]>
Commit | Line | Data |
---|---|---|
83b5d2f5 | 1 | #include "cache.h" |
b2141fc1 | 2 | #include "config.h" |
83b5d2f5 | 3 | #include "grep.h" |
cbd53a21 | 4 | #include "object-store.h" |
60ecac98 | 5 | #include "userdiff.h" |
6bfce93e | 6 | #include "xdiff-interface.h" |
335ec3bf JK |
7 | #include "diff.h" |
8 | #include "diffcore.h" | |
5c1ebcca | 9 | #include "commit.h" |
793dc676 | 10 | #include "quote.h" |
3ac68a93 | 11 | #include "help.h" |
83b5d2f5 | 12 | |
07a7d656 | 13 | static int grep_source_load(struct grep_source *gs); |
acd00ea0 NTND |
14 | static int grep_source_is_binary(struct grep_source *gs, |
15 | struct index_state *istate); | |
07a7d656 | 16 | |
7687a054 JH |
17 | static struct grep_opt grep_defaults; |
18 | ||
513f2b0b CMAB |
19 | #ifdef USE_LIBPCRE2 |
20 | static pcre2_general_context *pcre2_global_context; | |
21 | ||
22 | static void *pcre2_malloc(PCRE2_SIZE size, MAYBE_UNUSED void *memory_data) | |
23 | { | |
24 | return malloc(size); | |
25 | } | |
26 | ||
27 | static void pcre2_free(void *pointer, MAYBE_UNUSED void *memory_data) | |
28 | { | |
867fc7f3 | 29 | free(pointer); |
513f2b0b CMAB |
30 | } |
31 | #endif | |
32 | ||
fa151dc5 NTND |
33 | static const char *color_grep_slots[] = { |
34 | [GREP_COLOR_CONTEXT] = "context", | |
35 | [GREP_COLOR_FILENAME] = "filename", | |
36 | [GREP_COLOR_FUNCTION] = "function", | |
37 | [GREP_COLOR_LINENO] = "lineNumber", | |
d036d667 | 38 | [GREP_COLOR_COLUMNNO] = "column", |
fa151dc5 NTND |
39 | [GREP_COLOR_MATCH_CONTEXT] = "matchContext", |
40 | [GREP_COLOR_MATCH_SELECTED] = "matchSelected", | |
41 | [GREP_COLOR_SELECTED] = "selected", | |
42 | [GREP_COLOR_SEP] = "separator", | |
43 | }; | |
44 | ||
379642bc BW |
45 | static void std_output(struct grep_opt *opt, const void *buf, size_t size) |
46 | { | |
47 | fwrite(buf, size, 1, stdout); | |
48 | } | |
49 | ||
75e5e9c3 SB |
50 | static void color_set(char *dst, const char *color_bytes) |
51 | { | |
52 | xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes); | |
53 | } | |
54 | ||
7687a054 JH |
55 | /* |
56 | * Initialize the grep_defaults template with hardcoded defaults. | |
57 | * We could let the compiler do this, but without C99 initializers | |
58 | * the code gets unwieldy and unreadable, so... | |
59 | */ | |
38bbc2ea | 60 | void init_grep_defaults(struct repository *repo) |
7687a054 JH |
61 | { |
62 | struct grep_opt *opt = &grep_defaults; | |
918d4e1c JH |
63 | static int run_once; |
64 | ||
65 | if (run_once) | |
66 | return; | |
67 | run_once++; | |
7687a054 JH |
68 | |
69 | memset(opt, 0, sizeof(*opt)); | |
38bbc2ea | 70 | opt->repo = repo; |
7687a054 JH |
71 | opt->relative = 1; |
72 | opt->pathname = 1; | |
7687a054 JH |
73 | opt->max_depth = -1; |
74 | opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED; | |
fa151dc5 NTND |
75 | color_set(opt->colors[GREP_COLOR_CONTEXT], ""); |
76 | color_set(opt->colors[GREP_COLOR_FILENAME], ""); | |
77 | color_set(opt->colors[GREP_COLOR_FUNCTION], ""); | |
78 | color_set(opt->colors[GREP_COLOR_LINENO], ""); | |
d036d667 | 79 | color_set(opt->colors[GREP_COLOR_COLUMNNO], ""); |
fa151dc5 NTND |
80 | color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED); |
81 | color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED); | |
82 | color_set(opt->colors[GREP_COLOR_SELECTED], ""); | |
83 | color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN); | |
9d8db06e | 84 | opt->only_matching = 0; |
7687a054 | 85 | opt->color = -1; |
379642bc | 86 | opt->output = std_output; |
7687a054 JH |
87 | } |
88 | ||
89 | static int parse_pattern_type_arg(const char *opt, const char *arg) | |
90 | { | |
91 | if (!strcmp(arg, "default")) | |
92 | return GREP_PATTERN_TYPE_UNSPECIFIED; | |
93 | else if (!strcmp(arg, "basic")) | |
94 | return GREP_PATTERN_TYPE_BRE; | |
95 | else if (!strcmp(arg, "extended")) | |
96 | return GREP_PATTERN_TYPE_ERE; | |
97 | else if (!strcmp(arg, "fixed")) | |
98 | return GREP_PATTERN_TYPE_FIXED; | |
99 | else if (!strcmp(arg, "perl")) | |
100 | return GREP_PATTERN_TYPE_PCRE; | |
101 | die("bad %s argument: %s", opt, arg); | |
102 | } | |
103 | ||
3ac68a93 NTND |
104 | define_list_config_array_extra(color_grep_slots, {"match"}); |
105 | ||
7687a054 JH |
106 | /* |
107 | * Read the configuration file once and store it in | |
108 | * the grep_defaults template. | |
109 | */ | |
110 | int grep_config(const char *var, const char *value, void *cb) | |
111 | { | |
112 | struct grep_opt *opt = &grep_defaults; | |
fa151dc5 | 113 | const char *slot; |
7687a054 JH |
114 | |
115 | if (userdiff_config(var, value) < 0) | |
116 | return -1; | |
117 | ||
118 | if (!strcmp(var, "grep.extendedregexp")) { | |
c7e38551 | 119 | opt->extended_regexp_option = git_config_bool(var, value); |
7687a054 JH |
120 | return 0; |
121 | } | |
122 | ||
123 | if (!strcmp(var, "grep.patterntype")) { | |
124 | opt->pattern_type_option = parse_pattern_type_arg(var, value); | |
125 | return 0; | |
126 | } | |
127 | ||
128 | if (!strcmp(var, "grep.linenumber")) { | |
129 | opt->linenum = git_config_bool(var, value); | |
130 | return 0; | |
131 | } | |
6653fec3 TB |
132 | if (!strcmp(var, "grep.column")) { |
133 | opt->columnnum = git_config_bool(var, value); | |
134 | return 0; | |
135 | } | |
7687a054 | 136 | |
6453f7b3 AS |
137 | if (!strcmp(var, "grep.fullname")) { |
138 | opt->relative = !git_config_bool(var, value); | |
139 | return 0; | |
140 | } | |
141 | ||
7687a054 JH |
142 | if (!strcmp(var, "color.grep")) |
143 | opt->color = git_config_colorbool(var, value); | |
fa151dc5 NTND |
144 | if (!strcmp(var, "color.grep.match")) { |
145 | if (grep_config("color.grep.matchcontext", value, cb) < 0) | |
146 | return -1; | |
147 | if (grep_config("color.grep.matchselected", value, cb) < 0) | |
148 | return -1; | |
149 | } else if (skip_prefix(var, "color.grep.", &slot)) { | |
150 | int i = LOOKUP_CONFIG(color_grep_slots, slot); | |
151 | char *color; | |
152 | ||
153 | if (i < 0) | |
154 | return -1; | |
155 | color = opt->colors[i]; | |
7687a054 JH |
156 | if (!value) |
157 | return config_error_nonbool(var); | |
f6c5a296 | 158 | return color_parse(value, color); |
7687a054 JH |
159 | } |
160 | return 0; | |
161 | } | |
162 | ||
163 | /* | |
164 | * Initialize one instance of grep_opt and copy the | |
165 | * default values from the template we read the configuration | |
166 | * information in an earlier call to git_config(grep_config). | |
57d46604 CMAB |
167 | * |
168 | * If using PCRE, make sure that the library is configured | |
169 | * to use the same allocator as Git (e.g. nedmalloc on Windows). | |
513f2b0b CMAB |
170 | * |
171 | * Any allocated memory needs to be released in grep_destroy(). | |
7687a054 | 172 | */ |
38bbc2ea | 173 | void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix) |
7687a054 JH |
174 | { |
175 | struct grep_opt *def = &grep_defaults; | |
fa151dc5 | 176 | int i; |
7687a054 | 177 | |
513f2b0b CMAB |
178 | #if defined(USE_LIBPCRE2) |
179 | if (!pcre2_global_context) | |
180 | pcre2_global_context = pcre2_general_context_create( | |
181 | pcre2_malloc, pcre2_free, NULL); | |
182 | #endif | |
183 | ||
57d46604 CMAB |
184 | #ifdef USE_LIBPCRE1 |
185 | pcre_malloc = malloc; | |
186 | pcre_free = free; | |
187 | #endif | |
188 | ||
7687a054 | 189 | memset(opt, 0, sizeof(*opt)); |
38bbc2ea | 190 | opt->repo = repo; |
7687a054 JH |
191 | opt->prefix = prefix; |
192 | opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0; | |
193 | opt->pattern_tail = &opt->pattern_list; | |
194 | opt->header_tail = &opt->header_list; | |
195 | ||
9d8db06e | 196 | opt->only_matching = def->only_matching; |
7687a054 JH |
197 | opt->color = def->color; |
198 | opt->extended_regexp_option = def->extended_regexp_option; | |
199 | opt->pattern_type_option = def->pattern_type_option; | |
200 | opt->linenum = def->linenum; | |
017c0fcf | 201 | opt->columnnum = def->columnnum; |
7687a054 JH |
202 | opt->max_depth = def->max_depth; |
203 | opt->pathname = def->pathname; | |
7687a054 | 204 | opt->relative = def->relative; |
379642bc | 205 | opt->output = def->output; |
7687a054 | 206 | |
fa151dc5 NTND |
207 | for (i = 0; i < NR_GREP_COLORS; i++) |
208 | color_set(opt->colors[i], def->colors[i]); | |
7687a054 | 209 | } |
07a7d656 | 210 | |
513f2b0b CMAB |
211 | void grep_destroy(void) |
212 | { | |
213 | #ifdef USE_LIBPCRE2 | |
214 | pcre2_general_context_free(pcre2_global_context); | |
215 | #endif | |
216 | } | |
217 | ||
8465541e | 218 | static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt) |
c5c31d33 | 219 | { |
07a3d411 ÆAB |
220 | /* |
221 | * When committing to the pattern type by setting the relevant | |
222 | * fields in grep_opt it's generally not necessary to zero out | |
223 | * the fields we're not choosing, since they won't have been | |
224 | * set by anything. The extended_regexp_option field is the | |
225 | * only exception to this. | |
226 | * | |
227 | * This is because in the process of parsing grep.patternType | |
228 | * & grep.extendedRegexp we set opt->pattern_type_option and | |
229 | * opt->extended_regexp_option, respectively. We then | |
230 | * internally use opt->extended_regexp_option to see if we're | |
231 | * compiling an ERE. It must be unset if that's not actually | |
232 | * the case. | |
233 | */ | |
234 | if (pattern_type != GREP_PATTERN_TYPE_ERE && | |
235 | opt->extended_regexp_option) | |
236 | opt->extended_regexp_option = 0; | |
237 | ||
c5c31d33 JH |
238 | switch (pattern_type) { |
239 | case GREP_PATTERN_TYPE_UNSPECIFIED: | |
240 | /* fall through */ | |
241 | ||
242 | case GREP_PATTERN_TYPE_BRE: | |
c5c31d33 JH |
243 | break; |
244 | ||
245 | case GREP_PATTERN_TYPE_ERE: | |
07a3d411 | 246 | opt->extended_regexp_option = 1; |
c5c31d33 JH |
247 | break; |
248 | ||
249 | case GREP_PATTERN_TYPE_FIXED: | |
250 | opt->fixed = 1; | |
c5c31d33 JH |
251 | break; |
252 | ||
253 | case GREP_PATTERN_TYPE_PCRE: | |
94da9193 | 254 | #ifdef USE_LIBPCRE2 |
94da9193 ÆAB |
255 | opt->pcre2 = 1; |
256 | #else | |
257 | /* | |
