filter.c 23.7 KB
Newer Older
1
/*
Edward Thomson committed
2
 * Copyright (C) the libgit2 contributors. All rights reserved.
3 4 5 6 7
 *
 * This file is part of libgit2, distributed under the GNU GPL v2 with
 * a Linking Exception. For full terms see the included COPYING file.
 */

8 9
#include "filter.h"

10
#include "common.h"
11
#include "futils.h"
12
#include "hash.h"
13
#include "repository.h"
14
#include "runtime.h"
15
#include "git2/sys/filter.h"
16
#include "git2/config.h"
17
#include "blob.h"
18
#include "attr_file.h"
19
#include "array.h"
20

21 22 23 24 25
struct git_filter_source {
	git_repository *repo;
	const char     *path;
	git_oid         oid;  /* zero if unknown (which is likely) */
	uint16_t        filemode; /* zero if unknown */
26
	git_filter_mode_t mode;
27
	uint32_t        flags;
28 29
};

30
typedef struct {
31
	const char *filter_name;
32 33 34 35 36 37 38
	git_filter *filter;
	void *payload;
} git_filter_entry;

struct git_filter_list {
	git_array_t(git_filter_entry) filters;
	git_filter_source source;
39
	git_buf *temp_buf;
40 41 42 43
	char path[GIT_FLEX_ARRAY];
};

typedef struct {
44
	char *filter_name;
45
	git_filter *filter;
46
	int priority;
47
	int initialized;
48 49 50
	size_t nattrs, nmatches;
	char *attrdata;
	const char *attrs[GIT_FLEX_ARRAY];
51 52
} git_filter_def;

53 54 55 56 57 58 59
static int filter_def_priority_cmp(const void *a, const void *b)
{
	int pa = ((const git_filter_def *)a)->priority;
	int pb = ((const git_filter_def *)b)->priority;
	return (pa < pb) ? -1 : (pa > pb) ? 1 : 0;
}

60 61
struct git_filter_registry {
	git_rwlock lock;
62
	git_vector filters;
63 64
};

65
static struct git_filter_registry filter_registry;
66

67
static void git_filter_global_shutdown(void);
Russell Belfer committed
68

69

70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
static int filter_def_scan_attrs(
	git_buf *attrs, size_t *nattr, size_t *nmatch, const char *attr_str)
{
	const char *start, *scan = attr_str;
	int has_eq;

	*nattr = *nmatch = 0;

	if (!scan)
		return 0;

	while (*scan) {
		while (git__isspace(*scan)) scan++;

		for (start = scan, has_eq = 0; *scan && !git__isspace(*scan); ++scan) {
			if (*scan == '=')
				has_eq = 1;
		}

		if (scan > start) {
			(*nattr)++;
Russell Belfer committed
91
			if (has_eq || *start == '-' || *start == '+' || *start == '!')
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
				(*nmatch)++;

			if (has_eq)
				git_buf_putc(attrs, '=');
			git_buf_put(attrs, start, scan - start);
			git_buf_putc(attrs, '\0');
		}
	}

	return 0;
}

static void filter_def_set_attrs(git_filter_def *fdef)
{
	char *scan = fdef->attrdata;
	size_t i;

	for (i = 0; i < fdef->nattrs; ++i) {
		const char *name, *value;

		switch (*scan) {
		case '=':
			name = scan + 1;
			for (scan++; *scan != '='; scan++) /* find '=' */;
			*scan++ = '\0';
			value = scan;
			break;
		case '-':
			name = scan + 1; value = git_attr__false; break;
		case '+':
			name = scan + 1; value = git_attr__true;  break;
		case '!':
			name = scan + 1; value = git_attr__unset; break;
		default:
			name = scan;     value = NULL; break;
		}

		fdef->attrs[i] = name;
		fdef->attrs[i + fdef->nattrs] = value;

		scan += strlen(scan) + 1;
	}
}

136 137 138 139
static int filter_def_name_key_check(const void *key, const void *fdef)
{
	const char *name =
		fdef ? ((const git_filter_def *)fdef)->filter_name : NULL;
140 141 142 143 144 145 146
	return name ? git__strcmp(key, name) : -1;
}

static int filter_def_filter_key_check(const void *key, const void *fdef)
{
	const void *filter = fdef ? ((const git_filter_def *)fdef)->filter : NULL;
	return (key == filter) ? 0 : -1;
147 148
}

