t7810: add test for grep -W and trailing empty context lines
[git/git.git] / grep.c
CommitLineData
83b5d2f5 1#include "cache.h"
83b5d2f5 2#include "grep.h"
60ecac98 3#include "userdiff.h"
6bfce93e 4#include "xdiff-interface.h"
335ec3bf
JK
5#include "diff.h"
6#include "diffcore.h"
83b5d2f5 7
07a7d656
JH
8static int grep_source_load(struct grep_source *gs);
9static int grep_source_is_binary(struct grep_source *gs);
10
7687a054
JH
11static struct grep_opt grep_defaults;
12
13/*
14 * Initialize the grep_defaults template with hardcoded defaults.
15 * We could let the compiler do this, but without C99 initializers
16 * the code gets unwieldy and unreadable, so...
17 */
18void init_grep_defaults(void)
19{
20 struct grep_opt *opt = &grep_defaults;
918d4e1c
JH
21 static int run_once;
22
23 if (run_once)
24 return;
25 run_once++;
7687a054
JH
26
27 memset(opt, 0, sizeof(*opt));
28 opt->relative = 1;
29 opt->pathname = 1;
30 opt->regflags = REG_NEWLINE;
31 opt->max_depth = -1;
32 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
33 opt->extended_regexp_option = 0;
7ce4fb94
JK
34 color_set(opt->color_context, "");
35 color_set(opt->color_filename, "");
36 color_set(opt->color_function, "");
37 color_set(opt->color_lineno, "");
38 color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
39 color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
40 color_set(opt->color_selected, "");
41 color_set(opt->color_sep, GIT_COLOR_CYAN);
7687a054
JH
42 opt->color = -1;
43}
44
45static int parse_pattern_type_arg(const char *opt, const char *arg)
46{
47 if (!strcmp(arg, "default"))
48 return GREP_PATTERN_TYPE_UNSPECIFIED;
49 else if (!strcmp(arg, "basic"))
50 return GREP_PATTERN_TYPE_BRE;
51 else if (!strcmp(arg, "extended"))
52 return GREP_PATTERN_TYPE_ERE;
53 else if (!strcmp(arg, "fixed"))
54 return GREP_PATTERN_TYPE_FIXED;
55 else if (!strcmp(arg, "perl"))
56 return GREP_PATTERN_TYPE_PCRE;
57 die("bad %s argument: %s", opt, arg);
58}
59
60/*
61 * Read the configuration file once and store it in
62 * the grep_defaults template.
63 */
64int grep_config(const char *var, const char *value, void *cb)
65{
66 struct grep_opt *opt = &grep_defaults;
67 char *color = NULL;
68
69 if (userdiff_config(var, value) < 0)
70 return -1;
71
72 if (!strcmp(var, "grep.extendedregexp")) {
73 if (git_config_bool(var, value))
74 opt->extended_regexp_option = 1;
75 else
76 opt->extended_regexp_option = 0;
77 return 0;
78 }
79
80 if (!strcmp(var, "grep.patterntype")) {
81 opt->pattern_type_option = parse_pattern_type_arg(var, value);
82 return 0;
83 }
84
85 if (!strcmp(var, "grep.linenumber")) {
86 opt->linenum = git_config_bool(var, value);
87 return 0;
88 }
89
6453f7b3
AS
90 if (!strcmp(var, "grep.fullname")) {
91 opt->relative = !git_config_bool(var, value);
92 return 0;
93 }
94
7687a054
JH
95 if (!strcmp(var, "color.grep"))
96 opt->color = git_config_colorbool(var, value);
97 else if (!strcmp(var, "color.grep.context"))
98 color = opt->color_context;
99 else if (!strcmp(var, "color.grep.filename"))
100 color = opt->color_filename;
101 else if (!strcmp(var, "color.grep.function"))
102 color = opt->color_function;
103 else if (!strcmp(var, "color.grep.linenumber"))
104 color = opt->color_lineno;
79a77109
RS
105 else if (!strcmp(var, "color.grep.matchcontext"))
106 color = opt->color_match_context;
107 else if (!strcmp(var, "color.grep.matchselected"))
108 color = opt->color_match_selected;
7687a054
JH
109 else if (!strcmp(var, "color.grep.selected"))
110 color = opt->color_selected;
111 else if (!strcmp(var, "color.grep.separator"))
112 color = opt->color_sep;
79a77109
RS
113 else if (!strcmp(var, "color.grep.match")) {
114 int rc = 0;
115 if (!value)
116 return config_error_nonbool(var);
bf1f639e
JH
117 rc |= color_parse(value, opt->color_match_context);
118 rc |= color_parse(value, opt->color_match_selected);
79a77109
RS
119 return rc;
120 }
7687a054
JH
121
122 if (color) {
123 if (!value)
124 return config_error_nonbool(var);
f6c5a296 125 return color_parse(value, color);
7687a054
JH
126 }
127 return 0;
128}
129
130/*
131 * Initialize one instance of grep_opt and copy the
132 * default values from the template we read the configuration
133 * information in an earlier call to git_config(grep_config).
134 */
135void grep_init(struct grep_opt *opt, const char *prefix)
136{
137 struct grep_opt *def = &grep_defaults;
138
139 memset(opt, 0, sizeof(*opt));
140 opt->prefix = prefix;
141 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
142 opt->pattern_tail = &opt->pattern_list;
143 opt->header_tail = &opt->header_list;
144
145 opt->color = def->color;
146 opt->extended_regexp_option = def->extended_regexp_option;
147 opt->pattern_type_option = def->pattern_type_option;
148 opt->linenum = def->linenum;
149 opt->max_depth = def->max_depth;
150 opt->pathname = def->pathname;
151 opt->regflags = def->regflags;
152 opt->relative = def->relative;
153
7ce4fb94
JK
154 color_set(opt->color_context, def->color_context);
155 color_set(opt->color_filename, def->color_filename);
156 color_set(opt->color_function, def->color_function);
157 color_set(opt->color_lineno, def->color_lineno);
158 color_set(opt->color_match_context, def->color_match_context);
159 color_set(opt->color_match_selected, def->color_match_selected);
160 color_set(opt->color_selected, def->color_selected);
161 color_set(opt->color_sep, def->color_sep);
7687a054 162}
07a7d656 163
c5c31d33
JH
164void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
165{
166 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
167 grep_set_pattern_type_option(pattern_type, opt);
168 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
169 grep_set_pattern_type_option(opt->pattern_type_option, opt);
170 else if (opt->extended_regexp_option)
171 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
172}
173
174void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
175{
176 switch (pattern_type) {
177 case GREP_PATTERN_TYPE_UNSPECIFIED:
178 /* fall through */
179
180 case GREP_PATTERN_TYPE_BRE:
181 opt->fixed = 0;
182 opt->pcre = 0;
183 opt->regflags &= ~REG_EXTENDED;
184 break;
185
186 case GREP_PATTERN_TYPE_ERE:
187 opt->fixed = 0;
188 opt->pcre = 0;
189 opt->regflags |= REG_EXTENDED;
190 break;
191
192 case GREP_PATTERN_TYPE_FIXED:
193 opt->fixed = 1;
194 opt->pcre = 0;
195 opt->regflags &= ~REG_EXTENDED;
196 break;
197
198 case GREP_PATTERN_TYPE_PCRE:
199 opt->fixed = 0;
200 opt->pcre = 1;
201 opt->regflags &= ~REG_EXTENDED;
202 break;
203 }
204}
205
fc456751
RS
206static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
