tar-tree.c

   1 /*
   2  * Copyright (c) 2005, 2006 Rene Scharfe
   3  */
   4 #include <time.h>
   5 #include "cache.h"
   6 #include "diff.h"
   7 #include "commit.h"
   8 #include "strbuf.h"
   9 #include "tar.h"
  10
  11 #define RECORDSIZE      (512)
  12 #define BLOCKSIZE       (RECORDSIZE * 20)
  13
  14 #define EXT_HEADER_PATH         1
  15 #define EXT_HEADER_LINKPATH     2
  16
  17 static const char tar_tree_usage[] = "git-tar-tree <key> [basedir]";
  18
  19 static char block[BLOCKSIZE];
  20 static unsigned long offset;
  21
  22 static const char *basedir;
  23 static time_t archive_time;
  24
  25 struct path_prefix {
  26         struct path_prefix *prev;
  27         const char *name;
  28 };
  29
  30 /* tries hard to write, either succeeds or dies in the attempt */
  31 static void reliable_write(void *buf, unsigned long size)
  32 {
  33         while (size > 0) {
  34                 long ret = xwrite(1, buf, size);
  35                 if (ret < 0) {
  36                         if (errno == EPIPE)
  37                                 exit(0);
  38                         die("git-tar-tree: %s", strerror(errno));
  39                 } else if (!ret) {
  40                         die("git-tar-tree: disk full?");
  41                 }
  42                 size -= ret;
  43                 buf += ret;
  44         }
  45 }
  46
  47 /* writes out the whole block, but only if it is full */
  48 static void write_if_needed(void)
  49 {
  50         if (offset == BLOCKSIZE) {
  51                 reliable_write(block, BLOCKSIZE);
  52                 offset = 0;
  53         }
  54 }
  55
  56 /* acquire the next record from the buffer; user must call write_if_needed() */
  57 static char *get_record(void)
  58 {
  59         char *p = block + offset;
  60         memset(p, 0, RECORDSIZE);
  61         offset += RECORDSIZE;
  62         return p;
  63 }
  64
  65 /*
  66  * The end of tar archives is marked by 1024 nul bytes and after that
  67  * follows the rest of the block (if any).
  68  */
  69 static void write_trailer(void)
  70 {
  71         get_record();
  72         write_if_needed();
  73         get_record();
  74         write_if_needed();
  75         while (offset) {
  76                 get_record();
  77                 write_if_needed();
  78         }
  79 }
  80
  81 /*
  82  * queues up writes, so that all our write(2) calls write exactly one
  83  * full block; pads writes to RECORDSIZE
  84  */
  85 static void write_blocked(void *buf, unsigned long size)
  86 {
  87         unsigned long tail;
  88
  89         if (offset) {
  90                 unsigned long chunk = BLOCKSIZE - offset;
  91                 if (size < chunk)
  92                         chunk = size;
  93                 memcpy(block + offset, buf, chunk);
  94                 size -= chunk;
  95                 offset += chunk;
  96                 buf += chunk;
  97                 write_if_needed();
  98         }
  99         while (size >= BLOCKSIZE) {
 100                 reliable_write(buf, BLOCKSIZE);
 101                 size -= BLOCKSIZE;
 102                 buf += BLOCKSIZE;
 103         }
 104         if (size) {
 105                 memcpy(block + offset, buf, size);
 106                 buf += size;
 107                 offset += size;
 108         }
 109         tail = offset % RECORDSIZE;
 110         if (tail)  {
 111                 memset(block + offset, 0, RECORDSIZE - tail);
 112                 offset += RECORDSIZE - tail;
 113         }
 114         write_if_needed();
 115 }
 116
 117 static void strbuf_append_string(struct strbuf *sb, const char *s)
 118 {
 119         int slen = strlen(s);
 120         int total = sb->len + slen;
 121         if (total > sb->alloc) {
 122                 sb->buf = xrealloc(sb->buf, total);
 123                 sb->alloc = total;
 124         }
 125         memcpy(sb->buf + sb->len, s, slen);
 126         sb->len = total;
 127 }
 128
 129 /*
 130  * pax extended header records have the format "%u %s=%s\n".  %u contains
 131  * the size of the whole string (including the %u), the first %s is the
 132  * keyword, the second one is the value.  This function constructs such a
 133  * string and appends it to a struct strbuf.
