]> git.proxmox.com Git - libgit2.git/blame - src/diff_driver.c
Upload to experimental
[libgit2.git] / src / diff_driver.c
CommitLineData
114f5a6c
RB
1/*
2 * Copyright (C) the libgit2 contributors. All rights reserved.
3 *
4 * This file is part of libgit2, distributed under the GNU GPL v2 with
5 * a Linking Exception. For full terms see the included COPYING file.
6 */
eae0bfdc
PP
7
8#include "diff_driver.h"
114f5a6c
RB
9
10#include "git2/attr.h"
11
22a2d3d5 12#include "common.h"
114f5a6c 13#include "diff.h"
114f5a6c 14#include "strmap.h"
114f5a6c 15#include "map.h"
9f77b3f6 16#include "config.h"
22a2d3d5 17#include "regexp.h"
5dc98298 18#include "repository.h"
114f5a6c
RB
19
20typedef enum {
21 DIFF_DRIVER_AUTO = 0,
5dc98298
RB
22 DIFF_DRIVER_BINARY = 1,
23 DIFF_DRIVER_TEXT = 2,
e579e0f7 24 DIFF_DRIVER_PATTERNLIST = 3
114f5a6c
RB
25} git_diff_driver_t;
26
a5a38643 27typedef struct {
22a2d3d5 28 git_regexp re;
a5a38643
RB
29 int flags;
30} git_diff_driver_pattern;
31
114f5a6c 32enum {
a5a38643 33 REG_NEGATE = (1 << 15) /* get out of the way of existing flags */
114f5a6c
RB
34};
35
36/* data for finding function context for a given file type */
37struct git_diff_driver {
38 git_diff_driver_t type;
5dc98298
RB
39 uint32_t binary_flags;
40 uint32_t other_flags;
a5a38643 41 git_array_t(git_diff_driver_pattern) fn_patterns;
22a2d3d5 42 git_regexp word_pattern;
42e6cf78 43 char name[GIT_FLEX_ARRAY];
114f5a6c
RB
44};
45
c7c260a5 46#include "userdiff.h"
2c65602e 47
114f5a6c
RB
48struct git_diff_driver_registry {
49 git_strmap *drivers;
114f5a6c
RB
50};
51
5dc98298
RB
52#define FORCE_DIFFABLE (GIT_DIFF_FORCE_TEXT | GIT_DIFF_FORCE_BINARY)
53
e579e0f7
MB
54static git_diff_driver diff_driver_auto = { DIFF_DRIVER_AUTO, 0, 0 };
55static git_diff_driver diff_driver_binary = { DIFF_DRIVER_BINARY, GIT_DIFF_FORCE_BINARY, 0 };
56static git_diff_driver diff_driver_text = { DIFF_DRIVER_TEXT, GIT_DIFF_FORCE_TEXT, 0 };
114f5a6c 57
8fd74c08 58git_diff_driver_registry *git_diff_driver_registry_new(void)
114f5a6c 59{
3eadfecd
RB
60 git_diff_driver_registry *reg =
61 git__calloc(1, sizeof(git_diff_driver_registry));
62 if (!reg)
63 return NULL;
64
22a2d3d5 65 if (git_strmap_new(&reg->drivers) < 0) {
3eadfecd
RB
66 git_diff_driver_registry_free(reg);
67 return NULL;
68 }
69
70 return reg;
114f5a6c
RB
71}
72
73void git_diff_driver_registry_free(git_diff_driver_registry *reg)
74{
5dc98298
RB
75 git_diff_driver *drv;
76
3eadfecd
RB
77 if (!reg)
78 return;
79
5dc98298 80 git_strmap_foreach_value(reg->drivers, drv, git_diff_driver_free(drv));
3eadfecd 81 git_strmap_free(reg->drivers);
114f5a6c
RB
82 git__free(reg);
83}
84
a5a38643
RB
85static int diff_driver_add_patterns(
86 git_diff_driver *drv, const char *regex_str, int regex_flags)
5dc98298 87{
a5a38643
RB
88 int error = 0;
89 const char *scan, *end;
90 git_diff_driver_pattern *pat = NULL;
e579e0f7 91 git_str buf = GIT_STR_INIT;
a5a38643
RB
92
93 for (scan = regex_str; scan; scan = end) {
94 /* get pattern to fill in */
95 if ((pat = git_array_alloc(drv->fn_patterns)) == NULL) {
e451cd5c 96 return -1;
a5a38643 97 }
5dc98298 98
a5a38643
RB
99 pat->flags = regex_flags;
100 if (*scan == '!') {
101 pat->flags |= REG_NEGATE;
102 ++scan;
103 }
104
105 if ((end = strchr(scan, '\n')) != NULL) {
e579e0f7 106 error = git_str_set(&buf, scan, end - scan);
a5a38643
RB
107 end++;
108 } else {
e579e0f7 109 error = git_str_sets(&buf, scan);
a5a38643
RB
110 }
111 if (error < 0)
112 break;
113
