Store permalink data, implement Special:FindComment/GoToComment

Depends-On: I90656cc74bb1cb1f2f3c82ad51cfb164cb8a4a4b Bug: T296801 Change-Id: I84187b303aa10a242c872088403f808df3d1f940
2024-11-23 16:06:53 +00:00 · 2022-02-17 00:29:10 +01:00 · 2022-02-17 00:29:10 +01:00 · 0024a94ba7
parent 91a94fe9bb
commit 0024a94ba7
52 changed files with 2701 additions and 4 deletions
--- a/DiscussionTools.alias.php
+++ b/DiscussionTools.alias.php
@ -11,6 +11,8 @@ $specialPageAliases = [];
 /** English (English) */
 $specialPageAliases['en'] = [
 	'TopicSubscriptions' => [ 'TopicSubscriptions' ],
+	'FindComment' => [ 'FindComment' ],
+	'GoToComment' => [ 'GoToComment' ],
 ];

 /** čeština (Czech) */
--- a/composer.json
+++ b/composer.json
@ -18,8 +18,11 @@
 		],
 		"dbschema": [
 			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_subscription.json --sql sql/mysql/discussiontools_subscription.sql --type mysql",
+			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_persistent.json --sql sql/mysql/discussiontools_persistent.sql --type mysql",
 			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_subscription.json --sql sql/postgres/discussiontools_subscription.sql --type postgres",
-			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_subscription.json --sql sql/sqlite/discussiontools_subscription.sql --type sqlite"
+			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_persistent.json --sql sql/postgres/discussiontools_persistent.sql --type postgres",
+			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_subscription.json --sql sql/sqlite/discussiontools_subscription.sql --type sqlite",
+			"php ../../maintenance/generateSchemaSql.php --json sql/discussiontools_persistent.json --sql sql/sqlite/discussiontools_persistent.sql --type sqlite"
 		],
 		"phan": "phan -d . --long-progress-bar",
 		"phpcs": "phpcs -sp --cache"
--- a/extension.json
+++ b/extension.json
@ -470,6 +470,19 @@
 				"LinkRenderer",
 				"LinkBatchFactory"
 			]
+		},
+		"FindComment": {
+			"class": "\\MediaWiki\\Extension\\DiscussionTools\\SpecialFindComment",
+			"services": [
+				"DiscussionTools.ThreadItemStore",
+				"DiscussionTools.ThreadItemFormatter"
+			]
+		},
+		"GoToComment": {
+			"class": "\\MediaWiki\\Extension\\DiscussionTools\\SpecialGoToComment",
+			"services": [
+				"DiscussionTools.ThreadItemStore"
+			]
 		}
 	},
 	"Hooks": {
@ -477,6 +490,7 @@
 		"EchoGetBundleRules": "\\MediaWiki\\Extension\\DiscussionTools\\Hooks\\EchoHooks::onEchoGetBundleRules",
 		"EchoGetEventsForRevision": "\\MediaWiki\\Extension\\DiscussionTools\\Hooks\\EchoHooks::onEchoGetEventsForRevision",
 		"MinervaNeueTalkPageOverlay": "\\MediaWiki\\Extension\\DiscussionTools\\Hooks\\MobileHooks::onMinervaNeueTalkPageOverlay",
+		"RevisionDataUpdates": "dataupdates",
 		"LoadExtensionSchemaUpdates": "installer",
 		"ParserAfterTidy": "parser",
 		"ArticleViewHeader": "page",
@ -494,6 +508,12 @@
 		"RecentChange_save": "tags"
 	},
 	"HookHandlers": {
+		"dataupdates": {
+			"class": "MediaWiki\\Extension\\DiscussionTools\\Hooks\\DataUpdatesHooks",
+			"services": [
+				"DiscussionTools.ThreadItemStore"
+			]
+		},
 		"installer": {
 			"class": "MediaWiki\\Extension\\DiscussionTools\\Hooks\\InstallerHooks"
 		},
--- a/i18n/en.json
+++ b/i18n/en.json
@ -35,6 +35,15 @@
 	"discussiontools-error-noswitchtove-table": "table syntax",
 	"discussiontools-error-noswitchtove-template": "template syntax",
 	"discussiontools-error-noswitchtove-title": "Visual mode disabled",
+	"discussiontools-findcomment-gotocomment": "If it only appears in the current revision of one page, you can [[Special:GoToComment/$1|go directly to the comment using this link]]. Otherwise it will redirect to these search results.",
+	"discussiontools-findcomment-label-idorname": "Comment ID or name",
+	"discussiontools-findcomment-label-search": "Search",
+	"discussiontools-findcomment-noresults": "No results.",
+	"discussiontools-findcomment-results-id": "Comment with the given ID has appeared on the following {{PLURAL:$1|page|pages}}:",
+	"discussiontools-findcomment-results-name": "Comment with the given name has appeared on the following {{PLURAL:$1|page|pages}}:",
+	"discussiontools-findcomment-results-notcurrent": "(not in current revision)",
+	"discussiontools-findcomment-results-transcluded": "(transcluded from another page)",
+	"discussiontools-findcomment-title": "Find comment",
 	"discussiontools-limitreport-errorreqid": "DiscussionTools error request ID",
 	"discussiontools-limitreport-timeusage": "DiscussionTools time usage",
 	"discussiontools-limitreport-timeusage-value": "$1 {{PLURAL:$1|second|seconds}}",
--- a/i18n/qqq.json
+++ b/i18n/qqq.json
@ -47,6 +47,15 @@
 	"discussiontools-error-noswitchtove-table": "Type of syntax detected, used as a parameter in {{msg-mw|discussiontools-error-noswitchtove}}.",
 	"discussiontools-error-noswitchtove-template": "Type of syntax detected, used as a parameter in {{msg-mw|discussiontools-error-noswitchtove}}.",
 	"discussiontools-error-noswitchtove-title": "Type of syntax detected, used as a parameter in {{msg-mw|discussiontools-error-noswitchtove}}.",
+	"discussiontools-findcomment-gotocomment": "Message shown after submitting the form on [[Special:FindComment]]. Follows {{msg-mw|discussiontools-findcomment-results-id}} or {{msg-mw|discussiontools-findcomment-results-name}} and a list of results.\n\nParameters:\n* $1 – comment ID, can only be used in link target",
+	"discussiontools-findcomment-label-idorname": "Text field label on [[Special:FindComment]]",
+	"discussiontools-findcomment-label-search": "Button label on [[Special:FindComment]]\n\n'ID' and 'name' are technical terms, see https://www.mediawiki.org/wiki/Extension:DiscussionTools/How_it_works#Assigning_ID_and_name",
+	"discussiontools-findcomment-noresults": "Message shown after submitting the form on [[Special:FindComment]].",
+	"discussiontools-findcomment-results-id": "Message shown after submitting the form on [[Special:FindComment]]. Followed by a bulleted list.\n\nParameters:\n* $1 – number of results\n\n'ID' and 'name' are technical terms, see https://www.mediawiki.org/wiki/Extension:DiscussionTools/How_it_works#Assigning_ID_and_name",
+	"discussiontools-findcomment-results-name": "Message shown after submitting the form on [[Special:FindComment]]. Followed by a bulleted list.\n\nParameters:\n* $1 – number of results\n\n'ID' and 'name' are technical terms, see https://www.mediawiki.org/wiki/Extension:DiscussionTools/How_it_works#Assigning_ID_and_name",
+	"discussiontools-findcomment-results-notcurrent": "Additional label for a result on [[Special:FindComment]], shown following a link to a page.",
+	"discussiontools-findcomment-results-transcluded": "Additional label for a result on [[Special:FindComment]], shown following a link to a page.",
+	"discussiontools-findcomment-title": "Page title for [[Special:FindComment]], also shown on the list on [[Special:SpecialPages]]",
 	"discussiontools-limitreport-errorreqid": "Label for the ID of the web request in which a DiscussionTools error has occurred.",
 	"discussiontools-limitreport-timeusage": "Label for the time usage (duration) of DiscussionTools in the parser limit report. Followed by {{msg-mw|discussiontools-limitreport-timeusage-value}}.\n\nSimilar to:\n* {{msg-mw|limitreport-cputime}}\n* {{msg-mw|limitreport-walltime}}\n* {{msg-mw|scribunto-limitreport-timeusage}}",
 	"discussiontools-limitreport-timeusage-value": "Follows {{msg-mw|discussiontools-limitreport-timeusage}}.\n\nParameters:\n* $1 - the usage in seconds\n{{Identical|Second}}",
--- a/includes/Hooks/DataUpdatesHooks.php
+++ b/includes/Hooks/DataUpdatesHooks.php
@ -0,0 +1,52 @@
+<?php
+/**
+ * DiscussionTools data updates hooks
+ *
+ * @file
+ * @ingroup Extensions
+ * @license MIT
+ */
+
+namespace MediaWiki\Extension\DiscussionTools\Hooks;
+
+use DeferrableUpdate;
+use MediaWiki\Extension\DiscussionTools\ThreadItemStore;
+use MediaWiki\Revision\RenderedRevision;
+use MediaWiki\Storage\Hook\RevisionDataUpdatesHook;
+use MWCallableUpdate;
+use Title;
+
+class DataUpdatesHooks implements RevisionDataUpdatesHook {
+
+	/** @var ThreadItemStore */
+	private $threadItemStore;
+
+	/**
+	 * @param ThreadItemStore $threadItemStore
+	 */
+	public function __construct(
+		ThreadItemStore $threadItemStore
+	) {
+		$this->threadItemStore = $threadItemStore;
+	}
+
+	/**
+	 * @param Title $title
+	 * @param RenderedRevision $renderedRevision
+	 * @param DeferrableUpdate[] &$updates
+	 * @return bool|void
+	 */
+	public function onRevisionDataUpdates( $title, $renderedRevision, &$updates ) {
+		// This doesn't trigger on action=purge, only on automatic purge after editing a template or
+		// transcluded page, and API action=purge&forcelinkupdate=1.
+
+		// TODO Deduplicate work between this and the Echo hook (make it use Parsoid too)
+		$rev = $renderedRevision->getRevision();
+		if ( HookUtils::isAvailableForTitle( $title ) ) {
+			$updates[] = new MWCallableUpdate( function () use ( $rev ) {
+				$threadItemSet = HookUtils::parseRevisionParsoidHtml( $rev );
+				$this->threadItemStore->insertThreadItems( $rev, $threadItemSet );
+			}, __METHOD__ );
+		}
+	}
+}
--- a/includes/Hooks/HookUtils.php
+++ b/includes/Hooks/HookUtils.php
@ -11,13 +11,22 @@ namespace MediaWiki\Extension\DiscussionTools\Hooks;

 use ExtensionRegistry;
 use IContextSource;
+use MediaWiki\Extension\DiscussionTools\CommentUtils;
+use MediaWiki\Extension\DiscussionTools\ContentThreadItemSet;
+use MediaWiki\Extension\VisualEditor\ParsoidHelper;
 use MediaWiki\Linker\LinkTarget;
 use MediaWiki\MediaWikiServices;
+use MediaWiki\Revision\RevisionRecord;
 use MediaWiki\User\UserIdentity;
+use MWException;
 use OutputPage;
+use Psr\Log\NullLogger;
 use RequestContext;
 use Title;
+use TitleValue;
 use Wikimedia\Assert\Assert;
+use Wikimedia\Parsoid\Utils\DOMCompat;
+use Wikimedia\Parsoid\Utils\DOMUtils;

 class HookUtils {
 	public const REPLYTOOL = 'replytool';
@ -67,6 +76,43 @@ class HookUtils {
 		return static::$propCache[ $id ][ $prop ];
 	}

