diff options
Diffstat (limited to 'includes/Revision.php')
-rw-r--r-- | includes/Revision.php | 714 |
1 files changed, 555 insertions, 159 deletions
diff --git a/includes/Revision.php b/includes/Revision.php index 20cc8f58..233eac01 100644 --- a/includes/Revision.php +++ b/includes/Revision.php @@ -25,6 +25,10 @@ */ class Revision implements IDBAccessObject { protected $mId; + + /** + * @var int|null + */ protected $mPage; protected $mUserText; protected $mOrigUserText; @@ -38,8 +42,29 @@ class Revision implements IDBAccessObject { protected $mComment; protected $mText; protected $mTextRow; + + /** + * @var null|Title + */ protected $mTitle; protected $mCurrent; + protected $mContentModel; + protected $mContentFormat; + + /** + * @var Content|null|bool + */ + protected $mContent; + + /** + * @var null|ContentHandler + */ + protected $mContentHandler; + + /** + * @var int + */ + protected $mQueryFlags = 0; // Revision deletion constants const DELETED_TEXT = 1; @@ -83,21 +108,21 @@ class Revision implements IDBAccessObject { * @param $flags Integer Bitfield (optional) * @return Revision or null */ - public static function newFromTitle( $title, $id = 0, $flags = null ) { + public static function newFromTitle( $title, $id = 0, $flags = 0 ) { $conds = array( 'page_namespace' => $title->getNamespace(), - 'page_title' => $title->getDBkey() + 'page_title' => $title->getDBkey() ); if ( $id ) { // Use the specified ID $conds['rev_id'] = $id; + return self::newFromConds( $conds, (int)$flags ); } else { // Use a join to get the latest revision $conds[] = 'rev_id=page_latest'; - // Callers assume this will be up-to-date - $flags = is_int( $flags ) ? $flags : self::READ_LATEST; // b/c + $db = wfGetDB( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_SLAVE ); + return self::loadFromConds( $db, $conds, $flags ); } - return self::newFromConds( $conds, (int)$flags ); } /** @@ -106,7 +131,7 @@ class Revision implements IDBAccessObject { * Returns null if no such revision can be found. * * $flags include: - * Revision::READ_LATEST : Select the data from the master + * Revision::READ_LATEST : Select the data from the master (since 1.20) * Revision::READ_LOCKING : Select & lock the data from the master * * @param $revId Integer @@ -114,15 +139,13 @@ class Revision implements IDBAccessObject { * @param $flags Integer Bitfield (optional) * @return Revision or null */ - public static function newFromPageId( $pageId, $revId = 0, $flags = null ) { + public static function newFromPageId( $pageId, $revId = 0, $flags = 0 ) { $conds = array( 'page_id' => $pageId ); if ( $revId ) { $conds['rev_id'] = $revId; } else { // Use a join to get the latest revision $conds[] = 'rev_id = page_latest'; - // Callers assume this will be up-to-date - $flags = is_int( $flags ) ? $flags : self::READ_LATEST; // b/c } return self::newFromConds( $conds, (int)$flags ); } @@ -135,9 +158,12 @@ class Revision implements IDBAccessObject { * @param $row * @param $overrides array * + * @throws MWException * @return Revision */ public static function newFromArchiveRow( $row, $overrides = array() ) { + global $wgContentHandlerUseDB; + $attribs = $overrides + array( 'page' => isset( $row->ar_page_id ) ? $row->ar_page_id : null, 'id' => isset( $row->ar_rev_id ) ? $row->ar_rev_id : null, @@ -150,7 +176,22 @@ class Revision implements IDBAccessObject { 'deleted' => $row->ar_deleted, 'len' => $row->ar_len, 'sha1' => isset( $row->ar_sha1 ) ? $row->ar_sha1 : null, + 'content_model' => isset( $row->ar_content_model ) ? $row->ar_content_model : null, + 'content_format' => isset( $row->ar_content_format ) ? $row->ar_content_format : null, ); + + if ( !$wgContentHandlerUseDB ) { + unset( $attribs['content_model'] ); + unset( $attribs['content_format'] ); + } + + if ( !isset( $attribs['title'] ) + && isset( $row->ar_namespace ) + && isset( $row->ar_title ) ) { + + $attribs['title'] = Title::makeTitle( $row->ar_namespace, $row->ar_title ); + } + if ( isset( $row->ar_text ) && !$row->ar_text_id ) { // Pre-1.5 ar_text row $attribs['text'] = self::getRevisionText( $row, 'ar_' ); @@ -194,8 +235,8 @@ class Revision implements IDBAccessObject { * @return Revision or null */ public static function loadFromPageId( $db, $pageid, $id = 0 ) { - $conds = array( 'rev_page' => intval( $pageid ), 'page_id' => intval( $pageid ) ); - if( $id ) { + $conds = array( 'rev_page' => intval( $pageid ), 'page_id' => intval( $pageid ) ); + if ( $id ) { $conds['rev_id'] = intval( $id ); } else { $conds[] = 'rev_id=page_latest'; @@ -214,15 +255,17 @@ class Revision implements IDBAccessObject { * @return Revision or null */ public static function loadFromTitle( $db, $title, $id = 0 ) { - if( $id ) { + if ( $id ) { $matchId = intval( $id ); } else { $matchId = 'page_latest'; } return self::loadFromConds( $db, - array( "rev_id=$matchId", - 