X-Git-Url: http://git.heureux-cyclage.org/?a=blobdiff_plain;f=maintenance%2FrenderDump.php;h=78c5b6f31e0e8909e766db3d404add4ebf834ad2;hb=b17e4b4a24a317b5ea8b0c7687c695cce783c154;hp=f8b97e7c9f05dcb07ee87d67b1133d88d190d6c0;hpb=72a4abe588a7511ed5a92a5e30f889c60d824c1a;p=lhc%2Fweb%2Fwiklou.git
diff --git a/maintenance/renderDump.php b/maintenance/renderDump.php
index f8b97e7c9f..78c5b6f31e 100644
--- a/maintenance/renderDump.php
+++ b/maintenance/renderDump.php
@@ -24,44 +24,81 @@
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
- * @addtogroup Maintenance
+ * @file
+ * @ingroup Maintenance
*/
-$optionsWithArgs = array( 'report' );
+require_once( dirname( __FILE__ ) . '/Maintenance.php' );
-require_once( 'commandLine.inc' );
+class DumpRenderer extends Maintenance {
-class DumpRenderer {
- function __construct( $dir ) {
- $this->stderr = fopen( "php://stderr", "wt" );
- $this->outputDirectory = $dir;
- $this->count = 0;
+ private $count = 0;
+ private $outputDirectory, $startTime;
+
+ public function __construct() {
+ parent::__construct();
+ $this->mDescription = "Take page text out of an XML dump file and render basic HTML out to files";
+ $this->addOption( 'output-dir', 'The directory to output the HTML files to', true, true );
+ $this->addOption( 'prefix', 'Prefix for the rendered files (defaults to wiki)', false, true );
+ $this->addOption( 'parser', 'Use an alternative parser class', false, true );
}
- function handleRevision( $rev ) {
+ public function execute() {
+ $this->outputDirectory = $this->getOption( 'output-dir' );
+ $this->prefix = $this->getOption( 'prefix', 'wiki' );
+ $this->startTime = wfTime();
+
+ if ( $this->hasOption( 'parser' ) ) {
+ global $wgParserConf;
+ $wgParserConf['class'] = $this->getOption( 'parser' );
+ $this->prefix .= "-{$wgParserConf['class']}";
+ }
+
+ $source = new ImportStreamSource( $this->getStdin() );
+ $importer = new WikiImporter( $source );
+
+ $importer->setRevisionCallback(
+ array( &$this, 'handleRevision' ) );
+
+ $importer->doImport();
+
+ $delta = wfTime() - $this->startTime;
+ $this->error( "Rendered {$this->count} pages in " . round($delta, 2) . " seconds " );
+ if ($delta > 0)
+ $this->error( round($this->count / $delta, 2) . " pages/sec" );
+ $this->error( "\n" );
+ }
+
+ /**
+ * Callback function for each revision, turn into HTML and save
+ * @param $rev Revision
+ */
+ public function handleRevision( $rev ) {
+ global $wgParserConf;
+
$title = $rev->getTitle();
- if (!$title) {
- fprintf( $this->stderr, "Got bogus revision with null title!" );
+ if ( !$title ) {
+ $this->error( "Got bogus revision with null title!" );
return;
}
$display = $title->getPrefixedText();
-
+
$this->count++;
-
+
$sanitized = rawurlencode( $display );
- $filename = sprintf( "%s/wiki-%07d-%s.html",
+ $filename = sprintf( "%s/%s-%07d-%s.html",
$this->outputDirectory,
+ $this->prefix,
$this->count,
$sanitized );
- fprintf( $this->stderr, "%s\n", $filename, $display );
-
- // fixme
+ $this->output( sprintf( "%s\n", $filename, $display ) );
+
$user = new User();
- $parser = new Parser();
+ $parser = new $wgParserConf['class']();
$options = ParserOptions::newFromUser( $user );
-
+
$output = $parser->parse( $rev->getText(), $title, $options );
-
+
file_put_contents( $filename,
"\n" .
@@ -69,33 +106,13 @@ class DumpRenderer {
"\n" .
"\n" .
"" . htmlspecialchars( $display ) . "\n" .
- "\n" .
+ "\n" .
"\n" .
$output->getText() .
"\n" .
"