http-fetch.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "pack.h"
   4 #include "fetch.h"
   5 #include "http.h"
   6
   7 #define PREV_BUF_SIZE 4096
   8 #define RANGE_HEADER_SIZE 30
   9
  10 static int got_alternates = -1;
  11
  12 static struct curl_slist *no_pragma_header;
  13
  14 struct alt_base
  15 {
  16         char *base;
  17         int got_indices;
  18         struct packed_git *packs;
  19         struct alt_base *next;
  20 };
  21
  22 static struct alt_base *alt = NULL;
  23
  24 enum object_request_state {
  25         WAITING,
  26         ABORTED,
  27         ACTIVE,
  28         COMPLETE,
  29 };
  30
  31 struct object_request
  32 {
  33         unsigned char sha1[20];
  34         struct alt_base *repo;
  35         char *url;
  36         char filename[PATH_MAX];
  37         char tmpfile[PATH_MAX];
  38         int local;
  39         enum object_request_state state;
  40         CURLcode curl_result;
  41         char errorstr[CURL_ERROR_SIZE];
  42         long http_code;
  43         unsigned char real_sha1[20];
  44         SHA_CTX c;
  45         z_stream stream;
  46         int zret;
  47         int rename;
  48         struct active_request_slot *slot;
  49         struct object_request *next;
  50 };
  51
  52 struct alternates_request {
  53         char *base;
  54         char *url;
  55         struct buffer *buffer;
  56         struct active_request_slot *slot;
  57         int http_specific;
  58 };
  59
  60 static struct object_request *object_queue_head = NULL;
  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         struct object_request *obj_req = (struct object_request *)data;
  69         do {
  70                 ssize_t retval = write(obj_req->local,
  71                                        ptr + posn, size - posn);
  72                 if (retval < 0)
  73                         return posn;
  74                 posn += retval;
  75         } while (posn < size);
  76
  77         obj_req->stream.avail_in = size;
  78         obj_req->stream.next_in = ptr;
  79         do {
  80                 obj_req->stream.next_out = expn;
  81                 obj_req->stream.avail_out = sizeof(expn);
  82                 obj_req->zret = inflate(&obj_req->stream, Z_SYNC_FLUSH);
  83                 SHA1_Update(&obj_req->c, expn,
  84                             sizeof(expn) - obj_req->stream.avail_out);
  85         } while (obj_req->stream.avail_in && obj_req->zret == Z_OK);
  86         data_received++;
  87         return size;
  88 }
  89
  90 static void fetch_alternates(char *base);
  91
  92 static void process_object_response(void *callback_data);
  93
  94 static void start_object_request(struct object_request *obj_req)
  95 {
  96         char *hex = sha1_to_hex(obj_req->sha1);
  97         char prevfile[PATH_MAX];
  98         char *url;
  99         char *posn;
 100         int prevlocal;
 101         unsigned char prev_buf[PREV_BUF_SIZE];
 102         ssize_t prev_read = 0;
 103         long prev_posn = 0;
 104         char range[RANGE_HEADER_SIZE];
 105         struct curl_slist *range_header = NULL;
 106         struct active_request_slot *slot;
 107
 108         snprintf(prevfile, sizeof(prevfile), "%s.prev", obj_req->filename);
 109         unlink(prevfile);
 110         rename(obj_req->tmpfile, prevfile);
 111         unlink(obj_req->tmpfile);
 112
 113         if (obj_req->local != -1)
 114                 error("fd leakage in start: %d", obj_req->local);
 115         obj_req->local = open(obj_req->tmpfile,
 116                               O_WRONLY | O_CREAT | O_EXCL, 0666);
 117         /* This could have failed due to the "lazy directory creation";
 118          * try to mkdir the last path component.