'page_namespace' => $title->getNamespace(), - 'page_title' => $title->getDBkey() ) + array( + "rev_id=$matchId", + 'page_namespace' => $title->getNamespace(), + 'page_title' => $title->getDBkey() + ) ); } @@ -238,9 +281,11 @@ class Revision implements IDBAccessObject { */ public static function loadFromTimestamp( $db, $title, $timestamp ) { return self::loadFromConds( $db, - array( 'rev_timestamp' => $db->timestamp( $timestamp ), - 'page_namespace' => $title->getNamespace(), - 'page_title' => $title->getDBkey() ) + array( + 'rev_timestamp' => $db->timestamp( $timestamp ), + 'page_namespace' => $title->getNamespace(), + 'page_title' => $title->getDBkey() + ) ); } @@ -260,6 +305,9 @@ class Revision implements IDBAccessObject { $rev = self::loadFromConds( $dbw, $conditions, $flags ); } } + if ( $rev ) { + $rev->mQueryFlags = $flags; + } return $rev; } @@ -274,9 +322,9 @@ class Revision implements IDBAccessObject { */ private static function loadFromConds( $db, $conditions, $flags = 0 ) { $res = self::fetchFromConds( $db, $conditions, $flags ); - if( $res ) { + if ( $res ) { $row = $res->fetchObject(); - if( $row ) { + if ( $row ) { $ret = new Revision( $row ); return $ret; } @@ -296,9 +344,11 @@ class Revision implements IDBAccessObject { public static function fetchRevision( $title ) { return self::fetchFromConds( wfGetDB( DB_SLAVE ), - array( 'rev_id=page_latest', - 'page_namespace' => $title->getNamespace(), - 'page_title' => $title->getDBkey() ) + array( + 'rev_id=page_latest', + 'page_namespace' => $title->getNamespace(), + 'page_title' => $title->getDBkey() + ) ); } @@ -343,7 +393,7 @@ class Revision implements IDBAccessObject { } /** - * Return the value of a select() page conds array for the paeg table. + * Return the value of a select() page conds array for the page table. * This will assure that the revision(s) are not orphaned from live pages. * @since 1.19 * @return Array @@ -358,7 +408,9 @@ class Revision implements IDBAccessObject { * @return array */ public static function selectFields() { - return array( + global $wgContentHandlerUseDB; + + $fields = array( 'rev_id', 'rev_page', 'rev_text_id', @@ -370,8 +422,15 @@ class Revision implements IDBAccessObject { 'rev_deleted', 'rev_len', 'rev_parent_id', - 'rev_sha1' + 'rev_sha1', ); + + if ( $wgContentHandlerUseDB ) { + $fields[] = 'rev_content_format'; + $fields[] = 'rev_content_model'; + } + + return $fields; } /** @@ -436,10 +495,11 @@ class Revision implements IDBAccessObject { * Constructor * * @param $row Mixed: either a database row or an array + * @throws MWException * @access private */ function __construct( $row ) { - if( is_object( $row ) ) { + if ( is_object( $row ) ) { $this->mId = intval( $row->rev_id ); $this->mPage = intval( $row->rev_page ); $this->mTextId = intval( $row->rev_text_id ); @@ -449,13 +509,13 @@ class Revision implements IDBAccessObject { $this->mTimestamp = $row->rev_timestamp; $this->mDeleted = intval( $row->rev_deleted ); - if( !isset( $row->rev_parent_id ) ) { - $this->mParentId = is_null( $row->rev_parent_id ) ? null : 0; + if ( !isset( $row->rev_parent_id ) ) { + $this->mParentId = null; } else { - $this->mParentId = intval( $row->rev_parent_id ); + $this->mParentId = intval( $row->rev_parent_id ); } - if( !isset( $row->rev_len ) || is_null( $row->rev_len ) ) { + if ( !isset( $row->rev_len ) ) { $this->mSize = null; } else { $this->mSize = intval( $row->rev_len ); @@ -467,7 +527,7 @@ class Revision implements IDBAccessObject { $this->mSha1 = $row->rev_sha1; } - if( isset( $row->page_latest ) ) { + if ( isset( $row->page_latest ) ) { $this->mCurrent = ( $row->rev_id == $row->page_latest ); $this->mTitle = Title::newFromRow( $row ); } else { @@ -475,9 +535,21 @@ class Revision implements IDBAccessObject { $this->mTitle = null; } + if ( !isset( $row->rev_content_model ) || is_null( $row->rev_content_model ) ) { + $this->mContentModel = null; # determine on demand if needed + } else { + $this->mContentModel = strval( $row->rev_content_model ); + } + + if ( !isset( $row->rev_content_format ) || is_null( $row->rev_content_format ) ) { + $this->mContentFormat = null; # determine on demand if needed + } else { + $this->mContentFormat = strval( $row->rev_content_format ); + } + // Lazy extraction... - $this->mText = null; - if( isset( $row->old_text ) ) { + $this->mText = null; + if ( isset( $row->old_text ) ) { $this->mTextRow = $row; } else { // 'text' table row entry will be lazy-loaded @@ -492,10 +564,24 @@ class Revision implements IDBAccessObject { $this->mUserText = $row->user_name; // logged-in user } $this->mOrigUserText = $row->rev_user_text; - } elseif( is_array( $row ) ) { + } elseif ( is_array( $row ) ) { // Build a new revision to be saved... global $wgUser; // ugh + # if we have a content object, use it to set the model and