archive-zip: support archives bigger than 4GB
[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
c061a149
RS
14/* We only care about the "buf" part here. */
15static struct strbuf zip_dir;
e4fbbfe9 16
af95749f 17static uintmax_t zip_offset;
88329ca8
RS
18static uint64_t zip_dir_entries;
19
20static unsigned int max_creator_version;
e4fbbfe9 21
88182bab
RS
22#define ZIP_STREAM (1 << 3)
23#define ZIP_UTF8 (1 << 11)
e4fbbfe9
RS
24
25struct zip_local_header {
26 unsigned char magic[4];
27 unsigned char version[2];
28 unsigned char flags[2];
29 unsigned char compression_method[2];
30 unsigned char mtime[2];
31 unsigned char mdate[2];
32 unsigned char crc32[4];
33 unsigned char compressed_size[4];
34 unsigned char size[4];
35 unsigned char filename_length[2];
36 unsigned char extra_length[2];
0ea865ce 37 unsigned char _end[1];
e4fbbfe9
RS
38};
39
2158f883
RS
40struct zip_data_desc {
41 unsigned char magic[4];
42 unsigned char crc32[4];
43 unsigned char compressed_size[4];
44 unsigned char size[4];
45 unsigned char _end[1];
46};
47
e4fbbfe9
RS
48struct zip_dir_trailer {
49 unsigned char magic[4];
50 unsigned char disk[2];
51 unsigned char directory_start_disk[2];
52 unsigned char entries_on_this_disk[2];
53 unsigned char entries[2];
54 unsigned char size[4];
55 unsigned char offset[4];
56 unsigned char comment_length[2];
0ea865ce 57 unsigned char _end[1];
e4fbbfe9
RS
58};
59
227bf598
RS
60struct zip_extra_mtime {
61 unsigned char magic[2];
62 unsigned char extra_size[2];
63 unsigned char flags[1];
64 unsigned char mtime[4];
65 unsigned char _end[1];
66};
67
88329ca8
RS
68struct zip64_dir_trailer {
69 unsigned char magic[4];
70 unsigned char record_size[8];
71 unsigned char creator_version[2];
72 unsigned char version[2];
73 unsigned char disk[4];
74 unsigned char directory_start_disk[4];
75 unsigned char entries_on_this_disk[8];
76 unsigned char entries[8];
77 unsigned char size[8];
78 unsigned char offset[8];
79 unsigned char _end[1];
80};
81
82struct zip64_dir_trailer_locator {
83 unsigned char magic[4];
84 unsigned char disk[4];
85 unsigned char offset[8];
86 unsigned char number_of_disks[4];
87 unsigned char _end[1];
88};
89
0ea865ce
RS
90/*
91 * On ARM, padding is added at the end of the struct, so a simple
92 * sizeof(struct ...) reports two bytes more than the payload size
93 * we're interested in.
