grep: remove redundant and verbose re-assignments to 0
[git/git.git] / grep.c
CommitLineData
83b5d2f5 1#include "cache.h"
b2141fc1 2#include "config.h"
83b5d2f5 3#include "grep.h"
60ecac98 4#include "userdiff.h"
6bfce93e 5#include "xdiff-interface.h"
335ec3bf
JK
6#include "diff.h"
7#include "diffcore.h"
5c1ebcca 8#include "commit.h"
793dc676 9#include "quote.h"
83b5d2f5 10
07a7d656
JH
11static int grep_source_load(struct grep_source *gs);
12static int grep_source_is_binary(struct grep_source *gs);
13
7687a054
JH
14static struct grep_opt grep_defaults;
15
379642bc
BW
16static void std_output(struct grep_opt *opt, const void *buf, size_t size)
17{
18 fwrite(buf, size, 1, stdout);
19}
20
7687a054
JH
21/*
22 * Initialize the grep_defaults template with hardcoded defaults.
23 * We could let the compiler do this, but without C99 initializers
24 * the code gets unwieldy and unreadable, so...
25 */
26void init_grep_defaults(void)
27{
28 struct grep_opt *opt = &grep_defaults;
918d4e1c
JH
29 static int run_once;
30
31 if (run_once)
32 return;
33 run_once++;
7687a054
JH
34
35 memset(opt, 0, sizeof(*opt));
36 opt->relative = 1;
37 opt->pathname = 1;
38 opt->regflags = REG_NEWLINE;
39 opt->max_depth = -1;
40 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
7ce4fb94
JK
41 color_set(opt->color_context, "");
42 color_set(opt->color_filename, "");
43 color_set(opt->color_function, "");
44 color_set(opt->color_lineno, "");
45 color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
46 color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
47 color_set(opt->color_selected, "");
48 color_set(opt->color_sep, GIT_COLOR_CYAN);
7687a054 49 opt->color = -1;
379642bc 50 opt->output = std_output;
7687a054
JH
51}
52
53static int parse_pattern_type_arg(const char *opt, const char *arg)
54{
55 if (!strcmp(arg, "default"))
56 return GREP_PATTERN_TYPE_UNSPECIFIED;
57 else if (!strcmp(arg, "basic"))
58 return GREP_PATTERN_TYPE_BRE;
59 else if (!strcmp(arg, "extended"))
60 return GREP_PATTERN_TYPE_ERE;
61 else if (!strcmp(arg, "fixed"))
62 return GREP_PATTERN_TYPE_FIXED;
63 else if (!strcmp(arg, "perl"))
64 return GREP_PATTERN_TYPE_PCRE;
65 die("bad %s argument: %s", opt, arg);
66}
67
68/*
69 * Read the configuration file once and store it in
70 * the grep_defaults template.
71 */
72int grep_config(const char *var, const char *value, void *cb)
73{
74 struct grep_opt *opt = &grep_defaults;
75 char *color = NULL;
76
77 if (userdiff_config(var, value) < 0)
78 return -1;
79
80 if (!strcmp(var, "grep.extendedregexp")) {
c7e38551 81 opt->extended_regexp_option = git_config_bool(var, value);
7687a054
JH
82 return 0;
83 }
84
85 if (!strcmp(var, "grep.patterntype")) {
86 opt->pattern_type_option = parse_pattern_type_arg(var, value);
87 return 0;
88 }
89
90 if (!strcmp(var, "grep.linenumber")) {
91 opt->linenum = git_config_bool(var, value);
92 return 0;
93 }
94
6453f7b3
AS
95 if (!strcmp(var, "grep.fullname")) {
96 opt->relative = !git_config_bool(var, value);
97 return 0;
98 }
99
7687a054
JH
100 if (!strcmp(var, "color.grep"))
101 opt->color = git_config_colorbool(var, value);
102 else if (!strcmp(var, "color.grep.context"))
103 color = opt->color_context;
104 else if (!strcmp(var, "color.grep.filename"))
105 color = opt->color_filename;
106 else if (!strcmp(var, "color.grep.function"))
107 color = opt->color_function;
108 else if (!strcmp(var, "color.grep.linenumber"))
109 color = opt->color_lineno;
79a77109
RS
110 else if (!strcmp(var, "color.grep.matchcontext"))
111 color = opt->color_match_context;
112 else if (!strcmp(var, "color.grep.matchselected"))
113 color = opt->color_match_selected;
7687a054
JH
114 else if (!strcmp(var, "color.grep.selected"))
115 color = opt->color_selected;
116 else if (!strcmp(var, "color.grep.separator"))
117 color = opt->color_sep;
79a77109
RS
118 else if (!strcmp(var, "color.grep.match")) {
119 int rc = 0;
120 if (!value)
121 return config_error_nonbool(var);
bf1f639e
JH
122 rc |= color_parse(value, opt->color_match_context);
123 rc |= color_parse(value, opt->color_match_selected);
79a77109
RS
124 return rc;
125 }
7687a054
JH
126
127 if (color) {
128 if (!value)
129 return config_error_nonbool(var);
f6c5a296 130 return color_parse(value, color);
7687a054
JH
131 }
132 return 0;
133}
134
135/*
136 * Initialize one instance of grep_opt and copy the
137 * default values from the template we read the configuration
138 * information in an earlier call to git_config(grep_config).
139 */
140void grep_init(struct grep_opt *opt, const char *prefix)
141{
142 struct grep_opt *def = &grep_defaults;
143
144 memset(opt, 0, sizeof(*opt));
145 opt->prefix = prefix;
146 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
147 opt->pattern_tail = &opt->pattern_list;
148 opt->header_tail = &opt->header_list;
149
150 opt->color = def->color;
151 opt->extended_regexp_option = def->extended_regexp_option;
152 opt->pattern_type_option = def->pattern_type_option;
153 opt->linenum = def->linenum;
154 opt->max_depth = def->max_depth;
155 opt->pathname = def->pathname;
156 opt->regflags = def->regflags;
157 opt->relative = def->relative;
379642bc 158 opt->output = def->output;
7687a054 159
7ce4fb94
JK
160 color_set(opt->color_context, def->color_context);
161 color_set(opt->color_filename, def->color_filename);
162 color_set(opt->color_function, def->color_function);
163 color_set(opt->color_lineno, def->color_lineno);
164 color_set(opt->color_match_context, def->color_match_context);
165 color_set(opt->color_match_selected, def->color_match_selected);
166 color_set(opt->color_selected, def->color_selected);
167 color_set(opt->color_sep, def->color_sep);
7687a054 168}
07a7d656 169
8465541e 170static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
c5c31d33
JH
171{
172 switch (pattern_type) {
173 case GREP_PATTERN_TYPE_UNSPECIFIED:
174 /* fall through */
175
176 case GREP_PATTERN_TYPE_BRE:
c5c31d33
JH
177 break;
178
179 case GREP_PATTERN_TYPE_ERE:
c5c31d33
JH
180 opt->regflags |= REG_EXTENDED;
181 break;
182
183 case GREP_PATTERN_TYPE_FIXED:
184 opt->fixed = 1;
c5c31d33
JH
185 break;
186
187 case GREP_PATTERN_TYPE_PCRE:
94da9193 188#ifdef USE_LIBPCRE2
94da9193
ÆAB
189 opt->pcre2 = 1;
190#else
191 /*
192 * It's important that pcre1 always be assigned to
193 * even when there's no USE_LIBPCRE* defined. We still
194 * call the PCRE stub function, it just dies with
195 * "cannot use Perl-compatible regexes[...]".