207 const char *origin, int no,
208 enum grep_pat_token t,
209 enum grep_header_field field)
a4d7d2c6
JH
210{
211 struct grep_pat *p = xcalloc(1, sizeof(*p));
526a858a 212 p->pattern = xmemdupz(pat, patlen);
fc456751
RS
213 p->patternlen = patlen;
214 p->origin = origin;
215 p->no = no;
216 p->token = t;
a4d7d2c6 217 p->field = field;
fc456751
RS
218 return p;
219}
220
2b3873ff
RS
221static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
222{
223 **tail = p;
224 *tail = &p->next;
a4d7d2c6 225 p->next = NULL;
526a858a
RS
226
227 switch (p->token) {
228 case GREP_PATTERN: /* atom */
229 case GREP_PATTERN_HEAD:
230 case GREP_PATTERN_BODY:
231 for (;;) {
232 struct grep_pat *new_pat;
233 size_t len = 0;
234 char *cp = p->pattern + p->patternlen, *nl = NULL;
235 while (++len <= p->patternlen) {
236 if (*(--cp) == '\n') {
237 nl = cp;
238 break;
239 }
240 }
241 if (!nl)
242 break;
243 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
244 p->no, p->token, p->field);
245 new_pat->next = p->next;
246 if (!p->next)
247 *tail = &new_pat->next;
248 p->next = new_pat;
249 *nl = '\0';
250 p->patternlen -= len;
251 }
252 break;
253 default:
254 break;
255 }
2b3873ff
RS
256}
257
fc456751
RS
258void append_header_grep_pattern(struct grep_opt *opt,
259 enum grep_header_field field, const char *pat)
260{
261 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
262 GREP_PATTERN_HEAD, field);
baa6378f
JH
263 if (field == GREP_HEADER_REFLOG)
264 opt->use_reflog_filter = 1;
2b3873ff 265 do_append_grep_pat(&opt->header_tail, p);
a4d7d2c6
JH
266}
267
83b5d2f5
JH
268void append_grep_pattern(struct grep_opt *opt, const char *pat,
269 const char *origin, int no, enum grep_pat_token t)
ed40a095
RS
270{
271 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
272}
273
274void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
275 const char *origin, int no, enum grep_pat_token t)
83b5d2f5 276{
fc456751 277 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
2b3873ff 278 do_append_grep_pat(&opt->pattern_tail, p);
83b5d2f5
JH
279}
280
5b594f45
FK
281struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
282{
283 struct grep_pat *pat;
284 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
285 *ret = *opt;
286
287 ret->pattern_list = NULL;
288 ret->pattern_tail = &ret->pattern_list;
289
290 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
291 {
292 if(pat->token == GREP_PATTERN_HEAD)
293 append_header_grep_pattern(ret, pat->field,
294 pat->pattern);
295 else
ed40a095
RS
296 append_grep_pat(ret, pat->pattern, pat->patternlen,
297 pat->origin, pat->no, pat->token);
5b594f45
FK
298 }
299
300 return ret;
301}
302
a30c148a
MK
303static NORETURN void compile_regexp_failed(const struct grep_pat *p,
304 const char *error)
305{
306 char where[1024];
307
308 if (p->no)
19bdd3e7 309 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
a30c148a 310 else if (p->origin)
19bdd3e7 311 xsnprintf(where, sizeof(where), "%s, ", p->origin);
a30c148a
MK
312 else
313 where[0] = 0;
314
315 die("%s'%s': %s", where, p->pattern, error);
316}
317
63e7e9d8
MK
318#ifdef USE_LIBPCRE
319static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
320{
321 const char *error;
322 int erroffset;
fba4f125 323 int options = PCRE_MULTILINE;
63e7e9d8
MK
324
325 if (opt->ignore_case)
326 options |= PCRE_CASELESS;
327
328 p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
329 NULL);
330 if (!p->pcre_regexp)
331 compile_regexp_failed(p, error);
332
333 p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
334 if (!p->pcre_extra_info && error)
335 die("%s", error);
336}
337
338static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
339 regmatch_t *match, int eflags)
340{
341 int ovector[30], ret, flags = 0;
342
343 if (eflags & REG_NOTBOL)
344 flags |= PCRE_NOTBOL;
345
346 ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
347 0, flags, ovector, ARRAY_SIZE(ovector));
348 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
349 die("pcre_exec failed with error code %d", ret);
350 if (ret > 0) {
351 ret = 0;
352 match->rm_so = ovector[0];
353 match->rm_eo = ovector[1];
354 }
355
356 return ret;
357}
358
359static void free_pcre_regexp(struct grep_pat *p)
360{
361 pcre_free(p->pcre_regexp);
362 pcre_free(p->pcre_extra_info);
363}
364#else /* !USE_LIBPCRE */
365static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
366{
367 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
368}
369
370static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
371 regmatch_t *match, int eflags)
372{
373 return 1;
374}
375
376static void free_pcre_regexp(struct grep_pat *p)
377{
378}
379#endif /* !USE_LIBPCRE */
380
9eceddee
FK
381static int is_fixed(const char *s, size_t len)
382{
383 size_t i;
384
385 /* regcomp cannot accept patterns with NULs so we
386 * consider any pattern containing a NUL fixed.
387 */
388 if (memchr(s, 0, len))
389 return 1;
390
391 for (i = 0; i < len; i++) {
392 if (is_regex_special(s[i]))
393 return 0;
394 }
395
396 return 1;
397}
398
83b5d2f5
JH
399static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
400{
c822255c
RS
401 int err;
402
d7eb527d 403 p->word_regexp = opt->word_regexp;
5183bf67 404 p->ignore_case = opt->ignore_case;
d7eb527d 405
9eceddee
FK
406 if (opt->fixed || is_fixed(p->pattern, p->patternlen))
407 p->fixed = 1;
408 else
409 p->fixed = 0;
410
411 if (p->fixed) {
0f871cf5
JH
412 if (opt->regflags & REG_ICASE || p->ignore_case)
413 p->kws = kwsalloc(tolower_trans_tbl);
414 else
9eceddee 415 p->kws = kwsalloc(NULL);
9eceddee
FK
416 kwsincr(p->kws, p->pattern, p->patternlen);
417 kwsprep(p->kws);
c822255c 418 return;
9eceddee 419 }
c822255c 420
63e7e9d8
MK
421 if (opt->pcre) {
422 compile_pcre_regexp(p, opt);
423 return;
424 }
425
c822255c 426 err = regcomp(&p->regexp, p->pattern, opt->regflags);
83b5d2f5
JH
427 if (err) {
428 char errbuf[1024];
83b5d2f5
JH
429 regerror(err, &p->regexp, errbuf, 1024);
430 regfree(&p->regexp);
a30c148a 431 compile_regexp_failed(p, errbuf);
83b5d2f5
JH
432 }
433}
434
0ab7befa 435static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
436static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
437{
438 struct grep_pat *p;
439 struct grep_expr *x;
440
441 p = *list;
c922b01f
LT
442 if (!p)
443 return NULL;
83b5d2f5
JH
444 switch (p->token) {
445 case GREP_PATTERN: /* atom */
480c1ca6
JH
446 case GREP_PATTERN_HEAD:
447 case GREP_PATTERN_BODY:
83b5d2f5
JH
448 x = xcalloc(1, sizeof (struct grep_expr));
449 x->node = GREP_NODE_ATOM;
450 x->u.atom = p;
451 *list = p->next;
452 return x;
453 case GREP_OPEN_PAREN:
454 *list = p->next;
0ab7befa 455 x = compile_pattern_or(list);
83b5d2f5
JH
456 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
457 die("unmatched parenthesis");
458 *list = (*list)->next;
459 return x;
460 default:
461 return NULL;
462 }
463}
464
465static struct grep_expr *compile_pattern_not(struct grep_pat **list)
466{
467 struct grep_pat *p;
468 struct grep_expr *x;
469
470 p = *list;
c922b01f
LT
471 if (!p)
472 return NULL;
83b5d2f5
JH
473 switch (p->token) {
474 case GREP_NOT:
475 if (!p->next)
476 die("--not not followed by pattern expression");
477 *list = p->next;
478 x = xcalloc(1, sizeof (struct grep_expr));
479 x->node = GREP_NODE_NOT;
480 x->u.unary = compile_pattern_not(list);
481 if (!x->u.unary)
482 die("--not followed by non pattern expression");
483 return x;
484 default:
485 return compile_pattern_atom(list);
486 }
487}
488
489static struct grep_expr *compile_pattern_and(struct grep_pat **list)
490{
491 struct grep_pat *p;
492 struct grep_expr *x, *y, *z;
493
494 x = compile_pattern_not(list);
495 p = *list;
496 if (p && p->token == GREP_AND) {
497 if (!p->next)
498 die("--and not followed by pattern expression");
499 *list = p->next;
500 y = compile_pattern_and(list);
501 if (!y)
502 die("--and not followed by pattern expression");
503 z = xcalloc(1, sizeof (struct grep_expr));
504 z->node = GREP_NODE_AND;
505 z->u.binary.left = x;
506 z->u.binary.right = y;
507 return z;
508 }
509 return x;
510}
511
512static struct grep_expr *compile_pattern_or(struct grep_pat **list)
513{
514 struct grep_pat *p;
515 struct grep_expr *x, *y, *z;
516
517 x = compile_pattern_and(list);
518 p = *list;
519 if (x && p && p->token != GREP_CLOSE_PAREN) {
520 y = compile_pattern_or(list);
521 if (!y)
522 die("not a pattern expression %s", p->pattern);
523 z = xcalloc(1, sizeof (struct grep_expr));
524 z->node = GREP_NODE_OR;
525 z->u.binary.left = x;
526 z->u.binary.right = y;
527 return z;
528 }
529 return x;
530}
531
532static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
533{
534 return compile_pattern_or(list);
535}
536
17bf35a3
JH
537static void indent(int in)
538{
539 while (in-- > 0)
540 fputc(' ', stderr);
541}
542
543static void dump_grep_pat(struct grep_pat *p)
544{
545 switch (p->token) {
546 case GREP_AND: fprintf(stderr, "*and*"); break;
547 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
548 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
549 case GREP_NOT: fprintf(stderr, "*not*"); break;
550 case GREP_OR: fprintf(stderr, "*or*"); break;
551
552 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
553 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
554 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
555 }
556
557 switch (p->token) {
558 default: break;
559 case GREP_PATTERN_HEAD:
560 fprintf(stderr, "<head %d>", p->field); break;
561 case GREP_PATTERN_BODY:
562 fprintf(stderr, "<body>"); break;
563 }
564 switch (p->token) {
565 default: break;
566 case GREP_PATTERN_HEAD:
567 case GREP_PATTERN_BODY:
568 case GREP_PATTERN:
569 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
570 break;
571 }
572 fputc('\n', stderr);
573}
574
575static void dump_grep_expression_1(struct grep_expr *x, int in)
576{
577 indent(in);
578 switch (x->node) {
579 case GREP_NODE_TRUE:
580 fprintf(stderr, "true\n");
581 break;
582 case GREP_NODE_ATOM:
583 dump_grep_pat(x->u.atom);
584 break;
585 case GREP_NODE_NOT:
586 fprintf(stderr, "(not\n");
587 dump_grep_expression_1(x->u.unary, in+1);
588 indent(in);
589 fprintf(stderr, ")\n");
590 break;
591 case GREP_NODE_AND:
592 fprintf(stderr, "(and\n");
593 dump_grep_expression_1(x->u.binary.left, in+1);
594 dump_grep_expression_1(x->u.binary.right, in+1);
595 indent(in);
596 fprintf(stderr, ")\n");
597 break;
598 case GREP_NODE_OR:
599 fprintf(stderr, "(or\n");
600 dump_grep_expression_1(x->u.binary.left, in+1);
601 dump_grep_expression_1(x->u.binary.right, in+1);
602 indent(in);
603 fprintf(stderr, ")\n");
604 break;
605 }
606}
607
07a7d656 608static void dump_grep_expression(struct grep_opt *opt)
17bf35a3
JH
609{
610 struct grep_expr *x = opt->pattern_expression;
611
612 if (opt->all_match)
613 fprintf(stderr, "[all-match]\n");
614 dump_grep_expression_1(x, 0);
615 fflush(NULL);
616}
617
5aaeb733
JH
618static struct grep_expr *grep_true_expr(void)
619{
620 struct grep_expr *z = xcalloc(1, sizeof(*z));
621 z->node = GREP_NODE_TRUE;
622 return z;
623}
624
625static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
626{
627 struct grep_expr *z = xcalloc(1, sizeof(*z));
628 z->node = GREP_NODE_OR;
629 z->u.binary.left = left;
630 z->u.binary.right = right;
631 return z;
632}
633
95ce9ce2 634static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
83b5d2f5
JH
635{
636 struct grep_pat *p;
95ce9ce2 637 struct grep_expr *header_expr;
5aaeb733
JH
638 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
639 enum grep_header_field fld;
83b5d2f5 640
95ce9ce2
JH
641 if (!opt->header_list)
642 return NULL;
2385f246 643
95ce9ce2
JH
644 for (p = opt->header_list; p; p = p->next) {
645 if (p->token != GREP_PATTERN_HEAD)
646 die("bug: a non-header pattern in grep header list.");
3ce3ffb8
AP
647 if (p->field < GREP_HEADER_FIELD_MIN ||
648 GREP_HEADER_FIELD_MAX <= p->field)
95ce9ce2
JH
649 die("bug: unknown header field %d", p->field);
650 compile_regexp(p, opt);
80235ba7 651 }
5aaeb733
JH
652
653 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
654 header_group[fld] = NULL;
655
656 for (p = opt->header_list; p; p = p->next) {
657 struct grep_expr *h;
658 struct grep_pat *pp = p;
659
660 h = compile_pattern_atom(&pp);
661 if (!h || pp != p->next)
662 die("bug: malformed header expr");
663 if (!header_group[p->field]) {
664 header_group[p->field] = h;
665 continue;
666 }
667 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
668 }
669
670 header_expr = NULL;
671
672 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
673 if (!header_group[fld])
674 continue;
675 if (!header_expr)
676 header_expr = grep_true_expr();
677 header_expr = grep_or_expr(header_group[fld], header_expr);
678 }
95ce9ce2
JH
679 return header_expr;
680}
681
13e4fc7e
JH
682static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
683{
684 struct grep_expr *z = x;
685
686 while (x) {
687 assert(x->node == GREP_NODE_OR);
688 if (x->u.binary.right &&
689 x->u.binary.right->node == GREP_NODE_TRUE) {
690 x->u.binary.right = y;
691 break;
692 }
693 x = x->u.binary.right;
694 }
695 return z;
696}
697
17bf35a3 698static void compile_grep_patterns_real(struct grep_opt *opt)
95ce9ce2
JH
699{
