http-walker.con commit Merge branch 'ss/clone-guess-dir-name-simplify' into maint (de67af4)
   1#include "cache.h"
   2#include "commit.h"
   3#include "walker.h"
   4#include "http.h"
   5
   6struct alt_base {
   7        char *base;
   8        int got_indices;
   9        struct packed_git *packs;
  10        struct alt_base *next;
  11};
  12
  13enum object_request_state {
  14        WAITING,
  15        ABORTED,
  16        ACTIVE,
  17        COMPLETE
  18};
  19
  20struct object_request {
  21        struct walker *walker;
  22        unsigned char sha1[20];
  23        struct alt_base *repo;
  24        enum object_request_state state;
  25        struct http_object_request *req;
  26        struct object_request *next;
  27};
  28
  29struct alternates_request {
  30        struct walker *walker;
  31        const char *base;
  32        char *url;
  33        struct strbuf *buffer;
  34        struct active_request_slot *slot;
  35        int http_specific;
  36};
  37
  38struct walker_data {
  39        const char *url;
  40        int got_alternates;
  41        struct alt_base *alt;
  42};
  43
  44static struct object_request *object_queue_head;
  45
  46static void fetch_alternates(struct walker *walker, const char *base);
  47
  48static void process_object_response(void *callback_data);
  49
  50static void start_object_request(struct walker *walker,
  51                                 struct object_request *obj_req)
  52{
  53        struct active_request_slot *slot;
  54        struct http_object_request *req;
  55
  56        req = new_http_object_request(obj_req->repo->base, obj_req->sha1);
  57        if (req == NULL) {
  58                obj_req->state = ABORTED;
  59                return;
  60        }
  61        obj_req->req = req;
  62
  63        slot = req->slot;
  64        slot->callback_func = process_object_response;
  65        slot->callback_data = obj_req;
  66
  67        /* Try to get the request started, abort the request on error */
  68        obj_req->state = ACTIVE;
  69        if (!start_active_slot(slot)) {
  70                obj_req->state = ABORTED;
  71                release_http_object_request(req);
  72                return;
  73        }
  74}
  75
  76static void finish_object_request(struct object_request *obj_req)
  77{
  78        if (finish_http_object_request(obj_req->req))
  79                return;
  80
  81        if (obj_req->req->rename == 0)
  82                walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
  83}
  84
  85static void process_object_response(void *callback_data)
  86{
  87        struct object_request *obj_req =
  88                (struct object_request *)callback_data;
  89        struct walker *walker = obj_req->walker;
  90        struct walker_data *data = walker->data;
  91        struct alt_base *alt = data->alt;
  92
  93        process_http_object_request(obj_req->req);
  94        obj_req->state = COMPLETE;
  95
  96        /* Use alternates if necessary */
  97        if (missing_target(obj_req->req)) {
  98                fetch_alternates(walker, alt->base);
  99                if (obj_req->repo->next != NULL) {
 100                        obj_req->repo =
 101                                obj_req->repo->next;
 102                        release_http_object_request(obj_req->req);
 103                        start_object_request(walker, obj_req);
 104                        return;
 105                }
 106        }
 107
 108        finish_object_request(obj_req);
 109}
 110
 111static void release_object_request(struct object_request *obj_req)
 112{
 113        struct object_request *entry = object_queue_head;
 114
 115        if (obj_req->req !=NULL && obj_req->req->localfile != -1)
 116                error("fd leakage in release: %d", obj_req->req->localfile);
 117        if (obj_req == object_queue_head) {
 118                object_queue_head = obj_req->next;
 119        } else {
 120                while (entry->next != NULL && entry->next != obj_req)
 121                        entry = entry->next;
 122                if (entry->next == obj_req)
 123                        entry->next = entry->next->next;
 124        }
 125
 126        free(obj_req);
 127}
 128
 129#ifdef USE_CURL_MULTI
 130static int fill_active_slot(struct walker *walker)
 131{
 132        struct object_request *obj_req;
 133
 134        for (obj_req = object_queue_head; obj_req; obj_req = obj_req->next) {
 135                if (obj_req->state == WAITING) {
 136                        if (has_sha1_file(obj_req->sha1))
 137                                obj_req->state = COMPLETE;
 138                        else {
 139                                start_object_request(walker, obj_req);
 140                                return 1;
 141                        }
 142                }
 143        }
 144        return 0;
 145}
 146#endif
 147
 148static void prefetch(struct walker *walker, unsigned char *sha1)
 149{
 150        struct object_request *newreq;
 151        struct object_request *tail;
 152        struct walker_data *data = walker->data;
 153
 154        newreq = xmalloc(sizeof(*newreq));
