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