git.sesse.net Git - cubemap/blob - httpinput.cpp

   1 #include <assert.h>
   2 #include <errno.h>
   3 #include <fcntl.h>
   4 #include <math.h>
   5 #include <netdb.h>
   6 #include <netinet/in.h>
   7 #include <poll.h>
   8 #include <stdint.h>
   9 #include <stdlib.h>
  10 #include <string.h>
  11 #include <sys/ioctl.h>
  12 #include <sys/socket.h>
  13 #include <sys/time.h>
  14 #include <sys/types.h>
  15 #include <sys/wait.h>
  16 #include <time.h>
  17 #include <unistd.h>
  18 #include <math.h>
  19 #include <spawn.h>
  20 #include <map>
  21 #include <string>
  22 #include <utility>
  23 #include <vector>
  24
  25 #include "httpinput.h"
  26 #include "log.h"
  27 #include "metacube2.h"
  28 #include "parse.h"
  29 #include "serverpool.h"
  30 #include "state.pb.h"
  31 #include "stream.h"
  32 #include "timespec.h"
  33 #include "util.h"
  34 #include "version.h"
  35
  36 using namespace std;
  37
  38 namespace {
  39
  40 string host_header(const string &host, const string &port)
  41 {
  42         if (port == "http" || atoi(port.c_str()) == 80) {
  43                 return host;
  44         } else {
  45                 return host + ":" + port;
  46         }
  47 }
  48
  49 }  // namespace
  50
  51 extern ServerPool *servers;
  52
  53 HTTPInput::HTTPInput(const string &url, Input::Encoding encoding)
  54         : state(NOT_CONNECTED),
  55           url(url),
  56           encoding(encoding)
  57 {
  58         stats.url = url;
  59         stats.bytes_received = 0;
  60         stats.data_bytes_received = 0;
  61         stats.metadata_bytes_received = 0;
  62         stats.connect_time = -1;
  63         stats.latency_sec = HUGE_VAL;
  64 }
  65
  66 HTTPInput::HTTPInput(const InputProto &serialized)
  67         : state(State(serialized.state())),
  68           url(serialized.url()),
  69           encoding(serialized.is_metacube_encoded() ?
  70                    Input::INPUT_ENCODING_METACUBE :
  71                    Input::INPUT_ENCODING_RAW),
  72           request(serialized.request()),
  73           request_bytes_sent(serialized.request_bytes_sent()),
  74           response(serialized.response()),
  75           http_header(serialized.http_header()),
  76           stream_header(serialized.stream_header()),
  77           has_metacube_header(serialized.has_metacube_header()),
  78           sock(serialized.sock())
  79 {
  80         // Set back the close-on-exec flag for the socket.
  81         // (This can't leak into a child, since we haven't been started yet.)
  82         fcntl(sock, F_SETFD, 1);
  83
  84         pending_data.resize(serialized.pending_data().size());
  85         memcpy(&pending_data[0], serialized.pending_data().data(), serialized.pending_data().size());
  86
  87         string protocol, user;
  88         parse_url(url, &protocol, &user, &host, &port, &path);  // Don't care if it fails.
  89
  90         stats.url = url;
  91         stats.bytes_received = serialized.bytes_received();
  92         stats.data_bytes_received = serialized.data_bytes_received();
  93         stats.metadata_bytes_received = serialized.metadata_bytes_received();
  94         if (serialized.has_connect_time()) {
  95                 stats.connect_time = serialized.connect_time();
  96         } else {
  97                 stats.connect_time = time(nullptr);
  98         }
  99         if (serialized.has_latency_sec()) {
 100                 stats.latency_sec = serialized.latency_sec();
 101         } else {
 102                 stats.latency_sec = HUGE_VAL;
 103         }
 104
 105         last_verbose_connection.tv_sec = -3600;
 106         last_verbose_connection.tv_nsec = 0;
 107 }
 108
 109 void HTTPInput::close_socket()
 110 {
 111         if (sock != -1) {
 112                 safe_close(sock);
 113                 sock = -1;
 114         }
 115
 116         lock_guard<mutex> lock(stats_mutex);
 117         stats.connect_time = -1;
 118 }
 119
 120 InputProto HTTPInput::serialize() const
 121 {
 122         // Unset the close-on-exec flag for the socket.
