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 #if LIBCURL_VERSION_NUM < 0x070c04
  22 #define NO_CURL_EASY_DUPHANDLE
  23 #endif
  24
  25 #define PREV_BUF_SIZE 4096
  26 #define RANGE_HEADER_SIZE 30
  27
  28 static int got_alternates = 0;
  29 static int active_requests = 0;
  30 static int data_received;
  31
  32 #ifdef USE_CURL_MULTI
  33 static int max_requests = -1;
  34 static CURLM *curlm;
  35 #endif
  36 #ifndef NO_CURL_EASY_DUPHANDLE
  37 static CURL *curl_default;
  38 #endif
  39 static struct curl_slist *pragma_header;
  40 static struct curl_slist *no_pragma_header;
  41 static struct curl_slist *no_range_header;
  42 static char curl_errorstr[CURL_ERROR_SIZE];
  43
  44 struct alt_base
  45 {
  46         char *base;
  47         int got_indices;
  48         struct packed_git *packs;
  49         struct alt_base *next;
  50 };
  51
  52 static struct alt_base *alt = NULL;
  53
  54 enum transfer_state {
  55         WAITING,
  56         ABORTED,
  57         ACTIVE,
  58         COMPLETE,
  59 };
  60
  61 struct transfer_request
  62 {
  63         unsigned char sha1[20];
  64         struct alt_base *repo;
  65         char *url;
  66         char filename[PATH_MAX];
  67         char tmpfile[PATH_MAX];
  68         int local;
  69         enum transfer_state state;
  70         CURLcode curl_result;
  71         char errorstr[CURL_ERROR_SIZE];
  72         long http_code;
  73         unsigned char real_sha1[20];
  74         SHA_CTX c;
  75         z_stream stream;
  76         int zret;
  77         int rename;
  78         struct active_request_slot *slot;
  79         struct transfer_request *next;
  80 };
  81
  82 struct active_request_slot
  83 {
  84         CURL *curl;
  85         FILE *local;
  86         int in_use;
  87         int done;
  88         CURLcode curl_result;
  89         long http_code;
  90         struct active_request_slot *next;
  91 };
  92
  93 static struct transfer_request *request_queue_head = NULL;
  94 static struct active_request_slot *active_queue_head = NULL;
  95
  96 static int curl_ssl_verify = -1;
  97 static char *ssl_cert = NULL;
  98 #if LIBCURL_VERSION_NUM >= 0x070902
  99 static char *ssl_key = NULL;
 100 #endif
 101 #if LIBCURL_VERSION_NUM >= 0x070908
 102 static char *ssl_capath = NULL;
 103 #endif
 104 static char *ssl_cainfo = NULL;
 105 static long curl_low_speed_limit = -1;
 106 static long curl_low_speed_time = -1;
 107
 108 struct buffer
 109 {
 110         size_t posn;
 111         size_t size;
 112         void *buffer;
 113 };
 114
 115 static int http_options(const char *var, const char *value)
 116 {
 117         if (!strcmp("http.sslverify", var)) {
 118                 if (curl_ssl_verify == -1) {
 119                         curl_ssl_verify = git_config_bool(var, value);
 120                 }
 121                 return 0;
 122         }
 123
 124         if (!strcmp("http.sslcert", var)) {
 125                 if (ssl_cert == NULL) {
 126                         ssl_cert = xmalloc(strlen(value)+1);
 127                         strcpy(ssl_cert, value);
 128                 }
 129                 return 0;
 130         }
 131 #if LIBCURL_VERSION_NUM >= 0x070902
 132         if (!strcmp("http.sslkey", var)) {
 133                 if (ssl_key == NULL) {
 134                         ssl_key = xmalloc(strlen(value)+1);
 135                         strcpy(ssl_key, value);
 136                 }
 137                 return 0;
 138         }
 139 #endif
 140 #if LIBCURL_VERSION_NUM >= 0x070908
 141         if (!strcmp("http.sslcapath", var)) {
 142                 if (ssl_capath == NULL) {
 143                         ssl_capath = xmalloc(strlen(value)+1);
 144                         strcpy(ssl_capath, value);
 145                 }
 146                 return 0;
 147         }
 148 #endif
 149         if (!strcmp("http.sslcainfo", var)) {
 150                 if (ssl_cainfo == NULL) {
 151                         ssl_cainfo = xmalloc(strlen(value)+1);
 152                         strcpy(ssl_cainfo, value);
 153                 }
 154                 return 0;
 155         }
 156
 157 #ifdef USE_CURL_MULTI
 158         if (!strcmp("http.maxrequests", var)) {
 159                 if (max_requests == -1)
 160                         max_requests = git_config_int(var, value);
 161                 return 0;
 162         }
 163 #endif
 164
 165         if (!strcmp("http.lowspeedlimit", var)) {
 166                 if (curl_low_speed_limit == -1)
 167                         curl_low_speed_limit = (long)git_config_int(var, value);
 168                 return 0;
 169         }
 170         if (!strcmp("http.lowspeedtime", var)) {
 171                 if (curl_low_speed_time == -1)
 172                         curl_low_speed_time = (long)git_config_int(var, value);
 173                 return 0;
 174         }
 175
 176         /* Fall back on the default ones */
 177         return git_default_config(var, value);
 178 }
 179
 180 static size_t fwrite_buffer(void *ptr, size_t eltsize, size_t nmemb,
 181                             struct buffer *buffer)
 182 {
 183         size_t size = eltsize * nmemb;
 184         if (size > buffer->size - buffer->posn)
 185                 size = buffer->size - buffer->posn;
 186         memcpy(buffer->buffer + buffer->posn, ptr, size);
 187         buffer->posn += size;
 188         data_received++;
 189         return size;
 190 }
 191
 192 static size_t fwrite_buffer_dynamic(const void *ptr, size_t eltsize,
 193                                     size_t nmemb, struct buffer *buffer)
 194 {
 195         size_t size = eltsize * nmemb;
 196         if (size > buffer->size - buffer->posn) {
 197                 buffer->size = buffer->size * 3 / 2;
 198                 if (buffer->size < buffer->posn + size)
 199                         buffer->size = buffer->posn + size;
 200                 buffer->buffer = xrealloc(buffer->buffer, buffer->size);
 201         }
 202         memcpy(buffer->buffer + buffer->posn, ptr, size);
 203         buffer->posn += size;
 204         data_received++;
 205         return size;
 206 }
 207
 208 static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
 209                                void *data)
 210 {
 211         unsigned char expn[4096];
 212         size_t size = eltsize * nmemb;
 213         int posn = 0;
 214         struct transfer_request *request = (struct transfer_request *)data;
 215         do {
 216                 ssize_t retval = write(request->local,
 217                                        ptr + posn, size - posn);
