# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
# http://www.gnu.org/copyleft/gpl.html
+/**
+ * @defgroup Dump Dump
+ */
/**
- *
- * @addtogroup SpecialPage
+ * @ingroup SpecialPage Dump
*/
class WikiExporter {
var $list_authors = false ; # Return distinct author list (when not returning full history)
var $author_list = "" ;
+ var $dumpUploads = false;
+
const FULL = 0;
const CURRENT = 1;
+ const LOGS = 2;
const BUFFER = 0;
const STREAM = 1;
* make additional queries to pull source data while the
* main query is still running.
*
- * @param Database $db
- * @param mixed $history one of WikiExporter::FULL or WikiExporter::CURRENT, or an
- * associative array:
- * offset: non-inclusive offset at which to start the query
- * limit: maximum number of rows to return
- * dir: "asc" or "desc" timestamp order
- * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
+ * @param $db Database
+ * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
+ * or an associative array:
+ * offset: non-inclusive offset at which to start the query
+ * limit: maximum number of rows to return
+ * dir: "asc" or "desc" timestamp order
+ * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
*/
function __construct( &$db, $history = WikiExporter::CURRENT,
$buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
* various row objects and XML output for filtering. Filters
* can be chained or used as callbacks.
*
- * @param mixed $callback
+ * @param $sink mixed
*/
function setOutputSink( &$sink ) {
$this->sink =& $sink;
/**
* Dumps a series of page and revision records for those pages
* in the database falling within the page_id range given.
- * @param int $start Inclusive lower limit (this id is included)
- * @param int $end Exclusive upper limit (this id is not included)
+ * @param $start Int: inclusive lower limit (this id is included)
+ * @param $end Int: Exclusive upper limit (this id is not included)
* If 0, no upper limit.
*/
function pagesByRange( $start, $end ) {
}
return $this->dumpFrom( $condition );
}
+
+ function allLogs() {
+ return $this->dumpFrom( '' );
+ }
+
+ function logsByRange( $start, $end ) {
+ $condition = 'log_id >= ' . intval( $start );
+ if( $end ) {
+ $condition .= ' AND log_id < ' . intval( $end );
+ }
+ return $this->dumpFrom( $condition );
+ }
/**
- * @param Title $title
+ * @param $title Title
*/
function pageByTitle( $title ) {
return $this->dumpFrom(
'page_namespace=' . $title->getNamespace() .
- ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) );
+ ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
}
function pageByName( $name ) {
$this->author_list = "<contributors>";
//rev_deleted
$nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
-
+
$sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND $nothidden AND " . $cond ;
$result = $this->db->query( $sql, $fname );
$resultset = $this->db->resultObject( $result );
while( $row = $resultset->fetchObject() ) {
- $this->author_list .= "<contributor>" .
- "<username>" .
- htmlentities( $row->rev_user_text ) .
- "</username>" .
- "<id>" .
+ $this->author_list .= "<contributor>" .
+ "<username>" .
+ htmlentities( $row->rev_user_text ) .
+ "</username>" .
+ "<id>" .
$row->rev_user .
- "</id>" .
+ "</id>" .
"</contributor>";
}
wfProfileOut( $fname );
function dumpFrom( $cond = '' ) {
$fname = 'WikiExporter::dumpFrom';
wfProfileIn( $fname );
-
+
+ # For logs dumps...
+ if( $this->history & self::LOGS ) {
+ $where = array( 'user_id = log_user' );
+ # Hide private logs
+ $where[] = LogEventsList::getExcludeClause( $this->db );
+ if( $cond ) $where[] = $cond;
+ $result = $this->db->select( array('logging','user'),
+ '*',
+ $where,
+ $fname,
+ array( 'ORDER BY' => 'log_id')
+ );
+ $wrapper = $this->db->resultObject( $result );
+ $this->outputLogStream( $wrapper );
+ wfProfileOut( $fname );
+ return;
+ }
+ # For page dumps...
$page = $this->db->tableName( 'page' );
$revision = $this->db->tableName( 'revision' );
$text = $this->db->tableName( 'text' );
}
$result = $this->db->query( $sql, $fname );
$wrapper = $this->db->resultObject( $result );
- $this->outputStream( $wrapper );
+ $this->outputPageStream( $wrapper );
if ( $this->list_authors ) {
- $this->outputStream( $wrapper );
+ $this->outputPageStream( $wrapper );
}
if( $this->buffer == WikiExporter::STREAM ) {
* separate database connection not managed by LoadBalancer; some
* blob storage types will make queries to pull source data.