 123         // (This can't leak into a child, since there's only one thread left.)
 124         fcntl(sock, F_SETFD, 0);
 125
 126         InputProto serialized;
 127         serialized.set_state(state);
 128         serialized.set_url(url);
 129         serialized.set_request(request);
 130         serialized.set_request_bytes_sent(request_bytes_sent);
 131         serialized.set_response(response);
 132         serialized.set_http_header(http_header);
 133         serialized.set_stream_header(stream_header);
 134         serialized.set_pending_data(string(pending_data.begin(), pending_data.end()));
 135         serialized.set_has_metacube_header(has_metacube_header);
 136         serialized.set_sock(sock);
 137         serialized.set_bytes_received(stats.bytes_received);
 138         serialized.set_data_bytes_received(stats.data_bytes_received);
 139         if (isfinite(stats.latency_sec)) {
 140                 serialized.set_latency_sec(stats.latency_sec);
 141         }
 142         serialized.set_connect_time(stats.connect_time);
 143         if (encoding == Input::INPUT_ENCODING_METACUBE) {
 144                 serialized.set_is_metacube_encoded(true);
 145         } else {
 146                 assert(encoding == Input::INPUT_ENCODING_RAW);
 147                 serialized.set_is_metacube_encoded(false);
 148         }
 149         return serialized;
 150 }
 151
 152 int HTTPInput::lookup_and_connect(const string &host, const string &port)
 153 {
 154         addrinfo *ai;
 155         int err = getaddrinfo(host.c_str(), port.c_str(), nullptr, &ai);
 156         if (err != 0) {
 157                 if (!suppress_logging) {
 158                         log(WARNING, "[%s] Lookup of '%s' failed (%s).",
 159                                 url.c_str(), host.c_str(), gai_strerror(err));
 160                 }
 161                 return -1;
 162         }
 163
 164         addrinfo *base_ai = ai;
 165
 166         // Connect to everything in turn until we have a socket.
 167         for ( ; ai && !should_stop(); ai = ai->ai_next) {
 168                 // Now do a non-blocking connect. This is important because we want to be able to be
 169                 // woken up, even though it's rather cumbersome.
 170                 int sock = socket(ai->ai_family, SOCK_STREAM | SOCK_NONBLOCK | SOCK_CLOEXEC, IPPROTO_TCP);
 171                 if (sock == -1) {
 172                         // Could be e.g. EPROTONOSUPPORT. The show must go on.
 173                         continue;
 174                 }
 175
 176                 // Do a non-blocking connect.
 177                 do {
 178                         err = connect(sock, ai->ai_addr, ai->ai_addrlen);
 179                 } while (err == -1 && errno == EINTR);
 180
 181                 if (err == -1 && errno != EINPROGRESS) {
 182                         log_perror("connect");
 183                         safe_close(sock);
 184                         continue;
 185                 }
 186
 187                 // Wait for the connect to complete, or an error to happen.
 188                 for ( ;; ) {
 189                         bool complete = wait_for_activity(sock, POLLIN | POLLOUT, nullptr);
 190                         if (should_stop()) {
 191                                 safe_close(sock);
 192                                 return -1;
 193                         }
 194                         if (complete) {
 195                                 break;
 196                         }
 197                 }
 198
 199                 // Check whether it ended in an error or not.
 200                 socklen_t err_size = sizeof(err);
 201                 if (getsockopt(sock, SOL_SOCKET, SO_ERROR, &err, &err_size) == -1) {
 202                         log_perror("getsockopt");
 203                         safe_close(sock);
 204                         continue;
 205                 }
 206
 207                 errno = err;
 208
 209                 if (err == 0) {
 210                         // Successful connect.
