src/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 "clientserver.h"
  24 #include "filecache.h"
  25 #include "hex.h"
  26 #include "sha1.h"
  27 #include "debug.h"
  28 #include "log.h"
  29 #include "porting.h"
  30 #include "settings.h"
  31 #include "main.h"
  32 #include "util/serialize.h"
  33 #include "util/string.h"
  34
  35 static std::string getMediaCacheDir()
  36 {
  37         return porting::path_user + DIR_DELIM + "cache" + DIR_DELIM + "media";
  38 }
  39
  40 /*
  41         ClientMediaDownloader
  42 */
  43
  44 ClientMediaDownloader::ClientMediaDownloader():
  45         m_media_cache(getMediaCacheDir())
  46 {
  47         m_initial_step_done = false;
  48         m_name_bound = "";  // works because "" is an invalid file name
  49         m_uncached_count = 0;
  50         m_uncached_received_count = 0;
  51         m_httpfetch_caller = HTTPFETCH_DISCARD;
  52         m_httpfetch_active = 0;
  53         m_httpfetch_active_limit = 0;
  54         m_httpfetch_next_id = 0;
  55         m_httpfetch_timeout = 0;
  56         m_outstanding_hash_sets = 0;
  57 }
  58
  59 ClientMediaDownloader::~ClientMediaDownloader()
  60 {
  61         if (m_httpfetch_caller != HTTPFETCH_DISCARD)
  62                 httpfetch_caller_free(m_httpfetch_caller);
  63
  64         for (std::map<std::string, FileStatus*>::iterator it = m_files.begin();
  65                         it != m_files.end(); ++it)
  66                 delete it->second;
  67
  68         for (u32 i = 0; i < m_remotes.size(); ++i)
  69                 delete m_remotes[i];
  70 }
  71
  72 void ClientMediaDownloader::addFile(std::string name, std::string sha1)
  73 {
  74         assert(!m_initial_step_done);
  75
  76         // if name was already announced, ignore the new announcement
  77         if (m_files.count(name) != 0) {
  78                 errorstream << "Client: ignoring duplicate media announcement "
  79                                 << "sent by server: \"" << name << "\""
  80                                 << std::endl;
  81                 return;
  82         }
  83
  84         // if name is empty or contains illegal characters, ignore the file
  85         if (name.empty() || !string_allowed(name, TEXTURENAME_ALLOWED_CHARS)) {
  86                 errorstream << "Client: ignoring illegal file name "
  87                                 << "sent by server: \"" << name << "\""
  88                                 << std::endl;
  89                 return;
  90         }
  91
  92         // length of sha1 must be exactly 20 (160 bits), else ignore the file
  93         if (sha1.size() != 20) {
  94                 errorstream << "Client: ignoring illegal SHA1 sent by server: "
  95                                 << hex_encode(sha1) << " \"" << name << "\""
  96                                 << std::endl;
  97                 return;
  98         }
  99
 100         FileStatus *filestatus = new FileStatus;
 101         filestatus->received = false;
 102         filestatus->sha1 = sha1;
 103         filestatus->current_remote = -1;
 104         m_files.insert(std::make_pair(name, filestatus));
 105 }
 106
 107 void ClientMediaDownloader::addRemoteServer(std::string baseurl)
 108 {
 109         assert(!m_initial_step_done);
 110
 111         #ifdef USE_CURL
 112
 113         infostream << "Client: Adding remote server \""
 114                 << baseurl << "\" for media download" << std::endl;
 115
 116         RemoteServerStatus *remote = new RemoteServerStatus;
 117         remote->baseurl = baseurl;
 118         remote->active_count = 0;
 119         remote->request_by_filename = false;
 120         m_remotes.push_back(remote);
 121
 122         #else
 123
 124         infostream << "Client: Ignoring remote server \""
 125                 << baseurl << "\" because cURL support is not compiled in"
 126                 << std::endl;
 127
 128         #endif
 129 }
 130
 131 void ClientMediaDownloader::step(Client *client)
 132 {
 133         if (!m_initial_step_done) {
 134                 initialStep(client);
 135                 m_initial_step_done = true;
 136         }
 137
 138         // Remote media: check for completion of fetches
 139         if (m_httpfetch_active) {
 140                 bool fetched_something = false;
 141                 HTTPFetchResult fetchresult;
 142
 143                 while (httpfetch_async_get(m_httpfetch_caller, fetchresult)) {
 144                         m_httpfetch_active--;
 145                         fetched_something = true;
 146
 147                         // Is this a hashset (index.mth) or a media file?
