SanitizerFactory::newTextCatLanguageDetector() - Code Metrics - onoi/tesa - Measure and Improve Code Quality continuously with Scrutinizer

SanitizerFactory::newTextCatLanguageDetector() A
last analyzed 2018-01-27 17:32 UTC

↳ Parent: SanitizerFactory

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	2
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	3
ccs	2
cts	2
cp	1
rs	10
c	0
b	0
f	0
cc	1
eloc	2
nc	1
nop	0
crap	1

<?php

namespace Onoi\Tesa;

use Onoi\Tesa\StopwordAnalyzer\StopwordAnalyzer;
use Onoi\Tesa\StopwordAnalyzer\NullStopwordAnalyzer;
use Onoi\Tesa\StopwordAnalyzer\CdbStopwordAnalyzer;
use Onoi\Tesa\StopwordAnalyzer\ArrayStopwordAnalyzer;
use Onoi\Tesa\Synonymizer\Synonymizer;
use Onoi\Tesa\Synonymizer\NullSynonymizer;
use Onoi\Tesa\LanguageDetector\NullLanguageDetector;
use Onoi\Tesa\LanguageDetector\TextCatLanguageDetector;
use Onoi\Tesa\Tokenizer\CJKSimpleCharacterRegExTokenizer;
use Onoi\Tesa\Tokenizer\Tokenizer;
use Onoi\Tesa\Tokenizer\GenericRegExTokenizer;
use Onoi\Tesa\Tokenizer\JaCompoundGroupTokenizer;
use Onoi\Tesa\Tokenizer\IcuWordBoundaryTokenizer;
use Onoi\Tesa\Tokenizer\NGramTokenizer;
use Onoi\Tesa\Tokenizer\JaTinySegmenterTokenizer;
use Onoi\Tesa\Tokenizer\PunctuationRegExTokenizer;

/**
 * @license GNU GPL v2+
 * @since 0.1
 *
 * @author mwjames
 */
class SanitizerFactory {

	/**
	 * @since 0.1
	 *
	 * @return Sanitizer
	 */
	public function newSanitizer( $text = '' ) {
		return new Sanitizer( $text );
	}

	/* StopwordAnalyzer */

	/**
	 * @since 0.1
	 *
	 * @param string|null $languageCode
	 *
	 * @return StopwordAnalyzer
	 */
	public function newStopwordAnalyzerByLanguage( $languageCode = null ) {

		if ( $languageCode === null ) {
			return $this->newNullStopwordAnalyzer();
		}

		$cdbStopwordAnalyzer = $this->newCdbStopwordAnalyzer(
			$languageCode
		);

		return $cdbStopwordAnalyzer->isAvailable() ? $cdbStopwordAnalyzer : $this->newNullStopwordAnalyzer();;
	}

	/**
	 * @since 0.1
	 *
	 * @return StopwordAnalyzer
	 */
	public function newCdbStopwordAnalyzer( $languageCode = null ) {
		return new CdbStopwordAnalyzer( CdbStopwordAnalyzer::getTargetByLanguage( $languageCode ) );
	}

	/**
	 * @since 0.1
	 *
	 * @param array $stopwords;
/**
 * @param array $germany
 * @param array $ireland
 */
function finale($germany, $island) {
    return "2:1";
}
	 *
	 * @return StopwordAnalyzer
	 */
	public function newArrayStopwordAnalyzer( array $stopwords = array() ) {
		return new ArrayStopwordAnalyzer( $stopwords );
	}

	/**
	 * @since 0.1
	 *
	 * @return StopwordAnalyzer
	 */
	public function newNullStopwordAnalyzer() {
		return new NullStopwordAnalyzer();
	}

	/**
	 * @since 0.1
	 *
	 * @param string|null $languageCode
	 *
	 * @return Synonymizer
	 */
	public function newSynonymizerByLanguage( $languageCode = null ) {

		if ( $languageCode === null ) {
			return $this->newNullSynonymizer();
		}

		return $this->newNullSynonymizer();;
	}

	/* Synonymizer */

	/**
	 * @since 0.1
	 *
	 * @return Synonymizer
	 */
	public function newNullSynonymizer() {
		return new NullSynonymizer();
	}

	/* LanguageDetector */

	/**
	 * @since 0.1
	 *
	 * @return NullLanguageDetector
	 */
	public function newNullLanguageDetector() {
		return new NullLanguageDetector();
	}