149 150
/* Note: callers must lock the registry before calling this function */
static int filter_registry_insert(
151 152 153
	const char *name, git_filter *filter, int priority)
{
	git_filter_def *fdef;
154
	size_t nattr = 0, nmatch = 0, alloc_len;
155 156 157 158 159
	git_buf attrs = GIT_BUF_INIT;

	if (filter_def_scan_attrs(&attrs, &nattr, &nmatch, filter->attributes) < 0)
		return -1;

160 161 162
	GIT_ERROR_CHECK_ALLOC_MULTIPLY(&alloc_len, nattr, 2);
	GIT_ERROR_CHECK_ALLOC_MULTIPLY(&alloc_len, alloc_len, sizeof(char *));
	GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, sizeof(git_filter_def));
163 164

	fdef = git__calloc(1, alloc_len);
165
	GIT_ERROR_CHECK_ALLOC(fdef);
166

167
	fdef->filter_name = git__strdup(name);
168
	GIT_ERROR_CHECK_ALLOC(fdef->filter_name);
169

170 171 172 173 174 175 176 177
	fdef->filter      = filter;
	fdef->priority    = priority;
	fdef->nattrs      = nattr;
	fdef->nmatches    = nmatch;
	fdef->attrdata    = git_buf_detach(&attrs);

	filter_def_set_attrs(fdef);

178
	if (git_vector_insert(&filter_registry.filters, fdef) < 0) {
179
		git__free(fdef->filter_name);
180 181 182 183 184
		git__free(fdef->attrdata);
		git__free(fdef);
		return -1;
	}

185
	git_vector_sort(&filter_registry.filters);
186 187 188
	return 0;
}

189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
int git_filter_global_init(void)
{
	git_filter *crlf = NULL, *ident = NULL;
	int error = 0;

	if (git_rwlock_init(&filter_registry.lock) < 0)
		return -1;

	if ((error = git_vector_init(&filter_registry.filters, 2,
		filter_def_priority_cmp)) < 0)
		goto done;

	if ((crlf = git_crlf_filter_new()) == NULL ||
		filter_registry_insert(
			GIT_FILTER_CRLF, crlf, GIT_FILTER_CRLF_PRIORITY) < 0 ||
		(ident = git_ident_filter_new()) == NULL ||
		filter_registry_insert(
			GIT_FILTER_IDENT, ident, GIT_FILTER_IDENT_PRIORITY) < 0)
		error = -1;

209
	error = git_runtime_shutdown_register(git_filter_global_shutdown);
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268

done:
	if (error) {
		git_filter_free(crlf);
		git_filter_free(ident);
	}

	return error;
}

static void git_filter_global_shutdown(void)
{
	size_t pos;
	git_filter_def *fdef;

	if (git_rwlock_wrlock(&filter_registry.lock) < 0)
		return;

	git_vector_foreach(&filter_registry.filters, pos, fdef) {
		if (fdef->filter && fdef->filter->shutdown) {
			fdef->filter->shutdown(fdef->filter);
			fdef->initialized = false;
		}

		git__free(fdef->filter_name);
		git__free(fdef->attrdata);
		git__free(fdef);
	}

	git_vector_free(&filter_registry.filters);

	git_rwlock_wrunlock(&filter_registry.lock);
	git_rwlock_free(&filter_registry.lock);
}

/* Note: callers must lock the registry before calling this function */
static int filter_registry_find(size_t *pos, const char *name)
{
	return git_vector_search2(
		pos, &filter_registry.filters, filter_def_name_key_check, name);
}

/* Note: callers must lock the registry before calling this function */
static git_filter_def *filter_registry_lookup(size_t *pos, const char *name)
{
	git_filter_def *fdef = NULL;

	if (!filter_registry_find(pos, name))
		fdef = git_vector_get(&filter_registry.filters, *pos);

	return fdef;
}


int git_filter_register(
	const char *name, git_filter *filter, int priority)
{
	int error;