 119          */
 120         if (obj_req->local < 0 && errno == ENOENT) {
 121                 char *dir = strrchr(obj_req->tmpfile, '/');
 122                 if (dir) {
 123                         *dir = 0;
 124                         mkdir(obj_req->tmpfile, 0777);
 125                         *dir = '/';
 126                 }
 127                 obj_req->local = open(obj_req->tmpfile,
 128                                       O_WRONLY | O_CREAT | O_EXCL, 0666);
 129         }
 130
 131         if (obj_req->local < 0) {
 132                 obj_req->state = ABORTED;
 133                 error("Couldn't create temporary file %s for %s: %s\n",
 134                       obj_req->tmpfile, obj_req->filename, strerror(errno));
 135                 return;
 136         }
 137
 138         memset(&obj_req->stream, 0, sizeof(obj_req->stream));
 139
 140         inflateInit(&obj_req->stream);
 141
 142         SHA1_Init(&obj_req->c);
 143
 144         url = xmalloc(strlen(obj_req->repo->base) + 50);
 145         obj_req->url = xmalloc(strlen(obj_req->repo->base) + 50);
 146         strcpy(url, obj_req->repo->base);
 147         posn = url + strlen(obj_req->repo->base);
 148         strcpy(posn, "objects/");
 149         posn += 8;
 150         memcpy(posn, hex, 2);
 151         posn += 2;
 152         *(posn++) = '/';
 153         strcpy(posn, hex + 2);
 154         strcpy(obj_req->url, url);
 155
 156         /* If a previous temp file is present, process what was already
 157            fetched. */
 158         prevlocal = open(prevfile, O_RDONLY);
 159         if (prevlocal != -1) {
 160                 do {
 161                         prev_read = read(prevlocal, prev_buf, PREV_BUF_SIZE);
 162                         if (prev_read>0) {
 163                                 if (fwrite_sha1_file(prev_buf,
 164                                                      1,
 165                                                      prev_read,
 166                                                      obj_req) == prev_read) {
 167                                         prev_posn += prev_read;
 168                                 } else {
 169                                         prev_read = -1;
 170                                 }
 171                         }
 172                 } while (prev_read > 0);
 173                 close(prevlocal);
 174         }
 175         unlink(prevfile);
 176
 177         /* Reset inflate/SHA1 if there was an error reading the previous temp
 178            file; also rewind to the beginning of the local file. */
 179         if (prev_read == -1) {
 180                 memset(&obj_req->stream, 0, sizeof(obj_req->stream));
 181                 inflateInit(&obj_req->stream);
 182                 SHA1_Init(&obj_req->c);
 183                 if (prev_posn>0) {
 184                         prev_posn = 0;
 185                         lseek(obj_req->local, SEEK_SET, 0);
 186                         ftruncate(obj_req->local, 0);
 187                 }
 188         }
 189
 190         slot = get_active_slot();
 191         slot->callback_func = process_object_response;
 192         slot->callback_data = obj_req;
 193         obj_req->slot = slot;
 194
 195         curl_easy_setopt(slot->curl, CURLOPT_FILE, obj_req);
 196         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 197         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, obj_req->errorstr);
 198         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 199         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 200
 201         /* If we have successfully processed data from a previous fetch
 202            attempt, only fetch the data we don't already have. */
 203         if (prev_posn>0) {
 204                 if (get_verbosely)
 205                         fprintf(stderr,
 206                                 "Resuming fetch of object %s at byte %ld\n",
 207                                 hex, prev_posn);
 208                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 209                 range_header = curl_slist_append(range_header, range);
 210                 curl_easy_setopt(slot->curl,
 211                                  CURLOPT_HTTPHEADER, range_header);
 212         }
 213
 214         /* Try to get the request started, abort the request on error */
 215         obj_req->state = ACTIVE;
 216         if (!start_active_slot(slot)) {
 217                 obj_req->state = ABORTED;
 218                 obj_req->slot = NULL;
 219                 close(obj_req->local); obj_req->local = -1;
 220                 free(obj_req->url);
 221                 return;
 222         }
 223 }
 224
 225 static void finish_object_request(struct object_request *obj_req)
 226 {
 227         struct stat st;
 228
 229         fchmod(obj_req->local, 0444);
 230         close(obj_req->local); obj_req->local = -1;
 231
 232         if (obj_req->http_code == 416) {
 233                 fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 234         } else if (obj_req->curl_result != CURLE_OK) {
 235                 if (stat(obj_req->tmpfile, &st) == 0)
 236                         if (st.st_size == 0)
 237                                 unlink(obj_req->tmpfile);
 238                 return;
 239         }
 240
 241         inflateEnd(&obj_req->stream);
 242         SHA1_Final(obj_req->real_sha1, &obj_req->c);
 243         if (obj_req->zret != Z_STREAM_END) {