+	/**
+	 * Parse a revision by using the discussion parser on the HTML provided by Parsoid.
+	 *
+	 * @param RevisionRecord $revRecord
+	 * @return ContentThreadItemSet
+	 */
+	public static function parseRevisionParsoidHtml( RevisionRecord $revRecord ): ContentThreadItemSet {
+		$services = MediaWikiServices::getInstance();
+		$parsoidHelper = new ParsoidHelper(
+			$services->getMainConfig(),
+			new NullLogger(),
+			false
+		);
+
+		// Get HTML for the revision
+		$status = $parsoidHelper->requestRestbasePageHtml( $revRecord );
+
+		if ( !$status->isOK() ) {
+			[ 'message' => $msg, 'params' => $params ] = $status->getErrors()[0];
+			throw new MWException( wfMessage( $msg, ...$params )->text() );
+		}
+
+		$title = TitleValue::newFromPage( $revRecord->getPage() );
+
+		$response = $status->getValue();
+		$html = $response['body'];
+
+		// Run the discussion parser on it
+		$doc = DOMUtils::parseHTML( $html );
+		$container = DOMCompat::getBody( $doc );
+
+		CommentUtils::unwrapParsoidSections( $container );
+
+		$parser = $services->getService( 'DiscussionTools.CommentParser' );
+		return $parser->parse( $container, $title );
+	}
+
 	/**
 	 * Check if a DiscussionTools feature is available to this user
 	 *
--- a/includes/Hooks/InstallerHooks.php
+++ b/includes/Hooks/InstallerHooks.php
@ -30,5 +30,9 @@ class InstallerHooks implements
 			'discussiontools_subscription',
 			"$base/../sql/$type/discussiontools_subscription.sql"
 		);
+		$updater->addExtensionTable(
+			'discussiontools_items',
+			"$base/../sql/$type/discussiontools_persistent.sql"
+		);
 	}
 }
--- a/includes/ServiceWiring.php
+++ b/includes/ServiceWiring.php
@ -29,5 +29,22 @@ return [
 			$services->getReadOnlyMode(),
 			$services->getUserFactory()
 		);
-	}
+	},
+	'DiscussionTools.ThreadItemStore' => static function ( MediaWikiServices $services ): ThreadItemStore {
+		return new ThreadItemStore(
+			$services->getConfigFactory(),
+			$services->getDBLoadBalancerFactory(),
+			$services->getReadOnlyMode(),
+			$services->getPageStore(),
+			$services->getRevisionStore(),
+			$services->getTitleFormatter(),
+			$services->getActorStore()
+		);
+	},
+	'DiscussionTools.ThreadItemFormatter' => static function ( MediaWikiServices $services ): ThreadItemFormatter {
+		return new ThreadItemFormatter(
+			$services->getTitleFormatter(),
+			$services->getLinkRenderer()
+		);
+	},
 ];
--- a/includes/SpecialFindComment.php
+++ b/includes/SpecialFindComment.php
@ -0,0 +1,128 @@
+<?php
+
+namespace MediaWiki\Extension\DiscussionTools;
+
+use FormSpecialPage;
+use Html;
+use HTMLForm;
+
+class SpecialFindComment extends FormSpecialPage {
+
+	/** @var ThreadItemStore */
+	private $threadItemStore;
+
+	/** @var ThreadItemFormatter */
+	private $threadItemFormatter;
+
+	/**
+	 * @param ThreadItemStore $threadItemStore
+	 * @param ThreadItemFormatter $threadItemFormatter
+	 */
+	public function __construct(
+		ThreadItemStore $threadItemStore,
+		ThreadItemFormatter $threadItemFormatter
+	) {
+		parent::__construct( 'FindComment' );
+		$this->threadItemStore = $threadItemStore;
+		$this->threadItemFormatter = $threadItemFormatter;
+	}
+
+	/**
+	 * @inheritDoc
+	 */
+	protected function getFormFields() {
+		return [
+			'idorname' => [
+				'label-message' => 'discussiontools-findcomment-label-idorname',
+				'name' => 'idorname',
+				'type' => 'text',
+				'default' => $this->par,
+			],
+		];
+	}
+
+	/**
+	 * @inheritDoc
+	 */
+	protected function getDisplayFormat() {
+		return 'ooui';
+	}
+
+	/**
+	 * @inheritDoc
+	 */
+	protected function alterForm( HTMLForm $form ) {
+		$form->setMethod( 'GET' );
+		$form->setWrapperLegend( true );
+		$form->setSubmitTextMsg( 'discussiontools-findcomment-label-search' );
+		// Remove subpage when submitting
+		$form->setTitle( $this->getPageTitle() );
+	}
+
+	private $idOrName;
+
+	/**
+	 * @inheritDoc
+	 */
+	public function onSubmit( array $data ) {
+		$this->idOrName = $data['idorname'];
+		// Always display the form again
+		return false;
+	}
+
+	/**
+	 * @inheritDoc
+	 */
+	public function execute( $par ) {
+		parent::execute( $par );
+
+		$out = $this->getOutput();
+		$results = false;
+
+		if ( $this->idOrName ) {
+			$byId = $this->threadItemStore->findNewestRevisionsById( $this->idOrName );
+			if ( $byId ) {
+				$this->displayItems( $byId, 'discussiontools-findcomment-results-id' );
+				$results = true;
+			}
+
+			$byName = $this->threadItemStore->findNewestRevisionsByName( $this->idOrName );
+			if ( $byName ) {
+				$this->displayItems( $byName, 'discussiontools-findcomment-results-name' );
+				$results = true;
+			}
+		}
+
+		if ( $results ) {
+			$out->addHTML( Html::rawElement( 'p', [],
+				$this->msg( 'discussiontools-findcomment-gotocomment', $this->idOrName )->parse() ) );
+		} elseif ( $this->idOrName ) {
+			$out->addHTML( Html::rawElement( 'p', [],
+				$this->msg( 'discussiontools-findcomment-noresults' )->parse() ) );
+		}
+	}
+
+	/**
+	 * @param array $threadItems
+	 * @param string $msgKey
+	 */
+	private function displayItems( array $threadItems, string $msgKey ) {
+		$out = $this->getOutput();
+
+		$list = [];
+		foreach ( $threadItems as $item ) {
+			$line = $this->threadItemFormatter->formatLine( $item, $this );
+			$list[] = Html::rawElement( 'li', [], $line );
+		}
+
+		$out->addHTML( Html::rawElement( 'p', [], $this->msg( $msgKey, count( $list ) )->parse() ) );
+		$out->addHTML( Html::rawElement( 'ul', [], implode( '', $list ) ) );
+	}
+
+	/**
+	 * @inheritDoc
+	 */
+	public function getDescription() {
+		return $this->msg( 'discussiontools-findcomment-title' )->text();
+	}
+}
--- a/includes/SpecialGoToComment.php
+++ b/includes/SpecialGoToComment.php
@ -0,0 +1,59 @@
+<?php
+
+namespace MediaWiki\Extension\DiscussionTools;
+
+use RedirectSpecialPage;
+use SpecialPage;
+use Title;
+
+class SpecialGoToComment extends RedirectSpecialPage {
+
+	/** @var ThreadItemStore */
+	private $threadItemStore;
+
+	/**
+	 * @param ThreadItemStore $threadItemStore
+	 */
+	public function __construct(
+		ThreadItemStore $threadItemStore
+	) {
+		parent::__construct( 'GoToComment' );
+		$this->threadItemStore = $threadItemStore;
+	}
+
+	/**
+	 * @inheritDoc
+	 */
+	public function getRedirect( $subpage ) {
+		$results = [];
+
+		// Search for all thread items with the given ID or name, returning results from the latest
+		// revision of each page they appeared on.
+		//
+		// If there is exactly one result which is not transcluded from another page and in the current
+		// revision of its page, redirect to it.
+		//
+		// Otherwise, redirect to full search results on Special:FindComment.
+
+		if ( $subpage ) {
+			$threadItems = $this->threadItemStore->findNewestRevisionsById( $subpage );
+			foreach ( $threadItems as $item ) {
+				if ( $item->getRevision()->isCurrent() && !is_string( $item->getTranscludedFrom() ) ) {
+					$results[] = $item;
+				}
+			}
+			$threadItems = $this->threadItemStore->findNewestRevisionsByName( $subpage );
+			foreach ( $threadItems as $item ) {
+				if ( $item->getRevision()->isCurrent() && !is_string( $item->getTranscludedFrom() ) ) {
+					$results[] = $item;
+				}
+			}
+		}
+
+		if ( count( $results ) === 1 ) {
+			return Title::castFromPageIdentity( $results[0]->getPage() )->createFragmentTarget( $results[0]->getId() );
+		} else {
+			return SpecialPage::getTitleFor( 'FindComment', $subpage ?: false );
+		}
+	}
+}
--- a/includes/ThreadItem/DatabaseCommentItem.php
+++ b/includes/ThreadItem/DatabaseCommentItem.php
@ -3,6 +3,8 @@
 namespace MediaWiki\Extension\DiscussionTools\ThreadItem;

 use DateTimeImmutable;
+use MediaWiki\Page\ProperPageIdentity;
+use MediaWiki\Revision\RevisionRecord;

 class DatabaseCommentItem extends DatabaseThreadItem implements CommentItem {
 	use CommentItemTrait {
@ -16,6 +18,8 @@ class DatabaseCommentItem extends DatabaseThreadItem implements CommentItem {
 	private $author;

 	/**
+	 * @param ProperPageIdentity $page
+	 * @param RevisionRecord $rev
 	 * @param string $name
 	 * @param string $id
 	 * @param DatabaseThreadItem|null $parent
@ -25,10 +29,11 @@ class DatabaseCommentItem extends DatabaseThreadItem implements CommentItem {
 	 * @param string $author
 	 */
 	public function __construct(
+		ProperPageIdentity $page, RevisionRecord $rev,
 		string $name, string $id, ?DatabaseThreadItem $parent, $transcludedFrom, int $level,
 		string $timestamp, string $author
 	) {
-		parent::__construct( 'comment', $name, $id, $parent, $transcludedFrom, $level );
+		parent::__construct( $page, $rev, 'comment', $name, $id, $parent, $transcludedFrom, $level );
 		$this->timestamp = $timestamp;
 		$this->author = $author;
 	}
--- a/includes/ThreadItem/DatabaseHeadingItem.php
+++ b/includes/ThreadItem/DatabaseHeadingItem.php
@ -2,6 +2,9 @@

 namespace MediaWiki\Extension\DiscussionTools\ThreadItem;

+use MediaWiki\Page\ProperPageIdentity;
+use MediaWiki\Revision\RevisionRecord;
+
 class DatabaseHeadingItem extends DatabaseThreadItem implements HeadingItem {
 	use HeadingItemTrait;

@ -14,6 +17,8 @@ class DatabaseHeadingItem extends DatabaseThreadItem implements HeadingItem {
 	private const PLACEHOLDER_HEADING_LEVEL = 99;

