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(), array( 'caller' => $caller ) );
  84                         return false;
  85                 }
  86         }
  87
  88         /**
  89          * Simple wrapper for Http::request( 'GET' )
  90          * @see Http::request()
  91          * @since 1.25 Second parameter $timeout removed. Second parameter
  92          * is now $options which can be given a 'timeout'
  93          *
  94          * @param string $url
  95          * @param array $options
  96          * @param string $caller The method making this request, for profiling
  97          * @return string
  98          */
  99         public static function get( $url, $options = array(), $caller = __METHOD__ ) {
 100                 $args = func_get_args();
 101                 if ( isset( $args[1] ) && ( is_string( $args[1] ) || is_numeric( $args[1] ) ) ) {
 102                         // Second was used to be the timeout
 103                         // And third parameter used to be $options
 104                         wfWarn( "Second parameter should not be a timeout.", 2 );
 105                         $options = isset( $args[2] ) && is_array( $args[2] ) ?
 106                                 $args[2] : array();
 107                         $options['timeout'] = $args[1];
 108                         $caller = __METHOD__;
 109                 }
 110                 return Http::request( 'GET', $url, $options, $caller );
 111         }
 112
 113         /**
 114          * Simple wrapper for Http::request( 'POST' )
 115          * @see Http::request()
 116          *
 117          * @param string $url
 118          * @param array $options
 119          * @param string $caller The method making this request, for profiling
 120          * @return string
 121          */
 122         public static function post( $url, $options = array(), $caller = __METHOD__ ) {
 123                 return Http::request( 'POST', $url, $options, $caller );
 124         }
 125
 126         /**
 127          * Check if the URL can be served by localhost
 128          *
 129          * @param string $url Full url to check
 130          * @return bool
 131          */
 132         public static function isLocalURL( $url ) {
 133                 global $wgCommandLineMode, $wgLocalVirtualHosts, $wgConf;
 134
 135                 if ( $wgCommandLineMode ) {
 136                         return false;
 137                 }
 138
 139                 // Extract host part
 140                 $matches = array();
 141                 if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
 142                         $host = $matches[1];
 143                         // Split up dotwise
 144                         $domainParts = explode( '.', $host );
 145                         // Check if this domain or any superdomain is listed as a local virtual host
 146                         $domainParts = array_reverse( $domainParts );
 147
 148                         $domain = '';
 149                         $countParts = count( $domainParts );
 150                         for ( $i = 0; $i < $countParts; $i++ ) {
 151                                 $domainPart = $domainParts[$i];
 152                                 if ( $i == 0 ) {
 153                                         $domain = $domainPart;
 154                                 } else {
 155                                         $domain = $domainPart . '.' . $domain;
 156                                 }
 157
 158                                 if ( in_array( $domain, $wgLocalVirtualHosts )
 159                                         || $wgConf->isLocalVHost( $domain )
 160                                 ) {
 161                                         return true;
 162                                 }
 163                         }
 164                 }
 165
 166                 return false;
 167         }
 168
 169         /**
 170          * A standard user-agent we can use for external requests.
 171          * @return string
 172          */
 173         public static function userAgent() {
 174                 global $wgVersion;
 175                 return "MediaWiki/$wgVersion";
 176         }
 177
 178         /**
 179          * Checks that the given URI is a valid one. Hardcoding the
 180          * protocols, because we only want protocols that both cURL
 181          * and php support.
 182          *
 183          * file:// should not be allowed here for security purpose (r67684)
 184          *
 185          * @todo FIXME this is wildly inaccurate and fails to actually check most stuff
 186          *
 187          * @param string $uri URI to check for validity
 188          * @return bool
 189          */
 190         public static function isValidURI( $uri ) {
 191                 return preg_match(
 192                         '/^https?:\/\/[^\/\s]\S*$/D',
 193                         $uri
 194                 );
 195         }
 196
 197         /**
 198          * Gets the relevant proxy from $wgHTTPProxy/http_proxy (when set).
 199          *
 200          * @return mixed The proxy address or an empty string if not set.
 201          */
 202         public static function getProxy() {
 203                 global $wgHTTPProxy;
 204
 205                 if ( $wgHTTPProxy ) {
 206                         return $wgHTTPProxy;
 207                 }
 208
 209                 $envHttpProxy = getenv( "http_proxy" );
 210                 if ( $envHttpProxy ) {
 211                         return $envHttpProxy;
 212                 }
 213
 214                 return "";
 215         }
 216 }
 217
 218 /**
 219  * This wrapper class will call out to curl (if available) or fallback
 220  * to regular PHP if necessary for handling internal HTTP requests.
