archive-zip: add tests for big ZIP archives
[git/git.git] / archive-zip.c
CommitLineData
e4fbbfe9
RS
1/*
2 * Copyright (c) 2006 Rene Scharfe
3 */
e4fbbfe9 4#include "cache.h"
ec06bff5 5#include "archive.h"
2158f883 6#include "streaming.h"
88182bab 7#include "utf8.h"
4aff646d
RS
8#include "userdiff.h"
9#include "xdiff-interface.h"
e4fbbfe9 10
e4fbbfe9
RS
11static int zip_date;
12static int zip_time;
13
14static unsigned char *zip_dir;
15static unsigned int zip_dir_size;
16
17static unsigned int zip_offset;
18static unsigned int zip_dir_offset;
88329ca8
RS
19static uint64_t zip_dir_entries;
20
21static unsigned int max_creator_version;
e4fbbfe9
RS
22
23#define ZIP_DIRECTORY_MIN_SIZE (1024 * 1024)
88182bab
RS
24#define ZIP_STREAM (1 << 3)
25#define ZIP_UTF8 (1 << 11)
e4fbbfe9
RS
26
27struct zip_local_header {
28 unsigned char magic[4];
29 unsigned char version[2];
30 unsigned char flags[2];
31 unsigned char compression_method[2];
32 unsigned char mtime[2];
33 unsigned char mdate[2];
34 unsigned char crc32[4];
35 unsigned char compressed_size[4];
36 unsigned char size[4];
37 unsigned char filename_length[2];
38 unsigned char extra_length[2];
0ea865ce 39 unsigned char _end[1];
e4fbbfe9
RS
40};
41
2158f883
RS
42struct zip_data_desc {
43 unsigned char magic[4];
44 unsigned char crc32[4];
45 unsigned char compressed_size[4];
46 unsigned char size[4];
47 unsigned char _end[1];
48};
49
e4fbbfe9
RS
50struct zip_dir_header {
51 unsigned char magic[4];
52 unsigned char creator_version[2];
53 unsigned char version[2];
54 unsigned char flags[2];
55 unsigned char compression_method[2];
56 unsigned char mtime[2];
57 unsigned char mdate[2];
58 unsigned char crc32[4];
59 unsigned char compressed_size[4];
60 unsigned char size[4];
61 unsigned char filename_length[2];
62 unsigned char extra_length[2];
63 unsigned char comment_length[2];
64 unsigned char disk[2];
65 unsigned char attr1[2];
66 unsigned char attr2[4];
67 unsigned char offset[4];
0ea865ce 68 unsigned char _end[1];
e4fbbfe9
RS
69};
70
71struct zip_dir_trailer {
72 unsigned char magic[4];
73 unsigned char disk[2];
74 unsigned char directory_start_disk[2];
75 unsigned char entries_on_this_disk[2];
76 unsigned char entries[2];
77 unsigned char size[4];
78 unsigned char offset[4];
79 unsigned char comment_length[2];
0ea865ce 80 unsigned char _end[1];
e4fbbfe9
RS
81};
82
227bf598
RS
83struct zip_extra_mtime {
84 unsigned char magic[2];
85 unsigned char extra_size[2];
86 unsigned char flags[1];
87 unsigned char mtime[4];
88 unsigned char _end[1];
89};
90
88329ca8
RS
91struct zip64_dir_trailer {
92 unsigned char magic[4];
93 unsigned char record_size[8];
94 unsigned char creator_version[2];
95 unsigned char version[2];
96 unsigned char disk[4];
97 unsigned char directory_start_disk[4];
98 unsigned char entries_on_this_disk[8];
99 unsigned char entries[8];
100 unsigned char size[8];
101 unsigned char offset[8];
102 unsigned char _end[1];
103};
104
105struct zip64_dir_trailer_locator {
106 unsigned char magic[4];
107 unsigned char disk[4];
108 unsigned char offset[8];
109 unsigned char number_of_disks[4];
110 unsigned char _end[1];
111};
112
0ea865ce
RS
113/*
114 * On ARM, padding is added at the end of the struct, so a simple
115 * sizeof(struct ...) reports two bytes more than the payload size
116 * we're interested in.
