t6036: fix broken "merge fails but has appropriate contents" tests
[git/debian.git] / archive-zip.c
blob74f3fe9103420571c9f22273ae44ebb9d5715c78
1 /*
2 * Copyright (c) 2006 Rene Scharfe
3 */
4 #include "cache.h"
5 #include "config.h"
6 #include "archive.h"
7 #include "streaming.h"
8 #include "utf8.h"
9 #include "userdiff.h"
10 #include "xdiff-interface.h"
12 static int zip_date;
13 static int zip_time;
15 /* We only care about the "buf" part here. */
16 static struct strbuf zip_dir;
18 static uintmax_t zip_offset;
19 static uint64_t zip_dir_entries;
21 static unsigned int max_creator_version;
23 #define ZIP_STREAM (1 << 3)
24 #define ZIP_UTF8 (1 << 11)
26 struct zip_local_header {
27 unsigned char magic[4];
28 unsigned char version[2];
29 unsigned char flags[2];
30 unsigned char compression_method[2];
31 unsigned char mtime[2];
32 unsigned char mdate[2];
33 unsigned char crc32[4];
34 unsigned char compressed_size[4];
35 unsigned char size[4];
36 unsigned char filename_length[2];
37 unsigned char extra_length[2];
38 unsigned char _end[1];
41 struct zip_data_desc {
42 unsigned char magic[4];
43 unsigned char crc32[4];
44 unsigned char compressed_size[4];
45 unsigned char size[4];
46 unsigned char _end[1];
49 struct zip64_data_desc {
50 unsigned char magic[4];
51 unsigned char crc32[4];
52 unsigned char compressed_size[8];
53 unsigned char size[8];
54 unsigned char _end[1];
57 struct zip_dir_trailer {
58 unsigned char magic[4];
59 unsigned char disk[2];
60 unsigned char directory_start_disk[2];
61 unsigned char entries_on_this_disk[2];
62 unsigned char entries[2];
63 unsigned char size[4];
64 unsigned char offset[4];
65 unsigned char comment_length[2];
66 unsigned char _end[1];
69 struct zip_extra_mtime {
70 unsigned char magic[2];
71 unsigned char extra_size[2];
72 unsigned char flags[1];
73 unsigned char mtime[4];
74 unsigned char _end[1];
77 struct zip64_extra {
78 unsigned char magic[2];
79 unsigned char extra_size[2];
80 unsigned char size[8];
81 unsigned char compressed_size[8];
82 unsigned char _end[1];
85 struct zip64_dir_trailer {
86 unsigned char magic[4];
87 unsigned char record_size[8];
88 unsigned char creator_version[2];
89 unsigned char version[2];
90 unsigned char disk[4];
91 unsigned char directory_start_disk[4];
92 unsigned char entries_on_this_disk[8];
93 unsigned char entries[8];
94 unsigned char size[8];
95 unsigned char offset[8];
96 unsigned char _end[1];
99 struct zip64_dir_trailer_locator {
100 unsigned char magic[4];
101 unsigned char disk[4];
102 unsigned char offset[8];
103 unsigned char number_of_disks[4];
104 unsigned char _end[1];
108 * On ARM, padding is added at the end of the struct, so a simple
109 * sizeof(struct ...) reports two bytes more than the payload size
110 * we're interested in.
