src/client/clientmedia.cpp

   1 /*
   2 Minetest
   3 Copyright (C) 2013 celeron55, Perttu Ahola <celeron55@gmail.com>
   4
   5 This program is free software; you can redistribute it and/or modify
   6 it under the terms of the GNU Lesser General Public License as published by
   7 the Free Software Foundation; either version 2.1 of the License, or
   8 (at your option) any later version.
   9
  10 This program is distributed in the hope that it will be useful,
  11 but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 GNU Lesser General Public License for more details.
  14
  15 You should have received a copy of the GNU Lesser General Public License along
  16 with this program; if not, write to the Free Software Foundation, Inc.,
  17 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  18 */
  19
  20 #include "clientmedia.h"
  21 #include "httpfetch.h"
  22 #include "client.h"
  23 #include "filecache.h"
  24 #include "filesys.h"
  25 #include "log.h"
  26 #include "porting.h"
  27 #include "settings.h"
  28 #include "util/hex.h"
  29 #include "util/serialize.h"
  30 #include "util/sha1.h"
  31 #include "util/string.h"
  32
  33 static std::string getMediaCacheDir()
  34 {
  35         return porting::path_cache + DIR_DELIM + "media";
  36 }
  37
  38 bool clientMediaUpdateCache(const std::string &raw_hash, const std::string &filedata)
  39 {
  40         FileCache media_cache(getMediaCacheDir());
  41         std::string sha1_hex = hex_encode(raw_hash);
  42         if (!media_cache.exists(sha1_hex))
  43                 return media_cache.update(sha1_hex, filedata);
  44         return true;
  45 }
  46
  47 /*
  48         ClientMediaDownloader
  49 */
  50
  51 ClientMediaDownloader::ClientMediaDownloader():
  52         m_media_cache(getMediaCacheDir()),
  53         m_httpfetch_caller(HTTPFETCH_DISCARD)
  54 {
  55 }
  56
  57 ClientMediaDownloader::~ClientMediaDownloader()
  58 {
  59         if (m_httpfetch_caller != HTTPFETCH_DISCARD)
  60                 httpfetch_caller_free(m_httpfetch_caller);
  61
  62         for (auto &file_it : m_files)
  63                 delete file_it.second;
  64
  65         for (auto &remote : m_remotes)
  66                 delete remote;
  67 }
  68
  69 void ClientMediaDownloader::addFile(const std::string &name, const std::string &sha1)
  70 {
  71         assert(!m_initial_step_done); // pre-condition
  72
  73         // if name was already announced, ignore the new announcement
  74         if (m_files.count(name) != 0) {
  75                 errorstream << "Client: ignoring duplicate media announcement "
  76                                 << "sent by server: \"" << name << "\""
  77                                 << std::endl;
  78                 return;
  79         }
  80
  81         // if name is empty or contains illegal characters, ignore the file
  82         if (name.empty() || !string_allowed(name, TEXTURENAME_ALLOWED_CHARS)) {
  83                 errorstream << "Client: ignoring illegal file name "
  84                                 << "sent by server: \"" << name << "\""
  85                                 << std::endl;
  86                 return;
  87         }
  88
  89         // length of sha1 must be exactly 20 (160 bits), else ignore the file
  90         if (sha1.size() != 20) {
  91                 errorstream << "Client: ignoring illegal SHA1 sent by server: "
  92                                 << hex_encode(sha1) << " \"" << name << "\""
  93                                 << std::endl;
  94                 return;
  95         }
  96
  97         FileStatus *filestatus = new FileStatus();
  98         filestatus->received = false;
  99         filestatus->sha1 = sha1;
 100         filestatus->current_remote = -1;
 101         m_files.insert(std::make_pair(name, filestatus));
 102 }
 103
 104 void ClientMediaDownloader::addRemoteServer(const std::string &baseurl)
 105 {
 106         assert(!m_initial_step_done);   // pre-condition
 107
 108         #ifdef USE_CURL
 109
 110         if (g_settings->getBool("enable_remote_media_server")) {
 111                 infostream << "Client: Adding remote server \""
 112                         << baseurl << "\" for media download" << std::endl;
 113
 114                 RemoteServerStatus *remote = new RemoteServerStatus();
 115                 remote->baseurl = baseurl;
 116                 remote->active_count = 0;
 117                 m_remotes.push_back(remote);
 118         }
 119
 120         #else
 121
 122         infostream << "Client: Ignoring remote server \""
 123                 << baseurl << "\" because cURL support is not compiled in"
 124                 << std::endl;
 125
 126         #endif
 127 }
 128
 129 void ClientMediaDownloader::step(Client *client)
 130 {
 131         if (!m_initial_step_done) {
 132                 initialStep(client);
 133                 m_initial_step_done = true;
 134         }
 135
 136         // Remote media: check for completion of fetches
 137         if (m_httpfetch_active) {
 138                 bool fetched_something = false;
 139                 HTTPFetchResult fetch_result;
 140
 141                 while (httpfetch_async_get(m_httpfetch_caller, fetch_result)) {
 142                         m_httpfetch_active--;
 143                         fetched_something = true;
 144
 145                         // Is this a hashset (index.mth) or a media file?
