tar-tree.c

   1 #include <time.h>
   2 #include "cache.h"
   3
   4 #define RECORDSIZE      (512)
   5 #define BLOCKSIZE       (RECORDSIZE * 20)
   6
   7 #define TYPEFLAG_AUTO           '\0'
   8 #define TYPEFLAG_REG            '0'
   9 #define TYPEFLAG_LNK            '2'
  10 #define TYPEFLAG_DIR            '5'
  11 #define TYPEFLAG_GLOBAL_HEADER  'g'
  12 #define TYPEFLAG_EXT_HEADER     'x'
  13
  14 #define EXT_HEADER_PATH         1
  15 #define EXT_HEADER_LINKPATH     2
  16
  17 static const char *tar_tree_usage = "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 = write(1, buf, size);
  35                 if (ret < 0) {
  36                         if (errno == EAGAIN)
  37                                 continue;
  38                         if (errno == EPIPE)
  39                                 exit(0);
  40                         die("tar-tree: %s", strerror(errno));
  41                 } else if (!ret) {
  42                         die("tar-tree: disk full?");
  43                 }
  44                 size -= ret;
  45                 buf += ret;
  46         }
  47 }
  48
  49 /* writes out the whole block, but only if it is full */
  50 static void write_if_needed(void)
  51 {
  52         if (offset == BLOCKSIZE) {
  53                 reliable_write(block, BLOCKSIZE);
  54                 offset = 0;
  55         }
  56 }
  57
  58 /* acquire the next record from the buffer; user must call write_if_needed() */
  59 static char *get_record(void)
  60 {
  61         char *p = block + offset;
  62         memset(p, 0, RECORDSIZE);
  63         offset += RECORDSIZE;
  64         return p;
  65 }
  66
  67 /*
  68  * The end of tar archives is marked by 1024 nul bytes and after that
  69  * follows the rest of the block (if any).
  70  */
  71 static void write_trailer(void)
  72 {
  73         memset(block + offset, 0, RECORDSIZE);
  74         offset += RECORDSIZE;
  75         write_if_needed();
  76         memset(block + offset, 0, RECORDSIZE);
  77         offset += RECORDSIZE;
  78         write_if_needed();
  79         if (offset) {
  80                 memset(block + offset, 0, BLOCKSIZE - offset);
  81                 reliable_write(block, BLOCKSIZE);
  82                 offset = 0;
  83         }
  84 }
  85
  86 /*
  87  * queues up writes, so that all our write(2) calls write exactly one
  88  * full block; pads writes to RECORDSIZE
  89  */
  90 static void write_blocked(void *buf, unsigned long size)
  91 {
  92         unsigned long tail;
  93
  94         if (offset) {
  95                 unsigned long chunk = BLOCKSIZE - offset;
  96                 if (size < chunk)
  97                         chunk = size;
  98                 memcpy(block + offset, buf, chunk);
  99                 size -= chunk;
 100                 offset += chunk;
 101                 buf += chunk;
 102                 write_if_needed();
 103         }
 104         while (size >= BLOCKSIZE) {
 105                 reliable_write(buf, BLOCKSIZE);
 106                 size -= BLOCKSIZE;
 107                 buf += BLOCKSIZE;
 108         }
 109         if (size) {
 110                 memcpy(block + offset, buf, size);
 111                 buf += size;
 112                 offset += size;
 113         }
 114         tail = offset % RECORDSIZE;
 115         if (tail)  {
 116                 memset(block + offset, 0, RECORDSIZE - tail);
 117                 offset += RECORDSIZE - tail;
 118         }
 119         write_if_needed();
 120 }
 121
 122 static void append_string(char **p, const char *s)
 123 {
 124         unsigned int len = strlen(s);
 125         memcpy(*p, s, len);
 126         *p += len;
 127 }
 128
 129 static void append_char(char **p, char c)
 130 {
 131         **p = c;
 132         *p += 1;
 133 }
 134
 135 static void append_path_prefix(char **buffer, struct path_prefix *prefix)
 136 {
 137         if (!prefix)
 138                 return;
 139         append_path_prefix(buffer, prefix->prev);
 140         append_string(buffer, prefix->name);