258 | * It's important that pcre1 always be assigned to | |
259 | * even when there's no USE_LIBPCRE* defined. We still | |
260 | * call the PCRE stub function, it just dies with | |
261 | * "cannot use Perl-compatible regexes[...]". | |
262 | */ | |
6d4b5747 | 263 | opt->pcre1 = 1; |
94da9193 | 264 | #endif |
c5c31d33 JH |
265 | break; |
266 | } | |
267 | } | |
268 | ||
8465541e JH |
269 | void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt) |
270 | { | |
271 | if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED) | |
272 | grep_set_pattern_type_option(pattern_type, opt); | |
273 | else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED) | |
274 | grep_set_pattern_type_option(opt->pattern_type_option, opt); | |
275 | else if (opt->extended_regexp_option) | |
07a3d411 ÆAB |
276 | /* |
277 | * This branch *must* happen after setting from the | |
278 | * opt->pattern_type_option above, we don't want | |
279 | * grep.extendedRegexp to override grep.patternType! | |
280 | */ | |
8465541e JH |
281 | grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt); |
282 | } | |
283 | ||
fc456751 RS |
284 | static struct grep_pat *create_grep_pat(const char *pat, size_t patlen, |
285 | const char *origin, int no, | |
286 | enum grep_pat_token t, | |
287 | enum grep_header_field field) | |
a4d7d2c6 JH |
288 | { |
289 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
526a858a | 290 | p->pattern = xmemdupz(pat, patlen); |
fc456751 RS |
291 | p->patternlen = patlen; |
292 | p->origin = origin; | |
293 | p->no = no; | |
294 | p->token = t; | |
a4d7d2c6 | 295 | p->field = field; |
fc456751 RS |
296 | return p; |
297 | } | |
298 | ||
2b3873ff RS |
299 | static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p) |
300 | { | |
301 | **tail = p; | |
302 | *tail = &p->next; | |
a4d7d2c6 | 303 | p->next = NULL; |
526a858a RS |
304 | |
305 | switch (p->token) { | |
306 | case GREP_PATTERN: /* atom */ | |
307 | case GREP_PATTERN_HEAD: | |
308 | case GREP_PATTERN_BODY: | |
309 | for (;;) { | |
310 | struct grep_pat *new_pat; | |
311 | size_t len = 0; | |
312 | char *cp = p->pattern + p->patternlen, *nl = NULL; | |
313 | while (++len <= p->patternlen) { | |
314 | if (*(--cp) == '\n') { | |
315 | nl = cp; | |
316 | break; | |
317 | } | |
318 | } | |
319 | if (!nl) | |
320 | break; | |
321 | new_pat = create_grep_pat(nl + 1, len - 1, p->origin, | |
322 | p->no, p->token, p->field); | |
323 | new_pat->next = p->next; | |
324 | if (!p->next) | |
325 | *tail = &new_pat->next; | |
326 | p->next = new_pat; | |
327 | *nl = '\0'; | |
328 | p->patternlen -= len; | |
329 | } | |
330 | break; | |
331 | default: | |
332 | break; | |
333 | } | |
2b3873ff RS |
334 | } |
335 | ||
fc456751 RS |
336 | void append_header_grep_pattern(struct grep_opt *opt, |
337 | enum grep_header_field field, const char *pat) | |
338 | { | |
339 | struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0, | |
340 | GREP_PATTERN_HEAD, field); | |
baa6378f JH |
341 | if (field == GREP_HEADER_REFLOG) |
342 | opt->use_reflog_filter = 1; | |
2b3873ff | 343 | do_append_grep_pat(&opt->header_tail, p); |
a4d7d2c6 JH |
344 | } |
345 | ||
83b5d2f5 JH |
346 | void append_grep_pattern(struct grep_opt *opt, const char *pat, |
347 | const char *origin, int no, enum grep_pat_token t) | |
ed40a095 RS |
348 | { |
349 | append_grep_pat(opt, pat, strlen(pat), origin, no, t); | |
350 | } | |
351 | ||
352 | void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen, | |
353 | const char *origin, int no, enum grep_pat_token t) | |
83b5d2f5 | 354 | { |
fc456751 | 355 | struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0); |
2b3873ff | 356 | do_append_grep_pat(&opt->pattern_tail, p); |
83b5d2f5 JH |
357 | } |
358 | ||
5b594f45 FK |
359 | struct grep_opt *grep_opt_dup(const struct grep_opt *opt) |
360 | { | |
361 | struct grep_pat *pat; | |
362 | struct grep_opt *ret = xmalloc(sizeof(struct grep_opt)); | |
363 | *ret = *opt; | |
364 | ||
365 | ret->pattern_list = NULL; | |
366 | ret->pattern_tail = &ret->pattern_list; | |
367 | ||
368 | for(pat = opt->pattern_list; pat != NULL; pat = pat->next) | |
369 | { | |
370 | if(pat->token == GREP_PATTERN_HEAD) | |
371 | append_header_grep_pattern(ret, pat->field, | |
372 | pat->pattern); | |
373 | else | |
ed40a095 RS |
374 | append_grep_pat(ret, pat->pattern, pat->patternlen, |
375 | pat->origin, pat->no, pat->token); | |
5b594f45 FK |
376 | } |
377 | ||
378 | return ret; | |
379 | } | |
380 | ||
a30c148a MK |
381 | static NORETURN void compile_regexp_failed(const struct grep_pat *p, |
382 | const char *error) | |
383 | { | |
384 | char where[1024]; | |
385 | ||
386 | if (p->no) | |
19bdd3e7 | 387 | xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no); |
a30c148a | 388 | else if (p->origin) |
19bdd3e7 | 389 | xsnprintf(where, sizeof(where), "%s, ", p->origin); |
a30c148a MK |
390 | else |
391 | where[0] = 0; | |
392 | ||
393 | die("%s'%s': %s", where, p->pattern, error); | |
394 | } | |
395 | ||
543f1c0c ÆAB |
396 | static int is_fixed(const char *s, size_t len) |
397 | { | |
398 | size_t i; | |
399 | ||
400 | for (i = 0; i < len; i++) { | |
401 | if (is_regex_special(s[i])) | |
402 | return 0; | |
403 | } | |
404 | ||
405 | return 1; | |
406 | } | |
407 | ||
3485bea1 | 408 | #ifdef USE_LIBPCRE1 |
6d4b5747 | 409 | static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt) |
63e7e9d8 MK |
410 | { |
411 | const char *error; | |
412 | int erroffset; | |
fba4f125 | 413 | int options = PCRE_MULTILINE; |
ff61681b | 414 | int study_options = 0; |
63e7e9d8 | 415 | |
9d9babb8 | 416 | if (opt->ignore_case) { |
44570188 | 417 | if (!opt->ignore_locale && has_non_ascii(p->pattern)) |
6d4b5747 | 418 | p->pcre1_tables = pcre_maketables(); |
63e7e9d8 | 419 | options |= PCRE_CASELESS; |
9d9babb8 | 420 | } |
44570188 | 421 | if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern)) |
18547aac | 422 | options |= PCRE_UTF8; |
63e7e9d8 | 423 | |
6d4b5747 ÆAB |
424 | p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset, |
425 | p->pcre1_tables); | |
426 | if (!p->pcre1_regexp) | |
63e7e9d8 MK |
427 | compile_regexp_failed(p, error); |
428 | ||
ff61681b | 429 | #if defined(PCRE_CONFIG_JIT) && !defined(NO_LIBPCRE1_JIT) |
fbaceaac | 430 | pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on); |
c581e4a7 BB |
431 | if (opt->debug) |
432 | fprintf(stderr, "pcre1_jit_on=%d\n", p->pcre1_jit_on); | |
ff61681b CMAB |
433 | |
434 | if (p->pcre1_jit_on) | |
435 | study_options = PCRE_STUDY_JIT_COMPILE; | |
fbaceaac | 436 | #endif |
ff61681b CMAB |
437 | |
438 | p->pcre1_extra_info = pcre_study(p->pcre1_regexp, study_options, &error); | |
439 | if (!p->pcre1_extra_info && error) | |
440 | die("%s", error); | |
63e7e9d8 MK |
441 | } |
442 | ||
6d4b5747 | 443 | static int pcre1match(struct grep_pat *p, const char *line, const char *eol, |
63e7e9d8 MK |
444 | regmatch_t *match, int eflags) |
445 | { | |
ad7c543e | 446 | int ovector[30], ret, flags = PCRE_NO_UTF8_CHECK; |
63e7e9d8 MK |
447 | |
448 | if (eflags & REG_NOTBOL) | |
449 | flags |= PCRE_NOTBOL; | |
450 | ||
685668fa ÆAB |
451 | ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line, |
452 | eol - line, 0, flags, ovector, | |
453 | ARRAY_SIZE(ovector)); | |
fbaceaac | 454 | |
63e7e9d8 MK |
455 | if (ret < 0 && ret != PCRE_ERROR_NOMATCH) |
456 | die("pcre_exec failed with error code %d", ret); | |
457 | if (ret > 0) { | |
458 | ret = 0; | |
459 | match->rm_so = ovector[0]; | |
460 | match->rm_eo = ovector[1]; | |
461 | } | |
462 | ||
463 | return ret; | |
464 | } | |
465 | ||
6d4b5747 | 466 | static void free_pcre1_regexp(struct grep_pat *p) |
63e7e9d8 | 467 | { |
6d4b5747 | 468 | pcre_free(p->pcre1_regexp); |
ff61681b | 469 | #ifdef PCRE_CONFIG_JIT |
685668fa | 470 | if (p->pcre1_jit_on) |
fbaceaac | 471 | pcre_free_study(p->pcre1_extra_info); |
685668fa | 472 | else |
fbaceaac | 473 | #endif |
fbaceaac | 474 | pcre_free(p->pcre1_extra_info); |
6d4b5747 | 475 | pcre_free((void *)p->pcre1_tables); |
63e7e9d8 | 476 | } |
3485bea1 | 477 | #else /* !USE_LIBPCRE1 */ |
6d4b5747 | 478 | static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt) |
63e7e9d8 MK |
479 | { |
480 | die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE"); | |
481 | } | |
482 | ||
6d4b5747 | 483 | static int pcre1match(struct grep_pat *p, const char *line, const char *eol, |
63e7e9d8 MK |
484 | regmatch_t *match, int eflags) |
485 | { | |
486 | return 1; | |
487 | } | |
488 | ||
6d4b5747 | 489 | static void free_pcre1_regexp(struct grep_pat *p) |
63e7e9d8 MK |
490 | { |
491 | } | |
3485bea1 | 492 | #endif /* !USE_LIBPCRE1 */ |
63e7e9d8 | 493 | |
94da9193 ÆAB |
494 | #ifdef USE_LIBPCRE2 |
495 | static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt) | |
496 | { | |
497 | int error; | |
498 | PCRE2_UCHAR errbuf[256]; | |
499 | PCRE2_SIZE erroffset; | |
500 | int options = PCRE2_MULTILINE; | |
94da9193 | 501 | int jitret; |
a25b9085 ÆAB |
502 | int patinforet; |
503 | size_t jitsizearg; | |
94da9193 ÆAB |
504 | |
505 | assert(opt->pcre2); | |
506 | ||
507 | p->pcre2_compile_context = NULL; | |
508 | ||
513f2b0b | 509 | /* pcre2_global_context is initialized in append_grep_pattern */ |
94da9193 | 510 | if (opt->ignore_case) { |
44570188 | 511 | if (!opt->ignore_locale && has_non_ascii(p->pattern)) { |
513f2b0b CMAB |
512 | if (!pcre2_global_context) |
513 | BUG("pcre2_global_context uninitialized"); | |
10da030a | 514 | p->pcre2_tables = pcre2_maketables(pcre2_global_context); |
94da9193 | 515 | p->pcre2_compile_context = pcre2_compile_context_create(NULL); |
10da030a CMAB |
516 | pcre2_set_character_tables(p->pcre2_compile_context, |
517 | p->pcre2_tables); | |
94da9193 ÆAB |
518 | } |
519 | options |= PCRE2_CASELESS; | |
520 | } | |
870eea81 ÆAB |
521 | if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern) && |
522 | !(!opt->ignore_case && (p->fixed || p->is_fixed))) | |
94da9193 ÆAB |
523 | options |= PCRE2_UTF; |
524 | ||
525 | p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern, | |
526 | p->patternlen, options, &error, &erroffset, | |
