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 <map>
  15 #include <string>
  16 #include <utility>
  17 #include <vector>
  18
  19 #include "httpinput.h"
  20 #include "log.h"
  21 #include "metacube2.h"
  22 #include "mutexlock.h"
  23 #include "parse.h"
  24 #include "serverpool.h"
  25 #include "state.pb.h"
  26 #include "stream.h"
  27 #include "util.h"
  28 #include "version.h"
  29
  30 using namespace std;
  31
  32 extern ServerPool *servers;
  33
  34 namespace {
  35
  36 // Compute b-a.
  37 timespec clock_diff(const timespec &a, const timespec &b)
  38 {
  39         timespec ret;
  40         ret.tv_sec = b.tv_sec - a.tv_sec;
  41         ret.tv_nsec = b.tv_nsec - a.tv_nsec;
  42         if (ret.tv_nsec < 0) {
  43                 ret.tv_sec--;
  44                 ret.tv_nsec += 1000000000;
  45         }
  46         assert(ret.tv_nsec >= 0);
  47         return ret;
  48 }
  49
  50 }  // namespace
  51
  52 HTTPInput::HTTPInput(const string &url)
  53         : state(NOT_CONNECTED),
  54           url(url),
  55           has_metacube_header(false),
  56           sock(-1)
  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.connect_time = -1;
  63 }
  64
  65 HTTPInput::HTTPInput(const InputProto &serialized)
  66         : state(State(serialized.state())),
  67           url(serialized.url()),
  68           request(serialized.request()),
  69           request_bytes_sent(serialized.request_bytes_sent()),
  70           response(serialized.response()),
  71           http_header(serialized.http_header()),
  72           stream_header(serialized.stream_header()),
  73           has_metacube_header(serialized.has_metacube_header()),
  74           sock(serialized.sock())
  75 {
  76         pending_data.resize(serialized.pending_data().size());
  77         memcpy(&pending_data[0], serialized.pending_data().data(), serialized.pending_data().size());
  78
  79         string protocol;
  80         parse_url(url, &protocol, &host, &port, &path);  // Don't care if it fails.
  81
  82         pthread_mutex_init(&stats_mutex, NULL);
  83         stats.url = url;
  84         stats.bytes_received = serialized.bytes_received();
  85         stats.data_bytes_received = serialized.data_bytes_received();
  86         if (serialized.has_connect_time()) {
  87                 stats.connect_time = serialized.connect_time();
  88         } else {
  89                 stats.connect_time = time(NULL);
  90         }
  91 }
  92
  93 void HTTPInput::close_socket()
  94 {
  95         if (sock != -1) {
  96                 safe_close(sock);
  97         }
  98
  99         MutexLock lock(&stats_mutex);
 100         stats.connect_time = -1;
 101 }
 102
 103 InputProto HTTPInput::serialize() const
 104 {
 105         InputProto serialized;
 106         serialized.set_state(state);
 107         serialized.set_url(url);
 108         serialized.set_request(request);
 109         serialized.set_request_bytes_sent(request_bytes_sent);
 110         serialized.set_response(response);
 111         serialized.set_http_header(http_header);
 112         serialized.set_stream_header(stream_header);
 113         serialized.set_pending_data(string(pending_data.begin(), pending_data.end()));
 114         serialized.set_has_metacube_header(has_metacube_header);
 115         serialized.set_sock(sock);
 116         serialized.set_bytes_received(stats.bytes_received);
 117         serialized.set_data_bytes_received(stats.data_bytes_received);
 118         serialized.set_connect_time(stats.connect_time);
 119         return serialized;
 120 }
 121
 122 int HTTPInput::lookup_and_connect(const string &host, const string &port)
 123 {
 124         addrinfo *ai;
 125         int err = getaddrinfo(host.c_str(), port.c_str(), NULL, &ai);
 126         if (err != 0) {
 127                 log(WARNING, "[%s] Lookup of '%s' failed (%s).",
 128                         url.c_str(), host.c_str(), gai_strerror(err));
 129                 return -1;
 130         }
 131
 132         addrinfo *base_ai = ai;
 133
 134         // Connect to everything in turn until we have a socket.
 135         for ( ; ai && !should_stop(); ai = ai->ai_next) {
 136                 int sock = socket(ai->ai_family, SOCK_STREAM, IPPROTO_TCP);
 137                 if (sock == -1) {
 138                         // Could be e.g. EPROTONOSUPPORT. The show must go on.
 139                         continue;
 140                 }
 141
 142                 // Now do a non-blocking connect. This is important because we want to be able to be
 143                 // woken up, even though it's rather cumbersome.
