Merge branch 'hv/config-from-blob'
[git] / archive-zip.c
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
9 static int zip_date;
10 static int zip_time;
11
12 static unsigned char *zip_dir;
13 static unsigned int zip_dir_size;
14
15 static unsigned int zip_offset;
16 static unsigned int zip_dir_offset;
17 static unsigned int zip_dir_entries;
18
19 #define ZIP_DIRECTORY_MIN_SIZE  (1024 * 1024)
20 #define ZIP_STREAM      (1 <<  3)
21 #define ZIP_UTF8        (1 << 11)
22
23 struct 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];
35         unsigned char _end[1];
36 };
37
38 struct 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
46 struct 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];
64         unsigned char _end[1];
65 };
66
67 struct 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];
76         unsigned char _end[1];
77 };
78
79 struct 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
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)
93 #define ZIP_DATA_DESC_SIZE      offsetof(struct zip_data_desc, _end)
94 #define ZIP_DIR_HEADER_SIZE     offsetof(struct zip_dir_header, _end)
95 #define ZIP_DIR_TRAILER_SIZE    offsetof(struct zip_dir_trailer, _end)
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))
99
100 static void copy_le16(unsigned char *dest, unsigned int n)
101 {
102         dest[0] = 0xff & n;
103         dest[1] = 0xff & (n >> 010);
104 }
105
106 static 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
114 static void *zlib_deflate_raw(void *data, unsigned long size,
115                               int compression_level,
116                               unsigned long *compressed_size)
117 {
118         git_zstream stream;
119         unsigned long maxsize;
120         void *buffer;
121         int result;
122
123         memset(&stream, 0, sizeof(stream));
124         git_deflate_init_raw(&stream, compression_level);
125         maxsize = git_deflate_bound(&stream, size);
126         buffer = xmalloc(maxsize);
127
128         stream.next_in = data;
129         stream.avail_in = size;
130         stream.next_out = buffer;
131         stream.avail_out = maxsize;
132
133         do {
134                 result = git_deflate(&stream, Z_FINISH);
135         } while (result == Z_OK);
136
137         if (result != Z_STREAM_END) {
138                 free(buffer);
139                 return NULL;
140         }
141
142         git_deflate_end(&stream);
143         *compressed_size = stream.total_out;
144
145         return buffer;
146 }
147
148 static void write_zip_data_desc(unsigned long size,
149                                 unsigned long compressed_size,
150                                 unsigned long crc)
151 {
152         struct zip_data_desc trailer;
153
154         copy_le32(trailer.magic, 0x08074b50);
155         copy_le32(trailer.crc32, crc);
156         copy_le32(trailer.compressed_size, compressed_size);
157         copy_le32(trailer.size, size);
158         write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
159 }
160
161 static void set_zip_dir_data_desc(struct zip_dir_header *header,
162                                   unsigned long size,
163                                   unsigned long compressed_size,
164                                   unsigned long crc)
165 {
166         copy_le32(header->crc32, crc);
167         copy_le32(header->compressed_size, compressed_size);
168         copy_le32(header->size, size);
169 }
170
171 static void set_zip_header_data_desc(struct zip_local_header *header,
172                                      unsigned long size,
173                                      unsigned long compressed_size,
174                                      unsigned long crc)
175 {
176         copy_le32(header->crc32, crc);
177         copy_le32(header->compressed_size, compressed_size);
178         copy_le32(header->size, size);
179 }
180
181 static int has_only_ascii(const char *s)
182 {
183         for (;;) {
184                 int c = *s++;
185                 if (c == '\0')
186                         return 1;
187                 if (!isascii(c))
188                         return 0;
189         }
190 }
191
192 #define STREAM_BUFFER_SIZE (1024 * 16)
193
194 static int write_zip_entry(struct archiver_args *args,
195                            const unsigned char *sha1,
196                            const char *path, size_t pathlen,
197                            unsigned int mode)
198 {
199         struct zip_local_header header;
200         struct zip_dir_header dirent;
201         struct zip_extra_mtime extra;
202         unsigned long attr2;
203         unsigned long compressed_size;
204         unsigned long crc;
205         unsigned long direntsize;
206         int method;
207         unsigned char *out;
208         void *deflated = NULL;
209         void *buffer;
210         struct git_istream *stream = NULL;
211         unsigned long flags = 0;
212         unsigned long size;
213
214         crc = crc32(0, NULL, 0);
215
216         if (!has_only_ascii(path)) {
217                 if (is_utf8(path))
