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_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
 114                                void *data)
 115 {
 116         unsigned char expn[4096];
 117         size_t size = eltsize * nmemb;
 118         int posn = 0;
 119         struct transfer_request *request = (struct transfer_request *)data;
 120         do {
 121                 ssize_t retval = write(request->local,
 122                                        ptr + posn, size - posn);
 123                 if (retval < 0)
 124                         return posn;
 125                 posn += retval;
 126         } while (posn < size);
 127
 128         request->stream.avail_in = size;
 129         request->stream.next_in = ptr;
 130         do {
 131                 request->stream.next_out = expn;
 132                 request->stream.avail_out = sizeof(expn);
 133                 request->zret = inflate(&request->stream, Z_SYNC_FLUSH);
 134                 SHA1_Update(&request->c, expn,
 135                             sizeof(expn) - request->stream.avail_out);
 136         } while (request->stream.avail_in && request->zret == Z_OK);
 137         data_received++;
 138         return size;
 139 }
 140
 141 #ifdef USE_CURL_MULTI
 142 void process_curl_messages();
 143 void process_request_queue();
 144 #endif
 145
 146 struct active_request_slot *get_active_slot()
 147 {
 148         struct active_request_slot *slot = active_queue_head;
 149         struct active_request_slot *newslot;
 150
 151 #ifdef USE_CURL_MULTI
 152         int num_transfers;
 153
 154         /* Wait for a slot to open up if the queue is full */
 155         while (active_requests >= max_requests) {
 156                 curl_multi_perform(curlm, &num_transfers);
 157                 if (num_transfers < active_requests) {
 158                         process_curl_messages();
 159                 }
 160         }
 161 #endif
 162
 163         while (slot != NULL && slot->in_use) {
 164                 slot = slot->next;
 165         }
 166         if (slot == NULL) {
 167                 newslot = xmalloc(sizeof(*newslot));
 168                 newslot->curl = curl_easy_duphandle(curl_default);
 169                 newslot->in_use = 0;
 170                 newslot->next = NULL;
 171
 172                 slot = active_queue_head;
 173                 if (slot == NULL) {
 174                         active_queue_head = newslot;
 175                 } else {
 176                         while (slot->next != NULL) {
 177                                 slot = slot->next;
 178                         }
 179                         slot->next = newslot;
 180                 }
 181                 slot = newslot;
 182         }
 183
 184         active_requests++;
 185         slot->in_use = 1;
 186         slot->done = 0;
 187         slot->local = NULL;
 188         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, pragma_header);
 189         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_range_header);
 190         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 191
 192         return slot;
 193 }
 194
 195 int start_active_slot(struct active_request_slot *slot)
 196 {
 197 #ifdef USE_CURL_MULTI
 198         CURLMcode curlm_result = curl_multi_add_handle(curlm, slot->curl);
 199
 200         if (curlm_result != CURLM_OK &&
 201             curlm_result != CURLM_CALL_MULTI_PERFORM) {
 202                 active_requests--;
 203                 slot->in_use = 0;
 204                 return 0;
 205         }
 206 #endif
 207         return 1;
 208 }
 209
 210 void run_active_slot(struct active_request_slot *slot)
 211 {
 212 #ifdef USE_CURL_MULTI
 213         int num_transfers;
 214         long last_pos = 0;
 215         long current_pos;
 216         fd_set readfds;
 217         fd_set writefds;
 218         fd_set excfds;
 219         int max_fd;
 220         struct timeval select_timeout;
 221         CURLMcode curlm_result;
 222
 223         while (!slot->done) {
 224                 data_received = 0;
 225                 do {
 226                         curlm_result = curl_multi_perform(curlm,
 227                                                           &num_transfers);
 228                 } while (curlm_result == CURLM_CALL_MULTI_PERFORM);
 229                 if (num_transfers < active_requests) {
 230                         process_curl_messages();
 231                         process_request_queue();
 232                 }
 233
 234                 if (!data_received && slot->local != NULL) {
 235                         current_pos = ftell(slot->local);
 236                         if (current_pos > last_pos)
 237                                 data_received++;
 238                         last_pos = current_pos;
 239                 }
 240
 241                 if (!slot->done && !data_received) {
 242                         max_fd = 0;
 243                         FD_ZERO(&readfds);
 244                         FD_ZERO(&writefds);
 245                         FD_ZERO(&excfds);
 246                         select_timeout.tv_sec = 0;
 247                         select_timeout.tv_usec = 50000;
