Last active
August 29, 2015 14:00
-
-
Save ppKrauss/11200364 to your computer and use it in GitHub Desktop.
Javascript of http://www.corpuswiki.org translated to PHP and adapted for "restrict universe" option and object orientation.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
<?php | |
/** | |
* Detect language by trigram sample analysis. | |
* Adapted by ppkrauss from http://www.corpuswiki.org/langdetect.html | |
*/ | |
class LangDetect { | |
private $models; | |
private $universeConfig; | |
private $universeOnDetection; | |
private $isoTr3to2; | |
public $lang; | |
public $lang_alpha2; | |
public $score; | |
public $score_before; | |
public $score_befDiffPerc; | |
public $num_langs; | |
public function __construct($jsonFile='js/langtrigrams.json', $universe=false) { | |
$this->models = json_decode(file_get_contents($jsonFile),true); | |
$this->universeConfig = ( is_array($universe) && count($universe) )? $universe: false; | |
$this->isoTr3to2=array( // translate ISO 3166-1 codes (from alpha-3 to alpha-2) | |
'eng'=>'en', 'por'=>'pt', 'spa'=>'es', 'ita'=>'it', 'fra'=>'fr', 'afr'=>'af', | |
); | |
} | |
public function detect( $content, $universe='' ) { // array('eng','por','spa','ita', 'fra', 'afr') | |
if ($universe==='') $universe = $this->universeConfig; | |
$listScore = $this->scoreLangs($content,$universe); // do compareLang and updates $universeOnDetection | |
$this->score = 0; | |
for($i=0; $i<count($listScore); $i++) | |
if ( $this->score < $listScore[$i] ) { | |
$this->lang = $this->universeOnDetection[$i]; | |
$this->lang_alpha2 = $this->isoTr3to2[$this->lang]; | |
$this->score_before = $this->score; | |
$this->score = $listScore[$i]; | |
$this->score_befDiffPerc = round( 100.0*($this->score - $this->score_before)/$this->score ); | |
// print "\n\t\t-- DEBUG RESULT: bestlang={$this->lang}, score={$this->score} = {$this->models[$i]['iso']}"; | |
} | |
return $this->lang; | |
} | |
private function scoreLangs($content,$universe='') { | |
$score = array(); | |
$this->universeOnDetection = array(); | |
$mindist = $this->num_langs = 0; | |
$contentModel = $this->createOrderedModel($content); | |
foreach ($this->models as $model) if (!$universe || in_array($model['iso'],$universe)) { | |
$s = $this->compareLang($contentModel, $model['trigrams']); | |
$score[] = $s; | |
$this->universeOnDetection[] = $model['iso']; | |
$this->num_langs++; | |
} | |
return $score; | |
} | |
private function createOrderedModel($content) { | |
// Create a list of trigrams in content sorted by frequency. | |
$trigrams = array(); // assoc | |
$sortedTrigrams = array(); // array | |
$content = preg_replace('/[^\p{L}\']+/us',' ',$content); // only-UTF8-words filter | |
$contentArr = preg_split("//us", strtolower($content), -1, PREG_SPLIT_NO_EMPTY); //str_split_unicode | |
for ($i = 0, $l = count($contentArr) - 2; $i < $l; $i++) { | |
$trigramKey = $contentArr[$i] . $contentArr[$i + 1] . $contentArr[$i + 2]; | |
if (!isset($trigrams[$trigramKey])) | |
$trigrams[$trigramKey] = 1; | |
else | |
$trigrams[$trigramKey] += 1; | |
} | |
// convert object to array: | |
foreach ($trigrams as $k=>$v) | |
$sortedTrigrams[count($sortedTrigrams)] = [$k, $v]; | |
// sort array results: | |
uksort($sortedTrigrams, function($objA, $objB) { | |
return $objB[1] - $objA[1]; // sort high-to-low | |
} | |
); | |
return $sortedTrigrams; | |
} | |
private function compareLang($model, $known_model) { | |
// Calculate the distance to the known model. | |
$dist = 0; | |
if (!is_array($model) || !count($model)) die("\npau com MODEL"); | |
for ($i = 0, $l = count($model); $i < $l; $i++) { | |
if (isset($known_model[$model[$i][0]]) && $known_model[$model[$i][0]]) | |
$dist += abs($model[$i][1] - $known_model[$model[$i][0]]); | |
else | |
$dist += 300; | |
} | |
$score = 1 - ( $dist / (300*count($model))); | |
return $score; | |
} | |
} // class | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Some references and alternatives:
THEORY REF: https://en.wikipedia.org/wiki/Trigram https://en.wikipedia.org/wiki/Language_model#N-gram_models
ALTERNATIVES: https://en.wikipedia.org/wiki/Cache_language_model https://en.wikipedia.org/wiki/Katz%27s_back-off_model