 218                 if (retval < 0)
 219                         return posn;
 220                 posn += retval;
 221         } while (posn < size);
 222
 223         request->stream.avail_in = size;
 224         request->stream.next_in = ptr;
 225         do {
 226                 request->stream.next_out = expn;
 227                 request->stream.avail_out = sizeof(expn);
 228                 request->zret = inflate(&request->stream, Z_SYNC_FLUSH);
 229                 SHA1_Update(&request->c, expn,
 230                             sizeof(expn) - request->stream.avail_out);
 231         } while (request->stream.avail_in && request->zret == Z_OK);
 232         data_received++;
 233         return size;
 234 }
 235
 236 #ifdef USE_CURL_MULTI
 237 static void process_curl_messages(void);
 238 static void process_request_queue(void);
 239 #endif
 240 static int fetch_alternates(char *base);
 241
 242 static CURL* get_curl_handle(void)
 243 {
 244         CURL* result = curl_easy_init();
 245
 246         curl_easy_setopt(result, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
 247 #if LIBCURL_VERSION_NUM >= 0x070907
 248         curl_easy_setopt(result, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
 249 #endif
 250
 251         if (ssl_cert != NULL)
 252                 curl_easy_setopt(result, CURLOPT_SSLCERT, ssl_cert);
 253 #if LIBCURL_VERSION_NUM >= 0x070902
 254         if (ssl_key != NULL)
 255                 curl_easy_setopt(result, CURLOPT_SSLKEY, ssl_key);
 256 #endif
 257 #if LIBCURL_VERSION_NUM >= 0x070908
 258         if (ssl_capath != NULL)
 259                 curl_easy_setopt(result, CURLOPT_CAPATH, ssl_capath);
 260 #endif
 261         if (ssl_cainfo != NULL)
 262                 curl_easy_setopt(result, CURLOPT_CAINFO, ssl_cainfo);
 263         curl_easy_setopt(result, CURLOPT_FAILONERROR, 1);
 264
 265         if (curl_low_speed_limit > 0 && curl_low_speed_time > 0) {
 266                 curl_easy_setopt(result, CURLOPT_LOW_SPEED_LIMIT,
 267                                  curl_low_speed_limit);
 268                 curl_easy_setopt(result, CURLOPT_LOW_SPEED_TIME,
 269                                  curl_low_speed_time);
 270         }
 271
 272         curl_easy_setopt(result, CURLOPT_FOLLOWLOCATION, 1);
 273
 274         return result;
 275 }
 276
 277 static struct active_request_slot *get_active_slot(void)
 278 {
 279         struct active_request_slot *slot = active_queue_head;
 280         struct active_request_slot *newslot;
 281
 282 #ifdef USE_CURL_MULTI
 283         int num_transfers;
 284
 285         /* Wait for a slot to open up if the queue is full */
 286         while (active_requests >= max_requests) {
 287                 curl_multi_perform(curlm, &num_transfers);
 288                 if (num_transfers < active_requests) {
 289                         process_curl_messages();
 290                 }
 291         }
 292 #endif
 293
 294         while (slot != NULL && slot->in_use) {
 295                 slot = slot->next;
 296         }
 297         if (slot == NULL) {
 298                 newslot = xmalloc(sizeof(*newslot));
 299                 newslot->curl = NULL;
 300                 newslot->in_use = 0;
 301                 newslot->next = NULL;
 302
 303                 slot = active_queue_head;
 304                 if (slot == NULL) {
 305                         active_queue_head = newslot;
 306                 } else {
 307                         while (slot->next != NULL) {
 308                                 slot = slot->next;
 309                         }
 310                         slot->next = newslot;
 311                 }
 312                 slot = newslot;
 313         }
 314
 315         if (slot->curl == NULL) {
 316 #ifdef NO_CURL_EASY_DUPHANDLE
 317                 slot->curl = get_curl_handle();
 318 #else
 319                 slot->curl = curl_easy_duphandle(curl_default);
 320 #endif
 321         }
 322
 323         active_requests++;
 324         slot->in_use = 1;
 325         slot->done = 0;
 326         slot->local = NULL;
 327         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, pragma_header);
 328         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_range_header);
 329         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 330
 331         return slot;
 332 }
 333
 334 static int start_active_slot(struct active_request_slot *slot)
 335 {
 336 #ifdef USE_CURL_MULTI
 337         CURLMcode curlm_result = curl_multi_add_handle(curlm, slot->curl);
 338
 339         if (curlm_result != CURLM_OK &&
 340             curlm_result != CURLM_CALL_MULTI_PERFORM) {
 341                 active_requests--;
 342                 slot->in_use = 0;
 343                 return 0;
 344         }
 345 #endif
 346         return 1;
 347 }
 348
 349 static void run_active_slot(struct active_request_slot *slot)
 350 {
 351 #ifdef USE_CURL_MULTI
 352         int num_transfers;
 353         long last_pos = 0;
 354         long current_pos;
 355         fd_set readfds;
 356         fd_set writefds;
 357         fd_set excfds;
 358         int max_fd;
 359         struct timeval select_timeout;
 360         CURLMcode curlm_result;
 361
 362         while (!slot->done) {
 363                 data_received = 0;
 364                 do {
 365                         curlm_result = curl_multi_perform(curlm,
 366                                                           &num_transfers);
 367                 } while (curlm_result == CURLM_CALL_MULTI_PERFORM);
 368                 if (num_transfers < active_requests) {
 369                         process_curl_messages();
 370                         process_request_queue();
 371                 }
 372
 373                 if (!data_received && slot->local != NULL) {
 374                         current_pos = ftell(slot->local);
 375                         if (current_pos > last_pos)
 376                                 data_received++;
 377                         last_pos = current_pos;
 378                 }
 379
 380                 if (!slot->done && !data_received) {
 381                         max_fd = 0;
 382                         FD_ZERO(&readfds);
 383                         FD_ZERO(&writefds);
 384                         FD_ZERO(&excfds);
 385                         select_timeout.tv_sec = 0;
 386                         select_timeout.tv_usec = 50000;
 387                         select(max_fd, &readfds, &writefds,
 388                                &excfds, &select_timeout);
 389                 }
