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 <unistd.h>
  12 #include <map>
  13 #include <string>
  14 #include <utility>
  15 #include <vector>
  16
  17 #include "httpinput.h"
  18 #include "log.h"
  19 #include "metacube.h"
  20 #include "parse.h"
  21 #include "serverpool.h"
  22 #include "state.pb.h"
  23 #include "version.h"
  24
  25 using namespace std;
  26
  27 extern ServerPool *servers;
  28
  29 HTTPInput::HTTPInput(const string &url)
  30         : state(NOT_CONNECTED),
  31           url(url),
  32           has_metacube_header(false),
  33           sock(-1)
  34 {
  35 }
  36
  37 HTTPInput::HTTPInput(const InputProto &serialized)
  38         : state(State(serialized.state())),
  39           url(serialized.url()),
  40           request(serialized.request()),
  41           request_bytes_sent(serialized.request_bytes_sent()),
  42           response(serialized.response()),
  43           http_header(serialized.http_header()),
  44           has_metacube_header(serialized.has_metacube_header()),
  45           sock(serialized.sock())
  46 {
  47         pending_data.resize(serialized.pending_data().size());
  48         memcpy(&pending_data[0], serialized.pending_data().data(), serialized.pending_data().size());
  49
  50         string protocol;
  51         parse_url(url, &protocol, &host, &port, &path);  // Don't care if it fails.
  52
  53         // Older versions stored the extra \r\n in the HTTP header.
  54         // Strip it if we find it.
  55         if (http_header.size() >= 4 &&
  56             memcmp(http_header.data() + http_header.size() - 4, "\r\n\r\n", 4) == 0) {
  57                 http_header.resize(http_header.size() - 2);
  58         }
  59 }
  60
  61 void HTTPInput::close_socket()
  62 {
  63         int ret;
  64         do {
  65                 ret = close(sock);
  66         } while (ret == -1 && errno == EINTR);
  67
  68         if (ret == -1) {
  69                 log_perror("close()");
  70         }
  71 }
  72
  73 InputProto HTTPInput::serialize() const
  74 {
  75         InputProto serialized;
  76         serialized.set_state(state);
  77         serialized.set_url(url);
  78         serialized.set_request(request);
  79         serialized.set_request_bytes_sent(request_bytes_sent);
  80         serialized.set_response(response);
  81         serialized.set_http_header(http_header);
  82         serialized.set_pending_data(string(pending_data.begin(), pending_data.end()));
  83         serialized.set_has_metacube_header(has_metacube_header);
  84         serialized.set_sock(sock);
  85         return serialized;
  86 }
  87
  88 int HTTPInput::lookup_and_connect(const string &host, const string &port)
  89 {
  90         addrinfo *ai;
  91         int err = getaddrinfo(host.c_str(), port.c_str(), NULL, &ai);
  92         if (err == -1) {
  93                 log(WARNING, "[%s] Lookup of '%s' failed (%s).",
  94                         url.c_str(), host.c_str(), gai_strerror(err));
  95                 freeaddrinfo(ai);
  96                 return -1;
  97         }
  98
  99         addrinfo *base_ai = ai;
 100
 101         // Connect to everything in turn until we have a socket.
 102         while (ai && !should_stop) {
 103                 int sock = socket(ai->ai_family, SOCK_STREAM, IPPROTO_TCP);
 104                 if (sock == -1) {
 105                         // Could be e.g. EPROTONOSUPPORT. The show must go on.
 106                         continue;
 107                 }
 108
 109                 do {
 110                         err = connect(sock, ai->ai_addr, ai->ai_addrlen);
 111                 } while (err == -1 && errno == EINTR);
 112
 113                 if (err != -1) {
 114                         freeaddrinfo(base_ai);
 115                         return sock;
 116                 }
 117
 118                 do {
 119                         err = close(sock);
 120                 } while (err == -1 && errno == EINTR);
 121
 122                 if (err == -1) {
 123                         log_perror("close");
 124                         // Can still continue.
 125                 }
 126
 127                 ai = ai->ai_next;
 128         }
 129
 130         // Give the last one as error.
