<?php

$data = file_get_contents('raw.txt');

$data = preg_replace('~^.*<!-- TOC END -->\\s+~s', '', $data);

$data = preg_replace('~\\s*<!-- wikipage \\w+ -->\\s*~s', '', $data);

$data = preg_replace('~\\s*<h1>(?:<a[^>]+>)?([^<]+)(?:</a>)?</h1>\\s*~', "\n\n$1\n", $data);

$data = preg_replace('~\\s*<h2>(?:<a[^>]+>)?([^<]+)(?:</a>)?</h2>\\s*~', "\n\n$1\n", $data);

$data = preg_replace('~\\s*<h3>(?:<a[^>]+>)?([^<]+)(?:</a>)?</h3>\\s*~', "\n\n$1", $data);

$data = preg_replace('~\\s*<h\\d>(?:<a[^>]+>)?([^<]+)(?:</a>)?</h\\d>\\s*~', "\n\n$1", $data);

$data = preg_replace('~<div class="level\\d+">|</div>|<p>|</p>|</?(?:acronym|tbody|br|dl|dt|dd)[^>]*>|</li>~', '', $data);

$data = preg_replace('~<a [^>]*href="(?:https://web.archive.org/web/\\d+/)?([^"]+)"[^>]+>(.*?)</a>~', '$2', $data);

$data = preg_replace('~(?:\\[url=[^>]+\\])?<img [^>]*src="[^"]*(/[^/"]+)"[^>]+/?>(?:\\[/url\\])?~', '{Image (attachment) $1}', $data);

$data = preg_replace('~<embed [^>]*src="[^"]*(/[^/"]+)"[^>]+/?>~', '{attachment $1}', $data);

$data = preg_replace('~<(?trong|b)>(.*?)</(?trong|b)>~s', '$1', $data);

$data = preg_replace('~<(?:em|i)>(.*?)</(?:em|i)>~s', '$1', $data);

$data = preg_replace('~<(?:u)>(.*?)</(?:u)>~s', '$1', $data);

$data = preg_replace('~<(?:code)>(.*?)</(?:code)>~s', '
Code :
$1
', $data);

$data = preg_replace('~<(?:blockquote)>(.*?)</(?:blockquote)>~s', '
Citation :$1
', $data);

$data = preg_replace('~<hr/?>~', '


', $data);

$data = preg_replace('~<(table|tr|td|th)[^>]*>\\s*(.*?)\\s*</(?:\1)>~s', '[$1]$2[/$1]', $data);

$data = preg_replace('~<(table|tr|td|th)[^>]*>\\s*(.*?)\\s*</(?:\1)>~s', '[$1]$2[/$1]', $data);

$data = preg_replace('~<(table|tr|td|th)[^>]*>\\s*(.*?)\\s*</(?:\1)>~s', '[$1]$2[/$1]', $data);

$data = str_replace(array('“', '”'), '"', $data);



$data = preg_replace_callback('~<ul>(.*?)</ul>~s', static function ($match): string {

return '

' . preg_replace('~<li class="level\\d+"><div class="li">\\s*~', '


', $match[1])

 . '



';

}, $data);



$data = preg_replace_callback('~<pre class="code( [^"]+)?">(.*?)</pre>~s', static function ($match): string {

return '[code' . (($match[1] ?? false) ? "=" . trim(trim($match[1]), '=') : '') . ']'

. preg_replace('~<[^>]+>~', '', $match[2])

. '[/ code]';

}, $data);



$data = preg_replace('~\\n{3,}~', "\n\n", $data);



$data = mb_convert_encoding(html_entity_decode($data), 'utf-8');



if (strpos($data, '<') !== false) {

echo "REMAINING < IN DATA!\n";

}



file_put_contents('out.txt', $data);