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                 size = 0;
 236         } else if (S_ISREG(mode) || S_ISLNK(mode)) {
 237                 enum object_type type = sha1_object_info(sha1, &size);
 238
 239                 method = 0;
 240                 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
 241                         (mode & 0111) ? ((mode) << 16) : 0;
 242                 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
 243                         method = 8;
 244
 245                 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
 246                     size > big_file_threshold) {
 247                         stream = open_istream(sha1, &type, &size, NULL);
 248                         if (!stream)
 249                                 return error("cannot stream blob %s",
 250                                              sha1_to_hex(sha1));
 251                         flags |= ZIP_STREAM;
 252                         out = buffer = NULL;
 253                 } else {
 254                         buffer = sha1_file_to_archive(args, path, sha1, mode,
 255                                                       &type, &size);
 256                         if (!buffer)
 257                                 return error("cannot read %s",
 258                                              sha1_to_hex(sha1));
 259                         crc = crc32(crc, buffer, size);
 260                         out = buffer;
 261                 }
 262                 compressed_size = (method == 0) ? size : 0;
 263         } else {
 264                 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
 265                                 sha1_to_hex(sha1));
 266         }
 267
 268         if (buffer && method == 8) {
 269                 out = deflated = zlib_deflate_raw(buffer, size,
 270                                                   args->compression_level,
 271                                                   &compressed_size);
 272                 if (!out || compressed_size >= size) {
 273                         out = buffer;
 274                         method = 0;
 275                         compressed_size = size;
 276                 }
 277         }
 278
 279         copy_le16(extra.magic, 0x5455);
 280         copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
 281         extra.flags[0] = 1;     /* just mtime */
 282         copy_le32(extra.mtime, args->time);
 283
 284         /* make sure we have enough free space in the dictionary */
 285         direntsize = ZIP_DIR_HEADER_SIZE + pathlen + ZIP_EXTRA_MTIME_SIZE;
 286         while (zip_dir_size < zip_dir_offset + direntsize) {
 287                 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
 288                 zip_dir = xrealloc(zip_dir, zip_dir_size);
 289         }
 290
 291         copy_le32(dirent.magic, 0x02014b50);
 292         copy_le16(dirent.creator_version,
 293                 S_ISLNK(mode) || (S_ISREG(mode) && (mode & 0111)) ? 0x0317 : 0);
 294         copy_le16(dirent.version, 10);
 295         copy_le16(dirent.flags, flags);
 296         copy_le16(dirent.compression_method, method);
 297         copy_le16(dirent.mtime, zip_time);
 298         copy_le16(dirent.mdate, zip_date);
 299         set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 300         copy_le16(dirent.filename_length, pathlen);
 301         copy_le16(dirent.extra_length, ZIP_EXTRA_MTIME_SIZE);
 302         copy_le16(dirent.comment_length, 0);
 303         copy_le16(dirent.disk, 0);
 304         copy_le16(dirent.attr1, 0);
 305         copy_le32(dirent.attr2, attr2);
 306         copy_le32(dirent.offset, zip_offset);
 307
 308         copy_le32(header.magic, 0x04034b50);
 309         copy_le16(header.version, 10);
 310         copy_le16(header.flags, flags);
 311         copy_le16(header.compression_method, method);
 312         copy_le16(header.mtime, zip_time);
 313         copy_le16(header.mdate, zip_date);
 314         set_zip_header_data_desc(&header, size, compressed_size, crc);
 315         copy_le16(header.filename_length, pathlen);
 316         copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
 317         write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
 318         zip_offset += ZIP_LOCAL_HEADER_SIZE;
 319         write_or_die(1, path, pathlen);
 320         zip_offset += pathlen;
 321         write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
 322         zip_offset += ZIP_EXTRA_MTIME_SIZE;
 323         if (stream && method == 0) {
 324                 unsigned char buf[STREAM_BUFFER_SIZE];
 325                 ssize_t readlen;
 326
 327                 for (;;) {
 328                         readlen = read_istream(stream, buf, sizeof(buf));
 329                         if (readlen <= 0)
 330                                 break;
 331                         crc = crc32(crc, buf, readlen);
 332                         write_or_die(1, buf, readlen);
 333                 }
 334                 close_istream(stream);
 335                 if (readlen)
 336                         return readlen;
 337
 338                 compressed_size = size;
 339                 zip_offset += compressed_size;
 340
 341                 write_zip_data_desc(size, compressed_size, crc);