196 */
6d4b5747 197 opt->pcre1 = 1;
94da9193 198#endif
c5c31d33
JH
199 break;
200 }
201}
202
8465541e
JH
203void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
204{
205 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
206 grep_set_pattern_type_option(pattern_type, opt);
207 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
208 grep_set_pattern_type_option(opt->pattern_type_option, opt);
209 else if (opt->extended_regexp_option)
210 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
211}
212
fc456751
RS
213static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
214 const char *origin, int no,
215 enum grep_pat_token t,
216 enum grep_header_field field)
a4d7d2c6
JH
217{
218 struct grep_pat *p = xcalloc(1, sizeof(*p));
526a858a 219 p->pattern = xmemdupz(pat, patlen);
fc456751
RS
220 p->patternlen = patlen;
221 p->origin = origin;
222 p->no = no;
223 p->token = t;
a4d7d2c6 224 p->field = field;
fc456751
RS
225 return p;
226}
227
2b3873ff
RS
228static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
229{
230 **tail = p;
231 *tail = &p->next;
a4d7d2c6 232 p->next = NULL;
526a858a
RS
233
234 switch (p->token) {
235 case GREP_PATTERN: /* atom */
236 case GREP_PATTERN_HEAD:
237 case GREP_PATTERN_BODY:
238 for (;;) {
239 struct grep_pat *new_pat;
240 size_t len = 0;
241 char *cp = p->pattern + p->patternlen, *nl = NULL;
242 while (++len <= p->patternlen) {
243 if (*(--cp) == '\n') {
244 nl = cp;
245 break;
246 }
247 }
248 if (!nl)
249 break;
250 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
251 p->no, p->token, p->field);
252 new_pat->next = p->next;
253 if (!p->next)
254 *tail = &new_pat->next;
255 p->next = new_pat;
256 *nl = '\0';
257 p->patternlen -= len;
258 }
259 break;
260 default:
261 break;
262 }
2b3873ff
RS
263}
264
fc456751
RS
265void append_header_grep_pattern(struct grep_opt *opt,
266 enum grep_header_field field, const char *pat)
267{
268 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
269 GREP_PATTERN_HEAD, field);
baa6378f
JH
270 if (field == GREP_HEADER_REFLOG)
271 opt->use_reflog_filter = 1;
2b3873ff 272 do_append_grep_pat(&opt->header_tail, p);
a4d7d2c6
JH
273}
274
83b5d2f5
JH
275void append_grep_pattern(struct grep_opt *opt, const char *pat,
276 const char *origin, int no, enum grep_pat_token t)
ed40a095
RS
277{
278 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
279}
280
281void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
282 const char *origin, int no, enum grep_pat_token t)
83b5d2f5 283{
fc456751 284 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
2b3873ff 285 do_append_grep_pat(&opt->pattern_tail, p);
83b5d2f5
JH
286}
287
5b594f45
FK
288struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
289{
290 struct grep_pat *pat;
291 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
292 *ret = *opt;
293
294 ret->pattern_list = NULL;
295 ret->pattern_tail = &ret->pattern_list;
296
297 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
298 {
299 if(pat->token == GREP_PATTERN_HEAD)
300 append_header_grep_pattern(ret, pat->field,
301 pat->pattern);
302 else
ed40a095
RS
303 append_grep_pat(ret, pat->pattern, pat->patternlen,
304 pat->origin, pat->no, pat->token);
5b594f45
FK
305 }
306
307 return ret;
308}
309
a30c148a
MK
310static NORETURN void compile_regexp_failed(const struct grep_pat *p,
311 const char *error)
312{
313 char where[1024];
314
315 if (p->no)
19bdd3e7 316 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
a30c148a 317 else if (p->origin)
19bdd3e7 318 xsnprintf(where, sizeof(where), "%s, ", p->origin);
a30c148a
MK
319 else
320 where[0] = 0;
321
322 die("%s'%s': %s", where, p->pattern, error);
323}
324
543f1c0c
ÆAB
325static int is_fixed(const char *s, size_t len)
326{
327 size_t i;
328
329 for (i = 0; i < len; i++) {
330 if (is_regex_special(s[i]))
331 return 0;
332 }
333
334 return 1;
335}
336
219e65b6
ÆAB
337static int has_null(const char *s, size_t len)
338{
339 /*
340 * regcomp cannot accept patterns with NULs so when using it
341 * we consider any pattern containing a NUL fixed.
342 */
343 if (memchr(s, 0, len))
344 return 1;
345
346 return 0;
347}
348
3485bea1 349#ifdef USE_LIBPCRE1
6d4b5747 350static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
63e7e9d8
MK
351{
352 const char *error;
353 int erroffset;
fba4f125 354 int options = PCRE_MULTILINE;
63e7e9d8 355
9d9babb8
NTND
356 if (opt->ignore_case) {
357 if (has_non_ascii(p->pattern))
6d4b5747 358 p->pcre1_tables = pcre_maketables();
63e7e9d8 359 options |= PCRE_CASELESS;
9d9babb8 360 }
18547aac
NTND
361 if (is_utf8_locale() && has_non_ascii(p->pattern))
362 options |= PCRE_UTF8;
63e7e9d8 363
6d4b5747
ÆAB
364 p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
365 p->pcre1_tables);
366 if (!p->pcre1_regexp)
63e7e9d8
MK
367 compile_regexp_failed(p, error);
368
fbaceaac 369 p->pcre1_extra_info = pcre_study(p->pcre1_regexp, PCRE_STUDY_JIT_COMPILE, &error);
6d4b5747 370 if (!p->pcre1_extra_info && error)
63e7e9d8 371 die("%s", error);
fbaceaac 372
e87de7ca 373#ifdef GIT_PCRE1_USE_JIT
fbaceaac
ÆAB
374 pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
375 if (p->pcre1_jit_on == 1) {
376 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
377 if (!p->pcre1_jit_stack)
378 die("Couldn't allocate PCRE JIT stack");
379 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
380 } else if (p->pcre1_jit_on != 0) {
381 die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
382 p->pcre1_jit_on);
383 }
384#endif
63e7e9d8
MK
385}
386
6d4b5747 387static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
63e7e9d8
MK
388 regmatch_t *match, int eflags)
389{
390 int ovector[30], ret, flags = 0;
391
392 if (eflags & REG_NOTBOL)
393 flags |= PCRE_NOTBOL;
394
e87de7ca 395#ifdef GIT_PCRE1_USE_JIT
fbaceaac
ÆAB
396 if (p->pcre1_jit_on) {
397 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
398 eol - line, 0, flags, ovector,
399 ARRAY_SIZE(ovector), p->pcre1_jit_stack);
400 } else
401#endif
402 {
403 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
404 eol - line, 0, flags, ovector,
405 ARRAY_SIZE(ovector));
406 }
407
63e7e9d8
MK
408 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
409 die("pcre_exec failed with error code %d", ret);
410 if (ret > 0) {
411 ret = 0;
412 match->rm_so = ovector[0];
413 match->rm_eo = ovector[1];
414 }
415
416 return ret;
417}
418
6d4b5747 419static void free_pcre1_regexp(struct grep_pat *p)
63e7e9d8 420{
6d4b5747 421 pcre_free(p->pcre1_regexp);
e87de7ca 422#ifdef GIT_PCRE1_USE_JIT
fbaceaac
ÆAB
423 if (p->pcre1_jit_on) {
424 pcre_free_study(p->pcre1_extra_info);
425 pcre_jit_stack_free(p->pcre1_jit_stack);
426 } else
427#endif
428 {
429 pcre_free(p->pcre1_extra_info);
430 }
6d4b5747 431 pcre_free((void *)p->pcre1_tables);
63e7e9d8 432}
3485bea1 433#else /* !USE_LIBPCRE1 */
6d4b5747 434static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
63e7e9d8
MK
435{
436 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
437}
438
6d4b5747 439static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
63e7e9d8
MK
440 regmatch_t *match, int eflags)
441{
442 return 1;
443}
444
6d4b5747 445static void free_pcre1_regexp(struct grep_pat *p)
63e7e9d8
MK
446{
447}
3485bea1 448#endif /* !USE_LIBPCRE1 */
63e7e9d8 449
94da9193
ÆAB
450#ifdef USE_LIBPCRE2
451static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
452{
453 int error;
454 PCRE2_UCHAR errbuf[256];
455 PCRE2_SIZE erroffset;
456 int options = PCRE2_MULTILINE;
457 const uint8_t *character_tables = NULL;
458 int jitret;
459
460 assert(opt->pcre2);
461
462 p->pcre2_compile_context = NULL;
463
464 if (opt->ignore_case) {
465 if (has_non_ascii(p->pattern)) {
466 character_tables = pcre2_maketables(NULL);
467 p->pcre2_compile_context = pcre2_compile_context_create(NULL);
468 pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
469 }
470 options |= PCRE2_CASELESS;
471 }
472 if (is_utf8_locale() && has_non_ascii(p->pattern))
473 options |= PCRE2_UTF;
474
475 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
476 p->patternlen, options, &error, &erroffset,
477 p->pcre2_compile_context);
478
479 if (p->pcre2_pattern) {
480 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
481 if (!p->pcre2_match_data)
482 die("Couldn't allocate PCRE2 match data");
483 } else {
484 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
485 compile_regexp_failed(p, (const char *)&errbuf);
486 }
487
488 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
489 if (p->pcre2_jit_on == 1) {
490 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
491 if (jitret)
492 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
493 p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
494 if (!p->pcre2_jit_stack)
495 die("Couldn't allocate PCRE2 JIT stack");
496 p->pcre2_match_context = pcre2_match_context_create(NULL);
674ad936 497 if (!p->pcre2_match_context)
94da9193
ÆAB
498 die("Couldn't allocate PCRE2 match context");
499 pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
500 } else if (p->pcre2_jit_on != 0) {
501 die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
502 p->pcre1_jit_on);
503 }
504}
505
506static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
507 regmatch_t *match, int eflags)
508{
509 int ret, flags = 0;
510 PCRE2_SIZE *ovector;
511 PCRE2_UCHAR errbuf[256];
512
513 if (eflags & REG_NOTBOL)
514 flags |= PCRE2_NOTBOL;
515
516 if (p->pcre2_jit_on)
517 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
518 eol - line, 0, flags, p->pcre2_match_data,
519 NULL);
520 else
521 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
522 eol - line, 0, flags, p->pcre2_match_data,
523 NULL);
524
525 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
526 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
527 die("%s failed with error code %d: %s",
528 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
529 errbuf);
530 }
531 if (ret > 0) {
532 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
533 ret = 0;
534 match->rm_so = (int)ovector[0];
535 match->rm_eo = (int)ovector[1];
536 }
537
538 return ret;
539}
540
541static void free_pcre2_pattern(struct grep_pat *p)
542{
543 pcre2_compile_context_free(p->pcre2_compile_context);
544 pcre2_code_free(p->pcre2_pattern);
545 pcre2_match_data_free(p->pcre2_match_data);
546 pcre2_jit_stack_free(p->pcre2_jit_stack);
547 pcre2_match_context_free(p->pcre2_match_context);
548}
549#else /* !USE_LIBPCRE2 */
550static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
551{
552 /*
553 * Unreachable until USE_LIBPCRE2 becomes synonymous with
554 * USE_LIBPCRE. See the sibling comment in
555 * grep_set_pattern_type_option().
