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 static unsigned char *zip_dir;
  15 static unsigned int zip_dir_size;
  16
  17 static unsigned int zip_offset;
  18 static unsigned int zip_dir_offset;
  19 static unsigned int zip_dir_entries;
  20
  21 #define ZIP_DIRECTORY_MIN_SIZE  (1024 * 1024)
  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 zip_dir_header {
  49         unsigned char magic[4];
  50         unsigned char creator_version[2];
  51         unsigned char version[2];
  52         unsigned char flags[2];
  53         unsigned char compression_method[2];
  54         unsigned char mtime[2];
  55         unsigned char mdate[2];
  56         unsigned char crc32[4];
  57         unsigned char compressed_size[4];
  58         unsigned char size[4];
  59         unsigned char filename_length[2];
  60         unsigned char extra_length[2];
  61         unsigned char comment_length[2];
  62         unsigned char disk[2];
  63         unsigned char attr1[2];
  64         unsigned char attr2[4];
  65         unsigned char offset[4];
  66         unsigned char _end[1];
  67 };
  68
  69 struct zip_dir_trailer {
  70         unsigned char magic[4];
  71         unsigned char disk[2];
  72         unsigned char directory_start_disk[2];
  73         unsigned char entries_on_this_disk[2];
  74         unsigned char entries[2];
  75         unsigned char size[4];
  76         unsigned char offset[4];
  77         unsigned char comment_length[2];
  78         unsigned char _end[1];
  79 };
  80
  81 struct zip_extra_mtime {
  82         unsigned char magic[2];
  83         unsigned char extra_size[2];
  84         unsigned char flags[1];
  85         unsigned char mtime[4];
  86         unsigned char _end[1];
  87 };
  88
  89 /*
  90  * On ARM, padding is added at the end of the struct, so a simple
  91  * sizeof(struct ...) reports two bytes more than the payload size
  92  * we're interested in.
  93  */
  94 #define ZIP_LOCAL_HEADER_SIZE   offsetof(struct zip_local_header, _end)
  95 #define ZIP_DATA_DESC_SIZE      offsetof(struct zip_data_desc, _end)
  96 #define ZIP_DIR_HEADER_SIZE     offsetof(struct zip_dir_header, _end)
  97 #define ZIP_DIR_TRAILER_SIZE    offsetof(struct zip_dir_trailer, _end)
  98 #define ZIP_EXTRA_MTIME_SIZE    offsetof(struct zip_extra_mtime, _end)
  99 #define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
 100         (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
 101
 102 static void copy_le16(unsigned char *dest, unsigned int n)
 103 {
 104         dest[0] = 0xff & n;
 105         dest[1] = 0xff & (n >> 010);
 106 }
 107
 108 static void copy_le32(unsigned char *dest, unsigned int n)
 109 {
 110         dest[0] = 0xff & n;
 111         dest[1] = 0xff & (n >> 010);
 112         dest[2] = 0xff & (n >> 020);
 113         dest[3] = 0xff & (n >> 030);
 114 }
 115
 116 static void *zlib_deflate_raw(void *data, unsigned long size,
 117                               int compression_level,
 118                               unsigned long *compressed_size)
 119 {
 120         git_zstream stream;
 121         unsigned long maxsize;
 122         void *buffer;
 123         int result;
 124
 125         git_deflate_init_raw(&stream, compression_level);
 126         maxsize = git_deflate_bound(&stream, size);
 127         buffer = xmalloc(maxsize);
 128
 129         stream.next_in = data;
 130         stream.avail_in = size;
 131         stream.next_out = buffer;
 132         stream.avail_out = maxsize;
 133
 134         do {
 135                 result = git_deflate(&stream, Z_FINISH);
 136         } while (result == Z_OK);
 137
 138         if (result != Z_STREAM_END) {
 139                 free(buffer);
 140                 return NULL;
 141         }
 142
 143         git_deflate_end(&stream);
 144         *compressed_size = stream.total_out;
 145
 146         return buffer;
 147 }
 148
 149 static void write_zip_data_desc(unsigned long size,
 150                                 unsigned long compressed_size,
 151                                 unsigned long crc)
 152 {
 153         struct zip_data_desc trailer;
 154
 155         copy_le32(trailer.magic, 0x08074b50);
 156         copy_le32(trailer.crc32, crc);
 157         copy_le32(trailer.compressed_size, compressed_size);
 158         copy_le32(trailer.size, size);
 159         write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
 160 }
 161
 162 static void set_zip_dir_data_desc(struct zip_dir_header *header,
 163                                   unsigned long size,
 164                                   unsigned long compressed_size,
 165                                   unsigned long crc)
 166 {
