Gotweb

Blob

Date:: Fri Dec 1 23:51:11 2017 UTC
Message:: fix previous
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 		}
239 		memcpy(buf + totlen, zb.outbuf, outlen);
240 		totlen += outlen;
241 		i++;
242 	} while (strchr(zb.outbuf, '\0') == NULL);
243 
244 	err = parse_object_header(obj, buf, totlen);
245 done:
246 	inflate_end(&zb);
247 	fclose(f);
248 	return err;
249 }
250 
251 static const struct got_error *
252 object_path(char **path, struct got_object_id *id,
253     struct got_repository *repo)
254 {
255 	const struct got_error *err = NULL;
256 	char hex[SHA1_DIGEST_STRING_LENGTH];
257 	char *path_objects = got_repo_get_path_objects(repo);
258 
259 	if (path_objects == NULL)
260 		return got_error(GOT_ERR_NO_MEM);
261 
262 	got_object_id_str(id, hex, sizeof(hex));
263 
264 	if (asprintf(path, "%s/%.2x/%s", path_objects,
265 	    id->sha1[0], hex + 2) == -1)
266 		err = got_error(GOT_ERR_NO_MEM);
267 
268 	free(path_objects);
269 	return err;
270 }
271 
272 const struct got_error *
273 got_object_open(struct got_object **obj, struct got_repository *repo,
274     struct got_object_id *id)
275 {
276 	const struct got_error *err = NULL;
277 	char *path = NULL;
278 
279 	err = object_path(&path, id, repo);
280 	if (err)
281 		return err;
282 
283 	err = read_object_header(obj, repo, path);
284 	if (err == NULL)
285 		memcpy((*obj)->id.sha1, id->sha1, SHA1_DIGEST_LENGTH);
286 done:
287 	free(path);
288 	return err;
289 }
290 
291 void
292 got_object_close(struct got_object *obj)
293 {
294 	free(obj);
295 }
296 
297 static int
298 commit_object_valid(struct got_commit_object *commit)
299 {
300 	int i;
301 	int n;
302 
303 	if (commit == NULL)
304 		return 0;
305 
306 	n = 0;
307 	for (i = 0; i < SHA1_DIGEST_LENGTH; i++) {
308 		if (commit->tree_id.sha1[i] == 0)
309 			n++;
310 	}
311 	if (n == SHA1_DIGEST_LENGTH)
312 		return 0;
313 
314 	return 1;
315 }
316 
317 static const struct got_error *
318 parse_commit_object(struct got_commit_object **commit, char *buf, size_t len)
319 {
320 	const struct got_error *err = NULL;
321 	char *s = buf;
322 	size_t tlen;
323 	ssize_t remain = (ssize_t)len;
324  
325 	*commit = calloc(1, sizeof(**commit));
326 	if (*commit == NULL)
327 		return got_error(GOT_ERR_NO_MEM);
328 
329 	SIMPLEQ_INIT(&(*commit)->parent_ids);
330 
331 	tlen = strlen(GOT_COMMIT_TAG_TREE);
332 	if (strncmp(s, GOT_COMMIT_TAG_TREE, tlen) == 0) {
333 		remain -= tlen;
334 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
335 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
336 			goto done;
337 		}
338 		s += tlen;
339 		if (!got_parse_sha1_digest((*commit)->tree_id.sha1, s)) {
340 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
341 			goto done;
342 		}
343 		remain -= SHA1_DIGEST_STRING_LENGTH;
344 		s += SHA1_DIGEST_STRING_LENGTH;
345 	} else {
346 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
347 		goto done;
348 	}
349 
350 	tlen = strlen(GOT_COMMIT_TAG_PARENT);
351 	while (strncmp(s, GOT_COMMIT_TAG_PARENT, tlen) == 0) {
352 		struct got_parent_id *pid;
353 
354 		remain -= tlen;
355 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
356 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
357 			goto done;
358 		}	
359 
360 		pid = calloc(1, sizeof(*pid));
361 		if (pid == NULL) {
362 			err = got_error(GOT_ERR_NO_MEM);
363 			goto done;
364 		}
365 		s += tlen;
366 		if (!got_parse_sha1_digest(pid->id.sha1, s)) {
367 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
368 			goto done;
369 		}
370 		SIMPLEQ_INSERT_TAIL(&(*commit)->parent_ids, pid, entry);
371 		(*commit)->nparents++;
372 
373 		s += SHA1_DIGEST_STRING_LENGTH;
374 	}
375 
376 	tlen = strlen(GOT_COMMIT_TAG_AUTHOR);
377 	if (strncmp(s, GOT_COMMIT_TAG_AUTHOR, tlen) == 0) {
378 		char *p;
379 
380 		remain -= tlen;
381 		if (remain <= 0) {
382 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
383 			goto done;
384 		}
385 		s += tlen;
386 		p = strchr(s, '\n');
387 		if (p == NULL) {
388 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
389 			goto done;
390 		}
391 		*p = '\0';