556 */
557 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
558}
559
560static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
561 regmatch_t *match, int eflags)
562{
563 return 1;
564}
565
566static void free_pcre2_pattern(struct grep_pat *p)
567{
568}
569#endif /* !USE_LIBPCRE2 */
570
793dc676
NTND
571static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
572{
573 struct strbuf sb = STRBUF_INIT;
574 int err;
e0b9f8ae 575 int regflags = opt->regflags;
793dc676
NTND
576
577 basic_regex_quote_buf(&sb, p->pattern);
793dc676
NTND
578 if (opt->ignore_case)
579 regflags |= REG_ICASE;
580 err = regcomp(&p->regexp, sb.buf, regflags);
581 if (opt->debug)
582 fprintf(stderr, "fixed %s\n", sb.buf);
583 strbuf_release(&sb);
584 if (err) {
585 char errbuf[1024];
586 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
587 regfree(&p->regexp);
588 compile_regexp_failed(p, errbuf);
589 }
590}
591
83b5d2f5
JH
592static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
593{
5c1ebcca 594 int icase, ascii_only;
c822255c
RS
595 int err;
596
d7eb527d 597 p->word_regexp = opt->word_regexp;
5183bf67 598 p->ignore_case = opt->ignore_case;
5c1ebcca
NTND
599 icase = opt->regflags & REG_ICASE || p->ignore_case;
600 ascii_only = !has_non_ascii(p->pattern);
d7eb527d 601
793dc676
NTND
602 /*
603 * Even when -F (fixed) asks us to do a non-regexp search, we
604 * may not be able to correctly case-fold when -i
605 * (ignore-case) is asked (in which case, we'll synthesize a
606 * regexp to match the pattern that matches regexp special
607 * characters literally, while ignoring case differences). On
608 * the other hand, even without -F, if the pattern does not
609 * have any regexp special characters and there is no need for
610 * case-folding search, we can internally turn it into a
611 * simple string match using kws. p->fixed tells us if we
612 * want to use kws.
613 */
219e65b6
ÆAB
614 if (opt->fixed ||
615 has_null(p->pattern, p->patternlen) ||
616 is_fixed(p->pattern, p->patternlen))
793dc676 617 p->fixed = !icase || ascii_only;
9eceddee
FK
618
619 if (p->fixed) {
695f95ba 620 p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
9eceddee
FK
621 kwsincr(p->kws, p->pattern, p->patternlen);
622 kwsprep(p->kws);
c822255c 623 return;
793dc676
NTND
624 } else if (opt->fixed) {
625 /*
626 * We come here when the pattern has the non-ascii
627 * characters we cannot case-fold, and asked to
628 * ignore-case.
629 */
630 compile_fixed_regexp(p, opt);
631 return;
9eceddee 632 }
c822255c 633
94da9193
ÆAB
634 if (opt->pcre2) {
635 compile_pcre2_pattern(p, opt);
636 return;
637 }
638
6d4b5747
ÆAB
639 if (opt->pcre1) {
640 compile_pcre1_regexp(p, opt);
63e7e9d8
MK
641 return;
642 }
643
c822255c 644 err = regcomp(&p->regexp, p->pattern, opt->regflags);
83b5d2f5
JH
645 if (err) {
646 char errbuf[1024];
83b5d2f5
JH
647 regerror(err, &p->regexp, errbuf, 1024);
648 regfree(&p->regexp);
a30c148a 649 compile_regexp_failed(p, errbuf);
83b5d2f5
JH
650 }
651}
652
0ab7befa 653static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
654static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
655{
656 struct grep_pat *p;
657 struct grep_expr *x;
658
659 p = *list;
c922b01f
LT
660 if (!p)
661 return NULL;
83b5d2f5
JH
662 switch (p->token) {
663 case GREP_PATTERN: /* atom */
480c1ca6
JH
664 case GREP_PATTERN_HEAD:
665 case GREP_PATTERN_BODY:
83b5d2f5
JH
666 x = xcalloc(1, sizeof (struct grep_expr));
667 x->node = GREP_NODE_ATOM;
668 x->u.atom = p;
669 *list = p->next;
670 return x;
671 case GREP_OPEN_PAREN:
672 *list = p->next;
0ab7befa 673 x = compile_pattern_or(list);
83b5d2f5
JH
674 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
675 die("unmatched parenthesis");
676 *list = (*list)->next;
677 return x;
678 default:
679 return NULL;
680 }
681}
682
683static struct grep_expr *compile_pattern_not(struct grep_pat **list)
684{
685 struct grep_pat *p;
686 struct grep_expr *x;
687
688 p = *list;
c922b01f
LT
689 if (!p)
690 return NULL;
83b5d2f5
JH
691 switch (p->token) {
692 case GREP_NOT:
693 if (!p->next)
694 die("--not not followed by pattern expression");
695 *list = p->next;
696 x = xcalloc(1, sizeof (struct grep_expr));
697 x->node = GREP_NODE_NOT;
698 x->u.unary = compile_pattern_not(list);
699 if (!x->u.unary)
700 die("--not followed by non pattern expression");
701 return x;
702 default:
703 return compile_pattern_atom(list);
704 }
705}
706
707static struct grep_expr *compile_pattern_and(struct grep_pat **list)
708{
709 struct grep_pat *p;
710 struct grep_expr *x, *y, *z;
711
712 x = compile_pattern_not(list);
713 p = *list;
714 if (p && p->token == GREP_AND) {
715 if (!p->next)
716 die("--and not followed by pattern expression");
717 *list = p->next;
718 y = compile_pattern_and(list);
719 if (!y)
720 die("--and not followed by pattern expression");
721 z = xcalloc(1, sizeof (struct grep_expr));
722 z->node = GREP_NODE_AND;
723 z->u.binary.left = x;
724 z->u.binary.right = y;
725 return z;
726 }
727 return x;
728}
729
730static struct grep_expr *compile_pattern_or(struct grep_pat **list)
731{
732 struct grep_pat *p;
733 struct grep_expr *x, *y, *z;
734
735 x = compile_pattern_and(list);
736 p = *list;
737 if (x && p && p->token != GREP_CLOSE_PAREN) {
738 y = compile_pattern_or(list);
739 if (!y)
740 die("not a pattern expression %s", p->pattern);
741 z = xcalloc(1, sizeof (struct grep_expr));
742 z->node = GREP_NODE_OR;
743 z->u.binary.left = x;
744 z->u.binary.right = y;
745 return z;
746 }
747 return x;
748}
749
750static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
751{
752 return compile_pattern_or(list);
753}
754
17bf35a3
JH
755static void indent(int in)
756{
757 while (in-- > 0)
758 fputc(' ', stderr);
759}
760
761static void dump_grep_pat(struct grep_pat *p)
762{
763 switch (p->token) {
764 case GREP_AND: fprintf(stderr, "*and*"); break;
765 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
766 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
767 case GREP_NOT: fprintf(stderr, "*not*"); break;
768 case GREP_OR: fprintf(stderr, "*or*"); break;
769
770 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
771 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
772 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
773 }
774
775 switch (p->token) {
776 default: break;
777 case GREP_PATTERN_HEAD:
778 fprintf(stderr, "<head %d>", p->field); break;
779 case GREP_PATTERN_BODY:
780 fprintf(stderr, "<body>"); break;
781 }
782 switch (p->token) {
783 default: break;
784 case GREP_PATTERN_HEAD:
785 case GREP_PATTERN_BODY:
786 case GREP_PATTERN:
787 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
788 break;
789 }
790 fputc('\n', stderr);
791}
792
793static void dump_grep_expression_1(struct grep_expr *x, int in)
794{
795 indent(in);
796 switch (x->node) {
797 case GREP_NODE_TRUE:
798 fprintf(stderr, "true\n");
799 break;
800 case GREP_NODE_ATOM:
801 dump_grep_pat(x->u.atom);
802 break;
803 case GREP_NODE_NOT:
804 fprintf(stderr, "(not\n");
805 dump_grep_expression_1(x->u.unary, in+1);
806 indent(in);
807 fprintf(stderr, ")\n");
808 break;
809 case GREP_NODE_AND:
810 fprintf(stderr, "(and\n");
811 dump_grep_expression_1(x->u.binary.left, in+1);
812 dump_grep_expression_1(x->u.binary.right, in+1);
813 indent(in);
814 fprintf(stderr, ")\n");
815 break;
816 case GREP_NODE_OR:
