BrillTagger::isPronoun() - Code Metrics - Inspection of "Refactor" - ekinhbayar/BrillTagger - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Branch — master (32d80d)

by Ekin

created 2017-03-19 13:44 UTC

BrillTagger::isPronoun() A

↳ Parent: BrillTagger

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

<?php
/**
 * Part Of Speech Tagging
 * Brill Tagger
 *
 * @category   BrillTagger
 * @author     Ekin H. Bayar <[email protected]>
 * @version    0.1.0
 */

namespace BrillTagger;

class BrillTagger
{
    private $dictionary = LEXICON;

    /**
     * @param $text
     * @return array
     */
    public function tag($text) {

        preg_match_all("/[\w\d\.'%@]+/", $text, $matches);

        $tags = [];
        $i = 0;

        foreach ($matches[0] as $token) {
            # default to a common noun
            $tags[$i] = ['token' => $token, 'tag' => 'NN'];

            # remove trailing full stops
            if (substr(trim($token), -1) == '.') {
                $token = preg_replace('/\.+$/', '', $token);
            }

            # get from dictionary if set
            if ($this->tokenExists($token)) {
                $tags[$i]['tag'] = $this->dictionary[strtolower($token)][0];
            }

            $tags[$i]['tag'] = $this->transformNumerics($tags[$i]['tag'], $token);

            # Anything that ends 'ly' is an adverb
            if ($this->isAdverb($token)) {
                $tags[$i]['tag'] = 'RB';
            }

            if ($this->isNoun($tags[$i]['tag'])) {
                $tags[$i]['tag'] = $this->transformNoun($tags[$i]['tag'], $token);
            }

            if ($i > 0) {
                $tags[$i]['tag'] = $this->transformNounToVerb($tags, $i, $token);
                $tags[$i]['tag'] = $this->transformVerbToNoun($tags, $i);
            }

            $i++;
        }

        return $tags;
    }

    /**
     * @param string $token
     * @return bool
     */
    public function tokenExists($token) {
        return isset($this->dictionary[strtolower($token)]);
    }

    /**
     * @param string $tag
     * @return bool
     */
    public function isNoun($tag) {
        return substr(trim($tag), 0, 1) == 'N';
    }

    /**
     * @param string $tag
     * @return bool
     */
    public function isSingularNoun($tag) {
        return $tag == 'NN';
    }

    /**
     * @param string $tag
     * @param string $token
     * @return bool
     */
    public function isPluralNoun($tag, $token) {
        return ($this->isNoun($tag) && substr($token, -1) == 's');
    }

    /**
     * @param string $tag
     * @return bool
     */
    public function isVerb($tag) {
        return substr(trim($tag), 0, 2) == 'VB';
    }

    /**
     * @param string $tag
     * @return bool
     */
    public function isPronoun($tag) {
        return substr(trim($tag), 0, 1) == 'P';
    }

    /**
     * @param string $token
     * @return bool
     */
    public function isPastTenseVerb($token) {
        return in_array('VBN', $this->dictionary[strtolower($token)]);
    }

    /**
     * @param string $token
     * @return bool
     */
    public function isPresentTenseVerb($token) {
        return in_array('VBZ', $this->dictionary[strtolower($token)]);
    }

    /** it him me us you 'em thee we'uns
     * @param string $tag
     * @return bool
     */
    public function isAccusativePronoun($tag) {
        return $tag === 'PPO';
    }

    /** it he she thee
     * @param string $tag
     * @return bool
     */
    public function isThirdPersonPronoun($tag) {
        return $tag === 'PPS';
    }

    /** they we I you ye thou you'uns
     * @param string $tag
     * @return bool
     */
    public function isSingularPersonalPronoun($tag) {
        return $tag === 'PPSS';
    }

    /** itself himself myself yourself herself oneself ownself
     * @param string $tag
     * @return bool
     */
    public function isSingularReflexivePronoun($tag) {
        return $tag === 'PPL';
    }

    /** themselves ourselves yourselves
     * @param string $tag
     * @return bool
     */
    public function isPluralReflexivePronoun($tag) {
        return $tag === 'PPLS';
    }