 	/**
+	 * @param ProperPageIdentity $page
+	 * @param RevisionRecord $rev
 	 * @param string $name
 	 * @param string $id
 	 * @param DatabaseThreadItem|null $parent
@ -22,10 +27,11 @@ class DatabaseHeadingItem extends DatabaseThreadItem implements HeadingItem {
 	 * @param ?int $headingLevel Heading level (1-6). Use null for a placeholder heading.
 	 */
 	public function __construct(
+		ProperPageIdentity $page, RevisionRecord $rev,
 		string $name, string $id, ?DatabaseThreadItem $parent, $transcludedFrom, int $level,
 		?int $headingLevel
 	) {
-		parent::__construct( 'heading', $name, $id, $parent, $transcludedFrom, $level );
+		parent::__construct( $page, $rev, 'heading', $name, $id, $parent, $transcludedFrom, $level );
 		$this->placeholderHeading = $headingLevel === null;
 		$this->headingLevel = $this->placeholderHeading ? static::PLACEHOLDER_HEADING_LEVEL : $headingLevel;
 	}
--- a/includes/ThreadItem/DatabaseThreadItem.php
+++ b/includes/ThreadItem/DatabaseThreadItem.php
@ -3,10 +3,16 @@
 namespace MediaWiki\Extension\DiscussionTools\ThreadItem;

 use JsonSerializable;
+use MediaWiki\Page\ProperPageIdentity;
+use MediaWiki\Revision\RevisionRecord;

 class DatabaseThreadItem implements JsonSerializable, ThreadItem {
 	use ThreadItemTrait;

+	/** @var ProperPageIdentity */
+	private $page;
+	/** @var RevisionRecord */
+	private $rev;
 	/** @var string */
 	private $type;
 	/** @var string */
@ -23,6 +29,8 @@ class DatabaseThreadItem implements JsonSerializable, ThreadItem {
 	private $level;

 	/**
+	 * @param ProperPageIdentity $page
+	 * @param RevisionRecord $rev
 	 * @param string $type
 	 * @param string $name
 	 * @param string $id
@ -31,8 +39,11 @@ class DatabaseThreadItem implements JsonSerializable, ThreadItem {
 	 * @param int $level
 	 */
 	public function __construct(
+		ProperPageIdentity $page, RevisionRecord $rev,
 		string $type, string $name, string $id, ?DatabaseThreadItem $parent, $transcludedFrom, int $level
 	) {
+		$this->page = $page;
+		$this->rev = $rev;
 		$this->name = $name;
 		$this->id = $id;
 		$this->type = $type;
@ -41,6 +52,20 @@ class DatabaseThreadItem implements JsonSerializable, ThreadItem {
 		$this->level = $level;
 	}