 390         }
 391 #else
 392         slot->curl_result = curl_easy_perform(slot->curl);
 393         active_requests--;
 394 #endif
 395 }
 396
 397 static void start_request(struct transfer_request *request)
 398 {
 399         char *hex = sha1_to_hex(request->sha1);
 400         char prevfile[PATH_MAX];
 401         char *url;
 402         char *posn;
 403         int prevlocal;
 404         unsigned char prev_buf[PREV_BUF_SIZE];
 405         ssize_t prev_read = 0;
 406         long prev_posn = 0;
 407         char range[RANGE_HEADER_SIZE];
 408         struct curl_slist *range_header = NULL;
 409         struct active_request_slot *slot;
 410
 411         snprintf(prevfile, sizeof(prevfile), "%s.prev", request->filename);
 412         unlink(prevfile);
 413         rename(request->tmpfile, prevfile);
 414         unlink(request->tmpfile);
 415
 416         request->local = open(request->tmpfile,
 417                               O_WRONLY | O_CREAT | O_EXCL, 0666);
 418         /* This could have failed due to the "lazy directory creation";
 419          * try to mkdir the last path component.
 420          */
 421         if (request->local < 0 && errno == ENOENT) {
 422                 char *dir = strrchr(request->tmpfile, '/');
 423                 if (dir) {
 424                         *dir = 0;
 425                         mkdir(request->tmpfile, 0777);
 426                         *dir = '/';
 427                 }
 428                 request->local = open(request->tmpfile,
 429                                       O_WRONLY | O_CREAT | O_EXCL, 0666);
 430         }
 431
 432         if (request->local < 0) {
 433                 request->state = ABORTED;
 434                 error("Couldn't create temporary file %s for %s: %s\n",
 435                       request->tmpfile, request->filename, strerror(errno));
 436                 return;
 437         }
 438
 439         memset(&request->stream, 0, sizeof(request->stream));
 440
 441         inflateInit(&request->stream);
 442
 443         SHA1_Init(&request->c);
 444
 445         url = xmalloc(strlen(request->repo->base) + 50);
 446         request->url = xmalloc(strlen(request->repo->base) + 50);
 447         strcpy(url, request->repo->base);
 448         posn = url + strlen(request->repo->base);
 449         strcpy(posn, "objects/");
 450         posn += 8;
 451         memcpy(posn, hex, 2);
 452         posn += 2;
 453         *(posn++) = '/';
 454         strcpy(posn, hex + 2);
 455         strcpy(request->url, url);
 456
 457         /* If a previous temp file is present, process what was already
 458            fetched. */
 459         prevlocal = open(prevfile, O_RDONLY);
 460         if (prevlocal != -1) {
 461                 do {
 462                         prev_read = read(prevlocal, prev_buf, PREV_BUF_SIZE);
 463                         if (prev_read>0) {
 464                                 if (fwrite_sha1_file(prev_buf,
 465                                                      1,
 466                                                      prev_read,
 467                                                      request) == prev_read) {
 468                                         prev_posn += prev_read;
 469                                 } else {
 470                                         prev_read = -1;
 471                                 }
 472                         }
 473                 } while (prev_read > 0);
 474                 close(prevlocal);
 475         }
 476         unlink(prevfile);
 477
 478         /* Reset inflate/SHA1 if there was an error reading the previous temp
 479            file; also rewind to the beginning of the local file. */
 480         if (prev_read == -1) {
 481                 memset(&request->stream, 0, sizeof(request->stream));
 482                 inflateInit(&request->stream);
 483                 SHA1_Init(&request->c);
 484                 if (prev_posn>0) {
 485                         prev_posn = 0;
 486                         lseek(request->local, SEEK_SET, 0);
 487                         ftruncate(request->local, 0);
 488                 }
 489         }
 490
 491         slot = get_active_slot();
 492         curl_easy_setopt(slot->curl, CURLOPT_FILE, request);
 493         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 494         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, request->errorstr);
 495         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 496         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 497
 498         /* If we have successfully processed data from a previous fetch
 499            attempt, only fetch the data we don't already have. */
 500         if (prev_posn>0) {
 501                 if (get_verbosely)
 502                         fprintf(stderr,
 503                                 "Resuming fetch of object %s at byte %ld\n",
 504                                 hex, prev_posn);
 505                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 506                 range_header = curl_slist_append(range_header, range);
 507                 curl_easy_setopt(slot->curl,
 508                                  CURLOPT_HTTPHEADER, range_header);
 509         }
 510
 511         /* Try to get the request started, abort the request on error */
 512         if (!start_active_slot(slot)) {
 513                 request->state = ABORTED;
 514                 close(request->local);
 515                 free(request->url);
 516                 return;
 517         }
 518
 519         request->slot = slot;
 520         request->state = ACTIVE;
 521 }
 522
 523 static void finish_request(struct transfer_request *request)
 524 {
 525         struct stat st;
 526
 527         fchmod(request->local, 0444);
 528         close(request->local);
 529
 530         if (request->http_code == 416) {
 531                 fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 532         } else if (request->curl_result != CURLE_OK) {
 533                 if (stat(request->tmpfile, &st) == 0)
 534                         if (st.st_size == 0)
 535                                 unlink(request->tmpfile);
 536                 return;
 537         }
 538
 539         inflateEnd(&request->stream);
 540         SHA1_Final(request->real_sha1, &request->c);
 541         if (request->zret != Z_STREAM_END) {
 542                 unlink(request->tmpfile);
 543                 return;
 544         }
 545         if (memcmp(request->sha1, request->real_sha1, 20)) {
 546                 unlink(request->tmpfile);
