Andrew's git / gitweb.git / http-walker.c
?
http-walker.con commit send-email: 'References:' should only reference what is sent (15da108)
   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
  10struct alt_base
  11{
  12        char *base;
  13        int got_indices;
  14        struct packed_git *packs;
  15        struct alt_base *next;
  16};
  17
  18enum object_request_state {
  19        WAITING,
  20        ABORTED,
  21        ACTIVE,
  22        COMPLETE,
  23};
  24
  25struct 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        git_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
  47struct 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
  56struct walker_data {
  57        const char *url;
  58        int got_alternates;
  59        struct alt_base *alt;
  60        struct curl_slist *no_pragma_header;
  61};
  62
  63static struct object_request *object_queue_head;
  64
  65static 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 = git_inflate(&obj_req->stream, Z_SYNC_FLUSH);
  86                git_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
  93static void fetch_alternates(struct walker *walker, const char *base);
  94
  95static void process_object_response(void *callback_data);
  96
  97static 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        git_inflate_init(&obj_req->stream);
 146
 147        git_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                git_inflate_init(&obj_req->stream);
 187                git_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
 230static void finish_object_request(struct object_request *obj_req)
 231{
 232        struct stat st;
 233
 234        close(obj_req->local); obj_req->local = -1;
 235
 236        if (obj_req->http_code == 416) {
 237                fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 238        } else if (obj_req->curl_result != CURLE_OK) {
 239                if (stat(obj_req->tmpfile, &st) == 0)
 240                        if (st.st_size == 0)
 241                                unlink(obj_req->tmpfile);
 242                return;
 243        }
 244
 245        git_inflate_end(&obj_req->stream);
 246        git_SHA1_Final(obj_req->real_sha1, &obj_req->c);
 247        if (obj_req->zret != Z_STREAM_END) {
 248                unlink(obj_req->tmpfile);
 249                return;
 250        }
 251        if (hashcmp(obj_req->sha1, obj_req->real_sha1)) {
 252                unlink(obj_req->tmpfile);
 253                return;
 254        }
 255        obj_req->rename =
 256                move_temp_to_file(obj_req->tmpfile, obj_req->filename);
 257
 258        if (obj_req->rename == 0)
 259                walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
 260}
 261
 262static void process_object_response(void *callback_data)
 263{
 264        struct object_request *obj_req =
 265                (struct object_request *)callback_data;
 266        struct walker *walker = obj_req->walker;
 267        struct walker_data *data = walker->data;
 268        struct alt_base *alt = data->alt;
 269
 270        obj_req->curl_result = obj_req->slot->curl_result;
 271        obj_req->http_code = obj_req->slot->http_code;
 272        obj_req->slot = NULL;
 273        obj_req->state = COMPLETE;
 274
 275        /* Use alternates if necessary */
 276        if (missing_target(obj_req)) {
 277                fetch_alternates(walker, alt->base);
 278                if (obj_req->repo->next != NULL) {
 279                        obj_req->repo =
 280                                obj_req->repo->next;
 281                        close(obj_req->local);
 282                        obj_req->local = -1;
 283                        start_object_request(walker, obj_req);
 284                        return;
 285                }
 286        }
 287
 288        finish_object_request(obj_req);
 289}
 290
 291static void release_object_request(struct object_request *obj_req)
 292{
 293        struct object_request *entry = object_queue_head;
 294
 295        if (obj_req->local != -1)
 296                error("fd leakage in release: %d", obj_req->local);
 297        if (obj_req == object_queue_head) {
 298                object_queue_head = obj_req->next;
 299        } else {
 300                while (entry->next != NULL && entry->next != obj_req)
 301                        entry = entry->next;
 302                if (entry->next == obj_req)
 303                        entry->next = entry->next->next;
 304        }
 305
 306        free(obj_req->url);
 307        free(obj_req);
 308}
 309
 310#ifdef USE_CURL_MULTI
 311static int fill_active_slot(struct walker *walker)
 312{
 313        struct object_request *obj_req;
 314
 315        for (obj_req = object_queue_head; obj_req; obj_req = obj_req->next) {