    /** ours mine his her/hers their/theirs our its my your/yours out thy thine
     * @param string $tag
     * @return bool
     */
    public function isPossessivePronoun($tag) {
        return in_array($tag, ['PP$$', 'PP$']);
    }

    /**
     * @param string $token
     * @return bool
     */
    public function isAdjective($token) {
        return (substr($token, -2) == 'al' || in_array('JJ', $this->dictionary[strtolower($token)]));
    }

    /**
     * @param string $token
     * @return bool
     */
    public function isGerund($token) {
        return substr($token, -3) == 'ing';
    }

    /**
     * @param string $token
     * @return bool
     */
    public function isPastParticiple($token) {
        return substr($token, -2) == 'ed';
    }

    /**
     * @param string $token
     * @return bool
     */
    public function isAdverb($token) {
        return substr($token, -2) == 'ly';
    }

    /** Common noun to adj. if it ends with 'al',
     * to gerund if 'ing', to past tense if 'ed'
     *
     * @param string $tag
     * @param string $token
     * @return string
     */
    public function transformNoun($tag, $token) {

        if ($this->isAdjective($token)) {
            $tag = 'JJ';
        } elseif ($this->isGerund($token)) {
            $tag = 'VBG';
        } elseif ($this->isPastParticiple($token)) {
            $tag = 'VBN';
        } elseif ($token === 'I') {
            $tag = 'PPSS';
        } elseif ($this->isPluralNoun($tag, $token)) {
            $tag = 'NNS';
        }

        # Convert noun to number if . appears
        if (strpos($token, '.') !== false) {
            $tag = 'CD';
        }

        return $tag;
    }

    /**
     * @param array $tags
     * @param int $i
     * @param string $token
     * @return mixed
     */
    public function transformNounToVerb($tags, $i, $token) {
        # Noun to verb if the word before is 'would'
        if ($this->isSingularNoun($tags[$i]['tag']) && strtolower($tags[$i-1]['token']) == 'would') {
            $tags[$i]['tag'] = 'VB';
        }

        # If we get noun noun, and the 2nd can be a verb, convert to verb
        if ($this->isNoun($tags[$i]['tag']) &&
            $this->isNoun($tags[$i-1]['tag']) &&
            $this->tokenExists($token)
        ) {
            if ($this->isPastTenseVerb($token)) {
                $tags[$i]['tag'] = 'VBN';
            } elseif ($this->isPresentTenseVerb($token)) {
                $tags[$i]['tag'] = 'VBZ';
            }
        }

        return $tags[$i]['tag'];
    }

    /**
     * @param array $tags
     * @param int $i
     * @return mixed
     */
    public function transformVerbToNoun($tags, $i) {
        # Converts verbs after 'the' to nouns
        if ($tags[$i-1]['tag'] == 'DT' && $this->isVerb($tags[$i]['tag'])) {
            $tags[$i]['tag'] = 'NN';
        }

        return $tags[$i]['tag'];
    }

    /**
     * @param string $tag
     * @param string $token
     * @return string
     */
    public function transformNumerics($tag, $token) {
        # tag numerals, cardinals, money (NNS)
        if (preg_match(NUMERAL, $token)) {
            $tag = 'NNS';
        }

        # tag years
        if (preg_match(YEAR, $token, $matches)) {
            $tag = (isset($matches['nns'])) ? 'NNS' : 'CD';
        }

