|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpSpellcheck\Spellchecker; |
|
6
|
|
|
|
|
7
|
|
|
use PhpSpellcheck\Exception\ProcessHasErrorOutputException; |
|
8
|
|
|
use PhpSpellcheck\Misspelling; |
|
9
|
|
|
use PhpSpellcheck\Utils\CommandLine; |
|
10
|
|
|
use PhpSpellcheck\Utils\IspellOutputParser; |
|
11
|
|
|
use PhpSpellcheck\Utils\ProcessRunner; |
|
12
|
|
|
use Symfony\Component\Process\Process; |
|
13
|
|
|
use Webmozart\Assert\Assert; |
|
14
|
|
|
|
|
15
|
|
|
class Ispell implements SpellcheckerInterface |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* @var string[]|null |
|
19
|
|
|
*/ |
|
20
|
|
|
private $supportedLanguages; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var CommandLine |
|
24
|
|
|
*/ |
|
25
|
|
|
private $ispellCommandLine; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var CommandLine|null |
|
29
|
|
|
*/ |
|
30
|
|
|
private $shellEntryPoint; |
|
31
|
|
|
|
|
32
|
5 |
|
public function __construct(CommandLine $ispellCommandLine, ?CommandLine $shellEntryPoint = null) |
|
33
|
|
|
{ |
|
34
|
5 |
|
$this->ispellCommandLine = $ispellCommandLine; |
|
35
|
5 |
|
$this->shellEntryPoint = $shellEntryPoint; |
|
36
|
5 |
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* {@inheritdoc} |
|
40
|
|
|
*/ |
|
41
|
3 |
|
public function check(string $text, array $languages = [], array $context = []): iterable |
|
42
|
|
|
{ |
|
43
|
3 |
|
Assert::greaterThan($languages, 1, 'Ispell spellchecker doesn\'t support multiple languages check'); |
|
44
|
|
|
|
|
45
|
3 |
|
$cmd = $this->ispellCommandLine->addArg('-a'); |
|
46
|
|
|
|
|
47
|
3 |
|
if (!empty($languages)) { |
|
48
|
2 |
|
$cmd = $cmd->addArgs(['-d', implode(',', $languages)]); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
3 |
|
$process = new Process($cmd->getArgs()); |
|
52
|
|
|
|
|
53
|
|
|
// Add prefix characters putting Ispell's type of spellcheckers in terse-mode, |
|
54
|
|
|
// ignoring correct words and thus speeding execution |
|
55
|
3 |
|
$process->setInput('!' . PHP_EOL . self::preprocessInputForPipeMode($text) . PHP_EOL . '%'); |
|
56
|
|
|
|
|
57
|
3 |
|
$output = ProcessRunner::run($process)->getOutput(); |
|
58
|
|
|
|
|
59
|
3 |
|
if ($process->getErrorOutput() !== '') { |
|
60
|
1 |
|
throw new ProcessHasErrorOutputException($process->getErrorOutput(), $text, $process->getCommandLine()); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
2 |
|
$misspellings = IspellOutputParser::parseMisspellings($output, $context); |
|
64
|
|
|
|
|
65
|
|
|
return self::postprocessMisspellings($misspellings); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
public function getCommandLine(): CommandLine |
|
69
|
|
|
{ |
|
70
|
|
|
return $this->ispellCommandLine; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
2 |
|
* {@inheritdoc} |
|
75
|
|
|
*/ |
|
76
|
|
|
public function getSupportedLanguages(): iterable |
|
77
|
|
|
{ |
|
78
|
2 |
|
if ($this->supportedLanguages === null) { |
|
79
|
2 |
|
$shellEntryPoint = $this->shellEntryPoint ?? new CommandLine([]); |
|
80
|
2 |
|
$whichCommand = clone $shellEntryPoint; |
|
81
|
2 |
|
$process = new Process( |
|
82
|
|
|
$whichCommand |
|
83
|
2 |
|
->addArg('which') |
|
84
|
2 |
|
->addArg('ispell') |
|
85
|
2 |
|
->getArgs() |
|
86
|
|
|
); |
|
87
|
2 |