527 | p->pcre2_compile_context); | |
528 | ||
529 | if (p->pcre2_pattern) { | |
530 | p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL); | |
531 | if (!p->pcre2_match_data) | |
532 | die("Couldn't allocate PCRE2 match data"); | |
533 | } else { | |
534 | pcre2_get_error_message(error, errbuf, sizeof(errbuf)); | |
535 | compile_regexp_failed(p, (const char *)&errbuf); | |
536 | } | |
537 | ||
538 | pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on); | |
c581e4a7 BB |
539 | if (opt->debug) |
540 | fprintf(stderr, "pcre2_jit_on=%d\n", p->pcre2_jit_on); | |
04bef50c | 541 | if (p->pcre2_jit_on) { |
94da9193 ÆAB |
542 | jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE); |
543 | if (jitret) | |
544 | die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret); | |
a25b9085 ÆAB |
545 | |
546 | /* | |
547 | * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just | |
548 | * tells us whether the library itself supports JIT, | |
549 | * but to see whether we're going to be actually using | |
550 | * JIT we need to extract PCRE2_INFO_JITSIZE from the | |
551 | * pattern *after* we do pcre2_jit_compile() above. | |
552 | * | |
553 | * This is because if the pattern contains the | |
554 | * (*NO_JIT) verb (see pcre2syntax(3)) | |
555 | * pcre2_jit_compile() will exit early with 0. If we | |
556 | * then proceed to call pcre2_jit_match() further down | |
557 | * the line instead of pcre2_match() we'll either | |
558 | * segfault (pre PCRE 10.31) or run into a fatal error | |
559 | * (post PCRE2 10.31) | |
560 | */ | |
561 | patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg); | |
562 | if (patinforet) | |
563 | BUG("pcre2_pattern_info() failed: %d", patinforet); | |
564 | if (jitsizearg == 0) { | |
565 | p->pcre2_jit_on = 0; | |
c581e4a7 BB |
566 | if (opt->debug) |
567 | fprintf(stderr, "pcre2_jit_on=%d: (*NO_JIT) in regex\n", | |
568 | p->pcre2_jit_on); | |
a25b9085 ÆAB |
569 | return; |
570 | } | |
94da9193 ÆAB |
571 | } |
572 | } | |
573 | ||
574 | static int pcre2match(struct grep_pat *p, const char *line, const char *eol, | |
575 | regmatch_t *match, int eflags) | |
576 | { | |
577 | int ret, flags = 0; | |
578 | PCRE2_SIZE *ovector; | |
579 | PCRE2_UCHAR errbuf[256]; | |
580 | ||
581 | if (eflags & REG_NOTBOL) | |
582 | flags |= PCRE2_NOTBOL; | |
583 | ||
584 | if (p->pcre2_jit_on) | |
585 | ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line, | |
586 | eol - line, 0, flags, p->pcre2_match_data, | |
587 | NULL); | |
588 | else | |
589 | ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line, | |
590 | eol - line, 0, flags, p->pcre2_match_data, | |
591 | NULL); | |
592 | ||
593 | if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) { | |
594 | pcre2_get_error_message(ret, errbuf, sizeof(errbuf)); | |
595 | die("%s failed with error code %d: %s", | |
596 | (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret, | |
597 | errbuf); | |
598 | } | |
599 | if (ret > 0) { | |
600 | ovector = pcre2_get_ovector_pointer(p->pcre2_match_data); | |
601 | ret = 0; | |
602 | match->rm_so = (int)ovector[0]; | |
603 | match->rm_eo = (int)ovector[1]; | |
604 | } | |
605 | ||
606 | return ret; | |
607 | } | |
608 | ||
609 | static void free_pcre2_pattern(struct grep_pat *p) | |
610 | { | |
611 | pcre2_compile_context_free(p->pcre2_compile_context); | |
612 | pcre2_code_free(p->pcre2_pattern); | |
613 | pcre2_match_data_free(p->pcre2_match_data); | |
10da030a | 614 | free((void *)p->pcre2_tables); |
94da9193 ÆAB |
615 | } |
616 | #else /* !USE_LIBPCRE2 */ | |
617 | static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt) | |
618 | { | |
619 | /* | |
620 | * Unreachable until USE_LIBPCRE2 becomes synonymous with | |
621 | * USE_LIBPCRE. See the sibling comment in | |
622 | * grep_set_pattern_type_option(). | |
623 | */ | |
624 | die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE"); | |
625 | } | |
626 | ||
627 | static int pcre2match(struct grep_pat *p, const char *line, const char *eol, | |
628 | regmatch_t *match, int eflags) | |
629 | { | |
630 | return 1; | |
631 | } | |
632 | ||
633 | static void free_pcre2_pattern(struct grep_pat *p) | |
634 | { | |
635 | } | |
94da9193 | 636 | |
793dc676 NTND |
637 | static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt) |
638 | { | |
639 | struct strbuf sb = STRBUF_INIT; | |
640 | int err; | |
1ceababc | 641 | int regflags = 0; |
793dc676 NTND |
642 | |
643 | basic_regex_quote_buf(&sb, p->pattern); | |
793dc676 NTND |
644 | if (opt->ignore_case) |
645 | regflags |= REG_ICASE; | |
646 | err = regcomp(&p->regexp, sb.buf, regflags); | |
647 | if (opt->debug) | |
648 | fprintf(stderr, "fixed %s\n", sb.buf); | |
649 | strbuf_release(&sb); | |
650 | if (err) { | |
651 | char errbuf[1024]; | |
652 | regerror(err, &p->regexp, errbuf, sizeof(errbuf)); | |
793dc676 NTND |
653 | compile_regexp_failed(p, errbuf); |
654 | } | |
655 | } | |
b65abcaf | 656 | #endif /* !USE_LIBPCRE2 */ |
793dc676 | 657 | |
83b5d2f5 JH |
658 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) |
659 | { | |
c822255c | 660 | int err; |
07a3d411 | 661 | int regflags = REG_NEWLINE; |
c822255c | 662 | |
d7eb527d | 663 | p->word_regexp = opt->word_regexp; |
5183bf67 | 664 | p->ignore_case = opt->ignore_case; |
48de2a76 | 665 | p->fixed = opt->fixed; |
d7eb527d | 666 | |
45d1f37c ÆAB |
667 | if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2) |
668 | die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2")); | |
669 | ||
09872f64 | 670 | p->is_fixed = is_fixed(p->pattern, p->patternlen); |
8a599983 ÆAB |
671 | #ifdef USE_LIBPCRE2 |
672 | if (!p->fixed && !p->is_fixed) { | |
673 | const char *no_jit = "(*NO_JIT)"; | |
674 | const int no_jit_len = strlen(no_jit); | |
675 | if (starts_with(p->pattern, no_jit) && | |
676 | is_fixed(p->pattern + no_jit_len, | |
677 | p->patternlen - no_jit_len)) | |
678 | p->is_fixed = 1; | |
679 | } | |
680 | #endif | |
09872f64 | 681 | if (p->fixed || p->is_fixed) { |
b65abcaf ÆAB |
682 | #ifdef USE_LIBPCRE2 |
683 | opt->pcre2 = 1; | |
09872f64 | 684 | if (p->is_fixed) { |
b65abcaf ÆAB |
685 | compile_pcre2_pattern(p, opt); |
686 | } else { | |
687 | /* | |
688 | * E.g. t7811-grep-open.sh relies on the | |
689 | * pattern being restored. | |
690 | */ | |
691 | char *old_pattern = p->pattern; | |
692 | size_t old_patternlen = p->patternlen; | |
693 | struct strbuf sb = STRBUF_INIT; | |
694 | ||
695 | /* | |
696 | * There is the PCRE2_LITERAL flag, but it's | |
697 | * only in PCRE v2 10.30 and later. Needing to | |
698 | * ifdef our way around that and dealing with | |
699 | * it + PCRE2_MULTILINE being an error is more | |
700 | * complex than just quoting this ourselves. | |
701 | */ | |
702 | strbuf_add(&sb, "\\Q", 2); | |
703 | strbuf_add(&sb, p->pattern, p->patternlen); | |
704 | strbuf_add(&sb, "\\E", 2); | |
705 | ||
706 | p->pattern = sb.buf; | |
707 | p->patternlen = sb.len; | |
708 | compile_pcre2_pattern(p, opt); | |
709 | p->pattern = old_pattern; | |
710 | p->patternlen = old_patternlen; | |
711 | strbuf_release(&sb); | |
712 | } | |
713 | #else /* !USE_LIBPCRE2 */ | |
793dc676 | 714 | compile_fixed_regexp(p, opt); |
b65abcaf | 715 | #endif /* !USE_LIBPCRE2 */ |
793dc676 | 716 | return; |
9eceddee | 717 | } |
c822255c | 718 | |
94da9193 ÆAB |
719 | if (opt->pcre2) { |
720 | compile_pcre2_pattern(p, opt); | |
721 | return; | |
722 | } | |
723 | ||
6d4b5747 ÆAB |
724 | if (opt->pcre1) { |
725 | compile_pcre1_regexp(p, opt); | |
63e7e9d8 MK |
726 | return; |
727 | } | |
728 | ||
07a3d411 ÆAB |
729 | if (p->ignore_case) |
730 | regflags |= REG_ICASE; | |
731 | if (opt->extended_regexp_option) | |
732 | regflags |= REG_EXTENDED; | |
733 | err = regcomp(&p->regexp, p->pattern, regflags); | |
83b5d2f5 JH |
734 | if (err) { |
735 | char errbuf[1024]; | |
83b5d2f5 | 736 | regerror(err, &p->regexp, errbuf, 1024); |
a30c148a | 737 | compile_regexp_failed(p, errbuf); |
83b5d2f5 JH |
738 | } |
739 | } | |
740 | ||
0ab7befa | 741 | static struct grep_expr *compile_pattern_or(struct grep_pat **); |
83b5d2f5 JH |
742 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) |
743 | { | |
744 | struct grep_pat *p; | |
745 | struct grep_expr *x; | |
746 | ||
747 | p = *list; | |
c922b01f LT |
748 | if (!p) |
749 | return NULL; | |
83b5d2f5 JH |
750 | switch (p->token) { |
751 | case GREP_PATTERN: /* atom */ | |
480c1ca6 JH |
752 | case GREP_PATTERN_HEAD: |
753 | case GREP_PATTERN_BODY: | |
83b5d2f5 JH |
754 | x = xcalloc(1, sizeof (struct grep_expr)); |
755 | x->node = GREP_NODE_ATOM; | |
756 | x->u.atom = p; | |
757 | *list = p->next; | |
758 | return x; | |
759 | case GREP_OPEN_PAREN: | |
760 | *list = p->next; | |
0ab7befa | 761 | x = compile_pattern_or(list); |
83b5d2f5 JH |
762 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) |
763 | die("unmatched parenthesis"); | |
764 | *list = (*list)->next; | |
765 | return x; | |
766 | default: | |
767 | return NULL; | |
768 | } | |
769 | } | |
770 | ||
771 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
772 | { | |
773 | struct grep_pat *p; | |
774 | struct grep_expr *x; | |
775 | ||
776 | p = *list; | |
c922b01f LT |
777 | if (!p) |
778 | return NULL; | |
83b5d2f5 JH |
779 | switch (p->token) { |
780 | case GREP_NOT: | |
781 | if (!p->next) | |
782 | die("--not not followed by pattern expression"); | |
783 | *list = p->next; | |
784 | x = xcalloc(1, sizeof (struct grep_expr)); | |
785 | x->node = GREP_NODE_NOT; | |
786 | x->u.unary = compile_pattern_not(list); | |
787 | if (!x->u.unary) | |
788 | die("--not followed by non pattern expression"); | |
789 | return x; | |
790 | default: | |
791 | return compile_pattern_atom(list); | |
792 | } | |
793 | } | |
794 | ||
795 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
796 | { | |
797 | struct grep_pat *p; | |
798 | struct grep_expr *x, *y, *z; | |
799 | ||
800 | x = compile_pattern_not(list); | |
801 | p = *list; | |
802 | if (p && p->token == GREP_AND) { | |
803 | if (!p->next) | |
804 | die("--and not followed by pattern expression"); | |