 244                 unlink(obj_req->tmpfile);
 245                 return;
 246         }
 247         if (memcmp(obj_req->sha1, obj_req->real_sha1, 20)) {
 248                 unlink(obj_req->tmpfile);
 249                 return;
 250         }
 251         obj_req->rename =
 252                 move_temp_to_file(obj_req->tmpfile, obj_req->filename);
 253
 254         if (obj_req->rename == 0)
 255                 pull_say("got %s\n", sha1_to_hex(obj_req->sha1));
 256 }
 257
 258 static void process_object_response(void *callback_data)
 259 {
 260         struct object_request *obj_req =
 261                 (struct object_request *)callback_data;
 262
 263         obj_req->curl_result = obj_req->slot->curl_result;
 264         obj_req->http_code = obj_req->slot->http_code;
 265         obj_req->slot = NULL;
 266         obj_req->state = COMPLETE;
 267
 268         /* Use alternates if necessary */
 269         if (obj_req->http_code == 404 ||
 270             obj_req->curl_result == CURLE_FILE_COULDNT_READ_FILE) {
 271                 fetch_alternates(alt->base);
 272                 if (obj_req->repo->next != NULL) {
 273                         obj_req->repo =
 274                                 obj_req->repo->next;
 275                         close(obj_req->local);
 276                         obj_req->local = -1;
 277                         start_object_request(obj_req);
 278                         return;
 279                 }
 280         }
 281
 282         finish_object_request(obj_req);
 283 }
 284
 285 static void release_object_request(struct object_request *obj_req)
 286 {
 287         struct object_request *entry = object_queue_head;
 288
 289         if (obj_req->local != -1)
 290                 error("fd leakage in release: %d", obj_req->local);
 291         if (obj_req == object_queue_head) {
 292                 object_queue_head = obj_req->next;
 293         } else {
 294                 while (entry->next != NULL && entry->next != obj_req)
 295                         entry = entry->next;
 296                 if (entry->next == obj_req)
 297                         entry->next = entry->next->next;
 298         }
 299
 300         free(obj_req->url);
 301         free(obj_req);
 302 }
 303
 304 #ifdef USE_CURL_MULTI
 305 void fill_active_slots(void)
 306 {
 307         struct object_request *obj_req = object_queue_head;
 308         struct active_request_slot *slot = active_queue_head;
 309         int num_transfers;
 310
 311         while (active_requests < max_requests && obj_req != NULL) {
 312                 if (obj_req->state == WAITING) {
 313                         if (has_sha1_file(obj_req->sha1))
 314                                 obj_req->state = COMPLETE;
 315                         else
 316                                 start_object_request(obj_req);
 317                         curl_multi_perform(curlm, &num_transfers);
 318                 }
 319                 obj_req = obj_req->next;
 320         }
 321
 322         while (slot != NULL) {
 323                 if (!slot->in_use && slot->curl != NULL) {
 324                         curl_easy_cleanup(slot->curl);
 325                         slot->curl = NULL;
 326                 }
 327                 slot = slot->next;
 328         }
 329 }
 330 #endif
 331
 332 void prefetch(unsigned char *sha1)
 333 {
 334         struct object_request *newreq;
 335         struct object_request *tail;
 336         char *filename = sha1_file_name(sha1);
 337
 338         newreq = xmalloc(sizeof(*newreq));
 339         memcpy(newreq->sha1, sha1, 20);
 340         newreq->repo = alt;
 341         newreq->url = NULL;
 342         newreq->local = -1;
 343         newreq->state = WAITING;
 344         snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 345         snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 346                  "%s.temp", filename);
 347         newreq->next = NULL;
 348
 349         if (object_queue_head == NULL) {
 350                 object_queue_head = newreq;
 351         } else {
 352                 tail = object_queue_head;
 353                 while (tail->next != NULL) {
 354                         tail = tail->next;
 355                 }
 356                 tail->next = newreq;
 357         }
 358
 359 #ifdef USE_CURL_MULTI
 360         fill_active_slots();
 361         step_active_slots();
 362 #endif
 363 }
 364
 365 static int fetch_index(struct alt_base *repo, unsigned char *sha1)
 366 {
 367         char *hex = sha1_to_hex(sha1);
 368         char *filename;
 369         char *url;
 370         char tmpfile[PATH_MAX];
 371         long prev_posn = 0;
 372         char range[RANGE_HEADER_SIZE];
 373         struct curl_slist *range_header = NULL;
 374
 375         FILE *indexfile;
 376         struct active_request_slot *slot;
 377         struct slot_results results;
 378
 379         if (has_pack_index(sha1))
 380                 return 0;
 381
 382         if (get_verbosely)
 383                 fprintf(stderr, "Getting index for pack %s\n", hex);
 384
 385         url = xmalloc(strlen(repo->base) + 64);
 386         sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 387
 388         filename = sha1_pack_index_name(sha1);
 389         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 390         indexfile = fopen(tmpfile, "a");
 391         if (!indexfile)