 211                         freeaddrinfo(base_ai);
 212                         return sock;
 213                 }
 214
 215                 safe_close(sock);
 216         }
 217
 218         // Give the last one as error.
 219         if (!suppress_logging) {
 220                 log(WARNING, "[%s] Connect to '%s' failed (%s)",
 221                         url.c_str(), host.c_str(), strerror(errno));
 222         }
 223         freeaddrinfo(base_ai);
 224         return -1;
 225 }
 226
 227 int HTTPInput::open_child_process(const string &cmdline)
 228 {
 229         int devnullfd = open("/dev/null", O_RDONLY | O_CLOEXEC);
 230         if (devnullfd == -1) {
 231                 log_perror("/dev/null");
 232                 return -1;
 233         }
 234
 235         int pipefd[2];
 236         if (pipe2(pipefd, O_CLOEXEC) == -1) {
 237                 log_perror("pipe2()");
 238                 close(devnullfd);
 239                 return -1;
 240         }
 241
 242         // Point stdout to us, stdin to /dev/null, and stderr remains where it is
 243         // (probably the systemd log). All other file descriptors should be marked
 244         // as close-on-exec, and should thus not leak into the child.
 245         posix_spawn_file_actions_t actions;
 246         posix_spawn_file_actions_init(&actions);
 247         posix_spawn_file_actions_adddup2(&actions, devnullfd, 0);
 248         posix_spawn_file_actions_adddup2(&actions, pipefd[1], 1);
 249
 250         pid_t child_pid;
 251         char * const argv[] = {
 252                 strdup("/bin/sh"),
 253                 strdup("-c"),
 254                 strdup(path.c_str()),
 255                 nullptr
 256         };
 257         int err = posix_spawn(&child_pid, "/bin/sh", &actions, /*attrp=*/nullptr, argv, /*envp=*/nullptr);
 258         posix_spawn_file_actions_destroy(&actions);
 259         free(argv[0]);
 260         free(argv[1]);
 261         free(argv[2]);
 262         close(pipefd[1]);
 263
 264         if (err == 0) {
 265                 return pipefd[0];
 266         } else {
 267                 log_perror(cmdline.c_str());
 268                 close(pipefd[0]);
 269                 return -1;
 270         }
 271 }
 272
 273 bool HTTPInput::parse_response(const string &request)
 274 {
 275         vector<string> lines = split_lines(response);
 276         if (lines.empty()) {
 277                 if (!suppress_logging) {
 278                         log(WARNING, "[%s] Empty HTTP response from input.", url.c_str());
 279                 }
 280                 return false;
 281         }
 282
 283         vector<string> first_line_tokens = split_tokens(lines[0]);
 284         if (first_line_tokens.size() < 2) {
 285                 if (!suppress_logging) {
 286                         log(WARNING, "[%s] Malformed response line '%s' from input.",
 287                                 url.c_str(), lines[0].c_str());
 288                 }
 289                 return false;
 290         }
 291
 292         int response = atoi(first_line_tokens[1].c_str());
 293         if (response != 200) {
 294                 if (!suppress_logging) {
 295                         log(WARNING, "[%s] Non-200 response '%s' from input.",
 296                                 url.c_str(), lines[0].c_str());
 297                 }
 298                 return false;
 299         }
 300
 301         HTTPHeaderMultimap parameters = extract_headers(lines, url);
 302
 303         // Remove “Content-encoding: metacube”.
 304         const auto encoding_it = parameters.find("Content-Encoding");
 305         if (encoding_it != parameters.end() && encoding_it->second == "metacube") {
 306                 parameters.erase(encoding_it);
 307         }
 308
 309         // Change “Server: foo” to “Server: metacube/0.1 (reflecting: foo)”
 310         // XXX: Use a Via: instead?
