src/Network/HTTPRequest.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU APGL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Network;
  23
  24 use DOMDocument;
  25 use DomXPath;
  26 use Friendica\App;
  27 use Friendica\Core\Config\IConfig;
  28 use Friendica\Core\System;
  29 use Friendica\Util\Network;
  30 use Friendica\Util\Profiler;
  31 use GuzzleHttp\Client;
  32 use GuzzleHttp\Exception\RequestException;
  33 use GuzzleHttp\Exception\TransferException;
  34 use Psr\Http\Message\RequestInterface;
  35 use Psr\Http\Message\ResponseInterface;
  36 use Psr\Http\Message\UriInterface;
  37 use Psr\Log\LoggerInterface;
  38
  39 /**
  40  * Performs HTTP requests to a given URL
  41  */
  42 class HTTPRequest implements IHTTPRequest
  43 {
  44         /** @var LoggerInterface */
  45         private $logger;
  46         /** @var Profiler */
  47         private $profiler;
  48         /** @var IConfig */
  49         private $config;
  50         /** @var string */
  51         private $baseUrl;
  52
  53         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, App\BaseURL $baseUrl)
  54         {
  55                 $this->logger   = $logger;
  56                 $this->profiler = $profiler;
  57                 $this->config   = $config;
  58                 $this->baseUrl  = $baseUrl->get();
  59         }
  60
  61         /**
  62          * {@inheritDoc}
  63          */
  64         public function get(string $url, bool $binary = false, array $opts = [])
  65         {
  66                 $stamp1 = microtime(true);
  67
  68                 if (strlen($url) > 1000) {
  69                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  70                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  71                 }
  72
  73                 $parts2     = [];
  74                 $parts      = parse_url($url);
  75                 $path_parts = explode('/', $parts['path'] ?? '');
  76                 foreach ($path_parts as $part) {
  77                         if (strlen($part) <> mb_strlen($part)) {
  78                                 $parts2[] = rawurlencode($part);
  79                         } else {
  80                                 $parts2[] = $part;
  81                         }
  82                 }
  83                 $parts['path'] = implode('/', $parts2);
  84                 $url           = Network::unparseURL($parts);
  85
  86                 if (Network::isUrlBlocked($url)) {
  87                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  88                         return CurlResult::createErrorCurl($url);
  89                 }
  90
  91                 $curlOptions = [];
  92
  93                 $curlOptions[CURLOPT_HEADER] = true;
  94
  95                 if (!empty($opts['cookiejar'])) {
  96                         $curlOptions[CURLOPT_COOKIEJAR] = $opts["cookiejar"];
  97                         $curlOptions[CURLOPT_COOKIEFILE] = $opts["cookiejar"];
  98                 }
  99
 100                 // These settings aren't needed. We're following the location already.
 101                 //      $curlOptions[CURLOPT_FOLLOWLOCATION] =true;
 102                 //      $curlOptions[CURLOPT_MAXREDIRS] = 5;
 103
 104                 if (!empty($opts['accept_content'])) {
 105                         $curlOptions[CURLOPT_HTTPHEADER][] = ['Accept: ' . $opts['accept_content']];
 106                 }
 107
 108                 if (!empty($opts['header'])) {
 109                         $curlOptions[CURLOPT_HTTPHEADER][] = $opts['header'];
 110                 }
 111
 112                 $curlOptions[CURLOPT_RETURNTRANSFER] = true;
 113                 $curlOptions[CURLOPT_USERAGENT] = $this->getUserAgent();
 114
 115                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 116
 117                 if ($range > 0) {
 118                         $curlOptions[CURLOPT_RANGE] = '0-' . $range;
 119                 }
 120
 121                 // Without this setting it seems as if some webservers send compressed content
 122                 // This seems to confuse curl so that it shows this uncompressed.
