|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace UnicornFail\Emoji\Parser; |
|
6
|
|
|
|
|
7
|
|
|
use UnicornFail\Emoji\Environment\EnvironmentInterface; |
|
8
|
|
|
use UnicornFail\Emoji\Event\DocumentParsedEvent; |
|
9
|
|
|
use UnicornFail\Emoji\Event\DocumentPreParsedEvent; |
|
10
|
|
|
use UnicornFail\Emoji\Input\Input; |
|
11
|
|
|
use UnicornFail\Emoji\Node\Block\Document; |
|
12
|
|
|
use UnicornFail\Emoji\Node\Inline\Emoticon; |
|
13
|
|
|
use UnicornFail\Emoji\Node\Inline\HtmlEntity; |
|
14
|
|
|
use UnicornFail\Emoji\Node\Inline\Shortcode; |
|
15
|
|
|
use UnicornFail\Emoji\Node\Inline\Text; |
|
16
|
|
|
use UnicornFail\Emoji\Node\Inline\Unicode; |
|
17
|
|
|
|
|
18
|
|
|
class EmojiParser implements EmojiParserInterface |
|
19
|
|
|
{ |
|
20
|
|
|
public const T_DATASETS = [ |
|
21
|
|
|
Lexer::T_EMOTICON => 'emoticon', |
|
22
|
|
|
Lexer::T_HTML_ENTITY => 'htmlEntity', |
|
23
|
|
|
Lexer::T_SHORTCODE => 'shortcodes', |
|
24
|
|
|
Lexer::T_UNICODE => 'unicode', |
|
25
|
|
|
]; |
|
26
|
|
|
|
|
27
|
|
|
/** @var EnvironmentInterface */ |
|
28
|
|
|
private $environment; |
|
29
|
|
|
|
|
30
|
|
|
/** @var Lexer */ |
|
31
|
|
|
private $lexer; |
|
32
|
|
|
|
|
33
|
|
|
public function __construct(EnvironmentInterface $environment, ?Lexer $lexer = null) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->environment = $environment; |
|
36
|
|
|
$this->lexer = $lexer ?? new Lexer($environment); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
public function parse(string $input): Document |
|
40
|
|
|
{ |
|
41
|
|
|
$preParsedEvent = new DocumentPreParsedEvent(new Document(), new Input($input)); |
|
42
|
|
|
$this->environment->dispatch($preParsedEvent); |
|
43
|
|
|
|
|
44
|
|
|
$document = $preParsedEvent->getDocument(); |
|
45
|
|
|
$input = $preParsedEvent->getInput(); |
|
46
|
|
|
|
|
47
|
|
|
$lineCount = $input->getLineCount(); |
|
48
|
|
|
foreach ($input->getLines() as $lineNumber => $line) { |
|
49
|
|
|
$this->parseLine($line, $document); |
|
50
|
|
|
|
|
51
|
|
|
if ($lineNumber < $lineCount) { |
|
52
|
|
|
$document->appendChild(new Text("\n")); |
|
53
|
|
|
} |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
// If the original content ended with a new line, mimic the same. |
|
57
|
|
|
if (\preg_match_all('/.*\n|\r\n$/sm', $input->getContent()) === 1) { |
|
58
|
|
|
$document->appendChild(new Text("\n")); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
$this->environment->dispatch(new DocumentParsedEvent($document)); |
|
62
|
|
|
|
|
63
|
|
|
return $document; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
protected function parseLine(string $line, Document $document): void |
|
67
|
|
|
{ |
|
68
|
|
|
$this->lexer->setInput($line); |
|
69
|
|
|
$this->lexer->moveNext(); |
|
70
|
|
|
|
|
71
|
|
|
while (true) { |
|
72
|
|
|
if (! $this->lexer->lookahead) { |
|
73
|
|
|
break; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
$this->lexer->moveNext(); |
|
77
|
|
|
|
|
78
|
|
|
$type = (int) ($this->lexer->token['type'] ?? Lexer::T_TEXT); |
|
79
|
|
|
$value = (string) ($this->lexer->token['value'] ?? ''); |
|
80
|
|
|
|
|
81
|
|
|
if ($node = $type === Lexer::T_TEXT ? $this->parseText($value) : $this->parseEmoji($type, $value)) { |
|
82
|
|
|
$document->appendChild($node); |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
/** |
|
88
|
|
|
* @return Emoticon|HtmlEntity|Shortcode|Unicode|null |
|
89
|
|
|
*/ |
|
90
|
|
|
protected function parseEmoji(int $type, string $value) |
|
91
|
|
|
{ |
|
92
|
|
|
// Immediately return if not a valid type. |
|
93
|
|
|
if (! isset(self::T_DATASETS[$type])) { |
|
94
|
|
|
return null; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
$dataset = $this->environment->getRuntimeDataset()->indexBy(self::T_DATASETS[$type]); |
|
98
|
|
|
$emoji = $dataset->offsetGet($value); |
|
99
|
|
|
|
|
100
|
|
|
// Return if not an emoji. |
|
101
|
|
|
if (! $emoji) { |
|
102
|
|
|
return null; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
switch ($type) { |
|
106
|
|
|
case Lexer::T_EMOTICON: |
|
107
|
|
|
return new Emoticon($value, $emoji); |
|
108
|
|
|
case Lexer::T_HTML_ENTITY: |
|
109
|
|
|
return new HtmlEntity($value, $emoji); |
|
110
|
|
|
case Lexer::T_SHORTCODE: |
|
111
|
|
|
return new Shortcode($value, $emoji); |
|
112
|
|
|
case Lexer::T_UNICODE: |
|
113
|
|
|
return new Unicode($value, $emoji); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
return null; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
protected function parseText(string $value): ?Text |
|
120
|
|
|
{ |
|
121
|
|
|
$text = ''; |
|
122
|
|
|
while (true) { |
|
123
|
|
|
$text .= $value; |
|
124
|
|
|
if ($this->lexer->lookahead === null || $this->lexer->lookahead['type'] !== Lexer::T_TEXT) { |
|
125
|
|
|
break; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
$value = (string) ($this->lexer->lookahead['value'] ?? ''); |
|
129
|
|
|
|
|
130
|
|
|
$this->lexer->moveNext(); |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
return $text |
|
134
|
|
|
? new Text($text) |
|
135
|
|
|
: null; |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
unicorn-fail /
emoji