 392                 return error("Unable to open local file %s for pack index",
 393                              filename);
 394
 395         slot = get_active_slot();
 396         slot->results = &results;
 397         curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 398         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 399         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 400         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 401         slot->local = indexfile;
 402
 403         /* If there is data present from a previous transfer attempt,
 404            resume where it left off */
 405         prev_posn = ftell(indexfile);
 406         if (prev_posn>0) {
 407                 if (get_verbosely)
 408                         fprintf(stderr,
 409                                 "Resuming fetch of index for pack %s at byte %ld\n",
 410                                 hex, prev_posn);
 411                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 412                 range_header = curl_slist_append(range_header, range);
 413                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 414         }
 415
 416         if (start_active_slot(slot)) {
 417                 run_active_slot(slot);
 418                 if (results.curl_result != CURLE_OK) {
 419                         fclose(indexfile);
 420                         return error("Unable to get pack index %s\n%s", url,
 421                                      curl_errorstr);
 422                 }
 423         } else {
 424                 fclose(indexfile);
 425                 return error("Unable to start request");
 426         }
 427
 428         fclose(indexfile);
 429
 430         return move_temp_to_file(tmpfile, filename);
 431 }
 432
 433 static int setup_index(struct alt_base *repo, unsigned char *sha1)
 434 {
 435         struct packed_git *new_pack;
 436         if (has_pack_file(sha1))
 437                 return 0; // don't list this as something we can get
 438
 439         if (fetch_index(repo, sha1))
 440                 return -1;
 441
 442         new_pack = parse_pack_index(sha1);
 443         new_pack->next = repo->packs;
 444         repo->packs = new_pack;
 445         return 0;
 446 }
 447
 448 static void process_alternates_response(void *callback_data)
 449 {
 450         struct alternates_request *alt_req =
 451                 (struct alternates_request *)callback_data;
 452         struct active_request_slot *slot = alt_req->slot;
 453         struct alt_base *tail = alt;
 454         char *base = alt_req->base;
 455         static const char null_byte = '\0';
 456         char *data;
 457         int i = 0;
 458
 459         if (alt_req->http_specific) {
 460                 if (slot->curl_result != CURLE_OK ||
 461                     !alt_req->buffer->posn) {
 462
 463                         /* Try reusing the slot to get non-http alternates */
 464                         alt_req->http_specific = 0;
 465                         sprintf(alt_req->url, "%s/objects/info/alternates",
 466                                 base);
 467                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 468                                          alt_req->url);
 469                         active_requests++;
 470                         slot->in_use = 1;
 471                         if (!start_active_slot(slot)) {
 472                                 got_alternates = -1;
 473                                 slot->in_use = 0;
 474                         }
 475                         return;
 476                 }
 477         } else if (slot->curl_result != CURLE_OK) {
 478                 if (slot->http_code != 404 &&
 479                     slot->curl_result != CURLE_FILE_COULDNT_READ_FILE) {
 480                         got_alternates = -1;
 481                         return;
 482                 }
 483         }
 484
 485         fwrite_buffer(&null_byte, 1, 1, alt_req->buffer);
 486         alt_req->buffer->posn--;
 487         data = alt_req->buffer->buffer;
 488
 489         while (i < alt_req->buffer->posn) {
 490                 int posn = i;
 491                 while (posn < alt_req->buffer->posn && data[posn] != '\n')
 492                         posn++;
 493                 if (data[posn] == '\n') {
 494                         int okay = 0;
 495                         int serverlen = 0;
 496                         struct alt_base *newalt;
 497                         char *target = NULL;
 498                         if (data[i] == '/') {
 499                                 serverlen = strchr(base + 8, '/') - base;
 500                                 okay = 1;
 501                         } else if (!memcmp(data + i, "../", 3)) {
 502                                 i += 3;
 503                                 serverlen = strlen(base);
 504                                 while (i + 2 < posn &&
 505                                        !memcmp(data + i, "../", 3)) {
 506                                         do {
 507                                                 serverlen--;
 508                                         } while (serverlen &&
 509                                                  base[serverlen - 1] != '/');
 510                                         i += 3;
 511                                 }
 512                                 // If the server got removed, give up.
