Andrew's git / gitweb.git / http-fetch.c
?
http-fetch.con commit revised^2: git-daemon extra paranoia, and path DWIM (3e04c62)
   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
  28static int active_requests = 0;
  29static int data_received;
  30
  31#ifdef USE_CURL_MULTI
  32static int max_requests = -1;
  33static CURLM *curlm;
  34#endif
  35#ifndef NO_CURL_EASY_DUPHANDLE
  36static CURL *curl_default;
  37#endif
  38static struct curl_slist *pragma_header;
  39static struct curl_slist *no_pragma_header;
  40static struct curl_slist *no_range_header;
  41static char curl_errorstr[CURL_ERROR_SIZE];
  42
  43struct alt_base
  44{
  45        char *base;
  46        int got_indices;
  47        struct packed_git *packs;
  48        struct alt_base *next;
  49};
  50
  51static struct alt_base *alt = NULL;
  52
  53enum transfer_state {
  54        WAITING,
  55        ABORTED,
  56        ACTIVE,
  57        COMPLETE,
  58};
  59
  60struct transfer_request
  61{
  62        unsigned char sha1[20];
  63        struct alt_base *repo;
  64        char *url;
  65        char filename[PATH_MAX];
  66        char tmpfile[PATH_MAX];
  67        int local;
  68        enum transfer_state state;
  69        CURLcode curl_result;
  70        char errorstr[CURL_ERROR_SIZE];
  71        long http_code;
  72        unsigned char real_sha1[20];
  73        SHA_CTX c;
  74        z_stream stream;
  75        int zret;
  76        int rename;
  77        struct active_request_slot *slot;
  78        struct transfer_request *next;
  79};
  80
  81struct active_request_slot
  82{
  83        CURL *curl;
  84        FILE *local;
  85        int in_use;
  86        int done;
  87        CURLcode curl_result;
  88        struct active_request_slot *next;
  89};
  90
  91static struct transfer_request *request_queue_head = NULL;
  92static struct active_request_slot *active_queue_head = NULL;
  93
  94static int curl_ssl_verify = -1;
  95static char *ssl_cert = NULL;
  96#if LIBCURL_VERSION_NUM >= 0x070902
  97static char *ssl_key = NULL;
  98#endif
  99#if LIBCURL_VERSION_NUM >= 0x070908
 100static char *ssl_capath = NULL;
 101#endif
 102static char *ssl_cainfo = NULL;
 103
 104struct buffer
 105{
 106        size_t posn;
 107        size_t size;
 108        void *buffer;
 109};
 110
 111static int http_options(const char *var, const char *value)
 112{
 113        if (!strcmp("http.sslverify", var)) {
 114                if (curl_ssl_verify == -1) {
 115                        curl_ssl_verify = git_config_bool(var, value);
 116                }
 117                return 0;
 118        }
 119
 120        if (!strcmp("http.sslcert", var)) {
 121                if (ssl_cert == NULL) {
 122                        ssl_cert = xmalloc(strlen(value)+1);
 123                        strcpy(ssl_cert, value);
 124                }
 125                return 0;
 126        }
 127#if LIBCURL_VERSION_NUM >= 0x070902
 128        if (!strcmp("http.sslkey", var)) {
 129                if (ssl_key == NULL) {
 130                        ssl_key = xmalloc(strlen(value)+1);
 131                        strcpy(ssl_key, value);
 132                }
 133                return 0;
 134        }
 135#endif
 136#if LIBCURL_VERSION_NUM >= 0x070908
 137        if (!strcmp("http.sslcapath", var)) {
 138                if (ssl_capath == NULL) {
 139                        ssl_capath = xmalloc(strlen(value)+1);
 140                        strcpy(ssl_capath, value);
 141                }
 142                return 0;
 143        }
 144#endif
 145        if (!strcmp("http.sslcainfo", var)) {
 146                if (ssl_cainfo == NULL) {
 147                        ssl_cainfo = xmalloc(strlen(value)+1);
 148                        strcpy(ssl_cainfo, value);
 149                }
 150                return 0;
 151        }
 152
 153#ifdef USE_CURL_MULTI   
 154        if (!strcmp("http.maxrequests", var)) {
 155                if (max_requests == -1)
 156                        max_requests = git_config_int(var, value);
 157                return 0;
 158        }
 159#endif
 160
 161        /* Fall back on the default ones */
 162        return git_default_config(var, value);
 163}
 164
 165static size_t fwrite_buffer(void *ptr, size_t eltsize, size_t nmemb,
 166                            struct buffer *buffer)
 167{
 168        size_t size = eltsize * nmemb;
 169        if (size > buffer->size - buffer->posn)
 170                size = buffer->size - buffer->posn;
 171        memcpy(buffer->buffer + buffer->posn, ptr, size);
 172        buffer->posn += size;
 173        data_received++;
 174        return size;
 175}
 176
 177static size_t fwrite_buffer_dynamic(const void *ptr, size_t eltsize,
 178                                    size_t nmemb, struct buffer *buffer)
 179{
 180        size_t size = eltsize * nmemb;
 181        if (size > buffer->size - buffer->posn) {
 182                buffer->size = buffer->size * 3 / 2;
 183                if (buffer->size < buffer->posn + size)
 184                        buffer->size = buffer->posn + size;
 185                buffer->buffer = xrealloc(buffer->buffer, buffer->size);
 186        }
 187        memcpy(buffer->buffer + buffer->posn, ptr, size);
 188        buffer->posn += size;
 189        data_received++;
 190        return size;
 191}
 192
 193static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
 194                               void *data)
 195{
 196        unsigned char expn[4096];
 197        size_t size = eltsize * nmemb;
 198        int posn = 0;
 199        struct transfer_request *request = (struct transfer_request *)data;
 200        do {
 201                ssize_t retval = write(request->local,
 202                                       ptr + posn, size - posn);
 203                if (retval < 0)
 204                        return posn;
 205                posn += retval;
 206        } while (posn < size);
 207
 208        request->stream.avail_in = size;
 209        request->stream.next_in = ptr;
