Add deleted archive titles search

Allows search engine to suggest deleted titles for undelete search.
Note that the titles are still verified against the archive table,
to ensure search engine is not out-of-date.

Bug: T109561
Change-Id: Id6099fe9fbf18481068a6f0a329bbde0d218135f

diff --git a/docs/hooks.txt b/docs/hooks.txt
index a38f9bb..bee4477 100644 (file)
--- a/docs/hooks.txt
+++ b/docs/hooks.txt
@@ -736,7 +736,10 @@ $current: the reverted revision
 $create: Whether or not the restoration caused the page to be created (i.e. it
   didn't exist before).
 $comment: The comment associated with the undeletion.
-$oldPageId: ID of page previously deleted (from archive table)
+$oldPageId: ID of page previously deleted (from archive table). This ID will be used
+  for the restored page.
+$restoredPages: Set of page IDs that have revisions restored for this undelete,
+  with keys being page IDs and values are 'true'.
 
 'ArticleUndeleteLogEntry': When a log entry is generated but not yet saved.
 $pageArchive: the PageArchive object

diff --git a/includes/page/PageArchive.php b/includes/page/PageArchive.php
index c2bfb07..188088f 100644 (file)
--- a/includes/page/PageArchive.php
+++ b/includes/page/PageArchive.php
@@ -67,6 +67,56 @@ class PageArchive {
                return self::listPages( $dbr, '' );
        }
 
+       /**
+        * List deleted pages recorded in the archive matching the
+        * given term, using search engine archive.
+        * Returns result wrapper with (ar_namespace, ar_title, count) fields.
+        *
+        * @param string $term Search term
+        * @return ResultWrapper
+        */
+       public static function listPagesBySearch( $term ) {
+               $title = Title::newFromText( $term );
+               if ( $title ) {
+                       $ns = $title->getNamespace();
+                       $termMain = $title->getText();
+                       $termDb = $title->getDBkey();
+               } else {
+                       // Prolly won't work too good
+                       // @todo handle bare namespace names cleanly?
+                       $ns = 0;
+                       $termMain = $termDb = $term;
+               }
+
+               // Try search engine first
+               $engine = MediaWikiServices::getInstance()->newSearchEngine();
+               $engine->setLimitOffset( 100 );
+               $engine->setNamespaces( [ $ns ] );
+               $results = $engine->searchArchiveTitle( $termMain );
+               if ( !$results->isOK() ) {
+                       $results = [];
+               } else {
+                       $results = $results->getValue();
+               }
+
+               if ( !$results ) {
+                       // Fall back to regular prefix search
+                       return self::listPagesByPrefix( $term );
+               }
+
+               $dbr = wfGetDB( DB_REPLICA );
+               $condTitles = array_unique( array_map( function ( Title $t ) {
+                       return $t->getDBkey();
+               }, $results ) );
+               $conds = [
+                       'ar_namespace' => $ns,
+                       $dbr->makeList( [ 'ar_title' => $condTitles ], LIST_OR ) . " OR ar_title " .
+                       $dbr->buildLike( $termDb, $dbr->anyString() )
+               ];
+
+               return self::listPages( $dbr, $conds );
+       }
+
        /**
         * List deleted pages recorded in the archive table matching the
         * given title prefix.
@@ -133,6 +183,7 @@ class PageArchive {
                $fields = [
                        'ar_minor_edit', 'ar_timestamp', 'ar_user', 'ar_user_text',
                        'ar_comment', 'ar_len', 'ar_deleted', 'ar_rev_id', 'ar_sha1',
+                       'ar_page_id'
                ];
 
                if ( $this->config->get( 'ContentHandlerUseDB' ) ) {
@@ -620,7 +671,7 @@ class PageArchive {
                $restored = 0; // number of revisions restored
                /** @var Revision $revision */
                $revision = null;
-
+               $restoredPages = [];
                // If there are no restorable revisions, we can skip most of the steps.
                if ( $latestRestorableRow === null ) {
                        $failedRevisionCount = $rev_count;
@@ -677,6 +728,7 @@ class PageArchive {
 
                                Hooks::run( 'ArticleRevisionUndeleted',
                                        [ &$this->title, $revision, $row->ar_page_id ] );
+                               $restoredPages[$row->ar_page_id] = true;
                        }
 
                        // Now that it's safely stored, take it out of the archive
@@ -717,7 +769,8 @@ class PageArchive {
                                );
                        }
 