 248                         select(max_fd, &readfds, &writefds,
 249                                &excfds, &select_timeout);
 250                 }
 251         }
 252 #else
 253         slot->curl_result = curl_easy_perform(slot->curl);
 254         active_requests--;
 255 #endif
 256 }
 257
 258 void start_request(struct transfer_request *request)
 259 {
 260         char *hex = sha1_to_hex(request->sha1);
 261         char prevfile[PATH_MAX];
 262         char *url;
 263         char *posn;
 264         int prevlocal;
 265         unsigned char prev_buf[PREV_BUF_SIZE];
 266         ssize_t prev_read = 0;
 267         long prev_posn = 0;
 268         char range[RANGE_HEADER_SIZE];
 269         struct curl_slist *range_header = NULL;
 270         struct active_request_slot *slot;
 271
 272         snprintf(prevfile, sizeof(prevfile), "%s.prev", request->filename);
 273         unlink(prevfile);
 274         rename(request->tmpfile, prevfile);
 275         unlink(request->tmpfile);
 276
 277         request->local = open(request->tmpfile,
 278                               O_WRONLY | O_CREAT | O_EXCL, 0666);
 279         /* This could have failed due to the "lazy directory creation";
 280          * try to mkdir the last path component.
 281          */
 282         if (request->local < 0 && errno == ENOENT) {
 283                 char *dir = strrchr(request->tmpfile, '/');
 284                 if (dir) {
 285                         *dir = 0;
 286                         mkdir(request->tmpfile, 0777);
 287                         *dir = '/';
 288                 }
 289                 request->local = open(request->tmpfile,
 290                                       O_WRONLY | O_CREAT | O_EXCL, 0666);
 291         }
 292
 293         if (request->local < 0) {
 294                 request->state = ABORTED;
 295                 error("Couldn't create temporary file %s for %s: %s\n",
 296                       request->tmpfile, request->filename, strerror(errno));
 297                 return;
 298         }
 299
 300         memset(&request->stream, 0, sizeof(request->stream));
 301
 302         inflateInit(&request->stream);
 303
 304         SHA1_Init(&request->c);
 305
 306         url = xmalloc(strlen(request->repo->base) + 50);
 307         request->url = xmalloc(strlen(request->repo->base) + 50);
 308         strcpy(url, request->repo->base);
 309         posn = url + strlen(request->repo->base);
 310         strcpy(posn, "objects/");
 311         posn += 8;
 312         memcpy(posn, hex, 2);
 313         posn += 2;
 314         *(posn++) = '/';
 315         strcpy(posn, hex + 2);
 316         strcpy(request->url, url);
 317
 318         /* If a previous temp file is present, process what was already
 319            fetched. */
 320         prevlocal = open(prevfile, O_RDONLY);
 321         if (prevlocal != -1) {
 322                 do {
 323                         prev_read = read(prevlocal, prev_buf, PREV_BUF_SIZE);
 324                         if (prev_read>0) {
 325                                 if (fwrite_sha1_file(prev_buf,
 326                                                      1,
 327                                                      prev_read,
 328                                                      request) == prev_read) {
 329                                         prev_posn += prev_read;
 330                                 } else {
 331                                         prev_read = -1;
 332                                 }
 333                         }
 334                 } while (prev_read > 0);
 335                 close(prevlocal);
 336         }
 337         unlink(prevfile);
 338
 339         /* Reset inflate/SHA1 if there was an error reading the previous temp
 340            file; also rewind to the beginning of the local file. */
 341         if (prev_read == -1) {
 342                 memset(&request->stream, 0, sizeof(request->stream));
 343                 inflateInit(&request->stream);
 344                 SHA1_Init(&request->c);
 345                 if (prev_posn>0) {
 346                         prev_posn = 0;
 347                         lseek(request->local, SEEK_SET, 0);
 348                         ftruncate(request->local, 0);
 349                 }
 350         }
 351
 352         slot = get_active_slot();
 353         curl_easy_setopt(slot->curl, CURLOPT_FILE, request);
 354         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 355         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, request->errorstr);
 356         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 357         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 358
 359         /* If we have successfully processed data from a previous fetch
 360            attempt, only fetch the data we don't already have. */
 361         if (prev_posn>0) {
 362                 if (get_verbosely)
 363                         fprintf(stderr,
 364                                 "Resuming fetch of object %s at byte %ld\n",
 365                                 hex, prev_posn);