 210        do {
 211                request->stream.next_out = expn;
 212                request->stream.avail_out = sizeof(expn);
 213                request->zret = inflate(&request->stream, Z_SYNC_FLUSH);
 214                SHA1_Update(&request->c, expn,
 215                            sizeof(expn) - request->stream.avail_out);
 216        } while (request->stream.avail_in && request->zret == Z_OK);
 217        data_received++;
 218        return size;
 219}
 220
 221#ifdef USE_CURL_MULTI
 222static void process_curl_messages(void);
 223static void process_request_queue(void);
 224#endif
 225
 226static CURL* get_curl_handle(void)
 227{
 228        CURL* result = curl_easy_init();
 229
 230        curl_easy_setopt(result, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
 231#if LIBCURL_VERSION_NUM >= 0x070907
 232        curl_easy_setopt(result, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
 233#endif
 234
 235        if (ssl_cert != NULL)
 236                curl_easy_setopt(result, CURLOPT_SSLCERT, ssl_cert);
 237#if LIBCURL_VERSION_NUM >= 0x070902
 238        if (ssl_key != NULL)
 239                curl_easy_setopt(result, CURLOPT_SSLKEY, ssl_key);
 240#endif
 241#if LIBCURL_VERSION_NUM >= 0x070908
 242        if (ssl_capath != NULL)
 243                curl_easy_setopt(result, CURLOPT_CAPATH, ssl_capath);
 244#endif
 245        if (ssl_cainfo != NULL)
 246                curl_easy_setopt(result, CURLOPT_CAINFO, ssl_cainfo);
 247        curl_easy_setopt(result, CURLOPT_FAILONERROR, 1);
 248
 249        return result;
 250}
 251
 252static struct active_request_slot *get_active_slot(void)
 253{
 254        struct active_request_slot *slot = active_queue_head;
 255        struct active_request_slot *newslot;
 256
 257#ifdef USE_CURL_MULTI
 258        int num_transfers;
 259
 260        /* Wait for a slot to open up if the queue is full */
 261        while (active_requests >= max_requests) {
 262                curl_multi_perform(curlm, &num_transfers);
 263                if (num_transfers < active_requests) {
 264                        process_curl_messages();
 265                }
 266        }
 267#endif
 268
 269        while (slot != NULL && slot->in_use) {
 270                slot = slot->next;
 271        }
 272        if (slot == NULL) {
 273                newslot = xmalloc(sizeof(*newslot));
 274#ifdef NO_CURL_EASY_DUPHANDLE
 275                newslot->curl = get_curl_handle();
 276#else
 277                newslot->curl = curl_easy_duphandle(curl_default);
 278#endif
 279                newslot->in_use = 0;
 280                newslot->next = NULL;
 281
 282                slot = active_queue_head;
 283                if (slot == NULL) {
 284                        active_queue_head = newslot;
 285                } else {
 286                        while (slot->next != NULL) {
 287                                slot = slot->next;
 288                        }
 289                        slot->next = newslot;
 290                }
 291                slot = newslot;
 292        }
 293
 294        active_requests++;
 295        slot->in_use = 1;
 296        slot->done = 0;
 297        slot->local = NULL;
 298        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, pragma_header);
 299        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_range_header);
 300        curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 301
 302        return slot;
 303}
 304
 305static int start_active_slot(struct active_request_slot *slot)
 306{
 307#ifdef USE_CURL_MULTI
 308        CURLMcode curlm_result = curl_multi_add_handle(curlm, slot->curl);
 309
 310        if (curlm_result != CURLM_OK &&
 311            curlm_result != CURLM_CALL_MULTI_PERFORM) {
 312                active_requests--;
 313                slot->in_use = 0;
 314                return 0;
 315        }
 316#endif
 317        return 1;
 318}
 319
 320static void run_active_slot(struct active_request_slot *slot)
 321{
 322#ifdef USE_CURL_MULTI
 323        int num_transfers;
 324        long last_pos = 0;
 325        long current_pos;
 326        fd_set readfds;
 327        fd_set writefds;
 328        fd_set excfds;
 329        int max_fd;
 330        struct timeval select_timeout;
 331        CURLMcode curlm_result;
 332
 333        while (!slot->done) {
 334                data_received = 0;
 335                do {
 336                        curlm_result = curl_multi_perform(curlm,
 337                                                          &num_transfers);
 338                } while (curlm_result == CURLM_CALL_MULTI_PERFORM);
 339                if (num_transfers < active_requests) {
 340                        process_curl_messages();
 341                        process_request_queue();
 342                }
 343
 344                if (!data_received && slot->local != NULL) {
 345                        current_pos = ftell(slot->local);
 346                        if (current_pos > last_pos)
 347                                data_received++;
 348                        last_pos = current_pos;
 349                }
 350
 351                if (!slot->done && !data_received) {
 352                        max_fd = 0;
 353                        FD_ZERO(&readfds);
 354                        FD_ZERO(&writefds);
 355                        FD_ZERO(&excfds);
 356                        select_timeout.tv_sec = 0;
 357                        select_timeout.tv_usec = 50000;
 358                        select(max_fd, &readfds, &writefds,
 359                               &excfds, &select_timeout);
 360                }
 361        }
 362#else
 363        slot->curl_result = curl_easy_perform(slot->curl);
 364        active_requests--;
 365#endif
 366}
 367
 368static void start_request(struct transfer_request *request)
 369{
 370        char *hex = sha1_to_hex(request->sha1);
 371        char prevfile[PATH_MAX];
 372        char *url;
 373        char *posn;
 374        int prevlocal;
