|
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\AbstractToken; |
|
10
|
|
|
use UnicornFail\Emoji\Token\Emoticon; |
|
11
|
|
|
use UnicornFail\Emoji\Token\HtmlEntity; |
|
12
|
|
|
use UnicornFail\Emoji\Token\Shortcode; |
|
13
|
|
|
use UnicornFail\Emoji\Token\Text; |
|
14
|
|
|
use UnicornFail\Emoji\Token\Unicode; |
|
15
|
|
|
use UnicornFail\Emoji\Util\Normalize; |
|
16
|
|
|
|
|
17
|
|
|
class Parser implements ParserInterface |
|
18
|
|
|
{ |
|
19
|
|
|
private const TYPE_METHODS = [ |
|
20
|
|
|
Lexer::T_TEXT => 'parseText', |
|
21
|
|
|
Lexer::T_EMOTICON => 'parseEmoticon', |
|
22
|
|
|
Lexer::T_HTML_ENTITY => 'parseHtmlEntity', |
|
23
|
|
|
Lexer::T_SHORTCODE => 'parseShortcode', |
|
24
|
|
|
Lexer::T_UNICODE => 'parseUnicode', |
|
25
|
|
|
]; |
|
26
|
|
|
|
|
27
|
|
|
/** @var ConfigurationInterface */ |
|
28
|
|
|
private $configuration; |
|
29
|
|
|
|
|
30
|
|
|
/** @var Dataset */ |
|
31
|
|
|
private $dataset; |
|
32
|
|
|
|
|
33
|
|
|
/** @var Lexer */ |
|
34
|
|
|
private $lexer; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @param mixed[]|\Traversable $configuration |
|
38
|
|
|
*/ |
|
39
|
606 |
|
public function __construct(?iterable $configuration = null, ?Dataset $dataset = null, ?Lexer $lexer = null) |
|
40
|
|
|
{ |
|
41
|
606 |
|
$this->configuration = Configuration::create($configuration); |
|
42
|
606 |
|
$locale = $this->configuration->get('locale'); |
|
43
|
606 |
|
$preset = $this->configuration->get('preset'); |
|
44
|
606 |
|
$this->dataset = $dataset ?? self::loadLocalePreset($locale, $preset); |
|
45
|
192 |
|
$this->lexer = $lexer ?? new Lexer($this->configuration); |
|
46
|
192 |
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @param string[] $presets |
|
50
|
|
|
*/ |
|
51
|
606 |
|
protected static function loadLocalePreset(string $locale = 'en', array $presets = ShortcodeInterface::DEFAULT_PRESETS): Dataset |
|
52
|
|
|
{ |
|
53
|
606 |
|
$throwables = []; |
|
54
|
606 |
|
$presets = \array_filter($presets); |
|
55
|
606 |
|
$remaining = $presets; |
|
56
|
606 |
|
while (\count($remaining) > 0) { |
|
57
|
606 |
|
$preset = \array_shift($remaining); |
|
58
|
|
|
try { |
|
59
|
606 |
|
return Dataset::unarchive(\sprintf('%s/%s/%s.gz', Dataset::DIRECTORY, $locale, $preset)); |
|
60
|
414 |
|
} catch (\Throwable $throwable) { |
|
61
|
414 |
|
$throwables[$preset] = $throwable; |
|
62
|
|
|
} |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
414 |
|
throw new LocalePresetException($locale, $throwables); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
81 |
|
public function getConfiguration(): ConfigurationInterface |
|
69
|
|
|
{ |
|
70
|
81 |
|
return $this->configuration; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @return AbstractToken[] |
|
75
|
|
|
*/ |
|
76
|
81 |
|
public function parse(string $input): array |
|
77
|
|
|
{ |
|
78
|
81 |
|
$this->lexer->setInput($input); |
|
79
|
81 |
|
$this->lexer->moveNext(); |
|
80
|
|
|
|
|
81
|
81 |
|
return $this->parseTokens(); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @return AbstractToken[] |
|
86
|
|
|
*/ |
|
87
|
81 |
|
protected function parseTokens(): array |
|
88
|
|
|
{ |
|
89
|
81 |
|
$tokens = []; |
|
90
|
81 |
|
while (true) { |
|
91
|
81 |
|
if (! $this->lexer->lookahead) { |
|
92
|
81 |
|
break; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
81 |
|
$this->lexer->moveNext(); |
|
96