218                         flags |= ZIP_UTF8;
219                 else
220                         warning("Path is not valid UTF-8: %s", path);
221         }
222
223         if (pathlen > 0xffff) {
224                 return error("path too long (%d chars, SHA1: %s): %s",
225                                 (int)pathlen, sha1_to_hex(sha1), path);
226         }
227
228         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
229                 method = 0;
230                 attr2 = 16;
231                 out = NULL;
232                 size = 0;
233                 compressed_size = 0;
234                 buffer = NULL;
235         } else if (S_ISREG(mode) || S_ISLNK(mode)) {
236                 enum object_type type = sha1_object_info(sha1, &size);
237
238                 method = 0;
239                 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
240                         (mode & 0111) ? ((mode) << 16) : 0;
241                 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
242                         method = 8;
243
244                 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
245                     size > big_file_threshold) {
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                 }
261                 compressed_size = (method == 0) ? size : 0;
262         } else {
263                 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
264                                 sha1_to_hex(sha1));
265         }
266
267         if (buffer && method == 8) {
268                 out = deflated = zlib_deflate_raw(buffer, size,
269                                                   args->compression_level,
270                                                   &compressed_size);
271                 if (!out || compressed_size >= size) {
272                         out = buffer;
273                         method = 0;
274                         compressed_size = size;
275                 }
276         }
277
278         copy_le16(extra.magic, 0x5455);
279         copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
280         extra.flags[0] = 1;     /* just mtime */
281         copy_le32(extra.mtime, args->time);
282
283         /* make sure we have enough free space in the dictionary */
284         direntsize = ZIP_DIR_HEADER_SIZE + pathlen + ZIP_EXTRA_MTIME_SIZE;
285         while (zip_dir_size < zip_dir_offset + direntsize) {
286                 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
287                 zip_dir = xrealloc(zip_dir, zip_dir_size);
288         }
289
290         copy_le32(dirent.magic, 0x02014b50);
291         copy_le16(dirent.creator_version,
292                 S_ISLNK(mode) || (S_ISREG(mode) && (mode & 0111)) ? 0x0317 : 0);
293         copy_le16(dirent.version, 10);
294         copy_le16(dirent.flags, flags);
295         copy_le16(dirent.compression_method, method);
296         copy_le16(dirent.mtime, zip_time);
297         copy_le16(dirent.mdate, zip_date);
298         set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
299         copy_le16(dirent.filename_length, pathlen);
300         copy_le16(dirent.extra_length, ZIP_EXTRA_MTIME_SIZE);
301         copy_le16(dirent.comment_length, 0);
302         copy_le16(dirent.disk, 0);
303         copy_le16(dirent.attr1, 0);
304         copy_le32(dirent.attr2, attr2);
305         copy_le32(dirent.offset, zip_offset);
306
307         copy_le32(header.magic, 0x04034b50);
308         copy_le16(header.version, 10);
309         copy_le16(header.flags, flags);
310         copy_le16(header.compression_method, method);
311         copy_le16(header.mtime, zip_time);
312         copy_le16(header.mdate, zip_date);
313         set_zip_header_data_desc(&header, size, compressed_size, crc);
314         copy_le16(header.filename_length, pathlen);
315         copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
316         write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
317         zip_offset += ZIP_LOCAL_HEADER_SIZE;
318         write_or_die(1, path, pathlen);
319         zip_offset += pathlen;
320         write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
321         zip_offset += ZIP_EXTRA_MTIME_SIZE;
322         if (stream && method == 0) {
323                 unsigned char buf[STREAM_BUFFER_SIZE];
324                 ssize_t readlen;
325
326                 for (;;) {
327                         readlen = read_istream(stream, buf, sizeof(buf));
328                         if (readlen <= 0)
329                                 break;
330                         crc = crc32(crc, buf, readlen);
331                         write_or_die(1, buf, readlen);
332                 }
333                 close_istream(stream);
334                 if (readlen)
335                         return readlen;
336
337                 compressed_size = size;
338                 zip_offset += compressed_size;
339
340                 write_zip_data_desc(size, compressed_size, crc);
341                 zip_offset += ZIP_DATA_DESC_SIZE;
342
343                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
