http-fetch.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "pack.h"
   4 #include "fetch.h"
   5
   6 #include <curl/curl.h>
   7 #include <curl/easy.h>
   8
   9 #if LIBCURL_VERSION_NUM >= 0x070908
  10 #define USE_CURL_MULTI
  11 #define DEFAULT_MAX_REQUESTS 5
  12 #endif
  13
  14 #if LIBCURL_VERSION_NUM < 0x070704
  15 #define curl_global_cleanup() do { /* nothing */ } while(0)
  16 #endif
  17 #if LIBCURL_VERSION_NUM < 0x070800
  18 #define curl_global_init(a) do { /* nothing */ } while(0)
  19 #endif
  20
  21 #define PREV_BUF_SIZE 4096
  22 #define RANGE_HEADER_SIZE 30
  23
  24 static int active_requests = 0;
  25 static int data_received;
  26
  27 #ifdef USE_CURL_MULTI
  28 static int max_requests = DEFAULT_MAX_REQUESTS;
  29 static CURLM *curlm;
  30 #endif
  31 static CURL *curl_default;
  32 static struct curl_slist *pragma_header;
  33 static struct curl_slist *no_pragma_header;
  34 static struct curl_slist *no_range_header;
  35 static char curl_errorstr[CURL_ERROR_SIZE];
  36
  37 struct alt_base
  38 {
  39         char *base;
  40         int got_indices;
  41         struct packed_git *packs;
  42         struct alt_base *next;
  43 };
  44
  45 static struct alt_base *alt = NULL;
  46
  47 enum transfer_state {
  48         WAITING,
  49         ABORTED,
  50         ACTIVE,
  51         COMPLETE,
  52 };
  53
  54 struct transfer_request
  55 {
  56         unsigned char sha1[20];
  57         struct alt_base *repo;
  58         char *url;
  59         char filename[PATH_MAX];
  60         char tmpfile[PATH_MAX];
  61         int local;
  62         enum transfer_state state;
  63         CURLcode curl_result;
  64         char errorstr[CURL_ERROR_SIZE];
  65         long http_code;
  66         unsigned char real_sha1[20];
  67         SHA_CTX c;
  68         z_stream stream;
  69         int zret;
  70         int rename;
  71         struct active_request_slot *slot;
  72         struct transfer_request *next;
  73 };
  74
  75 struct active_request_slot
  76 {
  77         CURL *curl;
  78         FILE *local;
  79         int in_use;
  80         int done;
  81         CURLcode curl_result;
  82         struct active_request_slot *next;
  83 };
  84
  85 static struct transfer_request *request_queue_head = NULL;
  86 static struct active_request_slot *active_queue_head = NULL;
  87
  88 static int curl_ssl_verify;
  89 static char *ssl_cert;
  90 static char *ssl_key;
  91 static char *ssl_capath;
  92 static char *ssl_cainfo;
  93
  94 struct buffer
  95 {
  96         size_t posn;
  97         size_t size;
  98         void *buffer;
  99 };
 100
 101 static size_t fwrite_buffer(void *ptr, size_t eltsize, size_t nmemb,
 102                             struct buffer *buffer)
 103 {
 104         size_t size = eltsize * nmemb;
 105         if (size > buffer->size - buffer->posn)
 106                 size = buffer->size - buffer->posn;
 107         memcpy(buffer->buffer + buffer->posn, ptr, size);
 108         buffer->posn += size;
 109         data_received++;
 110         return size;
 111 }
 112
 113 static size_t fwrite_buffer_dynamic(const void *ptr, size_t eltsize,
 114                                     size_t nmemb, struct buffer *buffer)
 115 {
 116         size_t size = eltsize * nmemb;
 117         if (size > buffer->size - buffer->posn) {
 118                 buffer->size = buffer->size * 3 / 2;
 119                 if (buffer->size < buffer->posn + size)
 120                         buffer->size = buffer->posn + size;
 121                 buffer->buffer = xrealloc(buffer->buffer, buffer->size);
 122         }
 123         memcpy(buffer->buffer + buffer->posn, ptr, size);
 124         buffer->posn += size;
 125         data_received++;
 126         return size;
 127 }
 128
 129 static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
 130                                void *data)
 131 {
 132         unsigned char expn[4096];
 133         size_t size = eltsize * nmemb;
 134         int posn = 0;
 135         struct transfer_request *request = (struct transfer_request *)data;
 136         do {
 137                 ssize_t retval = write(request->local,
 138                                        ptr + posn, size - posn);
 139                 if (retval < 0)
 140                         return posn;
 141                 posn += retval;
 142         } while (posn < size);
 143
 144         request->stream.avail_in = size;
 145         request->stream.next_in = ptr;
 146         do {
 147                 request->stream.next_out = expn;
 148                 request->stream.avail_out = sizeof(expn);
 149                 request->zret = inflate(&request->stream, Z_SYNC_FLUSH);
 150                 SHA1_Update(&request->c, expn,
 151                             sizeof(expn) - request->stream.avail_out);
 152         } while (request->stream.avail_in && request->zret == Z_OK);
 153         data_received++;
 154         return size;
 155 }
 156
 157 #ifdef USE_CURL_MULTI
 158 void process_curl_messages();
 159 void process_request_queue();
 160 #endif
 161
