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