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           child_pid(serialized.child_pid())
  80 {
  81         // Set back the close-on-exec flag for the socket.
  82         // (This can't leak into a child, since we haven't been started yet.)
  83         fcntl(sock, F_SETFD, O_CLOEXEC);
  84
  85         pending_data.resize(serialized.pending_data().size());
  86         memcpy(&pending_data[0], serialized.pending_data().data(), serialized.pending_data().size());
  87
  88         string protocol, user;
  89         parse_url(url, &protocol, &user, &host, &port, &path);  // Don't care if it fails.
  90
  91         stats.url = url;
  92         stats.bytes_received = serialized.bytes_received();
  93         stats.data_bytes_received = serialized.data_bytes_received();
  94         stats.metadata_bytes_received = serialized.metadata_bytes_received();
  95         if (serialized.has_connect_time()) {
  96                 stats.connect_time = serialized.connect_time();
  97         } else {
  98                 stats.connect_time = time(nullptr);
  99         }
 100         if (serialized.has_latency_sec()) {
 101                 stats.latency_sec = serialized.latency_sec();
 102         } else {
 103                 stats.latency_sec = HUGE_VAL;
 104         }
 105
 106         last_verbose_connection.tv_sec = -3600;
 107         last_verbose_connection.tv_nsec = 0;
 108 }
 109
 110 void HTTPInput::close_socket()
 111 {
 112         if (sock != -1) {
 113                 safe_close(sock);
 114                 sock = -1;
 115         }
 116         if (child_pid != -1) {
 117                 // Kill the child process group, forcibly.
 118                 // TODO: Consider using a pidfd on newer kernels, so that we're guaranteed
 119                 // never to kill the wrong process.
 120                 kill(-child_pid, SIGKILL);
 121         }
 122         child_pid = -1;
 123
 124         lock_guard<mutex> lock(stats_mutex);
 125         stats.connect_time = -1;
 126 }
 127
 128 InputProto HTTPInput::serialize() const
 129 {
 130         // Unset the close-on-exec flag for the socket.
 131         // (This can't leak into a child, since there's only one thread left.)
 132         fcntl(sock, F_SETFD, 0);
 133
 134         InputProto serialized;
 135         serialized.set_state(state);
 136         serialized.set_url(url);
 137         serialized.set_request(request);
 138         serialized.set_request_bytes_sent(request_bytes_sent);
 139         serialized.set_response(response);
 140         serialized.set_http_header(http_header);
 141         serialized.set_stream_header(stream_header);
 142         serialized.set_pending_data(string(pending_data.begin(), pending_data.end()));
 143         serialized.set_has_metacube_header(has_metacube_header);
 144         serialized.set_sock(sock);
 145         serialized.set_child_pid(child_pid);
 146         serialized.set_bytes_received(stats.bytes_received);
 147         serialized.set_data_bytes_received(stats.data_bytes_received);
 148         if (isfinite(stats.latency_sec)) {
 149                 serialized.set_latency_sec(stats.latency_sec);
 150         }
 151         serialized.set_connect_time(stats.connect_time);
 152         if (encoding == Input::INPUT_ENCODING_METACUBE) {
 153                 serialized.set_is_metacube_encoded(true);
 154         } else {
 155                 assert(encoding == Input::INPUT_ENCODING_RAW);
 156                 serialized.set_is_metacube_encoded(false);
 157         }
 158         return serialized;
 159 }
 160
 161 int HTTPInput::lookup_and_connect(const string &host, const string &port)
 162 {
 163         addrinfo *ai;
 164         int err = getaddrinfo(host.c_str(), port.c_str(), nullptr, &ai);
 165         if (err != 0) {
 166                 if (!suppress_logging) {
 167                         log(WARNING, "[%s] Lookup of '%s' failed (%s).",
 168                                 url.c_str(), host.c_str(), gai_strerror(err));
 169                 }
 170                 freeaddrinfo(ai);
 171                 return -1;
 172         }
 173
 174         addrinfo *base_ai = ai;
 175
 176         // Connect to everything in turn until we have a socket.
 177         for ( ; ai && !should_stop(); ai = ai->ai_next) {
 178                 // Now do a non-blocking connect. This is important because we want to be able to be
 179                 // woken up, even though it's rather cumbersome.
