http-walker.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "pack.h"
   4 #include "walker.h"
   5 #include "http.h"
   6
   7 #define PREV_BUF_SIZE 4096
   8 #define RANGE_HEADER_SIZE 30
   9
  10 struct alt_base
  11 {
  12         char *base;
  13         int got_indices;
  14         struct packed_git *packs;
  15         struct alt_base *next;
  16 };
  17
  18 enum object_request_state {
  19         WAITING,
  20         ABORTED,
  21         ACTIVE,
  22         COMPLETE,
  23 };
  24
  25 struct object_request
  26 {
  27         struct walker *walker;
  28         unsigned char sha1[20];
  29         struct alt_base *repo;
  30         char *url;
  31         char filename[PATH_MAX];
  32         char tmpfile[PATH_MAX];
  33         int local;
  34         enum object_request_state state;
  35         CURLcode curl_result;
  36         char errorstr[CURL_ERROR_SIZE];
  37         long http_code;
  38         unsigned char real_sha1[20];
  39         SHA_CTX c;
  40         z_stream stream;
  41         int zret;
  42         int rename;
  43         struct active_request_slot *slot;
  44         struct object_request *next;
  45 };
  46
  47 struct alternates_request {
  48         struct walker *walker;
  49         const char *base;
  50         char *url;
  51         struct strbuf *buffer;
  52         struct active_request_slot *slot;
  53         int http_specific;
  54 };
  55
  56 struct walker_data {
  57         const char *url;
  58         int got_alternates;
  59         struct alt_base *alt;
  60         struct curl_slist *no_pragma_header;
  61 };
  62
  63 static struct object_request *object_queue_head;
  64
  65 static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
  66                                void *data)
  67 {
  68         unsigned char expn[4096];
  69         size_t size = eltsize * nmemb;
  70         int posn = 0;
  71         struct object_request *obj_req = (struct object_request *)data;
  72         do {
  73                 ssize_t retval = xwrite(obj_req->local,
  74                                        (char *) ptr + posn, size - posn);
  75                 if (retval < 0)
  76                         return posn;
  77                 posn += retval;
  78         } while (posn < size);
  79
  80         obj_req->stream.avail_in = size;
  81         obj_req->stream.next_in = ptr;
  82         do {
  83                 obj_req->stream.next_out = expn;
  84                 obj_req->stream.avail_out = sizeof(expn);
  85                 obj_req->zret = inflate(&obj_req->stream, Z_SYNC_FLUSH);
  86                 SHA1_Update(&obj_req->c, expn,
  87                             sizeof(expn) - obj_req->stream.avail_out);
  88         } while (obj_req->stream.avail_in && obj_req->zret == Z_OK);
  89         data_received++;
  90         return size;
  91 }
  92
  93 static void fetch_alternates(struct walker *walker, const char *base);
  94
  95 static void process_object_response(void *callback_data);
  96
  97 static void start_object_request(struct walker *walker,
  98                                  struct object_request *obj_req)
  99 {
 100         char *hex = sha1_to_hex(obj_req->sha1);
 101         char prevfile[PATH_MAX];
 102         char *url;
 103         char *posn;
 104         int prevlocal;
 105         unsigned char prev_buf[PREV_BUF_SIZE];
 106         ssize_t prev_read = 0;
 107         long prev_posn = 0;
 108         char range[RANGE_HEADER_SIZE];
 109         struct curl_slist *range_header = NULL;
 110         struct active_request_slot *slot;
 111         struct walker_data *data = walker->data;
 112
 113         snprintf(prevfile, sizeof(prevfile), "%s.prev", obj_req->filename);
 114         unlink(prevfile);
 115         rename(obj_req->tmpfile, prevfile);
 116         unlink(obj_req->tmpfile);
 117
 118         if (obj_req->local != -1)
 119                 error("fd leakage in start: %d", obj_req->local);
 120         obj_req->local = open(obj_req->tmpfile,
 121                               O_WRONLY | O_CREAT | O_EXCL, 0666);
 122         /* This could have failed due to the "lazy directory creation";
 123          * try to mkdir the last path component.
