<?php
+// Some smart guy removed XMLReader's global constants from PHP 5.1
+// and replaced them with class constants. Breaking source compatibility
+// is SUPER awesome, and I love languages which do this constantly!
+$xmlReaderConstants = array(
+ "NONE",
+ "ELEMENT",
+ "ATTRIBUTE",
+ "TEXT",
+ "CDATA",
+ "ENTITY_REF",
+ "ENTITY",
+ "PI",
+ "COMMENT",
+ "DOC",
+ "DOC_TYPE",
+ "DOC_FRAGMENT",
+ "NOTATION",
+ "WHITESPACE",
+ "SIGNIFICANT_WHITESPACE",
+ "END_ELEMENT",
+ "END_ENTITY",
+ "XML_DECLARATION",
+ "LOADDTD",
+ "DEFAULTATTRS",
+ "VALIDATE",
+ "SUBST_ENTITIES" );
+foreach ( $xmlReaderConstants as $name ) {
+ $fullName = "XMLREADER_$name";
+ $newName = "XMLReader::$name";
+ if ( !defined( $fullName ) ) {
+ if ( defined( $newName ) ) {
+ define( $fullName, constant( $newName ) );
+ } else {
+ // broken or missing the extension...
+ }
+ }
+}
+
/**
* Readahead helper for making large MediaWiki data dumps;
* reads in a previous XML dump to sequentially prefetch text
* - text contents are immutable and should not change once
* recorded, so the previous dump is a reliable source
*
- * Requires PHP 5 and the XMLReader PECL extension.
+ * Requires the XMLReader PECL extension.
+ * @ingroup Maintenance
*/
class BaseDump {
var $reader = null;
var $atEnd = false;
+ var $atPageEnd = false;
var $lastPage = 0;
var $lastRev = 0;
-
+
function BaseDump( $infile ) {
$this->reader = new XMLReader();
$this->reader->open( $infile );
}
-
+
/**
* Attempts to fetch the text of a particular page revision
* from the dump stream. May return null if the page is
* unavailable.
*
- * @param int $page ID number of page to read
- * @param int $rev ID number of revision to read
+ * @param $page Integer: ID number of page to read
+ * @param $rev Integer: ID number of revision to read
* @return string or null
*/
function prefetch( $page, $rev ) {
- while( $this->lastPage < $page && !$this->atEnd ) {
+ $page = intval( $page );
+ $rev = intval( $rev );
+ while ( $this->lastPage < $page && !$this->atEnd ) {
+ $this->debug( "BaseDump::prefetch at page $this->lastPage, looking for $page" );
$this->nextPage();
}
- if( $this->lastPage > $page || $this->atEnd ) {
- $this->debug( "BaseDump::prefetch already past page $page looking for rev $rev\n" );
+ if ( $this->lastPage > $page || $this->atEnd ) {
+ $this->debug( "BaseDump::prefetch already past page $page looking for rev $rev [$this->lastPage, $this->lastRev]" );
return null;
}
- while( $this->lastRev < $rev && !$this->atEnd ) {
+ while ( $this->lastRev < $rev && !$this->atEnd && !$this->atPageEnd ) {
+ $this->debug( "BaseDump::prefetch at page $this->lastPage, rev $this->lastRev, looking for $page, $rev" );
$this->nextRev();
}
- if( $this->lastRev == $rev ) {
- $this->debug( "BaseDump::prefetch hit on $page, $rev\n" );
+ if ( $this->lastRev == $rev && !$this->atEnd ) {
+ $this->debug( "BaseDump::prefetch hit on $page, $rev [$this->lastPage, $this->lastRev]" );
return $this->nextText();
} else {
- $this->debug( "BaseDump::prefetch already past rev $rev on page $page\n" );
+ $this->debug( "BaseDump::prefetch already past rev $rev on page $page [$this->lastPage, $this->lastRev]" );
return null;
}
}
-
+
function debug( $str ) {
- wfDebug( $str );
- //global $dumper;
- //$dumper->progress( $str );
+ wfDebug( $str . "\n" );
+ // global $dumper;
+ // $dumper->progress( $str );
}
-
+
/**
* @access private
*/
function nextPage() {
- $this->skipTo( 'page' );
- $this->skipTo( 'id' );
- $this->lastPage = intval( $this->nodeContents() );
- $this->lastRev = 0;
+ if ( $this->skipTo( 'page', 'mediawiki' ) ) {
+ if ( $this->skipTo( 'id' ) ) {
+ $this->lastPage = intval( $this->nodeContents() );
+ $this->lastRev = 0;
+ $this->atPageEnd = false;
+ }
+ } else {
+ $this->atEnd = true;
+ }
}
-
+
/**
* @access private
*/
function nextRev() {
- $this->skipTo( 'revision' );
- $this->skipTo( 'id' );
- $this->lastRev = intval( $this->nodeContents() );
+ if ( $this->skipTo( 'revision' ) ) {
+ if ( $this->skipTo( 'id' ) ) {
+ $this->lastRev = intval( $this->nodeContents() );
+ }
+ } else {
+ $this->atPageEnd = true;
+ }
}
-
+
/**
* @access private
*/
$this->skipTo( 'text' );
return strval( $this->nodeContents() );
}
-
+
/**
* @access private
*/
- function skipTo( $name ) {
- while( $this->reader->read() ) {
- if( $this->reader->nodeType == XMLREADER_ELEMENT &&
+ function skipTo( $name, $parent = 'page' ) {
+ if ( $this->atEnd ) {
+ return false;
+ }
+ while ( $this->reader->read() ) {
+ if ( $this->reader->nodeType == XMLREADER_ELEMENT &&
$this->reader->name == $name ) {
return true;
}
+ if ( $this->reader->nodeType == XMLREADER_END_ELEMENT &&
+ $this->reader->name == $parent ) {
+ $this->debug( "BaseDump::skipTo found </$parent> searching for <$name>" );
+ return false;
+ }
}
return $this->close();
}
-
+
/**
* Shouldn't something like this be built-in to XMLReader?
* Fetches text contents of the current element, assuming
* no sub-elements or such scary things.
- * @return string
+ *
+ * @return String
* @access private
*/
function nodeContents() {
- if( $this->reader->isEmptyElement ) {
+ if ( $this->atEnd ) {
+ return null;
+ }
+ if ( $this->reader->isEmptyElement ) {
return "";
}
$buffer = "";
- while( $this->reader->read() ) {
+ while ( $this->reader->read() ) {
switch( $this->reader->nodeType ) {
case XMLREADER_TEXT:
// case XMLREADER_WHITESPACE:
}
return $this->close();
}
-
+
/**
* @access private
*/
function close() {
$this->reader->close();
- $this->reader = null;
$this->atEnd = true;
- return false;
+ return null;
}
}
-
-?>