 375        unsigned char prev_buf[PREV_BUF_SIZE];
 376        ssize_t prev_read = 0;
 377        long prev_posn = 0;
 378        char range[RANGE_HEADER_SIZE];
 379        struct curl_slist *range_header = NULL;
 380        struct active_request_slot *slot;
 381
 382        snprintf(prevfile, sizeof(prevfile), "%s.prev", request->filename);
 383        unlink(prevfile);
 384        rename(request->tmpfile, prevfile);
 385        unlink(request->tmpfile);
 386
 387        request->local = open(request->tmpfile,
 388                              O_WRONLY | O_CREAT | O_EXCL, 0666);
 389        /* This could have failed due to the "lazy directory creation";
 390         * try to mkdir the last path component.
 391         */
 392        if (request->local < 0 && errno == ENOENT) {
 393                char *dir = strrchr(request->tmpfile, '/');
 394                if (dir) {
 395                        *dir = 0;
 396                        mkdir(request->tmpfile, 0777);
 397                        *dir = '/';
 398                }
 399                request->local = open(request->tmpfile,
 400                                      O_WRONLY | O_CREAT | O_EXCL, 0666);
 401        }
 402
 403        if (request->local < 0) {
 404                request->state = ABORTED;
 405                error("Couldn't create temporary file %s for %s: %s\n",
 406                      request->tmpfile, request->filename, strerror(errno));
 407                return;
 408        }
 409
 410        memset(&request->stream, 0, sizeof(request->stream));
 411
 412        inflateInit(&request->stream);
 413
 414        SHA1_Init(&request->c);
 415
 416        url = xmalloc(strlen(request->repo->base) + 50);
 417        request->url = xmalloc(strlen(request->repo->base) + 50);
 418        strcpy(url, request->repo->base);
 419        posn = url + strlen(request->repo->base);
 420        strcpy(posn, "objects/");
 421        posn += 8;
 422        memcpy(posn, hex, 2);
 423        posn += 2;
 424        *(posn++) = '/';
 425        strcpy(posn, hex + 2);
 426        strcpy(request->url, url);
 427
 428        /* If a previous temp file is present, process what was already
 429           fetched. */
 430        prevlocal = open(prevfile, O_RDONLY);
 431        if (prevlocal != -1) {
 432                do {
 433                        prev_read = read(prevlocal, prev_buf, PREV_BUF_SIZE);
 434                        if (prev_read>0) {
 435                                if (fwrite_sha1_file(prev_buf,
 436                                                     1,
 437                                                     prev_read,
 438                                                     request) == prev_read) {
 439                                        prev_posn += prev_read;
 440                                } else {
 441                                        prev_read = -1;
 442                                }
 443                        }
 444                } while (prev_read > 0);
 445                close(prevlocal);
 446        }
 447        unlink(prevfile);
 448
 449        /* Reset inflate/SHA1 if there was an error reading the previous temp
 450           file; also rewind to the beginning of the local file. */
 451        if (prev_read == -1) {
 452                memset(&request->stream, 0, sizeof(request->stream));
 453                inflateInit(&request->stream);
 454                SHA1_Init(&request->c);
 455                if (prev_posn>0) {
 456                        prev_posn = 0;
 457                        lseek(request->local, SEEK_SET, 0);
 458                        ftruncate(request->local, 0);
 459                }
 460        }
 461
 462        slot = get_active_slot();
 463        curl_easy_setopt(slot->curl, CURLOPT_FILE, request);
 464        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 465        curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, request->errorstr);
 466        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 467        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 468
 469        /* If we have successfully processed data from a previous fetch
 470           attempt, only fetch the data we don't already have. */
 471        if (prev_posn>0) {
 472                if (get_verbosely)
 473                        fprintf(stderr,
 474                                "Resuming fetch of object %s at byte %ld\n",
 475                                hex, prev_posn);
 476                sprintf(range, "Range: bytes=%ld-", prev_posn);
 477                range_header = curl_slist_append(range_header, range);
 478                curl_easy_setopt(slot->curl,
 479                                 CURLOPT_HTTPHEADER, range_header);
 480        }
 481
 482        /* Try to get the request started, abort the request on error */
 483        if (!start_active_slot(slot)) {
 484                request->state = ABORTED;
 485                close(request->local);
 486                free(request->url);
 487                return;
 488        }
 489        
 490        request->slot = slot;
 491        request->state = ACTIVE;
 492}
 493
 494static void finish_request(struct transfer_request *request)
 495{
 496        fchmod(request->local, 0444);
 497        close(request->local);
 498
 499        if (request->http_code == 416) {
 500                fprintf(stderr, "Warning: requested range invalid; we may already have all the data.\n");
 501        } else if (request->curl_result != CURLE_OK) {
 502                return;
 503        }
 504
 505        inflateEnd(&request->stream);
 506        SHA1_Final(request->real_sha1, &request->c);
 507        if (request->zret != Z_STREAM_END) {
 508                unlink(request->tmpfile);
 509                return;
 510        }
 511        if (memcmp(request->sha1, request->real_sha1, 20)) {
 512                unlink(request->tmpfile);
 513                return;
 514        }
 515        request->rename =
