http-fetch.c

   1 #include "cache.h"
   2 #include "commit.h"
   3
   4 #include "fetch.h"
   5
   6 #include <curl/curl.h>
   7 #include <curl/easy.h>
   8
   9 #if LIBCURL_VERSION_NUM < 0x070704
  10 #define curl_global_cleanup() do { /* nothing */ } while(0)
  11 #endif
  12 #if LIBCURL_VERSION_NUM < 0x070800
  13 #define curl_global_init(a) do { /* nothing */ } while(0)
  14 #endif
  15
  16 static CURL *curl;
  17 static struct curl_slist *no_pragma_header;
  18 static char curl_errorstr[CURL_ERROR_SIZE];
  19
  20 static char *initial_base;
  21
  22 struct alt_base
  23 {
  24         char *base;
  25         int got_indices;
  26         struct packed_git *packs;
  27         struct alt_base *next;
  28 };
  29
  30 struct alt_base *alt = NULL;
  31
  32 static SHA_CTX c;
  33 static z_stream stream;
  34
  35 static int local;
  36 static int zret;
  37
  38 static int curl_ssl_verify;
  39
  40 struct buffer
  41 {
  42         size_t posn;
  43         size_t size;
  44         void *buffer;
  45 };
  46
  47 static size_t fwrite_buffer(void *ptr, size_t eltsize, size_t nmemb,
  48                             struct buffer *buffer)
  49 {
  50         size_t size = eltsize * nmemb;
  51         if (size > buffer->size - buffer->posn)
  52                 size = buffer->size - buffer->posn;
  53         memcpy(buffer->buffer + buffer->posn, ptr, size);
  54         buffer->posn += size;
  55         return size;
  56 }
  57
  58 static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
  59                                void *data)
  60 {
  61         unsigned char expn[4096];
  62         size_t size = eltsize * nmemb;
  63         int posn = 0;
  64         do {
  65                 ssize_t retval = write(local, ptr + posn, size - posn);
  66                 if (retval < 0)
  67                         return posn;
  68                 posn += retval;
  69         } while (posn < size);
  70
  71         stream.avail_in = size;
  72         stream.next_in = ptr;
  73         do {
  74                 stream.next_out = expn;
  75                 stream.avail_out = sizeof(expn);
  76                 zret = inflate(&stream, Z_SYNC_FLUSH);
  77                 SHA1_Update(&c, expn, sizeof(expn) - stream.avail_out);
  78         } while (stream.avail_in && zret == Z_OK);
  79         return size;
  80 }
  81
  82 void prefetch(unsigned char *sha1)
  83 {
  84 }
  85
  86 static int got_alternates = 0;
  87
  88 static int fetch_index(struct alt_base *repo, unsigned char *sha1)
  89 {
  90         char *filename;
  91         char *url;
  92
  93         FILE *indexfile;
  94
  95         if (has_pack_index(sha1))
  96                 return 0;
  97
  98         if (get_verbosely)
  99                 fprintf(stderr, "Getting index for pack %s\n",
 100                         sha1_to_hex(sha1));
 101
 102         url = xmalloc(strlen(repo->base) + 64);
 103         sprintf(url, "%s/objects/pack/pack-%s.idx",
 104                 repo->base, sha1_to_hex(sha1));
 105
 106         filename = sha1_pack_index_name(sha1);
 107         indexfile = fopen(filename, "w");
 108         if (!indexfile)
 109                 return error("Unable to open local file %s for pack index",
 110                              filename);
 111
 112         curl_easy_setopt(curl, CURLOPT_FILE, indexfile);
 113         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite);
 114         curl_easy_setopt(curl, CURLOPT_URL, url);
 115         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 116         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 117
 118         if (curl_easy_perform(curl)) {
 119                 fclose(indexfile);
 120                 return error("Unable to get pack index %s\n%s", url,
 121                              curl_errorstr);
 122         }
 123
 124         fclose(indexfile);
 125         return 0;
 126 }
 127
 128 static int setup_index(struct alt_base *repo, unsigned char *sha1)
 129 {
 130         struct packed_git *new_pack;
 131         if (has_pack_file(sha1))
 132                 return 0; // don't list this as something we can get
 133
 134         if (fetch_index(repo, sha1))
 135                 return -1;
 136
 137         new_pack = parse_pack_index(sha1);
 138         new_pack->next = repo->packs;
 139         repo->packs = new_pack;
 140         return 0;
 141 }
 142
 143 static int fetch_alternates(char *base)
 144 {
 145         int ret = 0;
 146         struct buffer buffer;
 147         char *url;
 148         char *data;
 149         int i = 0;
 150         int http_specific = 1;
 151         if (got_alternates)
 152                 return 0;
 153         data = xmalloc(4096);
 154         buffer.size = 4095;
 155         buffer.posn = 0;
 156         buffer.buffer = data;
 157
 158         if (get_verbosely)
 159                 fprintf(stderr, "Getting alternates list\n");