 180                 int sock = socket(ai->ai_family, SOCK_STREAM | SOCK_NONBLOCK | SOCK_CLOEXEC, IPPROTO_TCP);
 181                 if (sock == -1) {
 182                         // Could be e.g. EPROTONOSUPPORT. The show must go on.
 183                         continue;
 184                 }
 185
 186                 // Do a non-blocking connect.
 187                 do {
 188                         err = connect(sock, ai->ai_addr, ai->ai_addrlen);
 189                 } while (err == -1 && errno == EINTR);
 190
 191                 if (err == -1 && errno != EINPROGRESS) {
 192                         log_perror("connect");
 193                         safe_close(sock);
 194                         continue;
 195                 }
 196
 197                 // Wait for the connect to complete, or an error to happen.
 198                 for ( ;; ) {
 199                         bool complete = wait_for_activity(sock, POLLIN | POLLOUT, nullptr);
 200                         if (should_stop()) {
 201                                 safe_close(sock);
 202                                 freeaddrinfo(base_ai);
 203                                 return -1;
 204                         }
 205                         if (complete) {
 206                                 break;
 207                         }
 208                 }
 209
 210                 // Check whether it ended in an error or not.
 211                 socklen_t err_size = sizeof(err);
 212                 if (getsockopt(sock, SOL_SOCKET, SO_ERROR, &err, &err_size) == -1) {
 213                         log_perror("getsockopt");
 214                         safe_close(sock);
 215                         continue;
 216                 }
 217
 218                 errno = err;
 219
 220                 if (err == 0) {
 221                         // Successful connect.
 222                         freeaddrinfo(base_ai);
 223                         return sock;
 224                 }
 225
 226                 safe_close(sock);
 227         }
 228
 229         // Give the last one as error.
 230         if (!suppress_logging) {
 231                 log(WARNING, "[%s] Connect to '%s' failed (%s)",
 232                         url.c_str(), host.c_str(), strerror(errno));
 233         }
 234         freeaddrinfo(base_ai);
 235         return -1;
 236 }
 237
 238 int HTTPInput::open_child_process(const string &cmdline)
 239 {
 240         int devnullfd = open("/dev/null", O_RDONLY | O_CLOEXEC);
 241         if (devnullfd == -1) {
 242                 log_perror("/dev/null");
 243                 return -1;
 244         }
 245
 246         int pipefd[2];
 247         if (pipe2(pipefd, O_CLOEXEC) == -1) {
 248                 log_perror("pipe2()");
 249                 close(devnullfd);
 250                 return -1;
 251         }
 252
 253         // Point stdout to us, stdin to /dev/null, and stderr remains where it is
 254         // (probably the systemd log). All other file descriptors should be marked
 255         // as close-on-exec, and should thus not leak into the child.
 256         posix_spawn_file_actions_t actions;
 257         posix_spawn_file_actions_init(&actions);
 258         posix_spawn_file_actions_adddup2(&actions, devnullfd, 0);
 259         posix_spawn_file_actions_adddup2(&actions, pipefd[1], 1);
 260
 261         // Make the process a leader of its own process group, so that we can easily
 262         // kill it and any of its child processes (unless it's started new process
 263         // groups itself, of course).
 264         posix_spawnattr_t attr;
 265         posix_spawnattr_init(&attr);
 266         posix_spawnattr_setflags(&attr, POSIX_SPAWN_SETPGROUP);
 267         posix_spawnattr_setpgroup(&attr, 0);
 268
 269         char * const argv[] = {
 270                 strdup("/bin/sh"),
 271                 strdup("-c"),
 272                 strdup(path.c_str()),
 273                 nullptr
 274         };
 275         int err = posix_spawn(&child_pid, "/bin/sh", &actions, &attr, argv, /*envp=*/nullptr);
 276         posix_spawn_file_actions_destroy(&actions);
 277         posix_spawnattr_destroy(&attr);
 278         free(argv[0]);
 279         free(argv[1]);
 280         free(argv[2]);
 281         close(devnullfd);
 282         close(pipefd[1]);
 283
 284         if (err == 0) {
 285                 return pipefd[0];
 286         } else {
 287                 child_pid = -1;
 288                 log_perror(cmdline.c_str());
 289                 close(pipefd[0]);
 290                 return -1;
 291         }
 292 }
 293
 294 bool HTTPInput::parse_response(const string &request)
 295 {
 296         vector<string> lines = split_lines(response);
 297         if (lines.empty()) {
 298                 if (!suppress_logging) {
 299                         log(WARNING, "[%s] Empty HTTP response from input.", url.c_str());
 300                 }
 301                 return false;
 302         }
 303
 304         vector<string> first_line_tokens = split_tokens(lines[0]);
 305         if (first_line_tokens.size() < 2) {
 306                 if (!suppress_logging) {
 307                         log(WARNING, "[%s] Malformed response line '%s' from input.",
 308                                 url.c_str(), lines[0].c_str());
 309                 }
 310                 return false;
 311         }
 312
 313         int response = atoi(first_line_tokens[1].c_str());
 314         if (response != 200) {
 315                 if (!suppress_logging) {
 316                         log(WARNING, "[%s] Non-200 response '%s' from input.",
 317                                 url.c_str(), lines[0].c_str());
 318                 }
 319                 return false;
 320         }
 321
 322         HTTPHeaderMultimap parameters = extract_headers(lines, url);
 323
 324         // Remove “Content-encoding: metacube”.
