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

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