|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace UCD\Infrastructure\Repository\CharacterRepository\XMLRepository\ElementParser; |
|
4
|
|
|
|
|
5
|
|
|
use UCD\Unicode\Codepoint; |
|
6
|
|
|
use UCD\Unicode\CodepointAssigned; |
|
7
|
|
|
use UCD\Exception\RuntimeException; |
|
8
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\XMLRepository\ElementParser; |
|
9
|
|
|
|
|
10
|
|
|
class CodepointAssignedParser implements ElementParser |
|
11
|
|
|
{ |
|
12
|
|
|
const TAG_NAME_CHAR = 'char'; |
|
13
|
|
|
const TAG_NAME_NON_CHAR = 'noncharacter'; |
|
14
|
|
|
const TAG_NAME_SURROGATE = 'surrogate'; |
|
15
|
|
|
|
|
16
|
|
|
const ATTR_CODEPOINT = 'cp'; |
|
17
|
|
|
const ATTR_CODEPOINT_FIRST = 'first-cp'; |
|
18
|
|
|
const ATTR_CODEPOINT_LAST = 'last-cp'; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var CharacterParser |
|
22
|
|
|
*/ |
|
23
|
|
|
private $characterParser; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var NonCharacterParser |
|
27
|
|
|
*/ |
|
28
|
|
|
private $nonCharacterParser; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @var SurrogateParser |
|
32
|
|
|
*/ |
|
33
|
|
|
private $surrogateParser; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @param CharacterParser $characterParser |
|
37
|
|
|
* @param NonCharacterParser $nonCharacterParser |
|
38
|
|
|
* @param SurrogateParser $surrogateParser |
|
39
|
|
|
*/ |
|
40
|
|
|
public function __construct( |
|
41
|
|
|
CharacterParser $characterParser, |
|
42
|
|
|
NonCharacterParser $nonCharacterParser, |
|
43
|
|
|
SurrogateParser $surrogateParser |
|
44
|
|
|
) { |
|
45
|
|
|
$this->characterParser = $characterParser; |
|
46
|
|
|
$this->nonCharacterParser = $nonCharacterParser; |
|
47
|
|
|
$this->surrogateParser = $surrogateParser; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* @param \DOMElement $element |
|
52
|
|
|
* @return CodepointAssigned[] |
|
53
|
|
|
*/ |
|
54
|
|
|
public function parseElement(\DOMElement $element) |
|
55
|
|
|
{ |
|
56
|
|
|
$parser = $this->getParserForElement($element); |
|
57
|
|
|
$codepointValues = $this->extractCodepoints($element); |
|
58
|
|
|
|
|
59
|
|
|
foreach ($codepointValues as $codepointValue) { |
|
60
|
|
|
$codepoint = Codepoint::fromInt($codepointValue); |
|
61
|
|
|
yield $parser->parseElement($element, $codepoint); |
|
62
|
|
|
} |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* @param \DOMElement $element |
|
67
|
|
|
* @return CodepointAwareParser |
|
68
|
|
|
* @throws RuntimeException |
|
69
|
|
|
*/ |
|
70
|
|
|
private function getParserForElement(\DOMElement $element) |
|
71
|
|
|
{ |
|
72
|
|
|
if ($element->tagName === self::TAG_NAME_CHAR) { |
|
73
|
|
|
return $this->characterParser; |
|
74
|
|
|
} elseif ($element->tagName === self::TAG_NAME_NON_CHAR) { |
|
75
|
|
|
return $this->nonCharacterParser; |
|
76
|
|
|
} elseif ($element->tagName === self::TAG_NAME_SURROGATE) { |
|
77
|
|
|
return $this->surrogateParser; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
throw new RuntimeException(sprintf('No parser found for %s', $element->tagName)); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* @param \DOMElement $element |
|
85
|
|
|
* @return \int[] |
|
86
|
|
|
*/ |
|
87
|
|
|
private function extractCodepoints(\DOMElement $element) |
|
88
|
|
|
{ |
|
89
|
|
|
if ($element->hasAttribute(self::ATTR_CODEPOINT)) { |
|
90
|
|
|
$first = hexdec($element->getAttribute(self::ATTR_CODEPOINT)); |
|
91
|
|
|
$last = $first; |
|
92
|
|
|
} else { |
|
93
|
|
|
$first = hexdec($element->getAttribute(self::ATTR_CODEPOINT_FIRST)); |
|
94
|
|
|
$last = hexdec($element->getAttribute(self::ATTR_CODEPOINT_LAST)); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
return range($first, $last); |
|
98
|
|
|
} |
|
99
|
|
|
} |
nick-jones /
php-ucd
