op public repos

Blob

Date:: Fri Jan 12 19:59:51 2018 UTC
Message:: Extract non-deltified objects from pack files.
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018 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/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 
21 #include <dirent.h>
22 #include <errno.h>
23 #include <stdio.h>
24 #include <stdint.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <limits.h>
28 #include <sha1.h>
29 #include <endian.h>
30 #include <zlib.h>
31 
32 #include "got_error.h"
33 #include "got_object.h"
34 #include "got_repository.h"
35 #include "got_sha1.h"
36 #include "pack.h"
37 #include "path.h"
38 
39 #define GOT_PACK_PREFIX		"pack-"
40 #define GOT_PACKFILE_SUFFIX	".pack"
41 #define GOT_PACKIDX_SUFFIX		".idx"
42 #define GOT_PACKFILE_NAMELEN	(strlen(GOT_PACK_PREFIX) + \
43 				SHA1_DIGEST_STRING_LENGTH - 1 + \
44 				strlen(GOT_PACKFILE_SUFFIX))
45 #define GOT_PACKIDX_NAMELEN	(strlen(GOT_PACK_PREFIX) + \
46 				SHA1_DIGEST_STRING_LENGTH - 1 + \
47 				strlen(GOT_PACKIDX_SUFFIX))
48 
49 #ifndef MIN
50 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
51 #endif
52 
53 static const struct got_error *
54 verify_fanout_table(uint32_t *fanout_table)
55 {
56 	int i;
57 
58 	for (i = 0; i < 0xff - 1; i++) {
59 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
60 			return got_error(GOT_ERR_BAD_PACKIDX);
61 	}
62 
63 	return NULL;
64 }
65 
66 static const struct got_error *
67 get_packfile_size(size_t *size, const char *path_idx)
68 {
69 	struct stat sb;
70 	char *path_pack;
71 	char base_path[PATH_MAX];
72 	char *dot;
73 
74 	if (strlcpy(base_path, path_idx, PATH_MAX) > PATH_MAX)
75 		return got_error(GOT_ERR_NO_SPACE);
76 
77 	dot = strrchr(base_path, '.');
78 	if (dot == NULL)
79 		return got_error(GOT_ERR_BAD_PATH);
80 	*dot = '\0';
81 	if (asprintf(&path_pack, "%s.pack", base_path) == -1)
82 		return got_error(GOT_ERR_NO_MEM);
83 
84 	if (stat(path_pack, &sb) != 0) {
85 		free(path_pack);
86 		return got_error(GOT_ERR_IO);
87 
88 	}
89 
90 	free(path_pack);
91 	*size = sb.st_size;
92 	return 0;
93 }
94 
95 const struct got_error *
96 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
97 {
98 	struct got_packidx_v2_hdr *p;
99 	FILE *f;
100 	const struct got_error *err = NULL;
101 	size_t n, nobj, packfile_size;
102 	SHA1_CTX ctx;
103 	uint8_t sha1[SHA1_DIGEST_LENGTH];
104 
105 	SHA1Init(&ctx);
106 
107 	f = fopen(path, "rb");
108 	if (f == NULL)
109 		return got_error(GOT_ERR_BAD_PATH);
110 
111 	err = get_packfile_size(&packfile_size, path);
112 	if (err)
113 		return err;
114 
115 	p = calloc(1, sizeof(*p));
116 	if (p == NULL) {
117 		err = got_error(GOT_ERR_NO_MEM);
118 		goto done;
119 	}
120 
121 	n = fread(&p->magic, sizeof(p->magic), 1, f);
122 	if (n != 1) {
123 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
124 		goto done;
125 	}
126 
127 	if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
128 		err = got_error(GOT_ERR_BAD_PACKIDX);
129 		goto done;
130 	}
131 
132 	SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
133 
134 	n = fread(&p->version, sizeof(p->version), 1, f);
135 	if (n != 1) {
136 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
137 		goto done;
138 	}
139 
140 	if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
141 		err = got_error(GOT_ERR_BAD_PACKIDX);
142 		goto done;
143 	}
144 
145 	SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
146 
147 	n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
148 	if (n != 1) {
149 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
150 		goto done;
151 	}
152 
153 	err = verify_fanout_table(p->fanout_table);
154 	if (err)
155 		goto done;
156 
157 	SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
158 
159 	nobj = betoh32(p->fanout_table[0xff]);
160 
161 	p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
162 	if (p->sorted_ids == NULL) {
163 		err = got_error(GOT_ERR_NO_MEM);
164 		goto done;
165 	}
166 
167 	n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
168 	if (n != nobj) {
169 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
170 		goto done;
171 	}
172 
173 	SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
174 	    nobj * sizeof(*p->sorted_ids));
175 
176 	p->crc32 = calloc(nobj, sizeof(*p->crc32));
177 	if (p->crc32 == NULL) {
178 		err = got_error(GOT_ERR_NO_MEM);
179 		goto done;
180 	}
181 
182 	n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
183 	if (n != nobj) {
184 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
185 		goto done;
186 	}
187 
188 	SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
189 
190 	p->offsets = calloc(nobj, sizeof(*p->offsets));
