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 static 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 static char *ssl_cert;
  40 static char *ssl_key;
  41 static char *ssl_capath;
  42 static char *ssl_cainfo;
  43
  44 struct buffer
  45 {
  46         size_t posn;
  47         size_t size;
  48         void *buffer;
  49 };
  50
  51 static size_t fwrite_buffer(void *ptr, size_t eltsize, size_t nmemb,
  52                             struct buffer *buffer)
  53 {
  54         size_t size = eltsize * nmemb;
  55         if (size > buffer->size - buffer->posn)
  56                 size = buffer->size - buffer->posn;
  57         memcpy(buffer->buffer + buffer->posn, ptr, size);
  58         buffer->posn += size;
  59         return size;
  60 }
  61
  62 static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
  63                                void *data)
  64 {
  65         unsigned char expn[4096];
  66         size_t size = eltsize * nmemb;
  67         int posn = 0;
  68         do {
  69                 ssize_t retval = write(local, ptr + posn, size - posn);
  70                 if (retval < 0)
  71                         return posn;
  72                 posn += retval;
  73         } while (posn < size);
  74
  75         stream.avail_in = size;
  76         stream.next_in = ptr;
  77         do {
  78                 stream.next_out = expn;
  79                 stream.avail_out = sizeof(expn);
  80                 zret = inflate(&stream, Z_SYNC_FLUSH);
  81                 SHA1_Update(&c, expn, sizeof(expn) - stream.avail_out);
  82         } while (stream.avail_in && zret == Z_OK);
  83         return size;
  84 }
  85
  86 void prefetch(unsigned char *sha1)
  87 {
  88 }
  89
  90 static int got_alternates = 0;
  91
  92 static int fetch_index(struct alt_base *repo, unsigned char *sha1)
  93 {
  94         char *filename;
  95         char *url;
  96
  97         FILE *indexfile;
  98
  99         if (has_pack_index(sha1))
 100                 return 0;
 101
 102         if (get_verbosely)
 103                 fprintf(stderr, "Getting index for pack %s\n",
 104                         sha1_to_hex(sha1));
 105
 106         url = xmalloc(strlen(repo->base) + 64);
 107         sprintf(url, "%s/objects/pack/pack-%s.idx",
 108                 repo->base, sha1_to_hex(sha1));
 109
 110         filename = sha1_pack_index_name(sha1);
 111         indexfile = fopen(filename, "w");
 112         if (!indexfile)
 113                 return error("Unable to open local file %s for pack index",
 114                              filename);
 115
 116         curl_easy_setopt(curl, CURLOPT_FILE, indexfile);
 117         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite);
 118         curl_easy_setopt(curl, CURLOPT_URL, url);
 119         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 120         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 121
 122         if (curl_easy_perform(curl)) {
 123                 fclose(indexfile);
 124                 return error("Unable to get pack index %s\n%s", url,
 125                              curl_errorstr);
 126         }
 127
 128         fclose(indexfile);
 129         return 0;
 130 }
 131
 132 static int setup_index(struct alt_base *repo, unsigned char *sha1)
 133 {
 134         struct packed_git *new_pack;
 135         if (has_pack_file(sha1))
 136                 return 0; // don't list this as something we can get
 137
 138         if (fetch_index(repo, sha1))
 139                 return -1;
 140
 141         new_pack = parse_pack_index(sha1);
 142         new_pack->next = repo->packs;
 143         repo->packs = new_pack;
 144         return 0;
 145 }
 146
 147 static int fetch_alternates(char *base)
 148 {
 149         int ret = 0;
 150         struct buffer buffer;
 151         char *url;
 152         char *data;
 153         int i = 0;
 154         int http_specific = 1;
 155         if (got_alternates)
 156                 return 0;
 157         data = xmalloc(4096);
 158         buffer.size = 4095;
 159         buffer.posn = 0;
 160         buffer.buffer = data;
 161
 162         if (get_verbosely)
 163                 fprintf(stderr, "Getting alternates list\n");
 164
 165         url = xmalloc(strlen(base) + 31);
 166         sprintf(url, "%s/objects/info/http-alternates", base);
 167
 168         curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 169         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 170         curl_easy_setopt(curl, CURLOPT_URL, url);
 171
 172         if (curl_easy_perform(curl) || !buffer.posn) {
 173                 http_specific = 0;
 174
 175                 sprintf(url, "%s/objects/info/alternates", base);
 176
 177                 curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 178                 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 179                 curl_easy_setopt(curl, CURLOPT_URL, url);
 180
 181                 if (curl_easy_perform(curl)) {
 182                         return 0;
 183                 }
 184         }
 185
 186         data[buffer.posn] = '\0';
 187
 188         while (i < buffer.posn) {
 189                 int posn = i;
 190                 while (posn < buffer.posn && data[posn] != '\n')
 191                         posn++;
 192                 if (data[posn] == '\n') {
 193                         int okay = 0;
 194                         int serverlen = 0;
 195                         struct alt_base *newalt;
 196                         char *target = NULL;
 197                         if (data[i] == '/') {
 198                                 serverlen = strchr(base + 8, '/') - base;
 199                                 okay = 1;
 200                         } else if (!memcmp(data + i, "../", 3)) {
 201                                 i += 3;
 202                                 serverlen = strlen(base);
 203                                 while (i + 2 < posn &&
 204                                        !memcmp(data + i, "../", 3)) {
 205                                         do {
 206                                                 serverlen--;
 207                                         } while (serverlen &&
 208                                                  base[serverlen - 1] != '/');
 209                                         i += 3;
 210                                 }
 211                                 // If the server got removed, give up.