 141         append_char(buffer, '/');
 142 }
 143
 144 static unsigned int path_prefix_len(struct path_prefix *prefix)
 145 {
 146         if (!prefix)
 147                 return 0;
 148         return path_prefix_len(prefix->prev) + strlen(prefix->name) + 1;
 149 }
 150
 151 static void append_path(char **p, int is_dir, const char *basepath,
 152                         struct path_prefix *prefix, const char *path)
 153 {
 154         if (basepath) {
 155                 append_string(p, basepath);
 156                 append_char(p, '/');
 157         }
 158         append_path_prefix(p, prefix);
 159         append_string(p, path);
 160         if (is_dir)
 161                 append_char(p, '/');
 162 }
 163
 164 static unsigned int path_len(int is_dir, const char *basepath,
 165                              struct path_prefix *prefix, const char *path)
 166 {
 167         unsigned int len = 0;
 168         if (basepath)
 169                 len += strlen(basepath) + 1;
 170         len += path_prefix_len(prefix) + strlen(path);
 171         if (is_dir)
 172                 len++;
 173         return len;
 174 }
 175
 176 static void append_extended_header_prefix(char **p, unsigned int size,
 177                                           const char *keyword)
 178 {
 179         int len = sprintf(*p, "%u %s=", size, keyword);
 180         *p += len;
 181 }
 182
 183 static unsigned int extended_header_len(const char *keyword,
 184                                         unsigned int valuelen)
 185 {
 186         /* "%u %s=%s\n" */
 187         unsigned int len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 188         if (len > 9)
 189                 len++;
 190         if (len > 99)
 191                 len++;
 192         return len;
 193 }
 194
 195 static void append_extended_header(char **p, const char *keyword,
 196                                    const char *value, unsigned int len)
 197 {
 198         unsigned int size = extended_header_len(keyword, len);
 199         append_extended_header_prefix(p, size, keyword);
 200         memcpy(*p, value, len);
 201         *p += len;
 202         append_char(p, '\n');
 203 }
 204
 205 static void write_header(const char *, char, const char *, struct path_prefix *,
 206                          const char *, unsigned int, void *, unsigned long);
 207
 208 /* stores a pax extended header directly in the block buffer */
 209 static void write_extended_header(const char *headerfilename, int is_dir,
 210                                   unsigned int flags, const char *basepath,
 211                                   struct path_prefix *prefix,
 212                                   const char *path, unsigned int namelen,
 213                                   void *content, unsigned int contentsize)
 214 {
 215         char *p;
 216         unsigned int pathlen, size, linkpathlen = 0;
 217
 218         size = pathlen = extended_header_len("path", namelen);
 219         if (flags & EXT_HEADER_LINKPATH) {
 220                 linkpathlen = extended_header_len("linkpath", contentsize);
 221                 size += linkpathlen;
 222         }
 223         if (size > RECORDSIZE)
 224                 die("tar-tree: extended header too big, wtf?");
 225         write_header(NULL, TYPEFLAG_EXT_HEADER, NULL, NULL, headerfilename,
 226                      0100600, NULL, size);
 227
 228         p = get_record();
 229         append_extended_header_prefix(&p, pathlen, "path");
 230         append_path(&p, is_dir, basepath, prefix, path);
 231         append_char(&p, '\n');
 232         if (flags & EXT_HEADER_LINKPATH)
 233                 append_extended_header(&p, "linkpath", content, contentsize);
 234         write_if_needed();
 235 }
 236
 237 static void write_global_extended_header(const char *sha1)
 238 {
 239         char *p;
 240         unsigned int size;
 241
 242         size = extended_header_len("comment", 40);
 243         write_header(NULL, TYPEFLAG_GLOBAL_HEADER, NULL, NULL,
