op public repos

Blob

Date:: Fri Dec 1 23:45:51 2017 UTC
Message:: don't overwrite already decompressed data
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2017 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/stat.h>
18 #include <sys/queue.h>
19 
20 #include <stdio.h>
21 #include <stdlib.h>
22 #include <string.h>
23 #include <sha1.h>
24 #include <zlib.h>
25 #include <ctype.h>
26 #include <limits.h>
27 
28 #include "got_error.h"
29 #include "got_object.h"
30 #include "got_repository.h"
31 #include "got_sha1.h"
32 
33 #ifndef MIN
34 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
35 #endif
36 
37 #ifndef nitems
38 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
39 #endif
40 
41 #define GOT_OBJ_TAG_COMMIT	"commit"
42 #define GOT_OBJ_TAG_TREE	"tree"
43 #define GOT_OBJ_TAG_BLOB	"blob"
44 
45 #define GOT_COMMIT_TAG_TREE		"tree "
46 #define GOT_COMMIT_TAG_PARENT		"parent "
47 #define GOT_COMMIT_TAG_AUTHOR		"author "
48 #define GOT_COMMIT_TAG_COMMITTER	"committer "
49 
50 char *
51 got_object_id_str(struct got_object_id *id, char *buf, size_t size)
52 {
53 	char *p = buf;
54 	char hex[3];
55 	int i;
56 
57 	if (size < SHA1_DIGEST_STRING_LENGTH)
58 		return NULL;
59 
60 	for (i = 0; i < SHA1_DIGEST_LENGTH; i++) {
61 		snprintf(hex, sizeof(hex), "%.2x", id->sha1[i]);
62 		p[0] = hex[0];
63 		p[1] = hex[1];
64 		p += 2;
65 	}
66 	p[0] = '\0';
67 
68 	return buf;
69 }
70 
71 static void
72 inflate_end(struct got_zstream_buf *zb)
73 {
74 	free(zb->inbuf);
75 	free(zb->outbuf);
76 	inflateEnd(&zb->z);
77 }
78 
79 static const struct got_error *
80 inflate_init(struct got_zstream_buf *zb, size_t bufsize)
81 {
82 	const struct got_error *err = NULL;
83 
84 	memset(zb, 0, sizeof(*zb));
85 
86 	zb->z.zalloc = Z_NULL;
87 	zb->z.zfree = Z_NULL;
88 	if (inflateInit(&zb->z) != Z_OK) {
89 		err = got_error(GOT_ERR_IO);
90 		goto done;
91 	}
92 
93 	zb->inlen = zb->outlen = bufsize;
94 
95 	zb->inbuf = calloc(1, zb->inlen);
96 	if (zb->inbuf == NULL) {
97 		err = got_error(GOT_ERR_NO_MEM);
98 		goto done;
99 	}
100 
101 	zb->outbuf = calloc(1, zb->outlen);
102 	if (zb->outbuf == NULL) {
103 		err = got_error(GOT_ERR_NO_MEM);
104 		goto done;
105 	}
106 
107 done:
108 	if (err)
109 		inflate_end(zb);
110 	return err;
111 }
112 
113 static const struct got_error *
114 inflate_read(struct got_zstream_buf *zb, FILE *f, size_t *outlenp)
115 {
116 	size_t last_total_out = zb->z.total_out;
117 	z_stream *z = &zb->z;
118 	int n, ret;
119 
120 	z->next_out = zb->outbuf;
121 	z->avail_out = zb->outlen;
122 
123 	do {
124 		if (z->avail_in == 0) {
125 			int i;
126 			n = fread(zb->inbuf, 1, zb->inlen, f);
127 			if (n == 0) {
128 				if (ferror(f))
129 					return got_error(GOT_ERR_IO);
130 				*outlenp = 0;
131 				return NULL;
132 			}
133 			z->next_in = zb->inbuf;
134 			z->avail_in = n;
135 		}
136 		ret = inflate(z, Z_SYNC_FLUSH);
137 	} while (ret == Z_OK && z->avail_out > 0);
138 
139 	if (ret != Z_OK) {
140 		if (ret != Z_STREAM_END)
141 			return got_error(GOT_ERR_DECOMPRESSION);
142 		zb->flags |= GOT_ZSTREAM_F_HAVE_MORE;
143 	}
144 
145 	*outlenp = z->total_out - last_total_out;
146 	return NULL;
147 }
148 
149 static const struct got_error *
150 parse_object_header(struct got_object **obj, char *buf, size_t len)
151 {
152 	const char *obj_tags[] = {
153 		GOT_OBJ_TAG_COMMIT,
154 		GOT_OBJ_TAG_TREE,
155 		GOT_OBJ_TAG_BLOB
156 	};
157 	const int obj_types[] = {
158 		GOT_OBJ_TYPE_COMMIT,
159 		GOT_OBJ_TYPE_TREE,
160 		GOT_OBJ_TYPE_BLOB,
