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 = [], $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( false, false, 'en' ),
  84                                 [ '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|bool false on error
  99          */
 100         public static function get( $url, $options = [], $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] : [];
 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|bool false on error
 122          */
 123         public static function post( $url, $options = [], $caller = __METHOD__ ) {
 124                 return Http::request( 'POST', $url, $options, $caller );
 125         }
 126
 127         /**
 128          * A standard user-agent we can use for external requests.
 129          * @return string
 130          */
 131         public static function userAgent() {
 132                 global $wgVersion;
 133                 return "MediaWiki/$wgVersion";
 134         }
 135
 136         /**
 137          * Checks that the given URI is a valid one. Hardcoding the
 138          * protocols, because we only want protocols that both cURL
 139          * and php support.
 140          *
 141          * file:// should not be allowed here for security purpose (r67684)
 142          *
 143          * @todo FIXME this is wildly inaccurate and fails to actually check most stuff
 144          *
 145          * @param string $uri URI to check for validity
 146          * @return bool
 147          */
 148         public static function isValidURI( $uri ) {
 149                 return preg_match(
 150                         '/^https?:\/\/[^\/\s]\S*$/D',
 151                         $uri
 152                 );
 153         }
 154
 155         /**
 156          * Gets the relevant proxy from $wgHTTPProxy
 157          *
 158          * @return mixed The proxy address or an empty string if not set.
 159          */
 160         public static function getProxy() {
 161                 global $wgHTTPProxy;
 162
 163                 if ( $wgHTTPProxy ) {
 164                         return $wgHTTPProxy;
 165                 }
 166
 167                 return "";
 168         }
 169 }
 170
 171 /**
 172  * This wrapper class will call out to curl (if available) or fallback
 173  * to regular PHP if necessary for handling internal HTTP requests.
 174  *
 175  * Renamed from HttpRequest to MWHttpRequest to avoid conflict with
 176  * PHP's HTTP extension.
 177  */
 178 class MWHttpRequest {
 179         const SUPPORTS_FILE_POSTS = false;
 180
 181         protected $content;
 182         protected $timeout = 'default';
 183         protected $headersOnly = null;
 184         protected $postData = null;
 185         protected $proxy = null;
 186         protected $noProxy = false;
 187         protected $sslVerifyHost = true;
 188         protected $sslVerifyCert = true;
 189         protected $caInfo = null;
 190         protected $method = "GET";
 191         protected $reqHeaders = [];
 192         protected $url;
 193         protected $parsedUrl;
 194         protected $callback;
 195         protected $maxRedirects = 5;
 196         protected $followRedirects = false;
 197
 198         /**
 199          * @var CookieJar
 200          */
 201         protected $cookieJar;
 202
 203         protected $headerList = [];
 204         protected $respVersion = "0.9";
 205         protected $respStatus = "200 Ok";
 206         protected $respHeaders = [];
 207
 208         public $status;
 209
 210         /**
 211          * @var Profiler
 212          */
 213         protected $profiler;
 214
 215         /**
 216          * @var string
 217          */
 218         protected $profileName;
 219
 220         /**
 221          * @param string $url Url to use. If protocol-relative, will be expanded to an http:// URL
 222          * @param array $options (optional) extra params to pass (see Http::request())
 223          * @param string $caller The method making this request, for profiling
 224          * @param Profiler $profiler An instance of the profiler for profiling, or null
 225          */
 226         protected function __construct(
 227                 $url, $options = [], $caller = __METHOD__, $profiler = null
 228         ) {
 229                 global $wgHTTPTimeout, $wgHTTPConnectTimeout;
 230
 231                 $this->url = wfExpandUrl( $url, PROTO_HTTP );
 232                 $this->parsedUrl = wfParseUrl( $this->url );
 233
 234                 if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) {
 235                         $this->status = Status::newFatal( 'http-invalid-url', $url );
 236                 } else {
 237                         $this->status = Status::newGood( 100 ); // continue
 238                 }
 239
 240                 if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) {
 241                         $this->timeout = $options['timeout'];
 242                 } else {
 243                         $this->timeout = $wgHTTPTimeout;
 244                 }
 245                 if ( isset( $options['connectTimeout'] ) && $options['connectTimeout'] != 'default' ) {
 246                         $this->connectTimeout = $options['connectTimeout'];
 247                 } else {
 248                         $this->connectTimeout = $wgHTTPConnectTimeout;
 249                 }
 250                 if ( isset( $options['userAgent'] ) ) {
 251                         $this->setUserAgent( $options['userAgent'] );
 252                 }
 253
 254                 $members = [ "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
 255                                 "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" ];
 256
 257                 foreach ( $members as $o ) {
 258                         if ( isset( $options[$o] ) ) {
 259                                 // ensure that MWHttpRequest::method is always
 260                                 // uppercased. Bug 36137
 261                                 if ( $o == 'method' ) {
 262                                         $options[$o] = strtoupper( $options[$o] );
 263                                 }
 264                                 $this->$o = $options[$o];
 265                         }
 266                 }
 267
 268                 if ( $this->noProxy ) {
 269                         $this->proxy = ''; // noProxy takes precedence