	/**
	 * @since 0.1
	 *
	 * @return TextCatLanguageDetector
	 */
	public function newTextCatLanguageDetector() {
		return new TextCatLanguageDetector();
	}

	/* Tokenizer */

	/**
	 * @since 0.1
	 *
	 * @param string $text
	 * @param string|null $languageCode
	 *
	 * @return Tokenizer
	 */
	public function newPreferredTokenizerByLanguage( $text, $languageCode = null ) {

		$tokenizer = $this->newIcuWordBoundaryTokenizer();

		if ( !$tokenizer->isAvailable() && CharacterExaminer::contains( CharacterExaminer::CJK_UNIFIED, $text ) ) {
			return $this->newCJKMatchableTokenizer( $text );
		} elseif( !$tokenizer->isAvailable() ) {
			return $this->newGenericRegExTokenizer( $tokenizer );
		}

		$tokenizer->setLocale( $languageCode );

		$tokenizer->setWordTokenizerAttribute(

			!CharacterExaminer::contains( CharacterExaminer::CJK_UNIFIED, $text )
		);

		return $this->newGenericRegExTokenizer( $tokenizer );
	}

	/**
	 * @since 0.1
	 *
	 * @param string $text
	 *
	 * @return Tokenizer
	 */
	public function newCJKMatchableTokenizer( $text ) {

		$tokenizer = null;

		if ( CharacterExaminer::contains( CharacterExaminer::HIRAGANA_KATAKANA, $text ) ) {
			$tokenizer = $this->newJaTinySegmenterTokenizer();
		} else {
			$tokenizer = $this->newNGramTokenizer( $tokenizer );
		}

		$tokenizer = $this->newCJKSimpleCharacterRegExTokenizer( $tokenizer );

		return $this->newGenericRegExTokenizer( $tokenizer );
	}

	/**
	 * @since 0.1
	 *
	 * @param Tokenizer|null $tokenizer
	 *
	 * @return Tokenizer
	 */
	public function newIcuWordBoundaryTokenizer( Tokenizer $tokenizer = null ) {
		return new IcuWordBoundaryTokenizer( $tokenizer );
	}

	/**
	 * @since 0.1
	 *
	 * @param Tokenizer|null $tokenizer
	 *
	 * @return Tokenizer
	 */
	public function newGenericRegExTokenizer( Tokenizer $tokenizer = null ) {
		return new GenericRegExTokenizer( $tokenizer );
	}

	/**
	 * @since 0.1
	 *
	 * @param Tokenizer|null $tokenizer
	 *
	 * @return Tokenizer
	 */
	public function newPunctuationRegExTokenizer( Tokenizer $tokenizer = null ) {
		return new PunctuationRegExTokenizer( $tokenizer );
	}

	/**
	 * @since 0.1
	 *
	 * @return Tokenizer
	 */
	public function newJaCompoundGroupTokenizer( Tokenizer $tokinizer = null ) {
		return new JaCompoundGroupTokenizer( $tokinizer );
	}

	/**
	 * @since 0.1
	 *
	 * @return Tokenizer
	 */
	public function newJaTinySegmenterTokenizer( Tokenizer $tokinizer = null ) {
		return new JaTinySegmenterTokenizer( $tokinizer );
	}

	/**
	 * @since 0.1
	 *
	 * @return Tokenizer
	 */
	public function newCJKSimpleCharacterRegExTokenizer( Tokenizer $tokinizer = null ) {
		return new CJKSimpleCharacterRegExTokenizer( $tokinizer );
	}

	/**
	 * @since 0.1
	 *
	 * @return Tokenizer
	 */
	public function newNGramTokenizer( Tokenizer $tokinizer = null, $ngram = 2 ) {
		return new NGramTokenizer( $tokinizer, $ngram );
	}