 516                move_temp_to_file(request->tmpfile, request->filename);
 517
 518        if (request->rename == 0)
 519                pull_say("got %s\n", sha1_to_hex(request->sha1));
 520}
 521
 522static void release_request(struct transfer_request *request)
 523{
 524        struct transfer_request *entry = request_queue_head;
 525
 526        if (request == request_queue_head) {
 527                request_queue_head = request->next;
 528        } else {
 529                while (entry->next != NULL && entry->next != request)
 530                        entry = entry->next;
 531                if (entry->next == request)
 532                        entry->next = entry->next->next;
 533        }
 534
 535        free(request->url);
 536        free(request);
 537}
 538
 539#ifdef USE_CURL_MULTI
 540void process_curl_messages(void)
 541{
 542        int num_messages;
 543        struct active_request_slot *slot;
 544        struct transfer_request *request = NULL;
 545        CURLMsg *curl_message = curl_multi_info_read(curlm, &num_messages);
 546
 547        while (curl_message != NULL) {
 548                if (curl_message->msg == CURLMSG_DONE) {
 549                        slot = active_queue_head;
 550                        while (slot != NULL &&
 551                               slot->curl != curl_message->easy_handle)
 552                                slot = slot->next;
 553                        if (slot != NULL) {
 554                                curl_multi_remove_handle(curlm, slot->curl);
 555                                active_requests--;
 556                                slot->done = 1;
 557                                slot->in_use = 0;
 558                                slot->curl_result = curl_message->data.result;
 559                                request = request_queue_head;
 560                                while (request != NULL &&
 561                                       request->slot != slot)
 562                                        request = request->next;
 563                        } else {
 564                                fprintf(stderr, "Received DONE message for unknown request!\n");
 565                        }
 566                        if (request != NULL) {
 567                                request->curl_result =
 568                                        curl_message->data.result;
 569                                curl_easy_getinfo(slot->curl,
 570                                                  CURLINFO_HTTP_CODE,
 571                                                  &request->http_code);
 572                                request->slot = NULL;
 573
 574                                /* Use alternates if necessary */
 575                                if (request->http_code == 404 &&
 576                                    request->repo->next != NULL) {
 577                                        request->repo = request->repo->next;
 578                                        start_request(request);
 579                                } else {
 580                                        finish_request(request);
 581                                        request->state = COMPLETE;
 582                                }
 583                        }
 584                } else {
 585                        fprintf(stderr, "Unknown CURL message received: %d\n",
 586                                (int)curl_message->msg);
 587                }
 588                curl_message = curl_multi_info_read(curlm, &num_messages);
 589        }
 590}
 591
 592void process_request_queue(void)
 593{
 594        struct transfer_request *request = request_queue_head;
 595        int num_transfers;
 596
 597        while (active_requests < max_requests && request != NULL) {
 598                if (request->state == WAITING) {
 599                        if (has_sha1_file(request->sha1))
 600                                release_request(request);
 601                        else
 602                                start_request(request);
 603                        curl_multi_perform(curlm, &num_transfers);
 604                }
 605                request = request->next;
 606        }
 607}
 608#endif
 609
 610void prefetch(unsigned char *sha1)
 611{
 612        struct transfer_request *newreq;
 613        struct transfer_request *tail;
 614        char *filename = sha1_file_name(sha1);
 615
 616        newreq = xmalloc(sizeof(*newreq));
 617        memcpy(newreq->sha1, sha1, 20);
 618        newreq->repo = alt;
 619        newreq->url = NULL;
 620        newreq->local = -1;
 621        newreq->state = WAITING;
 622        snprintf(newreq->filename, sizeof(newreq->filename), "%s", filename);
 623        snprintf(newreq->tmpfile, sizeof(newreq->tmpfile),
 624                 "%s.temp", filename);
 625        newreq->next = NULL;
 626
 627        if (request_queue_head == NULL) {
 628                request_queue_head = newreq;
 629        } else {
 630                tail = request_queue_head;
 631                while (tail->next != NULL) {
 632                        tail = tail->next;
 633                }
 634                tail->next = newreq;
 635        }
 636#ifdef USE_CURL_MULTI
 637        process_request_queue();
 638        process_curl_messages();
 639#endif
 640}
 641
 642static int fetch_index(struct alt_base *repo, unsigned char *sha1)
 643{
 644        char *hex = sha1_to_hex(sha1);
 645        char *filename;
 646        char *url;
 647        char tmpfile[PATH_MAX];
 648        long prev_posn = 0;
 649        char range[RANGE_HEADER_SIZE];
 650        struct curl_slist *range_header = NULL;
 651
 652        FILE *indexfile;
 653        struct active_request_slot *slot;
 654
 655        if (has_pack_index(sha1))
 656                return 0;
 657
 658        if (get_verbosely)
 659                fprintf(stderr, "Getting index for pack %s\n", hex);
 660        
 661        url = xmalloc(strlen(repo->base) + 64);
 662        sprintf(url, "%s/objects/pack/pack-%s.idx", repo->base, hex);
