UploadFromUrl.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Upload;
 
use MediaWiki\Context\RequestContext;
use MediaWiki\HookContainer\HookRunner;
use MediaWiki\Http\MWHttpRequest;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use MediaWiki\Permissions\Authority;
use MediaWiki\Request\WebRequest;
use MediaWiki\Status\Status;
 
class UploadFromUrl extends UploadBase {
    protected $mUrl;
 
    protected $mTmpHandle;
 
    protected static $allowedUrls = [];
 
    public static function isAllowed( Authority $performer ) {
        if ( !$performer->isAllowed( 'upload_by_url' ) ) {
            return 'upload_by_url';
        }
 
        return parent::isAllowed( $performer );
    }
 
    public static function isEnabled() {
        $allowCopyUploads = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::AllowCopyUploads );
 
        return $allowCopyUploads && parent::isEnabled();
    }
 
    public static function isAllowedHost( $url ) {
        $domains = self::getAllowedHosts();
        if ( !count( $domains ) ) {
            return true;
        }
        $parsedUrl = wfGetUrlUtils()->parse( $url );
        if ( !$parsedUrl ) {
            return false;
        }
        $valid = false;
        foreach ( $domains as $domain ) {
            // See if the domain for the upload matches this allowed domain
            $domainPieces = explode( '.', $domain );
            $uploadDomainPieces = explode( '.', $parsedUrl['host'] );
            if ( count( $domainPieces ) === count( $uploadDomainPieces ) ) {
                $valid = true;
                // See if all the pieces match or not (excluding wildcards)
                foreach ( $domainPieces as $index => $piece ) {
                    if ( $piece !== '*' && $piece !== $uploadDomainPieces[$index] ) {
                        $valid = false;
                    }
                }
                if ( $valid ) {
                    // We found a match, so quit comparing against the list
                    break;
                }
            }
            /* Non-wildcard test
            if ( $parsedUrl['host'] === $domain ) {
                $valid = true;
                break;
            }
            */
        }
 
        return $valid;
    }
 
    public static function getCacheKey( $params ) {
        if ( !isset( $params['filename'] ) || !isset( $params['url'] ) ) {
            return "";
        } else {
            // We use sha1 here to ensure we have a fixed-length string of printable
            // characters. There is no cryptography involved, so we just need a
            // relatively fast function.
            return sha1( sprintf( "%s|||%s", $params['filename'], $params['url'] ) );
        }
    }
 
    public static function getCacheKeyFromRequest( &$request ) {
        $uploadCacheKey = $request->getText( 'wpCacheKey', $request->getText( 'key', '' ) );
        if ( $uploadCacheKey !== '' ) {
            return $uploadCacheKey;
        }
        $desiredDestName = $request->getText( 'wpDestFile' );
        if ( !$desiredDestName ) {
            $desiredDestName = $request->getText( 'wpUploadFileURL' );
        }
        return self::getCacheKey(
            [
                'filename' => $desiredDestName,
                'url' => trim( $request->getVal( 'wpUploadFileURL' ) )
            ]
        );
    }
 
    public static function getAllowedHosts(): array {
        $config = MediaWikiServices::getInstance()->getMainConfig();
        $domains = $config->get( MainConfigNames::CopyUploadsDomains );
 
        if ( $config->get( MainConfigNames::CopyUploadAllowOnWikiDomainConfig ) ) {
            $page = wfMessage( 'copyupload-allowed-domains' )->inContentLanguage()->plain();
 
            foreach ( explode( "\n", $page ) as $line ) {
                // Strip comments
                $line = preg_replace( "/^\\s*([^#]*)\\s*((.*)?)$/", "\\1", $line );
                // Trim whitespace
                $line = trim( $line );
 
                if ( $line !== '' ) {
                    $domains[] = $line;
                }
            }
        }
 
        return $domains;
    }
 
    public static function isAllowedUrl( $url ) {
        if ( !isset( self::$allowedUrls[$url] ) ) {
            $allowed = true;
            ( new HookRunner( MediaWikiServices::getInstance()->getHookContainer() ) )
                ->onIsUploadAllowedFromUrl( $url, $allowed );
            self::$allowedUrls[$url] = $allowed;
        }
 
        return self::$allowedUrls[$url];
    }
 
    public function getUrl() {
        return $this->mUrl;
    }
 
    public function initialize( $name, $url, $initTempFile = true ) {
        $this->mUrl = $url;
 
        $tempPath = $initTempFile ? $this->makeTemporaryFile() : null;
        $fileSize = $initTempFile ? 0 : null;
        # File size and removeTempFile will be filled in later
        $this->initializePathInfo( $name, $tempPath, $fileSize, false );
    }
 
    public function initializeFromRequest( &$request ) {
        $desiredDestName = $request->getText( 'wpDestFile' );
        if ( !$desiredDestName ) {
            $desiredDestName = $request->getText( 'wpUploadFileURL' );
        }
        $this->initialize(
            $desiredDestName,
            trim( $request->getVal( 'wpUploadFileURL' ) )
        );
    }
 
    public static function isValidRequest( $request ) {
        $user = RequestContext::getMain()->getUser();
 
