tar-tree.c

   1 #include <time.h>
   2 #include "cache.h"
   3
   4 #define RECORDSIZE      (512)
   5 #define BLOCKSIZE       (RECORDSIZE * 20)
   6
   7 static const char *tar_tree_usage = "tar-tree <key> [basedir]";
   8
   9 static char block[BLOCKSIZE];
  10 static unsigned long offset;
  11
  12 static const char *basedir;
  13 static time_t archive_time;
  14
  15 struct path_prefix {
  16         struct path_prefix *prev;
  17         const char *name;
  18 };
  19
  20 /* tries hard to write, either succeeds or dies in the attempt */
  21 static void reliable_write(void *buf, unsigned long size)
  22 {
  23         while (size > 0) {
  24                 long ret = write(1, buf, size);
  25                 if (ret < 0) {
  26                         if (errno == EAGAIN)
  27                                 continue;
  28                         if (errno == EPIPE)
  29                                 exit(0);
  30                         die("tar-tree: %s", strerror(errno));
  31                 } else if (!ret) {
  32                         die("tar-tree: disk full?");
  33                 }
  34                 size -= ret;
  35                 buf += ret;
  36         }
  37 }
  38
  39 /* writes out the whole block, but only if it is full */
  40 static void write_if_needed(void)
  41 {
  42         if (offset == BLOCKSIZE) {
  43                 reliable_write(block, BLOCKSIZE);
  44                 offset = 0;
  45         }
  46 }
  47
  48 /*
  49  * The end of tar archives is marked by 1024 nul bytes and after that
  50  * follows the rest of the block (if any).
  51  */
  52 static void write_trailer(void)
  53 {
  54         memset(block + offset, 0, RECORDSIZE);
  55         offset += RECORDSIZE;
  56         write_if_needed();
  57         memset(block + offset, 0, RECORDSIZE);
  58         offset += RECORDSIZE;
  59         write_if_needed();
  60         if (offset) {
  61                 memset(block + offset, 0, BLOCKSIZE - offset);
  62                 reliable_write(block, BLOCKSIZE);
  63                 offset = 0;
  64         }
  65 }
  66
  67 /*
  68  * queues up writes, so that all our write(2) calls write exactly one
  69  * full block; pads writes to RECORDSIZE
  70  */
  71 static void write_blocked(void *buf, unsigned long size)
  72 {
  73         unsigned long tail;
  74
  75         if (offset) {
  76                 unsigned long chunk = BLOCKSIZE - offset;
  77                 if (size < chunk)
  78                         chunk = size;
  79                 memcpy(block + offset, buf, chunk);
  80                 size -= chunk;
  81                 offset += chunk;
  82                 buf += chunk;
  83                 write_if_needed();
  84         }
  85         while (size >= BLOCKSIZE) {
  86                 reliable_write(buf, BLOCKSIZE);
  87                 size -= BLOCKSIZE;
  88                 buf += BLOCKSIZE;
  89         }
  90         if (size) {
  91                 memcpy(block + offset, buf, size);
  92                 buf += size;
  93                 offset += size;
  94         }
  95         tail = offset % RECORDSIZE;
  96         if (tail)  {
  97                 memset(block + offset, 0, RECORDSIZE - tail);
  98                 offset += RECORDSIZE - tail;
  99         }
 100         write_if_needed();
 101 }
 102
 103 static void append_string(char **p, const char *s)
 104 {
 105         unsigned int len = strlen(s);
 106         memcpy(*p, s, len);
 107         *p += len;
 108 }
 109
 110 static void append_char(char **p, char c)
 111 {
 112         **p = c;
 113         *p += 1;
 114 }
 115
 116 static void append_long(char **p, long n)
 117 {
 118         int len = sprintf(*p, "%ld", n);
 119         *p += len;
 120 }
 121
 122 static void append_path_prefix(char **buffer, struct path_prefix *prefix)
 123 {
 124         if (!prefix)
 125                 return;
 126         append_path_prefix(buffer, prefix->prev);
 127         append_string(buffer, prefix->name);
 128         append_char(buffer, '/');
 129 }
 130
 131 static unsigned int path_prefix_len(struct path_prefix *prefix)
 132 {
 133         if (!prefix)
 134                 return 0;
 135         return path_prefix_len(prefix->prev) + strlen(prefix->name) + 1;
 136 }
 137
 138 static void append_path(char **p, int is_dir, const char *basepath,
 139                         struct path_prefix *prefix, const char *path)