94 */
95#define ZIP_LOCAL_HEADER_SIZE offsetof(struct zip_local_header, _end)
2158f883 96#define ZIP_DATA_DESC_SIZE offsetof(struct zip_data_desc, _end)
0ea865ce
RS
97#define ZIP_DIR_HEADER_SIZE offsetof(struct zip_dir_header, _end)
98#define ZIP_DIR_TRAILER_SIZE offsetof(struct zip_dir_trailer, _end)
227bf598
RS
99#define ZIP_EXTRA_MTIME_SIZE offsetof(struct zip_extra_mtime, _end)
100#define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
101 (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
88329ca8
RS
102#define ZIP64_DIR_TRAILER_SIZE offsetof(struct zip64_dir_trailer, _end)
103#define ZIP64_DIR_TRAILER_RECORD_SIZE \
104 (ZIP64_DIR_TRAILER_SIZE - \
105 offsetof(struct zip64_dir_trailer, creator_version))
106#define ZIP64_DIR_TRAILER_LOCATOR_SIZE \
107 offsetof(struct zip64_dir_trailer_locator, _end)
0ea865ce 108
e4fbbfe9
RS
109static void copy_le16(unsigned char *dest, unsigned int n)
110{
111 dest[0] = 0xff & n;
112 dest[1] = 0xff & (n >> 010);
113}
114
115static void copy_le32(unsigned char *dest, unsigned int n)
116{
117 dest[0] = 0xff & n;
118 dest[1] = 0xff & (n >> 010);
119 dest[2] = 0xff & (n >> 020);
120 dest[3] = 0xff & (n >> 030);
121}
122
88329ca8
RS
123static void copy_le64(unsigned char *dest, uint64_t n)
124{
125 dest[0] = 0xff & n;
126 dest[1] = 0xff & (n >> 010);
127 dest[2] = 0xff & (n >> 020);
128 dest[3] = 0xff & (n >> 030);
129 dest[4] = 0xff & (n >> 040);
130 dest[5] = 0xff & (n >> 050);
131 dest[6] = 0xff & (n >> 060);
132 dest[7] = 0xff & (n >> 070);
133}
134
135static uint64_t clamp_max(uint64_t n, uint64_t max, int *clamped)
136{
137 if (n <= max)
138 return n;
139 *clamped = 1;
140 return max;
141}
142
143static void copy_le16_clamp(unsigned char *dest, uint64_t n, int *clamped)
144{
145 copy_le16(dest, clamp_max(n, 0xffff, clamped));
146}
147
af95749f
RS
148static void copy_le32_clamp(unsigned char *dest, uint64_t n, int *clamped)
149{
150 copy_le32(dest, clamp_max(n, 0xffffffff, clamped));
151}
152
3c78fd80
RS
153static int strbuf_add_le(struct strbuf *sb, size_t size, uintmax_t n)
154{
155 while (size-- > 0) {
156 strbuf_addch(sb, n & 0xff);
157 n >>= 8;
158 }
159 return -!!n;
160}
161
af95749f
RS
162static uint32_t clamp32(uintmax_t n)
163{
164 const uintmax_t max = 0xffffffff;
165 return (n < max) ? n : max;
166}
167
c3c2e1a0
RS
168static void *zlib_deflate_raw(void *data, unsigned long size,
169 int compression_level,
170 unsigned long *compressed_size)
e4fbbfe9 171{
ef49a7a0 172 git_zstream stream;
e4fbbfe9
RS
173 unsigned long maxsize;
174 void *buffer;
175 int result;
176
c3c2e1a0 177 git_deflate_init_raw(&stream, compression_level);
225a6f10 178 maxsize = git_deflate_bound(&stream, size);
e4fbbfe9
RS
179 buffer = xmalloc(maxsize);
180
181 stream.next_in = data;
182 stream.avail_in = size;
183 stream.next_out = buffer;
184 stream.avail_out = maxsize;
185
186 do {
55bb5c91 187 result = git_deflate(&stream, Z_FINISH);
e4fbbfe9
RS
188 } while (result == Z_OK);
189
190 if (result != Z_STREAM_END) {
191 free(buffer);
192 return NULL;
193 }
194
55bb5c91 195 git_deflate_end(&stream);
e4fbbfe9
RS
196 *compressed_size = stream.total_out;
197
198 return buffer;
