archive-zip: support files bigger than 4GB
[git.git] / archive-zip.c
blob44ed78f163e38449862663bf89153945e274a6c5
1 /*
2 * Copyright (c) 2006 Rene Scharfe
3 */
4 #include "cache.h"
5 #include "archive.h"
6 #include "streaming.h"
7 #include "utf8.h"
8 #include "userdiff.h"
9 #include "xdiff-interface.h"
11 static int zip_date;
12 static int zip_time;
14 /* We only care about the "buf" part here. */
15 static struct strbuf zip_dir;
17 static uintmax_t zip_offset;
18 static uint64_t zip_dir_entries;
20 static unsigned int max_creator_version;
22 #define ZIP_STREAM (1 << 3)
23 #define ZIP_UTF8 (1 << 11)
25 struct 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];
37 unsigned char _end[1];
40 struct 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];
48 struct zip64_data_desc {
49 unsigned char magic[4];
50 unsigned char crc32[4];
51 unsigned char compressed_size[8];
52 unsigned char size[8];
53 unsigned char _end[1];
56 struct zip_dir_trailer {
57 unsigned char magic[4];
58 unsigned char disk[2];
59 unsigned char directory_start_disk[2];
60 unsigned char entries_on_this_disk[2];
61 unsigned char entries[2];
62 unsigned char size[4];
63 unsigned char offset[4];
64 unsigned char comment_length[2];
65 unsigned char _end[1];
68 struct zip_extra_mtime {
69 unsigned char magic[2];
70 unsigned char extra_size[2];
71 unsigned char flags[1];
72 unsigned char mtime[4];
73 unsigned char _end[1];
76 struct zip64_extra {
77 unsigned char magic[2];
78 unsigned char extra_size[2];
79 unsigned char size[8];
80 unsigned char compressed_size[8];
81 unsigned char _end[1];
84 struct zip64_dir_trailer {
85 unsigned char magic[4];
86 unsigned char record_size[8];
87 unsigned char creator_version[2];
88 unsigned char version[2];
89 unsigned char disk[4];
90 unsigned char directory_start_disk[4];
91 unsigned char entries_on_this_disk[8];
92 unsigned char entries[8];
93 unsigned char size[8];
94 unsigned char offset[8];
95 unsigned char _end[1];
98 struct zip64_dir_trailer_locator {
99 unsigned char magic[4];
100 unsigned char disk[4];
101 unsigned char offset[8];
102 unsigned char number_of_disks[4];
103 unsigned char _end[1];
107 * On ARM, padding is added at the end of the struct, so a simple
108 * sizeof(struct ...) reports two bytes more than the payload size
109 * we're interested in.
111 #define ZIP_LOCAL_HEADER_SIZE offsetof(struct zip_local_header, _end)
112 #define ZIP_DATA_DESC_SIZE offsetof(struct zip_data_desc, _end)
113 #define ZIP64_DATA_DESC_SIZE offsetof(struct zip64_data_desc, _end)
114 #define ZIP_DIR_HEADER_SIZE offsetof(struct zip_dir_header, _end)
115 #define ZIP_DIR_TRAILER_SIZE offsetof(struct zip_dir_trailer, _end)
116 #define ZIP_EXTRA_MTIME_SIZE offsetof(struct zip_extra_mtime, _end)
117 #define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
118 (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
119 #define ZIP64_EXTRA_SIZE offsetof(struct zip64_extra, _end)
120 #define ZIP64_EXTRA_PAYLOAD_SIZE \
121 (ZIP64_EXTRA_SIZE - offsetof(struct zip64_extra, size))
122 #define ZIP64_DIR_TRAILER_SIZE offsetof(struct zip64_dir_trailer, _end)
123 #define ZIP64_DIR_TRAILER_RECORD_SIZE \
124 (ZIP64_DIR_TRAILER_SIZE - \
125 offsetof(struct zip64_dir_trailer, creator_version))
126 #define ZIP64_DIR_TRAILER_LOCATOR_SIZE \
127 offsetof(struct zip64_dir_trailer_locator, _end)
129 static void copy_le16(unsigned char *dest, unsigned int n)
131 dest[0] = 0xff & n;
132 dest[1] = 0xff & (n >> 010);
135 static void copy_le32(unsigned char *dest, unsigned int n)