817 fprintf(stderr, "(or\n");
818 dump_grep_expression_1(x->u.binary.left, in+1);
819 dump_grep_expression_1(x->u.binary.right, in+1);
820 indent(in);
821 fprintf(stderr, ")\n");
822 break;
823 }
824}
825
07a7d656 826static void dump_grep_expression(struct grep_opt *opt)
17bf35a3
JH
827{
828 struct grep_expr *x = opt->pattern_expression;
829
830 if (opt->all_match)
831 fprintf(stderr, "[all-match]\n");
832 dump_grep_expression_1(x, 0);
833 fflush(NULL);
834}
835
5aaeb733
JH
836static struct grep_expr *grep_true_expr(void)
837{
838 struct grep_expr *z = xcalloc(1, sizeof(*z));
839 z->node = GREP_NODE_TRUE;
840 return z;
841}
842
843static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
844{
845 struct grep_expr *z = xcalloc(1, sizeof(*z));
846 z->node = GREP_NODE_OR;
847 z->u.binary.left = left;
848 z->u.binary.right = right;
849 return z;
850}
851
95ce9ce2 852static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
83b5d2f5
JH
853{
854 struct grep_pat *p;
95ce9ce2 855 struct grep_expr *header_expr;
5aaeb733
JH
856 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
857 enum grep_header_field fld;
83b5d2f5 858
95ce9ce2
JH
859 if (!opt->header_list)
860 return NULL;
2385f246 861
95ce9ce2
JH
862 for (p = opt->header_list; p; p = p->next) {
863 if (p->token != GREP_PATTERN_HEAD)
ef1177d1 864 die("BUG: a non-header pattern in grep header list.");
3ce3ffb8
AP
865 if (p->field < GREP_HEADER_FIELD_MIN ||
866 GREP_HEADER_FIELD_MAX <= p->field)
ef1177d1 867 die("BUG: unknown header field %d", p->field);
95ce9ce2 868 compile_regexp(p, opt);
80235ba7 869 }
5aaeb733
JH
870
871 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
872 header_group[fld] = NULL;
873
874 for (p = opt->header_list; p; p = p->next) {
875 struct grep_expr *h;
876 struct grep_pat *pp = p;
877
878 h = compile_pattern_atom(&pp);
879 if (!h || pp != p->next)
ef1177d1 880 die("BUG: malformed header expr");
5aaeb733
JH
881 if (!header_group[p->field]) {
882 header_group[p->field] = h;
883 continue;
884 }
885 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
886 }
887
888 header_expr = NULL;
889
890 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
891 if (!header_group[fld])
892 continue;
893 if (!header_expr)
894 header_expr = grep_true_expr();
895 header_expr = grep_or_expr(header_group[fld], header_expr);
896 }
95ce9ce2
JH
897 return header_expr;
898}
899
13e4fc7e
JH
900static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
901{
902 struct grep_expr *z = x;
903
904 while (x) {
905 assert(x->node == GREP_NODE_OR);
906 if (x->u.binary.right &&
907 x->u.binary.right->node == GREP_NODE_TRUE) {
908 x->u.binary.right = y;
909 break;
910 }
911 x = x->u.binary.right;
912 }
913 return z;
914}
915
17bf35a3 916static void compile_grep_patterns_real(struct grep_opt *opt)
95ce9ce2
JH
917{
918 struct grep_pat *p;
919 struct grep_expr *header_expr = prep_header_patterns(opt);
0ab7befa 920
83b5d2f5 921 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
922 switch (p->token) {
923 case GREP_PATTERN: /* atom */
924 case GREP_PATTERN_HEAD:
925 case GREP_PATTERN_BODY:
c822255c 926 compile_regexp(p, opt);
480c1ca6
JH
927 break;
928 default:
83b5d2f5 929 opt->extended = 1;
480c1ca6
JH
930 break;
931 }
83b5d2f5
JH
932 }
933
80235ba7
JH
934 if (opt->all_match || header_expr)
935 opt->extended = 1;
17bf35a3 936 else if (!opt->extended && !opt->debug)
83b5d2f5
JH
937 return;
938
83b5d2f5 939 p = opt->pattern_list;
ba150a3f
MB
940 if (p)
941 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
942 if (p)
943 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
944
945 if (!header_expr)
946 return;
947
5aaeb733 948 if (!opt->pattern_expression)
80235ba7 949 opt->pattern_expression = header_expr;
13e4fc7e
JH
950 else if (opt->all_match)
951 opt->pattern_expression = grep_splice_or(header_expr,
952 opt->pattern_expression);
5aaeb733
JH
953 else
954 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
955 header_expr);
80235ba7 956 opt->all_match = 1;
83b5d2f5
JH
957}
958
17bf35a3
JH
959void compile_grep_patterns(struct grep_opt *opt)
960{
961 compile_grep_patterns_real(opt);
962 if (opt->debug)
963 dump_grep_expression(opt);
964}
965
b48fb5b6
JH
966static void free_pattern_expr(struct grep_expr *x)
967{
968 switch (x->node) {
5aaeb733 969 case GREP_NODE_TRUE:
b48fb5b6
JH
970 case GREP_NODE_ATOM:
971 break;
972 case GREP_NODE_NOT:
973 free_pattern_expr(x->u.unary);
974 break;
975 case GREP_NODE_AND:
976 case GREP_NODE_OR:
977 free_pattern_expr(x->u.binary.left);
978 free_pattern_expr(x->u.binary.right);
979 break;
980 }
981 free(x);
982}
983
984void free_grep_patterns(struct grep_opt *opt)
985{
986 struct grep_pat *p, *n;
987
988 for (p = opt->pattern_list; p; p = n) {
989 n = p->next;
990 switch (p->token) {
991 case GREP_PATTERN: /* atom */
992 case GREP_PATTERN_HEAD:
993 case GREP_PATTERN_BODY:
9eceddee
FK
994 if (p->kws)
995 kwsfree(p->kws);
6d4b5747
ÆAB
996 else if (p->pcre1_regexp)
997 free_pcre1_regexp(p);
94da9193
ÆAB
998 else if (p->pcre2_pattern)
999 free_pcre2_pattern(p);
63e7e9d8
MK
1000 else
1001 regfree(&p->regexp);
526a858a 1002 free(p->pattern);
b48fb5b6
JH
1003 break;
1004 default:
1005 break;
1006 }
1007 free(p);
1008 }
1009
1010 if (!opt->extended)
1011 return;
1012 free_pattern_expr(opt->pattern_expression);
1013}
1014
83b5d2f5
JH
1015static char *end_of_line(char *cp, unsigned long *left)
1016{
1017 unsigned long l = *left;
1018 while (l && *cp != '\n') {
1019 l--;
1020 cp++;
1021 }
1022 *left = l;
1023 return cp;
1024}
1025
1026static int word_char(char ch)
1027{
1028 return isalnum(ch) || ch == '_';
1029}
1030
55f638bd
ML
1031static void output_color(struct grep_opt *opt, const void *data, size_t size,
1032 const char *color)
1033{
daa0c3d9 1034 if (want_color(opt->color) && color && color[0]) {
55f638bd
ML
1035 opt->output(opt, color, strlen(color));
1036 opt->output(opt, data, size);
1037 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1038 } else
1039 opt->output(opt, data, size);
1040}
1041
1042static void output_sep(struct grep_opt *opt, char sign)
1043{
1044 if (opt->null_following_name)
1045 opt->output(opt, "\0", 1);
1046 else
1047 output_color(opt, &sign, 1, opt->color_sep);
1048}
1049
83caecca
RZ
1050static void show_name(struct grep_opt *opt, const char *name)
1051{
55f638bd 1052 output_color(opt, name, strlen(name), opt->color_filename);
5b594f45 1053 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
1054}
1055
ed40a095
RS
1056static int fixmatch(struct grep_pat *p, char *line, char *eol,
1057 regmatch_t *match)
83b5d2f5 1058{
9eceddee
FK
1059 struct kwsmatch kwsm;
1060 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1061 if (offset == -1) {
83b5d2f5
JH
1062 match->rm_so = match->rm_eo = -1;
1063 return REG_NOMATCH;
9eceddee
FK
1064 } else {
1065 match->rm_so = offset;
1066 match->rm_eo = match->rm_so + kwsm.size[0];
83b5d2f5
JH
1067 return 0;
1068 }
1069}
1070
97e77784
MK
1071static int patmatch(struct grep_pat *p, char *line, char *eol,
1072 regmatch_t *match, int eflags)
1073{
1074 int hit;
1075
1076 if (p->fixed)
1077 hit = !fixmatch(p, line, eol, match);
6d4b5747
ÆAB
1078 else if (p->pcre1_regexp)
1079 hit = !pcre1match(p, line, eol, match, eflags);
94da9193
ÆAB
1080 else if (p->pcre2_pattern)
1081 hit = !pcre2match(p, line, eol, match, eflags);