 270                 }
 271
 272                 // Profile based on what's calling us
 273                 $this->profiler = $profiler;
 274                 $this->profileName = $caller;
 275         }
 276
 277         /**
 278          * Simple function to test if we can make any sort of requests at all, using
 279          * cURL or fopen()
 280          * @return bool
 281          */
 282         public static function canMakeRequests() {
 283                 return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
 284         }
 285
 286         /**
 287          * Generate a new request object
 288          * @param string $url Url to use
 289          * @param array $options (optional) extra params to pass (see Http::request())
 290          * @param string $caller The method making this request, for profiling
 291          * @throws MWException
 292          * @return CurlHttpRequest|PhpHttpRequest
 293          * @see MWHttpRequest::__construct
 294          */
 295         public static function factory( $url, $options = null, $caller = __METHOD__ ) {
 296                 if ( !Http::$httpEngine ) {
 297                         Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
 298                 } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
 299                         throw new MWException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' .
 300                                 ' Http::$httpEngine is set to "curl"' );
 301                 }
 302
 303                 switch ( Http::$httpEngine ) {
 304                         case 'curl':
 305                                 return new CurlHttpRequest( $url, $options, $caller, Profiler::instance() );
 306                         case 'php':
 307                                 if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
 308                                         throw new MWException( __METHOD__ . ': allow_url_fopen ' .
 309                                                 'needs to be enabled for pure PHP http requests to ' .
 310                                                 'work. If possible, curl should be used instead. See ' .
 311                                                 'http://php.net/curl.'
 312                                         );
 313                                 }
 314                                 return new PhpHttpRequest( $url, $options, $caller, Profiler::instance() );
 315                         default:
 316                                 throw new MWException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' );
 317                 }
 318         }
 319
 320         /**
 321          * Get the body, or content, of the response to the request
 322          *
 323          * @return string
 324          */
 325         public function getContent() {
 326                 return $this->content;
 327         }
 328
 329         /**
 330          * Set the parameters of the request
 331          *
 332          * @param array $args
 333          * @todo overload the args param
 334          */
 335         public function setData( $args ) {
 336                 $this->postData = $args;
 337         }
 338
 339         /**
 340          * Take care of setting up the proxy (do nothing if "noProxy" is set)
 341          *
 342          * @return void
 343          */
 344         public function proxySetup() {
 345                 // If there is an explicit proxy set and proxies are not disabled, then use it
 346                 if ( $this->proxy && !$this->noProxy ) {
 347                         return;
 348                 }
 349
 350                 // Otherwise, fallback to $wgHTTPProxy if this is not a machine
 351                 // local URL and proxies are not disabled
 352                 if ( self::isLocalURL( $this->url ) || $this->noProxy ) {
 353                         $this->proxy = '';
 354                 } else {
 355                         $this->proxy = Http::getProxy();
 356                 }
 357         }
 358
 359         /**
 360          * Check if the URL can be served by localhost
 361          *
 362          * @param string $url Full url to check
 363          * @return bool
 364          */
 365         private static function isLocalURL( $url ) {
 366                 global $wgCommandLineMode, $wgLocalVirtualHosts;
 367
 368                 if ( $wgCommandLineMode ) {
 369                         return false;
 370                 }
 371
 372                 // Extract host part
 373                 $matches = [];
 374                 if ( preg_match( '!^https?://([\w.-]+)[/:].*$!', $url, $matches ) ) {
 375                         $host = $matches[1];
 376                         // Split up dotwise
 377                         $domainParts = explode( '.', $host );
 378                         // Check if this domain or any superdomain is listed as a local virtual host
 379                         $domainParts = array_reverse( $domainParts );
 380
 381                         $domain = '';
 382                         $countParts = count( $domainParts );
 383                         for ( $i = 0; $i < $countParts; $i++ ) {
 384                                 $domainPart = $domainParts[$i];
 385                                 if ( $i == 0 ) {
 386                                         $domain = $domainPart;
 387                                 } else {
 388                                         $domain = $domainPart . '.' . $domain;
 389                                 }
 390
 391                                 if ( in_array( $domain, $wgLocalVirtualHosts ) ) {
 392                                         return true;
 393                                 }
 394                         }
 395                 }
 396
 397                 return false;
 398         }
 399
 400         /**
 401          * Set the user agent
 402          * @param string $UA
 403          */
 404         public function setUserAgent( $UA ) {
 405                 $this->setHeader( 'User-Agent', $UA );
 406         }
 407
 408         /**
 409          * Set an arbitrary header
 410          * @param string $name
 411          * @param string $value
 412          */
 413         public function setHeader( $name, $value ) {
 414                 // I feel like I should normalize the case here...