 342                 zip_offset += ZIP_DATA_DESC_SIZE;
 343
 344                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 345         } else if (stream && method == 8) {
 346                 unsigned char buf[STREAM_BUFFER_SIZE];
 347                 ssize_t readlen;
 348                 git_zstream zstream;
 349                 int result;
 350                 size_t out_len;
 351                 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
 352
 353                 memset(&zstream, 0, sizeof(zstream));
 354                 git_deflate_init_raw(&zstream, args->compression_level);
 355
 356                 compressed_size = 0;
 357                 zstream.next_out = compressed;
 358                 zstream.avail_out = sizeof(compressed);
 359
 360                 for (;;) {
 361                         readlen = read_istream(stream, buf, sizeof(buf));
 362                         if (readlen <= 0)
 363                                 break;
 364                         crc = crc32(crc, buf, readlen);
 365
 366                         zstream.next_in = buf;
 367                         zstream.avail_in = readlen;
 368                         result = git_deflate(&zstream, 0);
 369                         if (result != Z_OK)
 370                                 die("deflate error (%d)", result);
 371                         out_len = zstream.next_out - compressed;
 372
 373                         if (out_len > 0) {
 374                                 write_or_die(1, compressed, out_len);
 375                                 compressed_size += out_len;
 376                                 zstream.next_out = compressed;
 377                                 zstream.avail_out = sizeof(compressed);
 378                         }
 379
 380                 }
 381                 close_istream(stream);
 382                 if (readlen)
 383                         return readlen;
 384
 385                 zstream.next_in = buf;
 386                 zstream.avail_in = 0;
 387                 result = git_deflate(&zstream, Z_FINISH);
 388                 if (result != Z_STREAM_END)
 389                         die("deflate error (%d)", result);
 390
 391                 git_deflate_end(&zstream);
 392                 out_len = zstream.next_out - compressed;
 393                 write_or_die(1, compressed, out_len);
 394                 compressed_size += out_len;
 395                 zip_offset += compressed_size;
 396
 397                 write_zip_data_desc(size, compressed_size, crc);
 398                 zip_offset += ZIP_DATA_DESC_SIZE;
 399
 400                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 401         } else if (compressed_size > 0) {
 402                 write_or_die(1, out, compressed_size);
 403                 zip_offset += compressed_size;
 404         }
 405
 406         free(deflated);
 407         free(buffer);
 408
 409         memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
 410         zip_dir_offset += ZIP_DIR_HEADER_SIZE;
 411         memcpy(zip_dir + zip_dir_offset, path, pathlen);
 412         zip_dir_offset += pathlen;
 413         memcpy(zip_dir + zip_dir_offset, &extra, ZIP_EXTRA_MTIME_SIZE);
 414         zip_dir_offset += ZIP_EXTRA_MTIME_SIZE;
 415         zip_dir_entries++;
 416
 417         return 0;
 418 }
 419
 420 static void write_zip_trailer(const unsigned char *sha1)
 421 {
 422         struct zip_dir_trailer trailer;
 423
 424         copy_le32(trailer.magic, 0x06054b50);
 425         copy_le16(trailer.disk, 0);
 426         copy_le16(trailer.directory_start_disk, 0);
 427         copy_le16(trailer.entries_on_this_disk, zip_dir_entries);
 428         copy_le16(trailer.entries, zip_dir_entries);
 429         copy_le32(trailer.size, zip_dir_offset);
 430         copy_le32(trailer.offset, zip_offset);
 431         copy_le16(trailer.comment_length, sha1 ? 40 : 0);
 432
 433         write_or_die(1, zip_dir, zip_dir_offset);
 434         write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
 435         if (sha1)
 436                 write_or_die(1, sha1_to_hex(sha1), 40);
 437 }
 438
 439 static void dos_time(time_t *time, int *dos_date, int *dos_time)
 440 {
 441         struct tm *t = localtime(time);
 442
 443         *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
 444                     (t->tm_year + 1900 - 1980) * 512;
 445         *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
 446 }
 447
 448 static int write_zip_archive(const struct archiver *ar,
 449                              struct archiver_args *args)
 450 {
 451         int err;
 452
 453         dos_time(&args->time, &zip_date, &zip_time);
 454
 455         zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
 456         zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
 457
 458         err = write_archive_entries(args, write_zip_entry);
 459         if (!err)
 460                 write_zip_trailer(args->commit_sha1);
 461
 462         free(zip_dir);
 463
 464         return err;
 465 }
 466
 467 static struct archiver zip_archiver = {
 468         "zip",
 469         write_zip_archive,
 470         ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
 471 };
 472
 473 void init_zip_archiver(void)
 474 {
 475         register_archiver(&zip_archiver);
 476 }