 366                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 367                 range_header = curl_slist_append(range_header, range);
 368                 curl_easy_setopt(slot->curl,
 369                                  CURLOPT_HTTPHEADER, range_header);
 370         }
 371
 372         /* Try to get the request started, abort the request on error */
 373         if (!start_active_slot(slot)) {
 374                 request->state = ABORTED;
 375                 close(request->local);
 376                 free(request->url);
 377                 return;
 378         }
 379
 380         request->slot = slot;
 381         request->state = ACTIVE;
 382 }
 383
 384 void finish_request(struct transfer_request *request)
 385 {
 386         fchmod(request->local, 0444);
 387         close(request->local);
 388
 389         if (request->http_code == 416) {
 390                 fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 391         } else if (request->curl_result != CURLE_OK) {
 392                 return;
 393         }
 394
 395         inflateEnd(&request->stream);
 396         SHA1_Final(request->real_sha1, &request->c);
 397         if (request->zret != Z_STREAM_END) {
 398                 unlink(request->tmpfile);
 399                 return;
 400         }
 401         if (memcmp(request->sha1, request->real_sha1, 20)) {
 402                 unlink(request->tmpfile);
 403                 return;
 404         }
 405         request->rename =
 406                 move_temp_to_file(request->tmpfile, request->filename);
 407
 408         if (request->rename == 0)
 409                 pull_say("got %s\n", sha1_to_hex(request->sha1));
 410 }
 411
 412 void release_request(struct transfer_request *request)
 413 {
 414         struct transfer_request *entry = request_queue_head;
 415
 416         if (request == request_queue_head) {
 417                 request_queue_head = request->next;
 418         } else {
 419                 while (entry->next != NULL && entry->next != request)
 420                         entry = entry->next;
 421                 if (entry->next == request)
 422                         entry->next = entry->next->next;
 423         }
 424
 425         free(request->url);
 426         free(request);
 427 }
 428
 429 #ifdef USE_CURL_MULTI
 430 void process_curl_messages()
 431 {
 432         int num_messages;
 433         struct active_request_slot *slot;
 434         struct transfer_request *request = NULL;
 435         CURLMsg *curl_message = curl_multi_info_read(curlm, &num_messages);
 436
 437         while (curl_message != NULL) {
 438                 if (curl_message->msg == CURLMSG_DONE) {
 439                         slot = active_queue_head;
 440                         while (slot != NULL &&
 441                                slot->curl != curl_message->easy_handle)
 442                                 slot = slot->next;
 443                         if (slot != NULL) {
 444                                 curl_multi_remove_handle(curlm, slot->curl);
 445                                 active_requests--;
 446                                 slot->done = 1;
 447                                 slot->in_use = 0;
 448                                 slot->curl_result = curl_message->data.result;
 449                                 request = request_queue_head;
 450                                 while (request != NULL &&
 451                                        request->slot != slot)
 452                                         request = request->next;
 453                         } else {
 454                                 fprintf(stderr, "Received DONE message for unknown request!\n");
 455                         }
 456                         if (request != NULL) {
 457                                 request->curl_result =
 458                                         curl_message->data.result;
 459                                 curl_easy_getinfo(slot->curl,
 460                                                   CURLINFO_HTTP_CODE,
 461                                                   &request->http_code);
 462                                 request->slot = NULL;
 463
 464                                 /* Use alternates if necessary */
 465                                 if (request->http_code == 404 &&
 466                                     request->repo->next != NULL) {
 467                                         request->repo = request->repo->next;
 468                                         start_request(request);
 469                                 } else {
 470                                         finish_request(request);
 471                                         request->state = COMPLETE;
 472                                 }
 473                         }
 474                 } else {
 475                         fprintf(stderr, "Unknown CURL message received: %d\n",
 476                                 (int)curl_message->msg);
 477                 }
 478                 curl_message = curl_multi_info_read(curlm, &num_messages);
 479         }
 480 }
 481
 482 void process_request_queue()
 483 {
 484         struct transfer_request *request = request_queue_head;
 485         int num_transfers;