 415                 $this->reqHeaders[$name] = $value;
 416         }
 417
 418         /**
 419          * Get an array of the headers
 420          * @return array
 421          */
 422         public function getHeaderList() {
 423                 $list = [];
 424
 425                 if ( $this->cookieJar ) {
 426                         $this->reqHeaders['Cookie'] =
 427                                 $this->cookieJar->serializeToHttpRequest(
 428                                         $this->parsedUrl['path'],
 429                                         $this->parsedUrl['host']
 430                                 );
 431                 }
 432
 433                 foreach ( $this->reqHeaders as $name => $value ) {
 434                         $list[] = "$name: $value";
 435                 }
 436
 437                 return $list;
 438         }
 439
 440         /**
 441          * Set a read callback to accept data read from the HTTP request.
 442          * By default, data is appended to an internal buffer which can be
 443          * retrieved through $req->getContent().
 444          *
 445          * To handle data as it comes in -- especially for large files that
 446          * would not fit in memory -- you can instead set your own callback,
 447          * in the form function($resource, $buffer) where the first parameter
 448          * is the low-level resource being read (implementation specific),
 449          * and the second parameter is the data buffer.
 450          *
 451          * You MUST return the number of bytes handled in the buffer; if fewer
 452          * bytes are reported handled than were passed to you, the HTTP fetch
 453          * will be aborted.
 454          *
 455          * @param callable $callback
 456          * @throws MWException
 457          */
 458         public function setCallback( $callback ) {
 459                 if ( !is_callable( $callback ) ) {
 460                         throw new MWException( 'Invalid MwHttpRequest callback' );
 461                 }
 462                 $this->callback = $callback;
 463         }
 464
 465         /**
 466          * A generic callback to read the body of the response from a remote
 467          * server.
 468          *
 469          * @param resource $fh
 470          * @param string $content
 471          * @return int
 472          */
 473         public function read( $fh, $content ) {
 474                 $this->content .= $content;
 475                 return strlen( $content );
 476         }
 477
 478         /**
 479          * Take care of whatever is necessary to perform the URI request.
 480          *
 481          * @return Status
 482          */
 483         public function execute() {
 484
 485                 $this->content = "";
 486
 487                 if ( strtoupper( $this->method ) == "HEAD" ) {
 488                         $this->headersOnly = true;
 489                 }
 490
 491                 $this->proxySetup(); // set up any proxy as needed
 492
 493                 if ( !$this->callback ) {
 494                         $this->setCallback( [ $this, 'read' ] );
 495                 }
 496
 497                 if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
 498                         $this->setUserAgent( Http::userAgent() );
 499                 }
 500
 501         }
 502
 503         /**
 504          * Parses the headers, including the HTTP status code and any
 505          * Set-Cookie headers.  This function expects the headers to be
 506          * found in an array in the member variable headerList.
 507          */
 508         protected function parseHeader() {
 509
 510                 $lastname = "";
 511
 512                 foreach ( $this->headerList as $header ) {
 513                         if ( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
 514                                 $this->respVersion = $match[1];
 515                                 $this->respStatus = $match[2];
 516                         } elseif ( preg_match( "#^[ \t]#", $header ) ) {
 517                                 $last = count( $this->respHeaders[$lastname] ) - 1;
 518                                 $this->respHeaders[$lastname][$last] .= "\r\n$header";
 519                         } elseif ( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
 520                                 $this->respHeaders[strtolower( $match[1] )][] = $match[2];
 521                                 $lastname = strtolower( $match[1] );
 522                         }
 523                 }
 524
 525                 $this->parseCookies();
 526
 527         }
 528
 529         /**
 530          * Sets HTTPRequest status member to a fatal value with the error
 531          * message if the returned integer value of the status code was
 532          * not successful (< 300) or a redirect (>=300 and < 400).  (see
 533          * RFC2616, section 10,
 534          * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a
 535          * list of status codes.)