 663        
 664        filename = sha1_pack_index_name(sha1);
 665        snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 666        indexfile = fopen(tmpfile, "a");
 667        if (!indexfile)
 668                return error("Unable to open local file %s for pack index",
 669                             filename);
 670
 671        slot = get_active_slot();
 672        curl_easy_setopt(slot->curl, CURLOPT_FILE, indexfile);
 673        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 674        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 675        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 676        slot->local = indexfile;
 677
 678        /* If there is data present from a previous transfer attempt,
 679           resume where it left off */
 680        prev_posn = ftell(indexfile);
 681        if (prev_posn>0) {
 682                if (get_verbosely)
 683                        fprintf(stderr,
 684                                "Resuming fetch of index for pack %s at byte %ld\n",
 685                                hex, prev_posn);
 686                sprintf(range, "Range: bytes=%ld-", prev_posn);
 687                range_header = curl_slist_append(range_header, range);
 688                curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 689        }
 690
 691        if (start_active_slot(slot)) {
 692                run_active_slot(slot);
 693                if (slot->curl_result != CURLE_OK) {
 694                        fclose(indexfile);
 695                        return error("Unable to get pack index %s\n%s", url,
 696                                     curl_errorstr);
 697                }
 698        } else {
 699                return error("Unable to start request");
 700        }
 701
 702        fclose(indexfile);
 703
 704        return move_temp_to_file(tmpfile, filename);
 705}
 706
 707static int setup_index(struct alt_base *repo, unsigned char *sha1)
 708{
 709        struct packed_git *new_pack;
 710        if (has_pack_file(sha1))
 711                return 0; // don't list this as something we can get
 712
 713        if (fetch_index(repo, sha1))
 714                return -1;
 715
 716        new_pack = parse_pack_index(sha1);
 717        new_pack->next = repo->packs;
 718        repo->packs = new_pack;
 719        return 0;
 720}
 721
 722static int fetch_alternates(char *base)
 723{
 724        int ret = 0;
 725        struct buffer buffer;
 726        char *url;
 727        char *data;
 728        int i = 0;
 729        int http_specific = 1;
 730        struct alt_base *tail = alt;
 731        static const char null_byte = '\0';
 732
 733        struct active_request_slot *slot;
 734
 735        data = xmalloc(4096);
 736        buffer.size = 4096;
 737        buffer.posn = 0;
 738        buffer.buffer = data;
 739
 740        if (get_verbosely)
 741                fprintf(stderr, "Getting alternates list\n");
 742        
 743        url = xmalloc(strlen(base) + 31);
 744        sprintf(url, "%s/objects/info/http-alternates", base);
 745
 746        slot = get_active_slot();
 747        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 748        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 749                         fwrite_buffer_dynamic);
 750        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 751        if (start_active_slot(slot)) {
 752                run_active_slot(slot);
 753                if (slot->curl_result != CURLE_OK || !buffer.posn) {
 754                        http_specific = 0;
 755
 756                        sprintf(url, "%s/objects/info/alternates", base);
 757
 758                        slot = get_active_slot();
 759                        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 760                        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 761                                         fwrite_buffer_dynamic);
 762                        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 763                        if (start_active_slot(slot)) {
 764                                run_active_slot(slot);
 765                                if (slot->curl_result != CURLE_OK) {
 766                                        free(buffer.buffer);
 767                                        return 0;
 768                                }
 769                        }
 770                }
 771        } else {
 772                free(buffer.buffer);
 773                return 0;
 774        }
 775
 776        fwrite_buffer_dynamic(&null_byte, 1, 1, &buffer);
 777        buffer.posn--;
 778        data = buffer.buffer;
 779
 780        while (i < buffer.posn) {
 781                int posn = i;
 782                while (posn < buffer.posn && data[posn] != '\n')
 783                        posn++;
 784                if (data[posn] == '\n') {
 785                        int okay = 0;
 786                        int serverlen = 0;
 787                        struct alt_base *newalt;
 788                        char *target = NULL;
 789                        if (data[i] == '/') {
 790                                serverlen = strchr(base + 8, '/') - base;
 791                                okay = 1;
 792                        } else if (!memcmp(data + i, "../", 3)) {
 793                                i += 3;
 794                                serverlen = strlen(base);
 795                                while (i + 2 < posn && 
 796                                       !memcmp(data + i, "../", 3)) {
 797                                        do {
 798                                                serverlen--;
 799                                        } while (serverlen &&
 800                                                 base[serverlen - 1] != '/');
 801                                        i += 3;
 802                                }
 803                                // If the server got removed, give up.