112 #define ZIP_LOCAL_HEADER_SIZE offsetof(struct zip_local_header, _end)
113 #define ZIP_DATA_DESC_SIZE offsetof(struct zip_data_desc, _end)
114 #define ZIP64_DATA_DESC_SIZE offsetof(struct zip64_data_desc, _end)
115 #define ZIP_DIR_HEADER_SIZE offsetof(struct zip_dir_header, _end)
116 #define ZIP_DIR_TRAILER_SIZE offsetof(struct zip_dir_trailer, _end)
117 #define ZIP_EXTRA_MTIME_SIZE offsetof(struct zip_extra_mtime, _end)
118 #define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
119 (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
120 #define ZIP64_EXTRA_SIZE offsetof(struct zip64_extra, _end)
121 #define ZIP64_EXTRA_PAYLOAD_SIZE \
122 (ZIP64_EXTRA_SIZE - offsetof(struct zip64_extra, size))
123 #define ZIP64_DIR_TRAILER_SIZE offsetof(struct zip64_dir_trailer, _end)
124 #define ZIP64_DIR_TRAILER_RECORD_SIZE \
125 (ZIP64_DIR_TRAILER_SIZE - \
126 offsetof(struct zip64_dir_trailer, creator_version))
127 #define ZIP64_DIR_TRAILER_LOCATOR_SIZE \
128 offsetof(struct zip64_dir_trailer_locator, _end)
130 static void copy_le16(unsigned char *dest, unsigned int n)
132 dest[0] = 0xff & n;
133 dest[1] = 0xff & (n >> 010);
136 static void copy_le32(unsigned char *dest, unsigned int n)
138 dest[0] = 0xff & n;
139 dest[1] = 0xff & (n >> 010);
140 dest[2] = 0xff & (n >> 020);
141 dest[3] = 0xff & (n >> 030);
144 static void copy_le64(unsigned char *dest, uint64_t n)
146 dest[0] = 0xff & n;
147 dest[1] = 0xff & (n >> 010);
148 dest[2] = 0xff & (n >> 020);
149 dest[3] = 0xff & (n >> 030);
150 dest[4] = 0xff & (n >> 040);
151 dest[5] = 0xff & (n >> 050);
152 dest[6] = 0xff & (n >> 060);
153 dest[7] = 0xff & (n >> 070);
156 static uint64_t clamp_max(uint64_t n, uint64_t max, int *clamped)
158 if (n <= max)
159 return n;
160 *clamped = 1;
161 return max;
164 static void copy_le16_clamp(unsigned char *dest, uint64_t n, int *clamped)
166 copy_le16(dest, clamp_max(n, 0xffff, clamped));
169 static void copy_le32_clamp(unsigned char *dest, uint64_t n, int *clamped)
171 copy_le32(dest, clamp_max(n, 0xffffffff, clamped));
174 static int strbuf_add_le(struct strbuf *sb, size_t size, uintmax_t n)
176 while (size-- > 0) {
177 strbuf_addch(sb, n & 0xff);
178 n >>= 8;
180 return -!!n;
183 static uint32_t clamp32(uintmax_t n)
185 const uintmax_t max = 0xffffffff;
186 return (n < max) ? n : max;
189 static void *zlib_deflate_raw(void *data, unsigned long size,
190 int compression_level,
191 unsigned long *compressed_size)
193 git_zstream stream;
194 unsigned long maxsize;
195 void *buffer;
196 int result;
198 git_deflate_init_raw(&stream, compression_level);
199 maxsize = git_deflate_bound(&stream, size);
200 buffer = xmalloc(maxsize);
202 stream.next_in = data;
203 stream.avail_in = size;
204 stream.next_out = buffer;
205 stream.avail_out = maxsize;
207 do {
208 result = git_deflate(&stream, Z_FINISH);
209 } while (result == Z_OK);
211 if (result != Z_STREAM_END) {
212 free(buffer);
213 return NULL;
216 git_deflate_end(&stream);
217 *compressed_size = stream.total_out;
219 return buffer;
222 static void write_zip_data_desc(unsigned long size,
223 unsigned long compressed_size,
224 unsigned long crc)
226 if (size >= 0xffffffff || compressed_size >= 0xffffffff) {
227 struct zip64_data_desc trailer;
228 copy_le32(trailer.magic, 0x08074b50);
229 copy_le32(trailer.crc32, crc);
230 copy_le64(trailer.compressed_size, compressed_size);
231 copy_le64(trailer.size, size);
232 write_or_die(1, &trailer, ZIP64_DATA_DESC_SIZE);
233 zip_offset += ZIP64_DATA_DESC_SIZE;
234 } else {
235 struct zip_data_desc trailer;
236 copy_le32(trailer.magic, 0x08074b50);
237 copy_le32(trailer.crc32, crc);
238 copy_le32(trailer.compressed_size, compressed_size);
239 copy_le32(trailer.size, size);
240 write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
241 zip_offset += ZIP_DATA_DESC_SIZE;
245 static void set_zip_header_data_desc(struct zip_local_header *header,
246 unsigned long size,
