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, user;
  80         parse_url(url, &protocol, &user, &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                 sock = -1;
  98         }
  99
 100         MutexLock lock(&stats_mutex);
 101         stats.connect_time = -1;
 102 }
 103
 104 InputProto HTTPInput::serialize() const
 105 {
 106         InputProto serialized;
 107         serialized.set_state(state);
 108         serialized.set_url(url);
 109         serialized.set_request(request);
 110         serialized.set_request_bytes_sent(request_bytes_sent);
 111         serialized.set_response(response);
 112         serialized.set_http_header(http_header);
 113         serialized.set_stream_header(stream_header);
 114         serialized.set_pending_data(string(pending_data.begin(), pending_data.end()));
 115         serialized.set_has_metacube_header(has_metacube_header);
 116         serialized.set_sock(sock);
 117         serialized.set_bytes_received(stats.bytes_received);
 118         serialized.set_data_bytes_received(stats.data_bytes_received);
 119         serialized.set_connect_time(stats.connect_time);
 120         return serialized;
 121 }
 122
 123 int HTTPInput::lookup_and_connect(const string &host, const string &port)
 124 {
 125         addrinfo *ai;
 126         int err = getaddrinfo(host.c_str(), port.c_str(), NULL, &ai);
 127         if (err != 0) {
 128                 log(WARNING, "[%s] Lookup of '%s' failed (%s).",
 129                         url.c_str(), host.c_str(), gai_strerror(err));
 130                 return -1;
 131         }
 132
 133         addrinfo *base_ai = ai;
 134
 135         // Connect to everything in turn until we have a socket.
 136         for ( ; ai && !should_stop(); ai = ai->ai_next) {
 137                 int sock = socket(ai->ai_family, SOCK_STREAM, IPPROTO_TCP);
 138                 if (sock == -1) {
 139                         // Could be e.g. EPROTONOSUPPORT. The show must go on.
 140                         continue;
 141                 }
 142
 143                 // Now do a non-blocking connect. This is important because we want to be able to be
 144                 // woken up, even though it's rather cumbersome.
 145
 146                 // Set the socket as nonblocking.
 147                 int one = 1;
 148                 if (ioctl(sock, FIONBIO, &one) == -1) {
 149                         log_perror("ioctl(FIONBIO)");
 150                         safe_close(sock);
 151                         return -1;
 152                 }
 153
 154                 // Do a non-blocking connect.
 155                 do {
 156                         err = connect(sock, ai->ai_addr, ai->ai_addrlen);
 157                 } while (err == -1 && errno == EINTR);
 158
 159                 if (err == -1 && errno != EINPROGRESS) {
 160                         log_perror("connect");
 161                         safe_close(sock);
 162                         continue;
 163                 }
 164
 165                 // Wait for the connect to complete, or an error to happen.
 166                 for ( ;; ) {
 167                         bool complete = wait_for_activity(sock, POLLIN | POLLOUT, NULL);
 168                         if (should_stop()) {
 169                                 safe_close(sock);
 170                                 return -1;
 171                         }
 172                         if (complete) {
 173                                 break;
 174                         }
 175                 }
 176
 177                 // Check whether it ended in an error or not.
 178                 socklen_t err_size = sizeof(err);
 179                 if (getsockopt(sock, SOL_SOCKET, SO_ERROR, &err, &err_size) == -1) {
 180                         log_perror("getsockopt");
 181                         safe_close(sock);
 182                         continue;
 183                 }
 184
 185                 errno = err;
 186
 187                 if (err == 0) {
 188                         // Successful connect.
 189                         freeaddrinfo(base_ai);
 190                         return sock;
 191                 }
 192
 193                 safe_close(sock);
 194         }
 195
 196         // Give the last one as error.
