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