 124          */
 125         if (obj_req->local < 0 && errno == ENOENT) {
 126                 char *dir = strrchr(obj_req->tmpfile, '/');
 127                 if (dir) {
 128                         *dir = 0;
 129                         mkdir(obj_req->tmpfile, 0777);
 130                         *dir = '/';
 131                 }
 132                 obj_req->local = open(obj_req->tmpfile,
 133                                       O_WRONLY | O_CREAT | O_EXCL, 0666);
 134         }
 135
 136         if (obj_req->local < 0) {
 137                 obj_req->state = ABORTED;
 138                 error("Couldn't create temporary file %s for %s: %s",
 139                       obj_req->tmpfile, obj_req->filename, strerror(errno));
 140                 return;
 141         }
 142
 143         memset(&obj_req->stream, 0, sizeof(obj_req->stream));
 144
 145         inflateInit(&obj_req->stream);
 146
 147         SHA1_Init(&obj_req->c);
 148
 149         url = xmalloc(strlen(obj_req->repo->base) + 51);
 150         obj_req->url = xmalloc(strlen(obj_req->repo->base) + 51);
 151         strcpy(url, obj_req->repo->base);
 152         posn = url + strlen(obj_req->repo->base);
 153         strcpy(posn, "/objects/");
 154         posn += 9;
 155         memcpy(posn, hex, 2);
 156         posn += 2;
 157         *(posn++) = '/';
 158         strcpy(posn, hex + 2);
 159         strcpy(obj_req->url, url);
 160
 161         /* If a previous temp file is present, process what was already
 162            fetched. */
 163         prevlocal = open(prevfile, O_RDONLY);
 164         if (prevlocal != -1) {
 165                 do {
 166                         prev_read = xread(prevlocal, prev_buf, PREV_BUF_SIZE);
 167                         if (prev_read>0) {
 168                                 if (fwrite_sha1_file(prev_buf,
 169                                                      1,
 170                                                      prev_read,
 171                                                      obj_req) == prev_read) {
 172                                         prev_posn += prev_read;
 173                                 } else {
 174                                         prev_read = -1;
 175                                 }
 176                         }
 177                 } while (prev_read > 0);
 178                 close(prevlocal);
 179         }
 180         unlink(prevfile);
 181
 182         /* Reset inflate/SHA1 if there was an error reading the previous temp
 183            file; also rewind to the beginning of the local file. */
 184         if (prev_read == -1) {
 185                 memset(&obj_req->stream, 0, sizeof(obj_req->stream));
 186                 inflateInit(&obj_req->stream);
 187                 SHA1_Init(&obj_req->c);
 188                 if (prev_posn>0) {
 189                         prev_posn = 0;
 190                         lseek(obj_req->local, 0, SEEK_SET);
 191                         ftruncate(obj_req->local, 0);
 192                 }
 193         }
 194
 195         slot = get_active_slot();
 196         slot->callback_func = process_object_response;
 197         slot->callback_data = obj_req;
 198         obj_req->slot = slot;
 199
 200         curl_easy_setopt(slot->curl, CURLOPT_FILE, obj_req);
 201         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 202         curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, obj_req->errorstr);
 203         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 204         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, data->no_pragma_header);
 205
 206         /* If we have successfully processed data from a previous fetch
 207            attempt, only fetch the data we don't already have. */
 208         if (prev_posn>0) {
 209                 if (walker->get_verbosely)
 210                         fprintf(stderr,
 211                                 "Resuming fetch of object %s at byte %ld\n",
 212                                 hex, prev_posn);
 213                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 214                 range_header = curl_slist_append(range_header, range);
 215                 curl_easy_setopt(slot->curl,
 216                                  CURLOPT_HTTPHEADER, range_header);
 217         }
 218
 219         /* Try to get the request started, abort the request on error */
 220         obj_req->state = ACTIVE;
 221         if (!start_active_slot(slot)) {
 222                 obj_req->state = ABORTED;
 223                 obj_req->slot = NULL;
 224                 close(obj_req->local); obj_req->local = -1;
 225                 free(obj_req->url);
 226                 return;
 227         }
 228 }
 229
 230 static void finish_object_request(struct object_request *obj_req)
 231 {
 232         struct stat st;
 233
 234         fchmod(obj_req->local, 0444);
 235         close(obj_req->local); obj_req->local = -1;
 236
 237         if (obj_req->http_code == 416) {
 238                 fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 239         } else if (obj_req->curl_result != CURLE_OK) {
 240                 if (stat(obj_req->tmpfile, &st) == 0)
 241                         if (st.st_size == 0)
 242                                 unlink(obj_req->tmpfile);
 243                 return;
 244         }
 245
