 if (!x->u.binary.left->hit)
2021 return 0;
2022 x = x->u.binary.right;
2023 }
2024}
2025
e1327023 2026int grep_source(struct grep_opt *opt, struct grep_source *gs)
0ab7befa
JH
2027{
2028 /*
2029 * we do not have to do the two-pass grep when we do not check
2030 * buffer-wide "all-match".
2031 */
2032 if (!opt->all_match)
e1327023 2033 return grep_source_1(opt, gs, 0);
0ab7befa
JH
2034
2035 /* Otherwise the toplevel "or" terms hit a bit differently.
2036 * We first clear hit markers from them.
2037 */
2038 clr_hit_marker(opt->pattern_expression);
e1327023 2039 grep_source_1(opt, gs, 1);
0ab7befa
JH
2040
2041 if (!chk_hit_marker(opt->pattern_expression))
2042 return 0;
2043
e1327023
JK
2044 return grep_source_1(opt, gs, 0);
2045}
2046
c876d6da 2047int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
e1327023
JK
2048{
2049 struct grep_source gs;
2050 int r;
2051
55c61688 2052 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
e1327023
JK
2053 gs.buf = buf;
2054 gs.size = size;
2055
2056 r = grep_source(opt, &gs);
2057
2058 grep_source_clear(&gs);
2059 return r;
2060}
2061
2062void grep_source_init(struct grep_source *gs, enum grep_source_type type,
55c61688
NTND
2063 const char *name, const char *path,
2064 const void *identifier)
e1327023
JK
2065{
2066 gs->type = type;
8c53f071
JK
2067 gs->name = xstrdup_or_null(name);
2068 gs->path = xstrdup_or_null(path);
e1327023
JK
2069 gs->buf = NULL;
2070 gs->size = 0;
94ad9d9e 2071 gs->driver = NULL;
e1327023
JK
2072
2073 switch (type) {
2074 case GREP_SOURCE_FILE:
2075 gs->identifier = xstrdup(identifier);
2076 break;
1c41c82b
BW
2077 case GREP_SOURCE_OID:
2078 gs->identifier = oiddup(identifier);
e1327023
JK
2079 break;
2080 case GREP_SOURCE_BUF:
2081 gs->identifier = NULL;
4538eef5 2082 break;
e1327023
JK
2083 }
2084}
2085
2086void grep_source_clear(struct grep_source *gs)
2087{
88ce3ef6
ÆAB
2088 FREE_AND_NULL(gs->name);
2089 FREE_AND_NULL(gs->path);
2090 FREE_AND_NULL(gs->identifier);
e1327023
JK
2091 grep_source_clear_data(gs);
2092}
2093
2094void grep_source_clear_data(struct grep_source *gs)
2095{
2096 switch (gs->type) {
2097 case GREP_SOURCE_FILE:
1c41c82b 2098 case GREP_SOURCE_OID:
6a83d902 2099 FREE_AND_NULL(gs->buf);
e1327023
JK
2100 gs->size = 0;
2101 break;
2102 case GREP_SOURCE_BUF:
2103 /* leave user-provided buf intact */
2104 break;
2105 }
2106}
2107
1c41c82b 2108static int grep_source_load_oid(struct grep_source *gs)
e1327023
JK
2109{
2110 enum object_type type;
2111
2112 grep_read_lock();
b4f5aca4 2113 gs->buf = read_object_file(gs->identifier, &type, &gs->size);
e1327023
JK
2114 grep_read_unlock();
2115
2116 if (!gs->buf)
2117 return error(_("'%s': unable to read %s"),
2118 gs->name,
1c41c82b 2119 oid_to_hex(gs->identifier));
e1327023
JK
2120 return 0;
2121}
2122
2123static int grep_source_load_file(struct grep_source *gs)
2124{
2125 const char *filename = gs->identifier;
2126 struct stat st;
2127 char *data;
2128 size_t size;
2129 int i;
2130
2131 if (lstat(filename, &st) < 0) {
2132 err_ret:
2133 if (errno != ENOENT)
7645d8f1 2134 error_errno(_("failed to stat '%s'"), filename);
e1327023
JK
2135 return -1;
2136 }
2137 if (!S_ISREG(st.st_mode))
2138 return -1;
2139 size = xsize_t(st.st_size);
2140 i = open(filename, O_RDONLY);
2141 if (i < 0)
2142 goto err_ret;
3733e694 2143 data = xmallocz(size);
e1327023 2144 if (st.st_size != read_in_full(i, data, size)) {
7645d8f1 2145 error_errno(_("'%s': short read"), filename);
e1327023
JK
2146 close(i);
2147 free(data);
2148 return -1;
2149 }
2150 close(i);
e1327023
JK
2151
2152 gs->buf = data;
2153 gs->size = size;
2154 return 0;
2155}
2156
3083301e 2157static int grep_source_load(struct grep_source *gs)
e1327023
JK
2158{
2159 if (gs->buf)
2160 return 0;
2161
2162 switch (gs->type) {
2163 case GREP_SOURCE_FILE:
2164 return grep_source_load_file(gs);
1c41c82b
BW
2165 case GREP_SOURCE_OID:
2166 return grep_source_load_oid(gs);
e1327023
JK
2167 case GREP_SOURCE_BUF:
2168 return gs->buf ? 0 : -1;
2169 }
033abf97 2170 BUG("invalid grep_source type to load");
0ab7befa 2171}
94ad9d9e 2172
acd00ea0
NTND
2173void grep_source_load_driver(struct grep_source *gs,
2174 struct index_state *istate)
94ad9d9e
JK
2175{
2176 if (gs->driver)
2177 return;
2178
2179 grep_attr_lock();
55c61688 2180 if (gs->path)
acd00ea0 2181 gs->driver = userdiff_find_by_path(istate, gs->path);
94ad9d9e
JK
2182 if (!gs->driver)
2183 gs->driver = userdiff_find_by_name("default");
2184 grep_attr_unlock();
2185}
41b59bfc 2186
acd00ea0
NTND
2187static int grep_source_is_binary(struct grep_source *gs,
2188 struct index_state *istate)
41b59bfc 2189{
acd00ea0 2190 grep_source_load_driver(gs, istate);
41b59bfc
JK
2191 if (gs->driver->binary != -1)
2192 return gs->driver->binary;
2193
2194 if (!grep_source_load(gs))
2195 return buffer_is_binary(gs->buf, gs->size);
2196
2197 return 0;
2198}