 513                                 okay = strchr(base, ':') - base + 3 <
 514                                         serverlen;
 515                         } else if (alt_req->http_specific) {
 516                                 char *colon = strchr(data + i, ':');
 517                                 char *slash = strchr(data + i, '/');
 518                                 if (colon && slash && colon < data + posn &&
 519                                     slash < data + posn && colon < slash) {
 520                                         okay = 1;
 521                                 }
 522                         }
 523                         // skip 'objects' at end
 524                         if (okay) {
 525                                 target = xmalloc(serverlen + posn - i - 6);
 526                                 strncpy(target, base, serverlen);
 527                                 strncpy(target + serverlen, data + i,
 528                                         posn - i - 7);
 529                                 target[serverlen + posn - i - 7] = '\0';
 530                                 if (get_verbosely)
 531                                         fprintf(stderr,
 532                                                 "Also look at %s\n", target);
 533                                 newalt = xmalloc(sizeof(*newalt));
 534                                 newalt->next = NULL;
 535                                 newalt->base = target;
 536                                 newalt->got_indices = 0;
 537                                 newalt->packs = NULL;
 538                                 while (tail->next != NULL)
 539                                         tail = tail->next;
 540                                 tail->next = newalt;
 541                         }
 542                 }
 543                 i = posn + 1;
 544         }
 545
 546         got_alternates = 1;
 547 }
 548
 549 static void fetch_alternates(char *base)
 550 {
 551         struct buffer buffer;
 552         char *url;
 553         char *data;
 554         struct active_request_slot *slot;
 555         struct alternates_request alt_req;
 556
 557         /* If another request has already started fetching alternates,
 558            wait for them to arrive and return to processing this request's
 559            curl message */
 560 #ifdef USE_CURL_MULTI
 561         while (got_alternates == 0) {
 562                 step_active_slots();
 563         }
 564 #endif
 565
 566         /* Nothing to do if they've already been fetched */
 567         if (got_alternates == 1)
 568                 return;
 569
 570         /* Start the fetch */
 571         got_alternates = 0;
 572
 573         data = xmalloc(4096);
 574         buffer.size = 4096;
 575         buffer.posn = 0;
 576         buffer.buffer = data;
 577
 578         if (get_verbosely)
 579                 fprintf(stderr, "Getting alternates list for %s\n", base);
 580
 581         url = xmalloc(strlen(base) + 31);
 582         sprintf(url, "%s/objects/info/http-alternates", base);
 583
 584         /* Use a callback to process the result, since another request
 585            may fail and need to have alternates loaded before continuing */
 586         slot = get_active_slot();
 587         slot->callback_func = process_alternates_response;
 588         slot->callback_data = &alt_req;
 589
 590         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 591         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 592         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 593
 594         alt_req.base = base;
 595         alt_req.url = url;
 596         alt_req.buffer = &buffer;
 597         alt_req.http_specific = 1;
 598         alt_req.slot = slot;
 599
 600         if (start_active_slot(slot))
 601                 run_active_slot(slot);
 602         else
 603                 got_alternates = -1;
 604
 605         free(data);
 606         free(url);
 607 }
 608
 609 static int fetch_indices(struct alt_base *repo)
 610 {
 611         unsigned char sha1[20];
 612         char *url;
 613         struct buffer buffer;
 614         char *data;
 615         int i = 0;
 616
 617         struct active_request_slot *slot;
 618         struct slot_results results;
 619
 620         if (repo->got_indices)
 621                 return 0;
 622
 623         data = xmalloc(4096);
 624         buffer.size = 4096;
 625         buffer.posn = 0;
 626         buffer.buffer = data;
 627
 628         if (get_verbosely)
