includes/HttpFunctions.php

   1 <?php
   2 /**
   3  * Various HTTP related functions.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup HTTP
  22  */
  23
  24 /**
  25  * @defgroup HTTP HTTP
  26  */
  27
  28 use MediaWiki\Logger\LoggerFactory;
  29
  30 /**
  31  * Various HTTP related functions
  32  * @ingroup HTTP
  33  */
  34 class Http {
  35         static public $httpEngine = false;
  36
  37         /**
  38          * Perform an HTTP request
  39          *
  40          * @param string $method HTTP method. Usually GET/POST
  41          * @param string $url Full URL to act on. If protocol-relative, will be expanded to an http:// URL
  42          * @param array $options Options to pass to MWHttpRequest object.
  43          *      Possible keys for the array:
  44          *    - timeout             Timeout length in seconds
  45          *    - connectTimeout      Timeout for connection, in seconds (curl only)
  46          *    - postData            An array of key-value pairs or a url-encoded form data
  47          *    - proxy               The proxy to use.
  48          *                          Otherwise it will use $wgHTTPProxy (if set)
  49          *                          Otherwise it will use the environment variable "http_proxy" (if set)
  50          *    - noProxy             Don't use any proxy at all. Takes precedence over proxy value(s).
  51          *    - sslVerifyHost       Verify hostname against certificate
  52          *    - sslVerifyCert       Verify SSL certificate
  53          *    - caInfo              Provide CA information
  54          *    - maxRedirects        Maximum number of redirects to follow (defaults to 5)
  55          *    - followRedirects     Whether to follow redirects (defaults to false).
  56          *                                  Note: this should only be used when the target URL is trusted,
  57          *                                  to avoid attacks on intranet services accessible by HTTP.
  58          *    - userAgent           A user agent, if you want to override the default
  59          *                          MediaWiki/$wgVersion
  60          * @param string $caller The method making this request, for profiling
  61          * @return string|bool (bool)false on failure or a string on success
  62          */
  63         public static function request( $method, $url, $options = array(), $caller = __METHOD__ ) {
  64                 wfDebug( "HTTP: $method: $url\n" );
  65
  66                 $options['method'] = strtoupper( $method );
  67
  68                 if ( !isset( $options['timeout'] ) ) {
  69                         $options['timeout'] = 'default';
  70                 }
  71                 if ( !isset( $options['connectTimeout'] ) ) {
  72                         $options['connectTimeout'] = 'default';
  73                 }
  74
  75                 $req = MWHttpRequest::factory( $url, $options, $caller );
  76                 $status = $req->execute();
  77
  78                 if ( $status->isOK() ) {
  79                         return $req->getContent();
  80                 } else {
  81                         $errors = $status->getErrorsByType( 'error' );
  82                         $logger = LoggerFactory::getInstance( 'http' );
  83                         $logger->warning( $status->getWikiText(),
  84                                 array( 'error' => $errors, 'caller' => $caller, 'content' => $req->getContent() ) );
  85                         return false;
  86                 }
  87         }
  88
  89         /**
  90          * Simple wrapper for Http::request( 'GET' )
  91          * @see Http::request()
  92          * @since 1.25 Second parameter $timeout removed. Second parameter
  93          * is now $options which can be given a 'timeout'
  94          *
  95          * @param string $url
  96          * @param array $options
  97          * @param string $caller The method making this request, for profiling
  98          * @return string
  99          */
 100         public static function get( $url, $options = array(), $caller = __METHOD__ ) {
 101                 $args = func_get_args();
 102                 if ( isset( $args[1] ) && ( is_string( $args[1] ) || is_numeric( $args[1] ) ) ) {
 103                         // Second was used to be the timeout
 104                         // And third parameter used to be $options
 105                         wfWarn( "Second parameter should not be a timeout.", 2 );
 106                         $options = isset( $args[2] ) && is_array( $args[2] ) ?