 131         log(WARNING, "[%s] Connect to '%s' failed (%s)",
 132                 url.c_str(), host.c_str(), strerror(errno));
 133         freeaddrinfo(base_ai);
 134         return -1;
 135 }
 136
 137 bool HTTPInput::parse_response(const std::string &request)
 138 {
 139         vector<string> lines = split_lines(response);
 140         if (lines.empty()) {
 141                 log(WARNING, "[%s] Empty HTTP response from input.", url.c_str());
 142                 return false;
 143         }
 144
 145         vector<string> first_line_tokens = split_tokens(lines[0]);
 146         if (first_line_tokens.size() < 2) {
 147                 log(WARNING, "[%s] Malformed response line '%s' from input.",
 148                         url.c_str(), lines[0].c_str());
 149                 return false;
 150         }
 151
 152         int response = atoi(first_line_tokens[1].c_str());
 153         if (response != 200) {
 154                 log(WARNING, "[%s] Non-200 response '%s' from input.",
 155                         url.c_str(), lines[0].c_str());
 156                 return false;
 157         }
 158
 159         multimap<string, string> parameters;
 160         for (size_t i = 1; i < lines.size(); ++i) {
 161                 size_t split = lines[i].find(":");
 162                 if (split == string::npos) {
 163                         log(WARNING, "[%s] Ignoring malformed HTTP response line '%s'",
 164                                 url.c_str(), lines[i].c_str());
 165                         continue;
 166                 }
 167
 168                 string key(lines[i].begin(), lines[i].begin() + split);
 169
 170                 // Skip any spaces after the colon.
 171                 do {
 172                         ++split;
 173                 } while (split < lines[i].size() && lines[i][split] == ' ');
 174
 175                 string value(lines[i].begin() + split, lines[i].end());
 176
 177                 // Remove “Content-encoding: metacube”.
 178                 // TODO: Make case-insensitive.
 179                 if (key == "Content-encoding" && value == "metacube") {
 180                         continue;
 181                 }
 182
 183                 parameters.insert(make_pair(key, value));
 184         }
 185
 186         // Change “Server: foo” to “Server: metacube/0.1 (reflecting: foo)”
 187         // TODO: Make case-insensitive.
 188         // XXX: Use a Via: instead?
 189         if (parameters.count("Server") == 0) {
 190                 parameters.insert(make_pair("Server", SERVER_IDENTIFICATION));
 191         } else {
 192                 for (multimap<string, string>::iterator it = parameters.begin();
 193                      it != parameters.end();
 194                      ++it) {
 195                         if (it->first != "Server") {
 196                                 continue;
 197                         }
 198                         it->second = SERVER_IDENTIFICATION " (reflecting: " + it->second + ")";
 199                 }
 200         }
 201
 202         // Set “Connection: close”.
 203         // TODO: Make case-insensitive.
 204         parameters.erase("Connection");
 205         parameters.insert(make_pair("Connection", "close"));
 206
 207         // Construct the new HTTP header.
 208         http_header = "HTTP/1.0 200 OK\r\n";
 209         for (multimap<string, string>::iterator it = parameters.begin();
 210              it != parameters.end();
 211              ++it) {
 212                 http_header.append(it->first + ": " + it->second + "\r\n");
 213         }
 214
 215         for (size_t i = 0; i < stream_ids.size(); ++i) {
 216                 servers->set_header(stream_ids[i], http_header, "");
 217         }
 218
 219         return true;
 220 }
 221
 222 void HTTPInput::do_work()
 223 {
 224         while (!should_stop) {
 225                 if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 226                         // Since we are non-blocking, we need to wait for the right state first.
 227                         // Wait up to 50 ms, then check should_stop.
 228                         pollfd pfd;
 229                         pfd.fd = sock;
 230                         pfd.events = (state == SENDING_REQUEST) ? POLLOUT : POLLIN;
 231                         pfd.events |= POLLRDHUP;
 232
 233                         int nfds = poll(&pfd, 1, 50);
 234                         if (nfds == 0 || (nfds == -1 && errno == EINTR)) {
 235                                 continue;
 236                         }
 237                         if (nfds == -1) {
 238                                 log_perror("poll");
 239                                 state = CLOSING_SOCKET;
 240                         }
 241                 }
 242
 243                 switch (state) {
 244                 case NOT_CONNECTED:
 245                         request.clear();
 246                         request_bytes_sent = 0;
 247                         response.clear();
 248                         pending_data.clear();
 249                         for (size_t i = 0; i < stream_ids.size(); ++i) {
 250                                 servers->set_header(stream_ids[i], "", "");
 251                         }
 252
 253                         {
 254                                 string protocol;  // Thrown away.
