diff_driver.c 12.2 KB
Newer Older
1 2 3 4 5 6
/*
 * Copyright (C) the libgit2 contributors. All rights reserved.
 *
 * This file is part of libgit2, distributed under the GNU GPL v2 with
 * a Linking Exception. For full terms see the included COPYING file.
 */
7 8

#include "diff_driver.h"
9 10 11

#include "git2/attr.h"

12
#include "common.h"
13 14 15 16
#include "diff.h"
#include "strmap.h"
#include "map.h"
#include "buf_text.h"
17
#include "config.h"
18
#include "regexp.h"
19
#include "repository.h"
20 21 22

typedef enum {
	DIFF_DRIVER_AUTO = 0,
23 24 25
	DIFF_DRIVER_BINARY = 1,
	DIFF_DRIVER_TEXT = 2,
	DIFF_DRIVER_PATTERNLIST = 3,
26 27
} git_diff_driver_t;

28
typedef struct {
29
	git_regexp re;
30 31 32
	int flags;
} git_diff_driver_pattern;

33
enum {
34
	REG_NEGATE = (1 << 15) /* get out of the way of existing flags */
35 36 37 38 39
};

/* data for finding function context for a given file type */
struct git_diff_driver {
	git_diff_driver_t type;
40 41
	uint32_t binary_flags;
	uint32_t other_flags;
42
	git_array_t(git_diff_driver_pattern) fn_patterns;
43
	git_regexp  word_pattern;
44
	char name[GIT_FLEX_ARRAY];
45 46
};

47
#include "userdiff.h"
48

49 50 51 52
struct git_diff_driver_registry {
	git_strmap *drivers;
};

53 54
#define FORCE_DIFFABLE (GIT_DIFF_FORCE_TEXT | GIT_DIFF_FORCE_BINARY)

55
static git_diff_driver global_drivers[3] = {
56 57 58
	{ DIFF_DRIVER_AUTO,   0, 0, },
	{ DIFF_DRIVER_BINARY, GIT_DIFF_FORCE_BINARY, 0 },
	{ DIFF_DRIVER_TEXT,   GIT_DIFF_FORCE_TEXT, 0 },
59 60
};

61
git_diff_driver_registry *git_diff_driver_registry_new(void)
62
{
63 64 65 66 67
	git_diff_driver_registry *reg =
		git__calloc(1, sizeof(git_diff_driver_registry));
	if (!reg)
		return NULL;

68
	if (git_strmap_new(&reg->drivers) < 0) {
69 70 71 72 73
		git_diff_driver_registry_free(reg);
		return NULL;
	}

	return reg;
74 75 76 77
}

void git_diff_driver_registry_free(git_diff_driver_registry *reg)
{
78 79
	git_diff_driver *drv;

80 81 82
	if (!reg)
		return;

83
	git_strmap_foreach_value(reg->drivers, drv, git_diff_driver_free(drv));
84
	git_strmap_free(reg->drivers);
85 86 87
	git__free(reg);
}

88 89
static int diff_driver_add_patterns(
	git_diff_driver *drv, const char *regex_str, int regex_flags)
90
{
91 92 93 94 95 96 97 98
	int error = 0;
	const char *scan, *end;
	git_diff_driver_pattern *pat = NULL;
	git_buf buf = GIT_BUF_INIT;

	for (scan = regex_str; scan; scan = end) {
		/* get pattern to fill in */
		if ((pat = git_array_alloc(drv->fn_patterns)) == NULL) {
99
			return -1;
100
		}
101

102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
		pat->flags = regex_flags;
		if (*scan == '!') {
			pat->flags |= REG_NEGATE;
			++scan;
		}

		if ((end = strchr(scan, '\n')) != NULL) {
			error = git_buf_set(&buf, scan, end - scan);
			end++;
		} else {
			error = git_buf_sets(&buf, scan);
		}
		if (error < 0)
			break;

117
		if ((error = git_regexp_compile(&pat->re, buf.ptr, regex_flags)) != 0) {
118 119 120
			/*
			 * TODO: issue a warning
			 */
121
		}
122 123
	}

124 125
	if (error && pat != NULL)
		(void)git_array_pop(drv->fn_patterns); /* release last item */
126
	git_buf_dispose(&buf);
127