161 	};
162 	int type = 0;
163 	size_t size = 0, hdrlen = 0;
164 	int i;
165 	char *p = strchr(buf, '\0');
166 
167 	if (p == NULL)
168 		return got_error(GOT_ERR_BAD_OBJ_HDR);
169 
170 	hdrlen = strlen(buf) + 1 /* '\0' */;
171 
172 	for (i = 0; i < nitems(obj_tags); i++) {
173 		const char *tag = obj_tags[i];
174 		size_t tlen = strlen(tag);
175 		const char *errstr;
176 
177 		if (strncmp(buf, tag, tlen) != 0)
178 			continue;
179 
180 		type = obj_types[i];
181 		if (len <= tlen)
182 			return got_error(GOT_ERR_BAD_OBJ_HDR);
183 		size = strtonum(buf + tlen, 0, LONG_MAX, &errstr);
184 		if (errstr != NULL)
185 			return got_error(GOT_ERR_BAD_OBJ_HDR);
186 		break;
187 	}
188 
189 	if (type == 0)
190 		return got_error(GOT_ERR_BAD_OBJ_HDR);
191 
192 	*obj = calloc(1, sizeof(**obj));
193 	(*obj)->type = type;
194 	(*obj)->hdrlen = hdrlen;
195 	(*obj)->size = size;
196 	return NULL;
197 }
198 
199 static const struct got_error *
200 read_object_header(struct got_object **obj, struct got_repository *repo,
201     const char *path)
202 {
203 	const struct got_error *err;
204 	FILE *f;
205 	struct got_zstream_buf zb;
206 	char *buf;
207 	size_t len;
208 	const size_t zbsize = 64;
209 	size_t outlen, totlen;
210 	int i, ret;
211 
212 	f = fopen(path, "rb");
213 	if (f == NULL)
214 		return got_error(GOT_ERR_BAD_PATH);
215 
216 	buf = calloc(zbsize, sizeof(char));
217 	if (buf == NULL)
218 		return got_error(GOT_ERR_NO_MEM);
219 
220 	err = inflate_init(&zb, zbsize);
221 	if (err) {
222 		fclose(f);
223 		return err;
224 	}
225 
226 	i = 0;
227 	totlen = 0;
228 	do {
229 		err = inflate_read(&zb, f, &outlen);
230 		if (err)
231 			goto done;
232 		if (strchr(zb.outbuf, '\0') == NULL) {
233 			buf = recallocarray(buf, 1 + i, 2 + i, zbsize);
234 			if (buf == NULL) {
235 				err = got_error(GOT_ERR_NO_MEM);
236 				goto done;
237 			}
238 			buf += totlen;
239 		}
240 		memcpy(buf, zb.outbuf, outlen);
241 		totlen += outlen;
242 		i++;
243 	} while (strchr(zb.outbuf, '\0') == NULL);
244 
245 	err = parse_object_header(obj, buf, totlen);
246 done:
247 	inflate_end(&zb);
248 	fclose(f);
249 	return err;
250 }
251 
252 static const struct got_error *
253 object_path(char **path, struct got_object_id *id,
254     struct got_repository *repo)
255 {
256 	const struct got_error *err = NULL;
257 	char hex[SHA1_DIGEST_STRING_LENGTH];
258 	char *path_objects = got_repo_get_path_objects(repo);
259 
260 	if (path_objects == NULL)
261 		return got_error(GOT_ERR_NO_MEM);
262 
263 	got_object_id_str(id, hex, sizeof(hex));
264 
265 	if (asprintf(path, "%s/%.2x/%s", path_objects,
266 	    id->sha1[0], hex + 2) == -1)
267 		err = got_error(GOT_ERR_NO_MEM);
268 
269 	free(path_objects);
270 	return err;
271 }
272 
273 const struct got_error *
274 got_object_open(struct got_object **obj, struct got_repository *repo,
275     struct got_object_id *id)
276 {
277 	const struct got_error *err = NULL;
278 	char *path = NULL;
279 
280 	err = object_path(&path, id, repo);
281 	if (err)
282 		return err;
283 
284 	err = read_object_header(obj, repo, path);
285 	if (err == NULL)
286 		memcpy((*obj)->id.sha1, id->sha1, SHA1_DIGEST_LENGTH);
287 done:
288 	free(path);
289 	return err;
290 }
291 
292 void
293 got_object_close(struct got_object *obj)
294 {
295 	free(obj);
296 }
297 
298 static int
299 commit_object_valid(struct got_commit_object *commit)
300 {
301 	int i;
302 	int n;
303 
304 	if (commit == NULL)
305 		return 0;
306 
307 	n = 0;
308 	for (i = 0; i < SHA1_DIGEST_LENGTH; i++) {