 221  *
 222  * Renamed from HttpRequest to MWHttpRequest to avoid conflict with
 223  * PHP's HTTP extension.
 224  */
 225 class MWHttpRequest {
 226         const SUPPORTS_FILE_POSTS = false;
 227
 228         protected $content;
 229         protected $timeout = 'default';
 230         protected $headersOnly = null;
 231         protected $postData = null;
 232         protected $proxy = null;
 233         protected $noProxy = false;
 234         protected $sslVerifyHost = true;
 235         protected $sslVerifyCert = true;
 236         protected $caInfo = null;
 237         protected $method = "GET";
 238         protected $reqHeaders = array();
 239         protected $url;
 240         protected $parsedUrl;
 241         protected $callback;
 242         protected $maxRedirects = 5;
 243         protected $followRedirects = false;
 244
 245         /**
 246          * @var CookieJar
 247          */
 248         protected $cookieJar;
 249
 250         protected $headerList = array();
 251         protected $respVersion = "0.9";
 252         protected $respStatus = "200 Ok";
 253         protected $respHeaders = array();
 254
 255         public $status;
 256
 257         /**
 258          * @var Profiler
 259          */
 260         protected $profiler;
 261
 262         /**
 263          * @var string
 264          */
 265         protected $profileName;
 266
 267         /**
 268          * @param string $url Url to use. If protocol-relative, will be expanded to an http:// URL
 269          * @param array $options (optional) extra params to pass (see Http::request())
 270          * @param string $caller The method making this request, for profiling
 271          * @param Profiler $profiler An instance of the profiler for profiling, or null
 272          */
 273         protected function __construct(
 274                 $url, $options = array(), $caller = __METHOD__, $profiler = null
 275         ) {
 276                 global $wgHTTPTimeout, $wgHTTPConnectTimeout;
 277
 278                 $this->url = wfExpandUrl( $url, PROTO_HTTP );
 279                 $this->parsedUrl = wfParseUrl( $this->url );
 280
 281                 if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) {
 282                         $this->status = Status::newFatal( 'http-invalid-url', $url );
 283                 } else {
 284                         $this->status = Status::newGood( 100 ); // continue
 285                 }
 286
 287                 if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) {
 288                         $this->timeout = $options['timeout'];
 289                 } else {
 290                         $this->timeout = $wgHTTPTimeout;
 291                 }
 292                 if ( isset( $options['connectTimeout'] ) && $options['connectTimeout'] != 'default' ) {
 293                         $this->connectTimeout = $options['connectTimeout'];
 294                 } else {
 295                         $this->connectTimeout = $wgHTTPConnectTimeout;
 296                 }
 297                 if ( isset( $options['userAgent'] ) ) {
 298                         $this->setUserAgent( $options['userAgent'] );
 299                 }
 300
 301                 $members = array( "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
 302                                 "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" );
 303
 304                 foreach ( $members as $o ) {
 305                         if ( isset( $options[$o] ) ) {
 306                                 // ensure that MWHttpRequest::method is always
 307                                 // uppercased. Bug 36137
 308                                 if ( $o == 'method' ) {
 309                                         $options[$o] = strtoupper( $options[$o] );
 310                                 }
 311                                 $this->$o = $options[$o];
 312                         }
 313                 }
 314
 315                 if ( $this->noProxy ) {
 316                         $this->proxy = ''; // noProxy takes precedence
 317                 }
 318
 319                 // Profile based on what's calling us
 320                 $this->profiler = $profiler;
 321                 $this->profileName = $caller;
 322         }
 323
 324         /**
 325          * Simple function to test if we can make any sort of requests at all, using
 326          * cURL or fopen()
 327          * @return bool
 328          */
 329         public static function canMakeRequests() {
 330                 return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
 331         }
 332
 333         /**
 334          * Generate a new request object
 335          * @param string $url Url to use
 336          * @param array $options (optional) extra params to pass (see Http::request())
 337          * @param string $caller The method making this request, for profiling
 338          * @throws MWException
 339          * @return CurlHttpRequest|PhpHttpRequest
 340          * @see MWHttpRequest::__construct
 341          */
 342         public static function factory( $url, $options = null, $caller = __METHOD__ ) {
 343                 if ( !Http::$httpEngine ) {
 344                         Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
 345                 } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
 346                         throw new MWException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' .