805 | *list = p->next; | |
806 | y = compile_pattern_and(list); | |
807 | if (!y) | |
808 | die("--and not followed by pattern expression"); | |
809 | z = xcalloc(1, sizeof (struct grep_expr)); | |
810 | z->node = GREP_NODE_AND; | |
811 | z->u.binary.left = x; | |
812 | z->u.binary.right = y; | |
813 | return z; | |
814 | } | |
815 | return x; | |
816 | } | |
817 | ||
818 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
819 | { | |
820 | struct grep_pat *p; | |
821 | struct grep_expr *x, *y, *z; | |
822 | ||
823 | x = compile_pattern_and(list); | |
824 | p = *list; | |
825 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
826 | y = compile_pattern_or(list); | |
827 | if (!y) | |
828 | die("not a pattern expression %s", p->pattern); | |
829 | z = xcalloc(1, sizeof (struct grep_expr)); | |
830 | z->node = GREP_NODE_OR; | |
831 | z->u.binary.left = x; | |
832 | z->u.binary.right = y; | |
833 | return z; | |
834 | } | |
835 | return x; | |
836 | } | |
837 | ||
838 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
839 | { | |
840 | return compile_pattern_or(list); | |
841 | } | |
842 | ||
17bf35a3 JH |
843 | static void indent(int in) |
844 | { | |
845 | while (in-- > 0) | |
846 | fputc(' ', stderr); | |
847 | } | |
848 | ||
849 | static void dump_grep_pat(struct grep_pat *p) | |
850 | { | |
851 | switch (p->token) { | |
852 | case GREP_AND: fprintf(stderr, "*and*"); break; | |
853 | case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break; | |
854 | case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break; | |
855 | case GREP_NOT: fprintf(stderr, "*not*"); break; | |
856 | case GREP_OR: fprintf(stderr, "*or*"); break; | |
857 | ||
858 | case GREP_PATTERN: fprintf(stderr, "pattern"); break; | |
859 | case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break; | |
860 | case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break; | |
861 | } | |
862 | ||
863 | switch (p->token) { | |
864 | default: break; | |
865 | case GREP_PATTERN_HEAD: | |
866 | fprintf(stderr, "<head %d>", p->field); break; | |
867 | case GREP_PATTERN_BODY: | |
868 | fprintf(stderr, "<body>"); break; | |
869 | } | |
870 | switch (p->token) { | |
871 | default: break; | |
872 | case GREP_PATTERN_HEAD: | |
873 | case GREP_PATTERN_BODY: | |
874 | case GREP_PATTERN: | |
875 | fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern); | |
876 | break; | |
877 | } | |
878 | fputc('\n', stderr); | |
879 | } | |
880 | ||
881 | static void dump_grep_expression_1(struct grep_expr *x, int in) | |
882 | { | |
883 | indent(in); | |
884 | switch (x->node) { | |
885 | case GREP_NODE_TRUE: | |
886 | fprintf(stderr, "true\n"); | |
887 | break; | |
888 | case GREP_NODE_ATOM: | |
889 | dump_grep_pat(x->u.atom); | |
890 | break; | |
891 | case GREP_NODE_NOT: | |
892 | fprintf(stderr, "(not\n"); | |
893 | dump_grep_expression_1(x->u.unary, in+1); | |
894 | indent(in); | |
895 | fprintf(stderr, ")\n"); | |
896 | break; | |
897 | case GREP_NODE_AND: | |
898 | fprintf(stderr, "(and\n"); | |
899 | dump_grep_expression_1(x->u.binary.left, in+1); | |
900 | dump_grep_expression_1(x->u.binary.right, in+1); | |
901 | indent(in); | |
902 | fprintf(stderr, ")\n"); | |
903 | break; | |
904 | case GREP_NODE_OR: | |
905 | fprintf(stderr, "(or\n"); | |
906 | dump_grep_expression_1(x->u.binary.left, in+1); | |
907 | dump_grep_expression_1(x->u.binary.right, in+1); | |
908 | indent(in); | |
909 | fprintf(stderr, ")\n"); | |
910 | break; | |
911 | } | |
912 | } | |
913 | ||
07a7d656 | 914 | static void dump_grep_expression(struct grep_opt *opt) |
17bf35a3 JH |
915 | { |
916 | struct grep_expr *x = opt->pattern_expression; | |
917 | ||
918 | if (opt->all_match) | |
919 | fprintf(stderr, "[all-match]\n"); | |
920 | dump_grep_expression_1(x, 0); | |
921 | fflush(NULL); | |
922 | } | |
923 | ||
5aaeb733 JH |
924 | static struct grep_expr *grep_true_expr(void) |
925 | { | |
926 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
927 | z->node = GREP_NODE_TRUE; | |
928 | return z; | |
929 | } | |
930 | ||
931 | static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right) | |
932 | { | |
933 | struct grep_expr *z = xcalloc(1, sizeof(*z)); | |
934 | z->node = GREP_NODE_OR; | |
935 | z->u.binary.left = left; | |
936 | z->u.binary.right = right; | |
937 | return z; | |
938 | } | |
939 | ||
95ce9ce2 | 940 | static struct grep_expr *prep_header_patterns(struct grep_opt *opt) |
83b5d2f5 JH |
941 | { |
942 | struct grep_pat *p; | |
95ce9ce2 | 943 | struct grep_expr *header_expr; |
5aaeb733 JH |
944 | struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]); |
945 | enum grep_header_field fld; | |
83b5d2f5 | 946 | |
95ce9ce2 JH |
947 | if (!opt->header_list) |
948 | return NULL; | |
2385f246 | 949 | |
95ce9ce2 JH |
950 | for (p = opt->header_list; p; p = p->next) { |
951 | if (p->token != GREP_PATTERN_HEAD) | |
033abf97 | 952 | BUG("a non-header pattern in grep header list."); |
3ce3ffb8 AP |
953 | if (p->field < GREP_HEADER_FIELD_MIN || |
954 | GREP_HEADER_FIELD_MAX <= p->field) | |
033abf97 | 955 | BUG("unknown header field %d", p->field); |
95ce9ce2 | 956 | compile_regexp(p, opt); |
80235ba7 | 957 | } |
5aaeb733 JH |
958 | |
959 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) | |
960 | header_group[fld] = NULL; | |
961 | ||
962 | for (p = opt->header_list; p; p = p->next) { | |
963 | struct grep_expr *h; | |
964 | struct grep_pat *pp = p; | |
965 | ||
966 | h = compile_pattern_atom(&pp); | |
967 | if (!h || pp != p->next) | |
033abf97 | 968 | BUG("malformed header expr"); |
5aaeb733 JH |
969 | if (!header_group[p->field]) { |
970 | header_group[p->field] = h; | |
971 | continue; | |
972 | } | |
973 | header_group[p->field] = grep_or_expr(h, header_group[p->field]); | |
974 | } | |
975 | ||
976 | header_expr = NULL; | |
977 | ||
978 | for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) { | |
979 | if (!header_group[fld]) | |
980 | continue; | |
981 | if (!header_expr) | |
982 | header_expr = grep_true_expr(); | |
983 | header_expr = grep_or_expr(header_group[fld], header_expr); | |
984 | } | |
95ce9ce2 JH |
985 | return header_expr; |
986 | } | |
987 | ||
13e4fc7e JH |
988 | static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y) |
989 | { | |
990 | struct grep_expr *z = x; | |
991 | ||
992 | while (x) { | |
993 | assert(x->node == GREP_NODE_OR); | |
994 | if (x->u.binary.right && | |
995 | x->u.binary.right->node == GREP_NODE_TRUE) { | |
996 | x->u.binary.right = y; | |
997 | break; | |
998 | } | |
999 | x = x->u.binary.right; | |
1000 | } | |
1001 | return z; | |
1002 | } | |
1003 | ||
17bf35a3 | 1004 | static void compile_grep_patterns_real(struct grep_opt *opt) |
95ce9ce2 JH |
1005 | { |
1006 | struct grep_pat *p; | |
1007 | struct grep_expr *header_expr = prep_header_patterns(opt); | |
0ab7befa | 1008 | |
83b5d2f5 | 1009 | for (p = opt->pattern_list; p; p = p->next) { |
480c1ca6 JH |
1010 | switch (p->token) { |
1011 | case GREP_PATTERN: /* atom */ | |
1012 | case GREP_PATTERN_HEAD: | |
1013 | case GREP_PATTERN_BODY: | |
c822255c | 1014 | compile_regexp(p, opt); |
480c1ca6 JH |
1015 | break; |
1016 | default: | |
83b5d2f5 | 1017 | opt->extended = 1; |
480c1ca6 JH |
1018 | break; |
1019 | } | |
83b5d2f5 JH |
1020 | } |
1021 | ||
80235ba7 JH |
1022 | if (opt->all_match || header_expr) |
1023 | opt->extended = 1; | |
17bf35a3 | 1024 | else if (!opt->extended && !opt->debug) |
83b5d2f5 JH |
1025 | return; |
1026 | ||
83b5d2f5 | 1027 | p = opt->pattern_list; |
ba150a3f MB |
1028 | if (p) |
1029 | opt->pattern_expression = compile_pattern_expr(&p); | |
83b5d2f5 JH |
1030 | if (p) |
1031 | die("incomplete pattern expression: %s", p->pattern); | |
80235ba7 JH |
1032 | |
1033 | if (!header_expr) | |
1034 | return; | |
1035 | ||
5aaeb733 | 1036 | if (!opt->pattern_expression) |
80235ba7 | 1037 | opt->pattern_expression = header_expr; |
13e4fc7e JH |
1038 | else if (opt->all_match) |
1039 | opt->pattern_expression = grep_splice_or(header_expr, | |
1040 | opt->pattern_expression); | |
5aaeb733 JH |
1041 | else |
1042 | opt->pattern_expression = grep_or_expr(opt->pattern_expression, | |
1043 | header_expr); | |
80235ba7 | 1044 | opt->all_match = 1; |
83b5d2f5 JH |
1045 | } |
1046 | ||
17bf35a3 JH |
1047 | void compile_grep_patterns(struct grep_opt *opt) |
1048 | { | |
1049 | compile_grep_patterns_real(opt); | |
1050 | if (opt->debug) | |
1051 | dump_grep_expression(opt); | |
1052 | } | |
1053 | ||
b48fb5b6 JH |
1054 | static void free_pattern_expr(struct grep_expr *x) |
1055 | { | |
1056 | switch (x->node) { | |
5aaeb733 | 1057 | case GREP_NODE_TRUE: |
b48fb5b6 JH |
1058 | case GREP_NODE_ATOM: |
1059 | break; | |
1060 | case GREP_NODE_NOT: | |
1061 | free_pattern_expr(x->u.unary); | |
1062 | break; | |
1063 | case GREP_NODE_AND: | |
1064 | case GREP_NODE_OR: | |
1065 | free_pattern_expr(x->u.binary.left); | |
1066 | free_pattern_expr(x->u.binary.right); | |
1067 | break; | |
1068 | } | |
1069 | free(x); | |
1070 | } | |
1071 | ||
1072 | void free_grep_patterns(struct grep_opt *opt) | |
1073 | { | |
1074 | struct grep_pat *p, *n; | |
1075 | ||
1076 | for (p = opt->pattern_list; p; p = n) { | |
1077 | n = p->next; | |
1078 | switch (p->token) { | |
1079 | case GREP_PATTERN: /* atom */ | |
1080 | case GREP_PATTERN_HEAD: | |
1081 | case GREP_PATTERN_BODY: | |
48de2a76 | 1082 | if (p->pcre1_regexp) |
6d4b5747 | 1083 | free_pcre1_regexp(p); |
94da9193 ÆAB |
1084 | else if (p->pcre2_pattern) |
1085 | free_pcre2_pattern(p); | |
63e7e9d8 MK |
1086 | else |
1087 | regfree(&p->regexp); | |
526a858a | 1088 | free(p->pattern); |
b48fb5b6 JH |
1089 | break; |
1090 | default: | |
1091 | break; | |
1092 | } | |
1093 | free(p); | |
1094 | } | |
1095 | ||
1096 | if (!opt->extended) | |
1097 | return; | |
1098 | free_pattern_expr(opt->pattern_expression); | |
1099 | } | |
1100 | ||
83b5d2f5 JH |
1101 | static char *end_of_line(char *cp, unsigned long *left) |
1102 | { | |
1103 | unsigned long l = *left; | |
1104 | while (l && *cp != '\n') { | |
1105 | l--; | |
1106 | cp++; | |
1107 | } | |