117 */
118#define ZIP_LOCAL_HEADER_SIZE offsetof(struct zip_local_header, _end)
2158f883 119#define ZIP_DATA_DESC_SIZE offsetof(struct zip_data_desc, _end)
0ea865ce
RS
120#define ZIP_DIR_HEADER_SIZE offsetof(struct zip_dir_header, _end)
121#define ZIP_DIR_TRAILER_SIZE offsetof(struct zip_dir_trailer, _end)
227bf598
RS
122#define ZIP_EXTRA_MTIME_SIZE offsetof(struct zip_extra_mtime, _end)
123#define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
124 (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
88329ca8
RS
125#define ZIP64_DIR_TRAILER_SIZE offsetof(struct zip64_dir_trailer, _end)
126#define ZIP64_DIR_TRAILER_RECORD_SIZE \
127 (ZIP64_DIR_TRAILER_SIZE - \
128 offsetof(struct zip64_dir_trailer, creator_version))
129#define ZIP64_DIR_TRAILER_LOCATOR_SIZE \
130 offsetof(struct zip64_dir_trailer_locator, _end)
0ea865ce 131
e4fbbfe9
RS
132static void copy_le16(unsigned char *dest, unsigned int n)
133{
134 dest[0] = 0xff & n;
135 dest[1] = 0xff & (n >> 010);
136}
137
138static void copy_le32(unsigned char *dest, unsigned int n)
139{
140 dest[0] = 0xff & n;
141 dest[1] = 0xff & (n >> 010);
142 dest[2] = 0xff & (n >> 020);
143 dest[3] = 0xff & (n >> 030);
144}
145
88329ca8
RS
146static void copy_le64(unsigned char *dest, uint64_t n)
147{
148 dest[0] = 0xff & n;
149 dest[1] = 0xff & (n >> 010);
150 dest[2] = 0xff & (n >> 020);
151 dest[3] = 0xff & (n >> 030);
152 dest[4] = 0xff & (n >> 040);
153 dest[5] = 0xff & (n >> 050);
154 dest[6] = 0xff & (n >> 060);
155 dest[7] = 0xff & (n >> 070);
156}
157
158static uint64_t clamp_max(uint64_t n, uint64_t max, int *clamped)
159{
160 if (n <= max)
161 return n;
162 *clamped = 1;
163 return max;
164}
165
166static void copy_le16_clamp(unsigned char *dest, uint64_t n, int *clamped)
167{
168 copy_le16(dest, clamp_max(n, 0xffff, clamped));
169}
170
c3c2e1a0
RS
171static void *zlib_deflate_raw(void *data, unsigned long size,
172 int compression_level,
173 unsigned long *compressed_size)
e4fbbfe9 174{
ef49a7a0 175 git_zstream stream;
e4fbbfe9
RS
176 unsigned long maxsize;
177 void *buffer;
178 int result;
179
c3c2e1a0 180 git_deflate_init_raw(&stream, compression_level);
225a6f10 181 maxsize = git_deflate_bound(&stream, size);
e4fbbfe9
RS
182 buffer = xmalloc(maxsize);
183
184 stream.next_in = data;
185 stream.avail_in = size;
186 stream.next_out = buffer;
187 stream.avail_out = maxsize;
188
189 do {
55bb5c91 190 result = git_deflate(&stream, Z_FINISH);
e4fbbfe9
RS
191 } while (result == Z_OK);
192
193 if (result != Z_STREAM_END) {
194 free(buffer);
195 return NULL;
196 }
197
55bb5c91 198 git_deflate_end(&stream);
e4fbbfe9
RS
199 *compressed_size = stream.total_out;
200
201 return buffer;
202}
203
2158f883
RS
204static void write_zip_data_desc(unsigned long size,
205 unsigned long compressed_size,
206 unsigned long crc)
207{
208 struct zip_data_desc trailer;
209
210 copy_le32(trailer.magic, 0x08074b50);
211 copy_le32(trailer.crc32, crc);
212 copy_le32(trailer.compressed_size, compressed_size);
213 copy_le32(trailer.size, size);
214 write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
215}
216
ebf5374a
RS
217static void set_zip_dir_data_desc(struct zip_dir_header *header,