 134  */
 135 static void strbuf_append_ext_header(struct strbuf *sb, const char *keyword,
 136                                      const char *value, unsigned int valuelen)
 137 {
 138         char *p;
 139         int len, total, tmp;
 140
 141         /* "%u %s=%s\n" */
 142         len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 143         for (tmp = len; tmp > 9; tmp /= 10)
 144                 len++;
 145
 146         total = sb->len + len;
 147         if (total > sb->alloc) {
 148                 sb->buf = xrealloc(sb->buf, total);
 149                 sb->alloc = total;
 150         }
 151
 152         p = sb->buf;
 153         p += sprintf(p, "%u %s=", len, keyword);
 154         memcpy(p, value, valuelen);
 155         p += valuelen;
 156         *p = '\n';
 157         sb->len = total;
 158 }
 159
 160 static unsigned int ustar_header_chksum(const struct ustar_header *header)
 161 {
 162         char *p = (char *)header;
 163         unsigned int chksum = 0;
 164         while (p < header->chksum)
 165                 chksum += *p++;
 166         chksum += sizeof(header->chksum) * ' ';
 167         p += sizeof(header->chksum);
 168         while (p < (char *)header + sizeof(struct ustar_header))
 169                 chksum += *p++;
 170         return chksum;
 171 }
 172
 173 static void write_entry(const unsigned char *sha1, struct strbuf *path,
 174                         unsigned int mode, void *buffer, unsigned long size)
 175 {
 176         struct ustar_header header;
 177         struct strbuf ext_header;
 178
 179         memset(&header, 0, sizeof(header));
 180         ext_header.buf = NULL;
 181         ext_header.len = ext_header.alloc = 0;
 182
 183         if (!sha1) {
 184                 *header.typeflag = TYPEFLAG_GLOBAL_HEADER;
 185                 mode = 0100666;
 186                 strcpy(header.name, "pax_global_header");
 187         } else if (!path) {
 188                 *header.typeflag = TYPEFLAG_EXT_HEADER;
 189                 mode = 0100666;
 190                 sprintf(header.name, "%s.paxheader", sha1_to_hex(sha1));
 191         } else {
 192                 if (S_ISDIR(mode)) {
 193                         *header.typeflag = TYPEFLAG_DIR;
 194                         mode |= 0777;
 195                 } else if (S_ISLNK(mode)) {
 196                         *header.typeflag = TYPEFLAG_LNK;
 197                         mode |= 0777;
 198                 } else if (S_ISREG(mode)) {
 199                         *header.typeflag = TYPEFLAG_REG;
 200                         mode |= (mode & 0100) ? 0777 : 0666;
 201                 } else {
 202                         error("unsupported file mode: 0%o (SHA1: %s)",
 203                               mode, sha1_to_hex(sha1));
 204                         return;
 205                 }
 206                 if (path->len > sizeof(header.name)) {
 207                         sprintf(header.name, "%s.data", sha1_to_hex(sha1));
 208                         strbuf_append_ext_header(&ext_header, "path",
 209                                                  path->buf, path->len);
 210                 } else
 211                         memcpy(header.name, path->buf, path->len);
 212         }
 213
 214         if (S_ISLNK(mode) && buffer) {
 215                 if (size > sizeof(header.linkname)) {
 216                         sprintf(header.linkname, "see %s.paxheader",
 217                                 sha1_to_hex(sha1));
 218                         strbuf_append_ext_header(&ext_header, "linkpath",
 219                                                  buffer, size);
 220                 } else
 221                         memcpy(header.linkname, buffer, size);
 222         }
 223
 224         sprintf(header.mode, "%07o", mode & 07777);
 225         sprintf(header.size, "%011lo", S_ISREG(mode) ? size : 0);
 226         sprintf(header.mtime, "%011lo", archive_time);
 227
 228         /* XXX: should we provide more meaningful info here? */
 229         sprintf(header.uid, "%07o", 0);