97e77784 1082 else
b7d36ffc
JS
1083 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1084 eflags);
97e77784
MK
1085
1086 return hit;
1087}
1088
a4d7d2c6
JH
1089static int strip_timestamp(char *bol, char **eol_p)
1090{
1091 char *eol = *eol_p;
1092 int ch;
1093
1094 while (bol < --eol) {
1095 if (*eol != '>')
1096 continue;
1097 *eol_p = ++eol;
1098 ch = *eol;
1099 *eol = '\0';
1100 return ch;
1101 }
1102 return 0;
1103}
1104
1105static struct {
1106 const char *field;
1107 size_t len;
1108} header_field[] = {
1109 { "author ", 7 },
1110 { "committer ", 10 },
72fd13f7 1111 { "reflog ", 7 },
a4d7d2c6
JH
1112};
1113
d7eb527d 1114static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
1115 enum grep_context ctx,
1116 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
1117{
1118 int hit = 0;
a4d7d2c6 1119 int saved_ch = 0;
e701fadb 1120 const char *start = bol;
83b5d2f5 1121
480c1ca6
JH
1122 if ((p->token != GREP_PATTERN) &&
1123 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1124 return 0;
1125
a4d7d2c6
JH
1126 if (p->token == GREP_PATTERN_HEAD) {
1127 const char *field;
1128 size_t len;
1129 assert(p->field < ARRAY_SIZE(header_field));
1130 field = header_field[p->field].field;
1131 len = header_field[p->field].len;
1132 if (strncmp(bol, field, len))
1133 return 0;
1134 bol += len;
ad4813b3
NTND
1135 switch (p->field) {
1136 case GREP_HEADER_AUTHOR:
1137 case GREP_HEADER_COMMITTER:
1138 saved_ch = strip_timestamp(bol, &eol);
1139 break;
1140 default:
1141 break;
1142 }
a4d7d2c6
JH
1143 }
1144
83b5d2f5 1145 again:
97e77784 1146 hit = patmatch(p, bol, eol, pmatch, eflags);
83b5d2f5 1147
d7eb527d 1148 if (hit && p->word_regexp) {
83b5d2f5 1149 if ((pmatch[0].rm_so < 0) ||
84201eae 1150 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
1151 (pmatch[0].rm_eo < 0) ||
1152 (eol - bol) < pmatch[0].rm_eo)
1153 die("regexp returned nonsense");
1154
1155 /* Match beginning must be either beginning of the
1156 * line, or at word boundary (i.e. the last char must
1157 * not be a word char). Similarly, match end must be
1158 * either end of the line, or at word boundary
1159 * (i.e. the next char must not be a word char).
1160 */
fb62eb7f 1161 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
1162 !word_char(bol[pmatch[0].rm_so-1])) &&
1163 ((pmatch[0].rm_eo == (eol-bol)) ||
1164 !word_char(bol[pmatch[0].rm_eo])) )
1165 ;
1166 else
1167 hit = 0;
1168
84201eae
RS
1169 /* Words consist of at least one character. */
1170 if (pmatch->rm_so == pmatch->rm_eo)
1171 hit = 0;
1172
83b5d2f5
JH
1173 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1174 /* There could be more than one match on the
1175 * line, and the first match might not be
1176 * strict word match. But later ones could be!
fb62eb7f
RS
1177 * Forward to the next possible start, i.e. the
1178 * next position following a non-word char.
83b5d2f5
JH
1179 */
1180 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
1181 while (word_char(bol[-1]) && bol < eol)
1182 bol++;
dbb6a4ad 1183 eflags |= REG_NOTBOL;
fb62eb7f
RS
1184 if (bol < eol)
1185 goto again;
83b5d2f5
JH
1186 }
1187 }
a4d7d2c6
JH
1188 if (p->token == GREP_PATTERN_HEAD && saved_ch)
1189 *eol = saved_ch;
e701fadb
RS
1190 if (hit) {
1191 pmatch[0].rm_so += bol - start;
1192 pmatch[0].rm_eo += bol - start;
1193 }
83b5d2f5
JH
1194 return hit;
1195}
1196
d7eb527d
RS
1197static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1198 enum grep_context ctx, int collect_hits)
83b5d2f5 1199{
0ab7befa 1200 int h = 0;
79212772 1201 regmatch_t match;
0ab7befa 1202
c922b01f
LT
1203 if (!x)
1204 die("Not a valid grep expression");
83b5d2f5 1205 switch (x->node) {
5aaeb733
JH
1206 case GREP_NODE_TRUE:
1207 h = 1;
1208 break;
83b5d2f5 1209 case GREP_NODE_ATOM:
79212772 1210 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
83b5d2f5
JH
1211 break;
1212 case GREP_NODE_NOT:
d7eb527d 1213 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
0ab7befa 1214 break;
83b5d2f5 1215 case GREP_NODE_AND:
d7eb527d 1216 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
252d560d 1217 return 0;
d7eb527d 1218 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
0ab7befa 1219 break;
83b5d2f5 1220 case GREP_NODE_OR:
0ab7befa 1221 if (!collect_hits)
d7eb527d 1222 return (match_expr_eval(x->u.binary.left,
0ab7befa 1223 bol, eol, ctx, 0) ||
d7eb527d 1224 match_expr_eval(x->u.binary.right,
0ab7befa 1225 bol, eol, ctx, 0));
d7eb527d 1226 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
0ab7befa 1227 x->u.binary.left->hit |= h;
d7eb527d 1228 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
0ab7befa
JH
1229 break;
1230 default:
d7530708 1231 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 1232 }
0ab7befa
JH
1233 if (collect_hits)
1234 x->hit |= h;
1235 return h;
83b5d2f5
JH
1236}
1237
480c1ca6 1238static int match_expr(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1239 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1240{
1241 struct grep_expr *x = opt->pattern_expression;
d7eb527d 1242 return match_expr_eval(x, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1243}
1244
480c1ca6 1245static int match_line(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1246 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1247{
1248 struct grep_pat *p;
79212772
RS
1249 regmatch_t match;
1250
83b5d2f5 1251 if (opt->extended)
0ab7befa
JH
1252 return match_expr(opt, bol, eol, ctx, collect_hits);
1253
1254 /* we do not call with collect_hits without being extended */
83b5d2f5 1255 for (p = opt->pattern_list; p; p = p->next) {
79212772 1256 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
83b5d2f5
JH
1257 return 1;
1258 }
1259 return 0;
1260}
1261
7e8f59d5
RS
1262static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1263 enum grep_context ctx,
1264 regmatch_t *pmatch, int eflags)
1265{
1266 regmatch_t match;
1267
1268 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1269 return 0;
1270 if (match.rm_so < 0 || match.rm_eo < 0)
1271 return 0;
1272 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1273 if (match.rm_so > pmatch->rm_so)
1274 return 1;
1275 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1276 return 1;
1277 }
1278 pmatch->rm_so = match.rm_so;
1279 pmatch->rm_eo = match.rm_eo;
1280 return 1;
1281}
1282
1283static int next_match(struct grep_opt *opt, char *bol, char *eol,
1284 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1285{
1286 struct grep_pat *p;
1287 int hit = 0;
1288
1289 pmatch->rm_so = pmatch->rm_eo = -1;
1290 if (bol < eol) {
1291 for (p = opt->pattern_list; p; p = p->next) {
1292 switch (p->token) {
1293 case GREP_PATTERN: /* atom */
1294 case GREP_PATTERN_HEAD:
1295 case GREP_PATTERN_BODY:
1296 hit |= match_next_pattern(p, bol, eol, ctx,
1297 pmatch, eflags);
1298 break;
1299 default:
1300 break;
1301 }
1302 }
1303 }
1304 return hit;
1305}
1306
1307static void show_line(struct grep_opt *opt, char *bol, char *eol,
1308 const char *name, unsigned lno, char sign)
1309{
1310 int rest = eol - bol;
79a77109 1311 const char *match_color, *line_color = NULL;
7e8f59d5 1312
a8f0e764
RS
1313 if (opt->file_break && opt->last_shown == 0) {
1314 if (opt->show_hunk_mark)
1315 opt->output(opt, "\n", 1);
ba8ea749 1316 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
046802d0 1317 if (opt->last_shown == 0) {
55f638bd
ML
