export.php

来自「php 开发的内容管理系统」· PHP 代码 · 共 737 行 · 第 1/2 页

	/**
	 * Closes a <page> section on the output stream.
	 *
	 * @access private
	 */
	function closePage() {
		return "  </page>\n";
	}

	/**
	 * Dumps a <revision> section on the output stream, with
	 * data filled in from the given database row.
	 *
	 * @param object $row
	 * @return string
	 * @access private
	 */
	function writeRevision( $row ) {
		$fname = 'WikiExporter::dumpRev';
		wfProfileIn( $fname );

		$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";

		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";
		}

		if( $row->rev_minor_edit ) {
			$out .=  "      <minor/>\n";
		}
		if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
			$out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
		} elseif( $row->rev_comment != '' ) {
			$out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
		}

		if( $row->rev_deleted & Revision::DELETED_TEXT ) {
			$out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
		} elseif( isset( $row->old_text ) ) {
			// Raw text from the database may have invalid chars
			$text = strval( Revision::getRevisionText( $row ) );
			$out .= "      " . wfElementClean( 'text',
				array( 'xml:space' => 'preserve' ),
				strval( $text ) ) . "\n";
		} else {
			// Stub output
			$out .= "      " . wfElement( 'text',
				array( 'id' => $row->rev_text_id ),
				"" ) . "\n";
		}

		$out .= "    </revision>\n";

		wfProfileOut( $fname );
		return $out;
	}

}


/**
 * Base class for output stream; prints to stdout or buffer or whereever.
 */
class DumpOutput {
	function writeOpenStream( $string ) {
		$this->write( $string );
	}

	function writeCloseStream( $string ) {
		$this->write( $string );
	}

	function writeOpenPage( $page, $string ) {
		$this->write( $string );
	}

	function writeClosePage( $string ) {
		$this->write( $string );
	}

	function writeRevision( $rev, $string ) {
		$this->write( $string );
	}

	/**
	 * Override to write to a different stream type.
	 * @return bool
	 */
	function write( $string ) {
		print $string;
	}
}

/**
 * Stream outputter to send data to a file.
 */
class DumpFileOutput extends DumpOutput {
	var $handle;

	function DumpFileOutput( $file ) {
		$this->handle = fopen( $file, "wt" );
	}

	function write( $string ) {
		fputs( $this->handle, $string );
	}
}

/**
 * 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.
 */
class DumpPipeOutput extends DumpFileOutput {
	function DumpPipeOutput( $command, $file = null ) {
		if( !is_null( $file ) ) {
			$command .=  " > " . wfEscapeShellArg( $file );
		}
		$this->handle = popen( $command, "w" );
	}
}

/**
 * Sends dump output via the gzip compressor.
 */
class DumpGZipOutput extends DumpPipeOutput {
	function DumpGZipOutput( $file ) {
		parent::DumpPipeOutput( "gzip", $file );
	}
}

/**
 * Sends dump output via the bgzip2 compressor.
 */
class DumpBZip2Output extends DumpPipeOutput {
	function DumpBZip2Output( $file ) {
		parent::DumpPipeOutput( "bzip2", $file );
	}
}

/**
 * Sends dump output via the p7zip compressor.
 */
class Dump7ZipOutput extends DumpPipeOutput {
	function Dump7ZipOutput( $file ) {
		$command = "7za a -bd -si " . wfEscapeShellArg( $file );
		// Suppress annoying useless crap from p7zip
		// Unfortunately this could suppress real error messages too
		$command .= " >/dev/null 2>&1";
		parent::DumpPipeOutput( $command );
	}
}



/**
 * Dump output filter class.
 * This just does output filtering and streaming; XML formatting is done
 * higher up, so be careful in what you do.
 */
class DumpFilter {
	function DumpFilter( &$sink ) {
		$this->sink =& $sink;
	}

	function writeOpenStream( $string ) {
		$this->sink->writeOpenStream( $string );
	}

	function writeCloseStream( $string ) {
		$this->sink->writeCloseStream( $string );
	}

	function writeOpenPage( $page, $string ) {
		$this->sendingThisPage = $this->pass( $page, $string );
		if( $this->sendingThisPage ) {
			$this->sink->writeOpenPage( $page, $string );
		}
	}

	function writeClosePage( $string ) {
		if( $this->sendingThisPage ) {
			$this->sink->writeClosePage( $string );
			$this->sendingThisPage = false;
		}
	}

	function writeRevision( $rev, $string ) {
		if( $this->sendingThisPage ) {
			$this->sink->writeRevision( $rev, $string );
		}
	}