type + if ( !empty( $row['content'] ) ) { + // @todo when is that set? test with external store setup! check out insertOn() [dk] + if ( !empty( $row['text_id'] ) ) { + throw new MWException( "Text already stored in external store (id {$row['text_id']}), " . + "can't serialize content object" ); + } + + $row['content_model'] = $row['content']->getModel(); + # note: mContentFormat is initializes later accordingly + # note: content is serialized later in this method! + # also set text to null? + } + $this->mId = isset( $row['id'] ) ? intval( $row['id'] ) : null; $this->mPage = isset( $row['page'] ) ? intval( $row['page'] ) : null; $this->mTextId = isset( $row['text_id'] ) ? intval( $row['text_id'] ) : null; @@ -508,21 +594,67 @@ class Revision implements IDBAccessObject { $this->mParentId = isset( $row['parent_id'] ) ? intval( $row['parent_id'] ) : null; $this->mSha1 = isset( $row['sha1'] ) ? strval( $row['sha1'] ) : null; + $this->mContentModel = isset( $row['content_model'] ) ? strval( $row['content_model'] ) : null; + $this->mContentFormat = isset( $row['content_format'] ) ? strval( $row['content_format'] ) : null; + // Enforce spacing trimming on supplied text $this->mComment = isset( $row['comment'] ) ? trim( strval( $row['comment'] ) ) : null; $this->mText = isset( $row['text'] ) ? rtrim( strval( $row['text'] ) ) : null; $this->mTextRow = null; - $this->mTitle = null; # Load on demand if needed - $this->mCurrent = false; - # If we still have no length, see it we have the text to figure it out + $this->mTitle = isset( $row['title'] ) ? $row['title'] : null; + + // if we have a Content object, override mText and mContentModel + if ( !empty( $row['content'] ) ) { + if ( !( $row['content'] instanceof Content ) ) { + throw new MWException( '`content` field must contain a Content object.' ); + } + + $handler = $this->getContentHandler(); + $this->mContent = $row['content']; + + $this->mContentModel = $this->mContent->getModel(); + $this->mContentHandler = null; + + $this->mText = $handler->serializeContent( $row['content'], $this->getContentFormat() ); + } elseif ( !is_null( $this->mText ) ) { + $handler = $this->getContentHandler(); + $this->mContent = $handler->unserializeContent( $this->mText ); + } + + // If we have a Title object, make sure it is consistent with mPage. + if ( $this->mTitle && $this->mTitle->exists() ) { + if ( $this->mPage === null ) { + // if the page ID wasn't known, set it now + $this->mPage = $this->mTitle->getArticleID(); + } elseif ( $this->mTitle->getArticleID() !== $this->mPage ) { + // Got different page IDs. This may be legit (e.g. during undeletion), + // but it seems worth mentioning it in the log. + wfDebug( "Page ID " . $this->mPage . " mismatches the ID " . + $this->mTitle->getArticleID() . " provided by the Title object." ); + } + } + + $this->mCurrent = false; + + // If we still have no length, see it we have the text to figure it out if ( !$this->mSize ) { - $this->mSize = is_null( $this->mText ) ? null : strlen( $this->mText ); + if ( !is_null( $this->mContent ) ) { + $this->mSize = $this->mContent->getSize(); + } else { + #NOTE: this should never happen if we have either text or content object! + $this->mSize = null; + } } - # Same for sha1 + + // Same for sha1 if ( $this->mSha1 === null ) { $this->mSha1 = is_null( $this->mText ) ? null : self::base36Sha1( $this->mText ); } + + // force lazy init + $this->getContentModel(); + $this->getContentFormat(); } else { throw new MWException( 'Revision constructor passed invalid row format.' ); } @@ -592,21 +724,26 @@ class Revision implements IDBAccessObject { * @return Title|null */ public function getTitle() { - if( isset( $this->mTitle ) ) { + if ( isset( $this->mTitle ) ) { return $this->mTitle; } - if( !is_null( $this->mId ) ) { //rev_id is defined as NOT NULL + if ( !is_null( $this->mId ) ) { //rev_id is defined as NOT NULL, but this revision may not yet have been inserted. $dbr = wfGetDB( DB_SLAVE ); $row = $dbr->selectRow( array( 'page', 'revision' ), self::selectPageFields(), array( 'page_id=rev_page', - 'rev_id' => $this->mId ), + 'rev_id' => $this->mId ), __METHOD__ ); if ( $row ) { $this->mTitle = Title::newFromRow( $row ); } } + + if ( !