includes/HttpFunctions.php

   1 <?php
   2 /**
   3  * Various HTTP related functions.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup HTTP
  22  */
  23
  24 /**
  25  * @defgroup HTTP HTTP
  26  */
  27
  28 use MediaWiki\Logger\LoggerFactory;
  29
  30 /**
  31  * Various HTTP related functions
  32  * @ingroup HTTP
  33  */
  34 class Http {
  35         static public $httpEngine = false;
  36
  37         /**
  38          * Perform an HTTP request
  39          *
  40          * @param string $method HTTP method. Usually GET/POST
  41          * @param string $url Full URL to act on. If protocol-relative, will be expanded to an http:// URL
  42          * @param array $options Options to pass to MWHttpRequest object.
  43          *      Possible keys for the array:
  44          *    - timeout             Timeout length in seconds
  45          *    - connectTimeout      Timeout for connection, in seconds (curl only)
  46          *    - postData            An array of key-value pairs or a url-encoded form data
  47          *    - proxy               The proxy to use.
  48          *                          Otherwise it will use $wgHTTPProxy (if set)
  49          *                          Otherwise it will use the environment variable "http_proxy" (if set)
  50          *    - noProxy             Don't use any proxy at all. Takes precedence over proxy value(s).
  51          *    - sslVerifyHost       Verify hostname against certificate
  52          *    - sslVerifyCert       Verify SSL certificate
  53          *    - caInfo              Provide CA information
  54          *    - maxRedirects        Maximum number of redirects to follow (defaults to 5)
  55          *    - followRedirects     Whether to follow redirects (defaults to false).
  56          *                                  Note: this should only be used when the target URL is trusted,
  57          *                                  to avoid attacks on intranet services accessible by HTTP.
  58          *    - userAgent           A user agent, if you want to override the default
  59          *                          MediaWiki/$wgVersion
  60          * @param string $caller The method making this request, for profiling
  61          * @return string|bool (bool)false on failure or a string on success
  62          */
  63         public static function request( $method, $url, $options = [], $caller = __METHOD__ ) {
  64                 wfDebug( "HTTP: $method: $url\n" );
  65
  66                 $options['method'] = strtoupper( $method );
  67
  68                 if ( !isset( $options['timeout'] ) ) {
  69                         $options['timeout'] = 'default';
  70                 }
  71                 if ( !isset( $options['connectTimeout'] ) ) {
  72                         $options['connectTimeout'] = 'default';
  73                 }
  74
  75                 $req = MWHttpRequest::factory( $url, $options, $caller );
  76                 $status = $req->execute();
  77
  78                 if ( $status->isOK() ) {
  79                         return $req->getContent();
  80                 } else {
  81                         $errors = $status->getErrorsByType( 'error' );
  82                         $logger = LoggerFactory::getInstance( 'http' );
  83                         $logger->warning( $status->getWikiText( null, null, 'en' ),
  84                                 [ 'error' => $errors, 'caller' => $caller, 'content' => $req->getContent() ] );
  85                         return false;
  86                 }
  87         }
  88
  89         /**
  90          * Simple wrapper for Http::request( 'GET' )
  91          * @see Http::request()
  92          * @since 1.25 Second parameter $timeout removed. Second parameter
  93          * is now $options which can be given a 'timeout'
  94          *
  95          * @param string $url
  96          * @param array $options
  97          * @param string $caller The method making this request, for profiling
  98          * @return string|bool false on error
  99          */
 100         public static function get( $url, $options = [], $caller = __METHOD__ ) {
 101                 $args = func_get_args();
 102                 if ( isset( $args[1] ) && ( is_string( $args[1] ) || is_numeric( $args[1] ) ) ) {
 103                         // Second was used to be the timeout
 104                         // And third parameter used to be $options
 105                         wfWarn( "Second parameter should not be a timeout.", 2 );
 106                         $options = isset( $args[2] ) && is_array( $args[2] ) ?
 107                                 $args[2] : [];
 108                         $options['timeout'] = $args[1];
 109                         $caller = __METHOD__;
 110                 }
 111                 return Http::request( 'GET', $url, $options, $caller );
 112         }
 113
 114         /**
 115          * Simple wrapper for Http::request( 'POST' )
 116          * @see Http::request()
 117          *
 118          * @param string $url
 119          * @param array $options
 120          * @param string $caller The method making this request, for profiling
 121          * @return string|bool false on error
 122          */
 123         public static function post( $url, $options = [], $caller = __METHOD__ ) {
 124                 return Http::request( 'POST', $url, $options, $caller );
 125         }
 126
 127         /**
 128          * Check if the URL can be served by localhost
 129          *
 130          * @param string $url Full url to check
 131          * @return bool
 132          */
 133         public static function isLocalURL( $url ) {
 134                 global $wgCommandLineMode, $wgLocalVirtualHosts;
 135
 136                 if ( $wgCommandLineMode ) {
 137                         return false;
 138                 }
 139
 140                 // Extract host part
 141                 $matches = [];
 142                 if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
 143                         $host = $matches[1];
 144                         // Split up dotwise
 145                         $domainParts = explode( '.', $host );
 146                         // Check if this domain or any superdomain is listed as a local virtual host
 147                         $domainParts = array_reverse( $domainParts );
 148
 149                         $domain = '';
 150                         $countParts = count( $domainParts );
 151                         for ( $i = 0; $i < $countParts; $i++ ) {
 152                                 $domainPart = $domainParts[$i];
 153                                 if ( $i == 0 ) {
 154                                         $domain = $domainPart;
 155                                 } else {
 156                                         $domain = $domainPart . '.' . $domain;
 157                                 }
 158
 159                                 if ( in_array( $domain, $wgLocalVirtualHosts ) ) {
 160                                         return true;
 161                                 }
 162                         }
 163                 }
 164
 165                 return false;
 166         }
 167
 168         /**
 169          * A standard user-agent we can use for external requests.