 486
 487         while (active_requests < max_requests && request != NULL) {
 488                 if (request->state == WAITING) {
 489                         if (has_sha1_file(request->sha1))
 490                                 release_request(request);
 491                         else
 492                                 start_request(request);
 493                         curl_multi_perform(curlm, &num_transfers);
 494                 }
 495                 request = request->next;
 496         }
 497 }
 498 #endif
 499
 500 void prefetch(unsigned char *sha1)
 501 {
 502         struct transfer_request *newreq;
 503         struct transfer_request *tail;
 504         char *filename = sha1_file_name(sha1);
 505
 506         newreq = xmalloc(sizeof(*newreq));
 507         memcpy(newreq->sha1, sha1, 20);
 508         newreq->repo = alt;
 509         newreq->url = NULL;
 510         newreq->local = -1;
 511         newreq->state = WAITING;
 512         snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 513         snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 514                  "%s.temp", filename);
 515         newreq->next = NULL;
 516
 517         if (request_queue_head == NULL) {
 518                 request_queue_head = newreq;
 519         } else {
 520                 tail = request_queue_head;
 521                 while (tail->next != NULL) {
 522                         tail = tail->next;
 523                 }
 524                 tail->next = newreq;
 525         }
 526 #ifdef USE_CURL_MULTI
 527         process_request_queue();
 528         process_curl_messages();
 529 #endif
 530 }
 531
 532 static int fetch_index(struct alt_base *repo, unsigned char *sha1)
 533 {
 534         char *hex = sha1_to_hex(sha1);
 535         char *filename;
 536         char *url;
 537         char tmpfile[PATH_MAX];
 538         long prev_posn = 0;
 539         char range[RANGE_HEADER_SIZE];
 540         struct curl_slist *range_header = NULL;
 541
 542         FILE *indexfile;
 543         struct active_request_slot *slot;
 544
 545         if (has_pack_index(sha1))
 546                 return 0;
 547
 548         if (get_verbosely)
 549                 fprintf(stderr, "Getting index for pack %s\n", hex);
 550
 551         url = xmalloc(strlen(repo->base) + 64);
 552         sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 553
 554         filename = sha1_pack_index_name(sha1);
 555         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 556         indexfile = fopen(tmpfile, "a");
 557         if (!indexfile)
 558                 return error("Unable to open local file %s for pack index",
 559                              filename);
 560
 561         slot = get_active_slot();
 562         curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 563         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 564         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 565         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 566         slot->local = indexfile;
 567
 568         /* If there is data present from a previous transfer attempt,
 569            resume where it left off */
 570         prev_posn = ftell(indexfile);
 571         if (prev_posn>0) {
 572                 if (get_verbosely)
 573                         fprintf(stderr,
 574                                 "Resuming fetch of index for pack %s at byte %ld\n",
 575                                 hex, prev_posn);
 576                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 577                 range_header = curl_slist_append(range_header, range);
 578                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 579         }
 580
 581         if (start_active_slot(slot)) {
 582                 run_active_slot(slot);
 583                 if (slot->curl_result != CURLE_OK) {
 584                         fclose(indexfile);
 585                         return error("Unable to get pack index %s\n%s", url,
 586                                      curl_errorstr);
 587                 }
 588         } else {
 589                 return error("Unable to start request");
 590         }
 591
 592         fclose(indexfile);
 593
 594         return move_temp_to_file(tmpfile, filename);
 595 }
 596
 597 static int setup_index(struct alt_base *repo, unsigned char *sha1)
 598 {
 599         struct packed_git *new_pack;
 600         if (has_pack_file(sha1))
 601                 return 0; // don't list this as something we can get
 602
 603         if (fetch_index(repo, sha1))
 604                 return -1;
 605
 606         new_pack = parse_pack_index(sha1);
 607         new_pack->next = repo->packs;
 608         repo->packs = new_pack;
 609         return 0;
 610 }
 611
 612 static int fetch_alternates(char *base)
 613 {
 614         int ret = 0;
 615         struct buffer buffer;
 616         char *url;
 617         char *data;
 618         int i = 0;
 619         int http_specific = 1;
 620         struct alt_base *tail = alt;
 621
 622         struct active_request_slot *slot;
 623
 624         data = xmalloc(4096);
 625         buffer.size = 4095;