218 unsigned long size,
219 unsigned long compressed_size,
220 unsigned long crc)
221{
222 copy_le32(header->crc32, crc);
223 copy_le32(header->compressed_size, compressed_size);
224 copy_le32(header->size, size);
225}
226
227static void set_zip_header_data_desc(struct zip_local_header *header,
228 unsigned long size,
229 unsigned long compressed_size,
230 unsigned long crc)
231{
232 copy_le32(header->crc32, crc);
233 copy_le32(header->compressed_size, compressed_size);
234 copy_le32(header->size, size);
235}
236
88182bab
RS
237static int has_only_ascii(const char *s)
238{
239 for (;;) {
240 int c = *s++;
241 if (c == '\0')
242 return 1;
243 if (!isascii(c))
244 return 0;
245 }
246}
247
4aff646d
RS
248static int entry_is_binary(const char *path, const void *buffer, size_t size)
249{
250 struct userdiff_driver *driver = userdiff_find_by_path(path);
251 if (!driver)
252 driver = userdiff_find_by_name("default");
253 if (driver->binary != -1)
254 return driver->binary;
255 return buffer_is_binary(buffer, size);
256}
257
2158f883
RS
258#define STREAM_BUFFER_SIZE (1024 * 16)
259
562e25ab 260static int write_zip_entry(struct archiver_args *args,
9cb513b7
NTND
261 const unsigned char *sha1,
262 const char *path, size_t pathlen,
263 unsigned int mode)
e4fbbfe9
RS
264{
265 struct zip_local_header header;
266 struct zip_dir_header dirent;
227bf598 267 struct zip_extra_mtime extra;
bb52d22e 268 unsigned long attr2;
e4fbbfe9 269 unsigned long compressed_size;
e4fbbfe9
RS
270 unsigned long crc;
271 unsigned long direntsize;
e4fbbfe9 272 int method;
e4fbbfe9 273 unsigned char *out;
e4fbbfe9 274 void *deflated = NULL;
9cb513b7 275 void *buffer;
2158f883
RS
276 struct git_istream *stream = NULL;
277 unsigned long flags = 0;
9cb513b7 278 unsigned long size;
4aff646d
RS
279 int is_binary = -1;
280 const char *path_without_prefix = path + args->baselen;
0f747f9d 281 unsigned int creator_version = 0;
e4fbbfe9 282
38f4d138 283 crc = crc32(0, NULL, 0);
e4fbbfe9 284
88182bab
RS
285 if (!has_only_ascii(path)) {
286 if (is_utf8(path))
287 flags |= ZIP_UTF8;
288 else
289 warning("Path is not valid UTF-8: %s", path);
290 }
291
e4fbbfe9 292 if (pathlen > 0xffff) {
562e25ab
RS
293 return error("path too long (%d chars, SHA1: %s): %s",
294 (int)pathlen, sha1_to_hex(sha1), path);
e4fbbfe9
RS
295 }
296
302b9282 297 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
e4fbbfe9 298 method = 0;
62cdce17 299 attr2 = 16;
e4fbbfe9 300 out = NULL;
60df6bd1 301 size = 0;
e4fbbfe9 302 compressed_size = 0;
9cb513b7 303 buffer = NULL;
62cdce17 304 } else if (S_ISREG(mode) || S_ISLNK(mode)) {
2158f883 305 enum object_type type = sha1_object_info(sha1, &size);
9cb513b7 306
62cdce17 307 method = 0;
bb52d22e
JH
308 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
309 (mode & 0111) ? ((mode) << 16) : 0;
0f747f9d
RS
310 if (S_ISLNK(mode) || (mode & 0111))
311 creator_version = 0x0317;
2158f883 312 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
62cdce17 313 method = 8;
2158f883
RS
314
315 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
c743c215 316 size > big_file_threshold) {
2158f883
RS