 160
 161         url = xmalloc(strlen(base) + 31);
 162         sprintf(url, "%s/objects/info/http-alternates", base);
 163
 164         curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 165         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 166         curl_easy_setopt(curl, CURLOPT_URL, url);
 167
 168         if (curl_easy_perform(curl) || !buffer.posn) {
 169                 http_specific = 0;
 170
 171                 sprintf(url, "%s/objects/info/alternates", base);
 172
 173                 curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 174                 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 175                 curl_easy_setopt(curl, CURLOPT_URL, url);
 176
 177                 if (curl_easy_perform(curl)) {
 178                         return 0;
 179                 }
 180         }
 181
 182         data[buffer.posn] = '\0';
 183
 184         while (i < buffer.posn) {
 185                 int posn = i;
 186                 while (posn < buffer.posn && data[posn] != '\n')
 187                         posn++;
 188                 if (data[posn] == '\n') {
 189                         int okay = 0;
 190                         int serverlen = 0;
 191                         struct alt_base *newalt;
 192                         char *target = NULL;
 193                         if (data[i] == '/') {
 194                                 serverlen = strchr(base + 8, '/') - base;
 195                                 okay = 1;
 196                         } else if (!memcmp(data + i, "../", 3)) {
 197                                 i += 3;
 198                                 serverlen = strlen(base);
 199                                 while (i + 2 < posn &&
 200                                        !memcmp(data + i, "../", 3)) {
 201                                         do {
 202                                                 serverlen--;
 203                                         } while (serverlen &&
 204                                                  base[serverlen - 1] != '/');
 205                                         i += 3;
 206                                 }
 207                                 // If the server got removed, give up.
 208                                 okay = strchr(base, ':') - base + 3 <
 209                                         serverlen;
 210                         } else if (http_specific) {
 211                                 char *colon = strchr(data + i, ':');
 212                                 char *slash = strchr(data + i, '/');
 213                                 if (colon && slash && colon < data + posn &&
 214                                     slash < data + posn && colon < slash) {
 215                                         okay = 1;
 216                                 }
 217                         }
 218                         // skip 'objects' at end
 219                         if (okay) {
 220                                 target = xmalloc(serverlen + posn - i - 6);
 221                                 strncpy(target, base, serverlen);
 222                                 strncpy(target + serverlen, data + i,
 223                                         posn - i - 7);
 224                                 target[serverlen + posn - i - 7] = '\0';
 225                                 if (get_verbosely)
 226                                         fprintf(stderr,
 227                                                 "Also look at %s\n", target);
 228                                 newalt = xmalloc(sizeof(*newalt));
 229                                 newalt->next = alt;
 230                                 newalt->base = target;
 231                                 newalt->got_indices = 0;
 232                                 newalt->packs = NULL;
 233                                 alt = newalt;
 234                                 ret++;
 235                         }
 236                 }
 237                 i = posn + 1;
 238         }
 239         got_alternates = 1;
 240
 241         return ret;
 242 }
 243
 244 static int fetch_indices(struct alt_base *repo)
 245 {
 246         unsigned char sha1[20];
 247         char *url;
 248         struct buffer buffer;
 249         char *data;
 250         int i = 0;
 251
 252         if (repo->got_indices)
 253                 return 0;
 254
 255         data = xmalloc(4096);
 256         buffer.size = 4096;
 257         buffer.posn = 0;
 258         buffer.buffer = data;
 259
 260         if (get_verbosely)
 261                 fprintf(stderr, "Getting pack list\n");
 262
 263         url = xmalloc(strlen(repo->base) + 21);
 264         sprintf(url, "%s/objects/info/packs", repo->base);
 265
 266         curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 267         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 268         curl_easy_setopt(curl, CURLOPT_URL, url);