269 270
	GIT_ASSERT_ARG(name);
	GIT_ASSERT_ARG(filter);
271 272

	if (git_rwlock_wrlock(&filter_registry.lock) < 0) {
273
		git_error_set(GIT_ERROR_OS, "failed to lock filter registry");
274 275 276 277
		return -1;
	}

	if (!filter_registry_find(NULL, name)) {
278 279
		git_error_set(
			GIT_ERROR_FILTER, "attempt to reregister existing filter '%s'", name);
280 281 282 283 284 285 286 287 288 289 290
		error = GIT_EEXISTS;
		goto done;
	}

	error = filter_registry_insert(name, filter, priority);

done:
	git_rwlock_wrunlock(&filter_registry.lock);
	return error;
}

291 292 293 294
int git_filter_unregister(const char *name)
{
	size_t pos;
	git_filter_def *fdef;
295
	int error = 0;
296

297
	GIT_ASSERT_ARG(name);
298

299
	/* cannot unregister default filters */
Russell Belfer committed
300
	if (!strcmp(GIT_FILTER_CRLF, name) || !strcmp(GIT_FILTER_IDENT, name)) {
301
		git_error_set(GIT_ERROR_FILTER, "cannot unregister filter '%s'", name);
302 303 304
		return -1;
	}

305
	if (git_rwlock_wrlock(&filter_registry.lock) < 0) {
306
		git_error_set(GIT_ERROR_OS, "failed to lock filter registry");
307 308 309
		return -1;
	}

310
	if ((fdef = filter_registry_lookup(&pos, name)) == NULL) {
311
		git_error_set(GIT_ERROR_FILTER, "cannot find filter '%s' to unregister", name);
312 313
		error = GIT_ENOTFOUND;
		goto done;
314 315
	}

316
	git_vector_remove(&filter_registry.filters, pos);
317

318
	if (fdef->initialized && fdef->filter && fdef->filter->shutdown) {
319
		fdef->filter->shutdown(fdef->filter);
320 321
		fdef->initialized = false;
	}
322

323
	git__free(fdef->filter_name);
324 325 326
	git__free(fdef->attrdata);
	git__free(fdef);

327 328 329
done:
	git_rwlock_wrunlock(&filter_registry.lock);
	return error;
330 331
}

332 333 334 335
static int filter_initialize(git_filter_def *fdef)
{
	int error = 0;

336 337 338
	if (!fdef->initialized && fdef->filter && fdef->filter->initialize) {
		if ((error = fdef->filter->initialize(fdef->filter)) < 0)
			return error;
339 340 341 342 343 344
	}

	fdef->initialized = true;
	return 0;
}

345 346 347
git_filter *git_filter_lookup(const char *name)
{
	size_t pos;
348
	git_filter_def *fdef;
349
	git_filter *filter = NULL;
350

351
	if (git_rwlock_rdlock(&filter_registry.lock) < 0) {
352
		git_error_set(GIT_ERROR_OS, "failed to lock filter registry");
353
		return NULL;
354
	}
355

356 357 358
	if ((fdef = filter_registry_lookup(&pos, name)) == NULL ||
		(!fdef->initialized && filter_initialize(fdef) < 0))
		goto done;
359

360
	filter = fdef->filter;
361

362 363 364
done:
	git_rwlock_rdunlock(&filter_registry.lock);
	return filter;
365 366
}

Russell Belfer committed
367 368 369 370 371
void git_filter_free(git_filter *filter)
{
	git__free(filter);
}

372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388
git_repository *git_filter_source_repo(const git_filter_source *src)
{
	return src->repo;
}

const char *git_filter_source_path(const git_filter_source *src)
{
	return src->path;
}

uint16_t git_filter_source_filemode(const git_filter_source *src)
{
	return src->filemode;
}

const git_oid *git_filter_source_id(const git_filter_source *src)
{
389
	return git_oid_is_zero(&src->oid) ? NULL : &src->oid;
390 391
}

392 393 394 395 396
git_filter_mode_t git_filter_source_mode(const git_filter_source *src)
{
	return src->mode;
}

397
uint32_t git_filter_source_flags(const git_filter_source *src)
398
{
399
	return src->flags;
400 401
}

402
static int filter_list_new(
403
	git_filter_list **out, const git_filter_source *src)
404
{
405
	git_filter_list *fl = NULL;
406
	size_t pathlen = src->path ? strlen(src->path) : 0, alloclen;
407

408 409
	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, sizeof(git_filter_list), pathlen);
	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);
410

411
	fl = git__calloc(1, alloclen);
412
	GIT_ERROR_CHECK_ALLOC(fl);
413 414 415 416 417

	if (src->path)
		memcpy(fl->path, src->path, pathlen);
	fl->source.repo = src->repo;
	fl->source.path = fl->path;
418
	fl->source.mode = src->mode;
419
	fl->source.flags = src->flags;
420 421 422 423 424