344         } else if (stream && method == 8) {
345                 unsigned char buf[STREAM_BUFFER_SIZE];
346                 ssize_t readlen;
347                 git_zstream zstream;
348                 int result;
349                 size_t out_len;
350                 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
351
352                 memset(&zstream, 0, sizeof(zstream));
353                 git_deflate_init_raw(&zstream, args->compression_level);
354
355                 compressed_size = 0;
356                 zstream.next_out = compressed;
357                 zstream.avail_out = sizeof(compressed);
358
359                 for (;;) {
360                         readlen = read_istream(stream, buf, sizeof(buf));
361                         if (readlen <= 0)
362                                 break;
363                         crc = crc32(crc, buf, readlen);
364
365                         zstream.next_in = buf;
366                         zstream.avail_in = readlen;
367                         result = git_deflate(&zstream, 0);
368                         if (result != Z_OK)
369                                 die("deflate error (%d)", result);
370                         out_len = zstream.next_out - compressed;
371
372                         if (out_len > 0) {
373                                 write_or_die(1, compressed, out_len);
374                                 compressed_size += out_len;
375                                 zstream.next_out = compressed;
376                                 zstream.avail_out = sizeof(compressed);
377                         }
378
379                 }
380                 close_istream(stream);
381                 if (readlen)
382                         return readlen;
383
384                 zstream.next_in = buf;
385                 zstream.avail_in = 0;
386                 result = git_deflate(&zstream, Z_FINISH);
387                 if (result != Z_STREAM_END)
388                         die("deflate error (%d)", result);
389
390                 git_deflate_end(&zstream);
391                 out_len = zstream.next_out - compressed;
392                 write_or_die(1, compressed, out_len);
393                 compressed_size += out_len;
394                 zip_offset += compressed_size;
395
396                 write_zip_data_desc(size, compressed_size, crc);
397                 zip_offset += ZIP_DATA_DESC_SIZE;
398
399                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
400         } else if (compressed_size > 0) {
401                 write_or_die(1, out, compressed_size);
402                 zip_offset += compressed_size;
403         }
404
405         free(deflated);
406         free(buffer);
407
408         memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
409         zip_dir_offset += ZIP_DIR_HEADER_SIZE;
410         memcpy(zip_dir + zip_dir_offset, path, pathlen);
411         zip_dir_offset += pathlen;
412         memcpy(zip_dir + zip_dir_offset, &extra, ZIP_EXTRA_MTIME_SIZE);
413         zip_dir_offset += ZIP_EXTRA_MTIME_SIZE;
414         zip_dir_entries++;
415
416         return 0;
417 }
418
419 static void write_zip_trailer(const unsigned char *sha1)
420 {
421         struct zip_dir_trailer trailer;
422
423         copy_le32(trailer.magic, 0x06054b50);
424         copy_le16(trailer.disk, 0);
425         copy_le16(trailer.directory_start_disk, 0);
426         copy_le16(trailer.entries_on_this_disk, zip_dir_entries);
427         copy_le16(trailer.entries, zip_dir_entries);
428         copy_le32(trailer.size, zip_dir_offset);
429         copy_le32(trailer.offset, zip_offset);
430         copy_le16(trailer.comment_length, sha1 ? 40 : 0);
431
432         write_or_die(1, zip_dir, zip_dir_offset);
433         write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
434         if (sha1)
435                 write_or_die(1, sha1_to_hex(sha1), 40);
436 }
437
438 static void dos_time(time_t *time, int *dos_date, int *dos_time)
439 {
440         struct tm *t = localtime(time);
441
442         *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
443                     (t->tm_year + 1900 - 1980) * 512;
444         *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
445 }
446
447 static int write_zip_archive(const struct archiver *ar,
448                              struct archiver_args *args)
449 {
450         int err;
451
452         dos_time(&args->time, &zip_date, &zip_time);
453
454         zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
455         zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
456
457         err = write_archive_entries(args, write_zip_entry);
458         if (!err)
459                 write_zip_trailer(args->commit_sha1);
460
461         free(zip_dir);
462
463         return err;
464 }
465
466 static struct archiver zip_archiver = {
467         "zip",
468         write_zip_archive,
469         ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
470 };
471
472 void init_zip_archiver(void)
473 {
474         register_archiver(&zip_archiver);
475 }