 547                 return;
 548         }
 549         request->rename =
 550                 move_temp_to_file(request->tmpfile, request->filename);
 551
 552         if (request->rename == 0)
 553                 pull_say("got %s\n", sha1_to_hex(request->sha1));
 554 }
 555
 556 static void release_request(struct transfer_request *request)
 557 {
 558         struct transfer_request *entry = request_queue_head;
 559
 560         if (request == request_queue_head) {
 561                 request_queue_head = request->next;
 562         } else {
 563                 while (entry->next != NULL && entry->next != request)
 564                         entry = entry->next;
 565                 if (entry->next == request)
 566                         entry->next = entry->next->next;
 567         }
 568
 569         free(request->url);
 570         free(request);
 571 }
 572
 573 #ifdef USE_CURL_MULTI
 574 static void process_curl_messages(void)
 575 {
 576         int num_messages;
 577         struct active_request_slot *slot;
 578         struct transfer_request *request = NULL;
 579         CURLMsg *curl_message = curl_multi_info_read(curlm, &num_messages);
 580
 581         while (curl_message != NULL) {
 582                 if (curl_message->msg == CURLMSG_DONE) {
 583                         int curl_result = curl_message->data.result;
 584                         slot = active_queue_head;
 585                         while (slot != NULL &&
 586                                slot->curl != curl_message->easy_handle)
 587                                 slot = slot->next;
 588                         if (slot != NULL) {
 589                                 curl_multi_remove_handle(curlm, slot->curl);
 590                                 active_requests--;
 591                                 slot->done = 1;
 592                                 slot->in_use = 0;
 593                                 slot->curl_result = curl_result;
 594                                 curl_easy_getinfo(slot->curl,
 595                                                   CURLINFO_HTTP_CODE,
 596                                                   &slot->http_code);
 597                                 request = request_queue_head;
 598                                 while (request != NULL &&
 599                                        request->slot != slot)
 600                                         request = request->next;
 601                         } else {
 602                                 fprintf(stderr, "Received DONE message for unknown request!\n");
 603                         }
 604                         if (request != NULL) {
 605                                 request->curl_result = curl_result;
 606                                 request->http_code = slot->http_code;
 607                                 request->slot = NULL;
 608                                 request->state = COMPLETE;
 609
 610                                 /* Use alternates if necessary */
 611                                 if (request->http_code == 404) {
 612                                         fetch_alternates(alt->base);
 613                                         if (request->repo->next != NULL) {
 614                                                 request->repo =
 615                                                         request->repo->next;
 616                                                 start_request(request);
 617                                         }
 618                                 } else {
 619                                         finish_request(request);
 620                                 }
 621                         }
 622                 } else {
 623                         fprintf(stderr, "Unknown CURL message received: %d\n",
 624                                 (int)curl_message->msg);
 625                 }
 626                 curl_message = curl_multi_info_read(curlm, &num_messages);
 627         }
 628 }
 629
 630 static void process_request_queue(void)
 631 {
 632         struct transfer_request *request = request_queue_head;
 633         struct active_request_slot *slot = active_queue_head;
 634         int num_transfers;
 635
 636         while (active_requests < max_requests && request != NULL) {
 637                 if (request->state == WAITING) {
 638                         if (has_sha1_file(request->sha1))
 639                                 release_request(request);
 640                         else
 641                                 start_request(request);
 642                         curl_multi_perform(curlm, &num_transfers);
 643                 }
 644                 request = request->next;
 645         }
 646
 647         while (slot != NULL) {
 648                 if (!slot->in_use && slot->curl != NULL) {
 649                         curl_easy_cleanup(slot->curl);
 650                         slot->curl = NULL;
 651                 }
 652                 slot = slot->next;
 653         }
 654 }
 655 #endif
 656
 657 void prefetch(unsigned char *sha1)
 658 {
 659         struct transfer_request *newreq;
 660         struct transfer_request *tail;
 661         char *filename = sha1_file_name(sha1);
 662
 663         newreq = xmalloc(sizeof(*newreq));
 664         memcpy(newreq->sha1, sha1, 20);
 665         newreq->repo = alt;
 666         newreq->url = NULL;
 667         newreq->local = -1;
 668         newreq->state = WAITING;
 669         snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 670         snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 671                  "%s.temp", filename);
 672         newreq->next = NULL;
 673
 674         if (request_queue_head == NULL) {
 675                 request_queue_head = newreq;
 676         } else {
 677                 tail = request_queue_head;
 678                 while (tail->next != NULL) {
 679                         tail = tail->next;
 680                 }
 681                 tail->next = newreq;
 682         }
 683 #ifdef USE_CURL_MULTI
 684         process_request_queue();
 685         process_curl_messages();
 686 #endif
 687 }
 688
 689 static int fetch_index(struct alt_base *repo, unsigned char *sha1)
 690 {
 691         char *hex = sha1_to_hex(sha1);
 692         char *filename;
 693         char *url;
 694         char tmpfile[PATH_MAX];
 695         long prev_posn = 0;
 696         char range[RANGE_HEADER_SIZE];
 697         struct curl_slist *range_header = NULL;
 698
 699         FILE *indexfile;
 700         struct active_request_slot *slot;
 701
