Some experimental support for including upload data in XML export, as specced. Usable...

Some issues:
* doesn't respond to the current v full-history modes
* won't pick up file entries with no corresponding page
* on default config, you may end up with 'localhost' in URLs :P
* synchronization and mutability problems
** DB lookups are made in the middle of the dump and may not be in sync with state at the beginning of the dump.
** URL to current version at time of dump could point to a different, future version in future

diff --git a/includes/Export.php b/includes/Export.php
index 69d88fc..47a8823 100644 (file)
--- a/includes/Export.php
+++ b/includes/Export.php
@@ -25,6 +25,8 @@
 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;
@@ -263,7 +265,11 @@ class WikiExporter {
                                $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 );
@@ -274,7 +280,12 @@ class WikiExporter {
                        $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();
@@ -415,20 +426,12 @@ class XmlDumpWriter {
                $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 ) {
@@ -460,6 +463,52 @@ class XmlDumpWriter {
                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";
+       }
 
 }
 

diff --git a/maintenance/backup.inc b/maintenance/backup.inc
index 94fb48c..3fcb981 100644 (file)
--- a/maintenance/backup.inc
+++ b/maintenance/backup.inc
@@ -40,6 +40,7 @@ class BackupDumper {
        var $endId      = 0;
        var $sink       = null; // Output filters
        var $stubText   = false; // include rev_text_id instead of text; for 2-pass dump
+       var $dumpUploads = false;
 
        function BackupDumper( $args ) {
                $this->stderr = fopen( "php://stderr", "wt" );
@@ -177,6 +178,7 @@ class BackupDumper {
 
                $db = $this->backupDb();
                $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
+               $exporter->dumpUploads = $this->dumpUploads;
 
                $wrapper = new ExportProgressFilter( $this->sink, $this );
                $exporter->setOutputSink( $wrapper );

diff --git a/maintenance/dumpBackup.php b/maintenance/dumpBackup.php
index 9ec5674..6590fd4 100644 (file)
--- a/maintenance/dumpBackup.php
+++ b/maintenance/dumpBackup.php
@@ -54,6 +54,7 @@ if( isset( $options['end'] ) ) {
 }
 $dumper->skipHeader = isset( $options['skip-header'] );
 $dumper->skipFooter = isset( $options['skip-footer'] );
+$dumper->dumpUploads = isset( $options['uploads'] );
 
 $textMode = isset( $options['stub'] ) ? WikiExporter::STUB : WikiExporter::TEXT;
 
@@ -83,6 +84,7 @@ Options:
   --skip-header Don't output the <mediawiki> header
   --skip-footer Don't output the </mediawiki> footer
   --stub      Don't perform old_text lookups; for 2-pass dump
+  --uploads   Include upload records (experimental)
 
 Fancy stuff:
   --plugin=<class>[:<file>]   Load a dump plugin class