From 9db190c7e736ec8d063187d4241b59feaf7dc2d1 Mon Sep 17 00:00:00 2001 From: Pierre Schmitz Date: Wed, 22 Jun 2011 11:28:20 +0200 Subject: update to MediaWiki 1.17.0 --- includes/upload/UploadBase.php | 372 ++++++++++++++++++++++----------- includes/upload/UploadFromFile.php | 61 ++++-- includes/upload/UploadFromStash.php | 17 +- includes/upload/UploadFromUrl.php | 221 ++++++++++++++------ includes/upload/UploadStash.php | 397 ++++++++++++++++++++++++++++++++++++ 5 files changed, 861 insertions(+), 207 deletions(-) create mode 100644 includes/upload/UploadStash.php (limited to 'includes/upload') diff --git a/includes/upload/UploadBase.php b/includes/upload/UploadBase.php index 5d955b36..546b9db8 100644 --- a/includes/upload/UploadBase.php +++ b/includes/upload/UploadBase.php @@ -25,14 +25,40 @@ abstract class UploadBase { const EMPTY_FILE = 3; const MIN_LENGTH_PARTNAME = 4; const ILLEGAL_FILENAME = 5; - const OVERWRITE_EXISTING_FILE = 7; + const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyPermissions() const FILETYPE_MISSING = 8; const FILETYPE_BADTYPE = 9; const VERIFICATION_ERROR = 10; + + # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR const UPLOAD_VERIFICATION_ERROR = 11; const HOOK_ABORTED = 11; + const FILE_TOO_LARGE = 12; const SESSION_VERSION = 2; + const SESSION_KEYNAME = 'wsUploadData'; + + static public function getSessionKeyname() { + return self::SESSION_KEYNAME; + } + + public function getVerificationErrorCode( $error ) { + $code_to_status = array(self::EMPTY_FILE => 'empty-file', + self::FILE_TOO_LARGE => 'file-too-large', + self::FILETYPE_MISSING => 'filetype-missing', + self::FILETYPE_BADTYPE => 'filetype-banned', + self::MIN_LENGTH_PARTNAME => 'filename-tooshort', + self::ILLEGAL_FILENAME => 'illegal-filename', + self::OVERWRITE_EXISTING_FILE => 'overwrite', + self::VERIFICATION_ERROR => 'verification-error', + self::HOOK_ABORTED => 'hookaborted', + ); + if( isset( $code_to_status[$error] ) ) { + return $code_to_status[$error]; + } + + return 'unknown-error'; + } /** * Returns true if uploads are enabled. @@ -57,8 +83,10 @@ abstract class UploadBase { * Can be overriden by subclasses. */ public static function isAllowed( $user ) { - if( !$user->isAllowed( 'upload' ) ) { - return 'upload'; + foreach ( array( 'upload', 'edit' ) as $permission ) { + if ( !$user->isAllowed( $permission ) ) { + return $permission; + } } return true; } @@ -143,15 +171,37 @@ abstract class UploadBase { } /** - * Return the file size + * Return true if the file is empty + * @return bool */ public function isEmptyFile() { return empty( $this->mFileSize ); } /** - * @param string $srcPath the source path - * @returns the real path if it was a virtual URL + * Return the file size + * @return integer + */ + public function getFileSize() { + return $this->mFileSize; + } + + /** + * Append a file to the Repo file + * + * @param $srcPath String: path to source file + * @param $toAppendPath String: path to the Repo file that will be appended to. + * @return Status Status + */ + protected function appendToUploadFile( $srcPath, $toAppendPath ) { + $repo = RepoGroup::singleton()->getLocalRepo(); + $status = $repo->append( $srcPath, $toAppendPath ); + return $status; + } + + /** + * @param $srcPath String: the source path + * @return the real path if it was a virtual URL */ function getRealPath( $srcPath ) { $repo = RepoGroup::singleton()->getLocalRepo(); @@ -163,7 +213,7 @@ abstract class UploadBase { /** * Verify whether the upload is sane. - * Returns self::OK or else an array with error information + * @return mixed self::OK or else an array with error information */ public function verifyUpload() { /** @@ -173,6 +223,17 @@ abstract class UploadBase { return array( 'status' => self::EMPTY_FILE ); } + /** + * Honor $wgMaxUploadSize + */ + global $wgMaxUploadSize; + if( $this->mFileSize > $wgMaxUploadSize ) { + return array( + 'status' => self::FILE_TOO_LARGE, + 'max' => $wgMaxUploadSize, + ); + } + /** * Look at the contents of the file; if we can recognize the * type but it's corrupt or data of the wrong type, we should @@ -180,15 +241,36 @@ abstract class UploadBase { */ $verification = $this->verifyFile(); if( $verification !== true ) { - if( !is_array( $verification ) ) { - $verification = array( $verification ); - } return array( 'status' => self::VERIFICATION_ERROR, 'details' => $verification ); } + /** + * Make sure this file can be created + */ + $result = $this->validateName(); + if( $result !== true ) { + return $result; + } + + $error = ''; + if( !wfRunHooks( 'UploadVerification', + array( $this->mDestName, $this->mTempPath, &$error ) ) ) { + return array( 'status' => self::HOOK_ABORTED, 'error' => $error ); + } + + return array( 'status' => self::OK ); + } + + /** + * Verify that the name is valid and, if necessary, that we can overwrite + * + * @return mixed true if valid, otherwise and array with 'status' + * and other keys + **/ + protected function validateName() { $nt = $this->getTitle(); if( is_null( $nt ) ) { $result = array( 'status' => $this->mTitleError ); @@ -202,41 +284,16 @@ abstract class UploadBase { } $this->mDestName = $this->getLocalFile()->getName(); - /** - * In some cases we may forbid overwriting of existing files. - */ - $overwrite = $this->checkOverwrite(); - if( $overwrite !