public function save($con = null)
{
// if the first save
$going_to_save = !$this->getId();
parent::save($con);
// if it is the first then connect to google
// for the queries and get the results.
if ($going_to_save) {
$query_result = new QueryResultBulk();
$query_result->setQuery($this);
//$searchEngine = new GoogleRegexp();
//$count = $searchEngine->search($this->getQuery());
$searchEngine = new GoogleHitFetcher();
$count = $searchEngine->fetch($this->getQuery());
$query_result->setResultSize($count);
$query_result->save();
}
}
protected function execute($arguments = array(), $options = array())
{
$stop_watch = new StopWatch();
$stop_watch->start();
logline('Started processing.');
$searchEngine = new GoogleHitFetcher();
$databaseManager = new sfDatabaseManager($this->configuration);
$queries = QueryPeer::doSelect(new Criteria());
logline(sprintf("There are %s queries to process.", sizeof($queries)));
foreach ($queries as $query) {
$qr = new QueryResultBulk();
$qr->setQuery($query);
$result_size = $searchEngine->fetch($query->getQuery());
$qr->setResultSize($result_size);
logline(sprintf("Found %s results for %s.", $result_size, $query->getQuery()));
$qr->save();
}
logline(sprintf("Finished processing."));
$stop_watch->end();
logline(sprintf('Execution time: %s seconds.', $stop_watch->getTime()));
logline(sprintf('!!!!!CAN RUN %s TIMES A DAY!!!!!', 24 * 60 * 60 / $stop_watch->getTime()));
}
<?php
include dirname(__FILE__) . '/../bootstrap/unit.php';
$t = new lime_test(4, new lime_output_color());
ProjectConfiguration::getApplicationConfiguration('frontend', 'test', true);
$hitFetcher = new GoogleHitFetcher();
$hit = $hitFetcher->extractHit(file_get_contents(dirname(__FILE__) . '/google.umut.utkan.html'));
$t->is($hit, 789, 'Result count has to be 789 for file google.umut.utkan.html');
$hit = $hitFetcher->extractHit(file_get_contents(dirname(__FILE__) . '/google.less.than.10.html'));
$t->is($hit, 1, 'Result count has to be 1 for google.less.than.10.html');
$hit = $hitFetcher->extractHit(file_get_contents(dirname(__FILE__) . '/google.nothing.html'));
$t->is($hit, 0, 'Result count has to be 0 for google.nothing.html');
$hit = $hitFetcher->extractHit(file_get_contents(dirname(__FILE__) . '/google.very.much.results.html'));
$t->is($hit, 95700000, 'Result count has to be 95700000 for google.very.much.results.html');
