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 /*
 118  * pax extended header records have the format "%u %s=%s\n".  %u contains
 119  * the size of the whole string (including the %u), the first %s is the
 120  * keyword, the second one is the value.  This function constructs such a
 121  * string and appends it to a struct strbuf.
 122  */
 123 static void strbuf_append_ext_header(struct strbuf *sb, const char *keyword,
 124                                      const char *value, unsigned int valuelen)
 125 {
 126         char *p;
 127         int len, total, tmp;
 128
 129         /* "%u %s=%s\n" */
 130         len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 131         for (tmp = len; tmp > 9; tmp /= 10)
 132                 len++;
 133
 134         total = sb->len + len;
 135         if (total > sb->alloc) {
 136                 sb->buf = xrealloc(sb->buf, total);
 137                 sb->alloc = total;
 138         }
 139
 140         p = sb->buf;
 141         p += sprintf(p, "%u %s=", len, keyword);
 142         memcpy(p, value, valuelen);
 143         p += valuelen;
 144         *p = '\n';
 145         sb->len = total;
 146 }
 147
 148 static unsigned int ustar_header_chksum(const struct ustar_header *header)
 149 {
 150         char *p = (char *)header;
 151         unsigned int chksum = 0;
 152         while (p < header->chksum)
 153                 chksum += *p++;
 154         chksum += sizeof(header->chksum) * ' ';
 155         p += sizeof(header->chksum);
 156         while (p < (char *)header + sizeof(struct ustar_header))
 157                 chksum += *p++;
 158         return chksum;
 159 }
 160
 161 static void write_entry(const unsigned char *sha1, struct strbuf *path,
 162                         unsigned int mode, void *buffer, unsigned long size)
 163 {
 164         struct ustar_header header;
 165         struct strbuf ext_header;
 166
 167         memset(&header, 0, sizeof(header));
 168         ext_header.buf = NULL;
 169         ext_header.len = ext_header.alloc = 0;
 170
 171         if (!sha1) {
 172                 *header.typeflag = TYPEFLAG_GLOBAL_HEADER;
 173                 mode = 0100666;
 174                 strcpy(header.name, "pax_global_header");
 175         } else if (!path) {
 176                 *header.typeflag = TYPEFLAG_EXT_HEADER;
 177                 mode = 0100666;
 178                 sprintf(header.name, "%s.paxheader", sha1_to_hex(sha1));
 179         } else {
 180                 if (S_ISDIR(mode)) {
 181                         *header.typeflag = TYPEFLAG_DIR;
 182                         mode |= 0777;
 183                 } else if (S_ISLNK(mode)) {
 184                         *header.typeflag = TYPEFLAG_LNK;
 185                         mode |= 0777;
 186                 } else if (S_ISREG(mode)) {
 187                         *header.typeflag = TYPEFLAG_REG;
 188                         mode |= (mode & 0100) ? 0777 : 0666;
 189                 } else {
 190                         error("unsupported file mode: 0%o (SHA1: %s)",
 191                               mode, sha1_to_hex(sha1));
 192                         return;
 193                 }
 194                 if (path->len > sizeof(header.name)) {
 195                         sprintf(header.name, "%s.data", sha1_to_hex(sha1));
 196                         strbuf_append_ext_header(&ext_header, "path",
 197                                                  path->buf, path->len);
 198                 } else
 199                         memcpy(header.name, path->buf, path->len);
 200         }
 201
 202         if (S_ISLNK(mode) && buffer) {
 203                 if (size > sizeof(header.linkname)) {
 204                         sprintf(header.linkname, "see %s.paxheader",
 205                                 sha1_to_hex(sha1));
 206                         strbuf_append_ext_header(&ext_header, "linkpath",
 207                                                  buffer, size);
 208                 } else
 209                         memcpy(header.linkname, buffer, size);
 210         }
 211