1318 if (opt->show_hunk_mark) {
1319 output_color(opt, "--", 2, opt->color_sep);
1320 opt->output(opt, "\n", 1);
07b838f0 1321 }
55f638bd
ML
1322 } else if (lno > opt->last_shown + 1) {
1323 output_color(opt, "--", 2, opt->color_sep);
1324 opt->output(opt, "\n", 1);
1325 }
5dd06d38 1326 }
1d84f72e
RS
1327 if (opt->heading && opt->last_shown == 0) {
1328 output_color(opt, name, strlen(name), opt->color_filename);
1329 opt->output(opt, "\n", 1);
1330 }
5dd06d38
RS
1331 opt->last_shown = lno;
1332
1d84f72e 1333 if (!opt->heading && opt->pathname) {
55f638bd
ML
1334 output_color(opt, name, strlen(name), opt->color_filename);
1335 output_sep(opt, sign);
5b594f45
FK
1336 }
1337 if (opt->linenum) {
1338 char buf[32];
1a168e5c 1339 xsnprintf(buf, sizeof(buf), "%d", lno);
55f638bd
ML
1340 output_color(opt, buf, strlen(buf), opt->color_lineno);
1341 output_sep(opt, sign);
5b594f45 1342 }
7e8f59d5
RS
1343 if (opt->color) {
1344 regmatch_t match;
1345 enum grep_context ctx = GREP_CONTEXT_BODY;
1346 int ch = *eol;
1347 int eflags = 0;
1348
79a77109
RS
1349 if (sign == ':')
1350 match_color = opt->color_match_selected;
1351 else
1352 match_color = opt->color_match_context;
00588bb5
ML
1353 if (sign == ':')
1354 line_color = opt->color_selected;
1355 else if (sign == '-')
1356 line_color = opt->color_context;
1357 else if (sign == '=')
1358 line_color = opt->color_function;
7e8f59d5
RS
1359 *eol = '\0';
1360 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
1361 if (match.rm_so == match.rm_eo)
1362 break;
5b594f45 1363
00588bb5 1364 output_color(opt, bol, match.rm_so, line_color);
55f638bd 1365 output_color(opt, bol + match.rm_so,
79a77109 1366 match.rm_eo - match.rm_so, match_color);
7e8f59d5
RS
1367 bol += match.rm_eo;
1368 rest -= match.rm_eo;
1369 eflags = REG_NOTBOL;
1370 }
1371 *eol = ch;
1372 }
00588bb5 1373 output_color(opt, bol, rest, line_color);
5b594f45 1374 opt->output(opt, "\n", 1);
7e8f59d5
RS
1375}
1376
0579f91d 1377#ifndef NO_PTHREADS
78db6ea9
JK
1378int grep_use_locks;
1379
0579f91d
TR
1380/*
1381 * This lock protects access to the gitattributes machinery, which is
1382 * not thread-safe.
1383 */
1384pthread_mutex_t grep_attr_mutex;
1385
78db6ea9 1386static inline void grep_attr_lock(void)
0579f91d 1387{
78db6ea9 1388 if (grep_use_locks)
0579f91d
TR
1389 pthread_mutex_lock(&grep_attr_mutex);
1390}
1391
78db6ea9 1392static inline void grep_attr_unlock(void)
0579f91d 1393{
78db6ea9 1394 if (grep_use_locks)
0579f91d
TR
1395 pthread_mutex_unlock(&grep_attr_mutex);
1396}
b3aeb285
JK
1397
1398/*
1399 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1400 */
1401pthread_mutex_t grep_read_mutex;
1402
0579f91d 1403#else
78db6ea9
JK
1404#define grep_attr_lock()
1405#define grep_attr_unlock()
0579f91d
TR
1406#endif
1407
e1327023 1408static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
2944e4e6 1409{
60ecac98 1410 xdemitconf_t *xecfg = opt->priv;
0579f91d 1411 if (xecfg && !xecfg->find_func) {
94ad9d9e
JK
1412 grep_source_load_driver(gs);
1413 if (gs->driver->funcname.pattern) {
1414 const struct userdiff_funcname *pe = &gs->driver->funcname;
0579f91d
TR
1415 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1416 } else {
1417 xecfg = opt->priv = NULL;
1418 }
1419 }
1420
1421 if (xecfg) {
60ecac98
RS
1422 char buf[1];
1423 return xecfg->find_func(bol, eol - bol, buf, 1,
1424 xecfg->find_func_priv) >= 0;
1425 }
1426
2944e4e6
RS
1427 if (bol == eol)
1428 return 0;
1429 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1430 return 1;
1431 return 0;
1432}
1433
e1327023
JK
1434static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1435 char *bol, unsigned lno)
2944e4e6 1436{
e1327023 1437 while (bol > gs->buf) {
2944e4e6
RS
1438 char *eol = --bol;
1439
e1327023 1440 while (bol > gs->buf && bol[-1] != '\n')
2944e4e6
RS
1441 bol--;
1442 lno--;
1443
1444 if (lno <= opt->last_shown)
1445 break;
1446
e1327023
JK
1447 if (match_funcname(opt, gs, bol, eol)) {
1448 show_line(opt, bol, eol, gs->name, lno, '=');
2944e4e6
RS
1449 break;
1450 }
1451 }
1452}
1453
e1327023 1454static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
ba8ea749 1455 char *bol, char *end, unsigned lno)
49de3216 1456{
2944e4e6 1457 unsigned cur = lno, from = 1, funcname_lno = 0;
ba8ea749
RS
1458 int funcname_needed = !!opt->funcname;
1459
e1327023 1460 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
ba8ea749 1461 funcname_needed = 2;
49de3216
RS
1462
1463 if (opt->pre_context < lno)
1464 from = lno - opt->pre_context;
1465 if (from <= opt->last_shown)
1466 from = opt->last_shown + 1;
1467
1468 /* Rewind. */
e1327023 1469 while (bol > gs->buf &&
ba8ea749 1470 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
2944e4e6
RS
1471 char *eol = --bol;
1472
e1327023 1473 while (bol > gs->buf && bol[-1] != '\n')
49de3216
RS
1474 bol--;
1475 cur--;
e1327023 1476 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
2944e4e6
RS
1477 funcname_lno = cur;
1478 funcname_needed = 0;
1479 }
49de3216
RS
1480 }
1481
2944e4e6
RS
1482 /* We need to look even further back to find a function signature. */
1483 if (opt->funcname && funcname_needed)
e1327023 1484 show_funcname_line(opt, gs, bol, cur);
2944e4e6 1485
49de3216
RS
1486 /* Back forward. */
1487 while (cur < lno) {
2944e4e6 1488 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
1489
1490 while (*eol != '\n')
1491 eol++;
e1327023 1492 show_line(opt, bol, eol, gs->name, cur, sign);
49de3216
RS
1493 bol = eol + 1;
1494 cur++;
1495 }
1496}
1497
a26345b6
JH
1498static int should_lookahead(struct grep_opt *opt)
1499{
1500 struct grep_pat *p;
1501
1502 if (opt->extended)
1503 return 0; /* punt for too complex stuff */
1504 if (opt->invert)
1505 return 0;
1506 for (p = opt->pattern_list; p; p = p->next) {
1507 if (p->token != GREP_PATTERN)
1508 return 0; /* punt for "header only" and stuff */
1509 }
1510 return 1;
1511}
1512
1513static int look_ahead(struct grep_opt *opt,
1514 unsigned long *left_p,
1515 unsigned *lno_p,
1516 char **bol_p)
1517{
1518 unsigned lno = *lno_p;
1519 char *bol = *bol_p;
1520 struct grep_pat *p;
1521 char *sp, *last_bol;
1522 regoff_t earliest = -1;
1523
1524 for (p = opt->pattern_list; p; p = p->next) {
1525 int hit;
1526 regmatch_t m;
1527
97e77784 1528 hit = patmatch(p, bol, bol + *left_p, &m, 0);
a26345b6
JH
1529 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1530 continue;
1531 if (earliest < 0 || m.rm_so < earliest)
1532 earliest = m.rm_so;
1533 }
1534
1535 if (earliest < 0) {
1536 *bol_p = bol + *left_p;
1537 *left_p = 0;
1538 return 1;
1539 }
1540 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1541 ; /* find the beginning of the line */
1542 last_bol = sp;
1543
1544 for (sp = bol; sp < last_bol; sp++) {
1545 if (*sp == '\n')
1546 lno++;
1547 }
1548 *left_p -= last_bol - bol;
1549 *bol_p = last_bol;
1550 *lno_p = lno;
1551 return 0;
1552}
1553
335ec3bf
JK
1554static int fill_textconv_grep(struct userdiff_driver *driver,
1555 struct grep_source *gs)
1556{
1557 struct diff_filespec *df;
1558 char *buf;
1559 size_t size;
1560
1561 if (!driver || !driver->textconv)
1562 return grep_source_load(gs);
1563
1564 /*
1565 * The textconv interface is intimately tied to diff_filespecs, so we
1566 * have to pretend to be one. If we could unify the grep_source
1567 * and diff_filespec structs, this mess could just go away.