199}
200
2158f883
RS
201static void write_zip_data_desc(unsigned long size,
202 unsigned long compressed_size,
203 unsigned long crc)
204{
205 struct zip_data_desc trailer;
206
207 copy_le32(trailer.magic, 0x08074b50);
208 copy_le32(trailer.crc32, crc);
209 copy_le32(trailer.compressed_size, compressed_size);
210 copy_le32(trailer.size, size);
211 write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
212}
213
ebf5374a
RS
214static void set_zip_header_data_desc(struct zip_local_header *header,
215 unsigned long size,
216 unsigned long compressed_size,
217 unsigned long crc)
218{
219 copy_le32(header->crc32, crc);
220 copy_le32(header->compressed_size, compressed_size);
221 copy_le32(header->size, size);
222}
223
88182bab
RS
224static int has_only_ascii(const char *s)
225{
226 for (;;) {
227 int c = *s++;
228 if (c == '\0')
229 return 1;
230 if (!isascii(c))
231 return 0;
232 }
233}
234
4aff646d
RS
235static int entry_is_binary(const char *path, const void *buffer, size_t size)
236{
237 struct userdiff_driver *driver = userdiff_find_by_path(path);
238 if (!driver)
239 driver = userdiff_find_by_name("default");
240 if (driver->binary != -1)
241 return driver->binary;
242 return buffer_is_binary(buffer, size);
243}
244
2158f883
RS
245#define STREAM_BUFFER_SIZE (1024 * 16)
246
562e25ab 247static int write_zip_entry(struct archiver_args *args,
9cb513b7
NTND
248 const unsigned char *sha1,
249 const char *path, size_t pathlen,
250 unsigned int mode)
e4fbbfe9
RS
251{
252 struct zip_local_header header;
3c78fd80 253 uintmax_t offset = zip_offset;
227bf598 254 struct zip_extra_mtime extra;
bb52d22e 255 unsigned long attr2;
e4fbbfe9 256 unsigned long compressed_size;
e4fbbfe9 257 unsigned long crc;
e4fbbfe9 258 int method;
e4fbbfe9 259 unsigned char *out;
e4fbbfe9 260 void *deflated = NULL;
9cb513b7 261 void *buffer;
2158f883
RS
262 struct git_istream *stream = NULL;
263 unsigned long flags = 0;
9cb513b7 264 unsigned long size;
4aff646d
RS
265 int is_binary = -1;
266 const char *path_without_prefix = path + args->baselen;
0f747f9d 267 unsigned int creator_version = 0;
af95749f
RS
268 size_t zip_dir_extra_size = ZIP_EXTRA_MTIME_SIZE;
269 size_t zip64_dir_extra_payload_size = 0;
e4fbbfe9 270
38f4d138 271 crc = crc32(0, NULL, 0);
e4fbbfe9 272
88182bab
RS
273 if (!has_only_ascii(path)) {
274 if (is_utf8(path))
275 flags |= ZIP_UTF8;
276 else
277 warning("Path is not valid UTF-8: %s", path);
278 }
279
e4fbbfe9 280 if (pathlen > 0xffff) {
562e25ab
RS
281 return error("path too long (%d chars, SHA1: %s): %s",
282 (int)pathlen, sha1_to_hex(sha1), path);
e4fbbfe9
RS
283 }
284
302b9282 285 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
e4fbbfe9 286 method = 0;
62cdce17 287 attr2 = 16;
e4fbbfe9 288 out = NULL;
60df6bd1 289 size = 0;
e4fbbfe9 290 compressed_size = 0;
9cb513b7 291 buffer = NULL;
62cdce17 292 } else if (S_ISREG(mode) || S_ISLNK(mode)) {
2158f883 293 enum object_type type = sha1_object_info(sha1, &size);
9cb513b7 294
62cdce17 295 method = 0;
bb52d22e
JH
296 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
297 (mode & 0111) ? ((mode) << 16) : 0;
0f747f9d
RS
298 if (S_ISLNK(mode) || (mode & 0111))
