MWHttpRequest.php

Go to the documentation of this file.

<?php
use MediaWiki\Logger\LoggerFactory;
use Psr\Log\LoggerInterface;
use Psr\Log\LoggerAwareInterface;
use Psr\Log\NullLogger;

class MWHttpRequest implements LoggerAwareInterface {
    const SUPPORTS_FILE_POSTS = false;

    protected $content;
    protected $timeout = 'default';
    protected $headersOnly = null;
    protected $postData = null;
    protected $proxy = null;
    protected $noProxy = false;
    protected $sslVerifyHost = true;
    protected $sslVerifyCert = true;
    protected $caInfo = null;
    protected $method = "GET";
    protected $reqHeaders = [];
    protected $url;
    protected $parsedUrl;
    protected $callback;
    protected $maxRedirects = 5;
    protected $followRedirects = false;

    protected $cookieJar;

    protected $headerList = [];
    protected $respVersion = "0.9";
    protected $respStatus = "200 Ok";
    protected $respHeaders = [];

    public $status;

    protected $profiler;

    protected $profileName;

    protected $logger;

    protected function __construct(
        $url, $options = [], $caller = __METHOD__, $profiler = null
    ) {
        global $wgHTTPTimeout, $wgHTTPConnectTimeout;

        $this->url = wfExpandUrl( $url, PROTO_HTTP );
        $this->parsedUrl = wfParseUrl( $this->url );

        if ( isset( $options['logger'] ) ) {
            $this->logger = $options['logger'];
        } else {
            $this->logger = new NullLogger();
        }

        if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) {
            $this->status = Status::newFatal( 'http-invalid-url', $url );
        } else {
            $this->status = Status::newGood( 100 ); // continue
        }

        if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) {
            $this->timeout = $options['timeout'];
        } else {
            $this->timeout = $wgHTTPTimeout;
        }
        if ( isset( $options['connectTimeout'] ) && $options['connectTimeout'] != 'default' ) {
            $this->connectTimeout = $options['connectTimeout'];
        } else {
            $this->connectTimeout = $wgHTTPConnectTimeout;
        }
        if ( isset( $options['userAgent'] ) ) {
            $this->setUserAgent( $options['userAgent'] );
        }

        $members = [ "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
                "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" ];

        foreach ( $members as $o ) {
            if ( isset( $options[$o] ) ) {
                // ensure that MWHttpRequest::method is always
                // uppercased. Bug 36137
                if ( $o == 'method' ) {
                    $options[$o] = strtoupper( $options[$o] );
                }
                $this->$o = $options[$o];
            }
        }

        if ( $this->noProxy ) {
            $this->proxy = ''; // noProxy takes precedence
        }

        // Profile based on what's calling us
        $this->profiler = $profiler;
        $this->profileName = $caller;
    }

    public function setLogger( LoggerInterface $logger ) {
        $this->logger = $logger;
    }

    public static function canMakeRequests() {
        return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
    }

    public static function factory( $url, $options = null, $caller = __METHOD__ ) {
        if ( !Http::$httpEngine ) {
            Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
        } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
            throw new MWException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' .
                ' Http::$httpEngine is set to "curl"' );
        }

        if ( !is_array( $options ) ) {
            $options = [];
        }

        if ( !isset( $options['logger'] ) ) {
            $options['logger'] = LoggerFactory::getInstance( 'http' );
        }

        switch ( Http::$httpEngine ) {
            case 'curl':
                return new CurlHttpRequest( $url, $options, $caller, Profiler::instance() );
            case 'php':
                if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
                    throw new MWException( __METHOD__ . ': allow_url_fopen ' .
                        'needs to be enabled for pure PHP http requests to ' .
                        'work. If possible, curl should be used instead. See ' .
                        'http://php.net/curl.'
                    );
                }
                return new PhpHttpRequest( $url, $options, $caller, Profiler::instance() );
            default:
                throw new MWException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' );
        }
    }

    public function getContent() {
        return $this->content;
    }

    public function setData( $args ) {
        $this->postData = $args;
    }

    public function proxySetup() {
        // If there is an explicit proxy set and proxies are not disabled, then use it
        if ( $this->proxy && !$this->noProxy ) {
            return;
        }

        // Otherwise, fallback to $wgHTTPProxy if this is not a machine
        // local URL and proxies are not disabled
        if ( self::isLocalURL( $this->url ) || $this->noProxy ) {
            $this->proxy = '';
        } else {
            $this->proxy = Http::getProxy();
        }
    }

    private static function isLocalURL( $url ) {
        global $wgCommandLineMode, $wgLocalVirtualHosts;

        if ( $wgCommandLineMode ) {
            return false;
        }

        // Extract host part
        $matches = [];
        if ( preg_match( '!^https?://([\w.-]+)[/:].*$!', $url, $matches ) ) {
            $host = $matches[1];
            // Split up dotwise
            $domainParts = explode( '.', $host );
            // Check if this domain or any superdomain is listed as a local virtual host
            $domainParts = array_reverse( $domainParts );