700 struct grep_pat *p;
701 struct grep_expr *header_expr = prep_header_patterns(opt);
0ab7befa 702
83b5d2f5 703 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
704 switch (p->token) {
705 case GREP_PATTERN: /* atom */
706 case GREP_PATTERN_HEAD:
707 case GREP_PATTERN_BODY:
c822255c 708 compile_regexp(p, opt);
480c1ca6
JH
709 break;
710 default:
83b5d2f5 711 opt->extended = 1;
480c1ca6
JH
712 break;
713 }
83b5d2f5
JH
714 }
715
80235ba7
JH
716 if (opt->all_match || header_expr)
717 opt->extended = 1;
17bf35a3 718 else if (!opt->extended && !opt->debug)
83b5d2f5
JH
719 return;
720
83b5d2f5 721 p = opt->pattern_list;
ba150a3f
MB
722 if (p)
723 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
724 if (p)
725 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
726
727 if (!header_expr)
728 return;
729
5aaeb733 730 if (!opt->pattern_expression)
80235ba7 731 opt->pattern_expression = header_expr;
13e4fc7e
JH
732 else if (opt->all_match)
733 opt->pattern_expression = grep_splice_or(header_expr,
734 opt->pattern_expression);
5aaeb733
JH
735 else
736 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
737 header_expr);
80235ba7 738 opt->all_match = 1;
83b5d2f5
JH
739}
740
17bf35a3
JH
741void compile_grep_patterns(struct grep_opt *opt)
742{
743 compile_grep_patterns_real(opt);
744 if (opt->debug)
745 dump_grep_expression(opt);
746}
747
b48fb5b6
JH
748static void free_pattern_expr(struct grep_expr *x)
749{
750 switch (x->node) {
5aaeb733 751 case GREP_NODE_TRUE:
b48fb5b6
JH
752 case GREP_NODE_ATOM:
753 break;
754 case GREP_NODE_NOT:
755 free_pattern_expr(x->u.unary);
756 break;
757 case GREP_NODE_AND:
758 case GREP_NODE_OR:
759 free_pattern_expr(x->u.binary.left);
760 free_pattern_expr(x->u.binary.right);
761 break;
762 }
763 free(x);
764}
765
766void free_grep_patterns(struct grep_opt *opt)
767{
768 struct grep_pat *p, *n;
769
770 for (p = opt->pattern_list; p; p = n) {
771 n = p->next;
772 switch (p->token) {
773 case GREP_PATTERN: /* atom */
774 case GREP_PATTERN_HEAD:
775 case GREP_PATTERN_BODY:
9eceddee
FK
776 if (p->kws)
777 kwsfree(p->kws);
778 else if (p->pcre_regexp)
63e7e9d8
MK
779 free_pcre_regexp(p);
780 else
781 regfree(&p->regexp);
526a858a 782 free(p->pattern);
b48fb5b6
JH
783 break;
784 default:
785 break;
786 }
787 free(p);
788 }
789
790 if (!opt->extended)
791 return;
792 free_pattern_expr(opt->pattern_expression);
793}
794
83b5d2f5
JH
795static char *end_of_line(char *cp, unsigned long *left)
796{
797 unsigned long l = *left;
798 while (l && *cp != '\n') {
799 l--;
800 cp++;
801 }
802 *left = l;
803 return cp;
804}
805
806static int word_char(char ch)
807{
808 return isalnum(ch) || ch == '_';
809}
810
55f638bd
ML
811static void output_color(struct grep_opt *opt, const void *data, size_t size,
812 const char *color)
813{
daa0c3d9 814 if (want_color(opt->color) && color && color[0]) {
55f638bd
ML
815 opt->output(opt, color, strlen(color));
816 opt->output(opt, data, size);
817 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
818 } else
819 opt->output(opt, data, size);
820}
821
822static void output_sep(struct grep_opt *opt, char sign)
823{
824 if (opt->null_following_name)
825 opt->output(opt, "\0", 1);
826 else
827 output_color(opt, &sign, 1, opt->color_sep);
828}
829
83caecca
RZ
830static void show_name(struct grep_opt *opt, const char *name)
831{
55f638bd 832 output_color(opt, name, strlen(name), opt->color_filename);
5b594f45 833 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
834}
835
ed40a095
RS
836static int fixmatch(struct grep_pat *p, char *line, char *eol,
837 regmatch_t *match)
83b5d2f5 838{
9eceddee
FK
839 struct kwsmatch kwsm;
840 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
841 if (offset == -1) {
83b5d2f5
JH
842 match->rm_so = match->rm_eo = -1;
843 return REG_NOMATCH;
9eceddee
FK
844 } else {
845 match->rm_so = offset;
846 match->rm_eo = match->rm_so + kwsm.size[0];
83b5d2f5
JH
847 return 0;
848 }
849}
850
f96e5673
RS
851static int regmatch(const regex_t *preg, char *line, char *eol,
852 regmatch_t *match, int eflags)
853{
854#ifdef REG_STARTEND
855 match->rm_so = 0;
856 match->rm_eo = eol - line;
857 eflags |= REG_STARTEND;
858#endif
859 return regexec(preg, line, 1, match, eflags);
860}
861
97e77784
MK
862static int patmatch(struct grep_pat *p, char *line, char *eol,
863 regmatch_t *match, int eflags)
864{
865 int hit;
866
867 if (p->fixed)
868 hit = !fixmatch(p, line, eol, match);
63e7e9d8
MK
869 else if (p->pcre_regexp)
870 hit = !pcrematch(p, line, eol, match, eflags);
97e77784
MK
871 else
872 hit = !regmatch(&p->regexp, line, eol, match, eflags);
873
874 return hit;
875}
876
a4d7d2c6
JH
877static int strip_timestamp(char *bol, char **eol_p)
878{
879 char *eol = *eol_p;
880 int ch;
881
882 while (bol < --eol) {
883 if (*eol != '>')
884 continue;
885 *eol_p = ++eol;
886 ch = *eol;
887 *eol = '\0';
888 return ch;
889 }
890 return 0;
891}
892
893static struct {
894 const char *field;
895 size_t len;
896} header_field[] = {
897 { "author ", 7 },
898 { "committer ", 10 },
72fd13f7 899 { "reflog ", 7 },
a4d7d2c6
JH
900};
901
d7eb527d 902static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
903 enum grep_context ctx,
904 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
905{
906 int hit = 0;
a4d7d2c6 907 int saved_ch = 0;
e701fadb 908 const char *start = bol;
83b5d2f5 909
480c1ca6
JH
910 if ((p->token != GREP_PATTERN) &&
911 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
912 return 0;
913
a4d7d2c6
JH
914 if (p->token == GREP_PATTERN_HEAD) {
915 const char *field;
916 size_t len;
917 assert(p->field < ARRAY_SIZE(header_field));
918 field = header_field[p->field].field;
919 len = header_field[p->field].len;
920 if (strncmp(bol, field, len))
921 return 0;
922 bol += len;
ad4813b3
NTND
923 switch (p->field) {
924 case GREP_HEADER_AUTHOR:
925 case GREP_HEADER_COMMITTER:
926 saved_ch = strip_timestamp(bol, &eol);
927 break;
928 default:
929 break;
930 }
a4d7d2c6
JH
931 }
932
83b5d2f5 933 again:
97e77784 934 hit = patmatch(p, bol, eol, pmatch, eflags);
83b5d2f5 935
d7eb527d 936 if (hit && p->word_regexp) {
83b5d2f5 937 if ((pmatch[0].rm_so < 0) ||
84201eae 938 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
939 (pmatch[0].rm_eo < 0) ||
940 (eol - bol) < pmatch[0].rm_eo)
941 die("regexp returned nonsense");
942
943 /* Match beginning must be either beginning of the
944 * line, or at word boundary (i.e. the last char must
945 * not be a word char). Similarly, match end must be
946 * either end of the line, or at word boundary
947 * (i.e. the next char must not be a word char).