 311         if (parameters.count("Server") == 0) {
 312                 parameters.insert(make_pair("Server", SERVER_IDENTIFICATION));
 313         } else {
 314                 for (auto &key_and_value : parameters) {
 315                         if (key_and_value.first != "Server") {
 316                                 continue;
 317                         }
 318                         key_and_value.second = SERVER_IDENTIFICATION " (reflecting: " + key_and_value.second + ")";
 319                 }
 320         }
 321
 322         // Erase “Connection: close”; we'll set it on the sending side if needed.
 323         parameters.erase("Connection");
 324
 325         // Construct the new HTTP header.
 326         http_header = "HTTP/1.0 200 OK\r\n";
 327         for (const auto &key_and_value : parameters) {
 328                 http_header.append(key_and_value.first + ": " + key_and_value.second + "\r\n");
 329         }
 330
 331         for (int stream_index : stream_indices) {
 332                 servers->set_header(stream_index, http_header, stream_header);
 333         }
 334
 335         return true;
 336 }
 337
 338 void HTTPInput::do_work()
 339 {
 340         timespec last_activity;
 341
 342         // TODO: Make the timeout persist across restarts.
 343         if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 344                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 345                 assert(err != -1);
 346         }
 347
 348         while (!should_stop()) {
 349                 if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 350                         // Give the socket 30 seconds since last activity before we time out.
 351                         static const int timeout_secs = 30;
 352
 353                         timespec now;
 354                         int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &now);
 355                         assert(err != -1);
 356
 357                         timespec elapsed = clock_diff(last_activity, now);
 358                         if (elapsed.tv_sec >= timeout_secs) {
 359                                 // Timeout!
 360                                 if (!suppress_logging) {
 361                                         log(ERROR, "[%s] Timeout after %d seconds, closing.", url.c_str(), elapsed.tv_sec);
 362                                 }
 363                                 state = CLOSING_SOCKET;
 364                                 continue;
 365                         }
 366
 367                         // Basically calculate (30 - (now - last_activity)) = (30 + (last_activity - now)).
 368                         // Add a second of slack to account for differences between clocks.
 369                         timespec timeout = clock_diff(now, last_activity);
 370                         timeout.tv_sec += timeout_secs + 1;
 371                         assert(timeout.tv_sec > 0 || (timeout.tv_sec >= 0 && timeout.tv_nsec > 0));
 372
 373                         bool activity = wait_for_activity(sock, (state == SENDING_REQUEST) ? POLLOUT : POLLIN, &timeout);
 374                         if (activity) {
 375                                 err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 376                                 assert(err != -1);
 377                         } else {
 378                                 // OK. Most likely, should_stop was set, or we have timed out.
 379                                 continue;
 380                         }
 381                 }
 382
 383                 switch (state) {
 384                 case NOT_CONNECTED: {
 385                         // Reap any exited children.
 386                         int wstatus;
 387                         while (waitpid(-1, &wstatus, WNOHANG) != -1 || errno == EINTR) ;
 388
 389                         request.clear();
 390                         request_bytes_sent = 0;
 391                         response.clear();
 392                         pending_data.clear();
 393                         has_metacube_header = false;
 394                         for (int stream_index : stream_indices) {
 395                                 // Don't zero out the header; it might still be of use to HLS clients.
 396                                 servers->set_unavailable(stream_index);
 397                         }
 398
 399                         string protocol;
 400                         {
 401                                 string user;  // Thrown away.
 402                                 if (!parse_url(url, &protocol, &user, &host, &port, &path)) {
 403                                         if (!suppress_logging) {
 404                                                 log(WARNING, "[%s] Failed to parse URL '%s'", url.c_str(), url.c_str());
 405                                         }
 406                                         break;
 407                                 }
 408
 409                                 // Remove the brackets around IPv6 address literals.
 410                                 // TODO: See if we can join this with the code in parse_ip_address(),
 411                                 // or maybe even more it into parse_url().
 412                                 if (!host.empty() && host[0] == '[' && host[host.size() - 1] == ']') {
 413                                         host = host.substr(1, host.size() - 2);
 414                                 }
 415                         }
 416
 417                         if (suppress_logging) {
 418                                 // See if there's more than one minute since last time we made a connection
 419                                 // with logging enabled. If so, turn it on again.
