|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace UnicornFail\Emoji; |
|
6
|
|
|
|
|
7
|
|
|
use UnicornFail\Emoji\Emojibase\ShortcodeInterface; |
|
8
|
|
|
use UnicornFail\Emoji\Exception\LocalePresetException; |
|
9
|
|
|
use UnicornFail\Emoji\Token\EmojiTokenInterface; |
|
10
|
|
|
use UnicornFail\Emoji\Token\Text; |
|
11
|
|
|
use UnicornFail\Emoji\Token\TokenInterface; |
|
12
|
|
|
|
|
13
|
|
|
class Parser implements ParserInterface |
|
14
|
|
|
{ |
|
15
|
|
|
public const T_EMOJI_TOKENS = [ |
|
16
|
|
|
Lexer::T_EMOTICON => '\UnicornFail\Emoji\Token\Emoticon', |
|
17
|
|
|
Lexer::T_HTML_ENTITY => 'UnicornFail\Emoji\Token\HtmlEntity', |
|
18
|
|
|
Lexer::T_SHORTCODE => 'UnicornFail\Emoji\Token\Shortcode', |
|
19
|
|
|
Lexer::T_UNICODE => 'UnicornFail\Emoji\Token\Unicode', |
|
20
|
|
|
]; |
|
21
|
|
|
|
|
22
|
|
|
public const T_DATASETS = [ |
|
23
|
|
|
Lexer::T_EMOTICON => 'emoticon', |
|
24
|
|
|
Lexer::T_HTML_ENTITY => 'htmlEntity', |
|
25
|
|
|
Lexer::T_SHORTCODE => 'shortcodes', |
|
26
|
|
|
Lexer::T_UNICODE => 'unicode', |
|
27
|
|
|
]; |
|
28
|
|
|
|
|
29
|
|
|
/** @var ConfigurationInterface */ |
|
30
|
|
|
private $configuration; |
|
31
|
|
|
|
|
32
|
|
|
/** @var Dataset */ |
|
33
|
|
|
private $dataset; |
|
34
|
|
|
|
|
35
|
|
|
/** @var Lexer */ |
|
36
|
|
|
private $lexer; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @param mixed[]|\Traversable $configuration |
|
40
|
|
|
*/ |
|
41
|
606 |
|
public function __construct(?iterable $configuration = null, ?Dataset $dataset = null, ?Lexer $lexer = null) |
|
42
|
|
|
{ |
|
43
|
606 |
|
$this->configuration = Configuration::create($configuration); |
|
44
|
|
|
|
|
45
|
606 |
|
$locale = $this->configuration->get('locale'); |
|
46
|
|
|
\assert(\is_string($locale)); |
|
47
|
|
|
|
|
48
|
|
|
/** @var string[] $preset */ |
|
49
|
606 |
|
$preset = $this->configuration->get('preset'); |
|
50
|
|
|
|
|
51
|
606 |
|
$this->dataset = $dataset ?? self::loadLocalePreset($locale, $preset); |
|
52
|
192 |
|
$this->lexer = $lexer ?? new Lexer($this->configuration); |
|
53
|
192 |
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @param string[] $presets |
|
57
|
|
|
*/ |
|
58
|
606 |
|
protected static function loadLocalePreset(string $locale = 'en', array $presets = ShortcodeInterface::DEFAULT_PRESETS): Dataset |
|
59
|
|
|
{ |
|
60
|
606 |
|
$throwables = []; |
|
61
|
606 |
|
$presets = \array_filter($presets); |
|
62
|
606 |
|
$remaining = $presets; |
|
63
|
606 |
|
while (\count($remaining) > 0) { |
|
64
|
606 |
|
$preset = \array_shift($remaining); |
|
65
|
|
|
try { |
|
66
|
606 |
|
return Dataset::unarchive(\sprintf('%s/%s/%s.gz', Dataset::DIRECTORY, $locale, $preset)); |
|
67
|
414 |
|
} catch (\Throwable $throwable) { |
|
68
|
414 |
|
$throwables[$preset] = $throwable; |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
414 |
|
throw new LocalePresetException($locale, $throwables); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
81 |
|
public function getConfiguration(): ConfigurationInterface |
|
76
|
|
|
{ |
|
77
|
81 |
|
return $this->configuration; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @return TokenInterface[] |
|
82
|
|
|
*/ |
|
83
|
81 |
|
public function parse(string $input): array |
|
84
|
|
|
{ |
|
85
|
81 |
|
$this->lexer->setInput($input); |
|
86
|
81 |
|
$this->lexer->moveNext(); |
|
87
|
|
|
|
|
88
|
81 |
|
return $this->parseTokens(); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @return TokenInterface[] |
|
93
|
|
|
*/ |
|
94
|
81 |
|
protected function parseTokens(): array |
|
95
|
|
|
{ |
|
96
|
81 |
|
$tokens = []; |
|
97
|
81 |
|
while (true) { |
|
98
|
81 |
|
if (! $this->lexer->lookahead) { |
|
99
|
81 |
|
break; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
81 |
|
$this->lexer->moveNext(); |
|
103
|
|
|
|
|
104
|
81 |
|
$type = (int) ($this->lexer->token['type'] ?? Lexer::T_TEXT); |
|
105
|
81 |
|
$value = (string) ($this->lexer->token['value'] ?? ''); |
|
106
|
|
|
|
|
107
|
81 |
|
if ($token = $type === Lexer::T_TEXT ? $this->parseTextToken($value) : $this->parseToken($type, $value)) { |
|
108
|
81 |
|
$tokens[] = $token; |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
81 |
|
return $tokens; |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
81 |
|
protected function parseToken(int $type, string $value): ?EmojiTokenInterface |
|
116
|
|
|
{ |
|
117
|
81 |
|
$token = null; |
|
118
|
|
|
|
|
119
|
|
|
// Immediately return if not a valid type. |
|
120
|
81 |
|
if (isset(self::T_DATASETS[$type]) || isset(self::T_EMOJI_TOKENS[$type])) { |
|
121
|
81 |
|
$dataset = $this->dataset->indexBy(self::T_DATASETS[$type]); |
|
122
|
|
|
|
|
123
|
81 |
|
$tokenClass = self::T_EMOJI_TOKENS[$type]; |
|
124
|
|
|
|
|
125
|
81 |
|
if ($emoji = $dataset->offsetGet($value)) { |
|
126
|
|
|
// Clone the configuration here. This is necessary so it can be passed to tokens, |
|
127
|
|
|
// which may be rendered at a later time; when the configuration may have changed. |
|
128
|
|
|
/** @var EmojiTokenInterface $token */ |
|
129
|
81 |
|
$token = new $tokenClass($value, clone $this->configuration, $emoji); |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
81 |
|
return $token; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
66 |
|
protected function parseTextToken(string $value): ?Text |
|
137
|
|
|
{ |
|
138
|
66 |
|
$text = ''; |
|
139
|
66 |
|
while (true) { |
|
140
|
66 |
|
$text .= $value; |
|
141
|
66 |
|
if ($this->lexer->lookahead === null || $this->lexer->lookahead['type'] !== Lexer::T_TEXT) { |
|
142
|
66 |
|
break; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
66 |
|
$value = (string) ($this->lexer->lookahead['value'] ?? ''); |
|
146
|
|
|
|
|
147
|
66 |
|
$this->lexer->moveNext(); |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
66 |
|
return $text |
|
151
|
66 |
|
? new Text($text) |
|
152
|
66 |
|
: null; |
|
153
|
|
|
} |
|
154
|
|
|
} |
|
155
|
|
|
|
unicorn-fail /
emoji
