<?php
/**
- * MediaWiki page data importer
+ * MediaWiki page data importer.
*
* Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
* http://www.mediawiki.org/
class WikiImporter {
private $reader = null;
private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
- private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
- private $mDebug;
+ private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
+ private $mNoticeCallback, $mDebug;
private $mImportUploads, $mImageBasePath;
private $mNoUpdates = false;
/**
* Creates an ImportXMLReader drawing from the source provided
+ * @param $source
*/
function __construct( $source ) {
$this->reader = new XMLReader();
$id = UploadSourceAdapter::registerSource( $source );
if (defined( 'LIBXML_PARSEHUGE' ) ) {
$this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
- }
- else {
+ } else {
$this->reader->open( "uploadsource://$id" );
}
wfDebug( "IMPORT: $data\n" );
}
- private function notice( $data ) {
- global $wgCommandLineMode;
- if( $wgCommandLineMode ) {
- print "$data\n";
- } else {
- global $wgOut;
- $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
+ private function notice( $msg /*, $param, ...*/ ) {
+ $params = func_get_args();
+ array_shift( $params );
+
+ if ( is_callable( $this->mNoticeCallback ) ) {
+ call_user_func( $this->mNoticeCallback, $msg, $params );
+ } else { # No ImportReporter -> CLI
+ echo wfMessage( $msg, $params )->text() . "\n";
}
}
/**
* Set debug mode...
+ * @param $debug bool
*/
function setDebug( $debug ) {
$this->mDebug = $debug;
}
-
+
/**
* Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
+ * @param $noupdates bool
*/
function setNoUpdates( $noupdates ) {
$this->mNoUpdates = $noupdates;
}
+ /**
+ * Set a callback that displays notice messages
+ *
+ * @param $callback callback
+ * @return callback
+ */
+ public function setNoticeCallback( $callback ) {
+ return wfSetVar( $this->mNoticeCallback, $callback );
+ }
+
/**
* Sets the action to perform as each new page in the stream is reached.
* @param $callback callback
/**
* Set a target namespace to override the defaults
+ * @param $namespace
+ * @return bool
*/
public function setTargetNamespace( $namespace ) {
if( is_null( $namespace ) ) {
return false;
}
}
-
+
+ /**
+ * Set a target root page under which all pages are imported
+ * @param $rootpage
+ * @return status object
+ */
+ public function setTargetRootPage( $rootpage ) {
+ $status = Status::newGood();
+ if( is_null( $rootpage ) ) {
+ // No rootpage
+ $this->mTargetRootPage = null;
+ } elseif( $rootpage !== '' ) {
+ $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
+ $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace ) ? $this->mTargetNamespace : NS_MAIN );
+ if( !$title || $title->isExternal() ) {
+ $status->fatal( 'import-rootpage-invalid' );
+ } else {
+ if( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
+ global $wgContLang;
+
+ $displayNSText = $title->getNamespace() == NS_MAIN
+ ? wfMessage( 'blanknamespace' )->text()
+ : $wgContLang->getNsText( $title->getNamespace() );
+ $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
+ } else {
+ // set namespace to 'all', so the namespace check in processTitle() can passed
+ $this->setTargetNamespace( null );
+ $this->mTargetRootPage = $title->getPrefixedDBKey();
+ }
+ }
+ }
+ return $status;
+ }
+
/**
- *
+ * @param $dir
*/
public function setImageBasePath( $dir ) {
$this->mImageBasePath = $dir;
}
+
+ /**
+ * @param $import
+ */
public function setImportUploads( $import ) {
$this->mImportUploads = $import;
}
/**
* Default per-revision callback, performs the import.
* @param $revision WikiRevision
+ * @return bool
*/
public function importRevision( $revision ) {
$dbw = wfGetDB( DB_MASTER );
/**
* Default per-revision callback, performs the import.
* @param $rev WikiRevision
+ * @return bool
*/
public function importLogItem( $rev ) {
$dbw = wfGetDB( DB_MASTER );
/**
* Dummy for now...
+ * @param $revision
+ * @return bool
*/
public function importUpload( $revision ) {
$dbw = wfGetDB( DB_MASTER );
/**
* Mostly for hook use
+ * @param $title
+ * @param $origTitle
+ * @param $revCount
+ * @param $sRevCount
+ * @param $pageInfo
+ * @return
*/
public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
$args = func_get_args();
}
/**
- * Notify the callback function when a new <page> is reached.
+ * Notify the callback function when a new "<page>" is reached.
* @param $title Title
*/
function pageCallback( $title ) {
}
/**
- * Notify the callback function when a </page> is closed.
+ * Notify the callback function when a "</page>" is closed.
