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