Refactor

ekinhbayar · Oct 31, 2016 · 32d80da · 32d80da
1 parent 322b046
commit 32d80da
Show file tree

Hide file tree

Showing 3 changed files with 292 additions and 75 deletions.
diff --git a/README.md b/README.md
@@ -1,6 +1,5 @@
 # Brill Tagger 
-[![Build Status](https://travis-ci.org/ekinhbayar/BrillTagger.svg?branch=master)](https://travis-ci.org/ekinhbayar/BrillTagger)
-[![Code Coverage](https://scrutinizer-ci.com/g/ekinhbayar/BrillTagger/badges/coverage.png?b=unit-tests)](https://scrutinizer-ci.com/g/ekinhbayar/BrillTagger/?branch=unit-tests)
+[![Build Status](https://travis-ci.org/ekinhbayar/BrillTagger.svg?branch=master)](https://travis-ci.org/ekinhbayar/BrillTagger) [![Code Coverage](https://scrutinizer-ci.com/g/ekinhbayar/BrillTagger/badges/coverage.png?b=unit-tests)](https://scrutinizer-ci.com/g/ekinhbayar/BrillTagger/?branch=master) [![Scrutinizer Code Quality](https://scrutinizer-ci.com/g/ekinhbayar/BrillTagger/badges/quality-score.png?b=unit-tests)](https://scrutinizer-ci.com/g/ekinhbayar/BrillTagger/?branch=master)
 
 PHP implementation of a Brill Tagger, an inductive part-of-speech tagging method.
 Uses a lexicon generated from [Brown Corpus](https://github.com/ekinhbayar/brown-corpus).

diff --git a/src/BrillTagger.php b/src/BrillTagger.php
@@ -14,6 +14,10 @@ class BrillTagger
 {
  private $dictionary = LEXICON;
 
+ /**
+ * @param $text
+ * @return array
+ */
  public function tag($text) {
 
  preg_match_all("/[\w\d\.'%@]+/", $text, $matches);
@@ -31,83 +35,24 @@ public function tag($text) {
  }
 
  # get from dictionary if set
- if (isset($this->dictionary[strtolower($token)])) {
+ if ($this->tokenExists($token)) {
  $tags[$i]['tag'] = $this->dictionary[strtolower($token)][0];
  }
 
- # Converts verbs after 'the' to nouns
- if ($i > 0) {
- if ($tags[$i-1]['tag'] == 'DT' && $this->isVerb($tags[$i]['tag'])) {
- $tags[$i]['tag'] = 'NN';
- }
- }
-
- # Convert noun to number if . appears
- if ($tags[$i]['tag'][0] == 'N' && strpos($token, '.') !== false) {
- $tags[$i]['tag'] = 'CD';
- }
-
- # manually tag numerals, cardinals, money (NNS)
- if (preg_match(NUMERAL, $token)) {
- $tags[$i]['tag'] = 'NNS';
- }
-
- # manually tag years
- if (preg_match(YEAR, $token, $matches)) {
- $tags[$i]['tag'] = (isset($matches['nns'])) ? 'NNS' : 'CD';
- }
-
- # manually tag percentages
- if (preg_match(PERCENTAGE, $token)) {
- $tags[$i]['tag'] = 'NN';
- }
-
- # Convert noun to past participle if ends with 'ed'
- if ($tags[$i]['tag'][0] == 'N' && substr($token, -2) == 'ed') {
- $tags[$i]['tag'] = 'VBN';
- }
+ $tags[$i]['tag'] = $this->transformNumerics($tags[$i]['tag'], $token);
 
  # Anything that ends 'ly' is an adverb
- if (substr($token, -2) == 'ly') {
+ if ($this->isAdverb($token)) {
  $tags[$i]['tag'] = 'RB';
  }
 
- # Common noun to adjective if it ends with 'al', to gerund if 'ing'
  if ($this->isNoun($tags[$i]['tag'])) {
- if (substr($token, -2) == 'al') {
- $tags[$i]['tag'] = 'JJ';
- } elseif (substr($token, -3) == 'ing') {
- $tags[$i]['tag'] = 'VBG';
- } elseif ($token === 'I') {
- $tags[$i]['tag'] = 'PPSS';
- }
+ $tags[$i]['tag'] = $this->transformNoun($tags[$i]['tag'], $token);
  }
 
- # Noun to verb if the word before is 'would'
  if ($i > 0) {
- if ($tags[$i]['tag'] == 'NN' && strtolower($tags[$i-1]['token']) == 'would') {
- $tags[$i]['tag'] = 'VB';
- }
- }
-
- # Noun to plural if it ends with an 's'
- if ($tags[$i]['tag'] == 'NN' && substr($token, -1) == 's') {
- $tags[$i]['tag'] = 'NNS';
- }
-
- # If we get noun noun, and the 2nd can be a verb, convert to verb
- if ($i > 0) {
-
- if ($this->isNoun($tags[$i]['tag'])
- && $this->isNoun($tags[$i-1]['tag'])
- && isset($this->dictionary[strtolower($token)])
- ) {
- if (in_array('VBN', $this->dictionary[strtolower($token)])) {
- $tags[$i]['tag'] = 'VBN';
- } else if (in_array('VBZ', $this->dictionary[strtolower($token)])) {
- $tags[$i]['tag'] = 'VBZ';
- }
- }
+ $tags[$i]['tag'] = $this->transformNounToVerb($tags, $i, $token);
+ $tags[$i]['tag'] = $this->transformVerbToNoun($tags, $i);
  }
 
  $i++;
@@ -116,46 +61,242 @@ public function tag($text) {
  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 in_array($tag, ['NN', 'NNS']);
+ return substr(trim($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';
  }
 
- # it him me us you 'em thee we'uns
+ /**
+ * @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
+ /** it he she thee
+ * @param string $tag
+ * @return bool
+ */
  public function isThirdPersonPronoun($tag) {
  return $tag === 'PPS';
  }
 
- # they we I you ye thou you'uns
+ /** 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
+ /** itself himself myself yourself herself oneself ownself
+ * @param string $tag
+ * @return bool
+ */
  public function isSingularReflexivePronoun($tag) {
  return $tag === 'PPL';
  }
 
- # themselves ourselves yourselves
+ /** 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
+ /** 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$$', 'PP$']);
+ return in_array($tag, ['PP$$', '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;
+ }
 }