128 129
	/* We want to ignore bad patterns, so return success regardless */
	return 0;
130 131 132 133
}

static int diff_driver_xfuncname(const git_config_entry *entry, void *payload)
{
134
	return diff_driver_add_patterns(payload, entry->value, 0);
135 136 137 138
}

static int diff_driver_funcname(const git_config_entry *entry, void *payload)
{
139
	return diff_driver_add_patterns(payload, entry->value, 0);
140 141 142 143 144 145 146 147 148 149 150 151 152 153
}

static git_diff_driver_registry *git_repository_driver_registry(
	git_repository *repo)
{
	if (!repo->diff_drivers) {
		git_diff_driver_registry *reg = git_diff_driver_registry_new();
		reg = git__compare_and_swap(&repo->diff_drivers, NULL, reg);

		if (reg != NULL) /* if we race, free losing allocation */
			git_diff_driver_registry_free(reg);
	}

	if (!repo->diff_drivers)
154
		git_error_set(GIT_ERROR_REPOSITORY, "unable to create diff driver registry");
155 156 157 158

	return repo->diff_drivers;
}

159 160 161 162 163
static int diff_driver_alloc(
	git_diff_driver **out, size_t *namelen_out, const char *name)
{
	git_diff_driver *driver;
	size_t driverlen = sizeof(git_diff_driver),
164 165
		namelen = strlen(name),
		alloclen;
166

167 168
	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, driverlen, namelen);
	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);
169

170
	driver = git__calloc(1, alloclen);
171
	GIT_ERROR_CHECK_ALLOC(driver);
172 173 174 175 176 177 178 179 180 181 182

	memcpy(driver->name, name, namelen);

	*out = driver;

	if (namelen_out)
		*namelen_out = namelen;

	return 0;
}

183 184 185 186 187 188 189
static int git_diff_driver_builtin(
	git_diff_driver **out,
	git_diff_driver_registry *reg,
	const char *driver_name)
{
	git_diff_driver_definition *ddef = NULL;
	git_diff_driver *drv = NULL;
190
	int error = 0;
191
	size_t idx;
192 193 194 195 196 197 198 199 200 201

	for (idx = 0; idx < ARRAY_SIZE(builtin_defs); ++idx) {
		if (!strcasecmp(driver_name, builtin_defs[idx].name)) {
			ddef = &builtin_defs[idx];
			break;
		}
	}
	if (!ddef)
		goto done;

202 203
	if ((error = diff_driver_alloc(&drv, NULL, ddef->name)) < 0)
		goto done;
204 205 206 207 208

	drv->type = DIFF_DRIVER_PATTERNLIST;

	if (ddef->fns &&
		(error = diff_driver_add_patterns(
209
			drv, ddef->fns, ddef->flags)) < 0)
210 211 212
		goto done;

	if (ddef->words &&
213
	    (error = git_regexp_compile(&drv->word_pattern, ddef->words, ddef->flags)) < 0)
214 215
		goto done;

216 217
	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
		goto done;
218 219

done:
220
	if (error && drv)
221
		git_diff_driver_free(drv);
222
	else
223 224 225 226 227
		*out = drv;

	return error;
}

228
static int git_diff_driver_load(
229
	git_diff_driver **out, git_repository *repo, const char *driver_name)
230
{
231
	int error = 0;
232
	git_diff_driver_registry *reg;
233 234
	git_diff_driver *drv;
	size_t namelen;
235
	git_config *cfg = NULL;
236
	git_buf name = GIT_BUF_INIT;
237
	git_config_entry *ce = NULL;
238
	bool found_driver = false;
239

240
	if ((reg = git_repository_driver_registry(repo)) == NULL)
241
		return -1;
242

243 244
	if ((drv = git_strmap_get(reg->drivers, driver_name)) != NULL) {
		*out = drv;
245
		return 0;
246 247
	}

248 249 250
	if ((error = diff_driver_alloc(&drv, &namelen, driver_name)) < 0)
		goto done;

251 252
	drv->type = DIFF_DRIVER_AUTO;

253
	/* if you can't read config for repo, just use default driver */
254
	if (git_repository_config_snapshot(&cfg, repo) < 0) {
255
		git_error_clear();
256
		goto done;
257 258 259
	}

	if ((error = git_buf_printf(&name, "diff.%s.binary", driver_name)) < 0)
260
		goto done;
261 262 263

	switch (git_config__get_bool_force(cfg, name.ptr, -1)) {
	case true:
264 265
		/* if diff.<driver>.binary is true, just return the binary driver */
		*out = &global_drivers[DIFF_DRIVER_BINARY];
266
		goto done;
267
	case false:
268 269 270
		/* if diff.<driver>.binary is false, force binary checks off */
		/* but still may have custom function context patterns, etc. */
		drv->binary_flags = GIT_DIFF_FORCE_TEXT;
271
		found_driver = true;
272 273
		break;
	default:
274
		/* diff.<driver>.binary unspecified or "auto", so just continue */
275
		break;
276 277 278 279
	}

