public static function crawl($source)
{
$crawler = new DomCrawler($source);
Log::log("Generating tree..");
$node_tree = Parser::parse($crawler, 0, "", false);
Log::log("Calculating components..");
$augmented_tree = Parser::augment($node_tree[0]);
Log::log("Calculating path..");
$marked_tree = Parser::identify_parent_path($augmented_tree);
file_put_contents("a-tree", print_r($marked_tree, true));
Log::log("Scanning the tree for a pattern..");
Scanner::scan($marked_tree);
// Test di funzionamento delle funzioni get_node_from_path e get_parent_from_path
//Log::log( Tree::get_node_from_path( $marked_tree, "0,0" )[ 'tag' ] );
//Log::log( Tree::get_parent_from_path( $marked_tree, "0,0" )[ 'tag' ] );
}
public static function scan_for_pattern($root, $node = "", $priority = [])
{
if (@(!$node)) {
$node = $root;
}
$children = Tree::get_tagged_children($node);
if (@(!!$children)) {
if (Node::get_class($node) == "posts-list-inner") {
//$elig = Scanner::are_eligible( $root, $node, $children );
//Log::log( "Children count is ".count( $children )." ".( $elig ? "and" : "but" )." they are ".( $elig ? "" : "not" )." eligible." );
}
if (Scanner::are_eligible($root, $node, $children) && !Scanner::has_been_prioritized($node['path'])) {
$node['count'] = count($children);
$id = Node::get_id($node);
if ($id !== false) {
$node['id'] = "#" . $id;
}
$class = Node::get_class($node);
if ($class !== false) {
$node['class'] = "." . $class;
}
$priority[] = $node;
Scanner::$been_prioritized[] = $node['path'];
}
foreach ($children as $child) {
$priority = Scanner::scan_for_pattern($root, $child, $priority);
}
}
return $priority;
}