191 	if (p->offsets == NULL) {
192 		err = got_error(GOT_ERR_NO_MEM);
193 		goto done;
194 	}
195 
196 	n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
197 	if (n != nobj) {
198 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
199 		goto done;
200 	}
201 
202 	SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
203 
204 	/* Large file offsets are contained only in files > 2GB. */
205 	if (packfile_size <= 0x80000000)
206 		goto checksum;
207 
208 	p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
209 	if (p->large_offsets == NULL) {
210 		err = got_error(GOT_ERR_NO_MEM);
211 		goto done;
212 	}
213 
214 	n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
215 	if (n != nobj) {
216 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
217 		goto done;
218 	}
219 
220 	SHA1Update(&ctx, (uint8_t*)p->large_offsets,
221 	    nobj * sizeof(*p->large_offsets));
222 
223 checksum:
224 	n = fread(&p->trailer, sizeof(p->trailer), 1, f);
225 	if (n != 1) {
226 		err = got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKIDX);
227 		goto done;
228 	}
229 
230 	SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
231 	SHA1Final(sha1, &ctx);
232 	if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
233 		err = got_error(GOT_ERR_PACKIDX_CSUM);
234 done:
235 	fclose(f);
236 	if (err)
237 		got_packidx_close(p);
238 	else
239 		*packidx = p;
240 	return err;
241 }
242 
243 void
244 got_packidx_close(struct got_packidx_v2_hdr *packidx)
245 {
246 	free(packidx->sorted_ids);
247 	free(packidx->offsets);
248 	free(packidx->crc32);
249 	free(packidx->large_offsets);
250 	free(packidx);
251 }
252 
253 static int
254 is_packidx_filename(const char *name, size_t len)
255 {
256 	if (len != GOT_PACKIDX_NAMELEN)
257 		return 0;
258 
259 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
260 		return 0;
261 
262 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
263 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
264 		return 0;
265 
266 	return 1;
267 }
268 
269 static off_t
270 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
271 {
272 	uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
273 	uint32_t offset = betoh32(packidx->offsets[idx]);
274 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
275 		uint64_t loffset;
276 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
277 		if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
278 			return -1;
279 		loffset = betoh64(packidx->large_offsets[idx]);
280 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
281 	}
282 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
283 }
284 
285 static int
286 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id)
287 {
288 	u_int8_t id0 = id->sha1[0];
289 	uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
290 	int i = 0;
291 
292 	if (id0 > 0)
293 		i = betoh32(packidx->fanout_table[id0 - 1]);
294 
295 	while (i < totobj) {
296 		struct got_object_id *oid = &packidx->sorted_ids[i++];
297 		uint32_t offset;
298 
299 		if (got_object_id_cmp(id, oid) < 0)
300 			continue;
301 		if (got_object_id_cmp(id, oid) > 0)
302 			break;
303 
304 		return i;
305 	}
306 
307 	return -1;
308 }
309 
310 const struct got_error *
311 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
312 {
313 	const struct got_error *err = NULL;
314 	uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
315 	struct got_packfile_hdr hdr;
316 	size_t n;
317 
318 	n = fread(&hdr, sizeof(hdr), 1, f);
319 	if (n != 1)
320 		return got_error(ferror(f) ? GOT_ERR_IO : GOT_ERR_BAD_PACKFILE);
321 
322 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
323 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
324 	    betoh32(hdr.nobjects) != totobj)
325 		err = got_error(GOT_ERR_BAD_PACKFILE);
326 
327 	return err;
328 }
329 
330 static const struct got_error *
331 dump_packed_object(FILE **f, FILE *packfile, off_t offset)
332 {
333 	const struct got_error *err = NULL;
334 	const char *template = "/tmp/got.XXXXXXXXXX";
335 	uint64_t size = 0;
336 	uint8_t type = 0;
337 	uint8_t sizeN;
338 	int i;
339 	size_t n;
340 	const char *type_tag;
341 
342 	*f = got_opentemp();
343 	if (*f == NULL) {
344 		err = got_error(GOT_ERR_FILE_OPEN);
345 		goto done;
346 	}
347 
348 	if (fseeko(packfile, offset, SEEK_SET) != 0) {
349 		err = got_error(errno == EIO ? GOT_ERR_IO : GOT_ERR_BAD_PATH);
350 		goto done;
351 	}
352 
353 	i = 0;
354 	do {
355 		/* We do not support size values which don't fit in 64 bit. */
356 		if (i > 9) {
357 			err = got_error(GOT_ERR_NO_SPACE);
358 			goto done;
359 		}
360 
361 		n = fread(&sizeN, sizeof(sizeN), 1, packfile);
362 		if (n != 1) {
363 			err = got_error(ferror(packfile) ?