 246         inflateEnd(&obj_req->stream);
 247         SHA1_Final(obj_req->real_sha1, &obj_req->c);
 248         if (obj_req->zret != Z_STREAM_END) {
 249                 unlink(obj_req->tmpfile);
 250                 return;
 251         }
 252         if (hashcmp(obj_req->sha1, obj_req->real_sha1)) {
 253                 unlink(obj_req->tmpfile);
 254                 return;
 255         }
 256         obj_req->rename =
 257                 move_temp_to_file(obj_req->tmpfile, obj_req->filename);
 258
 259         if (obj_req->rename == 0)
 260                 walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
 261 }
 262
 263 static void process_object_response(void *callback_data)
 264 {
 265         struct object_request *obj_req =
 266                 (struct object_request *)callback_data;
 267         struct walker *walker = obj_req->walker;
 268         struct walker_data *data = walker->data;
 269         struct alt_base *alt = data->alt;
 270
 271         obj_req->curl_result = obj_req->slot->curl_result;
 272         obj_req->http_code = obj_req->slot->http_code;
 273         obj_req->slot = NULL;
 274         obj_req->state = COMPLETE;
 275
 276         /* Use alternates if necessary */
 277         if (missing_target(obj_req)) {
 278                 fetch_alternates(walker, alt->base);
 279                 if (obj_req->repo->next != NULL) {
 280                         obj_req->repo =
 281                                 obj_req->repo->next;
 282                         close(obj_req->local);
 283                         obj_req->local = -1;
 284                         start_object_request(walker, obj_req);
 285                         return;
 286                 }
 287         }
 288
 289         finish_object_request(obj_req);
 290 }
 291
 292 static void release_object_request(struct object_request *obj_req)
 293 {
 294         struct object_request *entry = object_queue_head;
 295
 296         if (obj_req->local != -1)
 297                 error("fd leakage in release: %d", obj_req->local);
 298         if (obj_req == object_queue_head) {
 299                 object_queue_head = obj_req->next;
 300         } else {
 301                 while (entry->next != NULL && entry->next != obj_req)
 302                         entry = entry->next;
 303                 if (entry->next == obj_req)
 304                         entry->next = entry->next->next;
 305         }
 306
 307         free(obj_req->url);
 308         free(obj_req);
 309 }
 310
 311 #ifdef USE_CURL_MULTI
 312 static int fill_active_slot(struct walker *walker)
 313 {
 314         struct object_request *obj_req;
 315
 316         for (obj_req = object_queue_head; obj_req; obj_req = obj_req->next) {
 317                 if (obj_req->state == WAITING) {
 318                         if (has_sha1_file(obj_req->sha1))
 319                                 obj_req->state = COMPLETE;
 320                         else {
 321                                 start_object_request(walker, obj_req);
 322                                 return 1;
 323                         }
 324                 }
 325         }
 326         return 0;
 327 }
 328 #endif
 329
 330 static void prefetch(struct walker *walker, unsigned char *sha1)
 331 {
 332         struct object_request *newreq;
 333         struct object_request *tail;
 334         struct walker_data *data = walker->data;
 335         char *filename = sha1_file_name(sha1);
 336
 337         newreq = xmalloc(sizeof(*newreq));
 338         newreq->walker = walker;
 339         hashcpy(newreq->sha1, sha1);
 340         newreq->repo = data->alt;
 341         newreq->url = NULL;
 342         newreq->local = -1;
 343         newreq->state = WAITING;
 344         snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 345         snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 346                  "%s.temp", filename);
 347         newreq->slot = NULL;
 348         newreq->next = NULL;
 349
 350         if (object_queue_head == NULL) {
 351                 object_queue_head = newreq;
 352         } else {
 353                 tail = object_queue_head;
 354                 while (tail->next != NULL) {
 355                         tail = tail->next;
 356                 }
 357                 tail->next = newreq;
 358         }
 359
 360 #ifdef USE_CURL_MULTI
 361         fill_active_slots();
 362         step_active_slots();
 363 #endif
 364 }
 365
 366 static int fetch_index(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 367 {
 368         char *hex = sha1_to_hex(sha1);
 369         char *filename;
 370         char *url;
 371         char tmpfile[PATH_MAX];
 372         long prev_posn = 0;
 373         char range[RANGE_HEADER_SIZE];
 374         struct curl_slist *range_header = NULL;
 375         struct walker_data *data = walker->data;
 376
 377         FILE *indexfile;
 378         struct active_request_slot *slot;
 379         struct slot_results results;
 380
 381         if (has_pack_index(sha1))
 382                 return 0;
 383
 384         if (walker->get_verbosely)
 385                 fprintf(stderr, "Getting index for pack %s\n", hex);
 386
 387         url = xmalloc(strlen(repo->base) + 64);
 388         sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 389
 390         filename = sha1_pack_index_name(sha1);