 316                if (obj_req->state == WAITING) {
 317                        if (has_sha1_file(obj_req->sha1))
 318                                obj_req->state = COMPLETE;
 319                        else {
 320                                start_object_request(walker, obj_req);
 321                                return 1;
 322                        }
 323                }
 324        }
 325        return 0;
 326}
 327#endif
 328
 329static void prefetch(struct walker *walker, unsigned char *sha1)
 330{
 331        struct object_request *newreq;
 332        struct object_request *tail;
 333        struct walker_data *data = walker->data;
 334        char *filename = sha1_file_name(sha1);
 335
 336        newreq = xmalloc(sizeof(*newreq));
 337        newreq->walker = walker;
 338        hashcpy(newreq->sha1, sha1);
 339        newreq->repo = data->alt;
 340        newreq->url = NULL;
 341        newreq->local = -1;
 342        newreq->state = WAITING;
 343        snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 344        snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 345                 "%s.temp", filename);
 346        newreq->slot = NULL;
 347        newreq->next = NULL;
 348
 349        if (object_queue_head == NULL) {
 350                object_queue_head = newreq;
 351        } else {
 352                tail = object_queue_head;
 353                while (tail->next != NULL) {
 354                        tail = tail->next;
 355                }
 356                tail->next = newreq;
 357        }
 358
 359#ifdef USE_CURL_MULTI
 360        fill_active_slots();
 361        step_active_slots();
 362#endif
 363}
 364
 365static int fetch_index(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 366{
 367        char *hex = sha1_to_hex(sha1);
 368        char *filename;
 369        char *url;
 370        char tmpfile[PATH_MAX];
 371        long prev_posn = 0;
 372        char range[RANGE_HEADER_SIZE];
 373        struct curl_slist *range_header = NULL;
 374        struct walker_data *data = walker->data;
 375
 376        FILE *indexfile;
 377        struct active_request_slot *slot;
 378        struct slot_results results;
 379
 380        if (has_pack_index(sha1))
 381                return 0;
 382
 383        if (walker->get_verbosely)
 384                fprintf(stderr, "Getting index for pack %s\n", hex);
 385
 386        url = xmalloc(strlen(repo->base) + 64);
 387        sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 388
 389        filename = sha1_pack_index_name(sha1);
 390        snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 391        indexfile = fopen(tmpfile, "a");
 392        if (!indexfile)
 393                return error("Unable to open local file %s for pack index",
 394                             tmpfile);
 395
 396        slot = get_active_slot();
 397        slot->results = &results;
 398        curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 399        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 400        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 401        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, data->no_pragma_header);
 402        slot->local = indexfile;
 403
 404        /* If there is data present from a previous transfer attempt,
 405           resume where it left off */
 406        prev_posn = ftell(indexfile);
 407        if (prev_posn>0) {
 408                if (walker->get_verbosely)
 409                        fprintf(stderr,
 410                                "Resuming fetch of index for pack %s at byte %ld\n",
 411                                hex, prev_posn);
 412                sprintf(range, "Range: bytes=%ld-", prev_posn);
 413                range_header = curl_slist_append(range_header, range);
 414                curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 415        }
 416
 417        if (start_active_slot(slot)) {
 418                run_active_slot(slot);
 419                if (results.curl_result != CURLE_OK) {
 420                        fclose(indexfile);
 421                        return error("Unable to get pack index %s\n%s", url,
 422                                     curl_errorstr);
 423                }
 424        } else {
 425                fclose(indexfile);
 426                return error("Unable to start request");
 427        }
 428
 429        fclose(indexfile);
 430
 431        return move_temp_to_file(tmpfile, filename);
 432}
 433
 434static int setup_index(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 435{
 436        struct packed_git *new_pack;
 437        if (has_pack_file(sha1))
 438                return 0; /* don't list this as something we can get */
 439
 440        if (fetch_index(walker, repo, sha1))
