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_DIR            '5'
  10 #define TYPEFLAG_GLOBAL_HEADER  'g'
  11 #define TYPEFLAG_EXT_HEADER     'x'
  12
  13 static const char *tar_tree_usage = "tar-tree <key> [basedir]";
  14
  15 static char block[BLOCKSIZE];
  16 static unsigned long offset;
  17
  18 static const char *basedir;
  19 static time_t archive_time;
  20
  21 struct path_prefix {
  22         struct path_prefix *prev;
  23         const char *name;
  24 };
  25
  26 /* tries hard to write, either succeeds or dies in the attempt */
  27 static void reliable_write(void *buf, unsigned long size)
  28 {
  29         while (size > 0) {
  30                 long ret = write(1, buf, size);
  31                 if (ret < 0) {
  32                         if (errno == EAGAIN)
  33                                 continue;
  34                         if (errno == EPIPE)
  35                                 exit(0);
  36                         die("tar-tree: %s", strerror(errno));
  37                 } else if (!ret) {
  38                         die("tar-tree: disk full?");
  39                 }
  40                 size -= ret;
  41                 buf += ret;
  42         }
  43 }
  44
  45 /* writes out the whole block, but only if it is full */
  46 static void write_if_needed(void)
  47 {
  48         if (offset == BLOCKSIZE) {
  49                 reliable_write(block, BLOCKSIZE);
  50                 offset = 0;
  51         }
  52 }
  53
  54 /* acquire the next record from the buffer; user must call write_if_needed() */
  55 static char *get_record(void)
  56 {
  57         char *p = block + offset;
  58         memset(p, 0, RECORDSIZE);
  59         offset += RECORDSIZE;
  60         return p;
  61 }
  62
  63 /*
  64  * The end of tar archives is marked by 1024 nul bytes and after that
  65  * follows the rest of the block (if any).
  66  */
  67 static void write_trailer(void)
  68 {
  69         memset(block + offset, 0, RECORDSIZE);
  70         offset += RECORDSIZE;
  71         write_if_needed();
  72         memset(block + offset, 0, RECORDSIZE);
  73         offset += RECORDSIZE;
  74         write_if_needed();
  75         if (offset) {
  76                 memset(block + offset, 0, BLOCKSIZE - offset);
  77                 reliable_write(block, BLOCKSIZE);
  78                 offset = 0;
  79         }
  80 }
  81
  82 /*
  83  * queues up writes, so that all our write(2) calls write exactly one
  84  * full block; pads writes to RECORDSIZE
  85  */
  86 static void write_blocked(void *buf, unsigned long size)
  87 {
  88         unsigned long tail;
  89
  90         if (offset) {
  91                 unsigned long chunk = BLOCKSIZE - offset;
  92                 if (size < chunk)
  93                         chunk = size;
  94                 memcpy(block + offset, buf, chunk);
  95                 size -= chunk;
  96                 offset += chunk;
  97                 buf += chunk;
  98                 write_if_needed();
  99         }
 100         while (size >= BLOCKSIZE) {
 101                 reliable_write(buf, BLOCKSIZE);
 102                 size -= BLOCKSIZE;
 103                 buf += BLOCKSIZE;
 104         }
 105         if (size) {
 106                 memcpy(block + offset, buf, size);
 107                 buf += size;
 108                 offset += size;
 109         }
 110         tail = offset % RECORDSIZE;
 111         if (tail)  {
 112                 memset(block + offset, 0, RECORDSIZE - tail);
 113                 offset += RECORDSIZE - tail;
 114         }
 115         write_if_needed();
 116 }
 117
 118 static void append_string(char **p, const char *s)
 119 {
 120         unsigned int len = strlen(s);
 121         memcpy(*p, s, len);
 122         *p += len;
 123 }
 124
 125 static void append_char(char **p, char c)
 126 {
 127         **p = c;
 128         *p += 1;
 129 }
 130
 131 static void append_long(char **p, long n)
 132 {
 133         int len = sprintf(*p, "%ld", n);
 134         *p += len;
 135 }
 136
 137 static void append_path_prefix(char **buffer, struct path_prefix *prefix)
 138 {
 139         if (!prefix)
 140                 return;
 141         append_path_prefix(buffer, prefix->prev);
 142         append_string(buffer, prefix->name);
 143         append_char(buffer, '/');
 144 }
 145
 146 static unsigned int path_prefix_len(struct path_prefix *prefix)
 147 {
 148         if (!prefix)
 149                 return 0;
 150         return path_prefix_len(prefix->prev) + strlen(prefix->name) + 1;
 151 }
 152
 153 static void append_path(char **p, int is_dir, const char *basepath,