+	/**
+	 * @return ProperPageIdentity
+	 */
+	public function getPage(): ProperPageIdentity {
+		return $this->page;
+	}
+
+	/**
+	 * @return RevisionRecord
+	 */
+	public function getRevision(): RevisionRecord {
+		return $this->rev;
+	}
+
 	/**
 	 * @inheritDoc
 	 */
--- a/includes/ThreadItemFormatter.php
+++ b/includes/ThreadItemFormatter.php
@ -0,0 +1,76 @@
+<?php
+
+namespace MediaWiki\Extension\DiscussionTools;
+
+use MediaWiki\Extension\DiscussionTools\ThreadItem\DatabaseThreadItem;
+use MediaWiki\Linker\LinkRenderer;
+use MessageLocalizer;
+use TitleFormatter;
+use TitleValue;
+
+/**
+ * Displays links to comments and headings represented as ThreadItems.
+ */
+class ThreadItemFormatter {
+	/** @var TitleFormatter */
+	private $titleFormatter;
+
+	/** @var LinkRenderer */
+	private $linkRenderer;
+
+	/**
+	 * @param TitleFormatter $titleFormatter
+	 * @param LinkRenderer $linkRenderer
+	 */
+	public function __construct(
+		TitleFormatter $titleFormatter,
+		LinkRenderer $linkRenderer
+	) {
+		$this->titleFormatter = $titleFormatter;
+		$this->linkRenderer = $linkRenderer;
+	}
+
+	/**
+	 * Make a link to a thread item on the page.
+	 *
+	 * @param DatabaseThreadItem $item
+	 * @return string
+	 */
+	public function makeLink( DatabaseThreadItem $item ): string {
+		$title = TitleValue::newFromPage( $item->getPage() )->createFragmentTarget( $item->getId() );
+
+		$query = [];
+		if ( !$item->getRevision()->isCurrent() ) {
+			$query['oldid'] = $item->getRevision()->getId();
+		}
+
+		$text = $this->titleFormatter->getPrefixedText( $title );
+		$link = $this->linkRenderer->makeLink( $title, $text, [], $query );
+
+		return $link;
+	}
+
+	/**
+	 * Make a link to a thread item on the page, with additional information (used on special pages).
+	 *
+	 * @param DatabaseThreadItem $item
+	 * @param MessageLocalizer $context
+	 * @return string
+	 */
+	public function formatLine( DatabaseThreadItem $item, MessageLocalizer $context ): string {
+		$contents = [];
+
+		$contents[] = $this->makeLink( $item );
+
+		if ( !$item->getRevision()->isCurrent() ) {
+			$contents[] = $context->msg( 'discussiontools-findcomment-results-notcurrent' )->escaped();
+		}
+
+		if ( is_string( $item->getTranscludedFrom() ) ) {
+			$contents[] = $context->msg( 'discussiontools-findcomment-results-transcluded' )->escaped();
+		}
+
+		return implode( $context->msg( 'word-separator' )->escaped(), $contents );
+	}
+
+}
--- a/includes/ThreadItemStore.php
+++ b/includes/ThreadItemStore.php
@ -0,0 +1,569 @@
+<?php
+
+namespace MediaWiki\Extension\DiscussionTools;
+
+use ConfigFactory;
+use MediaWiki\Extension\DiscussionTools\ThreadItem\CommentItem;
+use MediaWiki\Extension\DiscussionTools\ThreadItem\DatabaseCommentItem;
+use MediaWiki\Extension\DiscussionTools\ThreadItem\DatabaseHeadingItem;
+use MediaWiki\Extension\DiscussionTools\ThreadItem\DatabaseThreadItem;
+use MediaWiki\Extension\DiscussionTools\ThreadItem\HeadingItem;
+use MediaWiki\Page\PageStore;
+use MediaWiki\Revision\RevisionRecord;
+use MediaWiki\Revision\RevisionStore;
+use MediaWiki\User\ActorStore;
+use MWTimestamp;
+use ReadOnlyMode;
+use stdClass;
+use TitleFormatter;
+use Wikimedia\Rdbms\ILBFactory;
+use Wikimedia\Rdbms\ILoadBalancer;
+use Wikimedia\Rdbms\IMaintainableDatabase;
+use Wikimedia\Rdbms\IResultWrapper;
+use Wikimedia\Rdbms\SelectQueryBuilder;
+
+/**
+ * Stores and fetches ThreadItemSets from the database.
+ */
+class ThreadItemStore {
+	/** @var ConfigFactory */
+	private $configFactory;
+
+	/** @var ILoadBalancer */
+	private $loadBalancer;
+
+	/** @var ReadOnlyMode */
+	private $readOnlyMode;
+
+	/** @var PageStore */
+	private $pageStore;
+
+	/** @var RevisionStore */
+	private $revStore;
+
+	/** @var TitleFormatter */
+	private $titleFormatter;
+
+	/** @var ActorStore */
+	private $actorStore;
+
+	/**
+	 * @param ConfigFactory $configFactory
+	 * @param ILBFactory $lbFactory
+	 * @param ReadOnlyMode $readOnlyMode
+	 * @param PageStore $pageStore
+	 * @param RevisionStore $revStore
+	 * @param TitleFormatter $titleFormatter
+	 * @param ActorStore $actorStore
+	 */
+	public function __construct(
+		ConfigFactory $configFactory,
+		ILBFactory $lbFactory,
+		ReadOnlyMode $readOnlyMode,
+		PageStore $pageStore,
+		RevisionStore $revStore,
+		TitleFormatter $titleFormatter,
+		ActorStore $actorStore
+	) {
+		$this->configFactory = $configFactory;
+		$this->loadBalancer = $lbFactory->getMainLB();
+		$this->readOnlyMode = $readOnlyMode;
+		$this->pageStore = $pageStore;
+		$this->revStore = $revStore;
+		$this->titleFormatter = $titleFormatter;
+		$this->actorStore = $actorStore;
+	}
+
+	/**
+	 * Returns true if the tables necessary for this feature haven't been created yet,
+	 * to allow failing softly in that case.
+	 *
+	 * @return bool
+	 */
+	private function isDisabled(): bool {
+		static $tablesCreated = null;
+		if ( $tablesCreated === null ) {
+			$tablesCreated = $this->getConnectionRef( DB_REPLICA )->tableExists( 'discussiontools_items', __METHOD__ );
+		}
+		return !$tablesCreated;
+	}
+
+	/**
+	 * @param int $dbIndex DB_PRIMARY or DB_REPLICA
+	 *
+	 * @return IMaintainableDatabase
+	 */
+	private function getConnectionRef( int $dbIndex ): IMaintainableDatabase {
+		return $this->loadBalancer->getConnectionRef( $dbIndex, [ 'watchlist' ] );
+	}
+
+	/**
+	 * Find the thread items with the given name in the newest revision of every page in which they
+	 * have appeared.
+	 *
+	 * @param string|string[] $itemName
+	 * @return DatabaseThreadItem[]
+	 */
+	public function findNewestRevisionsByName( $itemName ): array {
+		if ( $this->isDisabled() ) {
+			return [];
+		}
+
+		$queryBuilder = $this->getIdsNamesBuilder()
+			->where( [
+				'it_itemname' => $itemName,
+				// Disallow querying for headings of sections that contain no comments.
+				// They all share the same name, so this would return a huge useless list on most wikis.
+				// (But we still store them, as we might need this data elsewhere.)
+				"it_itemname != 'h-'",
+			] );
+
+		$result = $this->fetchItemsResultSet( $queryBuilder );
+		$revs = $this->fetchRevisionAndPageForItems( $result );
+
+		$threadItems = [];
+		foreach ( $result as $row ) {
+			$threadItem = $this->getThreadItemFromRow( $row, null, $revs );
+			if ( $threadItem ) {
+				$threadItems[] = $threadItem;
+			}
+		}
+		return $threadItems;
+	}
+
+	/**
+	 * Find the thread items with the given ID in the newest revision of every page in which they have
+	 * appeared.
+	 *
+	 * @param string|string[] $itemId
+	 * @return DatabaseThreadItem[]
+	 */
+	public function findNewestRevisionsById( $itemId ): array {
+		if ( $this->isDisabled() ) {
+			return [];
+		}
+
+		$queryBuilder = $this->getIdsNamesBuilder();
+
+		// First find the name associated with the ID; then find by name. Otherwise we wouldn't find the
+		// latest revision in case comment ID changed, e.g. the comment was moved elsewhere on the page.
+		$itemNameQueryBuilder = $this->getIdsNamesBuilder()
+			->where( [ 'itid_itemid' => $itemId ] )
+			->field( 'it_itemname' );
+			// I think there may be more than 1 only in case of headings?
+			// For comments, any ID corresponds to just 1 name.
+			// Not sure how bad it is to not have limit( 1 ) here?
+			// It might scan a bunch of rows...
+			// ->limit( 1 );
+
+		$queryBuilder
+			->where( [
+				'it_itemname IN (' . $itemNameQueryBuilder->getSQL() . ')',
+				"it_itemname != 'h-'",
+			] );
+
+		$result = $this->fetchItemsResultSet( $queryBuilder );
+		$revs = $this->fetchRevisionAndPageForItems( $result );
+
+		$threadItems = [];
+		foreach ( $result as $row ) {
+			$threadItem = $this->getThreadItemFromRow( $row, null, $revs );
+			if ( $threadItem ) {
+				$threadItems[] = $threadItem;
+			}
+		}
+		return $threadItems;
+	}
+
+	/**
+	 * @param SelectQueryBuilder $queryBuilder
+	 * @return IResultWrapper
+	 */
+	private function fetchItemsResultSet( SelectQueryBuilder $queryBuilder ): IResultWrapper {
+		$queryBuilder
+			->fields( [
+				'itr_id',
+				'it_itemname',
+				'it_timestamp',
+				'it_actor',
+				'itid_itemid',
+				'itr_parent_id',
+				'itr_transcludedfrom',
+				'itr_level',
+				'itr_headinglevel',
+				'itr_revision_id',
+			] )
+			// PageStore fields for the transcluded-from page
+			->leftJoin( 'page', null, [ 'page_id = itr_transcludedfrom' ] )
+			->fields( $this->pageStore->getSelectFields() )
+			// ActorStore fields for the author
+			->leftJoin( 'actor', null, [ 'actor_id = it_actor' ] )
+			->fields( [ 'actor_id', 'actor_name', 'actor_user' ] )
+			// Parent item ID (the string, not just the primary key)
+			->leftJoin(
+				$this->getIdsNamesBuilder()
+					->fields( [
+						'itr_parent__itr_id' => 'itr_id',
+						'itr_parent__itid_itemid' => 'itid_itemid',
+					] ),
+				null,
+				[ 'itr_parent_id = itr_parent__itr_id' ]
+			)
+			->field( 'itr_parent__itid_itemid' );
+
+		return $queryBuilder->fetchResultSet();
+	}
+
+	/**
+	 * @param IResultWrapper $result
+	 * @return stdClass[]
+	 */
+	private function fetchRevisionAndPageForItems( IResultWrapper $result ): array {
+		// This could theoretically be done in the same query as fetchItemsResultSet(),
+		// but the resulting query would be two screens long
+		// and we'd have to alias a lot of fields to avoid conflicts.
+		$revs = [];
+		foreach ( $result as $row ) {
+			$revs[ $row->itr_revision_id ] = null;
+		}
+		$revQueryBuilder = $this->getConnectionRef( DB_REPLICA )->newSelectQueryBuilder()
+			->queryInfo( $this->revStore->getQueryInfo( [ 'page' ] ) )
+			->fields( $this->pageStore->getSelectFields() )
+			->where( $revs ? [ 'rev_id' => array_keys( $revs ) ] : '0=1' );
+		$revResult = $revQueryBuilder->fetchResultSet();
+		foreach ( $revResult as $row ) {
+			$revs[ $row->rev_id ] = $row;
+		}
+		return $revs;
+	}
+
+	/**
+	 * @param stdClass $row
+	 * @param DatabaseThreadItemSet|null $set
+	 * @param array $revs
+	 * @return DatabaseThreadItem|null
+	 */
+	private function getThreadItemFromRow(
+		stdClass $row, ?DatabaseThreadItemSet $set, array $revs
+	): ?DatabaseThreadItem {
+		if ( $revs[ $row->itr_revision_id ] === null ) {
+			// We didn't find the 'revision' table row at all, this revision is deleted.
+			// (The page may or may not have other non-deleted revisions.)
+			// Pretend the thread item doesn't exist to avoid leaking data to users who shouldn't see it.
+			// TODO Allow privileged users to see it (we'd need to query from 'archive')
+			return null;
+		}
+
+		$revRow = $revs[$row->itr_revision_id];
+		$page = $this->pageStore->newPageRecordFromRow( $revRow );
+		$rev = $this->revStore->newRevisionFromRow( $revRow );
+		if ( $rev->isDeleted( RevisionRecord::DELETED_TEXT ) ) {
+			// This revision is revision-deleted.
+			// TODO Allow privileged users to see it
+			return null;
+		}
+
+		if ( $set && $row->itr_parent__itid_itemid ) {
+			$parent = $set->findCommentById( $row->itr_parent__itid_itemid );
+		} else {
+			$parent = null;
+		}
+
+		$transcludedFrom = $row->itr_transcludedfrom === null ? false : (
+			$row->itr_transcludedfrom === '0' ? true :
+				$this->titleFormatter->getPrefixedText(
+					$this->pageStore->newPageRecordFromRow( $row )
+				)
+		);
+
+		if ( $row->it_timestamp !== null && $row->it_actor !== null ) {
+			$author = $this->actorStore->newActorFromRow( $row )->getName();
+
+			$item = new DatabaseCommentItem(
+				$page,
+				$rev,
+				$row->it_itemname,
+				$row->itid_itemid,
+				$parent,
+				$transcludedFrom,
+				(int)$row->itr_level,
+				$row->it_timestamp,
+				$author
+			);
+		} else {
+			$item = new DatabaseHeadingItem(
+				$page,
+				$rev,
+				$row->it_itemname,
+				$row->itid_itemid,
+				$parent,
+				$transcludedFrom,
+				(int)$row->itr_level,
+				$row->itr_headinglevel === null ? null : (int)$row->itr_headinglevel
+			);
+		}
+
+		if ( $parent ) {
+			$parent->addReply( $item );
+		}
+		return $item;
+	}
+
+	/**
+	 * Find the thread item set for the given revision, assuming that it is the current revision of
+	 * its page.
+	 *
+	 * @param int $revId
+	 * @return DatabaseThreadItemSet
+	 */
+	public function findThreadItemsInCurrentRevision( int $revId ): DatabaseThreadItemSet {
+		if ( $this->isDisabled() ) {
+			return new DatabaseThreadItemSet();
+		}
+
+		$queryBuilder = $this->getIdsNamesBuilder();
+		$queryBuilder
+			->where( [ 'itr_revision_id' => $revId ] )
+			// We must process parents before their children in the loop later
