archive-zip: fix compressed size for stored export-subst files
[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"
e4fbbfe9 8
e4fbbfe9
RS
9static int zip_date;
10static int zip_time;
11
12static unsigned char *zip_dir;
13static unsigned int zip_dir_size;
14
15static unsigned int zip_offset;
16static unsigned int zip_dir_offset;
17static unsigned int zip_dir_entries;
18
19#define ZIP_DIRECTORY_MIN_SIZE (1024 * 1024)
88182bab
RS
20#define ZIP_STREAM (1 << 3)
21#define ZIP_UTF8 (1 << 11)
e4fbbfe9
RS
22
23struct zip_local_header {
24 unsigned char magic[4];
25 unsigned char version[2];
26 unsigned char flags[2];
27 unsigned char compression_method[2];
28 unsigned char mtime[2];
29 unsigned char mdate[2];
30 unsigned char crc32[4];
31 unsigned char compressed_size[4];
32 unsigned char size[4];
33 unsigned char filename_length[2];
34 unsigned char extra_length[2];
0ea865ce 35 unsigned char _end[1];
e4fbbfe9
RS
36};
37
2158f883
RS
38struct zip_data_desc {
39 unsigned char magic[4];
40 unsigned char crc32[4];
41 unsigned char compressed_size[4];
42 unsigned char size[4];
43 unsigned char _end[1];
44};
45
e4fbbfe9
RS
46struct zip_dir_header {
47 unsigned char magic[4];
48 unsigned char creator_version[2];
49 unsigned char version[2];
50 unsigned char flags[2];
51 unsigned char compression_method[2];
52 unsigned char mtime[2];
53 unsigned char mdate[2];
54 unsigned char crc32[4];
55 unsigned char compressed_size[4];
56 unsigned char size[4];
57 unsigned char filename_length[2];
58 unsigned char extra_length[2];
59 unsigned char comment_length[2];
60 unsigned char disk[2];
61 unsigned char attr1[2];
62 unsigned char attr2[4];
63 unsigned char offset[4];
0ea865ce 64 unsigned char _end[1];
e4fbbfe9
RS
65};
66
67struct zip_dir_trailer {
68 unsigned char magic[4];
69 unsigned char disk[2];
70 unsigned char directory_start_disk[2];
71 unsigned char entries_on_this_disk[2];
72 unsigned char entries[2];
73 unsigned char size[4];
74 unsigned char offset[4];
75 unsigned char comment_length[2];
0ea865ce 76 unsigned char _end[1];
e4fbbfe9
RS
77};
78
227bf598
RS
79struct zip_extra_mtime {
80 unsigned char magic[2];
81 unsigned char extra_size[2];
82 unsigned char flags[1];
83 unsigned char mtime[4];
84 unsigned char _end[1];
85};
86
0ea865ce
RS
87/*
88 * On ARM, padding is added at the end of the struct, so a simple
89 * sizeof(struct ...) reports two bytes more than the payload size
90 * we're interested in.
91 */
92#define ZIP_LOCAL_HEADER_SIZE offsetof(struct zip_local_header, _end)
2158f883 93#define ZIP_DATA_DESC_SIZE offsetof(struct zip_data_desc, _end)
0ea865ce
RS
94#define ZIP_DIR_HEADER_SIZE offsetof(struct zip_dir_header, _end)
95#define ZIP_DIR_TRAILER_SIZE offsetof(struct zip_dir_trailer, _end)
227bf598
RS
96#define ZIP_EXTRA_MTIME_SIZE offsetof(struct zip_extra_mtime, _end)
97#define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
98 (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
0ea865ce 99
e4fbbfe9
RS
100static void copy_le16(unsigned char *dest, unsigned int n)
101{
102 dest[0] = 0xff & n;
103 dest[1] = 0xff & (n >> 010);
104}
105
106static void copy_le32(unsigned char *dest, unsigned int n)
107{
108 dest[0] = 0xff & n;
109 dest[1] = 0xff & (n >> 010);
110 dest[2] = 0xff & (n >> 020);
111 dest[3] = 0xff & (n >> 030);
112}
113
114static void *zlib_deflate(void *data, unsigned long size,
3a176c6c 115 int compression_level, unsigned long *compressed_size)
e4fbbfe9 116{
ef49a7a0 117 git_zstream stream;
e4fbbfe9
RS
118 unsigned long maxsize;
119 void *buffer;
120 int result;
121
122 memset(&stream, 0, sizeof(stream));
55bb5c91 123 git_deflate_init(&stream, compression_level);
225a6f10 124 maxsize = git_deflate_bound(&stream, size);
e4fbbfe9
RS
125 buffer = xmalloc(maxsize);
126