299 creator_version = 0x0317;
2158f883 300 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
62cdce17 301 method = 8;
2158f883
RS
302
303 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
c743c215 304 size > big_file_threshold) {
2158f883
RS
305 stream = open_istream(sha1, &type, &size, NULL);
306 if (!stream)
307 return error("cannot stream blob %s",
308 sha1_to_hex(sha1));
309 flags |= ZIP_STREAM;
310 out = buffer = NULL;
311 } else {
312 buffer = sha1_file_to_archive(args, path, sha1, mode,
313 &type, &size);
314 if (!buffer)
315 return error("cannot read %s",
316 sha1_to_hex(sha1));
317 crc = crc32(crc, buffer, size);
4aff646d
RS
318 is_binary = entry_is_binary(path_without_prefix,
319 buffer, size);
2158f883
RS
320 out = buffer;
321 }
d3c1472f 322 compressed_size = (method == 0) ? size : 0;
e4fbbfe9 323 } else {
562e25ab
RS
324 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
325 sha1_to_hex(sha1));
e4fbbfe9
RS
326 }
327
88329ca8
RS
328 if (creator_version > max_creator_version)
329 max_creator_version = creator_version;
330
2158f883 331 if (buffer && method == 8) {
c3c2e1a0
RS
332 out = deflated = zlib_deflate_raw(buffer, size,
333 args->compression_level,
334 &compressed_size);
335 if (!out || compressed_size >= size) {
336 out = buffer;
e4fbbfe9
RS
337 method = 0;
338 compressed_size = size;
339 }
340 }
341
227bf598
RS
342 copy_le16(extra.magic, 0x5455);
343 copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
344 extra.flags[0] = 1; /* just mtime */
345 copy_le32(extra.mtime, args->time);
346
e4fbbfe9 347 copy_le32(header.magic, 0x04034b50);
cf72fb07 348 copy_le16(header.version, 10);
2158f883 349 copy_le16(header.flags, flags);
e4fbbfe9
RS
350 copy_le16(header.compression_method, method);
351 copy_le16(header.mtime, zip_time);
352 copy_le16(header.mdate, zip_date);
5ea2c847 353 set_zip_header_data_desc(&header, size, compressed_size, crc);
e4fbbfe9 354 copy_le16(header.filename_length, pathlen);
227bf598 355 copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
0ea865ce
RS
356 write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
357 zip_offset += ZIP_LOCAL_HEADER_SIZE;
e4fbbfe9
RS
358 write_or_die(1, path, pathlen);
359 zip_offset += pathlen;
227bf598
RS
360 write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
361 zip_offset += ZIP_EXTRA_MTIME_SIZE;
2158f883
RS
362 if (stream && method == 0) {
363 unsigned char buf[STREAM_BUFFER_SIZE];
364 ssize_t readlen;
365
366 for (;;) {
367 readlen = read_istream(stream, buf, sizeof(buf));
368 if (readlen <= 0)
369 break;
370 crc = crc32(crc, buf, readlen);
4aff646d
RS
371 if (is_binary == -1)
372 is_binary = entry_is_binary(path_without_prefix,
373 buf, readlen);
2158f883
RS
374 write_or_die(1, buf, readlen);
375 }
376 close_istream(stream);
377 if (readlen)
378 return readlen;
379
380 compressed_size = size;
381 zip_offset += compressed_size;
382
383 write_zip_data_desc(size, compressed_size, crc);
384 zip_offset += ZIP_DATA_DESC_SIZE;
c743c215
RS
385 } else if (stream && method == 8) {
386 unsigned char buf[STREAM_BUFFER_SIZE];
387 ssize_t readlen;
388 git_zstream zstream;
389 int result;
390 size_t out_len;