 325         const auto encoding_it = parameters.find("Content-Encoding");
 326         if (encoding_it != parameters.end() && encoding_it->second == "metacube") {
 327                 parameters.erase(encoding_it);
 328         }
 329
 330         // Change “Server: foo” to “Server: metacube/0.1 (reflecting: foo)”
 331         // XXX: Use a Via: instead?
 332         if (parameters.count("Server") == 0) {
 333                 parameters.insert(make_pair("Server", SERVER_IDENTIFICATION));
 334         } else {
 335                 for (auto &key_and_value : parameters) {
 336                         if (key_and_value.first != "Server") {
 337                                 continue;
 338                         }
 339                         key_and_value.second = SERVER_IDENTIFICATION " (reflecting: " + key_and_value.second + ")";
 340                 }
 341         }
 342
 343         // Erase “Connection: close”; we'll set it on the sending side if needed.
 344         parameters.erase("Connection");
 345
 346         // Construct the new HTTP header.
 347         http_header = "HTTP/1.0 200 OK\r\n";
 348         for (const auto &key_and_value : parameters) {
 349                 http_header.append(key_and_value.first + ": " + key_and_value.second + "\r\n");
 350         }
 351
 352         for (int stream_index : stream_indices) {
 353                 servers->set_header(stream_index, http_header, stream_header);
 354         }
 355
 356         return true;
 357 }
 358
 359 void HTTPInput::do_work()
 360 {
 361         timespec last_activity;
 362
 363         // TODO: Make the timeout persist across restarts.
 364         if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 365                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 366                 assert(err != -1);
 367         }
 368
 369         while (!should_stop()) {
 370                 if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 371                         // Give the socket 30 seconds since last activity before we time out.
 372                         static const int timeout_secs = 30;
 373
 374                         timespec now;
 375                         int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &now);
 376                         assert(err != -1);
 377
 378                         timespec elapsed = clock_diff(last_activity, now);
 379                         if (elapsed.tv_sec >= timeout_secs) {
 380                                 // Timeout!
 381                                 if (!suppress_logging) {
 382                                         log(ERROR, "[%s] Timeout after %d seconds, closing.", url.c_str(), elapsed.tv_sec);
 383                                 }
 384                                 state = CLOSING_SOCKET;
 385                                 continue;
 386                         }
 387
 388                         // Basically calculate (30 - (now - last_activity)) = (30 + (last_activity - now)).
 389                         // Add a second of slack to account for differences between clocks.
 390                         timespec timeout = clock_diff(now, last_activity);
 391                         timeout.tv_sec += timeout_secs + 1;
 392                         assert(timeout.tv_sec > 0 || (timeout.tv_sec >= 0 && timeout.tv_nsec > 0));
 393
 394                         bool activity = wait_for_activity(sock, (state == SENDING_REQUEST) ? POLLOUT : POLLIN, &timeout);
 395                         if (activity) {
 396                                 err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 397                                 assert(err != -1);
 398                         } else {
 399                                 // OK. Most likely, should_stop was set, or we have timed out.
 400                                 continue;
 401                         }
 402                 }
 403
 404                 switch (state) {
 405                 case NOT_CONNECTED: {
 406                         // Reap any exited children.