317 stream = open_istream(sha1, &type, &size, NULL);
318 if (!stream)
319 return error("cannot stream blob %s",
320 sha1_to_hex(sha1));
321 flags |= ZIP_STREAM;
322 out = buffer = NULL;
323 } else {
324 buffer = sha1_file_to_archive(args, path, sha1, mode,
325 &type, &size);
326 if (!buffer)
327 return error("cannot read %s",
328 sha1_to_hex(sha1));
329 crc = crc32(crc, buffer, size);
4aff646d
RS
330 is_binary = entry_is_binary(path_without_prefix,
331 buffer, size);
2158f883
RS
332 out = buffer;
333 }
d3c1472f 334 compressed_size = (method == 0) ? size : 0;
e4fbbfe9 335 } else {
562e25ab
RS
336 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
337 sha1_to_hex(sha1));
e4fbbfe9
RS
338 }
339
88329ca8
RS
340 if (creator_version > max_creator_version)
341 max_creator_version = creator_version;
342
2158f883 343 if (buffer && method == 8) {
c3c2e1a0
RS
344 out = deflated = zlib_deflate_raw(buffer, size,
345 args->compression_level,
346 &compressed_size);
347 if (!out || compressed_size >= size) {
348 out = buffer;
e4fbbfe9
RS
349 method = 0;
350 compressed_size = size;
351 }
352 }
353
227bf598
RS
354 copy_le16(extra.magic, 0x5455);
355 copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
356 extra.flags[0] = 1; /* just mtime */
357 copy_le32(extra.mtime, args->time);
358
e4fbbfe9 359 /* make sure we have enough free space in the dictionary */
227bf598 360 direntsize = ZIP_DIR_HEADER_SIZE + pathlen + ZIP_EXTRA_MTIME_SIZE;
e4fbbfe9
RS
361 while (zip_dir_size < zip_dir_offset + direntsize) {
362 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
363 zip_dir = xrealloc(zip_dir, zip_dir_size);
364 }
365
366 copy_le32(dirent.magic, 0x02014b50);
0f747f9d 367 copy_le16(dirent.creator_version, creator_version);
cf72fb07 368 copy_le16(dirent.version, 10);
2158f883 369 copy_le16(dirent.flags, flags);
e4fbbfe9
RS
370 copy_le16(dirent.compression_method, method);
371 copy_le16(dirent.mtime, zip_time);
372 copy_le16(dirent.mdate, zip_date);
ebf5374a 373 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
e4fbbfe9 374 copy_le16(dirent.filename_length, pathlen);
227bf598 375 copy_le16(dirent.extra_length, ZIP_EXTRA_MTIME_SIZE);
e4fbbfe9
RS
376 copy_le16(dirent.comment_length, 0);
377 copy_le16(dirent.disk, 0);
62cdce17 378 copy_le32(dirent.attr2, attr2);
e4fbbfe9 379 copy_le32(dirent.offset, zip_offset);
e4fbbfe9
RS
380
381 copy_le32(header.magic, 0x04034b50);
cf72fb07 382 copy_le16(header.version, 10);
2158f883 383 copy_le16(header.flags, flags);
e4fbbfe9
RS
384 copy_le16(header.compression_method, method);
385 copy_le16(header.mtime, zip_time);
386 copy_le16(header.mdate, zip_date);
5ea2c847 387 set_zip_header_data_desc(&header, size, compressed_size, crc);
e4fbbfe9 388 copy_le16(header.filename_length, pathlen);
227bf598 389 copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
0ea865ce
RS
390 write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
391 zip_offset += ZIP_LOCAL_HEADER_SIZE;
e4fbbfe9
RS
392 write_or_die(1, path, pathlen);
393 zip_offset += pathlen;
227bf598
RS
394 write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
395 zip_offset += ZIP_EXTRA_MTIME_SIZE;
2158f883
RS
