add option to XML dump stubs of page ranges with explicit rev_id ordering

tested for stubs, text, logging with and without start/end values,
with and without orderrevs, seems to work as expected, with the
appropriate changes to the query.

Bug: T29112
Change-Id: I94ca4a06235bdbed384bb997deb7432bb5aaa5b9

diff --git a/includes/export/WikiExporter.php b/includes/export/WikiExporter.php
index 54de26d..c1f2d59 100644 (file)
--- a/includes/export/WikiExporter.php
+++ b/includes/export/WikiExporter.php
@@ -134,13 +134,21 @@ class WikiExporter {
         * @param int $start Inclusive lower limit (this id is included)
         * @param int $end Exclusive upper limit (this id is not included)
         *   If 0, no upper limit.
+        * @param bool $orderRevs order revisions within pages in ascending order
         */
-       public function pagesByRange( $start, $end ) {
-               $condition = 'page_id >= ' . intval( $start );
-               if ( $end ) {
-                       $condition .= ' AND page_id < ' . intval( $end );
+       public function pagesByRange( $start, $end, $orderRevs ) {
+               if ( $orderRevs ) {
+                       $condition = 'rev_page >= ' . intval( $start );
+                       if ( $end ) {
+                               $condition .= ' AND rev_page < ' . intval( $end );
+                       }
+               } else {
+                       $condition = 'page_id >= ' . intval( $start );
+                       if ( $end ) {
+                               $condition .= ' AND page_id < ' . intval( $end );
+                       }
                }
-               $this->dumpFrom( $condition );
+               $this->dumpFrom( $condition, $orderRevs );
        }
 
        /**
@@ -245,7 +253,7 @@ class WikiExporter {
         * @throws MWException
         * @throws Exception
         */
-       protected function dumpFrom( $cond = '' ) {
+       protected function dumpFrom( $cond = '', $orderRevs = false ) {
                # For logging dumps...
                if ( $this->history & self::LOGS ) {
                        $where = [ 'user_id = log_user' ];
@@ -332,7 +340,16 @@ class WikiExporter {
                                }
                        } elseif ( $this->history & WikiExporter::FULL ) {
                                # Full history dumps...
-                               $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
+                               # query optimization for history stub dumps
+                               if ( $this->text == WikiExporter::STUB && $orderRevs ) {
+                                       $tables = [ 'revision', 'page' ];
+                                       $opts[] = 'STRAIGHT_JOIN';
+                                       $opts['ORDER BY'] = [ 'rev_page ASC', 'rev_id ASC' ];
+                                       $opts['USE INDEX']['revision'] = 'rev_page_id';
+                                       $join['page'] = [ 'INNER JOIN', 'rev_page=page_id' ];
+                               } else {
+                                       $join['revision'] = [ 'INNER JOIN', 'page_id=rev_page' ];
+                               }
                        } elseif ( $this->history & WikiExporter::CURRENT ) {
                                # Latest revision dumps...
                                if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
@@ -369,7 +386,6 @@ class WikiExporter {
                        if ( $this->buffer == WikiExporter::STREAM ) {
                                $prev = $this->db->bufferResults( false );
                        }
-
                        $result = null; // Assuring $result is not undefined, if exception occurs early
                        try {
                                Hooks::run( 'ModifyExportQuery',

diff --git a/maintenance/backup.inc b/maintenance/backup.inc
index 3271fd6..db3af92 100644 (file)
--- a/maintenance/backup.inc
+++ b/maintenance/backup.inc
@@ -41,6 +41,7 @@ class BackupDumper extends Maintenance {
        public $revEndId = 0;
        public $dumpUploads = false;
        public $dumpUploadFileContents = false;
+       public $orderRevs = false;
 
        protected $reportingInterval = 100;
        protected $pageCount = 0;
@@ -271,7 +272,7 @@ class BackupDumper extends Maintenance {
                } elseif ( is_null( $this->pages ) ) {
                        # Page dumps: all or by page ID range
                        if ( $this->startId || $this->endId ) {
-                               $exporter->pagesByRange( $this->startId, $this->endId );
+                               $exporter->pagesByRange( $this->startId, $this->endId, $this->orderRevs );
                        } elseif ( $this->revStartId || $this->revEndId ) {
                                $exporter->revsByRange( $this->revStartId, $this->revEndId );
                        } else {

diff --git a/maintenance/dumpBackup.php b/maintenance/dumpBackup.php
index d4255a0..9bf1222 100644 (file)
--- a/maintenance/dumpBackup.php
+++ b/maintenance/dumpBackup.php
@@ -50,6 +50,8 @@ TEXT
                $this->addOption( 'stable', 'Dump stable versions of pages' );
                $this->addOption( 'revrange', 'Dump range of revisions specified by revstart and ' .
                        'revend parameters' );
+               $this->addOption( 'orderrevs', 'Dump revisions in ascending revision order ' .
+                       '(implies dump of a range of pages)' );
                $this->addOption( 'pagelist',
                        'Dump only pages included in the file', false, true );
                // Options
@@ -127,6 +129,7 @@ TEXT
                $this->skipFooter = $this->hasOption( 'skip-footer' );
                $this->dumpUploads = $this->hasOption( 'uploads' );
                $this->dumpUploadFileContents = $this->hasOption( 'include-files' );
+               $this->orderRevs = $this->hasOption( 'orderrevs' );
        }
 }