Andrew's git - gitweb.git/blob

77f530c95d64193aa3b00dacd651aa0fa17c36e8
   1#include "cache.h"
   2#include "commit.h"
   3
   4#include "fetch.h"
   5
   6#include <curl/curl.h>
   7#include <curl/easy.h>
   8
   9#if LIBCURL_VERSION_NUM < 0x070704
  10#define curl_global_cleanup() do { /* nothing */ } while(0)
  11#endif
  12#if LIBCURL_VERSION_NUM < 0x070800
  13#define curl_global_init(a) do { /* nothing */ } while(0)
  14#endif
  15
  16static CURL *curl;
  17static struct curl_slist *no_pragma_header;
  18
  19static char *initial_base;
  20
  21struct alt_base
  22{
  23        char *base;
  24        int got_indices;
  25        struct packed_git *packs;
  26        struct alt_base *next;
  27};
  28
  29struct alt_base *alt = NULL;
  30
  31static SHA_CTX c;
  32static z_stream stream;
  33
  34static int local;
  35static int zret;
  36
  37static int curl_ssl_verify;
  38
  39struct buffer
  40{
  41        size_t posn;
  42        size_t size;
  43        void *buffer;
  44};
  45
  46static size_t fwrite_buffer(void *ptr, size_t eltsize, size_t nmemb,
  47                            struct buffer *buffer)
  48{
  49        size_t size = eltsize * nmemb;
  50        if (size > buffer->size - buffer->posn)
  51                size = buffer->size - buffer->posn;
  52        memcpy(buffer->buffer + buffer->posn, ptr, size);
  53        buffer->posn += size;
  54        return size;
  55}
  56
  57static size_t fwrite_sha1_file(void *ptr, size_t eltsize, size_t nmemb,
  58                               void *data)
  59{
  60        unsigned char expn[4096];
  61        size_t size = eltsize * nmemb;
  62        int posn = 0;
  63        do {
  64                ssize_t retval = write(local, ptr + posn, size - posn);
  65                if (retval < 0)
  66                        return posn;
  67                posn += retval;
  68        } while (posn < size);
  69
  70        stream.avail_in = size;
  71        stream.next_in = ptr;
  72        do {
  73                stream.next_out = expn;
  74                stream.avail_out = sizeof(expn);
  75                zret = inflate(&stream, Z_SYNC_FLUSH);
  76                SHA1_Update(&c, expn, sizeof(expn) - stream.avail_out);
  77        } while (stream.avail_in && zret == Z_OK);
  78        return size;
  79}
  80
  81void prefetch(unsigned char *sha1)
  82{
  83}
  84
  85static int got_alternates = 0;
  86
  87static int fetch_index(struct alt_base *repo, unsigned char *sha1)
  88{
  89        char *filename;
  90        char *url;
  91
  92        FILE *indexfile;
  93
  94        if (has_pack_index(sha1))
  95                return 0;
  96
  97        if (get_verbosely)
  98                fprintf(stderr, "Getting index for pack %s\n",
  99                        sha1_to_hex(sha1));
 100        
 101        url = xmalloc(strlen(repo->base) + 64);
 102        sprintf(url, "%s/objects/pack/pack-%s.idx",
 103                repo->base, sha1_to_hex(sha1));
 104        
 105        filename = sha1_pack_index_name(sha1);
 106        indexfile = fopen(filename, "w");
 107        if (!indexfile)
 108                return error("Unable to open local file %s for pack index",
 109                             filename);
 110
 111        curl_easy_setopt(curl, CURLOPT_FILE, indexfile);
 112        curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite);
 113        curl_easy_setopt(curl, CURLOPT_URL, url);
 114        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 115        
 116        if (curl_easy_perform(curl)) {
 117                fclose(indexfile);
 118                return error("Unable to get pack index %s", url);
 119        }
 120
 121        fclose(indexfile);
 122        return 0;
 123}
 124
 125static int setup_index(struct alt_base *repo, unsigned char *sha1)
 126{
 127        struct packed_git *new_pack;
 128        if (has_pack_file(sha1))
 129                return 0; // don't list this as something we can get
 130
 131        if (fetch_index(repo, sha1))
 132                return -1;
 133
 134        new_pack = parse_pack_index(sha1);
 135        new_pack->next = repo->packs;
 136        repo->packs = new_pack;
 137        return 0;
 138}
 139
 140static int fetch_alternates(char *base)
 141{
 142        int ret = 0;
 143        struct buffer buffer;
 144        char *url;
 145        char *data;
 146        int i = 0;
 147        int http_specific = 1;
 148        if (got_alternates)
 149                return 0;