 230         sprintf(header.gid, "%07o", 0);
 231         strncpy(header.uname, "git", 31);
 232         strncpy(header.gname, "git", 31);
 233         sprintf(header.devmajor, "%07o", 0);
 234         sprintf(header.devminor, "%07o", 0);
 235
 236         memcpy(header.magic, "ustar", 6);
 237         memcpy(header.version, "00", 2);
 238
 239         sprintf(header.chksum, "%07o", ustar_header_chksum(&header));
 240
 241         if (ext_header.len > 0) {
 242                 write_entry(sha1, NULL, 0, ext_header.buf, ext_header.len);
 243                 free(ext_header.buf);
 244         }
 245         write_blocked(&header, sizeof(header));
 246         if (S_ISREG(mode) && buffer && size > 0)
 247                 write_blocked(buffer, size);
 248 }
 249
 250 static void append_string(char **p, const char *s)
 251 {
 252         unsigned int len = strlen(s);
 253         memcpy(*p, s, len);
 254         *p += len;
 255 }
 256
 257 static void append_char(char **p, char c)
 258 {
 259         **p = c;
 260         *p += 1;
 261 }
 262
 263 static void append_path_prefix(char **buffer, struct path_prefix *prefix)
 264 {
 265         if (!prefix)
 266                 return;
 267         append_path_prefix(buffer, prefix->prev);
 268         append_string(buffer, prefix->name);
 269         append_char(buffer, '/');
 270 }
 271
 272 static unsigned int path_prefix_len(struct path_prefix *prefix)
 273 {
 274         if (!prefix)
 275                 return 0;
 276         return path_prefix_len(prefix->prev) + strlen(prefix->name) + 1;
 277 }
 278
 279 static void append_path(char **p, int is_dir, const char *basepath,
 280                         struct path_prefix *prefix, const char *path)
 281 {
 282         if (basepath) {
 283                 append_string(p, basepath);
 284                 append_char(p, '/');
 285         }
 286         append_path_prefix(p, prefix);
 287         append_string(p, path);
 288         if (is_dir)
 289                 append_char(p, '/');
 290 }
 291
 292 static unsigned int path_len(int is_dir, const char *basepath,
 293                              struct path_prefix *prefix, const char *path)
 294 {
 295         unsigned int len = 0;
 296         if (basepath)
 297                 len += strlen(basepath) + 1;
 298         len += path_prefix_len(prefix) + strlen(path);
 299         if (is_dir)
 300                 len++;
 301         return len;
 302 }
 303
 304 static void append_extended_header_prefix(char **p, unsigned int size,
 305                                           const char *keyword)
 306 {
 307         int len = sprintf(*p, "%u %s=", size, keyword);
 308         *p += len;
 309 }
 310
 311 static unsigned int extended_header_len(const char *keyword,
 312                                         unsigned int valuelen)
 313 {
 314         /* "%u %s=%s\n" */
 315         unsigned int len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 316         if (len > 9)
 317                 len++;
 318         if (len > 99)
 319                 len++;
 320         return len;
 321 }
 322
 323 static void append_extended_header(char **p, const char *keyword,
 324                                    const char *value, unsigned int len)
 325 {
 326         unsigned int size = extended_header_len(keyword, len);
 327         append_extended_header_prefix(p, size, keyword);
 328         memcpy(*p, value, len);
 329         *p += len;
 330         append_char(p, '\n');
 331 }
 332
 333 static void write_header(const unsigned char *, char, const char *, struct path_prefix *,
 334                          const char *, unsigned int, void *, unsigned long);
 335
 336 /* stores a pax extended header directly in the block buffer */
 337 static void write_extended_header(const char *headerfilename, int is_dir,
 338                                   unsigned int flags, const char *basepath,
 339                                   struct path_prefix *prefix,
 340                                   const char *path, unsigned int namelen,
 341                                   void *content, unsigned int contentsize)