309 		if (commit->tree_id.sha1[i] == 0)
310 			n++;
311 	}
312 	if (n == SHA1_DIGEST_LENGTH)
313 		return 0;
314 
315 	return 1;
316 }
317 
318 static const struct got_error *
319 parse_commit_object(struct got_commit_object **commit, char *buf, size_t len)
320 {
321 	const struct got_error *err = NULL;
322 	char *s = buf;
323 	size_t tlen;
324 	ssize_t remain = (ssize_t)len;
325  
326 	*commit = calloc(1, sizeof(**commit));
327 	if (*commit == NULL)
328 		return got_error(GOT_ERR_NO_MEM);
329 
330 	SIMPLEQ_INIT(&(*commit)->parent_ids);
331 
332 	tlen = strlen(GOT_COMMIT_TAG_TREE);
333 	if (strncmp(s, GOT_COMMIT_TAG_TREE, tlen) == 0) {
334 		remain -= tlen;
335 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
336 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
337 			goto done;
338 		}
339 		s += tlen;
340 		if (!got_parse_sha1_digest((*commit)->tree_id.sha1, s)) {
341 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
342 			goto done;
343 		}
344 		remain -= SHA1_DIGEST_STRING_LENGTH;
345 		s += SHA1_DIGEST_STRING_LENGTH;
346 	} else {
347 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
348 		goto done;
349 	}
350 
351 	tlen = strlen(GOT_COMMIT_TAG_PARENT);
352 	while (strncmp(s, GOT_COMMIT_TAG_PARENT, tlen) == 0) {
353 		struct got_parent_id *pid;
354 
355 		remain -= tlen;
356 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
357 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
358 			goto done;
359 		}	
360 
361 		pid = calloc(1, sizeof(*pid));
362 		if (pid == NULL) {
363 			err = got_error(GOT_ERR_NO_MEM);
364 			goto done;
365 		}
366 		s += tlen;
367 		if (!got_parse_sha1_digest(pid->id.sha1, s)) {
368 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
369 			goto done;
370 		}
371 		SIMPLEQ_INSERT_TAIL(&(*commit)->parent_ids, pid, entry);
372 		(*commit)->nparents++;
373 
374 		s += SHA1_DIGEST_STRING_LENGTH;
375 	}
376 
377 	tlen = strlen(GOT_COMMIT_TAG_AUTHOR);
378 	if (strncmp(s, GOT_COMMIT_TAG_AUTHOR, tlen) == 0) {
379 		char *p;
380 
381 		remain -= tlen;
382 		if (remain <= 0) {
383 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
384 			goto done;
385 		}
386 		s += tlen;
387 		p = strchr(s, '\n');
388 		if (p == NULL) {
389 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
390 			goto done;
391 		}
392 		*p = '\0';
393 		(*commit)->author = strdup(s);
394 		if ((*commit)->author == NULL) {
395 			err = got_error(GOT_ERR_NO_MEM);
396 			goto done;
397 		}
398 		s += strlen((*commit)->author) + 1;
399 	}
400 
401 	tlen = strlen(GOT_COMMIT_TAG_COMMITTER);
402 	if (strncmp(s, GOT_COMMIT_TAG_COMMITTER, tlen) == 0) {
403 		char *p;
404 
405 		remain -= tlen;
406 		if (remain <= 0) {
407 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
408 			goto done;
409 		}
410 		s += tlen;
411 		p = strchr(s, '\n');
412 		if (p == NULL) {
413 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
414 			goto done;
415 		}
416 		*p = '\0';
417 		(*commit)->committer = strdup(s);
418 		if ((*commit)->committer == NULL) {
419 			err = got_error(GOT_ERR_NO_MEM);
420 			goto done;
421 		}
422 		s += strlen((*commit)->committer) + 1;
423 	}
424 
425 	(*commit)->logmsg = strdup(s);
426 done:
427 	if (err)
428 		got_object_commit_close(*commit);
429 	return err;
430 }
431 
432 static void
433 tree_entry_close(struct got_tree_entry *te)
434 {
435 	free(te->name);
436 	free(te);
437 }
438 
439 static const struct got_error *
440 parse_tree_entry(struct got_tree_entry **te, size_t *elen, char *buf,