 146                         if (fetch_result.request_id < m_remotes.size())
 147                                 remoteHashSetReceived(fetch_result);
 148                         else
 149                                 remoteMediaReceived(fetch_result, client);
 150                 }
 151
 152                 if (fetched_something)
 153                         startRemoteMediaTransfers();
 154
 155                 // Did all remote transfers end and no new ones can be started?
 156                 // If so, request still missing files from the minetest server
 157                 // (Or report that we have all files.)
 158                 if (m_httpfetch_active == 0) {
 159                         if (m_uncached_received_count < m_uncached_count) {
 160                                 infostream << "Client: Failed to remote-fetch "
 161                                         << (m_uncached_count-m_uncached_received_count)
 162                                         << " files. Requesting them"
 163                                         << " the usual way." << std::endl;
 164                         }
 165                         startConventionalTransfers(client);
 166                 }
 167         }
 168 }
 169
 170 void ClientMediaDownloader::initialStep(Client *client)
 171 {
 172         // Check media cache
 173         m_uncached_count = m_files.size();
 174         for (auto &file_it : m_files) {
 175                 std::string name = file_it.first;
 176                 FileStatus *filestatus = file_it.second;
 177                 const std::string &sha1 = filestatus->sha1;
 178
 179                 std::ostringstream tmp_os(std::ios_base::binary);
 180                 bool found_in_cache = m_media_cache.load(hex_encode(sha1), tmp_os);
 181
 182                 // If found in cache, try to load it from there
 183                 if (found_in_cache) {
 184                         bool success = checkAndLoad(name, sha1,
 185                                         tmp_os.str(), true, client);
 186                         if (success) {
 187                                 filestatus->received = true;
 188                                 m_uncached_count--;
 189                         }
 190                 }
 191         }
 192
 193         assert(m_uncached_received_count == 0);
 194
 195         // Create the media cache dir if we are likely to write to it
 196         if (m_uncached_count != 0) {
 197                 bool did = fs::CreateAllDirs(getMediaCacheDir());
 198                 if (!did) {
 199                         errorstream << "Client: "
 200                                 << "Could not create media cache directory: "
 201                                 << getMediaCacheDir()
 202                                 << std::endl;
 203                 }
 204         }
 205
 206         // If we found all files in the cache, report this fact to the server.
 207         // If the server reported no remote servers, immediately start
 208         // conventional transfers. Note: if cURL support is not compiled in,
 209         // m_remotes is always empty, so "!USE_CURL" is redundant but may
 210         // reduce the size of the compiled code
 211         if (!USE_CURL || m_uncached_count == 0 || m_remotes.empty()) {
 212                 startConventionalTransfers(client);
 213         }
 214         else {
 215                 // Otherwise start off by requesting each server's sha1 set
 216
 217                 // This is the first time we use httpfetch, so alloc a caller ID
 218                 m_httpfetch_caller = httpfetch_caller_alloc();
 219                 m_httpfetch_timeout = g_settings->getS32("curl_timeout");
 220
 221                 // Set the active fetch limit to curl_parallel_limit or 84,
 222                 // whichever is greater. This gives us some leeway so that
 223                 // inefficiencies in communicating with the httpfetch thread
 224                 // don't slow down fetches too much. (We still want some limit
 225                 // so that when the first remote server returns its hash set,
 226                 // not all files are requested from that server immediately.)
