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

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