Mobi to XDXF

<?php

$re = '/<idx:entry.*?idx:orth value="([^"]+)">(.*?)<\/idx:entry>/m';

$str=file_get_contents('book.html');

$str=str_replace('<font size="3">en&sup2;','',$str);
$str=str_replace('<font size="3">pl&sup1;','',$str);
$str=str_replace('<font size="3">pl&sup3;','',$str);
$str=str_replace('<blockquote>','',$str);
$str=str_replace('</blockquote>','',$str);
$str=str_replace('<div height="0"></div> <div height="0"></div>','<br/>',$str);
$str=str_replace('<div height="0"></div>','<br/>',$str);

file_put_contents('book2.html',$str);


preg_match_all($re, $str, $matches, PREG_SET_ORDER, 0);

$xml1='<?xml version="1.0" encoding="UTF-8" ?>
<xdxf lang_from="ENG" lang_to="POL" format="visual">
    <meta_info>
        <title>Dictionary ENG-POL</title>
        <full_title>Dictionary ENG-POL</full_title>
        <description>Slownik angielsko polski</description>
        <file_ver>001</file_version>
        <creation_date>09-02-2020</creation_date>
    </meta_info>
    <lexicon>';
$szab="<ar><k>%s</k><def>%s</def></ar>\n";
$xml2='</lexicon>
</xdxf>';

$fp = fopen('slownik.xdxf', 'w');
fwrite($fp, $xml1);


foreach($matches as $m){
    $slowo=$m[1];
    $opis=$m[2];

    if(!empty($slowo) && !empty($opis)){
        $opis=str_replace('<br/>','__br__',$opis);
        $opis=str_replace('<b>','__b__',$opis);
        $opis=str_replace('</b>','__/b__',$opis);
        $opis=str_replace('<i>','__i__',$opis);
        $opis=str_replace('</i>','__/i__',$opis);
        $opis=str_replace('<h2>','__b__',$opis);
        $opis=str_replace('</h2>','__/b__',$opis);
        $opis=html_entity_decode(strip_tags($opis));
        $opis=htmlentities($opis,ENT_XML1);
        $opis=str_replace('__br__','<br/>',$opis);
        if(substr_count($opis,'__b__')==substr_count($opis,'__/b__')){
            $opis=str_replace('__b__','<b>',$opis);
            $opis=str_replace('__/b__','</b>',$opis);
        }
        if(substr_count($opis,'__i__')==substr_count($opis,'__/i__')){
            $opis=str_replace('__i__','<i>',$opis);
            $opis=str_replace('__/i__','</i>',$opis);
        }
        fwrite($fp, sprintf($szab,$slowo,$opis));

    }

}


fwrite($fp, $xml2);
fclose($fp);