 155        newreq->walker = walker;
 156        hashcpy(newreq->sha1, sha1);
 157        newreq->repo = data->alt;
 158        newreq->state = WAITING;
 159        newreq->req = NULL;
 160        newreq->next = NULL;
 161
 162        http_is_verbose = walker->get_verbosely;
 163
 164        if (object_queue_head == NULL) {
 165                object_queue_head = newreq;
 166        } else {
 167                tail = object_queue_head;
 168                while (tail->next != NULL)
 169                        tail = tail->next;
 170                tail->next = newreq;
 171        }
 172
 173#ifdef USE_CURL_MULTI
 174        fill_active_slots();
 175        step_active_slots();
 176#endif
 177}
 178
 179static void process_alternates_response(void *callback_data)
 180{
 181        struct alternates_request *alt_req =
 182                (struct alternates_request *)callback_data;
 183        struct walker *walker = alt_req->walker;
 184        struct walker_data *cdata = walker->data;
 185        struct active_request_slot *slot = alt_req->slot;
 186        struct alt_base *tail = cdata->alt;
 187        const char *base = alt_req->base;
 188        const char null_byte = '\0';
 189        char *data;
 190        int i = 0;
 191
 192        if (alt_req->http_specific) {
 193                if (slot->curl_result != CURLE_OK ||
 194                    !alt_req->buffer->len) {
 195
 196                        /* Try reusing the slot to get non-http alternates */
 197                        alt_req->http_specific = 0;
 198                        sprintf(alt_req->url, "%s/objects/info/alternates",
 199                                base);
 200                        curl_easy_setopt(slot->curl, CURLOPT_URL,
 201                                         alt_req->url);
 202                        active_requests++;
 203                        slot->in_use = 1;
 204                        if (slot->finished != NULL)
 205                                (*slot->finished) = 0;
 206                        if (!start_active_slot(slot)) {
 207                                cdata->got_alternates = -1;
 208                                slot->in_use = 0;
 209                                if (slot->finished != NULL)
 210                                        (*slot->finished) = 1;
 211                        }
 212                        return;
 213                }
 214        } else if (slot->curl_result != CURLE_OK) {
 215                if (!missing_target(slot)) {
 216                        cdata->got_alternates = -1;
 217                        return;
 218                }
 219        }
 220
 221        fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
 222        alt_req->buffer->len--;
 223        data = alt_req->buffer->buf;
 224
 225        while (i < alt_req->buffer->len) {
 226                int posn = i;
 227                while (posn < alt_req->buffer->len && data[posn] != '\n')
 228                        posn++;
 229                if (data[posn] == '\n') {
 230                        int okay = 0;
 231                        int serverlen = 0;
 232                        struct alt_base *newalt;
 233                        if (data[i] == '/') {
 234                                /*
 235                                 * This counts
 236                                 * http://git.host/pub/scm/linux.git/
 237                                 * -----------here^
 238                                 * so memcpy(dst, base, serverlen) will
 239                                 * copy up to "...git.host".
 240                                 */
 241                                const char *colon_ss = strstr(base,"://");
 242                                if (colon_ss) {
 243                                        serverlen = (strchr(colon_ss + 3, '/')
 244                                                     - base);
 245                                        okay = 1;
 246                                }
 247                        } else if (!memcmp(data + i, "../", 3)) {
 248                                /*
 249                                 * Relative URL; chop the corresponding
 250                                 * number of subpath from base (and ../
 251                                 * from data), and concatenate the result.
 252                                 *
 253                                 * The code first drops ../ from data, and
 254                                 * then drops one ../ from data and one path
 255                                 * from base.  IOW, one extra ../ is dropped
 256                                 * from data than path is dropped from base.
 257                                 *
 258                                 * This is not wrong.  The alternate in
 259                                 *     http://git.host/pub/scm/linux.git/
 260                                 * to borrow from
 261                                 *     http://git.host/pub/scm/linus.git/
 262                                 * is ../../linus.git/objects/.  You need
 263                                 * two ../../ to borrow from your direct
 264                                 * neighbour.
