X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FCommentStore.php;h=0d679d37db02b8c480d8cbf806e86afe33c5eae0;hb=f9814f7709e75d8a50a6b363d454da26a047339d;hp=0c86c1e871a9b40e4871105faedf4124ad38395d;hpb=9bd7c7f66058fe26e130759090e7a73beee4d3d0;p=lhc%2Fweb%2Fwiklou.git

diff --git a/includes/CommentStore.php b/includes/CommentStore.php
index 0c86c1e871..0d679d37db 100644
--- a/includes/CommentStore.php
+++ b/includes/CommentStore.php
@@ -29,6 +29,26 @@ use Wikimedia\Rdbms\IDatabase;
  */
 class CommentStore {
 
+	/**
+	 * Maximum length of a comment in UTF-8 characters. Longer comments will be truncated.
+	 * @note This must be at least 255 and not greater than floor( MAX_COMMENT_LENGTH / 4 ).
+	 */
+	const COMMENT_CHARACTER_LIMIT = 1000;
+
+	/**
+	 * Maximum length of a comment in bytes. Longer comments will be truncated.
+	 * @note This value is determined by the size of the underlying database field,
+	 *  currently BLOB in MySQL/MariaDB.
+	 */
+	const MAX_COMMENT_LENGTH = 65535;
+
+	/**
+	 * Maximum length of serialized data in bytes. Longer data will result in an exception.
+	 * @note This value is determined by the size of the underlying database field,
+	 *  currently BLOB in MySQL/MariaDB.
+	 */
+	const MAX_DATA_LENGTH = 65535;
+
 	/**
 	 * Define fields that use temporary tables for transitional purposes
 	 * @var array Keys are '$key', values are arrays with four fields:
@@ -68,15 +88,21 @@ class CommentStore {
 	/** @var array|null Cache for `self::getJoin()` */
 	protected $joinCache = null;
 
+	/** @var Language Language to use for comment truncation */
+	protected $lang;
+
 	/**
 	 * @param string $key A key such as "rev_comment" identifying the comment
 	 *  field being fetched.
+	 * @param Language $lang Language to use for comment truncation. Defaults
+	 *  to $wgContLang.
 	 */
-	public function __construct( $key ) {
-		global $wgCommentTableSchemaMigrationStage;
+	public function __construct( $key, Language $lang = null ) {
+		global $wgCommentTableSchemaMigrationStage, $wgContLang;
 
 		$this->key = $key;
 		$this->stage = $wgCommentTableSchemaMigrationStage;
+		$this->lang = $lang ?: $wgContLang;
 	}
 
 	/**
@@ -355,25 +381,18 @@ class CommentStore {
 	 * @return CommentStoreComment
 	 */
 	public function createComment( IDatabase $dbw, $comment, array $data = null ) {
-		global $wgContLang;
-
-		if ( !$comment instanceof CommentStoreComment ) {
-			if ( $data !== null ) {
-				foreach ( $data as $k => $v ) {
-					if ( substr( $k, 0, 1 ) === '_' ) {
-						throw new InvalidArgumentException( 'Keys in $data beginning with "_" are reserved' );
-					}
-				}
-			}
-			if ( $comment instanceof Message ) {
-				$message = clone $comment;
-				$text = $message->inLanguage( $wgContLang ) // Avoid $wgForceUIMsgAsContentMsg
-					->setInterfaceMessageFlag( true )
-					->text();
-				$comment = new CommentStoreComment( null, $text, $message, $data );
-			} else {
-				$comment = new CommentStoreComment( null, $comment, null, $data );
+		$comment = CommentStoreComment::newUnsavedComment( $comment, $data );
+
+		# Truncate comment in a Unicode-sensitive manner
+		$comment->text = $this->lang->truncate( $comment->text, self::MAX_COMMENT_LENGTH );
+		if ( mb_strlen( $comment->text, 'UTF-8' ) > self::COMMENT_CHARACTER_LIMIT ) {
+			$ellipsis = wfMessage( 'ellipsis' )->inLanguage( $this->lang )->escaped();
+			if ( mb_strlen( $ellipsis ) >= self::COMMENT_CHARACTER_LIMIT ) {
+				// WTF?
+				$ellipsis = '...';
 			}
+			$maxLength = self::COMMENT_CHARACTER_LIMIT - mb_strlen( $ellipsis, 'UTF-8' );
+			$comment->text = mb_substr( $comment->text, 0, $maxLength, 'UTF-8' ) . $ellipsis;
 		}
 
 		if ( $this->stage > MIGRATION_OLD && !$comment->id ) {
@@ -386,6 +405,11 @@ class CommentStore {
 			}
 			if ( $dbData !== null ) {
 				$dbData = FormatJson::encode( (object)$dbData, false, FormatJson::ALL_OK );
+				$len = strlen( $dbData );
+				if ( $len > self::MAX_DATA_LENGTH ) {
+					$max = self::MAX_DATA_LENGTH;
+					throw new OverflowException( "Comment data is too long ($len bytes, maximum is $max)" );
+				}
 			}
 
 			$hash = self::hash( $comment->text, $dbData );
@@ -400,11 +424,9 @@ class CommentStore {
 				__METHOD__
 			);
 			if ( !$comment->id ) {
-				$comment->id = $dbw->nextSequenceValue( 'comment_comment_id_seq' );
 				$dbw->insert(
 					'comment',
 					[
-						'comment_id' => $comment->id,
 						'comment_hash' => $hash,
 						'comment_text' => $comment->text,
 						'comment_data' => $dbData,
@@ -432,7 +454,7 @@ class CommentStore {
 		$comment = $this->createComment( $dbw, $comment, $data );
 
 		if ( $this->stage <= MIGRATION_WRITE_BOTH ) {
-			$fields[$this->key] = $comment->text;
+			$fields[$this->key] = $this->lang->truncate( $comment->text, 255 );
 		}
 
 		if ( $this->stage >= MIGRATION_WRITE_BOTH ) {
@@ -459,7 +481,7 @@ class CommentStore {
 	}
 
 	/**
-	 * Prepare for the insertion of a row with a comment
+	 * Insert a comment in preparation for a row that references it
 	 *
 	 * @note It's recommended to include both the call to this method and the
 	 *  row insert in the same transaction.
@@ -478,7 +500,7 @@ class CommentStore {
 	}
 
 	/**
-	 * Prepare for the insertion of a row with a comment and temporary table
+	 * Insert a comment in a temporary table in preparation for a row that references it
 	 *
 	 * This is currently needed for "rev_comment" and "img_description". In the
 	 * future that requirement will be removed.