 629                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 630
 631         url = xmalloc(strlen(repo->base) + 21);
 632         sprintf(url, "%s/objects/info/packs", repo->base);
 633
 634         slot = get_active_slot();
 635         slot->results = &results;
 636         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 637         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 638         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 639         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 640         if (start_active_slot(slot)) {
 641                 run_active_slot(slot);
 642                 if (results.curl_result != CURLE_OK) {
 643                         if (results.http_code == 404 ||
 644                             results.curl_result == CURLE_FILE_COULDNT_READ_FILE) {
 645                                 repo->got_indices = 1;
 646                                 free(buffer.buffer);
 647                                 return 0;
 648                         } else {
 649                                 repo->got_indices = 0;
 650                                 free(buffer.buffer);
 651                                 return error("%s", curl_errorstr);
 652                         }
 653                 }
 654         } else {
 655                 repo->got_indices = 0;
 656                 free(buffer.buffer);
 657                 return error("Unable to start request");
 658         }
 659
 660         data = buffer.buffer;
 661         while (i < buffer.posn) {
 662                 switch (data[i]) {
 663                 case 'P':
 664                         i++;
 665                         if (i + 52 <= buffer.posn &&
 666                             !strncmp(data + i, " pack-", 6) &&
 667                             !strncmp(data + i + 46, ".pack\n", 6)) {
 668                                 get_sha1_hex(data + i + 6, sha1);
 669                                 setup_index(repo, sha1);
 670                                 i += 51;
 671                                 break;
 672                         }
 673                 default:
 674                         while (i < buffer.posn && data[i] != '\n')
 675                                 i++;
 676                 }
 677                 i++;
 678         }
 679
 680         free(buffer.buffer);
 681         repo->got_indices = 1;
 682         return 0;
 683 }
 684
 685 static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 686 {
 687         char *url;
 688         struct packed_git *target;
 689         struct packed_git **lst;
 690         FILE *packfile;
 691         char *filename;
 692         char tmpfile[PATH_MAX];
 693         int ret;
 694         long prev_posn = 0;
 695         char range[RANGE_HEADER_SIZE];
 696         struct curl_slist *range_header = NULL;
 697
 698         struct active_request_slot *slot;
 699         struct slot_results results;
 700
 701         if (fetch_indices(repo))
 702                 return -1;
 703         target = find_sha1_pack(sha1, repo->packs);
 704         if (!target)
 705                 return -1;
 706
 707         if (get_verbosely) {
 708                 fprintf(stderr, "Getting pack %s\n",
 709                         sha1_to_hex(target->sha1));
 710                 fprintf(stderr, " which contains %s\n",
 711                         sha1_to_hex(sha1));
 712         }
 713
 714         url = xmalloc(strlen(repo->base) + 65);
 715         sprintf(url, "%s/objects/pack/pack-%s.pack",
 716                 repo->base, sha1_to_hex(target->sha1));
 717
 718         filename = sha1_pack_name(target->sha1);
 719         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 720         packfile = fopen(tmpfile, "a");
 721         if (!packfile)
 722                 return error("Unable to open local file %s for pack",
 723                              filename);
 724
 725         slot = get_active_slot();
 726         slot->results = &results;
 727         curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
 728         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 729         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 730         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 731         slot->local = packfile;
 732
 733         /* If there is data present from a previous transfer attempt,
 734            resume where it left off */
 735         prev_posn = ftell(packfile);
 736         if (prev_posn>0) {
 737                 if (get_verbosely)
 738                         fprintf(stderr,