        # tag percentages
        if (preg_match(PERCENTAGE, $token)) {
            $tag = 'NN';
        }

        return $tag;
    }
}


1		<?php
2		/**
3		* Part Of Speech Tagging
4		* Brill Tagger
5		*
6		* @category BrillTagger
7		* @author Ekin H. Bayar <[email protected]>
8		* @version 0.1.0
9		*/
10
11		namespace BrillTagger;
12
13		class BrillTagger
14		{
15		private $dictionary = LEXICON;
16
17		/**
18		* @param $text
19		* @return array
20		*/
21	21	public function tag($text) {
22
23	21	preg_match_all("/[\w\d\.'%@]+/", $text, $matches);
24
25	21	$tags = [];
26	21	$i = 0;
27
28	21	foreach ($matches[0] as $token) {
29		# default to a common noun
30	21	$tags[$i] = ['token' => $token, 'tag' => 'NN'];
31
32		# remove trailing full stops
33	21	if (substr(trim($token), -1) == '.') {
34	12	$token = preg_replace('/\.+$/', '', $token);
35		}
36
37		# get from dictionary if set
38	21	if ($this->tokenExists($token)) {
39	21	$tags[$i]['tag'] = $this->dictionary[strtolower($token)][0];
40		}
41
42	21	$tags[$i]['tag'] = $this->transformNumerics($tags[$i]['tag'], $token);
43
44		# Anything that ends 'ly' is an adverb
45	21	if ($this->isAdverb($token)) {
46	1	$tags[$i]['tag'] = 'RB';
47		}
48
49	21	if ($this->isNoun($tags[$i]['tag'])) {
50	18	$tags[$i]['tag'] = $this->transformNoun($tags[$i]['tag'], $token);
51		}
52
53	21	if ($i > 0) {
54	12	$tags[$i]['tag'] = $this->transformNounToVerb($tags, $i, $token);
55	12	$tags[$i]['tag'] = $this->transformVerbToNoun($tags, $i);
56		}
57
58	21	$i++;
59		}
60
61	21	return $tags;
62		}
63
64		/**
65		* @param string $token
66		* @return bool
67		*/
68	22	public function tokenExists($token) {
69	22	return isset($this->dictionary[strtolower($token)]);
70		}
71
72		/**
73		* @param string $tag
74		* @return bool
75		*/
76	21	public function isNoun($tag) {
77	21	return substr(trim($tag), 0, 1) == 'N';
78		}
79
80		/**
81		* @param string $tag
82		* @return bool
83		*/
84	13	public function isSingularNoun($tag) {
85	13	return $tag == 'NN';
86		}
87
88		/**
89		* @param string $tag
90		* @param string $token
91		* @return bool
92		*/
93	15	public function isPluralNoun($tag, $token) {
94	15	return ($this->isNoun($tag) && substr($token, -1) == 's');
95		}
96
97		/**
98		* @param string $tag
99		* @return bool
100		*/
101	2	public function isVerb($tag) {
102	2	return substr(trim($tag), 0, 2) == 'VB';
103		}
104
105		/**
106		* @param string $tag
107		* @return bool
108		*/
109	1	public function isPronoun($tag) {
110	1	return substr(trim($tag), 0, 1) == 'P';
111		}
112
113		/**
114		* @param string $token
115		* @return bool
116		*/
117	4	public function isPastTenseVerb($token) {
118	4	return in_array('VBN', $this->dictionary[strtolower($token)]);
119		}
120
121		/**
122		* @param string $token
123		* @return bool
124		*/
125	4	public function isPresentTenseVerb($token) {
126	4	return in_array('VBZ', $this->dictionary[strtolower($token)]);
127		}
128
129		/** it him me us you 'em thee we'uns
130		* @param string $tag
131		* @return bool
132		*/
133	1	public function isAccusativePronoun($tag) {
134	1	return $tag === 'PPO';
135		}
136
137		/** it he she thee
138		* @param string $tag
139		* @return bool
140		*/
141	1	public function isThirdPersonPronoun($tag) {
142	1	return $tag === 'PPS';
143		}
144
145		/** they we I you ye thou you'uns
146		* @param string $tag
147		* @return bool
148		*/
149	1	public function isSingularPersonalPronoun($tag) {
150	1	return $tag === 'PPSS';
151		}
152
153		/** itself himself myself yourself herself oneself ownself
154		* @param string $tag
155		* @return bool
156		*/
157	1	public function isSingularReflexivePronoun($tag) {
158	1	return $tag === 'PPL';
159		}