247 unsigned long compressed_size,
248 unsigned long crc)
250 copy_le32(header->crc32, crc);
251 copy_le32(header->compressed_size, compressed_size);
252 copy_le32(header->size, size);
255 static int has_only_ascii(const char *s)
257 for (;;) {
258 int c = *s++;
259 if (c == '\0')
260 return 1;
261 if (!isascii(c))
262 return 0;
266 static int entry_is_binary(const char *path, const void *buffer, size_t size)
268 struct userdiff_driver *driver = userdiff_find_by_path(path);
269 if (!driver)
270 driver = userdiff_find_by_name("default");
271 if (driver->binary != -1)
272 return driver->binary;
273 return buffer_is_binary(buffer, size);
276 #define STREAM_BUFFER_SIZE (1024 * 16)
278 static int write_zip_entry(struct archiver_args *args,
279 const struct object_id *oid,
280 const char *path, size_t pathlen,
281 unsigned int mode)
283 struct zip_local_header header;
284 uintmax_t offset = zip_offset;
285 struct zip_extra_mtime extra;
286 struct zip64_extra extra64;
287 size_t header_extra_size = ZIP_EXTRA_MTIME_SIZE;
288 int need_zip64_extra = 0;
289 unsigned long attr2;
290 unsigned long compressed_size;
291 unsigned long crc;
292 int method;
293 unsigned char *out;
294 void *deflated = NULL;
295 void *buffer;
296 struct git_istream *stream = NULL;
297 unsigned long flags = 0;
298 unsigned long size;
299 int is_binary = -1;
300 const char *path_without_prefix = path + args->baselen;
301 unsigned int creator_version = 0;
302 unsigned int version_needed = 10;
303 size_t zip_dir_extra_size = ZIP_EXTRA_MTIME_SIZE;
304 size_t zip64_dir_extra_payload_size = 0;
306 crc = crc32(0, NULL, 0);
308 if (!has_only_ascii(path)) {
309 if (is_utf8(path))
310 flags |= ZIP_UTF8;
311 else
312 warning("Path is not valid UTF-8: %s", path);
315 if (pathlen > 0xffff) {
316 return error("path too long (%d chars, SHA1: %s): %s",
317 (int)pathlen, oid_to_hex(oid), path);
320 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
321 method = 0;
322 attr2 = 16;
323 out = NULL;
324 size = 0;
325 compressed_size = 0;
326 buffer = NULL;
327 } else if (S_ISREG(mode) || S_ISLNK(mode)) {
328 enum object_type type = oid_object_info(the_repository, oid,
329 &size);
331 method = 0;
332 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
333 (mode & 0111) ? ((mode) << 16) : 0;
334 if (S_ISLNK(mode) || (mode & 0111))
335 creator_version = 0x0317;
336 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
337 method = 8;
339 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
340 size > big_file_threshold) {
341 stream = open_istream(oid, &type, &size, NULL);
342 if (!stream)
343 return error("cannot stream blob %s",
344 oid_to_hex(oid));
345 flags |= ZIP_STREAM;
346 out = buffer = NULL;
347 } else {
348 buffer = object_file_to_archive(args, path, oid, mode,
349 &type, &size);
350 if (!buffer)
351 return error("cannot read %s",
352 oid_to_hex(oid));
353 crc = crc32(crc, buffer, size);
354 is_binary = entry_is_binary(path_without_prefix,
355 buffer, size);
356 out = buffer;
358 compressed_size = (method == 0) ? size : 0;
359 } else {
360 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
361 oid_to_hex(oid));
364 if (creator_version > max_creator_version)
365 max_creator_version = creator_version;
367 if (buffer && method == 8) {
368 out = deflated = zlib_deflate_raw(buffer, size,
369 args->compression_level,
370 &compressed_size);
371 if (!out || compressed_size >= size) {
372 out = buffer;
373 method = 0;
374 compressed_size = size;
378 copy_le16(extra.magic, 0x5455);
379 copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
380 extra.flags[0] = 1; /* just mtime */
381 copy_le32(extra.mtime, args->time);
383 if (size > 0xffffffff || compressed_size > 0xffffffff)
384 need_zip64_extra = 1;
385 if (stream && size > 0x7fffffff)
386 need_zip64_extra = 1;
388 if (need_zip64_extra)
389 version_needed = 45;
391 copy_le32(header.magic, 0x04034b50);