22a2d3d5 114 if ((error = git_regexp_compile(&pat->re, buf.ptr, regex_flags)) != 0) {
e451cd5c
CMN
115 /*
116 * TODO: issue a warning
117 */
a5a38643 118 }
5dc98298
RB
119 }
120
a5a38643
RB
121 if (error && pat != NULL)
122 (void)git_array_pop(drv->fn_patterns); /* release last item */
e579e0f7 123 git_str_dispose(&buf);
5dc98298 124
e451cd5c
CMN
125 /* We want to ignore bad patterns, so return success regardless */
126 return 0;
5dc98298
RB
127}
128
129static int diff_driver_xfuncname(const git_config_entry *entry, void *payload)
130{
22a2d3d5 131 return diff_driver_add_patterns(payload, entry->value, 0);
5dc98298
RB
132}
133
134static int diff_driver_funcname(const git_config_entry *entry, void *payload)
135{
a5a38643 136 return diff_driver_add_patterns(payload, entry->value, 0);
5dc98298
RB
137}
138
139static git_diff_driver_registry *git_repository_driver_registry(
140 git_repository *repo)
141{
c25aa7cd
PP
142 git_diff_driver_registry *reg = git_atomic_load(repo->diff_drivers), *newreg;
143 if (reg)
144 return reg;
5dc98298 145
c25aa7cd
PP
146 newreg = git_diff_driver_registry_new();
147 if (!newreg) {
ac3d33df 148 git_error_set(GIT_ERROR_REPOSITORY, "unable to create diff driver registry");
c25aa7cd
PP
149 return newreg;
150 }
151 reg = git_atomic_compare_and_swap(&repo->diff_drivers, NULL, newreg);
152 if (!reg) {
153 reg = newreg;
154 } else {
155 /* if we race, free losing allocation */
156 git_diff_driver_registry_free(newreg);
157 }
158 return reg;
5dc98298
RB
159}
160
392702ee
ET
161static int diff_driver_alloc(
162 git_diff_driver **out, size_t *namelen_out, const char *name)
163{
164 git_diff_driver *driver;
165 size_t driverlen = sizeof(git_diff_driver),
f1453c59
ET
166 namelen = strlen(name),
167 alloclen;
392702ee 168
ac3d33df
JK
169 GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, driverlen, namelen);
170 GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);
392702ee 171
f1453c59 172 driver = git__calloc(1, alloclen);
ac3d33df 173 GIT_ERROR_CHECK_ALLOC(driver);
392702ee
ET
174
175 memcpy(driver->name, name, namelen);
176
177 *out = driver;
178
179 if (namelen_out)
180 *namelen_out = namelen;
181
182 return 0;
183}
184
a5a38643
RB
185static int git_diff_driver_builtin(
186 git_diff_driver **out,
187 git_diff_driver_registry *reg,
188 const char *driver_name)
189{
a5a38643
RB
190 git_diff_driver_definition *ddef = NULL;
191 git_diff_driver *drv = NULL;
22a2d3d5 192 int error = 0;
392702ee 193 size_t idx;
a5a38643
RB
194
195 for (idx = 0; idx < ARRAY_SIZE(builtin_defs); ++idx) {
196 if (!strcasecmp(driver_name, builtin_defs[idx].name)) {
197 ddef = &builtin_defs[idx];
198 break;
199 }
200 }
201 if (!ddef)
202 goto done;
203
392702ee
ET
204 if ((error = diff_driver_alloc(&drv, NULL, ddef->name)) < 0)
205 goto done;
a5a38643
RB
206
207 drv->type = DIFF_DRIVER_PATTERNLIST;
a5a38643
RB
208
209 if (ddef->fns &&
210 (error = diff_driver_add_patterns(
22a2d3d5 211 drv, ddef->fns, ddef->flags)) < 0)
a5a38643
RB
212 goto done;
213
214 if (ddef->words &&
22a2d3d5 215 (error = git_regexp_compile(&drv->word_pattern, ddef->words, ddef->flags)) < 0)
a5a38643 216 goto done;
a5a38643 217
22a2d3d5
UG
218 if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
219 goto done;
a5a38643
RB
220
221done:
2c65602e 222 if (error && drv)
a5a38643 223 git_diff_driver_free(drv);
2c65602e 224 else
a5a38643 225 *out = drv;
a5a38643
RB
226
227 return error;
228}
229
3eadfecd 230static int git_diff_driver_load(
5dc98298 231 git_diff_driver **out, git_repository *repo, const char *driver_name)