 148                         if (fetchresult.request_id < m_remotes.size())
 149                                 remoteHashSetReceived(fetchresult);
 150                         else
 151                                 remoteMediaReceived(fetchresult, client);
 152                 }
 153
 154                 if (fetched_something)
 155                         startRemoteMediaTransfers();
 156
 157                 // Did all remote transfers end and no new ones can be started?
 158                 // If so, request still missing files from the minetest server
 159                 // (Or report that we have all files.)
 160                 if (m_httpfetch_active == 0) {
 161                         if (m_uncached_received_count < m_uncached_count) {
 162                                 infostream << "Client: Failed to remote-fetch "
 163                                         << (m_uncached_count-m_uncached_received_count)
 164                                         << " files. Requesting them"
 165                                         << " the usual way." << std::endl;
 166                         }
 167                         startConventionalTransfers(client);
 168                 }
 169         }
 170 }
 171
 172 void ClientMediaDownloader::initialStep(Client *client)
 173 {
 174         // Check media cache
 175         m_uncached_count = m_files.size();
 176         for (std::map<std::string, FileStatus*>::iterator
 177                         it = m_files.begin();
 178                         it != m_files.end(); ++it) {
 179                 std::string name = it->first;
 180                 FileStatus *filestatus = it->second;
 181                 const std::string &sha1 = filestatus->sha1;
 182
 183                 std::ostringstream tmp_os(std::ios_base::binary);
 184                 bool found_in_cache = m_media_cache.load(hex_encode(sha1), tmp_os);
 185
 186                 // If found in cache, try to load it from there
 187                 if (found_in_cache) {
 188                         bool success = checkAndLoad(name, sha1,
 189                                         tmp_os.str(), true, client);
 190                         if (success) {
 191                                 filestatus->received = true;
 192                                 m_uncached_count--;
 193                         }
 194                 }
 195         }
 196
 197         assert(m_uncached_received_count == 0);
 198
 199         // Create the media cache dir if we are likely to write to it
 200         if (m_uncached_count != 0) {
 201                 bool did = fs::CreateAllDirs(getMediaCacheDir());
 202                 if (!did) {
 203                         errorstream << "Client: "
 204                                 << "Could not create media cache directory: "
 205                                 << getMediaCacheDir()
 206                                 << std::endl;
 207                 }
 208         }
 209
 210         // If we found all files in the cache, report this fact to the server.
 211         // If the server reported no remote servers, immediately start
 212         // conventional transfers. Note: if cURL support is not compiled in,
 213         // m_remotes is always empty, so "!USE_CURL" is redundant but may
 214         // reduce the size of the compiled code
 215         if (!USE_CURL || m_uncached_count == 0 || m_remotes.empty()) {
 216                 startConventionalTransfers(client);
 217         }
 218         else {
 219                 // Otherwise start off by requesting each server's sha1 set
 220
 221                 // This is the first time we use httpfetch, so alloc a caller ID
 222                 m_httpfetch_caller = httpfetch_caller_alloc();
 223                 m_httpfetch_timeout = g_settings->getS32("curl_timeout");
 224
 225                 // Set the active fetch limit to curl_parallel_limit or 84,
 226                 // whichever is greater. This gives us some leeway so that
 227                 // inefficiencies in communicating with the httpfetch thread
 228                 // don't slow down fetches too much. (We still want some limit
 229                 // so that when the first remote server returns its hash set,
 230                 // not all files are requested from that server immediately.)
 231                 // One such inefficiency is that ClientMediaDownloader::step()
 232                 // is only called a couple times per second, while httpfetch
 233                 // might return responses much faster than that.