 107                                 $args[2] : array();
 108                         $options['timeout'] = $args[1];
 109                         $caller = __METHOD__;
 110                 }
 111                 return Http::request( 'GET', $url, $options, $caller );
 112         }
 113
 114         /**
 115          * Simple wrapper for Http::request( 'POST' )
 116          * @see Http::request()
 117          *
 118          * @param string $url
 119          * @param array $options
 120          * @param string $caller The method making this request, for profiling
 121          * @return string
 122          */
 123         public static function post( $url, $options = array(), $caller = __METHOD__ ) {
 124                 return Http::request( 'POST', $url, $options, $caller );
 125         }
 126
 127         /**
 128          * Check if the URL can be served by localhost
 129          *
 130          * @param string $url Full url to check
 131          * @return bool
 132          */
 133         public static function isLocalURL( $url ) {
 134                 global $wgCommandLineMode, $wgLocalVirtualHosts, $wgConf;
 135
 136                 if ( $wgCommandLineMode ) {
 137                         return false;
 138                 }
 139
 140                 // Extract host part
 141                 $matches = array();
 142                 if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
 143                         $host = $matches[1];
 144                         // Split up dotwise
 145                         $domainParts = explode( '.', $host );
 146                         // Check if this domain or any superdomain is listed as a local virtual host
 147                         $domainParts = array_reverse( $domainParts );
 148
 149                         $domain = '';
 150                         $countParts = count( $domainParts );
 151                         for ( $i = 0; $i < $countParts; $i++ ) {
 152                                 $domainPart = $domainParts[$i];
 153                                 if ( $i == 0 ) {
 154                                         $domain = $domainPart;
 155                                 } else {
 156                                         $domain = $domainPart . '.' . $domain;
 157                                 }
 158
 159                                 if ( in_array( $domain, $wgLocalVirtualHosts )
 160                                         || $wgConf->isLocalVHost( $domain )
 161                                 ) {
 162                                         return true;
 163                                 }
 164                         }
 165                 }
 166
 167                 return false;
 168         }
 169
 170         /**
 171          * A standard user-agent we can use for external requests.
 172          * @return string
 173          */
 174         public static function userAgent() {
 175                 global $wgVersion;
 176                 return "MediaWiki/$wgVersion";
 177         }
 178
 179         /**
 180          * Checks that the given URI is a valid one. Hardcoding the
 181          * protocols, because we only want protocols that both cURL
 182          * and php support.
 183          *
 184          * file:// should not be allowed here for security purpose (r67684)
 185          *
 186          * @todo FIXME this is wildly inaccurate and fails to actually check most stuff
 187          *
 188          * @param string $uri URI to check for validity
 189          * @return bool
 190          */
 191         public static function isValidURI( $uri ) {
 192                 return preg_match(
 193                         '/^https?:\/\/[^\/\s]\S*$/D',
 194                         $uri
 195                 );
 196         }
 197 }
 198
 199 /**
 200  * This wrapper class will call out to curl (if available) or fallback
 201  * to regular PHP if necessary for handling internal HTTP requests.
 202  *
 203  * Renamed from HttpRequest to MWHttpRequest to avoid conflict with
 204  * PHP's HTTP extension.
 205  */
 206 class MWHttpRequest {
 207         const SUPPORTS_FILE_POSTS = false;
 208
 209         protected $content;
 210         protected $timeout = 'default';
 211         protected $headersOnly = null;
 212         protected $postData = null;
 213         protected $proxy = null;
 214         protected $noProxy = false;
 215         protected $sslVerifyHost = true;
 216         protected $sslVerifyCert = true;
 217         protected $caInfo = null;
 218         protected $method = "GET";
 219         protected $reqHeaders = array();
 220         protected $url;
 221         protected $parsedUrl;
 222         protected $callback;
 223         protected $maxRedirects = 5;
 224         protected $followRedirects = false;
 225
 226         /**
 227          * @var CookieJar
 228          */
 229         protected $cookieJar;
 230
 231         protected $headerList = array();
 232         protected $respVersion = "0.9";
 233         protected $respStatus = "200 Ok";
 234         protected $respHeaders = array();
 235
 236         public $status;
 237
 238         /**
 239          * @var Profiler
 240          */
 241         protected $profiler;
 242
 243         /**
 244          * @var string
 245          */
 246         protected $profileName;
 247
 248         /**
 249          * @param string $url Url to use. If protocol-relative, will be expanded to an http:// URL
 250          * @param array $options (optional) extra params to pass (see Http::request())
 251          * @param string $caller The method making this request, for profiling
 252          * @param Profiler $profiler An instance of the profiler for profiling, or null
 253          */
 254         protected function __construct(
 255                 $url, $options = array(), $caller = __METHOD__, $profiler = null
 256         ) {
 257                 global $wgHTTPTimeout, $wgHTTPConnectTimeout;
 258
 259                 $this->url = wfExpandUrl( $url, PROTO_HTTP );
 260                 $this->parsedUrl = wfParseUrl( $this->url );
 261
 262                 if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) {
 263                         $this->status = Status::newFatal( 'http-invalid-url', $url );
 264                 } else {
 265                         $this->status = Status::newGood( 100 ); // continue
 266                 }
 267
 268                 if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) {
 269                         $this->timeout = $options['timeout'];
 270                 } else {
 271                         $this->timeout = $wgHTTPTimeout;
 272                 }
 273                 if ( isset( $options['connectTimeout'] ) && $options['connectTimeout'] != 'default' ) {
 274                         $this->connectTimeout = $options['connectTimeout'];
 275                 } else {