 154                         struct path_prefix *prefix, const char *path)
 155 {
 156         if (basepath) {
 157                 append_string(p, basepath);
 158                 append_char(p, '/');
 159         }
 160         append_path_prefix(p, prefix);
 161         append_string(p, path);
 162         if (is_dir)
 163                 append_char(p, '/');
 164 }
 165
 166 static unsigned int path_len(int is_dir, const char *basepath,
 167                              struct path_prefix *prefix, const char *path)
 168 {
 169         unsigned int len = 0;
 170         if (basepath)
 171                 len += strlen(basepath) + 1;
 172         len += path_prefix_len(prefix) + strlen(path);
 173         if (is_dir)
 174                 len++;
 175         return len;
 176 }
 177
 178 static void write_header(const char *, char, const char *, struct path_prefix *,
 179                          const char *, unsigned int, unsigned long);
 180
 181 /* stores a pax extended header directly in the block buffer */
 182 static void write_extended_header(const char *headerfilename, int is_dir,
 183                                   const char *basepath,
 184                                   struct path_prefix *prefix,
 185                                   const char *path, unsigned int namelen)
 186 {
 187         char *p;
 188         unsigned int size = 1 + 6 + namelen + 1;
 189         if (size > 9)
 190                 size++;
 191         if (size > 99)
 192                 size++;
 193         if (size > RECORDSIZE)
 194                 die("tar-tree: extended header too big, wtf?");
 195         write_header(NULL, TYPEFLAG_EXT_HEADER, NULL, NULL, headerfilename,
 196                      0100600, size);
 197         p = get_record();
 198         append_long(&p, size);
 199         append_string(&p, " path=");
 200         append_path(&p, is_dir, basepath, prefix, path);
 201         append_char(&p, '\n');
 202         write_if_needed();
 203 }
 204
 205 static void write_global_extended_header(const char *sha1)
 206 {
 207         char *p;
 208         write_header(NULL, TYPEFLAG_GLOBAL_HEADER, NULL, NULL,
 209                      "pax_global_header", 0100600, 52);
 210         p = get_record();
 211         append_long(&p, 52);    /* 2 + 9 + 40 + 1 */
 212         append_string(&p, " comment=");
 213         append_string(&p, sha1_to_hex(sha1));
 214         append_char(&p, '\n');
 215         write_if_needed();
 216 }
 217
 218 /* stores a ustar header directly in the block buffer */
 219 static void write_header(const char *sha1, char typeflag, const char *basepath,
 220                          struct path_prefix *prefix, const char *path,
 221                          unsigned int mode, unsigned long size)
 222 {
 223         unsigned int namelen;
 224         char *p, *header = NULL;
 225         unsigned int checksum = 0;
 226         int i;
 227
 228         if (typeflag == TYPEFLAG_AUTO) {
 229                 if (S_ISDIR(mode))
 230                         typeflag = TYPEFLAG_DIR;
 231                 else
 232                         typeflag = TYPEFLAG_REG;
 233         }
 234
 235         namelen = path_len(S_ISDIR(mode), basepath, prefix, path);
 236         if (namelen > 500) {
 237                 die("tar-tree: name too log of object %s\n", sha1_to_hex(sha1));
 238         } else if (namelen > 100) {
 239                 char *sha1_hex = sha1_to_hex(sha1);
 240                 char headerfilename[51];
 241                 sprintf(headerfilename, "%s.paxheader", sha1_hex);
 242                 /* the extended header must be written before the normal one */
 243                 write_extended_header(headerfilename, S_ISDIR(mode), basepath,
 244                                       prefix, path, namelen);
 245
 246                 header = get_record();
 247                 sprintf(header, "%s.data", sha1_hex);
 248         } else {
 249                 p = header = get_record();
 250                 append_path(&p, S_ISDIR(mode), basepath, prefix, path);
 251         }
 252
 253         if (S_ISDIR(mode))
 254                 mode |= 0755;   /* GIT doesn't store permissions of dirs */
 255         sprintf(&header[100], "%07o", mode & 07777);
 256
 257         /* XXX: should we provide more meaningful info here? */
 258         sprintf(&header[108], "%07o", 0);       /* uid */
 259         sprintf(&header[116], "%07o", 0);       /* gid */
 260         strncpy(&header[265], "git", 31);       /* uname */
 261         strncpy(&header[297], "git", 31);       /* gname */
 262
 263         sprintf(&header[124], "%011lo", S_ISDIR(mode) ? 0 : size);