 150        data = xmalloc(4096);
 151        buffer.size = 4095;
 152        buffer.posn = 0;
 153        buffer.buffer = data;
 154
 155        if (get_verbosely)
 156                fprintf(stderr, "Getting alternates list\n");
 157        
 158        url = xmalloc(strlen(base) + 31);
 159        sprintf(url, "%s/objects/info/http-alternates", base);
 160
 161        curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 162        curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 163        curl_easy_setopt(curl, CURLOPT_URL, url);
 164
 165        if (curl_easy_perform(curl) || !buffer.posn) {
 166                http_specific = 0;
 167
 168                sprintf(url, "%s/objects/info/alternates", base);
 169                
 170                curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 171                curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 172                curl_easy_setopt(curl, CURLOPT_URL, url);
 173                
 174                if (curl_easy_perform(curl)) {
 175                        return 0;
 176                }
 177        }
 178
 179        data[buffer.posn] = '\0';
 180
 181        while (i < buffer.posn) {
 182                int posn = i;
 183                while (posn < buffer.posn && data[posn] != '\n')
 184                        posn++;
 185                if (data[posn] == '\n') {
 186                        int okay = 0;
 187                        int serverlen = 0;
 188                        struct alt_base *newalt;
 189                        char *target = NULL;
 190                        if (data[i] == '/') {
 191                                serverlen = strchr(base + 8, '/') - base;
 192                                okay = 1;
 193                        } else if (!memcmp(data + i, "../", 3)) {
 194                                i += 3;
 195                                serverlen = strlen(base);
 196                                while (i + 2 < posn && 
 197                                       !memcmp(data + i, "../", 3)) {
 198                                        do {
 199                                                serverlen--;
 200                                        } while (serverlen &&
 201                                                 base[serverlen - 1] != '/');
 202                                        i += 3;
 203                                }
 204                                // If the server got removed, give up.
 205                                okay = strchr(base, ':') - base + 3 < 
 206                                        serverlen;
 207                        } else if (http_specific) {
 208                                char *colon = strchr(data + i, ':');
 209                                char *slash = strchr(data + i, '/');
 210                                if (colon && slash && colon < data + posn &&
 211                                    slash < data + posn && colon < slash) {
 212                                        okay = 1;
 213                                }
 214                        }
 215                        // skip 'objects' at end
 216                        if (okay) {
 217                                target = xmalloc(serverlen + posn - i - 6);
 218                                strncpy(target, base, serverlen);
 219                                strncpy(target + serverlen, data + i,
 220                                        posn - i - 7);
 221                                target[serverlen + posn - i - 7] = '\0';
 222                                if (get_verbosely)
 223                                        fprintf(stderr, 
 224                                                "Also look at %s\n", target);
 225                                newalt = xmalloc(sizeof(*newalt));
 226                                newalt->next = alt;
 227                                newalt->base = target;
 228                                newalt->got_indices = 0;
 229                                newalt->packs = NULL;
 230                                alt = newalt;
 231                                ret++;
 232                        }
 233                }
 234                i = posn + 1;
 235        }
 236        got_alternates = 1;
 237        
 238        return ret;
 239}
 240
 241static int fetch_indices(struct alt_base *repo)
 242{
 243        unsigned char sha1[20];
 244        char *url;
 245        struct buffer buffer;