 276                         $this->connectTimeout = $wgHTTPConnectTimeout;
 277                 }
 278                 if ( isset( $options['userAgent'] ) ) {
 279                         $this->setUserAgent( $options['userAgent'] );
 280                 }
 281
 282                 $members = array( "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
 283                                 "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" );
 284
 285                 foreach ( $members as $o ) {
 286                         if ( isset( $options[$o] ) ) {
 287                                 // ensure that MWHttpRequest::method is always
 288                                 // uppercased. Bug 36137
 289                                 if ( $o == 'method' ) {
 290                                         $options[$o] = strtoupper( $options[$o] );
 291                                 }
 292                                 $this->$o = $options[$o];
 293                         }
 294                 }
 295
 296                 if ( $this->noProxy ) {
 297                         $this->proxy = ''; // noProxy takes precedence
 298                 }
 299
 300                 // Profile based on what's calling us
 301                 $this->profiler = $profiler;
 302                 $this->profileName = $caller;
 303         }
 304
 305         /**
 306          * Simple function to test if we can make any sort of requests at all, using
 307          * cURL or fopen()
 308          * @return bool
 309          */
 310         public static function canMakeRequests() {
 311                 return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
 312         }
 313
 314         /**
 315          * Generate a new request object
 316          * @param string $url Url to use
 317          * @param array $options (optional) extra params to pass (see Http::request())
 318          * @param string $caller The method making this request, for profiling
 319          * @throws MWException
 320          * @return CurlHttpRequest|PhpHttpRequest
 321          * @see MWHttpRequest::__construct
 322          */
 323         public static function factory( $url, $options = null, $caller = __METHOD__ ) {
 324                 if ( !Http::$httpEngine ) {
 325                         Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
 326                 } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
 327                         throw new MWException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' .
 328                                 ' Http::$httpEngine is set to "curl"' );
 329                 }
 330
 331                 switch ( Http::$httpEngine ) {
 332                         case 'curl':
 333                                 return new CurlHttpRequest( $url, $options, $caller, Profiler::instance() );
 334                         case 'php':
 335                                 if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
 336                                         throw new MWException( __METHOD__ . ': allow_url_fopen ' .
 337                                                 'needs to be enabled for pure PHP http requests to ' .
 338                                                 'work. If possible, curl should be used instead. See ' .
 339                                                 'http://php.net/curl.'
 340                                         );
 341                                 }
 342                                 return new PhpHttpRequest( $url, $options, $caller, Profiler::instance() );
 343                         default:
 344                                 throw new MWException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' );
 345                 }
 346         }
 347
 348         /**
 349          * Get the body, or content, of the response to the request
 350          *
 351          * @return string
 352          */
 353         public function getContent() {
 354                 return $this->content;
 355         }
 356
 357         /**
 358          * Set the parameters of the request
 359          *
 360          * @param array $args
 361          * @todo overload the args param
 362          */
 363         public function setData( $args ) {
 364                 $this->postData = $args;
 365         }
 366
 367         /**
 368          * Take care of setting up the proxy (do nothing if "noProxy" is set)
 369          *
 370          * @return void
 371          */
 372         public function proxySetup() {
 373                 global $wgHTTPProxy;
 374
 375                 // If there is an explicit proxy set and proxies are not disabled, then use it
 376                 if ( $this->proxy && !$this->noProxy ) {
 377                         return;
 378                 }
 379
 380                 // Otherwise, fallback to $wgHTTPProxy/http_proxy (when set) if this is not a machine
 381                 // local URL and proxies are not disabled
 382                 if ( Http::isLocalURL( $this->url ) || $this->noProxy ) {
 383                         $this->proxy = '';
 384                 } elseif ( $wgHTTPProxy ) {
 385                         $this->proxy = $wgHTTPProxy;
 386                 } elseif ( getenv( "http_proxy" ) ) {
 387                         $this->proxy = getenv( "http_proxy" );
 388                 }
 389         }
 390
 391         /**
 392          * Set the user agent
 393          * @param string $UA
 394          */
 395         public function setUserAgent( $UA ) {
 396                 $this->setHeader( 'User-Agent', $UA );
 397         }
 398
 399         /**
 400          * Set an arbitrary header
 401          * @param string $name
 402          * @param string $value
 403          */
 404         public function setHeader( $name, $value ) {
 405                 // I feel like I should normalize the case here...
 406                 $this->reqHeaders[$name] = $value;
 407         }
 408
 409         /**
 410          * Get an array of the headers
 411          * @return array
 412          */
 413         public function getHeaderList() {
 414                 $list = array();
 415
 416                 if ( $this->cookieJar ) {
 417                         $this->reqHeaders['Cookie'] =
 418                                 $this->cookieJar->serializeToHttpRequest(
 419                                         $this->parsedUrl['path'],
 420                                         $this->parsedUrl['host']
 421                                 );
 422                 }
 423
 424                 foreach ( $this->reqHeaders as $name => $value ) {
 425                         $list[] = "$name: $value";
 426                 }
 427
 428                 return $list;
 429         }
 430
 431         /**
 432          * Set a read callback to accept data read from the HTTP request.