137 dest[0] = 0xff & n;
138 dest[1] = 0xff & (n >> 010);
139 dest[2] = 0xff & (n >> 020);
140 dest[3] = 0xff & (n >> 030);
143 static void copy_le64(unsigned char *dest, uint64_t n)
145 dest[0] = 0xff & n;
146 dest[1] = 0xff & (n >> 010);
147 dest[2] = 0xff & (n >> 020);
148 dest[3] = 0xff & (n >> 030);
149 dest[4] = 0xff & (n >> 040);
150 dest[5] = 0xff & (n >> 050);
151 dest[6] = 0xff & (n >> 060);
152 dest[7] = 0xff & (n >> 070);
155 static uint64_t clamp_max(uint64_t n, uint64_t max, int *clamped)
157 if (n <= max)
158 return n;
159 *clamped = 1;
160 return max;
163 static void copy_le16_clamp(unsigned char *dest, uint64_t n, int *clamped)
165 copy_le16(dest, clamp_max(n, 0xffff, clamped));
168 static void copy_le32_clamp(unsigned char *dest, uint64_t n, int *clamped)
170 copy_le32(dest, clamp_max(n, 0xffffffff, clamped));
173 static int strbuf_add_le(struct strbuf *sb, size_t size, uintmax_t n)
175 while (size-- > 0) {
176 strbuf_addch(sb, n & 0xff);
177 n >>= 8;
179 return -!!n;
182 static uint32_t clamp32(uintmax_t n)
184 const uintmax_t max = 0xffffffff;
185 return (n < max) ? n : max;
188 static void *zlib_deflate_raw(void *data, unsigned long size,
189 int compression_level,
190 unsigned long *compressed_size)
192 git_zstream stream;
193 unsigned long maxsize;
194 void *buffer;
195 int result;
197 git_deflate_init_raw(&stream, compression_level);
198 maxsize = git_deflate_bound(&stream, size);
199 buffer = xmalloc(maxsize);
201 stream.next_in = data;
202 stream.avail_in = size;
203 stream.next_out = buffer;
204 stream.avail_out = maxsize;
206 do {
207 result = git_deflate(&stream, Z_FINISH);
208 } while (result == Z_OK);
210 if (result != Z_STREAM_END) {
211 free(buffer);
212 return NULL;
215 git_deflate_end(&stream);
216 *compressed_size = stream.total_out;
218 return buffer;
221 static void write_zip_data_desc(unsigned long size,
222 unsigned long compressed_size,
223 unsigned long crc)
225 if (size >= 0xffffffff || compressed_size >= 0xffffffff) {
226 struct zip64_data_desc trailer;
227 copy_le32(trailer.magic, 0x08074b50);
228 copy_le32(trailer.crc32, crc);
229 copy_le64(trailer.compressed_size, compressed_size);
230 copy_le64(trailer.size, size);
231 write_or_die(1, &trailer, ZIP64_DATA_DESC_SIZE);
232 zip_offset += ZIP64_DATA_DESC_SIZE;
233 } else {
234 struct zip_data_desc trailer;
235 copy_le32(trailer.magic, 0x08074b50);
236 copy_le32(trailer.crc32, crc);
237 copy_le32(trailer.compressed_size, compressed_size);
238 copy_le32(trailer.size, size);
239 write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
240 zip_offset += ZIP_DATA_DESC_SIZE;
244 static void set_zip_header_data_desc(struct zip_local_header *header,
245 unsigned long size,
246 unsigned long compressed_size,
247 unsigned long crc)
249 copy_le32(header->crc32, crc);
250 copy_le32(header->compressed_size, compressed_size);
251 copy_le32(header->size, size);
254 static int has_only_ascii(const char *s)
256 for (;;) {
257 int c = *s++;
258 if (c == '\0')
259 return 1;
260 if (!isascii(c))
261 return 0;
265 static int entry_is_binary(const char *path, const void *buffer, size_t size)
267 struct userdiff_driver *driver = userdiff_find_by_path(path);
268 if (!driver)
269 driver = userdiff_find_by_name("default");
270 if (driver->binary != -1)
271 return driver->binary;
272 return buffer_is_binary(buffer, size);
275 #define STREAM_BUFFER_SIZE (1024 * 16)
277 static int write_zip_entry(struct archiver_args *args,
278 const unsigned char *sha1,
279 const char *path, size_t pathlen,
280 unsigned int mode)