 170          * @return string
 171          */
 172         public static function userAgent() {
 173                 global $wgVersion;
 174                 return "MediaWiki/$wgVersion";
 175         }
 176
 177         /**
 178          * Checks that the given URI is a valid one. Hardcoding the
 179          * protocols, because we only want protocols that both cURL
 180          * and php support.
 181          *
 182          * file:// should not be allowed here for security purpose (r67684)
 183          *
 184          * @todo FIXME this is wildly inaccurate and fails to actually check most stuff
 185          *
 186          * @param string $uri URI to check for validity
 187          * @return bool
 188          */
 189         public static function isValidURI( $uri ) {
 190                 return preg_match(
 191                         '/^https?:\/\/[^\/\s]\S*$/D',
 192                         $uri
 193                 );
 194         }
 195 }
 196
 197 /**
 198  * This wrapper class will call out to curl (if available) or fallback
 199  * to regular PHP if necessary for handling internal HTTP requests.
 200  *
 201  * Renamed from HttpRequest to MWHttpRequest to avoid conflict with
 202  * PHP's HTTP extension.
 203  */
 204 class MWHttpRequest {
 205         const SUPPORTS_FILE_POSTS = false;
 206
 207         protected $content;
 208         protected $timeout = 'default';
 209         protected $headersOnly = null;
 210         protected $postData = null;
 211         protected $proxy = null;
 212         protected $noProxy = false;
 213         protected $sslVerifyHost = true;
 214         protected $sslVerifyCert = true;
 215         protected $caInfo = null;
 216         protected $method = "GET";
 217         protected $reqHeaders = [];
 218         protected $url;
 219         protected $parsedUrl;
 220         protected $callback;
 221         protected $maxRedirects = 5;
 222         protected $followRedirects = false;
 223
 224         /**
 225          * @var CookieJar
 226          */
 227         protected $cookieJar;
 228
 229         protected $headerList = [];
 230         protected $respVersion = "0.9";
 231         protected $respStatus = "200 Ok";
 232         protected $respHeaders = [];
 233
 234         public $status;
 235
 236         /**
 237          * @var Profiler
 238          */
 239         protected $profiler;
 240
 241         /**
 242          * @var string
 243          */
 244         protected $profileName;
 245
 246         /**
 247          * @param string $url Url to use. If protocol-relative, will be expanded to an http:// URL
 248          * @param array $options (optional) extra params to pass (see Http::request())
 249          * @param string $caller The method making this request, for profiling
 250          * @param Profiler $profiler An instance of the profiler for profiling, or null
 251          */
 252         protected function __construct(
 253                 $url, $options = [], $caller = __METHOD__, $profiler = null
 254         ) {
 255                 global $wgHTTPTimeout, $wgHTTPConnectTimeout;
 256
 257                 $this->url = wfExpandUrl( $url, PROTO_HTTP );
 258                 $this->parsedUrl = wfParseUrl( $this->url );
 259
 260                 if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) {
 261                         $this->status = Status::newFatal( 'http-invalid-url', $url );
 262                 } else {
 263                         $this->status = Status::newGood( 100 ); // continue
 264                 }
 265
 266                 if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) {
 267                         $this->timeout = $options['timeout'];
 268                 } else {
 269                         $this->timeout = $wgHTTPTimeout;
 270                 }
 271                 if ( isset( $options['connectTimeout'] ) && $options['connectTimeout'] != 'default' ) {
 272                         $this->connectTimeout = $options['connectTimeout'];
 273                 } else {
 274                         $this->connectTimeout = $wgHTTPConnectTimeout;
 275                 }
 276                 if ( isset( $options['userAgent'] ) ) {
 277                         $this->setUserAgent( $options['userAgent'] );
 278                 }
 279
 280                 $members = [ "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
 281                                 "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" ];
 282
 283                 foreach ( $members as $o ) {
 284                         if ( isset( $options[$o] ) ) {
 285                                 // ensure that MWHttpRequest::method is always
 286                                 // uppercased. Bug 36137
 287                                 if ( $o == 'method' ) {
 288                                         $options[$o] = strtoupper( $options[$o] );
 289                                 }
 290                                 $this->$o = $options[$o];
 291                         }
 292                 }
 293
 294                 if ( $this->noProxy ) {
 295                         $this->proxy = ''; // noProxy takes precedence
 296                 }
 297
 298                 // Profile based on what's calling us
 299                 $this->profiler = $profiler;
 300                 $this->profileName = $caller;
 301         }
 302
 303         /**
 304          * Simple function to test if we can make any sort of requests at all, using
 305          * cURL or fopen()
 306          * @return bool
 307          */
 308         public static function canMakeRequests() {
 309                 return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
 310         }
 311
 312         /**
 313          * Generate a new request object
 314          * @param string $url Url to use
 315          * @param array $options (optional) extra params to pass (see Http::request())
 316          * @param string $caller The method making this request, for profiling
 317          * @throws MWException
 318          * @return CurlHttpRequest|PhpHttpRequest
 319          * @see MWHttpRequest::__construct
 320          */
 321         public static function factory( $url, $options = null, $caller = __METHOD__ ) {
 322                 if ( !Http::$httpEngine ) {
 323                         Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
 324                 } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
 325                         throw new MWException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' .