1568 */
1569 df = alloc_filespec(gs->path);
1570 switch (gs->type) {
1c41c82b 1571 case GREP_SOURCE_OID:
335ec3bf
JK
1572 fill_filespec(df, gs->identifier, 1, 0100644);
1573 break;
1574 case GREP_SOURCE_FILE:
f9704c2d 1575 fill_filespec(df, &null_oid, 0, 0100644);
335ec3bf
JK
1576 break;
1577 default:
1578 die("BUG: attempt to textconv something without a path?");
1579 }
1580
1581 /*
1582 * fill_textconv is not remotely thread-safe; it may load objects
1583 * behind the scenes, and it modifies the global diff tempfile
1584 * structure.
1585 */
1586 grep_read_lock();
1587 size = fill_textconv(driver, df, &buf);
1588 grep_read_unlock();
1589 free_filespec(df);
1590
1591 /*
1592 * The normal fill_textconv usage by the diff machinery would just keep
1593 * the textconv'd buf separate from the diff_filespec. But much of the
1594 * grep code passes around a grep_source and assumes that its "buf"
1595 * pointer is the beginning of the thing we are searching. So let's
1596 * install our textconv'd version into the grep_source, taking care not
1597 * to leak any existing buffer.
1598 */
1599 grep_source_clear_data(gs);
1600 gs->buf = buf;
1601 gs->size = size;
1602
1603 return 0;
1604}
1605
4aa2c475
RS
1606static int is_empty_line(const char *bol, const char *eol)
1607{
1608 while (bol < eol && isspace(*bol))
1609 bol++;
1610 return bol == eol;
1611}
1612
e1327023 1613static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
83b5d2f5 1614{
e1327023 1615 char *bol;
4aa2c475 1616 char *peek_bol = NULL;
e1327023 1617 unsigned long left;
83b5d2f5 1618 unsigned lno = 1;
83b5d2f5 1619 unsigned last_hit = 0;
83b5d2f5 1620 int binary_match_only = 0;
83b5d2f5 1621 unsigned count = 0;
a26345b6 1622 int try_lookahead = 0;
ba8ea749 1623 int show_function = 0;
335ec3bf 1624 struct userdiff_driver *textconv = NULL;
480c1ca6 1625 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 1626 xdemitconf_t xecfg;
83b5d2f5 1627
5b594f45
FK
1628 if (!opt->output)
1629 opt->output = std_output;
1630
ba8ea749
RS
1631 if (opt->pre_context || opt->post_context || opt->file_break ||
1632 opt->funcbody) {
08303c36
RS
1633 /* Show hunk marks, except for the first file. */
1634 if (opt->last_shown)
1635 opt->show_hunk_mark = 1;
1636 /*
1637 * If we're using threads then we can't easily identify
1638 * the first file. Always put hunk marks in that case
1639 * and skip the very first one later in work_done().
1640 */
1641 if (opt->output != std_output)
1642 opt->show_hunk_mark = 1;
1643 }
431d6e7b
RS
1644 opt->last_shown = 0;
1645
335ec3bf
JK
1646 if (opt->allow_textconv) {
1647 grep_source_load_driver(gs);
1648 /*
1649 * We might set up the shared textconv cache data here, which
1650 * is not thread-safe.
1651 */
1652 grep_attr_lock();
1653 textconv = userdiff_get_textconv(gs->driver);
1654 grep_attr_unlock();
1655 }
1656
1657 /*
1658 * We know the result of a textconv is text, so we only have to care
1659 * about binary handling if we are not using it.
1660 */
1661 if (!textconv) {
1662 switch (opt->binary) {
1663 case GREP_BINARY_DEFAULT:
1664 if (grep_source_is_binary(gs))
1665 binary_match_only = 1;
1666 break;
1667 case GREP_BINARY_NOMATCH:
1668 if (grep_source_is_binary(gs))
1669 return 0; /* Assume unmatch */
1670 break;
1671 case GREP_BINARY_TEXT:
1672 break;
1673 default:
ef1177d1 1674 die("BUG: unknown binary handling mode");
335ec3bf 1675 }
83b5d2f5
JH
1676 }
1677
60ecac98 1678 memset(&xecfg, 0, sizeof(xecfg));
0579f91d
TR
1679 opt->priv = &xecfg;
1680
a26345b6 1681 try_lookahead = should_lookahead(opt);
60ecac98 1682
335ec3bf 1683 if (fill_textconv_grep(textconv, gs) < 0)
08265798
JK
1684 return 0;
1685
e1327023
JK
1686 bol = gs->buf;
1687 left = gs->size;
83b5d2f5
JH
1688 while (left) {
1689 char *eol, ch;
0ab7befa 1690 int hit;
83b5d2f5 1691
a26345b6 1692 /*
8997da38 1693 * look_ahead() skips quickly to the line that possibly
a26345b6
JH
1694 * has the next hit; don't call it if we need to do
1695 * something more than just skipping the current line
1696 * in response to an unmatch for the current line. E.g.
1697 * inside a post-context window, we will show the current
1698 * line as a context around the previous hit when it
1699 * doesn't hit.
1700 */
1701 if (try_lookahead
1702 && !(last_hit
ba8ea749
RS
1703 && (show_function ||
1704 lno <= last_hit + opt->post_context))
a26345b6
JH
1705 && look_ahead(opt, &left, &lno, &bol))
1706 break;
83b5d2f5
JH
1707 eol = end_of_line(bol, &left);
1708 ch = *eol;
1709 *eol = 0;
1710
480c1ca6
JH
1711 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1712 ctx = GREP_CONTEXT_BODY;
1713
0ab7befa 1714 hit = match_line(opt, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1715 *eol = ch;
1716
0ab7befa
JH
1717 if (collect_hits)
1718 goto next_line;
1719
83b5d2f5
JH
1720 /* "grep -v -e foo -e bla" should list lines
1721 * that do not have either, so inversion should
1722 * be done outside.
1723 */
1724 if (opt->invert)
1725 hit = !hit;
1726 if (opt->unmatch_name_only) {
1727 if (hit)
1728 return 0;
1729 goto next_line;
1730 }
1731 if (hit) {
1732 count++;
1733 if (opt->status_only)
1734 return 1;
321ffcc0 1735 if (opt->name_only) {
e1327023 1736 show_name(opt, gs->name);
321ffcc0
RS
1737 return 1;
1738 }
c30c10cf
RS
1739 if (opt->count)
1740 goto next_line;
83b5d2f5 1741 if (binary_match_only) {
5b594f45 1742 opt->output(opt, "Binary file ", 12);
e1327023 1743 output_color(opt, gs->name, strlen(gs->name),
55f638bd 1744 opt->color_filename);
5b594f45 1745 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
1746 return 1;
1747 }
83b5d2f5
JH
1748 /* Hit at this line. If we haven't shown the
1749 * pre-context lines, we would need to show them.
83b5d2f5 1750 */
ba8ea749 1751 if (opt->pre_context || opt->funcbody)
e1327023 1752 show_pre_context(opt, gs, bol, eol, lno);
2944e4e6 1753 else if (opt->funcname)
e1327023
JK
1754 show_funcname_line(opt, gs, bol, lno);
1755 show_line(opt, bol, eol, gs->name, lno, ':');
5dd06d38 1756 last_hit = lno;
ba8ea749
RS
1757 if (opt->funcbody)
1758 show_function = 1;
1759 goto next_line;
83b5d2f5 1760 }
4aa2c475
RS
1761 if (show_function && (!peek_bol || peek_bol < bol)) {
1762 unsigned long peek_left = left;
1763 char *peek_eol = eol;
1764
1765 /*
1766 * Trailing empty lines are not interesting.
1767 * Peek past them to see if they belong to the
1768 * body of the current function.
1769 */
1770 peek_bol = bol;
1771 while (is_empty_line(peek_bol, peek_eol)) {
1772 peek_bol = peek_eol + 1;
1773 peek_eol = end_of_line(peek_bol, &peek_left);
1774 }
1775
1776 if (match_funcname(opt, gs, peek_bol, peek_eol))
1777 show_function = 0;
1778 }
ba8ea749
RS
1779 if (show_function ||
1780 (last_hit && lno <= last_hit + opt->post_context)) {
83b5d2f5
JH
1781 /* If the last hit is within the post context,
1782 * we need to show this line.