392 		(*commit)->author = strdup(s);
393 		if ((*commit)->author == NULL) {
394 			err = got_error(GOT_ERR_NO_MEM);
395 			goto done;
396 		}
397 		s += strlen((*commit)->author) + 1;
398 	}
399 
400 	tlen = strlen(GOT_COMMIT_TAG_COMMITTER);
401 	if (strncmp(s, GOT_COMMIT_TAG_COMMITTER, tlen) == 0) {
402 		char *p;
403 
404 		remain -= tlen;
405 		if (remain <= 0) {
406 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
407 			goto done;
408 		}
409 		s += tlen;
410 		p = strchr(s, '\n');
411 		if (p == NULL) {
412 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
413 			goto done;
414 		}
415 		*p = '\0';
416 		(*commit)->committer = strdup(s);
417 		if ((*commit)->committer == NULL) {
418 			err = got_error(GOT_ERR_NO_MEM);
419 			goto done;
420 		}
421 		s += strlen((*commit)->committer) + 1;
422 	}
423 
424 	(*commit)->logmsg = strdup(s);
425 done:
426 	if (err)
427 		got_object_commit_close(*commit);
428 	return err;
429 }
430 
431 static void
432 tree_entry_close(struct got_tree_entry *te)
433 {
434 	free(te->name);
435 	free(te);
436 }
437 
438 static const struct got_error *
439 parse_tree_entry(struct got_tree_entry **te, size_t *elen, char *buf,
440     size_t maxlen)
441 {
442 	char *p = buf, *space;
443 	const struct got_error *err = NULL;
444 	char hex[SHA1_DIGEST_STRING_LENGTH];
445 
446 	*te = calloc(1, sizeof(**te));
447 	if (*te == NULL)
448 		return got_error(GOT_ERR_NO_MEM);
449 
450 	*elen = strlen(buf) + 1;
451 	if (*elen > maxlen) {
452 		free(*te);
453 		return got_error(GOT_ERR_BAD_OBJ_DATA);
454 	}
455 
456 	space = strchr(buf, ' ');
457 	if (space == NULL) {
458 		free(*te);
459 		return got_error(GOT_ERR_BAD_OBJ_DATA);
460 	}
461 	while (*p != ' ') {
462 		if (*p < '0' && *p > '7') {
463 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
464 			goto done;
465 		}
466 		(*te)->mode <<= 3;
467 		(*te)->mode |= *p - '0';
468 		p++;
469 	}
470 
471 	(*te)->name = strdup(space + 1);
472 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH) {
473 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
474 		goto done;
475 	}
476 	buf += strlen(buf) + 1;
477 	memcpy((*te)->id.sha1, buf, SHA1_DIGEST_LENGTH);
478 	*elen += SHA1_DIGEST_LENGTH;
479 done:
480 	if (err)
481 		tree_entry_close(*te);
482 	return err;
483 }
484 
485 static const struct got_error *
486 parse_tree_object(struct got_tree_object **tree, struct got_repository *repo,
487     char *buf, size_t len)
488 {
489 	size_t remain = len;
490 	int nentries;
491 
492 	*tree = calloc(1, sizeof(**tree));
493 	if (*tree == NULL)
494 		return got_error(GOT_ERR_NO_MEM);
495 
496 	SIMPLEQ_INIT(&(*tree)->entries);
497 
498 	while (remain > 0) {
499 		struct got_tree_entry *te;
500 		size_t elen;
501 
502 		parse_tree_entry(&te, &elen, buf, remain);
503 		(*tree)->nentries++;
504 		SIMPLEQ_INSERT_TAIL(&(*tree)->entries, te, entry);
505 		buf += elen;
506 		remain -= elen;
507 	}
508 
509 	if (remain != 0) {
510 		got_object_tree_close(*tree);
511 		return got_error(GOT_ERR_BAD_OBJ_DATA);
512 	}
513 
514 	return NULL;
515 }
516 
517 static const struct got_error *
518 read_commit_object(struct got_commit_object **commit,
519     struct got_repository *repo, struct got_object *obj, const char *path)
520 {
521 	const struct got_error *err = NULL;
522 	FILE *f;
523 	struct got_zstream_buf zb;
524 	size_t len;
525 	char *p;
526 	int i, ret;
527 
528 	f = fopen(path, "rb");
529 	if (f == NULL)
530 		return got_error(GOT_ERR_BAD_PATH);
531 
532 	err = inflate_init(&zb, 8192);
533 	if (err) {
534 		fclose(f);
535 		return err;
536 	}
537 
538 	do {
539 		err = inflate_read(&zb, f, &len);
540 		if (err || len == 0)
541 			break;
542 	} while (len < obj->hdrlen + obj->size);
543 
544 	if (len < obj->hdrlen + obj->size) {
545 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
546 		goto done;
547 	}
548 
549 	/* Skip object header. */
550 	len -= obj->hdrlen;
551 	err = parse_commit_object(commit, zb.outbuf + obj->hdrlen, len);
552 done:
553 	inflate_end(&zb);
554 	fclose(f);
555 	return err;
556 }
557 
558 const struct got_error *