 420                                 timespec now;
 421                                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &now);
 422                                 assert(err != -1);
 423
 424                                 double elapsed = now.tv_sec - last_verbose_connection.tv_sec +
 425                                         1e-9 * (now.tv_nsec - last_verbose_connection.tv_nsec);
 426                                 if (elapsed > 60.0) {
 427                                         suppress_logging = false;
 428                                 }
 429                         }
 430                         if (!suppress_logging) {
 431                                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_verbose_connection);
 432                                 assert(err != -1);
 433                         }
 434                         ++num_connection_attempts;
 435                         if (protocol == "pipe") {
 436                                 sock = open_child_process(path.c_str());
 437
 438                                 if (sock != -1) {
 439                                         // Construct a minimal HTTP header.
 440                                         http_header = "HTTP/1.0 200 OK\r\n";
 441                                         for (int stream_index : stream_indices) {
 442                                                 servers->set_header(stream_index, http_header, stream_header);
 443                                         }
 444                                         state = RECEIVING_DATA;
 445                                 }
 446                         } else {
 447                                 sock = lookup_and_connect(host, port);
 448                                 if (sock != -1) {
 449                                         // Yay, successful connect.
 450                                         state = SENDING_REQUEST;
 451                                         request = "GET " + path + " HTTP/1.0\r\nHost: " + host_header(host, port) + "\r\nUser-Agent: cubemap\r\n\r\n";
 452                                         request_bytes_sent = 0;
 453                                 }
 454                         }
 455                         if (sock != -1) {
 456                                 lock_guard<mutex> lock(stats_mutex);
 457                                 stats.connect_time = time(nullptr);
 458                                 clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 459                         }
 460                         break;
 461                 }
 462                 case SENDING_REQUEST: {
 463                         size_t to_send = request.size() - request_bytes_sent;
 464                         int ret;
 465
 466                         do {
 467                                 ret = write(sock, request.data() + request_bytes_sent, to_send);
 468                         } while (ret == -1 && errno == EINTR);
 469
 470                         if (ret == -1) {
 471                                 log_perror("write");
 472                                 state = CLOSING_SOCKET;
 473                                 continue;
 474                         }
 475
 476                         assert(ret >= 0);
 477                         request_bytes_sent += ret;
 478
 479                         if (request_bytes_sent == request.size()) {
 480                                 state = RECEIVING_HEADER;
 481                         }
 482                         break;
 483                 }
 484                 case RECEIVING_HEADER: {
 485                         char buf[4096];
 486                         int ret;
 487
 488                         do {
 489                                 ret = read(sock, buf, sizeof(buf));
 490                         } while (ret == -1 && errno == EINTR);
 491
 492                         if (ret == -1) {
 493                                 log_perror("read");
 494                                 state = CLOSING_SOCKET;
 495                                 continue;
 496                         }
 497
 498                         if (ret == 0) {
 499                                 // This really shouldn't happen...
 500                                 if (!suppress_logging) {
 501                                         log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 502                                                    url.c_str());
 503                                 }
 504                                 state = CLOSING_SOCKET;
 505                                 continue;
 506                         }
 507
 508                         RequestParseStatus status = wait_for_double_newline(&response, buf, ret);
 509
 510                         if (status == RP_OUT_OF_SPACE) {
 511                                 if (!suppress_logging) {
 512                                         log(WARNING, "[%s] Server sent overlong HTTP response!", url.c_str());
 513                                 }
 514                                 state = CLOSING_SOCKET;
 515                                 continue;
 516                         } else if (status == RP_NOT_FINISHED_YET) {
 517                                 continue;
 518                         }
 519
 520                         // OK, so we're fine, but there might be some of the actual data after the response.
 521                         // We'll need to deal with that separately.