3eadfecd 232{
9f77b3f6 233 int error = 0;
5dc98298 234 git_diff_driver_registry *reg;
22a2d3d5
UG
235 git_diff_driver *drv;
236 size_t namelen;
392702ee 237 git_config *cfg = NULL;
e579e0f7 238 git_str name = GIT_STR_INIT;
9a97f49e 239 git_config_entry *ce = NULL;
42e6cf78 240 bool found_driver = false;
5dc98298 241
a5a38643 242 if ((reg = git_repository_driver_registry(repo)) == NULL)
5dc98298 243 return -1;
a5a38643 244
22a2d3d5
UG
245 if ((drv = git_strmap_get(reg->drivers, driver_name)) != NULL) {
246 *out = drv;
a5a38643 247 return 0;
5dc98298
RB
248 }
249
392702ee
ET
250 if ((error = diff_driver_alloc(&drv, &namelen, driver_name)) < 0)
251 goto done;
252
a37aa82e 253 drv->type = DIFF_DRIVER_AUTO;
a37aa82e 254
5dc98298 255 /* if you can't read config for repo, just use default driver */
ac99d86b 256 if (git_repository_config_snapshot(&cfg, repo) < 0) {
ac3d33df 257 git_error_clear();
a5a38643 258 goto done;
5dc98298
RB
259 }
260
e579e0f7 261 if ((error = git_str_printf(&name, "diff.%s.binary", driver_name)) < 0)
54faddd2 262 goto done;
9f77b3f6
RB
263
264 switch (git_config__get_bool_force(cfg, name.ptr, -1)) {
265 case true:
5dc98298 266 /* if diff.<driver>.binary is true, just return the binary driver */
e579e0f7 267 *out = &diff_driver_binary;
54faddd2 268 goto done;
9f77b3f6 269 case false:
5dc98298
RB
270 /* if diff.<driver>.binary is false, force binary checks off */
271 /* but still may have custom function context patterns, etc. */
272 drv->binary_flags = GIT_DIFF_FORCE_TEXT;
42e6cf78 273 found_driver = true;
9f77b3f6
RB
274 break;
275 default:
a5a38643 276 /* diff.<driver>.binary unspecified or "auto", so just continue */
9f77b3f6 277 break;
5dc98298
RB
278 }
279
280 /* TODO: warn if diff.<name>.command or diff.<name>.textconv are set */
281
e579e0f7
MB
282 git_str_truncate(&name, namelen + strlen("diff.."));
283 if ((error = git_str_PUTS(&name, "xfuncname")) < 0)
22a2d3d5
UG
284 goto done;
285
4efa3290 286 if ((error = git_config_get_multivar_foreach(
5dc98298
RB
287 cfg, name.ptr, NULL, diff_driver_xfuncname, drv)) < 0) {
288 if (error != GIT_ENOTFOUND)
54faddd2 289 goto done;
ac3d33df 290 git_error_clear(); /* no diff.<driver>.xfuncname, so just continue */
5dc98298 291 }
3eadfecd 292
e579e0f7
MB
293 git_str_truncate(&name, namelen + strlen("diff.."));
294 if ((error = git_str_PUTS(&name, "funcname")) < 0)
22a2d3d5
UG
295 goto done;
296
4efa3290 297 if ((error = git_config_get_multivar_foreach(
5dc98298
RB
298 cfg, name.ptr, NULL, diff_driver_funcname, drv)) < 0) {
299 if (error != GIT_ENOTFOUND)
54faddd2 300 goto done;
ac3d33df 301 git_error_clear(); /* no diff.<driver>.funcname, so just continue */
5dc98298
RB
302 }
303
304 /* if we found any patterns, set driver type to use correct callback */
42e6cf78 305 if (git_array_size(drv->fn_patterns) > 0) {
5dc98298 306 drv->type = DIFF_DRIVER_PATTERNLIST;
42e6cf78
RB
307 found_driver = true;
308 }
5dc98298 309
e579e0f7
MB
310 git_str_truncate(&name, namelen + strlen("diff.."));
311 if ((error = git_str_PUTS(&name, "wordregex")) < 0)
22a2d3d5
UG
312 goto done;
313
9f77b3f6 314 if ((error = git_config__lookup_entry(&ce, cfg, name.ptr, false)) < 0)
54faddd2 315 goto done;
9f77b3f6
RB
316 if (!ce || !ce->value)
317 /* no diff.<driver>.wordregex, so just continue */;
22a2d3d5 318 else if (!(error = git_regexp_compile(&drv->word_pattern, ce->value, 0)))
42e6cf78 319 found_driver = true;
9f77b3f6
RB
320 else {
321 /* TODO: warn about bad regex instead of failure */