 144
 145                 // Set the socket as nonblocking.
 146                 int one = 1;
 147                 if (ioctl(sock, FIONBIO, &one) == -1) {
 148                         log_perror("ioctl(FIONBIO)");
 149                         safe_close(sock);
 150                         return -1;
 151                 }
 152
 153                 // Do a non-blocking connect.
 154                 do {
 155                         err = connect(sock, ai->ai_addr, ai->ai_addrlen);
 156                 } while (err == -1 && errno == EINTR);
 157
 158                 if (err == -1 && errno != EINPROGRESS) {
 159                         log_perror("connect");
 160                         safe_close(sock);
 161                         continue;
 162                 }
 163
 164                 // Wait for the connect to complete, or an error to happen.
 165                 for ( ;; ) {
 166                         bool complete = wait_for_activity(sock, POLLIN | POLLOUT, NULL);
 167                         if (should_stop()) {
 168                                 safe_close(sock);
 169                                 return -1;
 170                         }
 171                         if (complete) {
 172                                 break;
 173                         }
 174                 }
 175
 176                 // Check whether it ended in an error or not.
 177                 socklen_t err_size = sizeof(err);
 178                 if (getsockopt(sock, SOL_SOCKET, SO_ERROR, &err, &err_size) == -1) {
 179                         log_perror("getsockopt");
 180                         safe_close(sock);
 181                         continue;
 182                 }
 183
 184                 errno = err;
 185
 186                 if (err == 0) {
 187                         // Successful connect.
 188                         freeaddrinfo(base_ai);
 189                         return sock;
 190                 }
 191
 192                 safe_close(sock);
 193         }
 194
 195         // Give the last one as error.
 196         log(WARNING, "[%s] Connect to '%s' failed (%s)",
 197                 url.c_str(), host.c_str(), strerror(errno));
 198         freeaddrinfo(base_ai);
 199         return -1;
 200 }
 201
 202 bool HTTPInput::parse_response(const std::string &request)
 203 {
 204         vector<string> lines = split_lines(response);
 205         if (lines.empty()) {
 206                 log(WARNING, "[%s] Empty HTTP response from input.", url.c_str());
 207                 return false;
 208         }
 209
 210         vector<string> first_line_tokens = split_tokens(lines[0]);
 211         if (first_line_tokens.size() < 2) {
 212                 log(WARNING, "[%s] Malformed response line '%s' from input.",
 213                         url.c_str(), lines[0].c_str());
 214                 return false;
 215         }
 216
 217         int response = atoi(first_line_tokens[1].c_str());
 218         if (response != 200) {
 219                 log(WARNING, "[%s] Non-200 response '%s' from input.",
 220                         url.c_str(), lines[0].c_str());
 221                 return false;
 222         }
 223
 224         multimap<string, string> parameters;
 225         for (size_t i = 1; i < lines.size(); ++i) {
 226                 size_t split = lines[i].find(":");
 227                 if (split == string::npos) {
 228                         log(WARNING, "[%s] Ignoring malformed HTTP response line '%s'",
 229                                 url.c_str(), lines[i].c_str());
 230                         continue;
 231                 }
 232
 233                 string key(lines[i].begin(), lines[i].begin() + split);
 234
 235                 // Skip any spaces after the colon.
 236                 do {
 237                         ++split;
 238                 } while (split < lines[i].size() && lines[i][split] == ' ');
 239
 240                 string value(lines[i].begin() + split, lines[i].end());
 241
 242                 // Remove “Content-encoding: metacube”.
 243                 // TODO: Make case-insensitive.
 244                 if (key == "Content-encoding" && value == "metacube") {
 245                         continue;
 246                 }
 247
 248                 parameters.insert(make_pair(key, value));
 249         }
 250
 251         // Change “Server: foo” to “Server: metacube/0.1 (reflecting: foo)”
 252         // TODO: Make case-insensitive.
 253         // XXX: Use a Via: instead?
 254         if (parameters.count("Server") == 0) {
 255                 parameters.insert(make_pair("Server", SERVER_IDENTIFICATION));
 256         } else {
 257                 for (multimap<string, string>::iterator it = parameters.begin();
 258                      it != parameters.end();
 259                      ++it) {
 260                         if (it->first != "Server") {
 261                                 continue;
 262                         }
 263                         it->second = SERVER_IDENTIFICATION " (reflecting: " + it->second + ")";
 264                 }
 265         }
 266
 267         // Set “Connection: close”.
 268         // TODO: Make case-insensitive.
 269         parameters.erase("Connection");
 270         parameters.insert(make_pair("Connection", "close"));
 271
 272         // Construct the new HTTP header.
