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_path_prefix(char **buffer, struct path_prefix *prefix)
 132 {
 133         if (!prefix)
 134                 return;
 135         append_path_prefix(buffer, prefix->prev);
 136         append_string(buffer, prefix->name);
 137         append_char(buffer, '/');
 138 }
 139
 140 static unsigned int path_prefix_len(struct path_prefix *prefix)
 141 {
 142         if (!prefix)
 143                 return 0;
 144         return path_prefix_len(prefix->prev) + strlen(prefix->name) + 1;
 145 }
 146
 147 static void append_path(char **p, int is_dir, const char *basepath,
 148                         struct path_prefix *prefix, const char *path)
 149 {
 150         if (basepath) {
 151                 append_string(p, basepath);
 152                 append_char(p, '/');
 153         }
 154         append_path_prefix(p, prefix);
 155         append_string(p, path);
 156         if (is_dir)
 157                 append_char(p, '/');
 158 }
 159
 160 static unsigned int path_len(int is_dir, const char *basepath,
 161                              struct path_prefix *prefix, const char *path)
 162 {
 163         unsigned int len = 0;
 164         if (basepath)
 165                 len += strlen(basepath) + 1;
 166         len += path_prefix_len(prefix) + strlen(path);
 167         if (is_dir)
 168                 len++;
 169         return len;
 170 }
 171
 172 static void append_extended_header_prefix(char **p, unsigned int size,
 173                                           const char *keyword)
 174 {
 175         int len = sprintf(*p, "%u %s=", size, keyword);
 176         *p += len;
 177 }
 178
 179 static unsigned int extended_header_len(const char *keyword,
 180                                         unsigned int valuelen)
 181 {
 182         /* "%u %s=%s\n" */
 183         unsigned int len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
 184         if (len > 9)
 185                 len++;
 186         if (len > 99)
 187                 len++;
 188         return len;
 189 }
 190
 191 static void append_extended_header(char **p, const char *keyword,
 192                                    const char *value, unsigned int len)
 193 {
 194         unsigned int size = extended_header_len(keyword, len);
 195         append_extended_header_prefix(p, size, keyword);
 196         memcpy(*p, value, len);
 197         *p += len;
 198         append_char(p, '\n');
 199 }
 200
 201 static void write_header(const char *, char, const char *, struct path_prefix *,
 202                          const char *, unsigned int, void *, unsigned long);
 203
 204 /* stores a pax extended header directly in the block buffer */
 205 static void write_extended_header(const char *headerfilename, int is_dir,
 206                                   unsigned int flags, const char *basepath,
 207                                   struct path_prefix *prefix,
 208                                   const char *path, unsigned int namelen,
 209                                   void *content, unsigned int contentsize)
 210 {
 211         char *p;
 212         unsigned int pathlen, size;
 213
 214         size = pathlen = extended_header_len("path", namelen);
 215         if (size > RECORDSIZE)
 216                 die("tar-tree: extended header too big, wtf?");
 217         write_header(NULL, TYPEFLAG_EXT_HEADER, NULL, NULL, headerfilename,
 218                      0100600, NULL, size);
 219
 220         p = get_record();
 221         append_extended_header_prefix(&p, pathlen, "path");
 222         append_path(&p, is_dir, basepath, prefix, path);
 223         append_char(&p, '\n');
 224         write_if_needed();
 225 }
 226
 227 static void write_global_extended_header(const char *sha1)
 228 {
 229         char *p;
 230         unsigned int size;
 231
 232         size = extended_header_len("comment", 40);
 233         write_header(NULL, TYPEFLAG_GLOBAL_HEADER, NULL, NULL,
 234                      "pax_global_header", 0100600, NULL, size);
 235
 236         p = get_record();
 237         append_extended_header(&p, "comment", sha1_to_hex(sha1), 40);
 238         write_if_needed();
 239 }
 240
 241 /* stores a ustar header directly in the block buffer */
 242 static void write_header(const char *sha1, char typeflag, const char *basepath,
 243                          struct path_prefix *prefix, const char *path,
 244                          unsigned int mode, void *buffer, unsigned long size)
 245 {
 246         unsigned int namelen;
 247         char *p, *header = NULL;
 248         unsigned int checksum = 0;
 249         int i;
 250
 251         if (typeflag == TYPEFLAG_AUTO) {
 252                 if (S_ISDIR(mode))
 253                         typeflag = TYPEFLAG_DIR;
 254                 else
 255                         typeflag = TYPEFLAG_REG;
 256         }
 257
 258         namelen = path_len(S_ISDIR(mode), basepath, prefix, path);
 259         if (namelen > 500) {
 260                 die("tar-tree: name too log of object %s\n", sha1_to_hex(sha1));
 261         } else if (namelen > 100) {
 262                 char *sha1_hex = sha1_to_hex(sha1);
 263                 char headerfilename[51];
 264                 sprintf(headerfilename, "%s.paxheader", sha1_hex);
 265                 /* the extended header must be written before the normal one */
 266                 write_extended_header(headerfilename, S_ISDIR(mode),
 267                                       0, basepath, prefix, path,
 268                                       namelen, buffer, size);