+			->orderBy( 'itr_id', SelectQueryBuilder::SORT_ASC );
+
+		$result = $this->fetchItemsResultSet( $queryBuilder );
+		$revs = $this->fetchRevisionAndPageForItems( $result );
+
+		$set = new DatabaseThreadItemSet();
+		foreach ( $result as $row ) {
+			$threadItem = $this->getThreadItemFromRow( $row, $set, $revs );
+			if ( $threadItem ) {
+				$set->addThreadItem( $threadItem );
+				$set->updateIdAndNameMaps( $threadItem );
+			}
+		}
+		return $set;
+	}
+
+	/**
+	 * @return SelectQueryBuilder
+	 */
+	private function getIdsNamesBuilder(): SelectQueryBuilder {
+		$dbr = $this->getConnectionRef( DB_REPLICA );
+
+		$queryBuilder = $dbr->newSelectQueryBuilder()
+			->from( 'discussiontools_items' )
+			->join( 'discussiontools_item_pages', null, [ 'itp_items_id = it_id' ] )
+			->join( 'discussiontools_item_revisions', null, [
+				'itr_items_id = it_id',
+				// Only the latest revision of the items with each name
+				'itr_revision_id = itp_newest_revision_id',
+			] )
+			->join( 'discussiontools_item_ids', null, [ 'itid_id = itr_itemid_id' ] );
+
+		return $queryBuilder;
+	}
+
+	/**
+	 * Store the thread item set.
+	 *
+	 * @param RevisionRecord $rev
+	 * @param ThreadItemSet $threadItemSet
+	 * @return bool
+	 */
+	public function insertThreadItems( RevisionRecord $rev, ThreadItemSet $threadItemSet ): bool {
+		if ( $this->isDisabled() || $this->readOnlyMode->isReadOnly() ) {
+			return false;
+		}
+
+		$dbw = $this->getConnectionRef( DB_PRIMARY );
+		$didInsert = false;
+		$method = __METHOD__;
+
+		$dbw->doAtomicSection( $method, function ( $dbw ) use ( $method, $rev, $threadItemSet, &$didInsert ) {
+			$itemRevisionsIds = [];
+			foreach ( $threadItemSet->getThreadItems() as $item ) {
+				$itemIdsId = $dbw->selectField(
+					'discussiontools_item_ids',
+					'itid_id',
+					[ 'itid_itemid' => $item->getId() ],
+					$method
+				);
+				if ( $itemIdsId === false ) {
+					$dbw->insert(
+						'discussiontools_item_ids',
+						[
+							'itid_itemid' => $item->getId(),
+						],
+						$method
+					);
+					$itemIdsId = $dbw->insertId();
+					$didInsert = true;
+				}
+
+				$itemsId = $dbw->selectField(
+					'discussiontools_items',
+					'it_id',
+					[ 'it_itemname' => $item->getName() ],
+					$method
+				);
+				if ( $itemsId === false ) {
+					$dbw->insert(
+						'discussiontools_items',
+						[
+							'it_itemname' => $item->getName(),
+						] +
+						( $item instanceof CommentItem ? [
+							'it_timestamp' =>
+								$dbw->timestamp( $item->getTimestampString() ),
+							'it_actor' =>
+								$this->actorStore->findActorIdByName( $item->getAuthor(), $dbw ),
+						] : [] ),
+						$method
+					);
+					$itemsId = $dbw->insertId();
+					$didInsert = true;
+				}
+
+				$itemRevisionsId = $dbw->selectField(
+					'discussiontools_item_revisions',
+					'itr_id',
+					[
+						'itr_itemid_id' => $itemIdsId,
+						'itr_revision_id' => $rev->getId(),
+					],
+					$method
+				);
+				if ( $itemRevisionsId === false ) {
+					$transcl = $item->getTranscludedFrom();
+					$dbw->insert(
+						'discussiontools_item_revisions',
+						[
+							'itr_itemid_id' => $itemIdsId,
+							'itr_revision_id' => $rev->getId(),
+							'itr_items_id' => $itemsId,
+							'itr_parent_id' =>
+								// This assumes that parent items were processed first
+								$item->getParent() ? $itemRevisionsIds[ $item->getParent()->getId() ] : null,
+							'itr_transcludedfrom' =>
+								$transcl === false ? null : (
+									$transcl === true ? 0 :
+										$this->pageStore->getPageByText( $transcl )->getId()
+								),
+							'itr_level' => $item->getLevel(),
+						] +
+						( $item instanceof HeadingItem ? [
+							'itr_headinglevel' => $item->isPlaceholderHeading() ? null : $item->getHeadingLevel(),
+						] : [] ),
+						$method
+					);
+					$itemRevisionsId = $dbw->insertId();
+					$didInsert = true;
+				}
+				$itemRevisionsIds[ $item->getId() ] = $itemRevisionsId;
+
+				// Update (or insert) the references to oldest/newest item revision.
+				// The page revision we're processing is usually the newest one, but it doesn't have to be
+				// (in case of backfilling using the maintenance script, or in case of revisions being
+				// imported), so we need all these funky queries to see if we need to update oldest/newest.
+
+				// This should be a single upsert query (INSERT ... ON DUPLICATE KEY UPDATE), however it
+				// doesn't work in practice:
+				//
+				// - Attempt 1:
+				//   https://gerrit.wikimedia.org/r/c/mediawiki/extensions/DiscussionTools/+/771974/14/includes/ThreadItemStore.php#451
+				//
+				//   This is the same logic as below in SQL: only doing a single comparison of the timestamp
+				//   of the current revision to the existing data in discussiontools_item_pages.
+				//   It worked great on my machine, `mysql --version`:
+				//     mysql  Ver 8.0.29-0ubuntu0.20.04.3 for Linux on x86_64 ((Ubuntu))
+				//   but it failed in Wikimedia CI, `mysql --version`:
+				//     mysql  Ver 15.1 Distrib 10.3.34-MariaDB, for debian-linux-gnu (x86_64) using readline 5.2
+				//   …with the error:
+				//     "Error 1054: Unknown column 'itp_oldest_revision_id' in 'where clause'".
+				//   Apparently it doesn't like dependent subqueries in the UPDATE part of an upsert.
+				//   I'm not sure if it's a bug in MariaDB or if you're not supposed to do that.
+				//
+				// - Attempt 2:
+				//   https://gerrit.wikimedia.org/r/c/mediawiki/extensions/DiscussionTools/+/771974/15/includes/ThreadItemStore.php#451
+				//
+				//   This avoids the dependent subquery: instead of comparing to the existing data in
+				//   discussiontools_item_pages, it just takes the IDs with min/max timestamp from
+				//   discussiontools_item_revisions/revision. This should be a simple lookup from an index,
+				//   but apparently it doesn't work that way and is significantly slower (the maintenance
+				//   script went from 13 minutes to 18 minutes when processing a few thousands of revisions
+				//   on my local testing wiki).
+				//
+				// In the end, the solution below using multiple queries is just as fast as the original,
+				// and only a little more verbose.
+
+				$itemPagesRow = $dbw->newSelectQueryBuilder()
+					->from( 'discussiontools_item_pages' )
+					->join( 'revision', 'revision_oldest', [ 'itp_oldest_revision_id = revision_oldest.rev_id' ] )
+					->join( 'revision', 'revision_newest', [ 'itp_newest_revision_id = revision_newest.rev_id' ] )
+					->field( 'itp_id' )
+					->field( 'revision_oldest.rev_timestamp', 'oldest_rev_timestamp' )
+					->field( 'revision_newest.rev_timestamp', 'newest_rev_timestamp' )
+					->where( [
+						'itp_items_id' => $itemsId,
+						'itp_page_id' => $rev->getPageId(),
+					] )
+					->fetchRow();
+				if ( $itemPagesRow === false ) {
+					$dbw->insert(
+						'discussiontools_item_pages',
+						[
+							'itp_items_id' => $itemsId,
+							'itp_page_id' => $rev->getPageId(),
+							'itp_oldest_revision_id' => $rev->getId(),
+							'itp_newest_revision_id' => $rev->getId(),
+						],
+						$method
+					);
+				} else {
+					$existingTime = ( new MWTimestamp( $itemPagesRow->oldest_rev_timestamp ) )->getTimestamp( TS_MW );
+					if ( $existingTime >= $rev->getTimestamp() ) {
+						$dbw->update(
+							'discussiontools_item_pages',
+							[ 'itp_oldest_revision_id' => $rev->getId() ],
+							[ 'itp_id' => $itemPagesRow->itp_id ],
+							$method
+						);
+					}
+					$existingTime = ( new MWTimestamp( $itemPagesRow->newest_rev_timestamp ) )->getTimestamp( TS_MW );
+					if ( $existingTime <= $rev->getTimestamp() ) {
+						$dbw->update(
+							'discussiontools_item_pages',
+							[ 'itp_newest_revision_id' => $rev->getId() ],
+							[ 'itp_id' => $itemPagesRow->itp_id ],
+							$method
+						);
+					}
+				}
+
+				// Delete rows that we don't care about, to save space (item revisions with the same ID and
+				// name as the one we just inserted, which are not the oldest or newest revision).
+				$oldestRevisionSql = $dbw->selectSQLText(
+					'discussiontools_item_pages',
+					'itp_oldest_revision_id',
+					[ 'itp_items_id' => $itemsId ],
+					$method
+				);
+				$newestRevisionSql = $dbw->selectSQLText(
+					'discussiontools_item_pages',
+					'itp_newest_revision_id',
+					[ 'itp_items_id' => $itemsId ],
+					$method
+				);
+				$dbw->delete(
+					'discussiontools_item_revisions',
+					[
+						'itr_itemid_id' => $itemIdsId,
+						'itr_items_id' => $itemsId,
+						"itr_revision_id NOT IN ($oldestRevisionSql)",
+						"itr_revision_id NOT IN ($newestRevisionSql)",
+					],
+					$method
+				);
+			}
+		}, $dbw::ATOMIC_CANCELABLE );
+
+		return $didInsert;
+	}
+}
--- a/maintenance/persistRevisionThreadItems.php
+++ b/maintenance/persistRevisionThreadItems.php
@ -0,0 +1,123 @@
+<?php
+
+namespace MediaWiki\Extension\DiscussionTools\Maintenance;
+
+use MediaWiki\Extension\DiscussionTools\Hooks\HookUtils;
+use MediaWiki\Extension\DiscussionTools\ThreadItemStore;
+use MediaWiki\MediaWikiServices;
+use MediaWiki\Revision\RevisionStore;
+use MWExceptionRenderer;
+use stdClass;
+use TableCleanup;
+use Throwable;
+use Title;
+
+// Security: Disable all stream wrappers and reenable individually as needed
+foreach ( stream_get_wrappers() as $wrapper ) {
+	stream_wrapper_unregister( $wrapper );
+}
+
+stream_wrapper_restore( 'file' );
+$basePath = getenv( 'MW_INSTALL_PATH' );
+if ( $basePath ) {
+	if ( !is_dir( $basePath )
+		|| strpos( $basePath, '.' ) !== false
+		|| strpos( $basePath, '~' ) !== false
+	) {
+		die( "Bad MediaWiki install path: $basePath\n" );
+	}
+} else {
+	$basePath = __DIR__ . '/../../..';
+}
+require_once "$basePath/maintenance/Maintenance.php";
+// Autoloader isn't set up yet until we do `require_once RUN_MAINTENANCE_IF_MAIN`…
+// but our class needs to exist at that point D:
+require_once "$basePath/maintenance/TableCleanup.php";
+
+class PersistRevisionThreadItems extends TableCleanup {
+
+	/** @var ThreadItemStore */
+	private $itemStore;
+
+	/** @var RevisionStore */
+	private $revStore;
+
+	public function __construct() {
+		parent::__construct();
+		$this->requireExtension( 'DiscussionTools' );
+		$this->addDescription( 'Persist thread item information for the given pages/revisions' );
+		$this->addOption( 'rev', 'Revision ID to process', false, true, false, true );
+		$this->addOption( 'page', 'Page title to process', false, true, false, true );
+		$this->addOption( 'all', 'Process the whole wiki', false, false, false, false );
+	}
+
+	public function execute() {
+		$services = MediaWikiServices::getInstance();
+
+		$this->itemStore = $services->getService( 'DiscussionTools.ThreadItemStore' );
+		$this->revStore = $services->getRevisionStore();
+
+		if ( $this->getOption( 'all' ) ) {
+			$conds = [];
+
+		} elseif ( $this->getOption( 'page' ) ) {
+			$linkBatch = $services->getLinkBatchFactory()->newLinkBatch();
+			foreach ( $this->getOption( 'page' ) as $page ) {
+				$linkBatch->addObj( Title::newFromText( $page ) );
+			}
+			$pageIds = array_map( static function ( $page ) {
+				return $page->getId();
+			}, $linkBatch->getPageIdentities() );
+
+			$conds = [ 'rev_page' => $pageIds ];
+
+		} elseif ( $this->getOption( 'rev' ) ) {
+			$conds = [ 'rev_id' => $this->getOption( 'rev' ) ];
+		} else {
+			$this->error( "One of 'all', 'page', or 'rev' required" );
+			$this->maybeHelp( true );
+			return;
+		}
+
+		$this->runTable( [
+			'table' => 'revision',
+			'conds' => $conds,
+			'index' => [ 'rev_page', 'rev_timestamp', 'rev_id' ],
+			'callback' => 'processRow',
+		] );
+	}
+
+	/**
+	 * @param stdClass $row Database table row
+	 */
+	protected function processRow( stdClass $row ) {
+		$changed = false;
+		try {
+			// HACK (because we don't query the table this data ordinarily comes from,
+			// and we don't care about edit summaries here)
+			$row->rev_comment_text = '';
+			$row->rev_comment_data = null;
+			$row->rev_comment_cid = null;
+
+			$rev = $this->revStore->newRevisionFromRow( $row );
+			$title = Title::newFromLinkTarget(
+				$rev->getPageAsLinkTarget()
+			);
+			if ( HookUtils::isAvailableForTitle( $title ) ) {
+				$threadItemSet = HookUtils::parseRevisionParsoidHtml( $rev );
+
+				if ( !$this->dryrun ) {
+					// Store permalink data
+					$changed = $this->itemStore->insertThreadItems( $rev, $threadItemSet );
+				}
+			}
+		} catch ( Throwable $e ) {