1108 | *left = l; | |
1109 | return cp; | |
1110 | } | |
1111 | ||
1112 | static int word_char(char ch) | |
1113 | { | |
1114 | return isalnum(ch) || ch == '_'; | |
1115 | } | |
1116 | ||
55f638bd ML |
1117 | static void output_color(struct grep_opt *opt, const void *data, size_t size, |
1118 | const char *color) | |
1119 | { | |
daa0c3d9 | 1120 | if (want_color(opt->color) && color && color[0]) { |
55f638bd ML |
1121 | opt->output(opt, color, strlen(color)); |
1122 | opt->output(opt, data, size); | |
1123 | opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET)); | |
1124 | } else | |
1125 | opt->output(opt, data, size); | |
1126 | } | |
1127 | ||
1128 | static void output_sep(struct grep_opt *opt, char sign) | |
1129 | { | |
1130 | if (opt->null_following_name) | |
1131 | opt->output(opt, "\0", 1); | |
1132 | else | |
fa151dc5 | 1133 | output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]); |
55f638bd ML |
1134 | } |
1135 | ||
83caecca RZ |
1136 | static void show_name(struct grep_opt *opt, const char *name) |
1137 | { | |
fa151dc5 | 1138 | output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]); |
5b594f45 | 1139 | opt->output(opt, opt->null_following_name ? "\0" : "\n", 1); |
83caecca RZ |
1140 | } |
1141 | ||
97e77784 MK |
1142 | static int patmatch(struct grep_pat *p, char *line, char *eol, |
1143 | regmatch_t *match, int eflags) | |
1144 | { | |
1145 | int hit; | |
1146 | ||
48de2a76 | 1147 | if (p->pcre1_regexp) |
6d4b5747 | 1148 | hit = !pcre1match(p, line, eol, match, eflags); |
94da9193 ÆAB |
1149 | else if (p->pcre2_pattern) |
1150 | hit = !pcre2match(p, line, eol, match, eflags); | |
97e77784 | 1151 | else |
b7d36ffc JS |
1152 | hit = !regexec_buf(&p->regexp, line, eol - line, 1, match, |
1153 | eflags); | |
97e77784 MK |
1154 | |
1155 | return hit; | |
1156 | } | |
1157 | ||
a4d7d2c6 JH |
1158 | static int strip_timestamp(char *bol, char **eol_p) |
1159 | { | |
1160 | char *eol = *eol_p; | |
1161 | int ch; | |
1162 | ||
1163 | while (bol < --eol) { | |
1164 | if (*eol != '>') | |
1165 | continue; | |
1166 | *eol_p = ++eol; | |
1167 | ch = *eol; | |
1168 | *eol = '\0'; | |
1169 | return ch; | |
1170 | } | |
1171 | return 0; | |
1172 | } | |
1173 | ||
1174 | static struct { | |
1175 | const char *field; | |
1176 | size_t len; | |
1177 | } header_field[] = { | |
1178 | { "author ", 7 }, | |
1179 | { "committer ", 10 }, | |
72fd13f7 | 1180 | { "reflog ", 7 }, |
a4d7d2c6 JH |
1181 | }; |
1182 | ||
d7eb527d | 1183 | static int match_one_pattern(struct grep_pat *p, char *bol, char *eol, |
79212772 RS |
1184 | enum grep_context ctx, |
1185 | regmatch_t *pmatch, int eflags) | |
83b5d2f5 JH |
1186 | { |
1187 | int hit = 0; | |
a4d7d2c6 | 1188 | int saved_ch = 0; |
e701fadb | 1189 | const char *start = bol; |
83b5d2f5 | 1190 | |
480c1ca6 JH |
1191 | if ((p->token != GREP_PATTERN) && |
1192 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
1193 | return 0; | |
1194 | ||
a4d7d2c6 JH |
1195 | if (p->token == GREP_PATTERN_HEAD) { |
1196 | const char *field; | |
1197 | size_t len; | |
1198 | assert(p->field < ARRAY_SIZE(header_field)); | |
1199 | field = header_field[p->field].field; | |
1200 | len = header_field[p->field].len; | |
1201 | if (strncmp(bol, field, len)) | |
1202 | return 0; | |
1203 | bol += len; | |
ad4813b3 NTND |
1204 | switch (p->field) { |
1205 | case GREP_HEADER_AUTHOR: | |
1206 | case GREP_HEADER_COMMITTER: | |
1207 | saved_ch = strip_timestamp(bol, &eol); | |
1208 | break; | |
1209 | default: | |
1210 | break; | |
1211 | } | |
a4d7d2c6 JH |
1212 | } |
1213 | ||
83b5d2f5 | 1214 | again: |
97e77784 | 1215 | hit = patmatch(p, bol, eol, pmatch, eflags); |
83b5d2f5 | 1216 | |
d7eb527d | 1217 | if (hit && p->word_regexp) { |
83b5d2f5 | 1218 | if ((pmatch[0].rm_so < 0) || |
84201eae | 1219 | (eol - bol) < pmatch[0].rm_so || |
83b5d2f5 JH |
1220 | (pmatch[0].rm_eo < 0) || |
1221 | (eol - bol) < pmatch[0].rm_eo) | |
1222 | die("regexp returned nonsense"); | |
1223 | ||
1224 | /* Match beginning must be either beginning of the | |
1225 | * line, or at word boundary (i.e. the last char must | |
1226 | * not be a word char). Similarly, match end must be | |
1227 | * either end of the line, or at word boundary | |
1228 | * (i.e. the next char must not be a word char). | |
1229 | */ | |
fb62eb7f | 1230 | if ( ((pmatch[0].rm_so == 0) || |
83b5d2f5 JH |
1231 | !word_char(bol[pmatch[0].rm_so-1])) && |
1232 | ((pmatch[0].rm_eo == (eol-bol)) || | |
1233 | !word_char(bol[pmatch[0].rm_eo])) ) | |
1234 | ; | |
1235 | else | |
1236 | hit = 0; | |
1237 | ||
84201eae RS |
1238 | /* Words consist of at least one character. */ |
1239 | if (pmatch->rm_so == pmatch->rm_eo) | |
1240 | hit = 0; | |
1241 | ||
83b5d2f5 JH |
1242 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { |
1243 | /* There could be more than one match on the | |
1244 | * line, and the first match might not be | |
1245 | * strict word match. But later ones could be! | |
fb62eb7f RS |
1246 | * Forward to the next possible start, i.e. the |
1247 | * next position following a non-word char. | |
83b5d2f5 JH |
1248 | */ |
1249 | bol = pmatch[0].rm_so + bol + 1; | |
fb62eb7f RS |
1250 | while (word_char(bol[-1]) && bol < eol) |
1251 | bol++; | |
dbb6a4ad | 1252 | eflags |= REG_NOTBOL; |
fb62eb7f RS |
1253 | if (bol < eol) |
1254 | goto again; | |
83b5d2f5 JH |
1255 | } |
1256 | } | |
a4d7d2c6 JH |
1257 | if (p->token == GREP_PATTERN_HEAD && saved_ch) |
1258 | *eol = saved_ch; | |
e701fadb RS |
1259 | if (hit) { |
1260 | pmatch[0].rm_so += bol - start; | |
1261 | pmatch[0].rm_eo += bol - start; | |
1262 | } | |
83b5d2f5 JH |
1263 | return hit; |
1264 | } | |
1265 | ||
68d686e6 TB |
1266 | static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol, |
1267 | char *eol, enum grep_context ctx, ssize_t *col, | |
1268 | ssize_t *icol, int collect_hits) | |
83b5d2f5 | 1269 | { |
0ab7befa JH |
1270 | int h = 0; |
1271 | ||
c922b01f LT |
1272 | if (!x) |
1273 | die("Not a valid grep expression"); | |
83b5d2f5 | 1274 | switch (x->node) { |
5aaeb733 JH |
1275 | case GREP_NODE_TRUE: |
1276 | h = 1; | |
1277 | break; | |
83b5d2f5 | 1278 | case GREP_NODE_ATOM: |
68d686e6 TB |
1279 | { |
1280 | regmatch_t tmp; | |
1281 | h = match_one_pattern(x->u.atom, bol, eol, ctx, | |
1282 | &tmp, 0); | |
1283 | if (h && (*col < 0 || tmp.rm_so < *col)) | |
1284 | *col = tmp.rm_so; | |
1285 | } | |
83b5d2f5 JH |
1286 | break; |
1287 | case GREP_NODE_NOT: | |
68d686e6 TB |
1288 | /* |
1289 | * Upon visiting a GREP_NODE_NOT, col and icol become swapped. | |
1290 | */ | |
1291 | h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col, | |
1292 | 0); | |
0ab7befa | 1293 | break; |
83b5d2f5 | 1294 | case GREP_NODE_AND: |
017c0fcf | 1295 | h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col, |
68d686e6 | 1296 | icol, 0); |
017c0fcf TB |
1297 | if (h || opt->columnnum) { |
1298 | /* | |
1299 | * Don't short-circuit AND when given --column, since a | |
1300 | * NOT earlier in the tree may turn this into an OR. In | |
1301 | * this case, see the below comment. | |
1302 | */ | |
1303 | h &= match_expr_eval(opt, x->u.binary.right, bol, eol, | |
1304 | ctx, col, icol, 0); | |
1305 | } | |
0ab7befa | 1306 | break; |
83b5d2f5 | 1307 | case GREP_NODE_OR: |
017c0fcf TB |
1308 | if (!(collect_hits || opt->columnnum)) { |
1309 | /* | |
1310 | * Don't short-circuit OR when given --column (or | |
1311 | * collecting hits) to ensure we don't skip a later | |
1312 | * child that would produce an earlier match. | |
1313 | */ | |
68d686e6 TB |
1314 | return (match_expr_eval(opt, x->u.binary.left, bol, eol, |
1315 | ctx, col, icol, 0) || | |
1316 | match_expr_eval(opt, x->u.binary.right, bol, | |
1317 | eol, ctx, col, icol, 0)); | |
017c0fcf | 1318 | } |
68d686e6 TB |
1319 | h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col, |
1320 | icol, 0); | |
017c0fcf TB |
1321 | if (collect_hits) |
1322 | x->u.binary.left->hit |= h; | |
68d686e6 | 1323 | h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col, |
017c0fcf | 1324 | icol, collect_hits); |
0ab7befa JH |
1325 | break; |
1326 | default: | |
d7530708 | 1327 | die("Unexpected node type (internal error) %d", x->node); |
83b5d2f5 | 1328 | } |
0ab7befa JH |
1329 | if (collect_hits) |
1330 | x->hit |= h; | |
1331 | return h; | |
83b5d2f5 JH |
1332 | } |
1333 | ||
480c1ca6 | 1334 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, |
68d686e6 TB |
1335 | enum grep_context ctx, ssize_t *col, |
1336 | ssize_t *icol, int collect_hits) | |
83b5d2f5 JH |
1337 | { |
1338 | struct grep_expr *x = opt->pattern_expression; | |
68d686e6 | 1339 | return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits); |
83b5d2f5 JH |
1340 | } |
1341 | ||
480c1ca6 | 1342 | static int match_line(struct grep_opt *opt, char *bol, char *eol, |
68d686e6 | 1343 | ssize_t *col, ssize_t *icol, |
0ab7befa | 1344 | enum grep_context ctx, int collect_hits) |
83b5d2f5 JH |
1345 | { |
1346 | struct grep_pat *p; | |
017c0fcf | 1347 | int hit = 0; |
79212772 | 1348 | |
83b5d2f5 | 1349 | if (opt->extended) |
68d686e6 TB |
1350 | return match_expr(opt, bol, eol, ctx, col, icol, |
1351 | collect_hits); | |
0ab7befa JH |
1352 | |
1353 | /* we do not call with collect_hits without being extended */ | |
83b5d2f5 | 1354 | for (p = opt->pattern_list; p; p = p->next) { |
68d686e6 TB |
1355 | regmatch_t tmp; |
1356 | if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) { | |
017c0fcf TB |
1357 | hit |= 1; |
1358 | if (!opt->columnnum) { | |
1359 | /* | |
1360 | * Without --column, any single match on a line | |
1361 | * is enough to know that it needs to be | |
1362 | * printed. With --column, scan _all_ patterns | |
1363 | * to find the earliest. | |
1364 | */ | |
1365 | break; | |
1366 | } | |
1367 | if (*col < 0 || tmp.rm_so < *col) | |
1368 | *col = tmp.rm_so; | |
68d686e6 | 1369 | } |
83b5d2f5 | 1370 | } |
017c0fcf | 1371 | return hit; |
83b5d2f5 JH |
1372 | } |
1373 | ||
7e8f59d5 RS |
1374 | static int match_next_pattern(struct grep_pat *p, char *bol, char *eol, |