== true ) { - return array( - 'status' => self::OVERWRITE_EXISTING_FILE, - 'overwrite' => $overwrite - ); - } - - $error = ''; - if( !wfRunHooks( 'UploadVerification', - array( $this->mDestName, $this->mTempPath, &$error ) ) ) { - // This status needs another name... - return array( 'status' => self::HOOK_ABORTED, 'error' => $error ); - } - - return array( 'status' => self::OK ); + return true; } /** - * Verifies that it's ok to include the uploaded file + * Verify the mime type * - * @return mixed true of the file is verified, a string or array otherwise. + * @param $mime string representing the mime + * @return mixed true if the file is verified, an array otherwise */ - protected function verifyFile() { - $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension ); - $this->checkMacBinary(); - - # magically determine mime type - $magic = MimeMagic::singleton(); - $mime = $magic->guessMimeType( $this->mTempPath, false ); - - # check mime type, if desired + protected function verifyMimeType( $mime ) { global $wgVerifyMimeType; if ( $wgVerifyMimeType ) { wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n"); @@ -253,6 +310,8 @@ abstract class UploadBase { $fp = fopen( $this->mTempPath, 'rb' ); $chunk = fread( $fp, 256 ); fclose( $fp ); + + $magic = MimeMagic::singleton(); $extMime = $magic->guessTypesForExtension( $this->mFinalExtension ); $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime ); foreach ( $ieTypes as $ieType ) { @@ -262,13 +321,36 @@ abstract class UploadBase { } } + return true; + } + + /** + * Verifies that it's ok to include the uploaded file + * + * @return mixed true of the file is verified, array otherwise. + */ + protected function verifyFile() { + # get the title, even though we are doing nothing with it, because + # we need to populate mFinalExtension + $this->getTitle(); + + $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension ); + $this->checkMacBinary(); + + # check mime type, if desired + $mime = $this->mFileProps[ 'file-mime' ]; + $status = $this->verifyMimeType( $mime ); + if ( $status !== true ) { + return $status; + } + # check for htmlish code and javascript if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) { - return 'uploadscripted'; + return array( 'uploadscripted' ); } if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) { - if( self::detectScriptInSvg( $this->mTempPath ) ) { - return 'uploadscripted'; + if( $this->detectScriptInSvg( $this->mTempPath ) ) { + return array( 'uploadscripted' ); } } @@ -279,14 +361,33 @@ abstract class UploadBase { if ( $virus ) { return array( 'uploadvirus', $virus ); } + + $handler = MediaHandler::getHandler( $mime ); + if ( $handler ) { + $handlerStatus = $handler->verifyUpload( $this->mTempPath ); + if ( !$handlerStatus->isOK() ) { + $errors = $handlerStatus->getErrorsArray(); + return reset( $errors ); + } + } + + wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) ); + if ( $status !== true ) { + return $status; + } + wfDebug( __METHOD__ . ": all clear; passing.\n" ); return true; } /** - * Check whether the user can edit, upload and create the image. + * Check whether the user can edit, upload and create the image. This + * checks only against the current title; if it returns errors, it may + * very well be that another title will not give errors. Therefore + * isAllowed() should be called as well for generic is-user-blocked or + * can-user-upload checking. * - * @param User $user the user to verify the permissions against + * @param $user the User object to verify the permissions against * @return mixed An array as returned by getUserPermissionsErrors or true * in case the user has proper permissions. */ @@ -301,19 +402,29 @@ abstract class UploadBase { } $permErrors = $nt->getUserPermissionsErrors( 'edit', $user ); $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user ); - $permErrorsCreate = ( $nt->exists() ? array() : $nt->getUserPermissionsErrors( 'create', $user ) ); + if ( !$nt->exists() ) { + $permErrorsCreate = $nt->getUserPermissionsErrors( 'createpage', $user ); + } else { + $permErrorsCreate = array(); + } if( $permErrors || $permErrorsUpload || $permErrorsCreate ) { $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) ); $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) ); return $permErrors; } + + $overwriteError = $this->checkOverwrite( $user ); + if ( $overwriteError !== true ) { + return array( $overwriteError ); + } + return true; } /** * Check for non fatal problems with the file * - * @return array Array of warnings + * @return Array of warnings */ public function checkWarnings() { $warnings = array(); @@ -321,7 +432,6 @@ abstract class UploadBase { $localFile = $this->getLocalFile(); $filename = $localFile->getName(); $n = strrpos( $filename, '.' ); - $partname = $n ? substr( $filename, 0, $n ) : $filename; /** * Check whether the resulting filename is different from the desired one, @@ -386,15 +496,21 @@ abstract class UploadBase { * @return mixed Status indicating the whether the upload succeeded. */ public function performUpload( $comment, $pageText, $watch, $user ) { - wfDebug( "\n\n\performUpload: sum:" . $comment . ' c: ' . $pageText . ' w:' . $watch ); - $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText, - File::DELETE_SOURCE, $this->mFileProps, false, $user ); - - if( $status->isGood() && $watch ) { - $user->addWatch( $this->getLocalFile()->getTitle() ); - } + $status = $this->getLocalFile()->upload( + $this->mTempPath, + $comment, + $pageText, + File::DELETE_SOURCE, + $this->mFileProps, + false, + $user + ); if( $status->isGood() ) { + if ( $watch ) { + $user->addWatch( $this->getLocalFile()->getTitle() ); + } + wfRunHooks( 'UploadComplete', array( &$this ) ); } @@ -417,9 +533,7 @@ abstract class UploadBase { * filter out illegal characters, and try to make a legible name * out of it. We'll strip some silently that Title would die on. */ - $basename = $this->mDesiredDestName; - - $this->mFilteredName = wfStripIllegalFilenameChars( $basename ); + $this->mFilteredName = wfStripIllegalFilenameChars( $this->mDesiredDestName ); /* Normalize to title form before we do any further processing */ $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName ); if( is_null( $nt ) ) { @@ -466,11 +580,6 @@ abstract class UploadBase { return $this->mTitle = null; } - $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName ); - if( is_null( $nt ) ) { - $this->mTitleError = self::ILLEGAL_FILENAME; - return $this->mTitle = null; - } return $this->mTitle = $nt; } @@ -486,15 +595,18 @@ abstract class UploadBase { } /** + * NOTE: Probably should be deprecated in favor of UploadStash, but this is sometimes + * called outside that context. + * * Stash a file in a temporary directory for later processing * after the user has confirmed it. * * If the user doesn't explicitly cancel or accept, these files * can accumulate in the temp directory. * - * @param string $saveName - the destination filename - * @param string $tempSrc - the source temporary file to save - * @return string - full path the stashed file, or false on failure + * @param $saveName String: the destination filename + * @param $tempSrc String: the source temporary file to save + * @return String: full path the stashed file, or false on failure */ protected function saveTempUploadedFile( $saveName, $tempSrc ) { $repo = RepoGroup::singleton()->getLocalRepo(); @@ -503,39 +615,35 @@ abstract class UploadBase { } /** - * Stash a file in a temporary directory for later processing, - * and save the necessary descriptive info into the session. - * Returns a key value which will be passed through a form - * to pick up the path info on a later invocation. + * If the user does not supply all necessary information in the first upload form submission (either by accident or + * by design) then we may want to stash the file temporarily, get more information, and publish the file later. + * + * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info + * into the user's session. + * This method returns the file object, which also has a 'sessionKey' property which can be passed through a form or + * API request to find this stashed file again. * - * @return int Session key + * @param $key String: (optional) the session key used to find the file info again. If not supplied, a key will be autogenerated. + * @return File: stashed file */ - public function stashSession() { - $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath ); - if( !$status->isOK() ) { - # Couldn't save the file. - return false; - } - if( !isset( $_SESSION ) ) { - session_start(); // start up the session (might have been previously closed to prevent php session locking) - } - $key = $this->getSessionKey(); - $_SESSION['wsUploadData'][$key] = array( - 'mTempPath' => $status->value, - 'mFileSize' => $this->mFileSize, - 'mFileProps' => $this->mFileProps, - 'version' => self::SESSION_VERSION, + public function stashSessionFile( $key = null ) { + $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash(); + $data = array( + 'mFileProps' => $this->mFileProps ); - return $key; + $file = $stash->stashFile( $this->mTempPath, $data, $key ); + $this->mLocalFile = $file; + return $file; } /** - * Generate a random session key from stash in cases where we want to start an upload without much information + * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashSessionFile(). + * + * @param $key String: (optional) the session key used to find the file info again. If not supplied, a key will be autogenerated. + * @return String: session key */ - protected function getSessionKey() { - $key = mt_rand( 0, 0x7fffffff ); - $_SESSION['wsUploadData'][$key] = array(); - return $key; + public function stashSession( $key = null ) { + return $this->stashSessionFile( $key )->getSessionKey(); } /** @@ -571,9 +679,9 @@ abstract class UploadBase { * Perform case-insensitive match against a list of file extensions. * Returns true if the extension is in the list. * - * @param string $ext - * @param array $list - * @return bool + * @param $ext String + * @param $list Array + * @return Boolean */ public static function checkFileExtension( $ext, $list ) { return in_array( strtolower( $ext ), $list ); @@ -583,9 +691,9 @@ abstract class UploadBase { * Perform case-insensitive match against a list of file extensions. * Returns true if any of the extensions are in the list. * - * @param array $ext - * @param array $list - * @return bool + * @param $ext Array + * @param $list Array + * @return Boolean */ public static function checkFileExtensionList( $ext, $list ) { foreach( $ext as $e ) { @@ -599,9 +707,9 @@ abstract class UploadBase { /** * Checks if the mime type of the uploaded file matches the file extension. * - * @param string $mime the mime type of the uploaded file - * @param string $extension The filename extension that the file is to be served with - * @return bool + * @param $mime String: the mime type of the uploaded file + * @param $extension String: the filename extension that the file is to be served with + * @return Boolean */ public static function verifyExtension( $mime, $extension ) { $magic = MimeMagic::singleton(); @@ -640,10 +748,10 @@ abstract class UploadBase { * potentially harmful. The present implementation will produce false * positives in some situations. * - * @param string $file Pathname to the temporary upload file - * @param string $mime The mime type of the file - * @param string $extension The extension of the file - * @return bool true if the file contains something looking like embedded scripts + * @param $file String: pathname to the temporary upload file + * @param $mime String: the mime type of the file + * @param $extension String: the extension of the file + * @return Boolean: true if the file contains something looking like embedded scripts */ public static function detectScript( $file, $mime, $extension ) { global $wgAllowTitlesInSVG; @@ -790,7 +898,7 @@ abstract class UploadBase { * This relies on the $wgAntivirus and $wgAntivirusSetup variables. * $wgAntivirusRequired may be used to deny upload if the scan fails. * - * @param string $file Pathname to the temporary upload file + * @param $file String: pathname to the temporary upload file * @return mixed false if not virus is found, NULL if the scan fails or is disabled, * or a string containing feedback from the virus scanner if a virus was found. * If textual feedback is missing but a virus was found, this function returns true. @@ -805,7 +913,8 @@ abstract class UploadBase { if ( !$wgAntivirusSetup[$wgAntivirus] ) { wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" ); - $wgOut->wrapWikiMsg( "
