mediawiki-extensions-DiscussionTools/includes/CommentItem.php

<?php

namespace MediaWiki\Extension\DiscussionTools;

use DateTimeImmutable;
use MediaWiki\MediaWikiServices;
use MWException;
use Sanitizer;
use Title;
use Wikimedia\Parsoid\DOM\DocumentFragment;
use Wikimedia\Parsoid\DOM\Text;
use Wikimedia\Parsoid\Utils\DOMCompat;
use Wikimedia\Parsoid\Utils\DOMUtils;

class CommentItem extends ThreadItem {
	private $signatureRanges;
	private $timestamp;
	private $author;

	/**
	 * @param int $level
	 * @param ImmutableRange $range
	 * @param ImmutableRange[] $signatureRanges Objects describing the extent of signatures (plus
	 *  timestamps) for this comment. There is always at least one signature, but there may be
	 *  multiple. The author and timestamp of the comment is determined from the first signature.
	 *  The last node in every signature range is a node containing the timestamp.
	 * @param DateTimeImmutable $timestamp
	 * @param string $author Comment author's username
	 */
	public function __construct(
		int $level, ImmutableRange $range,
		array $signatureRanges, DateTimeImmutable $timestamp, string $author
	) {
		parent::__construct( 'comment', $level, $range );
		$this->signatureRanges = $signatureRanges;
		$this->timestamp = $timestamp;
		$this->author = $author;
	}

	/**
	 * @inheritDoc
	 */
	public function jsonSerialize( bool $deep = false, ?callable $callback = null ): array {
		return array_merge( parent::jsonSerialize( $deep, $callback ), [
			'timestamp' => $this->getTimestampString(),
			'author' => $this->author,
		] );
	}

	/**
	 * @return array JSON-serializable array
	 */
	public function jsonSerializeForDiff(): array {
		$data = $this->jsonSerialize();

		$heading = $this->getHeading();
		$data['headingId'] = $heading->getId();
		$subscribableHeading = $this->getSubscribableHeading();
		$data['subscribableHeadingId'] = $subscribableHeading ? $subscribableHeading->getId() : null;

		return $data;
	}

	/**
	 * Get the HTML of this comment's body
	 *
	 * @param bool $stripTrailingSeparator Strip a trailing separator between the body and
	 *  the signature which consists of whitespace and hyphens e.g. ' --'
	 * @return DocumentFragment Cloned fragment of the body content
	 */
	private function getBodyFragment( bool $stripTrailingSeparator = false ): DocumentFragment {
		$fragment = $this->getBodyRange()->cloneContents();
		CommentModifier::unwrapFragment( $fragment );

		if ( $stripTrailingSeparator ) {
			// Find a trailing text node
			$lastChild = $fragment->lastChild;
			while (
				$lastChild &&
				!( $lastChild instanceof Text )
			) {
				$lastChild = $lastChild->lastChild;
			}
			if (
				$lastChild instanceof Text &&
				preg_match( '/[\s\-~\x{2010}-\x{2015}\x{2043}\x{2060}]+$/u', $lastChild->nodeValue ?? '', $matches )
			) {
				$lastChild->nodeValue =
					substr( $lastChild->nodeValue ?? '', 0, -strlen( $matches[0] ) );
			}
		}
		return $fragment;
	}

	/**
	 * Get the HTML of this comment's body
	 *
	 *
	 * @param bool $stripTrailingSeparator See getBodyFragment
	 * @return string HTML
	 */
	public function getBodyHTML( bool $stripTrailingSeparator = false ): string {
		$fragment = $this->getBodyFragment( $stripTrailingSeparator );
		return DOMUtils::getFragmentInnerHTML( $fragment );
	}

	/**
	 * Get the text of this comment's body
	 *
	 * @param bool $stripTrailingSeparator See getBodyFragment
	 * @return string Text
	 */
	public function getBodyText( bool $stripTrailingSeparator = false ): string {
		$html = $this->getBodyHTML( $stripTrailingSeparator );
		return Sanitizer::stripAllTags( $html );
	}

