<?php
function documentNotUTF8($content)
{
$exprs = array(
'#<\?xml[^>]+encoding=([\'"])[Uu][Tt][Ff]-8\1#U',
'#<meta\s+(?:(?:name\s*=\s*([\'"])http_equiv\1\s*)|(?:content\s*=\s*([\'"])text/html\s*;.*charset\s*=\s*utf-?8.*\2\s*)){2}#Ui',
);
foreach ($exprs as $expr) {
if ( $match = preg_match($expr, $content) ) {
break;
}
}
return (bool) $match;
}
$content = array(
<<<HTML
<?xml version="1.0" encoding="foo"?>
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
,
<<<HTML
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<META NAME="HTTP_EQUIV" CONTENT="text/html; charset=UTF-8">
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
,
<<<HTML
<?xml version="1.0" encoding="foo" bar="baz"?>
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
,
<<<HTML
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<META CONTENT="text/html; charset=UTF-8" NAME="HTTP_EQUIV">
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
,
<<<HTML
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<META NAME="HTTP_EQUIV" CONTENT="text/html; charset=UTF-8">
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
,
<<<HTML
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<META NAME="HTTP_EQUIV" CONTENT="text/html; stuff=things, charset=UTF-8, otherparam=foo">
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
,
<<<HTML
<HTML>
<!-- $Revision: 317718 $ -->
<HEAD>
<TITLE>PHP Manual</TITLE>
<LINK REL="STYLESHEET" HREF="style.css">
</HEAD>
HTML
);
foreach ($content as $str) {
var_dump(documentNotUTF8($content));
}
preferences:
39.53 ms | 402 KiB | 5 Q