 227                 // One such inefficiency is that ClientMediaDownloader::step()
 228                 // is only called a couple times per second, while httpfetch
 229                 // might return responses much faster than that.
 230                 // Note that httpfetch strictly enforces curl_parallel_limit
 231                 // but at no inter-thread communication cost. This however
 232                 // doesn't help with the aforementioned inefficiencies.
 233                 // The signifance of 84 is that it is 2*6*9 in base 13.
 234                 m_httpfetch_active_limit = g_settings->getS32("curl_parallel_limit");
 235                 m_httpfetch_active_limit = MYMAX(m_httpfetch_active_limit, 84);
 236
 237                 // Write a list of hashes that we need. This will be POSTed
 238                 // to the server using Content-Type: application/octet-stream
 239                 std::string required_hash_set = serializeRequiredHashSet();
 240
 241                 // minor fixme: this loop ignores m_httpfetch_active_limit
 242
 243                 // another minor fixme, unlikely to matter in normal usage:
 244                 // these index.mth fetches do (however) count against
 245                 // m_httpfetch_active_limit when starting actual media file
 246                 // requests, so if there are lots of remote servers that are
 247                 // not responding, those will stall new media file transfers.
 248
 249                 for (u32 i = 0; i < m_remotes.size(); ++i) {
 250                         assert(m_httpfetch_next_id == i);
 251
 252                         RemoteServerStatus *remote = m_remotes[i];
 253                         actionstream << "Client: Contacting remote server \""
 254                                 << remote->baseurl << "\"" << std::endl;
 255
 256                         HTTPFetchRequest fetch_request;
 257                         fetch_request.url =
 258                                 remote->baseurl + MTHASHSET_FILE_NAME;
 259                         fetch_request.caller = m_httpfetch_caller;
 260                         fetch_request.request_id = m_httpfetch_next_id; // == i
 261                         fetch_request.timeout = m_httpfetch_timeout;
 262                         fetch_request.connect_timeout = m_httpfetch_timeout;
 263                         fetch_request.method = HTTP_POST;
 264                         fetch_request.raw_data = required_hash_set;
 265                         fetch_request.extra_headers.emplace_back(
 266                                 "Content-Type: application/octet-stream");
 267
 268                         // Encapsulate possible IPv6 plain address in []
 269                         std::string addr = client->getAddressName();
 270                         if (addr.find(':', 0) != std::string::npos)
 271                                 addr = '[' + addr + ']';
 272                         fetch_request.extra_headers.emplace_back(
 273                                 std::string("Referer: minetest://") +
 274                                 addr + ":" +
 275                                 std::to_string(client->getServerAddress().getPort()));
 276
 277                         httpfetch_async(fetch_request);
 278
 279                         m_httpfetch_active++;
 280                         m_httpfetch_next_id++;
 281                         m_outstanding_hash_sets++;
 282                 }
 283         }
 284 }
 285
 286 void ClientMediaDownloader::remoteHashSetReceived(
 287                 const HTTPFetchResult &fetch_result)
 288 {
 289         u32 remote_id = fetch_result.request_id;
 290         assert(remote_id < m_remotes.size());
 291         RemoteServerStatus *remote = m_remotes[remote_id];
 292
 293         m_outstanding_hash_sets--;
 294
 295         if (fetch_result.succeeded) {
 296                 try {
 297                         // Server sent a list of file hashes that are
 298                         // available on it, try to parse the list
 299
 300                         std::set<std::string> sha1_set;
 301                         deSerializeHashSet(fetch_result.data, sha1_set);
 302
 303                         // Parsing succeeded: For every file that is
 304                         // available on this server, add this server
 305                         // to the available_remotes array
 306
 307                         for(std::map<std::string, FileStatus*>::iterator
 308                                         it = m_files.upper_bound(m_name_bound);
 309                                         it != m_files.end(); ++it) {
 310                                 FileStatus *f = it->second;
 311                                 if (!f->received && sha1_set.count(f->sha1))
 312                                         f->available_remotes.push_back(remote_id);
 313                         }
 314                 }
 315                 catch (SerializationError &e) {
 316                         infostream << "Client: Remote server \""
 317                                 << remote->baseurl << "\" sent invalid hash set: "
 318                                 << e.what() << std::endl;
 319                 }
 320         }
 321 }
 322
 323 void ClientMediaDownloader::remoteMediaReceived(
 324                 const HTTPFetchResult &fetch_result,
 325                 Client *client)
 326 {
 327         // Some remote server sent us a file.
