|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace UnicornFail\Emoji; |
|
6
|
|
|
|
|
7
|
|
|
use Doctrine\Common\Lexer\AbstractLexer; |
|
8
|
|
|
use UnicornFail\Emoji\Emojibase\RegexInterface; |
|
9
|
|
|
|
|
10
|
|
|
class Lexer extends AbstractLexer implements RegexInterface |
|
11
|
|
|
{ |
|
12
|
|
|
public const T_TEXT = 0; |
|
13
|
|
|
|
|
14
|
|
|
public const T_EMOTICON = 1; |
|
15
|
|
|
|
|
16
|
|
|
public const T_HTML_ENTITY = 2; |
|
17
|
|
|
|
|
18
|
|
|
public const T_SHORTCODE = 3; |
|
19
|
|
|
|
|
20
|
|
|
public const T_UNICODE = 4; |
|
21
|
|
|
|
|
22
|
|
|
public const TYPES = [ |
|
23
|
|
|
self::T_TEXT => 'text', |
|
24
|
|
|
self::T_EMOTICON => 'emoticon', |
|
25
|
|
|
self::T_HTML_ENTITY => 'htmlEntity', |
|
26
|
|
|
self::T_SHORTCODE => 'shortcode', |
|
27
|
|
|
self::T_UNICODE => 'unicode', |
|
28
|
|
|
]; |
|
29
|
|
|
|
|
30
|
|
|
/** @var Configuration|ConfigurationInterface */ |
|
31
|
|
|
private $configuration; |
|
32
|
|
|
|
|
33
|
192 |
|
public function __construct(ConfigurationInterface $configuration) |
|
34
|
|
|
{ |
|
35
|
192 |
|
$this->configuration = $configuration; |
|
36
|
192 |
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* {@inheritDoc} |
|
40
|
|
|
* |
|
41
|
|
|
* @return string[] |
|
42
|
|
|
*/ |
|
43
|
81 |
|
protected function getCatchablePatterns() |
|
44
|
|
|
{ |
|
45
|
|
|
$patterns = [ |
|
46
|
81 |
|
self::CODEPOINT_EMOJI_LOOSE_REGEX, |
|
47
|
81 |
|
self::HTML_ENTITY_REGEX, |
|
48
|
81 |
|
$this->configuration->get('native') ? self::SHORTCODE_NATIVE_REGEX : self::SHORTCODE_REGEX, |
|
49
|
|
|
]; |
|
50
|
81 |
|
if ($this->configuration->get('convertEmoticons')) { |
|
51
|
81 |
|
$patterns[] = self::EMOTICON_REGEX; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
// Some regex patterns from the constants include the delimiter and modifiers. Because the |
|
55
|
|
|
// lexer joins these expressions together as an OR group (|), they must be removed. |
|
56
|
81 |
|
foreach ($patterns as &$pattern) { |
|
57
|
81 |
|
$pattern = \trim(\rtrim($pattern, 'imsxeADSUXJu'), '/'); |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
81 |
|
return $patterns; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* {@inheritDoc} |
|
65
|
|
|
* |
|
66
|
|
|
* @return string[] |
|
67
|
|
|
*/ |
|
68
|
81 |
|
protected function getNonCatchablePatterns() |
|
69
|
|
|
{ |
|
70
|
81 |
|
return []; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* {@inheritDoc} |
|
75
|
|
|
* |
|
76
|
|
|
* @return int |
|
77
|
|
|
* |
|
78
|
|
|
* @noinspection PhpParameterByRefIsNotUsedAsReferenceInspection |
|
79
|
|
|
*/ |
|
80
|
81 |
|
protected function getType(&$value) |
|
81
|
|
|
{ |
|
82
|
81 |
|
if (\preg_match(self::HTML_ENTITY_REGEX, $value)) { |
|
83
|
15 |
|
return self::T_HTML_ENTITY; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
// @phpstan-ignore-next-line |
|
87
|
81 |
|
if (\preg_match($this->configuration->get('native') ? self::SHORTCODE_NATIVE_REGEX : self::SHORTCODE_REGEX, $value)) { |
|
88
|
54 |
|
return self::T_SHORTCODE; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
69 |
|
if (\preg_match(self::EMOTICON_REGEX, $value)) { |
|
92
|
18 |
|
return self::T_EMOTICON; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
69 |
|
if (\preg_match(self::CODEPOINT_EMOJI_LOOSE_REGEX, $value)) { |
|
96
|
42 |
|
return self::T_UNICODE; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
66 |
|
return self::T_TEXT; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
unicorn-fail /
emoji
