split_html

/**
* @param $delimiter string splitting token like "--page break--"
* @param $html string the html code to split
* return array
**/
function split_html($delimiter, $html)
{
$out = explode($delimiter, $html);
$numParts = count($out);
for($i=0; $i<$numParts; $i++) {
$unclosedTags = get_unclosed_tags($out[$i]);
if (!empty($unclosedTags)) {
foreach($unclosedTags as $tag) {
$out[$i] .= "</{$tag}>";
if ($numParts > $i+1) {
$out[$i+1] = "<{$tag}>".$out[$i+1];
}
}
}
}
return $out;
}
/**
* Helper function, return the unclosed tag-names of $html in order of their appearance
**/
function get_unclosed_tags($html)
{
$open=array();
$close=array();
preg_match_all("/<([^\/> ]+)[^>\/]*>/", $html, $open);
preg_match_all("/<\/([^>]+)>/", $html, $close);
$closed = $close[1];
$stillOpen = array_filter($open[1], function($a) use (&$closed) {
$i=array_search($a, $closed);
if ($i!==false) {
unset($closed[$i]);
return false;
}
return true;
});
return $stillOpen;
}