 536          */
 537         protected function setStatus() {
 538                 if ( !$this->respHeaders ) {
 539                         $this->parseHeader();
 540                 }
 541
 542                 if ( (int)$this->respStatus > 399 ) {
 543                         list( $code, $message ) = explode( " ", $this->respStatus, 2 );
 544                         $this->status->fatal( "http-bad-status", $code, $message );
 545                 }
 546         }
 547
 548         /**
 549          * Get the integer value of the HTTP status code (e.g. 200 for "200 Ok")
 550          * (see RFC2616, section 10, http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
 551          * for a list of status codes.)
 552          *
 553          * @return int
 554          */
 555         public function getStatus() {
 556                 if ( !$this->respHeaders ) {
 557                         $this->parseHeader();
 558                 }
 559
 560                 return (int)$this->respStatus;
 561         }
 562
 563         /**
 564          * Returns true if the last status code was a redirect.
 565          *
 566          * @return bool
 567          */
 568         public function isRedirect() {
 569                 if ( !$this->respHeaders ) {
 570                         $this->parseHeader();
 571                 }
 572
 573                 $status = (int)$this->respStatus;
 574
 575                 if ( $status >= 300 && $status <= 303 ) {
 576                         return true;
 577                 }
 578
 579                 return false;
 580         }
 581
 582         /**
 583          * Returns an associative array of response headers after the
 584          * request has been executed.  Because some headers
 585          * (e.g. Set-Cookie) can appear more than once the, each value of
 586          * the associative array is an array of the values given.
 587          *
 588          * @return array
 589          */
 590         public function getResponseHeaders() {
 591                 if ( !$this->respHeaders ) {
 592                         $this->parseHeader();
 593                 }
 594
 595                 return $this->respHeaders;
 596         }
 597
 598         /**
 599          * Returns the value of the given response header.
 600          *
 601          * @param string $header
 602          * @return string
 603          */
 604         public function getResponseHeader( $header ) {
 605                 if ( !$this->respHeaders ) {
 606                         $this->parseHeader();
 607                 }
 608
 609                 if ( isset( $this->respHeaders[strtolower( $header )] ) ) {
 610                         $v = $this->respHeaders[strtolower( $header )];
 611                         return $v[count( $v ) - 1];
 612                 }
 613
 614                 return null;
 615         }
 616
 617         /**
 618          * Tells the MWHttpRequest object to use this pre-loaded CookieJar.
 619          *
 620          * @param CookieJar $jar
 621          */
 622         public function setCookieJar( $jar ) {
 623                 $this->cookieJar = $jar;
 624         }
 625
 626         /**
 627          * Returns the cookie jar in use.
 628          *
 629          * @return CookieJar
 630          */
 631         public function getCookieJar() {
 632                 if ( !$this->respHeaders ) {
 633                         $this->parseHeader();
 634                 }
 635
 636                 return $this->cookieJar;
 637         }
 638
 639         /**
 640          * Sets a cookie. Used before a request to set up any individual
 641          * cookies. Used internally after a request to parse the
 642          * Set-Cookie headers.
 643          * @see Cookie::set
 644          * @param string $name
 645          * @param mixed $value
 646          * @param array $attr
 647          */
 648         public function setCookie( $name, $value = null, $attr = null ) {
 649                 if ( !$this->cookieJar ) {
 650                         $this->cookieJar = new CookieJar;
 651                 }
 652
 653                 $this->cookieJar->setCookie( $name, $value, $attr );
 654         }
 655
 656         /**
 657          * Parse the cookies in the response headers and store them in the cookie jar.
 658          */
 659         protected function parseCookies() {
 660
 661                 if ( !$this->cookieJar ) {
 662                         $this->cookieJar = new CookieJar;
 663                 }
 664
 665                 if ( isset( $this->respHeaders['set-cookie'] ) ) {
 666                         $url = parse_url( $this->getFinalUrl() );
 667                         foreach ( $this->respHeaders['set-cookie'] as $cookie ) {
 668                                 $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
 669                         }
 670                 }
 671
 672         }
 673
 674         /**
 675          * Returns the final URL after all redirections.
