full-text-rss/libraries/html5php/HTML5/Parser/CharacterReference.php

64 lines
1.6 KiB
PHP
Raw Permalink Normal View History

2014-09-15 20:24:06 +00:00
<?php
2015-06-14 00:03:20 +00:00
namespace Masterminds\HTML5\Parser;
2014-09-15 20:24:06 +00:00
2015-06-14 00:03:20 +00:00
use Masterminds\HTML5\Entities;
2014-09-15 20:24:06 +00:00
/**
* Manage entity references.
*
* This is a simple resolver for HTML5 character reference entitites.
2015-06-14 00:03:20 +00:00
* See \Masterminds\HTML5\Entities for the list of supported entities.
2014-09-15 20:24:06 +00:00
*/
2015-06-14 00:03:20 +00:00
class CharacterReference
{
2014-09-15 20:24:06 +00:00
2015-06-14 00:03:20 +00:00
protected static $numeric_mask = array(
0x0,
0x2FFFF,
0,
0xFFFF
);
2014-09-15 20:24:06 +00:00
2015-06-14 00:03:20 +00:00
/**
* Given a name (e.g.
* 'amp'), lookup the UTF-8 character ('&')
*
* @param string $name
* The name to look up.
* @return string The character sequence. In UTF-8 this may be more than one byte.
*/
public static function lookupName($name)
{
// Do we really want to return NULL here? or FFFD
return isset(Entities::$byName[$name]) ? Entities::$byName[$name] : null;
}
2014-09-15 20:24:06 +00:00
2015-06-14 00:03:20 +00:00
/**
* Given a Unicode codepoint, return the UTF-8 character.
*
* (NOT USED ANYWHERE)
*/
/*
* public static function lookupCode($codePoint) { return 'POINT'; }
*/
2014-09-15 20:24:06 +00:00
2015-06-14 00:03:20 +00:00
/**
* Given a decimal number, return the UTF-8 character.
*/
public static function lookupDecimal($int)
{
$entity = '&#' . $int . ';';
// UNTESTED: This may fail on some planes. Couldn't find full documentation
// on the value of the mask array.
return mb_decode_numericentity($entity, static::$numeric_mask, 'utf-8');
}
2014-09-15 20:24:06 +00:00
2015-06-14 00:03:20 +00:00
/**
* Given a hexidecimal number, return the UTF-8 character.
*/
public static function lookupHex($hexdec)
{
return static::lookupDecimal(hexdec($hexdec));
}
2014-09-15 20:24:06 +00:00
}