$this->mTitle && !is_null( $this->mPage ) && $this->mPage > 0 ) { + $this->mTitle = Title::newFromID( $this->mPage ); + } + return $this->mTitle; } @@ -642,9 +779,9 @@ class Revision implements IDBAccessObject { * @return Integer */ public function getUser( $audience = self::FOR_PUBLIC, User $user = null ) { - if( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_USER ) ) { + if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_USER ) ) { return 0; - } elseif( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_USER, $user ) ) { + } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_USER, $user ) ) { return 0; } else { return $this->mUser; @@ -674,9 +811,9 @@ class Revision implements IDBAccessObject { * @return string */ public function getUserText( $audience = self::FOR_PUBLIC, User $user = null ) { - if( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_USER ) ) { + if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_USER ) ) { return ''; - } elseif( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_USER, $user ) ) { + } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_USER, $user ) ) { return ''; } else { return $this->getRawUserText(); @@ -714,9 +851,9 @@ class Revision implements IDBAccessObject { * @return String */ function getComment( $audience = self::FOR_PUBLIC, User $user = null ) { - if( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_COMMENT ) ) { + if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_COMMENT ) ) { return ''; - } elseif( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_COMMENT, $user ) ) { + } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_COMMENT, $user ) ) { return ''; } else { return $this->mComment; @@ -740,28 +877,41 @@ class Revision implements IDBAccessObject { } /** - * @return Integer rcid of the unpatrolled row, zero if there isn't one + * @return integer rcid of the unpatrolled row, zero if there isn't one */ public function isUnpatrolled() { - if( $this->mUnpatrolled !== null ) { + if ( $this->mUnpatrolled !== null ) { return $this->mUnpatrolled; } + $rc = $this->getRecentChange(); + if ( $rc && $rc->getAttribute( 'rc_patrolled' ) == 0 ) { + $this->mUnpatrolled = $rc->getAttribute( 'rc_id' ); + } else { + $this->mUnpatrolled = 0; + } + return $this->mUnpatrolled; + } + + /** + * Get the RC object belonging to the current revision, if there's one + * + * @since 1.22 + * @return RecentChange|null + */ + public function getRecentChange() { $dbr = wfGetDB( DB_SLAVE ); - $this->mUnpatrolled = $dbr->selectField( 'recentchanges', - 'rc_id', - array( // Add redundant user,timestamp condition so we can use the existing index - 'rc_user_text' => $this->getRawUserText(), - 'rc_timestamp' => $dbr->timestamp( $this->getTimestamp() ), - 'rc_this_oldid' => $this->getId(), - 'rc_patrolled' => 0 + return RecentChange::newFromConds( + array( + 'rc_user_text' => $this->getRawUserText(), + 'rc_timestamp' => $dbr->timestamp( $this->getTimestamp() ), + 'rc_this_oldid' => $this->getId() ), __METHOD__ ); - return (int)$this->mUnpatrolled; } /** - * @param $field int one of DELETED_* bitfield constants + * @param int $field one of DELETED_* bitfield constants * * @return Boolean */ @@ -789,15 +939,39 @@ class Revision implements IDBAccessObject { * Revision::RAW get the text regardless of permissions * @param $user User object to check for, only if FOR_THIS_USER is passed * to the $audience parameter + * + * @deprecated in 1.21, use getContent() instead + * @todo Replace usage in core * @return String */ public function getText( $audience = self::FOR_PUBLIC, User $user = null ) { - if( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_TEXT ) ) { - return ''; - } elseif( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_TEXT, $user ) ) { - return ''; + ContentHandler::deprecated( __METHOD__, '1.21' ); + + $content = $this->getContent( $audience, $user ); + return ContentHandler::getContentText( $content ); # returns the raw content text, if applicable + } + + /** + * Fetch revision content if it's available to the specified audience. + * If the specified audience does not have the ability to view this + * revision, null will be returned. + * + * @param $audience Integer: one of: + * Revision::FOR_PUBLIC to be displayed to all users + * Revision::FOR_THIS_USER to be displayed to $wgUser + * Revision::RAW get the text regardless of permissions + * @param $user User object to check for, only if FOR_THIS_USER is passed + * to the $audience parameter + * @since 1.21 + * @return Content|null + */ + public function getContent( $audience = self::FOR_PUBLIC, User $user = null ) { + if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_TEXT ) ) { + return null; + } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_TEXT, $user ) ) { + return null; } else { - return $this->getRawText(); + return $this->getContentInternal(); } } @@ -816,16 +990,121 @@ class Revision implements IDBAccessObject { * Fetch revision text without regard for view restrictions * * @return String + * + * @deprecated since 1.21. Instead, use Revision::getContent( Revision::RAW ) + * or Revision::getSerializedData() as appropriate. */ public function getRawText() { - if( is_null( $this->mText ) ) { - // Revision text is immutable. Load on demand: + ContentHandler::deprecated( __METHOD__, "1.21" ); + return $this->getText( self::RAW ); + } + + /** + * Fetch original serialized data without regard for view restrictions + * + * @since 1.21 + * @return String + */ + public function getSerializedData() { + if ( is_null( $this->mText ) ) { $this->mText = $this->loadText(); } + return $this->mText; } /** + * Gets the content object for the revision (or null on failure). + * + * Note that for mutable Content objects, each call to this method will return a + * fresh clone. + * + * @since 1.21 + * @return Content|null the Revision's content, or null on failure. + */ + protected function getContentInternal() { + if ( is_null( $this->mContent ) ) { + // Revision is immutable. Load on demand: + if ( is_null( $this->mText ) ) { + $this->mText = $this->loadText(); + } + + if ( $this->mText !