	/* TODO: warn if diff.<name>.command or diff.<name>.textconv are set */

280
	git_buf_truncate(&name, namelen + strlen("diff.."));
281 282 283
	if ((error = git_buf_PUTS(&name, "xfuncname")) < 0)
		goto done;

284
	if ((error = git_config_get_multivar_foreach(
285 286
			cfg, name.ptr, NULL, diff_driver_xfuncname, drv)) < 0) {
		if (error != GIT_ENOTFOUND)
287
			goto done;
288
		git_error_clear(); /* no diff.<driver>.xfuncname, so just continue */
289
	}
290

291
	git_buf_truncate(&name, namelen + strlen("diff.."));
292 293 294
	if ((error = git_buf_PUTS(&name, "funcname")) < 0)
		goto done;

295
	if ((error = git_config_get_multivar_foreach(
296 297
			cfg, name.ptr, NULL, diff_driver_funcname, drv)) < 0) {
		if (error != GIT_ENOTFOUND)
298
			goto done;
299
		git_error_clear(); /* no diff.<driver>.funcname, so just continue */
300 301 302
	}

	/* if we found any patterns, set driver type to use correct callback */
303
	if (git_array_size(drv->fn_patterns) > 0) {
304
		drv->type = DIFF_DRIVER_PATTERNLIST;
305 306
		found_driver = true;
	}
307

308
	git_buf_truncate(&name, namelen + strlen("diff.."));
309 310 311
	if ((error = git_buf_PUTS(&name, "wordregex")) < 0)
		goto done;

312
	if ((error = git_config__lookup_entry(&ce, cfg, name.ptr, false)) < 0)
313
		goto done;
314 315
	if (!ce || !ce->value)
		/* no diff.<driver>.wordregex, so just continue */;
316
	else if (!(error = git_regexp_compile(&drv->word_pattern, ce->value, 0)))
317
		found_driver = true;
318 319 320
	else {
		/* TODO: warn about bad regex instead of failure */
		goto done;
321 322 323 324 325 326
	}

	/* TODO: look up diff.<driver>.algorithm to turn on minimal / patience
	 * diff in drv->other_flags
	 */

327
	/* if no driver config found at all, fall back on AUTO driver */
328
	if (!found_driver)
329
		goto done;
330 331

	/* store driver in registry */
332
	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
333
		goto done;
334

335 336
	*out = drv;

337
done:
338
	git_config_entry_free(ce);
339
	git_buf_dispose(&name);
340
	git_config_free(cfg);
341

342 343 344 345 346
	if (!*out) {
		int error2 = git_diff_driver_builtin(out, reg, driver_name);
		if (!error)
			error = error2;
	}
347 348 349 350

	if (drv && drv != *out)
		git_diff_driver_free(drv);

351
	return error;
352 353
}

354
int git_diff_driver_lookup(
355 356
	git_diff_driver **out, git_repository *repo,
	git_attr_session *attrsession, const char *path)
357
{
358
	int error = 0;
359
	const char *values[1], *attrs[] = { "diff" };
360 361

	assert(out);
362
	*out = NULL;
363 364

	if (!repo || !path || !strlen(path))
365
		/* just use the auto value */;
366 367
	else if ((error = git_attr_get_many_with_session(values, repo,
			attrsession, 0, path, 1, attrs)) < 0)
368
		/* return error below */;
369

370
	else if (GIT_ATTR_IS_UNSPECIFIED(values[0]))
371
		/* just use the auto value */;
372
	else if (GIT_ATTR_IS_FALSE(values[0]))
373
		*out = &global_drivers[DIFF_DRIVER_BINARY];
374
	else if (GIT_ATTR_IS_TRUE(values[0]))
375
		*out = &global_drivers[DIFF_DRIVER_TEXT];
376 377

	/* otherwise look for driver information in config and build driver */
378
	else if ((error = git_diff_driver_load(out, repo, values[0])) < 0) {
379 380
		if (error == GIT_ENOTFOUND) {
			error = 0;
381
			git_error_clear();
382
		}
383
	}
384

385 386 387
	if (!*out)
		*out = &global_drivers[DIFF_DRIVER_AUTO];