 407                         int wstatus, err;
 408                         do {
 409                                 err = waitpid(-1, &wstatus, WNOHANG);
 410                                 if (err == -1) {
 411                                         if (errno == EINTR) {
 412                                                 continue;
 413                                         }
 414                                         if (errno == ECHILD) {
 415                                                 break;
 416                                         }
 417                                         log_perror("waitpid");
 418                                         break;
 419                                 }
 420                         } while (err != 0);
 421                         child_pid = -1;
 422
 423                         request.clear();
 424                         request_bytes_sent = 0;
 425                         response.clear();
 426                         pending_data.clear();
 427                         has_metacube_header = false;
 428                         for (int stream_index : stream_indices) {
 429                                 // Don't zero out the header; it might still be of use to HLS clients.
 430                                 servers->set_unavailable(stream_index);
 431                         }
 432
 433                         string protocol;
 434                         {
 435                                 string user;  // Thrown away.
 436                                 if (!parse_url(url, &protocol, &user, &host, &port, &path)) {
 437                                         if (!suppress_logging) {
 438                                                 log(WARNING, "[%s] Failed to parse URL '%s'", url.c_str(), url.c_str());
 439                                         }
 440                                         break;
 441                                 }
 442
 443                                 // Remove the brackets around IPv6 address literals.
 444                                 // TODO: See if we can join this with the code in parse_ip_address(),
 445                                 // or maybe even more it into parse_url().
 446                                 if (!host.empty() && host[0] == '[' && host[host.size() - 1] == ']') {
 447                                         host = host.substr(1, host.size() - 2);
 448                                 }
 449                         }
 450
 451                         if (suppress_logging) {
 452                                 // See if there's more than one minute since last time we made a connection
 453                                 // with logging enabled. If so, turn it on again.
 454                                 timespec now;
 455                                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &now);
 456                                 assert(err != -1);
 457
 458                                 double elapsed = now.tv_sec - last_verbose_connection.tv_sec +
 459                                         1e-9 * (now.tv_nsec - last_verbose_connection.tv_nsec);
 460                                 if (elapsed > 60.0) {
 461                                         suppress_logging = false;
 462                                 }
 463                         }
 464                         if (!suppress_logging) {
 465                                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_verbose_connection);
 466                                 assert(err != -1);
 467                         }
 468                         ++num_connection_attempts;
 469                         if (protocol == "pipe") {
 470                                 sock = open_child_process(path.c_str());
 471
 472                                 if (sock != -1) {
 473                                         // Construct a minimal HTTP header.
 474                                         http_header = "HTTP/1.0 200 OK\r\n";
 475                                         for (int stream_index : stream_indices) {
 476                                                 servers->set_header(stream_index, http_header, stream_header);
 477                                         }
 478                                         state = RECEIVING_DATA;
 479                                 }
 480                         } else {
 481                                 sock = lookup_and_connect(host, port);
 482                                 if (sock != -1) {
 483                                         // Yay, successful connect.
 484                                         state = SENDING_REQUEST;
 485                                         request = "GET " + path + " HTTP/1.0\r\nHost: " + host_header(host, port) + "\r\nUser-Agent: cubemap\r\n\r\n";
 486                                         request_bytes_sent = 0;
 487                                 }
 488                         }
 489                         if (sock != -1) {
 490                                 lock_guard<mutex> lock(stats_mutex);
 491                                 stats.connect_time = time(nullptr);
 492                                 clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 493                         }
 494                         break;
 495                 }
 496                 case SENDING_REQUEST: {
 497                         size_t to_send = request.size() - request_bytes_sent;
 498                         int ret;
 499
 500                         do {
 501                                 ret = write(sock, request.data() + request_bytes_sent, to_send);
 502                         } while (ret == -1 && errno == EINTR);
 503
 504                         if (ret == -1) {
 505                                 log_perror("write");
 506                                 state = CLOSING_SOCKET;
 507                                 continue;
 508                         }
 509
 510                         assert(ret >= 0);
 511                         request_bytes_sent += ret;
 512
 513                         if (request_bytes_sent == request.size()) {
 514                                 state = RECEIVING_HEADER;
 515                         }
 516                         break;
 517                 }
 518                 case RECEIVING_HEADER: {
 519                         char buf[4096];
 520                         int ret;
 521
 522                         do {
 523                                 ret = read(sock, buf, sizeof(buf));
 524                         } while (ret == -1 && errno == EINTR);
 525
 526                         if (ret == -1) {
 527                                 log_perror("read");
 528                                 state = CLOSING_SOCKET;
 529                                 continue;
 530                         }
 531
 532                         if (ret == 0) {
 533                                 // This really shouldn't happen...