 269         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
 270         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 271
 272         if (curl_easy_perform(curl))
 273                 return error("%s", curl_errorstr);
 274
 275         while (i < buffer.posn) {
 276                 switch (data[i]) {
 277                 case 'P':
 278                         i++;
 279                         if (i + 52 < buffer.posn &&
 280                             !strncmp(data + i, " pack-", 6) &&
 281                             !strncmp(data + i + 46, ".pack\n", 6)) {
 282                                 get_sha1_hex(data + i + 6, sha1);
 283                                 setup_index(repo, sha1);
 284                                 i += 51;
 285                                 break;
 286                         }
 287                 default:
 288                         while (data[i] != '\n')
 289                                 i++;
 290                 }
 291                 i++;
 292         }
 293
 294         repo->got_indices = 1;
 295         return 0;
 296 }
 297
 298 static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 299 {
 300         char *url;
 301         struct packed_git *target;
 302         struct packed_git **lst;
 303         FILE *packfile;
 304         char *filename;
 305
 306         if (fetch_indices(repo))
 307                 return -1;
 308         target = find_sha1_pack(sha1, repo->packs);
 309         if (!target)
 310                 return -1;
 311
 312         if (get_verbosely) {
 313                 fprintf(stderr, "Getting pack %s\n",
 314                         sha1_to_hex(target->sha1));
 315                 fprintf(stderr, " which contains %s\n",
 316                         sha1_to_hex(sha1));
 317         }
 318
 319         url = xmalloc(strlen(repo->base) + 65);
 320         sprintf(url, "%s/objects/pack/pack-%s.pack",
 321                 repo->base, sha1_to_hex(target->sha1));
 322
 323         filename = sha1_pack_name(target->sha1);
 324         packfile = fopen(filename, "w");
 325         if (!packfile)
 326                 return error("Unable to open local file %s for pack",
 327                              filename);
 328
 329         curl_easy_setopt(curl, CURLOPT_FILE, packfile);
 330         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite);
 331         curl_easy_setopt(curl, CURLOPT_URL, url);
 332         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 333         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 334
 335         if (curl_easy_perform(curl)) {
 336                 fclose(packfile);
 337                 return error("Unable to get pack file %s\n%s", url,
 338                              curl_errorstr);
 339         }
 340
 341         fclose(packfile);
 342
 343         lst = &repo->packs;
 344         while (*lst != target)
 345                 lst = &((*lst)->next);
 346         *lst = (*lst)->next;
 347
 348         install_packed_git(target);
 349
 350         return 0;
 351 }
 352
 353 int fetch_object(struct alt_base *repo, unsigned char *sha1)
 354 {
 355         char *hex = sha1_to_hex(sha1);
 356         char *filename = sha1_file_name(sha1);
 357         unsigned char real_sha1[20];
 358         char tmpfile[PATH_MAX];
 359         int ret;
 360         char *url;
 361         char *posn;
 362
 363         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX",
 364                  get_object_directory());
 365
 366         local = mkstemp(tmpfile);
 367         if (local < 0)
 368                 return error("Couldn't create temporary file %s for %s: %s\n",
 369                              tmpfile, filename, strerror(errno));
 370
 371         memset(&stream, 0, sizeof(stream));
 372
 373         inflateInit(&stream);
 374
 375         SHA1_Init(&c);
 376
 377         curl_easy_setopt(curl, CURLOPT_FAILONERROR, 1);
 378         curl_easy_setopt(curl, CURLOPT_FILE, NULL);
 379         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 380         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 381         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 382
 383         url = xmalloc(strlen(repo->base) + 50);
 384         strcpy(url, repo->base);
 385         posn = url + strlen(repo->base);
 386         strcpy(posn, "objects/");
 387         posn += 8;
 388         memcpy(posn, hex, 2);
 389         posn += 2;
 390         *(posn++) = '/';
 391         strcpy(posn, hex + 2);
 392
 393         curl_easy_setopt(curl, CURLOPT_URL, url);
 394
 395         if (curl_easy_perform(curl)) {
 396                 unlink(filename);
 397                 return error("%s", curl_errorstr);
 398         }
 399
 400         fchmod(local, 0444);
 401         close(local);
 402         inflateEnd(&stream);