 347                                 ' Http::$httpEngine is set to "curl"' );
 348                 }
 349
 350                 switch ( Http::$httpEngine ) {
 351                         case 'curl':
 352                                 return new CurlHttpRequest( $url, $options, $caller, Profiler::instance() );
 353                         case 'php':
 354                                 if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
 355                                         throw new MWException( __METHOD__ . ': allow_url_fopen ' .
 356                                                 'needs to be enabled for pure PHP http requests to ' .
 357                                                 'work. If possible, curl should be used instead. See ' .
 358                                                 'http://php.net/curl.'
 359                                         );
 360                                 }
 361                                 return new PhpHttpRequest( $url, $options, $caller, Profiler::instance() );
 362                         default:
 363                                 throw new MWException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' );
 364                 }
 365         }
 366
 367         /**
 368          * Get the body, or content, of the response to the request
 369          *
 370          * @return string
 371          */
 372         public function getContent() {
 373                 return $this->content;
 374         }
 375
 376         /**
 377          * Set the parameters of the request
 378          *
 379          * @param array $args
 380          * @todo overload the args param
 381          */
 382         public function setData( $args ) {
 383                 $this->postData = $args;
 384         }
 385
 386         /**
 387          * Take care of setting up the proxy (do nothing if "noProxy" is set)
 388          *
 389          * @return void
 390          */
 391         public function proxySetup() {
 392                 // If there is an explicit proxy set and proxies are not disabled, then use it
 393                 if ( $this->proxy && !$this->noProxy ) {
 394                         return;
 395                 }
 396
 397                 // Otherwise, fallback to $wgHTTPProxy/http_proxy (when set) if this is not a machine
 398                 // local URL and proxies are not disabled
 399                 if ( Http::isLocalURL( $this->url ) || $this->noProxy ) {
 400                         $this->proxy = '';
 401                 } else {
 402                         $this->proxy = Http::getProxy();
 403                 }
 404         }
 405
 406         /**
 407          * Set the user agent
 408          * @param string $UA
 409          */
 410         public function setUserAgent( $UA ) {
 411                 $this->setHeader( 'User-Agent', $UA );
 412         }
 413
 414         /**
 415          * Set an arbitrary header
 416          * @param string $name
 417          * @param string $value
 418          */
 419         public function setHeader( $name, $value ) {
 420                 // I feel like I should normalize the case here...
 421                 $this->reqHeaders[$name] = $value;
 422         }
 423
 424         /**
 425          * Get an array of the headers
 426          * @return array
 427          */
 428         public function getHeaderList() {
 429                 $list = array();
 430
 431                 if ( $this->cookieJar ) {
 432                         $this->reqHeaders['Cookie'] =
 433                                 $this->cookieJar->serializeToHttpRequest(
 434                                         $this->parsedUrl['path'],
 435                                         $this->parsedUrl['host']
 436                                 );
 437                 }
 438
 439                 foreach ( $this->reqHeaders as $name => $value ) {
 440                         $list[] = "$name: $value";
 441                 }
 442
 443                 return $list;
 444         }
 445
 446         /**
 447          * Set a read callback to accept data read from the HTTP request.
 448          * By default, data is appended to an internal buffer which can be
 449          * retrieved through $req->getContent().
 450          *
 451          * To handle data as it comes in -- especially for large files that
 452          * would not fit in memory -- you can instead set your own callback,
 453          * in the form function($resource, $buffer) where the first parameter
 454          * is the low-level resource being read (implementation specific),
 455          * and the second parameter is the data buffer.
 456          *
 457          * You MUST return the number of bytes handled in the buffer; if fewer
 458          * bytes are reported handled than were passed to you, the HTTP fetch
 459          * will be aborted.
 460          *
 461          * @param callable $callback
 462          * @throws MWException
 463          */
 464         public function setCallback( $callback ) {
 465                 if ( !is_callable( $callback ) ) {
 466                         throw new MWException( 'Invalid MwHttpRequest callback' );
 467                 }
 468                 $this->callback = $callback;
 469         }
 470
 471         /**
 472          * A generic callback to read the body of the response from a remote
 473          * server.
 474          *
 475          * @param resource $fh
 476          * @param string $content
 477          * @return int
 478          */
 479         public function read( $fh, $content ) {
 480                 $this->content .= $content;
 481                 return strlen( $content );
 482         }
 483
 484         /**
 485          * Take care of whatever is necessary to perform the URI request.