 197         log(WARNING, "[%s] Connect to '%s' failed (%s)",
 198                 url.c_str(), host.c_str(), strerror(errno));
 199         freeaddrinfo(base_ai);
 200         return -1;
 201 }
 202
 203 bool HTTPInput::parse_response(const std::string &request)
 204 {
 205         vector<string> lines = split_lines(response);
 206         if (lines.empty()) {
 207                 log(WARNING, "[%s] Empty HTTP response from input.", url.c_str());
 208                 return false;
 209         }
 210
 211         vector<string> first_line_tokens = split_tokens(lines[0]);
 212         if (first_line_tokens.size() < 2) {
 213                 log(WARNING, "[%s] Malformed response line '%s' from input.",
 214                         url.c_str(), lines[0].c_str());
 215                 return false;
 216         }
 217
 218         int response = atoi(first_line_tokens[1].c_str());
 219         if (response != 200) {
 220                 log(WARNING, "[%s] Non-200 response '%s' from input.",
 221                         url.c_str(), lines[0].c_str());
 222                 return false;
 223         }
 224
 225         multimap<string, string> parameters = extract_headers(lines, url);
 226
 227         // Remove “Content-encoding: metacube”.
 228         // TODO: Make case-insensitive.
 229         multimap<string, string>::iterator encoding_it =
 230                 parameters.find("Content-encoding");
 231         if (encoding_it != parameters.end() && encoding_it->second == "metacube") {
 232                 parameters.erase(encoding_it);
 233         }
 234
 235         // Change “Server: foo” to “Server: metacube/0.1 (reflecting: foo)”
 236         // TODO: Make case-insensitive.
 237         // XXX: Use a Via: instead?
 238         if (parameters.count("Server") == 0) {
 239                 parameters.insert(make_pair("Server", SERVER_IDENTIFICATION));
 240         } else {
 241                 for (multimap<string, string>::iterator it = parameters.begin();
 242                      it != parameters.end();
 243                      ++it) {
 244                         if (it->first != "Server") {
 245                                 continue;
 246                         }
 247                         it->second = SERVER_IDENTIFICATION " (reflecting: " + it->second + ")";
 248                 }
 249         }
 250
 251         // Set “Connection: close”.
 252         // TODO: Make case-insensitive.
 253         parameters.erase("Connection");
 254         parameters.insert(make_pair("Connection", "close"));
 255
 256         // Construct the new HTTP header.
 257         http_header = "HTTP/1.0 200 OK\r\n";
 258         for (multimap<string, string>::iterator it = parameters.begin();
 259              it != parameters.end();
 260              ++it) {
 261                 http_header.append(it->first + ": " + it->second + "\r\n");
 262         }
 263
 264         for (size_t i = 0; i < stream_indices.size(); ++i) {
 265                 servers->set_header(stream_indices[i], http_header, stream_header);
 266         }
 267
 268         return true;
 269 }
 270
 271 void HTTPInput::do_work()
 272 {
 273         timespec last_activity;
 274
 275         // TODO: Make the timeout persist across restarts.
 276         if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 277                 int err = clock_gettime(CLOCK_MONOTONIC, &last_activity);
 278                 assert(err != -1);
 279         }
 280
 281         while (!should_stop()) {
 282                 if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 283                         // Give the socket 30 seconds since last activity before we time out.
 284                         static const int timeout_secs = 30;
 285
 286                         timespec now;
 287                         int err = clock_gettime(CLOCK_MONOTONIC, &now);
 288                         assert(err != -1);
 289
 290                         timespec elapsed = clock_diff(last_activity, now);
 291                         if (elapsed.tv_sec >= timeout_secs) {
 292                                 // Timeout!
 293                                 log(ERROR, "[%s] Timeout after %d seconds, closing.", url.c_str(), elapsed.tv_sec);
 294                                 state = CLOSING_SOCKET;
 295                                 continue;
 296                         }
 297
 298                         // Basically calculate (30 - (now - last_activity)) = (30 + (last_activity - now)).
 299                         // Add a second of slack to account for differences between clocks.