 433          * By default, data is appended to an internal buffer which can be
 434          * retrieved through $req->getContent().
 435          *
 436          * To handle data as it comes in -- especially for large files that
 437          * would not fit in memory -- you can instead set your own callback,
 438          * in the form function($resource, $buffer) where the first parameter
 439          * is the low-level resource being read (implementation specific),
 440          * and the second parameter is the data buffer.
 441          *
 442          * You MUST return the number of bytes handled in the buffer; if fewer
 443          * bytes are reported handled than were passed to you, the HTTP fetch
 444          * will be aborted.
 445          *
 446          * @param callable $callback
 447          * @throws MWException
 448          */
 449         public function setCallback( $callback ) {
 450                 if ( !is_callable( $callback ) ) {
 451                         throw new MWException( 'Invalid MwHttpRequest callback' );
 452                 }
 453                 $this->callback = $callback;
 454         }
 455
 456         /**
 457          * A generic callback to read the body of the response from a remote
 458          * server.
 459          *
 460          * @param resource $fh
 461          * @param string $content
 462          * @return int
 463          */
 464         public function read( $fh, $content ) {
 465                 $this->content .= $content;
 466                 return strlen( $content );
 467         }
 468
 469         /**
 470          * Take care of whatever is necessary to perform the URI request.
 471          *
 472          * @return Status
 473          */
 474         public function execute() {
 475
 476                 $this->content = "";
 477
 478                 if ( strtoupper( $this->method ) == "HEAD" ) {
 479                         $this->headersOnly = true;
 480                 }
 481
 482                 $this->proxySetup(); // set up any proxy as needed
 483
 484                 if ( !$this->callback ) {
 485                         $this->setCallback( array( $this, 'read' ) );
 486                 }
 487
 488                 if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
 489                         $this->setUserAgent( Http::userAgent() );
 490                 }
 491
 492         }
 493
 494         /**
 495          * Parses the headers, including the HTTP status code and any
 496          * Set-Cookie headers.  This function expects the headers to be
 497          * found in an array in the member variable headerList.
 498          */
 499         protected function parseHeader() {
 500
 501                 $lastname = "";
 502
 503                 foreach ( $this->headerList as $header ) {
 504                         if ( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
 505                                 $this->respVersion = $match[1];
 506                                 $this->respStatus = $match[2];
 507                         } elseif ( preg_match( "#^[ \t]#", $header ) ) {
 508                                 $last = count( $this->respHeaders[$lastname] ) - 1;
 509                                 $this->respHeaders[$lastname][$last] .= "\r\n$header";
 510                         } elseif ( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
 511                                 $this->respHeaders[strtolower( $match[1] )][] = $match[2];
 512                                 $lastname = strtolower( $match[1] );
 513                         }
 514                 }
 515
 516                 $this->parseCookies();
 517
 518         }
 519
 520         /**
 521          * Sets HTTPRequest status member to a fatal value with the error
 522          * message if the returned integer value of the status code was
 523          * not successful (< 300) or a redirect (>=300 and < 400).  (see
 524          * RFC2616, section 10,
 525          * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a
 526          * list of status codes.)
 527          */
 528         protected function setStatus() {
 529                 if ( !$this->respHeaders ) {
 530                         $this->parseHeader();
 531                 }
 532
 533                 if ( (int)$this->respStatus > 399 ) {
 534                         list( $code, $message ) = explode( " ", $this->respStatus, 2 );
 535                         $this->status->fatal( "http-bad-status", $code, $message );
 536                 }
 537         }
 538
 539         /**
 540          * Get the integer value of the HTTP status code (e.g. 200 for "200 Ok")
 541          * (see RFC2616, section 10, http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
 542          * for a list of status codes.)
 543          *
 544          * @return int
 545          */
 546         public function getStatus() {
 547                 if ( !$this->respHeaders ) {
 548                         $this->parseHeader();
 549                 }
 550
 551                 return (int)$this->respStatus;
 552         }
 553
 554         /**
 555          * Returns true if the last status code was a redirect.
 556          *
 557          * @return bool
 558          */
 559         public function isRedirect() {
 560                 if ( !$this->respHeaders ) {
 561                         $this->parseHeader();
 562                 }
 563
 564                 $status = (int)$this->respStatus;
 565
 566                 if ( $status >= 300 && $status <= 303 ) {
 567                         return true;
 568                 }
 569
 570                 return false;
 571         }
 572
 573         /**
 574          * Returns an associative array of response headers after the
 575          * request has been executed.  Because some headers
 576          * (e.g. Set-Cookie) can appear more than once the, each value of
 577          * the associative array is an array of the values given.
 578          *
 579          * @return array
 580          */
 581         public function getResponseHeaders() {
 582                 if ( !$this->respHeaders ) {
 583                         $this->parseHeader();
 584                 }
 585
 586                 return $this->respHeaders;
 587         }
 588
 589         /**
 590          * Returns the value of the given response header.