 123                 /// @todo  We could possibly set this value to "gzip" or something similar
 124                 $curlOptions[CURLOPT_ENCODING] = '';
 125
 126                 if (!empty($opts['headers'])) {
 127                         $curlOptions[CURLOPT_HTTPHEADER][] = $opts['headers'];
 128                 }
 129
 130                 if (!empty($opts['nobody'])) {
 131                         $curlOptions[CURLOPT_NOBODY] = $opts['nobody'];
 132                 }
 133
 134                 $curlOptions[CURLOPT_CONNECTTIMEOUT] = 10;
 135
 136                 if (!empty($opts['timeout'])) {
 137                         $curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 138                 } else {
 139                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 140                         $curlOptions[CURLOPT_TIMEOUT] = intval($curl_time);
 141                 }
 142
 143                 // by default we will allow self-signed certs
 144                 // but you can override this
 145
 146                 $check_cert = $this->config->get('system', 'verifyssl');
 147                 $curlOptions[CURLOPT_SSL_VERIFYPEER] = ($check_cert) ? true : false;
 148
 149                 if ($check_cert) {
 150                         $curlOptions[CURLOPT_SSL_VERIFYHOST] = 2;
 151                 }
 152
 153                 $proxy = $this->config->get('system', 'proxy');
 154
 155                 if (!empty($proxy)) {
 156                         $curlOptions[CURLOPT_HTTPPROXYTUNNEL] = 1;
 157                         $curlOptions[CURLOPT_PROXY] = $proxy;
 158                         $proxyuser = $this->config->get('system', 'proxyuser');
 159
 160                         if (!empty($proxyuser)) {
 161                                 $curlOptions[CURLOPT_PROXYUSERPWD] = $proxyuser;
 162                         }
 163                 }
 164
 165                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 166                         $curlOptions[CURLOPT_IPRESOLVE] = CURL_IPRESOLVE_V4;
 167                 }
 168
 169                 if ($binary) {
 170                         $curlOptions[CURLOPT_BINARYTRANSFER] = 1;
 171                 }
 172
 173                 $onRedirect = function(
 174                         RequestInterface $request,
 175                         ResponseInterface $response,
 176                         UriInterface $uri
 177                 ) {
 178                         $this->logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 179                 };
 180
 181                 $onHeaders = function (ResponseInterface $response) use ($opts) {
 182                         if (!empty($opts['content_length']) &&
 183                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 184                                 throw new TransferException('The file is too big!');
 185                         }
 186                 };
 187
 188                 $client = new Client([
 189                         'allow_redirect' => [
 190                                 'max' => 8,
 191                                 'on_redirect' => $onRedirect,
 192                                 'on_headers' => $onHeaders,
 193                                 'track_redirect' => true,
 194                                 'strict' => true,
 195                                 'referer' => true,
 196                         ],
 197                         'curl' => $curlOptions
 198                 ]);
 199
 200                 try {
 201                         $response = $client->get($url);
 202                         return new GuzzleResponse($response, $url);
 203                 } catch (TransferException $exception) {
 204                         if ($exception instanceof RequestException &&
 205                                 $exception->hasResponse()) {
 206                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 207                         } else {
 208                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 209                         }
 210                 } finally {
 211                         $this->profiler->saveTimestamp($stamp1, 'network');
 212                 }
 213         }
 214
 215         /**
 216          * {@inheritDoc}
 217          *
 218          * @param int $redirects The recursion counter for internal use - default 0
 219          *
 220          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 221          */
 222         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 223         {
 224                 $stamp1 = microtime(true);
 225
 226                 if (Network::isUrlBlocked($url)) {
 227                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 228                         return CurlResult::createErrorCurl($url);
 229                 }
 230
 231                 $ch = curl_init($url);
 232
 233                 if (($redirects > 8) || (!$ch)) {
 234                         return CurlResult::createErrorCurl($url);
 235                 }
 236
 237                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 238
 239                 curl_setopt($ch, CURLOPT_HEADER, true);
 240                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 241                 curl_setopt($ch, CURLOPT_POST, 1);
 242                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 243                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 244
 245                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 246                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 247                 }
 248
 249                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 250
 251                 if (intval($timeout)) {
 252                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 253                 } else {
 254                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 255                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 256                 }
 257
 258                 if (!empty($headers)) {
 259                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 260                 }
 261
 262                 $check_cert = $this->config->get('system', 'verifyssl');
 263                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 264
 265                 if ($check_cert) {
 266                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 267                 }
 268
 269                 $proxy = $this->config->get('system', 'proxy');
 270
 271                 if (!empty($proxy)) {
 272                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 273                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 274                         $proxyuser = $this->config->get('system', 'proxyuser');
 275                         if (!empty($proxyuser)) {
 276                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 277                         }
 278                 }
 279
 280                 // don't let curl abort the entire application
 281                 // if it throws any errors.