9f77b3f6 322 goto done;
5dc98298
RB
323 }
324
325 /* TODO: look up diff.<driver>.algorithm to turn on minimal / patience
326 * diff in drv->other_flags
327 */
328
54faddd2 329 /* if no driver config found at all, fall back on AUTO driver */
42e6cf78 330 if (!found_driver)
54faddd2 331 goto done;
42e6cf78
RB
332
333 /* store driver in registry */
22a2d3d5 334 if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
54faddd2 335 goto done;
42e6cf78 336
5dc98298 337 *out = drv;
5dc98298 338
54faddd2 339done:
9a97f49e 340 git_config_entry_free(ce);
e579e0f7 341 git_str_dispose(&name);
29c4cb09 342 git_config_free(cfg);
54faddd2 343
a5a38643
RB
344 if (!*out) {
345 int error2 = git_diff_driver_builtin(out, reg, driver_name);
346 if (!error)
347 error = error2;
348 }
54faddd2
RB
349
350 if (drv && drv != *out)
351 git_diff_driver_free(drv);
352
5dc98298 353 return error;
3eadfecd
RB
354}
355
114f5a6c 356int git_diff_driver_lookup(
eae0bfdc
PP
357 git_diff_driver **out, git_repository *repo,
358 git_attr_session *attrsession, const char *path)
114f5a6c 359{
3eadfecd 360 int error = 0;
eae0bfdc 361 const char *values[1], *attrs[] = { "diff" };
114f5a6c 362
c25aa7cd 363 GIT_ASSERT_ARG(out);
2c65602e 364 *out = NULL;
114f5a6c
RB
365
366 if (!repo || !path || !strlen(path))
2c65602e 367 /* just use the auto value */;
eae0bfdc
PP
368 else if ((error = git_attr_get_many_with_session(values, repo,
369 attrsession, 0, path, 1, attrs)) < 0)
2c65602e 370 /* return error below */;
eae0bfdc 371
22a2d3d5 372 else if (GIT_ATTR_IS_UNSPECIFIED(values[0]))
42e6cf78 373 /* just use the auto value */;
22a2d3d5 374 else if (GIT_ATTR_IS_FALSE(values[0]))
e579e0f7 375 *out = &diff_driver_binary;
22a2d3d5 376 else if (GIT_ATTR_IS_TRUE(values[0]))
e579e0f7 377 *out = &diff_driver_text;
114f5a6c
RB
378
379 /* otherwise look for driver information in config and build driver */
eae0bfdc 380 else if ((error = git_diff_driver_load(out, repo, values[0])) < 0) {
2c65602e
RB
381 if (error == GIT_ENOTFOUND) {
382 error = 0;
ac3d33df 383 git_error_clear();
2c65602e 384 }
3eadfecd 385 }
114f5a6c 386
42e6cf78 387 if (!*out)
e579e0f7 388 *out = &diff_driver_auto;
42e6cf78 389
2c65602e 390 return error;
114f5a6c
RB
391}
392
393void git_diff_driver_free(git_diff_driver *driver)
394{
c25aa7cd 395 git_diff_driver_pattern *pat;
5dc98298
RB
396
397 if (!driver)
398 return;
399
c25aa7cd
PP
400 while ((pat = git_array_pop(driver->fn_patterns)) != NULL)
401 git_regexp_dispose(&pat->re);
5dc98298
RB
402 git_array_clear(driver->fn_patterns);
403
22a2d3d5 404 git_regexp_dispose(&driver->word_pattern);
5dc98298
RB
405
406 git__free(driver);
114f5a6c
RB
407}
408
5dc98298
RB
409void git_diff_driver_update_options(
410 uint32_t *option_flags, git_diff_driver *driver)
114f5a6c 411{
5dc98298
RB
412 if ((*option_flags & FORCE_DIFFABLE) == 0)
413 *option_flags |= driver->binary_flags;
414
415 *option_flags |= driver->other_flags;
114f5a6c
RB
416}
417
418int git_diff_driver_content_is_binary(
419 git_diff_driver *driver, const char *content, size_t content_len)
420{
e579e0f7 421 git_str search = GIT_STR_INIT;
114f5a6c
RB
422
423 GIT_UNUSED(driver);
424
e579e0f7 425 git_str_attach_notowned(&search, content,
d4cf1675
ET
426 min(content_len, GIT_FILTER_BYTES_TO_CHECK_NUL));
427
114f5a6c
RB
428 /* TODO: provide encoding / binary detection callbacks that can
429 * be UTF-8 aware, etc. For now, instead of trying to be smart,
430 * let's just use the simple NUL-byte detection that core git uses.