 676          *
 677          * Relative values of the "Location" header are incorrect as
 678          * stated in RFC, however they do happen and modern browsers
 679          * support them.  This function loops backwards through all
 680          * locations in order to build the proper absolute URI - Marooned
 681          * at wikia-inc.com
 682          *
 683          * Note that the multiple Location: headers are an artifact of
 684          * CURL -- they shouldn't actually get returned this way. Rewrite
 685          * this when bug 29232 is taken care of (high-level redirect
 686          * handling rewrite).
 687          *
 688          * @return string
 689          */
 690         public function getFinalUrl() {
 691                 $headers = $this->getResponseHeaders();
 692
 693                 // return full url (fix for incorrect but handled relative location)
 694                 if ( isset( $headers['location'] ) ) {
 695                         $locations = $headers['location'];
 696                         $domain = '';
 697                         $foundRelativeURI = false;
 698                         $countLocations = count( $locations );
 699
 700                         for ( $i = $countLocations - 1; $i >= 0; $i-- ) {
 701                                 $url = parse_url( $locations[$i] );
 702
 703                                 if ( isset( $url['host'] ) ) {
 704                                         $domain = $url['scheme'] . '://' . $url['host'];
 705                                         break; // found correct URI (with host)
 706                                 } else {
 707                                         $foundRelativeURI = true;
 708                                 }
 709                         }
 710
 711                         if ( $foundRelativeURI ) {
 712                                 if ( $domain ) {
 713                                         return $domain . $locations[$countLocations - 1];
 714                                 } else {
 715                                         $url = parse_url( $this->url );
 716                                         if ( isset( $url['host'] ) ) {
 717                                                 return $url['scheme'] . '://' . $url['host'] .
 718                                                         $locations[$countLocations - 1];
 719                                         }
 720                                 }
 721                         } else {
 722                                 return $locations[$countLocations - 1];
 723                         }
 724                 }
 725
 726                 return $this->url;
 727         }
 728
 729         /**
 730          * Returns true if the backend can follow redirects. Overridden by the
 731          * child classes.
 732          * @return bool
 733          */
 734         public function canFollowRedirects() {
 735                 return true;
 736         }
 737 }
 738
 739 /**
 740  * MWHttpRequest implemented using internal curl compiled into PHP
 741  */
 742 class CurlHttpRequest extends MWHttpRequest {
 743         const SUPPORTS_FILE_POSTS = true;
 744
 745         protected $curlOptions = [];
 746         protected $headerText = "";
 747
 748         /**
 749          * @param resource $fh
 750          * @param string $content
 751          * @return int
 752          */
 753         protected function readHeader( $fh, $content ) {
 754                 $this->headerText .= $content;
 755                 return strlen( $content );
 756         }
 757
 758         public function execute() {
 759
 760                 parent::execute();
 761
 762                 if ( !$this->status->isOK() ) {
 763                         return $this->status;
 764                 }
 765
 766                 $this->curlOptions[CURLOPT_PROXY] = $this->proxy;
 767                 $this->curlOptions[CURLOPT_TIMEOUT] = $this->timeout;
 768
 769                 // Only supported in curl >= 7.16.2
 770                 if ( defined( 'CURLOPT_CONNECTTIMEOUT_MS' ) ) {
 771                         $this->curlOptions[CURLOPT_CONNECTTIMEOUT_MS] = $this->connectTimeout * 1000;
 772                 }
 773
 774                 $this->curlOptions[CURLOPT_HTTP_VERSION] = CURL_HTTP_VERSION_1_0;
 775                 $this->curlOptions[CURLOPT_WRITEFUNCTION] = $this->callback;
 776                 $this->curlOptions[CURLOPT_HEADERFUNCTION] = [ $this, "readHeader" ];
 777                 $this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
 778                 $this->curlOptions[CURLOPT_ENCODING] = ""; # Enable compression
 779
 780                 $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
 781
 782                 $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost ? 2 : 0;
 783                 $this->curlOptions[CURLOPT_SSL_VERIFYPEER] = $this->sslVerifyCert;
 784
 785                 if ( $this->caInfo ) {
 786                         $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
 787                 }
 788
 789                 if ( $this->headersOnly ) {
 790                         $this->curlOptions[CURLOPT_NOBODY] = true;
 791                         $this->curlOptions[CURLOPT_HEADER] = true;
 792                 } elseif ( $this->method == 'POST' ) {
 793                         $this->curlOptions[CURLOPT_POST] = true;
 794                         $postData = $this->postData;
 795                         // Don't interpret POST parameters starting with '@' as file uploads, because this
 796                         // makes it impossible to POST plain values starting with '@' (and causes security
 797                         // issues potentially exposing the contents of local files).