+			$this->output( "Error while processing revid=$row->rev_id, pageid=$row->rev_page\n" );
+			MWExceptionRenderer::output( $e, MWExceptionRenderer::AS_RAW );
+		}
+		$this->progress( (int)$changed );
+	}
+}
+
+$maintClass = PersistRevisionThreadItems::class;
+require_once RUN_MAINTENANCE_IF_MAIN;
--- a/sql/discussiontools_persistent.json
+++ b/sql/discussiontools_persistent.json
@ -0,0 +1,260 @@
+[
+	{
+		"name": "discussiontools_items",
+		"columns": [
+			{
+				"name": "it_id",
+				"type": "integer",
+				"options": {
+					"autoincrement": true,
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "it_itemname",
+				"comment": "Internal name used to identify this item across all pages and revisions where it might appear, see CommentParser::computeName()",
+				"type": "binary",
+				"options": {
+					"notnull": true,
+					"length": 255
+				}
+			},
+			{
+				"name": "it_timestamp",
+				"comment": "Date and time from signature, or null for headings",
+				"type": "mwtimestamp",
+				"options": {
+					"notnull": false
+				}
+			},
+			{
+				"name": "it_actor",
+				"comment": "Author from signature, or null for headings. (key to actor.actor_id)",
+				"type": "bigint",
+				"options": {
+					"unsigned": true,
+					"notnull": false
+				}
+			}
+		],
+		"indexes": [
+			{
+				"name": "it_itemname",
+				"comment": "",
+				"columns": [
+					"it_itemname"
+				],
+				"unique": true
+			}
+		],
+		"pk": [
+			"it_id"
+		]
+	},
+	{
+		"name": "discussiontools_item_ids",
+		"columns": [
+			{
+				"name": "itid_id",
+				"type": "integer",
+				"options": {
+					"autoincrement": true,
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itid_itemid",
+				"comment": "Internal ID used to identify this item in this revision, see CommentParser::computeId()",
+				"type": "binary",
+				"options": {
+					"notnull": true,
+					"length": 255
+				}
+			}
+		],
+		"indexes": [
+			{
+				"name": "itid_itemid",
+				"comment": "",
+				"columns": [
+					"itid_itemid"
+				],
+				"unique": true
+			}
+		],
+		"pk": [
+			"itid_id"
+		]
+	},
+	{
+		"name": "discussiontools_item_pages",
+		"comment": "Data in this table is redundant to discussiontools_item_revisions, but if we want to find all pages where a given comment has appeared, querying it from that table would require some awful joins and be expensive",
+		"columns": [
+			{
+				"name": "itp_id",
+				"type": "integer",
+				"options": {
+					"autoincrement": true,
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itp_items_id",
+				"comment": "(key to discussiontools_items.it_id)",
+				"type": "integer",
+				"options": {
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itp_page_id",
+				"comment": "(key to page.page_id)",
+				"type": "bigint",
+				"options": {
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itp_oldest_revision_id",
+				"comment": "(key to revision.rev_id)",
+				"type": "bigint",
+				"options": {
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itp_newest_revision_id",
+				"comment": "(key to revision.rev_id)",
+				"type": "bigint",
+				"options": {
+					"unsigned": true,
+					"notnull": true
+				}
+			}
+		],
+		"indexes": [
+			{
+				"name": "itp_items_id_page_id",
+				"comment": "insertThreadItems()",
+				"columns": [
+					"itp_items_id",
+					"itp_page_id"
+				],
+				"unique": true
+			},
+			{
+				"name": "itp_items_id_newest_revision_id",
+				"comment": "findNewestRevisionsByName(), findNewestRevisionsById()",
+				"columns": [
+					"itp_items_id",
+					"itp_newest_revision_id"
+				],
+				"unique": true
+			}
+		],
+		"pk": [
+			"itp_id"
+		]
+	},
+	{
+		"name": "discussiontools_item_revisions",
+		"comment": "",
+		"columns": [
+			{
+				"name": "itr_id",
+				"type": "integer",
+				"options": {
+					"autoincrement": true,
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itr_itemid_id",
+				"comment": "(key to discussiontools_item_ids.itid_id)",
+				"type": "integer",
+				"options": {
+					"notnull": true
+				}
+			},
+			{
+				"name": "itr_revision_id",
+				"comment": "(key to revision.rev_id)",
+				"type": "bigint",
+				"options": {
+					"unsigned": true,
+					"notnull": true
+				}
+			},
+			{
+				"name": "itr_items_id",
+				"comment": "(key to discussiontools_items.it_id)",
+				"type": "integer",
+				"options": {
+					"notnull": true
+				}
+			},
+			{
+				"name": "itr_parent_id",
+				"comment": "(key to discussiontools_item_revisions.itr_id)",
+				"type": "integer",
+				"options": {
+					"notnull": false
+				}
+			},
+			{
+				"name": "itr_transcludedfrom",
+				"comment": "Page where this item is transcluded from (null: not transcluded, 0: transcluded from unknown page) (key to page.page_id)",
+				"type": "bigint",
+				"options": {
+					"unsigned": true,
+					"notnull": false
+				}
+			},
+			{
+				"name": "itr_level",
+				"comment": "Indentation level (0 for headings, 1+ for comments)",
+				"type": "mwtinyint",
+				"options": {
+					"notnull": true
+				}
+			},
+			{
+				"name": "itr_headinglevel",
+				"comment": "Heading level (1-6), or null for placeholder headings and comments",
+				"type": "mwtinyint",
+				"options": {
+					"notnull": false
+				}
+			}
+		],
+		"indexes": [
+			{
+				"name": "itr_revision_id",
+				"comment": "findThreadItemsInCurrentRevision()",
+				"columns": [
+					"itr_revision_id"
+				],
+				"unique": false
+			},
+			{
+				"name": "itr_itemid_id_revision_id",
+				"comment": "findNewestRevisionsByName(), findNewestRevisionsById(), insertThreadItems()",
+				"columns": [
+					"itr_itemid_id",
+					"itr_revision_id"
+				],
+				"unique": true
+			}
+		],
+		"pk": [
+			"itr_id"
+		]
+	}
+]
--- a/sql/mysql/discussiontools_persistent.sql
+++ b/sql/mysql/discussiontools_persistent.sql
@ -0,0 +1,49 @@
+-- This file is automatically generated using maintenance/generateSchemaSql.php.
+-- Source: sql/discussiontools_persistent.json
+-- Do not modify this file directly.
+-- See https://www.mediawiki.org/wiki/Manual:Schema_changes
+CREATE TABLE /*_*/discussiontools_items (
+  it_id INT UNSIGNED AUTO_INCREMENT NOT NULL,
+  it_itemname VARBINARY(255) NOT NULL,
+  it_timestamp BINARY(14) DEFAULT NULL,
+  it_actor BIGINT UNSIGNED DEFAULT NULL,
+  UNIQUE INDEX it_itemname (it_itemname),
+  PRIMARY KEY(it_id)
+) /*$wgDBTableOptions*/;
+
+
+CREATE TABLE /*_*/discussiontools_item_ids (
+  itid_id INT UNSIGNED AUTO_INCREMENT NOT NULL,
+  itid_itemid VARBINARY(255) NOT NULL,
+  UNIQUE INDEX itid_itemid (itid_itemid),
+  PRIMARY KEY(itid_id)
+) /*$wgDBTableOptions*/;
+
+
+CREATE TABLE /*_*/discussiontools_item_pages (
+  itp_id INT UNSIGNED AUTO_INCREMENT NOT NULL,
+  itp_items_id INT UNSIGNED NOT NULL,
+  itp_page_id BIGINT UNSIGNED NOT NULL,
+  itp_oldest_revision_id BIGINT UNSIGNED NOT NULL,
+  itp_newest_revision_id BIGINT UNSIGNED NOT NULL,
+  UNIQUE INDEX itp_items_id_page_id (itp_items_id, itp_page_id),
+  UNIQUE INDEX itp_items_id_newest_revision_id (
+    itp_items_id, itp_newest_revision_id
+  ),
+  PRIMARY KEY(itp_id)
+) /*$wgDBTableOptions*/;
+
+
+CREATE TABLE /*_*/discussiontools_item_revisions (
+  itr_id INT UNSIGNED AUTO_INCREMENT NOT NULL,
+  itr_itemid_id INT NOT NULL,
+  itr_revision_id BIGINT UNSIGNED NOT NULL,
+  itr_items_id INT NOT NULL,
+  itr_parent_id INT DEFAULT NULL,
+  itr_transcludedfrom BIGINT UNSIGNED DEFAULT NULL,
+  itr_level TINYINT NOT NULL,
+  itr_headinglevel TINYINT DEFAULT NULL,
+  INDEX itr_revision_id (itr_revision_id),
+  UNIQUE INDEX itr_itemid_id_revision_id (itr_itemid_id, itr_revision_id),
+  PRIMARY KEY(itr_id)
+) /*$wgDBTableOptions*/;
--- a/sql/postgres/discussiontools_persistent.sql
+++ b/sql/postgres/discussiontools_persistent.sql
@ -0,0 +1,55 @@
+-- This file is automatically generated using maintenance/generateSchemaSql.php.
+-- Source: sql/discussiontools_persistent.json
+-- Do not modify this file directly.
+-- See https://www.mediawiki.org/wiki/Manual:Schema_changes
+CREATE TABLE discussiontools_items (
+  it_id SERIAL NOT NULL,
+  it_itemname TEXT NOT NULL,
+  it_timestamp TIMESTAMPTZ DEFAULT NULL,
+  it_actor BIGINT DEFAULT NULL,
+  PRIMARY KEY(it_id)
+);
+
+CREATE UNIQUE INDEX it_itemname ON discussiontools_items (it_itemname);
+
+
+CREATE TABLE discussiontools_item_ids (
+  itid_id SERIAL NOT NULL,
+  itid_itemid TEXT NOT NULL,
+  PRIMARY KEY(itid_id)
+);
+
+CREATE UNIQUE INDEX itid_itemid ON discussiontools_item_ids (itid_itemid);
+
+
+CREATE TABLE discussiontools_item_pages (
+  itp_id SERIAL NOT NULL,
+  itp_items_id INT NOT NULL,
+  itp_page_id BIGINT NOT NULL,
+  itp_oldest_revision_id BIGINT NOT NULL,
+  itp_newest_revision_id BIGINT NOT NULL,
+  PRIMARY KEY(itp_id)
+);
+
+CREATE UNIQUE INDEX itp_items_id_page_id ON discussiontools_item_pages (itp_items_id, itp_page_id);
+
+CREATE UNIQUE INDEX itp_items_id_newest_revision_id ON discussiontools_item_pages (
+  itp_items_id, itp_newest_revision_id
+);
+
+
+CREATE TABLE discussiontools_item_revisions (
+  itr_id SERIAL NOT NULL,
+  itr_itemid_id INT NOT NULL,
+  itr_revision_id BIGINT NOT NULL,
+  itr_items_id INT NOT NULL,
+  itr_parent_id INT DEFAULT NULL,
+  itr_transcludedfrom BIGINT DEFAULT NULL,
+  itr_level SMALLINT NOT NULL,
+  itr_headinglevel SMALLINT DEFAULT NULL,
+  PRIMARY KEY(itr_id)
+);
+
+CREATE INDEX itr_revision_id ON discussiontools_item_revisions (itr_revision_id);
+
+CREATE UNIQUE INDEX itr_itemid_id_revision_id ON discussiontools_item_revisions (itr_itemid_id, itr_revision_id);
--- a/sql/sqlite/discussiontools_persistent.sql
+++ b/sql/sqlite/discussiontools_persistent.sql
@ -0,0 +1,47 @@
+-- This file is automatically generated using maintenance/generateSchemaSql.php.
+-- Source: sql/discussiontools_persistent.json
+-- Do not modify this file directly.
+-- See https://www.mediawiki.org/wiki/Manual:Schema_changes
+CREATE TABLE /*_*/discussiontools_items (
+  it_id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
+  it_itemname BLOB NOT NULL, it_timestamp BLOB DEFAULT NULL,
+  it_actor BIGINT UNSIGNED DEFAULT NULL
+);
+
+CREATE UNIQUE INDEX it_itemname ON /*_*/discussiontools_items (it_itemname);
+
+
+CREATE TABLE /*_*/discussiontools_item_ids (
+  itid_id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
+  itid_itemid BLOB NOT NULL
+);
+
+CREATE UNIQUE INDEX itid_itemid ON /*_*/discussiontools_item_ids (itid_itemid);
+
+
+CREATE TABLE /*_*/discussiontools_item_pages (
+  itp_id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
+  itp_items_id INTEGER UNSIGNED NOT NULL,
+  itp_page_id BIGINT UNSIGNED NOT NULL,
+  itp_oldest_revision_id BIGINT UNSIGNED NOT NULL,
+  itp_newest_revision_id BIGINT UNSIGNED NOT NULL
+);
+
+CREATE UNIQUE INDEX itp_items_id_page_id ON /*_*/discussiontools_item_pages (itp_items_id, itp_page_id);
+
+CREATE UNIQUE INDEX itp_items_id_newest_revision_id ON /*_*/discussiontools_item_pages (
+  itp_items_id, itp_newest_revision_id
+);
+
+
+CREATE TABLE /*_*/discussiontools_item_revisions (
+  itr_id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
+  itr_itemid_id INTEGER NOT NULL, itr_revision_id BIGINT UNSIGNED NOT NULL,
+  itr_items_id INTEGER NOT NULL, itr_parent_id INTEGER DEFAULT NULL,
+  itr_transcludedfrom BIGINT UNSIGNED DEFAULT NULL,
+  itr_level SMALLINT NOT NULL, itr_headinglevel SMALLINT DEFAULT NULL
+);
+
+CREATE INDEX itr_revision_id ON /*_*/discussiontools_item_revisions (itr_revision_id);
+
+CREATE UNIQUE INDEX itr_itemid_id_revision_id ON /*_*/discussiontools_item_revisions (itr_itemid_id, itr_revision_id);
--- a/tests/cases/ThreadItemStore/1simple-example/discussiontools_item_ids.json
+++ b/tests/cases/ThreadItemStore/1simple-example/discussiontools_item_ids.json
@ -0,0 +1,18 @@
+[
+	{
+		"itid_id": "1",
+		"itid_itemid": "h-A-20220720010100"
+	},
+	{
+		"itid_id": "2",
+		"itid_itemid": "c-X-20220720010100-A"
+	},
+	{
+		"itid_id": "3",
+		"itid_itemid": "c-Y-20220720010200-X-20220720010100"
+	},
+	{
+		"itid_id": "4",
+		"itid_itemid": "c-Z-20220720010300-Y-20220720010200"
+	}
+]
--- a/tests/cases/ThreadItemStore/1simple-example/discussiontools_item_pages.json
+++ b/tests/cases/ThreadItemStore/1simple-example/discussiontools_item_pages.json
@ -0,0 +1,30 @@
+[
+	{
+		"itp_id": "1",
+		"itp_items_id": "1",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "2",
+		"itp_items_id": "2",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "3",
+		"itp_items_id": "3",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "4",