 591          *
 592          * @param string $header
 593          * @return string
 594          */
 595         public function getResponseHeader( $header ) {
 596                 if ( !$this->respHeaders ) {
 597                         $this->parseHeader();
 598                 }
 599
 600                 if ( isset( $this->respHeaders[strtolower( $header )] ) ) {
 601                         $v = $this->respHeaders[strtolower( $header )];
 602                         return $v[count( $v ) - 1];
 603                 }
 604
 605                 return null;
 606         }
 607
 608         /**
 609          * Tells the MWHttpRequest object to use this pre-loaded CookieJar.
 610          *
 611          * @param CookieJar $jar
 612          */
 613         public function setCookieJar( $jar ) {
 614                 $this->cookieJar = $jar;
 615         }
 616
 617         /**
 618          * Returns the cookie jar in use.
 619          *
 620          * @return CookieJar
 621          */
 622         public function getCookieJar() {
 623                 if ( !$this->respHeaders ) {
 624                         $this->parseHeader();
 625                 }
 626
 627                 return $this->cookieJar;
 628         }
 629
 630         /**
 631          * Sets a cookie. Used before a request to set up any individual
 632          * cookies. Used internally after a request to parse the
 633          * Set-Cookie headers.
 634          * @see Cookie::set
 635          * @param string $name
 636          * @param mixed $value
 637          * @param array $attr
 638          */
 639         public function setCookie( $name, $value = null, $attr = null ) {
 640                 if ( !$this->cookieJar ) {
 641                         $this->cookieJar = new CookieJar;
 642                 }
 643
 644                 $this->cookieJar->setCookie( $name, $value, $attr );
 645         }
 646
 647         /**
 648          * Parse the cookies in the response headers and store them in the cookie jar.
 649          */
 650         protected function parseCookies() {
 651
 652                 if ( !$this->cookieJar ) {
 653                         $this->cookieJar = new CookieJar;
 654                 }
 655
 656                 if ( isset( $this->respHeaders['set-cookie'] ) ) {
 657                         $url = parse_url( $this->getFinalUrl() );
 658                         foreach ( $this->respHeaders['set-cookie'] as $cookie ) {
 659                                 $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
 660                         }
 661                 }
 662
 663         }
 664
 665         /**
 666          * Returns the final URL after all redirections.
 667          *
 668          * Relative values of the "Location" header are incorrect as
 669          * stated in RFC, however they do happen and modern browsers
 670          * support them.  This function loops backwards through all
 671          * locations in order to build the proper absolute URI - Marooned
 672          * at wikia-inc.com
 673          *
 674          * Note that the multiple Location: headers are an artifact of
 675          * CURL -- they shouldn't actually get returned this way. Rewrite
 676          * this when bug 29232 is taken care of (high-level redirect
 677          * handling rewrite).
 678          *
 679          * @return string
 680          */
 681         public function getFinalUrl() {
 682                 $headers = $this->getResponseHeaders();
 683
 684                 // return full url (fix for incorrect but handled relative location)
 685                 if ( isset( $headers['location'] ) ) {
 686                         $locations = $headers['location'];
 687                         $domain = '';
 688                         $foundRelativeURI = false;
 689                         $countLocations = count( $locations );
 690
 691                         for ( $i = $countLocations - 1; $i >= 0; $i-- ) {
 692                                 $url = parse_url( $locations[$i] );
 693
 694                                 if ( isset( $url['host'] ) ) {
 695                                         $domain = $url['scheme'] . '://' . $url['host'];
 696                                         break; // found correct URI (with host)
 697                                 } else {
 698                                         $foundRelativeURI = true;
 699                                 }
 700                         }
 701
 702                         if ( $foundRelativeURI ) {
 703                                 if ( $domain ) {
 704                                         return $domain . $locations[$countLocations - 1];
 705                                 } else {
 706                                         $url = parse_url( $this->url );
 707                                         if ( isset( $url['host'] ) ) {
 708                                                 return $url['scheme'] . '://' . $url['host'] .
 709                                                         $locations[$countLocations - 1];
 710                                         }
 711                                 }
 712                         } else {
 713                                 return $locations[$countLocations - 1];
 714                         }
 715                 }
 716
 717                 return $this->url;
 718         }
 719
 720         /**
 721          * Returns true if the backend can follow redirects. Overridden by the
 722          * child classes.
