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