127 stream.next_in = data;
128 stream.avail_in = size;
129 stream.next_out = buffer;
130 stream.avail_out = maxsize;
131
132 do {
55bb5c91 133 result = git_deflate(&stream, Z_FINISH);
e4fbbfe9
RS
134 } while (result == Z_OK);
135
136 if (result != Z_STREAM_END) {
137 free(buffer);
138 return NULL;
139 }
140
55bb5c91 141 git_deflate_end(&stream);
e4fbbfe9
RS
142 *compressed_size = stream.total_out;
143
144 return buffer;
145}
146
2158f883
RS
147static void write_zip_data_desc(unsigned long size,
148 unsigned long compressed_size,
149 unsigned long crc)
150{
151 struct zip_data_desc trailer;
152
153 copy_le32(trailer.magic, 0x08074b50);
154 copy_le32(trailer.crc32, crc);
155 copy_le32(trailer.compressed_size, compressed_size);
156 copy_le32(trailer.size, size);
157 write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
158}
159
ebf5374a
RS
160static void set_zip_dir_data_desc(struct zip_dir_header *header,
161 unsigned long size,
162 unsigned long compressed_size,
163 unsigned long crc)
164{
165 copy_le32(header->crc32, crc);
166 copy_le32(header->compressed_size, compressed_size);
167 copy_le32(header->size, size);
168}
169
170static void set_zip_header_data_desc(struct zip_local_header *header,
171 unsigned long size,
172 unsigned long compressed_size,
173 unsigned long crc)
174{
175 copy_le32(header->crc32, crc);
176 copy_le32(header->compressed_size, compressed_size);
177 copy_le32(header->size, size);
178}
179
88182bab
RS
180static int has_only_ascii(const char *s)
181{
182 for (;;) {
183 int c = *s++;
184 if (c == '\0')
185 return 1;
186 if (!isascii(c))
187 return 0;
188 }
189}
190
2158f883
RS
191#define STREAM_BUFFER_SIZE (1024 * 16)
192
562e25ab 193static int write_zip_entry(struct archiver_args *args,
9cb513b7
NTND
194 const unsigned char *sha1,
195 const char *path, size_t pathlen,
196 unsigned int mode)
e4fbbfe9
RS
197{
198 struct zip_local_header header;
199 struct zip_dir_header dirent;
227bf598 200 struct zip_extra_mtime extra;
bb52d22e 201 unsigned long attr2;
e4fbbfe9 202 unsigned long compressed_size;
e4fbbfe9
RS
203 unsigned long crc;
204 unsigned long direntsize;
e4fbbfe9 205 int method;
e4fbbfe9 206 unsigned char *out;
e4fbbfe9 207 void *deflated = NULL;
9cb513b7 208 void *buffer;
2158f883
RS
209 struct git_istream *stream = NULL;
210 unsigned long flags = 0;
9cb513b7 211 unsigned long size;
e4fbbfe9 212
38f4d138 213 crc = crc32(0, NULL, 0);
e4fbbfe9 214
88182bab
RS
215 if (!has_only_ascii(path)) {
216 if (is_utf8(path))
217 flags |= ZIP_UTF8;
218 else
219 warning("Path is not valid UTF-8: %s", path);
220 }
221
e4fbbfe9 222 if (pathlen > 0xffff) {
562e25ab
RS
223 return error("path too long (%d chars, SHA1: %s): %s",
224 (int)pathlen, sha1_to_hex(sha1), path);
e4fbbfe9
RS
225 }
226
302b9282 227 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
e4fbbfe9 228 method = 0;
62cdce17 229 attr2 = 16;
e4fbbfe9 230 out = NULL;
60df6bd1 231 size = 0;
e4fbbfe9 232 compressed_size = 0;
9cb513b7
NTND
233 buffer = NULL;
234 size = 0;
62cdce17 235 } else if (S_ISREG(mode) || S_ISLNK(mode)) {
2158f883 236 enum object_type type = sha1_object_info(sha1, &size);
9cb513b7 237
62cdce17 238 method = 0;
bb52d22e
JH
239 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
240 (mode & 0111) ? ((mode) << 16) : 0;
2158f883 241 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
62cdce17 242 method = 8;
2158f883
RS
243
244 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
c743c215 245 size > big_file_threshold) {
2158f883
RS
246 stream = open_istream(sha1, &type, &size, NULL);
247 if (!stream)
248 return error("cannot stream blob %s",
249 sha1_to_hex(sha1));
250 flags |= ZIP_STREAM;
251 out = buffer = NULL;
252 } else {
253 buffer = sha1_file_to_archive(args, path, sha1, mode,
254 &type, &size);