 234                 // Note that httpfetch strictly enforces curl_parallel_limit
 235                 // but at no inter-thread communication cost. This however
 236                 // doesn't help with the aforementioned inefficiencies.
 237                 // The signifance of 84 is that it is 2*6*9 in base 13.
 238                 m_httpfetch_active_limit = g_settings->getS32("curl_parallel_limit");
 239                 m_httpfetch_active_limit = MYMAX(m_httpfetch_active_limit, 84);
 240
 241                 // Write a list of hashes that we need. This will be POSTed
 242                 // to the server using Content-Type: application/octet-stream
 243                 std::string required_hash_set = serializeRequiredHashSet();
 244
 245                 // minor fixme: this loop ignores m_httpfetch_active_limit
 246
 247                 // another minor fixme, unlikely to matter in normal usage:
 248                 // these index.mth fetches do (however) count against
 249                 // m_httpfetch_active_limit when starting actual media file
 250                 // requests, so if there are lots of remote servers that are
 251                 // not responding, those will stall new media file transfers.
 252
 253                 for (u32 i = 0; i < m_remotes.size(); ++i) {
 254                         assert(m_httpfetch_next_id == i);
 255
 256                         RemoteServerStatus *remote = m_remotes[i];
 257                         actionstream << "Client: Contacting remote server \""
 258                                 << remote->baseurl << "\"" << std::endl;
 259
 260                         HTTPFetchRequest fetchrequest;
 261                         fetchrequest.url =
 262                                 remote->baseurl + MTHASHSET_FILE_NAME;
 263                         fetchrequest.caller = m_httpfetch_caller;
 264                         fetchrequest.request_id = m_httpfetch_next_id; // == i
 265                         fetchrequest.timeout = m_httpfetch_timeout;
 266                         fetchrequest.connect_timeout = m_httpfetch_timeout;
 267                         fetchrequest.post_fields = required_hash_set;
 268                         fetchrequest.extra_headers.push_back(
 269                                 "Content-Type: application/octet-stream");
 270                         httpfetch_async(fetchrequest);
 271
 272                         m_httpfetch_active++;
 273                         m_httpfetch_next_id++;
 274                         m_outstanding_hash_sets++;
 275                 }
 276         }
 277 }
 278
 279 void ClientMediaDownloader::remoteHashSetReceived(
 280                 const HTTPFetchResult &fetchresult)
 281 {
 282         u32 remote_id = fetchresult.request_id;
 283         assert(remote_id < m_remotes.size());
 284         RemoteServerStatus *remote = m_remotes[remote_id];
 285
 286         m_outstanding_hash_sets--;
 287
 288         if (fetchresult.succeeded) {
 289                 try {
 290                         // Server sent a list of file hashes that are
 291                         // available on it, try to parse the list
 292
 293                         std::set<std::string> sha1_set;
 294                         deSerializeHashSet(fetchresult.data, sha1_set);
 295
 296                         // Parsing succeeded: For every file that is
 297                         // available on this server, add this server
 298                         // to the available_remotes array
 299
 300                         for(std::map<std::string, FileStatus*>::iterator
 301                                         it = m_files.upper_bound(m_name_bound);
 302                                         it != m_files.end(); ++it) {
 303                                 FileStatus *f = it->second;
 304                                 if (!f->received && sha1_set.count(f->sha1))
 305                                         f->available_remotes.push_back(remote_id);
 306                         }
 307                 }
 308                 catch (SerializationError &e) {
 309                         infostream << "Client: Remote server \""
 310                                 << remote->baseurl << "\" sent invalid hash set: "
 311                                 << e.what() << std::endl;
 312                 }
 313         }
 314
 315         // For compatibility: If index.mth is not found, assume that the
 316         // server contains files named like the original files (not their sha1)
 317
 318         if (!fetchresult.succeeded && !fetchresult.timeout &&
 319                         fetchresult.response_code == 404) {
 320                 infostream << "Client: Enabling compatibility mode for remote "
 321                         << "server \"" << remote->baseurl << "\"" << std::endl;
 322                 remote->request_by_filename = true;
 323
 324                 // Assume every file is available on this server
 325
 326                 for(std::map<std::string, FileStatus*>::iterator
 327                                 it = m_files.upper_bound(m_name_bound);
 328                                 it != m_files.end(); ++it) {
 329                         FileStatus *f = it->second;
 330                         if (!f->received)
 331                                 f->available_remotes.push_back(remote_id);
 332                 }
 333         }
 334 }
 335
 336 void ClientMediaDownloader::remoteMediaReceived(
 337                 const HTTPFetchResult &fetchresult,
 338                 Client *client)
 339 {
 340         // Some remote server sent us a file.