        $url = $request->getVal( 'wpUploadFileURL' );
 
        return $url
            && MediaWikiServices::getInstance()
                ->getPermissionManager()
                ->userHasRight( $user, 'upload_by_url' );
    }
 
    public function getSourceType() {
        return 'url';
    }
 
    public function fetchFile( $httpOptions = [] ) {
        $status = $this->canFetchFile();
        if ( !$status->isGood() ) {
            return $status;
        }
        return $this->reallyFetchFile( $httpOptions );
    }
 
    public function canFetchFile() {
        if ( !MWHttpRequest::isValidURI( $this->mUrl ) ) {
            return Status::newFatal( 'http-invalid-url', $this->mUrl );
        }
 
        if ( !self::isAllowedHost( $this->mUrl ) ) {
            return Status::newFatal( 'upload-copy-upload-invalid-domain' );
        }
        if ( !self::isAllowedUrl( $this->mUrl ) ) {
            return Status::newFatal( 'upload-copy-upload-invalid-url' );
        }
        return Status::newGood();
    }
 
    protected function makeTemporaryFile() {
        $tmpFile = MediaWikiServices::getInstance()->getTempFSFileFactory()
            ->newTempFSFile( 'URL', 'urlupload_' );
        $tmpFile->bind( $this );
 
        return $tmpFile->getPath();
    }
 
    public function saveTempFileChunk( $req, $buffer ) {
        wfDebugLog( 'fileupload', 'Received chunk of ' . strlen( $buffer ) . ' bytes' );
        $nbytes = fwrite( $this->mTmpHandle, $buffer );
 
        if ( $nbytes == strlen( $buffer ) ) {
            $this->mFileSize += $nbytes;
        } else {
            // Well... that's not good!
            wfDebugLog(
                'fileupload',
                'Short write ' . $nbytes . '/' . strlen( $buffer ) .
                ' bytes, aborting with ' . $this->mFileSize . ' uploaded so far'
            );
            fclose( $this->mTmpHandle );
            $this->mTmpHandle = false;
        }
 
        return $nbytes;
    }
 
    protected function reallyFetchFile( $httpOptions = [] ) {
        $copyUploadProxy = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::CopyUploadProxy );
        $copyUploadTimeout = MediaWikiServices::getInstance()->getMainConfig()
            ->get( MainConfigNames::CopyUploadTimeout );
 
        // Note the temporary file should already be created by makeTemporaryFile()
        $this->mTmpHandle = fopen( $this->mTempPath, 'wb' );
        if ( !$this->mTmpHandle ) {
            return Status::newFatal( 'tmp-create-error' );
        }
        wfDebugLog( 'fileupload', 'Temporary file created "' . $this->mTempPath . '"' );
 
        $this->mRemoveTempFile = true;
        $this->mFileSize = 0;
 
        $options = $httpOptions + [ 'followRedirects' => false ];
 
        if ( $copyUploadProxy !== false ) {
            $options['proxy'] = $copyUploadProxy;
        }
 
        if ( $copyUploadTimeout && !isset( $options['timeout'] ) ) {
            $options['timeout'] = $copyUploadTimeout;
        }
        wfDebugLog(
            'fileupload',
            'Starting download from "' . $this->mUrl . '" ' .
            '<' . implode( ',', array_keys( array_filter( $options ) ) ) . '>'
        );
 
        // Manually follow any redirects up to the limit and reset the output file before each new request to prevent
        // capturing the redirect response as part of the file.
        $attemptsLeft = $options['maxRedirects'] ?? 5;
        $targetUrl = $this->mUrl;
        $requestFactory = MediaWikiServices::getInstance()->getHttpRequestFactory();
        while ( $attemptsLeft > 0 ) {
            $req = $requestFactory->create( $targetUrl, $options, __METHOD__ );
            $req->setCallback( $this->saveTempFileChunk( ... ) );
            $status = $req->execute();
            if ( !$req->isRedirect() ) {
                break;
            }
            $targetUrl = $req->getFinalUrl();
            // Remove redirect response content from file.
            ftruncate( $this->mTmpHandle, 0 );
            rewind( $this->mTmpHandle );
            $attemptsLeft--;
        }
 
        if ( $attemptsLeft == 0 ) {
            return Status::newFatal( 'upload-too-many-redirects' );
        }
 
        if ( $this->mTmpHandle ) {
            // File got written ok...
            fclose( $this->mTmpHandle );
            $this->mTmpHandle = null;
        } else {
            // We encountered a write error during the download...
            return Status::newFatal( 'tmp-write-error' );
        }
 
        // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop
        if ( $status->isOK() ) {
            wfDebugLog( 'fileupload', 'Download by URL completed successfully.' );
        } else {
            // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop
            wfDebugLog( 'fileupload', $status->getWikiText( false, false, 'en' ) );
            wfDebugLog(
                'fileupload',
                // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop
                'Download by URL completed with HTTP status ' . $req->getStatus()
            );
        }
 
        // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop
        return $status;
    }
}
 
class_alias( UploadFromUrl::class, 'UploadFromUrl' );