 723          * @return bool
 724          */
 725         public function canFollowRedirects() {
 726                 return true;
 727         }
 728 }
 729
 730 /**
 731  * MWHttpRequest implemented using internal curl compiled into PHP
 732  */
 733 class CurlHttpRequest extends MWHttpRequest {
 734         const SUPPORTS_FILE_POSTS = true;
 735
 736         protected $curlOptions = array();
 737         protected $headerText = "";
 738
 739         /**
 740          * @param resource $fh
 741          * @param string $content
 742          * @return int
 743          */
 744         protected function readHeader( $fh, $content ) {
 745                 $this->headerText .= $content;
 746                 return strlen( $content );
 747         }
 748
 749         public function execute() {
 750
 751                 parent::execute();
 752
 753                 if ( !$this->status->isOK() ) {
 754                         return $this->status;
 755                 }
 756
 757                 $this->curlOptions[CURLOPT_PROXY] = $this->proxy;
 758                 $this->curlOptions[CURLOPT_TIMEOUT] = $this->timeout;
 759
 760                 // Only supported in curl >= 7.16.2
 761                 if ( defined( 'CURLOPT_CONNECTTIMEOUT_MS' ) ) {
 762                         $this->curlOptions[CURLOPT_CONNECTTIMEOUT_MS] = $this->connectTimeout * 1000;
 763                 }
 764
 765                 $this->curlOptions[CURLOPT_HTTP_VERSION] = CURL_HTTP_VERSION_1_0;
 766                 $this->curlOptions[CURLOPT_WRITEFUNCTION] = $this->callback;
 767                 $this->curlOptions[CURLOPT_HEADERFUNCTION] = array( $this, "readHeader" );
 768                 $this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
 769                 $this->curlOptions[CURLOPT_ENCODING] = ""; # Enable compression
 770
 771                 $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
 772
 773                 $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost ? 2 : 0;
 774                 $this->curlOptions[CURLOPT_SSL_VERIFYPEER] = $this->sslVerifyCert;
 775
 776                 if ( $this->caInfo ) {
 777                         $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
 778                 }
 779
 780                 if ( $this->headersOnly ) {
 781                         $this->curlOptions[CURLOPT_NOBODY] = true;
 782                         $this->curlOptions[CURLOPT_HEADER] = true;
 783                 } elseif ( $this->method == 'POST' ) {
 784                         $this->curlOptions[CURLOPT_POST] = true;
 785                         $this->curlOptions[CURLOPT_POSTFIELDS] = $this->postData;
 786                         // Suppress 'Expect: 100-continue' header, as some servers
 787                         // will reject it with a 417 and Curl won't auto retry
 788                         // with HTTP 1.0 fallback
 789                         $this->reqHeaders['Expect'] = '';
 790                 } else {
 791                         $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
 792                 }
 793
 794                 $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
 795
 796                 $curlHandle = curl_init( $this->url );
 797
 798                 if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
 799                         throw new MWException( "Error setting curl options." );
 800                 }
 801
 802                 if ( $this->followRedirects && $this->canFollowRedirects() ) {
 803                         MediaWiki\suppressWarnings();
 804                         if ( !curl_setopt( $curlHandle, CURLOPT_FOLLOWLOCATION, true ) ) {
 805                                 wfDebug( __METHOD__ . ": Couldn't set CURLOPT_FOLLOWLOCATION. " .
 806                                         "Probably safe_mode or open_basedir is set.\n" );
 807                                 // Continue the processing. If it were in curl_setopt_array,
 808                                 // processing would have halted on its entry
 809                         }
 810                         MediaWiki\restoreWarnings();
 811                 }
 812
 813                 if ( $this->profiler ) {
 814                         $profileSection = $this->profiler->scopedProfileIn(
 815                                 __METHOD__ . '-' . $this->profileName
 816                         );
 817                 }
 818
 819                 $curlRes = curl_exec( $curlHandle );
 820                 if ( curl_errno( $curlHandle ) == CURLE_OPERATION_TIMEOUTED ) {
 821                         $this->status->fatal( 'http-timed-out', $this->url );
 822                 } elseif ( $curlRes === false ) {
 823                         $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
 824                 } else {
 825                         $this->headerList = explode( "\r\n", $this->headerText );
 826                 }
 827
 828                 curl_close( $curlHandle );
 829
 830                 if ( $this->profiler ) {
 831                         $this->profiler->scopedProfileOut( $profileSection );
 832                 }
 833
 834                 $this->parseHeader();
 835                 $this->setStatus();
 836
 837                 return $this->status;
 838         }
 839
 840         /**
 841          * @return bool
 842          */
 843         public function canFollowRedirects() {
 844                 $curlVersionInfo = curl_version();
 845                 if ( $curlVersionInfo['version_number'] < 0x071304 ) {
 846                         wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
 847                         return false;
 848                 }
 849
 850                 if ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 851                         if ( strval( ini_get( 'open_basedir' ) ) !== '' || wfIniGetBool( 'safe_mode' ) ) {
 852                                 wfDebug( "Cannot follow redirects in safe mode\n" );
 853                                 return false;
 854                         }
 855                 }
 856
 857                 return true;
 858         }
 859 }
 860
 861 class PhpHttpRequest extends MWHttpRequest {
 862
 863         private $fopenErrors = array();
 864
 865         /**
 866          * @param string $url
 867          * @return string
 868          */
 869         protected function urlToTcp( $url ) {
 870                 $parsedUrl = parse_url( $url );
 871
 872                 return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
 873         }
 874
 875         /**
 876          * Returns an array with a 'capath' or 'cafile' key
 877          * that is suitable to be merged into the 'ssl' sub-array of
 878          * a stream context options array.