 626         buffer.posn = 0;
 627         buffer.buffer = data;
 628
 629         if (get_verbosely)
 630                 fprintf(stderr, "Getting alternates list\n");
 631
 632         url = xmalloc(strlen(base) + 31);
 633         sprintf(url, "%s/objects/info/http-alternates", base);
 634
 635         slot = get_active_slot();
 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         if (start_active_slot(slot)) {
 640                 run_active_slot(slot);
 641                 if (slot->curl_result != CURLE_OK || !buffer.posn) {
 642                         http_specific = 0;
 643
 644                         sprintf(url, "%s/objects/info/alternates", base);
 645
 646                         slot = get_active_slot();
 647                         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 648                         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 649                                          fwrite_buffer);
 650                         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 651                         if (start_active_slot(slot)) {
 652                                 run_active_slot(slot);
 653                                 if (slot->curl_result != CURLE_OK) {
 654                                         return 0;
 655                                 }
 656                         }
 657                 }
 658         } else {
 659                 return 0;
 660         }
 661
 662         data[buffer.posn] = '\0';
 663
 664         while (i < buffer.posn) {
 665                 int posn = i;
 666                 while (posn < buffer.posn && data[posn] != '\n')
 667                         posn++;
 668                 if (data[posn] == '\n') {
 669                         int okay = 0;
 670                         int serverlen = 0;
 671                         struct alt_base *newalt;
 672                         char *target = NULL;
 673                         if (data[i] == '/') {
 674                                 serverlen = strchr(base + 8, '/') - base;
 675                                 okay = 1;
 676                         } else if (!memcmp(data + i, "../", 3)) {
 677                                 i += 3;
 678                                 serverlen = strlen(base);
 679                                 while (i + 2 < posn &&
 680                                        !memcmp(data + i, "../", 3)) {
 681                                         do {
 682                                                 serverlen--;
 683                                         } while (serverlen &&
 684                                                  base[serverlen - 1] != '/');
 685                                         i += 3;
 686                                 }
 687                                 // If the server got removed, give up.
 688                                 okay = strchr(base, ':') - base + 3 <
 689                                         serverlen;
 690                         } else if (http_specific) {
 691                                 char *colon = strchr(data + i, ':');
 692                                 char *slash = strchr(data + i, '/');
 693                                 if (colon && slash && colon < data + posn &&
 694                                     slash < data + posn && colon < slash) {
 695                                         okay = 1;
 696                                 }
 697                         }
 698                         // skip 'objects' at end
 699                         if (okay) {
 700                                 target = xmalloc(serverlen + posn - i - 6);
 701                                 strncpy(target, base, serverlen);
 702                                 strncpy(target + serverlen, data + i,
 703                                         posn - i - 7);
 704                                 target[serverlen + posn - i - 7] = '\0';
 705                                 if (get_verbosely)
 706                                         fprintf(stderr,
 707                                                 "Also look at %s\n", target);
 708                                 newalt = xmalloc(sizeof(*newalt));
 709                                 newalt->next = NULL;
 710                                 newalt->base = target;
 711                                 newalt->got_indices = 0;
 712                                 newalt->packs = NULL;
 713                                 while (tail->next != NULL)
 714                                         tail = tail->next;
 715                                 tail->next = newalt;
 716                                 ret++;
 717                         }
 718                 }
 719                 i = posn + 1;
 720         }
 721
 722         return ret;
 723 }
 724
 725 static int fetch_indices(struct alt_base *repo)
 726 {
 727         unsigned char sha1[20];
 728         char *url;
 729         struct buffer buffer;
 730         char *data;
 731         int i = 0;
 732
 733         struct active_request_slot *slot;
 734
 735         if (repo->got_indices)
 736                 return 0;
 737
 738         data = xmalloc(4096);