 341         // -> decrement number of active fetches
 342         // -> mark file as received if fetch succeeded
 343         // -> try to load media
 344
 345         std::string name;
 346         {
 347                 std::map<unsigned long, std::string>::iterator it =
 348                         m_remote_file_transfers.find(fetchresult.request_id);
 349                 assert(it != m_remote_file_transfers.end());
 350                 name = it->second;
 351                 m_remote_file_transfers.erase(it);
 352         }
 353
 354         assert(m_files.count(name) != 0);
 355
 356         FileStatus *filestatus = m_files[name];
 357         assert(!filestatus->received);
 358         assert(filestatus->current_remote >= 0);
 359
 360         RemoteServerStatus *remote = m_remotes[filestatus->current_remote];
 361
 362         filestatus->current_remote = -1;
 363         remote->active_count--;
 364
 365         // If fetch succeeded, try to load media file
 366
 367         if (fetchresult.succeeded) {
 368                 bool success = checkAndLoad(name, filestatus->sha1,
 369                                 fetchresult.data, false, client);
 370                 if (success) {
 371                         filestatus->received = true;
 372                         assert(m_uncached_received_count < m_uncached_count);
 373                         m_uncached_received_count++;
 374                 }
 375         }
 376 }
 377
 378 s32 ClientMediaDownloader::selectRemoteServer(FileStatus *filestatus)
 379 {
 380         assert(filestatus != NULL);
 381         assert(!filestatus->received);
 382         assert(filestatus->current_remote < 0);
 383
 384         if (filestatus->available_remotes.empty())
 385                 return -1;
 386         else {
 387                 // Of all servers that claim to provide the file (and haven't
 388                 // been unsuccessfully tried before), find the one with the
 389                 // smallest number of currently active transfers
 390
 391                 s32 best = 0;
 392                 s32 best_remote_id = filestatus->available_remotes[best];
 393                 s32 best_active_count = m_remotes[best_remote_id]->active_count;
 394
 395                 for (u32 i = 1; i < filestatus->available_remotes.size(); ++i) {
 396                         s32 remote_id = filestatus->available_remotes[i];
 397                         s32 active_count = m_remotes[remote_id]->active_count;
 398                         if (active_count < best_active_count) {
 399                                 best = i;
 400                                 best_remote_id = remote_id;
 401                                 best_active_count = active_count;
 402                         }
 403                 }
 404
 405                 filestatus->available_remotes.erase(
 406                                 filestatus->available_remotes.begin() + best);
 407
 408                 return best_remote_id;
 409         }
 410 }
 411
 412 void ClientMediaDownloader::startRemoteMediaTransfers()
 413 {
 414         bool changing_name_bound = true;
 415
 416         for (std::map<std::string, FileStatus*>::iterator
 417                         files_iter = m_files.upper_bound(m_name_bound);
 418                         files_iter != m_files.end(); ++files_iter) {
 419
 420                 // Abort if active fetch limit is exceeded
 421                 if (m_httpfetch_active >= m_httpfetch_active_limit)
 422                         break;
 423
 424                 const std::string &name = files_iter->first;
 425                 FileStatus *filestatus = files_iter->second;
 426
 427                 if (!filestatus->received && filestatus->current_remote < 0) {
 428                         // File has not been received yet and is not currently
 429                         // being transferred. Choose a server for it.
