src/Network/HTTPClient.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, the Friendica project
   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\Core\Config\IConfig;
  27 use Friendica\Core\System;
  28 use Friendica\Util\Network;
  29 use Friendica\Util\Profiler;
  30 use GuzzleHttp\Client;
  31 use GuzzleHttp\Cookie\FileCookieJar;
  32 use GuzzleHttp\Exception\RequestException;
  33 use GuzzleHttp\Exception\TransferException;
  34 use GuzzleHttp\RequestOptions;
  35 use Psr\Http\Message\ResponseInterface;
  36 use Psr\Log\LoggerInterface;
  37
  38 /**
  39  * Performs HTTP requests to a given URL
  40  */
  41 class HTTPClient implements IHTTPClient
  42 {
  43         /** @var LoggerInterface */
  44         private $logger;
  45         /** @var Profiler */
  46         private $profiler;
  47         /** @var IConfig */
  48         private $config;
  49         /** @var string */
  50         private $userAgent;
  51         /** @var Client */
  52         private $client;
  53
  54         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, string $userAgent, Client $client)
  55         {
  56                 $this->logger    = $logger;
  57                 $this->profiler  = $profiler;
  58                 $this->config    = $config;
  59                 $this->userAgent = $userAgent;
  60                 $this->client    = $client;
  61         }
  62
  63         /**
  64          * @throws HTTPException\InternalServerErrorException
  65          */
  66         protected function request(string $method, string $url, array $opts = []): IHTTPResult
  67         {
  68                 $this->profiler->startRecording('network');
  69
  70                 if (Network::isLocalLink($url)) {
  71                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
  72                 }
  73
  74                 if (strlen($url) > 1000) {
  75                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  76                         $this->profiler->stopRecording();
  77                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  78                 }
  79
  80                 $parts2     = [];
  81                 $parts      = parse_url($url);
  82                 $path_parts = explode('/', $parts['path'] ?? '');
  83                 foreach ($path_parts as $part) {
  84                         if (strlen($part) <> mb_strlen($part)) {
  85                                 $parts2[] = rawurlencode($part);
  86                         } else {
  87                                 $parts2[] = $part;
  88                         }
  89                 }
  90                 $parts['path'] = implode('/', $parts2);
  91                 $url           = Network::unparseURL($parts);
  92
  93                 if (Network::isUrlBlocked($url)) {
  94                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  95                         $this->profiler->stopRecording();
  96                         return CurlResult::createErrorCurl($url);
  97                 }
  98
  99                 $conf = [];
 100
 101                 if (!empty($opts['cookiejar'])) {
 102                         $jar                           = new FileCookieJar($opts['cookiejar']);
 103                         $conf[RequestOptions::COOKIES] = $jar;
 104                 }
 105
 106                 $header = [];
 107
 108                 if (!empty($opts['accept_content'])) {
 109                         array_push($header, 'Accept: ' . $opts['accept_content']);
 110                 }
 111
 112                 if (!empty($opts['header'])) {
 113                         $header = array_merge($opts['header'], $header);
 114                 }
 115
 116                 if (!empty($opts['headers'])) {
 117                         $this->logger->notice('Wrong option \'headers\' used.');
 118                         $header = array_merge($opts['headers'], $header);
 119                 }
 120
 121                 $conf[RequestOptions::HEADERS] = array_merge($this->client->getConfig(RequestOptions::HEADERS), $header);
 122
 123                 if (!empty($opts['timeout'])) {
 124                         $conf[RequestOptions::TIMEOUT] = $opts['timeout'];
 125                 }
 126
 127                 $conf[RequestOptions::ON_HEADERS] = function (ResponseInterface $response) use ($opts) {
 128                         if (!empty($opts['content_length']) &&
 129                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 130                                 throw new TransferException('The file is too big!');
 131                         }
 132                 };
 133
 134                 try {
 135                         $response = $this->client->$method($url, $conf);
 136                         return new GuzzleResponse($response, $url);
 137                 } catch (TransferException $exception) {
 138                         if ($exception instanceof RequestException &&
 139                                 $exception->hasResponse()) {
 140                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), '');
 141                         } else {
 142                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), '');
 143                         }
 144                 } finally {
 145                         $this->profiler->stopRecording();
 146                 }
 147         }
 148
 149         /** {@inheritDoc}
 150          *
 151          * @throws HTTPException\InternalServerErrorException
 152          */
 153         public function head(string $url, array $opts = []): IHTTPResult
 154         {
 155                 return $this->request('head', $url, $opts);
 156         }
 157
 158         /**
 159          * {@inheritDoc}
 160          */
 161         public function get(string $url, array $opts = []): IHTTPResult
 162         {
 163                 return $this->request('get', $url, $opts);
 164         }
 165
 166         /**
 167          * {@inheritDoc}
 168          *
 169          * @param int $redirects The recursion counter for internal use - default 0
 170          *
 171          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 172          */
 173         public function post(string $url, $params, array $headers = [], int $timeout = 0, &$redirects = 0)
 174         {
 175                 $this->profiler->startRecording('network');
 176
 177                 if (Network::isLocalLink($url)) {
 178                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 179                 }
 180
 181                 if (Network::isUrlBlocked($url)) {
 182                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 183                         $this->profiler->stopRecording();
 184                         return CurlResult::createErrorCurl($url);
 185                 }
 186
 187                 $ch = curl_init($url);
 188
 189                 if (($redirects > 8) || (!$ch)) {
 190                         $this->profiler->stopRecording();
 191                         return CurlResult::createErrorCurl($url);
 192                 }
 193
 194                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 195
 196                 curl_setopt($ch, CURLOPT_HEADER, true);
 197                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 198                 curl_setopt($ch, CURLOPT_POST, 1);
 199                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 200                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 201
 202                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 203                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 204                 }
 205
 206                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 207
 208                 if (intval($timeout)) {
 209                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 210                 } else {
 211                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 212                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 213                 }
 214
 215                 if (!empty($headers)) {
 216                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 217                 }
 218
 219                 $check_cert = $this->config->get('system', 'verifyssl');
 220                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 221
 222                 if ($check_cert) {
 223                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 224                 }
 225
 226                 $proxy = $this->config->get('system', 'proxy');
 227
 228                 if (!empty($proxy)) {
 229                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 230                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 231                         $proxyuser = $this->config->get('system', 'proxyuser');
 232                         if (!empty($proxyuser)) {
 233                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 234                         }
 235                 }
 236
 237                 // don't let curl abort the entire application
 238                 // if it throws any errors.