1375 | enum grep_context ctx, | |
1376 | regmatch_t *pmatch, int eflags) | |
1377 | { | |
1378 | regmatch_t match; | |
1379 | ||
1380 | if (!match_one_pattern(p, bol, eol, ctx, &match, eflags)) | |
1381 | return 0; | |
1382 | if (match.rm_so < 0 || match.rm_eo < 0) | |
1383 | return 0; | |
1384 | if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) { | |
1385 | if (match.rm_so > pmatch->rm_so) | |
1386 | return 1; | |
1387 | if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo) | |
1388 | return 1; | |
1389 | } | |
1390 | pmatch->rm_so = match.rm_so; | |
1391 | pmatch->rm_eo = match.rm_eo; | |
1392 | return 1; | |
1393 | } | |
1394 | ||
1395 | static int next_match(struct grep_opt *opt, char *bol, char *eol, | |
1396 | enum grep_context ctx, regmatch_t *pmatch, int eflags) | |
1397 | { | |
1398 | struct grep_pat *p; | |
1399 | int hit = 0; | |
1400 | ||
1401 | pmatch->rm_so = pmatch->rm_eo = -1; | |
1402 | if (bol < eol) { | |
1403 | for (p = opt->pattern_list; p; p = p->next) { | |
1404 | switch (p->token) { | |
1405 | case GREP_PATTERN: /* atom */ | |
1406 | case GREP_PATTERN_HEAD: | |
1407 | case GREP_PATTERN_BODY: | |
1408 | hit |= match_next_pattern(p, bol, eol, ctx, | |
1409 | pmatch, eflags); | |
1410 | break; | |
1411 | default: | |
1412 | break; | |
1413 | } | |
1414 | } | |
1415 | } | |
1416 | return hit; | |
1417 | } | |
1418 | ||
c707ded3 TB |
1419 | static void show_line_header(struct grep_opt *opt, const char *name, |
1420 | unsigned lno, ssize_t cno, char sign) | |
7e8f59d5 | 1421 | { |
1d84f72e | 1422 | if (opt->heading && opt->last_shown == 0) { |
fa151dc5 | 1423 | output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]); |
1d84f72e RS |
1424 | opt->output(opt, "\n", 1); |
1425 | } | |
5dd06d38 RS |
1426 | opt->last_shown = lno; |
1427 | ||
1d84f72e | 1428 | if (!opt->heading && opt->pathname) { |
fa151dc5 | 1429 | output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]); |
55f638bd | 1430 | output_sep(opt, sign); |
5b594f45 FK |
1431 | } |
1432 | if (opt->linenum) { | |
1433 | char buf[32]; | |
1a168e5c | 1434 | xsnprintf(buf, sizeof(buf), "%d", lno); |
fa151dc5 | 1435 | output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]); |
55f638bd | 1436 | output_sep(opt, sign); |
5b594f45 | 1437 | } |
89252cd0 TB |
1438 | /* |
1439 | * Treat 'cno' as the 1-indexed offset from the start of a non-context | |
1440 | * line to its first match. Otherwise, 'cno' is 0 indicating that we are | |
1441 | * being called with a context line. | |
1442 | */ | |
1443 | if (opt->columnnum && cno) { | |
1444 | char buf[32]; | |
1445 | xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno); | |
d036d667 | 1446 | output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]); |
89252cd0 TB |
1447 | output_sep(opt, sign); |
1448 | } | |
c707ded3 TB |
1449 | } |
1450 | ||
1451 | static void show_line(struct grep_opt *opt, char *bol, char *eol, | |
1452 | const char *name, unsigned lno, ssize_t cno, char sign) | |
1453 | { | |
1454 | int rest = eol - bol; | |
9d8db06e TB |
1455 | const char *match_color = NULL; |
1456 | const char *line_color = NULL; | |
c707ded3 TB |
1457 | |
1458 | if (opt->file_break && opt->last_shown == 0) { | |
1459 | if (opt->show_hunk_mark) | |
1460 | opt->output(opt, "\n", 1); | |
1461 | } else if (opt->pre_context || opt->post_context || opt->funcbody) { | |
1462 | if (opt->last_shown == 0) { | |
1463 | if (opt->show_hunk_mark) { | |
87ece7ce | 1464 | output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]); |
c707ded3 TB |
1465 | opt->output(opt, "\n", 1); |
1466 | } | |
1467 | } else if (lno > opt->last_shown + 1) { | |
87ece7ce | 1468 | output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]); |
c707ded3 TB |
1469 | opt->output(opt, "\n", 1); |
1470 | } | |
1471 | } | |
9d8db06e TB |
1472 | if (!opt->only_matching) { |
1473 | /* | |
1474 | * In case the line we're being called with contains more than | |
1475 | * one match, leave printing each header to the loop below. | |
1476 | */ | |
1477 | show_line_header(opt, name, lno, cno, sign); | |
1478 | } | |
1479 | if (opt->color || opt->only_matching) { | |
7e8f59d5 RS |
1480 | regmatch_t match; |
1481 | enum grep_context ctx = GREP_CONTEXT_BODY; | |
1482 | int ch = *eol; | |
1483 | int eflags = 0; | |
1484 | ||
9d8db06e TB |
1485 | if (opt->color) { |
1486 | if (sign == ':') | |
87ece7ce | 1487 | match_color = opt->colors[GREP_COLOR_MATCH_SELECTED]; |
9d8db06e | 1488 | else |
87ece7ce | 1489 | match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT]; |
9d8db06e | 1490 | if (sign == ':') |
87ece7ce | 1491 | line_color = opt->colors[GREP_COLOR_SELECTED]; |
9d8db06e | 1492 | else if (sign == '-') |
87ece7ce | 1493 | line_color = opt->colors[GREP_COLOR_CONTEXT]; |
9d8db06e | 1494 | else if (sign == '=') |
87ece7ce | 1495 | line_color = opt->colors[GREP_COLOR_FUNCTION]; |
9d8db06e | 1496 | } |
7e8f59d5 RS |
1497 | *eol = '\0'; |
1498 | while (next_match(opt, bol, eol, ctx, &match, eflags)) { | |
1f5b9cc4 RS |
1499 | if (match.rm_so == match.rm_eo) |
1500 | break; | |
5b594f45 | 1501 | |
9d8db06e TB |
1502 | if (opt->only_matching) |
1503 | show_line_header(opt, name, lno, cno, sign); | |
1504 | else | |
1505 | output_color(opt, bol, match.rm_so, line_color); | |
55f638bd | 1506 | output_color(opt, bol + match.rm_so, |
79a77109 | 1507 | match.rm_eo - match.rm_so, match_color); |
9d8db06e TB |
1508 | if (opt->only_matching) |
1509 | opt->output(opt, "\n", 1); | |
7e8f59d5 | 1510 | bol += match.rm_eo; |
9d8db06e | 1511 | cno += match.rm_eo; |
7e8f59d5 RS |
1512 | rest -= match.rm_eo; |
1513 | eflags = REG_NOTBOL; | |
1514 | } | |
1515 | *eol = ch; | |
1516 | } | |
9d8db06e TB |
1517 | if (!opt->only_matching) { |
1518 | output_color(opt, bol, rest, line_color); | |
1519 | opt->output(opt, "\n", 1); | |
1520 | } | |
7e8f59d5 RS |
1521 | } |
1522 | ||
78db6ea9 JK |
1523 | int grep_use_locks; |
1524 | ||
0579f91d TR |
1525 | /* |
1526 | * This lock protects access to the gitattributes machinery, which is | |
1527 | * not thread-safe. | |
1528 | */ | |
1529 | pthread_mutex_t grep_attr_mutex; | |
1530 | ||
78db6ea9 | 1531 | static inline void grep_attr_lock(void) |
0579f91d | 1532 | { |
78db6ea9 | 1533 | if (grep_use_locks) |
0579f91d TR |
1534 | pthread_mutex_lock(&grep_attr_mutex); |
1535 | } | |
1536 | ||
78db6ea9 | 1537 | static inline void grep_attr_unlock(void) |
0579f91d | 1538 | { |
78db6ea9 | 1539 | if (grep_use_locks) |
0579f91d TR |
1540 | pthread_mutex_unlock(&grep_attr_mutex); |
1541 | } | |
b3aeb285 JK |
1542 | |
1543 | /* | |
1544 | * Same as git_attr_mutex, but protecting the thread-unsafe object db access. | |
1545 | */ | |
1546 | pthread_mutex_t grep_read_mutex; | |
1547 | ||
e1327023 | 1548 | static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol) |
2944e4e6 | 1549 | { |
60ecac98 | 1550 | xdemitconf_t *xecfg = opt->priv; |
0579f91d | 1551 | if (xecfg && !xecfg->find_func) { |
acd00ea0 | 1552 | grep_source_load_driver(gs, opt->repo->index); |
94ad9d9e JK |
1553 | if (gs->driver->funcname.pattern) { |
1554 | const struct userdiff_funcname *pe = &gs->driver->funcname; | |
0579f91d TR |
1555 | xdiff_set_find_func(xecfg, pe->pattern, pe->cflags); |
1556 | } else { | |
1557 | xecfg = opt->priv = NULL; | |
1558 | } | |
1559 | } | |
1560 | ||
1561 | if (xecfg) { | |
60ecac98 RS |
1562 | char buf[1]; |
1563 | return xecfg->find_func(bol, eol - bol, buf, 1, | |
1564 | xecfg->find_func_priv) >= 0; | |
1565 | } | |
1566 | ||
2944e4e6 RS |
1567 | if (bol == eol) |
1568 | return 0; | |
1569 | if (isalpha(*bol) || *bol == '_' || *bol == '$') | |
1570 | return 1; | |
1571 | return 0; | |
1572 | } | |
1573 | ||
e1327023 JK |
1574 | static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs, |
1575 | char *bol, unsigned lno) | |
2944e4e6 | 1576 | { |
e1327023 | 1577 | while (bol > gs->buf) { |
2944e4e6 RS |
1578 | char *eol = --bol; |
1579 | ||
e1327023 | 1580 | while (bol > gs->buf && bol[-1] != '\n') |
2944e4e6 RS |
1581 | bol--; |
1582 | lno--; | |
1583 | ||
1584 | if (lno <= opt->last_shown) | |
1585 | break; | |
1586 | ||
e1327023 | 1587 | if (match_funcname(opt, gs, bol, eol)) { |
89252cd0 | 1588 | show_line(opt, bol, eol, gs->name, lno, 0, '='); |
2944e4e6 RS |
1589 | break; |
1590 | } | |
1591 | } | |
1592 | } | |
1593 | ||
a5dc20b0 RS |
1594 | static int is_empty_line(const char *bol, const char *eol); |
1595 | ||
e1327023 | 1596 | static void show_pre_context(struct grep_opt *opt, struct grep_source *gs, |
ba8ea749 | 1597 | char *bol, char *end, unsigned lno) |
49de3216 | 1598 | { |
6653a01b | 1599 | unsigned cur = lno, from = 1, funcname_lno = 0, orig_from; |
a5dc20b0 | 1600 | int funcname_needed = !!opt->funcname, comment_needed = 0; |
ba8ea749 | 1601 | |
49de3216 RS |
1602 | if (opt->pre_context < lno) |
1603 | from = lno - opt->pre_context; | |
1604 | if (from <= opt->last_shown) | |
1605 | from = opt->last_shown + 1; | |
6653a01b | 1606 | orig_from = from; |
a5dc20b0 RS |
1607 | if (opt->funcbody) { |
1608 | if (match_funcname(opt, gs, bol, end)) | |
1609 | comment_needed = 1; | |
1610 | else | |
1611 | funcname_needed = 1; | |
6653a01b RS |
1612 | from = opt->last_shown + 1; |
1613 | } | |
49de3216 RS |
1614 | |
1615 | /* Rewind. */ | |
6653a01b | 1616 | while (bol > gs->buf && cur > from) { |
a5dc20b0 | 1617 | char *next_bol = bol; |
2944e4e6 RS |
1618 | char *eol = --bol; |
1619 | ||
e1327023 | 1620 | while (bol > gs->buf && bol[-1] != '\n') |
49de3216 RS |
1621 | bol--; |
1622 | cur--; | |
a5dc20b0 RS |
1623 | if (comment_needed && (is_empty_line(bol, eol) || |
1624 | match_funcname(opt, gs, bol, eol))) { | |
1625 | comment_needed = 0; | |
1626 | from = orig_from; | |
1627 | if (cur < from) { | |
1628 | cur++; | |
1629 | bol = next_bol; | |
1630 | break; | |
1631 | } | |
1632 | } | |
e1327023 | 1633 | if (funcname_needed && match_funcname(opt, gs, bol, eol)) { |
2944e4e6 RS |
1634 | funcname_lno = cur; |
1635 | funcname_needed = 0; | |
a5dc20b0 RS |
1636 | if (opt->funcbody) |
1637 | comment_needed = 1; | |
1638 | else | |