 246        char *data;
 247        int i = 0;
 248
 249        if (repo->got_indices)
 250                return 0;
 251
 252        data = xmalloc(4096);
 253        buffer.size = 4096;
 254        buffer.posn = 0;
 255        buffer.buffer = data;
 256
 257        if (get_verbosely)
 258                fprintf(stderr, "Getting pack list\n");
 259        
 260        url = xmalloc(strlen(repo->base) + 21);
 261        sprintf(url, "%s/objects/info/packs", repo->base);
 262
 263        curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 264        curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 265        curl_easy_setopt(curl, CURLOPT_URL, url);
 266        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
 267        
 268        if (curl_easy_perform(curl)) {
 269                return -1;
 270        }
 271
 272        while (i < buffer.posn) {
 273                switch (data[i]) {
 274                case 'P':
 275                        i++;
 276                        if (i + 52 < buffer.posn &&
 277                            !strncmp(data + i, " pack-", 6) &&
 278                            !strncmp(data + i + 46, ".pack\n", 6)) {
 279                                get_sha1_hex(data + i + 6, sha1);
 280                                setup_index(repo, sha1);
 281                                i += 51;
 282                                break;
 283                        }
 284                default:
 285                        while (data[i] != '\n')
 286                                i++;
 287                }
 288                i++;
 289        }
 290
 291        repo->got_indices = 1;
 292        return 0;
 293}
 294
 295static int fetch_pack(struct alt_base *repo, unsigned char *sha1)
 296{
 297        char *url;
 298        struct packed_git *target;
 299        struct packed_git **lst;
 300        FILE *packfile;
 301        char *filename;
 302
 303        if (fetch_indices(repo))
 304                return -1;
 305        target = find_sha1_pack(sha1, repo->packs);
 306        if (!target)
 307                return -1;
 308
 309        if (get_verbosely) {
 310                fprintf(stderr, "Getting pack %s\n",
 311                        sha1_to_hex(target->sha1));
 312                fprintf(stderr, " which contains %s\n",
 313                        sha1_to_hex(sha1));
 314        }
 315
 316        url = xmalloc(strlen(repo->base) + 65);
 317        sprintf(url, "%s/objects/pack/pack-%s.pack",
 318                repo->base, sha1_to_hex(target->sha1));
 319
 320        filename = sha1_pack_name(target->sha1);
 321        packfile = fopen(filename, "w");
 322        if (!packfile)
 323                return error("Unable to open local file %s for pack",
 324                             filename);
 325
 326        curl_easy_setopt(curl, CURLOPT_FILE, packfile);
 327        curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite);
 328        curl_easy_setopt(curl, CURLOPT_URL, url);
 329        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 330        
 331        if (curl_easy_perform(curl)) {
 332                fclose(packfile);
 333                return error("Unable to get pack file %s", url);
 334        }
 335
 336        fclose(packfile);
 337
 338        lst = &repo->packs;
 339        while (*lst != target)
 340                lst = &((*lst)->next);
 341        *lst = (*lst)->next;
 342
 343        install_packed_git(target);
 344
 345        return 0;
 346}
 347
 348int fetch_object(struct alt_base *repo, unsigned char *sha1)
 349{
 350        char *hex = sha1_to_hex(sha1);
 351        char *filename = sha1_file_name(sha1);
 352        unsigned char real_sha1[20];
 353        char *url;
 354        char *posn;
 355
 356        local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
 357
 358        if (local < 0)
 359                return error("Couldn't open local object %s\n", filename);
 360
 361        memset(&stream, 0, sizeof(stream));
 362
 363        inflateInit(&stream);
 364
 365        SHA1_Init(&c);
 366
 367        curl_easy_setopt(curl, CURLOPT_FAILONERROR, 1);
 368        curl_easy_setopt(curl, CURLOPT_FILE, NULL);
 369        curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
 370        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, no_pragma_header);
 371
 372        url = xmalloc(strlen(repo->base) + 50);
 373        strcpy(url, repo->base);
 374        posn = url + strlen(repo->base);
 375        strcpy(posn, "objects/");