}


1		<?php
2
3		namespace Onoi\Tesa;
4
5		use Onoi\Tesa\StopwordAnalyzer\StopwordAnalyzer;
6		use Onoi\Tesa\StopwordAnalyzer\NullStopwordAnalyzer;
7		use Onoi\Tesa\StopwordAnalyzer\CdbStopwordAnalyzer;
8		use Onoi\Tesa\StopwordAnalyzer\ArrayStopwordAnalyzer;
9		use Onoi\Tesa\Synonymizer\Synonymizer;
10		use Onoi\Tesa\Synonymizer\NullSynonymizer;
11		use Onoi\Tesa\LanguageDetector\NullLanguageDetector;
12		use Onoi\Tesa\LanguageDetector\TextCatLanguageDetector;
13		use Onoi\Tesa\Tokenizer\CJKSimpleCharacterRegExTokenizer;
14		use Onoi\Tesa\Tokenizer\Tokenizer;
15		use Onoi\Tesa\Tokenizer\GenericRegExTokenizer;
16		use Onoi\Tesa\Tokenizer\JaCompoundGroupTokenizer;
17		use Onoi\Tesa\Tokenizer\IcuWordBoundaryTokenizer;
18		use Onoi\Tesa\Tokenizer\NGramTokenizer;
19		use Onoi\Tesa\Tokenizer\JaTinySegmenterTokenizer;
20		use Onoi\Tesa\Tokenizer\PunctuationRegExTokenizer;
21
22		/**
23		* @license GNU GPL v2+
24		* @since 0.1
25		*
26		* @author mwjames
27		*/
28		class SanitizerFactory {
29
30		/**
31		* @since 0.1
32		*
33		* @return Sanitizer
34		*/
35	5	public function newSanitizer( $text = '' ) {
36	5	return new Sanitizer( $text );
37	1	}
38
39		/* StopwordAnalyzer */
40
41		/**
42		* @since 0.1
43		*
44		* @param string\|null $languageCode
45		*
46		* @return StopwordAnalyzer
47		*/
48	2	public function newStopwordAnalyzerByLanguage( $languageCode = null ) {
49
50	2	if ( $languageCode === null ) {
51	1	return $this->newNullStopwordAnalyzer();
52		}
53
54	1	$cdbStopwordAnalyzer = $this->newCdbStopwordAnalyzer(
55		$languageCode
56	1	);
57
58	1	return $cdbStopwordAnalyzer->isAvailable() ? $cdbStopwordAnalyzer : $this->newNullStopwordAnalyzer();;
59		}
60
61		/**
62		* @since 0.1
63		*
64		* @return StopwordAnalyzer
65		*/
66	6	public function newCdbStopwordAnalyzer( $languageCode = null ) {
67	6	return new CdbStopwordAnalyzer( CdbStopwordAnalyzer::getTargetByLanguage( $languageCode ) );
68		}
69
70		/**
71		* @since 0.1
72		*
73		* @param array $stopwords;
		0 ignored issues – show Documentation introduced 2016-08-02 14:35 UTC by Report Bug Copy Issue Report There is no parameter named `$stopwords;`. Did you maybe mean `$stopwords`? This check looks for PHPDoc comments describing methods or function parameters that do not exist on the corresponding method or function. It has, however, found a similar but not annotated parameter which might be a good fit. Consider the following example. The parameter `$ireland` is not defined by the method `finale(...)`. /** * @param array $germany * @param array $ireland */ function finale($germany, $island) { return "2:1"; } The most likely cause is that the parameter was changed, but the annotation was not. Loading history...
74		*
75		* @return StopwordAnalyzer
76		*/
77	1	public function newArrayStopwordAnalyzer( array $stopwords = array() ) {
78	1	return new ArrayStopwordAnalyzer( $stopwords );
79		}
80
81		/**
82		* @since 0.1
83		*
84		* @return StopwordAnalyzer
85		*/
86	2	public function newNullStopwordAnalyzer() {
87	2	return new NullStopwordAnalyzer();
88		}
89
90		/**
91		* @since 0.1
92		*
93		* @param string\|null $languageCode
94		*
95		* @return Synonymizer
96		*/
97	1	public function newSynonymizerByLanguage( $languageCode = null ) {
98
99	1	if ( $languageCode === null ) {
100	1	return $this->newNullSynonymizer();
101		}
102
103		return $this->newNullSynonymizer();;
104		}
105
106		/* Synonymizer */
107
108		/**
109		* @since 0.1
110		*
111		* @return Synonymizer
112		*/
113	6	public function newNullSynonymizer() {
114	6	return new NullSynonymizer();
115		}
116
117		/* LanguageDetector */
118
119		/**
120		* @since 0.1
121		*
122		* @return NullLanguageDetector
123		*/
124	1	public function newNullLanguageDetector() {
125	1	return new NullLanguageDetector();
126		}
127
128		/**
129		* @since 0.1
130		*
131		* @return TextCatLanguageDetector
132		*/