 441                return -1;
 442
 443        new_pack = parse_pack_index(sha1);
 444        if (!new_pack)
 445                return -1; /* parse_pack_index() already issued error message */
 446        new_pack->next = repo->packs;
 447        repo->packs = new_pack;
 448        return 0;
 449}
 450
 451static void process_alternates_response(void *callback_data)
 452{
 453        struct alternates_request *alt_req =
 454                (struct alternates_request *)callback_data;
 455        struct walker *walker = alt_req->walker;
 456        struct walker_data *cdata = walker->data;
 457        struct active_request_slot *slot = alt_req->slot;
 458        struct alt_base *tail = cdata->alt;
 459        const char *base = alt_req->base;
 460        static const char null_byte = '\0';
 461        char *data;
 462        int i = 0;
 463
 464        if (alt_req->http_specific) {
 465                if (slot->curl_result != CURLE_OK ||
 466                    !alt_req->buffer->len) {
 467
 468                        /* Try reusing the slot to get non-http alternates */
 469                        alt_req->http_specific = 0;
 470                        sprintf(alt_req->url, "%s/objects/info/alternates",
 471                                base);
 472                        curl_easy_setopt(slot->curl, CURLOPT_URL,
 473                                         alt_req->url);
 474                        active_requests++;
 475                        slot->in_use = 1;
 476                        if (slot->finished != NULL)
 477                                (*slot->finished) = 0;
 478                        if (!start_active_slot(slot)) {
 479                                cdata->got_alternates = -1;
 480                                slot->in_use = 0;
 481                                if (slot->finished != NULL)
 482                                        (*slot->finished) = 1;
 483                        }
 484                        return;
 485                }
 486        } else if (slot->curl_result != CURLE_OK) {
 487                if (!missing_target(slot)) {
 488                        cdata->got_alternates = -1;
 489                        return;
 490                }
 491        }
 492
 493        fwrite_buffer(&null_byte, 1, 1, alt_req->buffer);
 494        alt_req->buffer->len--;
 495        data = alt_req->buffer->buf;
 496
 497        while (i < alt_req->buffer->len) {
 498                int posn = i;
 499                while (posn < alt_req->buffer->len && data[posn] != '\n')
 500                        posn++;
 501                if (data[posn] == '\n') {
 502                        int okay = 0;
 503                        int serverlen = 0;
 504                        struct alt_base *newalt;
 505                        char *target = NULL;
 506                        if (data[i] == '/') {
 507                                /* This counts
 508                                 * http://git.host/pub/scm/linux.git/
 509                                 * -----------here^
 510                                 * so memcpy(dst, base, serverlen) will
 511                                 * copy up to "...git.host".
 512                                 */
 513                                const char *colon_ss = strstr(base,"://");
 514                                if (colon_ss) {
 515                                        serverlen = (strchr(colon_ss + 3, '/')
 516                                                     - base);
 517                                        okay = 1;
 518                                }
 519                        } else if (!memcmp(data + i, "../", 3)) {
 520                                /* Relative URL; chop the corresponding
 521                                 * number of subpath from base (and ../
 522                                 * from data), and concatenate the result.
 523                                 *
 524                                 * The code first drops ../ from data, and
 525                                 * then drops one ../ from data and one path
 526                                 * from base.  IOW, one extra ../ is dropped
 527                                 * from data than path is dropped from base.
 528                                 *
 529                                 * This is not wrong.  The alternate in
 530                                 *     http://git.host/pub/scm/linux.git/
 531                                 * to borrow from
 532                                 *     http://git.host/pub/scm/linus.git/
 533                                 * is ../../linus.git/objects/.  You need
 534                                 * two ../../ to borrow from your direct
 535                                 * neighbour.