 167         copy_le32(header->crc32, crc);
 168         copy_le32(header->compressed_size, compressed_size);
 169         copy_le32(header->size, size);
 170 }
 171
 172 static void set_zip_header_data_desc(struct zip_local_header *header,
 173                                      unsigned long size,
 174                                      unsigned long compressed_size,
 175                                      unsigned long crc)
 176 {
 177         copy_le32(header->crc32, crc);
 178         copy_le32(header->compressed_size, compressed_size);
 179         copy_le32(header->size, size);
 180 }
 181
 182 static int has_only_ascii(const char *s)
 183 {
 184         for (;;) {
 185                 int c = *s++;
 186                 if (c == '\0')
 187                         return 1;
 188                 if (!isascii(c))
 189                         return 0;
 190         }
 191 }
 192
 193 static int entry_is_binary(const char *path, const void *buffer, size_t size)
 194 {
 195         struct userdiff_driver *driver = userdiff_find_by_path(path);
 196         if (!driver)
 197                 driver = userdiff_find_by_name("default");
 198         if (driver->binary != -1)
 199                 return driver->binary;
 200         return buffer_is_binary(buffer, size);
 201 }
 202
 203 #define STREAM_BUFFER_SIZE (1024 * 16)
 204
 205 static int write_zip_entry(struct archiver_args *args,
 206                            const unsigned char *sha1,
 207                            const char *path, size_t pathlen,
 208                            unsigned int mode)
 209 {
 210         struct zip_local_header header;
 211         struct zip_dir_header dirent;
 212         struct zip_extra_mtime extra;
 213         unsigned long attr2;
 214         unsigned long compressed_size;
 215         unsigned long crc;
 216         unsigned long direntsize;
 217         int method;
 218         unsigned char *out;
 219         void *deflated = NULL;
 220         void *buffer;
 221         struct git_istream *stream = NULL;
 222         unsigned long flags = 0;
 223         unsigned long size;
 224         int is_binary = -1;
 225         const char *path_without_prefix = path + args->baselen;
 226
 227         crc = crc32(0, NULL, 0);
 228
 229         if (!has_only_ascii(path)) {
 230                 if (is_utf8(path))
 231                         flags |= ZIP_UTF8;
 232                 else
 233                         warning("Path is not valid UTF-8: %s", path);
 234         }
 235
 236         if (pathlen > 0xffff) {
 237                 return error("path too long (%d chars, SHA1: %s): %s",
 238                                 (int)pathlen, sha1_to_hex(sha1), path);
 239         }
 240
 241         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 242                 method = 0;
 243                 attr2 = 16;
 244                 out = NULL;
 245                 size = 0;
 246                 compressed_size = 0;
 247                 buffer = NULL;
 248         } else if (S_ISREG(mode) || S_ISLNK(mode)) {
 249                 enum object_type type = sha1_object_info(sha1, &size);
 250
 251                 method = 0;
 252                 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
 253                         (mode & 0111) ? ((mode) << 16) : 0;
 254                 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
 255                         method = 8;
 256
 257                 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
 258                     size > big_file_threshold) {
 259                         stream = open_istream(sha1, &type, &size, NULL);
 260                         if (!stream)
 261                                 return error("cannot stream blob %s",
 262                                              sha1_to_hex(sha1));
 263                         flags |= ZIP_STREAM;
 264                         out = buffer = NULL;
 265                 } else {
 266                         buffer = sha1_file_to_archive(args, path, sha1, mode,
 267                                                       &type, &size);
 268                         if (!buffer)
 269                                 return error("cannot read %s",
 270                                              sha1_to_hex(sha1));
 271                         crc = crc32(crc, buffer, size);
 272                         is_binary = entry_is_binary(path_without_prefix,
 273                                                     buffer, size);
 274                         out = buffer;
 275                 }
 276                 compressed_size = (method == 0) ? size : 0;
 277         } else {
 278                 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