282 struct zip_local_header header;
283 uintmax_t offset = zip_offset;
284 struct zip_extra_mtime extra;
285 struct zip64_extra extra64;
286 size_t header_extra_size = ZIP_EXTRA_MTIME_SIZE;
287 int need_zip64_extra = 0;
288 unsigned long attr2;
289 unsigned long compressed_size;
290 unsigned long crc;
291 int method;
292 unsigned char *out;
293 void *deflated = NULL;
294 void *buffer;
295 struct git_istream *stream = NULL;
296 unsigned long flags = 0;
297 unsigned long size;
298 int is_binary = -1;
299 const char *path_without_prefix = path + args->baselen;
300 unsigned int creator_version = 0;
301 size_t zip_dir_extra_size = ZIP_EXTRA_MTIME_SIZE;
302 size_t zip64_dir_extra_payload_size = 0;
304 crc = crc32(0, NULL, 0);
306 if (!has_only_ascii(path)) {
307 if (is_utf8(path))
308 flags |= ZIP_UTF8;
309 else
310 warning("Path is not valid UTF-8: %s", path);
313 if (pathlen > 0xffff) {
314 return error("path too long (%d chars, SHA1: %s): %s",
315 (int)pathlen, sha1_to_hex(sha1), path);
318 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
319 method = 0;
320 attr2 = 16;
321 out = NULL;
322 size = 0;
323 compressed_size = 0;
324 buffer = NULL;
325 } else if (S_ISREG(mode) || S_ISLNK(mode)) {
326 enum object_type type = sha1_object_info(sha1, &size);
328 method = 0;
329 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
330 (mode & 0111) ? ((mode) << 16) : 0;
331 if (S_ISLNK(mode) || (mode & 0111))
332 creator_version = 0x0317;
333 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
334 method = 8;
336 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
337 size > big_file_threshold) {
338 stream = open_istream(sha1, &type, &size, NULL);
339 if (!stream)
340 return error("cannot stream blob %s",
341 sha1_to_hex(sha1));
342 flags |= ZIP_STREAM;
343 out = buffer = NULL;
344 } else {
345 buffer = sha1_file_to_archive(args, path, sha1, mode,
346 &type, &size);
347 if (!buffer)
348 return error("cannot read %s",
349 sha1_to_hex(sha1));
350 crc = crc32(crc, buffer, size);
351 is_binary = entry_is_binary(path_without_prefix,
352 buffer, size);
353 out = buffer;
355 compressed_size = (method == 0) ? size : 0;
356 } else {
357 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
358 sha1_to_hex(sha1));
361 if (creator_version > max_creator_version)
362 max_creator_version = creator_version;
364 if (buffer && method == 8) {
365 out = deflated = zlib_deflate_raw(buffer, size,
366 args->compression_level,
367 &compressed_size);
368 if (!out || compressed_size >= size) {
369 out = buffer;
370 method = 0;
371 compressed_size = size;
375 copy_le16(extra.magic, 0x5455);
376 copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
377 extra.flags[0] = 1; /* just mtime */
378 copy_le32(extra.mtime, args->time);
380 if (size > 0xffffffff || compressed_size > 0xffffffff)
381 need_zip64_extra = 1;
382 if (stream && size > 0x7fffffff)
383 need_zip64_extra = 1;
385 copy_le32(header.magic, 0x04034b50);
386 copy_le16(header.version, 10);
387 copy_le16(header.flags, flags);
388 copy_le16(header.compression_method, method);
389 copy_le16(header.mtime, zip_time);
390 copy_le16(header.mdate, zip_date);
391 if (need_zip64_extra) {
392 set_zip_header_data_desc(&header, 0xffffffff, 0xffffffff, crc);
393 header_extra_size += ZIP64_EXTRA_SIZE;
394 } else {
395 set_zip_header_data_desc(&header, size, compressed_size, crc);
397 copy_le16(header.filename_length, pathlen);
398 copy_le16(header.extra_length, header_extra_size);
399 write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
