convert: convert get_cached_convert_stats_ascii to take an index
[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 #include "userdiff.h"
9 #include "xdiff-interface.h"
10
11 static int zip_date;
12 static int zip_time;
13
14 /* We only care about the "buf" part here. */
15 static struct strbuf zip_dir;
16
17 static uintmax_t zip_offset;
18 static uint64_t zip_dir_entries;
19
20 static unsigned int max_creator_version;
21
22 #define ZIP_STREAM      (1 <<  3)
23 #define ZIP_UTF8        (1 << 11)
24
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];
38 };
39
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];
46 };
47
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];
54 };
55
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];
66 };
67
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];
74 };
75
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];
82 };
83
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];
96 };
97
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];
104 };
105
106 /*
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.
110  */
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)
128
129 static void copy_le16(unsigned char *dest, unsigned int n)
130 {
131         dest[0] = 0xff & n;
132         dest[1] = 0xff & (n >> 010);
133 }
134
135 static void copy_le32(unsigned char *dest, unsigned int n)
136 {
137         dest[0] = 0xff & n;
138         dest[1] = 0xff & (n >> 010);
139         dest[2] = 0xff & (n >> 020);
140         dest[3] = 0xff & (n >> 030);
141 }
142
143 static void copy_le64(unsigned char *dest, uint64_t n)
144 {
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);
153 }
154
155 static uint64_t clamp_max(uint64_t n, uint64_t max, int *clamped)
156 {
157         if (n <= max)
158                 return n;
159         *clamped = 1;
160         return max;
161 }
162
163 static void copy_le16_clamp(unsigned char *dest, uint64_t n, int *clamped)
164 {
165         copy_le16(dest, clamp_max(n, 0xffff, clamped));
166 }
167
168 static void copy_le32_clamp(unsigned char *dest, uint64_t n, int *clamped)
169 {
170         copy_le32(dest, clamp_max(n, 0xffffffff, clamped));
171 }
172
173 static int strbuf_add_le(struct strbuf *sb, size_t size, uintmax_t n)
174 {
175         while (size-- > 0) {
176                 strbuf_addch(sb, n & 0xff);
177                 n >>= 8;
178         }
179         return -!!n;
180 }
181
182 static uint32_t clamp32(uintmax_t n)
183 {
184         const uintmax_t max = 0xffffffff;
185         return (n < max) ? n : max;
186 }
187
188 static void *zlib_deflate_raw(void *data, unsigned long size,
189                               int compression_level,
190                               unsigned long *compressed_size)
191 {
192         git_zstream stream;
193         unsigned long maxsize;
194         void *buffer;
195         int result;
196
197         git_deflate_init_raw(&stream, compression_level);
198         maxsize = git_deflate_bound(&stream, size);
199         buffer = xmalloc(maxsize);
200
201         stream.next_in = data;
202         stream.avail_in = size;
203         stream.next_out = buffer;
204         stream.avail_out = maxsize;
205
206         do {
207                 result = git_deflate(&stream, Z_FINISH);
208         } while (result == Z_OK);
209
210         if (result != Z_STREAM_END) {
211                 free(buffer);
212                 return NULL;
213         }
214
215         git_deflate_end(&stream);
216         *compressed_size = stream.total_out;
217
218         return buffer;
219 }
220
221 static void write_zip_data_desc(unsigned long size,
222                                 unsigned long compressed_size,
223                                 unsigned long crc)
224 {
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;
241         }
242 }
243
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)
248 {
249         copy_le32(header->crc32, crc);
250         copy_le32(header->compressed_size, compressed_size);
251         copy_le32(header->size, size);
252 }
253
254 static int has_only_ascii(const char *s)
255 {
256         for (;;) {
257                 int c = *s++;
258                 if (c == '\0')
259                         return 1;
260                 if (!isascii(c))
261                         return 0;
262         }
263 }
264
265 static int entry_is_binary(const char *path, const void *buffer, size_t size)
266 {
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);
273 }
274
275 #define STREAM_BUFFER_SIZE (1024 * 16)
276
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)
281 {
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         unsigned int version_needed = 10;
302         size_t zip_dir_extra_size = ZIP_EXTRA_MTIME_SIZE;
303         size_t zip64_dir_extra_payload_size = 0;
304
305         crc = crc32(0, NULL, 0);
306
307         if (!has_only_ascii(path)) {