 265                                 */
 266                                i += 3;
 267                                serverlen = strlen(base);
 268                                while (i + 2 < posn &&
 269                                       !memcmp(data + i, "../", 3)) {
 270                                        do {
 271                                                serverlen--;
 272                                        } while (serverlen &&
 273                                                 base[serverlen - 1] != '/');
 274                                        i += 3;
 275                                }
 276                                /* If the server got removed, give up. */
 277                                okay = strchr(base, ':') - base + 3 <
 278                                       serverlen;
 279                        } else if (alt_req->http_specific) {
 280                                char *colon = strchr(data + i, ':');
 281                                char *slash = strchr(data + i, '/');
 282                                if (colon && slash && colon < data + posn &&
 283                                    slash < data + posn && colon < slash) {
 284                                        okay = 1;
 285                                }
 286                        }
 287                        /* skip "objects\n" at end */
 288                        if (okay) {
 289                                struct strbuf target = STRBUF_INIT;
 290                                strbuf_add(&target, base, serverlen);
 291                                strbuf_add(&target, data + i, posn - i - 7);
 292                                if (walker->get_verbosely)
 293                                        fprintf(stderr, "Also look at %s\n",
 294                                                target.buf);
 295                                newalt = xmalloc(sizeof(*newalt));
 296                                newalt->next = NULL;
 297                                newalt->base = strbuf_detach(&target, NULL);
 298                                newalt->got_indices = 0;
 299                                newalt->packs = NULL;
 300
 301                                while (tail->next != NULL)
 302                                        tail = tail->next;
 303                                tail->next = newalt;
 304                        }
 305                }
 306                i = posn + 1;
 307        }
 308
 309        cdata->got_alternates = 1;
 310}
 311
 312static void fetch_alternates(struct walker *walker, const char *base)
 313{
 314        struct strbuf buffer = STRBUF_INIT;
 315        char *url;
 316        struct active_request_slot *slot;
 317        struct alternates_request alt_req;
 318        struct walker_data *cdata = walker->data;
 319
 320        /*
 321         * If another request has already started fetching alternates,
 322         * wait for them to arrive and return to processing this request's
 323         * curl message
 324         */
 325#ifdef USE_CURL_MULTI
 326        while (cdata->got_alternates == 0) {
 327                step_active_slots();
 328        }
 329#endif
 330
 331        /* Nothing to do if they've already been fetched */
 332        if (cdata->got_alternates == 1)
 333                return;
 334
 335        /* Start the fetch */
 336        cdata->got_alternates = 0;
 337
 338        if (walker->get_verbosely)
 339                fprintf(stderr, "Getting alternates list for %s\n", base);
 340
 341        url = xstrfmt("%s/objects/info/http-alternates", base);
 342
 343        /*
 344         * Use a callback to process the result, since another request
 345         * may fail and need to have alternates loaded before continuing
 346         */
 347        slot = get_active_slot();
 348        slot->callback_func = process_alternates_response;
 349        alt_req.walker = walker;
 350        slot->callback_data = &alt_req;
 351
 352        curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 353        curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 354        curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 355
 356        alt_req.base = base;
 357        alt_req.url = url;
 358        alt_req.buffer = &buffer;
 359        alt_req.http_specific = 1;
 360        alt_req.slot = slot;
 361
 362        if (start_active_slot(slot))
 363                run_active_slot(slot);
 364        else
 365                cdata->got_alternates = -1;
 366
 367        strbuf_release(&buffer);
 368        free(url);
 369}
 370
 371static int fetch_indices(struct walker *walker, struct alt_base *repo)
 372{
 373        int ret;
 374
 375        if (repo->got_indices)
 376                return 0;
 377
 378        if (walker->get_verbosely)
 379                fprintf(stderr, "Getting pack list for %s\n", repo->base);
 380
 381        switch (http_get_info_packs(repo->base, &repo->packs)) {
 382        case HTTP_OK:
 383        case HTTP_MISSING_TARGET:
 384                repo->got_indices = 1;
 385                ret = 0;
 386                break;
 387        default:
 388                repo->got_indices = 0;
 389                ret = -1;
 390        }
 391
 392        return ret;
 393}
 394
 395static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 396{
 397        struct packed_git *target;
 398        int ret;
 399        struct slot_results results;
 400        struct http_pack_request *preq;
 401
 402        if (fetch_indices(walker, repo))
 403                return -1;
 404        target = find_sha1_pack(sha1, repo->packs);
 405        if (!target)
 406                return -1;
 407
 408        if (walker->get_verbosely) {
 409                fprintf(stderr, "Getting pack %s\n",
 410                        sha1_to_hex(target->sha1));
 411                fprintf(stderr, " which contains %s\n",
 412                        sha1_to_hex(sha1));
 413        }
 414
 415        preq = new_http_pack_request(target, repo->base);
 416        if (preq == NULL)
 417                goto abort;
 418        preq->lst = &repo->packs;
 419        preq->slot->results = &results;
 420
 421        if (start_active_slot(preq->slot)) {
 422                run_active_slot(preq->slot);
 423                if (results.curl_result != CURLE_OK) {
 424                        error("Unable to get pack file %s\n%s", preq->url,
 425                              curl_errorstr);
 426                        goto abort;
 427                }
 428        } else {
 429                error("Unable to start request");
 430                goto abort;
 431        }
 432
 433        ret = finish_http_pack_request(preq);
 434        release_http_pack_request(preq);
 435        if (ret)
 436                return ret;
 437
 438        return 0;
 439
 440abort:
 441        return -1;
 442}
 443
 444static void abort_object_request(struct object_request *obj_req)
 445{
 446        release_object_request(obj_req);
 447}
 448
 449static int fetch_object(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 450{
 451        char *hex = sha1_to_hex(sha1);
 452        int ret = 0;
 453        struct object_request *obj_req = object_queue_head;
 454        struct http_object_request *req;
 455
 456        while (obj_req != NULL && hashcmp(obj_req->sha1, sha1))
 457                obj_req = obj_req->next;
 458        if (obj_req == NULL)
 459                return error("Couldn't find request for %s in the queue", hex);
 460
 461        if (has_sha1_file(obj_req->sha1)) {
 462                if (obj_req->req != NULL)
 463                        abort_http_object_request(obj_req->req);
 464                abort_object_request(obj_req);
 465                return 0;
 466        }
 467
 468#ifdef USE_CURL_MULTI
 469        while (obj_req->state == WAITING)
 470                step_active_slots();
 471#else
 472        start_object_request(walker, obj_req);
 473#endif
 474
 475        /*
 476         * obj_req->req might change when fetching alternates in the callback
 477         * process_object_response; therefore, the "shortcut" variable, req,
 478         * is used only after we're done with slots.