\n$1
", array( 'virus-badscanner', $wgAntivirus ) ); + $wgOut->wrapWikiMsg( "
\n$1\n
", + array( 'virus-badscanner', $wgAntivirus ) ); return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus"; } @@ -907,15 +1016,14 @@ abstract class UploadBase { * Check if there's an overwrite conflict and, if so, if restrictions * forbid this user from performing the upload. * - * @return mixed true on success, error string on failure + * @return mixed true on success, array on failure */ - private function checkOverwrite() { - global $wgUser; + private function checkOverwrite( $user ) { // First check whether the local file can be overwritten $file = $this->getLocalFile(); if( $file->exists() ) { - if( !self::userCanReUpload( $wgUser, $file ) ) { - return 'fileexists-forbidden'; + if( !self::userCanReUpload( $user, $file ) ) { + return array( 'fileexists-forbidden', $file->getName() ); } else { return true; } @@ -925,8 +1033,8 @@ abstract class UploadBase { * wfFindFile finds a file, it exists in a shared repository. */ $file = wfFindFile( $this->getTitle() ); - if ( $file && !$wgUser->isAllowed( 'reupload-shared' ) ) { - return 'fileexists-shared-forbidden'; + if ( $file && !$user->isAllowed( 'reupload-shared' ) ) { + return array( 'fileexists-shared-forbidden', $file->getName() ); } return true; @@ -935,9 +1043,9 @@ abstract class UploadBase { /** * Check if a user is the last uploader * - * @param User $user - * @param string $img, image name - * @return bool + * @param $user User object + * @param $img String: image name + * @return Boolean */ public static function userCanReUpload( User $user, $img ) { if( $user->isAllowed( 'reupload' ) ) { @@ -964,7 +1072,7 @@ abstract class UploadBase { * - File exists with normalized extension * - The file looks like a thumbnail and the original exists * - * @param File $file The file to check + * @param $file The File object to check * @return mixed False if the file does not exists, else an array */ public static function getExistsWarning( $file ) { @@ -1082,10 +1190,34 @@ abstract class UploadBase { return $blacklist; } + /** + * Gets image info about the file just uploaded. + * + * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if + * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that + * with the appropriate format, presumably. + * + * @param $result ApiResult: + * @return Array: image info + */ public function getImageInfo( $result ) { $file = $this->getLocalFile(); - $imParam = ApiQueryImageInfo::getPropertyNames(); - return ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result ); + // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here. + // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries. + if ( $file instanceof UploadStashFile ) { + $imParam = ApiQueryStashImageInfo::getPropertyNames(); + $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result ); + } else { + $imParam = ApiQueryImageInfo::getPropertyNames(); + $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result ); + } + return $info; } + + public function convertVerifyErrorToStatus( $error ) { + $code = $error['status']; + unset( $code['status'] ); + return Status::newFatal( $this->getVerificationErrorCode( $code ), $error ); + } } diff --git a/includes/upload/UploadFromFile.php b/includes/upload/UploadFromFile.php index 73581a61..e67ec191 100644 --- a/includes/upload/UploadFromFile.php +++ b/includes/upload/UploadFromFile.php @@ -1,32 +1,65 @@ getUpload( 'wpUploadFile' ); $desiredDestName = $request->getText( 'wpDestFile' ); if( !$desiredDestName ) - $desiredDestName = $request->getFileName( 'wpUploadFile' ); - return $this->initializePathInfo( - $desiredDestName, - $request->getFileTempName( 'wpUploadFile' ), - $request->getFileSize( 'wpUploadFile' ) - ); + $desiredDestName = $upload->getName(); + + return $this->initialize( $desiredDestName, $upload ); } + /** - * Entry point for upload from file. + * Initialize from a filename and a WebRequestUpload */ - function initialize( $name, $tempPath, $fileSize ) { - return $this->initializePathInfo( $name, $tempPath, $fileSize ); + function initialize( $name, $webRequestUpload ) { + $this->mUpload = $webRequestUpload; + return $this->initializePathInfo( $name, + $this->mUpload->getTempName(), $this->mUpload->getSize() ); } static function isValidRequest( $request ) { - return (bool)$request->getFileTempName( 'wpUploadFile' ); + # Allow all requests, even if no file is present, so that an error + # because a post_max_size or upload_max_filesize overflow + return true; + } + + public function verifyUpload() { + # Check for a post_max_size or upload_max_size overflow, so that a + # proper error can be shown to the user + if ( is_null( $this->mTempPath ) || $this->isEmptyFile() ) { + if ( $this->mUpload->isIniSizeOverflow() ) { + global $wgMaxUploadSize; + return array( + 'status' => UploadBase::FILE_TOO_LARGE, + 'max' => min( + $wgMaxUploadSize, + wfShorthandToInteger( ini_get( 'upload_max_filesize' ) ), + wfShorthandToInteger( ini_get( 'post_max_size' ) ) + ), + ); + } + } + + return parent::verifyUpload(); + } + + /** + * Get the path to the file underlying the upload + * @return String path to file + */ + public function getFileTempname() { + return $this->mUpload->getTempname(); } + + } diff --git a/includes/upload/UploadFromStash.php b/includes/upload/UploadFromStash.php index 17e922b0..156781e9 100644 --- a/includes/upload/UploadFromStash.php +++ b/includes/upload/UploadFromStash.php @@ -1,10 +1,9 @@ getSessionData( 'wsUploadData' ); + $sessionData = $request->getSessionData( UploadBase::SESSION_KEYNAME ); return self::isValidSessionKey( - $request->getInt( 'wpSessionKey' ), + $request->getText( 'wpSessionKey' ), $sessionData ); } @@ -45,8 +44,8 @@ class UploadFromStash extends UploadBase { } public function initializeFromRequest( &$request ) { - $sessionKey = $request->getInt( 'wpSessionKey' ); - $sessionData = $request->getSessionData('wsUploadData'); + $sessionKey = $request->getText( 'wpSessionKey' ); + $sessionData = $request->getSessionData( UploadBase::SESSION_KEYNAME ); $desiredDestName = $request->getText( 'wpDestFile' ); if( !