== null && $this->mText !== false ) { + // Unserialize content + $handler = $this->getContentHandler(); + $format = $this->getContentFormat(); + + $this->mContent = $handler->unserializeContent( $this->mText, $format ); + } else { + $this->mContent = false; // negative caching! + } + } + + // NOTE: copy() will return $this for immutable content objects + return $this->mContent ? $this->mContent->copy() : null; + } + + /** + * Returns the content model for this revision. + * + * If no content model was stored in the database, $this->getTitle()->getContentModel() is + * used to determine the content model to use. If no title is know, CONTENT_MODEL_WIKITEXT + * is used as a last resort. + * + * @return String the content model id associated with this revision, see the CONTENT_MODEL_XXX constants. + **/ + public function getContentModel() { + if ( !$this->mContentModel ) { + $title = $this->getTitle(); + $this->mContentModel = ( $title ? $title->getContentModel() : CONTENT_MODEL_WIKITEXT ); + + assert( !empty( $this->mContentModel ) ); + } + + return $this->mContentModel; + } + + /** + * Returns the content format for this revision. + * + * If no content format was stored in the database, the default format for this + * revision's content model is returned. + * + * @return String the content format id associated with this revision, see the CONTENT_FORMAT_XXX constants. + **/ + public function getContentFormat() { + if ( !$this->mContentFormat ) { + $handler = $this->getContentHandler(); + $this->mContentFormat = $handler->getDefaultFormat(); + + assert( !empty( $this->mContentFormat ) ); + } + + return $this->mContentFormat; + } + + /** + * Returns the content handler appropriate for this revision's content model. + * + * @throws MWException + * @return ContentHandler + */ + public function getContentHandler() { + if ( !$this->mContentHandler ) { + $model = $this->getContentModel(); + $this->mContentHandler = ContentHandler::getForModelID( $model ); + + $format = $this->getContentFormat(); + + if ( !$this->mContentHandler->isSupportedFormat( $format ) ) { + throw new MWException( "Oops, the content format $format is not supported for this content model, $model" ); + } + } + + return $this->mContentHandler; + } + + /** * @return String */ public function getTimestamp() { @@ -842,12 +1121,12 @@ class Revision implements IDBAccessObject { /** * Get previous revision for this title * - * @return Revision or null + * @return Revision|null */ public function getPrevious() { - if( $this->getTitle() ) { + if ( $this->getTitle() ) { $prev = $this->getTitle()->getPreviousRevisionID( $this->getId() ); - if( $prev ) { + if ( $prev ) { return self::newFromTitle( $this->getTitle(), $prev ); } } @@ -860,7 +1139,7 @@ class Revision implements IDBAccessObject { * @return Revision or null */ public function getNext() { - if( $this->getTitle() ) { + if ( $this->getTitle() ) { $next = $this->getTitle()->getNextRevisionID( $this->getId() ); if ( $next ) { return self::newFromTitle( $this->getTitle(), $next ); @@ -877,11 +1156,11 @@ class Revision implements IDBAccessObject { * @return Integer */ private function getPreviousRevisionId( $db ) { - if( is_null( $this->mPage ) ) { + if ( is_null( $this->mPage ) ) { return 0; } # Use page_latest if ID is not given - if( !$this->mId ) { + if ( !$this->mId ) { $prevId = $db->selectField( 'page', 'page_latest', array( 'page_id' => $this->mPage ), __METHOD__ ); @@ -900,23 +1179,27 @@ class Revision implements IDBAccessObject { * field must be included * * @param $row Object: the text data - * @param $prefix String: table prefix (default 'old_') + * @param string $prefix table prefix (default 'old_') + * @param string|false $wiki the name of the wiki to load the revision text from + * (same as the the wiki $row was loaded from) or false to indicate the local + * wiki (this is the default). Otherwise, it must be a symbolic wiki database + * identifier as understood by the LoadBalancer class. * @return String: text the text requested or false on failure */ - public static function getRevisionText( $row, $prefix = 'old_' ) { + public static function getRevisionText( $row, $prefix = 'old_', $wiki = false ) { wfProfileIn( __METHOD__ ); # Get data $textField = $prefix . 