 273         http_header = "HTTP/1.0 200 OK\r\n";
 274         for (multimap<string, string>::iterator it = parameters.begin();
 275              it != parameters.end();
 276              ++it) {
 277                 http_header.append(it->first + ": " + it->second + "\r\n");
 278         }
 279
 280         for (size_t i = 0; i < stream_indices.size(); ++i) {
 281                 servers->set_header(stream_indices[i], http_header, stream_header);
 282         }
 283
 284         return true;
 285 }
 286
 287 void HTTPInput::do_work()
 288 {
 289         timespec last_activity;
 290
 291         // TODO: Make the timeout persist across restarts.
 292         if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 293                 int err = clock_gettime(CLOCK_MONOTONIC, &last_activity);
 294                 assert(err != -1);
 295         }
 296
 297         while (!should_stop()) {
 298                 if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 299                         // Give the socket 30 seconds since last activity before we time out.
 300                         static const int timeout_secs = 30;
 301
 302                         timespec now;
 303                         int err = clock_gettime(CLOCK_MONOTONIC, &now);
 304                         assert(err != -1);
 305
 306                         timespec elapsed = clock_diff(last_activity, now);
 307                         if (elapsed.tv_sec >= timeout_secs) {
 308                                 // Timeout!
 309                                 log(ERROR, "[%s] Timeout after %d seconds, closing.", url.c_str(), elapsed.tv_sec);
 310                                 state = CLOSING_SOCKET;
 311                                 continue;
 312                         }
 313
 314                         // Basically calculate (30 - (now - last_activity)) = (30 + (last_activity - now)).
 315                         // Add a second of slack to account for differences between clocks.
 316                         timespec timeout = clock_diff(now, last_activity);
 317                         timeout.tv_sec += timeout_secs + 1;
 318                         assert(timeout.tv_sec > 0 || (timeout.tv_sec >= 0 && timeout.tv_nsec > 0));
 319
 320                         bool activity = wait_for_activity(sock, (state == SENDING_REQUEST) ? POLLOUT : POLLIN, &timeout);
 321                         if (activity) {
 322                                 err = clock_gettime(CLOCK_MONOTONIC, &last_activity);
 323                                 assert(err != -1);
 324                         } else {
 325                                 // OK. Most likely, should_stop was set, or we have timed out.
 326                                 continue;
 327                         }
 328                 }
 329
 330                 switch (state) {
 331                 case NOT_CONNECTED:
 332                         request.clear();
 333                         request_bytes_sent = 0;
 334                         response.clear();
 335                         pending_data.clear();
 336                         has_metacube_header = false;
 337                         for (size_t i = 0; i < stream_indices.size(); ++i) {
 338                                 servers->set_header(stream_indices[i], "", "");
 339                         }
 340
 341                         {
 342                                 string protocol;  // Thrown away.
 343                                 if (!parse_url(url, &protocol, &host, &port, &path)) {
 344                                         log(WARNING, "[%s] Failed to parse URL '%s'", url.c_str(), url.c_str());
 345                                         break;
 346                                 }
 347                         }
 348
 349                         sock = lookup_and_connect(host, port);
 350                         if (sock != -1) {
 351                                 // Yay, successful connect. Try to set it as nonblocking.
 352                                 int one = 1;
 353                                 if (ioctl(sock, FIONBIO, &one) == -1) {
 354                                         log_perror("ioctl(FIONBIO)");
 355                                         state = CLOSING_SOCKET;
 356                                 } else {
 357                                         state = SENDING_REQUEST;
 358                                         request = "GET " + path + " HTTP/1.0\r\nUser-Agent: cubemap\r\n\r\n";
 359                                         request_bytes_sent = 0;
 360                                 }
 361
 362                                 MutexLock lock(&stats_mutex);
 363                                 stats.connect_time = time(NULL);
 364                                 clock_gettime(CLOCK_MONOTONIC, &last_activity);
 365                         }
 366                         break;
 367                 case SENDING_REQUEST: {
 368                         size_t to_send = request.size() - request_bytes_sent;
 369                         int ret;
 370
 371                         do {
 372                                 ret = write(sock, request.data() + request_bytes_sent, to_send);
 373                         } while (ret == -1 && errno == EINTR);
 374
 375                         if (ret == -1) {
 376                                 log_perror("write");
 377                                 state = CLOSING_SOCKET;
 378                                 continue;
 379                         }
 380
 381                         assert(ret >= 0);
 382                         request_bytes_sent += ret;
 383
 384                         if (request_bytes_sent == request.size()) {
 385                                 state = RECEIVING_HEADER;
 386                         }
 387                         break;
 388                 }
 389                 case RECEIVING_HEADER: {
 390                         char buf[4096];
 391                         int ret;
 392
 393                         do {
 394                                 ret = read(sock, buf, sizeof(buf));
 395                         } while (ret == -1 && errno == EINTR);
 396
 397                         if (ret == -1) {
 398                                 log_perror("read");
 399                                 state = CLOSING_SOCKET;
 400                                 continue;
 401                         }
 402
 403                         if (ret == 0) {
 404                                 // This really shouldn't happen...