$desiredDestName ) @@ -65,7 +64,7 @@ class UploadFromStash extends UploadBase { /** * There is no need to stash the image twice */ - public function stashSession() { + public function stashSession( $key = null ) { if ( !empty( $this->mSessionKey ) ) return $this->mSessionKey; return parent::stashSession(); diff --git a/includes/upload/UploadFromUrl.php b/includes/upload/UploadFromUrl.php index 763dae38..c28fd7da 100644 --- a/includes/upload/UploadFromUrl.php +++ b/includes/upload/UploadFromUrl.php @@ -1,28 +1,30 @@ isAllowed( 'upload_by_url' ) ) + if ( !$user->isAllowed( 'upload_by_url' ) ) return 'upload_by_url'; return parent::isAllowed( $user ); } /** * Checks if the upload from URL feature is enabled + * @return bool */ public static function isEnabled() { global $wgAllowCopyUploads; @@ -31,14 +33,22 @@ class UploadFromUrl extends UploadBase { /** * Entry point for API upload + * + * @param $name string + * @param $url string + * @param $async mixed Whether the download should be performed + * asynchronous. False for synchronous, async or async-leavemessage for + * asynchronous download. */ - public function initialize( $name, $url, $na, $nb = false ) { - global $wgTmpDirectory; + public function initialize( $name, $url, $async = false ) { + global $wgAllowAsyncCopyUploads; - $localFile = tempnam( $wgTmpDirectory, 'WEBUPLOAD' ); - $this->initializePathInfo( $name, $localFile, 0, true ); + $this->mUrl = $url; + $this->mAsync = $wgAllowAsyncCopyUploads ? $async : false; - $this->mUrl = trim( $url ); + $tempPath = $this->mAsync ? null : $this->makeTemporaryFile(); + # File size and removeTempFile will be filled in later + $this->initializePathInfo( $name, $tempPath, 0, false ); } /** @@ -47,7 +57,7 @@ class UploadFromUrl extends UploadBase { */ public function initializeFromRequest( &$request ) { $desiredDestName = $request->getText( 'wpDestFile' ); - if( !$desiredDestName ) + if ( !$desiredDestName ) $desiredDestName = $request->getText( 'wpUploadFileURL' ); return $this->initialize( $desiredDestName, @@ -59,79 +69,162 @@ class UploadFromUrl extends UploadBase { /** * @param $request Object: WebRequest object */ - public static function isValidRequest( $request ){ - if( !$request->getVal( 'wpUploadFileURL' ) ) - return false; - // check that is a valid url: - return self::isValidUrl( $request->getVal( 'wpUploadFileURL' ) ); + public static function isValidRequest( $request ) { + global $wgUser; + + $url = $request->getVal( 'wpUploadFileURL' ); + return !empty( $url ) + && Http::isValidURI( $url ) + && $wgUser->isAllowed( 'upload_by_url' ); } - public static function isValidUrl( $url ) { - // Only allow HTTP or FTP for now - return (bool)preg_match( '!^(http://|ftp://)!', $url ); + + public function fetchFile() { + if ( !Http::isValidURI( $this->mUrl ) ) { + return Status::newFatal( 'http-invalid-url' ); + } + + if ( !$this->mAsync ) { + return $this->reallyFetchFile(); + } + return Status::newGood(); + } + /** + * Create a new temporary file in the URL subdirectory of wfTempDir(). + * + * @return string Path to the file + */ + protected function makeTemporaryFile() { + return tempnam( wfTempDir(), 'URL' ); } /** - * Do the real fetching stuff + * Callback: save a chunk of the result of a HTTP request to the temporary file + * + * @param $req mixed + * @param $buffer string + * @return int number of bytes handled */ - function fetchFile() { - if( !self::isValidUrl( $this->mUrl ) ) { - return Status::newFatal( 'upload-proto-error' ); + public function saveTempFileChunk( $req, $buffer ) { + $nbytes = fwrite( $this->mTmpHandle, $buffer ); + + if ( $nbytes == strlen( $buffer ) ) { + $this->mFileSize += $nbytes; + } else { + // Well... that's not good! + fclose( $this->mTmpHandle ); + $this->mTmpHandle = false; } - $res = $this->curlCopy(); - if( $res !== true ) { - return Status::newFatal( $res ); - } - return Status::newGood(); + + return $nbytes; } /** - * Safe copy from URL - * Returns true if there was an error, false otherwise + * Download the file, save it to the temporary file and update the file + * size and set $mRemoveTempFile to true. */ - private function curlCopy() { - global $wgOut; - - # Open temporary file - $this->mCurlDestHandle = @fopen( $this->mTempPath, "wb" ); - if( $this->mCurlDestHandle === false ) { - # Could not open temporary file to write in - return 'upload-file-error'; + protected function reallyFetchFile() { + if ( $this->mTempPath === false ) { + return Status::newFatal( 'tmp-create-error' ); } - $ch = curl_init(); - curl_setopt( $ch, CURLOPT_HTTP_VERSION, 1.0); # Probably not needed, but apparently can work around some bug - curl_setopt( $ch, CURLOPT_TIMEOUT, 10); # 10 seconds timeout - curl_setopt( $ch, CURLOPT_LOW_SPEED_LIMIT, 512); # 0.5KB per second minimum transfer speed - curl_setopt( $ch, CURLOPT_URL, $this->mUrl); - curl_setopt( $ch, CURLOPT_WRITEFUNCTION, array( $this, 'uploadCurlCallback' ) ); - curl_exec( $ch ); - $error = curl_errno( $ch ); - curl_close( $ch ); + // 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' ); + } + + $this->mRemoveTempFile = true; + $this->mFileSize = 0; - fclose( $this->mCurlDestHandle ); - unset( $this->mCurlDestHandle ); + $req = MWHttpRequest::factory( $this->mUrl ); + $req->setCallback( array( $this, 'saveTempFileChunk' ) ); + $status = $req->execute(); - if( $error ) - return "upload-curl-error$errornum"; + if ( $this->mTmpHandle ) { + // File got written ok... + fclose( $this->mTmpHandle ); + $this->mTmpHandle = null; + } else { + // We encountered a write error during the download... + return Status::newFatal( 'tmp-write-error' ); + } + + if ( !