 212                                 okay = strchr(base, ':') - base + 3 <
 213                                         serverlen;
 214                         } else if (http_specific) {
 215                                 char *colon = strchr(data + i, ':');
 216                                 char *slash = strchr(data + i, '/');
 217                                 if (colon && slash && colon < data + posn &&
 218                                     slash < data + posn && colon < slash) {
 219                                         okay = 1;
 220                                 }
 221                         }
 222                         // skip 'objects' at end
 223                         if (okay) {
 224                                 target = xmalloc(serverlen + posn - i - 6);
 225                                 strncpy(target, base, serverlen);
 226                                 strncpy(target + serverlen, data + i,
 227                                         posn - i - 7);
 228                                 target[serverlen + posn - i - 7] = '\0';
 229                                 if (get_verbosely)
 230                                         fprintf(stderr,
 231                                                 "Also look at %s\n", target);
 232                                 newalt = xmalloc(sizeof(*newalt));
 233                                 newalt->next = alt;
 234                                 newalt->base = target;
 235                                 newalt->got_indices = 0;
 236                                 newalt->packs = NULL;
 237                                 alt = newalt;
 238                                 ret++;
 239                         }
 240                 }
 241                 i = posn + 1;
 242         }
 243         got_alternates = 1;
 244
 245         return ret;
 246 }
 247
 248 static int fetch_indices(struct alt_base *repo)
 249 {
 250         unsigned char sha1[20];
 251         char *url;
 252         struct buffer buffer;
 253         char *data;
 254         int i = 0;
 255
 256         if (repo->got_indices)
 257                 return 0;
 258
 259         data = xmalloc(4096);
 260         buffer.size = 4096;
 261         buffer.posn = 0;
 262         buffer.buffer = data;
 263
 264         if (get_verbosely)
 265                 fprintf(stderr, "Getting pack list\n");
 266
 267         url = xmalloc(strlen(repo->base) + 21);
 268         sprintf(url, "%s/objects/info/packs", repo->base);
 269
 270         curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 271         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 272         curl_easy_setopt(curl, CURLOPT_URL, url);
 273         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
 274         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 275
 276         if (curl_easy_perform(curl))
 277                 return error("%s", curl_errorstr);
 278
 279         while (i < buffer.posn) {
 280                 switch (data[i]) {
 281                 case 'P':
 282                         i++;
 283                         if (i + 52 < buffer.posn &&
 284                             !strncmp(data + i, " pack-", 6) &&
 285                             !strncmp(data + i + 46, ".pack\n", 6)) {
 286                                 get_sha1_hex(data + i + 6, sha1);
 287                                 setup_index(repo, sha1);
 288                                 i += 51;
 289                                 break;
 290                         }
 291                 default:
 292                         while (data[i] != '\n')
 293                                 i++;
 294                 }
 295                 i++;
 296         }
 297
 298         repo->got_indices = 1;
 299         return 0;
 300 }
 301
 302 static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 303 {
 304         char *url;
 305         struct packed_git *target;
 306         struct packed_git **lst;
 307         FILE *packfile;
 308         char *filename;
 309
 310         if (fetch_indices(repo))
 311                 return -1;
 312         target = find_sha1_pack(sha1, repo->packs);
 313         if (!target)
 314                 return -1;
 315
 316         if (get_verbosely) {
 317                 fprintf(stderr, "Getting pack %s\n",
 318                         sha1_to_hex(target->sha1));