 804                                okay = strchr(base, ':') - base + 3 < 
 805                                        serverlen;
 806                        } else if (http_specific) {
 807                                char *colon = strchr(data + i, ':');
 808                                char *slash = strchr(data + i, '/');
 809                                if (colon && slash && colon < data + posn &&
 810                                    slash < data + posn && colon < slash) {
 811                                        okay = 1;
 812                                }
 813                        }
 814                        // skip 'objects' at end
 815                        if (okay) {
 816                                target = xmalloc(serverlen + posn - i - 6);
 817                                strncpy(target, base, serverlen);
 818                                strncpy(target + serverlen, data + i,
 819                                        posn - i - 7);
 820                                target[serverlen + posn - i - 7] = '\0';
 821                                if (get_verbosely)
 822                                        fprintf(stderr, 
 823                                                "Also look at %s\n", target);
 824                                newalt = xmalloc(sizeof(*newalt));
 825                                newalt->next = NULL;
 826                                newalt->base = target;
 827                                newalt->got_indices = 0;
 828                                newalt->packs = NULL;
 829                                while (tail->next != NULL)
 830                                        tail = tail->next;
 831                                tail->next = newalt;
 832                                ret++;
 833                        }
 834                }
 835                i = posn + 1;
 836        }
 837
 838        free(buffer.buffer);
 839        return ret;
 840}
 841
 842static int fetch_indices(struct alt_base *repo)
 843{
 844        unsigned char sha1[20];
 845        char *url;
 846        struct buffer buffer;
 847        char *data;
 848        int i = 0;
 849
 850        struct active_request_slot *slot;
 851
 852        if (repo->got_indices)
 853                return 0;
 854
 855        data = xmalloc(4096);
 856        buffer.size = 4096;
 857        buffer.posn = 0;
 858        buffer.buffer = data;
 859
 860        if (get_verbosely)
 861                fprintf(stderr, "Getting pack list\n");
 862        
 863        url = xmalloc(strlen(repo->base) + 21);
 864        sprintf(url, "%s/objects/info/packs", repo->base);
 865
 866        slot = get_active_slot();
 867        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 868        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
 869                         fwrite_buffer_dynamic);
 870        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 871        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
 872        if (start_active_slot(slot)) {
 873                run_active_slot(slot);
 874                if (slot->curl_result != CURLE_OK) {
 875                        free(buffer.buffer);
 876                        return error("%s", curl_errorstr);
 877                }
 878        } else {
 879                free(buffer.buffer);
 880                return error("Unable to start request");
 881        }
 882
 883        data = buffer.buffer;
 884        while (i < buffer.posn) {
 885                switch (data[i]) {
 886                case 'P':
 887                        i++;
 888                        if (i + 52 < buffer.posn &&
 889                            !strncmp(data + i, " pack-", 6) &&
 890                            !strncmp(data + i + 46, ".pack\n", 6)) {
 891                                get_sha1_hex(data + i + 6, sha1);
 892                                setup_index(repo, sha1);
 893                                i += 51;
 894                                break;
 895                        }
 896                default:
 897                        while (data[i] != '\n')
 898                                i++;
 899                }
 900                i++;
 901        }
 902
 903        free(buffer.buffer);
 904        repo->got_indices = 1;
 905        return 0;
 906}
 907
 908static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 909{
 910        char *url;
 911        struct packed_git *target;
 912        struct packed_git **lst;
 913        FILE *packfile;
 914        char *filename;
 915        char tmpfile[PATH_MAX];
 916        int ret;
 917        long prev_posn = 0;
 918        char range[RANGE_HEADER_SIZE];
 919        struct curl_slist *range_header = NULL;
 920
 921        struct active_request_slot *slot;
 922
 923        if (fetch_indices(repo))
 924                return -1;
 925        target = find_sha1_pack(sha1, repo->packs);
 926        if (!target)
 927                return -1;
 928
 929        if (get_verbosely) {
 930                fprintf(stderr, "Getting pack %s\n",
 931                        sha1_to_hex(target->sha1));
 932                fprintf(stderr, " which contains %s\n",
 933                        sha1_to_hex(sha1));
 934        }
 935
 936        url = xmalloc(strlen(repo->base) + 65);
 937        sprintf(url, "%s/objects/pack/pack-%s.pack",
 938                repo->base, sha1_to_hex(target->sha1));
 939
 940        filename = sha1_pack_name(target->sha1);
 941        snprintf(tmpfile, sizeof(tmpfile), "%s.temp", filename);
 942        packfile = fopen(tmpfile, "a");
 943        if (!packfile)
 944                return error("Unable to open local file %s for pack",
 945                             filename);
 946
 947        slot = get_active_slot();
 948        curl_easy_setopt(slot->curl, CURLOPT_FILE, packfile);