 255                                 if (!parse_url(url, &protocol, &host, &port, &path)) {
 256                                         log(WARNING, "[%s] Failed to parse URL '%s'", url.c_str(), url.c_str());
 257                                         break;
 258                                 }
 259                         }
 260
 261                         sock = lookup_and_connect(host, port);
 262                         if (sock != -1) {
 263                                 // Yay, successful connect. Try to set it as nonblocking.
 264                                 int one = 1;
 265                                 if (ioctl(sock, FIONBIO, &one) == -1) {
 266                                         log_perror("ioctl(FIONBIO)");
 267                                         state = CLOSING_SOCKET;
 268                                 } else {
 269                                         state = SENDING_REQUEST;
 270                                         request = "GET " + path + " HTTP/1.0\r\nUser-Agent: cubemap\r\n\r\n";
 271                                         request_bytes_sent = 0;
 272                                 }
 273                         }
 274                         break;
 275                 case SENDING_REQUEST: {
 276                         size_t to_send = request.size() - request_bytes_sent;
 277                         int ret;
 278
 279                         do {
 280                                 ret = write(sock, request.data() + request_bytes_sent, to_send);
 281                         } while (ret == -1 && errno == EINTR);
 282
 283                         if (ret == -1) {
 284                                 log_perror("write");
 285                                 state = CLOSING_SOCKET;
 286                                 continue;
 287                         }
 288
 289                         assert(ret >= 0);
 290                         request_bytes_sent += ret;
 291
 292                         if (request_bytes_sent == request.size()) {
 293                                 state = RECEIVING_HEADER;
 294                         }
 295                         break;
 296                 }
 297                 case RECEIVING_HEADER: {
 298                         char buf[4096];
 299                         int ret;
 300
 301                         do {
 302                                 ret = read(sock, buf, sizeof(buf));
 303                         } while (ret == -1 && errno == EINTR);
 304
 305                         if (ret == -1) {
 306                                 log_perror("read");
 307                                 state = CLOSING_SOCKET;
 308                                 continue;
 309                         }
 310
 311                         if (ret == 0) {
 312                                 // This really shouldn't happen...
 313                                 log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 314                                            url.c_str());
 315                                 state = CLOSING_SOCKET;
 316                                 continue;
 317                         }
 318
 319                         RequestParseStatus status = wait_for_double_newline(&response, buf, ret);
 320
 321                         if (status == RP_OUT_OF_SPACE) {
 322                                 log(WARNING, "[%s] Sever sent overlong HTTP response!", url.c_str());
 323                                 state = CLOSING_SOCKET;
 324                                 continue;
 325                         } else if (status == RP_NOT_FINISHED_YET) {
 326                                 continue;
 327                         }
 328
 329                         // OK, so we're fine, but there might be some of the actual data after the response.
 330                         // We'll need to deal with that separately.
 331                         string extra_data;
 332                         if (status == RP_EXTRA_DATA) {
 333                                 char *ptr = static_cast<char *>(
 334                                         memmem(response.data(), response.size(), "\r\n\r\n", 4));
 335                                 assert(ptr != NULL);
 336                                 extra_data = string(ptr, &response[0] + response.size());
 337                                 response.resize(ptr - response.data());
 338                         }
 339
 340                         if (!parse_response(response)) {
 341                                 state = CLOSING_SOCKET;
 342                                 continue;
 343                         }
 344
 345                         if (!extra_data.empty()) {
 346                                 process_data(&extra_data[0], extra_data.size());
 347                         }
 348
 349                         log(INFO, "[%s] Connected to '%s', receiving data.",
 350                                    url.c_str(), url.c_str());
 351                         state = RECEIVING_DATA;
 352                         break;
 353                 }
 354                 case RECEIVING_DATA: {
 355                         char buf[4096];
 356                         int ret;
 357
 358                         do {
 359                                 ret = read(sock, buf, sizeof(buf));
 360                         } while (ret == -1 && errno == EINTR);
 361
 362                         if (ret == -1) {
 363                                 log_perror("read");
 364                                 state = CLOSING_SOCKET;
 365                                 continue;
 366                         }
 367
 368                         if (ret == 0) {
 369                                 // This really shouldn't happen...