133	1	public function newTextCatLanguageDetector() {
134	1	return new TextCatLanguageDetector();
135		}
136
137		/* Tokenizer */
138
139		/**
140		* @since 0.1
141		*
142		* @param string $text
143		* @param string\|null $languageCode
144		*
145		* @return Tokenizer
146		*/
147	1	public function newPreferredTokenizerByLanguage( $text, $languageCode = null ) {
148
149	1	$tokenizer = $this->newIcuWordBoundaryTokenizer();
150
151	1	if ( !$tokenizer->isAvailable() && CharacterExaminer::contains( CharacterExaminer::CJK_UNIFIED, $text ) ) {
152		return $this->newCJKMatchableTokenizer( $text );
153	1	} elseif( !$tokenizer->isAvailable() ) {
154		return $this->newGenericRegExTokenizer( $tokenizer );
155		}
156
157	1	$tokenizer->setLocale( $languageCode );
158
159	1	$tokenizer->setWordTokenizerAttribute(
		0 ignored issues – show Bug introduced 2016-08-02 14:35 UTC by Report Bug Copy Issue Report The method `setWordTokenizerAttribute()` does not exist on `Onoi\Tesa\Tokenizer\Tokenizer`. Did you maybe mean `tokenize()`? This check marks calls to methods that do not seem to exist on an object. This is most likely the result of a method being renamed without all references to it being renamed likewise. Loading history...
160	1	!CharacterExaminer::contains( CharacterExaminer::CJK_UNIFIED, $text )
161	1	);
162
163	1	return $this->newGenericRegExTokenizer( $tokenizer );
164		}
165
166		/**
167		* @since 0.1
168		*
169		* @param string $text
170		*
171		* @return Tokenizer
172		*/
173	1	public function newCJKMatchableTokenizer( $text ) {
174
175	1	$tokenizer = null;
176
177	1	if ( CharacterExaminer::contains( CharacterExaminer::HIRAGANA_KATAKANA, $text ) ) {
178	1	$tokenizer = $this->newJaTinySegmenterTokenizer();
179	1	} else {
180	1	$tokenizer = $this->newNGramTokenizer( $tokenizer );
181		}
182
183	1	$tokenizer = $this->newCJKSimpleCharacterRegExTokenizer( $tokenizer );
184
185	1	return $this->newGenericRegExTokenizer( $tokenizer );
186		}
187
188		/**
189		* @since 0.1
190		*
191		* @param Tokenizer\|null $tokenizer
192		*
193		* @return Tokenizer
194		*/
195	2	public function newIcuWordBoundaryTokenizer( Tokenizer $tokenizer = null ) {
196	2	return new IcuWordBoundaryTokenizer( $tokenizer );
197		}
198
199		/**
200		* @since 0.1
201		*
202		* @param Tokenizer\|null $tokenizer
203		*
204		* @return Tokenizer
205		*/
206	8	public function newGenericRegExTokenizer( Tokenizer $tokenizer = null ) {
207	8	return new GenericRegExTokenizer( $tokenizer );
208		}
209
210		/**
211		* @since 0.1
212		*
213		* @param Tokenizer\|null $tokenizer
214		*
215		* @return Tokenizer
216		*/
217	2	public function newPunctuationRegExTokenizer( Tokenizer $tokenizer = null ) {
218	2	return new PunctuationRegExTokenizer( $tokenizer );
219		}
220
221		/**
222		* @since 0.1
223		*
224		* @return Tokenizer
225		*/
226	1	public function newJaCompoundGroupTokenizer( Tokenizer $tokinizer = null ) {
227	1	return new JaCompoundGroupTokenizer( $tokinizer );
228		}
229
230		/**
231		* @since 0.1
232		*
233		* @return Tokenizer
234		*/
235	3	public function newJaTinySegmenterTokenizer( Tokenizer $tokinizer = null ) {
236	3	return new JaTinySegmenterTokenizer( $tokinizer );
237		}
238
239		/**
240		* @since 0.1
241		*
242		* @return Tokenizer
243		*/
244	2	public function newCJKSimpleCharacterRegExTokenizer( Tokenizer $tokinizer = null ) {
245	2	return new CJKSimpleCharacterRegExTokenizer( $tokinizer );
246		}
247
248		/**
249		* @since 0.1
250		*
251		* @return Tokenizer
252		*/
253	3	public function newNGramTokenizer( Tokenizer $tokinizer = null, $ngram = 2 ) {
254	3	return new NGramTokenizer( $tokinizer, $ngram );
255		}
256
257		}
258

onoi / tesa

SanitizerFactory::newTextCatLanguageDetector() A last analyzed 2018-01-27 17:32 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

SanitizerFactory::newTextCatLanguageDetector() A
last analyzed 2018-01-27 17:32 UTC