 244                      "pax_global_header", 0100600, NULL, size);
 245
 246         p = get_record();
 247         append_extended_header(&p, "comment", sha1_to_hex(sha1), 40);
 248         write_if_needed();
 249 }
 250
 251 /* stores a ustar header directly in the block buffer */
 252 static void write_header(const char *sha1, char typeflag, const char *basepath,
 253                          struct path_prefix *prefix, const char *path,
 254                          unsigned int mode, void *buffer, unsigned long size)
 255 {
 256         unsigned int namelen;
 257         char *header = NULL;
 258         unsigned int checksum = 0;
 259         int i;
 260         unsigned int ext_header = 0;
 261
 262         if (typeflag == TYPEFLAG_AUTO) {
 263                 if (S_ISDIR(mode))
 264                         typeflag = TYPEFLAG_DIR;
 265                 else if (S_ISLNK(mode))
 266                         typeflag = TYPEFLAG_LNK;
 267                 else
 268                         typeflag = TYPEFLAG_REG;
 269         }
 270
 271         namelen = path_len(S_ISDIR(mode), basepath, prefix, path);
 272         if (namelen > 500)
 273                 die("tar-tree: name too log of object %s\n", sha1_to_hex(sha1));
 274         else if (namelen > 100)
 275                 ext_header |= EXT_HEADER_PATH;
 276         if (typeflag == TYPEFLAG_LNK && size > 100)
 277                 ext_header |= EXT_HEADER_LINKPATH;
 278
 279         /* the extended header must be written before the normal one */
 280         if (ext_header) {
 281                 char headerfilename[51];
 282                 sprintf(headerfilename, "%s.paxheader", sha1_to_hex(sha1));
 283                 write_extended_header(headerfilename, S_ISDIR(mode),
 284                                       ext_header, basepath, prefix, path,
 285                                       namelen, buffer, size);
 286         }
 287
 288         header = get_record();
 289
 290         if (ext_header) {
 291                 sprintf(header, "%s.data", sha1_to_hex(sha1));
 292         } else {
 293                 char *p = header;
 294                 append_path(&p, S_ISDIR(mode), basepath, prefix, path);
 295         }
 296
 297         if (typeflag == TYPEFLAG_LNK) {
 298                 if (ext_header & EXT_HEADER_LINKPATH) {
 299                         sprintf(&header[157], "see %s.paxheader",
 300                                 sha1_to_hex(sha1));
 301                 } else {
 302                         if (buffer)
 303                                 strncpy(&header[157], buffer, size);
 304                 }
 305         }
 306
 307         if (S_ISDIR(mode))
 308                 mode |= 0755;   /* GIT doesn't store permissions of dirs */
 309         if (S_ISLNK(mode))
 310                 mode |= 0777;   /* ... nor of symlinks */
 311         sprintf(&header[100], "%07o", mode & 07777);
 312
 313         /* XXX: should we provide more meaningful info here? */
 314         sprintf(&header[108], "%07o", 0);       /* uid */
 315         sprintf(&header[116], "%07o", 0);       /* gid */
 316         strncpy(&header[265], "git", 31);       /* uname */
 317         strncpy(&header[297], "git", 31);       /* gname */
 318
 319         if (S_ISDIR(mode) || S_ISLNK(mode))
 320                 size = 0;
 321         sprintf(&header[124], "%011lo", size);
 322         sprintf(&header[136], "%011lo", archive_time);
 323
 324         header[156] = typeflag;
 325
 326         memcpy(&header[257], "ustar", 6);
 327         memcpy(&header[263], "00", 2);
 328
 329         printf(&header[329], "%07o", 0);        /* devmajor */
 330         printf(&header[337], "%07o", 0);        /* devminor */
 331
 332         memset(&header[148], ' ', 8);
 333         for (i = 0; i < RECORDSIZE; i++)
 334                 checksum += header[i];
 335         sprintf(&header[148], "%07o", checksum & 0x1fffff);
 336
 337         write_if_needed();
 338 }
 339
 340 static void traverse_tree(void *buffer, unsigned long size,
 341                           struct path_prefix *prefix)