 328         // -> decrement number of active fetches
 329         // -> mark file as received if fetch succeeded
 330         // -> try to load media
 331
 332         std::string name;
 333         {
 334                 std::unordered_map<unsigned long, std::string>::iterator it =
 335                         m_remote_file_transfers.find(fetch_result.request_id);
 336                 assert(it != m_remote_file_transfers.end());
 337                 name = it->second;
 338                 m_remote_file_transfers.erase(it);
 339         }
 340
 341         sanity_check(m_files.count(name) != 0);
 342
 343         FileStatus *filestatus = m_files[name];
 344         sanity_check(!filestatus->received);
 345         sanity_check(filestatus->current_remote >= 0);
 346
 347         RemoteServerStatus *remote = m_remotes[filestatus->current_remote];
 348
 349         filestatus->current_remote = -1;
 350         remote->active_count--;
 351
 352         // If fetch succeeded, try to load media file
 353
 354         if (fetch_result.succeeded) {
 355                 bool success = checkAndLoad(name, filestatus->sha1,
 356                                 fetch_result.data, false, client);
 357                 if (success) {
 358                         filestatus->received = true;
 359                         assert(m_uncached_received_count < m_uncached_count);
 360                         m_uncached_received_count++;
 361                 }
 362         }
 363 }
 364
 365 s32 ClientMediaDownloader::selectRemoteServer(FileStatus *filestatus)
 366 {
 367         // Pre-conditions
 368         assert(filestatus != NULL);
 369         assert(!filestatus->received);
 370         assert(filestatus->current_remote < 0);
 371
 372         if (filestatus->available_remotes.empty())
 373                 return -1;
 374
 375         // Of all servers that claim to provide the file (and haven't
 376         // been unsuccessfully tried before), find the one with the
 377         // smallest number of currently active transfers
 378
 379         s32 best = 0;
 380         s32 best_remote_id = filestatus->available_remotes[best];
 381         s32 best_active_count = m_remotes[best_remote_id]->active_count;
 382
 383         for (u32 i = 1; i < filestatus->available_remotes.size(); ++i) {
 384                 s32 remote_id = filestatus->available_remotes[i];
 385                 s32 active_count = m_remotes[remote_id]->active_count;
 386                 if (active_count < best_active_count) {
 387                         best = i;
 388                         best_remote_id = remote_id;
 389                         best_active_count = active_count;
 390                 }
 391         }
 392
 393         filestatus->available_remotes.erase(
 394                         filestatus->available_remotes.begin() + best);
 395
 396         return best_remote_id;
 397
 398 }
 399
 400 void ClientMediaDownloader::startRemoteMediaTransfers()
 401 {
 402         bool changing_name_bound = true;
 403
 404         for (std::map<std::string, FileStatus*>::iterator
 405                         files_iter = m_files.upper_bound(m_name_bound);
 406                         files_iter != m_files.end(); ++files_iter) {
 407
 408                 // Abort if active fetch limit is exceeded
 409                 if (m_httpfetch_active >= m_httpfetch_active_limit)
 410                         break;
 411
 412                 const std::string &name = files_iter->first;
 413                 FileStatus *filestatus = files_iter->second;
 414
 415                 if (!filestatus->received && filestatus->current_remote < 0) {
 416                         // File has not been received yet and is not currently
 417                         // being transferred. Choose a server for it.