 212         sprintf(header.mode, "%07o", mode & 07777);
 213         sprintf(header.size, "%011lo", S_ISREG(mode) ? size : 0);
 214         sprintf(header.mtime, "%011lo", archive_time);
 215
 216         /* XXX: should we provide more meaningful info here? */
 217         sprintf(header.uid, "%07o", 0);
 218         sprintf(header.gid, "%07o", 0);
 219         strncpy(header.uname, "git", 31);
 220         strncpy(header.gname, "git", 31);
 221         sprintf(header.devmajor, "%07o", 0);
 222         sprintf(header.devminor, "%07o", 0);
 223
 224         memcpy(header.magic, "ustar", 6);
 225         memcpy(header.version, "00", 2);
 226
 227         sprintf(header.chksum, "%07o", ustar_header_chksum(&header));
 228
 229         if (ext_header.len > 0) {
 230                 write_entry(sha1, NULL, 0, ext_header.buf, ext_header.len);
 231                 free(ext_header.buf);
 232         }
 233         write_blocked(&header, sizeof(header));
 234         if (S_ISREG(mode) && buffer && size > 0)
 235                 write_blocked(buffer, size);
 236 }
 237
 238 static void append_string(char **p, const char *s)
 239 {
 240         unsigned int len = strlen(s);
 241         memcpy(*p, s, len);
 242         *p += len;
 243 }
 244
 245 static void append_char(char **p, char c)
 246 {
 247         **p = c;
 248         *p += 1;
 249 }
 250
 251 static void append_path_prefix(char **buffer, struct path_prefix *prefix)
 252 {
 253         if (!prefix)
 254                 return;
 255         append_path_prefix(buffer, prefix->prev);
 256         append_string(buffer, prefix->name);
 257         append_char(buffer, '/');
 258 }
 259
 260 static unsigned int path_prefix_len(struct path_prefix *prefix)
 261 {
 262         if (!prefix)
 263                 return 0;
 264         return path_prefix_len(prefix->prev) + strlen(prefix->name) + 1;
 265 }
 266
 267 static void append_path(char **p, int is_dir, const char *basepath,
 268                         struct path_prefix *prefix, const char *path)
 269 {
 270         if (basepath) {
 271                 append_string(p, basepath);
 272                 append_char(p, '/');
 273         }
 274         append_path_prefix(p, prefix);
 275         append_string(p, path);
 276         if (is_dir)
 277                 append_char(p, '/');
 278 }
 279
 280 static unsigned int path_len(int is_dir, const char *basepath,
 281                              struct path_prefix *prefix, const char *path)
 282 {
 283         unsigned int len = 0;
 284         if (basepath)
 285                 len += strlen(basepath) + 1;
 286         len += path_prefix_len(prefix) + strlen(path);
 287         if (is_dir)
 288                 len++;
 289         return len;
 290 }
 291
 292 static void append_extended_header_prefix(char **p, unsigned int size,
 293                                           const char *keyword)
 294 {
 295         int len = sprintf(*p, "%u %s=", size, keyword);
 296         *p += len;
 297 }
 298
 299 static unsigned int extended_header_len(const char *keyword,
 300                                         unsigned int valuelen)
 301 {
 302         /* "%u %s=%s\n" */
 303         unsigned int len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 304         if (len > 9)
 305                 len++;
 306         if (len > 99)
 307                 len++;
 308         return len;
 309 }
 310
 311 static void append_extended_header(char **p, const char *keyword,
 312                                    const char *value, unsigned int len)
 313 {
 314         unsigned int size = extended_header_len(keyword, len);
 315         append_extended_header_prefix(p, size, keyword);
 316         memcpy(*p, value, len);
 317         *p += len;
 318         append_char(p, '\n');
 319 }
 320
 321 static void write_header(const unsigned char *, char, const char *, struct path_prefix *,
 322                          const char *, unsigned int, void *, unsigned long);
 323
 324 /* stores a pax extended header directly in the block buffer */
 325 static void write_extended_header(const char *headerfilename, int is_dir,