948 */
fb62eb7f 949 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
950 !word_char(bol[pmatch[0].rm_so-1])) &&
951 ((pmatch[0].rm_eo == (eol-bol)) ||
952 !word_char(bol[pmatch[0].rm_eo])) )
953 ;
954 else
955 hit = 0;
956
84201eae
RS
957 /* Words consist of at least one character. */
958 if (pmatch->rm_so == pmatch->rm_eo)
959 hit = 0;
960
83b5d2f5
JH
961 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
962 /* There could be more than one match on the
963 * line, and the first match might not be
964 * strict word match. But later ones could be!
fb62eb7f
RS
965 * Forward to the next possible start, i.e. the
966 * next position following a non-word char.
83b5d2f5
JH
967 */
968 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
969 while (word_char(bol[-1]) && bol < eol)
970 bol++;
dbb6a4ad 971 eflags |= REG_NOTBOL;
fb62eb7f
RS
972 if (bol < eol)
973 goto again;
83b5d2f5
JH
974 }
975 }
a4d7d2c6
JH
976 if (p->token == GREP_PATTERN_HEAD && saved_ch)
977 *eol = saved_ch;
e701fadb
RS
978 if (hit) {
979 pmatch[0].rm_so += bol - start;
980 pmatch[0].rm_eo += bol - start;
981 }
83b5d2f5
JH
982 return hit;
983}
984
d7eb527d
RS
985static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
986 enum grep_context ctx, int collect_hits)
83b5d2f5 987{
0ab7befa 988 int h = 0;
79212772 989 regmatch_t match;
0ab7befa 990
c922b01f
LT
991 if (!x)
992 die("Not a valid grep expression");
83b5d2f5 993 switch (x->node) {
5aaeb733
JH
994 case GREP_NODE_TRUE:
995 h = 1;
996 break;
83b5d2f5 997 case GREP_NODE_ATOM:
79212772 998 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
83b5d2f5
JH
999 break;
1000 case GREP_NODE_NOT:
d7eb527d 1001 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
0ab7befa 1002 break;
83b5d2f5 1003 case GREP_NODE_AND:
d7eb527d 1004 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
252d560d 1005 return 0;
d7eb527d 1006 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
0ab7befa 1007 break;
83b5d2f5 1008 case GREP_NODE_OR:
0ab7befa 1009 if (!collect_hits)
d7eb527d 1010 return (match_expr_eval(x->u.binary.left,
0ab7befa 1011 bol, eol, ctx, 0) ||
d7eb527d 1012 match_expr_eval(x->u.binary.right,
0ab7befa 1013 bol, eol, ctx, 0));
d7eb527d 1014 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
0ab7befa 1015 x->u.binary.left->hit |= h;
d7eb527d 1016 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
0ab7befa
JH
1017 break;
1018 default:
d7530708 1019 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 1020 }
0ab7befa
JH
1021 if (collect_hits)
1022 x->hit |= h;
1023 return h;
83b5d2f5
JH
1024}
1025
480c1ca6 1026static int match_expr(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1027 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1028{
1029 struct grep_expr *x = opt->pattern_expression;
d7eb527d 1030 return match_expr_eval(x, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1031}
1032
480c1ca6 1033static int match_line(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1034 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1035{
1036 struct grep_pat *p;
79212772
RS
1037 regmatch_t match;
1038
83b5d2f5 1039 if (opt->extended)
0ab7befa
JH
1040 return match_expr(opt, bol, eol, ctx, collect_hits);
1041
1042 /* we do not call with collect_hits without being extended */
83b5d2f5 1043 for (p = opt->pattern_list; p; p = p->next) {
79212772 1044 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
83b5d2f5
JH
1045 return 1;
1046 }
1047 return 0;
1048}
1049
7e8f59d5
RS
1050static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1051 enum grep_context ctx,
1052 regmatch_t *pmatch, int eflags)
1053{
1054 regmatch_t match;
1055
1056 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1057 return 0;
1058 if (match.rm_so < 0 || match.rm_eo < 0)
1059 return 0;
1060 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1061 if (match.rm_so > pmatch->rm_so)
1062 return 1;
1063 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1064 return 1;
1065 }
1066 pmatch->rm_so = match.rm_so;
1067 pmatch->rm_eo = match.rm_eo;
1068 return 1;
1069}
1070
1071static int next_match(struct grep_opt *opt, char *bol, char *eol,
1072 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1073{
1074 struct grep_pat *p;
1075 int hit = 0;
1076
1077 pmatch->rm_so = pmatch->rm_eo = -1;
1078 if (bol < eol) {
1079 for (p = opt->pattern_list; p; p = p->next) {
1080 switch (p->token) {
1081 case GREP_PATTERN: /* atom */
1082 case GREP_PATTERN_HEAD:
1083 case GREP_PATTERN_BODY:
1084 hit |= match_next_pattern(p, bol, eol, ctx,
1085 pmatch, eflags);
1086 break;
1087 default:
1088 break;
1089 }
1090 }
1091 }
1092 return hit;
1093}
1094
1095static void show_line(struct grep_opt *opt, char *bol, char *eol,
1096 const char *name, unsigned lno, char sign)
1097{
1098 int rest = eol - bol;
79a77109 1099 const char *match_color, *line_color = NULL;
7e8f59d5 1100
a8f0e764
RS
1101 if (opt->file_break && opt->last_shown == 0) {
1102 if (opt->show_hunk_mark)
1103 opt->output(opt, "\n", 1);
ba8ea749 1104 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
046802d0 1105 if (opt->last_shown == 0) {
55f638bd
ML
1106 if (opt->show_hunk_mark) {
1107 output_color(opt, "--", 2, opt->color_sep);
1108 opt->output(opt, "\n", 1);
07b838f0 1109 }
55f638bd
ML
1110 } else if (lno > opt->last_shown + 1) {
1111 output_color(opt, "--", 2, opt->color_sep);
1112 opt->output(opt, "\n", 1);
1113 }
5dd06d38 1114 }
1d84f72e
RS
1115 if (opt->heading && opt->last_shown == 0) {
1116 output_color(opt, name, strlen(name), opt->color_filename);
1117 opt->output(opt, "\n", 1);
1118 }
5dd06d38
RS
1119 opt->last_shown = lno;
1120
1d84f72e 1121 if (!opt->heading && opt->pathname) {
55f638bd
ML
1122 output_color(opt, name, strlen(name), opt->color_filename);
1123 output_sep(opt, sign);
5b594f45
FK
1124 }
1125 if (opt->linenum) {
1126 char buf[32];
1127 snprintf(buf, sizeof(buf), "%d", lno);
55f638bd
ML
1128 output_color(opt, buf, strlen(buf), opt->color_lineno);
1129 output_sep(opt, sign);
5b594f45 1130 }
7e8f59d5
RS
1131 if (opt->color) {
1132 regmatch_t match;
1133 enum grep_context ctx = GREP_CONTEXT_BODY;
1134 int ch = *eol;
1135 int eflags = 0;
1136
79a77109
RS
1137 if (sign == ':')
1138 match_color = opt->color_match_selected;
1139 else
1140 match_color = opt->color_match_context;
00588bb5
ML
1141 if (sign == ':')
1142 line_color = opt->color_selected;
1143 else if (sign == '-')
1144 line_color = opt->color_context;
1145 else if (sign == '=')
1146 line_color = opt->color_function;
7e8f59d5
RS
1147 *eol = '\0';
1148 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
1149 if (match.rm_so == match.rm_eo)
1150 break;
5b594f45 1151
00588bb5 1152 output_color(opt, bol, match.rm_so, line_color);
55f638bd 1153 output_color(opt, bol + match.rm_so,
79a77109 1154 match.rm_eo - match.rm_so, match_color);
7e8f59d5
RS
1155 bol += match.rm_eo;
1156 rest -= match.rm_eo;
1157 eflags = REG_NOTBOL;
1158 }
1159 *eol = ch;
1160 }
00588bb5 1161 output_color(opt, bol, rest, line_color);
5b594f45 1162 opt->output(opt, "\n", 1);
7e8f59d5
RS
1163}
1164
0579f91d 1165#ifndef NO_PTHREADS
78db6ea9
JK
1166int grep_use_locks;
1167
0579f91d
TR
1168/*
1169 * This lock protects access to the gitattributes machinery, which is
1170 * not thread-safe.
