Levenshtein::findClosestUriToToken() - Code Metrics - Inspection of "Merge branch 'master' of https://github.com/linked..." - linkeddatacenter/BOTK-core - Measure and Improve Code Quality continuously with Scrutinizer

Test Setup Failed

Push — master ( 51a98e...17386c )

by Enrico

created 2019-08-27 14:40 UTC

Levenshtein::findClosestUriToToken() B

↳ Parent: Levenshtein

Complexity

Conditions	11
Paths	4

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	29
rs	7.3166
c	0
b	0
f	0
cc	11
nc	4
nop	2

How to fix Complexity

<?php
/*
 * Copyright (c) 2019 LinkedData.Center. All rights reserved.
 */
namespace BOTK\Reasoner;

/**
 * input sensitivity: an acceptable rate for changes (0-1)
 * Input stream specification:
 *   The input stream is divided in two part, the first part contains, one for row,
 *   a couple of object Uri,  string
 *   then there is an empty line
 *   the second part contains, one for row, a couple of an subject uri and a token to be searched in the string
 *   
 * Output stream:
 *   rules to insert in a graphName, RDF statements that links subject and object having  
 *   the minimun Levenshtein distance evaluated on string tokens
 */
class Levenshtein extends AbstractReasoner
{
    protected $sensitivity=0.2;
   
    public function setSensitivity( float $sensitivity )
    {
        assert( $sentistivity >= 0 && $sentistivity <=1 );

        
        $this->sensitivity = $sensitivity;
        return $this ;
    }
      
 		
    /**
	 * targets must be an hash of tokenized strings
     */
	protected function findClosestUriToToken( $token, array $targets) 
	{
	    $tokenLength=strlen($token);
	    $absSensitivity= round($tokenLength * $this->sensitivity);
	    $minLength=max( 1, $tokenLength - $absSensitivity);
	    $maxLength=$tokenLength + $absSensitivity;
	    $closestTarget = null;
	    $shortest = -1;
	    foreach($targets as $targetUri=>$targetTokens){
	        if( $shortest === 0 ) break; // a perfect match found.
	        foreach ($targetTokens as $targetToken ){
	            $targetTokenLength=strlen($targetToken);
	            if( ($targetTokenLength >= $minLength) && ($targetTokenLength <= $maxLength)) {
	                $lev = levenshtein($token, $targetToken);
	                if ($lev <= $shortest || $shortest < 0) {
	                    $closestTarget  = $targetUri;
	                    $shortest = $lev;
	                }
	                if( $shortest === 0 ) break; // a perfect match found.
	            }
	        }
	    }
	    
	    if( $closestTarget && ($shortest <= $absSensitivity) ){
	        return $closestTarget;
	    } else {
	        return false;
	    }
	}
	
	
	public function run ()
	{
	    // analyze first input stream part: read targets
	    // only tokens with length > sensitivity are considered.
	    $targets=[];
	    
	    while (($data=fgetcsv($this->inputStream)) && isset($data[1])) {
	        assert( $data[0] && $data[1] ) ;
	        
	        $subjectUri=$data[0];
	        $string=$data[1];
	        
	        // tokenize string
	        $tokenizedString=[];
	        $tok= strtok($string, ' ');
	        while ($tok !== false) {
	            // ignore too short tokens
	            if( floor(strlen($tok)*$this->sensitivity)>0) { $tokenizedString[]=strtolower($tok);}
                $tok = strtok(' ');
	        }
	        
	        if( !empty($tokenizedString) ) { $targets[$subjectUri]=$tokenizedString;}
	    }
	    
	    //analyze second input stream part: token to be searched in targets
	    fwrite($this->outputStream, "INSERT DATA { GRAPH <$this->graphName> {\n");
	    
	    while (($data = fgetcsv($this->inputStream)) !== FALSE) {
	       
	        assert( $data[0] && $data[1] ) ;
	        
	        list ($uri, $token)=$data;
	        if( $closestUri=$this->findClosestUriToToken( strtolower($token), $targets) ){
	            fprintf($this->outputStream, "<%s> <%s> <%s>.\n", $closestUri, $this->property, $uri);
	        }
	        