 534                                 if (!suppress_logging) {
 535                                         log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 536                                                    url.c_str());
 537                                 }
 538                                 state = CLOSING_SOCKET;
 539                                 continue;
 540                         }
 541
 542                         RequestParseStatus status = wait_for_double_newline(&response, buf, ret);
 543
 544                         if (status == RP_OUT_OF_SPACE) {
 545                                 if (!suppress_logging) {
 546                                         log(WARNING, "[%s] Server sent overlong HTTP response!", url.c_str());
 547                                 }
 548                                 state = CLOSING_SOCKET;
 549                                 continue;
 550                         } else if (status == RP_NOT_FINISHED_YET) {
 551                                 continue;
 552                         }
 553
 554                         // OK, so we're fine, but there might be some of the actual data after the response.
 555                         // We'll need to deal with that separately.
 556                         string extra_data;
 557                         if (status == RP_EXTRA_DATA) {
 558                                 char *ptr = static_cast<char *>(
 559                                         memmem(response.data(), response.size(), "\r\n\r\n", 4));
 560                                 assert(ptr != nullptr);
 561                                 extra_data = string(ptr + 4, &response[0] + response.size());
 562                                 response.resize(ptr - response.data());
 563                         }
 564
 565                         if (!parse_response(response)) {
 566                                 state = CLOSING_SOCKET;
 567                                 continue;
 568                         }
 569
 570                         if (!extra_data.empty()) {
 571                                 process_data(&extra_data[0], extra_data.size());
 572                         }
 573
 574                         if (!suppress_logging) {
 575                                 if (encoding == Input::INPUT_ENCODING_RAW) {
 576                                         log(INFO, "[%s] Connected to '%s', receiving raw data.",
 577                                                    url.c_str(), url.c_str());
 578                                 } else {
 579                                         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 580                                         log(INFO, "[%s] Connected to '%s', receiving data.",
 581                                                    url.c_str(), url.c_str());
 582                                 }
 583                         }
 584                         state = RECEIVING_DATA;
 585                         break;
 586                 }
 587                 case RECEIVING_DATA: {
 588                         char buf[4096];
 589                         int ret;
 590
 591                         do {
 592                                 ret = read(sock, buf, sizeof(buf));
 593                         } while (ret == -1 && errno == EINTR);
 594
 595                         if (ret == -1) {
 596                                 log_perror("read");
 597                                 state = CLOSING_SOCKET;
 598                                 continue;
 599                         }
 600
 601                         if (ret == 0) {
 602                                 // This really shouldn't happen...
 603                                 if (!suppress_logging) {
 604                                         log(ERROR, "[%s] Socket unexpectedly closed while reading data",
 605                                                    url.c_str());
 606                                 }
 607                                 state = CLOSING_SOCKET;
 608                                 continue;
 609                         }
 610
 611                         num_connection_attempts = 0;  // Reset, since we have a successful read.
 612                         if (suppress_logging) {
 613                                 // This was suppressed earlier, so print it out now.
 614                                 if (encoding == Input::INPUT_ENCODING_RAW) {
 615                                         log(INFO, "[%s] Connected to '%s', receiving raw data.",
 616                                                    url.c_str(), url.c_str());
 617                                 } else {
 618                                         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 619                                         log(INFO, "[%s] Connected to '%s', receiving data.",
 620                                                    url.c_str(), url.c_str());
 621                                 }
 622                                 suppress_logging = false;
 623                         }
 624
 625                         process_data(buf, ret);
 626                         break;
 627                 }
 628                 case CLOSING_SOCKET: {
 629                         close_socket();
 630                         state = NOT_CONNECTED;
 631                         break;
 632                 }
 633                 default:
 634                         assert(false);
 635                 }
 636
 637                 // If we are still in NOT_CONNECTED, either something went wrong,
 638                 // or the connection just got closed.
 639                 // The earlier steps have already given the error message, if any.