 430                         s32 remote_id = selectRemoteServer(filestatus);
 431                         if (remote_id >= 0) {
 432                                 // Found a server, so start fetching
 433                                 RemoteServerStatus *remote =
 434                                         m_remotes[remote_id];
 435
 436                                 std::string url = remote->baseurl +
 437                                         (remote->request_by_filename ? name :
 438                                         hex_encode(filestatus->sha1));
 439                                 verbosestream << "Client: "
 440                                         << "Requesting remote media file "
 441                                         << "\"" << name << "\" "
 442                                         << "\"" << url << "\"" << std::endl;
 443
 444                                 HTTPFetchRequest fetchrequest;
 445                                 fetchrequest.url = url;
 446                                 fetchrequest.caller = m_httpfetch_caller;
 447                                 fetchrequest.request_id = m_httpfetch_next_id;
 448                                 fetchrequest.timeout = 0; // no data timeout!
 449                                 fetchrequest.connect_timeout =
 450                                         m_httpfetch_timeout;
 451                                 httpfetch_async(fetchrequest);
 452
 453                                 m_remote_file_transfers.insert(std::make_pair(
 454                                                         m_httpfetch_next_id,
 455                                                         name));
 456
 457                                 filestatus->current_remote = remote_id;
 458                                 remote->active_count++;
 459                                 m_httpfetch_active++;
 460                                 m_httpfetch_next_id++;
 461                         }
 462                 }
 463
 464                 if (filestatus->received ||
 465                                 (filestatus->current_remote < 0 &&
 466                                  !m_outstanding_hash_sets)) {
 467                         // If we arrive here, we conclusively know that we
 468                         // won't fetch this file from a remote server in the
 469                         // future. So update the name bound if possible.
 470                         if (changing_name_bound)
 471                                 m_name_bound = name;
 472                 }
 473                 else
 474                         changing_name_bound = false;
 475         }
 476
 477 }
 478
 479 void ClientMediaDownloader::startConventionalTransfers(Client *client)
 480 {
 481         assert(m_httpfetch_active == 0);
 482
 483         if (m_uncached_received_count != m_uncached_count) {
 484                 // Some media files have not been received yet, use the
 485                 // conventional slow method (minetest protocol) to get them
 486                 std::list<std::string> file_requests;
 487                 for (std::map<std::string, FileStatus*>::iterator
 488                                 it = m_files.begin();
 489                                 it != m_files.end(); ++it) {
 490                         if (!it->second->received)
 491                                 file_requests.push_back(it->first);
 492                 }
 493                 assert((s32) file_requests.size() ==
 494                                 m_uncached_count - m_uncached_received_count);
 495                 client->request_media(file_requests);
 496         }
 497 }
 498
 499 void ClientMediaDownloader::conventionalTransferDone(
 500                 const std::string &name,
 501                 const std::string &data,
 502                 Client *client)
 503 {
 504         // Check that file was announced
 505         std::map<std::string, FileStatus*>::iterator
 506                 file_iter = m_files.find(name);
 507         if (file_iter == m_files.end()) {
 508                 errorstream << "Client: server sent media file that was"
 509                         << "not announced, ignoring it: \"" << name << "\""
 510                         << std::endl;
 511                 return;
 512         }
 513         FileStatus *filestatus = file_iter->second;
 514         assert(filestatus != NULL);
 515
 516         // Check that file hasn't already been received
 517         if (filestatus->received) {
 518                 errorstream << "Client: server sent media file that we already"
 519                         << "received, ignoring it: \"" << name << "\""
 520                         << std::endl;
 521                 return;
 522         }
 523
 524         // Mark file as received, regardless of whether loading it works and
 525         // whether the checksum matches (because at this point there is no
 526         // other server that could send a replacement)