 486          *
 487          * @return Status
 488          */
 489         public function execute() {
 490
 491                 $this->content = "";
 492
 493                 if ( strtoupper( $this->method ) == "HEAD" ) {
 494                         $this->headersOnly = true;
 495                 }
 496
 497                 $this->proxySetup(); // set up any proxy as needed
 498
 499                 if ( !$this->callback ) {
 500                         $this->setCallback( array( $this, 'read' ) );
 501                 }
 502
 503                 if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
 504                         $this->setUserAgent( Http::userAgent() );
 505                 }
 506
 507         }
 508
 509         /**
 510          * Parses the headers, including the HTTP status code and any
 511          * Set-Cookie headers.  This function expects the headers to be
 512          * found in an array in the member variable headerList.
 513          */
 514         protected function parseHeader() {
 515
 516                 $lastname = "";
 517
 518                 foreach ( $this->headerList as $header ) {
 519                         if ( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
 520                                 $this->respVersion = $match[1];
 521                                 $this->respStatus = $match[2];
 522                         } elseif ( preg_match( "#^[ \t]#", $header ) ) {
 523                                 $last = count( $this->respHeaders[$lastname] ) - 1;
 524                                 $this->respHeaders[$lastname][$last] .= "\r\n$header";
 525                         } elseif ( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
 526                                 $this->respHeaders[strtolower( $match[1] )][] = $match[2];
 527                                 $lastname = strtolower( $match[1] );
 528                         }
 529                 }
 530
 531                 $this->parseCookies();
 532
 533         }
 534
 535         /**
 536          * Sets HTTPRequest status member to a fatal value with the error
 537          * message if the returned integer value of the status code was
 538          * not successful (< 300) or a redirect (>=300 and < 400).  (see
 539          * RFC2616, section 10,
 540          * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a
 541          * list of status codes.)
 542          */
 543         protected function setStatus() {
 544                 if ( !$this->respHeaders ) {
 545                         $this->parseHeader();
 546                 }
 547
 548                 if ( (int)$this->respStatus > 399 ) {
 549                         list( $code, $message ) = explode( " ", $this->respStatus, 2 );
 550                         $this->status->fatal( "http-bad-status", $code, $message );
 551                 }
 552         }
 553
 554         /**
 555          * Get the integer value of the HTTP status code (e.g. 200 for "200 Ok")
 556          * (see RFC2616, section 10, http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
 557          * for a list of status codes.)
 558          *
 559          * @return int
 560          */
 561         public function getStatus() {
 562                 if ( !$this->respHeaders ) {
 563                         $this->parseHeader();
 564                 }
 565
 566                 return (int)$this->respStatus;
 567         }
 568
 569         /**
 570          * Returns true if the last status code was a redirect.
 571          *
 572          * @return bool
 573          */
 574         public function isRedirect() {
 575                 if ( !$this->respHeaders ) {
 576                         $this->parseHeader();
 577                 }
 578
 579                 $status = (int)$this->respStatus;
 580
 581                 if ( $status >= 300 && $status <= 303 ) {
 582                         return true;
 583                 }
 584
 585                 return false;
 586         }
 587
 588         /**
 589          * Returns an associative array of response headers after the
 590          * request has been executed.  Because some headers
 591          * (e.g. Set-Cookie) can appear more than once the, each value of
 592          * the associative array is an array of the values given.
 593          *
 594          * @return array
 595          */
 596         public function getResponseHeaders() {
 597                 if ( !$this->respHeaders ) {
 598                         $this->parseHeader();
 599                 }
 600
 601                 return $this->respHeaders;
 602         }
 603
 604         /**
 605          * Returns the value of the given response header.
 606          *
 607          * @param string $header
 608          * @return string
 609          */
 610         public function getResponseHeader( $header ) {
 611                 if ( !$this->respHeaders ) {
 612                         $this->parseHeader();
 613                 }
 614
 615                 if ( isset( $this->respHeaders[strtolower( $header )] ) ) {
 616                         $v = $this->respHeaders[strtolower( $header )];
 617                         return $v[count( $v ) - 1];
 618                 }
 619
 620                 return null;
 621         }
 622
 623         /**
 624          * Tells the MWHttpRequest object to use this pre-loaded CookieJar.
 625          *
 626          * @param CookieJar $jar
 627          */
 628         public function setCookieJar( $jar ) {
 629                 $this->cookieJar = $jar;
 630         }
 631
 632         /**
 633          * Returns the cookie jar in use.
 634          *
 635          * @return CookieJar
 636          */
 637         public function getCookieJar() {
 638                 if ( !$this->respHeaders ) {
 639                         $this->parseHeader();
 640                 }
 641
 642                 return $this->cookieJar;
 643         }
 644
 645         /**
 646          * Sets a cookie. Used before a request to set up any individual
 647          * cookies. Used internally after a request to parse the
 648          * Set-Cookie headers.