1639 | from = orig_from; | |
2944e4e6 | 1640 | } |
49de3216 RS |
1641 | } |
1642 | ||
2944e4e6 RS |
1643 | /* We need to look even further back to find a function signature. */ |
1644 | if (opt->funcname && funcname_needed) | |
e1327023 | 1645 | show_funcname_line(opt, gs, bol, cur); |
2944e4e6 | 1646 | |
49de3216 RS |
1647 | /* Back forward. */ |
1648 | while (cur < lno) { | |
2944e4e6 | 1649 | char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-'; |
49de3216 RS |
1650 | |
1651 | while (*eol != '\n') | |
1652 | eol++; | |
89252cd0 | 1653 | show_line(opt, bol, eol, gs->name, cur, 0, sign); |
49de3216 RS |
1654 | bol = eol + 1; |
1655 | cur++; | |
1656 | } | |
1657 | } | |
1658 | ||
a26345b6 JH |
1659 | static int should_lookahead(struct grep_opt *opt) |
1660 | { | |
1661 | struct grep_pat *p; | |
1662 | ||
1663 | if (opt->extended) | |
1664 | return 0; /* punt for too complex stuff */ | |
1665 | if (opt->invert) | |
1666 | return 0; | |
1667 | for (p = opt->pattern_list; p; p = p->next) { | |
1668 | if (p->token != GREP_PATTERN) | |
1669 | return 0; /* punt for "header only" and stuff */ | |
1670 | } | |
1671 | return 1; | |
1672 | } | |
1673 | ||
1674 | static int look_ahead(struct grep_opt *opt, | |
1675 | unsigned long *left_p, | |
1676 | unsigned *lno_p, | |
1677 | char **bol_p) | |
1678 | { | |
1679 | unsigned lno = *lno_p; | |
1680 | char *bol = *bol_p; | |
1681 | struct grep_pat *p; | |
1682 | char *sp, *last_bol; | |
1683 | regoff_t earliest = -1; | |
1684 | ||
1685 | for (p = opt->pattern_list; p; p = p->next) { | |
1686 | int hit; | |
1687 | regmatch_t m; | |
1688 | ||
97e77784 | 1689 | hit = patmatch(p, bol, bol + *left_p, &m, 0); |
a26345b6 JH |
1690 | if (!hit || m.rm_so < 0 || m.rm_eo < 0) |
1691 | continue; | |
1692 | if (earliest < 0 || m.rm_so < earliest) | |
1693 | earliest = m.rm_so; | |
1694 | } | |
1695 | ||
1696 | if (earliest < 0) { | |
1697 | *bol_p = bol + *left_p; | |
1698 | *left_p = 0; | |
1699 | return 1; | |
1700 | } | |
1701 | for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--) | |
1702 | ; /* find the beginning of the line */ | |
1703 | last_bol = sp; | |
1704 | ||
1705 | for (sp = bol; sp < last_bol; sp++) { | |
1706 | if (*sp == '\n') | |
1707 | lno++; | |
1708 | } | |
1709 | *left_p -= last_bol - bol; | |
1710 | *bol_p = last_bol; | |
1711 | *lno_p = lno; | |
1712 | return 0; | |
1713 | } | |
1714 | ||
38bbc2ea NTND |
1715 | static int fill_textconv_grep(struct repository *r, |
1716 | struct userdiff_driver *driver, | |
335ec3bf JK |
1717 | struct grep_source *gs) |
1718 | { | |
1719 | struct diff_filespec *df; | |
1720 | char *buf; | |
1721 | size_t size; | |
1722 | ||
1723 | if (!driver || !driver->textconv) | |
1724 | return grep_source_load(gs); | |
1725 | ||
1726 | /* | |
1727 | * The textconv interface is intimately tied to diff_filespecs, so we | |
1728 | * have to pretend to be one. If we could unify the grep_source | |
1729 | * and diff_filespec structs, this mess could just go away. | |
1730 | */ | |
1731 | df = alloc_filespec(gs->path); | |
1732 | switch (gs->type) { | |
1c41c82b | 1733 | case GREP_SOURCE_OID: |
335ec3bf JK |
1734 | fill_filespec(df, gs->identifier, 1, 0100644); |
1735 | break; | |
1736 | case GREP_SOURCE_FILE: | |
f9704c2d | 1737 | fill_filespec(df, &null_oid, 0, 0100644); |
335ec3bf JK |
1738 | break; |
1739 | default: | |
033abf97 | 1740 | BUG("attempt to textconv something without a path?"); |
335ec3bf JK |
1741 | } |
1742 | ||
1743 | /* | |
1744 | * fill_textconv is not remotely thread-safe; it may load objects | |
1745 | * behind the scenes, and it modifies the global diff tempfile | |
1746 | * structure. | |
1747 | */ | |
1748 | grep_read_lock(); | |
38bbc2ea | 1749 | size = fill_textconv(r, driver, df, &buf); |
335ec3bf JK |
1750 | grep_read_unlock(); |
1751 | free_filespec(df); | |
1752 | ||
1753 | /* | |
1754 | * The normal fill_textconv usage by the diff machinery would just keep | |
1755 | * the textconv'd buf separate from the diff_filespec. But much of the | |
1756 | * grep code passes around a grep_source and assumes that its "buf" | |
1757 | * pointer is the beginning of the thing we are searching. So let's | |
1758 | * install our textconv'd version into the grep_source, taking care not | |
1759 | * to leak any existing buffer. | |
1760 | */ | |
1761 | grep_source_clear_data(gs); | |
1762 | gs->buf = buf; | |
1763 | gs->size = size; | |
1764 | ||
1765 | return 0; | |
1766 | } | |
1767 | ||
4aa2c475 RS |
1768 | static int is_empty_line(const char *bol, const char *eol) |
1769 | { | |
1770 | while (bol < eol && isspace(*bol)) | |
1771 | bol++; | |
1772 | return bol == eol; | |
1773 | } | |
1774 | ||
e1327023 | 1775 | static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits) |
83b5d2f5 | 1776 | { |
e1327023 | 1777 | char *bol; |
4aa2c475 | 1778 | char *peek_bol = NULL; |
e1327023 | 1779 | unsigned long left; |
83b5d2f5 | 1780 | unsigned lno = 1; |
83b5d2f5 | 1781 | unsigned last_hit = 0; |
83b5d2f5 | 1782 | int binary_match_only = 0; |
83b5d2f5 | 1783 | unsigned count = 0; |
a26345b6 | 1784 | int try_lookahead = 0; |
ba8ea749 | 1785 | int show_function = 0; |
335ec3bf | 1786 | struct userdiff_driver *textconv = NULL; |
480c1ca6 | 1787 | enum grep_context ctx = GREP_CONTEXT_HEAD; |
60ecac98 | 1788 | xdemitconf_t xecfg; |
83b5d2f5 | 1789 | |
de99eb0c ES |
1790 | if (!opt->status_only && gs->name == NULL) |
1791 | BUG("grep call which could print a name requires " | |
1792 | "grep_source.name be non-NULL"); | |
1793 | ||
5b594f45 FK |
1794 | if (!opt->output) |
1795 | opt->output = std_output; | |
1796 | ||
ba8ea749 RS |
1797 | if (opt->pre_context || opt->post_context || opt->file_break || |
1798 | opt->funcbody) { | |
08303c36 RS |
1799 | /* Show hunk marks, except for the first file. */ |
1800 | if (opt->last_shown) | |
1801 | opt->show_hunk_mark = 1; | |
1802 | /* | |
1803 | * If we're using threads then we can't easily identify | |
1804 | * the first file. Always put hunk marks in that case | |
1805 | * and skip the very first one later in work_done(). | |
1806 | */ | |
1807 | if (opt->output != std_output) | |
1808 | opt->show_hunk_mark = 1; | |
1809 | } | |
431d6e7b RS |
1810 | opt->last_shown = 0; |
1811 | ||
335ec3bf | 1812 | if (opt->allow_textconv) { |
acd00ea0 | 1813 | grep_source_load_driver(gs, opt->repo->index); |
335ec3bf JK |
1814 | /* |
1815 | * We might set up the shared textconv cache data here, which | |
1816 | * is not thread-safe. | |
1817 | */ | |
1818 | grep_attr_lock(); | |
bd7ad45b | 1819 | textconv = userdiff_get_textconv(opt->repo, gs->driver); |
335ec3bf JK |
1820 | grep_attr_unlock(); |
1821 | } | |
1822 | ||
1823 | /* | |
1824 | * We know the result of a textconv is text, so we only have to care | |
1825 | * about binary handling if we are not using it. | |
1826 | */ | |
1827 | if (!textconv) { | |
1828 | switch (opt->binary) { | |
1829 | case GREP_BINARY_DEFAULT: | |
acd00ea0 | 1830 | if (grep_source_is_binary(gs, opt->repo->index)) |
335ec3bf JK |
1831 | binary_match_only = 1; |
1832 | break; | |
1833 | case GREP_BINARY_NOMATCH: | |
acd00ea0 | 1834 | if (grep_source_is_binary(gs, opt->repo->index)) |
335ec3bf JK |
1835 | return 0; /* Assume unmatch */ |
1836 | break; | |
1837 | case GREP_BINARY_TEXT: | |
1838 | break; | |
1839 | default: | |
033abf97 | 1840 | BUG("unknown binary handling mode"); |
335ec3bf | 1841 | } |
83b5d2f5 JH |
1842 | } |
1843 | ||
60ecac98 | 1844 | memset(&xecfg, 0, sizeof(xecfg)); |
0579f91d TR |
1845 | opt->priv = &xecfg; |
1846 | ||
a26345b6 | 1847 | try_lookahead = should_lookahead(opt); |
60ecac98 | 1848 | |
38bbc2ea | 1849 | if (fill_textconv_grep(opt->repo, textconv, gs) < 0) |
08265798 JK |
1850 | return 0; |
1851 | ||
e1327023 JK |
1852 | bol = gs->buf; |
1853 | left = gs->size; | |
83b5d2f5 JH |
1854 | while (left) { |
1855 | char *eol, ch; | |
0ab7befa | 1856 | int hit; |
89252cd0 | 1857 | ssize_t cno; |
68d686e6 | 1858 | ssize_t col = -1, icol = -1; |
83b5d2f5 | 1859 | |
a26345b6 | 1860 | /* |
8997da38 | 1861 | * look_ahead() skips quickly to the line that possibly |
a26345b6 JH |
1862 | * has the next hit; don't call it if we need to do |
1863 | * something more than just skipping the current line | |
1864 | * in response to an unmatch for the current line. E.g. | |
1865 | * inside a post-context window, we will show the current | |
1866 | * line as a context around the previous hit when it | |
1867 | * doesn't hit. | |
1868 | */ | |
1869 | if (try_lookahead | |
1870 | && !(last_hit | |
ba8ea749 RS |
1871 | && (show_function || |
1872 | lno <= last_hit + opt->post_context)) | |
a26345b6 JH |
1873 | && look_ahead(opt, &left, &lno, &bol)) |
1874 | break; | |
83b5d2f5 JH |
1875 | eol = end_of_line(bol, &left); |
1876 | ch = *eol; | |
1877 | *eol = 0; | |
1878 | ||
480c1ca6 JH |
1879 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) |
1880 | ctx = GREP_CONTEXT_BODY; | |
1881 | ||
68d686e6 | 1882 | hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits); |
83b5d2f5 JH |
1883 | *eol = ch; |
1884 | ||
0ab7befa JH |
1885 | if (collect_hits) |
1886 | goto next_line; | |
1887 | ||
83b5d2f5 JH |
1888 | /* "grep -v -e foo -e bla" should list lines |
1889 | * that do not have either, so inversion should | |
1890 | * be done outside. | |
1891 | */ | |
1892 | if (opt->invert) | |
1893 | hit = !hit; | |
1894 | if (opt->unmatch_name_only) { | |
1895 | if (hit) | |
1896 | return 0; | |
1897 | goto next_line; | |
1898 | } | |
1899 | if (hit) { | |
1900 | count++; | |
1901 | if (opt->status_only) | |
1902 | return 1; | |
321ffcc0 | 1903 | if (opt->name_only) { |
e1327023 | 1904 | show_name(opt, gs->name); |
321ffcc0 RS |
1905 | return 1; |
1906 | } | |
c30c10cf RS |
1907 | if (opt->count) |
1908 | goto next_line; | |
83b5d2f5 | 1909 | if (binary_match_only) { |
5b594f45 | 1910 | opt->output(opt, "Binary file ", 12); |
e1327023 | 1911 | output_color(opt, gs->name, strlen(gs->name), |
fa151dc5 | 1912 | opt->colors[GREP_COLOR_FILENAME]); |
5b594f45 | 1913 | opt->output(opt, " matches\n", 9); |