 342 {
 343         char *buffer, *p;
 344         unsigned int pathlen, size, linkpathlen = 0;
 345
 346         size = pathlen = extended_header_len("path", namelen);
 347         if (flags & EXT_HEADER_LINKPATH) {
 348                 linkpathlen = extended_header_len("linkpath", contentsize);
 349                 size += linkpathlen;
 350         }
 351         write_header(NULL, TYPEFLAG_EXT_HEADER, NULL, NULL, headerfilename,
 352                      0100600, NULL, size);
 353
 354         buffer = p = malloc(size);
 355         if (!buffer)
 356                 die("git-tar-tree: %s", strerror(errno));
 357         append_extended_header_prefix(&p, pathlen, "path");
 358         append_path(&p, is_dir, basepath, prefix, path);
 359         append_char(&p, '\n');
 360         if (flags & EXT_HEADER_LINKPATH)
 361                 append_extended_header(&p, "linkpath", content, contentsize);
 362         write_blocked(buffer, size);
 363         free(buffer);
 364 }
 365
 366 static void write_global_extended_header(const unsigned char *sha1)
 367 {
 368         struct strbuf ext_header;
 369         ext_header.buf = NULL;
 370         ext_header.len = ext_header.alloc = 0;
 371         strbuf_append_ext_header(&ext_header, "comment", sha1_to_hex(sha1), 40);
 372         write_entry(NULL, NULL, 0, ext_header.buf, ext_header.len);
 373         free(ext_header.buf);
 374 }
 375
 376 /* stores a ustar header directly in the block buffer */
 377 static void write_header(const unsigned char *sha1, char typeflag, const char *basepath,
 378                          struct path_prefix *prefix, const char *path,
 379                          unsigned int mode, void *buffer, unsigned long size)
 380 {
 381         unsigned int namelen;
 382         char *header = NULL;
 383         unsigned int checksum = 0;
 384         int i;
 385         unsigned int ext_header = 0;
 386
 387         if (typeflag == TYPEFLAG_AUTO) {
 388                 if (S_ISDIR(mode))
 389                         typeflag = TYPEFLAG_DIR;
 390                 else if (S_ISLNK(mode))
 391                         typeflag = TYPEFLAG_LNK;
 392                 else
 393                         typeflag = TYPEFLAG_REG;
 394         }
 395
 396         namelen = path_len(S_ISDIR(mode), basepath, prefix, path);
 397         if (namelen > 100)
 398                 ext_header |= EXT_HEADER_PATH;
 399         if (typeflag == TYPEFLAG_LNK && size > 100)
 400                 ext_header |= EXT_HEADER_LINKPATH;
 401
 402         /* the extended header must be written before the normal one */
 403         if (ext_header) {
 404                 char headerfilename[51];
 405                 sprintf(headerfilename, "%s.paxheader", sha1_to_hex(sha1));
 406                 write_extended_header(headerfilename, S_ISDIR(mode),
 407                                       ext_header, basepath, prefix, path,
 408                                       namelen, buffer, size);
 409         }
 410
 411         header = get_record();
 412
 413         if (ext_header) {
 414                 sprintf(header, "%s.data", sha1_to_hex(sha1));
 415         } else {
 416                 char *p = header;
 417                 append_path(&p, S_ISDIR(mode), basepath, prefix, path);
 418         }
 419
 420         if (typeflag == TYPEFLAG_LNK) {
 421                 if (ext_header & EXT_HEADER_LINKPATH) {
 422                         sprintf(&header[157], "see %s.paxheader",
 423                                 sha1_to_hex(sha1));
 424                 } else {
 425                         if (buffer)
 426                                 strncpy(&header[157], buffer, size);
 427                 }
 428         }
 429
 430         if (S_ISDIR(mode))
 431                 mode |= 0777;
 432         else if (S_ISREG(mode))
 433                 mode |= (mode & 0100) ? 0777 : 0666;
 434         else if (S_ISLNK(mode))
 435                 mode |= 0777;
 436         sprintf(&header[100], "%07o", mode & 07777);
 437