 326                                   unsigned int flags, const char *basepath,
 327                                   struct path_prefix *prefix,
 328                                   const char *path, unsigned int namelen,
 329                                   void *content, unsigned int contentsize)
 330 {
 331         char *buffer, *p;
 332         unsigned int pathlen, size, linkpathlen = 0;
 333
 334         size = pathlen = extended_header_len("path", namelen);
 335         if (flags & EXT_HEADER_LINKPATH) {
 336                 linkpathlen = extended_header_len("linkpath", contentsize);
 337                 size += linkpathlen;
 338         }
 339         write_header(NULL, TYPEFLAG_EXT_HEADER, NULL, NULL, headerfilename,
 340                      0100600, NULL, size);
 341
 342         buffer = p = malloc(size);
 343         if (!buffer)
 344                 die("git-tar-tree: %s", strerror(errno));
 345         append_extended_header_prefix(&p, pathlen, "path");
 346         append_path(&p, is_dir, basepath, prefix, path);
 347         append_char(&p, '\n');
 348         if (flags & EXT_HEADER_LINKPATH)
 349                 append_extended_header(&p, "linkpath", content, contentsize);
 350         write_blocked(buffer, size);
 351         free(buffer);
 352 }
 353
 354 static void write_global_extended_header(const unsigned char *sha1)
 355 {
 356         struct strbuf ext_header;
 357         ext_header.buf = NULL;
 358         ext_header.len = ext_header.alloc = 0;
 359         strbuf_append_ext_header(&ext_header, "comment", sha1_to_hex(sha1), 40);
 360         write_entry(NULL, NULL, 0, ext_header.buf, ext_header.len);
 361         free(ext_header.buf);
 362 }
 363
 364 /* stores a ustar header directly in the block buffer */
 365 static void write_header(const unsigned char *sha1, char typeflag, const char *basepath,
 366                          struct path_prefix *prefix, const char *path,
 367                          unsigned int mode, void *buffer, unsigned long size)
 368 {
 369         unsigned int namelen;
 370         char *header = NULL;
 371         unsigned int checksum = 0;
 372         int i;
 373         unsigned int ext_header = 0;
 374
 375         if (typeflag == TYPEFLAG_AUTO) {
 376                 if (S_ISDIR(mode))
 377                         typeflag = TYPEFLAG_DIR;
 378                 else if (S_ISLNK(mode))
 379                         typeflag = TYPEFLAG_LNK;
 380                 else
 381                         typeflag = TYPEFLAG_REG;
 382         }
 383
 384         namelen = path_len(S_ISDIR(mode), basepath, prefix, path);
 385         if (namelen > 100)
 386                 ext_header |= EXT_HEADER_PATH;
 387         if (typeflag == TYPEFLAG_LNK && size > 100)
 388                 ext_header |= EXT_HEADER_LINKPATH;
 389
 390         /* the extended header must be written before the normal one */
 391         if (ext_header) {
 392                 char headerfilename[51];
 393                 sprintf(headerfilename, "%s.paxheader", sha1_to_hex(sha1));
 394                 write_extended_header(headerfilename, S_ISDIR(mode),
 395                                       ext_header, basepath, prefix, path,
 396                                       namelen, buffer, size);
 397         }
 398
 399         header = get_record();
 400
 401         if (ext_header) {
 402                 sprintf(header, "%s.data", sha1_to_hex(sha1));
 403         } else {
 404                 char *p = header;
 405                 append_path(&p, S_ISDIR(mode), basepath, prefix, path);
 406         }
 407
 408         if (typeflag == TYPEFLAG_LNK) {
 409                 if (ext_header & EXT_HEADER_LINKPATH) {
 410                         sprintf(&header[157], "see %s.paxheader",
 411                                 sha1_to_hex(sha1));
 412                 } else {
 413                         if (buffer)
 414                                 strncpy(&header[157], buffer, size);
 415                 }
 416         }
 417
 418         if (S_ISDIR(mode))
 419                 mode |= 0777;
 420         else if (S_ISREG(mode))