255 if (!buffer)
256 return error("cannot read %s",
257 sha1_to_hex(sha1));
258 crc = crc32(crc, buffer, size);
259 out = buffer;
260 }
d3c1472f 261 compressed_size = (method == 0) ? size : 0;
e4fbbfe9 262 } else {
562e25ab
RS
263 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
264 sha1_to_hex(sha1));
e4fbbfe9
RS
265 }
266
2158f883 267 if (buffer && method == 8) {
3a176c6c
RS
268 deflated = zlib_deflate(buffer, size, args->compression_level,
269 &compressed_size);
e4fbbfe9
RS
270 if (deflated && compressed_size - 6 < size) {
271 /* ZLIB --> raw compressed data (see RFC 1950) */
272 /* CMF and FLG ... */
273 out = (unsigned char *)deflated + 2;
274 compressed_size -= 6; /* ... and ADLER32 */
275 } else {
276 method = 0;
277 compressed_size = size;
278 }
279 }
280
227bf598
RS
281 copy_le16(extra.magic, 0x5455);
282 copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
283 extra.flags[0] = 1; /* just mtime */
284 copy_le32(extra.mtime, args->time);
285
e4fbbfe9 286 /* make sure we have enough free space in the dictionary */
227bf598 287 direntsize = ZIP_DIR_HEADER_SIZE + pathlen + ZIP_EXTRA_MTIME_SIZE;
e4fbbfe9
RS
288 while (zip_dir_size < zip_dir_offset + direntsize) {
289 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
290 zip_dir = xrealloc(zip_dir, zip_dir_size);
291 }
292
293 copy_le32(dirent.magic, 0x02014b50);
bb52d22e
JH
294 copy_le16(dirent.creator_version,
295 S_ISLNK(mode) || (S_ISREG(mode) && (mode & 0111)) ? 0x0317 : 0);
cf72fb07 296 copy_le16(dirent.version, 10);
2158f883 297 copy_le16(dirent.flags, flags);
e4fbbfe9
RS
298 copy_le16(dirent.compression_method, method);
299 copy_le16(dirent.mtime, zip_time);
300 copy_le16(dirent.mdate, zip_date);
ebf5374a 301 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
e4fbbfe9 302 copy_le16(dirent.filename_length, pathlen);
227bf598 303 copy_le16(dirent.extra_length, ZIP_EXTRA_MTIME_SIZE);
e4fbbfe9
RS
304 copy_le16(dirent.comment_length, 0);
305 copy_le16(dirent.disk, 0);
306 copy_le16(dirent.attr1, 0);
62cdce17 307 copy_le32(dirent.attr2, attr2);
e4fbbfe9 308 copy_le32(dirent.offset, zip_offset);
e4fbbfe9
RS
309
310 copy_le32(header.magic, 0x04034b50);
cf72fb07 311 copy_le16(header.version, 10);
2158f883 312 copy_le16(header.flags, flags);
e4fbbfe9
RS
313 copy_le16(header.compression_method, method);
314 copy_le16(header.mtime, zip_time);
315 copy_le16(header.mdate, zip_date);
5ea2c847 316 set_zip_header_data_desc(&header, size, compressed_size, crc);
e4fbbfe9 317 copy_le16(header.filename_length, pathlen);
227bf598 318 copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
0ea865ce
RS
319 write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
320 zip_offset += ZIP_LOCAL_HEADER_SIZE;
e4fbbfe9
RS
321 write_or_die(1, path, pathlen);
322 zip_offset += pathlen;
227bf598
RS
323 write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
324 zip_offset += ZIP_EXTRA_MTIME_SIZE;
2158f883
RS
325 if (stream && method == 0) {
326 unsigned char buf[STREAM_BUFFER_SIZE];
327 ssize_t readlen;
328
329 for (;;) {
330 readlen = read_istream(stream, buf, sizeof(buf));
331 if (readlen <= 0)
332 break;
333 crc = crc32(crc, buf, readlen);
334 write_or_die(1, buf, readlen);
335 }
336 close_istream(stream);
337 if (readlen)
338 return readlen;
339
340 compressed_size = size;
341 zip_offset += compressed_size;
342
343 write_zip_data_desc(size, compressed_size, crc);
344 zip_offset += ZIP_DATA_DESC_SIZE;
345
c743c215
RS
346 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
347 } else if (stream && method == 8) {
348 unsigned char buf[STREAM_BUFFER_SIZE];
349 ssize_t readlen;
350 git_zstream zstream;
351 int result;
352 size_t out_len;
353 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
354
355 memset(&zstream, 0, sizeof(zstream));
