|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpSpellcheck\Spellchecker; |
|
6
|
|
|
|
|
7
|
|
|
use PhpSpellcheck\Exception\LogicException; |
|
8
|
|
|
use PhpSpellcheck\Exception\RuntimeException; |
|
9
|
|
|
use PhpSpellcheck\Misspelling; |
|
10
|
|
|
use PhpSpellcheck\Utils\TextEncoding; |
|
11
|
|
|
use Webmozart\Assert\Assert; |
|
12
|
|
|
|
|
13
|
|
|
class PHPPspell implements SpellcheckerInterface |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* @var int |
|
17
|
|
|
*/ |
|
18
|
|
|
private $mode; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var int |
|
22
|
|
|
*/ |
|
23
|
|
|
private $numberOfCharactersLowerLimit; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @link http://php.net/manual/en/function.pspell-config-mode.php |
|
27
|
|
|
* @link http://php.net/manual/en/function.pspell-config-ignore.php |
|
28
|
|
|
* |
|
29
|
|
|
* @param int $mode The mode parameter is the mode in which spellchecker will work. |
|
30
|
|
|
* @param int $numberOfCharactersLowerLimit Words less than n characters will be skipped |
|
31
|
|
|
*/ |
|
32
|
3 |
|
public function __construct(int $mode, int $numberOfCharactersLowerLimit = 0) |
|
33
|
|
|
{ |
|
34
|
3 |
|
if (!extension_loaded('pspell')) { |
|
35
|
|
|
throw new RuntimeException('Pspell extension must be loaded to use the PHPPspell spellchecker'); |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
3 |
|
Assert::greaterThanEq($numberOfCharactersLowerLimit, 0); |
|
39
|
|
|
|
|
40
|
3 |
|
$this->mode = $mode; |
|
41
|
3 |
|
$this->numberOfCharactersLowerLimit = $numberOfCharactersLowerLimit; |
|
42
|
3 |
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* {@inheritDoc} |
|
46
|
|
|
*/ |
|
47
|
2 |
|
public function check( |
|
48
|
|
|
string $text, |
|
49
|
|
|
array $languages = [], |
|
50
|
|
|
array $context = [], |
|
51
|
|
|
?string $encoding = TextEncoding::UTF8 |
|
52
|
|
|
): iterable { |
|
53
|
2 |
|
Assert::count($languages, 1, 'PHPPspell spellchecker doesn\'t support multiple languages check'); |
|
54
|
2 |
|
Assert::notNull($encoding, 'PHPPspell requires the encoding to be defined'); |
|
55
|
|
|
|
|
56
|
1 |
|
$pspellConfig = \Safe\pspell_config_create(current($languages), '', '', $encoding); |
|
57
|
1 |
|
\Safe\pspell_config_mode($pspellConfig, $this->mode); |
|
58
|
1 |
|
\Safe\pspell_config_ignore($pspellConfig, $this->numberOfCharactersLowerLimit); |
|
59
|
1 |
|
$dictionary = \Safe\pspell_new_config($pspellConfig); |
|
60
|
|
|
|
|
61
|
1 |
|
$lines = explode(PHP_EOL, $text); |
|
62
|
1 |
|
|
|
63
|
1 |
|
/** @var string $line */ |
|
64
|
1 |
|
foreach ($lines as $lineNumber => $line) { |
|
65
|
1 |
|
$words = explode(' ', \Safe\preg_replace("/(?!['’-])(\p{P}|\+|--)/u", '', $line)); |
|
66
|
1 |
|
foreach ($words as $key => $word) { |
|
67
|
|
|
if (!pspell_check($dictionary, $word)) { |
|
68
|
|
|
$suggestions = pspell_suggest($dictionary, $word); |
|
69
|
|
|
yield new Misspelling($word, 0, $lineNumber + 1, $suggestions, $context); |
|
70
|
1 |
|
} |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
1 |
|
/** |
|
76
|
|
|
* {@inheritDoc} |
|
77
|
1 |
|
*/ |
|
78
|
|
|
public function getSupportedLanguages(): iterable |
|
79
|
|
|
{ |
|
80
|
|
|
throw new LogicException('Retrieving supported dictionaries for PHPPspell spellchecker is not supported yet'); |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
tigitz /
php-spellchecker