+		"itp_items_id": "4",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "4"
+	}
+]
--- a/tests/cases/ThreadItemStore/1simple-example/discussiontools_item_revisions.json
+++ b/tests/cases/ThreadItemStore/1simple-example/discussiontools_item_revisions.json
@ -0,0 +1,72 @@
+[
+	{
+		"itr_id": "1",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "2",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "2",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "2",
+		"itr_items_id": "2",
+		"itr_parent_id": "1",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "5",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "3",
+		"itr_items_id": "3",
+		"itr_parent_id": "4",
+		"itr_transcludedfrom": null,
+		"itr_level": "2",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "6",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "4",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "7",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "4",
+		"itr_items_id": "2",
+		"itr_parent_id": "6",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "8",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "4",
+		"itr_items_id": "3",
+		"itr_parent_id": "7",
+		"itr_transcludedfrom": null,
+		"itr_level": "2",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "9",
+		"itr_itemid_id": "4",
+		"itr_revision_id": "4",
+		"itr_items_id": "4",
+		"itr_parent_id": "8",
+		"itr_transcludedfrom": null,
+		"itr_level": "3",
+		"itr_headinglevel": null
+	}
+]
--- a/tests/cases/ThreadItemStore/1simple-example/discussiontools_items.json
+++ b/tests/cases/ThreadItemStore/1simple-example/discussiontools_items.json
@ -0,0 +1,26 @@
+[
+	{
+		"it_id": "1",
+		"it_itemname": "h-X-20220720010100",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "2",
+		"it_itemname": "c-X-20220720010100",
+		"it_timestamp": "20220720010100",
+		"it_actor": "2"
+	},
+	{
+		"it_id": "3",
+		"it_itemname": "c-Y-20220720010200",
+		"it_timestamp": "20220720010200",
+		"it_actor": "3"
+	},
+	{
+		"it_id": "4",
+		"it_itemname": "c-Z-20220720010300",
+		"it_timestamp": "20220720010300",
+		"it_actor": "4"
+	}
+]
--- a/tests/cases/ThreadItemStore/1simple-example/dump.xml
+++ b/tests/cases/ThreadItemStore/1simple-example/dump.xml
@ -0,0 +1,33 @@
+<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.11/ http://www.mediawiki.org/xml/export-0.11.xsd" version="0.11" xml:lang="en">
+  <page>
+    <title>Talk:ThreadItemStore1</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 01:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T01:01:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Y</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 01:01, 20 July 2022 (UTC)
+:C. --[[User:Y]] 01:02, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T01:02:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 01:01, 20 July 2022 (UTC)
+:C. --[[User:Y]] 01:02, 20 July 2022 (UTC)
+::D. --[[User:Z]] 01:03, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T01:03:00Z</timestamp>
+    </revision>
+  </page>
+</mediawiki>
--- a/tests/cases/ThreadItemStore/2archived-section/discussiontools_item_ids.json
+++ b/tests/cases/ThreadItemStore/2archived-section/discussiontools_item_ids.json
@ -0,0 +1,10 @@
+[
+	{
+		"itid_id": "1",
+		"itid_itemid": "h-A-20220720020100"
+	},
+	{
+		"itid_id": "2",
+		"itid_itemid": "c-X-20220720020100-A"
+	}
+]
--- a/tests/cases/ThreadItemStore/2archived-section/discussiontools_item_pages.json
+++ b/tests/cases/ThreadItemStore/2archived-section/discussiontools_item_pages.json
@ -0,0 +1,30 @@
+[
+	{
+		"itp_id": "1",
+		"itp_items_id": "1",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "2"
+	},
+	{
+		"itp_id": "2",
+		"itp_items_id": "2",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "2"
+	},
+	{
+		"itp_id": "3",
+		"itp_items_id": "1",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "4",
+		"itp_items_id": "2",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "4"
+	}
+]
--- a/tests/cases/ThreadItemStore/2archived-section/discussiontools_item_revisions.json
+++ b/tests/cases/ThreadItemStore/2archived-section/discussiontools_item_revisions.json
@ -0,0 +1,42 @@
+[
+	{
+		"itr_id": "1",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "2",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "2",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "2",
+		"itr_items_id": "2",
+		"itr_parent_id": "1",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "3",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "4",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "4",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "4",
+		"itr_items_id": "2",
+		"itr_parent_id": "3",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	}
+]
--- a/tests/cases/ThreadItemStore/2archived-section/discussiontools_items.json
+++ b/tests/cases/ThreadItemStore/2archived-section/discussiontools_items.json
@ -0,0 +1,14 @@
+[
+	{
+		"it_id": "1",
+		"it_itemname": "h-X-20220720020100",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "2",
+		"it_itemname": "c-X-20220720020100",
+		"it_timestamp": "20220720020100",
+		"it_actor": "2"
+	}
+]
--- a/tests/cases/ThreadItemStore/2archived-section/dump.xml
+++ b/tests/cases/ThreadItemStore/2archived-section/dump.xml
@ -0,0 +1,33 @@
+<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.11/ http://www.mediawiki.org/xml/export-0.11.xsd" version="0.11" xml:lang="en">
+  <page>
+    <title>Talk:ThreadItemStore2</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 02:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T02:01:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve" />
+      <timestamp>2022-07-20T02:02:00Z</timestamp>
+    </revision>
+  </page>
+  <page>
+    <title>Talk:ThreadItemStore2/Archive</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 02:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T02:03:00Z</timestamp>
+    </revision>
+  </page>
+</mediawiki>
--- a/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_item_ids.json
+++ b/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_item_ids.json
@ -0,0 +1,18 @@
+[
+	{
+		"itid_id": "1",
+		"itid_itemid": "h-A-20220720030100"
+	},
+	{
+		"itid_id": "2",
+		"itid_itemid": "c-X-20220720030100-A"
+	},
+	{
+		"itid_id": "3",
+		"itid_itemid": "h-C-20220720030100"
+	},
+	{
+		"itid_id": "4",
+		"itid_itemid": "c-X-20220720030100-C"
+	}
+]
--- a/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_item_pages.json
+++ b/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_item_pages.json
@ -0,0 +1,30 @@
+[
+	{
+		"itp_id": "1",
+		"itp_items_id": "1",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "2"
+	},
+	{
+		"itp_id": "2",
+		"itp_items_id": "2",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "2"
+	},
+	{
+		"itp_id": "3",
+		"itp_items_id": "1",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "3"
+	},
+	{
+		"itp_id": "4",
+		"itp_items_id": "2",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "3"
+	}
+]
--- a/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_item_revisions.json
+++ b/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_item_revisions.json
@ -0,0 +1,42 @@
+[
+	{
+		"itr_id": "1",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "2",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "2",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "2",
+		"itr_items_id": "2",
+		"itr_parent_id": "1",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "3",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "3",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "4",
+		"itr_itemid_id": "4",
+		"itr_revision_id": "3",
+		"itr_items_id": "2",
+		"itr_parent_id": "3",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	}
+]
--- a/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_items.json
+++ b/tests/cases/ThreadItemStore/3indistinguishable-comments/discussiontools_items.json
@ -0,0 +1,14 @@
+[
+	{
+		"it_id": "1",
+		"it_itemname": "h-X-20220720030100",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "2",
+		"it_itemname": "c-X-20220720030100",
+		"it_timestamp": "20220720030100",
+		"it_actor": "2"
+	}
+]
--- a/tests/cases/ThreadItemStore/3indistinguishable-comments/dump.xml
+++ b/tests/cases/ThreadItemStore/3indistinguishable-comments/dump.xml
@ -0,0 +1,26 @@
+<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.11/ http://www.mediawiki.org/xml/export-0.11.xsd" version="0.11" xml:lang="en">
+  <page>
+    <title>Talk:ThreadItemStore3</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 03:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T03:01:00Z</timestamp>
+    </revision>
+  </page>
+  <page>
+    <title>Talk:ThreadItemStore3b</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== C ==
+D. --[[User:X]] 03:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T03:01:01Z</timestamp>
+    </revision>
+  </page>
+</mediawiki>
--- a/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_item_ids.json
+++ b/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_item_ids.json
@ -0,0 +1,14 @@
+[
+	{
+		"itid_id": "1",
+		"itid_itemid": "h-A-20220720040100"
+	},
+	{
+		"itid_id": "2",
+		"itid_itemid": "c-X-20220720040100-A"
+	},
+	{
+		"itid_id": "3",
+		"itid_itemid": "c-Y-20220720040200-X-20220720040100"
+	}
+]
--- a/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_item_pages.json
+++ b/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_item_pages.json
@ -0,0 +1,44 @@
+[
+	{
+		"itp_id": "1",
+		"itp_items_id": "1",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "3"
+	},
+	{
+		"itp_id": "2",
+		"itp_items_id": "2",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "3"
+	},
+	{
+		"itp_id": "3",
+		"itp_items_id": "3",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "3"
+	},
+	{
+		"itp_id": "4",
+		"itp_items_id": "1",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "5",
+		"itp_items_id": "2",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "6",
+		"itp_items_id": "3",
+		"itp_page_id": "3",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "4"
+	}
+]
--- a/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_item_revisions.json
+++ b/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_item_revisions.json
@ -0,0 +1,82 @@
+[
+	{
+		"itr_id": "1",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "2",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "2",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "2",
+		"itr_items_id": "2",
+		"itr_parent_id": "1",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "3",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "3",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "4",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "3",
+		"itr_items_id": "2",
+		"itr_parent_id": "3",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "5",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "3",
+		"itr_items_id": "3",
+		"itr_parent_id": "4",
+		"itr_transcludedfrom": null,
+		"itr_level": "2",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "6",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "4",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": "2",
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "7",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "4",
+		"itr_items_id": "2",
+		"itr_parent_id": "6",
+		"itr_transcludedfrom": "2",
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "8",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "4",
+		"itr_items_id": "3",
+		"itr_parent_id": "7",
+		"itr_transcludedfrom": "2",
+		"itr_level": "2",
+		"itr_headinglevel": null
+	}
+]
--- a/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_items.json
+++ b/tests/cases/ThreadItemStore/4transcluded-section/discussiontools_items.json
@ -0,0 +1,20 @@
+[
+	{
+		"it_id": "1",
+		"it_itemname": "h-X-20220720040100",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "2",
+		"it_itemname": "c-X-20220720040100",
+		"it_timestamp": "20220720040100",
+		"it_actor": "2"
+	},
+	{
+		"it_id": "3",
+		"it_itemname": "c-Y-20220720040200",
+		"it_timestamp": "20220720040200",
+		"it_actor": "3"
+	}
+]
--- a/tests/cases/ThreadItemStore/4transcluded-section/dump.xml
+++ b/tests/cases/ThreadItemStore/4transcluded-section/dump.xml
@ -0,0 +1,34 @@
+<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.11/ http://www.mediawiki.org/xml/export-0.11.xsd" version="0.11" xml:lang="en">
+  <page>