441     size_t maxlen)
442 {
443 	char *p = buf, *space;
444 	const struct got_error *err = NULL;
445 	char hex[SHA1_DIGEST_STRING_LENGTH];
446 
447 	*te = calloc(1, sizeof(**te));
448 	if (*te == NULL)
449 		return got_error(GOT_ERR_NO_MEM);
450 
451 	*elen = strlen(buf) + 1;
452 	if (*elen > maxlen) {
453 		free(*te);
454 		return got_error(GOT_ERR_BAD_OBJ_DATA);
455 	}
456 
457 	space = strchr(buf, ' ');
458 	if (space == NULL) {
459 		free(*te);
460 		return got_error(GOT_ERR_BAD_OBJ_DATA);
461 	}
462 	while (*p != ' ') {
463 		if (*p < '0' && *p > '7') {
464 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
465 			goto done;
466 		}
467 		(*te)->mode <<= 3;
468 		(*te)->mode |= *p - '0';
469 		p++;
470 	}
471 
472 	(*te)->name = strdup(space + 1);
473 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH) {
474 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
475 		goto done;
476 	}
477 	buf += strlen(buf) + 1;
478 	memcpy((*te)->id.sha1, buf, SHA1_DIGEST_LENGTH);
479 	*elen += SHA1_DIGEST_LENGTH;
480 done:
481 	if (err)
482 		tree_entry_close(*te);
483 	return err;
484 }
485 
486 static const struct got_error *
487 parse_tree_object(struct got_tree_object **tree, struct got_repository *repo,
488     char *buf, size_t len)
489 {
490 	size_t remain = len;
491 	int nentries;
492 
493 	*tree = calloc(1, sizeof(**tree));
494 	if (*tree == NULL)
495 		return got_error(GOT_ERR_NO_MEM);
496 
497 	SIMPLEQ_INIT(&(*tree)->entries);
498 
499 	while (remain > 0) {
500 		struct got_tree_entry *te;
501 		size_t elen;
502 
503 		parse_tree_entry(&te, &elen, buf, remain);
504 		(*tree)->nentries++;
505 		SIMPLEQ_INSERT_TAIL(&(*tree)->entries, te, entry);
506 		buf += elen;
507 		remain -= elen;
508 	}
509 
510 	if (remain != 0) {
511 		got_object_tree_close(*tree);
512 		return got_error(GOT_ERR_BAD_OBJ_DATA);
513 	}
514 
515 	return NULL;
516 }
517 
518 static const struct got_error *
519 read_commit_object(struct got_commit_object **commit,
520     struct got_repository *repo, struct got_object *obj, const char *path)
521 {
522 	const struct got_error *err = NULL;
523 	FILE *f;
524 	struct got_zstream_buf zb;
525 	size_t len;
526 	char *p;
527 	int i, ret;
528 
529 	f = fopen(path, "rb");
530 	if (f == NULL)
531 		return got_error(GOT_ERR_BAD_PATH);
532 
533 	err = inflate_init(&zb, 8192);
534 	if (err) {
535 		fclose(f);
536 		return err;
537 	}
538 
539 	do {
540 		err = inflate_read(&zb, f, &len);
541 		if (err || len == 0)
542 			break;
543 	} while (len < obj->hdrlen + obj->size);
544 
545 	if (len < obj->hdrlen + obj->size) {
546 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
547 		goto done;
548 	}
549 
550 	/* Skip object header. */
551 	len -= obj->hdrlen;
552 	err = parse_commit_object(commit, zb.outbuf + obj->hdrlen, len);
553 done:
554 	inflate_end(&zb);
555 	fclose(f);
556 	return err;
557 }
558 
559 const struct got_error *
560 got_object_commit_open(struct got_commit_object **commit,
561     struct got_repository *repo, struct got_object *obj)
562 {
563 	const struct got_error *err = NULL;
564 	char *path = NULL;
565 
566 	if (obj->type != GOT_OBJ_TYPE_COMMIT)
567 		return got_error(GOT_ERR_OBJ_TYPE);
568 
569 	err = object_path(&path, &obj->id, repo);
570 	if (err)
571 		return err;
572 
573 	err = read_commit_object(commit, repo, obj, path);
574 	free(path);
575 	return err;
576 }
577 
578 void
579 got_object_commit_close(struct got_commit_object *commit)
580 {
581 	struct got_parent_id *pid;
582 