 649          * @see Cookie::set
 650          * @param string $name
 651          * @param mixed $value
 652          * @param array $attr
 653          */
 654         public function setCookie( $name, $value = null, $attr = null ) {
 655                 if ( !$this->cookieJar ) {
 656                         $this->cookieJar = new CookieJar;
 657                 }
 658
 659                 $this->cookieJar->setCookie( $name, $value, $attr );
 660         }
 661
 662         /**
 663          * Parse the cookies in the response headers and store them in the cookie jar.
 664          */
 665         protected function parseCookies() {
 666
 667                 if ( !$this->cookieJar ) {
 668                         $this->cookieJar = new CookieJar;
 669                 }
 670
 671                 if ( isset( $this->respHeaders['set-cookie'] ) ) {
 672                         $url = parse_url( $this->getFinalUrl() );
 673                         foreach ( $this->respHeaders['set-cookie'] as $cookie ) {
 674                                 $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
 675                         }
 676                 }
 677
 678         }
 679
 680         /**
 681          * Returns the final URL after all redirections.
 682          *
 683          * Relative values of the "Location" header are incorrect as
 684          * stated in RFC, however they do happen and modern browsers
 685          * support them.  This function loops backwards through all
 686          * locations in order to build the proper absolute URI - Marooned
 687          * at wikia-inc.com
 688          *
 689          * Note that the multiple Location: headers are an artifact of
 690          * CURL -- they shouldn't actually get returned this way. Rewrite
 691          * this when bug 29232 is taken care of (high-level redirect
 692          * handling rewrite).
 693          *
 694          * @return string
 695          */
 696         public function getFinalUrl() {
 697                 $headers = $this->getResponseHeaders();
 698
 699                 // return full url (fix for incorrect but handled relative location)
 700                 if ( isset( $headers['location'] ) ) {
 701                         $locations = $headers['location'];
 702                         $domain = '';
 703                         $foundRelativeURI = false;
 704                         $countLocations = count( $locations );
 705
 706                         for ( $i = $countLocations - 1; $i >= 0; $i-- ) {
 707                                 $url = parse_url( $locations[$i] );
 708
 709                                 if ( isset( $url['host'] ) ) {
 710                                         $domain = $url['scheme'] . '://' . $url['host'];
 711                                         break; // found correct URI (with host)
 712                                 } else {
 713                                         $foundRelativeURI = true;
 714                                 }
 715                         }
 716
 717                         if ( $foundRelativeURI ) {
 718                                 if ( $domain ) {
 719                                         return $domain . $locations[$countLocations - 1];
 720                                 } else {
 721                                         $url = parse_url( $this->url );
 722                                         if ( isset( $url['host'] ) ) {
 723                                                 return $url['scheme'] . '://' . $url['host'] .
 724                                                         $locations[$countLocations - 1];
 725                                         }
 726                                 }
 727                         } else {
 728                                 return $locations[$countLocations - 1];
 729                         }
 730                 }
 731
 732                 return $this->url;
 733         }
 734
 735         /**
 736          * Returns true if the backend can follow redirects. Overridden by the
 737          * child classes.