 798                         // The PHP manual says this option was introduced in PHP 5.5 defaults to true in PHP 5.6,
 799                         // but we support lower versions, and the option doesn't exist in HHVM 5.6.99.
 800                         if ( defined( 'CURLOPT_SAFE_UPLOAD' ) ) {
 801                                 $this->curlOptions[CURLOPT_SAFE_UPLOAD] = true;
 802                         } elseif ( is_array( $postData ) ) {
 803                                 // In PHP 5.2 and later, '@' is interpreted as a file upload if POSTFIELDS
 804                                 // is an array, but not if it's a string. So convert $req['body'] to a string
 805                                 // for safety.
 806                                 $postData = wfArrayToCgi( $postData );
 807                         }
 808                         $this->curlOptions[CURLOPT_POSTFIELDS] = $postData;
 809
 810                         // Suppress 'Expect: 100-continue' header, as some servers
 811                         // will reject it with a 417 and Curl won't auto retry
 812                         // with HTTP 1.0 fallback
 813                         $this->reqHeaders['Expect'] = '';
 814                 } else {
 815                         $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
 816                 }
 817
 818                 $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
 819
 820                 $curlHandle = curl_init( $this->url );
 821
 822                 if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
 823                         throw new MWException( "Error setting curl options." );
 824                 }
 825
 826                 if ( $this->followRedirects && $this->canFollowRedirects() ) {
 827                         MediaWiki\suppressWarnings();
 828                         if ( !curl_setopt( $curlHandle, CURLOPT_FOLLOWLOCATION, true ) ) {
 829                                 wfDebug( __METHOD__ . ": Couldn't set CURLOPT_FOLLOWLOCATION. " .
 830                                         "Probably open_basedir is set.\n" );
 831                                 // Continue the processing. If it were in curl_setopt_array,
 832                                 // processing would have halted on its entry
 833                         }
 834                         MediaWiki\restoreWarnings();
 835                 }
 836
 837                 if ( $this->profiler ) {
 838                         $profileSection = $this->profiler->scopedProfileIn(
 839                                 __METHOD__ . '-' . $this->profileName
 840                         );
 841                 }
 842
 843                 $curlRes = curl_exec( $curlHandle );
 844                 if ( curl_errno( $curlHandle ) == CURLE_OPERATION_TIMEOUTED ) {
 845                         $this->status->fatal( 'http-timed-out', $this->url );
 846                 } elseif ( $curlRes === false ) {
 847                         $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
 848                 } else {
 849                         $this->headerList = explode( "\r\n", $this->headerText );
 850                 }
 851
 852                 curl_close( $curlHandle );
 853
 854                 if ( $this->profiler ) {
 855                         $this->profiler->scopedProfileOut( $profileSection );
 856                 }
 857
 858                 $this->parseHeader();
 859                 $this->setStatus();
 860
 861                 return $this->status;
 862         }
 863
 864         /**
 865          * @return bool
 866          */
 867         public function canFollowRedirects() {
 868                 $curlVersionInfo = curl_version();
 869                 if ( $curlVersionInfo['version_number'] < 0x071304 ) {
 870                         wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
 871                         return false;
 872                 }
 873
 874                 if ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 875                         if ( strval( ini_get( 'open_basedir' ) ) !== '' ) {
 876                                 wfDebug( "Cannot follow redirects when open_basedir is set\n" );
 877                                 return false;
 878                         }
 879                 }
 880
 881                 return true;
 882         }
 883 }
 884
 885 class PhpHttpRequest extends MWHttpRequest {
 886
 887         private $fopenErrors = [];
 888
 889         /**
 890          * @param string $url
 891          * @return string
 892          */
 893         protected function urlToTcp( $url ) {
 894                 $parsedUrl = parse_url( $url );
 895
 896                 return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
 897         }
 898
 899         /**
 900          * Returns an array with a 'capath' or 'cafile' key
 901          * that is suitable to be merged into the 'ssl' sub-array of
 902          * a stream context options array.
 903          * Uses the 'caInfo' option of the class if it is provided, otherwise uses the system
 904          * default CA bundle if PHP supports that, or searches a few standard locations.
