backfill_db.php

<?php
	include 'wikibot.classes.php';

	$wpapi = new wikipediaapi;
	$wpapi->login( 'ClueBot_NG', 'passwordhere' );

	function mysql_reconnect() {
		if( !@mysql_ping() ) {
			mysql_connect( 'bots-sql2', 'cb', 'passwordhere' );
			mysql_select_db( 'cluebotbig' );
		}
	}

	$continue = null;
	$id = False;
	$count = 0;

	while( ($id == False || $id > 936592) ) {
		$contribs = $wpapi->usercontribs( 'ClueBot_NG', 5000, $continue );
		foreach( $contribs as $edit ) {
			if( $id == False ) {
				$id = $edit['revid'];
			}

			if( $edit['revid'] > $id ) {
				$id = $edit['revid'];
			}

			if( !is_array( $edit ) || !isset( $edit['comment'] ) ) {
				continue;
			}

			if( preg_match( '/^Reverting possible vandalism by \[\[Special:Contributions\/.+\|(.+)\]\] to version by (.+)\. False positive\? \[\[User:ClueBot NG\/FalsePositives\|Report it\]\]\. Thanks, \[\[User:ClueBot NG\|ClueBot NG\]\]\. \((.+)\) \(Bot\)$/', $edit['comment'], $matches )  ) {
				$ruser = $matches[1];
				$oruser = $matches[2];
				$cbng_id = $matches[3];

				mysql_reconnect();
				$res = mysql_query(
					'SELECT * FROM `vandalism` WHERE `id` = "' . mysql_real_escape_string( $cbng_id ) . '" AND ' .
					'(`user` = "" OR `article` = "" OR `diff` = "" OR `new_id` = 0)'
				);

				if( mysql_num_rows( $res ) > 0 ) {
					$revs = $wpapi->revisions( $edit['title'], $count=3, $dir='older', $content=False, $revid=$edit['revid'] );
					$rev = $revs[1];
					$rev2 = $revs[2];

					if( !is_array( $rev ) || !is_array( $rev2 ) ) {
						continue;
					}

					$query = 'UPDATE `vandalism` SET';
					$query .= ' `article` = "' . mysql_real_escape_string( $edit['title'] ) . '",';
					$query .= ' `user` = "' . mysql_real_escape_string( $rev['user']  ) . '",';
					$query .= ' `diff` = "' . mysql_real_escape_string( 'http://en.wikipedia.org/w/index.php?diff=' . $rev['revid'] . '&oldid=' . $rev2['revid']  ) . '",';
					$query .= ' `new_id` = "' . mysql_real_escape_string( $rev['revid'] ) . '",';
					$query .= ' `old_id` = "' . mysql_real_escape_string( $rev2['revid'] ) . '"';
					$query .= ' WHERE `id` = "' . mysql_real_escape_string( $cbng_id ) . '" AND';
					$query .= ' (`user` = "" OR `article` = "" OR `diff` = "" OR `new_id` = 0 OR `old_id` = 0)';

					if( mysql_query( $query ) ) {
						print "Updated " . $cbng_id . "\n";
						$count++;
					}
				}
			}
		}
	}
	print $count . " updated\n";
?>