 326                                 ' Http::$httpEngine is set to "curl"' );
 327                 }
 328
 329                 switch ( Http::$httpEngine ) {
 330                         case 'curl':
 331                                 return new CurlHttpRequest( $url, $options, $caller, Profiler::instance() );
 332                         case 'php':
 333                                 if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
 334                                         throw new MWException( __METHOD__ . ': allow_url_fopen ' .
 335                                                 'needs to be enabled for pure PHP http requests to ' .
 336                                                 'work. If possible, curl should be used instead. See ' .
 337                                                 'http://php.net/curl.'
 338                                         );
 339                                 }
 340                                 return new PhpHttpRequest( $url, $options, $caller, Profiler::instance() );
 341                         default:
 342                                 throw new MWException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' );
 343                 }
 344         }
 345
 346         /**
 347          * Get the body, or content, of the response to the request
 348          *
 349          * @return string
 350          */
 351         public function getContent() {
 352                 return $this->content;
 353         }
 354
 355         /**
 356          * Set the parameters of the request
 357          *
 358          * @param array $args
 359          * @todo overload the args param
 360          */
 361         public function setData( $args ) {
 362                 $this->postData = $args;
 363         }
 364
 365         /**
 366          * Take care of setting up the proxy (do nothing if "noProxy" is set)
 367          *
 368          * @return void
 369          */
 370         public function proxySetup() {
 371                 global $wgHTTPProxy;
 372
 373                 // If there is an explicit proxy set and proxies are not disabled, then use it
 374                 if ( $this->proxy && !$this->noProxy ) {
 375                         return;
 376                 }
 377
 378                 // Otherwise, fallback to $wgHTTPProxy/http_proxy (when set) if this is not a machine
 379                 // local URL and proxies are not disabled
 380                 if ( Http::isLocalURL( $this->url ) || $this->noProxy ) {
 381                         $this->proxy = '';
 382                 } elseif ( $wgHTTPProxy ) {
 383                         $this->proxy = $wgHTTPProxy;
 384                 } elseif ( getenv( "http_proxy" ) ) {
 385                         $this->proxy = getenv( "http_proxy" );
 386                 }
 387         }
 388
 389         /**
 390          * Set the user agent
 391          * @param string $UA
 392          */
 393         public function setUserAgent( $UA ) {
 394                 $this->setHeader( 'User-Agent', $UA );
 395         }
 396
 397         /**
 398          * Set an arbitrary header
 399          * @param string $name
 400          * @param string $value
 401          */
 402         public function setHeader( $name, $value ) {
 403                 // I feel like I should normalize the case here...
 404                 $this->reqHeaders[$name] = $value;
 405         }
 406
 407         /**
 408          * Get an array of the headers
 409          * @return array
 410          */
 411         public function getHeaderList() {
 412                 $list = [];
 413
 414                 if ( $this->cookieJar ) {
 415                         $this->reqHeaders['Cookie'] =
 416                                 $this->cookieJar->serializeToHttpRequest(
 417                                         $this->parsedUrl['path'],
 418                                         $this->parsedUrl['host']
 419                                 );
 420                 }
 421
 422                 foreach ( $this->reqHeaders as $name => $value ) {
 423                         $list[] = "$name: $value";
 424                 }
 425
 426                 return $list;
 427         }
 428
 429         /**
 430          * Set a read callback to accept data read from the HTTP request.
 431          * By default, data is appended to an internal buffer which can be
 432          * retrieved through $req->getContent().
 433          *
 434          * To handle data as it comes in -- especially for large files that
 435          * would not fit in memory -- you can instead set your own callback,
 436          * in the form function($resource, $buffer) where the first parameter
 437          * is the low-level resource being read (implementation specific),
 438          * and the second parameter is the data buffer.