 738          * @return bool
 739          */
 740         public function canFollowRedirects() {
 741                 return true;
 742         }
 743 }
 744
 745 /**
 746  * MWHttpRequest implemented using internal curl compiled into PHP
 747  */
 748 class CurlHttpRequest extends MWHttpRequest {
 749         const SUPPORTS_FILE_POSTS = true;
 750
 751         protected $curlOptions = array();
 752         protected $headerText = "";
 753
 754         /**
 755          * @param resource $fh
 756          * @param string $content
 757          * @return int
 758          */
 759         protected function readHeader( $fh, $content ) {
 760                 $this->headerText .= $content;
 761                 return strlen( $content );
 762         }
 763
 764         public function execute() {
 765
 766                 parent::execute();
 767
 768                 if ( !$this->status->isOK() ) {
 769                         return $this->status;
 770                 }
 771
 772                 $this->curlOptions[CURLOPT_PROXY] = $this->proxy;
 773                 $this->curlOptions[CURLOPT_TIMEOUT] = $this->timeout;
 774
 775                 // Only supported in curl >= 7.16.2
 776                 if ( defined( 'CURLOPT_CONNECTTIMEOUT_MS' ) ) {
 777                         $this->curlOptions[CURLOPT_CONNECTTIMEOUT_MS] = $this->connectTimeout * 1000;
 778                 }
 779
 780                 $this->curlOptions[CURLOPT_HTTP_VERSION] = CURL_HTTP_VERSION_1_0;
 781                 $this->curlOptions[CURLOPT_WRITEFUNCTION] = $this->callback;
 782                 $this->curlOptions[CURLOPT_HEADERFUNCTION] = array( $this, "readHeader" );
 783                 $this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
 784                 $this->curlOptions[CURLOPT_ENCODING] = ""; # Enable compression
 785
 786                 $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
 787
 788                 $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost ? 2 : 0;
 789                 $this->curlOptions[CURLOPT_SSL_VERIFYPEER] = $this->sslVerifyCert;
 790
 791                 if ( $this->caInfo ) {
 792                         $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
 793                 }
 794
 795                 if ( $this->headersOnly ) {
 796                         $this->curlOptions[CURLOPT_NOBODY] = true;
 797                         $this->curlOptions[CURLOPT_HEADER] = true;
 798                 } elseif ( $this->method == 'POST' ) {
 799                         $this->curlOptions[CURLOPT_POST] = true;
 800                         $this->curlOptions[CURLOPT_POSTFIELDS] = $this->postData;
 801                         // Suppress 'Expect: 100-continue' header, as some servers
 802                         // will reject it with a 417 and Curl won't auto retry
 803                         // with HTTP 1.0 fallback
 804                         $this->reqHeaders['Expect'] = '';
 805                 } else {
 806                         $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
 807                 }
 808
 809                 $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
 810
 811                 $curlHandle = curl_init( $this->url );
 812
 813                 if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
 814                         throw new MWException( "Error setting curl options." );
 815                 }
 816
 817                 if ( $this->followRedirects && $this->canFollowRedirects() ) {
 818                         MediaWiki\suppressWarnings();
 819                         if ( !curl_setopt( $curlHandle, CURLOPT_FOLLOWLOCATION, true ) ) {
 820                                 wfDebug( __METHOD__ . ": Couldn't set CURLOPT_FOLLOWLOCATION. " .
 821                                         "Probably safe_mode or open_basedir is set.\n" );
 822                                 // Continue the processing. If it were in curl_setopt_array,
 823                                 // processing would have halted on its entry
 824                         }
 825                         MediaWiki\restoreWarnings();
 826                 }
 827
 828                 if ( $this->profiler ) {
 829                         $profileSection = $this->profiler->scopedProfileIn(
 830                                 __METHOD__ . '-' . $this->profileName
 831                         );
 832                 }
 833
 834                 $curlRes = curl_exec( $curlHandle );
 835                 if ( curl_errno( $curlHandle ) == CURLE_OPERATION_TIMEOUTED ) {
 836                         $this->status->fatal( 'http-timed-out', $this->url );
 837                 } elseif ( $curlRes === false ) {
 838                         $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
 839                 } else {
 840                         $this->headerList = explode( "\r\n", $this->headerText );
 841                 }
 842
 843                 curl_close( $curlHandle );
 844
 845                 if ( $this->profiler ) {
 846                         $this->profiler->scopedProfileOut( $profileSection );
 847                 }
 848
 849                 $this->parseHeader();
 850                 $this->setStatus();
 851
 852                 return $this->status;
 853         }
 854
 855         /**
 856          * @return bool
 857          */
 858         public function canFollowRedirects() {
 859                 $curlVersionInfo = curl_version();
 860                 if ( $curlVersionInfo['version_number'] < 0x071304 ) {
 861                         wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
 862                         return false;
 863                 }
 864
 865                 if ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 866                         if ( strval( ini_get( 'open_basedir' ) ) !== '' || wfIniGetBool( 'safe_mode' ) ) {
 867                                 wfDebug( "Cannot follow redirects in safe mode\n" );
 868                                 return false;
 869                         }
 870                 }
 871
 872                 return true;
 873         }
 874 }
 875
 876 class PhpHttpRequest extends MWHttpRequest {
 877
 878         private $fopenErrors = array();
 879
 880         /**
 881          * @param string $url
 882          * @return string
 883          */
 884         protected function urlToTcp( $url ) {
 885                 $parsedUrl = parse_url( $url );
 886
 887                 return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
 888         }
 889
 890         /**
 891          * Returns an array with a 'capath' or 'cafile' key
 892          * that is suitable to be merged into the 'ssl' sub-array of
 893          * a stream context options array.
 894          * Uses the 'caInfo' option of the class if it is provided, otherwise uses the system
 895          * default CA bundle if PHP supports that, or searches a few standard locations.