'text'; $flagsField = $prefix . 'flags'; - if( isset( $row->$flagsField ) ) { + if ( isset( $row->$flagsField ) ) { $flags = explode( ',', $row->$flagsField ); } else { $flags = array(); } - if( isset( $row->$textField ) ) { + if ( isset( $row->$textField ) ) { $text = $row->$textField; } else { wfProfileOut( __METHOD__ ); @@ -927,44 +1210,16 @@ class Revision implements IDBAccessObject { if ( in_array( 'external', $flags ) ) { $url = $text; $parts = explode( '://', $url, 2 ); - if( count( $parts ) == 1 || $parts[1] == '' ) { + if ( count( $parts ) == 1 || $parts[1] == '' ) { wfProfileOut( __METHOD__ ); return false; } - $text = ExternalStore::fetchFromURL( $url ); + $text = ExternalStore::fetchFromURL( $url, array( 'wiki' => $wiki ) ); } // If the text was fetched without an error, convert it if ( $text !== false ) { - if( in_array( 'gzip', $flags ) ) { - # Deal with optional compression of archived pages. - # This can be done periodically via maintenance/compressOld.php, and - # as pages are saved if $wgCompressRevisions is set. - $text = gzinflate( $text ); - } - - if( in_array( 'object', $flags ) ) { - # Generic compressed storage - $obj = unserialize( $text ); - if ( !is_object( $obj ) ) { - // Invalid object - wfProfileOut( __METHOD__ ); - return false; - } - $text = $obj->getText(); - } - - global $wgLegacyEncoding; - if( $text !== false && $wgLegacyEncoding - && !in_array( 'utf-8', $flags ) && !in_array( 'utf8', $flags ) ) - { - # Old revisions kept around in a legacy encoding? - # Upconvert on demand. - # ("utf8" checked for compatibility with some broken - # conversion scripts 2008-12-30) - global $wgContLang; - $text = $wgContLang->iconv( $wgLegacyEncoding, 'UTF-8', $text ); - } + $text = self::decompressRevisionText( $text, $flags ); } wfProfileOut( __METHOD__ ); return $text; @@ -988,8 +1243,8 @@ class Revision implements IDBAccessObject { # on load if $wgLegacyCharset is set in the future. $flags[] = 'utf-8'; - if( $wgCompressRevisions ) { - if( function_exists( 'gzdeflate' ) ) { + if ( $wgCompressRevisions ) { + if ( function_exists( 'gzdeflate' ) ) { $text = gzdeflate( $text ); $flags[] = 'gzip'; } else { @@ -1000,73 +1255,140 @@ class Revision implements IDBAccessObject { } /** + * Re-converts revision text according to it's flags. + * + * @param $text Mixed: reference to a text + * @param $flags array: compression flags + * @return String|bool decompressed text, or false on failure + */ + public static function decompressRevisionText( $text, $flags ) { + if ( in_array( 'gzip', $flags ) ) { + # Deal with optional compression of archived pages. + # This can be done periodically via maintenance/compressOld.php, and + # as pages are saved if $wgCompressRevisions is set. + $text = gzinflate( $text ); + } + + if ( in_array( 'object', $flags ) ) { + # Generic compressed storage + $obj = unserialize( $text ); + if ( !is_object( $obj ) ) { + // Invalid object + return false; + } + $text = $obj->getText(); + } + + global $wgLegacyEncoding; + if ( $text !== false && $wgLegacyEncoding + && !in_array( 'utf-8', $flags ) && !in_array( 'utf8', $flags ) ) + { + # Old revisions kept around in a legacy encoding? + # Upconvert on demand. + # ("utf8" checked for compatibility with some broken + # conversion scripts 2008-12-30) + global $wgContLang; + $text = $wgContLang->iconv( $wgLegacyEncoding, 'UTF-8', $text ); + } + + return $text; + } + + /** * Insert a new revision into the database, returning the new revision ID * number on success and dies horribly on failure. * * @param $dbw DatabaseBase: (master connection) + * @throws MWException * @return Integer */ public function insertOn( $dbw ) { - global $wgDefaultExternalStore; + global $wgDefaultExternalStore, $wgContentHandlerUseDB; wfProfileIn( __METHOD__ ); + $this->checkContentModel(); + $data = $this->mText; $flags = self::compressRevisionText( $data ); # Write to external storage if required - if( $wgDefaultExternalStore ) { + if ( $wgDefaultExternalStore ) { // Store and get the URL $data = ExternalStore::insertToDefault( $data ); - if( !$data ) { + if ( !$data ) { + wfProfileOut( __METHOD__ ); throw new MWException( "Unable to store text to external storage" ); } - if( $flags ) { + if ( $flags ) { $flags .= ','; } $flags .= 'external'; } # Record the text (or external storage URL) to the text table - if( !isset( $this->mTextId ) ) { + if ( !isset( $this->mTextId ) ) { $old_id = $dbw->nextSequenceValue( 'text_old_id_seq' ); $dbw->insert( 'text', array( - 'old_id' => $old_id, - 'old_text' => $data, + 'old_id' => $old_id, + 'old_text' => $data, 'old_flags' => $flags, ), __METHOD__ ); $this->mTextId = $dbw->insertId(); } - if ( $this->mComment === null ) $this->mComment = ""; + if ( $this->mComment === null ) { + $this->mComment = ""; + } # Record the edit in revisions $rev_id = isset( $this->mId ) ? $this->mId : $dbw->nextSequenceValue( 'revision_rev_id_seq' ); - $dbw->insert( 'revision', - array( - 'rev_id' => $rev_id, - 'rev_page' => $this->mPage, - 'rev_text_id' => $this->mTextId, - 'rev_comment' => $this->mComment, - 'rev_minor_edit' => $this->mMinorEdit ? 