391 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
392
c3c2e1a0 393 git_deflate_init_raw(&zstream, args->compression_level);
c743c215
RS
394
395 compressed_size = 0;
396 zstream.next_out = compressed;
397 zstream.avail_out = sizeof(compressed);
398
399 for (;;) {
400 readlen = read_istream(stream, buf, sizeof(buf));
401 if (readlen <= 0)
402 break;
403 crc = crc32(crc, buf, readlen);
4aff646d
RS
404 if (is_binary == -1)
405 is_binary = entry_is_binary(path_without_prefix,
406 buf, readlen);
c743c215
RS
407
408 zstream.next_in = buf;
409 zstream.avail_in = readlen;
410 result = git_deflate(&zstream, 0);
411 if (result != Z_OK)
412 die("deflate error (%d)", result);
c3c2e1a0 413 out_len = zstream.next_out - compressed;
c743c215
RS
414
415 if (out_len > 0) {
c3c2e1a0 416 write_or_die(1, compressed, out_len);
c743c215
RS
417 compressed_size += out_len;
418 zstream.next_out = compressed;
419 zstream.avail_out = sizeof(compressed);
420 }
421
422 }
423 close_istream(stream);
424 if (readlen)
425 return readlen;
426
427 zstream.next_in = buf;
428 zstream.avail_in = 0;
429 result = git_deflate(&zstream, Z_FINISH);
430 if (result != Z_STREAM_END)
431 die("deflate error (%d)", result);
432
433 git_deflate_end(&zstream);
c3c2e1a0
RS
434 out_len = zstream.next_out - compressed;
435 write_or_die(1, compressed, out_len);
c743c215
RS
436 compressed_size += out_len;
437 zip_offset += compressed_size;
438
439 write_zip_data_desc(size, compressed_size, crc);
440 zip_offset += ZIP_DATA_DESC_SIZE;
2158f883 441 } else if (compressed_size > 0) {
e4fbbfe9
RS
442 write_or_die(1, out, compressed_size);
443 zip_offset += compressed_size;
444 }
445
e4fbbfe9 446 free(deflated);
9cb513b7 447 free(buffer);
e4fbbfe9 448
af95749f
RS
449 if (offset > 0xffffffff) {
450 zip64_dir_extra_payload_size += 8;
451 zip_dir_extra_size += 2 + 2 + zip64_dir_extra_payload_size;
452 }
453
3c78fd80
RS
454 strbuf_add_le(&zip_dir, 4, 0x02014b50); /* magic */
455 strbuf_add_le(&zip_dir, 2, creator_version);
456 strbuf_add_le(&zip_dir, 2, 10); /* version */
457 strbuf_add_le(&zip_dir, 2, flags);
458 strbuf_add_le(&zip_dir, 2, method);
459 strbuf_add_le(&zip_dir, 2, zip_time);
460 strbuf_add_le(&zip_dir, 2, zip_date);
461 strbuf_add_le(&zip_dir, 4, crc);
462 strbuf_add_le(&zip_dir, 4, compressed_size);
463 strbuf_add_le(&zip_dir, 4, size);
464 strbuf_add_le(&zip_dir, 2, pathlen);
af95749f 465 strbuf_add_le(&zip_dir, 2, zip_dir_extra_size);
3c78fd80
RS
466 strbuf_add_le(&zip_dir, 2, 0); /* comment length */
467 strbuf_add_le(&zip_dir, 2, 0); /* disk */
468 strbuf_add_le(&zip_dir, 2, !is_binary);
469 strbuf_add_le(&zip_dir, 4, attr2);
af95749f 470 strbuf_add_le(&zip_dir, 4, clamp32(offset));
c061a149
RS
471 strbuf_add(&zip_dir, path, pathlen);
472 strbuf_add(&zip_dir, &extra, ZIP_EXTRA_MTIME_SIZE);
af95749f
RS
473 if (zip64_dir_extra_payload_size) {
474 strbuf_add_le(&zip_dir, 2, 0x0001); /* magic */
475 strbuf_add_le(&zip_dir, 2, zip64_dir_extra_payload_size);
476 if (offset >= 0xffffffff)
477 strbuf_add_le(&zip_dir, 8, offset);
478 }
ebf5374a
RS
479 zip_dir_entries++;
480
562e25ab 481 return 0;
e4fbbfe9
RS
482}
483
88329ca8
RS