 739         buffer.size = 4096;
 740         buffer.posn = 0;
 741         buffer.buffer = data;
 742
 743         if (get_verbosely)
 744                 fprintf(stderr, "Getting pack list\n");
 745
 746         url = xmalloc(strlen(repo->base) + 21);
 747         sprintf(url, "%s/objects/info/packs", repo->base);
 748
 749         slot = get_active_slot();
 750         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 751         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 752         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 753         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 754         if (start_active_slot(slot)) {
 755                 run_active_slot(slot);
 756                 if (slot->curl_result != CURLE_OK)
 757                         return error("%s", curl_errorstr);
 758         } else {
 759                 return error("Unable to start request");
 760         }
 761
 762         while (i < buffer.posn) {
 763                 switch (data[i]) {
 764                 case 'P':
 765                         i++;
 766                         if (i + 52 < buffer.posn &&
 767                             !strncmp(data + i, " pack-", 6) &&
 768                             !strncmp(data + i + 46, ".pack\n", 6)) {
 769                                 get_sha1_hex(data + i + 6, sha1);
 770                                 setup_index(repo, sha1);
 771                                 i += 51;
 772                                 break;
 773                         }
 774                 default:
 775                         while (data[i] != '\n')
 776                                 i++;
 777                 }
 778                 i++;
 779         }
 780
 781         repo->got_indices = 1;
 782         return 0;
 783 }
 784
 785 static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 786 {
 787         char *url;
 788         struct packed_git *target;
 789         struct packed_git **lst;
 790         FILE *packfile;
 791         char *filename;
 792         char tmpfile[PATH_MAX];
 793         int ret;
 794         long prev_posn = 0;
 795         char range[RANGE_HEADER_SIZE];
 796         struct curl_slist *range_header = NULL;
 797
 798         struct active_request_slot *slot;
 799
 800         if (fetch_indices(repo))
 801                 return -1;
 802         target = find_sha1_pack(sha1, repo->packs);
 803         if (!target)
 804                 return -1;
 805
 806         if (get_verbosely) {
 807                 fprintf(stderr, "Getting pack %s\n",
 808                         sha1_to_hex(target->sha1));
 809                 fprintf(stderr, " which contains %s\n",
 810                         sha1_to_hex(sha1));
 811         }
 812
 813         url = xmalloc(strlen(repo->base) + 65);
 814         sprintf(url, "%s/objects/pack/pack-%s.pack",
 815                 repo->base, sha1_to_hex(target->sha1));
 816
 817         filename = sha1_pack_name(target->sha1);
 818         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 819         packfile = fopen(tmpfile, "a");
 820         if (!packfile)
 821                 return error("Unable to open local file %s for pack",
 822                              filename);
 823
 824         slot = get_active_slot();
 825         curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
 826         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 827         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 828         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 829         slot->local = packfile;
 830
 831         /* If there is data present from a previous transfer attempt,
 832            resume where it left off */
 833         prev_posn = ftell(packfile);
 834         if (prev_posn>0) {
 835                 if (get_verbosely)
 836                         fprintf(stderr,
 837                                 "Resuming fetch of pack %s at byte %ld\n",
 838                                 sha1_to_hex(target->sha1), prev_posn);
 839                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 840                 range_header = curl_slist_append(range_header, range);
 841                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 842         }
 843
 844         if (start_active_slot(slot)) {
 845                 run_active_slot(slot);
 846                 if (slot->curl_result != CURLE_OK) {
 847                         fclose(packfile);
 848                         return error("Unable to get pack file %s\n%s", url,
 849                                      curl_errorstr);
 850                 }
 851         } else {
 852                 return error("Unable to start request");
 853         }
 854
 855         fclose(packfile);
 856
 857         ret = move_temp_to_file(tmpfile, filename);
 858         if (ret)
 859                 return ret;
 860
 861         lst = &repo->packs;
 862         while (*lst != target)
 863                 lst = &((*lst)->next);
 864         *lst = (*lst)->next;
 865
 866         if (verify_pack(target, 0))
 867                 return -1;
 868         install_packed_git(target);
 869
 870         return 0;
 871 }
 872