 640                 if (state == NOT_CONNECTED && !should_stop()) {
 641                         if (!suppress_logging) {
 642                                 log(INFO, "[%s] Waiting 0.2 seconds and restarting...", url.c_str());
 643                         }
 644
 645                         if (num_connection_attempts >= 3 && !suppress_logging) {
 646                                 log(INFO, "[%s] %d failed connection attempts, suppressing logging for one minute.",
 647                                         url.c_str(), num_connection_attempts);
 648                                 suppress_logging = true;
 649                         }
 650                         timespec timeout_ts;
 651                         timeout_ts.tv_sec = 0;
 652                         timeout_ts.tv_nsec = 200000000;
 653                         wait_for_wakeup(&timeout_ts);
 654                 }
 655         }
 656 }
 657
 658 void HTTPInput::process_data(char *ptr, size_t bytes)
 659 {
 660         {
 661                 lock_guard<mutex> lock(stats_mutex);
 662                 stats.bytes_received += bytes;
 663         }
 664
 665         if (encoding == Input::INPUT_ENCODING_RAW) {
 666                 for (int stream_index : stream_indices) {
 667                         servers->add_data(stream_index, ptr, bytes, /*metacube_flags=*/0, /*pts=*/RationalPTS());
 668                 }
 669                 return;
 670         }
 671
 672         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 673         pending_data.insert(pending_data.end(), ptr, ptr + bytes);
 674
 675         for ( ;; ) {
 676                 // If we don't have enough data (yet) for even the Metacube header, just return.
 677                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 678                         return;
 679                 }
 680
 681                 // Make sure we have the Metacube sync header at the start.
 682                 // We may need to skip over junk data (it _should_ not happen, though).
 683                 if (!has_metacube_header) {
 684                         char *ptr = static_cast<char *>(
 685                                 memmem(pending_data.data(), pending_data.size(),
 686                                        METACUBE2_SYNC, strlen(METACUBE2_SYNC)));
 687                         if (ptr == nullptr) {
 688                                 // OK, so we didn't find the sync marker. We know then that
 689                                 // we do not have the _full_ marker in the buffer, but we
 690                                 // could have N-1 bytes. Drop everything before that,
 691                                 // and then give up.
 692                                 drop_pending_data(pending_data.size() - (strlen(METACUBE2_SYNC) - 1));
 693                                 return;
 694                         } else {
 695                                 // Yay, we found the header. Drop everything (if anything) before it.
 696                                 drop_pending_data(ptr - pending_data.data());
 697                                 has_metacube_header = true;
 698
 699                                 // Re-check that we have the entire header; we could have dropped data.
 700                                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 701                                         return;
 702                                 }
 703                         }
 704                 }
 705
 706                 // Now it's safe to read the header.
 707                 metacube2_block_header hdr;
 708                 memcpy(&hdr, pending_data.data(), sizeof(hdr));
 709                 assert(memcmp(hdr.sync, METACUBE2_SYNC, sizeof(hdr.sync)) == 0);
 710                 uint32_t size = ntohl(hdr.size);
 711                 uint16_t flags = ntohs(hdr.flags);
 712                 uint16_t expected_csum = metacube2_compute_crc(&hdr);
 713
 714                 if (expected_csum != ntohs(hdr.csum)) {
 715                         log(WARNING, "[%s] Metacube checksum failed (expected 0x%x, got 0x%x), "
 716                                 "not reading block claiming to be %d bytes (flags=%x).",
 717                                 url.c_str(), expected_csum, ntohs(hdr.csum),
 718                                 size, flags);
 719
 720                         // Drop only the first byte, and let the rest of the code handle resync.
 721                         pending_data.erase(pending_data.begin(), pending_data.begin() + 1);
 722                         has_metacube_header = false;
 723                         continue;
 724                 }
 725                 if (size > 10485760) {
 726                         log(WARNING, "[%s] Metacube block of %d bytes (flags=%x); corrupted header??",
 727                                 url.c_str(), size, flags);
 728                 }
 729
 730                 // See if we have the entire block. If not, wait for more data.
 731                 if (pending_data.size() < sizeof(metacube2_block_header) + size) {
 732                         return;
 733                 }
 734
 735                 // See if this is a metadata block. If so, we don't want to send it on,
 736                 // but rather process it ourselves.
 737                 // TODO: Keep metadata when sending on to other Metacube users.
 738                 if (flags & METACUBE_FLAGS_METADATA) {
 739                         {
 740                                 lock_guard<mutex> lock(stats_mutex);
 741                                 stats.metadata_bytes_received += size;
 742                         }
 743                         process_metacube_metadata_block(hdr, pending_data.data() + sizeof(hdr), size);
 744                 } else {
 745                         // Send this block on to the servers.