 391         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 392         indexfile = fopen(tmpfile, "a");
 393         if (!indexfile)
 394                 return error("Unable to open local file %s for pack index",
 395                              tmpfile);
 396
 397         slot = get_active_slot();
 398         slot->results = &results;
 399         curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 400         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 401         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 402         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, data->no_pragma_header);
 403         slot->local = indexfile;
 404
 405         /* If there is data present from a previous transfer attempt,
 406            resume where it left off */
 407         prev_posn = ftell(indexfile);
 408         if (prev_posn>0) {
 409                 if (walker->get_verbosely)
 410                         fprintf(stderr,
 411                                 "Resuming fetch of index for pack %s at byte %ld\n",
 412                                 hex, prev_posn);
 413                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 414                 range_header = curl_slist_append(range_header, range);
 415                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 416         }
 417
 418         if (start_active_slot(slot)) {
 419                 run_active_slot(slot);
 420                 if (results.curl_result != CURLE_OK) {
 421                         fclose(indexfile);
 422                         return error("Unable to get pack index %s\n%s", url,
 423                                      curl_errorstr);
 424                 }
 425         } else {
 426                 fclose(indexfile);
 427                 return error("Unable to start request");
 428         }
 429
 430         fclose(indexfile);
 431
 432         return move_temp_to_file(tmpfile, filename);
 433 }
 434
 435 static int setup_index(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 436 {
 437         struct packed_git *new_pack;
 438         if (has_pack_file(sha1))
 439                 return 0; /* don't list this as something we can get */
 440
 441         if (fetch_index(walker, repo, sha1))
 442                 return -1;
 443
 444         new_pack = parse_pack_index(sha1);
 445         new_pack->next = repo->packs;
 446         repo->packs = new_pack;
 447         return 0;
 448 }
 449
 450 static void process_alternates_response(void *callback_data)
 451 {
 452         struct alternates_request *alt_req =
 453                 (struct alternates_request *)callback_data;
 454         struct walker *walker = alt_req->walker;
 455         struct walker_data *cdata = walker->data;
 456         struct active_request_slot *slot = alt_req->slot;
 457         struct alt_base *tail = cdata->alt;
 458         const char *base = alt_req->base;
 459         static const char null_byte = '\0';
 460         char *data;
 461         int i = 0;
 462
 463         if (alt_req->http_specific) {
 464                 if (slot->curl_result != CURLE_OK ||
 465                     !alt_req->buffer->len) {
 466
 467                         /* Try reusing the slot to get non-http alternates */
 468                         alt_req->http_specific = 0;
 469                         sprintf(alt_req->url, "%s/objects/info/alternates",
 470                                 base);
 471                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 472                                          alt_req->url);
 473                         active_requests++;
 474                         slot->in_use = 1;
 475                         if (slot->finished != NULL)
 476                                 (*slot->finished) = 0;
 477                         if (!start_active_slot(slot)) {
 478                                 cdata->got_alternates = -1;
 479                                 slot->in_use = 0;
 480                                 if (slot->finished != NULL)
 481                                         (*slot->finished) = 1;
 482                         }
 483                         return;
 484                 }
 485         } else if (slot->curl_result != CURLE_OK) {
 486                 if (!missing_target(slot)) {
 487                         cdata->got_alternates = -1;
 488                         return;
 489                 }
 490         }
 491
 492         fwrite_buffer(&null_byte, 1, 1, alt_req->buffer);
 493         alt_req->buffer->len--;
 494         data = alt_req->buffer->buf;
 495
 496         while (i < alt_req->buffer->len) {
 497                 int posn = i;
 498                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 499                         posn++;
 500                 if (data[posn] == '\n') {
 501                         int okay = 0;
 502                         int serverlen = 0;
 503                         struct alt_base *newalt;
 504                         char *target = NULL;
 505                         if (data[i] == '/') {
 506                                 /* This counts
 507                                  * http://git.host/pub/scm/linux.git/
 508                                  * -----------here^
 509                                  * so memcpy(dst, base, serverlen) will
 510                                  * copy up to "...git.host".