 421                 mode |= (mode & 0100) ? 0777 : 0666;
 422         else if (S_ISLNK(mode))
 423                 mode |= 0777;
 424         sprintf(&header[100], "%07o", mode & 07777);
 425
 426         /* XXX: should we provide more meaningful info here? */
 427         sprintf(&header[108], "%07o", 0);       /* uid */
 428         sprintf(&header[116], "%07o", 0);       /* gid */
 429         strncpy(&header[265], "git", 31);       /* uname */
 430         strncpy(&header[297], "git", 31);       /* gname */
 431
 432         if (S_ISDIR(mode) || S_ISLNK(mode))
 433                 size = 0;
 434         sprintf(&header[124], "%011lo", size);
 435         sprintf(&header[136], "%011lo", archive_time);
 436
 437         header[156] = typeflag;
 438
 439         memcpy(&header[257], "ustar", 6);
 440         memcpy(&header[263], "00", 2);
 441
 442         sprintf(&header[329], "%07o", 0);       /* devmajor */
 443         sprintf(&header[337], "%07o", 0);       /* devminor */
 444
 445         memset(&header[148], ' ', 8);
 446         for (i = 0; i < RECORDSIZE; i++)
 447                 checksum += header[i];
 448         sprintf(&header[148], "%07o", checksum & 0x1fffff);
 449
 450         write_if_needed();
 451 }
 452
 453 static void traverse_tree(struct tree_desc *tree,
 454                           struct path_prefix *prefix)
 455 {
 456         struct path_prefix this_prefix;
 457         this_prefix.prev = prefix;
 458
 459         while (tree->size) {
 460                 const char *name;
 461                 const unsigned char *sha1;
 462                 unsigned mode;
 463                 void *eltbuf;
 464                 char elttype[20];
 465                 unsigned long eltsize;
 466
 467                 sha1 = tree_entry_extract(tree, &name, &mode);
 468                 update_tree_entry(tree);
 469
 470                 eltbuf = read_sha1_file(sha1, elttype, &eltsize);
 471                 if (!eltbuf)
 472                         die("cannot read %s", sha1_to_hex(sha1));
 473                 write_header(sha1, TYPEFLAG_AUTO, basedir,
 474                              prefix, name, mode, eltbuf, eltsize);
 475                 if (S_ISDIR(mode)) {
 476                         struct tree_desc subtree;
 477                         subtree.buf = eltbuf;
 478                         subtree.size = eltsize;
 479                         this_prefix.name = name;
 480                         traverse_tree(&subtree, &this_prefix);
 481                 } else if (!S_ISLNK(mode)) {
 482                         write_blocked(eltbuf, eltsize);
 483                 }
 484                 free(eltbuf);
 485         }
 486 }
 487
 488 int main(int argc, char **argv)
 489 {
 490         unsigned char sha1[20], tree_sha1[20];
 491         struct commit *commit;
 492         struct tree_desc tree;
 493
 494         setup_git_directory();
 495
 496         switch (argc) {
 497         case 3:
 498                 basedir = argv[2];
 499                 /* FALLTHROUGH */
 500         case 2:
 501                 if (get_sha1(argv[1], sha1) < 0)
 502                         usage(tar_tree_usage);
 503                 break;
 504         default:
 505                 usage(tar_tree_usage);
 506         }
 507
 508         commit = lookup_commit_reference_gently(sha1, 1);
 509         if (commit) {
 510                 write_global_extended_header(commit->object.sha1);
 511                 archive_time = commit->date;
 512         }
 513         tree.buf = read_object_with_reference(sha1, "tree", &tree.size,
 514                                               tree_sha1);
 515         if (!tree.buf)
 516                 die("not a reference to a tag, commit or tree object: %s",
 517                     sha1_to_hex(sha1));
 518         if (!archive_time)
 519                 archive_time = time(NULL);
 520         if (basedir)
 521                 write_header(tree_sha1, TYPEFLAG_DIR, NULL, NULL,
 522                         basedir, 040777, NULL, 0);
 523         traverse_tree(&tree, NULL);
 524         write_trailer();
 525         return 0;
 526 }