1 : 0, - 'rev_user' => $this->mUser, - 'rev_user_text' => $this->mUserText, - 'rev_timestamp' => $dbw->timestamp( $this->mTimestamp ), - 'rev_deleted' => $this->mDeleted, - 'rev_len' => $this->mSize, - 'rev_parent_id' => is_null( $this->mParentId ) - ? $this->getPreviousRevisionId( $dbw ) - : $this->mParentId, - 'rev_sha1' => is_null( $this->mSha1 ) - ? self::base36Sha1( $this->mText ) - : $this->mSha1 - ), __METHOD__ + $row = array( + 'rev_id' => $rev_id, + 'rev_page' => $this->mPage, + 'rev_text_id' => $this->mTextId, + 'rev_comment' => $this->mComment, + 'rev_minor_edit' => $this->mMinorEdit ? 1 : 0, + 'rev_user' => $this->mUser, + 'rev_user_text' => $this->mUserText, + 'rev_timestamp' => $dbw->timestamp( $this->mTimestamp ), + 'rev_deleted' => $this->mDeleted, + 'rev_len' => $this->mSize, + 'rev_parent_id' => is_null( $this->mParentId ) + ? $this->getPreviousRevisionId( $dbw ) + : $this->mParentId, + 'rev_sha1' => is_null( $this->mSha1 ) + ? Revision::base36Sha1( $this->mText ) + : $this->mSha1, ); + if ( $wgContentHandlerUseDB ) { + //NOTE: Store null for the default model and format, to save space. + //XXX: Makes the DB sensitive to changed defaults. Make this behavior optional? Only in miser mode? + + $model = $this->getContentModel(); + $format = $this->getContentFormat(); + + $title = $this->getTitle(); + + if ( $title === null ) { + wfProfileOut( __METHOD__ ); + throw new MWException( "Insufficient information to determine the title of the revision's page!" ); + } + + $defaultModel = ContentHandler::getDefaultModelFor( $title ); + $defaultFormat = ContentHandler::getForModelID( $defaultModel )->getDefaultFormat(); + + $row['rev_content_model'] = ( $model === $defaultModel ) ? null : $model; + $row['rev_content_format'] = ( $format === $defaultFormat ) ? null : $format; + } + + $dbw->insert( 'revision', $row, __METHOD__ ); + $this->mId = !is_null( $rev_id ) ? $rev_id : $dbw->insertId(); wfRunHooks( 'RevisionInsertComplete', array( &$this, $data, $flags ) ); @@ -1075,6 +1397,52 @@ class Revision implements IDBAccessObject { return $this->mId; } + protected function checkContentModel() { + global $wgContentHandlerUseDB; + + $title = $this->getTitle(); //note: may return null for revisions that have not yet been inserted. + + $model = $this->getContentModel(); + $format = $this->getContentFormat(); + $handler = $this->getContentHandler(); + + if ( !$handler->isSupportedFormat( $format ) ) { + $t = $title->getPrefixedDBkey(); + + throw new MWException( "Can't use format $format with content model $model on $t" ); + } + + if ( !$wgContentHandlerUseDB && $title ) { + // if $wgContentHandlerUseDB is not set, all revisions must use the default content model and format. + + $defaultModel = ContentHandler::getDefaultModelFor( $title ); + $defaultHandler = ContentHandler::getForModelID( $defaultModel ); + $defaultFormat = $defaultHandler->getDefaultFormat(); + + if ( $this->getContentModel() != $defaultModel ) { + $t = $title->getPrefixedDBkey(); + + throw new MWException( "Can't save non-default content model with \$wgContentHandlerUseDB disabled: " + . "model is $model , default for $t is $defaultModel" ); + } + + if ( $this->getContentFormat() != $defaultFormat ) { + $t = $title->getPrefixedDBkey(); + + throw new MWException( "Can't use non-default content format with \$wgContentHandlerUseDB disabled: " + . "format is $format, default for $t is $defaultFormat" ); + } + } + + $content = $this->getContent( Revision::RAW ); + + if ( !$content || !$content->isValid() ) { + $t = $title->getPrefixedDBkey(); + + throw new MWException( "Content of $t is not valid! Content model is $model" ); + } + } + /** * Get the base 36 SHA-1 value for a string of text * @param $text String @@ -1088,7 +1456,7 @@ class Revision implements IDBAccessObject { * Lazy-load the revision's text. * Currently hardcoded to the 'text' table storage engine. * - * @return String + * @return String|bool the revision's text, or false on failure */ protected function loadText() { wfProfileIn( __METHOD__ ); @@ -1097,9 +1465,9 @@ class Revision implements IDBAccessObject { global $wgRevisionCacheExpiry, $wgMemc; $textId = $this->getTextId(); $key = wfMemcKey( 'revisiontext', 'textid', $textId ); - if( $wgRevisionCacheExpiry ) { + if ( $wgRevisionCacheExpiry ) { $text = $wgMemc->get( $key ); - if( is_string( $text ) ) { + if ( is_string( $text ) ) { wfDebug( __METHOD__ . ": got id $textId from cache\n" ); wfProfileOut( __METHOD__ ); return $text; @@ -1114,28 +1482,38 @@ class Revision implements IDBAccessObject { $row = null; } - if( !