Strip Tables from a page (inc. inner tables)

[celebx]

Hi!

I am looking for a solution to obtain tables from a page into an array, e.g. :

$tables = array('tables' =>array(0 => array('rows' => array(0 => array('fields' => array(0 => '... HTML ...'))))));

Anyone has an idear how to do it ?

(please note, it should work when a table contains tables inside a field!)

Greetz,
Snt

[celebx]

	function identify_tables($tables) {
		$table_info=array();
		foreach ($tables as $table_id => $table) {

		$tabletags_start = $this->multi_strpos("<table.*>", $table);
		$tabletags_end = $this->multi_strpos("</table>", $table);

		if (count($tabletags_start)>2) {
			print 'innerarrays '.count($tabletags_start).'<br>';
			continue 1;
		}

		preg_match("|<table(.*)>|U", $table, $settings);
		preg_match_all("'<tr[^>]*>.*?</tr>'si", strtolower($table), $rows);
		$table_info[$table_id]["settings"] = $settings[1];
		foreach ($rows[0] as $row_number => $row) {
			preg_match("|<tr(.*)>|U", $row, $settings);
			preg_match_all("'<td[^>]*>(.*)?</td>'U", strtolower($row), $fields);
			$row_info[$row_number]["settings"] = $settings[1];
			foreach ($fields[0] as $field_number => $field) {
				preg_match("|<td(.*)>|U", $field, $settings);
				$rowspan = '';$colspan = '';$width = '';$height = '';
				if (strpos($settings[1],'rowspan')) {
					preg_match("'rowspan=(.*) 'U", stripslashes($settings[1]), $rowspan);	
					$rowspan=$rowspan[1];
				}
				if (strpos($settings[1],'colspan')) {
					preg_match("'colspan=(.*) 'U", stripslashes($settings[1]), $colspan);	
					$colspan=$colspan[1];
				}
				if (strpos($settings[1],'width')) {
					preg_match("'width=(.*) 'U", stripslashes($settings[1]), $width);	
					$width=$width[1];
				}
				if (strpos($settings[1],'height')) {
					preg_match("'height=(.*) 'U", stripslashes($settings[1]), $height);	
					$height=$height[1];
				}
				$field_info[$field_number]["settings"]["rowspan"] = $rowspan;
				$field_info[$field_number]["settings"]["colspan"] = $colspan;
				$field_info[$field_number]["settings"]["width"]   = $width;
				$field_info[$field_number]["settings"]["height"]  = $height;

				$field_info[$field_number]["content"] 		  = $fields[1][$field_number];
			}

			$row_info[$row_number]["fields"]=$field_info;	
		}
		$table_info[$table_id]["rows"] = $row_info;


	}
	return $table_info;
}

function get_tables($content) {

	$tables=array();


	preg_match_all("'<table[^>]*>(.*)</table>'s", strip_tags(strtolower($content), '<table><tr><td><img><a>'), $tables);

print_r($tables);
//print_r(strtolower($content));

exit;

	$tabletags_start 	= $this->multi_strpos("<table.*>", $content);
	$tabletags_end 		= $this->multi_strpos("</table>", $content);



	$key=0;
	$run=0;
	while ($tabletags_start[$run]) {
		$n=$key;
		while (isset($tabletags_start[$n]) && ($tabletags_start[$n]<$tabletags_end[$key])) {
			$n++;
		}
		$start = $tabletags_start[$key];
		$end = $tabletags_end[$n];

		$tmp_table = substr($content, $start, ($end-$start));

		$tables[count($tables)] = $tmp_table.'</table>';
		$key=$n;
		$run++;
	}
	return $this->identify_tables($tables);
}