 536                                 */
 537                                i += 3;
 538                                serverlen = strlen(base);
 539                                while (i + 2 < posn &&
 540                                       !memcmp(data + i, "../", 3)) {
 541                                        do {
 542                                                serverlen--;
 543                                        } while (serverlen &&
 544                                                 base[serverlen - 1] != '/');
 545                                        i += 3;
 546                                }
 547                                /* If the server got removed, give up. */
 548                                okay = strchr(base, ':') - base + 3 <
 549                                        serverlen;
 550                        } else if (alt_req->http_specific) {
 551                                char *colon = strchr(data + i, ':');
 552                                char *slash = strchr(data + i, '/');
 553                                if (colon && slash && colon < data + posn &&
 554                                    slash < data + posn && colon < slash) {
 555                                        okay = 1;
 556                                }
 557                        }
 558                        /* skip "objects\n" at end */
 559                        if (okay) {
 560                                target = xmalloc(serverlen + posn - i - 6);
 561                                memcpy(target, base, serverlen);
 562                                memcpy(target + serverlen, data + i,
 563                                       posn - i - 7);
 564                                target[serverlen + posn - i - 7] = 0;
 565                                if (walker->get_verbosely)
 566                                        fprintf(stderr,
 567                                                "Also look at %s\n", target);
 568                                newalt = xmalloc(sizeof(*newalt));
 569                                newalt->next = NULL;
 570                                newalt->base = target;
 571                                newalt->got_indices = 0;
 572                                newalt->packs = NULL;
 573
 574                                while (tail->next != NULL)
 575                                        tail = tail->next;
 576                                tail->next = newalt;
 577                        }
 578                }
 579                i = posn + 1;
 580        }
 581
 582        cdata->got_alternates = 1;
 583}
 584
 585static void fetch_alternates(struct walker *walker, const char *base)
 586{
 587        struct strbuf buffer = STRBUF_INIT;
 588        char *url;
 589        struct active_request_slot *slot;
 590        struct alternates_request alt_req;
 591        struct walker_data *cdata = walker->data;
 592
 593        /* If another request has already started fetching alternates,
 594           wait for them to arrive and return to processing this request's
 595           curl message */
 596#ifdef USE_CURL_MULTI
 597        while (cdata->got_alternates == 0) {
 598                step_active_slots();
 599        }
 600#endif
 601
 602        /* Nothing to do if they've already been fetched */
 603        if (cdata->got_alternates == 1)
 604                return;
 605
 606        /* Start the fetch */
 607        cdata->got_alternates = 0;
 608
 609        if (walker->get_verbosely)
 610                fprintf(stderr, "Getting alternates list for %s\n", base);
 611
 612        url = xmalloc(strlen(base) + 31);
 613        sprintf(url, "%s/objects/info/http-alternates", base);
 614
 615        /* Use a callback to process the result, since another request
 616           may fail and need to have alternates loaded before continuing */
 617        slot = get_active_slot();
 618        slot->callback_func = process_alternates_response;
 619        alt_req.walker = walker;
 620        slot->callback_data = &alt_req;
 621
 622        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 623        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 624        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 625
 626        alt_req.base = base;
 627        alt_req.url = url;
 628        alt_req.buffer = &buffer;
 629        alt_req.http_specific = 1;
 630        alt_req.slot = slot;
 631
 632        if (start_active_slot(slot))
 633                run_active_slot(slot);
 634        else
 635                cdata->got_alternates = -1;
 636
 637        strbuf_release(&buffer);
 638        free(url);
 639}
 640
 641static int fetch_indices(struct walker *walker, struct alt_base *repo)
 642{
 643        unsigned char sha1[20];
 644        char *url;
 645        struct strbuf buffer = STRBUF_INIT;
 646        char *data;
 647        int i = 0;
 648        int ret = 0;
 649
 650        struct active_request_slot *slot;
 651        struct slot_results results;
 652
 653        if (repo->got_indices)
 654                return 0;
 655
 656        if (walker->get_verbosely)
 657                fprintf(stderr, "Getting pack list for %s\n", repo->base);
 658
 659        url = xmalloc(strlen(repo->base) + 21);