 511                                  */
 512                                 const char *colon_ss = strstr(base,"://");
 513                                 if (colon_ss) {
 514                                         serverlen = (strchr(colon_ss + 3, '/')
 515                                                      - base);
 516                                         okay = 1;
 517                                 }
 518                         } else if (!memcmp(data + i, "../", 3)) {
 519                                 /* Relative URL; chop the corresponding
 520                                  * number of subpath from base (and ../
 521                                  * from data), and concatenate the result.
 522                                  *
 523                                  * The code first drops ../ from data, and
 524                                  * then drops one ../ from data and one path
 525                                  * from base.  IOW, one extra ../ is dropped
 526                                  * from data than path is dropped from base.
 527                                  *
 528                                  * This is not wrong.  The alternate in
 529                                  *     http://git.host/pub/scm/linux.git/
 530                                  * to borrow from
 531                                  *     http://git.host/pub/scm/linus.git/
 532                                  * is ../../linus.git/objects/.  You need
 533                                  * two ../../ to borrow from your direct
 534                                  * neighbour.
 535                                  */
 536                                 i += 3;
 537                                 serverlen = strlen(base);
 538                                 while (i + 2 < posn &&
 539                                        !memcmp(data + i, "../", 3)) {
 540                                         do {
 541                                                 serverlen--;
 542                                         } while (serverlen &&
 543                                                  base[serverlen - 1] != '/');
 544                                         i += 3;
 545                                 }
 546                                 /* If the server got removed, give up. */
 547                                 okay = strchr(base, ':') - base + 3 <
 548                                         serverlen;
 549                         } else if (alt_req->http_specific) {
 550                                 char *colon = strchr(data + i, ':');
 551                                 char *slash = strchr(data + i, '/');
 552                                 if (colon && slash && colon < data + posn &&
 553                                     slash < data + posn && colon < slash) {
 554                                         okay = 1;
 555                                 }
 556                         }
 557                         /* skip "objects\n" at end */
 558                         if (okay) {
 559                                 target = xmalloc(serverlen + posn - i - 6);
 560                                 memcpy(target, base, serverlen);
 561                                 memcpy(target + serverlen, data + i,
 562                                        posn - i - 7);
 563                                 target[serverlen + posn - i - 7] = 0;
 564                                 if (walker->get_verbosely)
 565                                         fprintf(stderr,
 566                                                 "Also look at %s\n", target);
 567                                 newalt = xmalloc(sizeof(*newalt));
 568                                 newalt->next = NULL;
 569                                 newalt->base = target;
 570                                 newalt->got_indices = 0;
 571                                 newalt->packs = NULL;
 572
 573                                 while (tail->next != NULL)
 574                                         tail = tail->next;
 575                                 tail->next = newalt;
 576                         }
 577                 }
 578                 i = posn + 1;
 579         }
 580
 581         cdata->got_alternates = 1;
 582 }
 583
 584 static void fetch_alternates(struct walker *walker, const char *base)
 585 {
 586         struct strbuf buffer = STRBUF_INIT;
 587         char *url;
 588         struct active_request_slot *slot;
 589         struct alternates_request alt_req;
 590         struct walker_data *cdata = walker->data;
 591
 592         /* If another request has already started fetching alternates,
 593            wait for them to arrive and return to processing this request's
 594            curl message */
 595 #ifdef USE_CURL_MULTI
 596         while (cdata->got_alternates == 0) {
 597                 step_active_slots();
 598         }
 599 #endif
 600
 601         /* Nothing to do if they've already been fetched */
 602         if (cdata->got_alternates == 1)
 603                 return;
 604
 605         /* Start the fetch */
 606         cdata->got_alternates = 0;
 607
 608         if (walker->get_verbosely)
 609                 fprintf(stderr, "Getting alternates list for %s\n", base);
 610
 611         url = xmalloc(strlen(base) + 31);
 612         sprintf(url, "%s/objects/info/http-alternates", base);
 613
 614         /* Use a callback to process the result, since another request