	*out = fl;
	return 0;
}

425
static int filter_list_check_attributes(
426 427 428 429 430
	const char ***out,
	git_repository *repo,
	git_attr_session *attr_session,
	git_filter_def *fdef,
	const git_filter_source *src)
431 432
{
	const char **strs = git__calloc(fdef->nattrs, sizeof(const char *));
433 434 435 436
	uint32_t flags = 0;
	size_t i;
	int error;

437
	GIT_ERROR_CHECK_ALLOC(strs);
438

439 440 441
	if ((src->flags & GIT_FILTER_NO_SYSTEM_ATTRIBUTES) != 0)
		flags |= GIT_ATTR_CHECK_NO_SYSTEM;

442 443 444
	if ((src->flags & GIT_FILTER_ATTRIBUTES_FROM_HEAD) != 0)
		flags |= GIT_ATTR_CHECK_INCLUDE_HEAD;

445
	error = git_attr_get_many_with_session(
446
		strs, repo, attr_session, flags, src->path, fdef->nattrs, fdef->attrs);
447 448 449

	/* if no values were found but no matches are needed, it's okay! */
	if (error == GIT_ENOTFOUND && !fdef->nmatches) {
450
		git_error_clear();
Russell Belfer committed
451
		git__free((void *)strs);
452 453 454 455 456
		return 0;
	}

	for (i = 0; !error && i < fdef->nattrs; ++i) {
		const char *want = fdef->attrs[fdef->nattrs + i];
457
		git_attr_value_t want_type, found_type;
458 459 460 461 462 463 464

		if (!want)
			continue;

		want_type  = git_attr_value(want);
		found_type = git_attr_value(strs[i]);

465 466
		if (want_type != found_type)
			error = GIT_ENOTFOUND;
467
		else if (want_type == GIT_ATTR_VALUE_STRING &&
468 469
				strcmp(want, strs[i]) &&
				strcmp(want, "*"))
470 471 472 473
			error = GIT_ENOTFOUND;
	}

	if (error)
Russell Belfer committed
474
		git__free((void *)strs);
475 476 477 478 479 480
	else
		*out = strs;

	return error;
}

481
int git_filter_list_new(
482 483 484
	git_filter_list **out,
	git_repository *repo,
	git_filter_mode_t mode,
485
	uint32_t flags)
486 487 488 489 490
{
	git_filter_source src = { 0 };
	src.repo = repo;
	src.path = NULL;
	src.mode = mode;
491
	src.flags = flags;
492 493 494
	return filter_list_new(out, &src);
}

495
int git_filter_list__load_ext(
496 497
	git_filter_list **filters,
	git_repository *repo,
Russell Belfer committed
498
	git_blob *blob, /* can be NULL */
499
	const char *path,
500
	git_filter_mode_t mode,
501
	git_filter_options *filter_opts)
502 503 504 505 506
{
	int error = 0;
	git_filter_list *fl = NULL;
	git_filter_source src = { 0 };
	git_filter_entry *fe;
507 508
	size_t idx;
	git_filter_def *fdef;
509

510
	if (git_rwlock_rdlock(&filter_registry.lock) < 0) {
511
		git_error_set(GIT_ERROR_OS, "failed to lock filter registry");
512
		return -1;
513
	}
514 515 516

	src.repo = repo;
	src.path = path;
517
	src.mode = mode;
518 519
	src.flags = filter_opts->flags;

Russell Belfer committed
520 521
	if (blob)
		git_oid_cpy(&src.oid, git_blob_id(blob));
522

523
	git_vector_foreach(&filter_registry.filters, idx, fdef) {
524
		const char **values = NULL;
525 526 527 528
		void *payload = NULL;

		if (!fdef || !fdef->filter)
			continue;
529

530
		if (fdef->nattrs > 0) {
531
			error = filter_list_check_attributes(
532
				&values, repo, filter_opts->attr_session, fdef, &src);
533

534 535 536 537 538 539 540
			if (error == GIT_ENOTFOUND) {
				error = 0;
				continue;
			} else if (error < 0)
				break;
		}

541 542 543
		if (!fdef->initialized && (error = filter_initialize(fdef)) < 0)
			break;