 162 struct active_request_slot *get_active_slot()
 163 {
 164         struct active_request_slot *slot = active_queue_head;
 165         struct active_request_slot *newslot;
 166
 167 #ifdef USE_CURL_MULTI
 168         int num_transfers;
 169
 170         /* Wait for a slot to open up if the queue is full */
 171         while (active_requests >= max_requests) {
 172                 curl_multi_perform(curlm, &num_transfers);
 173                 if (num_transfers < active_requests) {
 174                         process_curl_messages();
 175                 }
 176         }
 177 #endif
 178
 179         while (slot != NULL && slot->in_use) {
 180                 slot = slot->next;
 181         }
 182         if (slot == NULL) {
 183                 newslot = xmalloc(sizeof(*newslot));
 184                 newslot->curl = curl_easy_duphandle(curl_default);
 185                 newslot->in_use = 0;
 186                 newslot->next = NULL;
 187
 188                 slot = active_queue_head;
 189                 if (slot == NULL) {
 190                         active_queue_head = newslot;
 191                 } else {
 192                         while (slot->next != NULL) {
 193                                 slot = slot->next;
 194                         }
 195                         slot->next = newslot;
 196                 }
 197                 slot = newslot;
 198         }
 199
 200         active_requests++;
 201         slot->in_use = 1;
 202         slot->done = 0;
 203         slot->local = NULL;
 204         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, pragma_header);
 205         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_range_header);
 206         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 207
 208         return slot;
 209 }
 210
 211 int start_active_slot(struct active_request_slot *slot)
 212 {
 213 #ifdef USE_CURL_MULTI
 214         CURLMcode curlm_result = curl_multi_add_handle(curlm, slot->curl);
 215
 216         if (curlm_result != CURLM_OK &&
 217             curlm_result != CURLM_CALL_MULTI_PERFORM) {
 218                 active_requests--;
 219                 slot->in_use = 0;
 220                 return 0;
 221         }
 222 #endif
 223         return 1;
 224 }
 225
 226 void run_active_slot(struct active_request_slot *slot)
 227 {
 228 #ifdef USE_CURL_MULTI
 229         int num_transfers;
 230         long last_pos = 0;
 231         long current_pos;
 232         fd_set readfds;
 233         fd_set writefds;
 234         fd_set excfds;
 235         int max_fd;
 236         struct timeval select_timeout;
 237         CURLMcode curlm_result;
 238
 239         while (!slot->done) {
 240                 data_received = 0;
 241                 do {
 242                         curlm_result = curl_multi_perform(curlm,
 243                                                           &num_transfers);
 244                 } while (curlm_result == CURLM_CALL_MULTI_PERFORM);
 245                 if (num_transfers < active_requests) {
 246                         process_curl_messages();
 247                         process_request_queue();
 248                 }
 249
 250                 if (!data_received && slot->local != NULL) {
 251                         current_pos = ftell(slot->local);
 252                         if (current_pos > last_pos)
 253                                 data_received++;
 254                         last_pos = current_pos;
 255                 }
 256
 257                 if (!slot->done && !data_received) {
 258                         max_fd = 0;
 259                         FD_ZERO(&readfds);
 260                         FD_ZERO(&writefds);
 261                         FD_ZERO(&excfds);
 262                         select_timeout.tv_sec = 0;
 263                         select_timeout.tv_usec = 50000;
 264                         select(max_fd, &readfds, &writefds,
 265                                &excfds, &select_timeout);
 266                 }
 267         }
 268 #else
 269         slot->curl_result = curl_easy_perform(slot->curl);
 270         active_requests--;
 271 #endif
 272 }
 273
 274 void start_request(struct transfer_request *request)
 275 {
 276         char *hex = sha1_to_hex(request->sha1);
 277         char prevfile[PATH_MAX];
 278         char *url;
 279         char *posn;
 280         int prevlocal;
 281         unsigned char prev_buf[PREV_BUF_SIZE];
 282         ssize_t prev_read = 0;
 283         long prev_posn = 0;
 284         char range[RANGE_HEADER_SIZE];
 285         struct curl_slist *range_header = NULL;
 286         struct active_request_slot *slot;
 287
 288         snprintf(prevfile, sizeof(prevfile), "%s.prev", request->filename);
 289         unlink(prevfile);
 290         rename(request->tmpfile, prevfile);
 291         unlink(request->tmpfile);
 292
 293         request->local = open(request->tmpfile,
 294                               O_WRONLY | O_CREAT | O_EXCL, 0666);
 295         /* This could have failed due to the "lazy directory creation";
 296          * try to mkdir the last path component.