 522                         string extra_data;
 523                         if (status == RP_EXTRA_DATA) {
 524                                 char *ptr = static_cast<char *>(
 525                                         memmem(response.data(), response.size(), "\r\n\r\n", 4));
 526                                 assert(ptr != nullptr);
 527                                 extra_data = string(ptr + 4, &response[0] + response.size());
 528                                 response.resize(ptr - response.data());
 529                         }
 530
 531                         if (!parse_response(response)) {
 532                                 state = CLOSING_SOCKET;
 533                                 continue;
 534                         }
 535
 536                         if (!extra_data.empty()) {
 537                                 process_data(&extra_data[0], extra_data.size());
 538                         }
 539
 540                         if (!suppress_logging) {
 541                                 if (encoding == Input::INPUT_ENCODING_RAW) {
 542                                         log(INFO, "[%s] Connected to '%s', receiving raw data.",
 543                                                    url.c_str(), url.c_str());
 544                                 } else {
 545                                         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 546                                         log(INFO, "[%s] Connected to '%s', receiving data.",
 547                                                    url.c_str(), url.c_str());
 548                                 }
 549                         }
 550                         state = RECEIVING_DATA;
 551                         break;
 552                 }
 553                 case RECEIVING_DATA: {
 554                         char buf[4096];
 555                         int ret;
 556
 557                         do {
 558                                 ret = read(sock, buf, sizeof(buf));
 559                         } while (ret == -1 && errno == EINTR);
 560
 561                         if (ret == -1) {
 562                                 log_perror("read");
 563                                 state = CLOSING_SOCKET;
 564                                 continue;
 565                         }
 566
 567                         if (ret == 0) {
 568                                 // This really shouldn't happen...
 569                                 if (!suppress_logging) {
 570                                         log(ERROR, "[%s] Socket unexpectedly closed while reading data",
 571                                                    url.c_str());
 572                                 }
 573                                 state = CLOSING_SOCKET;
 574                                 continue;
 575                         }
 576
 577                         num_connection_attempts = 0;  // Reset, since we have a successful read.
 578                         if (suppress_logging) {
 579                                 // This was suppressed earlier, so print it out now.
 580                                 if (encoding == Input::INPUT_ENCODING_RAW) {
 581                                         log(INFO, "[%s] Connected to '%s', receiving raw data.",
 582                                                    url.c_str(), url.c_str());
 583                                 } else {
 584                                         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 585                                         log(INFO, "[%s] Connected to '%s', receiving data.",
 586                                                    url.c_str(), url.c_str());
 587                                 }
 588                                 suppress_logging = false;
 589                         }
 590
 591                         process_data(buf, ret);
 592                         break;
 593                 }
 594                 case CLOSING_SOCKET: {
 595                         close_socket();
 596                         state = NOT_CONNECTED;
 597                         break;
 598                 }
 599                 default:
 600                         assert(false);
 601                 }
 602
 603                 // If we are still in NOT_CONNECTED, either something went wrong,
 604                 // or the connection just got closed.
 605                 // The earlier steps have already given the error message, if any.
 606                 if (state == NOT_CONNECTED && !should_stop()) {
 607                         if (!suppress_logging) {
 608                                 log(INFO, "[%s] Waiting 0.2 seconds and restarting...", url.c_str());
 609                         }
 610
 611                         if (num_connection_attempts >= 3 && !suppress_logging) {
 612                                 log(INFO, "[%s] %d failed connection attempts, suppressing logging for one minute.",
 613                                         url.c_str(), num_connection_attempts);
 614                                 suppress_logging = true;
 615                         }
 616                         timespec timeout_ts;
 617                         timeout_ts.tv_sec = 0;
 618                         timeout_ts.tv_nsec = 200000000;
 619                         wait_for_wakeup(&timeout_ts);
 620                 }
 621         }
 622 }
 623
 624 void HTTPInput::process_data(char *ptr, size_t bytes)
 625 {
 626         {
 627                 lock_guard<mutex> lock(stats_mutex);
 628                 stats.bytes_received += bytes;
 629         }
 630
 631         if (encoding == Input::INPUT_ENCODING_RAW) {
 632                 for (int stream_index : stream_indices) {
 633                         servers->add_data(stream_index, ptr, bytes, /*metacube_flags=*/0, /*pts=*/RationalPTS());
 634                 }
 635                 return;
 636         }
 637
 638         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 639         pending_data.insert(pending_data.end(), ptr, ptr + bytes);
 640
 641         for ( ;; ) {
 642                 // If we don't have enough data (yet) for even the Metacube header, just return.