 739                                 "Resuming fetch of pack %s at byte %ld\n",
 740                                 sha1_to_hex(target->sha1), prev_posn);
 741                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 742                 range_header = curl_slist_append(range_header, range);
 743                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 744         }
 745
 746         if (start_active_slot(slot)) {
 747                 run_active_slot(slot);
 748                 if (results.curl_result != CURLE_OK) {
 749                         fclose(packfile);
 750                         return error("Unable to get pack file %s\n%s", url,
 751                                      curl_errorstr);
 752                 }
 753         } else {
 754                 fclose(packfile);
 755                 return error("Unable to start request");
 756         }
 757
 758         fclose(packfile);
 759
 760         ret = move_temp_to_file(tmpfile, filename);
 761         if (ret)
 762                 return ret;
 763
 764         lst = &repo->packs;
 765         while (*lst != target)
 766                 lst = &((*lst)->next);
 767         *lst = (*lst)->next;
 768
 769         if (verify_pack(target, 0))
 770                 return -1;
 771         install_packed_git(target);
 772
 773         return 0;
 774 }
 775
 776 static void abort_object_request(struct object_request *obj_req)
 777 {
 778         if (obj_req->local >= 0) {
 779                 close(obj_req->local);
 780                 obj_req->local = -1;
 781         }
 782         unlink(obj_req->tmpfile);
 783         if (obj_req->slot) {
 784                 release_active_slot(obj_req->slot);
 785                 obj_req->slot = NULL;
 786         }
 787         release_object_request(obj_req);
 788 }
 789
 790 static int fetch_object(struct alt_base *repo, unsigned char *sha1)
 791 {
 792         char *hex = sha1_to_hex(sha1);
 793         int ret = 0;
 794         struct object_request *obj_req = object_queue_head;
 795
 796         while (obj_req != NULL && memcmp(obj_req->sha1, sha1, 20))
 797                 obj_req = obj_req->next;
 798         if (obj_req == NULL)
 799                 return error("Couldn't find request for %s in the queue", hex);
 800
 801         if (has_sha1_file(obj_req->sha1)) {
 802                 abort_object_request(obj_req);
 803                 return 0;
 804         }
 805
 806 #ifdef USE_CURL_MULTI
 807         while (obj_req->state == WAITING) {
 808                 step_active_slots();
 809         }
 810 #else
 811         start_object_request(obj_req);
 812 #endif
 813
 814         while (obj_req->state == ACTIVE) {
 815                 run_active_slot(obj_req->slot);
 816         }
 817         if (obj_req->local != -1) {
 818                 close(obj_req->local); obj_req->local = -1;
 819         }
 820
 821         if (obj_req->state == ABORTED) {
 822                 ret = error("Request for %s aborted", hex);
 823         } else if (obj_req->curl_result != CURLE_OK &&
 824                    obj_req->http_code != 416) {
 825                 if (obj_req->http_code == 404 ||
 826                     obj_req->curl_result == CURLE_FILE_COULDNT_READ_FILE)
 827                         ret = -1; /* Be silent, it is probably in a pack. */
 828                 else
 829                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 830                                     obj_req->errorstr, obj_req->curl_result,
 831                                     obj_req->http_code, hex);
 832         } else if (obj_req->zret != Z_STREAM_END) {
 833                 ret = error("File %s (%s) corrupt\n", hex, obj_req->url);
 834         } else if (memcmp(obj_req->sha1, obj_req->real_sha1, 20)) {
 835                 ret = error("File %s has bad hash\n", hex);
 836         } else if (obj_req->rename < 0) {
 837                 ret = error("unable to write sha1 filename %s",
 838                             obj_req->filename);
 839         }
 840
 841         release_object_request(obj_req);
 842         return ret;
 843 }
 844
 845 int fetch(unsigned char *sha1)
 846 {
 847         struct alt_base *altbase = alt;
 848
 849         if (!fetch_object(altbase, sha1))
 850                 return 0;
 851         while (altbase) {
 852                 if (!fetch_pack(altbase, sha1))
 853                         return 0;
 854                 fetch_alternates(alt->base);
 855                 altbase = altbase->next;
 856         }