 746                         {
 747                                 lock_guard<mutex> lock(stats_mutex);
 748                                 stats.data_bytes_received += size;
 749                         }
 750                         char *inner_data = pending_data.data() + sizeof(metacube2_block_header);
 751                         if (flags & METACUBE_FLAGS_HEADER) {
 752                                 stream_header = string(inner_data, inner_data + size);
 753                                 for (int stream_index : stream_indices) {
 754                                         servers->set_header(stream_index, http_header, stream_header);
 755                                 }
 756                         }
 757                         for (int stream_index : stream_indices) {
 758                                 servers->add_data(stream_index, inner_data, size, flags, next_block_pts);
 759                         }
 760                         next_block_pts = RationalPTS();
 761                 }
 762
 763                 // Consume the block. This isn't the most efficient way of dealing with things
 764                 // should we have many blocks, but these routines don't need to be too efficient
 765                 // anyway.
 766                 pending_data.erase(pending_data.begin(), pending_data.begin() + sizeof(metacube2_block_header) + size);
 767                 has_metacube_header = false;
 768         }
 769 }
 770
 771 void HTTPInput::drop_pending_data(size_t num_bytes)
 772 {
 773         if (num_bytes == 0) {
 774                 return;
 775         }
 776         log(WARNING, "[%s] Dropping %lld junk bytes; not a Metacube2 stream, or data was dropped from the middle of the stream.",
 777                 url.c_str(), (long long)num_bytes);
 778         assert(pending_data.size() >= num_bytes);
 779         pending_data.erase(pending_data.begin(), pending_data.begin() + num_bytes);
 780 }
 781
 782 void HTTPInput::add_destination(int stream_index)
 783 {
 784         stream_indices.push_back(stream_index);
 785         servers->set_header(stream_index, http_header, stream_header);
 786 }
 787
 788 InputStats HTTPInput::get_stats() const
 789 {
 790         lock_guard<mutex> lock(stats_mutex);
 791         return stats;
 792 }
 793
 794 void HTTPInput::process_metacube_metadata_block(const metacube2_block_header &hdr, const char *payload, uint32_t payload_size)
 795 {
 796         if (payload_size < sizeof(uint64_t)) {
 797                 log(WARNING, "[%s] Undersized Metacube metadata block (%d bytes); corrupted header?",
 798                         url.c_str(), payload_size);
 799                 return;
 800         }
 801
 802         uint64_t type = be64toh(*(const uint64_t *)payload);
 803         if (type == METACUBE_METADATA_TYPE_ENCODER_TIMESTAMP) {
 804                 timespec now;
 805                 clock_gettime(CLOCK_REALTIME, &now);
 806
 807                 const metacube2_timestamp_packet *pkt = (const metacube2_timestamp_packet *)payload;
 808                 if (payload_size != sizeof(*pkt)) {
 809                         log(WARNING, "[%s] Metacube timestamp block of wrong size (%d bytes); ignoring.",
 810                                 url.c_str(), payload_size);
 811                         return;
 812                 }
 813
 814                 double elapsed = now.tv_sec - be64toh(pkt->tv_sec) +
 815                         1e-9 * (now.tv_nsec - long(be64toh(pkt->tv_nsec)));
 816                 {
 817                         lock_guard<mutex> lock(stats_mutex);
 818                         stats.latency_sec = elapsed;
 819                 }
 820         } else if (type == METACUBE_METADATA_TYPE_NEXT_BLOCK_PTS) {
 821                 const metacube2_pts_packet *pkt = (const metacube2_pts_packet *)payload;
 822                 if (payload_size != sizeof(*pkt)) {
 823                         log(WARNING, "[%s] Metacube pts block of wrong size (%d bytes); ignoring.",
 824                                 url.c_str(), payload_size);
 825                         return;
 826                 }
 827                 next_block_pts.pts = be64toh(pkt->pts);
 828                 next_block_pts.timebase_num = be64toh(pkt->timebase_num);
 829                 next_block_pts.timebase_den = be64toh(pkt->timebase_den);
 830         } else {
 831                 // Unknown metadata block, ignore
 832                 log(INFO, "[%s] Metadata block %llu\n", url.c_str(), type);
 833                 return;
 834         }
 835 }