 479         */
 480        while (obj_req->state == ACTIVE)
 481                run_active_slot(obj_req->req->slot);
 482
 483        req = obj_req->req;
 484
 485        if (req->localfile != -1) {
 486                close(req->localfile);
 487                req->localfile = -1;
 488        }
 489
 490        if (obj_req->state == ABORTED) {
 491                ret = error("Request for %s aborted", hex);
 492        } else if (req->curl_result != CURLE_OK &&
 493                   req->http_code != 416) {
 494                if (missing_target(req))
 495                        ret = -1; /* Be silent, it is probably in a pack. */
 496                else
 497                        ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 498                                    req->errorstr, req->curl_result,
 499                                    req->http_code, hex);
 500        } else if (req->zret != Z_STREAM_END) {
 501                walker->corrupt_object_found++;
 502                ret = error("File %s (%s) corrupt", hex, req->url);
 503        } else if (hashcmp(obj_req->sha1, req->real_sha1)) {
 504                ret = error("File %s has bad hash", hex);
 505        } else if (req->rename < 0) {
 506                ret = error("unable to write sha1 filename %s",
 507                            sha1_file_name(req->sha1));
 508        }
 509
 510        release_http_object_request(req);
 511        release_object_request(obj_req);
 512        return ret;
 513}
 514
 515static int fetch(struct walker *walker, unsigned char *sha1)
 516{
 517        struct walker_data *data = walker->data;
 518        struct alt_base *altbase = data->alt;
 519
 520        if (!fetch_object(walker, altbase, sha1))
 521                return 0;
 522        while (altbase) {
 523                if (!http_fetch_pack(walker, altbase, sha1))
 524                        return 0;
 525                fetch_alternates(walker, data->alt->base);
 526                altbase = altbase->next;
 527        }
 528        return error("Unable to find %s under %s", sha1_to_hex(sha1),
 529                     data->alt->base);
 530}
 531
 532static int fetch_ref(struct walker *walker, struct ref *ref)
 533{
 534        struct walker_data *data = walker->data;
 535        return http_fetch_ref(data->alt->base, ref);
 536}
 537
 538static void cleanup(struct walker *walker)
 539{
 540        struct walker_data *data = walker->data;
 541        struct alt_base *alt, *alt_next;
 542
 543        if (data) {
 544                alt = data->alt;
 545                while (alt) {
 546                        alt_next = alt->next;
 547
 548                        free(alt->base);
 549                        free(alt);
 550
 551                        alt = alt_next;
 552                }
 553                free(data);
 554                walker->data = NULL;
 555        }
 556}
 557
 558struct walker *get_http_walker(const char *url)
 559{
 560        char *s;
 561        struct walker_data *data = xmalloc(sizeof(struct walker_data));
 562        struct walker *walker = xmalloc(sizeof(struct walker));
 563
 564        data->alt = xmalloc(sizeof(*data->alt));
 565        data->alt->base = xstrdup(url);
 566        for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 567                *s = 0;
 568
 569        data->alt->got_indices = 0;
 570        data->alt->packs = NULL;
 571        data->alt->next = NULL;
 572        data->got_alternates = -1;
 573
 574        walker->corrupt_object_found = 0;
 575        walker->fetch = fetch;
 576        walker->fetch_ref = fetch_ref;
 577        walker->prefetch = prefetch;
 578        walker->cleanup = cleanup;
 579        walker->data = data;
 580
 581#ifdef USE_CURL_MULTI
 582        add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 583#endif
 584
 585        return walker;
 586}