|
$process->mustRun(); |
|
88
|
2 |
|
$binaryPath = trim($process->getOutput()); |
|
89
|
|
|
|
|
90
|
2 |
|
$lsCommand = clone $shellEntryPoint; |
|
91
|
2 |
|
$process = new Process( |
|
92
|
|
|
$lsCommand |
|
93
|
2 |
|
->addArg('ls') |
|
94
|
2 |
|
->addArg(\dirname($binaryPath, 2) . '/lib/ispell') |
|
95
|
2 |
|
->getArgs() |
|
96
|
|
|
); |
|
97
|
2 |
|
$process->mustRun(); |
|
98
|
|
|
|
|
99
|
2 |
|
$listOfFiles = trim($process->getOutput()); |
|
100
|
|
|
|
|
101
|
2 |
|
$this->supportedLanguages = []; |
|
102
|
2 |
|
foreach (explode(PHP_EOL, $listOfFiles) as $file) { |
|
103
|
2 |
|
if (strpos($file, '.aff', -4) === false) { |
|
104
|
1 |
|
continue; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
2 |
|
yield \Safe\substr($file, 0, -4); |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
2 |
|
|
|
111
|
|
|
return $this->supportedLanguages; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
public static function create(?string $ispellCommandLineAsString): self |
|
115
|
2 |
|
{ |
|
116
|
|
|
return new self(new CommandLine($ispellCommandLineAsString ?? 'ispell')); |
|
117
|
|
|
} |
|
118
|
1 |
|
|
|
119
|
|
|
/** |
|
120
|
1 |
|
* Preprocess the source text so that aspell/ispell pipe mode instruction are ignored. |
|
121
|
|
|
* |
|
122
|
|
|
* in pipe mode some special characters at the beginning of the line are instructions for aspell/ispell |
|
123
|
|
|
* see http://aspell.net/man-html/Through-A-Pipe.html#Through-A-Pipe |
|
124
|
|
|
* users put these chars innocently at the beginning of lines |
|
125
|
|
|
* we must tell the spellchecker to not interpret them using the ^ symbol |
|
126
|
|
|
* |
|
127
|
|
|
* @param string $text the text to preprocess |
|
128
|
|
|
* |
|
129
|
|
|
* @return string the result of the preprocessing |
|
130
|
|
|
*/ |
|
131
|
|
|
public static function preprocessInputForPipeMode(string $text): string |
|
132
|
|
|
{ |
|
133
|
|
|
$lines = explode("\n", $text); |
|
134
|
|
|
|
|
135
|
|
|
$prefixedLines = array_map( |
|
136
|
|
|
function ($line) { |
|
137
|
|
|
return \strlen($line) === 0 ? "$line" : "^$line"; |
|
138
|
|
|
}, |
|
139
|
|
|
$lines |
|
140
|
|
|
); |
|
141
|
|
|
|
|
142
|
|
|
$preprocessedText = implode( |
|
143
|
|
|
"\n", |
|
144
|
|
|
$prefixedLines |
|
145
|
|
|
); |
|
146
|
|
|
|
|
147
|
|
|
return $preprocessedText; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/** |
|
151
|
|
|
* Adapts misspellings to compensate the additional caret added in self::preprocessInputForPipeMode. |
|
152
|
|
|
* |
|
153
|
|
|
* @param iterable<Misspelling> $misspellings |
|
154
|
|
|
* |
|
155
|
|
|
* @return iterable<Misspelling> copies of input misspells with corrected offset |
|
156
|
|
|
*/ |
|
157
|
|
|
public static function postprocessMisspellings(iterable $misspellings): iterable |
|
158
|
|
|
{ |
|
159
|
|
|
foreach ($misspellings as $m) { |
|
160
|
|
|
$offset = $m->getOffset(); |
|
161
|
|
|
if ($offset !== null) { |
|
162
|
|
|
$offset--; |
|
163
|
|
|
} |
|
164
|
|
|
yield new Misspelling( |
|
165
|
|
|
$m->getWord(), |
|
166
|
|
|
$offset, |
|
167
|
|
|
$m->getLineNumber(), |
|
168
|
|
|
$m->getSuggestions(), |
|
169
|
|
|
$m->getContext() |
|
170
|
|
|
); |
|
171
|
|
|
} |
|
172
|
|
|
} |
|
173
|
|
|
} |
|
174
|
|
|
|
tigitz /
php-spellchecker