 300                         timespec timeout = clock_diff(now, last_activity);
 301                         timeout.tv_sec += timeout_secs + 1;
 302                         assert(timeout.tv_sec > 0 || (timeout.tv_sec >= 0 && timeout.tv_nsec > 0));
 303
 304                         bool activity = wait_for_activity(sock, (state == SENDING_REQUEST) ? POLLOUT : POLLIN, &timeout);
 305                         if (activity) {
 306                                 err = clock_gettime(CLOCK_MONOTONIC, &last_activity);
 307                                 assert(err != -1);
 308                         } else {
 309                                 // OK. Most likely, should_stop was set, or we have timed out.
 310                                 continue;
 311                         }
 312                 }
 313
 314                 switch (state) {
 315                 case NOT_CONNECTED:
 316                         request.clear();
 317                         request_bytes_sent = 0;
 318                         response.clear();
 319                         pending_data.clear();
 320                         has_metacube_header = false;
 321                         for (size_t i = 0; i < stream_indices.size(); ++i) {
 322                                 servers->set_header(stream_indices[i], "", "");
 323                         }
 324
 325                         {
 326                                 string protocol, user;  // Thrown away.
 327                                 if (!parse_url(url, &protocol, &user, &host, &port, &path)) {
 328                                         log(WARNING, "[%s] Failed to parse URL '%s'", url.c_str(), url.c_str());
 329                                         break;
 330                                 }
 331                         }
 332
 333                         sock = lookup_and_connect(host, port);
 334                         if (sock != -1) {
 335                                 // Yay, successful connect. Try to set it as nonblocking.
 336                                 int one = 1;
 337                                 if (ioctl(sock, FIONBIO, &one) == -1) {
 338                                         log_perror("ioctl(FIONBIO)");
 339                                         state = CLOSING_SOCKET;
 340                                 } else {
 341                                         state = SENDING_REQUEST;
 342                                         request = "GET " + path + " HTTP/1.0\r\nUser-Agent: cubemap\r\n\r\n";
 343                                         request_bytes_sent = 0;
 344                                 }
 345
 346                                 MutexLock lock(&stats_mutex);
 347                                 stats.connect_time = time(NULL);
 348                                 clock_gettime(CLOCK_MONOTONIC, &last_activity);
 349                         }
 350                         break;
 351                 case SENDING_REQUEST: {
 352                         size_t to_send = request.size() - request_bytes_sent;
 353                         int ret;
 354
 355                         do {
 356                                 ret = write(sock, request.data() + request_bytes_sent, to_send);
 357                         } while (ret == -1 && errno == EINTR);
 358
 359                         if (ret == -1) {
 360                                 log_perror("write");
 361                                 state = CLOSING_SOCKET;
 362                                 continue;
 363                         }
 364
 365                         assert(ret >= 0);
 366                         request_bytes_sent += ret;
 367
 368                         if (request_bytes_sent == request.size()) {
 369                                 state = RECEIVING_HEADER;
 370                         }
 371                         break;
 372                 }
 373                 case RECEIVING_HEADER: {
 374                         char buf[4096];
 375                         int ret;
 376
 377                         do {
 378                                 ret = read(sock, buf, sizeof(buf));
 379                         } while (ret == -1 && errno == EINTR);
 380
 381                         if (ret == -1) {
 382                                 log_perror("read");
 383                                 state = CLOSING_SOCKET;
 384                                 continue;
 385                         }
 386
 387                         if (ret == 0) {
 388                                 // This really shouldn't happen...
 389                                 log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 390                                            url.c_str());
 391                                 state = CLOSING_SOCKET;
 392                                 continue;
 393                         }
 394
 395                         RequestParseStatus status = wait_for_double_newline(&response, buf, ret);
 396
 397                         if (status == RP_OUT_OF_SPACE) {
 398                                 log(WARNING, "[%s] Server sent overlong HTTP response!", url.c_str());
 399                                 state = CLOSING_SOCKET;
 400                                 continue;
 401                         } else if (status == RP_NOT_FINISHED_YET) {
 402                                 continue;
 403                         }
 404
 405                         // OK, so we're fine, but there might be some of the actual data after the response.