392 copy_le16(header.version, version_needed);
393 copy_le16(header.flags, flags);
394 copy_le16(header.compression_method, method);
395 copy_le16(header.mtime, zip_time);
396 copy_le16(header.mdate, zip_date);
397 if (need_zip64_extra) {
398 set_zip_header_data_desc(&header, 0xffffffff, 0xffffffff, crc);
399 header_extra_size += ZIP64_EXTRA_SIZE;
400 } else {
401 set_zip_header_data_desc(&header, size, compressed_size, crc);
403 copy_le16(header.filename_length, pathlen);
404 copy_le16(header.extra_length, header_extra_size);
405 write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
406 zip_offset += ZIP_LOCAL_HEADER_SIZE;
407 write_or_die(1, path, pathlen);
408 zip_offset += pathlen;
409 write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
410 zip_offset += ZIP_EXTRA_MTIME_SIZE;
411 if (need_zip64_extra) {
412 copy_le16(extra64.magic, 0x0001);
413 copy_le16(extra64.extra_size, ZIP64_EXTRA_PAYLOAD_SIZE);
414 copy_le64(extra64.size, size);
415 copy_le64(extra64.compressed_size, compressed_size);
416 write_or_die(1, &extra64, ZIP64_EXTRA_SIZE);
417 zip_offset += ZIP64_EXTRA_SIZE;
420 if (stream && method == 0) {
421 unsigned char buf[STREAM_BUFFER_SIZE];
422 ssize_t readlen;
424 for (;;) {
425 readlen = read_istream(stream, buf, sizeof(buf));
426 if (readlen <= 0)
427 break;
428 crc = crc32(crc, buf, readlen);
429 if (is_binary == -1)
430 is_binary = entry_is_binary(path_without_prefix,
431 buf, readlen);
432 write_or_die(1, buf, readlen);
434 close_istream(stream);
435 if (readlen)
436 return readlen;
438 compressed_size = size;
439 zip_offset += compressed_size;
441 write_zip_data_desc(size, compressed_size, crc);
442 } else if (stream && method == 8) {
443 unsigned char buf[STREAM_BUFFER_SIZE];
444 ssize_t readlen;
445 git_zstream zstream;
446 int result;
447 size_t out_len;
448 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
450 git_deflate_init_raw(&zstream, args->compression_level);
452 compressed_size = 0;
453 zstream.next_out = compressed;
454 zstream.avail_out = sizeof(compressed);
456 for (;;) {
457 readlen = read_istream(stream, buf, sizeof(buf));
458 if (readlen <= 0)
459 break;
460 crc = crc32(crc, buf, readlen);
461 if (is_binary == -1)
462 is_binary = entry_is_binary(path_without_prefix,
463 buf, readlen);
465 zstream.next_in = buf;
466 zstream.avail_in = readlen;
467 result = git_deflate(&zstream, 0);
468 if (result != Z_OK)
469 die("deflate error (%d)", result);
470 out_len = zstream.next_out - compressed;
472 if (out_len > 0) {
473 write_or_die(1, compressed, out_len);
474 compressed_size += out_len;
475 zstream.next_out = compressed;
476 zstream.avail_out = sizeof(compressed);
480 close_istream(stream);
481 if (readlen)
482 return readlen;
484 zstream.next_in = buf;
485 zstream.avail_in = 0;
486 result = git_deflate(&zstream, Z_FINISH);
487 if (result != Z_STREAM_END)
488 die("deflate error (%d)", result);
490 git_deflate_end(&zstream);
491 out_len = zstream.next_out - compressed;
492 write_or_die(1, compressed, out_len);
493 compressed_size += out_len;
494 zip_offset += compressed_size;
496 write_zip_data_desc(size, compressed_size, crc);
497 } else if (compressed_size > 0) {
498 write_or_die(1, out, compressed_size);
499 zip_offset += compressed_size;
502 free(deflated);
503 free(buffer);
505 if (compressed_size > 0xffffffff || size > 0xffffffff ||
506 offset > 0xffffffff) {
507 if (compressed_size >= 0xffffffff)
508 zip64_dir_extra_payload_size += 8;
509 if (size >= 0xffffffff)
510 zip64_dir_extra_payload_size += 8;
511 if (offset >= 0xffffffff)
512 zip64_dir_extra_payload_size += 8;
513 zip_dir_extra_size += 2 + 2 + zip64_dir_extra_payload_size;
516 strbuf_add_le(&zip_dir, 4, 0x02014b50); /* magic */
517 strbuf_add_le(&zip_dir, 2, creator_version);
518 strbuf_add_le(&zip_dir, 2, version_needed);
519 strbuf_add_le(&zip_dir, 2, flags);
520 strbuf_add_le(&zip_dir, 2, method);
521 strbuf_add_le(&zip_dir, 2, zip_time);
