Chris@0
|
1 <?php
|
Chris@0
|
2 namespace Masterminds\HTML5\Parser;
|
Chris@0
|
3
|
Chris@0
|
4 use Masterminds\HTML5\Entities;
|
Chris@0
|
5
|
Chris@0
|
6 /**
|
Chris@0
|
7 * Manage entity references.
|
Chris@0
|
8 *
|
Chris@0
|
9 * This is a simple resolver for HTML5 character reference entitites.
|
Chris@0
|
10 * See \Masterminds\HTML5\Entities for the list of supported entities.
|
Chris@0
|
11 */
|
Chris@0
|
12 class CharacterReference
|
Chris@0
|
13 {
|
Chris@0
|
14
|
Chris@0
|
15 protected static $numeric_mask = array(
|
Chris@0
|
16 0x0,
|
Chris@0
|
17 0x2FFFF,
|
Chris@0
|
18 0,
|
Chris@0
|
19 0xFFFF
|
Chris@0
|
20 );
|
Chris@0
|
21
|
Chris@0
|
22 /**
|
Chris@0
|
23 * Given a name (e.g.
|
Chris@0
|
24 * 'amp'), lookup the UTF-8 character ('&')
|
Chris@0
|
25 *
|
Chris@0
|
26 * @param string $name
|
Chris@0
|
27 * The name to look up.
|
Chris@0
|
28 * @return string The character sequence. In UTF-8 this may be more than one byte.
|
Chris@0
|
29 */
|
Chris@0
|
30 public static function lookupName($name)
|
Chris@0
|
31 {
|
Chris@0
|
32 // Do we really want to return NULL here? or FFFD
|
Chris@0
|
33 return isset(Entities::$byName[$name]) ? Entities::$byName[$name] : null;
|
Chris@0
|
34 }
|
Chris@0
|
35
|
Chris@0
|
36 /**
|
Chris@0
|
37 * Given a Unicode codepoint, return the UTF-8 character.
|
Chris@0
|
38 *
|
Chris@0
|
39 * (NOT USED ANYWHERE)
|
Chris@0
|
40 */
|
Chris@0
|
41 /*
|
Chris@0
|
42 * public static function lookupCode($codePoint) { return 'POINT'; }
|
Chris@0
|
43 */
|
Chris@0
|
44
|
Chris@0
|
45 /**
|
Chris@0
|
46 * Given a decimal number, return the UTF-8 character.
|
Chris@0
|
47 */
|
Chris@0
|
48 public static function lookupDecimal($int)
|
Chris@0
|
49 {
|
Chris@0
|
50 $entity = '&#' . $int . ';';
|
Chris@0
|
51 // UNTESTED: This may fail on some planes. Couldn't find full documentation
|
Chris@0
|
52 // on the value of the mask array.
|
Chris@0
|
53 return mb_decode_numericentity($entity, static::$numeric_mask, 'utf-8');
|
Chris@0
|
54 }
|
Chris@0
|
55
|
Chris@0
|
56 /**
|
Chris@0
|
57 * Given a hexidecimal number, return the UTF-8 character.
|
Chris@0
|
58 */
|
Chris@0
|
59 public static function lookupHex($hexdec)
|
Chris@0
|
60 {
|
Chris@0
|
61 return static::lookupDecimal(hexdec($hexdec));
|
Chris@0
|
62 }
|
Chris@0
|
63 }
|