400 zip_offset += ZIP_LOCAL_HEADER_SIZE;
401 write_or_die(1, path, pathlen);
402 zip_offset += pathlen;
403 write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
404 zip_offset += ZIP_EXTRA_MTIME_SIZE;
405 if (need_zip64_extra) {
406 copy_le16(extra64.magic, 0x0001);
407 copy_le16(extra64.extra_size, ZIP64_EXTRA_PAYLOAD_SIZE);
408 copy_le64(extra64.size, size);
409 copy_le64(extra64.compressed_size, compressed_size);
410 write_or_die(1, &extra64, ZIP64_EXTRA_SIZE);
411 zip_offset += ZIP64_EXTRA_SIZE;
414 if (stream && method == 0) {
415 unsigned char buf[STREAM_BUFFER_SIZE];
416 ssize_t readlen;
418 for (;;) {
419 readlen = read_istream(stream, buf, sizeof(buf));
420 if (readlen <= 0)
421 break;
422 crc = crc32(crc, buf, readlen);
423 if (is_binary == -1)
424 is_binary = entry_is_binary(path_without_prefix,
425 buf, readlen);
426 write_or_die(1, buf, readlen);
428 close_istream(stream);
429 if (readlen)
430 return readlen;
432 compressed_size = size;
433 zip_offset += compressed_size;
435 write_zip_data_desc(size, compressed_size, crc);
436 } else if (stream && method == 8) {
437 unsigned char buf[STREAM_BUFFER_SIZE];
438 ssize_t readlen;
439 git_zstream zstream;
440 int result;
441 size_t out_len;
442 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
444 git_deflate_init_raw(&zstream, args->compression_level);
446 compressed_size = 0;
447 zstream.next_out = compressed;
448 zstream.avail_out = sizeof(compressed);
450 for (;;) {
451 readlen = read_istream(stream, buf, sizeof(buf));
452 if (readlen <= 0)
453 break;
454 crc = crc32(crc, buf, readlen);
455 if (is_binary == -1)
456 is_binary = entry_is_binary(path_without_prefix,
457 buf, readlen);
459 zstream.next_in = buf;
460 zstream.avail_in = readlen;
461 result = git_deflate(&zstream, 0);
462 if (result != Z_OK)
463 die("deflate error (%d)", result);
464 out_len = zstream.next_out - compressed;
466 if (out_len > 0) {
467 write_or_die(1, compressed, out_len);
468 compressed_size += out_len;
469 zstream.next_out = compressed;
470 zstream.avail_out = sizeof(compressed);
474 close_istream(stream);
475 if (readlen)
476 return readlen;
478 zstream.next_in = buf;
479 zstream.avail_in = 0;
480 result = git_deflate(&zstream, Z_FINISH);
481 if (result != Z_STREAM_END)
482 die("deflate error (%d)", result);
484 git_deflate_end(&zstream);
485 out_len = zstream.next_out - compressed;
486 write_or_die(1, compressed, out_len);
487 compressed_size += out_len;
488 zip_offset += compressed_size;
490 write_zip_data_desc(size, compressed_size, crc);
491 } else if (compressed_size > 0) {
492 write_or_die(1, out, compressed_size);
493 zip_offset += compressed_size;
496 free(deflated);
497 free(buffer);
499 if (compressed_size > 0xffffffff || size > 0xffffffff ||
500 offset > 0xffffffff) {
501 if (compressed_size >= 0xffffffff)
502 zip64_dir_extra_payload_size += 8;
503 if (size >= 0xffffffff)
504 zip64_dir_extra_payload_size += 8;
505 if (offset >= 0xffffffff)
506 zip64_dir_extra_payload_size += 8;
507 zip_dir_extra_size += 2 + 2 + zip64_dir_extra_payload_size;
510 strbuf_add_le(&zip_dir, 4, 0x02014b50); /* magic */
511 strbuf_add_le(&zip_dir, 2, creator_version);
512 strbuf_add_le(&zip_dir, 2, 10); /* version */
513 strbuf_add_le(&zip_dir, 2, flags);
514 strbuf_add_le(&zip_dir, 2, method);
515 strbuf_add_le(&zip_dir, 2, zip_time);
516 strbuf_add_le(&zip_dir, 2, zip_date);
517 strbuf_add_le(&zip_dir, 4, crc);
518 strbuf_add_le(&zip_dir, 4, clamp32(compressed_size));
519 strbuf_add_le(&zip_dir, 4, clamp32(size));