 342 {
 343         struct path_prefix this_prefix;
 344         this_prefix.prev = prefix;
 345
 346         while (size) {
 347                 int namelen = strlen(buffer)+1;
 348                 void *eltbuf;
 349                 char elttype[20];
 350                 unsigned long eltsize;
 351                 unsigned char *sha1 = buffer + namelen;
 352                 char *path = strchr(buffer, ' ') + 1;
 353                 unsigned int mode;
 354
 355                 if (size < namelen + 20 || sscanf(buffer, "%o", &mode) != 1)
 356                         die("corrupt 'tree' file");
 357                 buffer = sha1 + 20;
 358                 size -= namelen + 20;
 359
 360                 eltbuf = read_sha1_file(sha1, elttype, &eltsize);
 361                 if (!eltbuf)
 362                         die("cannot read %s", sha1_to_hex(sha1));
 363                 write_header(sha1, TYPEFLAG_AUTO, basedir, prefix, path,
 364                              mode, eltbuf, eltsize);
 365                 if (!strcmp(elttype, "tree")) {
 366                         this_prefix.name = path;
 367                         traverse_tree(eltbuf, eltsize, &this_prefix);
 368                 } else if (!strcmp(elttype, "blob") && !S_ISLNK(mode)) {
 369                         write_blocked(eltbuf, eltsize);
 370                 }
 371                 free(eltbuf);
 372         }
 373 }
 374
 375 /* get commit time from committer line of commit object */
 376 time_t commit_time(void * buffer, unsigned long size)
 377 {
 378         time_t result = 0;
 379         char *p = buffer;
 380
 381         while (size > 0) {
 382                 char *endp = memchr(p, '\n', size);
 383                 if (!endp || endp == p)
 384                         break;
 385                 *endp = '\0';
 386                 if (endp - p > 10 && !memcmp(p, "committer ", 10)) {
 387                         char *nump = strrchr(p, '>');
 388                         if (!nump)
 389                                 break;
 390                         nump++;
 391                         result = strtoul(nump, &endp, 10);
 392                         if (*endp != ' ')
 393                                 result = 0;
 394                         break;
 395                 }
 396                 size -= endp - p - 1;
 397                 p = endp + 1;
 398         }
 399         return result;
 400 }
 401
 402 int main(int argc, char **argv)
 403 {
 404         unsigned char sha1[20];
 405         unsigned char commit_sha1[20];
 406         void *buffer;
 407         unsigned long size;
 408
 409         switch (argc) {
 410         case 3:
 411                 basedir = argv[2];
 412                 /* FALLTHROUGH */
 413         case 2:
 414                 if (get_sha1(argv[1], sha1) < 0)
 415                         usage(tar_tree_usage);
 416                 break;
 417         default:
 418                 usage(tar_tree_usage);
 419         }
 420
 421         sha1_file_directory = getenv(DB_ENVIRONMENT);
 422         if (!sha1_file_directory)
 423                 sha1_file_directory = DEFAULT_DB_ENVIRONMENT;
 424
 425         buffer = read_object_with_reference(sha1, "commit", &size, commit_sha1);
 426         if (buffer) {
 427                 write_global_extended_header(commit_sha1);
 428                 archive_time = commit_time(buffer, size);
 429                 free(buffer);
 430         }
 431         buffer = read_object_with_reference(sha1, "tree", &size, NULL);
 432         if (!buffer)
 433                 die("not a reference to a tag, commit or tree object: %s",
 434                     sha1_to_hex(sha1));
 435         if (!archive_time)
 436                 archive_time = time(NULL);
 437         if (basedir)
 438                 write_header("0", TYPEFLAG_DIR, NULL, NULL, basedir, 040755,
 439                              NULL, 0);
 440         traverse_tree(buffer, size, NULL);
 441         free(buffer);
 442         write_trailer();
 443         return 0;
 444 }