Merge "Cache countable statistics to prevent multiple counting on import"

diff --git a/includes/Import.php b/includes/Import.php
index 7eff5da..eb2ca77 100644 (file)
--- a/includes/Import.php
+++ b/includes/Import.php
@@ -42,6 +42,8 @@ class WikiImporter {
        private $config;
        /** @var ImportTitleFactory */
        private $importTitleFactory;
+       /** @var array */
+       private $countableCache = array();
 
        /**
         * Creates an ImportXMLReader drawing from the source provided
@@ -67,6 +69,7 @@ class WikiImporter {
                }
 
                // Default callbacks
+               $this->setPageCallback( array( $this, 'beforeImportPage' ) );
                $this->setRevisionCallback( array( $this, "importRevision" ) );
                $this->setUploadCallback( array( $this, 'importUpload' ) );
                $this->setLogItemCallback( array( $this, 'importLogItem' ) );
@@ -288,6 +291,19 @@ class WikiImporter {
                $this->mImportUploads = $import;
        }
 
+       /**
+        * Default per-page callback. Sets up some things related to site statistics
+        * @param array $titleAndForeignTitle Two-element array, with Title object at
+        * index 0 and ForeignTitle object at index 1
+        * @return bool
+        */
+       public function beforeImportPage( $titleAndForeignTitle ) {
+               $title = $titleAndForeignTitle[0];
+               $page = WikiPage::factory( $title );
+               $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
+               return true;
+       }
+
        /**
         * Default per-revision callback, performs the import.
         * @param WikiRevision $revision
@@ -349,6 +365,26 @@ class WikiImporter {
         */
        public function finishImportPage( $title, $foreignTitle, $revCount,
                        $sRevCount, $pageInfo ) {
+
+               // Update article count statistics (T42009)
+               // The normal counting logic in WikiPage->doEditUpdates() is designed for
+               // one-revision-at-a-time editing, not bulk imports. In this situation it
+               // suffers from issues of slave lag. We let WikiPage handle the total page
+               // and revision count, and we implement our own custom logic for the
+               // article (content page) count.
+               $page = WikiPage::factory( $title );
+               $page->loadPageData( 'fromdbmaster' );
+               $content = $page->getContent();
+               $editInfo = $page->prepareContentForEdit( $content );
+
+               $countable = $page->isCountable( $editInfo );
+               $oldcountable = $this->countableCache['title_' . $title->getPrefixedText()];
+               if ( isset( $oldcountable ) && $countable != $oldcountable ) {
+                       DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
+                               'articles' => ( (int)$countable - (int)$oldcountable )
+                       ) ) );
+               }
+
                $args = func_get_args();
                return Hooks::run( 'AfterImportPage', $args );
        }
@@ -1544,7 +1580,6 @@ class WikiRevision {
                                        $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
                                return false;
                        }
-                       $oldcountable = $page->isCountable();
                }
 
                # @todo FIXME: Use original rev_id optionally (better for backups)
@@ -1567,10 +1602,11 @@ class WikiRevision {
 
                if ( $changed !== false && !$this->mNoUpdates ) {
                        wfDebug( __METHOD__ . ": running updates\n" );
+                       // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
                        $page->doEditUpdates(
                                $revision,
                                $userObj,
-                               array( 'created' => $created, 'oldcountable' => $oldcountable )
+                               array( 'created' => $created, 'oldcountable' => 'no-change' )
                        );
                }
 

diff --git a/includes/page/WikiPage.php b/includes/page/WikiPage.php
index 2f900f9..d30f589 100644 (file)
--- a/includes/page/WikiPage.php
+++ b/includes/page/WikiPage.php
@@ -2154,10 +2154,12 @@ class WikiPage implements Page, IDBAccessObject {
         * - changed: boolean, whether the revision changed the content (default true)
         * - created: boolean, whether the revision created the page (default false)
         * - moved: boolean, whether the page was moved (default false)
-        * - oldcountable: boolean or null (default null):
+        * - oldcountable: boolean, null, or string 'no-change' (default null):
         *   - boolean: whether the page was counted as an article before that
         *     revision, only used in changed is true and created is false
-        *   - null: don't change the article count
+        *   - null: if created is false, don't update the article count; if created
+        *     is true, do update the article count
+        *   - 'no-change': don't update the article count, ever
         */
        public function doEditUpdates( Revision $revision, User $user, array $options = array() ) {
                global $wgEnableParserCache;
@@ -2212,7 +2214,9 @@ class WikiPage implements Page, IDBAccessObject {
                $title = $this->mTitle->getPrefixedDBkey();
                $shortTitle = $this->mTitle->getDBkey();
 
-               if ( !$options['changed'] && !$options['moved'] ) {
+               if ( $options['oldcountable'] === 'no-change' ||
+                       ( !$options['changed'] && !$options['moved'] )
+               ) {
                        $good = 0;
                } elseif ( $options['created'] ) {
                        $good = (int)$this->isCountable( $editInfo );