 615            may fail and need to have alternates loaded before continuing */
 616         slot = get_active_slot();
 617         slot->callback_func = process_alternates_response;
 618         alt_req.walker = walker;
 619         slot->callback_data = &alt_req;
 620
 621         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 622         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 623         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 624
 625         alt_req.base = base;
 626         alt_req.url = url;
 627         alt_req.buffer = &buffer;
 628         alt_req.http_specific = 1;
 629         alt_req.slot = slot;
 630
 631         if (start_active_slot(slot))
 632                 run_active_slot(slot);
 633         else
 634                 cdata->got_alternates = -1;
 635
 636         strbuf_release(&buffer);
 637         free(url);
 638 }
 639
 640 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 641 {
 642         unsigned char sha1[20];
 643         char *url;
 644         struct strbuf buffer = STRBUF_INIT;
 645         char *data;
 646         int i = 0;
 647
 648         struct active_request_slot *slot;
 649         struct slot_results results;
 650
 651         if (repo->got_indices)
 652                 return 0;
 653
 654         if (walker->get_verbosely)
 655                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 656
 657         url = xmalloc(strlen(repo->base) + 21);
 658         sprintf(url, "%s/objects/info/packs", repo->base);
 659
 660         slot = get_active_slot();
 661         slot->results = &results;
 662         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 663         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 664         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 665         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 666         if (start_active_slot(slot)) {
 667                 run_active_slot(slot);
 668                 if (results.curl_result != CURLE_OK) {
 669                         strbuf_release(&buffer);
 670                         if (missing_target(&results)) {
 671                                 repo->got_indices = 1;
 672                                 return 0;
 673                         } else {
 674                                 repo->got_indices = 0;
 675                                 return error("%s", curl_errorstr);
 676                         }
 677                 }
 678         } else {
 679                 repo->got_indices = 0;
 680                 strbuf_release(&buffer);
 681                 return error("Unable to start request");
 682         }
 683
 684         data = buffer.buf;
 685         while (i < buffer.len) {
 686                 switch (data[i]) {
 687                 case 'P':
 688                         i++;
 689                         if (i + 52 <= buffer.len &&
 690                             !prefixcmp(data + i, " pack-") &&
 691                             !prefixcmp(data + i + 46, ".pack\n")) {
 692                                 get_sha1_hex(data + i + 6, sha1);
 693                                 setup_index(walker, repo, sha1);
 694                                 i += 51;
 695                                 break;
 696                         }
 697                 default:
 698                         while (i < buffer.len && data[i] != '\n')
 699                                 i++;
 700                 }
 701                 i++;
 702         }
 703
 704         strbuf_release(&buffer);
 705         repo->got_indices = 1;
 706         return 0;
 707 }
 708
 709 static int fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 710 {
 711         char *url;
 712         struct packed_git *target;
 713         struct packed_git **lst;
 714         FILE *packfile;
 715         char *filename;
 716         char tmpfile[PATH_MAX];
 717         int ret;
 718         long prev_posn = 0;
 719         char range[RANGE_HEADER_SIZE];
 720         struct curl_slist *range_header = NULL;
 721         struct walker_data *data = walker->data;
 722
 723         struct active_request_slot *slot;
 724         struct slot_results results;
 725
 726         if (fetch_indices(walker, repo))
 727                 return -1;
 728         target = find_sha1_pack(sha1, repo->packs);
 729         if (!target)
 730                 return -1;
 731
 732         if (walker->get_verbosely) {
 733                 fprintf(stderr, "Getting pack %s\n",
 734                         sha1_to_hex(target->sha1));
 735                 fprintf(stderr, " which contains %s\n",
 736                         sha1_to_hex(sha1));
 737         }
 738
 739         url = xmalloc(strlen(repo->base) + 65);
 740         sprintf(url, "%s/objects/pack/pack-%s.pack",
 741                 repo->base, sha1_to_hex(target->sha1));
 742
 743         filename = sha1_pack_name(target->sha1);
 744         snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 745         packfile = fopen(tmpfile, "a");
 746         if (!packfile)
 747                 return error("Unable to open local file %s for pack",
 748                              tmpfile);
 749
 750         slot = get_active_slot();
 751         slot->results = &results;
 752         curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
 753         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 754         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 755         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, data->no_pragma_header);
