|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpSpellcheck\Spellchecker; |
|
6
|
|
|
|
|
7
|
|
|
use PhpSpellcheck\Exception\RuntimeException; |
|
8
|
|
|
use PhpSpellcheck\Misspelling; |
|
9
|
|
|
use PhpSpellcheck\MisspellingInterface; |
|
10
|
|
|
use Webmozart\Assert\Assert; |
|
11
|
|
|
|
|
12
|
|
|
class PHPPspell implements SpellcheckerInterface |
|
13
|
|
|
{ |
|
14
|
|
|
/** |
|
15
|
|
|
* @var int |
|
16
|
|
|
*/ |
|
17
|
|
|
private $mode; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var int |
|
21
|
|
|
*/ |
|
22
|
|
|
private $numberOfCharactersLowerLimit; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var Aspell |
|
26
|
|
|
*/ |
|
27
|
|
|
private $aspell; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @see http://php.net/manual/en/function.pspell-config-mode.php |
|
31
|
|
|
* @see http://php.net/manual/en/function.pspell-config-ignore.php |
|
32
|
3 |
|
* |
|
33
|
|
|
* @param int|null $mode the mode parameter is the mode in which the spellchecker will work |
|
34
|
3 |
|
* @param int $numberOfCharactersLowerLimit Words less than n characters will be skipped |
|
35
|
|
|
* @param Aspell|null $aspell Aspell spellchecker that pspell extension is using underneath. Used to help retrieve supported languages |
|
36
|
|
|
*/ |
|
37
|
|
|
public function __construct( |
|
38
|
3 |
|
?int $mode = null, |
|
39
|
|
|
int $numberOfCharactersLowerLimit = 0, |
|
40
|
3 |
|
Aspell $aspell = null |
|
41
|
3 |
|
) { |
|
42
|
3 |
|
if (!\extension_loaded('pspell')) { |
|
43
|
|
|
throw new RuntimeException('Pspell extension must be loaded to use the PHPPspell spellchecker'); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
if ($mode === null) { |
|
47
|
2 |
|
$mode = PSPELL_FAST; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
Assert::greaterThanEq($numberOfCharactersLowerLimit, 0); |
|
51
|
|
|
|
|
52
|
|
|
$this->mode = $mode; |
|
53
|
2 |
|
$this->numberOfCharactersLowerLimit = $numberOfCharactersLowerLimit; |
|
54
|
2 |
|
$this->aspell = $aspell ?? Aspell::create(); |
|
55
|
|
|
} |
|
56
|
1 |
|
|
|
57
|
1 |
|
/** |
|
58
|
1 |
|
* {@inheritdoc} |
|
59
|
1 |
|
*/ |
|
60
|
|
|
public function check( |
|
61
|
1 |
|
string $text, |
|
62
|
1 |
|
array $languages, |
|
63
|
1 |
|
array $context |
|
64
|
1 |
|
): iterable { |
|
65
|
1 |
|
if (PHP_VERSION_ID < 80100) { |
|
66
|
1 |
|
return $this->checkBefore81($text, $languages, $context); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
return $this->checkAfter81($text, $languages, $context); |
|
70
|
1 |
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* {@inheritdoc} |
|
74
|
|
|
*/ |
|
75
|
1 |
|
public function getSupportedLanguages(): iterable |
|
76
|
|
|
{ |
|
77
|
1 |
|
return $this->aspell->getSupportedLanguages(); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @param array<mixed> $context |
|
82
|
|
|
* @param array<string> $languages |
|
83
|
|
|
* |
|
84
|
|
|
* @return iterable<MisspellingInterface> |
|
85
|
|
|
*/ |
|
86
|
|
|
private function checkBefore81( |