356 git_deflate_init(&zstream, args->compression_level);
357
358 compressed_size = 0;
359 zstream.next_out = compressed;
360 zstream.avail_out = sizeof(compressed);
361
362 for (;;) {
363 readlen = read_istream(stream, buf, sizeof(buf));
364 if (readlen <= 0)
365 break;
366 crc = crc32(crc, buf, readlen);
367
368 zstream.next_in = buf;
369 zstream.avail_in = readlen;
370 result = git_deflate(&zstream, 0);
371 if (result != Z_OK)
372 die("deflate error (%d)", result);
373 out = compressed;
374 if (!compressed_size)
375 out += 2;
376 out_len = zstream.next_out - out;
377
378 if (out_len > 0) {
379 write_or_die(1, out, out_len);
380 compressed_size += out_len;
381 zstream.next_out = compressed;
382 zstream.avail_out = sizeof(compressed);
383 }
384
385 }
386 close_istream(stream);
387 if (readlen)
388 return readlen;
389
390 zstream.next_in = buf;
391 zstream.avail_in = 0;
392 result = git_deflate(&zstream, Z_FINISH);
393 if (result != Z_STREAM_END)
394 die("deflate error (%d)", result);
395
396 git_deflate_end(&zstream);
397 out = compressed;
398 if (!compressed_size)
399 out += 2;
400 out_len = zstream.next_out - out - 4;
401 write_or_die(1, out, out_len);
402 compressed_size += out_len;
403 zip_offset += compressed_size;
404
405 write_zip_data_desc(size, compressed_size, crc);
406 zip_offset += ZIP_DATA_DESC_SIZE;
407
2158f883
RS
408 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
409 } else if (compressed_size > 0) {
e4fbbfe9
RS
410 write_or_die(1, out, compressed_size);
411 zip_offset += compressed_size;
412 }
413
e4fbbfe9 414 free(deflated);
9cb513b7 415 free(buffer);
e4fbbfe9 416
ebf5374a
RS
417 memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
418 zip_dir_offset += ZIP_DIR_HEADER_SIZE;
419 memcpy(zip_dir + zip_dir_offset, path, pathlen);
420 zip_dir_offset += pathlen;
227bf598
RS
421 memcpy(zip_dir + zip_dir_offset, &extra, ZIP_EXTRA_MTIME_SIZE);
422 zip_dir_offset += ZIP_EXTRA_MTIME_SIZE;
ebf5374a
RS
423 zip_dir_entries++;
424
562e25ab 425 return 0;
e4fbbfe9
RS
426}
427
428static void write_zip_trailer(const unsigned char *sha1)
429{
430 struct zip_dir_trailer trailer;
431
432 copy_le32(trailer.magic, 0x06054b50);
433 copy_le16(trailer.disk, 0);
434 copy_le16(trailer.directory_start_disk, 0);
435 copy_le16(trailer.entries_on_this_disk, zip_dir_entries);
436 copy_le16(trailer.entries, zip_dir_entries);
437 copy_le32(trailer.size, zip_dir_offset);
438 copy_le32(trailer.offset, zip_offset);
439 copy_le16(trailer.comment_length, sha1 ? 40 : 0);
440
441 write_or_die(1, zip_dir, zip_dir_offset);
0ea865ce 442 write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
e4fbbfe9
RS
443 if (sha1)
444 write_or_die(1, sha1_to_hex(sha1), 40);
445}
446
447static void dos_time(time_t *time, int *dos_date, int *dos_time)
448{
449 struct tm *t = localtime(time);
450
451 *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
452 (t->tm_year + 1900 - 1980) * 512;
453 *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
454}
455
4d7c9898
JK
456static int write_zip_archive(const struct archiver *ar,
457 struct archiver_args *args)
ec06bff5 458{
562e25ab
RS
459 int err;
460
ec06bff5
FBH
461 dos_time(&args->time, &zip_date, &zip_time);
462
463 zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
464 zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
562e25ab
RS
465
466 err = write_archive_entries(args, write_zip_entry);
467 if (!err)
468 write_zip_trailer(args->commit_sha1);
ec06bff5
FBH
469
470 free(zip_dir);
471
562e25ab 472 return err;
ec06bff5 473}
13e0f88d
JK
474
475static struct archiver zip_archiver = {
476 "zip",
477 write_zip_archive,
7b97730b 478 ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
13e0f88d
JK
479};
480
481void init_zip_archiver(void)
482{
483 register_archiver(&zip_archiver);
484}