 438         /* XXX: should we provide more meaningful info here? */
 439         sprintf(&header[108], "%07o", 0);       /* uid */
 440         sprintf(&header[116], "%07o", 0);       /* gid */
 441         strncpy(&header[265], "git", 31);       /* uname */
 442         strncpy(&header[297], "git", 31);       /* gname */
 443
 444         if (S_ISDIR(mode) || S_ISLNK(mode))
 445                 size = 0;
 446         sprintf(&header[124], "%011lo", size);
 447         sprintf(&header[136], "%011lo", archive_time);
 448
 449         header[156] = typeflag;
 450
 451         memcpy(&header[257], "ustar", 6);
 452         memcpy(&header[263], "00", 2);
 453
 454         sprintf(&header[329], "%07o", 0);       /* devmajor */
 455         sprintf(&header[337], "%07o", 0);       /* devminor */
 456
 457         memset(&header[148], ' ', 8);
 458         for (i = 0; i < RECORDSIZE; i++)
 459                 checksum += header[i];
 460         sprintf(&header[148], "%07o", checksum & 0x1fffff);
 461
 462         write_if_needed();
 463 }
 464
 465 static void traverse_tree(struct tree_desc *tree, struct strbuf *path)
 466 {
 467         int pathlen = path->len;
 468
 469         while (tree->size) {
 470                 const char *name;
 471                 const unsigned char *sha1;
 472                 unsigned mode;
 473                 void *eltbuf;
 474                 char elttype[20];
 475                 unsigned long eltsize;
 476
 477                 sha1 = tree_entry_extract(tree, &name, &mode);
 478                 update_tree_entry(tree);
 479
 480                 eltbuf = read_sha1_file(sha1, elttype, &eltsize);
 481                 if (!eltbuf)
 482                         die("cannot read %s", sha1_to_hex(sha1));
 483
 484                 path->len = pathlen;
 485                 strbuf_append_string(path, name);
 486                 if (S_ISDIR(mode))
 487                         strbuf_append_string(path, "/");
 488
 489                 write_entry(sha1, path, mode, eltbuf, eltsize);
 490
 491                 if (S_ISDIR(mode)) {
 492                         struct tree_desc subtree;
 493                         subtree.buf = eltbuf;
 494                         subtree.size = eltsize;
 495                         traverse_tree(&subtree, path);
 496                 }
 497                 free(eltbuf);
 498         }
 499 }
 500
 501 int main(int argc, char **argv)
 502 {
 503         unsigned char sha1[20], tree_sha1[20];
 504         struct commit *commit;
 505         struct tree_desc tree;
 506         struct strbuf current_path;
 507
 508         current_path.buf = xmalloc(PATH_MAX);
 509         current_path.alloc = PATH_MAX;
 510         current_path.len = current_path.eof = 0;
 511
 512         setup_git_directory();
 513
 514         switch (argc) {
 515         case 3:
 516                 strbuf_append_string(&current_path, argv[2]);
 517                 strbuf_append_string(&current_path, "/");
 518                 /* FALLTHROUGH */
 519         case 2:
 520                 if (get_sha1(argv[1], sha1) < 0)
 521                         usage(tar_tree_usage);
 522                 break;
 523         default:
 524                 usage(tar_tree_usage);
 525         }
 526
 527         commit = lookup_commit_reference_gently(sha1, 1);
 528         if (commit) {
 529                 write_global_extended_header(commit->object.sha1);
 530                 archive_time = commit->date;
 531         } else
 532                 archive_time = time(NULL);
 533
 534         tree.buf = read_object_with_reference(sha1, "tree", &tree.size,
 535                                               tree_sha1);
 536         if (!tree.buf)
 537                 die("not a reference to a tag, commit or tree object: %s",
 538                     sha1_to_hex(sha1));
 539
 540         if (current_path.len > 0)
 541                 write_entry(tree_sha1, &current_path, 040777, NULL, 0);
 542         traverse_tree(&tree, &current_path);
 543         write_trailer();
 544         free(current_path.buf);
 545         return 0;
 546 }