431 */
432
e579e0f7
MB
433 /* previously was: if (git_str_is_binary(&search)) */
434 if (git_str_contains_nul(&search))
114f5a6c
RB
435 return 1;
436
437 return 0;
438}
439
5dc98298 440static int diff_context_line__simple(
e579e0f7 441 git_diff_driver *driver, git_str *line)
5dc98298 442{
a5a38643 443 char firstch = line->ptr[0];
5dc98298 444 GIT_UNUSED(driver);
a5a38643 445 return (git__isalpha(firstch) || firstch == '_' || firstch == '$');
5dc98298
RB
446}
447
448static int diff_context_line__pattern_match(
e579e0f7 449 git_diff_driver *driver, git_str *line)
5dc98298 450{
b8e86c62 451 size_t i, maxi = git_array_size(driver->fn_patterns);
22a2d3d5 452 git_regmatch pmatch[2];
5dc98298 453
b8e86c62 454 for (i = 0; i < maxi; ++i) {
a5a38643
RB
455 git_diff_driver_pattern *pat = git_array_get(driver->fn_patterns, i);
456
22a2d3d5 457 if (!git_regexp_search(&pat->re, line->ptr, 2, pmatch)) {
a5a38643
RB
458 if (pat->flags & REG_NEGATE)
459 return false;
b8e86c62
RB
460
461 /* use pmatch data to trim line data */
22a2d3d5 462 i = (pmatch[1].start >= 0) ? 1 : 0;
e579e0f7
MB
463 git_str_consume(line, git_str_cstr(line) + pmatch[i].start);
464 git_str_truncate(line, pmatch[i].end - pmatch[i].start);
465 git_str_rtrim(line);
b8e86c62 466
5dc98298 467 return true;
a5a38643 468 }
5dc98298
RB
469 }
470
471 return false;
472}
473
114f5a6c
RB
474static long diff_context_find(
475 const char *line,
476 long line_len,
477 char *out,
478 long out_size,
479 void *payload)
480{
5dc98298 481 git_diff_find_context_payload *ctxt = payload;
114f5a6c 482
e579e0f7 483 if (git_str_set(&ctxt->line, line, (size_t)line_len) < 0)
114f5a6c 484 return -1;
e579e0f7 485 git_str_rtrim(&ctxt->line);
114f5a6c 486
5dc98298 487 if (!ctxt->line.size)
114f5a6c
RB
488 return -1;
489
a5a38643 490 if (!ctxt->match_line || !ctxt->match_line(ctxt->driver, &ctxt->line))
5dc98298 491 return -1;
114f5a6c 492
a5f9b5f8 493 if (out_size > (long)ctxt->line.size)
584f2d30 494 out_size = (long)ctxt->line.size;
a5f9b5f8 495 memcpy(out, ctxt->line.ptr, (size_t)out_size);
114f5a6c 496
a5f9b5f8 497 return out_size;
5dc98298 498}
114f5a6c 499
5dc98298
RB
500void git_diff_find_context_init(
501 git_diff_find_context_fn *findfn_out,
502 git_diff_find_context_payload *payload_out,
503 git_diff_driver *driver)
504{
505 *findfn_out = driver ? diff_context_find : NULL;
506
507 memset(payload_out, 0, sizeof(*payload_out));
508 if (driver) {
509 payload_out->driver = driver;
510 payload_out->match_line = (driver->type == DIFF_DRIVER_PATTERNLIST) ?
511 diff_context_line__pattern_match : diff_context_line__simple;
e579e0f7 512 git_str_init(&payload_out->line, 0);
5dc98298 513 }
114f5a6c
RB
514}
515
5dc98298 516void git_diff_find_context_clear(git_diff_find_context_payload *payload)
114f5a6c 517{
5dc98298 518 if (payload) {
e579e0f7 519 git_str_dispose(&payload->line);
5dc98298
RB
520 payload->driver = NULL;
521 }
114f5a6c 522}