 643                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 644                         return;
 645                 }
 646
 647                 // Make sure we have the Metacube sync header at the start.
 648                 // We may need to skip over junk data (it _should_ not happen, though).
 649                 if (!has_metacube_header) {
 650                         char *ptr = static_cast<char *>(
 651                                 memmem(pending_data.data(), pending_data.size(),
 652                                        METACUBE2_SYNC, strlen(METACUBE2_SYNC)));
 653                         if (ptr == nullptr) {
 654                                 // OK, so we didn't find the sync marker. We know then that
 655                                 // we do not have the _full_ marker in the buffer, but we
 656                                 // could have N-1 bytes. Drop everything before that,
 657                                 // and then give up.
 658                                 drop_pending_data(pending_data.size() - (strlen(METACUBE2_SYNC) - 1));
 659                                 return;
 660                         } else {
 661                                 // Yay, we found the header. Drop everything (if anything) before it.
 662                                 drop_pending_data(ptr - pending_data.data());
 663                                 has_metacube_header = true;
 664
 665                                 // Re-check that we have the entire header; we could have dropped data.
 666                                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 667                                         return;
 668                                 }
 669                         }
 670                 }
 671
 672                 // Now it's safe to read the header.
 673                 metacube2_block_header hdr;
 674                 memcpy(&hdr, pending_data.data(), sizeof(hdr));
 675                 assert(memcmp(hdr.sync, METACUBE2_SYNC, sizeof(hdr.sync)) == 0);
 676                 uint32_t size = ntohl(hdr.size);
 677                 uint16_t flags = ntohs(hdr.flags);
 678                 uint16_t expected_csum = metacube2_compute_crc(&hdr);
 679
 680                 if (expected_csum != ntohs(hdr.csum)) {
 681                         log(WARNING, "[%s] Metacube checksum failed (expected 0x%x, got 0x%x), "
 682                                 "not reading block claiming to be %d bytes (flags=%x).",
 683                                 url.c_str(), expected_csum, ntohs(hdr.csum),
 684                                 size, flags);
 685
 686                         // Drop only the first byte, and let the rest of the code handle resync.
 687                         pending_data.erase(pending_data.begin(), pending_data.begin() + 1);
 688                         has_metacube_header = false;
 689                         continue;
 690                 }
 691                 if (size > 10485760) {
 692                         log(WARNING, "[%s] Metacube block of %d bytes (flags=%x); corrupted header??",
 693                                 url.c_str(), size, flags);
 694                 }
 695
 696                 // See if we have the entire block. If not, wait for more data.
 697                 if (pending_data.size() < sizeof(metacube2_block_header) + size) {
 698                         return;
 699                 }
 700
 701                 // See if this is a metadata block. If so, we don't want to send it on,
 702                 // but rather process it ourselves.
 703                 // TODO: Keep metadata when sending on to other Metacube users.
 704                 if (flags & METACUBE_FLAGS_METADATA) {
 705                         {
 706                                 lock_guard<mutex> lock(stats_mutex);
 707                                 stats.metadata_bytes_received += size;
 708                         }
 709                         process_metacube_metadata_block(hdr, pending_data.data() + sizeof(hdr), size);
 710                 } else {
 711                         // Send this block on to the servers.