 439          *
 440          * You MUST return the number of bytes handled in the buffer; if fewer
 441          * bytes are reported handled than were passed to you, the HTTP fetch
 442          * will be aborted.
 443          *
 444          * @param callable $callback
 445          * @throws MWException
 446          */
 447         public function setCallback( $callback ) {
 448                 if ( !is_callable( $callback ) ) {
 449                         throw new MWException( 'Invalid MwHttpRequest callback' );
 450                 }
 451                 $this->callback = $callback;
 452         }
 453
 454         /**
 455          * A generic callback to read the body of the response from a remote
 456          * server.
 457          *
 458          * @param resource $fh
 459          * @param string $content
 460          * @return int
 461          */
 462         public function read( $fh, $content ) {
 463                 $this->content .= $content;
 464                 return strlen( $content );
 465         }
 466
 467         /**
 468          * Take care of whatever is necessary to perform the URI request.
 469          *
 470          * @return Status
 471          */
 472         public function execute() {
 473
 474                 $this->content = "";
 475
 476                 if ( strtoupper( $this->method ) == "HEAD" ) {
 477                         $this->headersOnly = true;
 478                 }
 479
 480                 $this->proxySetup(); // set up any proxy as needed
 481
 482                 if ( !$this->callback ) {
 483                         $this->setCallback( [ $this, 'read' ] );
 484                 }
 485
 486                 if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
 487                         $this->setUserAgent( Http::userAgent() );
 488                 }
 489
 490         }
 491
 492         /**
 493          * Parses the headers, including the HTTP status code and any
 494          * Set-Cookie headers.  This function expects the headers to be
 495          * found in an array in the member variable headerList.
 496          */
 497         protected function parseHeader() {
 498
 499                 $lastname = "";
 500
 501                 foreach ( $this->headerList as $header ) {
 502                         if ( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
 503                                 $this->respVersion = $match[1];
 504                                 $this->respStatus = $match[2];
 505                         } elseif ( preg_match( "#^[ \t]#", $header ) ) {
 506                                 $last = count( $this->respHeaders[$lastname] ) - 1;
 507                                 $this->respHeaders[$lastname][$last] .= "\r\n$header";
 508                         } elseif ( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
 509                                 $this->respHeaders[strtolower( $match[1] )][] = $match[2];
 510                                 $lastname = strtolower( $match[1] );
 511                         }
 512                 }
 513
 514                 $this->parseCookies();
 515
 516         }
 517
 518         /**
 519          * Sets HTTPRequest status member to a fatal value with the error
 520          * message if the returned integer value of the status code was
 521          * not successful (< 300) or a redirect (>=300 and < 400).  (see
 522          * RFC2616, section 10,
 523          * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a
 524          * list of status codes.)
 525          */
 526         protected function setStatus() {
 527                 if ( !$this->respHeaders ) {
 528                         $this->parseHeader();
 529                 }
 530
 531                 if ( (int)$this->respStatus > 399 ) {
 532                         list( $code, $message ) = explode( " ", $this->respStatus, 2 );
 533                         $this->status->fatal( "http-bad-status", $code, $message );
 534                 }
 535         }
 536
 537         /**
 538          * Get the integer value of the HTTP status code (e.g. 200 for "200 Ok")
 539          * (see RFC2616, section 10, http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
 540          * for a list of status codes.)
 541          *
 542          * @return int
 543          */
 544         public function getStatus() {
 545                 if ( !$this->respHeaders ) {
 546                         $this->parseHeader();
 547                 }
 548
 549                 return (int)$this->respStatus;
 550         }
 551
 552         /**
 553          * Returns true if the last status code was a redirect.
 554          *
 555          * @return bool
 556          */
 557         public function isRedirect() {
 558                 if ( !$this->respHeaders ) {
 559                         $this->parseHeader();
 560                 }
 561
 562                 $status = (int)$this->respStatus;
 563
 564                 if ( $status >= 300 && $status <= 303 ) {
 565                         return true;
 566                 }
 567
 568                 return false;
 569         }
 570
 571         /**
 572          * Returns an associative array of response headers after the
 573          * request has been executed.  Because some headers
 574          * (e.g. Set-Cookie) can appear more than once the, each value of
 575          * the associative array is an array of the values given.
 576          *
 577          * @return array
 578          */
 579         public function getResponseHeaders() {
 580                 if ( !$this->respHeaders ) {
 581                         $this->parseHeader();
 582                 }
 583
 584                 return $this->respHeaders;
 585         }
 586
 587         /**
 588          * Returns the value of the given response header.
 589          *
 590          * @param string $header
 591          * @return string
 592          */
 593         public function getResponseHeader( $header ) {
 594                 if ( !$this->respHeaders ) {
 595                         $this->parseHeader();
 596                 }
 597
 598                 if ( isset( $this->respHeaders[strtolower( $header )] ) ) {
 599                         $v = $this->respHeaders[strtolower( $header )];
 600                         return $v[count( $v ) - 1];
 601                 }
 602
 603                 return null;
 604         }
 605
 606         /**
 607          * Tells the MWHttpRequest object to use this pre-loaded CookieJar.