522 strbuf_add_le(&zip_dir, 2, zip_date);
523 strbuf_add_le(&zip_dir, 4, crc);
524 strbuf_add_le(&zip_dir, 4, clamp32(compressed_size));
525 strbuf_add_le(&zip_dir, 4, clamp32(size));
526 strbuf_add_le(&zip_dir, 2, pathlen);
527 strbuf_add_le(&zip_dir, 2, zip_dir_extra_size);
528 strbuf_add_le(&zip_dir, 2, 0); /* comment length */
529 strbuf_add_le(&zip_dir, 2, 0); /* disk */
530 strbuf_add_le(&zip_dir, 2, !is_binary);
531 strbuf_add_le(&zip_dir, 4, attr2);
532 strbuf_add_le(&zip_dir, 4, clamp32(offset));
533 strbuf_add(&zip_dir, path, pathlen);
534 strbuf_add(&zip_dir, &extra, ZIP_EXTRA_MTIME_SIZE);
535 if (zip64_dir_extra_payload_size) {
536 strbuf_add_le(&zip_dir, 2, 0x0001); /* magic */
537 strbuf_add_le(&zip_dir, 2, zip64_dir_extra_payload_size);
538 if (size >= 0xffffffff)
539 strbuf_add_le(&zip_dir, 8, size);
540 if (compressed_size >= 0xffffffff)
541 strbuf_add_le(&zip_dir, 8, compressed_size);
542 if (offset >= 0xffffffff)
543 strbuf_add_le(&zip_dir, 8, offset);
545 zip_dir_entries++;
547 return 0;
550 static void write_zip64_trailer(void)
552 struct zip64_dir_trailer trailer64;
553 struct zip64_dir_trailer_locator locator64;
555 copy_le32(trailer64.magic, 0x06064b50);
556 copy_le64(trailer64.record_size, ZIP64_DIR_TRAILER_RECORD_SIZE);
557 copy_le16(trailer64.creator_version, max_creator_version);
558 copy_le16(trailer64.version, 45);
559 copy_le32(trailer64.disk, 0);
560 copy_le32(trailer64.directory_start_disk, 0);
561 copy_le64(trailer64.entries_on_this_disk, zip_dir_entries);
562 copy_le64(trailer64.entries, zip_dir_entries);
563 copy_le64(trailer64.size, zip_dir.len);
564 copy_le64(trailer64.offset, zip_offset);
566 copy_le32(locator64.magic, 0x07064b50);
567 copy_le32(locator64.disk, 0);
568 copy_le64(locator64.offset, zip_offset + zip_dir.len);
569 copy_le32(locator64.number_of_disks, 1);
571 write_or_die(1, &trailer64, ZIP64_DIR_TRAILER_SIZE);
572 write_or_die(1, &locator64, ZIP64_DIR_TRAILER_LOCATOR_SIZE);
575 static void write_zip_trailer(const unsigned char *sha1)
577 struct zip_dir_trailer trailer;
578 int clamped = 0;
580 copy_le32(trailer.magic, 0x06054b50);
581 copy_le16(trailer.disk, 0);
582 copy_le16(trailer.directory_start_disk, 0);
583 copy_le16_clamp(trailer.entries_on_this_disk, zip_dir_entries,
584 &clamped);
585 copy_le16_clamp(trailer.entries, zip_dir_entries, &clamped);
586 copy_le32(trailer.size, zip_dir.len);
587 copy_le32_clamp(trailer.offset, zip_offset, &clamped);
588 copy_le16(trailer.comment_length, sha1 ? GIT_SHA1_HEXSZ : 0);
590 write_or_die(1, zip_dir.buf, zip_dir.len);
591 if (clamped)
592 write_zip64_trailer();
593 write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
594 if (sha1)
595 write_or_die(1, sha1_to_hex(sha1), GIT_SHA1_HEXSZ);
598 static void dos_time(timestamp_t *timestamp, int *dos_date, int *dos_time)
600 time_t time;
601 struct tm *t;
603 if (date_overflows(*timestamp))
604 die("timestamp too large for this system: %"PRItime,
605 *timestamp);
606 time = (time_t)*timestamp;
607 t = localtime(&time);
608 *timestamp = time;
610 *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
611 (t->tm_year + 1900 - 1980) * 512;
612 *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
615 static int archive_zip_config(const char *var, const char *value, void *data)
617 return userdiff_config(var, value);
620 static int write_zip_archive(const struct archiver *ar,
621 struct archiver_args *args)
623 int err;
625 git_config(archive_zip_config, NULL);
627 dos_time(&args->time, &zip_date, &zip_time);
629 strbuf_init(&zip_dir, 0);
631 err = write_archive_entries(args, write_zip_entry);
632 if (!err)
633 write_zip_trailer(args->commit_sha1);
635 strbuf_release(&zip_dir);
637 return err;
640 static struct archiver zip_archiver = {
641 "zip",
642 write_zip_archive,
643 ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
646 void init_zip_archiver(void)
648 register_archiver(&zip_archiver);