 403         SHA1_Final(real_sha1, &c);
 404         if (zret != Z_STREAM_END) {
 405                 unlink(tmpfile);
 406                 return error("File %s (%s) corrupt\n", hex, url);
 407         }
 408         if (memcmp(sha1, real_sha1, 20)) {
 409                 unlink(tmpfile);
 410                 return error("File %s has bad hash\n", hex);
 411         }
 412         ret = link(tmpfile, filename);
 413         if (ret < 0) {
 414                 /* Same Coda hack as in write_sha1_file(sha1_file.c) */
 415                 ret = errno;
 416                 if (ret == EXDEV && !rename(tmpfile, filename))
 417                         goto out;
 418         }
 419         unlink(tmpfile);
 420         if (ret) {
 421                 if (ret != EEXIST)
 422                         return error("unable to write sha1 filename %s: %s",
 423                                      filename, strerror(ret));
 424         }
 425  out:
 426         pull_say("got %s\n", hex);
 427         return 0;
 428 }
 429
 430 int fetch(unsigned char *sha1)
 431 {
 432         struct alt_base *altbase = alt;
 433         while (altbase) {
 434                 if (!fetch_object(altbase, sha1))
 435                         return 0;
 436                 if (!fetch_pack(altbase, sha1))
 437                         return 0;
 438                 if (fetch_alternates(altbase->base) > 0) {
 439                         altbase = alt;
 440                         continue;
 441                 }
 442                 altbase = altbase->next;
 443         }
 444         return error("Unable to find %s under %s\n", sha1_to_hex(sha1),
 445                      initial_base);
 446 }
 447
 448 int fetch_ref(char *ref, unsigned char *sha1)
 449 {
 450         char *url, *posn;
 451         char hex[42];
 452         struct buffer buffer;
 453         char *base = initial_base;
 454         buffer.size = 41;
 455         buffer.posn = 0;
 456         buffer.buffer = hex;
 457         hex[41] = '\0';
 458
 459         curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 460         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 461         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
 462         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 463
 464         url = xmalloc(strlen(base) + 6 + strlen(ref));
 465         strcpy(url, base);
 466         posn = url + strlen(base);
 467         strcpy(posn, "refs/");
 468         posn += 5;
 469         strcpy(posn, ref);
 470
 471         curl_easy_setopt(curl, CURLOPT_URL, url);
 472
 473         if (curl_easy_perform(curl))
 474                 return error("Couldn't get %s for %s\n%s",
 475                              url, ref, curl_errorstr);
 476
 477         hex[40] = '\0';
 478         get_sha1_hex(hex, sha1);
 479         return 0;
 480 }
 481
 482 int main(int argc, char **argv)
 483 {
 484         char *commit_id;
 485         char *url;
 486         int arg = 1;
 487
 488         while (arg < argc && argv[arg][0] == '-') {
 489                 if (argv[arg][1] == 't') {
 490                         get_tree = 1;
 491                 } else if (argv[arg][1] == 'c') {
 492                         get_history = 1;
 493                 } else if (argv[arg][1] == 'a') {
 494                         get_all = 1;
 495                         get_tree = 1;
 496                         get_history = 1;
 497                 } else if (argv[arg][1] == 'v') {
 498                         get_verbosely = 1;
 499                 } else if (argv[arg][1] == 'w') {
 500                         write_ref = argv[arg + 1];
 501                         arg++;
 502                 } else if (!strcmp(argv[arg], "--recover")) {
 503                         get_recover = 1;
 504                 }
 505                 arg++;
 506         }
 507         if (argc < arg + 2) {
 508                 usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
 509                 return 1;
 510         }
 511         commit_id = argv[arg];
 512         url = argv[arg + 1];
 513
 514         curl_global_init(CURL_GLOBAL_ALL);
 515
 516         curl = curl_easy_init();
 517         no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
 518
 519         curl_ssl_verify = getenv("GIT_SSL_NO_VERIFY") ? 0 : 1;
 520         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
 521 #if LIBCURL_VERSION_NUM >= 0x070907
 522         curl_easy_setopt(curl, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
 523 #endif
 524
 525         alt = xmalloc(sizeof(*alt));
 526         alt->base = url;
 527         alt->got_indices = 0;
 528         alt->packs = NULL;
 529         alt->next = NULL;
 530         initial_base = url;
 531
 532         if (pull(commit_id))
 533                 return 1;
 534
 535         curl_slist_free_all(no_pragma_header);
 536         curl_global_cleanup();
 537         return 0;
 538 }