 756         slot->local = packfile;
 757
 758         /* If there is data present from a previous transfer attempt,
 759            resume where it left off */
 760         prev_posn = ftell(packfile);
 761         if (prev_posn>0) {
 762                 if (walker->get_verbosely)
 763                         fprintf(stderr,
 764                                 "Resuming fetch of pack %s at byte %ld\n",
 765                                 sha1_to_hex(target->sha1), prev_posn);
 766                 sprintf(range, "Range: bytes=%ld-", prev_posn);
 767                 range_header = curl_slist_append(range_header, range);
 768                 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 769         }
 770
 771         if (start_active_slot(slot)) {
 772                 run_active_slot(slot);
 773                 if (results.curl_result != CURLE_OK) {
 774                         fclose(packfile);
 775                         return error("Unable to get pack file %s\n%s", url,
 776                                      curl_errorstr);
 777                 }
 778         } else {
 779                 fclose(packfile);
 780                 return error("Unable to start request");
 781         }
 782
 783         target->pack_size = ftell(packfile);
 784         fclose(packfile);
 785
 786         ret = move_temp_to_file(tmpfile, filename);
 787         if (ret)
 788                 return ret;
 789
 790         lst = &repo->packs;
 791         while (*lst != target)
 792                 lst = &((*lst)->next);
 793         *lst = (*lst)->next;
 794
 795         if (verify_pack(target, 0))
 796                 return -1;
 797         install_packed_git(target);
 798
 799         return 0;
 800 }
 801
 802 static void abort_object_request(struct object_request *obj_req)
 803 {
 804         if (obj_req->local >= 0) {
 805                 close(obj_req->local);
 806                 obj_req->local = -1;
 807         }
 808         unlink(obj_req->tmpfile);
 809         if (obj_req->slot) {
 810                 release_active_slot(obj_req->slot);
 811                 obj_req->slot = NULL;
 812         }
 813         release_object_request(obj_req);
 814 }
 815
 816 static int fetch_object(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 817 {
 818         char *hex = sha1_to_hex(sha1);
 819         int ret = 0;
 820         struct object_request *obj_req = object_queue_head;
 821
 822         while (obj_req != NULL && hashcmp(obj_req->sha1, sha1))
 823                 obj_req = obj_req->next;
 824         if (obj_req == NULL)
 825                 return error("Couldn't find request for %s in the queue", hex);
 826
 827         if (has_sha1_file(obj_req->sha1)) {
 828                 abort_object_request(obj_req);
 829                 return 0;
 830         }
 831
 832 #ifdef USE_CURL_MULTI
 833         while (obj_req->state == WAITING) {
 834                 step_active_slots();
 835         }
 836 #else
 837         start_object_request(walker, obj_req);
 838 #endif
 839
 840         while (obj_req->state == ACTIVE) {
 841                 run_active_slot(obj_req->slot);
 842         }
 843         if (obj_req->local != -1) {
 844                 close(obj_req->local); obj_req->local = -1;
 845         }
 846
 847         if (obj_req->state == ABORTED) {
 848                 ret = error("Request for %s aborted", hex);
 849         } else if (obj_req->curl_result != CURLE_OK &&
 850                    obj_req->http_code != 416) {
 851                 if (missing_target(obj_req))
 852                         ret = -1; /* Be silent, it is probably in a pack. */
 853                 else
 854                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 855                                     obj_req->errorstr, obj_req->curl_result,
 856                                     obj_req->http_code, hex);
 857         } else if (obj_req->zret != Z_STREAM_END) {
 858                 walker->corrupt_object_found++;
 859                 ret = error("File %s (%s) corrupt", hex, obj_req->url);
 860         } else if (hashcmp(obj_req->sha1, obj_req->real_sha1)) {
 861                 ret = error("File %s has bad hash", hex);
 862         } else if (obj_req->rename < 0) {
 863                 ret = error("unable to write sha1 filename %s",
 864                             obj_req->filename);
 865         }
 866
 867         release_object_request(obj_req);
 868         return ret;
 869 }
 870
 871 static int fetch(struct walker *walker, unsigned char *sha1)
 872 {
 873         struct walker_data *data = walker->data;
 874         struct alt_base *altbase = data->alt;
 875
 876         if (!fetch_object(walker, altbase, sha1))
 877                 return 0;
 878         while (altbase) {
 879                 if (!fetch_pack(walker, altbase, sha1))
 880                         return 0;
 881                 fetch_alternates(walker, data->alt->base);
 882                 altbase = altbase->next;
 883         }
 884         return error("Unable to find %s under %s", sha1_to_hex(sha1),
 885                      data->alt->base);
 886 }
 887
 888 static inline int needs_quote(int ch)
 889 {
 890         if (((ch >= 'A') && (ch <= 'Z'))
 891                         || ((ch >= 'a') && (ch <= 'z'))
 892                         || ((ch >= '0') && (ch <= '9'))
 893                         || (ch == '/')
 894                         || (ch == '-')
 895                         || (ch == '.'))