 370                                 log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 371                                            url.c_str());
 372                                 state = CLOSING_SOCKET;
 373                                 continue;
 374                         }
 375
 376                         process_data(buf, ret);
 377                         break;
 378                 }
 379                 case CLOSING_SOCKET: {
 380                         int err;
 381                         do {
 382                                 err = close(sock);
 383                         } while (err == -1 && errno == EINTR);
 384
 385                         if (err == -1) {
 386                                 log_perror("close");
 387                         }
 388
 389                         state = NOT_CONNECTED;
 390                         break;
 391                 }
 392                 default:
 393                         assert(false);
 394                 }
 395
 396                 // If we are still in NOT_CONNECTED, either something went wrong,
 397                 // or the connection just got closed.
 398                 // The earlier steps have already given the error message, if any.
 399                 if (state == NOT_CONNECTED && !should_stop) {
 400                         log(INFO, "[%s] Waiting 0.2 second and restarting...", url.c_str());
 401                         usleep(200000);
 402                 }
 403         }
 404 }
 405
 406 void HTTPInput::process_data(char *ptr, size_t bytes)
 407 {
 408         pending_data.insert(pending_data.end(), ptr, ptr + bytes);
 409
 410         for ( ;; ) {
 411                 // If we don't have enough data (yet) for even the Metacube header, just return.
 412                 if (pending_data.size() < sizeof(metacube_block_header)) {
 413                         return;
 414                 }
 415
 416                 // Make sure we have the Metacube sync header at the start.
 417                 // We may need to skip over junk data (it _should_ not happen, though).
 418                 if (!has_metacube_header) {
 419                         char *ptr = static_cast<char *>(
 420                                 memmem(pending_data.data(), pending_data.size(),
 421                                        METACUBE_SYNC, strlen(METACUBE_SYNC)));
 422                         if (ptr == NULL) {
 423                                 // OK, so we didn't find the sync marker. We know then that
 424                                 // we do not have the _full_ marker in the buffer, but we
 425                                 // could have N-1 bytes. Drop everything before that,
 426                                 // and then give up.
 427                                 drop_pending_data(pending_data.size() - (strlen(METACUBE_SYNC) - 1));
 428                                 return;
 429                         } else {
 430                                 // Yay, we found the header. Drop everything (if anything) before it.
 431                                 drop_pending_data(ptr - pending_data.data());
 432                                 has_metacube_header = true;
 433
 434                                 // Re-check that we have the entire header; we could have dropped data.
 435                                 if (pending_data.size() < sizeof(metacube_block_header)) {
 436                                         return;
 437                                 }
 438                         }
 439                 }
 440
 441                 // Now it's safe to read the header.
 442                 metacube_block_header *hdr = reinterpret_cast<metacube_block_header *>(pending_data.data());
 443                 assert(memcmp(hdr->sync, METACUBE_SYNC, sizeof(hdr->sync)) == 0);
 444                 uint32_t size = ntohl(hdr->size);
 445                 uint32_t flags = ntohl(hdr->flags);
 446
 447                 // See if we have the entire block. If not, wait for more data.
 448                 if (pending_data.size() < sizeof(metacube_block_header) + size) {
 449                         return;
 450                 }
 451
 452                 // Send this block on to the data.
 453                 char *inner_data = pending_data.data() + sizeof(metacube_block_header);
 454                 if (flags & METACUBE_FLAGS_HEADER) {
 455                         string header(inner_data, inner_data + size);
 456                         for (size_t i = 0; i < stream_ids.size(); ++i) {
 457                                 servers->set_header(stream_ids[i], http_header, header);
 458                         }
 459                 } else {
 460                         for (size_t i = 0; i < stream_ids.size(); ++i) {
 461                                 servers->add_data(stream_ids[i], inner_data, size);
 462                         }
 463                 }
 464
 465                 // Consume the block. This isn't the most efficient way of dealing with things
 466                 // should we have many blocks, but these routines don't need to be too efficient
 467                 // anyway.
 468                 pending_data.erase(pending_data.begin(), pending_data.begin() + sizeof(metacube_block_header) + size);
 469                 has_metacube_header = false;
 470         }
 471 }
 472
 473 void HTTPInput::drop_pending_data(size_t num_bytes)
 474 {
 475         if (num_bytes == 0) {
 476                 return;
 477         }
 478         log(WARNING, "[%s] Dropping %lld junk bytes from stream, maybe it is not a Metacube stream?",
 479                 url.c_str(), (long long)num_bytes);
 480         pending_data.erase(pending_data.begin(), pending_data.begin() + num_bytes);
 481 }
 482