 905          * @return array
 906          * @throws DomainException
 907          */
 908         protected function getCertOptions() {
 909                 $certOptions = [];
 910                 $certLocations = [];
 911                 if ( $this->caInfo ) {
 912                         $certLocations = [ 'manual' => $this->caInfo ];
 913                 } elseif ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 914                         // @codingStandardsIgnoreStart Generic.Files.LineLength
 915                         // Default locations, based on
 916                         // https://www.happyassassin.net/2015/01/12/a-note-about-ssltls-trusted-certificate-stores-and-platforms/
 917                         // PHP 5.5 and older doesn't have any defaults, so we try to guess ourselves.
 918                         // PHP 5.6+ gets the CA location from OpenSSL as long as it is not set manually,
 919                         // so we should leave capath/cafile empty there.
 920                         // @codingStandardsIgnoreEnd
 921                         $certLocations = array_filter( [
 922                                 getenv( 'SSL_CERT_DIR' ),
 923                                 getenv( 'SSL_CERT_PATH' ),
 924                                 '/etc/pki/tls/certs/ca-bundle.crt', # Fedora et al
 925                                 '/etc/ssl/certs',  # Debian et al
 926                                 '/etc/pki/tls/certs/ca-bundle.trust.crt',
 927                                 '/etc/pki/ca-trust/extracted/pem/tls-ca-bundle.pem',
 928                                 '/System/Library/OpenSSL', # OSX
 929                         ] );
 930                 }
 931
 932                 foreach ( $certLocations as $key => $cert ) {
 933                         if ( is_dir( $cert ) ) {
 934                                 $certOptions['capath'] = $cert;
 935                                 break;
 936                         } elseif ( is_file( $cert ) ) {
 937                                 $certOptions['cafile'] = $cert;
 938                                 break;
 939                         } elseif ( $key === 'manual' ) {
 940                                 // fail more loudly if a cert path was manually configured and it is not valid
 941                                 throw new DomainException( "Invalid CA info passed: $cert" );
 942                         }
 943                 }
 944
 945                 return $certOptions;
 946         }
 947
 948         /**
 949          * Custom error handler for dealing with fopen() errors.
 950          * fopen() tends to fire multiple errors in succession, and the last one
 951          * is completely useless (something like "fopen: failed to open stream")
 952          * so normal methods of handling errors programmatically
 953          * like get_last_error() don't work.
 954          */
 955         public function errorHandler( $errno, $errstr ) {
 956                 $n = count( $this->fopenErrors ) + 1;
 957                 $this->fopenErrors += [ "errno$n" => $errno, "errstr$n" => $errstr ];
 958         }
 959
 960         public function execute() {
 961
 962                 parent::execute();
 963
 964                 if ( is_array( $this->postData ) ) {
 965                         $this->postData = wfArrayToCgi( $this->postData );
 966                 }
 967
 968                 if ( $this->parsedUrl['scheme'] != 'http'
 969                         && $this->parsedUrl['scheme'] != 'https' ) {
 970                         $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
 971                 }
 972
 973                 $this->reqHeaders['Accept'] = "*/*";
 974                 $this->reqHeaders['Connection'] = 'Close';
 975                 if ( $this->method == 'POST' ) {
 976                         // Required for HTTP 1.0 POSTs
 977                         $this->reqHeaders['Content-Length'] = strlen( $this->postData );
 978                         if ( !isset( $this->reqHeaders['Content-Type'] ) ) {
 979                                 $this->reqHeaders['Content-Type'] = "application/x-www-form-urlencoded";
 980                         }
 981                 }
 982
 983                 // Set up PHP stream context
 984                 $options = [
 985                         'http' => [
 986                                 'method' => $this->method,
 987                                 'header' => implode( "\r\n", $this->getHeaderList() ),
 988                                 'protocol_version' => '1.1',
 989                                 'max_redirects' => $this->followRedirects ? $this->maxRedirects : 0,
 990                                 'ignore_errors' => true,
 991                                 'timeout' => $this->timeout,
 992                                 // Curl options in case curlwrappers are installed
 993                                 'curl_verify_ssl_host' => $this->sslVerifyHost ? 2 : 0,
 994                                 'curl_verify_ssl_peer' => $this->sslVerifyCert,
 995                         ],
 996                         'ssl' => [
 997                                 'verify_peer' => $this->sslVerifyCert,
 998                                 'SNI_enabled' => true,
 999                                 'ciphers' => 'HIGH:!SSLv2:!SSLv3:-ADH:-kDH:-kECDH:-DSS',
1000                                 'disable_compression' => true,
1001                         ],
1002                 ];
1003
1004                 if ( $this->proxy ) {
1005                         $options['http']['proxy'] = $this->urlToTcp( $this->proxy );
1006                         $options['http']['request_fulluri'] = true;
1007                 }
1008
1009                 if ( $this->postData ) {
1010                         $options['http']['content'] = $this->postData;
1011                 }
1012
1013                 if ( $this->sslVerifyHost ) {
1014                         // PHP 5.6.0 deprecates CN_match, in favour of peer_name which
1015                         // actually checks SubjectAltName properly.