$row ) { + if ( !$row ) { // Text data is immutable; check slaves first. $dbr = wfGetDB( DB_SLAVE ); $row = $dbr->selectRow( 'text', array( 'old_text', 'old_flags' ), - array( 'old_id' => $this->getTextId() ), + array( 'old_id' => $textId ), __METHOD__ ); } - if( !$row && wfGetLB()->getServerCount() > 1 ) { - // Possible slave lag! + // Fallback to the master in case of slave lag. Also use FOR UPDATE if it was + // used to fetch this revision to avoid missing the row due to REPEATABLE-READ. + $forUpdate = ( $this->mQueryFlags & self::READ_LOCKING == self::READ_LOCKING ); + if ( !$row && ( $forUpdate || wfGetLB()->getServerCount() > 1 ) ) { $dbw = wfGetDB( DB_MASTER ); $row = $dbw->selectRow( 'text', array( 'old_text', 'old_flags' ), - array( 'old_id' => $this->getTextId() ), - __METHOD__ ); + array( 'old_id' => $textId ), + __METHOD__, + $forUpdate ? array( 'FOR UPDATE' ) : array() ); + } + + if ( !$row ) { + wfDebugLog( 'Revision', "No text row with ID '$textId' (revision {$this->getId()})." ); } $text = self::getRevisionText( $row ); + if ( $row && $text === false ) { + wfDebugLog( 'Revision', "No blob for text row '$textId' (revision {$this->getId()})." ); + } # No negative caching -- negative hits on text rows may be due to corrupted slave servers - if( $wgRevisionCacheExpiry && $text !== false ) { + if ( $wgRevisionCacheExpiry && $text !== false ) { $wgMemc->set( $key, $text, $wgRevisionCacheExpiry ); } @@ -1154,25 +1532,34 @@ class Revision implements IDBAccessObject { * * @param $dbw DatabaseBase * @param $pageId Integer: ID number of the page to read from - * @param $summary String: revision's summary + * @param string $summary revision's summary * @param $minor Boolean: whether the revision should be considered as minor * @return Revision|null on error */ public static function newNullRevision( $dbw, $pageId, $summary, $minor ) { + global $wgContentHandlerUseDB; + wfProfileIn( __METHOD__ ); + $fields = array( 'page_latest', 'page_namespace', 'page_title', + 'rev_text_id', 'rev_len', 'rev_sha1' ); + + if ( $wgContentHandlerUseDB ) { + $fields[] = 'rev_content_model'; + $fields[] = 'rev_content_format'; + } + $current = $dbw->selectRow( array( 'page', 'revision' ), - array( 'page_latest', 'page_namespace', 'page_title', - 'rev_text_id', 'rev_len', 'rev_sha1' ), + $fields, array( 'page_id' => $pageId, 'page_latest=rev_id', ), __METHOD__ ); - if( $current ) { - $revision = new Revision( array( + if ( $current ) { + $row = array( 'page' => $pageId, 'comment' => $summary, 'minor_edit' => $minor, @@ -1180,7 +1567,14 @@ class Revision implements IDBAccessObject { 'parent_id' => $current->page_latest, 'len' => $current->rev_len, 'sha1' => $current->rev_sha1 - ) ); + ); + + if ( $wgContentHandlerUseDB ) { + $row['content_model'] = $current->rev_content_model; + $row['content_format'] = $current->rev_content_format; + } + + $revision = new Revision( $row ); $revision->setTitle( Title::makeTitle( $current->page_namespace, $current->page_title ) ); } else { $revision = null; @@ -1217,7 +1611,7 @@ class Revision implements IDBAccessObject { * @return Boolean */ public static function userCanBitfield( $bitfield, $field, User $user = null ) { - if( $bitfield & $field ) { // aspect is deleted + if ( $bitfield & $field ) { // aspect is deleted if ( $bitfield & self::DELETED_RESTRICTED ) { $permission = 'suppressrevision'; } elseif ( $field & self::DELETED_TEXT ) { @@ -1245,7 +1639,7 @@ class Revision implements IDBAccessObject { */ static function getTimestampFromId( $title, $id ) { $dbr = wfGetDB( DB_SLAVE ); - // Casting fix for DB2 + // Casting fix for databases that can't take '' for rev_id if ( $id == '' ) { $id = 0; } @@ -1270,7 +1664,7 @@ class Revision implements IDBAccessObject { static function countByPageId( $db, $id ) { $row = $db->selectRow( 'revision', array( 'revCount' => 'COUNT(*)' ), array( 'rev_page' => $id ), __METHOD__ ); - if( $row ) { + if ( $row ) { return $row->revCount; } return 0; @@ -1285,7 +1679,7 @@ class Revision implements IDBAccessObject { */ static function countByTitle( $db, $title ) { $id = $title->getArticleID(); - if( $id ) { + if ( $id ) { return self::countByPageId( $db, $id ); } return 0; @@ -1307,7 +1701,9 @@ class Revision implements IDBAccessObject { * @return bool True if the given user was the only one to edit since the given timestamp */ public static function userWasLastToEdit( $db, $pageId, $userId, $since ) { - if ( !$userId ) return false; + if ( !$userId ) { + return false; + } if ( is_int( $db ) ) { $db = wfGetDB( $db ); @@ -1328,4 +1724,4 @@ class Revision implements IDBAccessObject { } return true; } -}
\ No newline at end of file +} |