 297          */
 298         if (request->local < 0 && errno == ENOENT) {
 299                 char *dir = strrchr(request->tmpfile, '/');
 300                 if (dir) {
 301                         *dir = 0;
 302                         mkdir(request->tmpfile, 0777);
 303                         *dir = '/';
 304                 }
 305                 request->local = open(request->tmpfile,
 306                                       O_WRONLY | O_CREAT | O_EXCL, 0666);
 307         }
 308
 309         if (request->local < 0) {
 310                 request->state = ABORTED;
 311                 error("Couldn't create temporary file %s for %s: %s\n",
 312                       request->tmpfile, request->filename, strerror(errno));
 313                 return;
 314         }
 315
 316         memset(&request->stream, 0, sizeof(request->stream));
 317
 318         inflateInit(&request->stream);
 319
 320         SHA1_Init(&request->c);
 321
 322         url = xmalloc(strlen(request->repo->base) + 50);
 323         request->url = xmalloc(strlen(request->repo->base) + 50);
 324         strcpy(url, request->repo->base);
 325         posn = url + strlen(request->repo->base);
 326         strcpy(posn, "objects/");
 327         posn += 8;
 328         memcpy(posn, hex, 2);
 329         posn += 2;
 330         *(posn++) = '/';
 331         strcpy(posn, hex + 2);
 332         strcpy(request->url, url);
 333
 334         /* If a previous temp file is present, process what was already
 335            fetched. */
 336         prevlocal = open(prevfile, O_RDONLY);
 337         if (prevlocal != -1) {
 338                 do {
 339                         prev_read = read(prevlocal, prev_buf, PREV_BUF_SIZE);
 340                         if (prev_read>0) {
 341                                 if (fwrite_sha1_file(prev_buf,
 342                                                      1,
 343                                                      prev_read,
 344                                                      request) == prev_read) {
 345                                         prev_posn += prev_read;
 346                                 } else {
 347                                         prev_read = -1;
 348                                 }
 349                         }
 350                 } while (prev_read > 0);
 351                 close(prevlocal);
 352         }
 353         unlink(prevfile);
 354
 355         /* Reset inflate/SHA1 if there was an error reading the previous temp
 356            file; also rewind to the beginning of the local file. */
 357         if (prev_read == -1) {
 358                 memset(&request->stream, 0, sizeof(request->stream));
 359                 inflateInit(&request->stream);
 360                 SHA1_Init(&request->c);
 361                 if (prev_posn>0) {
 362                         prev_posn = 0;
 363                         lseek(request->local, SEEK_SET, 0);
 364                         ftruncate(request->local, 0);
 365                 }
 366         }
 367
 368         slot = get_active_slot();
 369         curl_easy_setopt(slot->curl, CURLOPT_FILE, request);
 370         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 371         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, request->errorstr);
 372         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 373         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 374
 375         /* If we have successfully processed data from a previous fetch
 376            attempt, only fetch the data we don't already have. */
 377         if (prev_posn>0) {
 378                 if (get_verbosely)
 379                         fprintf(stderr,
 380                                 "Resuming fetch of object %s at byte %ld\n",
 381                                 hex, prev_posn);
 382                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 383                 range_header = curl_slist_append(range_header, range);
 384                 curl_easy_setopt(slot->curl,
 385                                  CURLOPT_HTTPHEADER, range_header);
 386         }
 387
 388         /* Try to get the request started, abort the request on error */
 389         if (!start_active_slot(slot)) {
 390                 request->state = ABORTED;
 391                 close(request->local);
 392                 free(request->url);
 393                 return;
 394         }
 395
 396         request->slot = slot;
 397         request->state = ACTIVE;
 398 }
 399
 400 void finish_request(struct transfer_request *request)
 401 {
 402         fchmod(request->local, 0444);
 403         close(request->local);
 404
 405         if (request->http_code == 416) {
 406                 fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 407         } else if (request->curl_result != CURLE_OK) {
 408                 return;
 409         }
 410
 411         inflateEnd(&request->stream);
 412         SHA1_Final(request->real_sha1, &request->c);
 413         if (request->zret != Z_STREAM_END) {
 414                 unlink(request->tmpfile);
 415                 return;
 416         }
 417         if (memcmp(request->sha1, request->real_sha1, 20)) {
 418                 unlink(request->tmpfile);
 419                 return;
 420         }
 421         request->rename =
 422                 move_temp_to_file(request->tmpfile, request->filename);
 423
 424         if (request->rename == 0)
 425                 pull_say("got %s\n", sha1_to_hex(request->sha1));
 426 }
 427
 428 void release_request(struct transfer_request *request)
 429 {
 430         struct transfer_request *entry = request_queue_head;
 431
 432         if (request == request_queue_head) {
 433                 request_queue_head = request->next;
 434         } else {
 435                 while (entry->next != NULL && entry->next != request)