 282
 283                 $s = @curl_exec($ch);
 284
 285                 $curl_info = curl_getinfo($ch);
 286
 287                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 288
 289                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 290                         $redirects++;
 291                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 292                         curl_close($ch);
 293                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 294                 }
 295
 296                 curl_close($ch);
 297
 298                 $this->profiler->saveTimestamp($stamp1, 'network');
 299
 300                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 301                 if ($curlResponse->getReturnCode() == 417) {
 302                         $redirects++;
 303
 304                         if (empty($headers)) {
 305                                 $headers = ['Expect:'];
 306                         } else {
 307                                 if (!in_array('Expect:', $headers)) {
 308                                         array_push($headers, 'Expect:');
 309                                 }
 310                         }
 311                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 312                         return $this->post($url, $params, $headers, $redirects, $timeout);
 313                 }
 314
 315                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 316
 317                 return $curlResponse;
 318         }
 319
 320         /**
 321          * {@inheritDoc}
 322          */
 323         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 324         {
 325                 if (Network::isUrlBlocked($url)) {
 326                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 327                         return $url;
 328                 }
 329
 330                 if (Network::isRedirectBlocked($url)) {
 331                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 332                         return $url;
 333                 }
 334
 335                 $url = Network::stripTrackingQueryParams($url);
 336
 337                 if ($depth > 10) {
 338                         return $url;
 339                 }
 340
 341                 $url = trim($url, "'");
 342
 343                 $stamp1 = microtime(true);
 344
 345                 $ch = curl_init();
 346                 curl_setopt($ch, CURLOPT_URL, $url);
 347                 curl_setopt($ch, CURLOPT_HEADER, 1);
 348                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 349                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 350                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 351                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 352                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 353
 354                 curl_exec($ch);
 355                 $curl_info = @curl_getinfo($ch);
 356                 $http_code = $curl_info['http_code'];
 357                 curl_close($ch);
 358
 359                 $this->profiler->saveTimestamp($stamp1, "network");
 360
 361                 if ($http_code == 0) {
 362                         return $url;
 363                 }
 364
 365                 if (in_array($http_code, ['301', '302'])) {
 366                         if (!empty($curl_info['redirect_url'])) {
 367                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 368                         } elseif (!empty($curl_info['location'])) {
 369                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 370                         }
 371                 }
 372
 373                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 374                 if (!$fetchbody) {
 375                         return $this->finalUrl($url, ++$depth, true);
 376                 }
 377
 378                 // if the file is too large then exit
 379                 if ($curl_info["download_content_length"] > 1000000) {
 380                         return $url;
 381                 }
 382
 383                 // if it isn't a HTML file then exit
 384                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 385                         return $url;
 386                 }
 387
 388                 $stamp1 = microtime(true);
 389
 390                 $ch = curl_init();
 391                 curl_setopt($ch, CURLOPT_URL, $url);
 392                 curl_setopt($ch, CURLOPT_HEADER, 0);
 393                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 394                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 395                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 396                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 397                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 398
 399                 $body = curl_exec($ch);
 400                 curl_close($ch);
 401
 402                 $this->profiler->saveTimestamp($stamp1, "network");
 403
 404                 if (trim($body) == "") {
 405                         return $url;
 406                 }
 407
 408                 // Check for redirect in meta elements
 409                 $doc = new DOMDocument();
 410                 @$doc->loadHTML($body);
 411
 412                 $xpath = new DomXPath($doc);
 413
 414                 $list = $xpath->query("//meta[@content]");
 415                 foreach ($list as $node) {
 416                         $attr = [];
 417                         if ($node->attributes->length) {
 418                                 foreach ($node->attributes as $attribute) {
 419                                         $attr[$attribute->name] = $attribute->value;
 420                                 }
 421                         }
 422
 423                         if (@$attr["http-equiv"] == 'refresh') {
 424                                 $path = $attr["content"];
 425                                 $pathinfo = explode(";", $path);
 426                                 foreach ($pathinfo as $value) {
 427                                         if (substr(strtolower($value), 0, 4) == "url=") {
 428                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 429                                         }
 430                                 }
 431                         }
 432                 }
 433
 434                 return $url;
 435         }
 436
 437         /**
 438          * {@inheritDoc}
 439          *
 440          * @param int $redirects The recursion counter for internal use - default 0
 441          *
 442          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 443          */
 444         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 445         {
 446                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 447
 448                 return $ret->getBody();
 449         }
 450
 451         /**
 452          * {@inheritDoc}
 453          *
 454          * @param int $redirects The recursion counter for internal use - default 0
 455          *
 456          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 457          */
 458         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 459         {
 460                 return $this->get(
 461                         $url,
 462                         $binary,
 463                         [
 464                                 'timeout'        => $timeout,
 465                                 'accept_content' => $accept_content,
 466                                 'cookiejar'      => $cookiejar
 467                         ]
 468                 );
 469         }
 470
 471         /**
 472          * {@inheritDoc}
 473          */
 474         public function getUserAgent()
 475         {
 476                 return
 477                         FRIENDICA_PLATFORM . " '" .
 478                         FRIENDICA_CODENAME . "' " .
 479                         FRIENDICA_VERSION . '-' .
 480                         DB_UPDATE_VERSION . '; ' .
 481                         $this->baseUrl;
 482         }
 483 }