 896                 return 0;
 897         return 1;
 898 }
 899
 900 static inline int hex(int v)
 901 {
 902         if (v < 10) return '0' + v;
 903         else return 'A' + v - 10;
 904 }
 905
 906 static char *quote_ref_url(const char *base, const char *ref)
 907 {
 908         const char *cp;
 909         char *dp, *qref;
 910         int len, baselen, ch;
 911
 912         baselen = strlen(base);
 913         len = baselen + 7; /* "/refs/" + NUL */
 914         for (cp = ref; (ch = *cp) != 0; cp++, len++)
 915                 if (needs_quote(ch))
 916                         len += 2; /* extra two hex plus replacement % */
 917         qref = xmalloc(len);
 918         memcpy(qref, base, baselen);
 919         memcpy(qref + baselen, "/refs/", 6);
 920         for (cp = ref, dp = qref + baselen + 6; (ch = *cp) != 0; cp++) {
 921                 if (needs_quote(ch)) {
 922                         *dp++ = '%';
 923                         *dp++ = hex((ch >> 4) & 0xF);
 924                         *dp++ = hex(ch & 0xF);
 925                 }
 926                 else
 927                         *dp++ = ch;
 928         }
 929         *dp = 0;
 930
 931         return qref;
 932 }
 933
 934 static int fetch_ref(struct walker *walker, char *ref, unsigned char *sha1)
 935 {
 936         char *url;
 937         struct strbuf buffer = STRBUF_INIT;
 938         struct walker_data *data = walker->data;
 939         const char *base = data->alt->base;
 940         struct active_request_slot *slot;
 941         struct slot_results results;
 942
 943         url = quote_ref_url(base, ref);
 944         slot = get_active_slot();
 945         slot->results = &results;
 946         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 947         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 948         curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 949         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 950         if (start_active_slot(slot)) {
 951                 run_active_slot(slot);
 952                 if (results.curl_result != CURLE_OK)
 953                         return error("Couldn't get %s for %s\n%s",
 954                                      url, ref, curl_errorstr);
 955         } else {
 956                 return error("Unable to start request");
 957         }
 958
 959         strbuf_rtrim(&buffer);
 960         if (buffer.len != 40)
 961                 return 1;
 962         return get_sha1_hex(buffer.buf, sha1);
 963 }
 964
 965 static void cleanup(struct walker *walker)
 966 {
 967         struct walker_data *data = walker->data;
 968         http_cleanup();
 969
 970         curl_slist_free_all(data->no_pragma_header);
 971 }
 972
 973 struct walker *get_http_walker(const char *url)
 974 {
 975         char *s;
 976         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 977         struct walker *walker = xmalloc(sizeof(struct walker));
 978
 979         http_init();
 980
 981         data->no_pragma_header = curl_slist_append(NULL, "Pragma:");
 982
 983         data->alt = xmalloc(sizeof(*data->alt));
 984         data->alt->base = xmalloc(strlen(url) + 1);
 985         strcpy(data->alt->base, url);
 986         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 987                 *s = 0;
 988
 989         data->alt->got_indices = 0;
 990         data->alt->packs = NULL;
 991         data->alt->next = NULL;
 992         data->got_alternates = -1;
 993
 994         walker->corrupt_object_found = 0;
 995         walker->fetch = fetch;
 996         walker->fetch_ref = fetch_ref;
 997         walker->prefetch = prefetch;
 998         walker->cleanup = cleanup;
 999         walker->data = data;
1000
1001 #ifdef USE_CURL_MULTI
1002         add_fill_function(walker, (int (*)(void *)) fill_active_slot);
1003 #endif
1004
1005         return walker;
1006 }