559 got_object_commit_open(struct got_commit_object **commit,
560     struct got_repository *repo, struct got_object *obj)
561 {
562 	const struct got_error *err = NULL;
563 	char *path = NULL;
564 
565 	if (obj->type != GOT_OBJ_TYPE_COMMIT)
566 		return got_error(GOT_ERR_OBJ_TYPE);
567 
568 	err = object_path(&path, &obj->id, repo);
569 	if (err)
570 		return err;
571 
572 	err = read_commit_object(commit, repo, obj, path);
573 	free(path);
574 	return err;
575 }
576 
577 void
578 got_object_commit_close(struct got_commit_object *commit)
579 {
580 	struct got_parent_id *pid;
581 
582 	while (!SIMPLEQ_EMPTY(&commit->parent_ids)) {
583 		pid = SIMPLEQ_FIRST(&commit->parent_ids);
584 		SIMPLEQ_REMOVE_HEAD(&commit->parent_ids, entry);
585 		free(pid);
586 	}
587 
588 	free(commit->author);
589 	free(commit->committer);
590 	free(commit->logmsg);
591 	free(commit);
592 }
593 
594 static const struct got_error *
595 read_tree_object(struct got_tree_object **tree,
596     struct got_repository *repo, struct got_object *obj, const char *path)
597 {
598 	const struct got_error *err = NULL;
599 	FILE *f;
600 	struct got_zstream_buf zb;
601 	size_t len;
602 	char *p;
603 	int i, ret;
604 
605 	f = fopen(path, "rb");
606 	if (f == NULL)
607 		return got_error(GOT_ERR_BAD_PATH);
608 
609 	err = inflate_init(&zb, 8192);
610 	if (err) {
611 		fclose(f);
612 		return err;
613 	}
614 
615 	do {
616 		err = inflate_read(&zb, f, &len);
617 		if (err || len == 0)
618 			break;
619 	} while (len < obj->hdrlen + obj->size);
620 
621 	if (len < obj->hdrlen + obj->size) {
622 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
623 		goto done;
624 	}
625 
626 	/* Skip object header. */
627 	len -= obj->hdrlen;
628 	err = parse_tree_object(tree, repo, zb.outbuf + obj->hdrlen, len);
629 done:
630 	inflate_end(&zb);
631 	fclose(f);
632 	return err;
633 }
634 
635 const struct got_error *
636 got_object_tree_open(struct got_tree_object **tree,
637     struct got_repository *repo, struct got_object *obj)
638 {
639 	const struct got_error *err = NULL;
640 	char *path = NULL;
641 
642 	if (obj->type != GOT_OBJ_TYPE_TREE)
643 		return got_error(GOT_ERR_OBJ_TYPE);
644 
645 	err = object_path(&path, &obj->id, repo);
646 	if (err)
647 		return err;
648 
649 	err = read_tree_object(tree, repo, obj, path);
650 	free(path);
651 	return err;
652 }
653 
654 void
655 got_object_tree_close(struct got_tree_object *tree)
656 {
657 	struct got_tree_entry *te;
658 
659 	while (!SIMPLEQ_EMPTY(&tree->entries)) {
660 		te = SIMPLEQ_FIRST(&tree->entries);
661 		SIMPLEQ_REMOVE_HEAD(&tree->entries, entry);
662 		tree_entry_close(te);
663 	}
664 
665 	free(tree);
666 }
667 
668 const struct got_error *
669 got_object_blob_open(struct got_blob_object **blob,
670     struct got_repository *repo, struct got_object *obj, size_t blocksize)
671 {
672 	const struct got_error *err = NULL;
673 	char *path;
674 
675 	if (obj->type != GOT_OBJ_TYPE_BLOB)
676 		return got_error(GOT_ERR_OBJ_TYPE);
677 
678 	if (blocksize < obj->hdrlen)
679 		return got_error(GOT_ERR_NO_SPACE);
680 
681 	err = object_path(&path, &obj->id, repo);
682 	if (err)
683 		return err;
684 
685 	*blob = calloc(1, sizeof(**blob));
686 	if (*blob == NULL) {
687 		free(path);
688 		return got_error(GOT_ERR_NO_MEM);
689 	}
690 
691 	(*blob)->f = fopen(path, "rb");
692 	if ((*blob)->f == NULL) {
693 		free(*blob);
694 		free(path);
695 		return got_error(GOT_ERR_BAD_PATH);
696 	}
697 
698 	err = inflate_init(&(*blob)->zb, blocksize);
699 	if (err != NULL) {
700 		fclose((*blob)->f);
701 		free(*blob);
702 		free(path);
703 		return err;
704 	}
705 
706 	(*blob)->hdrlen = obj->hdrlen;
707 	memcpy(&(*blob)->id.sha1, obj->id.sha1, SHA1_DIGEST_LENGTH);
708 
709 	free(path);
710 	return err;
711 }
712 
713 void
714 got_object_blob_close(struct got_blob_object *blob)
715 {
716 	inflate_end(&blob->zb);
717 	fclose(blob->f);
718 	free(blob);
719 }
720 
721 const struct got_error *
722 got_object_blob_read_block(struct got_blob_object *blob, size_t *outlenp)
723 {
724 	return inflate_read(&blob->zb, blob->f, outlenp);
725 }
726