* @param $title Title
* @param $origTitle Title
* @param $revCount Integer
/**
* Notify the callback function of a revision
- * @param $revision A WikiRevision object
+ * @param $revision WikiRevision object
+ * @return bool|mixed
*/
private function revisionCallback( $revision ) {
if ( isset( $this->mRevisionCallback ) ) {
/**
* Notify the callback function of a new log item
- * @param $revision A WikiRevision object
+ * @param $revision WikiRevision object
+ * @return bool|mixed
*/
private function logItemCallback( $revision ) {
if ( isset( $this->mLogItemCallback ) ) {
/**
* Primary entry point
+ * @return bool
*/
public function doImport() {
$this->reader->read();
return true;
}
+ /**
+ * @return bool
+ * @throws MWException
+ */
private function handleSiteInfo() {
// Site info is useful, but not actually used for dump imports.
// Includes a quick short-circuit to save performance.
$this->processLogItem( $logInfo );
}
+ /**
+ * @param $logInfo
+ * @return bool|mixed
+ */
private function processLogItem( $logInfo ) {
$revision = new WikiRevision;
$pageInfo );
}
+ /**
+ * @param $pageInfo array
+ */
private function handleRevision( &$pageInfo ) {
$this->debug( "Enter revision handler" );
$revisionInfo = array();
}
}
+ /**
+ * @param $pageInfo
+ * @param $revisionInfo
+ * @return bool|mixed
+ */
private function processRevision( $pageInfo, $revisionInfo ) {
$revision = new WikiRevision;
return $this->revisionCallback( $revision );
}
+ /**
+ * @param $pageInfo
+ * @return mixed
+ */
private function handleUpload( &$pageInfo ) {
$this->debug( "Enter upload handler" );
$uploadInfo = array();
$skip = true;
}
}
-
+
if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
$path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
if ( file_exists( $path ) ) {
return $this->processUpload( $pageInfo, $uploadInfo );
}
}
-
+
+ /**
+ * @param $contents
+ * @return string
+ */
private function dumpTemp( $contents ) {
$filename = tempnam( wfTempDir(), 'importupload' );
file_put_contents( $filename, $contents );
return $filename;
}
-
+ /**
+ * @param $pageInfo
+ * @param $uploadInfo
+ * @return mixed
+ */
private function processUpload( $pageInfo, $uploadInfo ) {
$revision = new WikiRevision;
$text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
return call_user_func( $this->mUploadCallback, $revision );
}
+ /**
+ * @return array
+ */
private function handleContributor() {
$fields = array( 'id', 'ip', 'username' );
$info = array();
}
/**
+ * @param $text string
* @return Array or false
*/
private function processTitle( $text ) {
$origTitle = Title::newFromText( $workTitle );
if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
- $title = Title::makeTitle( $this->mTargetNamespace,
+ # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
+ # and than dbKey can begin with a lowercase char
+ $title = Title::makeTitleSafe( $this->mTargetNamespace,
$origTitle->getDBkey() );
} else {
+ if( !is_null( $this->mTargetRootPage ) ) {
+ $workTitle = $this->mTargetRootPage . '/' . $workTitle;
+ }
$title = Title::newFromText( $workTitle );
}
if( is_null( $title ) ) {
# Invalid page title? Ignore the page
- $this->notice( "Skipping invalid page title '$workTitle'" );
+ $this->notice( 'import-error-invalid', $workTitle );
+ return false;
+ } elseif( $title->isExternal() ) {
+ $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
return false;
- } elseif( $title->getInterwiki() != '' ) {
- $this->notice( "Skipping interwiki page title '$workTitle'" );
+ } elseif( !$title->canExist() ) {
+ $this->notice( 'import-error-special', $title->getPrefixedText() );
return false;
} elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
# Do not import if the importing wiki user cannot edit this page
- $this->notice( wfMessage( 'import-error-edit', $title->getText() )->text() );
+ $this->notice( 'import-error-edit', $title->getPrefixedText() );
return false;
} elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
# Do not import if the importing wiki user cannot create this page
- $this->notice( wfMessage( 'import-error-create', $title->getText() )->text() );
+ $this->notice( 'import-error-create', $title->getPrefixedText() );
return false;
}
private $mBuffer;
private $mPosition;
+ /**
+ * @param $source
+ * @return string
+ */
static function registerSource( $source ) {
- $id = wfGenerateToken();
+ $id = wfRandomString();
self::$sourceRegistrations[$id] = $source;
return $id;
}
+ /**
+ * @param $path
+ * @param $mode