 436                         entry = entry->next;
 437                 if (entry->next == request)
 438                         entry->next = entry->next->next;
 439         }
 440
 441         free(request->url);
 442         free(request);
 443 }
 444
 445 #ifdef USE_CURL_MULTI
 446 void process_curl_messages()
 447 {
 448         int num_messages;
 449         struct active_request_slot *slot;
 450         struct transfer_request *request = NULL;
 451         CURLMsg *curl_message = curl_multi_info_read(curlm, &num_messages);
 452
 453         while (curl_message != NULL) {
 454                 if (curl_message->msg == CURLMSG_DONE) {
 455                         slot = active_queue_head;
 456                         while (slot != NULL &&
 457                                slot->curl != curl_message->easy_handle)
 458                                 slot = slot->next;
 459                         if (slot != NULL) {
 460                                 curl_multi_remove_handle(curlm, slot->curl);
 461                                 active_requests--;
 462                                 slot->done = 1;
 463                                 slot->in_use = 0;
 464                                 slot->curl_result = curl_message->data.result;
 465                                 request = request_queue_head;
 466                                 while (request != NULL &&
 467                                        request->slot != slot)
 468                                         request = request->next;
 469                         } else {
 470                                 fprintf(stderr, "Received DONE message for unknown request!\n");
 471                         }
 472                         if (request != NULL) {
 473                                 request->curl_result =
 474                                         curl_message->data.result;
 475                                 curl_easy_getinfo(slot->curl,
 476                                                   CURLINFO_HTTP_CODE,
 477                                                   &request->http_code);
 478                                 request->slot = NULL;
 479
 480                                 /* Use alternates if necessary */
 481                                 if (request->http_code == 404 &&
 482                                     request->repo->next != NULL) {
 483                                         request->repo = request->repo->next;
 484                                         start_request(request);
 485                                 } else {
 486                                         finish_request(request);
 487                                         request->state = COMPLETE;
 488                                 }
 489                         }
 490                 } else {
 491                         fprintf(stderr, "Unknown CURL message received: %d\n",
 492                                 (int)curl_message->msg);
 493                 }
 494                 curl_message = curl_multi_info_read(curlm, &num_messages);
 495         }
 496 }
 497
 498 void process_request_queue()
 499 {
 500         struct transfer_request *request = request_queue_head;
 501         int num_transfers;
 502
 503         while (active_requests < max_requests && request != NULL) {
 504                 if (request->state == WAITING) {
 505                         if (has_sha1_file(request->sha1))
 506                                 release_request(request);
 507                         else
 508                                 start_request(request);
 509                         curl_multi_perform(curlm, &num_transfers);
 510                 }
 511                 request = request->next;
 512         }
 513 }
 514 #endif
 515
 516 void prefetch(unsigned char *sha1)
 517 {
 518         struct transfer_request *newreq;
 519         struct transfer_request *tail;
 520         char *filename = sha1_file_name(sha1);
 521
 522         newreq = xmalloc(sizeof(*newreq));
 523         memcpy(newreq->sha1, sha1, 20);
 524         newreq->repo = alt;
 525         newreq->url = NULL;
 526         newreq->local = -1;
 527         newreq->state = WAITING;
 528         snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 529         snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 530                  "%s.temp", filename);
 531         newreq->next = NULL;
 532
 533         if (request_queue_head == NULL) {
 534                 request_queue_head = newreq;
 535         } else {
 536                 tail = request_queue_head;
 537                 while (tail->next != NULL) {
 538                         tail = tail->next;
 539                 }
 540                 tail->next = newreq;
 541         }
 542 #ifdef USE_CURL_MULTI
 543         process_request_queue();
 544         process_curl_messages();
 545 #endif
 546 }
 547
 548 static int fetch_index(struct alt_base *repo, unsigned char *sha1)
 549 {
 550         char *hex = sha1_to_hex(sha1);
 551         char *filename;
 552         char *url;
 553         char tmpfile[PATH_MAX];
 554         long prev_posn = 0;
 555         char range[RANGE_HEADER_SIZE];
 556         struct curl_slist *range_header = NULL;
 557
 558         FILE *indexfile;
 559         struct active_request_slot *slot;
 560
 561         if (has_pack_index(sha1))
 562                 return 0;
 563
 564         if (get_verbosely)
 565                 fprintf(stderr, "Getting index for pack %s\n", hex);
 566
 567         url = xmalloc(strlen(repo->base) + 64);
 568         sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 569
 570         filename = sha1_pack_index_name(sha1);
 571         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 572         indexfile = fopen(tmpfile, "a");
 573         if (!indexfile)