 418                         s32 remote_id = selectRemoteServer(filestatus);
 419                         if (remote_id >= 0) {
 420                                 // Found a server, so start fetching
 421                                 RemoteServerStatus *remote =
 422                                         m_remotes[remote_id];
 423
 424                                 std::string url = remote->baseurl +
 425                                         hex_encode(filestatus->sha1);
 426                                 verbosestream << "Client: "
 427                                         << "Requesting remote media file "
 428                                         << "\"" << name << "\" "
 429                                         << "\"" << url << "\"" << std::endl;
 430
 431                                 HTTPFetchRequest fetch_request;
 432                                 fetch_request.url = url;
 433                                 fetch_request.caller = m_httpfetch_caller;
 434                                 fetch_request.request_id = m_httpfetch_next_id;
 435                                 fetch_request.timeout = 0; // no data timeout!
 436                                 fetch_request.connect_timeout =
 437                                         m_httpfetch_timeout;
 438                                 httpfetch_async(fetch_request);
 439
 440                                 m_remote_file_transfers.insert(std::make_pair(
 441                                                         m_httpfetch_next_id,
 442                                                         name));
 443
 444                                 filestatus->current_remote = remote_id;
 445                                 remote->active_count++;
 446                                 m_httpfetch_active++;
 447                                 m_httpfetch_next_id++;
 448                         }
 449                 }
 450
 451                 if (filestatus->received ||
 452                                 (filestatus->current_remote < 0 &&
 453                                  !m_outstanding_hash_sets)) {
 454                         // If we arrive here, we conclusively know that we
 455                         // won't fetch this file from a remote server in the
 456                         // future. So update the name bound if possible.
 457                         if (changing_name_bound)
 458                                 m_name_bound = name;
 459                 }
 460                 else
 461                         changing_name_bound = false;
 462         }
 463
 464 }
 465
 466 void ClientMediaDownloader::startConventionalTransfers(Client *client)
 467 {
 468         assert(m_httpfetch_active == 0);        // pre-condition
 469
 470         if (m_uncached_received_count != m_uncached_count) {
 471                 // Some media files have not been received yet, use the
 472                 // conventional slow method (minetest protocol) to get them
 473                 std::vector<std::string> file_requests;
 474                 for (auto &file : m_files) {
 475                         if (!file.second->received)
 476                                 file_requests.push_back(file.first);
 477                 }
 478                 assert((s32) file_requests.size() ==
 479                                 m_uncached_count - m_uncached_received_count);
 480                 client->request_media(file_requests);
 481         }
 482 }
 483
 484 void ClientMediaDownloader::conventionalTransferDone(
 485                 const std::string &name,
 486                 const std::string &data,
 487                 Client *client)
 488 {
 489         // Check that file was announced
 490         std::map<std::string, FileStatus*>::iterator
 491                 file_iter = m_files.find(name);
 492         if (file_iter == m_files.end()) {
 493                 errorstream << "Client: server sent media file that was"
 494                         << "not announced, ignoring it: \"" << name << "\""
 495                         << std::endl;
 496                 return;
 497         }
 498         FileStatus *filestatus = file_iter->second;
 499         assert(filestatus != NULL);
 500
 501         // Check that file hasn't already been received
 502         if (filestatus->received) {
 503                 errorstream << "Client: server sent media file that we already"
 504                         << "received, ignoring it: \"" << name << "\""
 505                         << std::endl;
 506                 return;
 507         }
 508
 509         // Mark file as received, regardless of whether loading it works and
 510         // whether the checksum matches (because at this point there is no
 511         // other server that could send a replacement)
 512         filestatus->received = true;
 513         assert(m_uncached_received_count < m_uncached_count);