484static void write_zip64_trailer(void)
485{
486 struct zip64_dir_trailer trailer64;
487 struct zip64_dir_trailer_locator locator64;
488
489 copy_le32(trailer64.magic, 0x06064b50);
490 copy_le64(trailer64.record_size, ZIP64_DIR_TRAILER_RECORD_SIZE);
491 copy_le16(trailer64.creator_version, max_creator_version);
492 copy_le16(trailer64.version, 45);
493 copy_le32(trailer64.disk, 0);
494 copy_le32(trailer64.directory_start_disk, 0);
495 copy_le64(trailer64.entries_on_this_disk, zip_dir_entries);
496 copy_le64(trailer64.entries, zip_dir_entries);
c061a149 497 copy_le64(trailer64.size, zip_dir.len);
88329ca8
RS
498 copy_le64(trailer64.offset, zip_offset);
499
500 copy_le32(locator64.magic, 0x07064b50);
501 copy_le32(locator64.disk, 0);
c061a149 502 copy_le64(locator64.offset, zip_offset + zip_dir.len);
88329ca8
RS
503 copy_le32(locator64.number_of_disks, 1);
504
505 write_or_die(1, &trailer64, ZIP64_DIR_TRAILER_SIZE);
506 write_or_die(1, &locator64, ZIP64_DIR_TRAILER_LOCATOR_SIZE);
507}
508
e4fbbfe9
RS
509static void write_zip_trailer(const unsigned char *sha1)
510{
511 struct zip_dir_trailer trailer;
88329ca8 512 int clamped = 0;
e4fbbfe9
RS
513
514 copy_le32(trailer.magic, 0x06054b50);
515 copy_le16(trailer.disk, 0);
516 copy_le16(trailer.directory_start_disk, 0);
88329ca8
RS
517 copy_le16_clamp(trailer.entries_on_this_disk, zip_dir_entries,
518 &clamped);
519 copy_le16_clamp(trailer.entries, zip_dir_entries, &clamped);
c061a149 520 copy_le32(trailer.size, zip_dir.len);
af95749f 521 copy_le32_clamp(trailer.offset, zip_offset, &clamped);
aeecdcd4 522 copy_le16(trailer.comment_length, sha1 ? GIT_SHA1_HEXSZ : 0);
e4fbbfe9 523
c061a149 524 write_or_die(1, zip_dir.buf, zip_dir.len);
88329ca8
RS
525 if (clamped)
526 write_zip64_trailer();
0ea865ce 527 write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
e4fbbfe9 528 if (sha1)
aeecdcd4 529 write_or_die(1, sha1_to_hex(sha1), GIT_SHA1_HEXSZ);
e4fbbfe9
RS
530}
531
532static void dos_time(time_t *time, int *dos_date, int *dos_time)
533{
534 struct tm *t = localtime(time);
535
536 *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
537 (t->tm_year + 1900 - 1980) * 512;
538 *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
539}
540
965cba2e
JK
541static int archive_zip_config(const char *var, const char *value, void *data)
542{
543 return userdiff_config(var, value);
544}
545
4d7c9898
JK
546static int write_zip_archive(const struct archiver *ar,
547 struct archiver_args *args)
ec06bff5 548{
562e25ab
RS
549 int err;
550
965cba2e
JK
551 git_config(archive_zip_config, NULL);
552
ec06bff5
FBH
553 dos_time(&args->time, &zip_date, &zip_time);
554
c061a149 555 strbuf_init(&zip_dir, 0);
562e25ab
RS
556
557 err = write_archive_entries(args, write_zip_entry);
558 if (!err)
559 write_zip_trailer(args->commit_sha1);
ec06bff5 560
c061a149 561 strbuf_release(&zip_dir);
ec06bff5 562
562e25ab 563 return err;
ec06bff5 564}
13e0f88d
JK
565
566static struct archiver zip_archiver = {
567 "zip",
568 write_zip_archive,
7b97730b 569 ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
13e0f88d
JK
570};
571
572void init_zip_archiver(void)
573{
574 register_archiver(&zip_archiver);
575}