 279                                 sha1_to_hex(sha1));
 280         }
 281
 282         if (buffer && method == 8) {
 283                 out = deflated = zlib_deflate_raw(buffer, size,
 284                                                   args->compression_level,
 285                                                   &compressed_size);
 286                 if (!out || compressed_size >= size) {
 287                         out = buffer;
 288                         method = 0;
 289                         compressed_size = size;
 290                 }
 291         }
 292
 293         copy_le16(extra.magic, 0x5455);
 294         copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
 295         extra.flags[0] = 1;     /* just mtime */
 296         copy_le32(extra.mtime, args->time);
 297
 298         /* make sure we have enough free space in the dictionary */
 299         direntsize = ZIP_DIR_HEADER_SIZE + pathlen + ZIP_EXTRA_MTIME_SIZE;
 300         while (zip_dir_size < zip_dir_offset + direntsize) {
 301                 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
 302                 zip_dir = xrealloc(zip_dir, zip_dir_size);
 303         }
 304
 305         copy_le32(dirent.magic, 0x02014b50);
 306         copy_le16(dirent.creator_version,
 307                 S_ISLNK(mode) || (S_ISREG(mode) && (mode & 0111)) ? 0x0317 : 0);
 308         copy_le16(dirent.version, 10);
 309         copy_le16(dirent.flags, flags);
 310         copy_le16(dirent.compression_method, method);
 311         copy_le16(dirent.mtime, zip_time);
 312         copy_le16(dirent.mdate, zip_date);
 313         set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 314         copy_le16(dirent.filename_length, pathlen);
 315         copy_le16(dirent.extra_length, ZIP_EXTRA_MTIME_SIZE);
 316         copy_le16(dirent.comment_length, 0);
 317         copy_le16(dirent.disk, 0);
 318         copy_le32(dirent.attr2, attr2);
 319         copy_le32(dirent.offset, zip_offset);
 320
 321         copy_le32(header.magic, 0x04034b50);
 322         copy_le16(header.version, 10);
 323         copy_le16(header.flags, flags);
 324         copy_le16(header.compression_method, method);
 325         copy_le16(header.mtime, zip_time);
 326         copy_le16(header.mdate, zip_date);
 327         set_zip_header_data_desc(&header, size, compressed_size, crc);
 328         copy_le16(header.filename_length, pathlen);
 329         copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
 330         write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
 331         zip_offset += ZIP_LOCAL_HEADER_SIZE;
 332         write_or_die(1, path, pathlen);
 333         zip_offset += pathlen;
 334         write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
 335         zip_offset += ZIP_EXTRA_MTIME_SIZE;
 336         if (stream && method == 0) {
 337                 unsigned char buf[STREAM_BUFFER_SIZE];
 338                 ssize_t readlen;
 339
 340                 for (;;) {
 341                         readlen = read_istream(stream, buf, sizeof(buf));
 342                         if (readlen <= 0)
 343                                 break;
 344                         crc = crc32(crc, buf, readlen);
 345                         if (is_binary == -1)
 346                                 is_binary = entry_is_binary(path_without_prefix,
 347                                                             buf, readlen);
 348                         write_or_die(1, buf, readlen);
 349                 }
 350                 close_istream(stream);
 351                 if (readlen)
 352                         return readlen;
 353
 354                 compressed_size = size;
 355                 zip_offset += compressed_size;
 356
 357                 write_zip_data_desc(size, compressed_size, crc);
 358                 zip_offset += ZIP_DATA_DESC_SIZE;
 359
 360                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 361         } else if (stream && method == 8) {
 362                 unsigned char buf[STREAM_BUFFER_SIZE];
 363                 ssize_t readlen;
 364                 git_zstream zstream;
 365                 int result;
 366                 size_t out_len;
 367                 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
 368
 369                 git_deflate_init_raw(&zstream, args->compression_level);
 370
 371                 compressed_size = 0;
 372                 zstream.next_out = compressed;
 373                 zstream.avail_out = sizeof(compressed);
 374
 375                 for (;;) {
 376                         readlen = read_istream(stream, buf, sizeof(buf));