*
- * @param ResultWrapper $resultset
+ * @param $resultset ResultWrapper
* @access private
*/
- function outputStream( $resultset ) {
+ function outputPageStream( $resultset ) {
$last = null;
while( $row = $resultset->fetchObject() ) {
if( is_null( $last ) ||
$last->page_namespace != $row->page_namespace ||
$last->page_title != $row->page_title ) {
if( isset( $last ) ) {
- $output = $this->writer->closePage();
+ $output = '';
+ if( $this->dumpUploads ) {
+ $output .= $this->writer->writeUploads( $last );
+ }
+ $output .= $this->writer->closePage();
$this->sink->writeClosePage( $output );
}
$output = $this->writer->openPage( $row );
$this->sink->writeRevision( $row, $output );
}
if( isset( $last ) ) {
- $output = $this->author_list . $this->writer->closePage();
+ $output = '';
+ if( $this->dumpUploads ) {
+ $output .= $this->writer->writeUploads( $last );
+ }
+ $output .= $this->author_list;
+ $output .= $this->writer->closePage();
$this->sink->writeClosePage( $output );
}
$resultset->free();
}
+
+ function outputLogStream( $resultset ) {
+ while( $row = $resultset->fetchObject() ) {
+ $output = $this->writer->writeLogItem( $row );
+ $this->sink->writeLogItem( $row, $output );
+ }
+ $resultset->free();
+ }
}
/**
- * @addtogroup Dump
+ * @ingroup Dump
*/
class XmlDumpWriter {
* Opens a <page> section on the output stream, with data
* from the given database row.
*
- * @param object $row
+ * @param $row object
* @return string
* @access private
*/
* Dumps a <revision> section on the output stream, with
* data filled in from the given database row.
*
- * @param object $row
+ * @param $row object
* @return string
* @access private
*/
$out = " <revision>\n";
$out .= " " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
- $ts = wfTimestamp( TS_ISO_8601, $row->rev_timestamp );
- $out .= " " . wfElement( 'timestamp', null, $ts ) . "\n";
+ $out .= $this->writeTimestamp( $row->rev_timestamp );
if( $row->rev_deleted & Revision::DELETED_USER ) {
$out .= " " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
} else {
- $out .= " <contributor>\n";
- if( $row->rev_user ) {
- $out .= " " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n";
- $out .= " " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n";
- } else {
- $out .= " " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n";
- }
- $out .= " </contributor>\n";
+ $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
}
if( $row->rev_minor_edit ) {
wfProfileOut( $fname );
return $out;
}
+
+ /**
+ * Dumps a <logitem> section on the output stream, with
+ * data filled in from the given database row.
+ *
+ * @param $row object
+ * @return string
+ * @access private
+ */
+ function writeLogItem( $row ) {
+ $fname = 'WikiExporter::writeLogItem';
+ wfProfileIn( $fname );
+
+ $out = " <logitem>\n";
+ $out .= " " . wfElement( 'id', null, strval( $row->log_id ) ) . "\n";
+
+ $out .= $this->writeTimestamp( $row->log_timestamp );
+
+ if( $row->log_deleted & LogPage::DELETED_USER ) {
+ $out .= " " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
+ } else {
+ $out .= $this->writeContributor( $row->log_user, $row->user_name );
+ }
+
+ if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
+ $out .= " " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
+ } elseif( $row->log_comment != '' ) {
+ $out .= " " . wfElementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
+ }
+
+ $out .= " " . wfElement( 'type', null, strval( $row->log_type ) ) . "\n";
+ $out .= " " . wfElement( 'action', null, strval( $row->log_action ) ) . "\n";
+
+ if( $row->log_deleted & LogPage::DELETED_ACTION ) {
+ $out .= " " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
+ } else {
+ $title = Title::makeTitle( $row->log_namespace, $row->log_title );
+ $out .= " " . wfElementClean( 'title', null, $title->getPrefixedText() ) . "\n";
+ $out .= " " . wfElementClean( 'params',
+ array( 'xml:space' => 'preserve' ),
+ strval( $row->log_params ) ) . "\n";
+ }
+
+ $out .= " </logitem>\n";
+
+ wfProfileOut( $fname );
+ return $out;
+ }
+
+ function writeTimestamp( $timestamp ) {
