<?php

function htmlclean($html) {
  $html= str_replace('</blockquote>','',$html);
  $html= str_replace('<blockquote>','',$html);
  $html= preg_replace('/\\r\\n/', '', $html);
 $html= preg_replace_callback('/(\\n)\\n/', function ($m) {
                                         return "$m[1]";
                                      }, $html);
 $html= preg_replace('/\s{3,}/sm','',$html,PREG_SET_ORDER); //Mehr als zwei leerzeichen entfernen
 $html= preg_replace('/\s{2,}/sm',' ',$html,PREG_SET_ORDER); //Mehr als zwei leerzeichen entfernen
 # $html= trim($html);
  return $html;
}

$url=fopen("1mo1_bibeltext.htm", "r");
#$url=fopen("ps119_bibeltext.htm", "r");
#$url=fopen("ps1_bibeltext.htm", "r");

if ($url == NULL) {
        return "Fehler (1)";
        exit;
}
while(!feof($url)) {
  $buffer .= fgets($url, 4096);
}
# Headline H0
$pos_ab = "";
$pos_bis = "";
$buffer2 = "";
$pos_ab =strpos($buffer, '<li class="name">');
if($pos_ab != ''){
$buffer2 = substr($buffer,$pos_ab);
$pos_bis =strpos($buffer2, '</li>');
$name = substr($buffer2,0,$pos_bis);
$name ="<h1>".str_replace('<li class="name">', '',$name)."</h1>";
}else{
$name = "";
}


# H2
$pos_ab = "";
$pos_bis = "";
$buffer2 = "";
$pos_ab = preg_match('/<h1 id="h[0-9]">/', $buffer, $matches, PREG_OFFSET_CAPTURE) ? $matches[0][1] : '';
if($pos_ab != ''){
$buffer2 = substr($buffer,$pos_ab);
$pos_bis =strpos($buffer2, '</h1>');
$h2 = substr($buffer2,0,$pos_bis);
$h2 = "<h2>".preg_replace('/<h1 id="h[0-9]">/', '',$h2)."</h2>";
}else{
$h2 = "";
}


# H3
$pos_ab = "";
$pos_bis = "";
$buffer2 = "";
$pos_ab = preg_match('/<h2 id="h[0-9]">/', $buffer, $matches, PREG_OFFSET_CAPTURE) ? $matches[0][1] : '';
if($pos_ab != ''){
$buffer2 = substr($buffer,$pos_ab);
$pos_bis =strpos($buffer2, '</h2>');
$h3 = substr($buffer2,0,$pos_bis);
$h3 = "<h3>".preg_replace('/<h2 id="h[0-9]">/', '',$h3)."</h3>";
}else{
$h3 = "";
}

# Schleife �ber Verse
#<span class="verse" data-location="24:1:1:1">1</span>
$pos_ab = "";
$pos_bis = "";
$buffer2 = "";
# Bis zu 3 Kapitelstellen Psalm 150
$pos_ab = preg_match('/<p><span class="chapter">[0-9][0-9]?[0-9]?<\/span>/', $buffer, $matches, PREG_OFFSET_CAPTURE) ? $matches[0][1] : '';
if($pos_ab != ''){
$buffer2 = substr($buffer,$pos_ab);
$pos_bis =strpos($buffer2, '<nav class="bible-navigation bible-navigation--simple">');
$verse_ges = substr($buffer2,0,$pos_bis);

$verse_ges = preg_replace('/<span class="chapter">[0-9][0-9]?[0-9]?<\/span>/', '',$verse_ges);



$verse_ges = preg_replace_callback('/<p><span class="verse" data-location=".*">([0-9]*)<\/span>&nbsp;(.*\n*.*\n*.*\n*.*\n*)<\/p>/',
                                         function ($m) {
                                           return "||$m[1]|" . "<p>$m[2]</p>";
                                         },
                                         $verse_ges);

$verse_ges = htmlclean($verse_ges); 
$verse_ges = preg_split('/\|\|/', $verse_ges);


}else{
$verse_ges = "";
}





echo "<pre>\n";
echo print_r($verse_ges);
echo "</pre>";
?>