544
		if (fdef->filter->check)
545
			error = fdef->filter->check(
546
				fdef->filter, &payload, &src, values);
547

Russell Belfer committed
548
		git__free((void *)values);
549

Russell Belfer committed
550
		if (error == GIT_PASSTHROUGH)
551 552 553 554
			error = 0;
		else if (error < 0)
			break;
		else {
555 556
			if (!fl) {
				if ((error = filter_list_new(&fl, &src)) < 0)
557
					break;
558 559 560

				fl->temp_buf = filter_opts->temp_buf;
			}
561 562

			fe = git_array_alloc(fl->filters);
563
			GIT_ERROR_CHECK_ALLOC(fe);
564 565 566

			fe->filter = fdef->filter;
			fe->filter_name = fdef->filter_name;
567 568 569 570
			fe->payload = payload;
		}
	}

571 572
	git_rwlock_rdunlock(&filter_registry.lock);

573 574 575 576 577 578 579 580 581 582
	if (error && fl != NULL) {
		git_array_clear(fl->filters);
		git__free(fl);
		fl = NULL;
	}

	*filters = fl;
	return error;
}

583 584 585 586 587 588
int git_filter_list_load(
	git_filter_list **filters,
	git_repository *repo,
	git_blob *blob, /* can be NULL */
	const char *path,
	git_filter_mode_t mode,
589
	uint32_t flags)
590
{
591 592 593 594 595 596
	git_filter_options filter_opts = GIT_FILTER_OPTIONS_INIT;

	filter_opts.flags = flags;

	return git_filter_list__load_ext(
		filters, repo, blob, path, mode, &filter_opts);
597 598
}

599 600 601 602 603 604 605 606 607 608 609
void git_filter_list_free(git_filter_list *fl)
{
	uint32_t i;

	if (!fl)
		return;

	for (i = 0; i < git_array_size(fl->filters); ++i) {
		git_filter_entry *fe = git_array_get(fl->filters, i);
		if (fe->filter->cleanup)
			fe->filter->cleanup(fe->filter, fe->payload);
610 611
	}

612 613
	git_array_clear(fl->filters);
	git__free(fl);
614 615
}

616 617 618 619 620 621
int git_filter_list_contains(
	git_filter_list *fl,
	const char *name)
{
	size_t i;

622
	GIT_ASSERT_ARG(name);
623 624 625 626 627 628 629 630 631 632 633 634

	if (!fl)
		return 0;

	for (i = 0; i < fl->filters.size; i++) {
		if (strcmp(fl->filters.ptr[i].filter_name, name) == 0)
			return 1;
	}

	return 0;
}

635 636 637 638 639
int git_filter_list_push(
	git_filter_list *fl, git_filter *filter, void *payload)
{
	int error = 0;
	size_t pos;
640
	git_filter_def *fdef = NULL;
641 642
	git_filter_entry *fe;

643 644
	GIT_ASSERT_ARG(fl);
	GIT_ASSERT_ARG(filter);
645

646
	if (git_rwlock_rdlock(&filter_registry.lock) < 0) {
647
		git_error_set(GIT_ERROR_OS, "failed to lock filter registry");
648 649 650
		return -1;
	}

651
	if (git_vector_search2(
652 653 654 655 656 657 658
			&pos, &filter_registry.filters,
			filter_def_filter_key_check, filter) == 0)
		fdef = git_vector_get(&filter_registry.filters, pos);

	git_rwlock_rdunlock(&filter_registry.lock);

	if (fdef == NULL) {
659
		git_error_set(GIT_ERROR_FILTER, "cannot use an unregistered filter");
660 661 662 663 664 665 666
		return -1;
	}

	if (!fdef->initialized && (error = filter_initialize(fdef)) < 0)
		return error;

	fe = git_array_alloc(fl->filters);
667
	GIT_ERROR_CHECK_ALLOC(fe);
668 669 670 671 672 673
	fe->filter  = filter;
	fe->payload = payload;

	return 0;
}

Russell Belfer committed
674 675 676 677 678
size_t git_filter_list_length(const git_filter_list *fl)
{
	return fl ? git_array_size(fl->filters) : 0;
}