160
161		/** themselves ourselves yourselves
162		* @param string $tag
163		* @return bool
164		*/
165	1	public function isPluralReflexivePronoun($tag) {
166	1	return $tag === 'PPLS';
167		}
168
169		/** ours mine his her/hers their/theirs our its my your/yours out thy thine
170		* @param string $tag
171		* @return bool
172		*/
173	1	public function isPossessivePronoun($tag) {
174	1	return in_array($tag, ['PP$$', 'PP$']);
175		}
176
177		/**
178		* @param string $token
179		* @return bool
180		*/
181	19	public function isAdjective($token) {
182	19	return (substr($token, -2) == 'al' \|\| in_array('JJ', $this->dictionary[strtolower($token)]));
183		}
184
185		/**
186		* @param string $token
187		* @return bool
188		*/
189	19	public function isGerund($token) {
190	19	return substr($token, -3) == 'ing';
191		}
192
193		/**
194		* @param string $token
195		* @return bool
196		*/
197	19	public function isPastParticiple($token) {
198	19	return substr($token, -2) == 'ed';
199		}
200
201		/**
202		* @param string $token
203		* @return bool
204		*/
205	22	public function isAdverb($token) {
206	22	return substr($token, -2) == 'ly';
207		}
208
209		/** Common noun to adj. if it ends with 'al',
210		* to gerund if 'ing', to past tense if 'ed'
211		*
212		* @param string $tag
213		* @param string $token
214		* @return string
215		*/
216	18	public function transformNoun($tag, $token) {
217
218	18	if ($this->isAdjective($token)) {
219		$tag = 'JJ';
220	18	} elseif ($this->isGerund($token)) {
221	1	$tag = 'VBG';
222	18	} elseif ($this->isPastParticiple($token)) {
223		$tag = 'VBN';
224	18	} elseif ($token === 'I') {
225	5	$tag = 'PPSS';
226	15	} elseif ($this->isPluralNoun($tag, $token)) {
227	8	$tag = 'NNS';
228		}
229
230		# Convert noun to number if . appears
231	18	if (strpos($token, '.') !== false) {
232		$tag = 'CD';
233		}
234
235	18	return $tag;
236		}
237
238		/**
239		* @param array $tags
240		* @param int $i
241		* @param string $token
242		* @return mixed
243		*/
244	12	public function transformNounToVerb($tags, $i, $token) {
245		# Noun to verb if the word before is 'would'
246	12	if ($this->isSingularNoun($tags[$i]['tag']) && strtolower($tags[$i-1]['token']) == 'would') {
247		$tags[$i]['tag'] = 'VB';
248		}
249
250		# If we get noun noun, and the 2nd can be a verb, convert to verb
251	12	if ($this->isNoun($tags[$i]['tag']) &&
252	12	$this->isNoun($tags[$i-1]['tag']) &&
253	12	$this->tokenExists($token)
254		) {
255	3	if ($this->isPastTenseVerb($token)) {
256		$tags[$i]['tag'] = 'VBN';
257	3	} elseif ($this->isPresentTenseVerb($token)) {
258	3	$tags[$i]['tag'] = 'VBZ';
259		}
260		}
261
262	12	return $tags[$i]['tag'];
263		}
264
265		/**
266		* @param array $tags
267		* @param int $i
268		* @return mixed
269		*/
270	12	public function transformVerbToNoun($tags, $i) {
271		# Converts verbs after 'the' to nouns
272	12	if ($tags[$i-1]['tag'] == 'DT' && $this->isVerb($tags[$i]['tag'])) {
273		$tags[$i]['tag'] = 'NN';
274		}
275
276	12	return $tags[$i]['tag'];
277		}
278
279		/**
280		* @param string $tag
281		* @param string $token
282		* @return string
283		*/
284	21	public function transformNumerics($tag, $token) {
285		# tag numerals, cardinals, money (NNS)
286	21	if (preg_match(NUMERAL, $token)) {
287	1	$tag = 'NNS';
288		}
289
290		# tag years
291	21	if (preg_match(YEAR, $token, $matches)) {
292	1	$tag = (isset($matches['nns'])) ? 'NNS' : 'CD';
293		}
294
295		# tag percentages
296	21	if (preg_match(PERCENTAGE, $token)) {
297	1	$tag = 'NN';
298		}
299
300	21	return $tag;
301		}
302		}
303

ekinhbayar / BrillTagger

Branch — master (32d80d)

BrillTagger::isPronoun() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like