 608          *
 609          * @param CookieJar $jar
 610          */
 611         public function setCookieJar( $jar ) {
 612                 $this->cookieJar = $jar;
 613         }
 614
 615         /**
 616          * Returns the cookie jar in use.
 617          *
 618          * @return CookieJar
 619          */
 620         public function getCookieJar() {
 621                 if ( !$this->respHeaders ) {
 622                         $this->parseHeader();
 623                 }
 624
 625                 return $this->cookieJar;
 626         }
 627
 628         /**
 629          * Sets a cookie. Used before a request to set up any individual
 630          * cookies. Used internally after a request to parse the
 631          * Set-Cookie headers.
 632          * @see Cookie::set
 633          * @param string $name
 634          * @param mixed $value
 635          * @param array $attr
 636          */
 637         public function setCookie( $name, $value = null, $attr = null ) {
 638                 if ( !$this->cookieJar ) {
 639                         $this->cookieJar = new CookieJar;
 640                 }
 641
 642                 $this->cookieJar->setCookie( $name, $value, $attr );
 643         }
 644
 645         /**
 646          * Parse the cookies in the response headers and store them in the cookie jar.
 647          */
 648         protected function parseCookies() {
 649
 650                 if ( !$this->cookieJar ) {
 651                         $this->cookieJar = new CookieJar;
 652                 }
 653
 654                 if ( isset( $this->respHeaders['set-cookie'] ) ) {
 655                         $url = parse_url( $this->getFinalUrl() );
 656                         foreach ( $this->respHeaders['set-cookie'] as $cookie ) {
 657                                 $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
 658                         }
 659                 }
 660
 661         }
 662
 663         /**
 664          * Returns the final URL after all redirections.
 665          *
 666          * Relative values of the "Location" header are incorrect as
 667          * stated in RFC, however they do happen and modern browsers
 668          * support them.  This function loops backwards through all
 669          * locations in order to build the proper absolute URI - Marooned
 670          * at wikia-inc.com
 671          *
 672          * Note that the multiple Location: headers are an artifact of
 673          * CURL -- they shouldn't actually get returned this way. Rewrite
 674          * this when bug 29232 is taken care of (high-level redirect
 675          * handling rewrite).
 676          *
 677          * @return string
 678          */
 679         public function getFinalUrl() {
 680                 $headers = $this->getResponseHeaders();
 681
 682                 // return full url (fix for incorrect but handled relative location)
 683                 if ( isset( $headers['location'] ) ) {
 684                         $locations = $headers['location'];
 685                         $domain = '';
 686                         $foundRelativeURI = false;
 687                         $countLocations = count( $locations );
 688
 689                         for ( $i = $countLocations - 1; $i >= 0; $i-- ) {
 690                                 $url = parse_url( $locations[$i] );
 691
 692                                 if ( isset( $url['host'] ) ) {
 693                                         $domain = $url['scheme'] . '://' . $url['host'];
 694                                         break; // found correct URI (with host)
 695                                 } else {
 696                                         $foundRelativeURI = true;
 697                                 }
 698                         }
 699
 700                         if ( $foundRelativeURI ) {
 701                                 if ( $domain ) {
 702                                         return $domain . $locations[$countLocations - 1];
 703                                 } else {
 704                                         $url = parse_url( $this->url );
 705                                         if ( isset( $url['host'] ) ) {
 706                                                 return $url['scheme'] . '://' . $url['host'] .
 707                                                         $locations[$countLocations - 1];
 708                                         }
 709                                 }
 710                         } else {
 711                                 return $locations[$countLocations - 1];
 712                         }
 713                 }
 714
 715                 return $this->url;
 716         }
 717
 718         /**
 719          * Returns true if the backend can follow redirects. Overridden by the
 720          * child classes.