 660        sprintf(url, "%s/objects/info/packs", repo->base);
 661
 662        slot = get_active_slot();
 663        slot->results = &results;
 664        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 665        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 666        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 667        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 668        if (start_active_slot(slot)) {
 669                run_active_slot(slot);
 670                if (results.curl_result != CURLE_OK) {
 671                        if (missing_target(&results)) {
 672                                repo->got_indices = 1;
 673                                goto cleanup;
 674                        } else {
 675                                repo->got_indices = 0;
 676                                ret = error("%s", curl_errorstr);
 677                                goto cleanup;
 678                        }
 679                }
 680        } else {
 681                repo->got_indices = 0;
 682                ret = error("Unable to start request");
 683                goto cleanup;
 684        }
 685
 686        data = buffer.buf;
 687        while (i < buffer.len) {
 688                switch (data[i]) {
 689                case 'P':
 690                        i++;
 691                        if (i + 52 <= buffer.len &&
 692                            !prefixcmp(data + i, " pack-") &&
 693                            !prefixcmp(data + i + 46, ".pack\n")) {
 694                                get_sha1_hex(data + i + 6, sha1);
 695                                setup_index(walker, repo, sha1);
 696                                i += 51;
 697                                break;
 698                        }
 699                default:
 700                        while (i < buffer.len && data[i] != '\n')
 701                                i++;
 702                }
 703                i++;
 704        }
 705
 706        repo->got_indices = 1;
 707cleanup:
 708        strbuf_release(&buffer);
 709        free(url);
 710        return ret;
 711}
 712
 713static int fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 714{
 715        char *url;
 716        struct packed_git *target;
 717        struct packed_git **lst;
 718        FILE *packfile;
 719        char *filename;
 720        char tmpfile[PATH_MAX];
 721        int ret;
 722        long prev_posn = 0;
 723        char range[RANGE_HEADER_SIZE];
 724        struct curl_slist *range_header = NULL;
 725        struct walker_data *data = walker->data;
 726
 727        struct active_request_slot *slot;
 728        struct slot_results results;
 729
 730        if (fetch_indices(walker, repo))
 731                return -1;
 732        target = find_sha1_pack(sha1, repo->packs);
 733        if (!target)
 734                return -1;
 735
 736        if (walker->get_verbosely) {
 737                fprintf(stderr, "Getting pack %s\n",
 738                        sha1_to_hex(target->sha1));
 739                fprintf(stderr, " which contains %s\n",
 740                        sha1_to_hex(sha1));
 741        }
 742
 743        url = xmalloc(strlen(repo->base) + 65);
 744        sprintf(url, "%s/objects/pack/pack-%s.pack",
 745                repo->base, sha1_to_hex(target->sha1));
 746
 747        filename = sha1_pack_name(target->sha1);
 748        snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 749        packfile = fopen(tmpfile, "a");
 750        if (!packfile)
 751                return error("Unable to open local file %s for pack",
 752                             tmpfile);
 753
 754        slot = get_active_slot();
 755        slot->results = &results;
 756        curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
 757        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 758        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 759        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, data->no_pragma_header);
 760        slot->local = packfile;
 761
 762        /* If there is data present from a previous transfer attempt,
 763           resume where it left off */
 764        prev_posn = ftell(packfile);
 765        if (prev_posn>0) {
 766                if (walker->get_verbosely)
 767                        fprintf(stderr,
 768                                "Resuming fetch of pack %s at byte %ld\n",
 769                                sha1_to_hex(target->sha1), prev_posn);
 770                sprintf(range, "Range: bytes=%ld-", prev_posn);
 771                range_header = curl_slist_append(range_header, range);
 772                curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 773        }
 774
 775        if (start_active_slot(slot)) {
 776                run_active_slot(slot);
 777                if (results.curl_result != CURLE_OK) {
 778                        fclose(packfile);
 779                        return error("Unable to get pack file %s\n%s", url,
 780                                     curl_errorstr);
 781                }
 782        } else {
 783                fclose(packfile);
 784                return error("Unable to start request");
 785        }
 786
 787        target->pack_size = ftell(packfile);
 788        fclose(packfile);