 712                         {
 713                                 lock_guard<mutex> lock(stats_mutex);
 714                                 stats.data_bytes_received += size;
 715                         }
 716                         char *inner_data = pending_data.data() + sizeof(metacube2_block_header);
 717                         if (flags & METACUBE_FLAGS_HEADER) {
 718                                 stream_header = string(inner_data, inner_data + size);
 719                                 for (int stream_index : stream_indices) {
 720                                         servers->set_header(stream_index, http_header, stream_header);
 721                                 }
 722                         }
 723                         for (int stream_index : stream_indices) {
 724                                 servers->add_data(stream_index, inner_data, size, flags, next_block_pts);
 725                         }
 726                         next_block_pts = RationalPTS();
 727                 }
 728
 729                 // Consume the block. This isn't the most efficient way of dealing with things
 730                 // should we have many blocks, but these routines don't need to be too efficient
 731                 // anyway.
 732                 pending_data.erase(pending_data.begin(), pending_data.begin() + sizeof(metacube2_block_header) + size);
 733                 has_metacube_header = false;
 734         }
 735 }
 736
 737 void HTTPInput::drop_pending_data(size_t num_bytes)
 738 {
 739         if (num_bytes == 0) {
 740                 return;
 741         }
 742         log(WARNING, "[%s] Dropping %lld junk bytes; not a Metacube2 stream, or data was dropped from the middle of the stream.",
 743                 url.c_str(), (long long)num_bytes);
 744         assert(pending_data.size() >= num_bytes);
 745         pending_data.erase(pending_data.begin(), pending_data.begin() + num_bytes);
 746 }
 747
 748 void HTTPInput::add_destination(int stream_index)
 749 {
 750         stream_indices.push_back(stream_index);
 751         servers->set_header(stream_index, http_header, stream_header);
 752 }
 753
 754 InputStats HTTPInput::get_stats() const
 755 {
 756         lock_guard<mutex> lock(stats_mutex);
 757         return stats;
 758 }
 759
 760 void HTTPInput::process_metacube_metadata_block(const metacube2_block_header &hdr, const char *payload, uint32_t payload_size)
 761 {
 762         if (payload_size < sizeof(uint64_t)) {
 763                 log(WARNING, "[%s] Undersized Metacube metadata block (%d bytes); corrupted header?",
 764                         url.c_str(), payload_size);
 765                 return;
 766         }
 767
 768         uint64_t type = be64toh(*(const uint64_t *)payload);
 769         if (type == METACUBE_METADATA_TYPE_ENCODER_TIMESTAMP) {
 770                 timespec now;
 771                 clock_gettime(CLOCK_REALTIME, &now);
 772
 773                 const metacube2_timestamp_packet *pkt = (const metacube2_timestamp_packet *)payload;
 774                 if (payload_size != sizeof(*pkt)) {
 775                         log(WARNING, "[%s] Metacube timestamp block of wrong size (%d bytes); ignoring.",
 776                                 url.c_str(), payload_size);
 777                         return;
 778                 }
 779
 780                 double elapsed = now.tv_sec - be64toh(pkt->tv_sec) +
 781                         1e-9 * (now.tv_nsec - long(be64toh(pkt->tv_nsec)));
 782                 {
 783                         lock_guard<mutex> lock(stats_mutex);
 784                         stats.latency_sec = elapsed;
 785                 }
 786         } else if (type == METACUBE_METADATA_TYPE_NEXT_BLOCK_PTS) {
 787                 const metacube2_pts_packet *pkt = (const metacube2_pts_packet *)payload;
 788                 if (payload_size != sizeof(*pkt)) {
 789                         log(WARNING, "[%s] Metacube pts block of wrong size (%d bytes); ignoring.",
 790                                 url.c_str(), payload_size);
 791                         return;
 792                 }
 793                 next_block_pts.pts = be64toh(pkt->pts);
 794                 next_block_pts.timebase_num = be64toh(pkt->timebase_num);
 795                 next_block_pts.timebase_den = be64toh(pkt->timebase_den);
 796         } else {
 797                 // Unknown metadata block, ignore
 798                 log(INFO, "[%s] Metadata block %llu\n", url.c_str(), type);
 799                 return;
 800         }
 801 }