 140 {
 141         if (basepath) {
 142                 append_string(p, basepath);
 143                 append_char(p, '/');
 144         }
 145         append_path_prefix(p, prefix);
 146         append_string(p, path);
 147         if (is_dir)
 148                 append_char(p, '/');
 149 }
 150
 151 static unsigned int path_len(int is_dir, const char *basepath,
 152                              struct path_prefix *prefix, const char *path)
 153 {
 154         unsigned int len = 0;
 155         if (basepath)
 156                 len += strlen(basepath) + 1;
 157         len += path_prefix_len(prefix) + strlen(path);
 158         if (is_dir)
 159                 len++;
 160         return len;
 161 }
 162
 163 static void write_header(const char *, const char *, struct path_prefix *,
 164                          const char *, unsigned int, unsigned long);
 165
 166 /* stores a pax extended header directly in the block buffer */
 167 static void write_extended_header(const char *headerfilename, int is_dir,
 168                                   const char *basepath,
 169                                   struct path_prefix *prefix,
 170                                   const char *path, unsigned int namelen)
 171 {
 172         char *records, *p;
 173         unsigned int size = 1 + 6 + namelen + 1;
 174         if (size > 9)
 175                 size++;
 176         if (size > 99)
 177                 size++;
 178         if (size > RECORDSIZE)
 179                 die("tar-tree: extended header too big, wtf?");
 180         write_header(NULL, NULL, NULL, headerfilename, 0100600, size);
 181
 182         records = block + offset;
 183         memset(records, 0, RECORDSIZE);
 184         offset += RECORDSIZE;
 185         p = records;
 186         append_long(&p, size);
 187         append_string(&p, " path=");
 188         append_path(&p, is_dir, basepath, prefix, path);
 189         append_char(&p, '\n');
 190         write_if_needed();
 191 }
 192
 193 /* stores a ustar header directly in the block buffer */
 194 static void write_header(const char *sha1, const char *basepath,
 195                          struct path_prefix *prefix, const char *path,
 196                          unsigned int mode, unsigned long size)
 197 {
 198         unsigned int namelen;
 199         char *p, *header = NULL;
 200         unsigned int checksum = 0;
 201         int i;
 202
 203         namelen = path_len(S_ISDIR(mode), basepath, prefix, path);
 204         if (namelen > 500) {
 205                 die("tar-tree: name too log of object %s\n", sha1_to_hex(sha1));
 206         } else if (namelen > 100) {
 207                 char *sha1_hex = sha1_to_hex(sha1);
 208                 char headerfilename[51];
 209                 sprintf(headerfilename, "%s.paxheader", sha1_hex);
 210                 /* the extended header must be written before the normal one */
 211                 write_extended_header(headerfilename, S_ISDIR(mode), basepath,
 212                                       prefix, path, namelen);
 213
 214                 header = block + offset;
 215                 memset(header, 0, RECORDSIZE);
 216                 offset += RECORDSIZE;
 217                 sprintf(header, "%s.data", sha1_hex);
 218         } else {
 219                 header = block + offset;
 220                 memset(header, 0, RECORDSIZE);
 221                 offset += RECORDSIZE;
 222                 p = header;
 223                 append_path(&p, S_ISDIR(mode), basepath, prefix, path);
 224         }
 225
 226         if (S_ISDIR(mode))
 227                 mode |= 0755;   /* GIT doesn't store permissions of dirs */
 228         sprintf(&header[100], "%07o", mode & 07777);
 229
 230         /* XXX: should we provide more meaningful info here? */
 231         sprintf(&header[108], "%07o", 0);       /* uid */
 232         sprintf(&header[116], "%07o", 0);       /* gid */
 233         strncpy(&header[265], "git", 31);       /* uname */
 234         strncpy(&header[297], "git", 31);       /* gname */
 235
 236         sprintf(&header[124], "%011lo", S_ISDIR(mode) ? 0 : size);
 237         sprintf(&header[136], "%011lo", archive_time);
 238
 239         /* typeflag */
 240         if (!sha1)
 241                 header[156] = 'x';      /* extended header */
 242         else
 243                 header[156] = S_ISDIR(mode) ? '5' : '0';