|
87
|
|
|
string $text, |
|
88
|
|
|
array $languages, |
|
89
|
|
|
array $context |
|
90
|
|
|
): iterable { |
|
91
|
|
|
Assert::count($languages, 1, 'PHPPspell spellchecker doesn\'t support multi-language check'); |
|
92
|
|
|
|
|
93
|
|
|
$chosenLanguage = current($languages); |
|
94
|
|
|
$pspellConfig = \Safe\pspell_config_create(current($languages)); |
|
95
|
|
|
\Safe\pspell_config_mode($pspellConfig, $this->mode); |
|
96
|
|
|
\Safe\pspell_config_ignore($pspellConfig, $this->numberOfCharactersLowerLimit); |
|
97
|
|
|
$dictionary = \Safe\pspell_new_config($pspellConfig); |
|
98
|
|
|
|
|
99
|
|
|
$lines = explode(PHP_EOL, $text); |
|
100
|
|
|
|
|
101
|
|
|
/** @var string $line */ |
|
102
|
|
|
foreach ($lines as $lineNumber => $line) { |
|
103
|
|
|
$words = explode(' ', \Safe\preg_replace("/(?!['’-])(\p{P}|\+|--)/u", '', $line)); |
|
104
|
|
|
foreach ($words as $word) { |
|
105
|
|
|
if (!pspell_check($dictionary, $word)) { |
|
106
|
|
|
$suggestions = pspell_suggest($dictionary, $word); |
|
107
|
|
|
Assert::isArray( |
|
108
|
|
|
$suggestions, |
|
109
|
|
|
\Safe\sprintf('pspell_suggest method failed with language "%s" and word "%s"', $chosenLanguage, $word) |
|
110
|
|
|
); |
|
111
|
|
|
|
|
112
|
|
|
yield new Misspelling($word, 0, $lineNumber + 1, $suggestions, $context); |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* @param array<mixed> $context |
|
120
|
|
|
* @param array<string> $languages |
|
121
|
|
|
* |
|
122
|
|
|
* @return iterable<MisspellingInterface> |
|
123
|
|
|
*/ |
|
124
|
|
|
private function checkAfter81( |
|
125
|
|
|
string $text, |
|
126
|
|
|
array $languages, |
|
127
|
|
|
array $context |
|
128
|
|
|
): iterable { |
|
129
|
|
|
Assert::count($languages, 1, 'PHPPspell spellchecker doesn\'t support multi-language check'); |
|
130
|
|
|
|
|
131
|
|
|
$chosenLanguage = current($languages); |
|
132
|
|
|
$pspellConfig = pspell_config_create($chosenLanguage); |
|
133
|
|
|
pspell_config_mode($pspellConfig, $this->mode); |
|
134
|
|
|
pspell_config_ignore($pspellConfig, $this->numberOfCharactersLowerLimit); |
|
135
|
|
|
$dictionary = pspell_new_config($pspellConfig); |
|
136
|
|
|
|
|
137
|
|
|
$lines = explode(PHP_EOL, $text); |
|
138
|
|
|
|
|
139
|
|
|
/** @var string $line */ |
|
140
|
|
|
foreach ($lines as $lineNumber => $line) { |
|
141
|
|
|
$words = explode(' ', \Safe\preg_replace("/(?!['’-])(\p{P}|\+|--)/u", '', $line)); |
|
142
|
|
|
foreach ($words as $word) { |
|
143
|
|
|
if (!pspell_check($dictionary, $word)) { |
|
144
|
|
|
$suggestions = pspell_suggest($dictionary, $word); |
|
145
|
|
|
Assert::isArray( |
|
146
|
|
|
$suggestions, |
|
147
|
|
|
\Safe\sprintf('pspell_suggest method failed with language "%s" and word "%s"', $chosenLanguage, $word) |
|
148
|
|
|
); |
|
149
|
|
|
|
|
150
|
|
|
yield new Misspelling($word, 0, $lineNumber + 1, $suggestions, $context); |
|
151
|
|
|
} |
|
152
|
|
|
} |
|
153
|
|
|
} |
|
154
|
|
|
} |
|
155
|
|
|
} |
|
156
|
|
|
|
tigitz /
php-spellchecker
