|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace UnicornFail\Emoji\Parser; |
|
6
|
|
|
|
|
7
|
|
|
use UnicornFail\Emoji\Dataset\Emoji as DatasetEmoji; |
|
8
|
|
|
use UnicornFail\Emoji\Environment\EnvironmentInterface; |
|
9
|
|
|
use UnicornFail\Emoji\Event\DocumentParsedEvent; |
|
10
|
|
|
use UnicornFail\Emoji\Event\DocumentPreParsedEvent; |
|
11
|
|
|
use UnicornFail\Emoji\Lexer\EmojiLexer; |
|
12
|
|
|
use UnicornFail\Emoji\Node\Document; |
|
13
|
|
|
use UnicornFail\Emoji\Node\Emoji; |
|
14
|
|
|
use UnicornFail\Emoji\Node\Text; |
|
15
|
|
|
|
|
16
|
|
|
final class EmojiParser implements EmojiParserInterface |
|
17
|
|
|
{ |
|
18
|
|
|
public const INDICES = [ |
|
19
|
|
|
EmojiLexer::T_EMOTICON => 'emoticon', |
|
20
|
|
|
EmojiLexer::T_HTML_ENTITY => 'htmlEntity', |
|
21
|
|
|
EmojiLexer::T_SHORTCODE => 'shortcodes', |
|
22
|
|
|
EmojiLexer::T_UNICODE => 'unicode', |
|
23
|
|
|
]; |
|
24
|
|
|
|
|
25
|
|
|
/** @var EnvironmentInterface */ |
|
26
|
|
|
private $environment; |
|
27
|
|
|
|
|
28
|
|
|
/** @var EmojiLexer */ |
|
29
|
|
|
private $lexer; |
|
30
|
|
|
|
|
31
|
|
|
public function __construct(EnvironmentInterface $environment, ?EmojiLexer $lexer = null) |
|
32
|
|
|
{ |
|
33
|
|
|
$this->environment = $environment; |
|
34
|
|
|
$this->lexer = $lexer ?? new EmojiLexer($environment); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function getLexer(): EmojiLexer |
|
38
|
|
|
{ |
|
39
|
|
|
return $this->lexer; |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
public function parse(string $input): Document |
|
43
|
|
|
{ |
|
44
|
|
|
$preParsedEvent = new DocumentPreParsedEvent(new Document(), $input); |
|
45
|
|
|
$this->environment->dispatch($preParsedEvent); |
|
46
|
|
|
|
|
47
|
|
|
$document = $preParsedEvent->getDocument(); |
|
48
|
|
|
$input = $preParsedEvent->getInput(); |
|
49
|
|
|
|
|
50
|
|
|
$this->lexer->setInput($input); |
|
51
|
|
|
$this->lexer->moveNext(); |
|
52
|
|
|
|
|
53
|
|
|
while (true) { |
|
54
|
|
|
if (! $this->lexer->lookahead) { |
|
55
|
|
|
break; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
$this->lexer->moveNext(); |
|
59
|
|
|
|
|
60
|
|
|
/** @var array<string, mixed> $token */ |
|
61
|
|
|
$token = $this->lexer->token; |
|
62
|
|
|
|
|
63
|
|
|
$value = ''; |
|
64
|
|
|
if (((string) $token['value']) !== '') { |
|
65
|
|
|
$value = (string) ($token['value'] ?? ''); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
$type = (int) $token['type']; |
|
69
|
|
|
|
|
70
|
|
|
$node = null; |
|
71
|
|
|
if ($type !== EmojiLexer::T_TEXT && \in_array($type, EmojiLexer::TYPES, true)) { |
|
72
|
|
|
$node = $this->parseEmoji($type, $value); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
if ($node === null) { |
|
76
|
|
|
$node = $this->parseText($value); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
$document->appendNode($node); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
$this->environment->dispatch(new DocumentParsedEvent($document)); |
|
83
|
|
|
|
|
84
|
|
|
return $document; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
protected function parseEmoji(int $type, string $value): ?Emoji |
|
88
|
|
|
{ |
|
89
|
|
|
$dataset = $this->environment->getRuntimeDataset(self::INDICES[$type]); |
|
90
|
|
|
|
|
91
|
|
|
try { |
|
92
|
|
|
/** @var DatasetEmoji $emoji */ |
|
93
|
|
|
$emoji = $dataset->offsetGet($value); |
|
94
|
|
|
|
|
95
|
|
|
return new Emoji($type, $value, $emoji); |
|
96
|
|
|
} catch (\OutOfRangeException $exception) { |
|
97
|
|
|
return null; |
|
98
|
|
|
} |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
protected function parseText(string $value): Text |
|
102
|
|
|
{ |
|
103
|
|
|
$text = ''; |
|
104
|
|
|
while (true) { |
|
105
|
|
|
$text .= $value; |
|
106
|
|
|
if ($this->lexer->lookahead === null || $this->lexer->lookahead['type'] !== EmojiLexer::T_TEXT) { |
|
107
|
|
|
break; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
$value = (string) ($this->lexer->lookahead['value'] ?? ''); |
|
111
|
|
|
|
|
112
|
|
|
$this->lexer->moveNext(); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
return new Text($text); |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
|
|
|
unicorn-fail /
emoji