308                 if (is_utf8(path))
309                         flags |= ZIP_UTF8;
310                 else
311                         warning("Path is not valid UTF-8: %s", path);
312         }
313
314         if (pathlen > 0xffff) {
315                 return error("path too long (%d chars, SHA1: %s): %s",
316                                 (int)pathlen, sha1_to_hex(sha1), path);
317         }
318
319         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
320                 method = 0;
321                 attr2 = 16;
322                 out = NULL;
323                 size = 0;
324                 compressed_size = 0;
325                 buffer = NULL;
326         } else if (S_ISREG(mode) || S_ISLNK(mode)) {
327                 enum object_type type = sha1_object_info(sha1, &size);
328
329                 method = 0;
330                 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
331                         (mode & 0111) ? ((mode) << 16) : 0;
332                 if (S_ISLNK(mode) || (mode & 0111))
333                         creator_version = 0x0317;
334                 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
335                         method = 8;
336
337                 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
338                     size > big_file_threshold) {
339                         stream = open_istream(sha1, &type, &size, NULL);
340                         if (!stream)
341                                 return error("cannot stream blob %s",
342                                              sha1_to_hex(sha1));
343                         flags |= ZIP_STREAM;
344                         out = buffer = NULL;
345                 } else {
346                         buffer = sha1_file_to_archive(args, path, sha1, mode,
347                                                       &type, &size);
348                         if (!buffer)
349                                 return error("cannot read %s",
350                                              sha1_to_hex(sha1));
351                         crc = crc32(crc, buffer, size);
352                         is_binary = entry_is_binary(path_without_prefix,
353                                                     buffer, size);
354                         out = buffer;
355                 }
356                 compressed_size = (method == 0) ? size : 0;
357         } else {
358                 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
359                                 sha1_to_hex(sha1));
360         }
361
362         if (creator_version > max_creator_version)
363                 max_creator_version = creator_version;
364
365         if (buffer && method == 8) {
366                 out = deflated = zlib_deflate_raw(buffer, size,
367                                                   args->compression_level,
368                                                   &compressed_size);
369                 if (!out || compressed_size >= size) {
370                         out = buffer;
371                         method = 0;
372                         compressed_size = size;
373                 }
374         }
375
376         copy_le16(extra.magic, 0x5455);
377         copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
378         extra.flags[0] = 1;     /* just mtime */
379         copy_le32(extra.mtime, args->time);
380
381         if (size > 0xffffffff || compressed_size > 0xffffffff)
382                 need_zip64_extra = 1;
383         if (stream && size > 0x7fffffff)
384                 need_zip64_extra = 1;
385
386         if (need_zip64_extra)
387                 version_needed = 45;
388
389         copy_le32(header.magic, 0x04034b50);
390         copy_le16(header.version, version_needed);
391         copy_le16(header.flags, flags);
392         copy_le16(header.compression_method, method);
393         copy_le16(header.mtime, zip_time);
394         copy_le16(header.mdate, zip_date);
395         if (need_zip64_extra) {
396                 set_zip_header_data_desc(&header, 0xffffffff, 0xffffffff, crc);
397                 header_extra_size += ZIP64_EXTRA_SIZE;
398         } else {
399                 set_zip_header_data_desc(&header, size, compressed_size, crc);
400         }
401         copy_le16(header.filename_length, pathlen);
402         copy_le16(header.extra_length, header_extra_size);
403         write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
404         zip_offset += ZIP_LOCAL_HEADER_SIZE;
405         write_or_die(1, path, pathlen);
406         zip_offset += pathlen;
407         write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
408         zip_offset += ZIP_EXTRA_MTIME_SIZE;
409         if (need_zip64_extra) {
410                 copy_le16(extra64.magic, 0x0001);
411                 copy_le16(extra64.extra_size, ZIP64_EXTRA_PAYLOAD_SIZE);
412                 copy_le64(extra64.size, size);