 896          * @return array
 897          * @throws DomainException
 898          */
 899         protected function getCertOptions() {
 900                 $certOptions = array();
 901                 $certLocations = array();
 902                 if ( $this->caInfo ) {
 903                         $certLocations = array( 'manual' => $this->caInfo );
 904                 } elseif ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 905                         // @codingStandardsIgnoreStart Generic.Files.LineLength
 906                         // Default locations, based on
 907                         // https://www.happyassassin.net/2015/01/12/a-note-about-ssltls-trusted-certificate-stores-and-platforms/
 908                         // PHP 5.5 and older doesn't have any defaults, so we try to guess ourselves.
 909                         // PHP 5.6+ gets the CA location from OpenSSL as long as it is not set manually,
 910                         // so we should leave capath/cafile empty there.
 911                         // @codingStandardsIgnoreEnd
 912                         $certLocations = array_filter( array(
 913                                 getenv( 'SSL_CERT_DIR' ),
 914                                 getenv( 'SSL_CERT_PATH' ),
 915                                 '/etc/pki/tls/certs/ca-bundle.crt', # Fedora et al
 916                                 '/etc/ssl/certs',  # Debian et al
 917                                 '/etc/pki/tls/certs/ca-bundle.trust.crt',
 918                                 '/etc/pki/ca-trust/extracted/pem/tls-ca-bundle.pem',
 919                                 '/System/Library/OpenSSL', # OSX
 920                         ) );
 921                 }
 922
 923                 foreach ( $certLocations as $key => $cert ) {
 924                         if ( is_dir( $cert ) ) {
 925                                 $certOptions['capath'] = $cert;
 926                                 break;
 927                         } elseif ( is_file( $cert ) ) {
 928                                 $certOptions['cafile'] = $cert;
 929                                 break;
 930                         } elseif ( $key === 'manual' ) {
 931                                 // fail more loudly if a cert path was manually configured and it is not valid
 932                                 throw new DomainException( "Invalid CA info passed: $cert" );
 933                         }
 934                 }
 935
 936                 return $certOptions;
 937         }
 938
 939         /**
 940          * Custom error handler for dealing with fopen() errors.
 941          * fopen() tends to fire multiple errors in succession, and the last one
 942          * is completely useless (something like "fopen: failed to open stream")
 943          * so normal methods of handling errors programmatically
 944          * like get_last_error() don't work.
 945          */
 946         public function errorHandler( $errno, $errstr ) {
 947                 $n = count( $this->fopenErrors ) + 1;
 948                 $this->fopenErrors += array( "errno$n" => $errno, "errstr$n" => $errstr );
 949         }
 950
 951         public function execute() {
 952
 953                 parent::execute();
 954
 955                 if ( is_array( $this->postData ) ) {
 956                         $this->postData = wfArrayToCgi( $this->postData );
 957                 }
 958
 959                 if ( $this->parsedUrl['scheme'] != 'http'
 960                         && $this->parsedUrl['scheme'] != 'https' ) {
 961                         $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
 962                 }
 963
 964                 $this->reqHeaders['Accept'] = "*/*";
 965                 $this->reqHeaders['Connection'] = 'Close';
 966                 if ( $this->method == 'POST' ) {
 967                         // Required for HTTP 1.0 POSTs
 968                         $this->reqHeaders['Content-Length'] = strlen( $this->postData );
 969                         if ( !isset( $this->reqHeaders['Content-Type'] ) ) {
 970                                 $this->reqHeaders['Content-Type'] = "application/x-www-form-urlencoded";
 971                         }
 972                 }
 973
 974                 // Set up PHP stream context
 975                 $options = array(
 976                         'http' => array(
 977                                 'method' => $this->method,
 978                                 'header' => implode( "\r\n", $this->getHeaderList() ),
 979                                 'protocol_version' => '1.1',
 980                                 'max_redirects' => $this->followRedirects ? $this->maxRedirects : 0,
 981                                 'ignore_errors' => true,
 982                                 'timeout' => $this->timeout,
 983                                 // Curl options in case curlwrappers are installed
 984                                 'curl_verify_ssl_host' => $this->sslVerifyHost ? 2 : 0,
 985                                 'curl_verify_ssl_peer' => $this->sslVerifyCert,
 986                         ),
 987                         'ssl' => array(
 988                                 'verify_peer' => $this->sslVerifyCert,
 989                                 'SNI_enabled' => true,
 990                                 'ciphers' => 'HIGH:!SSLv2:!SSLv3:-ADH:-kDH:-kECDH:-DSS',
 991                                 'disable_compression' => true,
 992                         ),
 993                 );
 994
 995                 if ( $this->proxy ) {
 996                         $options['http']['proxy'] = $this->urlToTCP( $this->proxy );
 997                         $options['http']['request_fulluri'] = true;
 998                 }
 999
1000                 if ( $this->postData ) {
1001                         $options['http']['content'] = $this->postData;
1002                 }
1003
1004                 if ( $this->sslVerifyHost ) {
1005                         // PHP 5.6.0 deprecates CN_match, in favour of peer_name which
1006                         // actually checks SubjectAltName properly.