396 if (stream && method == 0) {
397 unsigned char buf[STREAM_BUFFER_SIZE];
398 ssize_t readlen;
399
400 for (;;) {
401 readlen = read_istream(stream, buf, sizeof(buf));
402 if (readlen <= 0)
403 break;
404 crc = crc32(crc, buf, readlen);
4aff646d
RS
405 if (is_binary == -1)
406 is_binary = entry_is_binary(path_without_prefix,
407 buf, readlen);
2158f883
RS
408 write_or_die(1, buf, readlen);
409 }
410 close_istream(stream);
411 if (readlen)
412 return readlen;
413
414 compressed_size = size;
415 zip_offset += compressed_size;
416
417 write_zip_data_desc(size, compressed_size, crc);
418 zip_offset += ZIP_DATA_DESC_SIZE;
419
c743c215
RS
420 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
421 } else if (stream && method == 8) {
422 unsigned char buf[STREAM_BUFFER_SIZE];
423 ssize_t readlen;
424 git_zstream zstream;
425 int result;
426 size_t out_len;
427 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
428
c3c2e1a0 429 git_deflate_init_raw(&zstream, args->compression_level);
c743c215
RS
430
431 compressed_size = 0;
432 zstream.next_out = compressed;
433 zstream.avail_out = sizeof(compressed);
434
435 for (;;) {
436 readlen = read_istream(stream, buf, sizeof(buf));
437 if (readlen <= 0)
438 break;
439 crc = crc32(crc, buf, readlen);
4aff646d
RS
440 if (is_binary == -1)
441 is_binary = entry_is_binary(path_without_prefix,
442 buf, readlen);
c743c215
RS
443
444 zstream.next_in = buf;
445 zstream.avail_in = readlen;
446 result = git_deflate(&zstream, 0);
447 if (result != Z_OK)
448 die("deflate error (%d)", result);
c3c2e1a0 449 out_len = zstream.next_out - compressed;
c743c215
RS
450
451 if (out_len > 0) {
c3c2e1a0 452 write_or_die(1, compressed, out_len);
c743c215
RS
453 compressed_size += out_len;
454 zstream.next_out = compressed;
455 zstream.avail_out = sizeof(compressed);
456 }
457
458 }
459 close_istream(stream);
460 if (readlen)
461 return readlen;
462
463 zstream.next_in = buf;
464 zstream.avail_in = 0;
465 result = git_deflate(&zstream, Z_FINISH);
466 if (result != Z_STREAM_END)
467 die("deflate error (%d)", result);
468
469 git_deflate_end(&zstream);
c3c2e1a0
RS
470 out_len = zstream.next_out - compressed;
471 write_or_die(1, compressed, out_len);
c743c215
RS
472 compressed_size += out_len;
473 zip_offset += compressed_size;
474
475 write_zip_data_desc(size, compressed_size, crc);
476 zip_offset += ZIP_DATA_DESC_SIZE;
477
2158f883
RS
478 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
479 } else if (compressed_size > 0) {
e4fbbfe9
RS
480 write_or_die(1, out, compressed_size);
481 zip_offset += compressed_size;
482 }
483
e4fbbfe9 484 free(deflated);
9cb513b7 485 free(buffer);
e4fbbfe9 486
4aff646d
RS
487 copy_le16(dirent.attr1, !is_binary);
488
ebf5374a
RS
489 memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
490 zip_dir_offset += ZIP_DIR_HEADER_SIZE;
491 memcpy(zip_dir + zip_dir_offset, path, pathlen);
492 zip_dir_offset += pathlen;
227bf598
RS
493 memcpy(zip_dir + zip_dir_offset, &extra, ZIP_EXTRA_MTIME_SIZE);
494 zip_dir_offset += ZIP_EXTRA_MTIME_SIZE;
ebf5374a
RS
495 zip_dir_entries++;
496
562e25ab 497 return 0;
e4fbbfe9
RS
498}
499
88329ca8
RS