679
struct buf_stream {
680
	git_writestream parent;
681 682 683 684 685
	git_buf *target;
	bool complete;
};

static int buf_stream_write(
686
	git_writestream *s, const char *buffer, size_t len)
687
{
688
	struct buf_stream *buf_stream = (struct buf_stream *)s;
689 690
	GIT_ASSERT_ARG(buf_stream);
	GIT_ASSERT(buf_stream->complete == 0);
691

692 693
	return git_buf_put(buf_stream->target, buffer, len);
}
694

695
static int buf_stream_close(git_writestream *s)
696 697
{
	struct buf_stream *buf_stream = (struct buf_stream *)s;
698
	GIT_ASSERT_ARG(buf_stream);
699

700
	GIT_ASSERT(buf_stream->complete == 0);
701
	buf_stream->complete = 1;
702

703 704
	return 0;
}
705

706
static void buf_stream_free(git_writestream *s)
707 708 709
{
	GIT_UNUSED(s);
}
710

711 712 713
static void buf_stream_init(struct buf_stream *writer, git_buf *target)
{
	memset(writer, 0, sizeof(struct buf_stream));
714

715 716 717
	writer->parent.write = buf_stream_write;
	writer->parent.close = buf_stream_close;
	writer->parent.free = buf_stream_free;
718
	writer->target = target;
719

720 721
	git_buf_clear(target);
}
722

723
int git_filter_list_apply_to_buffer(
724 725 726 727
	git_buf *out,
	git_filter_list *filters,
	const char *in,
	size_t in_len)
728 729 730 731
{
	struct buf_stream writer;
	int error;

732 733
	if ((error = git_buf_sanitize(out)) < 0)
		return error;
734

735
	buf_stream_init(&writer, out);
736

737
	if ((error = git_filter_list_stream_buffer(filters,
738
		in, in_len, &writer.parent)) < 0)
739 740
			return error;

741
	GIT_ASSERT(writer.complete);
742
	return error;
743
}
744

745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766
int git_filter_list__convert_buf(
	git_buf *out,
	git_filter_list *filters,
	git_buf *in)
{
	int error;

	if (!filters || git_filter_list_length(filters) == 0) {
		git_buf_swap(out, in);
		git_buf_dispose(in);
		return 0;
	}

	error = git_filter_list_apply_to_buffer(out, filters,
		in->ptr, in->size);

	if (!error)
		git_buf_dispose(in);

	return error;
}

767
int git_filter_list_apply_to_file(
768
	git_buf *out,
769 770 771 772
	git_filter_list *filters,
	git_repository *repo,
	const char *path)
{
773
	struct buf_stream writer;
774 775
	int error;

776
	buf_stream_init(&writer, out);
777

778
	if ((error = git_filter_list_stream_file(
Leo Yang committed
779
		filters, repo, path, &writer.parent)) < 0)
780
			return error;
781

782
	GIT_ASSERT(writer.complete);
783 784 785
	return error;
}

786 787
static int buf_from_blob(git_buf *out, git_blob *blob)
{
788
	git_object_size_t rawsize = git_blob_rawsize(blob);
789 790

	if (!git__is_sizet(rawsize)) {
791
		git_error_set(GIT_ERROR_OS, "blob is too large to filter");
792 793 794
		return -1;
	}

795
	git_buf_attach_notowned(out, git_blob_rawcontent(blob), (size_t)rawsize);
796 797 798
	return 0;
}

799
int git_filter_list_apply_to_blob(
800
	git_buf *out,
801 802 803
	git_filter_list *filters,
	git_blob *blob)
{
804 805
	struct buf_stream writer;
	int error;
806

807
	buf_stream_init(&writer, out);
808

809
	if ((error = git_filter_list_stream_blob(
Leo Yang committed
810
		filters, blob, &writer.parent)) < 0)
811
			return error;
812

813
	GIT_ASSERT(writer.complete);
814
	return error;
815 816 817
}

struct proxy_stream {
818
	git_writestream parent;
819 820 821 822
	git_filter *filter;
	const git_filter_source *source;
	void **payload;
	git_buf input;
823 824
	git_buf temp_buf;
	git_buf *output;
825
	git_writestream *target;
826 827 828
};

static int proxy_stream_write(
829
	git_writestream *s, const char *buffer, size_t len)
830 831
{
	struct proxy_stream *proxy_stream = (struct proxy_stream *)s;
832
	GIT_ASSERT_ARG(proxy_stream);
833 834 835 836

	return git_buf_put(&proxy_stream->input, buffer, len);
}

837
static int proxy_stream_close(git_writestream *s)
838 839 840
{
	struct proxy_stream *proxy_stream = (struct proxy_stream *)s;
	git_buf *writebuf;
841
	git_error_state error_state = {0};
842 843
	int error;

