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

$db = "jirafe";

$mongo = new Mongo();
$siteC = $mongo->selectCollection($db, 'sites');
$appC = $mongo->selectCollection($db, 'applications');
$userC = $mongo->selectCollection($db, 'users');
$tokenC = $mongo->selectCollection($db, 'tokens');

function findSitesWithoutApplications()
{
  global $siteC, $appC;

  $sites = iterator_to_array($siteC->find(array(), array('application' => true)));
  $appIds = iterator_to_array($appC->find(array(), array('_id' => true)));

  return array_filter($sites, function($site) use ($appIds) {
    return !isset($site['application']['$id'])
      || !isset($appIds[$site['application']['$id']]);
  });
}

function remove(array $objs, $collection)
{
  $ids = array_map(function($obj) { return $obj['_id']; }, $objs);

  $collection->remove(array('_id' => array('$in' => $ids)));
}

function findSoftdeleteBug($collection)
{
  $objs = iterator_to_array($collection->find(array(), array('url' => true)));

  return array_filter($objs, function($obj) {
    return !isset($obj['url']) || 0 !== strpos($obj['url'], 'http');
  });
}

function findUsersWithNoToken()
{
  global $userC, $tokenC;

  $users = iterator_to_array($userC->find(array(), array('_id' => true)));
  $tokens = iterator_to_array($tokenC->find(
    array('user' => array('$exists' => true)),
    array('user' => true)
  ));
  $tokenUserIds = array();
  foreach ($tokens as $token) {
    $tokenUserIds[$token['user']['$id']] = true;
  }

  return array_filter($users, function($user) use ($tokenUserIds) {
    return !isset($tokenUserIds[$user['_id']]);
  });
}

function findAppsWithNoToken()
{
  global $appC, $tokenC;

  $apps = iterator_to_array($appC->find(array(), array('_id' => true)));
  $tokens = iterator_to_array($tokenC->find(
    array('application' => array('$exists' => true)),
    array('application' => true)
  ));
  $tokenAppIds = array();
  foreach ($tokens as $token) {
    $tokenAppIds[$token['application']['$id']] = true;
  }

  return array_filter($apps, function($app) use ($tokenAppIds) {
    return !isset($tokenAppIds[$app['_id']]);
  });
}

function findTokensWithNoUserAndNoApp()
{
  global $appC, $userC, $tokenC;

  $appIds = iterator_to_array($appC->find(array(), array('_id' => true)));
  $userIds = iterator_to_array($userC->find(array(), array('_id' => true)));
  $tokens = iterator_to_array($tokenC->find(array(), array('user' => true, 'application' => true)));

  return array_filter($tokens, function($token) use($appIds, $userIds) {
    $tokenAppId = isset($token['application']['$id']) ? $token['application']['$id'] : null;
    $tokenUserId = isset($token['user']['$id']) ? $token['user']['$id'] : null;

    return ($tokenAppId == null || $tokenUserId == null)
      || (!isset($appIds[$tokenAppId]) && !isset($userIds[$tokenUserId]));
  });
}

function showUnfixableUrls(array $objs, $name)
{
  return sprintf("%d unfixable %s urls:n%s",
    count($objs),
    $name,
    implode(array_map(function($obj) {
      return sprintf("id: %s, url: %s", $obj['_id'], $obj['url']);
    }, $objs), "n")
  );
}

printf("There are %d sites in %d applicationsn", $siteC->count(), $appC->count());
print "n";

$appLessSites = findSitesWithoutApplications();

printf("Removing %d app-less sites n", count($appLessSites));
print "n";

remove($appLessSites, $siteC);

$softdeleteBugSites = findSoftdeleteBug($siteC);

printf("Removing %d sites which urls start with a number (softdelete bug)n", count($softdeleteBugSites));
print "n";

remove($softdeleteBugSites, $siteC);

$usersWithNoToken = findUsersWithNoToken();

printf("Removing %d token-less users n", count($usersWithNoToken));
print "n";

remove($usersWithNoToken, $userC);

$appsWithNoToken = findAppsWithNoToken();

printf("Removing %d token-less applications n", count($appsWithNoToken));
print "n";

remove($appsWithNoToken, $appC);

$orphanTokens = findTokensWithNoUserAndNoApp();

printf("Removing %d user-less AND application-less tokens n", count($orphanTokens));
print "n";

remove($orphanTokens, $tokenC);

print "Done.";