$status->isOk() ) { + return $status; + } - return true; + return $status; } /** - * Callback function for CURL-based web transfer - * Write data to file unless we've passed the length limit; - * if so, abort immediately. - * @access private + * Wrapper around the parent function in order to defer verifying the + * upload until the file really has been fetched. */ - function uploadCurlCallback( $ch, $data ) { - global $wgMaxUploadSize; - $length = strlen( $data ); - $this->mFileSize += $length; - if( $this->mFileSize > $wgMaxUploadSize ) { - return 0; + public function verifyUpload() { + if ( $this->mAsync ) { + return array( 'status' => UploadBase::OK ); } - fwrite( $this->mCurlDestHandle, $data ); - return $length; + return parent::verifyUpload(); } + + /** + * Wrapper around the parent function in order to defer checking warnings + * until the file really has been fetched. + */ + public function checkWarnings() { + if ( $this->mAsync ) { + $this->mIgnoreWarnings = false; + return array(); + } + return parent::checkWarnings(); + } + + /** + * Wrapper around the parent function in order to defer checking protection + * until we are sure that the file can actually be uploaded + */ + public function verifyPermissions( $user ) { + if ( $this->mAsync ) { + return true; + } + return parent::verifyPermissions( $user ); + } + + /** + * Wrapper around the parent function in order to defer uploading to the + * job queue for asynchronous uploads + */ + public function performUpload( $comment, $pageText, $watch, $user ) { + if ( $this->mAsync ) { + $sessionKey = $this->insertJob( $comment, $pageText, $watch, $user ); + + $status = new Status; + $status->error( 'async', $sessionKey ); + return $status; + } + + return parent::performUpload( $comment, $pageText, $watch, $user ); + } + + + protected function insertJob( $comment, $pageText, $watch, $user ) { + $sessionKey = $this->stashSession(); + $job = new UploadFromUrlJob( $this->getTitle(), array( + 'url' => $this->mUrl, + 'comment' => $comment, + 'pageText' => $pageText, + 'watch' => $watch, + 'userName' => $user->getName(), + 'leaveMessage' => $this->mAsync == 'async-leavemessage', + 'ignoreWarnings' => $this->mIgnoreWarnings, + 'sessionId' => session_id(), + 'sessionKey' => $sessionKey, + ) ); + $job->initializeSessionData(); + $job->insert(); + return $sessionKey; + } + + } diff --git a/includes/upload/UploadStash.php b/includes/upload/UploadStash.php new file mode 100644 index 00000000..1765925d --- /dev/null +++ b/includes/upload/UploadStash.php @@ -0,0 +1,397 @@ +file mapping, on the assumption that nobody else can access + * the session, even the uploading user. See SpecialUploadStash, which implements a web interface to some files stored this way. + * + */ +class UploadStash { + + // Format of the key for files -- has to be suitable as a filename itself (e.g. ab12cd34ef.jpg) + const KEY_FORMAT_REGEX = '/^[\w-]+\.\w*$/'; + + // repository that this uses to store temp files + // public because we sometimes need to get a LocalFile within the same repo. + public $repo; + + // array of initialized objects obtained from session (lazily initialized upon getFile()) + private $files = array(); + + // TODO: Once UploadBase starts using this, switch to use these constants rather than UploadBase::SESSION* + // const SESSION_VERSION = 2; + // const SESSION_KEYNAME = 'wsUploadData'; + + /** + * Represents the session which contains temporarily stored files. + * Designed to be compatible with the session stashing code in UploadBase (should replace it eventually) + * + * @param $repo FileRepo: optional -- repo in which to store files. Will choose LocalRepo if not supplied. + */ + public function __construct( $repo ) { + + // this might change based on wiki's configuration. + $this->repo = $repo; + + if ( ! isset( $_SESSION ) ) { + throw new UploadStashNotAvailableException( 'no session variable' ); + } + + if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME] ) ) { + $_SESSION[UploadBase::SESSION_KEYNAME] = array(); + } + + } + + /** + * Get a file and its metadata from the stash. + * May throw exception if session data cannot be parsed due to schema change, or key not found. + * + * @param $key Integer: key + * @throws UploadStashFileNotFoundException + * @throws UploadStashBadVersionException + * @return UploadStashFile + */ + public function getFile( $key ) { + if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) { + throw new UploadStashBadPathException( "key '$key' is not in a proper format" ); + } + + if ( !isset( $this->files[$key] ) ) { + if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME][$key] ) ) { + throw new UploadStashFileNotFoundException( "key '$key' not found in stash" ); + } + + $data = $_SESSION[UploadBase::SESSION_KEYNAME][$key]; + // guards against PHP class changing while session data doesn't + if ($data['version'] !