 702         if (has_pack_index(sha1))
 703                 return 0;
 704
 705         if (get_verbosely)
 706                 fprintf(stderr, "Getting index for pack %s\n", hex);
 707
 708         url = xmalloc(strlen(repo->base) + 64);
 709         sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 710
 711         filename = sha1_pack_index_name(sha1);
 712         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 713         indexfile = fopen(tmpfile, "a");
 714         if (!indexfile)
 715                 return error("Unable to open local file %s for pack index",
 716                              filename);
 717
 718         slot = get_active_slot();
 719         curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 720         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 721         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 722         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 723         slot->local = indexfile;
 724
 725         /* If there is data present from a previous transfer attempt,
 726            resume where it left off */
 727         prev_posn = ftell(indexfile);
 728         if (prev_posn>0) {
 729                 if (get_verbosely)
 730                         fprintf(stderr,
 731                                 "Resuming fetch of index for pack %s at byte %ld\n",
 732                                 hex, prev_posn);
 733                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 734                 range_header = curl_slist_append(range_header, range);
 735                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 736         }
 737
 738         if (start_active_slot(slot)) {
 739                 run_active_slot(slot);
 740                 if (slot->curl_result != CURLE_OK) {
 741                         fclose(indexfile);
 742                         return error("Unable to get pack index %s\n%s", url,
 743                                      curl_errorstr);
 744                 }
 745         } else {
 746                 return error("Unable to start request");
 747         }
 748
 749         fclose(indexfile);
 750
 751         return move_temp_to_file(tmpfile, filename);
 752 }
 753
 754 static int setup_index(struct alt_base *repo, unsigned char *sha1)
 755 {
 756         struct packed_git *new_pack;
 757         if (has_pack_file(sha1))
 758                 return 0; // don't list this as something we can get
 759
 760         if (fetch_index(repo, sha1))
 761                 return -1;
 762
 763         new_pack = parse_pack_index(sha1);
 764         new_pack->next = repo->packs;
 765         repo->packs = new_pack;
 766         return 0;
 767 }
 768
 769 static int fetch_alternates(char *base)
 770 {
 771         int ret = 0;
 772         struct buffer buffer;
 773         char *url;
 774         char *data;
 775         int i = 0;
 776         int http_specific = 1;
 777         struct alt_base *tail = alt;
 778         static const char null_byte = '\0';
 779
 780         struct active_request_slot *slot;
 781
 782         if (got_alternates)
 783                 return 0;
 784
 785         data = xmalloc(4096);
 786         buffer.size = 4096;
 787         buffer.posn = 0;
 788         buffer.buffer = data;
 789
 790         if (get_verbosely)
 791                 fprintf(stderr, "Getting alternates list\n");
 792
 793         url = xmalloc(strlen(base) + 31);
 794         sprintf(url, "%s/objects/info/http-alternates", base);
 795
 796         slot = get_active_slot();
 797         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 798         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 799                          fwrite_buffer_dynamic);
 800         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 801         if (start_active_slot(slot)) {
 802                 run_active_slot(slot);
 803                 if (slot->curl_result != CURLE_OK || !buffer.posn) {
 804                         http_specific = 0;
 805
 806                         sprintf(url, "%s/objects/info/alternates", base);
 807
 808                         slot = get_active_slot();
 809                         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 810                         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 811                                          fwrite_buffer_dynamic);
 812                         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 813                         if (start_active_slot(slot)) {
 814                                 run_active_slot(slot);
 815                                 if (slot->curl_result != CURLE_OK) {
 816                                         free(buffer.buffer);
 817                                         if (slot->http_code == 404)
 818                                                 got_alternates = 1;
 819                                         return 0;
 820                                 }
 821                         }
 822                 }
 823         } else {
 824                 free(buffer.buffer);
 825                 return 0;
 826         }
 827
 828         fwrite_buffer_dynamic(&null_byte, 1, 1, &buffer);
 829         buffer.posn--;
 830         data = buffer.buffer;
 831
 832         while (i < buffer.posn) {
 833                 int posn = i;
 834                 while (posn < buffer.posn && data[posn] != '\n')
 835                         posn++;
 836                 if (data[posn] == '\n') {
 837                         int okay = 0;
 838                         int serverlen = 0;
 839                         struct alt_base *newalt;
 840                         char *target = NULL;
 841                         if (data[i] == '/') {
 842                                 serverlen = strchr(base + 8, '/') - base;
 843                                 okay = 1;
 844                         } else if (!memcmp(data + i, "../", 3)) {
 845                                 i += 3;
 846                                 serverlen = strlen(base);
 847                                 while (i + 2 < posn &&
 848                                        !memcmp(data + i, "../", 3)) {
 849                                         do {
 850                                                 serverlen--;
 851                                         } while (serverlen &&
 852                                                  base[serverlen - 1] != '/');
 853                                         i += 3;
 854                                 }
 855                                 // If the server got removed, give up.
