refresh-dblist

#!/usr/bin/env php
<?php

/*
Master file of all wikis:
	all.dblist

Exceptions list:
	special.dblist

Generated by this script:
	wikibooks.dblist
	wikinews.dblist
	wikipedia.dblist
	wikiquote.dblist
	wikisource.dblist
	wiktionary.dblist
	wikimedia.dblist
	wikivoyage.dblist

The split-out files are used by the backup script.

Split by cluster:
	s1.dblist
	s2.dblist
	s3.dblist
	s4.dblist
	s5.dblist
	s6.dblist
	s7.dblist
*/

require_once( __DIR__ . '/multiversion/MWRealm.php' );
require_once( __DIR__ . '/multiversion/defines.php' );

$base = MEDIAWIKI_STAGING_DIR;

$defaultCluster = 's3';
$clusterOverrides = loadClusterList(getRealmSpecificFilename("$base/wmf-config/db.php"));
$clusterAssignments = array();

$all = array_filter(array_map('trim', file(getRealmSpecificFilename("$base/all.dblist"))));
listOut('all', $all, false);
printf("%d wikis listed in " . basename(getRealmSpecificFilename("$base/all.dblist")) . "...\n", count($all));

$special = array_filter(array_map('trim', file(getRealmSpecificFilename("$base/special.dblist"))));
listOut('special', $special, false);
printf("%d special wikis to be exempted from wikipedia group...\n", count($special));

$suffixes = array(
	"wiki" => "wikipedia",
	"wikinews" => "wikinews",
	"wikibooks" => "wikibooks",
	"wikiquote" => "wikiquote",
	"wikisource" => "wikisource",
	"wiktionary" => "wiktionary",
	"wikimedia" => "wikimedia",
	"wikiversity" => "wikiversity",
	"wikivoyage" => "wikivoyage" );

$groups = array();
$specialMatch = array();

foreach ($all as $wiki) {
	if (isset($clusterOverrides[$wiki])) {
		$cluster = $clusterOverrides[$wiki];
	} else {
		$cluster = $defaultCluster;
	}
	$clusterAssignments[$cluster][] = $wiki;

	if (in_array($wiki, $special)) {
		$specialMatch[] = $wiki;
		continue;
	}
	foreach ($suffixes as $suffix => $group) {
		if (substr($wiki, -strlen($suffix)) == $suffix) {
			$groups[$group][] = $wiki;
			continue 2;
		}
	}
	printf("Warning: wiki '%s' matches no known group.\n", $wiki);
}

if (count($special) != count($specialMatch)) {
	printf("Warning: expected %d special wikis, matched %d.\n",
		count($special), count($specialMatch));
}

foreach ($groups as $group => $list) {
	listOut($group, $list);
}

echo "Writing out db cluster assignments...\n";

foreach ($clusterAssignments as $cluster => $list) {
	listOut($cluster, $list);
}

echo "Done.\n";

function loadClusterList($file) {
	$wgDBname = $wgDBuser = $wgDBpassword = '';
	include( $file );

	return $wgLBFactoryConf['sectionsByDB'];
}

function listOut($group, $list, $print = true) {
	global $base;

	$filename = getRealmSpecificFilename("$base/$group.dblist");
	if ( $print ) {
		printf("Listing %d wikis in %s...\n", count($list), $filename);
	}

	$out = fopen("$filename.new", "xt");
	if ($out) {
		sort($list);
		foreach ($list as $wiki) {
			fwrite($out, "$wiki\n");
		}
		fclose($out);

		if (file_exists("$filename.bak")) {
			unlink("$filename.bak");
		}
		if (file_exists($filename)) {
			rename($filename, "$filename.bak");
		}
		rename("$filename.new", $filename);
		if (file_exists("$filename.bak")) {
			unlink("$filename.bak");
		}
	}
}