 406                         // We'll need to deal with that separately.
 407                         string extra_data;
 408                         if (status == RP_EXTRA_DATA) {
 409                                 char *ptr = static_cast<char *>(
 410                                         memmem(response.data(), response.size(), "\r\n\r\n", 4));
 411                                 assert(ptr != NULL);
 412                                 extra_data = string(ptr + 4, &response[0] + response.size());
 413                                 response.resize(ptr - response.data());
 414                         }
 415
 416                         if (!parse_response(response)) {
 417                                 state = CLOSING_SOCKET;
 418                                 continue;
 419                         }
 420
 421                         if (!extra_data.empty()) {
 422                                 process_data(&extra_data[0], extra_data.size());
 423                         }
 424
 425                         log(INFO, "[%s] Connected to '%s', receiving data.",
 426                                    url.c_str(), url.c_str());
 427                         state = RECEIVING_DATA;
 428                         break;
 429                 }
 430                 case RECEIVING_DATA: {
 431                         char buf[4096];
 432                         int ret;
 433
 434                         do {
 435                                 ret = read(sock, buf, sizeof(buf));
 436                         } while (ret == -1 && errno == EINTR);
 437
 438                         if (ret == -1) {
 439                                 log_perror("read");
 440                                 state = CLOSING_SOCKET;
 441                                 continue;
 442                         }
 443
 444                         if (ret == 0) {
 445                                 // This really shouldn't happen...
 446                                 log(ERROR, "[%s] Socket unexpectedly closed while reading data",
 447                                            url.c_str());
 448                                 state = CLOSING_SOCKET;
 449                                 continue;
 450                         }
 451
 452                         process_data(buf, ret);
 453                         break;
 454                 }
 455                 case CLOSING_SOCKET: {
 456                         close_socket();
 457                         state = NOT_CONNECTED;
 458                         break;
 459                 }
 460                 default:
 461                         assert(false);
 462                 }
 463
 464                 // If we are still in NOT_CONNECTED, either something went wrong,
 465                 // or the connection just got closed.
 466                 // The earlier steps have already given the error message, if any.
 467                 if (state == NOT_CONNECTED && !should_stop()) {
 468                         log(INFO, "[%s] Waiting 0.2 second and restarting...", url.c_str());
 469                         timespec timeout_ts;
 470                         timeout_ts.tv_sec = 0;
 471                         timeout_ts.tv_nsec = 200000000;
 472                         wait_for_wakeup(&timeout_ts);
 473                 }
 474         }
 475 }
 476
 477 void HTTPInput::process_data(char *ptr, size_t bytes)
 478 {
 479         pending_data.insert(pending_data.end(), ptr, ptr + bytes);
 480         {
 481                 MutexLock mutex(&stats_mutex);
 482                 stats.bytes_received += bytes;
 483         }
 484
 485         for ( ;; ) {
 486                 // If we don't have enough data (yet) for even the Metacube header, just return.
 487                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 488                         return;
 489                 }
 490
 491                 // Make sure we have the Metacube sync header at the start.
 492                 // We may need to skip over junk data (it _should_ not happen, though).
 493                 if (!has_metacube_header) {
 494                         char *ptr = static_cast<char *>(
 495                                 memmem(pending_data.data(), pending_data.size(),
 496                                        METACUBE2_SYNC, strlen(METACUBE2_SYNC)));
 497                         if (ptr == NULL) {
 498                                 // OK, so we didn't find the sync marker. We know then that
 499                                 // we do not have the _full_ marker in the buffer, but we
 500                                 // could have N-1 bytes. Drop everything before that,
 501                                 // and then give up.
 502                                 drop_pending_data(pending_data.size() - (strlen(METACUBE2_SYNC) - 1));
 503                                 return;
 504                         } else {
 505                                 // Yay, we found the header. Drop everything (if anything) before it.
 506                                 drop_pending_data(ptr - pending_data.data());
 507                                 has_metacube_header = true;
 508
 509                                 // Re-check that we have the entire header; we could have dropped data.