 264         sprintf(&header[136], "%011lo", archive_time);
 265
 266         header[156] = typeflag;
 267
 268         memcpy(&header[257], "ustar", 6);
 269         memcpy(&header[263], "00", 2);
 270
 271         printf(&header[329], "%07o", 0);        /* devmajor */
 272         printf(&header[337], "%07o", 0);        /* devminor */
 273
 274         memset(&header[148], ' ', 8);
 275         for (i = 0; i < RECORDSIZE; i++)
 276                 checksum += header[i];
 277         sprintf(&header[148], "%07o", checksum & 0x1fffff);
 278
 279         write_if_needed();
 280 }
 281
 282 static void traverse_tree(void *buffer, unsigned long size,
 283                           struct path_prefix *prefix)
 284 {
 285         struct path_prefix this_prefix;
 286         this_prefix.prev = prefix;
 287
 288         while (size) {
 289                 int namelen = strlen(buffer)+1;
 290                 void *eltbuf;
 291                 char elttype[20];
 292                 unsigned long eltsize;
 293                 unsigned char *sha1 = buffer + namelen;
 294                 char *path = strchr(buffer, ' ') + 1;
 295                 unsigned int mode;
 296
 297                 if (size < namelen + 20 || sscanf(buffer, "%o", &mode) != 1)
 298                         die("corrupt 'tree' file");
 299                 buffer = sha1 + 20;
 300                 size -= namelen + 20;
 301
 302                 eltbuf = read_sha1_file(sha1, elttype, &eltsize);
 303                 if (!eltbuf)
 304                         die("cannot read %s", sha1_to_hex(sha1));
 305                 write_header(sha1, TYPEFLAG_AUTO, basedir, prefix, path,
 306                              mode, eltsize);
 307                 if (!strcmp(elttype, "tree")) {
 308                         this_prefix.name = path;
 309                         traverse_tree(eltbuf, eltsize, &this_prefix);
 310                 } else if (!strcmp(elttype, "blob")) {
 311                         write_blocked(eltbuf, eltsize);
 312                 }
 313                 free(eltbuf);
 314         }
 315 }
 316
 317 /* get commit time from committer line of commit object */
 318 time_t commit_time(void * buffer, unsigned long size)
 319 {
 320         time_t result = 0;
 321         char *p = buffer;
 322
 323         while (size > 0) {
 324                 char *endp = memchr(p, '\n', size);
 325                 if (!endp || endp == p)
 326                         break;
 327                 *endp = '\0';
 328                 if (endp - p > 10 && !memcmp(p, "committer ", 10)) {
 329                         char *nump = strrchr(p, '>');
 330                         if (!nump)
 331                                 break;
 332                         nump++;
 333                         result = strtoul(nump, &endp, 10);
 334                         if (*endp != ' ')
 335                                 result = 0;
 336                         break;
 337                 }
 338                 size -= endp - p - 1;
 339                 p = endp + 1;
 340         }
 341         return result;
 342 }
 343
 344 int main(int argc, char **argv)
 345 {
 346         unsigned char sha1[20];
 347         unsigned char commit_sha1[20];
 348         void *buffer;
 349         unsigned long size;
 350
 351         switch (argc) {
 352         case 3:
 353                 basedir = argv[2];
 354                 /* FALLTHROUGH */
 355         case 2:
 356                 if (get_sha1(argv[1], sha1) < 0)
 357                         usage(tar_tree_usage);
 358                 break;
 359         default:
 360                 usage(tar_tree_usage);
 361         }
 362
 363         sha1_file_directory = getenv(DB_ENVIRONMENT);
 364         if (!sha1_file_directory)
 365                 sha1_file_directory = DEFAULT_DB_ENVIRONMENT;
 366
 367         buffer = read_object_with_reference(sha1, "commit", &size, commit_sha1);
 368         if (buffer) {
 369                 write_global_extended_header(commit_sha1);
 370                 archive_time = commit_time(buffer, size);
 371                 free(buffer);
 372         }
 373         buffer = read_object_with_reference(sha1, "tree", &size, NULL);
 374         if (!buffer)
 375                 die("not a reference to a tag, commit or tree object: %s",
 376                     sha1_to_hex(sha1));
 377         if (!archive_time)
 378                 archive_time = time(NULL);
 379         if (basedir)
 380                 write_header("0", TYPEFLAG_DIR, NULL, NULL, basedir, 040755, 0);
 381         traverse_tree(buffer, size, NULL);
 382         free(buffer);
 383         write_trailer();
 384         return 0;
 385 }