 376        posn += 8;
 377        memcpy(posn, hex, 2);
 378        posn += 2;
 379        *(posn++) = '/';
 380        strcpy(posn, hex + 2);
 381
 382        curl_easy_setopt(curl, CURLOPT_URL, url);
 383
 384        if (curl_easy_perform(curl)) {
 385                unlink(filename);
 386                return -1;
 387        }
 388
 389        close(local);
 390        inflateEnd(&stream);
 391        SHA1_Final(real_sha1, &c);
 392        if (zret != Z_STREAM_END) {
 393                unlink(filename);
 394                return error("File %s (%s) corrupt\n", hex, url);
 395        }
 396        if (memcmp(sha1, real_sha1, 20)) {
 397                unlink(filename);
 398                return error("File %s has bad hash\n", hex);
 399        }
 400        
 401        pull_say("got %s\n", hex);
 402        return 0;
 403}
 404
 405int fetch(unsigned char *sha1)
 406{
 407        struct alt_base *altbase = alt;
 408        while (altbase) {
 409                if (!fetch_object(altbase, sha1))
 410                        return 0;
 411                if (!fetch_pack(altbase, sha1))
 412                        return 0;
 413                if (fetch_alternates(altbase->base) > 0) {
 414                        altbase = alt;
 415                        continue;
 416                }
 417                altbase = altbase->next;
 418        }
 419        return error("Unable to find %s under %s\n", sha1_to_hex(sha1), 
 420                     initial_base);
 421}
 422
 423int fetch_ref(char *ref, unsigned char *sha1)
 424{
 425        char *url, *posn;
 426        char hex[42];
 427        struct buffer buffer;
 428        char *base = initial_base;
 429        buffer.size = 41;
 430        buffer.posn = 0;
 431        buffer.buffer = hex;
 432        hex[41] = '\0';
 433        
 434        curl_easy_setopt(curl, CURLOPT_FILE, &buffer);
 435        curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 436        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
 437
 438        url = xmalloc(strlen(base) + 6 + strlen(ref));
 439        strcpy(url, base);
 440        posn = url + strlen(base);
 441        strcpy(posn, "refs/");
 442        posn += 5;
 443        strcpy(posn, ref);
 444
 445        curl_easy_setopt(curl, CURLOPT_URL, url);
 446
 447        if (curl_easy_perform(curl))
 448                return error("Couldn't get %s for %s\n", url, ref);
 449
 450        hex[40] = '\0';
 451        get_sha1_hex(hex, sha1);
 452        return 0;
 453}
 454
 455int main(int argc, char **argv)
 456{
 457        char *commit_id;
 458        char *url;
 459        int arg = 1;
 460
 461        while (arg < argc && argv[arg][0] == '-') {
 462                if (argv[arg][1] == 't') {
 463                        get_tree = 1;
 464                } else if (argv[arg][1] == 'c') {
 465                        get_history = 1;
 466                } else if (argv[arg][1] == 'a') {
 467                        get_all = 1;
 468                        get_tree = 1;
 469                        get_history = 1;
 470                } else if (argv[arg][1] == 'v') {
 471                        get_verbosely = 1;
 472                } else if (argv[arg][1] == 'w') {
 473                        write_ref = argv[arg + 1];
 474                        arg++;
 475                }
 476                arg++;
 477        }
 478        if (argc < arg + 2) {
 479                usage("git-http-fetch [-c] [-t] [-a] [-d] [-v] [--recover] [-w ref] commit-id url");
 480                return 1;
 481        }
 482        commit_id = argv[arg];
 483        url = argv[arg + 1];
 484
 485        curl_global_init(CURL_GLOBAL_ALL);
 486
 487        curl = curl_easy_init();
 488        no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
 489
 490        curl_ssl_verify = getenv("GIT_SSL_NO_VERIFY") ? 0 : 1;
 491        curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, curl_ssl_verify);
 492#if LIBCURL_VERSION_NUM >= 0x070907
 493        curl_easy_setopt(curl, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
 494#endif
 495
 496        alt = xmalloc(sizeof(*alt));
 497        alt->base = url;
 498        alt->got_indices = 0;
 499        alt->packs = NULL;
 500        alt->next = NULL;
 501        initial_base = url;
 502
 503        if (pull(commit_id))
 504                return 1;
 505
 506        curl_slist_free_all(no_pragma_header);
 507        curl_global_cleanup();
 508        return 0;
 509}