 857         return error("Unable to find %s under %s\n", sha1_to_hex(sha1),
 858                      alt->base);
 859 }
 860
 861 static inline int needs_quote(int ch)
 862 {
 863         switch (ch) {
 864         case '/': case '-': case '.':
 865         case 'A'...'Z': case 'a'...'z': case '0'...'9':
 866                 return 0;
 867         default:
 868                 return 1;
 869         }
 870 }
 871
 872 static inline int hex(int v)
 873 {
 874         if (v < 10) return '0' + v;
 875         else return 'A' + v - 10;
 876 }
 877
 878 static char *quote_ref_url(const char *base, const char *ref)
 879 {
 880         const char *cp;
 881         char *dp, *qref;
 882         int len, baselen, ch;
 883
 884         baselen = strlen(base);
 885         len = baselen + 6; /* "refs/" + NUL */
 886         for (cp = ref; (ch = *cp) != 0; cp++, len++)
 887                 if (needs_quote(ch))
 888                         len += 2; /* extra two hex plus replacement % */
 889         qref = xmalloc(len);
 890         memcpy(qref, base, baselen);
 891         memcpy(qref + baselen, "refs/", 5);
 892         for (cp = ref, dp = qref + baselen + 5; (ch = *cp) != 0; cp++) {
 893                 if (needs_quote(ch)) {
 894                         *dp++ = '%';
 895                         *dp++ = hex((ch >> 4) & 0xF);
 896                         *dp++ = hex(ch & 0xF);
 897                 }
 898                 else
 899                         *dp++ = ch;
 900         }
 901         *dp = 0;
 902
 903         return qref;
 904 }
 905
 906 int fetch_ref(char *ref, unsigned char *sha1)
 907 {
 908         char *url;
 909         char hex[42];
 910         struct buffer buffer;
 911         char *base = alt->base;
 912         struct active_request_slot *slot;
 913         struct slot_results results;
 914         buffer.size = 41;
 915         buffer.posn = 0;
 916         buffer.buffer = hex;
 917         hex[41] = '\0';
 918
 919         url = quote_ref_url(base, ref);
 920         slot = get_active_slot();
 921         slot->results = &results;
 922         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 923         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 924         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 925         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 926         if (start_active_slot(slot)) {
 927                 run_active_slot(slot);
 928                 if (results.curl_result != CURLE_OK)
 929                         return error("Couldn't get %s for %s\n%s",
 930                                      url, ref, curl_errorstr);
 931         } else {
 932                 return error("Unable to start request");
 933         }
 934
 935         hex[40] = '\0';
 936         get_sha1_hex(hex, sha1);
 937         return 0;
 938 }
 939
 940 int main(int argc, char **argv)
 941 {
 942         char *commit_id;
 943         char *url;
 944         int arg = 1;
 945         int rc = 0;
 946
 947         setup_git_directory();
 948
 949         while (arg < argc && argv[arg][0] == '-') {
 950                 if (argv[arg][1] == 't') {
 951                         get_tree = 1;
 952                 } else if (argv[arg][1] == 'c') {
 953                         get_history = 1;
 954                 } else if (argv[arg][1] == 'a') {
 955                         get_all = 1;
 956                         get_tree = 1;
 957                         get_history = 1;
 958                 } else if (argv[arg][1] == 'v') {
 959                         get_verbosely = 1;
 960                 } else if (argv[arg][1] == 'w') {
 961                         write_ref = argv[arg + 1];
 962                         arg++;
 963                 } else if (!strcmp(argv[arg], "--recover")) {
 964                         get_recover = 1;
 965                 }
 966                 arg++;
 967         }
 968         if (argc < arg + 2) {
 969                 usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
 970                 return 1;
 971         }
 972         commit_id = argv[arg];
 973         url = argv[arg + 1];
 974
 975         http_init();
 976
 977         no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
 978
 979         alt = xmalloc(sizeof(*alt));
 980         alt->base = url;
 981         alt->got_indices = 0;
 982         alt->packs = NULL;
 983         alt->next = NULL;
 984
 985         if (pull(commit_id))
 986                 rc = 1;
 987
 988         curl_slist_free_all(no_pragma_header);
 989
 990         http_cleanup();
 991
 992         return rc;
 993 }