 405                                 log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 406                                            url.c_str());
 407                                 state = CLOSING_SOCKET;
 408                                 continue;
 409                         }
 410
 411                         RequestParseStatus status = wait_for_double_newline(&response, buf, ret);
 412
 413                         if (status == RP_OUT_OF_SPACE) {
 414                                 log(WARNING, "[%s] Sever sent overlong HTTP response!", url.c_str());
 415                                 state = CLOSING_SOCKET;
 416                                 continue;
 417                         } else if (status == RP_NOT_FINISHED_YET) {
 418                                 continue;
 419                         }
 420
 421                         // OK, so we're fine, but there might be some of the actual data after the response.
 422                         // We'll need to deal with that separately.
 423                         string extra_data;
 424                         if (status == RP_EXTRA_DATA) {
 425                                 char *ptr = static_cast<char *>(
 426                                         memmem(response.data(), response.size(), "\r\n\r\n", 4));
 427                                 assert(ptr != NULL);
 428                                 extra_data = string(ptr + 4, &response[0] + response.size());
 429                                 response.resize(ptr - response.data());
 430                         }
 431
 432                         if (!parse_response(response)) {
 433                                 state = CLOSING_SOCKET;
 434                                 continue;
 435                         }
 436
 437                         if (!extra_data.empty()) {
 438                                 process_data(&extra_data[0], extra_data.size());
 439                         }
 440
 441                         log(INFO, "[%s] Connected to '%s', receiving data.",
 442                                    url.c_str(), url.c_str());
 443                         state = RECEIVING_DATA;
 444                         break;
 445                 }
 446                 case RECEIVING_DATA: {
 447                         char buf[4096];
 448                         int ret;
 449
 450                         do {
 451                                 ret = read(sock, buf, sizeof(buf));
 452                         } while (ret == -1 && errno == EINTR);
 453
 454                         if (ret == -1) {
 455                                 log_perror("read");
 456                                 state = CLOSING_SOCKET;
 457                                 continue;
 458                         }
 459
 460                         if (ret == 0) {
 461                                 // This really shouldn't happen...
 462                                 log(ERROR, "[%s] Socket unexpectedly closed while reading data",
 463                                            url.c_str());
 464                                 state = CLOSING_SOCKET;
 465                                 continue;
 466                         }
 467
 468                         process_data(buf, ret);
 469                         break;
 470                 }
 471                 case CLOSING_SOCKET: {
 472                         close_socket();
 473                         state = NOT_CONNECTED;
 474                         break;
 475                 }
 476                 default:
 477                         assert(false);
 478                 }
 479
 480                 // If we are still in NOT_CONNECTED, either something went wrong,
 481                 // or the connection just got closed.
 482                 // The earlier steps have already given the error message, if any.
 483                 if (state == NOT_CONNECTED && !should_stop()) {
 484                         log(INFO, "[%s] Waiting 0.2 second and restarting...", url.c_str());
 485                         timespec timeout_ts;
 486                         timeout_ts.tv_sec = 0;
 487                         timeout_ts.tv_nsec = 200000000;
 488                         wait_for_wakeup(&timeout_ts);
 489                 }
 490         }
 491 }
 492
 493 void HTTPInput::process_data(char *ptr, size_t bytes)
 494 {
 495         pending_data.insert(pending_data.end(), ptr, ptr + bytes);
 496         {
 497                 MutexLock mutex(&stats_mutex);
 498                 stats.bytes_received += bytes;
 499         }
 500
 501         for ( ;; ) {
 502                 // If we don't have enough data (yet) for even the Metacube header, just return.
 503                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 504                         return;
 505                 }
 506
 507                 // Make sure we have the Metacube sync header at the start.
 508                 // We may need to skip over junk data (it _should_ not happen, though).
 509                 if (!has_metacube_header) {
 510                         char *ptr = static_cast<char *>(
 511                                 memmem(pending_data.data(), pending_data.size(),
 512                                        METACUBE2_SYNC, strlen(METACUBE2_SYNC)));
 513                         if (ptr == NULL) {
 514                                 // OK, so we didn't find the sync marker. We know then that
 515                                 // we do not have the _full_ marker in the buffer, but we
 516                                 // could have N-1 bytes. Drop everything before that,
 517                                 // and then give up.