583 	while (!SIMPLEQ_EMPTY(&commit->parent_ids)) {
584 		pid = SIMPLEQ_FIRST(&commit->parent_ids);
585 		SIMPLEQ_REMOVE_HEAD(&commit->parent_ids, entry);
586 		free(pid);
587 	}
588 
589 	free(commit->author);
590 	free(commit->committer);
591 	free(commit->logmsg);
592 	free(commit);
593 }
594 
595 static const struct got_error *
596 read_tree_object(struct got_tree_object **tree,
597     struct got_repository *repo, struct got_object *obj, const char *path)
598 {
599 	const struct got_error *err = NULL;
600 	FILE *f;
601 	struct got_zstream_buf zb;
602 	size_t len;
603 	char *p;
604 	int i, ret;
605 
606 	f = fopen(path, "rb");
607 	if (f == NULL)
608 		return got_error(GOT_ERR_BAD_PATH);
609 
610 	err = inflate_init(&zb, 8192);
611 	if (err) {
612 		fclose(f);
613 		return err;
614 	}
615 
616 	do {
617 		err = inflate_read(&zb, f, &len);
618 		if (err || len == 0)
619 			break;
620 	} while (len < obj->hdrlen + obj->size);
621 
622 	if (len < obj->hdrlen + obj->size) {
623 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
624 		goto done;
625 	}
626 
627 	/* Skip object header. */
628 	len -= obj->hdrlen;
629 	err = parse_tree_object(tree, repo, zb.outbuf + obj->hdrlen, len);
630 done:
631 	inflate_end(&zb);
632 	fclose(f);
633 	return err;
634 }
635 
636 const struct got_error *
637 got_object_tree_open(struct got_tree_object **tree,
638     struct got_repository *repo, struct got_object *obj)
639 {
640 	const struct got_error *err = NULL;
641 	char *path = NULL;
642 
643 	if (obj->type != GOT_OBJ_TYPE_TREE)
644 		return got_error(GOT_ERR_OBJ_TYPE);
645 
646 	err = object_path(&path, &obj->id, repo);
647 	if (err)
648 		return err;
649 
650 	err = read_tree_object(tree, repo, obj, path);
651 	free(path);
652 	return err;
653 }
654 
655 void
656 got_object_tree_close(struct got_tree_object *tree)
657 {
658 	struct got_tree_entry *te;
659 
660 	while (!SIMPLEQ_EMPTY(&tree->entries)) {
661 		te = SIMPLEQ_FIRST(&tree->entries);
662 		SIMPLEQ_REMOVE_HEAD(&tree->entries, entry);
663 		tree_entry_close(te);
664 	}
665 
666 	free(tree);
667 }
668 
669 const struct got_error *
670 got_object_blob_open(struct got_blob_object **blob,
671     struct got_repository *repo, struct got_object *obj, size_t blocksize)
672 {
673 	const struct got_error *err = NULL;
674 	char *path;
675 
676 	if (obj->type != GOT_OBJ_TYPE_BLOB)
677 		return got_error(GOT_ERR_OBJ_TYPE);
678 
679 	if (blocksize < obj->hdrlen)
680 		return got_error(GOT_ERR_NO_SPACE);
681 
682 	err = object_path(&path, &obj->id, repo);
683 	if (err)
684 		return err;
685 
686 	*blob = calloc(1, sizeof(**blob));
687 	if (*blob == NULL) {
688 		free(path);
689 		return got_error(GOT_ERR_NO_MEM);
690 	}
691 
692 	(*blob)->f = fopen(path, "rb");
693 	if ((*blob)->f == NULL) {
694 		free(*blob);
695 		free(path);
696 		return got_error(GOT_ERR_BAD_PATH);
697 	}
698 
699 	err = inflate_init(&(*blob)->zb, blocksize);
700 	if (err != NULL) {
701 		fclose((*blob)->f);
702 		free(*blob);
703 		free(path);
704 		return err;
705 	}
706 
707 	(*blob)->hdrlen = obj->hdrlen;
708 	memcpy(&(*blob)->id.sha1, obj->id.sha1, SHA1_DIGEST_LENGTH);
709 
710 	free(path);
711 	return err;
712 }
713 
714 void
715 got_object_blob_close(struct got_blob_object *blob)
716 {
717 	inflate_end(&blob->zb);
718 	fclose(blob->f);
719 	free(blob);
720 }
721 
722 const struct got_error *
723 got_object_blob_read_block(struct got_blob_object *blob, size_t *outlenp)
724 {
725 	return inflate_read(&blob->zb, blob->f, outlenp);
726 }
727