 721          * @return bool
 722          */
 723         public function canFollowRedirects() {
 724                 return true;
 725         }
 726 }
 727
 728 /**
 729  * MWHttpRequest implemented using internal curl compiled into PHP
 730  */
 731 class CurlHttpRequest extends MWHttpRequest {
 732         const SUPPORTS_FILE_POSTS = true;
 733
 734         protected $curlOptions = [];
 735         protected $headerText = "";
 736
 737         /**
 738          * @param resource $fh
 739          * @param string $content
 740          * @return int
 741          */
 742         protected function readHeader( $fh, $content ) {
 743                 $this->headerText .= $content;
 744                 return strlen( $content );
 745         }
 746
 747         public function execute() {
 748
 749                 parent::execute();
 750
 751                 if ( !$this->status->isOK() ) {
 752                         return $this->status;
 753                 }
 754
 755                 $this->curlOptions[CURLOPT_PROXY] = $this->proxy;
 756                 $this->curlOptions[CURLOPT_TIMEOUT] = $this->timeout;
 757
 758                 // Only supported in curl >= 7.16.2
 759                 if ( defined( 'CURLOPT_CONNECTTIMEOUT_MS' ) ) {
 760                         $this->curlOptions[CURLOPT_CONNECTTIMEOUT_MS] = $this->connectTimeout * 1000;
 761                 }
 762
 763                 $this->curlOptions[CURLOPT_HTTP_VERSION] = CURL_HTTP_VERSION_1_0;
 764                 $this->curlOptions[CURLOPT_WRITEFUNCTION] = $this->callback;
 765                 $this->curlOptions[CURLOPT_HEADERFUNCTION] = [ $this, "readHeader" ];
 766                 $this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
 767                 $this->curlOptions[CURLOPT_ENCODING] = ""; # Enable compression
 768
 769                 $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
 770
 771                 $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost ? 2 : 0;
 772                 $this->curlOptions[CURLOPT_SSL_VERIFYPEER] = $this->sslVerifyCert;
 773
 774                 if ( $this->caInfo ) {
 775                         $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
 776                 }
 777
 778                 if ( $this->headersOnly ) {
 779                         $this->curlOptions[CURLOPT_NOBODY] = true;
 780                         $this->curlOptions[CURLOPT_HEADER] = true;
 781                 } elseif ( $this->method == 'POST' ) {
 782                         $this->curlOptions[CURLOPT_POST] = true;
 783                         $postData = $this->postData;
 784                         // Don't interpret POST parameters starting with '@' as file uploads, because this
 785                         // makes it impossible to POST plain values starting with '@' (and causes security
 786                         // issues potentially exposing the contents of local files).
 787                         // The PHP manual says this option was introduced in PHP 5.5 defaults to true in PHP 5.6,
 788                         // but we support lower versions, and the option doesn't exist in HHVM 5.6.99.
 789                         if ( defined( 'CURLOPT_SAFE_UPLOAD' ) ) {
 790                                 $this->curlOptions[CURLOPT_SAFE_UPLOAD] = true;
 791                         } elseif ( is_array( $postData ) ) {
 792                                 // In PHP 5.2 and later, '@' is interpreted as a file upload if POSTFIELDS
 793                                 // is an array, but not if it's a string. So convert $req['body'] to a string
 794                                 // for safety.
 795                                 $postData = wfArrayToCgi( $postData );
 796                         }
 797                         $this->curlOptions[CURLOPT_POSTFIELDS] = $postData;
 798
 799                         // Suppress 'Expect: 100-continue' header, as some servers
 800                         // will reject it with a 417 and Curl won't auto retry
 801                         // with HTTP 1.0 fallback
 802                         $this->reqHeaders['Expect'] = '';
 803                 } else {
 804                         $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
 805                 }
 806
 807                 $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
 808
 809                 $curlHandle = curl_init( $this->url );
 810
 811                 if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
 812                         throw new MWException( "Error setting curl options." );
 813                 }
 814
 815                 if ( $this->followRedirects && $this->canFollowRedirects() ) {
 816                         MediaWiki\suppressWarnings();
 817                         if ( !curl_setopt( $curlHandle, CURLOPT_FOLLOWLOCATION, true ) ) {
 818                                 wfDebug( __METHOD__ . ": Couldn't set CURLOPT_FOLLOWLOCATION. " .
 819                                         "Probably open_basedir is set.\n" );
 820                                 // Continue the processing. If it were in curl_setopt_array,
 821                                 // processing would have halted on its entry
 822                         }
 823                         MediaWiki\restoreWarnings();
 824                 }
 825
 826                 if ( $this->profiler ) {
 827                         $profileSection = $this->profiler->scopedProfileIn(
 828                                 __METHOD__ . '-' . $this->profileName
 829                         );
 830                 }
 831
 832                 $curlRes = curl_exec( $curlHandle );
 833                 if ( curl_errno( $curlHandle ) == CURLE_OPERATION_TIMEOUTED ) {
 834                         $this->status->fatal( 'http-timed-out', $this->url );
 835                 } elseif ( $curlRes === false ) {
 836                         $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
 837                 } else {
 838                         $this->headerList = explode( "\r\n", $this->headerText );
 839                 }
 840
 841                 curl_close( $curlHandle );
 842
 843                 if ( $this->profiler ) {
 844                         $this->profiler->scopedProfileOut( $profileSection );
 845                 }
 846
 847                 $this->parseHeader();
 848                 $this->setStatus();
 849
 850                 return $this->status;
 851         }
 852
 853         /**
 854          * @return bool
 855          */
 856         public function canFollowRedirects() {
 857                 $curlVersionInfo = curl_version();
 858                 if ( $curlVersionInfo['version_number'] < 0x071304 ) {
 859                         wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
 860                         return false;
 861                 }
 862
 863                 if ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 864                         if ( strval( ini_get( 'open_basedir' ) ) !== '' ) {
 865                                 wfDebug( "Cannot follow redirects when open_basedir is set\n" );
 866                                 return false;
 867                         }
 868                 }
 869
 870                 return true;
 871         }
 872 }
 873
 874 class PhpHttpRequest extends MWHttpRequest {
 875
 876         private $fopenErrors = [];
 877
 878         /**
 879          * @param string $url
 880          * @return string
 881          */
 882         protected function urlToTcp( $url ) {
 883                 $parsedUrl = parse_url( $url );
 884
 885                 return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
 886         }
 887
 888         /**
 889          * Returns an array with a 'capath' or 'cafile' key
 890          * that is suitable to be merged into the 'ssl' sub-array of
 891          * a stream context options array.
