continue;
}
$content = file_get_contents("projects/{$project}/log/stat.log");
if (!preg_match("#tokens_count : (\\d+)#is", $content, $r)) {
continue;
}
$size = (int) $r[1];
if (!file_exists("projects/{$project}/datastore.sqlite")) {
continue;
}
$sqlite = new sqlite3("projects/{$project}/datastore.sqlite");
$res = $sqlite->query('SELECT * FROM hash WHERE key = "tokens"');
$initialSize = $res->fetchArray()['value'];
if ($initialSize == 0) {
print "php exakat project -v -p {$project}\n";
continue;
}
$res = array($project, $initialSize, $size, $buildRoot, $tokenizer, $analyze, $final);
$regression->addData(floor($final), [floor($size)]);
if ($initialSize < $size) {
print "Tokens grown : {$project}\n";
}
fputcsv($fp, $res);
++$total;
}
print "Did {$total} files\n";
// print "predict : ".."\n";
$coefficients = $regression->getCoefficients();
print_r($coefficients);
$intercept = round($regression->predict([0]), 2);
print "{$intercept}\n";