|
|
|
|
|
97
|
81 |
|
$currentToken = \array_merge([ |
|
98
|
|
|
'type' => Lexer::T_TEXT, |
|
99
|
|
|
'value' => '', |
|
100
|
81 |
|
], $this->lexer->token ?? []); |
|
101
|
|
|
|
|
102
|
81 |
|
$method = self::TYPE_METHODS[$currentToken['type']] ?? null; |
|
103
|
81 |
|
if ($method && ($token = $this->$method($currentToken['value'])) instanceof AbstractToken) { |
|
104
|
81 |
|
$tokens[] = $token; |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
81 |
|
return $tokens; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
18 |
|
protected function parseEmoticon(string $value): ?Emoticon |
|
112
|
|
|
{ |
|
113
|
18 |
|
$emoji = $this->dataset->indexBy('emoticon')->offsetGet($value); |
|
114
|
|
|
|
|
115
|
|
|
// Clone the configuration here. This is necessary so it can be passed to tokens, |
|
116
|
|
|
// which may be rendered at a later time; when the configuration may have changed. |
|
117
|
18 |
|
return $emoji ? new Emoticon(clone $this->configuration, $value, $emoji) : null; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
15 |
|
protected function parseHtmlEntity(string $value): ?HtmlEntity |
|
121
|
|
|
{ |
|
122
|
15 |
|
$emoji = $this->dataset->indexBy('htmlEntity')->offsetGet($value); |
|
123
|
|
|
|
|
124
|
|
|
// Clone the configuration here. This is necessary so it can be passed to tokens, |
|
125
|
|
|
// which may be rendered at a later time; when the configuration may have changed. |
|
126
|
15 |
|
$configuration = clone $this->configuration; |
|
127
|
|
|
|
|
128
|
15 |
|
return $emoji ? new HtmlEntity($configuration, $value, $emoji) : null; |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
54 |
|
protected function parseShortcode(string $value): ?Shortcode |
|
132
|
|
|
{ |
|
133
|
54 |
|
$emoji = $this->dataset->indexBy('shortcodes')->offsetGet(\current(Normalize::shortcodes($value))); |
|
134
|
|
|
|
|
135
|
|
|
// Clone the configuration here. This is necessary so it can be passed to tokens, |
|
136
|
|
|
// which may be rendered at a later time; when the configuration may have changed. |
|
137
|
54 |
|
return $emoji ? new Shortcode(clone $this->configuration, $value, $emoji) : null; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
42 |
|
protected function parseUnicode(string $value): ?Unicode |
|
141
|
|
|
{ |
|
142
|
42 |
|
$emoji = $this->dataset->indexBy('emoji')->offsetGet($value) ?: $this->dataset->indexBy('text')->offsetGet($value); |
|
143
|
|
|
|
|
144
|
|
|
// Clone the configuration here. This is necessary so it can be passed to tokens, |
|
145
|
|
|
// which may be rendered at a later time; when the configuration may have changed. |
|
146
|
42 |
|
return $emoji ? new Unicode(clone $this->configuration, $value, $emoji) : null; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
66 |
|
protected function parseText(string $value): ?Text |
|
150
|
|
|
{ |
|
151
|
66 |
|
$text = ''; |
|
152
|
66 |
|
while (true) { |
|
153
|
66 |
|
$text .= $value; |
|
154
|
66 |
|
if ($this->lexer->lookahead === null || $this->lexer->lookahead['type'] !== Lexer::T_TEXT) { |
|
155
|
66 |
|
break; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
66 |
|
$value = $this->lexer->lookahead['value'] ?? ''; |
|
159
|
66 |
|
$this->lexer->moveNext(); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
66 |
|
return $text ? new Text($text) : null; |
|
163
|
|
|
} |
|
164
|
|
|
} |
|
165
|
|
|
|