+ * @param $options
+ * @param $opened_path
+ * @return bool
+ */
function stream_open( $path, $mode, $options, &$opened_path ) {
$url = parse_url($path);
$id = $url['host'];
return true;
}
+ /**
+ * @param $count
+ * @return string
+ */
function stream_read( $count ) {
$return = '';
$leave = false;
return $return;
}
+ /**
+ * @param $data
+ * @return bool
+ */
function stream_write( $data ) {
return false;
}
+ /**
+ * @return mixed
+ */
function stream_tell() {
return $this->mPosition;
}
+ /**
+ * @return bool
+ */
function stream_eof() {
return $this->mSource->atEnd();
}
+ /**
+ * @return array
+ */
function url_stat() {
$result = array();
}
class XMLReader2 extends XMLReader {
+
+ /**
+ * @return bool|string
+ */
function nodeContents() {
if( $this->isEmptyElement ) {
return "";
*/
class WikiRevision {
var $importer = null;
+
+ /**
+ * @var Title
+ */
var $title = null;
var $id = 0;
var $timestamp = "20010115000000";
var $sha1base36 = false;
var $isTemp = false;
var $archiveName = '';
+ var $fileIsTemp;
private $mNoUpdates = false;
+ /**
+ * @param $title
+ * @throws MWException
+ */
function setTitle( $title ) {
if( is_object( $title ) ) {
$this->title = $title;
}
}
+ /**
+ * @param $id
+ */
function setID( $id ) {
$this->id = $id;
}
+ /**
+ * @param $ts
+ */
function setTimestamp( $ts ) {
# 2003-08-05T18:30:02Z
$this->timestamp = wfTimestamp( TS_MW, $ts );
}
+ /**
+ * @param $user
+ */
function setUsername( $user ) {
$this->user_text = $user;
}
+ /**
+ * @param $ip
+ */
function setUserIP( $ip ) {
$this->user_text = $ip;
}
+ /**
+ * @param $text
+ */
function setText( $text ) {
$this->text = $text;
}
+ /**
+ * @param $text
+ */
function setComment( $text ) {
$this->comment = $text;
}
+ /**
+ * @param $minor
+ */
function setMinor( $minor ) {
$this->minor = (bool)$minor;
}
+ /**
+ * @param $src
+ */
function setSrc( $src ) {
$this->src = $src;
}
+
+ /**
+ * @param $src
+ * @param $isTemp
+ */
function setFileSrc( $src, $isTemp ) {
$this->fileSrc = $src;
$this->fileIsTemp = $isTemp;
}
- function setSha1Base36( $sha1base36 ) {
+
+ /**
+ * @param $sha1base36
+ */
+ function setSha1Base36( $sha1base36 ) {
$this->sha1base36 = $sha1base36;
}
+ /**
+ * @param $filename
+ */
function setFilename( $filename ) {
$this->filename = $filename;
}
+
+ /**
+ * @param $archiveName
+ */
function setArchiveName( $archiveName ) {
$this->archiveName = $archiveName;
}
+ /**
+ * @param $size
+ */
function setSize( $size ) {
$this->size = intval( $size );
}
+ /**
+ * @param $type
+ */
function setType( $type ) {
$this->type = $type;
}
+ /**
+ * @param $action
+ */
function setAction( $action ) {
$this->action = $action;
}
+ /**
+ * @param $params
+ */
function setParams( $params ) {
$this->params = $params;
}
-
+
+ /**
+ * @param $noupdates
+ */
public function setNoUpdates( $noupdates ) {
$this->mNoUpdates = $noupdates;
}
return $this->title;
}
+ /**
+ * @return int
+ */
function getID() {
return $this->id;
}
+ /**
+ * @return string
+ */
function getTimestamp() {
return $this->timestamp;
}
+ /**
+ * @return string
+ */
function getUser() {
return $this->user_text;
}
+ /**
+ * @return string
+ */
function getText() {
return $this->text;
}
+ /**
+ * @return string
+ */
function getComment() {
return $this->comment;
}
+ /**
+ * @return bool
+ */
function getMinor() {
return $this->minor;
}
+ /**
+ * @return mixed
+ */
function getSrc() {
return $this->src;
}
+
+ /**
+ * @return bool|String
+ */
function getSha1() {
if ( $this->sha1base36 ) {
return wfBaseConvert( $this->sha1base36, 36, 16 );
}
return false;
}
+
+ /**
+ * @return string
+ */
function getFileSrc() {
return $this->fileSrc;
}
+
+ /**
+ * @return bool
+ */
function isTempSrc() {
return $this->isTemp;
}
+ /**
+ * @return mixed
+ */
function getFilename() {
return $this->filename;
}
+
+ /**
+ * @return string
+ */
function getArchiveName() {
return $this->archiveName;
}
+ /**
+ * @return mixed
+ */
function getSize() {
return $this->size;
}
+ /**
+ * @return string
+ */
function getType() {
return $this->type;
}
+ /**
+ * @return string
+ */
function getAction() {
return $this->action;
}
+ /**
+ * @return string
+ */
function getParams() {
return $this->params;
}
+ /**
+ * @return bool
+ */
function importOldRevision() {
$dbw = wfGetDB( DB_MASTER );
$linkCache = LinkCache::singleton();
$linkCache->clear();
- $article = new Article( $this->title );
- $pageId = $article->getId();
- if( $pageId == 0 ) {
+ $page = WikiPage::factory( $this->title );