 574                 return error("Unable to open local file %s for pack index",
 575                              filename);
 576
 577         slot = get_active_slot();
 578         curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 579         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 580         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 581         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 582         slot->local = indexfile;
 583
 584         /* If there is data present from a previous transfer attempt,
 585            resume where it left off */
 586         prev_posn = ftell(indexfile);
 587         if (prev_posn>0) {
 588                 if (get_verbosely)
 589                         fprintf(stderr,
 590                                 "Resuming fetch of index for pack %s at byte %ld\n",
 591                                 hex, prev_posn);
 592                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 593                 range_header = curl_slist_append(range_header, range);
 594                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 595         }
 596
 597         if (start_active_slot(slot)) {
 598                 run_active_slot(slot);
 599                 if (slot->curl_result != CURLE_OK) {
 600                         fclose(indexfile);
 601                         return error("Unable to get pack index %s\n%s", url,
 602                                      curl_errorstr);
 603                 }
 604         } else {
 605                 return error("Unable to start request");
 606         }
 607
 608         fclose(indexfile);
 609
 610         return move_temp_to_file(tmpfile, filename);
 611 }
 612
 613 static int setup_index(struct alt_base *repo, unsigned char *sha1)
 614 {
 615         struct packed_git *new_pack;
 616         if (has_pack_file(sha1))
 617                 return 0; // don't list this as something we can get
 618
 619         if (fetch_index(repo, sha1))
 620                 return -1;
 621
 622         new_pack = parse_pack_index(sha1);
 623         new_pack->next = repo->packs;
 624         repo->packs = new_pack;
 625         return 0;
 626 }
 627
 628 static int fetch_alternates(char *base)
 629 {
 630         int ret = 0;
 631         struct buffer buffer;
 632         char *url;
 633         char *data;
 634         int i = 0;
 635         int http_specific = 1;
 636         struct alt_base *tail = alt;
 637         static const char null_byte = '\0';
 638
 639         struct active_request_slot *slot;
 640
 641         data = xmalloc(4096);
 642         buffer.size = 4096;
 643         buffer.posn = 0;
 644         buffer.buffer = data;
 645
 646         if (get_verbosely)
 647                 fprintf(stderr, "Getting alternates list\n");
 648
 649         url = xmalloc(strlen(base) + 31);
 650         sprintf(url, "%s/objects/info/http-alternates", base);
 651
 652         slot = get_active_slot();
 653         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 654         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 655                          fwrite_buffer_dynamic);
 656         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 657         if (start_active_slot(slot)) {
 658                 run_active_slot(slot);
 659                 if (slot->curl_result != CURLE_OK || !buffer.posn) {
 660                         http_specific = 0;
 661
 662                         sprintf(url, "%s/objects/info/alternates", base);
 663
 664                         slot = get_active_slot();
 665                         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 666                         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 667                                          fwrite_buffer_dynamic);
 668                         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 669                         if (start_active_slot(slot)) {
 670                                 run_active_slot(slot);
 671                                 if (slot->curl_result != CURLE_OK) {
 672                                         free(buffer.buffer);
 673                                         return 0;
 674                                 }
 675                         }
 676                 }
 677         } else {
 678                 free(buffer.buffer);
 679                 return 0;
 680         }
 681
 682         fwrite_buffer_dynamic(&null_byte, 1, 1, &buffer);
 683         buffer.posn--;
 684         data = buffer.buffer;
 685
 686         while (i < buffer.posn) {
 687                 int posn = i;
 688                 while (posn < buffer.posn && data[posn] != '\n')
 689                         posn++;
 690                 if (data[posn] == '\n') {
 691                         int okay = 0;
 692                         int serverlen = 0;
 693                         struct alt_base *newalt;
 694                         char *target = NULL;
 695                         if (data[i] == '/') {
 696                                 serverlen = strchr(base + 8, '/') - base;
 697                                 okay = 1;
 698                         } else if (!memcmp(data + i, "../", 3)) {
 699                                 i += 3;
 700                                 serverlen = strlen(base);
 701                                 while (i + 2 < posn &&
 702                                        !memcmp(data + i, "../", 3)) {
 703                                         do {
 704                                                 serverlen--;
 705                                         } while (serverlen &&
 706                                                  base[serverlen - 1] != '/');
 707                                         i += 3;
 708                                 }
 709                                 // If the server got removed, give up.