 873 static int fetch_object(struct alt_base *repo, unsigned char *sha1)
 874 {
 875         char *hex = sha1_to_hex(sha1);
 876         int ret;
 877         struct transfer_request *request = request_queue_head;
 878
 879         while (request != NULL && memcmp(request->sha1, sha1, 20))
 880                 request = request->next;
 881         if (request == NULL)
 882                 return error("Couldn't find request for %s in the queue", hex);
 883
 884         if (has_sha1_file(request->sha1)) {
 885                 release_request(request);
 886                 return 0;
 887         }
 888
 889 #ifdef USE_CURL_MULTI
 890         int num_transfers;
 891         while (request->state == WAITING) {
 892                 curl_multi_perform(curlm, &num_transfers);
 893                 if (num_transfers < active_requests) {
 894                         process_curl_messages();
 895                         process_request_queue();
 896                 }
 897         }
 898 #else
 899         start_request(request);
 900 #endif
 901
 902         while (request->state == ACTIVE) {
 903                 run_active_slot(request->slot);
 904 #ifndef USE_CURL_MULTI
 905                 request->curl_result = request->slot->curl_result;
 906                 curl_easy_getinfo(request->slot->curl,
 907                                   CURLINFO_HTTP_CODE,
 908                                   &request->http_code);
 909                 request->slot = NULL;
 910
 911                 /* Use alternates if necessary */
 912                 if (request->http_code == 404 &&
 913                     request->repo->next != NULL) {
 914                         request->repo = request->repo->next;
 915                         start_request(request);
 916                 } else {
 917                         finish_request(request);
 918                         request->state = COMPLETE;
 919                 }
 920 #endif
 921         }
 922
 923         if (request->state == ABORTED) {
 924                 release_request(request);
 925                 return error("Request for %s aborted", hex);
 926         }
 927
 928         if (request->curl_result != CURLE_OK && request->http_code != 416) {
 929                 ret = error("%s", request->errorstr);
 930                 release_request(request);
 931                 return ret;
 932         }
 933
 934         if (request->zret != Z_STREAM_END) {
 935                 ret = error("File %s (%s) corrupt\n", hex, request->url);
 936                 release_request(request);
 937                 return ret;
 938         }
 939
 940         if (memcmp(request->sha1, request->real_sha1, 20)) {
 941                 release_request(request);
 942                 return error("File %s has bad hash\n", hex);
 943         }
 944
 945         if (request->rename < 0) {
 946                 ret = error("unable to write sha1 filename %s: %s",
 947                             request->filename,
 948                             strerror(request->rename));
 949                 release_request(request);
 950                 return ret;
 951         }
 952
 953         release_request(request);
 954         return 0;
 955 }
 956
 957 int fetch(unsigned char *sha1)
 958 {
 959         struct alt_base *altbase = alt;
 960
 961         if (!fetch_object(altbase, sha1))
 962                 return 0;
 963         while (altbase) {
 964                 if (!fetch_pack(altbase, sha1))
 965                         return 0;
 966                 altbase = altbase->next;
 967         }
 968         return error("Unable to find %s under %s\n", sha1_to_hex(sha1),
 969                      alt->base);
 970 }
 971
 972 static inline int needs_quote(int ch)
 973 {
 974         switch (ch) {
 975         case '/': case '-': case '.':
 976         case 'A'...'Z': case 'a'...'z': case '0'...'9':
 977                 return 0;
 978         default:
 979                 return 1;
 980         }
 981 }
 982
 983 static inline int hex(int v)
 984 {
 985         if (v < 10) return '0' + v;
 986         else return 'A' + v - 10;
 987 }
 988
 989 static char *quote_ref_url(const char *base, const char *ref)
 990 {
 991         const char *cp;
 992         char *dp, *qref;
 993         int len, baselen, ch;
 994
 995         baselen = strlen(base);
 996         len = baselen + 6; /* "refs/" + NUL */
 997         for (cp = ref; (ch = *cp) != 0; cp++, len++)
 998                 if (needs_quote(ch))
 999                         len += 2; /* extra two hex plus replacement % */
1000         qref = xmalloc(len);
1001         memcpy(qref, base, baselen);
1002         memcpy(qref + baselen, "refs/", 5);
1003         for (cp = ref, dp = qref + baselen + 5; (ch = *cp) != 0; cp++) {
1004                 if (needs_quote(ch)) {
1005                         *dp++ = '%';
1006                         *dp++ = hex((ch >> 4) & 0xF);
1007                         *dp++ = hex(ch & 0xF);
1008                 }
1009                 else
1010                         *dp++ = ch;
1011         }
1012         *dp = 0;
1013
1014         return qref;
1015 }
1016
1017 int fetch_ref(char *ref, unsigned char *sha1)
1018 {
1019         char *url;