	/**
	 * Get a list of all users mentioned
	 *
	 * @return Title[] Title objects for mentioned user pages
	 */
	public function getMentions(): array {
		$fragment = $this->getBodyRange()->cloneContents();
		// Note: DOMCompat::getElementsByTagName() doesn't take a DocumentFragment argument
		$links = DOMCompat::querySelectorAll( $fragment, 'a' );
		$users = [];
		foreach ( $links as $link ) {
			$href = $link->getAttribute( 'href' );
			if ( $href ) {
				$siteConfig = MediaWikiServices::getInstance()->getMainConfig();
				$title = Title::newFromText( CommentUtils::getTitleFromUrl( $href, $siteConfig ) );
				if ( $title && $title->getNamespace() === NS_USER ) {
					// TODO: Consider returning User objects
					$users[] = $title;
				}
			}
		}
		return array_unique( $users );
	}

	/**
	 * @return ImmutableRange[] Comment signature ranges
	 */
	public function getSignatureRanges(): array {
		return $this->signatureRanges;
	}

	/**
	 * @return ImmutableRange Range of the thread item's "body"
	 */
	public function getBodyRange(): ImmutableRange {
		// Exclude last signature from body
		$signatureRanges = $this->getSignatureRanges();
		$lastSignature = end( $signatureRanges );
		return $this->getRange()->setEnd( $lastSignature->startContainer, $lastSignature->startOffset );
	}

	/**
	 * @return DateTimeImmutable Comment timestamp
	 */
	public function getTimestamp(): DateTimeImmutable {
		return $this->timestamp;
	}

	/**
	 * Get the comment timestamp in the format used in IDs and names.
	 *
	 * Depending on the date of the comment, this may use one of two formats:
	 *
	 *  - For dates prior to 'DiscussionToolsTimestampFormatSwitchTime' (by default 2022-07-12):
	 *    Uses ISO 8601 date. Almost DateTimeInterface::RFC3339_EXTENDED, but ending with 'Z' instead
	 *    of '+00:00', like Date#toISOString in JavaScript.
	 *
	 *  - For dates on or after 'DiscussionToolsTimestampFormatSwitchTime' (by default 2022-07-12):
	 *    Uses MediaWiki timestamp (TS_MW in MediaWiki PHP code).
	 *
	 * @return string Comment timestamp in standard format
	 */
	public function getTimestampString(): string {
		$dtConfig = MediaWikiServices::getInstance()->getConfigFactory()->makeConfig( 'discussiontools' );
		$switchTime = new DateTimeImmutable(
			$dtConfig->get( 'DiscussionToolsTimestampFormatSwitchTime' )
		);
		$timestamp = $this->getTimestamp();
		if ( $timestamp < $switchTime ) {
			return $timestamp->format( 'Y-m-d\TH:i:s.v\Z' );
		} else {
			return $timestamp->format( 'YmdHis' );
		}
	}

	/**
	 * @return string Comment author
	 */
	public function getAuthor(): string {
		return $this->author;
	}

	/**
	 * @return HeadingItem Closest ancestor which is a HeadingItem
	 */
	public function getHeading(): HeadingItem {
		$parent = $this;
		while ( $parent instanceof CommentItem ) {
			$parent = $parent->getParent();
		}
		if ( !( $parent instanceof HeadingItem ) ) {
			throw new MWException( 'heading parent not found' );
		}
		return $parent;
	}

	/**
	 * @return HeadingItem|null Closest heading that can be used for topic subscriptions
	 */
	public function getSubscribableHeading(): ?HeadingItem {
		$heading = $this->getHeading();
		while ( $heading instanceof HeadingItem && !$heading->isSubscribable() ) {
			$heading = $heading->getParent();
		}
		return $heading instanceof HeadingItem ? $heading : null;
	}

	/**
	 * @param ImmutableRange $signatureRange Comment signature range to add
	 */
	public function addSignatureRange( ImmutableRange $signatureRange ): void {
		$this->signatureRanges[] = $signatureRange;
	}

	/**
	 * @param ImmutableRange[] $signatureRanges Comment signature ranges
	 */
	public function setSignatureRanges( array $signatureRanges ): void {
		$this->signatureRanges = $signatureRanges;
	}

	/**
	 * @param DateTimeImmutable $timestamp Comment timestamp
	 */
	public function setTimestamp( DateTimeImmutable $timestamp ): void {
		$this->timestamp = $timestamp;
	}

	/**
	 * @param string $author Comment author
	 */
	public function setAuthor( string $author ): void {
		$this->author = $author;
	}
}