520 strbuf_add_le(&zip_dir, 2, pathlen);
521 strbuf_add_le(&zip_dir, 2, zip_dir_extra_size);
522 strbuf_add_le(&zip_dir, 2, 0); /* comment length */
523 strbuf_add_le(&zip_dir, 2, 0); /* disk */
524 strbuf_add_le(&zip_dir, 2, !is_binary);
525 strbuf_add_le(&zip_dir, 4, attr2);
526 strbuf_add_le(&zip_dir, 4, clamp32(offset));
527 strbuf_add(&zip_dir, path, pathlen);
528 strbuf_add(&zip_dir, &extra, ZIP_EXTRA_MTIME_SIZE);
529 if (zip64_dir_extra_payload_size) {
530 strbuf_add_le(&zip_dir, 2, 0x0001); /* magic */
531 strbuf_add_le(&zip_dir, 2, zip64_dir_extra_payload_size);
532 if (size >= 0xffffffff)
533 strbuf_add_le(&zip_dir, 8, size);
534 if (compressed_size >= 0xffffffff)
535 strbuf_add_le(&zip_dir, 8, compressed_size);
536 if (offset >= 0xffffffff)
537 strbuf_add_le(&zip_dir, 8, offset);
539 zip_dir_entries++;
541 return 0;
544 static void write_zip64_trailer(void)
546 struct zip64_dir_trailer trailer64;
547 struct zip64_dir_trailer_locator locator64;
549 copy_le32(trailer64.magic, 0x06064b50);
550 copy_le64(trailer64.record_size, ZIP64_DIR_TRAILER_RECORD_SIZE);
551 copy_le16(trailer64.creator_version, max_creator_version);
552 copy_le16(trailer64.version, 45);
553 copy_le32(trailer64.disk, 0);
554 copy_le32(trailer64.directory_start_disk, 0);
555 copy_le64(trailer64.entries_on_this_disk, zip_dir_entries);
556 copy_le64(trailer64.entries, zip_dir_entries);
557 copy_le64(trailer64.size, zip_dir.len);
558 copy_le64(trailer64.offset, zip_offset);
560 copy_le32(locator64.magic, 0x07064b50);
561 copy_le32(locator64.disk, 0);
562 copy_le64(locator64.offset, zip_offset + zip_dir.len);
563 copy_le32(locator64.number_of_disks, 1);
565 write_or_die(1, &trailer64, ZIP64_DIR_TRAILER_SIZE);
566 write_or_die(1, &locator64, ZIP64_DIR_TRAILER_LOCATOR_SIZE);
569 static void write_zip_trailer(const unsigned char *sha1)
571 struct zip_dir_trailer trailer;
572 int clamped = 0;
574 copy_le32(trailer.magic, 0x06054b50);
575 copy_le16(trailer.disk, 0);
576 copy_le16(trailer.directory_start_disk, 0);
577 copy_le16_clamp(trailer.entries_on_this_disk, zip_dir_entries,
578 &clamped);
579 copy_le16_clamp(trailer.entries, zip_dir_entries, &clamped);
580 copy_le32(trailer.size, zip_dir.len);
581 copy_le32_clamp(trailer.offset, zip_offset, &clamped);
582 copy_le16(trailer.comment_length, sha1 ? GIT_SHA1_HEXSZ : 0);
584 write_or_die(1, zip_dir.buf, zip_dir.len);
585 if (clamped)
586 write_zip64_trailer();
587 write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
588 if (sha1)
589 write_or_die(1, sha1_to_hex(sha1), GIT_SHA1_HEXSZ);
592 static void dos_time(time_t *time, int *dos_date, int *dos_time)
594 struct tm *t = localtime(time);
596 *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
597 (t->tm_year + 1900 - 1980) * 512;
598 *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
601 static int archive_zip_config(const char *var, const char *value, void *data)
603 return userdiff_config(var, value);
606 static int write_zip_archive(const struct archiver *ar,
607 struct archiver_args *args)
609 int err;
611 git_config(archive_zip_config, NULL);
613 dos_time(&args->time, &zip_date, &zip_time);
615 strbuf_init(&zip_dir, 0);
617 err = write_archive_entries(args, write_zip_entry);
618 if (!err)
619 write_zip_trailer(args->commit_sha1);
621 strbuf_release(&zip_dir);
623 return err;
626 static struct archiver zip_archiver = {
627 "zip",
628 write_zip_archive,
629 ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
632 void init_zip_archiver(void)
634 register_archiver(&zip_archiver);