1016                         if ( version_compare( PHP_VERSION, '5.6.0', '>=' ) ) {
1017                                 $options['ssl']['peer_name'] = $this->parsedUrl['host'];
1018                         } else {
1019                                 $options['ssl']['CN_match'] = $this->parsedUrl['host'];
1020                         }
1021                 }
1022
1023                 $options['ssl'] += $this->getCertOptions();
1024
1025                 $context = stream_context_create( $options );
1026
1027                 $this->headerList = [];
1028                 $reqCount = 0;
1029                 $url = $this->url;
1030
1031                 $result = [];
1032
1033                 if ( $this->profiler ) {
1034                         $profileSection = $this->profiler->scopedProfileIn(
1035                                 __METHOD__ . '-' . $this->profileName
1036                         );
1037                 }
1038                 do {
1039                         $reqCount++;
1040                         $this->fopenErrors = [];
1041                         set_error_handler( [ $this, 'errorHandler' ] );
1042                         $fh = fopen( $url, "r", false, $context );
1043                         restore_error_handler();
1044
1045                         if ( !$fh ) {
1046                                 // HACK for instant commons.
1047                                 // If we are contacting (commons|upload).wikimedia.org
1048                                 // try again with CN_match for en.wikipedia.org
1049                                 // as php does not handle SubjectAltName properly
1050                                 // prior to "peer_name" option in php 5.6
1051                                 if ( isset( $options['ssl']['CN_match'] )
1052                                         && ( $options['ssl']['CN_match'] === 'commons.wikimedia.org'
1053                                                 || $options['ssl']['CN_match'] === 'upload.wikimedia.org' )
1054                                 ) {
1055                                         $options['ssl']['CN_match'] = 'en.wikipedia.org';
1056                                         $context = stream_context_create( $options );
1057                                         continue;
1058                                 }
1059                                 break;
1060                         }
1061
1062                         $result = stream_get_meta_data( $fh );
1063                         $this->headerList = $result['wrapper_data'];
1064                         $this->parseHeader();
1065
1066                         if ( !$this->followRedirects ) {
1067                                 break;
1068                         }
1069
1070                         # Handle manual redirection
1071                         if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
1072                                 break;
1073                         }
1074                         # Check security of URL
1075                         $url = $this->getResponseHeader( "Location" );
1076
1077                         if ( !Http::isValidURI( $url ) ) {
1078                                 wfDebug( __METHOD__ . ": insecure redirection\n" );
1079                                 break;
1080                         }
1081                 } while ( true );
1082                 if ( $this->profiler ) {
1083                         $this->profiler->scopedProfileOut( $profileSection );
1084                 }
1085
1086                 $this->setStatus();
1087
1088                 if ( $fh === false ) {
1089                         if ( $this->fopenErrors ) {
1090                                 LoggerFactory::getInstance( 'http' )->warning( __CLASS__
1091                                         . ': error opening connection: {errstr1}', $this->fopenErrors );
1092                         }
1093                         $this->status->fatal( 'http-request-error' );
1094                         return $this->status;
1095                 }
1096
1097                 if ( $result['timed_out'] ) {
1098                         $this->status->fatal( 'http-timed-out', $this->url );
1099                         return $this->status;
1100                 }
1101
1102                 // If everything went OK, or we received some error code
1103                 // get the response body content.
1104                 if ( $this->status->isOK() || (int)$this->respStatus >= 300 ) {
1105                         while ( !feof( $fh ) ) {
1106                                 $buf = fread( $fh, 8192 );
1107
1108                                 if ( $buf === false ) {
1109                                         $this->status->fatal( 'http-read-error' );
1110                                         break;
1111                                 }
1112
1113                                 if ( strlen( $buf ) ) {
1114                                         call_user_func( $this->callback, $fh, $buf );
1115                                 }
1116                         }
1117                 }
1118                 fclose( $fh );
1119
1120                 return $this->status;
1121         }
1122 }