+    <title>Talk:ThreadItemStore4/b</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 04:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T04:01:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Y</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 04:01, 20 July 2022 (UTC)
+:C. --[[User:Y]] 04:02, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T04:02:00Z</timestamp>
+    </revision>
+  </page>
+  <page>
+    <title>Talk:ThreadItemStore4</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve">{{Talk:ThreadItemStore4/b}}</text>
+      <timestamp>2022-07-20T04:03:00Z</timestamp>
+    </revision>
+  </page>
+</mediawiki>
--- a/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_item_ids.json
+++ b/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_item_ids.json
@ -0,0 +1,22 @@
+[
+	{
+		"itid_id": "1",
+		"itid_itemid": "h-A-20220720050100"
+	},
+	{
+		"itid_id": "2",
+		"itid_itemid": "c-X-20220720050100-A"
+	},
+	{
+		"itid_id": "3",
+		"itid_itemid": "c-Y-20220720050200-X-20220720050100"
+	},
+	{
+		"itid_id": "4",
+		"itid_itemid": "c-Z-20220720050300-Y-20220720050200"
+	},
+	{
+		"itid_id": "5",
+		"itid_itemid": "c-Z-20220720050300-X-20220720050100"
+	}
+]
--- a/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_item_pages.json
+++ b/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_item_pages.json
@ -0,0 +1,30 @@
+[
+	{
+		"itp_id": "1",
+		"itp_items_id": "1",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "5"
+	},
+	{
+		"itp_id": "2",
+		"itp_items_id": "2",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "5"
+	},
+	{
+		"itp_id": "3",
+		"itp_items_id": "3",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "5"
+	},
+	{
+		"itp_id": "4",
+		"itp_items_id": "4",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "4",
+		"itp_newest_revision_id": "5"
+	}
+]
--- a/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_item_revisions.json
+++ b/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_item_revisions.json
@ -0,0 +1,82 @@
+[
+	{
+		"itr_id": "1",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "2",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "2",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "2",
+		"itr_items_id": "2",
+		"itr_parent_id": "1",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "5",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "3",
+		"itr_items_id": "3",
+		"itr_parent_id": "4",
+		"itr_transcludedfrom": null,
+		"itr_level": "2",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "9",
+		"itr_itemid_id": "4",
+		"itr_revision_id": "4",
+		"itr_items_id": "4",
+		"itr_parent_id": "8",
+		"itr_transcludedfrom": null,
+		"itr_level": "3",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "10",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "5",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "11",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "5",
+		"itr_items_id": "2",
+		"itr_parent_id": "10",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "12",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "5",
+		"itr_items_id": "3",
+		"itr_parent_id": "11",
+		"itr_transcludedfrom": null,
+		"itr_level": "2",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "13",
+		"itr_itemid_id": "5",
+		"itr_revision_id": "5",
+		"itr_items_id": "4",
+		"itr_parent_id": "11",
+		"itr_transcludedfrom": null,
+		"itr_level": "2",
+		"itr_headinglevel": null
+	}
+]
--- a/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_items.json
+++ b/tests/cases/ThreadItemStore/5changed-comment-indentation/discussiontools_items.json
@ -0,0 +1,26 @@
+[
+	{
+		"it_id": "1",
+		"it_itemname": "h-X-20220720050100",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "2",
+		"it_itemname": "c-X-20220720050100",
+		"it_timestamp": "20220720050100",
+		"it_actor": "2"
+	},
+	{
+		"it_id": "3",
+		"it_itemname": "c-Y-20220720050200",
+		"it_timestamp": "20220720050200",
+		"it_actor": "3"
+	},
+	{
+		"it_id": "4",
+		"it_itemname": "c-Z-20220720050300",
+		"it_timestamp": "20220720050300",
+		"it_actor": "4"
+	}
+]
--- a/tests/cases/ThreadItemStore/5changed-comment-indentation/dump.xml
+++ b/tests/cases/ThreadItemStore/5changed-comment-indentation/dump.xml
@ -0,0 +1,43 @@
+<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.11/ http://www.mediawiki.org/xml/export-0.11.xsd" version="0.11" xml:lang="en">
+  <page>
+    <title>Talk:ThreadItemStore5</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 05:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T05:01:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Y</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 05:01, 20 July 2022 (UTC)
+:C. --[[User:Y]] 05:02, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T05:02:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 05:01, 20 July 2022 (UTC)
+:C. --[[User:Y]] 05:02, 20 July 2022 (UTC)
+::D. --[[User:Z]] 05:03, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T05:03:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 05:01, 20 July 2022 (UTC)
+:C. --[[User:Y]] 05:02, 20 July 2022 (UTC)
+:D. --[[User:Z]] 05:03, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T05:03:01Z</timestamp>
+    </revision>
+  </page>
+</mediawiki>
--- a/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_item_ids.json
+++ b/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_item_ids.json
@ -0,0 +1,22 @@
+[
+	{
+		"itid_id": "1",
+		"itid_itemid": "h-A-20220720060100"
+	},
+	{
+		"itid_id": "2",
+		"itid_itemid": "c-X-20220720060100-A"
+	},
+	{
+		"itid_id": "3",
+		"itid_itemid": "h-C-20220720060200"
+	},
+	{
+		"itid_id": "4",
+		"itid_itemid": "c-Y-20220720060200-C"
+	},
+	{
+		"itid_id": "5",
+		"itid_itemid": "h-C-A-20220720060200"
+	}
+]
--- a/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_item_pages.json
+++ b/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_item_pages.json
@ -0,0 +1,30 @@
+[
+	{
+		"itp_id": "1",
+		"itp_items_id": "1",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "2",
+		"itp_items_id": "2",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "2",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "3",
+		"itp_items_id": "3",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "4"
+	},
+	{
+		"itp_id": "4",
+		"itp_items_id": "4",
+		"itp_page_id": "2",
+		"itp_oldest_revision_id": "3",
+		"itp_newest_revision_id": "4"
+	}
+]
--- a/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_item_revisions.json
+++ b/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_item_revisions.json
@ -0,0 +1,82 @@
+[
+	{
+		"itr_id": "1",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "2",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "2",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "2",
+		"itr_items_id": "2",
+		"itr_parent_id": "1",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "5",
+		"itr_itemid_id": "3",
+		"itr_revision_id": "3",
+		"itr_items_id": "3",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "6",
+		"itr_itemid_id": "4",
+		"itr_revision_id": "3",
+		"itr_items_id": "4",
+		"itr_parent_id": "5",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "7",
+		"itr_itemid_id": "1",
+		"itr_revision_id": "4",
+		"itr_items_id": "1",
+		"itr_parent_id": null,
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "2"
+	},
+	{
+		"itr_id": "8",
+		"itr_itemid_id": "2",
+		"itr_revision_id": "4",
+		"itr_items_id": "2",
+		"itr_parent_id": "7",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	},
+	{
+		"itr_id": "9",
+		"itr_itemid_id": "5",
+		"itr_revision_id": "4",
+		"itr_items_id": "3",
+		"itr_parent_id": "7",
+		"itr_transcludedfrom": null,
+		"itr_level": "0",
+		"itr_headinglevel": "3"
+	},
+	{
+		"itr_id": "10",
+		"itr_itemid_id": "4",
+		"itr_revision_id": "4",
+		"itr_items_id": "4",
+		"itr_parent_id": "9",
+		"itr_transcludedfrom": null,
+		"itr_level": "1",
+		"itr_headinglevel": null
+	}
+]
--- a/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_items.json
+++ b/tests/cases/ThreadItemStore/6changed-heading-level/discussiontools_items.json
@ -0,0 +1,26 @@
+[
+	{
+		"it_id": "1",
+		"it_itemname": "h-X-20220720060100",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "2",
+		"it_itemname": "c-X-20220720060100",
+		"it_timestamp": "20220720060100",
+		"it_actor": "2"
+	},
+	{
+		"it_id": "3",
+		"it_itemname": "h-Y-20220720060200",
+		"it_timestamp": null,
+		"it_actor": null
+	},
+	{
+		"it_id": "4",
+		"it_itemname": "c-Y-20220720060200",
+		"it_timestamp": "20220720060200",
+		"it_actor": "3"
+	}
+]
--- a/tests/cases/ThreadItemStore/6changed-heading-level/dump.xml
+++ b/tests/cases/ThreadItemStore/6changed-heading-level/dump.xml
@ -0,0 +1,34 @@
+<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.11/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.11/ http://www.mediawiki.org/xml/export-0.11.xsd" version="0.11" xml:lang="en">
+  <page>
+    <title>Talk:ThreadItemStore6</title>
+    <ns>1</ns>
+    <revision>
+      <contributor>
+        <username>X</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 06:01, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T06:01:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Y</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 06:01, 20 July 2022 (UTC)
+== C ==
+D. --[[User:Y]] 06:02, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T06:02:00Z</timestamp>
+    </revision>
+    <revision>
+      <contributor>
+        <username>Z</username>
+      </contributor>
+      <text xml:space="preserve">== A ==
+B. --[[User:X]] 06:01, 20 July 2022 (UTC)
+=== C ===
+D. --[[User:Y]] 06:02, 20 July 2022 (UTC)</text>
+      <timestamp>2022-07-20T06:02:01Z</timestamp>
+    </revision>
+  </page>
+</mediawiki>
--- a/tests/phpunit/ThreadItemStoreTest.php
+++ b/tests/phpunit/ThreadItemStoreTest.php
@ -0,0 +1,104 @@
+<?php
+
+namespace MediaWiki\Extension\DiscussionTools\Tests;
+
+use ImportStringSource;
+use MediaWiki\MediaWikiServices;
+use TestUser;
+
+/**
+ * @coversDefaultClass \MediaWiki\Extension\DiscussionTools\ThreadItemStore
+ *
+ * @group DiscussionTools
+ * @group Database
+ */
+class ThreadItemStoreTest extends IntegrationTestCase {
+
+	/**
+	 * @inheritDoc
+	 */
+	public function getCliArg( $offset ) {
+		// Work around MySQL bug (T256006)
+		if ( $offset === 'use-normal-tables' ) {
+			return true;
+		}
+		return parent::getCliArg( $offset );
+	}
+
+	/** @var @inheritDoc */
+	protected $tablesUsed = [
+		'user',
+		'page',
+		'revision',
+		'discussiontools_items',
+		'discussiontools_item_pages',
+		'discussiontools_item_revisions',
+		'discussiontools_item_ids',
+	];
+
+	/**
+	 * @dataProvider provideInsertCases
+	 * @covers ::insertThreadItems
+	 */
+	public function testInsertThreadItems( string $dir ): void {
+		// Create users for the imported revisions
+		new TestUser( 'X' );
+		new TestUser( 'Y' );
+		new TestUser( 'Z' );
+
+		// Import revisions
+		$source = new ImportStringSource( static::getText( "$dir/dump.xml" ) );
+		$importer = MediaWikiServices::getInstance()
+			->getWikiImporterFactory()
+			->getWikiImporter( $source );
+		// `true` means to assign edits to the users we created above
+		$importer->setUsernamePrefix( 'import', true );
+		$importer->doImport();
+
+		// Check that expected data has been stored in the database
+		$expected = [];
+		$actual = [];
+		$tables = [
+			'discussiontools_items',
+			'discussiontools_item_pages',
+			'discussiontools_item_revisions',
+			'discussiontools_item_ids',
+		];
+		foreach ( $tables as $table ) {
+			$expected[$table] = static::getJson( "../$dir/$table.json", true );
+
+			$res = wfGetDb( DB_REPLICA )->select(
+				$table,
+				'*',
+				[],
+				__METHOD__,
+				[ 'ORDER BY' => 1 ]
+			);
+			foreach ( $res as $i => $row ) {
+				foreach ( $row as $key => $val ) {
+					$actual[$table][$i][$key] = $val;
+				}
+			}
+		}
+
+		// Optionally write updated content to the JSON files
+		if ( getenv( 'DISCUSSIONTOOLS_OVERWRITE_TESTS' ) ) {
+			foreach ( $tables as $table ) {
+				static::overwriteJsonFile( "../$dir/$table.json", $actual[$table] );
+			}
+		}
+
+		static::assertEquals( $expected, $actual );
+	}
+
+	public function provideInsertCases(): array {
+		return [
+			[ 'cases/ThreadItemStore/1simple-example' ],
+			[ 'cases/ThreadItemStore/2archived-section' ],
+			[ 'cases/ThreadItemStore/3indistinguishable-comments' ],
+			[ 'cases/ThreadItemStore/4transcluded-section' ],
+			[ 'cases/ThreadItemStore/5changed-comment-indentation' ],
+			[ 'cases/ThreadItemStore/6changed-heading-level' ],
+		];
+	}
+}