413                 copy_le64(extra64.compressed_size, compressed_size);
414                 write_or_die(1, &extra64, ZIP64_EXTRA_SIZE);
415                 zip_offset += ZIP64_EXTRA_SIZE;
416         }
417
418         if (stream && method == 0) {
419                 unsigned char buf[STREAM_BUFFER_SIZE];
420                 ssize_t readlen;
421
422                 for (;;) {
423                         readlen = read_istream(stream, buf, sizeof(buf));
424                         if (readlen <= 0)
425                                 break;
426                         crc = crc32(crc, buf, readlen);
427                         if (is_binary == -1)
428                                 is_binary = entry_is_binary(path_without_prefix,
429                                                             buf, readlen);
430                         write_or_die(1, buf, readlen);
431                 }
432                 close_istream(stream);
433                 if (readlen)
434                         return readlen;
435
436                 compressed_size = size;
437                 zip_offset += compressed_size;
438
439                 write_zip_data_desc(size, compressed_size, crc);
440         } else if (stream && method == 8) {
441                 unsigned char buf[STREAM_BUFFER_SIZE];
442                 ssize_t readlen;
443                 git_zstream zstream;
444                 int result;
445                 size_t out_len;
446                 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
447
448                 git_deflate_init_raw(&zstream, args->compression_level);
449
450                 compressed_size = 0;
451                 zstream.next_out = compressed;
452                 zstream.avail_out = sizeof(compressed);
453
454                 for (;;) {
455                         readlen = read_istream(stream, buf, sizeof(buf));
456                         if (readlen <= 0)
457                                 break;
458                         crc = crc32(crc, buf, readlen);
459                         if (is_binary == -1)
460                                 is_binary = entry_is_binary(path_without_prefix,
461                                                             buf, readlen);
462
463                         zstream.next_in = buf;
464                         zstream.avail_in = readlen;
465                         result = git_deflate(&zstream, 0);
466                         if (result != Z_OK)
467                                 die("deflate error (%d)", result);
468                         out_len = zstream.next_out - compressed;
469
470                         if (out_len > 0) {
471                                 write_or_die(1, compressed, out_len);
472                                 compressed_size += out_len;
473                                 zstream.next_out = compressed;
474                                 zstream.avail_out = sizeof(compressed);
475                         }
476
477                 }
478                 close_istream(stream);
479                 if (readlen)
480                         return readlen;
481
482                 zstream.next_in = buf;
483                 zstream.avail_in = 0;
484                 result = git_deflate(&zstream, Z_FINISH);
485                 if (result != Z_STREAM_END)
486                         die("deflate error (%d)", result);
487
488                 git_deflate_end(&zstream);
489                 out_len = zstream.next_out - compressed;
490                 write_or_die(1, compressed, out_len);
491                 compressed_size += out_len;
492                 zip_offset += compressed_size;
493
494                 write_zip_data_desc(size, compressed_size, crc);
495         } else if (compressed_size > 0) {
496                 write_or_die(1, out, compressed_size);
497                 zip_offset += compressed_size;
498         }
499
500         free(deflated);
501         free(buffer);
502
503         if (compressed_size > 0xffffffff || size > 0xffffffff ||
504             offset > 0xffffffff) {
505                 if (compressed_size >= 0xffffffff)
506                         zip64_dir_extra_payload_size += 8;
507                 if (size >= 0xffffffff)
508                         zip64_dir_extra_payload_size += 8;
509                 if (offset >= 0xffffffff)
510                         zip64_dir_extra_payload_size += 8;
511                 zip_dir_extra_size += 2 + 2 + zip64_dir_extra_payload_size;
512         }
513
514         strbuf_add_le(&zip_dir, 4, 0x02014b50); /* magic */
515         strbuf_add_le(&zip_dir, 2, creator_version);
516         strbuf_add_le(&zip_dir, 2, version_needed);
517         strbuf_add_le(&zip_dir, 2, flags);
518         strbuf_add_le(&zip_dir, 2, method);
519         strbuf_add_le(&zip_dir, 2, zip_time);
520         strbuf_add_le(&zip_dir, 2, zip_date);
521         strbuf_add_le(&zip_dir, 4, crc);