1783 */
e1327023 1784 show_line(opt, bol, eol, gs->name, lno, '-');
83b5d2f5 1785 }
83b5d2f5
JH
1786
1787 next_line:
1788 bol = eol + 1;
1789 if (!left)
1790 break;
1791 left--;
1792 lno++;
1793 }
1794
0ab7befa
JH
1795 if (collect_hits)
1796 return 0;
b48fb5b6 1797
83b5d2f5
JH
1798 if (opt->status_only)
1799 return 0;
1800 if (opt->unmatch_name_only) {
1801 /* We did not see any hit, so we want to show this */
e1327023 1802 show_name(opt, gs->name);
83b5d2f5
JH
1803 return 1;
1804 }
1805
60ecac98
RS
1806 xdiff_clear_find_func(&xecfg);
1807 opt->priv = NULL;
1808
83b5d2f5
JH
1809 /* NEEDSWORK:
1810 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1811 * which feels mostly useless but sometimes useful. Maybe
1812 * make it another option? For now suppress them.
1813 */
5b594f45
FK
1814 if (opt->count && count) {
1815 char buf[32];
f76d947a
RS
1816 if (opt->pathname) {
1817 output_color(opt, gs->name, strlen(gs->name),
1818 opt->color_filename);
1819 output_sep(opt, ':');
1820 }
1a168e5c 1821 xsnprintf(buf, sizeof(buf), "%u\n", count);
5b594f45 1822 opt->output(opt, buf, strlen(buf));
c30c10cf 1823 return 1;
5b594f45 1824 }
83b5d2f5
JH
1825 return !!last_hit;
1826}
1827
0ab7befa
JH
1828static void clr_hit_marker(struct grep_expr *x)
1829{
1830 /* All-hit markers are meaningful only at the very top level
1831 * OR node.
1832 */
1833 while (1) {
1834 x->hit = 0;
1835 if (x->node != GREP_NODE_OR)
1836 return;
1837 x->u.binary.left->hit = 0;
1838 x = x->u.binary.right;
1839 }
1840}
1841
1842static int chk_hit_marker(struct grep_expr *x)
1843{
1844 /* Top level nodes have hit markers. See if they all are hits */
1845 while (1) {
1846 if (x->node != GREP_NODE_OR)
1847 return x->hit;
1848 if (!x->u.binary.left->hit)
1849 return 0;
1850 x = x->u.binary.right;
1851 }
1852}
1853
e1327023 1854int grep_source(struct grep_opt *opt, struct grep_source *gs)
0ab7befa
JH
1855{
1856 /*
1857 * we do not have to do the two-pass grep when we do not check
1858 * buffer-wide "all-match".
1859 */
1860 if (!opt->all_match)
e1327023 1861 return grep_source_1(opt, gs, 0);
0ab7befa
JH
1862
1863 /* Otherwise the toplevel "or" terms hit a bit differently.
1864 * We first clear hit markers from them.
1865 */
1866 clr_hit_marker(opt->pattern_expression);
e1327023 1867 grep_source_1(opt, gs, 1);
0ab7befa
JH
1868
1869 if (!chk_hit_marker(opt->pattern_expression))
1870 return 0;
1871
e1327023
JK
1872 return grep_source_1(opt, gs, 0);
1873}
1874
c876d6da 1875int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
e1327023
JK
1876{
1877 struct grep_source gs;
1878 int r;
1879
55c61688 1880 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
e1327023
JK
1881 gs.buf = buf;
1882 gs.size = size;
1883
1884 r = grep_source(opt, &gs);
1885
1886 grep_source_clear(&gs);
1887 return r;
1888}
1889
1890void grep_source_init(struct grep_source *gs, enum grep_source_type type,
55c61688
NTND
1891 const char *name, const char *path,
1892 const void *identifier)
e1327023
JK
1893{
1894 gs->type = type;
8c53f071
JK
1895 gs->name = xstrdup_or_null(name);
1896 gs->path = xstrdup_or_null(path);
e1327023
JK
1897 gs->buf = NULL;
1898 gs->size = 0;
94ad9d9e 1899 gs->driver = NULL;
e1327023
JK
1900
1901 switch (type) {
1902 case GREP_SOURCE_FILE:
1903 gs->identifier = xstrdup(identifier);
1904 break;
4538eef5
BW
1905 case GREP_SOURCE_SUBMODULE:
1906 if (!identifier) {
1907 gs->identifier = NULL;
1908 break;
1909 }
1910 /*
1911 * FALL THROUGH
1912 * If the identifier is non-NULL (in the submodule case) it
1913 * will be a SHA1 that needs to be copied.
1914 */
1c41c82b
BW
1915 case GREP_SOURCE_OID:
1916 gs->identifier = oiddup(identifier);
e1327023
JK
1917 break;
1918 case GREP_SOURCE_BUF:
1919 gs->identifier = NULL;
4538eef5 1920 break;
e1327023
JK
1921 }
1922}
1923
1924void grep_source_clear(struct grep_source *gs)
1925{
88ce3ef6
ÆAB
1926 FREE_AND_NULL(gs->name);
1927 FREE_AND_NULL(gs->path);
1928 FREE_AND_NULL(gs->identifier);
e1327023
JK
1929 grep_source_clear_data(gs);
1930}
1931
1932void grep_source_clear_data(struct grep_source *gs)
1933{
1934 switch (gs->type) {
1935 case GREP_SOURCE_FILE:
1c41c82b 1936 case GREP_SOURCE_OID:
4538eef5 1937 case GREP_SOURCE_SUBMODULE:
6a83d902 1938 FREE_AND_NULL(gs->buf);
e1327023
JK
1939 gs->size = 0;
1940 break;
1941 case GREP_SOURCE_BUF:
1942 /* leave user-provided buf intact */
1943 break;
1944 }
1945}
1946
1c41c82b 1947static int grep_source_load_oid(struct grep_source *gs)
e1327023
JK
1948{
1949 enum object_type type;
1950
1951 grep_read_lock();
1952 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1953 grep_read_unlock();
1954
1955 if (!gs->buf)
1956 return error(_("'%s': unable to read %s"),
1957 gs->name,
1c41c82b 1958 oid_to_hex(gs->identifier));
e1327023
JK
1959 return 0;
1960}
1961
1962static int grep_source_load_file(struct grep_source *gs)
1963{
1964 const char *filename = gs->identifier;
1965 struct stat st;
1966 char *data;
1967 size_t size;
1968 int i;
1969
1970 if (lstat(filename, &st) < 0) {
1971 err_ret:
1972 if (errno != ENOENT)
7645d8f1 1973 error_errno(_("failed to stat '%s'"), filename);
e1327023
JK
1974 return -1;
1975 }
1976 if (!S_ISREG(st.st_mode))
1977 return -1;
1978 size = xsize_t(st.st_size);
1979 i = open(filename, O_RDONLY);
1980 if (i < 0)
1981 goto err_ret;
3733e694 1982 data = xmallocz(size);
e1327023 1983 if (st.st_size != read_in_full(i, data, size)) {
7645d8f1 1984 error_errno(_("'%s': short read"), filename);
e1327023
JK
1985 close(i);
1986 free(data);
1987 return -1;
1988 }
1989 close(i);
e1327023
JK
1990
1991 gs->buf = data;
1992 gs->size = size;
1993 return 0;
1994}
1995
3083301e 1996static int grep_source_load(struct grep_source *gs)
e1327023
JK
1997{
1998 if (gs->buf)
1999 return 0;
2000
2001 switch (gs->type) {
2002 case GREP_SOURCE_FILE:
2003 return grep_source_load_file(gs);
1c41c82b
BW
2004 case GREP_SOURCE_OID:
2005 return grep_source_load_oid(gs);
e1327023
JK
2006 case GREP_SOURCE_BUF:
2007 return gs->buf ? 0 : -1;
4538eef5
BW
2008 case GREP_SOURCE_SUBMODULE:
2009 break;
e1327023 2010 }
4538eef5 2011 die("BUG: invalid grep_source type to load");
0ab7befa 2012}
94ad9d9e
JK
2013
2014void grep_source_load_driver(struct grep_source *gs)
2015{
2016 if (gs->driver)
2017 return;
2018
2019 grep_attr_lock();
55c61688
NTND
2020 if (gs->path)
2021 gs->driver = userdiff_find_by_path(gs->path);
94ad9d9e
JK
2022 if (!gs->driver)
2023 gs->driver = userdiff_find_by_name("default");
2024 grep_attr_unlock();
2025}
41b59bfc 2026
3083301e 2027static int grep_source_is_binary(struct grep_source *gs)
41b59bfc
JK
2028{
2029 grep_source_load_driver(gs);
2030 if (gs->driver->binary != -1)
2031 return gs->driver->binary;
2032
2033 if (!grep_source_load(gs))
2034 return buffer_is_binary(gs->buf, gs->size);
2035
2036 return 0;
2037}