 892          * Uses the 'caInfo' option of the class if it is provided, otherwise uses the system
 893          * default CA bundle if PHP supports that, or searches a few standard locations.
 894          * @return array
 895          * @throws DomainException
 896          */
 897         protected function getCertOptions() {
 898                 $certOptions = [];
 899                 $certLocations = [];
 900                 if ( $this->caInfo ) {
 901                         $certLocations = [ 'manual' => $this->caInfo ];
 902                 } elseif ( version_compare( PHP_VERSION, '5.6.0', '<' ) ) {
 903                         // @codingStandardsIgnoreStart Generic.Files.LineLength
 904                         // Default locations, based on
 905                         // https://www.happyassassin.net/2015/01/12/a-note-about-ssltls-trusted-certificate-stores-and-platforms/
 906                         // PHP 5.5 and older doesn't have any defaults, so we try to guess ourselves.
 907                         // PHP 5.6+ gets the CA location from OpenSSL as long as it is not set manually,
 908                         // so we should leave capath/cafile empty there.
 909                         // @codingStandardsIgnoreEnd
 910                         $certLocations = array_filter( [
 911                                 getenv( 'SSL_CERT_DIR' ),
 912                                 getenv( 'SSL_CERT_PATH' ),
 913                                 '/etc/pki/tls/certs/ca-bundle.crt', # Fedora et al
 914                                 '/etc/ssl/certs',  # Debian et al
 915                                 '/etc/pki/tls/certs/ca-bundle.trust.crt',
 916                                 '/etc/pki/ca-trust/extracted/pem/tls-ca-bundle.pem',
 917                                 '/System/Library/OpenSSL', # OSX
 918                         ] );
 919                 }
 920
 921                 foreach ( $certLocations as $key => $cert ) {
 922                         if ( is_dir( $cert ) ) {
 923                                 $certOptions['capath'] = $cert;
 924                                 break;
 925                         } elseif ( is_file( $cert ) ) {
 926                                 $certOptions['cafile'] = $cert;
 927                                 break;
 928                         } elseif ( $key === 'manual' ) {
 929                                 // fail more loudly if a cert path was manually configured and it is not valid
 930                                 throw new DomainException( "Invalid CA info passed: $cert" );
 931                         }
 932                 }
 933
 934                 return $certOptions;
 935         }
 936
 937         /**
 938          * Custom error handler for dealing with fopen() errors.
 939          * fopen() tends to fire multiple errors in succession, and the last one
 940          * is completely useless (something like "fopen: failed to open stream")
 941          * so normal methods of handling errors programmatically
 942          * like get_last_error() don't work.
 943          */
 944         public function errorHandler( $errno, $errstr ) {
 945                 $n = count( $this->fopenErrors ) + 1;
 946                 $this->fopenErrors += [ "errno$n" => $errno, "errstr$n" => $errstr ];
 947         }
 948
 949         public function execute() {
 950
 951                 parent::execute();
 952
 953                 if ( is_array( $this->postData ) ) {
 954                         $this->postData = wfArrayToCgi( $this->postData );
 955                 }
 956
 957                 if ( $this->parsedUrl['scheme'] != 'http'
 958                         && $this->parsedUrl['scheme'] != 'https' ) {
 959                         $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
 960                 }
 961
 962                 $this->reqHeaders['Accept'] = "*/*";
 963                 $this->reqHeaders['Connection'] = 'Close';
 964                 if ( $this->method == 'POST' ) {
 965                         // Required for HTTP 1.0 POSTs
 966                         $this->reqHeaders['Content-Length'] = strlen( $this->postData );
 967                         if ( !isset( $this->reqHeaders['Content-Type'] ) ) {
 968                                 $this->reqHeaders['Content-Type'] = "application/x-www-form-urlencoded";
 969                         }
 970                 }
 971
 972                 // Set up PHP stream context
 973                 $options = [
 974                         'http' => [
 975                                 'method' => $this->method,
 976                                 'header' => implode( "\r\n", $this->getHeaderList() ),
 977                                 'protocol_version' => '1.1',
 978                                 'max_redirects' => $this->followRedirects ? $this->maxRedirects : 0,
 979                                 'ignore_errors' => true,
 980                                 'timeout' => $this->timeout,
 981                                 // Curl options in case curlwrappers are installed
 982                                 'curl_verify_ssl_host' => $this->sslVerifyHost ? 2 : 0,
 983                                 'curl_verify_ssl_peer' => $this->sslVerifyCert,
 984                         ],
 985                         'ssl' => [
 986                                 'verify_peer' => $this->sslVerifyCert,
 987                                 'SNI_enabled' => true,
 988                                 'ciphers' => 'HIGH:!SSLv2:!SSLv3:-ADH:-kDH:-kECDH:-DSS',
 989                                 'disable_compression' => true,
 990                         ],
 991                 ];
 992
 993                 if ( $this->proxy ) {
 994                         $options['http']['proxy'] = $this->urlToTCP( $this->proxy );
 995                         $options['http']['request_fulluri'] = true;
 996                 }
 997
 998                 if ( $this->postData ) {
 999                         $options['http']['content'] = $this->postData;
1000                 }
1001
1002                 if ( $this->sslVerifyHost ) {
1003                         // PHP 5.6.0 deprecates CN_match, in favour of peer_name which
1004                         // actually checks SubjectAltName properly.
