|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of dispositif/wikibot application |
|
4
|
|
|
* 2019 : Philippe M. <[email protected]> |
|
5
|
|
|
* For the full copyright and MIT license information, please view the LICENSE file. |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Infrastructure; |
|
11
|
|
|
|
|
12
|
|
|
use App\Domain\CorpusInterface; |
|
13
|
|
|
use DomainException; |
|
14
|
|
|
use Exception; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @unused |
|
18
|
|
|
* Dirty todo refac with FileManager and league/flysystem |
|
19
|
|
|
* todo : deleteElementFromCorpus, setCorpusFromFilename ??? |
|
20
|
|
|
* Class CorpusAdapter. |
|
21
|
|
|
*/ |
|
22
|
|
|
class CorpusAdapter extends FileManager implements CorpusInterface |
|
23
|
|
|
{ |
|
24
|
|
|
private $storage = []; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* ugly memory // todo refac Generator. |
|
28
|
|
|
* |
|
29
|
|
|
* @param string $element |
|
30
|
|
|
* @param string $corpusName |
|
31
|
|
|
* |
|
32
|
|
|
* @return bool |
|
33
|
|
|
* @throws Exception |
|
34
|
|
|
*/ |
|
35
|
|
|
public function inCorpus(string $element, string $corpusName): bool |
|
36
|
|
|
{ |
|
37
|
|
|
// corpus as array variable |
|
38
|
|
|
if (isset($this->storage[$corpusName])) { |
|
39
|
|
|
$corpData = $this->storage[$corpusName]; |
|
40
|
|
|
return is_array($corpData) && in_array($element, $corpData); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
// corpus as text |
|
44
|
|
|
$filename = $this->getCorpusFilename($corpusName); |
|
45
|
|
|
|
|
46
|
|
|
return $this->isStringInCSV($filename, $element); |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
private function getCorpusFilename(string $corpusName) |
|
50
|
|
|
{ |
|
51
|
|
|
// Corpus as text file |
|
52
|
|
|
if ('firstname' === $corpusName) { |
|
53
|
|
|
return __DIR__.'/../Domain/resources/corpus_firstname.txt'; |
|
54
|
|
|
} |
|
55
|
|
|
if ('all-titles' === $corpusName) { |
|
56
|
|
|
return __DIR__.'/frwiki-latest-all-titles-in-ns0.txt'; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
throw new DomainException("corpus $corpusName not defined"); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
public function setCorpusInStorage(string $corpusName, array $arrayContent): void |
|
63
|
|
|
{ |
|
64
|
|
|
$this->storage[$corpusName] = $arrayContent; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* dirty TODO. |
|
69
|
|
|
* |
|
70
|
|
|
* @param string $corpusName |
|
71
|
|
|
* @param string $element |
|
72
|
|
|
* |
|
73
|
|
|
* @return bool |
|
74
|
|
|
* @throws Exception |
|
75
|
|
|
*/ |
|
76
|
|
|
public function addNewElementToCorpus(string $corpusName, string $element): bool |
|
77
|
|
|
{ |
|
78
|
|
|
if (empty($element)) { |
|
79
|
|
|
return false; |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
//corpus as variable |
|
83
|
|
|
if (isset($this->storage[$corpusName])) { |
|
84
|
|
|
return $this->addNewElementToMemoryCorpus($corpusName, $element); |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
// else : corpus as file |
|
88
|
|
|
return $this->addNewElementToFileCorpus($corpusName, $element); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
private function addNewElementToFileCorpus(string $corpusName, string $element): bool |
|
92
|
|
|
{ |
|
93
|
|
|
if (empty($element)) { |
|
94
|
|
|
return false; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
// strip "/" |
|
98
|
|
|
$sanitizeCorpusName = preg_replace('#[^0-9a-z_]#i', '', $corpusName); |
|
99
|
|
|
$filename = __DIR__.'/../Domain/resources/'.$sanitizeCorpusName.'.txt'; |
|
100
|
|
|
|
|
101
|
|
|
// hack: create file or not ? |
|
102
|
|
|
if (!file_exists($filename)) { |
|
103
|
|
|
$newFile = @fopen($filename, 'w'); |
|
104
|
|
|
if ($newFile !== false) { |
|
105
|
|
|
fclose($newFile); |
|
106
|
|
|
} |
|
107
|
|
|
if (!file_exists($filename)) { |
|
108
|
|
|
throw new Exception('corpus filename does not exist'.$filename); |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
// check if the element already in the corpus |
|
113
|
|
|
if ($this->isStringInCSV($filename, $element)) { |
|
114
|
|
|
return false; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
$write = file_put_contents( |
|
118
|
|
|
$filename, |
|
119
|
|
|
utf8_encode($element)."\n", |
|
120
|
|
|
FILE_APPEND | LOCK_EX |
|
121
|
|
|
); |
|
122
|
|
|
|
|
123
|
|
|
return (bool) $write; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
private function addNewElementToMemoryCorpus(string $corpusName, string $element): bool |
|
127
|
|
|
{ |
|
128
|
|
|
if (isset($this->storage[$corpusName]) && is_array($this->storage[$corpusName])) { |
|
129
|
|
|
if (!in_array($element, $this->storage[$corpusName])) { |
|
130
|
|
|
$this->storage[$corpusName][] = $element; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
return true; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
return false; |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
|
Dispositif /
Wikibot