844
	GIT_ASSERT_ARG(proxy_stream);
845 846 847 848

	error = proxy_stream->filter->apply(
		proxy_stream->filter,
		proxy_stream->payload,
849
		proxy_stream->output,
850 851 852 853 854 855
		&proxy_stream->input,
		proxy_stream->source);

	if (error == GIT_PASSTHROUGH) {
		writebuf = &proxy_stream->input;
	} else if (error == 0) {
856 857 858
		if ((error = git_buf_sanitize(proxy_stream->output)) < 0)
			return error;

859
		writebuf = proxy_stream->output;
860
	} else {
861 862
		/* close stream before erroring out taking care
		 * to preserve the original error */
863
		git_error_state_capture(&error_state, error);
864
		proxy_stream->target->close(proxy_stream->target);
865
		git_error_state_restore(&error_state);
866 867 868 869 870 871 872 873 874 875
		return error;
	}

	if ((error = proxy_stream->target->write(
			proxy_stream->target, writebuf->ptr, writebuf->size)) == 0)
		error = proxy_stream->target->close(proxy_stream->target);

	return error;
}

876
static void proxy_stream_free(git_writestream *s)
877 878 879
{
	struct proxy_stream *proxy_stream = (struct proxy_stream *)s;

880 881 882 883 884
	if (proxy_stream) {
		git_buf_dispose(&proxy_stream->input);
		git_buf_dispose(&proxy_stream->temp_buf);
		git__free(proxy_stream);
	}
885 886 887
}

static int proxy_stream_init(
888
	git_writestream **out,
889
	git_filter *filter,
890
	git_buf *temp_buf,
891 892
	void **payload,
	const git_filter_source *source,
893
	git_writestream *target)
894 895
{
	struct proxy_stream *proxy_stream = git__calloc(1, sizeof(struct proxy_stream));
896
	GIT_ERROR_CHECK_ALLOC(proxy_stream);
897

898 899 900
	proxy_stream->parent.write = proxy_stream_write;
	proxy_stream->parent.close = proxy_stream_close;
	proxy_stream->parent.free = proxy_stream_free;
901 902 903 904
	proxy_stream->filter = filter;
	proxy_stream->payload = payload;
	proxy_stream->source = source;
	proxy_stream->target = target;
905
	proxy_stream->output = temp_buf ? temp_buf : &proxy_stream->temp_buf;
906

907 908 909
	if (temp_buf)
		git_buf_clear(temp_buf);

910
	*out = (git_writestream *)proxy_stream;
911 912 913 914
	return 0;
}

static int stream_list_init(
915
	git_writestream **out,
916 917
	git_vector *streams,
	git_filter_list *filters,
918
	git_writestream *target)
919
{
920
	git_writestream *last_stream = target;
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935
	size_t i;
	int error = 0;

	*out = NULL;

	if (!filters) {
		*out = target;
		return 0;
	}

	/* Create filters last to first to get the chaining direction */
	for (i = 0; i < git_array_size(filters->filters); ++i) {
		size_t filter_idx = (filters->source.mode == GIT_FILTER_TO_WORKTREE) ?
			git_array_size(filters->filters) - 1 - i : i;
		git_filter_entry *fe = git_array_get(filters->filters, filter_idx);
936
		git_writestream *filter_stream;
937

938
		GIT_ASSERT(fe->filter->stream || fe->filter->apply);
939

940 941 942
		/* If necessary, create a stream that proxies the traditional
		 * application.
		 */
943 944
		if (fe->filter->stream)
			error = fe->filter->stream(&filter_stream, fe->filter,
945
				&fe->payload, &filters->source, last_stream);
946 947 948 949 950
		else
			/* Create a stream that proxies the one-shot apply */
			error = proxy_stream_init(&filter_stream, fe->filter,
				filters->temp_buf, &fe->payload, &filters->source,
				last_stream);
951

952
		if (error < 0)
953
			goto out;
954 955

		git_vector_insert(streams, filter_stream);
956
		last_stream = filter_stream;
957 958
	}