 518                                 drop_pending_data(pending_data.size() - (strlen(METACUBE2_SYNC) - 1));
 519                                 return;
 520                         } else {
 521                                 // Yay, we found the header. Drop everything (if anything) before it.
 522                                 drop_pending_data(ptr - pending_data.data());
 523                                 has_metacube_header = true;
 524
 525                                 // Re-check that we have the entire header; we could have dropped data.
 526                                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 527                                         return;
 528                                 }
 529                         }
 530                 }
 531
 532                 // Now it's safe to read the header.
 533                 metacube2_block_header hdr;
 534                 memcpy(&hdr, pending_data.data(), sizeof(hdr));
 535                 assert(memcmp(hdr.sync, METACUBE2_SYNC, sizeof(hdr.sync)) == 0);
 536                 uint32_t size = ntohl(hdr.size);
 537                 uint16_t flags = ntohs(hdr.flags);
 538                 uint16_t expected_csum = metacube2_compute_crc(&hdr);
 539
 540                 if (expected_csum != ntohs(hdr.csum)) {
 541                         log(WARNING, "[%s] Metacube checksum failed (expected 0x%x, got 0x%x), "
 542                                 "not reading block claiming to be %d bytes (flags=%x).",
 543                                 url.c_str(), expected_csum, ntohs(hdr.csum),
 544                                 size, flags);
 545
 546                         // Drop only the first byte, and let the rest of the code handle resync.
 547                         pending_data.erase(pending_data.begin(), pending_data.begin() + 1);
 548                         has_metacube_header = false;
 549                         continue;
 550                 }
 551                 if (size > 262144) {
 552                         log(WARNING, "[%s] Metacube block of %d bytes (flags=%x); corrupted header?",
 553                                 url.c_str(), size, flags);
 554                 }
 555
 556                 // See if we have the entire block. If not, wait for more data.
 557                 if (pending_data.size() < sizeof(metacube2_block_header) + size) {
 558                         return;
 559                 }
 560
 561                 // Send this block on to the servers.
 562                 {
 563                         MutexLock lock(&stats_mutex);
 564                         stats.data_bytes_received += size;
 565                 }
 566                 char *inner_data = pending_data.data() + sizeof(metacube2_block_header);
 567                 if (flags & METACUBE_FLAGS_HEADER) {
 568                         stream_header = string(inner_data, inner_data + size);
 569                         for (size_t i = 0; i < stream_indices.size(); ++i) {
 570                                 servers->set_header(stream_indices[i], http_header, stream_header);
 571                         }
 572                 } else {
 573                         StreamStartSuitability suitable_for_stream_start;
 574                         if (flags & METACUBE_FLAGS_NOT_SUITABLE_FOR_STREAM_START) {
 575                                 suitable_for_stream_start = NOT_SUITABLE_FOR_STREAM_START;
 576                         } else {
 577                                 suitable_for_stream_start = SUITABLE_FOR_STREAM_START;
 578                         }
 579                         for (size_t i = 0; i < stream_indices.size(); ++i) {
 580                                 servers->add_data(stream_indices[i], inner_data, size, suitable_for_stream_start);
 581                         }
 582                 }
 583
 584                 // Consume the block. This isn't the most efficient way of dealing with things
 585                 // should we have many blocks, but these routines don't need to be too efficient
 586                 // anyway.
 587                 pending_data.erase(pending_data.begin(), pending_data.begin() + sizeof(metacube2_block_header) + size);
 588                 has_metacube_header = false;
 589         }
 590 }
 591
 592 void HTTPInput::drop_pending_data(size_t num_bytes)
 593 {
 594         if (num_bytes == 0) {
 595                 return;
 596         }
 597         log(WARNING, "[%s] Dropping %lld junk bytes from stream, maybe it is not a Metacube2 stream?",
 598                 url.c_str(), (long long)num_bytes);
 599         assert(pending_data.size() >= num_bytes);
 600         pending_data.erase(pending_data.begin(), pending_data.begin() + num_bytes);
 601 }
 602
 603 void HTTPInput::add_destination(int stream_index)
 604 {
 605         stream_indices.push_back(stream_index);
 606         servers->set_header(stream_index, http_header, stream_header);
 607 }
 608
 609 InputStats HTTPInput::get_stats() const
 610 {
 611         MutexLock lock(&stats_mutex);
 612         return stats;
 613 }