1005                         if ( version_compare( PHP_VERSION, '5.6.0', '>=' ) ) {
1006                                 $options['ssl']['peer_name'] = $this->parsedUrl['host'];
1007                         } else {
1008                                 $options['ssl']['CN_match'] = $this->parsedUrl['host'];
1009                         }
1010                 }
1011
1012                 $options['ssl'] += $this->getCertOptions();
1013
1014                 $context = stream_context_create( $options );
1015
1016                 $this->headerList = [];
1017                 $reqCount = 0;
1018                 $url = $this->url;
1019
1020                 $result = [];
1021
1022                 if ( $this->profiler ) {
1023                         $profileSection = $this->profiler->scopedProfileIn(
1024                                 __METHOD__ . '-' . $this->profileName
1025                         );
1026                 }
1027                 do {
1028                         $reqCount++;
1029                         $this->fopenErrors = [];
1030                         set_error_handler( [ $this, 'errorHandler' ] );
1031                         $fh = fopen( $url, "r", false, $context );
1032                         restore_error_handler();
1033
1034                         if ( !$fh ) {
1035                                 // HACK for instant commons.
1036                                 // If we are contacting (commons|upload).wikimedia.org
1037                                 // try again with CN_match for en.wikipedia.org
1038                                 // as php does not handle SubjectAltName properly
1039                                 // prior to "peer_name" option in php 5.6
1040                                 if ( isset( $options['ssl']['CN_match'] )
1041                                         && ( $options['ssl']['CN_match'] === 'commons.wikimedia.org'
1042                                                 || $options['ssl']['CN_match'] === 'upload.wikimedia.org' )
1043                                 ) {
1044                                         $options['ssl']['CN_match'] = 'en.wikipedia.org';
1045                                         $context = stream_context_create( $options );
1046                                         continue;
1047                                 }
1048                                 break;
1049                         }
1050
1051                         $result = stream_get_meta_data( $fh );
1052                         $this->headerList = $result['wrapper_data'];
1053                         $this->parseHeader();
1054
1055                         if ( !$this->followRedirects ) {
1056                                 break;
1057                         }
1058
1059                         # Handle manual redirection
1060                         if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
1061                                 break;
1062                         }
1063                         # Check security of URL
1064                         $url = $this->getResponseHeader( "Location" );
1065
1066                         if ( !Http::isValidURI( $url ) ) {
1067                                 wfDebug( __METHOD__ . ": insecure redirection\n" );
1068                                 break;
1069                         }
1070                 } while ( true );
1071                 if ( $this->profiler ) {
1072                         $this->profiler->scopedProfileOut( $profileSection );
1073                 }
1074
1075                 $this->setStatus();
1076
1077                 if ( $fh === false ) {
1078                         if ( $this->fopenErrors ) {
1079                                 LoggerFactory::getInstance( 'http' )->warning( __CLASS__
1080                                         . ': error opening connection: {errstr1}', $this->fopenErrors );
1081                         }
1082                         $this->status->fatal( 'http-request-error' );
1083                         return $this->status;
1084                 }
1085
1086                 if ( $result['timed_out'] ) {
1087                         $this->status->fatal( 'http-timed-out', $this->url );
1088                         return $this->status;
1089                 }
1090
1091                 // If everything went OK, or we received some error code
1092                 // get the response body content.
1093                 if ( $this->status->isOK() || (int)$this->respStatus >= 300 ) {
1094                         while ( !feof( $fh ) ) {
1095                                 $buf = fread( $fh, 8192 );
1096
1097                                 if ( $buf === false ) {
1098                                         $this->status->fatal( 'http-read-error' );
1099                                         break;
1100                                 }
1101
1102                                 if ( strlen( $buf ) ) {
1103                                         call_user_func( $this->callback, $fh, $buf );
1104                                 }
1105                         }
1106                 }
1107                 fclose( $fh );
1108
1109                 return $this->status;
1110         }
1111 }