            $domain = '';
            $countParts = count( $domainParts );
            for ( $i = 0; $i < $countParts; $i++ ) {
                $domainPart = $domainParts[$i];
                if ( $i == 0 ) {
                    $domain = $domainPart;
                } else {
                    $domain = $domainPart . '.' . $domain;
                }

                if ( in_array( $domain, $wgLocalVirtualHosts ) ) {
                    return true;
                }
            }
        }

        return false;
    }

    public function setUserAgent( $UA ) {
        $this->setHeader( 'User-Agent', $UA );
    }

    public function setHeader( $name, $value ) {
        // I feel like I should normalize the case here...
        $this->reqHeaders[$name] = $value;
    }

    public function getHeaderList() {
        $list = [];

        if ( $this->cookieJar ) {
            $this->reqHeaders['Cookie'] =
                $this->cookieJar->serializeToHttpRequest(
                    $this->parsedUrl['path'],
                    $this->parsedUrl['host']
                );
        }

        foreach ( $this->reqHeaders as $name => $value ) {
            $list[] = "$name: $value";
        }

        return $list;
    }

    public function setCallback( $callback ) {
        if ( !is_callable( $callback ) ) {
            throw new MWException( 'Invalid MwHttpRequest callback' );
        }
        $this->callback = $callback;
    }

    public function read( $fh, $content ) {
        $this->content .= $content;
        return strlen( $content );
    }

    public function execute() {

        $this->content = "";

        if ( strtoupper( $this->method ) == "HEAD" ) {
            $this->headersOnly = true;
        }

        $this->proxySetup(); // set up any proxy as needed

        if ( !$this->callback ) {
            $this->setCallback( [ $this, 'read' ] );
        }

        if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
            $this->setUserAgent( Http::userAgent() );
        }

    }

    protected function parseHeader() {

        $lastname = "";

        foreach ( $this->headerList as $header ) {
            if ( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
                $this->respVersion = $match[1];
                $this->respStatus = $match[2];
            } elseif ( preg_match( "#^[ \t]#", $header ) ) {
                $last = count( $this->respHeaders[$lastname] ) - 1;
                $this->respHeaders[$lastname][$last] .= "\r\n$header";
            } elseif ( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
                $this->respHeaders[strtolower( $match[1] )][] = $match[2];
                $lastname = strtolower( $match[1] );
            }
        }

        $this->parseCookies();

    }

    protected function setStatus() {
        if ( !$this->respHeaders ) {
            $this->parseHeader();
        }

        if ( (int)$this->respStatus > 399 ) {
            list( $code, $message ) = explode( " ", $this->respStatus, 2 );
            $this->status->fatal( "http-bad-status", $code, $message );
        }
    }

    public function getStatus() {
        if ( !$this->respHeaders ) {
            $this->parseHeader();
        }

        return (int)$this->respStatus;
    }

    public function isRedirect() {
        if ( !$this->respHeaders ) {
            $this->parseHeader();
        }

        $status = (int)$this->respStatus;

        if ( $status >= 300 && $status <= 303 ) {
            return true;
        }

        return false;
    }

    public function getResponseHeaders() {
        if ( !$this->respHeaders ) {
            $this->parseHeader();
        }

        return $this->respHeaders;
    }

    public function getResponseHeader( $header ) {
        if ( !$this->respHeaders ) {
            $this->parseHeader();
        }

        if ( isset( $this->respHeaders[strtolower( $header )] ) ) {
            $v = $this->respHeaders[strtolower( $header )];
            return $v[count( $v ) - 1];
        }

        return null;
    }

    public function setCookieJar( $jar ) {
        $this->cookieJar = $jar;
    }

    public function getCookieJar() {
        if ( !$this->respHeaders ) {
            $this->parseHeader();
        }

        return $this->cookieJar;
    }

    public function setCookie( $name, $value = null, $attr = null ) {
        if ( !$this->cookieJar ) {
            $this->cookieJar = new CookieJar;
        }

        $this->cookieJar->setCookie( $name, $value, $attr );
    }

    protected function parseCookies() {

        if ( !$this->cookieJar ) {
            $this->cookieJar = new CookieJar;
        }

        if ( isset( $this->respHeaders['set-cookie'] ) ) {
            $url = parse_url( $this->getFinalUrl() );
            foreach ( $this->respHeaders['set-cookie'] as $cookie ) {
                $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
            }
        }

    }

    public function getFinalUrl() {
        $headers = $this->getResponseHeaders();

        // return full url (fix for incorrect but handled relative location)
        if ( isset( $headers['location'] ) ) {
            $locations = $headers['location'];
            $domain = '';
            $foundRelativeURI = false;
            $countLocations = count( $locations );

            for ( $i = $countLocations - 1; $i >= 0; $i-- ) {
                $url = parse_url( $locations[$i] );

                if ( isset( $url['host'] ) ) {
                    $domain = $url['scheme'] . '://' . $url['host'];
                    break; // found correct URI (with host)
                } else {
                    $foundRelativeURI = true;
                }
            }

            if ( $foundRelativeURI ) {
                if ( $domain ) {
                    return $domain . $locations[$countLocations - 1];
                } else {
                    $url = parse_url( $this->url );
                    if ( isset( $url['host'] ) ) {
                        return $url['scheme'] . '://' . $url['host'] .
                            $locations[$countLocations - 1];
                    }
                }
            } else {
                return $locations[$countLocations - 1];
            }
        }

        return $this->url;
    }

    public function canFollowRedirects() {
        return true;
    }
}