== UploadBase::SESSION_VERSION ) { + throw new UploadStashBadVersionException( $data['version'] . " does not match current version " . UploadBase::SESSION_VERSION ); + } + + // separate the stashData into the path, and then the rest of the data + $path = $data['mTempPath']; + unset( $data['mTempPath'] ); + + $file = new UploadStashFile( $this, $this->repo, $path, $key, $data ); + if ( $file->getSize === 0 ) { + throw new UploadStashZeroLengthFileException( "File is zero length" ); + } + $this->files[$key] = $file; + + } + return $this->files[$key]; + } + + /** + * Stash a file in a temp directory and record that we did this in the session, along with other metadata. + * We store data in a flat key-val namespace because that's how UploadBase did it. This also means we have to + * ensure that the key-val pairs in $data do not overwrite other required fields. + * + * @param $path String: path to file you want stashed + * @param $data Array: optional, other data you want associated with the file. Do not use 'mTempPath', 'mFileProps', 'mFileSize', or 'version' as keys here + * @param $key String: optional, unique key for this file in this session. Used for directory hashing when storing, otherwise not important + * @throws UploadStashBadPathException + * @throws UploadStashFileException + * @return UploadStashFile: file, or null on failure + */ + public function stashFile( $path, $data = array(), $key = null ) { + if ( ! file_exists( $path ) ) { + wfDebug( "UploadStash: tried to stash file at '$path', but it doesn't exist\n" ); + throw new UploadStashBadPathException( "path doesn't exist" ); + } + $fileProps = File::getPropsFromPath( $path ); + + // we will be initializing from some tmpnam files that don't have extensions. + // most of MediaWiki assumes all uploaded files have good extensions. So, we fix this. + $extension = self::getExtensionForPath( $path ); + if ( ! preg_match( "/\\.\\Q$extension\\E$/", $path ) ) { + $pathWithGoodExtension = "$path.$extension"; + if ( ! rename( $path, $pathWithGoodExtension ) ) { + throw new UploadStashFileException( "couldn't rename $path to have a better extension at $pathWithGoodExtension" ); + } + $path = $pathWithGoodExtension; + } + + // If no key was supplied, use content hash. Also has the nice property of collapsing multiple identical files + // uploaded this session, which could happen if uploads had failed. + if ( is_null( $key ) ) { + $key = $fileProps['sha1'] . "." . $extension; + } + + if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) { + throw new UploadStashBadPathException( "key '$key' is not in a proper format" ); + } + + + // if not already in a temporary area, put it there + $status = $this->repo->storeTemp( basename( $path ), $path ); + + if( ! $status->isOK() ) { + // It is a convention in MediaWiki to only return one error per API exception, even if multiple errors + // are available. We use reset() to pick the "first" thing that was wrong, preferring errors to warnings. + // This is a bit lame, as we may have more info in the $status and we're throwing it away, but to fix it means + // redesigning API errors significantly. + // $status->value just contains the virtual URL (if anything) which is probably useless to the caller + $error = reset( $status->getErrorsArray() ); + if ( ! count( $error ) ) { + $error = reset( $status->getWarningsArray() ); + if ( ! count( $error ) ) { + $error = array( 'unknown', 'no error recorded' ); + } + } + throw new UploadStashFileException( "error storing file in '$path': " . implode( '; ', $error ) ); + } + $stashPath = $status->value; + + // required info we always store. Must trump any other application info in $data + // 'mTempPath', 'mFileSize', and 'mFileProps' are arbitrary names + // chosen for compatibility with UploadBase's way of doing this. + $requiredData = array( + 'mTempPath' => $stashPath, + 'mFileSize' => $fileProps['size'], + 'mFileProps' => $fileProps, + 'version' => UploadBase::SESSION_VERSION + ); + + // now, merge required info and extra data into the session. (The extra data changes from application to application. + // UploadWizard wants different things than say FirefoggChunkedUpload.) + wfDebug( __METHOD__ . " storing under $key\n" ); + $_SESSION[UploadBase::SESSION_KEYNAME][$key] = array_merge( $data, $requiredData ); + + return $this->getFile( $key ); + } + + /** + * Find or guess extension -- ensuring that our extension matches our mime type. + * Since these files are constructed from php tempnames they may not start off + * with an extension. + * XXX this is somewhat redundant with the checks that ApiUpload.php does with incoming + * uploads versus the desired filename. Maybe we can get that passed to us... + */ + public static function getExtensionForPath( $path ) { + // Does this have an extension? + $n = strrpos( $path, '.' ); + $extension = null; + if ( $n !== false ) { + $extension = $n ? substr( $path, $n + 1 ) : ''; + } else { + // If not, assume that it should be related to the mime type of the original file. + $magic = MimeMagic::singleton(); + $mimeType = $magic->guessMimeType( $path ); + $extensions = explode( ' ', MimeMagic::singleton()->getExtensionsForType( $mimeType ) ); + if ( count( $extensions ) ) { + $extension = $extensions[0]; + } + } + + if ( is_null( $extension ) ) { + throw new UploadStashFileException( "extension is null" ); + } + + return File::normalizeExtension( $extension ); + } + +} + +class UploadStashFile extends UnregisteredLocalFile { + private $sessionStash; + private $sessionKey; + private $sessionData; + private $urlName; + + /** + * A LocalFile wrapper around a file that has been temporarily stashed, so we can do things like create thumbnails for it + * Arguably UnregisteredLocalFile should be handling its own file repo but that class is a bit retarded currently + * + * @param $stash UploadStash: useful for obtaining config, stashing transformed files + * @param $repo FileRepo: repository where we should find the path + * @param $path String: path to file + * @param $key String: key to store the path and any stashed data under + * @param $data String: any other data we want stored with this file + * @throws UploadStashBadPathException + * @throws UploadStashFileNotFoundException + */ + public function __construct( $stash, $repo, $path, $key, $data ) { + $this->sessionStash = $stash; + $this->sessionKey = $key; + $this->sessionData = $data; + + // resolve mwrepo:// urls + if ( $repo->isVirtualUrl( $path ) ) { + $path = $repo->resolveVirtualUrl( $path ); + } + + // check if path appears to be sane, no parent traversals, and is in this repo's temp zone. + $repoTempPath = $repo->getZonePath( 'temp' ); + if ( ( ! $repo->validateFilename( $path ) ) || + ( strpos( $path, $repoTempPath ) !