rm protectionlevel from mVariableIDs per iAlex on codereview
[lhc/web/wiklou.git] / maintenance / rebuildrecentchanges.inc
index 4f64ce3..f846d2b 100644 (file)
-<?
-
-# Rebuild recent changes table.
-#
+<?php
+/**
+ * Rebuild recent changes table.
+ *
+ * @file
+ * @todo document
+ * @ingroup Maintenance
+ */
+
+/** Public entry; more passes might come in! :) */
+function rebuildRecentChangesTable() {
+       rebuildRecentChangesTablePass1();
+       rebuildRecentChangesTablePass2();
+       rebuildRecentChangesTablePass3();
+       rebuildRecentChangesTablePass4();
+}
 
+/** */
 function rebuildRecentChangesTablePass1()
 {
-       $sql = "DROP TABLE IF EXISTS recentchanges";
-       wfQuery( $sql, DB_WRITE );
-
-       $sql = "CREATE TABLE recentchanges (
-  rc_timestamp varchar(14) binary NOT NULL default '',
-  rc_cur_time varchar(14) binary NOT NULL default '',
-  rc_user int(10) unsigned NOT NULL default '0',
-  rc_user_text varchar(255) binary NOT NULL default '',
-  rc_namespace tinyint(3) unsigned NOT NULL default '0',
-  rc_title varchar(255) binary NOT NULL default '',
-  rc_comment varchar(255) binary NOT NULL default '',
-  rc_minor tinyint(3) unsigned NOT NULL default '0',
-  rc_bot tinyint(3) unsigned NOT NULL default '0',
-  rc_new tinyint(3) unsigned NOT NULL default '0',
-  rc_cur_id int(10) unsigned NOT NULL default '0',
-  rc_this_oldid int(10) unsigned NOT NULL default '0',
-  rc_last_oldid int(10) unsigned NOT NULL default '0',
-  INDEX rc_cur_id (rc_cur_id),
-  INDEX rc_cur_time (rc_cur_time),
-  INDEX rc_timestamp (rc_timestamp),
-  INDEX rc_namespace (rc_namespace),
-  INDEX rc_title (rc_title)
-) TYPE=MyISAM PACK_KEYS=1;";
-       wfQuery( $sql, DB_WRITE );
-
-       print( "Loading from CUR table...\n" );
-
-       $sql = "INSERT INTO recentchanges (rc_timestamp,rc_cur_time,rc_user," .
-         "rc_user_text,rc_namespace,rc_title,rc_comment,rc_minor,rc_bot,rc_new," .
-         "rc_cur_id,rc_this_oldid,rc_last_oldid) SELECT cur_timestamp," .
-         "cur_timestamp,cur_user,cur_user_text,cur_namespace,cur_title," .
-         "cur_comment,cur_minor_edit,0,cur_is_new,cur_id,0,0 FROM cur " .
-         "ORDER BY cur_timestamp DESC LIMIT 5000";
-       wfQuery( $sql, DB_WRITE );
-
-       print( "Loading from OLD table...\n" );
-
-       $sql = "INSERT INTO recentchanges (rc_timestamp,rc_cur_time,rc_user," .
-      "rc_user_text,rc_namespace,rc_title,rc_comment,rc_minor,rc_bot,rc_new," .
-      "rc_cur_id,rc_this_oldid,rc_last_oldid) SELECT old_timestamp,''," .
-         "old_user,old_user_text,old_namespace,old_title,old_comment," .
-         "old_minor_edit,0,0,0,old_id,0 FROM old ORDER BY old_timestamp " .
-         "DESC LIMIT 5000";
-       wfQuery( $sql, DB_WRITE );
-
-       $sql = "SELECT rc_timestamp FROM recentchanges " .
-         "ORDER BY rc_timestamp DESC LIMIT 5000,1";
-       $res = wfQuery( $sql, DB_WRITE );
-       $obj = wfFetchObject( $res );
-       $ts = $obj->rc_timestamp;
-
-       $sql = "DELETE FROM recentchanges WHERE rc_timestamp < '{$ts}'";
-       wfQuery( $sql, DB_WRITE );
-}
+       $dbw = wfGetDB( DB_MASTER );
 