364 			    GOT_ERR_IO : GOT_ERR_BAD_PACKFILE);
365 			goto done;
366 		}
367 		if (i == 0) {
368 			type = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
369 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
370 			size = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
371 		} else {
372 			size_t shift = 4 + 7 * (i - 1);
373 			size |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
374 		}
375 		i++;
376 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
377 
378 	if (type == GOT_OBJ_TYPE_OFFSET_DELTA)
379 		printf("object type OFFSET_DELTA not yet implemented\n");
380 	else if (type == GOT_OBJ_TYPE_REF_DELTA)
381 		printf("object type REF_DELTA not yet implemented\n");
382 	else if (type == GOT_OBJ_TYPE_TAG)
383 		printf("object type TAG not yet implemented\n");
384 
385 	type_tag = got_object_get_type_tag(type);
386 	if (type_tag == NULL) {
387 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
388 		goto done;
389 	}
390 
391 	fprintf(*f, "%s %llu", type_tag, size);
392 	fputc('\0', *f);
393 
394 	while (size > 0) {
395 		uint8_t data[2048];
396 		size_t len = MIN(size, sizeof(data));
397 
398 		n = fread(data, len, 1, packfile);
399 		if (n != 1) {
400 			err = got_error(ferror(packfile) ?
401 			    GOT_ERR_IO : GOT_ERR_BAD_PACKFILE);
402 			goto done;
403 		}
404 
405 		n = fwrite(data, len, 1, *f);
406 		if (n != 1) {
407 			err = got_error(ferror(*f) ?
408 			    GOT_ERR_IO : GOT_ERR_BAD_PACKFILE);
409 			goto done;
410 		}
411 
412 		size -= len;
413 	}
414 
415 	printf("object type is %d\n", type);
416 	rewind(*f);
417 done:
418 	if (err && *f)
419 		fclose(*f);
420 	return err;
421 }
422 static const struct got_error *
423 extract_object(FILE **f, const char *path_packdir,
424     struct got_packidx_v2_hdr *packidx, struct got_object_id *id)
425 {
426 	const struct got_error *err = NULL;
427 	int idx = get_object_idx(packidx, id);
428 	off_t offset;
429 	char *path_packfile;
430 	FILE *packfile;
431 	char hex[SHA1_DIGEST_STRING_LENGTH];
432 	char *sha1str;
433 
434 	*f = NULL;
435 	if (idx == -1) /* object not found in pack index */
436 		return NULL;
437 
438 	offset = get_object_offset(packidx, idx);
439 	if (offset == (uint64_t)-1)
440 		return got_error(GOT_ERR_BAD_PACKIDX);
441 
442 	sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
443 	    hex, sizeof(hex));
444 	if (sha1str == NULL)
445 		return got_error(GOT_ERR_PACKIDX_CSUM);
446 
447 	if (asprintf(&path_packfile, "%s/%s%s%s", path_packdir,
448 	    GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1)
449 		return got_error(GOT_ERR_NO_MEM);
450 
451 	packfile = fopen(path_packfile, "rb");
452 	if (packfile == NULL) {
453 		err = got_error(errno == EIO ? GOT_ERR_IO : GOT_ERR_BAD_PATH);
454 		goto done;
455 	}
456 
457 	err = read_packfile_hdr(packfile, packidx);
458 	if (err)
459 		goto done;
460 
461 	printf("Dumping object at offset %llu\n", offset);
462 	err = dump_packed_object(f, packfile, offset);
463 	if (err)
464 		goto done;
465 
466 done:
467 	free(path_packfile);
468 	if (packfile && fclose(packfile) == -1 && errno == EIO && err == 0)
469 		err = got_error(GOT_ERR_IO);
470 	return err;
471 }
472 
473 const struct got_error *
474 got_packfile_extract_object(FILE **f, struct got_object_id *id,
475     struct got_repository *repo)
476 {
477 	const struct got_error *err = NULL;
478 	DIR *packdir = NULL;
479 	struct dirent *dent;
480 	char *path_packdir = got_repo_get_path_objects_pack(repo);
481 
482 	if (path_packdir == NULL) {
483 		err = got_error(GOT_ERR_NO_MEM);
484 		goto done;
485 	}
486 
487 	packdir = opendir(path_packdir);
488 	if (packdir == NULL) {
489 		err = got_error(errno == EIO ? GOT_ERR_IO : GOT_ERR_BAD_PATH);
490 		goto done;
491 	}
492 
493 	while ((dent = readdir(packdir)) != NULL) {
494 		struct got_packidx_v2_hdr *packidx;
495 		char *path_packidx, *path_object;
496 
497 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
498 			continue;
499 
500 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
501 		    dent->d_name) == -1) {
502 			err = got_error(GOT_ERR_NO_MEM);
503 			goto done;
504 		}
505 
506 		err = got_packidx_open(&packidx, path_packidx);
507 		free(path_packidx);
508 		if (err)
509 			goto done;
510 
511 		err = extract_object(f, path_packdir, packidx, id);
512 		if (err)
513 			goto done;
514 		if (*f != NULL)
515 			break;
516 	}
517 
518 done:
519 	free(path_packdir);
520 	if (packdir && closedir(packdir) != 0 && errno == EIO && err == 0)
521 		err = got_error(GOT_ERR_IO);
522 	return err;
523 }