 856                                 okay = strchr(base, ':') - base + 3 <
 857                                         serverlen;
 858                         } else if (http_specific) {
 859                                 char *colon = strchr(data + i, ':');
 860                                 char *slash = strchr(data + i, '/');
 861                                 if (colon && slash && colon < data + posn &&
 862                                     slash < data + posn && colon < slash) {
 863                                         okay = 1;
 864                                 }
 865                         }
 866                         // skip 'objects' at end
 867                         if (okay) {
 868                                 target = xmalloc(serverlen + posn - i - 6);
 869                                 strncpy(target, base, serverlen);
 870                                 strncpy(target + serverlen, data + i,
 871                                         posn - i - 7);
 872                                 target[serverlen + posn - i - 7] = '\0';
 873                                 if (get_verbosely)
 874                                         fprintf(stderr,
 875                                                 "Also look at %s\n", target);
 876                                 newalt = xmalloc(sizeof(*newalt));
 877                                 newalt->next = NULL;
 878                                 newalt->base = target;
 879                                 newalt->got_indices = 0;
 880                                 newalt->packs = NULL;
 881                                 while (tail->next != NULL)
 882                                         tail = tail->next;
 883                                 tail->next = newalt;
 884                                 ret++;
 885                         }
 886                 }
 887                 i = posn + 1;
 888         }
 889
 890         got_alternates = 1;
 891         free(buffer.buffer);
 892         return ret;
 893 }
 894
 895 static int fetch_indices(struct alt_base *repo)
 896 {
 897         unsigned char sha1[20];
 898         char *url;
 899         struct buffer buffer;
 900         char *data;
 901         int i = 0;
 902
 903         struct active_request_slot *slot;
 904
 905         if (repo->got_indices)
 906                 return 0;
 907
 908         data = xmalloc(4096);
 909         buffer.size = 4096;
 910         buffer.posn = 0;
 911         buffer.buffer = data;
 912
 913         if (get_verbosely)
 914                 fprintf(stderr, "Getting pack list\n");
 915
 916         url = xmalloc(strlen(repo->base) + 21);
 917         sprintf(url, "%s/objects/info/packs", repo->base);
 918
 919         slot = get_active_slot();
 920         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 921         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 922                          fwrite_buffer_dynamic);
 923         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 924         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 925         if (start_active_slot(slot)) {
 926                 run_active_slot(slot);
 927                 if (slot->curl_result != CURLE_OK) {
 928                         free(buffer.buffer);
 929                         return error("%s", curl_errorstr);
 930                 }
 931         } else {
 932                 free(buffer.buffer);
 933                 return error("Unable to start request");
 934         }
 935
 936         data = buffer.buffer;
 937         while (i < buffer.posn) {
 938                 switch (data[i]) {
 939                 case 'P':
 940                         i++;
 941                         if (i + 52 < buffer.posn &&
 942                             !strncmp(data + i, " pack-", 6) &&
 943                             !strncmp(data + i + 46, ".pack\n", 6)) {
 944                                 get_sha1_hex(data + i + 6, sha1);
 945                                 setup_index(repo, sha1);
 946                                 i += 51;
 947                                 break;
 948                         }
 949                 default:
 950                         while (data[i] != '\n')
 951                                 i++;
 952                 }
 953                 i++;
 954         }
 955
 956         free(buffer.buffer);
 957         repo->got_indices = 1;
 958         return 0;
 959 }
 960
 961 static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 962 {
 963         char *url;
 964         struct packed_git *target;
 965         struct packed_git **lst;
 966         FILE *packfile;
 967         char *filename;
 968         char tmpfile[PATH_MAX];
 969         int ret;
 970         long prev_posn = 0;
 971         char range[RANGE_HEADER_SIZE];
 972         struct curl_slist *range_header = NULL;
 973
 974         struct active_request_slot *slot;
 975
 976         if (fetch_indices(repo))
 977                 return -1;
 978         target = find_sha1_pack(sha1, repo->packs);
 979         if (!target)
 980                 return -1;
 981
 982         if (get_verbosely) {
 983                 fprintf(stderr, "Getting pack %s\n",
 984                         sha1_to_hex(target->sha1));
 985                 fprintf(stderr, " which contains %s\n",
 986                         sha1_to_hex(sha1));
 987         }
 988
 989         url = xmalloc(strlen(repo->base) + 65);
 990         sprintf(url, "%s/objects/pack/pack-%s.pack",
 991                 repo->base, sha1_to_hex(target->sha1));
 992
 993         filename = sha1_pack_name(target->sha1);
 994         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 995         packfile = fopen(tmpfile, "a");
 996         if (!packfile)
 997                 return error("Unable to open local file %s for pack",
 998                              filename);
 999
1000         slot = get_active_slot();
1001         curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
1002         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
1003         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
1004         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
1005         slot->local = packfile;
1006
1007         /* If there is data present from a previous transfer attempt,
1008            resume where it left off */
1009         prev_posn = ftell(packfile);