-function rebuildRecentChangesTablePass2()
-{
-       $ns = $id = $count = 0;
-       $title = $ct =  "";
+       $dbw->delete( 'recentchanges', '*' );
 
-       print( "Updating links...\n" );
+       print( "Loading from page and revision tables...\n" );
 
-       $sql = "SELECT rc_namespace,rc_title,rc_timestamp FROM recentchanges " .
-         "ORDER BY rc_namespace,rc_title,rc_timestamp DESC";
-       $res = wfQuery( $sql, DB_WRITE );
+       global $wgRCMaxAge;
 
-       while ( $obj = wfFetchObject( $res ) ) {
-               if ( ! ( $ns == $obj->rc_namespace &&
-                          0 == strcmp( $title, wfStrencode( $obj->rc_title ) ) ) ) {
+       print( '$wgRCMaxAge=' . $wgRCMaxAge );
+       $days = $wgRCMaxAge / 24 / 3600;
+       if ( intval($days) == $days ) {
+                       print( " (" . $days . " days)\n" );
+       } else {
+                       print( " (approx. " .  intval($days) . " days)\n" );
+       }
 
-                       $ns = $obj->rc_namespace;
-                       $title = wfStrencode( $obj->rc_title );
+       $cutoff = time() - $wgRCMaxAge;
+       $dbw->insertSelect( 'recentchanges', array( 'page', 'revision' ),
+               array(
+                       'rc_timestamp'  => 'rev_timestamp',
+                       'rc_cur_time'   => 'rev_timestamp',
+                       'rc_user'       => 'rev_user',
+                       'rc_user_text'  => 'rev_user_text',
+                       'rc_namespace'  => 'page_namespace',
+                       'rc_title'      => 'page_title',
+                       'rc_comment'    => 'rev_comment',
+                       'rc_minor'      => 'rev_minor_edit',
+                       'rc_bot'        => 0,
+                       'rc_new'        => 'page_is_new',
+                       'rc_cur_id'     => 'page_id',
+                       'rc_this_oldid' => 'rev_id',
+                       'rc_last_oldid' => 0, // is this ok?
+                       'rc_type'       => $dbw->conditional( 'page_is_new != 0', RC_NEW, RC_EDIT ),
+                       'rc_deleted'    => 'rev_deleted'
+               ), array(
+                       'rev_timestamp > ' . $dbw->addQuotes( $dbw->timestamp( $cutoff ) ),
+                       'rev_page=page_id'
+               ), __METHOD__,
+               array(), // INSERT options
+               array( 'ORDER BY' => 'rev_timestamp DESC', 'LIMIT' => 5000 ) // SELECT options
+       );
+}
 