1171 */
1172pthread_mutex_t grep_attr_mutex;
1173
78db6ea9 1174static inline void grep_attr_lock(void)
0579f91d 1175{
78db6ea9 1176 if (grep_use_locks)
0579f91d
TR
1177 pthread_mutex_lock(&grep_attr_mutex);
1178}
1179
78db6ea9 1180static inline void grep_attr_unlock(void)
0579f91d 1181{
78db6ea9 1182 if (grep_use_locks)
0579f91d
TR
1183 pthread_mutex_unlock(&grep_attr_mutex);
1184}
b3aeb285
JK
1185
1186/*
1187 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1188 */
1189pthread_mutex_t grep_read_mutex;
1190
0579f91d 1191#else
78db6ea9
JK
1192#define grep_attr_lock()
1193#define grep_attr_unlock()
0579f91d
TR
1194#endif
1195
e1327023 1196static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
2944e4e6 1197{
60ecac98 1198 xdemitconf_t *xecfg = opt->priv;
0579f91d 1199 if (xecfg && !xecfg->find_func) {
94ad9d9e
JK
1200 grep_source_load_driver(gs);
1201 if (gs->driver->funcname.pattern) {
1202 const struct userdiff_funcname *pe = &gs->driver->funcname;
0579f91d
TR
1203 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1204 } else {
1205 xecfg = opt->priv = NULL;
1206 }
1207 }
1208
1209 if (xecfg) {
60ecac98
RS
1210 char buf[1];
1211 return xecfg->find_func(bol, eol - bol, buf, 1,
1212 xecfg->find_func_priv) >= 0;
1213 }
1214
2944e4e6
RS
1215 if (bol == eol)
1216 return 0;
1217 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1218 return 1;
1219 return 0;
1220}
1221
e1327023
JK
1222static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1223 char *bol, unsigned lno)
2944e4e6 1224{
e1327023 1225 while (bol > gs->buf) {
2944e4e6
RS
1226 char *eol = --bol;
1227
e1327023 1228 while (bol > gs->buf && bol[-1] != '\n')
2944e4e6
RS
1229 bol--;
1230 lno--;
1231
1232 if (lno <= opt->last_shown)
1233 break;
1234
e1327023
JK
1235 if (match_funcname(opt, gs, bol, eol)) {
1236 show_line(opt, bol, eol, gs->name, lno, '=');
2944e4e6
RS
1237 break;
1238 }
1239 }
1240}
1241
e1327023 1242static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
ba8ea749 1243 char *bol, char *end, unsigned lno)
49de3216 1244{
2944e4e6 1245 unsigned cur = lno, from = 1, funcname_lno = 0;
ba8ea749
RS
1246 int funcname_needed = !!opt->funcname;
1247
e1327023 1248 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
ba8ea749 1249 funcname_needed = 2;
49de3216
RS
1250
1251 if (opt->pre_context < lno)
1252 from = lno - opt->pre_context;
1253 if (from <= opt->last_shown)
1254 from = opt->last_shown + 1;
1255
1256 /* Rewind. */
e1327023 1257 while (bol > gs->buf &&
ba8ea749 1258 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
2944e4e6
RS
1259 char *eol = --bol;
1260
e1327023 1261 while (bol > gs->buf && bol[-1] != '\n')
49de3216
RS
1262 bol--;
1263 cur--;
e1327023 1264 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
2944e4e6
RS
1265 funcname_lno = cur;
1266 funcname_needed = 0;
1267 }
49de3216
RS
1268 }
1269
2944e4e6
RS
1270 /* We need to look even further back to find a function signature. */
1271 if (opt->funcname && funcname_needed)
e1327023 1272 show_funcname_line(opt, gs, bol, cur);
2944e4e6 1273
49de3216
RS
1274 /* Back forward. */
1275 while (cur < lno) {
2944e4e6 1276 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
1277
1278 while (*eol != '\n')
1279 eol++;
e1327023 1280 show_line(opt, bol, eol, gs->name, cur, sign);
49de3216
RS
1281 bol = eol + 1;
1282 cur++;
1283 }
1284}
1285
a26345b6
JH
1286static int should_lookahead(struct grep_opt *opt)
1287{
1288 struct grep_pat *p;
1289
1290 if (opt->extended)
1291 return 0; /* punt for too complex stuff */
1292 if (opt->invert)
1293 return 0;
1294 for (p = opt->pattern_list; p; p = p->next) {
1295 if (p->token != GREP_PATTERN)
1296 return 0; /* punt for "header only" and stuff */
1297 }
1298 return 1;
1299}
1300
1301static int look_ahead(struct grep_opt *opt,
1302 unsigned long *left_p,
1303 unsigned *lno_p,
1304 char **bol_p)
1305{
1306 unsigned lno = *lno_p;
1307 char *bol = *bol_p;
1308 struct grep_pat *p;
1309 char *sp, *last_bol;
1310 regoff_t earliest = -1;
1311
1312 for (p = opt->pattern_list; p; p = p->next) {
1313 int hit;
1314 regmatch_t m;
1315
97e77784 1316 hit = patmatch(p, bol, bol + *left_p, &m, 0);
a26345b6
JH
1317 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1318 continue;
1319 if (earliest < 0 || m.rm_so < earliest)
1320 earliest = m.rm_so;
1321 }
1322
1323 if (earliest < 0) {
1324 *bol_p = bol + *left_p;
1325 *left_p = 0;
1326 return 1;
1327 }
1328 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1329 ; /* find the beginning of the line */
1330 last_bol = sp;
1331
1332 for (sp = bol; sp < last_bol; sp++) {
1333 if (*sp == '\n')
1334 lno++;
1335 }
1336 *left_p -= last_bol - bol;
1337 *bol_p = last_bol;
1338 *lno_p = lno;
1339 return 0;
1340}
1341
5b594f45
FK
1342static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1343{
1344 fwrite(buf, size, 1, stdout);
1345}
1346
335ec3bf
JK
1347static int fill_textconv_grep(struct userdiff_driver *driver,
1348 struct grep_source *gs)
1349{
1350 struct diff_filespec *df;
1351 char *buf;
1352 size_t size;
1353
1354 if (!driver || !driver->textconv)
1355 return grep_source_load(gs);
1356
1357 /*
1358 * The textconv interface is intimately tied to diff_filespecs, so we
1359 * have to pretend to be one. If we could unify the grep_source
1360 * and diff_filespec structs, this mess could just go away.
1361 */
1362 df = alloc_filespec(gs->path);
1363 switch (gs->type) {
1364 case GREP_SOURCE_SHA1:
1365 fill_filespec(df, gs->identifier, 1, 0100644);
1366 break;
1367 case GREP_SOURCE_FILE:
1368 fill_filespec(df, null_sha1, 0, 0100644);
1369 break;
1370 default:
1371 die("BUG: attempt to textconv something without a path?");
1372 }
1373
1374 /*
1375 * fill_textconv is not remotely thread-safe; it may load objects
1376 * behind the scenes, and it modifies the global diff tempfile
1377 * structure.