 239
 240                 $s = @curl_exec($ch);
 241
 242                 $curl_info = curl_getinfo($ch);
 243
 244                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 245
 246                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 247                         $redirects++;
 248                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 249                         curl_close($ch);
 250                         $this->profiler->stopRecording();
 251                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 252                 }
 253
 254                 curl_close($ch);
 255
 256                 $this->profiler->stopRecording();
 257
 258                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 259                 if ($curlResponse->getReturnCode() == 417) {
 260                         $redirects++;
 261
 262                         if (empty($headers)) {
 263                                 $headers = ['Expect:'];
 264                         } else {
 265                                 if (!in_array('Expect:', $headers)) {
 266                                         array_push($headers, 'Expect:');
 267                                 }
 268                         }
 269                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 270                         return $this->post($url, $params, $headers, $redirects, $timeout);
 271                 }
 272
 273                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 274
 275                 return $curlResponse;
 276         }
 277
 278         /**
 279          * {@inheritDoc}
 280          */
 281         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 282         {
 283                 if (Network::isLocalLink($url)) {
 284                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 285                 }
 286
 287                 if (Network::isUrlBlocked($url)) {
 288                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 289                         return $url;
 290                 }
 291
 292                 if (Network::isRedirectBlocked($url)) {
 293                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 294                         return $url;
 295                 }
 296
 297                 $url = Network::stripTrackingQueryParams($url);
 298
 299                 if ($depth > 10) {
 300                         return $url;
 301                 }
 302
 303                 $url = trim($url, "'");
 304
 305                 $this->profiler->startRecording('network');
 306
 307                 $ch = curl_init();
 308                 curl_setopt($ch, CURLOPT_URL, $url);
 309                 curl_setopt($ch, CURLOPT_HEADER, 1);
 310                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 311                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 312                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 313                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 314                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 315
 316                 curl_exec($ch);
 317                 $curl_info = @curl_getinfo($ch);
 318                 $http_code = $curl_info['http_code'];
 319                 curl_close($ch);
 320
 321                 $this->profiler->stopRecording();
 322
 323                 if ($http_code == 0) {
 324                         return $url;
 325                 }
 326
 327                 if (in_array($http_code, ['301', '302'])) {
 328                         if (!empty($curl_info['redirect_url'])) {
 329                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 330                         } elseif (!empty($curl_info['location'])) {
 331                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 332                         }
 333                 }
 334
 335                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 336                 if (!$fetchbody) {
 337                         return $this->finalUrl($url, ++$depth, true);
 338                 }
 339
 340                 // if the file is too large then exit
 341                 if ($curl_info["download_content_length"] > 1000000) {
 342                         return $url;
 343                 }
 344
 345                 // if it isn't a HTML file then exit
 346                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 347                         return $url;
 348                 }
 349
 350                 $this->profiler->startRecording('network');
 351
 352                 $ch = curl_init();
 353                 curl_setopt($ch, CURLOPT_URL, $url);
 354                 curl_setopt($ch, CURLOPT_HEADER, 0);
 355                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 356                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 357                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 358                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 359                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 360
 361                 $body = curl_exec($ch);
 362                 curl_close($ch);
 363
 364                 $this->profiler->stopRecording();
 365
 366                 if (trim($body) == "") {
 367                         return $url;
 368                 }
 369
 370                 // Check for redirect in meta elements
 371                 $doc = new DOMDocument();
 372                 @$doc->loadHTML($body);
 373
 374                 $xpath = new DomXPath($doc);
 375
 376                 $list = $xpath->query("//meta[@content]");
 377                 foreach ($list as $node) {
 378                         $attr = [];
 379                         if ($node->attributes->length) {
 380                                 foreach ($node->attributes as $attribute) {
 381                                         $attr[$attribute->name] = $attribute->value;
 382                                 }
 383                         }
 384
 385                         if (@$attr["http-equiv"] == 'refresh') {
 386                                 $path = $attr["content"];
 387                                 $pathinfo = explode(";", $path);
 388                                 foreach ($pathinfo as $value) {
 389                                         if (substr(strtolower($value), 0, 4) == "url=") {
 390                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 391                                         }
 392                                 }
 393                         }
 394                 }
 395
 396                 return $url;
 397         }
 398
 399         /**
 400          * {@inheritDoc}
 401          */
 402         public function fetch(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '')
 403         {
 404                 $ret = $this->fetchFull($url, $timeout, $accept_content, $cookiejar);
 405
 406                 return $ret->getBody();
 407         }
 408
 409         /**
 410          * {@inheritDoc}
 411          */
 412         public function fetchFull(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '')
 413         {
 414                 return $this->get(
 415                         $url,
 416                         [
 417                                 'timeout'        => $timeout,
 418                                 'accept_content' => $accept_content,
 419                                 'cookiejar'      => $cookiejar
 420                         ]
 421                 );
 422         }
 423 }