 377                         if (readlen <= 0)
 378                                 break;
 379                         crc = crc32(crc, buf, readlen);
 380                         if (is_binary == -1)
 381                                 is_binary = entry_is_binary(path_without_prefix,
 382                                                             buf, readlen);
 383
 384                         zstream.next_in = buf;
 385                         zstream.avail_in = readlen;
 386                         result = git_deflate(&zstream, 0);
 387                         if (result != Z_OK)
 388                                 die("deflate error (%d)", result);
 389                         out_len = zstream.next_out - compressed;
 390
 391                         if (out_len > 0) {
 392                                 write_or_die(1, compressed, out_len);
 393                                 compressed_size += out_len;
 394                                 zstream.next_out = compressed;
 395                                 zstream.avail_out = sizeof(compressed);
 396                         }
 397
 398                 }
 399                 close_istream(stream);
 400                 if (readlen)
 401                         return readlen;
 402
 403                 zstream.next_in = buf;
 404                 zstream.avail_in = 0;
 405                 result = git_deflate(&zstream, Z_FINISH);
 406                 if (result != Z_STREAM_END)
 407                         die("deflate error (%d)", result);
 408
 409                 git_deflate_end(&zstream);
 410                 out_len = zstream.next_out - compressed;
 411                 write_or_die(1, compressed, out_len);
 412                 compressed_size += out_len;
 413                 zip_offset += compressed_size;
 414
 415                 write_zip_data_desc(size, compressed_size, crc);
 416                 zip_offset += ZIP_DATA_DESC_SIZE;
 417
 418                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 419         } else if (compressed_size > 0) {
 420                 write_or_die(1, out, compressed_size);
 421                 zip_offset += compressed_size;
 422         }
 423
 424         free(deflated);
 425         free(buffer);
 426
 427         copy_le16(dirent.attr1, !is_binary);
 428
 429         memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
 430         zip_dir_offset += ZIP_DIR_HEADER_SIZE;
 431         memcpy(zip_dir + zip_dir_offset, path, pathlen);
 432         zip_dir_offset += pathlen;
 433         memcpy(zip_dir + zip_dir_offset, &extra, ZIP_EXTRA_MTIME_SIZE);
 434         zip_dir_offset += ZIP_EXTRA_MTIME_SIZE;
 435         zip_dir_entries++;
 436
 437         return 0;
 438 }
 439
 440 static void write_zip_trailer(const unsigned char *sha1)
 441 {
 442         struct zip_dir_trailer trailer;
 443
 444         copy_le32(trailer.magic, 0x06054b50);
 445         copy_le16(trailer.disk, 0);
 446         copy_le16(trailer.directory_start_disk, 0);
 447         copy_le16(trailer.entries_on_this_disk, zip_dir_entries);
 448         copy_le16(trailer.entries, zip_dir_entries);
 449         copy_le32(trailer.size, zip_dir_offset);
 450         copy_le32(trailer.offset, zip_offset);
 451         copy_le16(trailer.comment_length, sha1 ? 40 : 0);
 452
 453         write_or_die(1, zip_dir, zip_dir_offset);
 454         write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
 455         if (sha1)
 456                 write_or_die(1, sha1_to_hex(sha1), 40);
 457 }
 458
 459 static void dos_time(time_t *time, int *dos_date, int *dos_time)
 460 {
 461         struct tm *t = localtime(time);
 462
 463         *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
 464                     (t->tm_year + 1900 - 1980) * 512;
 465         *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
 466 }
 467
 468 static int write_zip_archive(const struct archiver *ar,
 469                              struct archiver_args *args)
 470 {
 471         int err;
 472
 473         dos_time(&args->time, &zip_date, &zip_time);
 474
 475         zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
 476         zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
 477
 478         err = write_archive_entries(args, write_zip_entry);
 479         if (!err)
 480                 write_zip_trailer(args->commit_sha1);
 481
 482         free(zip_dir);
 483
 484         return err;
 485 }
 486
 487 static struct archiver zip_archiver = {
 488         "zip",
 489         write_zip_archive,
 490         ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
 491 };
 492
 493 void init_zip_archiver(void)
 494 {
 495         register_archiver(&zip_archiver);
 496 }