 514         m_uncached_received_count++;
 515
 516         // Check that received file matches announced checksum
 517         // If so, load it
 518         checkAndLoad(name, filestatus->sha1, data, false, client);
 519 }
 520
 521 bool ClientMediaDownloader::checkAndLoad(
 522                 const std::string &name, const std::string &sha1,
 523                 const std::string &data, bool is_from_cache, Client *client)
 524 {
 525         const char *cached_or_received = is_from_cache ? "cached" : "received";
 526         const char *cached_or_received_uc = is_from_cache ? "Cached" : "Received";
 527         std::string sha1_hex = hex_encode(sha1);
 528
 529         // Compute actual checksum of data
 530         std::string data_sha1;
 531         {
 532                 SHA1 data_sha1_calculator;
 533                 data_sha1_calculator.addBytes(data.c_str(), data.size());
 534                 unsigned char *data_tmpdigest = data_sha1_calculator.getDigest();
 535                 data_sha1.assign((char*) data_tmpdigest, 20);
 536                 free(data_tmpdigest);
 537         }
 538
 539         // Check that received file matches announced checksum
 540         if (data_sha1 != sha1) {
 541                 std::string data_sha1_hex = hex_encode(data_sha1);
 542                 infostream << "Client: "
 543                         << cached_or_received_uc << " media file "
 544                         << sha1_hex << " \"" << name << "\" "
 545                         << "mismatches actual checksum " << data_sha1_hex
 546                         << std::endl;
 547                 return false;
 548         }
 549
 550         // Checksum is ok, try loading the file
 551         bool success = client->loadMedia(data, name);
 552         if (!success) {
 553                 infostream << "Client: "
 554                         << "Failed to load " << cached_or_received << " media: "
 555                         << sha1_hex << " \"" << name << "\""
 556                         << std::endl;
 557                 return false;
 558         }
 559
 560         verbosestream << "Client: "
 561                 << "Loaded " << cached_or_received << " media: "
 562                 << sha1_hex << " \"" << name << "\""
 563                 << std::endl;
 564
 565         // Update cache (unless we just loaded the file from the cache)
 566         if (!is_from_cache)
 567                 m_media_cache.update(sha1_hex, data);
 568
 569         return true;
 570 }
 571
 572 /*
 573         Minetest Hashset File Format
 574
 575         All values are stored in big-endian byte order.
 576         [u32] signature: 'MTHS'
 577         [u16] version: 1
 578         For each hash in set:
 579                 [u8*20] SHA1 hash
 580
 581         Version changes:
 582         1 - Initial version
 583 */
 584
 585 std::string ClientMediaDownloader::serializeRequiredHashSet()
 586 {
 587         std::ostringstream os(std::ios::binary);
 588
 589         writeU32(os, MTHASHSET_FILE_SIGNATURE); // signature
 590         writeU16(os, 1);                        // version
 591
 592         // Write list of hashes of files that have not been
 593         // received (found in cache) yet
 594         for (std::map<std::string, FileStatus*>::iterator
 595                         it = m_files.begin();
 596                         it != m_files.end(); ++it) {
 597                 if (!it->second->received) {
 598                         FATAL_ERROR_IF(it->second->sha1.size() != 20, "Invalid SHA1 size");
 599                         os << it->second->sha1;
 600                 }
 601         }
 602
 603         return os.str();
 604 }
 605
 606 void ClientMediaDownloader::deSerializeHashSet(const std::string &data,
 607                 std::set<std::string> &result)
 608 {
 609         if (data.size() < 6 || data.size() % 20 != 6) {
 610                 throw SerializationError(
 611                                 "ClientMediaDownloader::deSerializeHashSet: "
 612                                 "invalid hash set file size");
 613         }
 614
 615         const u8 *data_cstr = (const u8*) data.c_str();
 616
 617         u32 signature = readU32(&data_cstr[0]);
 618         if (signature != MTHASHSET_FILE_SIGNATURE) {
 619                 throw SerializationError(
 620                                 "ClientMediaDownloader::deSerializeHashSet: "
 621                                 "invalid hash set file signature");
 622         }
 623
 624         u16 version = readU16(&data_cstr[4]);
 625         if (version != 1) {
 626                 throw SerializationError(
 627                                 "ClientMediaDownloader::deSerializeHashSet: "
 628                                 "unsupported hash set file version");
 629         }
 630
 631         for (u32 pos = 6; pos < data.size(); pos += 20) {
 632                 result.insert(data.substr(pos, 20));
 633         }
 634 }