1010         if (prev_posn>0) {
1011                 if (get_verbosely)
1012                         fprintf(stderr,
1013                                 "Resuming fetch of pack %s at byte %ld\n",
1014                                 sha1_to_hex(target->sha1), prev_posn);
1015                 sprintf(range, "Range: bytes=%ld-", prev_posn);
1016                 range_header = curl_slist_append(range_header, range);
1017                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
1018         }
1019
1020         if (start_active_slot(slot)) {
1021                 run_active_slot(slot);
1022                 if (slot->curl_result != CURLE_OK) {
1023                         fclose(packfile);
1024                         return error("Unable to get pack file %s\n%s", url,
1025                                      curl_errorstr);
1026                 }
1027         } else {
1028                 return error("Unable to start request");
1029         }
1030
1031         fclose(packfile);
1032
1033         ret = move_temp_to_file(tmpfile, filename);
1034         if (ret)
1035                 return ret;
1036
1037         lst = &repo->packs;
1038         while (*lst != target)
1039                 lst = &((*lst)->next);
1040         *lst = (*lst)->next;
1041
1042         if (verify_pack(target, 0))
1043                 return -1;
1044         install_packed_git(target);
1045
1046         return 0;
1047 }
1048
1049 static int fetch_object(struct alt_base *repo, unsigned char *sha1)
1050 {
1051         char *hex = sha1_to_hex(sha1);
1052         int ret;
1053         struct transfer_request *request = request_queue_head;
1054
1055         while (request != NULL && memcmp(request->sha1, sha1, 20))
1056                 request = request->next;
1057         if (request == NULL)
1058                 return error("Couldn't find request for %s in the queue", hex);
1059
1060         if (has_sha1_file(request->sha1)) {
1061                 release_request(request);
1062                 return 0;
1063         }
1064
1065 #ifdef USE_CURL_MULTI
1066         while (request->state == WAITING) {
1067                 int num_transfers;
1068                 curl_multi_perform(curlm, &num_transfers);
1069                 if (num_transfers < active_requests) {
1070                         process_curl_messages();
1071                         process_request_queue();
1072                 }
1073         }
1074 #else
1075         start_request(request);
1076 #endif
1077
1078         while (request->state == ACTIVE) {
1079                 run_active_slot(request->slot);
1080 #ifndef USE_CURL_MULTI
1081                 request->curl_result = request->slot->curl_result;
1082                 request->http_code = request->slot->http_code;
1083                 request->slot = NULL;
1084
1085                 /* Use alternates if necessary */
1086                 if (request->http_code == 404) {
1087                         fetch_alternates(alt->base);
1088                         if (request->repo->next != NULL) {
1089                                 request->repo = request->repo->next;
1090                                 start_request(request);
1091                         }
1092                 } else {
1093                         finish_request(request);
1094                         request->state = COMPLETE;
1095                 }
1096 #endif
1097         }
1098
1099         if (request->state == ABORTED) {
1100                 release_request(request);
1101                 return error("Request for %s aborted", hex);
1102         }
1103
1104         if (request->curl_result != CURLE_OK && request->http_code != 416) {
1105                 if (request->http_code == 404)
1106                         ret = -1; /* Be silent, it is probably in a pack. */
1107                 else
1108                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
1109                                     request->errorstr, request->curl_result,
1110                                     request->http_code, hex);
1111                 release_request(request);
1112                 return ret;
1113         }
1114
1115         if (request->zret != Z_STREAM_END) {
1116                 ret = error("File %s (%s) corrupt\n", hex, request->url);
1117                 release_request(request);
1118                 return ret;
1119         }
1120
1121         if (memcmp(request->sha1, request->real_sha1, 20)) {
1122                 release_request(request);
1123                 return error("File %s has bad hash\n", hex);
1124         }
1125
1126         if (request->rename < 0) {
1127                 ret = error("unable to write sha1 filename %s: %s",
1128                             request->filename,
1129                             strerror(request->rename));
1130                 release_request(request);
1131                 return ret;
1132         }
1133
1134         release_request(request);
1135         return 0;
1136 }
1137
1138 int fetch(unsigned char *sha1)
1139 {
1140         struct alt_base *altbase = alt;
1141
1142         if (!fetch_object(altbase, sha1))
1143                 return 0;
1144         while (altbase) {
1145                 if (!fetch_pack(altbase, sha1))
1146                         return 0;
1147                 fetch_alternates(alt->base);
1148                 altbase = altbase->next;
1149         }
1150         return error("Unable to find %s under %s\n", sha1_to_hex(sha1),
1151                      alt->base);
1152 }
1153
1154 static inline int needs_quote(int ch)
1155 {
1156         switch (ch) {
1157         case '/': case '-': case '.':
1158         case 'A'...'Z': case 'a'...'z': case '0'...'9':
1159                 return 0;
1160         default:
1161                 return 1;
1162         }
1163 }
1164
1165 static inline int hex(int v)
1166 {
1167         if (v < 10) return '0' + v;
1168         else return 'A' + v - 10;
1169 }
1170
1171 static char *quote_ref_url(const char *base, const char *ref)
1172 {
1173         const char *cp;
1174         char *dp, *qref;
1175         int len, baselen, ch;
1176
1177         baselen = strlen(base);
1178         len = baselen + 6; /* "refs/" + NUL */
1179         for (cp = ref; (ch = *cp) != 0; cp++, len++)
1180                 if (needs_quote(ch))
1181                         len += 2; /* extra two hex plus replacement % */
1182         qref = xmalloc(len);