 789
 790        ret = move_temp_to_file(tmpfile, filename);
 791        if (ret)
 792                return ret;
 793
 794        lst = &repo->packs;
 795        while (*lst != target)
 796                lst = &((*lst)->next);
 797        *lst = (*lst)->next;
 798
 799        if (verify_pack(target))
 800                return -1;
 801        install_packed_git(target);
 802
 803        return 0;
 804}
 805
 806static void abort_object_request(struct object_request *obj_req)
 807{
 808        if (obj_req->local >= 0) {
 809                close(obj_req->local);
 810                obj_req->local = -1;
 811        }
 812        unlink(obj_req->tmpfile);
 813        if (obj_req->slot) {
 814                release_active_slot(obj_req->slot);
 815                obj_req->slot = NULL;
 816        }
 817        release_object_request(obj_req);
 818}
 819
 820static int fetch_object(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 821{
 822        char *hex = sha1_to_hex(sha1);
 823        int ret = 0;
 824        struct object_request *obj_req = object_queue_head;
 825
 826        while (obj_req != NULL && hashcmp(obj_req->sha1, sha1))
 827                obj_req = obj_req->next;
 828        if (obj_req == NULL)
 829                return error("Couldn't find request for %s in the queue", hex);
 830
 831        if (has_sha1_file(obj_req->sha1)) {
 832                abort_object_request(obj_req);
 833                return 0;
 834        }
 835
 836#ifdef USE_CURL_MULTI
 837        while (obj_req->state == WAITING) {
 838                step_active_slots();
 839        }
 840#else
 841        start_object_request(walker, obj_req);
 842#endif
 843
 844        while (obj_req->state == ACTIVE) {
 845                run_active_slot(obj_req->slot);
 846        }
 847        if (obj_req->local != -1) {
 848                close(obj_req->local); obj_req->local = -1;
 849        }
 850
 851        if (obj_req->state == ABORTED) {
 852                ret = error("Request for %s aborted", hex);
 853        } else if (obj_req->curl_result != CURLE_OK &&
 854                   obj_req->http_code != 416) {
 855                if (missing_target(obj_req))
 856                        ret = -1; /* Be silent, it is probably in a pack. */
 857                else
 858                        ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 859                                    obj_req->errorstr, obj_req->curl_result,
 860                                    obj_req->http_code, hex);
 861        } else if (obj_req->zret != Z_STREAM_END) {
 862                walker->corrupt_object_found++;
 863                ret = error("File %s (%s) corrupt", hex, obj_req->url);
 864        } else if (hashcmp(obj_req->sha1, obj_req->real_sha1)) {
 865                ret = error("File %s has bad hash", hex);
 866        } else if (obj_req->rename < 0) {
 867                ret = error("unable to write sha1 filename %s",
 868                            obj_req->filename);
 869        }
 870
 871        release_object_request(obj_req);
 872        return ret;
 873}
 874
 875static int fetch(struct walker *walker, unsigned char *sha1)
 876{
 877        struct walker_data *data = walker->data;
 878        struct alt_base *altbase = data->alt;
 879
 880        if (!fetch_object(walker, altbase, sha1))
 881                return 0;
 882        while (altbase) {
 883                if (!fetch_pack(walker, altbase, sha1))
 884                        return 0;
 885                fetch_alternates(walker, data->alt->base);
 886                altbase = altbase->next;
 887        }
 888        return error("Unable to find %s under %s", sha1_to_hex(sha1),
 889                     data->alt->base);
 890}
 891
 892static int fetch_ref(struct walker *walker, struct ref *ref)
 893{
 894        struct walker_data *data = walker->data;
 895        return http_fetch_ref(data->alt->base, ref);
 896}
 897
 898static void cleanup(struct walker *walker)
 899{
 900        struct walker_data *data = walker->data;
 901        http_cleanup();
 902
 903        curl_slist_free_all(data->no_pragma_header);
 904}
 905
 906struct walker *get_http_walker(const char *url, struct remote *remote)
 907{
 908        char *s;
 909        struct walker_data *data = xmalloc(sizeof(struct walker_data));
 910        struct walker *walker = xmalloc(sizeof(struct walker));
 911
 912        http_init(remote);
 913
 914        data->no_pragma_header = curl_slist_append(NULL, "Pragma:");
 915
 916        data->alt = xmalloc(sizeof(*data->alt));
 917        data->alt->base = xmalloc(strlen(url) + 1);
 918        strcpy(data->alt->base, url);
 919        for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 920                *s = 0;
 921
 922        data->alt->got_indices = 0;
 923        data->alt->packs = NULL;
 924        data->alt->next = NULL;
 925        data->got_alternates = -1;
 926
 927        walker->corrupt_object_found = 0;
 928        walker->fetch = fetch;
 929        walker->fetch_ref = fetch_ref;
 930        walker->prefetch = prefetch;
 931        walker->cleanup = cleanup;
 932        walker->data = data;
 933
 934#ifdef USE_CURL_MULTI
 935        add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 936#endif
 937
 938        return walker;
 939}