	/**
	 * Override for page-based filter types.
	 * @return bool
	 */
	function pass( $page, $string ) {
		return true;
	}
}

/**
 * Simple dump output filter to exclude all talk pages.
 */
class DumpNotalkFilter extends DumpFilter {
	function pass( $page ) {
		return !Namespace::isTalk( $page->page_namespace );
	}
}

/**
 * Dump output filter to include or exclude pages in a given set of namespaces.
 */
class DumpNamespaceFilter extends DumpFilter {
	var $invert = false;
	var $namespaces = array();

	function DumpNamespaceFilter( &$sink, $param ) {
		parent::DumpFilter( $sink );

		$constants = array(
			"NS_MAIN"           => NS_MAIN,
			"NS_TALK"           => NS_TALK,
			"NS_USER"           => NS_USER,
			"NS_USER_TALK"      => NS_USER_TALK,
			"NS_PROJECT"        => NS_PROJECT,
			"NS_PROJECT_TALK"   => NS_PROJECT_TALK,
			"NS_IMAGE"          => NS_IMAGE,
			"NS_IMAGE_TALK"     => NS_IMAGE_TALK,
			"NS_MEDIAWIKI"      => NS_MEDIAWIKI,
			"NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
			"NS_TEMPLATE"       => NS_TEMPLATE,
			"NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
			"NS_HELP"           => NS_HELP,
			"NS_HELP_TALK"      => NS_HELP_TALK,
			"NS_CATEGORY"       => NS_CATEGORY,
			"NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );

		if( $param{0} == '!' ) {
			$this->invert = true;
			$param = substr( $param, 1 );
		}

		foreach( explode( ',', $param ) as $key ) {
			$key = trim( $key );
			if( isset( $constants[$key] ) ) {
				$ns = $constants[$key];
				$this->namespaces[$ns] = true;
			} elseif( is_numeric( $key ) ) {
				$ns = intval( $key );
				$this->namespaces[$ns] = true;
			} else {
				throw new MWException( "Unrecognized namespace key '$key'\n" );
			}
		}
	}

	function pass( $page ) {
		$match = isset( $this->namespaces[$page->page_namespace] );
		return $this->invert xor $match;
	}
}


/**
 * Dump output filter to include only the last revision in each page sequence.
 */
class DumpLatestFilter extends DumpFilter {
	var $page, $pageString, $rev, $revString;

	function writeOpenPage( $page, $string ) {
		$this->page = $page;
		$this->pageString = $string;
	}

	function writeClosePage( $string ) {
		if( $this->rev ) {
			$this->sink->writeOpenPage( $this->page, $this->pageString );
			$this->sink->writeRevision( $this->rev, $this->revString );
			$this->sink->writeClosePage( $string );
		}
		$this->rev = null;
		$this->revString = null;
		$this->page = null;
		$this->pageString = null;
	}

	function writeRevision( $rev, $string ) {
		if( $rev->rev_id == $this->page->page_latest ) {
			$this->rev = $rev;
			$this->revString = $string;
		}
	}
}

/**
 * Base class for output stream; prints to stdout or buffer or whereever.
 */
class DumpMultiWriter {
	function DumpMultiWriter( $sinks ) {
		$this->sinks = $sinks;
		$this->count = count( $sinks );
	}

	function writeOpenStream( $string ) {
		for( $i = 0; $i < $this->count; $i++ ) {
			$this->sinks[$i]->writeOpenStream( $string );
		}
	}

	function writeCloseStream( $string ) {
		for( $i = 0; $i < $this->count; $i++ ) {
			$this->sinks[$i]->writeCloseStream( $string );
		}
	}

	function writeOpenPage( $page, $string ) {
		for( $i = 0; $i < $this->count; $i++ ) {
			$this->sinks[$i]->writeOpenPage( $page, $string );
		}
	}

	function writeClosePage( $string ) {
		for( $i = 0; $i < $this->count; $i++ ) {
			$this->sinks[$i]->writeClosePage( $string );
		}
	}

	function writeRevision( $rev, $string ) {
		for( $i = 0; $i < $this->count; $i++ ) {
			$this->sinks[$i]->writeRevision( $rev, $string );
		}
	}
}

function xmlsafe( $string ) {
	$fname = 'xmlsafe';
	wfProfileIn( $fname );

	/**
	 * The page may contain old data which has not been properly normalized.
	 * Invalid UTF-8 sequences or forbidden control characters will make our
	 * XML output invalid, so be sure to strip them out.
	 */
	$string = UtfNormal::cleanUp( $string );

	$string = htmlspecialchars( $string );
	wfProfileOut( $fname );
	return $string;
}

?>