 879          * Uses the 'caInfo' option of the class if it is provided, otherwise uses the system
 880          * default CA bundle if PHP supports that, or searches a few standard locations.
 881          * @return array
 882          * @throws DomainException
 883          */
 884         protected function getCertOptions() {
 885                 $certOptions = array();
 886                 $certLocations = array();
 887                 if ( $this->caInfo ) {
 888                         $certLocations = array( 'manual' => $this->caInfo );
 889                 } elseif ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 890                         // @codingStandardsIgnoreStart Generic.Files.LineLength
 891                         // Default locations, based on
 892                         // https://www.happyassassin.net/2015/01/12/a-note-about-ssltls-trusted-certificate-stores-and-platforms/
 893                         // PHP 5.5 and older doesn't have any defaults, so we try to guess ourselves.
 894                         // PHP 5.6+ gets the CA location from OpenSSL as long as it is not set manually,
 895                         // so we should leave capath/cafile empty there.
 896                         // @codingStandardsIgnoreEnd
 897                         $certLocations = array_filter( array(
 898                                 getenv( 'SSL_CERT_DIR' ),
 899                                 getenv( 'SSL_CERT_PATH' ),
 900                                 '/etc/pki/tls/certs/ca-bundle.crt', # Fedora et al
 901                                 '/etc/ssl/certs',  # Debian et al
 902                                 '/etc/pki/tls/certs/ca-bundle.trust.crt',
 903                                 '/etc/pki/ca-trust/extracted/pem/tls-ca-bundle.pem',
 904                                 '/System/Library/OpenSSL', # OSX
 905                         ) );
 906                 }
 907
 908                 foreach ( $certLocations as $key => $cert ) {
 909                         if ( is_dir( $cert ) ) {
 910                                 $certOptions['capath'] = $cert;
 911                                 break;
 912                         } elseif ( is_file( $cert ) ) {
 913                                 $certOptions['cafile'] = $cert;
 914                                 break;
 915                         } elseif ( $key === 'manual' ) {
 916                                 // fail more loudly if a cert path was manually configured and it is not valid
 917                                 throw new DomainException( "Invalid CA info passed: $cert" );
 918                         }
 919                 }
 920
 921                 return $certOptions;
 922         }
 923
 924         /**
 925          * Custom error handler for dealing with fopen() errors.
 926          * fopen() tends to fire multiple errors in succession, and the last one
 927          * is completely useless (something like "fopen: failed to open stream")
 928          * so normal methods of handling errors programmatically
 929          * like get_last_error() don't work.
 930          */
 931         public function errorHandler( $errno, $errstr ) {
 932                 $n = count( $this->fopenErrors ) + 1;
 933                 $this->fopenErrors += array( "errno$n" => $errno, "errstr$n" => $errstr );
 934         }
 935
 936         public function execute() {
 937
 938                 parent::execute();
 939
 940                 if ( is_array( $this->postData ) ) {
 941                         $this->postData = wfArrayToCgi( $this->postData );
 942                 }
 943
 944                 if ( $this->parsedUrl['scheme'] != 'http'
 945                         && $this->parsedUrl['scheme'] != 'https' ) {
 946                         $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
 947                 }
 948
 949                 $this->reqHeaders['Accept'] = "*/*";
 950                 $this->reqHeaders['Connection'] = 'Close';
 951                 if ( $this->method == 'POST' ) {
 952                         // Required for HTTP 1.0 POSTs
 953                         $this->reqHeaders['Content-Length'] = strlen( $this->postData );
 954                         if ( !isset( $this->reqHeaders['Content-Type'] ) ) {
 955                                 $this->reqHeaders['Content-Type'] = "application/x-www-form-urlencoded";
 956                         }
 957                 }
 958
 959                 // Set up PHP stream context
 960                 $options = array(
 961                         'http' => array(
 962                                 'method' => $this->method,
 963                                 'header' => implode( "\r\n", $this->getHeaderList() ),
 964                                 'protocol_version' => '1.1',
 965                                 'max_redirects' => $this->followRedirects ? $this->maxRedirects : 0,
 966                                 'ignore_errors' => true,
 967                                 'timeout' => $this->timeout,
 968                                 // Curl options in case curlwrappers are installed
 969                                 'curl_verify_ssl_host' => $this->sslVerifyHost ? 2 : 0,
 970                                 'curl_verify_ssl_peer' => $this->sslVerifyCert,
 971                         ),
 972                         'ssl' => array(
 973                                 'verify_peer' => $this->sslVerifyCert,
 974                                 'SNI_enabled' => true,
 975                                 'ciphers' => 'HIGH:!SSLv2:!SSLv3:-ADH:-kDH:-kECDH:-DSS',
 976                                 'disable_compression' => true,
 977                         ),
 978                 );
 979
 980                 if ( $this->proxy ) {
 981                         $options['http']['proxy'] = $this->urlToTCP( $this->proxy );
 982                         $options['http']['request_fulluri'] = true;
 983                 }
 984
 985                 if ( $this->postData ) {
 986                         $options['http']['content'] = $this->postData;
 987                 }
 988
 989                 if ( $this->sslVerifyHost ) {
 990                         // PHP 5.6.0 deprecates CN_match, in favour of peer_name which
 991                         // actually checks SubjectAltName properly.