 319                 fprintf(stderr, " which contains %s\n",
 320                         sha1_to_hex(sha1));
 321         }
 322
 323         url = xmalloc(strlen(repo->base) + 65);
 324         sprintf(url, "%s/objects/pack/pack-%s.pack",
 325                 repo->base, sha1_to_hex(target->sha1));
 326
 327         filename = sha1_pack_name(target->sha1);
 328         packfile = fopen(filename, "w");
 329         if (!packfile)
 330                 return error("Unable to open local file %s for pack",
 331                              filename);
 332
 333         curl_easy_setopt(curl, CURLOPT_FILE, packfile);
 334         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite);
 335         curl_easy_setopt(curl, CURLOPT_URL, url);
 336         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 337         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 338
 339         if (curl_easy_perform(curl)) {
 340                 fclose(packfile);
 341                 return error("Unable to get pack file %s\n%s", url,
 342                              curl_errorstr);
 343         }
 344
 345         fclose(packfile);
 346
 347         lst = &repo->packs;
 348         while (*lst != target)
 349                 lst = &((*lst)->next);
 350         *lst = (*lst)->next;
 351
 352         install_packed_git(target);
 353
 354         return 0;
 355 }
 356
 357 static int fetch_object(struct alt_base *repo, unsigned char *sha1)
 358 {
 359         char *hex = sha1_to_hex(sha1);
 360         char *filename = sha1_file_name(sha1);
 361         unsigned char real_sha1[20];
 362         char tmpfile[PATH_MAX];
 363         int ret;
 364         char *url;
 365         char *posn;
 366
 367         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX",
 368                  get_object_directory());
 369
 370         local = mkstemp(tmpfile);
 371         if (local < 0)
 372                 return error("Couldn't create temporary file %s for %s: %s\n",
 373                              tmpfile, filename, strerror(errno));
 374
 375         memset(&stream, 0, sizeof(stream));
 376
 377         inflateInit(&stream);
 378
 379         SHA1_Init(&c);
 380
 381         curl_easy_setopt(curl, CURLOPT_FAILONERROR, 1);
 382         curl_easy_setopt(curl, CURLOPT_FILE, NULL);
 383         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 384         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 385         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 386
 387         url = xmalloc(strlen(repo->base) + 50);
 388         strcpy(url, repo->base);
 389         posn = url + strlen(repo->base);
 390         strcpy(posn, "objects/");
 391         posn += 8;
 392         memcpy(posn, hex, 2);
 393         posn += 2;
 394         *(posn++) = '/';
 395         strcpy(posn, hex + 2);
 396
 397         curl_easy_setopt(curl, CURLOPT_URL, url);
 398
 399         if (curl_easy_perform(curl)) {
 400                 unlink(filename);
 401                 return error("%s", curl_errorstr);
 402         }
 403
 404         fchmod(local, 0444);
 405         close(local);
 406         inflateEnd(&stream);
 407         SHA1_Final(real_sha1, &c);
 408         if (zret != Z_STREAM_END) {
 409                 unlink(tmpfile);
 410                 return error("File %s (%s) corrupt\n", hex, url);
 411         }
 412         if (memcmp(sha1, real_sha1, 20)) {
 413                 unlink(tmpfile);
 414                 return error("File %s has bad hash\n", hex);
 415         }
 416         ret = link(tmpfile, filename);
 417         if (ret < 0) {
 418                 /* Same Coda hack as in write_sha1_file(sha1_file.c) */
 419                 ret = errno;
 420                 if (ret == EXDEV && !rename(tmpfile, filename))
 421                         goto out;
 422         }
 423         unlink(tmpfile);
 424         if (ret) {
 425                 if (ret != EEXIST)
 426                         return error("unable to write sha1 filename %s: %s",
 427                                      filename, strerror(ret));
 428         }
 429  out:
 430         pull_say("got %s\n", hex);
 431         return 0;
 432 }
 433
 434 int fetch(unsigned char *sha1)
 435 {
 436         struct alt_base *altbase = alt;
 437         while (altbase) {
 438                 if (!fetch_object(altbase, sha1))
 439                         return 0;