 949        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
 950        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 951        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
 952        slot->local = packfile;
 953
 954        /* If there is data present from a previous transfer attempt,
 955           resume where it left off */
 956        prev_posn = ftell(packfile);
 957        if (prev_posn>0) {
 958                if (get_verbosely)
 959                        fprintf(stderr,
 960                                "Resuming fetch of pack %s at byte %ld\n",
 961                                sha1_to_hex(target->sha1), prev_posn);
 962                sprintf(range, "Range: bytes=%ld-", prev_posn);
 963                range_header = curl_slist_append(range_header, range);
 964                curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, range_header);
 965        }
 966
 967        if (start_active_slot(slot)) {
 968                run_active_slot(slot);
 969                if (slot->curl_result != CURLE_OK) {
 970                        fclose(packfile);
 971                        return error("Unable to get pack file %s\n%s", url,
 972                                     curl_errorstr);
 973                }
 974        } else {
 975                return error("Unable to start request");
 976        }
 977
 978        fclose(packfile);
 979
 980        ret = move_temp_to_file(tmpfile, filename);
 981        if (ret)
 982                return ret;
 983
 984        lst = &repo->packs;
 985        while (*lst != target)
 986                lst = &((*lst)->next);
 987        *lst = (*lst)->next;
 988
 989        if (verify_pack(target, 0))
 990                return -1;
 991        install_packed_git(target);
 992
 993        return 0;
 994}
 995
 996static int fetch_object(struct alt_base *repo, unsigned char *sha1)
 997{
 998        char *hex = sha1_to_hex(sha1);
 999        int ret;
1000        struct transfer_request *request = request_queue_head;
1001
1002        while (request != NULL && memcmp(request->sha1, sha1, 20))
1003                request = request->next;
1004        if (request == NULL)
1005                return error("Couldn't find request for %s in the queue", hex);
1006
1007        if (has_sha1_file(request->sha1)) {
1008                release_request(request);
1009                return 0;
1010        }
1011
1012#ifdef USE_CURL_MULTI
1013        while (request->state == WAITING) {
1014                int num_transfers;
1015                curl_multi_perform(curlm, &num_transfers);
1016                if (num_transfers < active_requests) {
1017                        process_curl_messages();
1018                        process_request_queue();
1019                }
1020        }
1021#else
1022        start_request(request);
1023#endif
1024
1025        while (request->state == ACTIVE) {
1026                run_active_slot(request->slot);
1027#ifndef USE_CURL_MULTI
1028                request->curl_result = request->slot->curl_result;
1029                curl_easy_getinfo(request->slot->curl,
1030                                  CURLINFO_HTTP_CODE,
1031                                  &request->http_code);
1032                request->slot = NULL;
1033
1034                /* Use alternates if necessary */
1035                if (request->http_code == 404 &&
1036                    request->repo->next != NULL) {
1037                        request->repo = request->repo->next;
1038                        start_request(request);
1039                } else {
1040                        finish_request(request);
1041                        request->state = COMPLETE;
1042                }
1043#endif
1044        }
1045
1046        if (request->state == ABORTED) {
1047                release_request(request);
1048                return error("Request for %s aborted", hex);
1049        }
1050
1051        if (request->curl_result != CURLE_OK && request->http_code != 416) {
1052                ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
1053                            request->errorstr, request->curl_result,
1054                            request->http_code, hex);
1055                release_request(request);
1056                return ret;
1057        }
1058
1059        if (request->zret != Z_STREAM_END) {
1060                ret = error("File %s (%s) corrupt\n", hex, request->url);
1061                release_request(request);
1062                return ret;
1063        }
1064
1065        if (memcmp(request->sha1, request->real_sha1, 20)) {
1066                release_request(request);
1067                return error("File %s has bad hash\n", hex);
1068        }
1069
1070        if (request->rename < 0) {
1071                ret = error("unable to write sha1 filename %s: %s",
1072                            request->filename,
1073                            strerror(request->rename));
1074                release_request(request);
1075                return ret;
1076        }
1077
1078        release_request(request);
1079        return 0;
1080}
1081
1082int fetch(unsigned char *sha1)
1083{
1084        struct alt_base *altbase = alt;
1085
1086        if (!fetch_object(altbase, sha1))
1087                return 0;
1088        while (altbase) {
1089                if (!fetch_pack(altbase, sha1))
1090                        return 0;
1091                altbase = altbase->next;
1092        }
1093        return error("Unable to find %s under %s\n", sha1_to_hex(sha1), 
1094                     alt->base);
1095}
1096
1097static inline int needs_quote(int ch)
1098{
1099        switch (ch) {
1100        case '/': case '-': case '.':
1101        case 'A'...'Z': case 'a'...'z': case '0'...'9':
1102                return 0;
1103        default:
1104                return 1;
1105        }
1106}
1107