+ $ts = wfTimestamp( TS_ISO_8601, $timestamp );
+ return " " . wfElement( 'timestamp', null, $ts ) . "\n";
+ }
+
+ function writeContributor( $id, $text ) {
+ $out = " <contributor>\n";
+ if( $id ) {
+ $out .= " " . wfElementClean( 'username', null, strval( $text ) ) . "\n";
+ $out .= " " . wfElement( 'id', null, strval( $id ) ) . "\n";
+ } else {
+ $out .= " " . wfElementClean( 'ip', null, strval( $text ) ) . "\n";
+ }
+ $out .= " </contributor>\n";
+ return $out;
+ }
+
+ /**
+ * Warning! This data is potentially inconsistent. :(
+ */
+ function writeUploads( $row ) {
+ if( $row->page_namespace == NS_IMAGE ) {
+ $img = wfFindFile( $row->page_title );
+ if( $img ) {
+ $out = '';
+ foreach( array_reverse( $img->getHistory() ) as $ver ) {
+ $out .= $this->writeUpload( $ver );
+ }
+ $out .= $this->writeUpload( $img );
+ return $out;
+ }
+ }
+ return '';
+ }
+
+ function writeUpload( $file ) {
+ return " <upload>\n" .
+ $this->writeTimestamp( $file->getTimestamp() ) .
+ $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
+ " " . wfElementClean( 'comment', null, $file->getDescription() ) . "\n" .
+ " " . wfElement( 'filename', null, $file->getName() ) . "\n" .
+ " " . wfElement( 'src', null, $file->getFullUrl() ) . "\n" .
+ " " . wfElement( 'size', null, $file->getSize() ) . "\n" .
+ " </upload>\n";
+ }
}
/**
* Base class for output stream; prints to stdout or buffer or whereever.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpOutput {
function writeOpenStream( $string ) {
function writeRevision( $rev, $string ) {
$this->write( $string );
}
+
+ function writeLogItem( $rev, $string ) {
+ $this->write( $string );
+ }
/**
* Override to write to a different stream type.
/**
* Stream outputter to send data to a file.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpFileOutput extends DumpOutput {
var $handle;
* Stream outputter to send data to a file via some filter program.
* Even if compression is available in a library, using a separate
* program can allow us to make use of a multi-processor system.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpPipeOutput extends DumpFileOutput {
function DumpPipeOutput( $command, $file = null ) {
/**
* Sends dump output via the gzip compressor.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpGZipOutput extends DumpPipeOutput {
function DumpGZipOutput( $file ) {
/**
* Sends dump output via the bgzip2 compressor.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpBZip2Output extends DumpPipeOutput {
function DumpBZip2Output( $file ) {
/**
* Sends dump output via the p7zip compressor.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class Dump7ZipOutput extends DumpPipeOutput {
function Dump7ZipOutput( $file ) {
* Dump output filter class.
* This just does output filtering and streaming; XML formatting is done
* higher up, so be careful in what you do.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpFilter {
function DumpFilter( &$sink ) {
$this->sink->writeRevision( $rev, $string );
}
}
+
+ function writeLogItem( $rev, $string ) {
+ $this->sink->writeRevision( $rev, $string );
+ }
/**
* Override for page-based filter types.
/**
* Simple dump output filter to exclude all talk pages.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpNotalkFilter extends DumpFilter {
function pass( $page ) {
- return !Namespace::isTalk( $page->page_namespace );
+ return !MWNamespace::isTalk( $page->page_namespace );
}
}
/**
* Dump output filter to include or exclude pages in a given set of namespaces.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpNamespaceFilter extends DumpFilter {
var $invert = false;
/**
* Dump output filter to include only the last revision in each page sequence.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpLatestFilter extends DumpFilter {
var $page, $pageString, $rev, $revString;
/**
* Base class for output stream; prints to stdout or buffer or whereever.
- * @addtogroup Dump
+ * @ingroup Dump
*/
class DumpMultiWriter {
function DumpMultiWriter( $sinks ) {
wfProfileOut( $fname );
return $string;
}
-
-
dépôts / lhc / web / wiklou.git / blobdiff