522         strbuf_add_le(&zip_dir, 4, clamp32(compressed_size));
523         strbuf_add_le(&zip_dir, 4, clamp32(size));
524         strbuf_add_le(&zip_dir, 2, pathlen);
525         strbuf_add_le(&zip_dir, 2, zip_dir_extra_size);
526         strbuf_add_le(&zip_dir, 2, 0);          /* comment length */
527         strbuf_add_le(&zip_dir, 2, 0);          /* disk */
528         strbuf_add_le(&zip_dir, 2, !is_binary);
529         strbuf_add_le(&zip_dir, 4, attr2);
530         strbuf_add_le(&zip_dir, 4, clamp32(offset));
531         strbuf_add(&zip_dir, path, pathlen);
532         strbuf_add(&zip_dir, &extra, ZIP_EXTRA_MTIME_SIZE);
533         if (zip64_dir_extra_payload_size) {
534                 strbuf_add_le(&zip_dir, 2, 0x0001);     /* magic */
535                 strbuf_add_le(&zip_dir, 2, zip64_dir_extra_payload_size);
536                 if (size >= 0xffffffff)
537                         strbuf_add_le(&zip_dir, 8, size);
538                 if (compressed_size >= 0xffffffff)
539                         strbuf_add_le(&zip_dir, 8, compressed_size);
540                 if (offset >= 0xffffffff)
541                         strbuf_add_le(&zip_dir, 8, offset);
542         }
543         zip_dir_entries++;
544
545         return 0;
546 }
547
548 static void write_zip64_trailer(void)
549 {
550         struct zip64_dir_trailer trailer64;
551         struct zip64_dir_trailer_locator locator64;
552
553         copy_le32(trailer64.magic, 0x06064b50);
554         copy_le64(trailer64.record_size, ZIP64_DIR_TRAILER_RECORD_SIZE);
555         copy_le16(trailer64.creator_version, max_creator_version);
556         copy_le16(trailer64.version, 45);
557         copy_le32(trailer64.disk, 0);
558         copy_le32(trailer64.directory_start_disk, 0);
559         copy_le64(trailer64.entries_on_this_disk, zip_dir_entries);
560         copy_le64(trailer64.entries, zip_dir_entries);
561         copy_le64(trailer64.size, zip_dir.len);
562         copy_le64(trailer64.offset, zip_offset);
563
564         copy_le32(locator64.magic, 0x07064b50);
565         copy_le32(locator64.disk, 0);
566         copy_le64(locator64.offset, zip_offset + zip_dir.len);
567         copy_le32(locator64.number_of_disks, 1);
568
569         write_or_die(1, &trailer64, ZIP64_DIR_TRAILER_SIZE);
570         write_or_die(1, &locator64, ZIP64_DIR_TRAILER_LOCATOR_SIZE);
571 }
572
573 static void write_zip_trailer(const unsigned char *sha1)
574 {
575         struct zip_dir_trailer trailer;
576         int clamped = 0;
577
578         copy_le32(trailer.magic, 0x06054b50);
579         copy_le16(trailer.disk, 0);
580         copy_le16(trailer.directory_start_disk, 0);
581         copy_le16_clamp(trailer.entries_on_this_disk, zip_dir_entries,
582                         &clamped);
583         copy_le16_clamp(trailer.entries, zip_dir_entries, &clamped);
584         copy_le32(trailer.size, zip_dir.len);
585         copy_le32_clamp(trailer.offset, zip_offset, &clamped);
586         copy_le16(trailer.comment_length, sha1 ? GIT_SHA1_HEXSZ : 0);
587
588         write_or_die(1, zip_dir.buf, zip_dir.len);
589         if (clamped)
590                 write_zip64_trailer();
591         write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
592         if (sha1)
593                 write_or_die(1, sha1_to_hex(sha1), GIT_SHA1_HEXSZ);
594 }
595
596 static void dos_time(timestamp_t *timestamp, int *dos_date, int *dos_time)
597 {
598         time_t time;
599         struct tm *t;
600
601         if (date_overflows(*timestamp))
602                 die("timestamp too large for this system: %"PRItime,
603                     *timestamp);
604         time = (time_t)*timestamp;
605         t = localtime(&time);
606         *timestamp = time;
607
608         *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
609                     (t->tm_year + 1900 - 1980) * 512;
610         *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
611 }
612
613 static int archive_zip_config(const char *var, const char *value, void *data)
614 {
615         return userdiff_config(var, value);
616 }
617
618 static int write_zip_archive(const struct archiver *ar,
619                              struct archiver_args *args)
620 {
621         int err;
622
623         git_config(archive_zip_config, NULL);
624
625         dos_time(&args->time, &zip_date, &zip_time);
626
627         strbuf_init(&zip_dir, 0);
628
629         err = write_archive_entries(args, write_zip_entry);
630         if (!err)
631                 write_zip_trailer(args->commit_sha1);
632
633         strbuf_release(&zip_dir);
634
635         return err;
636 }
637
638 static struct archiver zip_archiver = {
639         "zip",
640         write_zip_archive,
641         ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
642 };
643
644 void init_zip_archiver(void)
645 {
646         register_archiver(&zip_archiver);
647 }