83b5d2f5 JH |
1914 | return 1; |
1915 | } | |
83b5d2f5 JH |
1916 | /* Hit at this line. If we haven't shown the |
1917 | * pre-context lines, we would need to show them. | |
83b5d2f5 | 1918 | */ |
ba8ea749 | 1919 | if (opt->pre_context || opt->funcbody) |
e1327023 | 1920 | show_pre_context(opt, gs, bol, eol, lno); |
2944e4e6 | 1921 | else if (opt->funcname) |
e1327023 | 1922 | show_funcname_line(opt, gs, bol, lno); |
89252cd0 TB |
1923 | cno = opt->invert ? icol : col; |
1924 | if (cno < 0) { | |
1925 | /* | |
1926 | * A negative cno indicates that there was no | |
1927 | * match on the line. We are thus inverted and | |
1928 | * being asked to show all lines that _don't_ | |
1929 | * match a given expression. Therefore, set cno | |
1930 | * to 0 to suggest the whole line matches. | |
1931 | */ | |
1932 | cno = 0; | |
1933 | } | |
1934 | show_line(opt, bol, eol, gs->name, lno, cno + 1, ':'); | |
5dd06d38 | 1935 | last_hit = lno; |
ba8ea749 RS |
1936 | if (opt->funcbody) |
1937 | show_function = 1; | |
1938 | goto next_line; | |
83b5d2f5 | 1939 | } |
4aa2c475 RS |
1940 | if (show_function && (!peek_bol || peek_bol < bol)) { |
1941 | unsigned long peek_left = left; | |
1942 | char *peek_eol = eol; | |
1943 | ||
1944 | /* | |
1945 | * Trailing empty lines are not interesting. | |
1946 | * Peek past them to see if they belong to the | |
1947 | * body of the current function. | |
1948 | */ | |
1949 | peek_bol = bol; | |
1950 | while (is_empty_line(peek_bol, peek_eol)) { | |
1951 | peek_bol = peek_eol + 1; | |
1952 | peek_eol = end_of_line(peek_bol, &peek_left); | |
1953 | } | |
1954 | ||
1955 | if (match_funcname(opt, gs, peek_bol, peek_eol)) | |
1956 | show_function = 0; | |
1957 | } | |
ba8ea749 RS |
1958 | if (show_function || |
1959 | (last_hit && lno <= last_hit + opt->post_context)) { | |
83b5d2f5 JH |
1960 | /* If the last hit is within the post context, |
1961 | * we need to show this line. | |
1962 | */ | |
89252cd0 | 1963 | show_line(opt, bol, eol, gs->name, lno, col + 1, '-'); |
83b5d2f5 | 1964 | } |
83b5d2f5 JH |
1965 | |
1966 | next_line: | |
1967 | bol = eol + 1; | |
1968 | if (!left) | |
1969 | break; | |
1970 | left--; | |
1971 | lno++; | |
1972 | } | |
1973 | ||
0ab7befa JH |
1974 | if (collect_hits) |
1975 | return 0; | |
b48fb5b6 | 1976 | |
83b5d2f5 | 1977 | if (opt->status_only) |
e1f68c66 | 1978 | return opt->unmatch_name_only; |
83b5d2f5 JH |
1979 | if (opt->unmatch_name_only) { |
1980 | /* We did not see any hit, so we want to show this */ | |
e1327023 | 1981 | show_name(opt, gs->name); |
83b5d2f5 JH |
1982 | return 1; |
1983 | } | |
1984 | ||
60ecac98 RS |
1985 | xdiff_clear_find_func(&xecfg); |
1986 | opt->priv = NULL; | |
1987 | ||
83b5d2f5 JH |
1988 | /* NEEDSWORK: |
1989 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
1990 | * which feels mostly useless but sometimes useful. Maybe | |
1991 | * make it another option? For now suppress them. | |
1992 | */ | |
5b594f45 FK |
1993 | if (opt->count && count) { |
1994 | char buf[32]; | |
f76d947a RS |
1995 | if (opt->pathname) { |
1996 | output_color(opt, gs->name, strlen(gs->name), | |
fa151dc5 | 1997 | opt->colors[GREP_COLOR_FILENAME]); |
f76d947a RS |
1998 | output_sep(opt, ':'); |
1999 | } | |
1a168e5c | 2000 | xsnprintf(buf, sizeof(buf), "%u\n", count); |
5b594f45 | 2001 | opt->output(opt, buf, strlen(buf)); |
c30c10cf | 2002 | return 1; |
5b594f45 | 2003 | } |
83b5d2f5 JH |
2004 | return !!last_hit; |
2005 | } | |
2006 | ||
0ab7befa JH |
2007 | static void clr_hit_marker(struct grep_expr *x) |
2008 | { | |
2009 | /* All-hit markers are meaningful only at the very top level | |
2010 | * OR node. | |
2011 | */ | |
2012 | while (1) { | |
2013 | x->hit = 0; | |
2014 | if (x->node != GREP_NODE_OR) | |
2015 | return; | |
2016 | x->u.binary.left->hit = 0; | |
2017 | x = x->u.binary.right; | |
2018 | } | |
2019 | } | |
2020 | ||
2021 | static int chk_hit_marker(struct grep_expr *x) | |
2022 | { | |
2023 | /* Top level nodes have hit markers. See if they all are hits */ | |
2024 | while (1) { | |
2025 | if (x->node != GREP_NODE_OR) | |
2026 | return x->hit; | |
2027 | if (!x->u.binary.left->hit) | |
2028 | return 0; | |
2029 | x = x->u.binary.right; | |
2030 | } | |
2031 | } | |
2032 | ||
e1327023 | 2033 | int grep_source(struct grep_opt *opt, struct grep_source *gs) |
0ab7befa JH |
2034 | { |
2035 | /* | |
2036 | * we do not have to do the two-pass grep when we do not check | |
2037 | * buffer-wide "all-match". | |
2038 | */ | |
2039 | if (!opt->all_match) | |
e1327023 | 2040 | return grep_source_1(opt, gs, 0); |
0ab7befa JH |
2041 | |
2042 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
2043 | * We first clear hit markers from them. | |
2044 | */ | |
2045 | clr_hit_marker(opt->pattern_expression); | |
e1327023 | 2046 | grep_source_1(opt, gs, 1); |
0ab7befa JH |
2047 | |
2048 | if (!chk_hit_marker(opt->pattern_expression)) | |
2049 | return 0; | |
2050 | ||
e1327023 JK |
2051 | return grep_source_1(opt, gs, 0); |
2052 | } | |
2053 | ||
c876d6da | 2054 | int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size) |
e1327023 JK |
2055 | { |
2056 | struct grep_source gs; | |
2057 | int r; | |
2058 | ||
55c61688 | 2059 | grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL); |
e1327023 JK |
2060 | gs.buf = buf; |
2061 | gs.size = size; | |
2062 | ||
2063 | r = grep_source(opt, &gs); | |
2064 | ||
2065 | grep_source_clear(&gs); | |
2066 | return r; | |
2067 | } | |
2068 | ||
2069 | void grep_source_init(struct grep_source *gs, enum grep_source_type type, | |
55c61688 NTND |
2070 | const char *name, const char *path, |
2071 | const void *identifier) | |
e1327023 JK |
2072 | { |
2073 | gs->type = type; | |
8c53f071 JK |
2074 | gs->name = xstrdup_or_null(name); |
2075 | gs->path = xstrdup_or_null(path); | |
e1327023 JK |
2076 | gs->buf = NULL; |
2077 | gs->size = 0; | |
94ad9d9e | 2078 | gs->driver = NULL; |
e1327023 JK |
2079 | |
2080 | switch (type) { | |
2081 | case GREP_SOURCE_FILE: | |
2082 | gs->identifier = xstrdup(identifier); | |
2083 | break; | |
1c41c82b BW |
2084 | case GREP_SOURCE_OID: |
2085 | gs->identifier = oiddup(identifier); | |
e1327023 JK |
2086 | break; |
2087 | case GREP_SOURCE_BUF: | |
2088 | gs->identifier = NULL; | |
4538eef5 | 2089 | break; |
e1327023 JK |
2090 | } |
2091 | } | |
2092 | ||
2093 | void grep_source_clear(struct grep_source *gs) | |
2094 | { | |
88ce3ef6 ÆAB |
2095 | FREE_AND_NULL(gs->name); |
2096 | FREE_AND_NULL(gs->path); | |
2097 | FREE_AND_NULL(gs->identifier); | |
e1327023 JK |
2098 | grep_source_clear_data(gs); |
2099 | } | |
2100 | ||
2101 | void grep_source_clear_data(struct grep_source *gs) | |
2102 | { | |
2103 | switch (gs->type) { | |
2104 | case GREP_SOURCE_FILE: | |
1c41c82b | 2105 | case GREP_SOURCE_OID: |
6a83d902 | 2106 | FREE_AND_NULL(gs->buf); |
e1327023 JK |
2107 | gs->size = 0; |
2108 | break; | |
2109 | case GREP_SOURCE_BUF: | |
2110 | /* leave user-provided buf intact */ | |
2111 | break; | |
2112 | } | |
2113 | } | |
2114 | ||
1c41c82b | 2115 | static int grep_source_load_oid(struct grep_source *gs) |
e1327023 JK |
2116 | { |
2117 | enum object_type type; | |
2118 | ||
2119 | grep_read_lock(); | |
b4f5aca4 | 2120 | gs->buf = read_object_file(gs->identifier, &type, &gs->size); |
e1327023 JK |
2121 | grep_read_unlock(); |
2122 | ||
2123 | if (!gs->buf) | |
2124 | return error(_("'%s': unable to read %s"), | |
2125 | gs->name, | |
1c41c82b | 2126 | oid_to_hex(gs->identifier)); |
e1327023 JK |
2127 | return 0; |
2128 | } | |
2129 | ||
2130 | static int grep_source_load_file(struct grep_source *gs) | |
2131 | { | |
2132 | const char *filename = gs->identifier; | |
2133 | struct stat st; | |
2134 | char *data; | |
2135 | size_t size; | |
2136 | int i; | |
2137 | ||
2138 | if (lstat(filename, &st) < 0) { | |
2139 | err_ret: | |
2140 | if (errno != ENOENT) | |
7645d8f1 | 2141 | error_errno(_("failed to stat '%s'"), filename); |
e1327023 JK |
2142 | return -1; |
2143 | } | |
2144 | if (!S_ISREG(st.st_mode)) | |
2145 | return -1; | |
2146 | size = xsize_t(st.st_size); | |
2147 | i = open(filename, O_RDONLY); | |
2148 | if (i < 0) | |
2149 | goto err_ret; | |
3733e694 | 2150 | data = xmallocz(size); |
e1327023 | 2151 | if (st.st_size != read_in_full(i, data, size)) { |
7645d8f1 | 2152 | error_errno(_("'%s': short read"), filename); |
e1327023 JK |
2153 | close(i); |
2154 | free(data); | |
2155 | return -1; | |
2156 | } | |
2157 | close(i); | |
e1327023 JK |
2158 | |
2159 | gs->buf = data; | |
2160 | gs->size = size; | |
2161 | return 0; | |
2162 | } | |
2163 | ||
3083301e | 2164 | static int grep_source_load(struct grep_source *gs) |
e1327023 JK |
2165 | { |
2166 | if (gs->buf) | |
2167 | return 0; | |
2168 | ||
2169 | switch (gs->type) { | |
2170 | case GREP_SOURCE_FILE: | |
2171 | return grep_source_load_file(gs); | |
1c41c82b BW |
2172 | case GREP_SOURCE_OID: |
2173 | return grep_source_load_oid(gs); | |
e1327023 JK |
2174 | case GREP_SOURCE_BUF: |
2175 | return gs->buf ? 0 : -1; | |
2176 | } | |
033abf97 | 2177 | BUG("invalid grep_source type to load"); |
0ab7befa | 2178 | } |
94ad9d9e | 2179 | |
acd00ea0 NTND |
2180 | void grep_source_load_driver(struct grep_source *gs, |
2181 | struct index_state *istate) | |
94ad9d9e JK |
2182 | { |
2183 | if (gs->driver) | |
2184 | return; | |
2185 | ||
2186 | grep_attr_lock(); | |
55c61688 | 2187 | if (gs->path) |
acd00ea0 | 2188 | gs->driver = userdiff_find_by_path(istate, gs->path); |
94ad9d9e JK |
2189 | if (!gs->driver) |
2190 | gs->driver = userdiff_find_by_name("default"); | |
2191 | grep_attr_unlock(); | |
2192 | } | |
41b59bfc | 2193 | |
acd00ea0 NTND |
2194 | static int grep_source_is_binary(struct grep_source *gs, |
2195 | struct index_state *istate) | |
41b59bfc | 2196 | { |
acd00ea0 | 2197 | grep_source_load_driver(gs, istate); |
41b59bfc JK |
2198 | if (gs->driver->binary != -1) |
2199 | return gs->driver->binary; | |
2200 | ||
2201 | if (!grep_source_load(gs)) | |
2202 | return buffer_is_binary(gs->buf, gs->size); | |
2203 | ||
2204 | return 0; | |
2205 | } |