1378 */
1379 grep_read_lock();
1380 size = fill_textconv(driver, df, &buf);
1381 grep_read_unlock();
1382 free_filespec(df);
1383
1384 /*
1385 * The normal fill_textconv usage by the diff machinery would just keep
1386 * the textconv'd buf separate from the diff_filespec. But much of the
1387 * grep code passes around a grep_source and assumes that its "buf"
1388 * pointer is the beginning of the thing we are searching. So let's
1389 * install our textconv'd version into the grep_source, taking care not
1390 * to leak any existing buffer.
1391 */
1392 grep_source_clear_data(gs);
1393 gs->buf = buf;
1394 gs->size = size;
1395
1396 return 0;
1397}
1398
e1327023 1399static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
83b5d2f5 1400{
e1327023
JK
1401 char *bol;
1402 unsigned long left;
83b5d2f5 1403 unsigned lno = 1;
83b5d2f5 1404 unsigned last_hit = 0;
83b5d2f5 1405 int binary_match_only = 0;
83b5d2f5 1406 unsigned count = 0;
a26345b6 1407 int try_lookahead = 0;
ba8ea749 1408 int show_function = 0;
335ec3bf 1409 struct userdiff_driver *textconv = NULL;
480c1ca6 1410 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 1411 xdemitconf_t xecfg;
83b5d2f5 1412
5b594f45
FK
1413 if (!opt->output)
1414 opt->output = std_output;
1415
ba8ea749
RS
1416 if (opt->pre_context || opt->post_context || opt->file_break ||
1417 opt->funcbody) {
08303c36
RS
1418 /* Show hunk marks, except for the first file. */
1419 if (opt->last_shown)
1420 opt->show_hunk_mark = 1;
1421 /*
1422 * If we're using threads then we can't easily identify
1423 * the first file. Always put hunk marks in that case
1424 * and skip the very first one later in work_done().
1425 */
1426 if (opt->output != std_output)
1427 opt->show_hunk_mark = 1;
1428 }
431d6e7b
RS
1429 opt->last_shown = 0;
1430
335ec3bf
JK
1431 if (opt->allow_textconv) {
1432 grep_source_load_driver(gs);
1433 /*
1434 * We might set up the shared textconv cache data here, which
1435 * is not thread-safe.
1436 */
1437 grep_attr_lock();
1438 textconv = userdiff_get_textconv(gs->driver);
1439 grep_attr_unlock();
1440 }
1441
1442 /*
1443 * We know the result of a textconv is text, so we only have to care
1444 * about binary handling if we are not using it.
1445 */
1446 if (!textconv) {
1447 switch (opt->binary) {
1448 case GREP_BINARY_DEFAULT:
1449 if (grep_source_is_binary(gs))
1450 binary_match_only = 1;
1451 break;
1452 case GREP_BINARY_NOMATCH:
1453 if (grep_source_is_binary(gs))
1454 return 0; /* Assume unmatch */
1455 break;
1456 case GREP_BINARY_TEXT:
1457 break;
1458 default:
1459 die("bug: unknown binary handling mode");
1460 }
83b5d2f5
JH
1461 }
1462
60ecac98 1463 memset(&xecfg, 0, sizeof(xecfg));
0579f91d
TR
1464 opt->priv = &xecfg;
1465
a26345b6 1466 try_lookahead = should_lookahead(opt);
60ecac98 1467
335ec3bf 1468 if (fill_textconv_grep(textconv, gs) < 0)
08265798
JK
1469 return 0;
1470
e1327023
JK
1471 bol = gs->buf;
1472 left = gs->size;
83b5d2f5
JH
1473 while (left) {
1474 char *eol, ch;
0ab7befa 1475 int hit;
83b5d2f5 1476
a26345b6 1477 /*
8997da38 1478 * look_ahead() skips quickly to the line that possibly
a26345b6
JH
1479 * has the next hit; don't call it if we need to do
1480 * something more than just skipping the current line
1481 * in response to an unmatch for the current line. E.g.
1482 * inside a post-context window, we will show the current
1483 * line as a context around the previous hit when it
1484 * doesn't hit.
1485 */
1486 if (try_lookahead
1487 && !(last_hit
ba8ea749
RS
1488 && (show_function ||
1489 lno <= last_hit + opt->post_context))
a26345b6
JH
1490 && look_ahead(opt, &left, &lno, &bol))
1491 break;
83b5d2f5
JH
1492 eol = end_of_line(bol, &left);
1493 ch = *eol;
1494 *eol = 0;
1495
480c1ca6
JH
1496 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1497 ctx = GREP_CONTEXT_BODY;
1498
0ab7befa 1499 hit = match_line(opt, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1500 *eol = ch;
1501
0ab7befa
JH
1502 if (collect_hits)
1503 goto next_line;
1504
83b5d2f5
JH
1505 /* "grep -v -e foo -e bla" should list lines
1506 * that do not have either, so inversion should
1507 * be done outside.
1508 */
1509 if (opt->invert)
1510 hit = !hit;
1511 if (opt->unmatch_name_only) {
1512 if (hit)
1513 return 0;
1514 goto next_line;
1515 }
1516 if (hit) {
1517 count++;
1518 if (opt->status_only)
1519 return 1;
321ffcc0 1520 if (opt->name_only) {
e1327023 1521 show_name(opt, gs->name);
321ffcc0
RS
1522 return 1;
1523 }
c30c10cf
RS
1524 if (opt->count)
1525 goto next_line;
83b5d2f5 1526 if (binary_match_only) {
5b594f45 1527 opt->output(opt, "Binary file ", 12);
e1327023 1528 output_color(opt, gs->name, strlen(gs->name),
55f638bd 1529 opt->color_filename);
5b594f45 1530 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
1531 return 1;
1532 }
83b5d2f5
JH
1533 /* Hit at this line. If we haven't shown the
1534 * pre-context lines, we would need to show them.
83b5d2f5 1535 */
ba8ea749 1536 if (opt->pre_context || opt->funcbody)
e1327023 1537 show_pre_context(opt, gs, bol, eol, lno);
2944e4e6 1538 else if (opt->funcname)
e1327023
JK
1539 show_funcname_line(opt, gs, bol, lno);
1540 show_line(opt, bol, eol, gs->name, lno, ':');
5dd06d38 1541 last_hit = lno;
ba8ea749
RS
1542 if (opt->funcbody)
1543 show_function = 1;
1544 goto next_line;
83b5d2f5 1545 }
e1327023 1546 if (show_function && match_funcname(opt, gs, bol, eol))
ba8ea749
RS
1547 show_function = 0;
1548 if (show_function ||
1549 (last_hit && lno <= last_hit + opt->post_context)) {
83b5d2f5
JH
1550 /* If the last hit is within the post context,
1551 * we need to show this line.
1552 */
e1327023 1553 show_line(opt, bol, eol, gs->name, lno, '-');
83b5d2f5 1554 }
83b5d2f5
JH
1555
1556 next_line:
1557 bol = eol + 1;
1558 if (!left)
1559 break;
1560 left--;
1561 lno++;
1562 }
1563
0ab7befa
JH
1564 if (collect_hits)
1565 return 0;
b48fb5b6 1566
83b5d2f5
JH
1567 if (opt->status_only)
1568 return 0;
1569 if (opt->unmatch_name_only) {
1570 /* We did not see any hit, so we want to show this */
e1327023 1571 show_name(opt, gs->name);
83b5d2f5
JH
1572 return 1;
1573 }
1574
60ecac98
RS
1575 xdiff_clear_find_func(&xecfg);
1576 opt->priv = NULL;
1577
83b5d2f5
JH
1578 /* NEEDSWORK:
1579 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1580 * which feels mostly useless but sometimes useful. Maybe
1581 * make it another option? For now suppress them.