1020         char hex[42];
1021         struct buffer buffer;
1022         char *base = alt->base;
1023         struct active_request_slot *slot;
1024         buffer.size = 41;
1025         buffer.posn = 0;
1026         buffer.buffer = hex;
1027         hex[41] = '\0';
1028
1029         url = quote_ref_url(base, ref);
1030         slot = get_active_slot();
1031         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
1032         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
1033         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
1034         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
1035         if (start_active_slot(slot)) {
1036                 run_active_slot(slot);
1037                 if (slot->curl_result != CURLE_OK)
1038                         return error("Couldn't get %s for %s\n%s",
1039                                      url, ref, curl_errorstr);
1040         } else {
1041                 return error("Unable to start request");
1042         }
1043
1044         hex[40] = '\0';
1045         get_sha1_hex(hex, sha1);
1046         return 0;
1047 }
1048
1049 int main(int argc, char **argv)
1050 {
1051         char *commit_id;
1052         char *url;
1053         int arg = 1;
1054         struct active_request_slot *slot;
1055
1056         while (arg < argc && argv[arg][0] == '-') {
1057                 if (argv[arg][1] == 't') {
1058                         get_tree = 1;
1059                 } else if (argv[arg][1] == 'c') {
1060                         get_history = 1;
1061                 } else if (argv[arg][1] == 'a') {
1062                         get_all = 1;
1063                         get_tree = 1;
1064                         get_history = 1;
1065                 } else if (argv[arg][1] == 'v') {
1066                         get_verbosely = 1;
1067                 } else if (argv[arg][1] == 'w') {
1068                         write_ref = argv[arg + 1];
1069                         arg++;
1070                 } else if (!strcmp(argv[arg], "--recover")) {
1071                         get_recover = 1;
1072                 }
1073                 arg++;
1074         }
1075         if (argc < arg + 2) {
1076                 usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
1077                 return 1;
1078         }
1079         commit_id = argv[arg];
1080         url = argv[arg + 1];
1081
1082         curl_global_init(CURL_GLOBAL_ALL);
1083
1084 #ifdef USE_CURL_MULTI
1085         char *http_max_requests = getenv("GIT_HTTP_MAX_REQUESTS");
1086         if (http_max_requests != NULL)
1087                 max_requests = atoi(http_max_requests);
1088         if (max_requests < 1)
1089                 max_requests = DEFAULT_MAX_REQUESTS;
1090
1091         curlm = curl_multi_init();
1092         if (curlm == NULL) {
1093                 fprintf(stderr, "Error creating curl multi handle.\n");
1094                 return 1;
1095         }
1096 #endif
1097         pragma_header = curl_slist_append(pragma_header, "Pragma: no-cache");
1098         no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
1099         no_range_header = curl_slist_append(no_range_header, "Range:");
1100
1101         curl_default = curl_easy_init();
1102
1103         curl_ssl_verify = getenv("GIT_SSL_NO_VERIFY") ? 0 : 1;
1104         curl_easy_setopt(curl_default, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
1105 #if LIBCURL_VERSION_NUM >= 0x070907
1106         curl_easy_setopt(curl_default, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
1107 #endif
1108
1109         if ((ssl_cert = getenv("GIT_SSL_CERT")) != NULL) {
1110                 curl_easy_setopt(curl_default, CURLOPT_SSLCERT, ssl_cert);
1111         }
1112 #if LIBCURL_VERSION_NUM >= 0x070902
1113         if ((ssl_key = getenv("GIT_SSL_KEY")) != NULL) {
1114                 curl_easy_setopt(curl_default, CURLOPT_SSLKEY, ssl_key);
1115         }
1116 #endif
1117 #if LIBCURL_VERSION_NUM >= 0x070908
1118         if ((ssl_capath = getenv("GIT_SSL_CAPATH")) != NULL) {
1119                 curl_easy_setopt(curl_default, CURLOPT_CAPATH, ssl_capath);
1120         }
1121 #endif
1122         if ((ssl_cainfo = getenv("GIT_SSL_CAINFO")) != NULL) {
1123                 curl_easy_setopt(curl_default, CURLOPT_CAINFO, ssl_cainfo);
1124         }
1125         curl_easy_setopt(curl_default, CURLOPT_FAILONERROR, 1);
1126
1127         alt = xmalloc(sizeof(*alt));
1128         alt->base = url;
1129         alt->got_indices = 0;
1130         alt->packs = NULL;
1131         alt->next = NULL;
1132         fetch_alternates(alt->base);
1133
1134         if (pull(commit_id))
1135                 return 1;
1136
1137         curl_slist_free_all(pragma_header);
1138         curl_slist_free_all(no_pragma_header);
1139         curl_slist_free_all(no_range_header);
1140         curl_easy_cleanup(curl_default);
1141         slot = active_queue_head;
1142         while (slot != NULL) {
1143                 curl_easy_cleanup(slot->curl);
1144                 slot = slot->next;
1145         }
1146 #ifdef USE_CURL_MULTI
1147         curl_multi_cleanup(curlm);
1148 #endif
1149         curl_global_cleanup();
1150         return 0;
1151 }