 440                 if (!fetch_pack(altbase, sha1))
 441                         return 0;
 442                 if (fetch_alternates(altbase->base) > 0) {
 443                         altbase = alt;
 444                         continue;
 445                 }
 446                 altbase = altbase->next;
 447         }
 448         return error("Unable to find %s under %s\n", sha1_to_hex(sha1),
 449                      initial_base);
 450 }
 451
 452 int fetch_ref(char *ref, unsigned char *sha1)
 453 {
 454         char *url, *posn;
 455         char hex[42];
 456         struct buffer buffer;
 457         char *base = initial_base;
 458         buffer.size = 41;
 459         buffer.posn = 0;
 460         buffer.buffer = hex;
 461         hex[41] = '\0';
 462
 463         curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 464         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 465         curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
 466         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 467
 468         url = xmalloc(strlen(base) + 6 + strlen(ref));
 469         strcpy(url, base);
 470         posn = url + strlen(base);
 471         strcpy(posn, "refs/");
 472         posn += 5;
 473         strcpy(posn, ref);
 474
 475         curl_easy_setopt(curl, CURLOPT_URL, url);
 476
 477         if (curl_easy_perform(curl))
 478                 return error("Couldn't get %s for %s\n%s",
 479                              url, ref, curl_errorstr);
 480
 481         hex[40] = '\0';
 482         get_sha1_hex(hex, sha1);
 483         return 0;
 484 }
 485
 486 int main(int argc, char **argv)
 487 {
 488         char *commit_id;
 489         char *url;
 490         int arg = 1;
 491
 492         while (arg < argc && argv[arg][0] == '-') {
 493                 if (argv[arg][1] == 't') {
 494                         get_tree = 1;
 495                 } else if (argv[arg][1] == 'c') {
 496                         get_history = 1;
 497                 } else if (argv[arg][1] == 'a') {
 498                         get_all = 1;
 499                         get_tree = 1;
 500                         get_history = 1;
 501                 } else if (argv[arg][1] == 'v') {
 502                         get_verbosely = 1;
 503                 } else if (argv[arg][1] == 'w') {
 504                         write_ref = argv[arg + 1];
 505                         arg++;
 506                 } else if (!strcmp(argv[arg], "--recover")) {
 507                         get_recover = 1;
 508                 }
 509                 arg++;
 510         }
 511         if (argc < arg + 2) {
 512                 usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
 513                 return 1;
 514         }
 515         commit_id = argv[arg];
 516         url = argv[arg + 1];
 517
 518         curl_global_init(CURL_GLOBAL_ALL);
 519
 520         curl = curl_easy_init();
 521         no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
 522
 523         curl_ssl_verify = getenv("GIT_SSL_NO_VERIFY") ? 0 : 1;
 524         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
 525 #if LIBCURL_VERSION_NUM >= 0x070907
 526         curl_easy_setopt(curl, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
 527 #endif
 528
 529         if ((ssl_cert = getenv("GIT_SSL_CERT")) != NULL) {
 530                 curl_easy_setopt(curl, CURLOPT_SSLCERT, ssl_cert);
 531         }
 532 #if LIBCURL_VERSION_NUM >= 0x070902
 533         if ((ssl_key = getenv("GIT_SSL_KEY")) != NULL) {
 534                 curl_easy_setopt(curl, CURLOPT_SSLKEY, ssl_key);
 535         }
 536 #endif
 537 #if LIBCURL_VERSION_NUM >= 0x070908
 538         if ((ssl_capath = getenv("GIT_SSL_CAPATH")) != NULL) {
 539                 curl_easy_setopt(curl, CURLOPT_CAPATH, ssl_capath);
 540         }
 541 #endif
 542         if ((ssl_cainfo = getenv("GIT_SSL_CAINFO")) != NULL) {
 543                 curl_easy_setopt(curl, CURLOPT_CAINFO, ssl_cainfo);
 544         }
 545
 546         alt = xmalloc(sizeof(*alt));
 547         alt->base = url;
 548         alt->got_indices = 0;
 549         alt->packs = NULL;
 550         alt->next = NULL;
 551         initial_base = url;
 552
 553         if (pull(commit_id))
 554                 return 1;
 555
 556         curl_slist_free_all(no_pragma_header);
 557         curl_global_cleanup();
 558         return 0;
 559 }