-                       Hooks::run( 'ArticleUndelete', [ &$this->title, $created, $comment, $oldPageId ] );
+                       Hooks::run( 'ArticleUndelete',
+                               [ &$this->title, $created, $comment, $oldPageId, $restoredPages ] );
                        if ( $this->title->getNamespace() == NS_FILE ) {
                                DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->title, 'imagelinks' ) );
                        }

diff --git a/includes/search/SearchEngine.php b/includes/search/SearchEngine.php
index 9817b6c..6bb4e5a 100644 (file)
--- a/includes/search/SearchEngine.php
+++ b/includes/search/SearchEngine.php
@@ -72,6 +72,21 @@ abstract class SearchEngine {
                return null;
        }
 
+       /**
+        * Perform a title search in the article archive.
+        * NOTE: these results still should be filtered by
+        * matching against PageArchive, permissions checks etc
+        * The results returned by this methods are only sugegstions and
+        * may not end up being shown to the user.
+        *
+        * @param string $term Raw search term
+        * @return Status<Title[]>
+        * @since 1.29
+        */
+       function searchArchiveTitle( $term ) {
+               return Status::newGood( [] );
+       }
+
        /**
         * Perform a title-only search query and return a result set.
         * If title searches are not supported or disabled, return null.

diff --git a/includes/specials/SpecialUndelete.php b/includes/specials/SpecialUndelete.php
index dc5f877..eb4f0cc 100644 (file)
--- a/includes/specials/SpecialUndelete.php
+++ b/includes/specials/SpecialUndelete.php
@@ -21,6 +21,7 @@
  * @ingroup SpecialPage
  */
 
+use MediaWiki\MediaWikiServices;
 use Wikimedia\Rdbms\ResultWrapper;
 
 /**
@@ -45,6 +46,10 @@ class SpecialUndelete extends SpecialPage {
 
        /** @var Title */
        private $mTargetObj;
+       /**
+        * @var string Search prefix
+        */
+       private $mSearchPrefix;
 
        function __construct() {
                parent::__construct( 'Undelete', 'deletedhistory' );
@@ -235,6 +240,7 @@ class SpecialUndelete extends SpecialPage {
                        Xml::openElement( 'form', [ 'method' => 'get', 'action' => wfScript() ] ) .
                                Xml::fieldset( $this->msg( 'undelete-search-box' )->text() ) .
                                Html::hidden( 'title', $this->getPageTitle()->getPrefixedDBkey() ) .
+                               Html::hidden( 'fuzzy', $this->getRequest()->getVal( 'fuzzy' ) ) .
                                Html::rawElement(
                                        'label',
                                        [ 'for' => 'prefix' ],
@@ -245,15 +251,25 @@ class SpecialUndelete extends SpecialPage {
                                        20,
                                        $this->mSearchPrefix,
                                        [ 'id' => 'prefix', 'autofocus' => '' ]
-                               ) . ' ' .
-                               Xml::submitButton( $this->msg( 'undelete-search-submit' )->text() ) .
+                               ) .
+                               ' ' .
+                               Xml::submitButton(
+                                       $this->msg( 'undelete-search-submit' )->text(),
+                                       [ 'id' => 'searchUndelete' ]
+                               ) .
                                Xml::closeElement( 'fieldset' ) .
                                Xml::closeElement( 'form' )
                );
 
                # List undeletable articles
                if ( $this->mSearchPrefix ) {
-                       $result = PageArchive::listPagesByPrefix( $this->mSearchPrefix );
+                       // For now, we enable search engine match only when specifically asked to
+                       // by using fuzzy=1 parameter.
+                       if ( $this->getRequest()->getVal( "fuzzy", false ) ) {
+                               $result = PageArchive::listPagesBySearch( $this->mSearchPrefix );
+                       } else {
+                               $result = PageArchive::listPagesByPrefix( $this->mSearchPrefix );
+                       }
                        $this->showList( $result );
                }
        }
@@ -277,7 +293,7 @@ class SpecialUndelete extends SpecialPage {
 
                $linkRenderer = $this->getLinkRenderer();
                $undelete = $this->getPageTitle();
-               $out->addHTML( "<ul>\n" );
+               $out->addHTML( "<ul id='undeleteResultsList'>\n" );
                foreach ( $result as $row ) {
                        $title = Title::makeTitleSafe( $row->ar_namespace, $row->ar_title );
                        if ( $title !== null ) {
@@ -300,7 +316,7 @@ class SpecialUndelete extends SpecialPage {
                                );
                        }
                        $revs = $this->msg( 'undeleterevisions' )->numParams( $row->count )->parse();
-                       $out->addHTML( "<li>{$item} ({$revs})</li>\n" );
+                       $out->addHTML( "<li class='undeleteResult'>{$item} ({$revs})</li>\n" );
                }
                $result->free();
                $out->addHTML( "</ul>\n" );