 244
 245         memcpy(&header[257], "ustar", 6);
 246         memcpy(&header[263], "00", 2);
 247
 248         printf(&header[329], "%07o", 0);        /* devmajor */
 249         printf(&header[337], "%07o", 0);        /* devminor */
 250
 251         memset(&header[148], ' ', 8);
 252         for (i = 0; i < RECORDSIZE; i++)
 253                 checksum += header[i];
 254         sprintf(&header[148], "%07o", checksum & 0x1fffff);
 255
 256         write_if_needed();
 257 }
 258
 259 static void traverse_tree(void *buffer, unsigned long size,
 260                           struct path_prefix *prefix)
 261 {
 262         struct path_prefix this_prefix;
 263         this_prefix.prev = prefix;
 264
 265         while (size) {
 266                 int namelen = strlen(buffer)+1;
 267                 void *eltbuf;
 268                 char elttype[20];
 269                 unsigned long eltsize;
 270                 unsigned char *sha1 = buffer + namelen;
 271                 char *path = strchr(buffer, ' ') + 1;
 272                 unsigned int mode;
 273
 274                 if (size < namelen + 20 || sscanf(buffer, "%o", &mode) != 1)
 275                         die("corrupt 'tree' file");
 276                 buffer = sha1 + 20;
 277                 size -= namelen + 20;
 278
 279                 eltbuf = read_sha1_file(sha1, elttype, &eltsize);
 280                 if (!eltbuf)
 281                         die("cannot read %s", sha1_to_hex(sha1));
 282                 write_header(sha1, basedir, prefix, path, mode, eltsize);
 283                 if (!strcmp(elttype, "tree")) {
 284                         this_prefix.name = path;
 285                         traverse_tree(eltbuf, eltsize, &this_prefix);
 286                 } else if (!strcmp(elttype, "blob")) {
 287                         write_blocked(eltbuf, eltsize);
 288                 }
 289                 free(eltbuf);
 290         }
 291 }
 292
 293 /* get commit time from committer line of commit object */
 294 time_t commit_time(void * buffer, unsigned long size)
 295 {
 296         time_t result = 0;
 297         char *p = buffer;
 298
 299         while (size > 0) {
 300                 char *endp = memchr(p, '\n', size);
 301                 if (!endp || endp == p)
 302                         break;
 303                 *endp = '\0';
 304                 if (endp - p > 10 && !memcmp(p, "committer ", 10)) {
 305                         char *nump = strrchr(p, '>');
 306                         if (!nump)
 307                                 break;
 308                         nump++;
 309                         result = strtoul(nump, &endp, 10);
 310                         if (*endp != ' ')
 311                                 result = 0;
 312                         break;
 313                 }
 314                 size -= endp - p - 1;
 315                 p = endp + 1;
 316         }
 317         return result;
 318 }
 319
 320 int main(int argc, char **argv)
 321 {
 322         unsigned char sha1[20];
 323         void *buffer;
 324         unsigned long size;
 325
 326         switch (argc) {
 327         case 3:
 328                 basedir = argv[2];
 329                 /* FALLTHROUGH */
 330         case 2:
 331                 if (get_sha1_hex(argv[1], sha1) < 0)
 332                         usage(tar_tree_usage);
 333                 break;
 334         default:
 335                 usage(tar_tree_usage);
 336         }
 337
 338         sha1_file_directory = getenv(DB_ENVIRONMENT);
 339         if (!sha1_file_directory)
 340                 sha1_file_directory = DEFAULT_DB_ENVIRONMENT;
 341
 342         buffer = read_object_with_reference(sha1, "commit", &size, NULL);
 343         if (buffer) {
 344                 archive_time = commit_time(buffer, size);
 345                 free(buffer);
 346         }
 347         buffer = read_object_with_reference(sha1, "tree", &size, NULL);
 348         if (!buffer)
 349                 die("not a reference to a tag, commit or tree object: %s",
 350                     sha1_to_hex(sha1));
 351         if (!archive_time)
 352                 archive_time = time(NULL);
 353         if (basedir)
 354                 write_header("0", NULL, NULL, basedir, 040755, 0);
 355         traverse_tree(buffer, size, NULL);
 356         free(buffer);
 357         write_trailer();
 358         return 0;
 359 }