 710                                 okay = strchr(base, ':') - base + 3 <
 711                                         serverlen;
 712                         } else if (http_specific) {
 713                                 char *colon = strchr(data + i, ':');
 714                                 char *slash = strchr(data + i, '/');
 715                                 if (colon && slash && colon < data + posn &&
 716                                     slash < data + posn && colon < slash) {
 717                                         okay = 1;
 718                                 }
 719                         }
 720                         // skip 'objects' at end
 721                         if (okay) {
 722                                 target = xmalloc(serverlen + posn - i - 6);
 723                                 strncpy(target, base, serverlen);
 724                                 strncpy(target + serverlen, data + i,
 725                                         posn - i - 7);
 726                                 target[serverlen + posn - i - 7] = '\0';
 727                                 if (get_verbosely)
 728                                         fprintf(stderr,
 729                                                 "Also look at %s\n", target);
 730                                 newalt = xmalloc(sizeof(*newalt));
 731                                 newalt->next = NULL;
 732                                 newalt->base = target;
 733                                 newalt->got_indices = 0;
 734                                 newalt->packs = NULL;
 735                                 while (tail->next != NULL)
 736                                         tail = tail->next;
 737                                 tail->next = newalt;
 738                                 ret++;
 739                         }
 740                 }
 741                 i = posn + 1;
 742         }
 743
 744         free(buffer.buffer);
 745         return ret;
 746 }
 747
 748 static int fetch_indices(struct alt_base *repo)
 749 {
 750         unsigned char sha1[20];
 751         char *url;
 752         struct buffer buffer;
 753         char *data;
 754         int i = 0;
 755
 756         struct active_request_slot *slot;
 757
 758         if (repo->got_indices)
 759                 return 0;
 760
 761         data = xmalloc(4096);
 762         buffer.size = 4096;
 763         buffer.posn = 0;
 764         buffer.buffer = data;
 765
 766         if (get_verbosely)
 767                 fprintf(stderr, "Getting pack list\n");
 768
 769         url = xmalloc(strlen(repo->base) + 21);
 770         sprintf(url, "%s/objects/info/packs", repo->base);
 771
 772         slot = get_active_slot();
 773         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 774         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 775                          fwrite_buffer_dynamic);
 776         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 777         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 778         if (start_active_slot(slot)) {
 779                 run_active_slot(slot);
 780                 if (slot->curl_result != CURLE_OK) {
 781                         free(buffer.buffer);
 782                         return error("%s", curl_errorstr);
 783                 }
 784         } else {
 785                 free(buffer.buffer);
 786                 return error("Unable to start request");
 787         }
 788
 789         data = buffer.buffer;
 790         while (i < buffer.posn) {
 791                 switch (data[i]) {
 792                 case 'P':
 793                         i++;
 794                         if (i + 52 < buffer.posn &&
 795                             !strncmp(data + i, " pack-", 6) &&
 796                             !strncmp(data + i + 46, ".pack\n", 6)) {
 797                                 get_sha1_hex(data + i + 6, sha1);
 798                                 setup_index(repo, sha1);
 799                                 i += 51;
 800                                 break;
 801                         }
 802                 default:
 803                         while (data[i] != '\n')
 804                                 i++;
 805                 }
 806                 i++;
 807         }
 808
 809         free(buffer.buffer);
 810         repo->got_indices = 1;
 811         return 0;
 812 }
 813
 814 static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 815 {
 816         char *url;
 817         struct packed_git *target;
 818         struct packed_git **lst;
 819         FILE *packfile;
 820         char *filename;
 821         char tmpfile[PATH_MAX];
 822         int ret;
 823         long prev_posn = 0;
 824         char range[RANGE_HEADER_SIZE];
 825         struct curl_slist *range_header = NULL;
 826
 827         struct active_request_slot *slot;
 828
 829         if (fetch_indices(repo))
 830                 return -1;
 831         target = find_sha1_pack(sha1, repo->packs);
 832         if (!target)
 833                 return -1;
 834
 835         if (get_verbosely) {
 836                 fprintf(stderr, "Getting pack %s\n",
 837                         sha1_to_hex(target->sha1));
 838                 fprintf(stderr, " which contains %s\n",
 839                         sha1_to_hex(sha1));
 840         }
 841
 842         url = xmalloc(strlen(repo->base) + 65);
 843         sprintf(url, "%s/objects/pack/pack-%s.pack",
 844                 repo->base, sha1_to_hex(target->sha1));
 845
 846         filename = sha1_pack_name(target->sha1);
 847         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 848         packfile = fopen(tmpfile, "a");
 849         if (!packfile)
 850                 return error("Unable to open local file %s for pack",
 851                              filename);
 852
 853         slot = get_active_slot();
 854         curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