388
	return error;
389 390 391 392
}

void git_diff_driver_free(git_diff_driver *driver)
{
393 394 395 396 397
	size_t i;

	if (!driver)
		return;

398
	for (i = 0; i < git_array_size(driver->fn_patterns); ++i)
399
		git_regexp_dispose(& git_array_get(driver->fn_patterns, i)->re);
400 401
	git_array_clear(driver->fn_patterns);

402
	git_regexp_dispose(&driver->word_pattern);
403 404

	git__free(driver);
405 406
}

407 408
void git_diff_driver_update_options(
	uint32_t *option_flags, git_diff_driver *driver)
409
{
410 411 412 413
	if ((*option_flags & FORCE_DIFFABLE) == 0)
		*option_flags |= driver->binary_flags;

	*option_flags |= driver->other_flags;
414 415 416 417 418
}

int git_diff_driver_content_is_binary(
	git_diff_driver *driver, const char *content, size_t content_len)
{
419
	git_buf search = GIT_BUF_INIT;
420 421 422

	GIT_UNUSED(driver);

423 424 425
	git_buf_attach_notowned(&search, content,
		min(content_len, GIT_FILTER_BYTES_TO_CHECK_NUL));

426 427 428 429 430 431 432 433 434 435 436 437
	/* TODO: provide encoding / binary detection callbacks that can
	 * be UTF-8 aware, etc.  For now, instead of trying to be smart,
	 * let's just use the simple NUL-byte detection that core git uses.
	 */

	/* previously was: if (git_buf_text_is_binary(&search)) */
	if (git_buf_text_contains_nul(&search))
		return 1;

	return 0;
}

438
static int diff_context_line__simple(
439
	git_diff_driver *driver, git_buf *line)
440
{
441
	char firstch = line->ptr[0];
442
	GIT_UNUSED(driver);
443
	return (git__isalpha(firstch) || firstch == '_' || firstch == '$');
444 445 446
}

static int diff_context_line__pattern_match(
447
	git_diff_driver *driver, git_buf *line)
448
{
449
	size_t i, maxi = git_array_size(driver->fn_patterns);
450
	git_regmatch pmatch[2];
451

452
	for (i = 0; i < maxi; ++i) {
453 454
		git_diff_driver_pattern *pat = git_array_get(driver->fn_patterns, i);

455
		if (!git_regexp_search(&pat->re, line->ptr, 2, pmatch)) {
456 457
			if (pat->flags & REG_NEGATE)
				return false;
458 459

			/* use pmatch data to trim line data */
460 461 462
			i = (pmatch[1].start >= 0) ? 1 : 0;
			git_buf_consume(line, git_buf_cstr(line) + pmatch[i].start);
			git_buf_truncate(line, pmatch[i].end - pmatch[i].start);
463
			git_buf_rtrim(line);
464

465
			return true;
466
		}
467 468 469 470 471
	}

	return false;
}

472 473 474 475 476 477 478
static long diff_context_find(
	const char *line,
	long line_len,
	char *out,
	long out_size,
	void *payload)
{
479
	git_diff_find_context_payload *ctxt = payload;
480

481
	if (git_buf_set(&ctxt->line, line, (size_t)line_len) < 0)
482
		return -1;
483
	git_buf_rtrim(&ctxt->line);
484

485
	if (!ctxt->line.size)
486 487
		return -1;

488
	if (!ctxt->match_line || !ctxt->match_line(ctxt->driver, &ctxt->line))
489
		return -1;
490

491
	if (out_size > (long)ctxt->line.size)
Russell Belfer committed
492
		out_size = (long)ctxt->line.size;
493
	memcpy(out, ctxt->line.ptr, (size_t)out_size);
494

495
	return out_size;
496
}
497

498 499 500 501 502 503 504 505 506 507 508 509 510 511
void git_diff_find_context_init(
	git_diff_find_context_fn *findfn_out,
	git_diff_find_context_payload *payload_out,
	git_diff_driver *driver)
{
	*findfn_out = driver ? diff_context_find : NULL;

	memset(payload_out, 0, sizeof(*payload_out));
	if (driver) {
		payload_out->driver = driver;
		payload_out->match_line = (driver->type == DIFF_DRIVER_PATTERNLIST) ?
			diff_context_line__pattern_match : diff_context_line__simple;
		git_buf_init(&payload_out->line, 0);
	}
512 513
}

514
void git_diff_find_context_clear(git_diff_find_context_payload *payload)
515
{
516
	if (payload) {
517
		git_buf_dispose(&payload->line);
518 519
		payload->driver = NULL;
	}
520
}