 992                         if ( version_compare( PHP_VERSION, '5.6.0', '>=' ) ) {
 993                                 $options['ssl']['peer_name'] = $this->parsedUrl['host'];
 994                         } else {
 995                                 $options['ssl']['CN_match'] = $this->parsedUrl['host'];
 996                         }
 997                 }
 998
 999                 $options['ssl'] += $this->getCertOptions();
1000
1001                 $context = stream_context_create( $options );
1002
1003                 $this->headerList = array();
1004                 $reqCount = 0;
1005                 $url = $this->url;
1006
1007                 $result = array();
1008
1009                 if ( $this->profiler ) {
1010                         $profileSection = $this->profiler->scopedProfileIn(
1011                                 __METHOD__ . '-' . $this->profileName
1012                         );
1013                 }
1014                 do {
1015                         $reqCount++;
1016                         $this->fopenErrors = array();
1017                         set_error_handler( array( $this, 'errorHandler' ) );
1018                         $fh = fopen( $url, "r", false, $context );
1019                         restore_error_handler();
1020
1021                         if ( !$fh ) {
1022                                 // HACK for instant commons.
1023                                 // If we are contacting (commons|upload).wikimedia.org
1024                                 // try again with CN_match for en.wikipedia.org
1025                                 // as php does not handle SubjectAltName properly
1026                                 // prior to "peer_name" option in php 5.6
1027                                 if ( isset( $options['ssl']['CN_match'] )
1028                                         && ( $options['ssl']['CN_match'] === 'commons.wikimedia.org'
1029                                                 || $options['ssl']['CN_match'] === 'upload.wikimedia.org' )
1030                                 ) {
1031                                         $options['ssl']['CN_match'] = 'en.wikipedia.org';
1032                                         $context = stream_context_create( $options );
1033                                         continue;
1034                                 }
1035                                 break;
1036                         }
1037
1038                         $result = stream_get_meta_data( $fh );
1039                         $this->headerList = $result['wrapper_data'];
1040                         $this->parseHeader();
1041
1042                         if ( !$this->followRedirects ) {
1043                                 break;
1044                         }
1045
1046                         # Handle manual redirection
1047                         if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
1048                                 break;
1049                         }
1050                         # Check security of URL
1051                         $url = $this->getResponseHeader( "Location" );
1052
1053                         if ( !Http::isValidURI( $url ) ) {
1054                                 wfDebug( __METHOD__ . ": insecure redirection\n" );
1055                                 break;
1056                         }
1057                 } while ( true );
1058                 if ( $this->profiler ) {
1059                         $this->profiler->scopedProfileOut( $profileSection );
1060                 }
1061
1062                 $this->setStatus();
1063
1064                 if ( $fh === false ) {
1065                         if ( $this->fopenErrors ) {
1066                                 LoggerFactory::getInstance( 'http' )->warning( __CLASS__
1067                                         . ': error opening connection: {errstr1}', $this->fopenErrors );
1068                         }
1069                         $this->status->fatal( 'http-request-error' );
1070                         return $this->status;
1071                 }
1072
1073                 if ( $result['timed_out'] ) {
1074                         $this->status->fatal( 'http-timed-out', $this->url );
1075                         return $this->status;
1076                 }
1077
1078                 // If everything went OK, or we received some error code
1079                 // get the response body content.
1080                 if ( $this->status->isOK() || (int)$this->respStatus >= 300 ) {
1081                         while ( !feof( $fh ) ) {
1082                                 $buf = fread( $fh, 8192 );
1083
1084                                 if ( $buf === false ) {
1085                                         $this->status->fatal( 'http-read-error' );
1086                                         break;
1087                                 }
1088
1089                                 if ( strlen( $buf ) ) {
1090                                         call_user_func( $this->callback, $fh, $buf );
1091                                 }
1092                         }
1093                 }
1094                 fclose( $fh );
1095
1096                 return $this->status;
1097         }
1098 }