 855         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 856         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 857         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 858         slot->local = packfile;
 859
 860         /* If there is data present from a previous transfer attempt,
 861            resume where it left off */
 862         prev_posn = ftell(packfile);
 863         if (prev_posn>0) {
 864                 if (get_verbosely)
 865                         fprintf(stderr,
 866                                 "Resuming fetch of pack %s at byte %ld\n",
 867                                 sha1_to_hex(target->sha1), prev_posn);
 868                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 869                 range_header = curl_slist_append(range_header, range);
 870                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 871         }
 872
 873         if (start_active_slot(slot)) {
 874                 run_active_slot(slot);
 875                 if (slot->curl_result != CURLE_OK) {
 876                         fclose(packfile);
 877                         return error("Unable to get pack file %s\n%s", url,
 878                                      curl_errorstr);
 879                 }
 880         } else {
 881                 return error("Unable to start request");
 882         }
 883
 884         fclose(packfile);
 885
 886         ret = move_temp_to_file(tmpfile, filename);
 887         if (ret)
 888                 return ret;
 889
 890         lst = &repo->packs;
 891         while (*lst != target)
 892                 lst = &((*lst)->next);
 893         *lst = (*lst)->next;
 894
 895         if (verify_pack(target, 0))
 896                 return -1;
 897         install_packed_git(target);
 898
 899         return 0;
 900 }
 901
 902 static int fetch_object(struct alt_base *repo, unsigned char *sha1)
 903 {
 904         char *hex = sha1_to_hex(sha1);
 905         int ret;
 906         struct transfer_request *request = request_queue_head;
 907
 908         while (request != NULL && memcmp(request->sha1, sha1, 20))
 909                 request = request->next;
 910         if (request == NULL)
 911                 return error("Couldn't find request for %s in the queue", hex);
 912
 913         if (has_sha1_file(request->sha1)) {
 914                 release_request(request);
 915                 return 0;
 916         }
 917
 918 #ifdef USE_CURL_MULTI
 919         int num_transfers;
 920         while (request->state == WAITING) {
 921                 curl_multi_perform(curlm, &num_transfers);
 922                 if (num_transfers < active_requests) {
 923                         process_curl_messages();
 924                         process_request_queue();
 925                 }
 926         }
 927 #else
 928         start_request(request);
 929 #endif
 930
 931         while (request->state == ACTIVE) {
 932                 run_active_slot(request->slot);
 933 #ifndef USE_CURL_MULTI
 934                 request->curl_result = request->slot->curl_result;
 935                 curl_easy_getinfo(request->slot->curl,
 936                                   CURLINFO_HTTP_CODE,
 937                                   &request->http_code);
 938                 request->slot = NULL;
 939
 940                 /* Use alternates if necessary */
 941                 if (request->http_code == 404 &&
 942                     request->repo->next != NULL) {
 943                         request->repo = request->repo->next;
 944                         start_request(request);
 945                 } else {
 946                         finish_request(request);
 947                         request->state = COMPLETE;
 948                 }
 949 #endif
 950         }
 951
 952         if (request->state == ABORTED) {
 953                 release_request(request);
 954                 return error("Request for %s aborted", hex);
 955         }
 956
 957         if (request->curl_result != CURLE_OK && request->http_code != 416) {
 958                 ret = error("%s", request->errorstr);
 959                 release_request(request);
 960                 return ret;
 961         }
 962
 963         if (request->zret != Z_STREAM_END) {
 964                 ret = error("File %s (%s) corrupt\n", hex, request->url);
 965                 release_request(request);
 966                 return ret;
 967         }
 968
 969         if (memcmp(request->sha1, request->real_sha1, 20)) {
 970                 release_request(request);
 971                 return error("File %s has bad hash\n", hex);
 972         }
 973
 974         if (request->rename < 0) {
 975                 ret = error("unable to write sha1 filename %s: %s",
 976                             request->filename,
 977                             strerror(request->rename));
 978                 release_request(request);
 979                 return ret;
 980         }
 981
 982         release_request(request);
 983         return 0;
 984 }
 985
 986 int fetch(unsigned char *sha1)
 987 {
 988         struct alt_base *altbase = alt;
 989
 990         if (!fetch_object(altbase, sha1))
 991                 return 0;
 992         while (altbase) {
 993                 if (!fetch_pack(altbase, sha1))
 994                         return 0;
 995                 altbase = altbase->next;
 996         }
 997         return error("Unable to find %s under %s\n", sha1_to_hex(sha1),
 998                      alt->base);
 999 }
1000
1001 static inline int needs_quote(int ch)
1002 {
1003         switch (ch) {
1004         case '/': case '-': case '.':
1005         case 'A'...'Z': case 'a'...'z': case '0'...'9':
1006                 return 0;
1007         default:
1008                 return 1;
1009         }
1010 }
1011
1012 static inline int hex(int v)
1013 {
1014         if (v < 10) return '0' + v;
1015         else return 'A' + v - 10;
1016 }
1017
1018 static char *quote_ref_url(const char *base, const char *ref)
1019 {
1020         const char *cp;
1021         char *dp, *qref;
1022         int len, baselen, ch;
1023
1024         baselen = strlen(base);