1582 */
5b594f45
FK
1583 if (opt->count && count) {
1584 char buf[32];
f76d947a
RS
1585 if (opt->pathname) {
1586 output_color(opt, gs->name, strlen(gs->name),
1587 opt->color_filename);
1588 output_sep(opt, ':');
1589 }
55f638bd 1590 snprintf(buf, sizeof(buf), "%u\n", count);
5b594f45 1591 opt->output(opt, buf, strlen(buf));
c30c10cf 1592 return 1;
5b594f45 1593 }
83b5d2f5
JH
1594 return !!last_hit;
1595}
1596
0ab7befa
JH
1597static void clr_hit_marker(struct grep_expr *x)
1598{
1599 /* All-hit markers are meaningful only at the very top level
1600 * OR node.
1601 */
1602 while (1) {
1603 x->hit = 0;
1604 if (x->node != GREP_NODE_OR)
1605 return;
1606 x->u.binary.left->hit = 0;
1607 x = x->u.binary.right;
1608 }
1609}
1610
1611static int chk_hit_marker(struct grep_expr *x)
1612{
1613 /* Top level nodes have hit markers. See if they all are hits */
1614 while (1) {
1615 if (x->node != GREP_NODE_OR)
1616 return x->hit;
1617 if (!x->u.binary.left->hit)
1618 return 0;
1619 x = x->u.binary.right;
1620 }
1621}
1622
e1327023 1623int grep_source(struct grep_opt *opt, struct grep_source *gs)
0ab7befa
JH
1624{
1625 /*
1626 * we do not have to do the two-pass grep when we do not check
1627 * buffer-wide "all-match".
1628 */
1629 if (!opt->all_match)
e1327023 1630 return grep_source_1(opt, gs, 0);
0ab7befa
JH
1631
1632 /* Otherwise the toplevel "or" terms hit a bit differently.
1633 * We first clear hit markers from them.
1634 */
1635 clr_hit_marker(opt->pattern_expression);
e1327023 1636 grep_source_1(opt, gs, 1);
0ab7befa
JH
1637
1638 if (!chk_hit_marker(opt->pattern_expression))
1639 return 0;
1640
e1327023
JK
1641 return grep_source_1(opt, gs, 0);
1642}
1643
c876d6da 1644int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
e1327023
JK
1645{
1646 struct grep_source gs;
1647 int r;
1648
55c61688 1649 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
e1327023
JK
1650 gs.buf = buf;
1651 gs.size = size;
1652
1653 r = grep_source(opt, &gs);
1654
1655 grep_source_clear(&gs);
1656 return r;
1657}
1658
1659void grep_source_init(struct grep_source *gs, enum grep_source_type type,
55c61688
NTND
1660 const char *name, const char *path,
1661 const void *identifier)
e1327023
JK
1662{
1663 gs->type = type;
8c53f071
JK
1664 gs->name = xstrdup_or_null(name);
1665 gs->path = xstrdup_or_null(path);
e1327023
JK
1666 gs->buf = NULL;
1667 gs->size = 0;
94ad9d9e 1668 gs->driver = NULL;
e1327023
JK
1669
1670 switch (type) {
1671 case GREP_SOURCE_FILE:
1672 gs->identifier = xstrdup(identifier);
1673 break;
1674 case GREP_SOURCE_SHA1:
1675 gs->identifier = xmalloc(20);
50546b15 1676 hashcpy(gs->identifier, identifier);
e1327023
JK
1677 break;
1678 case GREP_SOURCE_BUF:
1679 gs->identifier = NULL;
1680 }
1681}
1682
1683void grep_source_clear(struct grep_source *gs)
1684{
1685 free(gs->name);
1686 gs->name = NULL;
55c61688
NTND
1687 free(gs->path);
1688 gs->path = NULL;
e1327023
JK
1689 free(gs->identifier);
1690 gs->identifier = NULL;
1691 grep_source_clear_data(gs);
1692}
1693
1694void grep_source_clear_data(struct grep_source *gs)
1695{
1696 switch (gs->type) {
1697 case GREP_SOURCE_FILE:
1698 case GREP_SOURCE_SHA1:
1699 free(gs->buf);
1700 gs->buf = NULL;
1701 gs->size = 0;
1702 break;
1703 case GREP_SOURCE_BUF:
1704 /* leave user-provided buf intact */
1705 break;
1706 }
1707}
1708
1709static int grep_source_load_sha1(struct grep_source *gs)
1710{
1711 enum object_type type;
1712
1713 grep_read_lock();
1714 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1715 grep_read_unlock();
1716
1717 if (!gs->buf)
1718 return error(_("'%s': unable to read %s"),
1719 gs->name,
1720 sha1_to_hex(gs->identifier));
1721 return 0;
1722}
1723
1724static int grep_source_load_file(struct grep_source *gs)
1725{
1726 const char *filename = gs->identifier;
1727 struct stat st;
1728 char *data;
1729 size_t size;
1730 int i;
1731
1732 if (lstat(filename, &st) < 0) {
1733 err_ret:
1734 if (errno != ENOENT)
1735 error(_("'%s': %s"), filename, strerror(errno));
1736 return -1;
1737 }
1738 if (!S_ISREG(st.st_mode))
1739 return -1;
1740 size = xsize_t(st.st_size);
1741 i = open(filename, O_RDONLY);
1742 if (i < 0)
1743 goto err_ret;
3733e694 1744 data = xmallocz(size);
e1327023
JK
1745 if (st.st_size != read_in_full(i, data, size)) {
1746 error(_("'%s': short read %s"), filename, strerror(errno));
1747 close(i);
1748 free(data);
1749 return -1;
1750 }
1751 close(i);
e1327023
JK
1752
1753 gs->buf = data;
1754 gs->size = size;
1755 return 0;
1756}
1757
3083301e 1758static int grep_source_load(struct grep_source *gs)
e1327023
JK
1759{
1760 if (gs->buf)
1761 return 0;
1762
1763 switch (gs->type) {
1764 case GREP_SOURCE_FILE:
1765 return grep_source_load_file(gs);
1766 case GREP_SOURCE_SHA1:
1767 return grep_source_load_sha1(gs);
1768 case GREP_SOURCE_BUF:
1769 return gs->buf ? 0 : -1;
1770 }
1771 die("BUG: invalid grep_source type");
0ab7befa 1772}
94ad9d9e
JK
1773
1774void grep_source_load_driver(struct grep_source *gs)
1775{
1776 if (gs->driver)
1777 return;
1778
1779 grep_attr_lock();
55c61688
NTND
1780 if (gs->path)
1781 gs->driver = userdiff_find_by_path(gs->path);
94ad9d9e
JK
1782 if (!gs->driver)
1783 gs->driver = userdiff_find_by_name("default");
1784 grep_attr_unlock();
1785}
41b59bfc 1786
3083301e 1787static int grep_source_is_binary(struct grep_source *gs)
41b59bfc
JK
1788{
1789 grep_source_load_driver(gs);
1790 if (gs->driver->binary != -1)
1791 return gs->driver->binary;
1792
1793 if (!grep_source_load(gs))
1794 return buffer_is_binary(gs->buf, gs->size);
1795
1796 return 0;
1797}