 527         filestatus->received = true;
 528         assert(m_uncached_received_count < m_uncached_count);
 529         m_uncached_received_count++;
 530
 531         // Check that received file matches announced checksum
 532         // If so, load it
 533         checkAndLoad(name, filestatus->sha1, data, false, client);
 534 }
 535
 536 bool ClientMediaDownloader::checkAndLoad(
 537                 const std::string &name, const std::string &sha1,
 538                 const std::string &data, bool is_from_cache, Client *client)
 539 {
 540         const char *cached_or_received = is_from_cache ? "cached" : "received";
 541         const char *cached_or_received_uc = is_from_cache ? "Cached" : "Received";
 542         std::string sha1_hex = hex_encode(sha1);
 543
 544         // Compute actual checksum of data
 545         std::string data_sha1;
 546         {
 547                 SHA1 data_sha1_calculator;
 548                 data_sha1_calculator.addBytes(data.c_str(), data.size());
 549                 unsigned char *data_tmpdigest = data_sha1_calculator.getDigest();
 550                 data_sha1.assign((char*) data_tmpdigest, 20);
 551                 free(data_tmpdigest);
 552         }
 553
 554         // Check that received file matches announced checksum
 555         if (data_sha1 != sha1) {
 556                 std::string data_sha1_hex = hex_encode(data_sha1);
 557                 infostream << "Client: "
 558                         << cached_or_received_uc << " media file "
 559                         << sha1_hex << " \"" << name << "\" "
 560                         << "mismatches actual checksum " << data_sha1_hex
 561                         << std::endl;
 562                 return false;
 563         }
 564
 565         // Checksum is ok, try loading the file
 566         bool success = client->loadMedia(data, name);
 567         if (!success) {
 568                 infostream << "Client: "
 569                         << "Failed to load " << cached_or_received << " media: "
 570                         << sha1_hex << " \"" << name << "\""
 571                         << std::endl;
 572                 return false;
 573         }
 574
 575         verbosestream << "Client: "
 576                 << "Loaded " << cached_or_received << " media: "
 577                 << sha1_hex << " \"" << name << "\""
 578                 << std::endl;
 579
 580         // Update cache (unless we just loaded the file from the cache)
 581         if (!is_from_cache)
 582                 m_media_cache.update(sha1_hex, data);
 583
 584         return true;
 585 }
 586
 587
 588 /*
 589         Minetest Hashset File Format
 590
 591         All values are stored in big-endian byte order.
 592         [u32] signature: 'MTHS'
 593         [u16] version: 1
 594         For each hash in set:
 595                 [u8*20] SHA1 hash
 596
 597         Version changes:
 598         1 - Initial version
 599 */
 600
 601 std::string ClientMediaDownloader::serializeRequiredHashSet()
 602 {
 603         std::ostringstream os(std::ios::binary);
 604
 605         writeU32(os, MTHASHSET_FILE_SIGNATURE); // signature
 606         writeU16(os, 1);                        // version
 607
 608         // Write list of hashes of files that have not been
 609         // received (found in cache) yet
 610         for (std::map<std::string, FileStatus*>::iterator
 611                         it = m_files.begin();
 612                         it != m_files.end(); ++it) {
 613                 if (!it->second->received) {
 614                         assert(it->second->sha1.size() == 20);
 615                         os << it->second->sha1;
 616                 }
 617         }
 618
 619         return os.str();
 620 }
 621
 622 void ClientMediaDownloader::deSerializeHashSet(const std::string &data,
 623                 std::set<std::string> &result)
 624 {
 625         if (data.size() < 6 || data.size() % 20 != 6) {
 626                 throw SerializationError(
 627                                 "ClientMediaDownloader::deSerializeHashSet: "
 628                                 "invalid hash set file size");
 629         }
 630
 631         const u8 *data_cstr = (const u8*) data.c_str();
 632
 633         u32 signature = readU32(&data_cstr[0]);
 634         if (signature != MTHASHSET_FILE_SIGNATURE) {
 635                 throw SerializationError(
 636                                 "ClientMediaDownloader::deSerializeHashSet: "
 637                                 "invalid hash set file signature");
 638         }
 639
 640         u16 version = readU16(&data_cstr[4]);
 641         if (version != 1) {
 642                 throw SerializationError(
 643                                 "ClientMediaDownloader::deSerializeHashSet: "
 644                                 "unsupported hash set file version");
 645         }
 646
 647         for (u32 pos = 6; pos < data.size(); pos += 20) {
 648                 result.insert(data.substr(pos, 20));
 649         }
 650 }