 269
 270                 header = get_record();
 271                 sprintf(header, "%s.data", sha1_hex);
 272         } else {
 273                 p = header = get_record();
 274                 append_path(&p, S_ISDIR(mode), basepath, prefix, path);
 275         }
 276
 277         if (S_ISDIR(mode))
 278                 mode |= 0755;   /* GIT doesn't store permissions of dirs */
 279         sprintf(&header[100], "%07o", mode & 07777);
 280
 281         /* XXX: should we provide more meaningful info here? */
 282         sprintf(&header[108], "%07o", 0);       /* uid */
 283         sprintf(&header[116], "%07o", 0);       /* gid */
 284         strncpy(&header[265], "git", 31);       /* uname */
 285         strncpy(&header[297], "git", 31);       /* gname */
 286
 287         sprintf(&header[124], "%011lo", S_ISDIR(mode) ? 0 : size);
 288         sprintf(&header[136], "%011lo", archive_time);
 289
 290         header[156] = typeflag;
 291
 292         memcpy(&header[257], "ustar", 6);
 293         memcpy(&header[263], "00", 2);
 294
 295         printf(&header[329], "%07o", 0);        /* devmajor */
 296         printf(&header[337], "%07o", 0);        /* devminor */
 297
 298         memset(&header[148], ' ', 8);
 299         for (i = 0; i < RECORDSIZE; i++)
 300                 checksum += header[i];
 301         sprintf(&header[148], "%07o", checksum & 0x1fffff);
 302
 303         write_if_needed();
 304 }
 305
 306 static void traverse_tree(void *buffer, unsigned long size,
 307                           struct path_prefix *prefix)
 308 {
 309         struct path_prefix this_prefix;
 310         this_prefix.prev = prefix;
 311
 312         while (size) {
 313                 int namelen = strlen(buffer)+1;
 314                 void *eltbuf;
 315                 char elttype[20];
 316                 unsigned long eltsize;
 317                 unsigned char *sha1 = buffer + namelen;
 318                 char *path = strchr(buffer, ' ') + 1;
 319                 unsigned int mode;
 320
 321                 if (size < namelen + 20 || sscanf(buffer, "%o", &mode) != 1)
 322                         die("corrupt 'tree' file");
 323                 buffer = sha1 + 20;
 324                 size -= namelen + 20;
 325
 326                 eltbuf = read_sha1_file(sha1, elttype, &eltsize);
 327                 if (!eltbuf)
 328                         die("cannot read %s", sha1_to_hex(sha1));
 329                 write_header(sha1, TYPEFLAG_AUTO, basedir, prefix, path,
 330                              mode, eltbuf, eltsize);
 331                 if (!strcmp(elttype, "tree")) {
 332                         this_prefix.name = path;
 333                         traverse_tree(eltbuf, eltsize, &this_prefix);
 334                 } else if (!strcmp(elttype, "blob")) {
 335                         write_blocked(eltbuf, eltsize);
 336                 }
 337                 free(eltbuf);
 338         }
 339 }
 340
 341 /* get commit time from committer line of commit object */
 342 time_t commit_time(void * buffer, unsigned long size)
 343 {
 344         time_t result = 0;
 345         char *p = buffer;
 346
 347         while (size > 0) {
 348                 char *endp = memchr(p, '\n', size);
 349                 if (!endp || endp == p)
 350                         break;
 351                 *endp = '\0';
 352                 if (endp - p > 10 && !memcmp(p, "committer ", 10)) {
 353                         char *nump = strrchr(p, '>');
 354                         if (!nump)
 355                                 break;
 356                         nump++;
 357                         result = strtoul(nump, &endp, 10);
 358                         if (*endp != ' ')
 359                                 result = 0;
 360                         break;
 361                 }
 362                 size -= endp - p - 1;
 363                 p = endp + 1;
 364         }
 365         return result;
 366 }
 367
 368 int main(int argc, char **argv)
 369 {
 370         unsigned char sha1[20];
 371         unsigned char commit_sha1[20];
 372         void *buffer;
 373         unsigned long size;
 374
 375         switch (argc) {
 376         case 3:
 377                 basedir = argv[2];
 378                 /* FALLTHROUGH */
 379         case 2:
 380                 if (get_sha1(argv[1], sha1) < 0)
 381                         usage(tar_tree_usage);
 382                 break;
 383         default:
 384                 usage(tar_tree_usage);
 385         }
 386
 387         sha1_file_directory = getenv(DB_ENVIRONMENT);
 388         if (!sha1_file_directory)
 389                 sha1_file_directory = DEFAULT_DB_ENVIRONMENT;
 390
 391         buffer = read_object_with_reference(sha1, "commit", &size, commit_sha1);
 392         if (buffer) {
 393                 write_global_extended_header(commit_sha1);
 394                 archive_time = commit_time(buffer, size);
 395                 free(buffer);
 396         }
 397         buffer = read_object_with_reference(sha1, "tree", &size, NULL);
 398         if (!buffer)
 399                 die("not a reference to a tag, commit or tree object: %s",
 400                     sha1_to_hex(sha1));
 401         if (!archive_time)
 402                 archive_time = time(NULL);
 403         if (basedir)
 404                 write_header("0", TYPEFLAG_DIR, NULL, NULL, basedir, 040755,
 405                              NULL, 0);
 406         traverse_tree(buffer, size, NULL);
 407         free(buffer);
 408         write_trailer();
 409         return 0;
 410 }