1025         len = baselen + 6; /* "refs/" + NUL */
1026         for (cp = ref; (ch = *cp) != 0; cp++, len++)
1027                 if (needs_quote(ch))
1028                         len += 2; /* extra two hex plus replacement % */
1029         qref = xmalloc(len);
1030         memcpy(qref, base, baselen);
1031         memcpy(qref + baselen, "refs/", 5);
1032         for (cp = ref, dp = qref + baselen + 5; (ch = *cp) != 0; cp++) {
1033                 if (needs_quote(ch)) {
1034                         *dp++ = '%';
1035                         *dp++ = hex((ch >> 4) & 0xF);
1036                         *dp++ = hex(ch & 0xF);
1037                 }
1038                 else
1039                         *dp++ = ch;
1040         }
1041         *dp = 0;
1042
1043         return qref;
1044 }
1045
1046 int fetch_ref(char *ref, unsigned char *sha1)
1047 {
1048         char *url;
1049         char hex[42];
1050         struct buffer buffer;
1051         char *base = alt->base;
1052         struct active_request_slot *slot;
1053         buffer.size = 41;
1054         buffer.posn = 0;
1055         buffer.buffer = hex;
1056         hex[41] = '\0';
1057
1058         url = quote_ref_url(base, ref);
1059         slot = get_active_slot();
1060         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
1061         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
1062         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
1063         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
1064         if (start_active_slot(slot)) {
1065                 run_active_slot(slot);
1066                 if (slot->curl_result != CURLE_OK)
1067                         return error("Couldn't get %s for %s\n%s",
1068                                      url, ref, curl_errorstr);
1069         } else {
1070                 return error("Unable to start request");
1071         }
1072
1073         hex[40] = '\0';
1074         get_sha1_hex(hex, sha1);
1075         return 0;
1076 }
1077
1078 int main(int argc, char **argv)
1079 {
1080         char *commit_id;
1081         char *url;
1082         int arg = 1;
1083         struct active_request_slot *slot;
1084
1085         while (arg < argc && argv[arg][0] == '-') {
1086                 if (argv[arg][1] == 't') {
1087                         get_tree = 1;
1088                 } else if (argv[arg][1] == 'c') {
1089                         get_history = 1;
1090                 } else if (argv[arg][1] == 'a') {
1091                         get_all = 1;
1092                         get_tree = 1;
1093                         get_history = 1;
1094                 } else if (argv[arg][1] == 'v') {
1095                         get_verbosely = 1;
1096                 } else if (argv[arg][1] == 'w') {
1097                         write_ref = argv[arg + 1];
1098                         arg++;
1099                 } else if (!strcmp(argv[arg], "--recover")) {
1100                         get_recover = 1;
1101                 }
1102                 arg++;
1103         }
1104         if (argc < arg + 2) {
1105                 usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
1106                 return 1;
1107         }
1108         commit_id = argv[arg];
1109         url = argv[arg + 1];
1110
1111         curl_global_init(CURL_GLOBAL_ALL);
1112
1113 #ifdef USE_CURL_MULTI
1114         char *http_max_requests = getenv("GIT_HTTP_MAX_REQUESTS");
1115         if (http_max_requests != NULL)
1116                 max_requests = atoi(http_max_requests);
1117         if (max_requests < 1)
1118                 max_requests = DEFAULT_MAX_REQUESTS;
1119
1120         curlm = curl_multi_init();
1121         if (curlm == NULL) {
1122                 fprintf(stderr, "Error creating curl multi handle.\n");
1123                 return 1;
1124         }
1125 #endif
1126         pragma_header = curl_slist_append(pragma_header, "Pragma: no-cache");
1127         no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
1128         no_range_header = curl_slist_append(no_range_header, "Range:");
1129
1130         curl_default = curl_easy_init();
1131
1132         curl_ssl_verify = getenv("GIT_SSL_NO_VERIFY") ? 0 : 1;
1133         curl_easy_setopt(curl_default, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
1134 #if LIBCURL_VERSION_NUM >= 0x070907
1135         curl_easy_setopt(curl_default, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
1136 #endif
1137
1138         if ((ssl_cert = getenv("GIT_SSL_CERT")) != NULL) {
1139                 curl_easy_setopt(curl_default, CURLOPT_SSLCERT, ssl_cert);
1140         }
1141 #if LIBCURL_VERSION_NUM >= 0x070902
1142         if ((ssl_key = getenv("GIT_SSL_KEY")) != NULL) {
1143                 curl_easy_setopt(curl_default, CURLOPT_SSLKEY, ssl_key);
1144         }
1145 #endif
1146 #if LIBCURL_VERSION_NUM >= 0x070908
1147         if ((ssl_capath = getenv("GIT_SSL_CAPATH")) != NULL) {
1148                 curl_easy_setopt(curl_default, CURLOPT_CAPATH, ssl_capath);
1149         }
1150 #endif
1151         if ((ssl_cainfo = getenv("GIT_SSL_CAINFO")) != NULL) {
1152                 curl_easy_setopt(curl_default, CURLOPT_CAINFO, ssl_cainfo);
1153         }
1154         curl_easy_setopt(curl_default, CURLOPT_FAILONERROR, 1);
1155
1156         alt = xmalloc(sizeof(*alt));
1157         alt->base = url;
1158         alt->got_indices = 0;
1159         alt->packs = NULL;
1160         alt->next = NULL;
1161         fetch_alternates(alt->base);
1162
1163         if (pull(commit_id))
1164                 return 1;
1165
1166         curl_slist_free_all(pragma_header);
1167         curl_slist_free_all(no_pragma_header);
1168         curl_slist_free_all(no_range_header);
1169         curl_easy_cleanup(curl_default);
1170         slot = active_queue_head;
1171         while (slot != NULL) {
1172                 curl_easy_cleanup(slot->curl);
1173                 slot = slot->next;
1174         }
1175 #ifdef USE_CURL_MULTI
1176         curl_multi_cleanup(curlm);
1177 #endif
1178         curl_global_cleanup();
1179         return 0;
1180 }