1183         memcpy(qref, base, baselen);
1184         memcpy(qref + baselen, "refs/", 5);
1185         for (cp = ref, dp = qref + baselen + 5; (ch = *cp) != 0; cp++) {
1186                 if (needs_quote(ch)) {
1187                         *dp++ = '%';
1188                         *dp++ = hex((ch >> 4) & 0xF);
1189                         *dp++ = hex(ch & 0xF);
1190                 }
1191                 else
1192                         *dp++ = ch;
1193         }
1194         *dp = 0;
1195
1196         return qref;
1197 }
1198
1199 int fetch_ref(char *ref, unsigned char *sha1)
1200 {
1201         char *url;
1202         char hex[42];
1203         struct buffer buffer;
1204         char *base = alt->base;
1205         struct active_request_slot *slot;
1206         buffer.size = 41;
1207         buffer.posn = 0;
1208         buffer.buffer = hex;
1209         hex[41] = '\0';
1210
1211         url = quote_ref_url(base, ref);
1212         slot = get_active_slot();
1213         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
1214         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
1215         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
1216         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
1217         if (start_active_slot(slot)) {
1218                 run_active_slot(slot);
1219                 if (slot->curl_result != CURLE_OK)
1220                         return error("Couldn't get %s for %s\n%s",
1221                                      url, ref, curl_errorstr);
1222         } else {
1223                 return error("Unable to start request");
1224         }
1225
1226         hex[40] = '\0';
1227         get_sha1_hex(hex, sha1);
1228         return 0;
1229 }
1230
1231 int main(int argc, char **argv)
1232 {
1233         char *commit_id;
1234         char *url;
1235         int arg = 1;
1236         struct active_request_slot *slot;
1237         char *low_speed_limit;
1238         char *low_speed_time;
1239         char *wait_url;
1240         int rc = 0;
1241
1242         while (arg < argc && argv[arg][0] == '-') {
1243                 if (argv[arg][1] == 't') {
1244                         get_tree = 1;
1245                 } else if (argv[arg][1] == 'c') {
1246                         get_history = 1;
1247                 } else if (argv[arg][1] == 'a') {
1248                         get_all = 1;
1249                         get_tree = 1;
1250                         get_history = 1;
1251                 } else if (argv[arg][1] == 'v') {
1252                         get_verbosely = 1;
1253                 } else if (argv[arg][1] == 'w') {
1254                         write_ref = argv[arg + 1];
1255                         arg++;
1256                 } else if (!strcmp(argv[arg], "--recover")) {
1257                         get_recover = 1;
1258                 }
1259                 arg++;
1260         }
1261         if (argc < arg + 2) {
1262                 usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
1263                 return 1;
1264         }
1265         commit_id = argv[arg];
1266         url = argv[arg + 1];
1267
1268         curl_global_init(CURL_GLOBAL_ALL);
1269
1270 #ifdef USE_CURL_MULTI
1271         {
1272                 char *http_max_requests = getenv("GIT_HTTP_MAX_REQUESTS");
1273                 if (http_max_requests != NULL)
1274                         max_requests = atoi(http_max_requests);
1275         }
1276
1277         curlm = curl_multi_init();
1278         if (curlm == NULL) {
1279                 fprintf(stderr, "Error creating curl multi handle.\n");
1280                 return 1;
1281         }
1282 #endif
1283
1284         if (getenv("GIT_SSL_NO_VERIFY"))
1285                 curl_ssl_verify = 0;
1286
1287         ssl_cert = getenv("GIT_SSL_CERT");
1288 #if LIBCURL_VERSION_NUM >= 0x070902
1289         ssl_key = getenv("GIT_SSL_KEY");
1290 #endif
1291 #if LIBCURL_VERSION_NUM >= 0x070908
1292         ssl_capath = getenv("GIT_SSL_CAPATH");
1293 #endif
1294         ssl_cainfo = getenv("GIT_SSL_CAINFO");
1295
1296         low_speed_limit = getenv("GIT_HTTP_LOW_SPEED_LIMIT");
1297         if (low_speed_limit != NULL)
1298                 curl_low_speed_limit = strtol(low_speed_limit, NULL, 10);
1299         low_speed_time = getenv("GIT_HTTP_LOW_SPEED_TIME");
1300         if (low_speed_time != NULL)
1301                 curl_low_speed_time = strtol(low_speed_time, NULL, 10);
1302
1303         git_config(http_options);
1304
1305         if (curl_ssl_verify == -1)
1306                 curl_ssl_verify = 1;
1307
1308 #ifdef USE_CURL_MULTI
1309         if (max_requests < 1)
1310                 max_requests = DEFAULT_MAX_REQUESTS;
1311 #endif
1312
1313         pragma_header = curl_slist_append(pragma_header, "Pragma: no-cache");
1314         no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
1315         no_range_header = curl_slist_append(no_range_header, "Range:");
1316
1317 #ifndef NO_CURL_EASY_DUPHANDLE
1318         curl_default = get_curl_handle();
1319 #endif
1320
1321         alt = xmalloc(sizeof(*alt));
1322         alt->base = url;
1323         alt->got_indices = 0;
1324         alt->packs = NULL;
1325         alt->next = NULL;
1326
1327         if (pull(commit_id))
1328                 rc = 1;
1329
1330         curl_slist_free_all(pragma_header);
1331         curl_slist_free_all(no_pragma_header);
1332         curl_slist_free_all(no_range_header);
1333 #ifndef NO_CURL_EASY_DUPHANDLE
1334         curl_easy_cleanup(curl_default);
1335 #endif
1336         slot = active_queue_head;
1337         while (slot != NULL) {
1338                 if (slot->in_use) {
1339                         if (get_verbosely) {
1340                                 curl_easy_getinfo(slot->curl,
1341                                                   CURLINFO_EFFECTIVE_URL,
1342                                                   &wait_url);
1343                                 fprintf(stderr, "Waiting for %s\n", wait_url);
1344                         }
1345                         run_active_slot(slot);
1346                 }
1347                 if (slot->curl != NULL)
1348                         curl_easy_cleanup(slot->curl);
1349                 slot = slot->next;
1350         }
1351 #ifdef USE_CURL_MULTI
1352         curl_multi_cleanup(curlm);
1353 #endif
1354         curl_global_cleanup();
1355         return rc;
1356 }