959 960 961 962 963 964 965
out:
	if (error)
		last_stream->close(last_stream);
	else
		*out = last_stream;

	return error;
966 967
}

968
static void filter_streams_free(git_vector *streams)
969
{
970
	git_writestream *stream;
971 972 973 974
	size_t i;

	git_vector_foreach(streams, i, stream)
		stream->free(stream);
975
	git_vector_free(streams);
976 977 978 979 980 981
}

int git_filter_list_stream_file(
	git_filter_list *filters,
	git_repository *repo,
	const char *path,
982
	git_writestream *target)
983
{
984
	char buf[FILTERIO_BUFSIZE];
985
	git_buf abspath = GIT_BUF_INIT;
986 987
	const char *base = repo ? git_repository_workdir(repo) : NULL;
	git_vector filter_streams = GIT_VECTOR_INIT;
988
	git_writestream *stream_start;
989
	ssize_t readlen;
990
	int fd = -1, error, initialized = 0;
991 992 993

	if ((error = stream_list_init(
			&stream_start, &filter_streams, filters, target)) < 0 ||
994 995
	    (error = git_path_join_unrooted(&abspath, path, base, NULL)) < 0 ||
	    (error = git_path_validate_workdir_buf(repo, &abspath)) < 0)
996
		goto done;
997

998
	initialized = 1;
999

1000
	if ((fd = git_futils_open_ro(abspath.ptr)) < 0) {
1001 1002 1003 1004
		error = fd;
		goto done;
	}

1005
	while ((readlen = p_read(fd, buf, sizeof(buf))) > 0) {
1006
		if ((error = stream_start->write(stream_start, buf, readlen)) < 0)
1007 1008 1009
			goto done;
	}

1010
	if (readlen < 0)
Edward Thomson committed
1011
		error = -1;
1012 1013

done:
1014 1015 1016
	if (initialized)
		error |= stream_start->close(stream_start);

1017 1018
	if (fd >= 0)
		p_close(fd);
1019
	filter_streams_free(&filter_streams);
1020
	git_buf_dispose(&abspath);
1021 1022 1023
	return error;
}

1024
int git_filter_list_stream_buffer(
1025
	git_filter_list *filters,
1026 1027
	const char *buffer,
	size_t len,
1028
	git_writestream *target)
1029 1030
{
	git_vector filter_streams = GIT_VECTOR_INIT;
1031
	git_writestream *stream_start;
1032
	int error, initialized = 0;
1033

1034 1035
	if ((error = stream_list_init(&stream_start, &filter_streams, filters, target)) < 0)
		goto out;
1036
	initialized = 1;
1037

1038
	if ((error = stream_start->write(stream_start, buffer, len)) < 0)
1039
		goto out;
1040 1041

out:
1042 1043 1044
	if (initialized)
		error |= stream_start->close(stream_start);

1045
	filter_streams_free(&filter_streams);
1046
	return error;
1047 1048 1049 1050 1051
}

int git_filter_list_stream_blob(
	git_filter_list *filters,
	git_blob *blob,
1052
	git_writestream *target)
1053 1054 1055 1056 1057
{
	git_buf in = GIT_BUF_INIT;

	if (buf_from_blob(&in, blob) < 0)
		return -1;
1058

Russell Belfer committed
1059 1060 1061
	if (filters)
		git_oid_cpy(&filters->source.oid, git_blob_id(blob));

1062
	return git_filter_list_stream_buffer(filters, in.ptr, in.size, target);
1063
}
1064 1065 1066 1067 1068 1069

int git_filter_init(git_filter *filter, unsigned int version)
{
	GIT_INIT_STRUCTURE_FROM_TEMPLATE(filter, version, git_filter, GIT_FILTER_INIT);
	return 0;
}
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085

#ifndef GIT_DEPRECATE_HARD

int git_filter_list_stream_data(
	git_filter_list *filters,
	git_buf *data,
	git_writestream *target)
{
	int error;

	if ((error = git_buf_sanitize(data)) < 0)
		return error;

	return git_filter_list_stream_buffer(filters, data->ptr, data->size, target);
}

1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096
int git_filter_list_apply_to_data(
	git_buf *tgt, git_filter_list *filters, git_buf *src)
{
	int error;

	if ((error = git_buf_sanitize(src)) < 0)
	    return error;

	return git_filter_list_apply_to_buffer(tgt, filters, src->ptr, src->size);
}

1097
#endif