+ if( !$page->exists() ) {
# must create the page...
- $pageId = $article->insertOn( $dbw );
+ $pageId = $page->insertOn( $dbw );
$created = true;
$oldcountable = null;
} else {
+ $pageId = $page->getId();
$created = false;
$prior = $dbw->selectField( 'revision', '1',
$this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
return false;
}
- $oldcountable = $article->isCountable();
+ $oldcountable = $page->isCountable();
}
# @todo FIXME: Use original rev_id optionally (better for backups)
'minor_edit' => $this->minor,
) );
$revision->insertOn( $dbw );
- $changed = $article->updateIfNewerOn( $dbw, $revision );
+ $changed = $page->updateIfNewerOn( $dbw, $revision );
if ( $changed !== false && !$this->mNoUpdates ) {
wfDebug( __METHOD__ . ": running updates\n" );
- $article->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
+ $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
}
return true;
}
+ /**
+ * @return mixed
+ */
function importLogItem() {
$dbw = wfGetDB( DB_MASTER );
# @todo FIXME: This will not record autoblocks
if( $prior ) {
wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
$this->timestamp . "\n" );
- return false;
+ return;
}
$log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
$data = array(
$dbw->insert( 'logging', $data, __METHOD__ );
}
+ /**
+ * @return bool
+ */
function importUpload() {
# Construct a file
$archiveName = $this->getArchiveName();
if ( $archiveName ) {
wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
- $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
- RepoGroup::singleton()->getLocalRepo(), $archiveName );
+ $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
+ RepoGroup::singleton()->getLocalRepo(), $archiveName );
} else {
$file = wfLocalFile( $this->getTitle() );
wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
$archiveName = $file->getTimestamp() . '!' . $file->getName();
- $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
+ $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
RepoGroup::singleton()->getLocalRepo(), $archiveName );
wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
}
wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
return false;
}
-
+
# Get the file source or download if necessary
$source = $this->getFileSrc();
$flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
}
$user = User::newFromName( $this->user_text );
-
+
# Do the actual upload
if ( $archiveName ) {
- $status = $file->uploadOld( $source, $archiveName,
+ $status = $file->uploadOld( $source, $archiveName,
$this->getTimestamp(), $this->getComment(), $user, $flags );
} else {
- $status = $file->upload( $source, $this->getComment(), $this->getComment(),
+ $status = $file->upload( $source, $this->getComment(), $this->getComment(),
$flags, false, $this->getTimestamp(), $user );
}
-
+
if ( $status->isGood() ) {
wfDebug( __METHOD__ . ": Succesful\n" );
return true;
}
}
+ /**
+ * @return bool|string
+ */
function downloadSource() {
global $wgEnableUploads;
if( !$wgEnableUploads ) {
$this->mRead = false;
}
+ /**
+ * @return bool
+ */
function atEnd() {
return $this->mRead;
}
+ /**
+ * @return bool|string
+ */
function readChunk() {
if( $this->atEnd() ) {
return false;
- } else {
- $this->mRead = true;
- return $this->mString;
}
+ $this->mRead = true;
+ return $this->mString;
}
}
$this->mHandle = $handle;
}
+ /**
+ * @return bool
+ */
function atEnd() {
return feof( $this->mHandle );
}
+ /**
+ * @return string
+ */
function readChunk() {
return fread( $this->mHandle, 32768 );
}
+ /**
+ * @param $filename string
+ * @return Status
+ */
static function newFromFile( $filename ) {
wfSuppressWarnings();
$file = fopen( $filename, 'rt' );
return Status::newGood( new ImportStreamSource( $file ) );
}
+ /**
+ * @param $fieldname string
+ * @return Status
+ */
static function newFromUpload( $fieldname = "xmlimport" ) {
$upload =& $_FILES[$fieldname];
}
}
+ /**
+ * @param $url
+ * @param $method string
+ * @return Status
+ */
static function newFromURL( $url, $method = 'GET' ) {
wfDebug( __METHOD__ . ": opening $url\n" );
# Use the standard HTTP fetch function; it times out
# quicker and sorts out user-agent problems which might
# otherwise prevent importing from large sites, such
# as the Wikimedia cluster, etc.
- $data = Http::request( $method, $url );
+ $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
if( $data !== false ) {
$file = tmpfile();
fwrite( $file, $data );
}
}
+ /**
+ * @param $interwiki
+ * @param $page
+ * @param $history bool
+ * @param $templates bool
+ * @param $pageLinkDepth int
+ * @return Status
+ */
public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
if( $page == '' ) {
return Status::newFatal( 'import-noarticle' );
dépôts / lhc / web / wiklou.git / blobdiff