500static void write_zip64_trailer(void)
501{
502 struct zip64_dir_trailer trailer64;
503 struct zip64_dir_trailer_locator locator64;
504
505 copy_le32(trailer64.magic, 0x06064b50);
506 copy_le64(trailer64.record_size, ZIP64_DIR_TRAILER_RECORD_SIZE);
507 copy_le16(trailer64.creator_version, max_creator_version);
508 copy_le16(trailer64.version, 45);
509 copy_le32(trailer64.disk, 0);
510 copy_le32(trailer64.directory_start_disk, 0);
511 copy_le64(trailer64.entries_on_this_disk, zip_dir_entries);
512 copy_le64(trailer64.entries, zip_dir_entries);
513 copy_le64(trailer64.size, zip_dir_offset);
514 copy_le64(trailer64.offset, zip_offset);
515
516 copy_le32(locator64.magic, 0x07064b50);
517 copy_le32(locator64.disk, 0);
518 copy_le64(locator64.offset, zip_offset + zip_dir_offset);
519 copy_le32(locator64.number_of_disks, 1);
520
521 write_or_die(1, &trailer64, ZIP64_DIR_TRAILER_SIZE);
522 write_or_die(1, &locator64, ZIP64_DIR_TRAILER_LOCATOR_SIZE);
523}
524
e4fbbfe9
RS
525static void write_zip_trailer(const unsigned char *sha1)
526{
527 struct zip_dir_trailer trailer;
88329ca8 528 int clamped = 0;
e4fbbfe9
RS
529
530 copy_le32(trailer.magic, 0x06054b50);
531 copy_le16(trailer.disk, 0);
532 copy_le16(trailer.directory_start_disk, 0);
88329ca8
RS
533 copy_le16_clamp(trailer.entries_on_this_disk, zip_dir_entries,
534 &clamped);
535 copy_le16_clamp(trailer.entries, zip_dir_entries, &clamped);
e4fbbfe9
RS
536 copy_le32(trailer.size, zip_dir_offset);
537 copy_le32(trailer.offset, zip_offset);
aeecdcd4 538 copy_le16(trailer.comment_length, sha1 ? GIT_SHA1_HEXSZ : 0);
e4fbbfe9
RS
539
540 write_or_die(1, zip_dir, zip_dir_offset);
88329ca8
RS
541 if (clamped)
542 write_zip64_trailer();
0ea865ce 543 write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
e4fbbfe9 544 if (sha1)
aeecdcd4 545 write_or_die(1, sha1_to_hex(sha1), GIT_SHA1_HEXSZ);
e4fbbfe9
RS
546}
547
548static void dos_time(time_t *time, int *dos_date, int *dos_time)
549{
550 struct tm *t = localtime(time);
551
552 *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
553 (t->tm_year + 1900 - 1980) * 512;
554 *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
555}
556
965cba2e
JK
557static int archive_zip_config(const char *var, const char *value, void *data)
558{
559 return userdiff_config(var, value);
560}
561
4d7c9898
JK
562static int write_zip_archive(const struct archiver *ar,
563 struct archiver_args *args)
ec06bff5 564{
562e25ab
RS
565 int err;
566
965cba2e
JK
567 git_config(archive_zip_config, NULL);
568
ec06bff5
FBH
569 dos_time(&args->time, &zip_date, &zip_time);
570
571 zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
572 zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
562e25ab
RS
573
574 err = write_archive_entries(args, write_zip_entry);
575 if (!err)
576 write_zip_trailer(args->commit_sha1);
ec06bff5
FBH
577
578 free(zip_dir);
579
562e25ab 580 return err;
ec06bff5 581}
13e0f88d
JK
582
583static struct archiver zip_archiver = {
584 "zip",
585 write_zip_archive,
7b97730b 586 ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
13e0f88d
JK
587};
588
589void init_zip_archiver(void)
590{
591 register_archiver(&zip_archiver);
592}