1108static inline int hex(int v)
1109{
1110        if (v < 10) return '0' + v;
1111        else return 'A' + v - 10;
1112}
1113
1114static char *quote_ref_url(const char *base, const char *ref)
1115{
1116        const char *cp;
1117        char *dp, *qref;
1118        int len, baselen, ch;
1119
1120        baselen = strlen(base);
1121        len = baselen + 6; /* "refs/" + NUL */
1122        for (cp = ref; (ch = *cp) != 0; cp++, len++)
1123                if (needs_quote(ch))
1124                        len += 2; /* extra two hex plus replacement % */
1125        qref = xmalloc(len);
1126        memcpy(qref, base, baselen);
1127        memcpy(qref + baselen, "refs/", 5);
1128        for (cp = ref, dp = qref + baselen + 5; (ch = *cp) != 0; cp++) {
1129                if (needs_quote(ch)) {
1130                        *dp++ = '%';
1131                        *dp++ = hex((ch >> 4) & 0xF);
1132                        *dp++ = hex(ch & 0xF);
1133                }
1134                else
1135                        *dp++ = ch;
1136        }
1137        *dp = 0;
1138
1139        return qref;
1140}
1141
1142int fetch_ref(char *ref, unsigned char *sha1)
1143{
1144        char *url;
1145        char hex[42];
1146        struct buffer buffer;
1147        char *base = alt->base;
1148        struct active_request_slot *slot;
1149        buffer.size = 41;
1150        buffer.posn = 0;
1151        buffer.buffer = hex;
1152        hex[41] = '\0';
1153        
1154        url = quote_ref_url(base, ref);
1155        slot = get_active_slot();
1156        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
1157        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
1158        curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, NULL);
1159        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
1160        if (start_active_slot(slot)) {
1161                run_active_slot(slot);
1162                if (slot->curl_result != CURLE_OK)
1163                        return error("Couldn't get %s for %s\n%s",
1164                                     url, ref, curl_errorstr);
1165        } else {
1166                return error("Unable to start request");
1167        }
1168
1169        hex[40] = '\0';
1170        get_sha1_hex(hex, sha1);
1171        return 0;
1172}
1173
1174int main(int argc, char **argv)
1175{
1176        char *commit_id;
1177        char *url;
1178        int arg = 1;
1179        struct active_request_slot *slot;
1180
1181        while (arg < argc && argv[arg][0] == '-') {
1182                if (argv[arg][1] == 't') {
1183                        get_tree = 1;
1184                } else if (argv[arg][1] == 'c') {
1185                        get_history = 1;
1186                } else if (argv[arg][1] == 'a') {
1187                        get_all = 1;
1188                        get_tree = 1;
1189                        get_history = 1;
1190                } else if (argv[arg][1] == 'v') {
1191                        get_verbosely = 1;
1192                } else if (argv[arg][1] == 'w') {
1193                        write_ref = argv[arg + 1];
1194                        arg++;
1195                } else if (!strcmp(argv[arg], "--recover")) {
1196                        get_recover = 1;
1197                }
1198                arg++;
1199        }
1200        if (argc < arg + 2) {
1201                usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
1202                return 1;
1203        }
1204        commit_id = argv[arg];
1205        url = argv[arg + 1];
1206
1207        curl_global_init(CURL_GLOBAL_ALL);
1208
1209#ifdef USE_CURL_MULTI
1210        {
1211                char *http_max_requests = getenv("GIT_HTTP_MAX_REQUESTS");
1212                if (http_max_requests != NULL)
1213                        max_requests = atoi(http_max_requests);
1214        }
1215
1216        curlm = curl_multi_init();
1217        if (curlm == NULL) {
1218                fprintf(stderr, "Error creating curl multi handle.\n");
1219                return 1;
1220        }
1221#endif
1222
1223        if (getenv("GIT_SSL_NO_VERIFY"))
1224                curl_ssl_verify = 0;
1225
1226        ssl_cert = getenv("GIT_SSL_CERT");
1227#if LIBCURL_VERSION_NUM >= 0x070902
1228        ssl_key = getenv("GIT_SSL_KEY");
1229#endif
1230#if LIBCURL_VERSION_NUM >= 0x070908
1231        ssl_capath = getenv("GIT_SSL_CAPATH");
1232#endif
1233        ssl_cainfo = getenv("GIT_SSL_CAINFO");
1234
1235        git_config(http_options);
1236
1237        if (curl_ssl_verify == -1)
1238                curl_ssl_verify = 1;
1239
1240#ifdef USE_CURL_MULTI
1241        if (max_requests < 1)
1242                max_requests = DEFAULT_MAX_REQUESTS;
1243#endif
1244
1245        pragma_header = curl_slist_append(pragma_header, "Pragma: no-cache");
1246        no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
1247        no_range_header = curl_slist_append(no_range_header, "Range:");
1248
1249#ifndef NO_CURL_EASY_DUPHANDLE
1250        curl_default = get_curl_handle();
1251#endif
1252
1253        alt = xmalloc(sizeof(*alt));
1254        alt->base = url;
1255        alt->got_indices = 0;
1256        alt->packs = NULL;
1257        alt->next = NULL;
1258        fetch_alternates(alt->base);
1259
1260        if (pull(commit_id))
1261                return 1;
1262
1263        curl_slist_free_all(pragma_header);
1264        curl_slist_free_all(no_pragma_header);
1265        curl_slist_free_all(no_range_header);
1266#ifndef NO_CURL_EASY_DUPHANDLE
1267        curl_easy_cleanup(curl_default);
1268#endif
1269        slot = active_queue_head;
1270        while (slot != NULL) {
1271                curl_easy_cleanup(slot->curl);
1272                slot = slot->next;
1273        }
1274#ifdef USE_CURL_MULTI
1275        curl_multi_cleanup(curlm);
1276#endif
1277        curl_global_cleanup();
1278        return 0;
1279}