-                       $sql = "SELECT cur_id,cur_timestamp FROM cur WHERE " .
-                         "cur_namespace={$ns} AND cur_title='{$title}'";
-                       $res2 = wfQuery( $sql, DB_WRITE );
-                       $obj2 = wfFetchObject( $res2 );
+function rebuildRecentChangesTablePass2()
+{
+       $dbw = wfGetDB( DB_MASTER );
+       list ($recentchanges, $revision) = $dbw->tableNamesN( 'recentchanges', 'revision' );
+
+       print( "Updating links and size differences...\n" );
+
+       # Fill in the rc_last_oldid field, which points to the previous edit
+       $sql = "SELECT rc_cur_id,rc_this_oldid,rc_timestamp FROM $recentchanges " .
+         "ORDER BY rc_cur_id,rc_timestamp";
+       $res = $dbw->query( $sql, DB_MASTER );
+
+       $lastCurId = 0;
+       $lastOldId = 0;
+       while ( $obj = $dbw->fetchObject( $res ) ) {
+               $new = 0;
+               if( $obj->rc_cur_id != $lastCurId ) {
+                       # Switch! Look up the previous last edit, if any
+                       $lastCurId = intval( $obj->rc_cur_id );
+                       $emit = $obj->rc_timestamp;
+                       $sql2 = "SELECT rev_id,rev_len FROM $revision " .
+                               "WHERE rev_page={$lastCurId} ".
+                               "AND rev_timestamp<'{$emit}' ORDER BY rev_timestamp DESC LIMIT 1";
+                       $res2 = $dbw->query( $sql2 );
+                       if( $row = $dbw->fetchObject( $res2 ) ) {
+                               $lastOldId = intval($row->rev_id);
+                               # Grab the last text size if available
+                               $lastSize = !is_null($row->rev_len) ? intval($row->rev_len) : 'NULL';
+                       } else {
+                               # No previous edit
+                               $lastOldId = 0;
+                               $lastSize = 'NULL';
+                               $new = 1; // probably true
+                       }
+                       $dbw->freeResult( $res2 );
+               }
+               if( $lastCurId == 0 ) {
+                       print "Uhhh, something wrong? No curid\n";
+               } else {
+                       # Grab the entry's text size
+                       $size = $dbw->selectField( 'revision', 'rev_len', array('rev_id' => $obj->rc_this_oldid ) );
+                       $size = !is_null($size) ? intval($size) : 'NULL';
+                       
+                       $sql3 = "UPDATE $recentchanges SET rc_last_oldid=$lastOldId,rc_new=$new,rc_type=$new," .
+                               "rc_old_len=$lastSize,rc_new_len=$size " .
+                               "WHERE rc_cur_id={$lastCurId} AND rc_this_oldid={$obj->rc_this_oldid}";
+                       $dbw->query( $sql3 );
+                       
+                       $lastOldId = intval( $obj->rc_this_oldid );
+               }
+       }
+       $dbw->freeResult( $res );
+}
 
-                       $id = $obj2->cur_id;
-                       $ct = $obj2->cur_timestamp;
+function rebuildRecentChangesTablePass3()
+{
+       $dbw = wfGetDB( DB_MASTER );
+
+       print( "Loading from user, page, and logging tables...\n" );
+       
+       global $wgRCMaxAge;
+       // Some logs don't go in RC. This can't really detect all of those.
+       // At least do the basics logs for a standard install...
+       // FIXME: this needs to be maintained
+       $basicRCLogs = array( 
+               'block',
+               'protect',
+               'rights',
+               'delete',
+               'upload',
+               'move',
+               'import',
+               'merge' );
+       // Escape...blah blah
+       $selectLogs = array();
+       foreach( $basicRCLogs as $logtype ) {
+               $safetype = $dbw->strencode( $logtype );
+               $selectLogs[] = "'$safetype'";
+       }
+       
+       $cutoff = time() - $wgRCMaxAge;
+       $dbw->insertSelect( 'recentchanges', array( 'logging', 'page', 'user' ),
+               array(
+                       'rc_timestamp'  => 'log_timestamp',
+                       'rc_cur_time'   => 'log_timestamp',
+                       'rc_user'       => 'log_user',
+                       'rc_user_text'  => 'user_name',
+                       'rc_namespace'  => 'log_namespace',
+                       'rc_title'      => 'log_title',
+                       'rc_comment'    => 'log_comment',
+                       'rc_minor'      => 0,
+                       'rc_bot'        => 0,
+                       'rc_patrolled'  => 1,
+                       'rc_new'        => 0,
+                       'rc_this_oldid' => 0,
+                       'rc_last_oldid' => 0,
+                       'rc_type'       => RC_LOG,
+                       'rc_cur_id'     => 'page_id',
+                       'rc_log_type'   => 'log_type',
+                       'rc_log_action' => 'log_action',
+                       'rc_logid'      => 'log_id',
+                       'rc_params'     => 'log_params',
+                       'rc_deleted'    => 'log_deleted'
+               ), array(
+                       'log_timestamp > ' . $dbw->addQuotes( $dbw->timestamp( $cutoff ) ),
+                       'log_user=user_id',
+                       'log_namespace=page_namespace',
+                       'log_title=page_title',
+                       'log_type IN(' . implode(',',$selectLogs) . ')'
+               ), __METHOD__,
+               array(), // INSERT options
+               array( 'ORDER BY' => 'log_timestamp DESC', 'LIMIT' => 5000 ) // SELECT options
+       );
+}
 