 510                                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 511                                         return;
 512                                 }
 513                         }
 514                 }
 515
 516                 // Now it's safe to read the header.
 517                 metacube2_block_header hdr;
 518                 memcpy(&hdr, pending_data.data(), sizeof(hdr));
 519                 assert(memcmp(hdr.sync, METACUBE2_SYNC, sizeof(hdr.sync)) == 0);
 520                 uint32_t size = ntohl(hdr.size);
 521                 uint16_t flags = ntohs(hdr.flags);
 522                 uint16_t expected_csum = metacube2_compute_crc(&hdr);
 523
 524                 if (expected_csum != ntohs(hdr.csum)) {
 525                         log(WARNING, "[%s] Metacube checksum failed (expected 0x%x, got 0x%x), "
 526                                 "not reading block claiming to be %d bytes (flags=%x).",
 527                                 url.c_str(), expected_csum, ntohs(hdr.csum),
 528                                 size, flags);
 529
 530                         // Drop only the first byte, and let the rest of the code handle resync.
 531                         pending_data.erase(pending_data.begin(), pending_data.begin() + 1);
 532                         has_metacube_header = false;
 533                         continue;
 534                 }
 535                 if (size > 1048576) {
 536                         log(WARNING, "[%s] Metacube block of %d bytes (flags=%x); corrupted header?",
 537                                 url.c_str(), size, flags);
 538                 }
 539
 540                 // See if we have the entire block. If not, wait for more data.
 541                 if (pending_data.size() < sizeof(metacube2_block_header) + size) {
 542                         return;
 543                 }
 544
 545                 // Send this block on to the servers.
 546                 {
 547                         MutexLock lock(&stats_mutex);
 548                         stats.data_bytes_received += size;
 549                 }
 550                 char *inner_data = pending_data.data() + sizeof(metacube2_block_header);
 551                 if (flags & METACUBE_FLAGS_HEADER) {
 552                         stream_header = string(inner_data, inner_data + size);
 553                         for (size_t i = 0; i < stream_indices.size(); ++i) {
 554                                 servers->set_header(stream_indices[i], http_header, stream_header);
 555                         }
 556                 } else {
 557                         StreamStartSuitability suitable_for_stream_start;
 558                         if (flags & METACUBE_FLAGS_NOT_SUITABLE_FOR_STREAM_START) {
 559                                 suitable_for_stream_start = NOT_SUITABLE_FOR_STREAM_START;
 560                         } else {
 561                                 suitable_for_stream_start = SUITABLE_FOR_STREAM_START;
 562                         }
 563                         for (size_t i = 0; i < stream_indices.size(); ++i) {
 564                                 servers->add_data(stream_indices[i], inner_data, size, suitable_for_stream_start);
 565                         }
 566                 }
 567
 568                 // Consume the block. This isn't the most efficient way of dealing with things
 569                 // should we have many blocks, but these routines don't need to be too efficient
 570                 // anyway.
 571                 pending_data.erase(pending_data.begin(), pending_data.begin() + sizeof(metacube2_block_header) + size);
 572                 has_metacube_header = false;
 573         }
 574 }
 575
 576 void HTTPInput::drop_pending_data(size_t num_bytes)
 577 {
 578         if (num_bytes == 0) {
 579                 return;
 580         }
 581         log(WARNING, "[%s] Dropping %lld junk bytes; not a Metacube2 stream, or data was dropped from the middle of the stream",
 582                 url.c_str(), (long long)num_bytes);
 583         assert(pending_data.size() >= num_bytes);
 584         pending_data.erase(pending_data.begin(), pending_data.begin() + num_bytes);
 585 }
 586
 587 void HTTPInput::add_destination(int stream_index)
 588 {
 589         stream_indices.push_back(stream_index);
 590         servers->set_header(stream_index, http_header, stream_header);
 591 }
 592
 593 InputStats HTTPInput::get_stats() const
 594 {
 595         MutexLock lock(&stats_mutex);
 596         return stats;
 597 }