1007                         if ( version_compare( PHP_VERSION, '5.6.0', '>=' ) ) {
1008                                 $options['ssl']['peer_name'] = $this->parsedUrl['host'];
1009                         } else {
1010                                 $options['ssl']['CN_match'] = $this->parsedUrl['host'];
1011                         }
1012                 }
1013
1014                 $options['ssl'] += $this->getCertOptions();
1015
1016                 $context = stream_context_create( $options );
1017
1018                 $this->headerList = array();
1019                 $reqCount = 0;
1020                 $url = $this->url;
1021
1022                 $result = array();
1023
1024                 if ( $this->profiler ) {
1025                         $profileSection = $this->profiler->scopedProfileIn(
1026                                 __METHOD__ . '-' . $this->profileName
1027                         );
1028                 }
1029                 do {
1030                         $reqCount++;
1031                         $this->fopenErrors = array();
1032                         set_error_handler( array( $this, 'errorHandler' ) );
1033                         $fh = fopen( $url, "r", false, $context );
1034                         restore_error_handler();
1035
1036                         if ( !$fh ) {
1037                                 // HACK for instant commons.
1038                                 // If we are contacting (commons|upload).wikimedia.org
1039                                 // try again with CN_match for en.wikipedia.org
1040                                 // as php does not handle SubjectAltName properly
1041                                 // prior to "peer_name" option in php 5.6
1042                                 if ( isset( $options['ssl']['CN_match'] )
1043                                         && ( $options['ssl']['CN_match'] === 'commons.wikimedia.org'
1044                                                 || $options['ssl']['CN_match'] === 'upload.wikimedia.org' )
1045                                 ) {
1046                                         $options['ssl']['CN_match'] = 'en.wikipedia.org';
1047                                         $context = stream_context_create( $options );
1048                                         continue;
1049                                 }
1050                                 break;
1051                         }
1052
1053                         $result = stream_get_meta_data( $fh );
1054                         $this->headerList = $result['wrapper_data'];
1055                         $this->parseHeader();
1056
1057                         if ( !$this->followRedirects ) {
1058                                 break;
1059                         }
1060
1061                         # Handle manual redirection
1062                         if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
1063                                 break;
1064                         }
1065                         # Check security of URL
1066                         $url = $this->getResponseHeader( "Location" );
1067
1068                         if ( !Http::isValidURI( $url ) ) {
1069                                 wfDebug( __METHOD__ . ": insecure redirection\n" );
1070                                 break;
1071                         }
1072                 } while ( true );
1073                 if ( $this->profiler ) {
1074                         $this->profiler->scopedProfileOut( $profileSection );
1075                 }
1076
1077                 $this->setStatus();
1078
1079                 if ( $fh === false ) {
1080                         if ( $this->fopenErrors ) {
1081                                 LoggerFactory::getInstance( 'http' )->warning( __CLASS__
1082                                         . ': error opening connection: {errstr1}', $this->fopenErrors );
1083                         }
1084                         $this->status->fatal( 'http-request-error' );
1085                         return $this->status;
1086                 }
1087
1088                 if ( $result['timed_out'] ) {
1089                         $this->status->fatal( 'http-timed-out', $this->url );
1090                         return $this->status;
1091                 }
1092
1093                 // If everything went OK, or we received some error code
1094                 // get the response body content.
1095                 if ( $this->status->isOK() || (int)$this->respStatus >= 300 ) {
1096                         while ( !feof( $fh ) ) {
1097                                 $buf = fread( $fh, 8192 );
1098
1099                                 if ( $buf === false ) {
1100                                         $this->status->fatal( 'http-read-error' );
1101                                         break;
1102                                 }
1103
1104                                 if ( strlen( $buf ) ) {
1105                                         call_user_func( $this->callback, $fh, $buf );
1106                                 }
1107                         }
1108                 }
1109                 fclose( $fh );
1110
1111                 return $this->status;
1112         }
1113 }