== 0 ) ) { + wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not valid\n" ); + throw new UploadStashBadPathException( 'path is not valid' ); + } + + // check if path exists! and is a plain file. + if ( ! $repo->fileExists( $path, FileRepo::FILES_ONLY ) ) { + wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not found\n" ); + throw new UploadStashFileNotFoundException( 'cannot find path, or not a plain file' ); + } + + + + parent::__construct( false, $repo, $path, false ); + + $this->name = basename( $this->path ); + } + + /** + * A method needed by the file transforming and scaling routines in File.php + * We do not necessarily care about doing the description at this point + * However, we also can't return the empty string, as the rest of MediaWiki demands this (and calls to imagemagick + * convert require it to be there) + * + * @return String: dummy value + */ + public function getDescriptionUrl() { + return $this->getUrl(); + } + + /** + * Get the path for the thumbnail (actually any transformation of this file) + * The actual argument is the result of thumbName although we seem to have + * buggy code elsewhere that expects a boolean 'suffix' + * + * @param $thumbName String: name of thumbnail (e.g. "120px-123456.jpg" ), or false to just get the path + * @return String: path thumbnail should take on filesystem, or containing directory if thumbname is false + */ + public function getThumbPath( $thumbName = false ) { + $path = dirname( $this->path ); + if ( $thumbName !== false ) { + $path .= "/$thumbName"; + } + return $path; + } + + /** + * Return the file/url base name of a thumbnail with the specified parameters + * + * @param $params Array: handler-specific parameters + * @return String: base name for URL, like '120px-12345.jpg', or null if there is no handler + */ + function thumbName( $params ) { + return $this->getParamThumbName( $this->getUrlName(), $params ); + } + + + /** + * Given the name of the original, i.e. Foo.jpg, and scaling parameters, returns filename with appropriate extension + * This is abstracted from getThumbName because we also use it to calculate the thumbname the file should have on + * remote image scalers + * + * @param String $urlName: A filename, like MyMovie.ogx + * @param Array $parameters: scaling parameters, like array( 'width' => '120' ); + * @return String|null parameterized thumb name, like 120px-MyMovie.ogx.jpg, or null if no handler found + */ + function getParamThumbName( $urlName, $params ) { + if ( !$this->getHandler() ) { + return null; + } + $extension = $this->getExtension(); + list( $thumbExt, ) = $this->handler->getThumbType( $extension, $this->getMimeType(), $params ); + $thumbName = $this->getHandler()->makeParamString( $params ) . '-' . $urlName; + if ( $thumbExt != $extension ) { + $thumbName .= ".$thumbExt"; + } + return $thumbName; + } + + /** + * Helper function -- given a 'subpage', return the local URL e.g. /wiki/Special:UploadStash/subpage + * @param {String} $subPage + * @return {String} local URL for this subpage in the Special:UploadStash space. + */ + private function getSpecialUrl( $subPage ) { + return SpecialPage::getTitleFor( 'UploadStash', $subPage )->getLocalURL(); + } + + + /** + * Get a URL to access the thumbnail + * This is required because the model of how files work requires that + * the thumbnail urls be predictable. However, in our model the URL is not based on the filename + * (that's hidden in the session) + * + * @param $thumbName String: basename of thumbnail file -- however, we don't want to use the file exactly + * @return String: URL to access thumbnail, or URL with partial path + */ + public function getThumbUrl( $thumbName = false ) { + wfDebug( __METHOD__ . " getting for $thumbName \n" ); + return $this->getSpecialUrl( 'thumb/' . $this->getUrlName() . '/' . $thumbName ); + } + + /** + * The basename for the URL, which we want to not be related to the filename. + * Will also be used as the lookup key for a thumbnail file. + * + * @return String: base url name, like '120px-123456.jpg' + */ + public function getUrlName() { + if ( ! $this->urlName ) { + $this->urlName = $this->sessionKey; + } + return $this->urlName; + } + + /** + * Return the URL of the file, if for some reason we wanted to download it + * We tend not to do this for the original file, but we do want thumb icons + * + * @return String: url + */ + public function getUrl() { + if ( !isset( $this->url ) ) { + $this->url = $this->getSpecialUrl( 'file/' . $this->getUrlName() ); + } + return $this->url; + } + + /** + * Parent classes use this method, for no obvious reason, to return the path (relative to wiki root, I assume). + * But with this class, the URL is unrelated to the path. + * + * @return String: url + */ + public function getFullUrl() { + return $this->getUrl(); + } + + + /** + * Getter for session key (the session-unique id by which this file's location & metadata is stored in the session) + * + * @return String: session key + */ + public function getSessionKey() { + return $this->sessionKey; + } + + /** + * Remove the associated temporary file + * @return Status: success + */ + public function remove() { + return $this->repo->freeTemp( $this->path ); + } + +} + +class UploadStashNotAvailableException extends MWException {}; +class UploadStashFileNotFoundException extends MWException {}; +class UploadStashBadPathException extends MWException {}; +class UploadStashBadVersionException extends MWException {}; +class UploadStashFileException extends MWException {}; +class UploadStashZeroLengthFileException extends MWException {}; + -- cgit v1.2.2