-                       $sql = "SELECT old_id FROM old WHERE old_namespace={$ns} " .
-                         "AND old_title='{$title}' ORDER BY old_timestamp DESC";
-                       $res2 = wfQuery( $sql, DB_WRITE );
-                       $numRows = wfNumRows( $res2 );
+function rebuildRecentChangesTablePass4()
+{
+       global $wgGroupPermissions, $wgUseRCPatrol;
+                       
+       $dbw = wfGetDB( DB_MASTER );
+       
+       list($recentchanges,$usergroups,$user) = $dbw->tableNamesN( 'recentchanges', 'user_groups', 'user' );
+
+       $botgroups = $autopatrolgroups = array();
+       foreach( $wgGroupPermissions as $group => $rights ) {
+               if( isset( $rights['bot'] ) && $rights['bot'] == true ) {
+                       $botgroups[] = $dbw->addQuotes( $group );
                }
+               if( $wgUseRCPatrol && isset( $rights['autopatrol'] ) && $rights['autopatrol'] == true ) {
+                       $autopatrolgroups[] = $dbw->addQuotes( $group );
+               }
+       }
+       # Flag our recent bot edits
+       if( !empty($botgroups) ) {
+               $botwhere = implode(',',$botgroups);
+               $botusers = array();
 
-               if ( 0 != $numRows-- ) {
-                       $obj2 = wfFetchObject( $res2 );
+               print( "Flagging bot account edits...\n" );
 
-                       $sql = "UPDATE recentchanges SET rc_cur_id={$id},rc_cur_time=" .
-                         "'{$ct}',rc_last_oldid={$obj2->old_id} WHERE " .
-                         "rc_namespace={$ns} AND rc_title='{$title}' AND " .
-                         "rc_timestamp='{$obj->rc_timestamp}'";
-                       wfQuery( $sql, DB_WRITE );
-               } else {
-                       $sql = "UPDATE recentchanges SET rc_cur_id={$id},rc_cur_time=" .
-                         "'{$ct}' WHERE rc_namespace={$ns} AND rc_title='{$title}' " .
-                         "AND rc_timestamp='{$obj->rc_timestamp}'";
-                       wfQuery( $sql, DB_WRITE );
+               # Find all users that are bots
+               $sql = "SELECT DISTINCT user_name FROM $usergroups, $user " .
+                       "WHERE ug_group IN($botwhere) AND user_id = ug_user";
+               $res = $dbw->query( $sql, DB_MASTER );
+
+               while( $obj = $dbw->fetchObject( $res ) ) {
+                       $botusers[] = $dbw->addQuotes( $obj->user_name );
+               }
+               # Fill in the rc_bot field
+               if( !empty($botusers) ) {
+                       $botwhere = implode(',',$botusers);
+                       $sql2 = "UPDATE $recentchanges SET rc_bot=1 " .
+                               "WHERE rc_user_text IN($botwhere)";
+                       $dbw->query( $sql2 );
                }
+       }
+       global $wgMiserMode;
+       # Flag our recent autopatrolled edits
+       if( !$wgMiserMode && !empty($autopatrolgroups) ) {
+               $patrolwhere = implode(',',$autopatrolgroups);
+               $patrolusers = array();
+
+               print( "Flagging auto-patrolled edits...\n" );
 
-               if ( 0 == ( ++$count % 500 ) ) {
-                       printf( "%d records processed.\n", $count );
+               # Find all users in RC with autopatrol rights
+               $sql = "SELECT DISTINCT user_name FROM $usergroups, $user " .
+                       "WHERE ug_group IN($patrolwhere) AND user_id = ug_user";
+               $res = $dbw->query( $sql, DB_MASTER );
+
+               while( $obj = $dbw->fetchObject( $res ) ) {
+                       $patrolusers[] = $dbw->addQuotes( $obj->user_name );
+               }
+               
+               # Fill in the rc_patrolled field
+               if( !empty($patrolusers) ) {
+                       $patrolwhere = implode(',',$patrolusers);
+                       $sql2 = "UPDATE $recentchanges SET rc_patrolled=1 " .
+                               "WHERE rc_user_text IN($patrolwhere)";
+                       $dbw->query( $sql2 );
                }
        }
+       
+       $dbw->freeResult( $res );
 }
-
-?>