	    }
	    fwrite($this->outputStream, "}}");	    
	}
}

1			<?php
2			/*
3			* Copyright (c) 2019 LinkedData.Center. All rights reserved.
4			*/
5			namespace BOTK\Reasoner;
6
7			/**
8			* input sensitivity: an acceptable rate for changes (0-1)
9			* Input stream specification:
10			* The input stream is divided in two part, the first part contains, one for row,
11			* a couple of object Uri, string
12			* then there is an empty line
13			* the second part contains, one for row, a couple of an subject uri and a token to be searched in the string
14			*
15			* Output stream:
16			* rules to insert in a graphName, RDF statements that links subject and object having
17			* the minimun Levenshtein distance evaluated on string tokens
18			*/
19			class Levenshtein extends AbstractReasoner
20			{
21			protected $sensitivity=0.2;
22
23			public function setSensitivity( float $sensitivity )
24			{
25			assert( $sentistivity >= 0 && $sentistivity <=1 );
			0 ignored issues – show Bug introduced 2019-08-27 14:42 UTC by Report Bug Copy Issue Report The variable `$sentistivity` does not exist. Did you forget to declare it? This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug. Loading history...
26
27			$this->sensitivity = $sensitivity;
28			return $this ;
29			}
30
31
32			/**
33			* targets must be an hash of tokenized strings
34			*/
35			protected function findClosestUriToToken( $token, array $targets)
36			{
37			$tokenLength=strlen($token);
38			$absSensitivity= round($tokenLength * $this->sensitivity);
39			$minLength=max( 1, $tokenLength - $absSensitivity);
40			$maxLength=$tokenLength + $absSensitivity;
41			$closestTarget = null;
42			$shortest = -1;
43			foreach($targets as $targetUri=>$targetTokens){
44			if( $shortest === 0 ) break; // a perfect match found.
45			foreach ($targetTokens as $targetToken ){
46			$targetTokenLength=strlen($targetToken);
47			if( ($targetTokenLength >= $minLength) && ($targetTokenLength <= $maxLength)) {
48			$lev = levenshtein($token, $targetToken);
49			if ($lev <= $shortest \|\| $shortest < 0) {
50			$closestTarget = $targetUri;
51			$shortest = $lev;
52			}
53			if( $shortest === 0 ) break; // a perfect match found.
54			}
55			}
56			}
57
58			if( $closestTarget && ($shortest <= $absSensitivity) ){
59			return $closestTarget;
60			} else {
61			return false;
62			}
63			}
64
65
66			public function run ()
67			{
68			// analyze first input stream part: read targets
69			// only tokens with length > sensitivity are considered.
70			$targets=[];
71
72			while (($data=fgetcsv($this->inputStream)) && isset($data[1])) {
73			assert( $data[0] && $data[1] ) ;
74
75			$subjectUri=$data[0];
76			$string=$data[1];
77
78			// tokenize string
79			$tokenizedString=[];
80			$tok= strtok($string, ' ');
81			while ($tok !== false) {
82			// ignore too short tokens
83			if( floor(strlen($tok)*$this->sensitivity)>0) { $tokenizedString[]=strtolower($tok);}
84			$tok = strtok(' ');
85			}
86
87			if( !empty($tokenizedString) ) { $targets[$subjectUri]=$tokenizedString;}
88			}
89
90			//analyze second input stream part: token to be searched in targets
91			fwrite($this->outputStream, "INSERT DATA { GRAPH <$this->graphName> {\n");
92
93			while (($data = fgetcsv($this->inputStream)) !== FALSE) {
94
95			assert( $data[0] && $data[1] ) ;
96
97			list ($uri, $token)=$data;
98			if( $closestUri=$this->findClosestUriToToken( strtolower($token), $targets) ){
99			fprintf($this->outputStream, "<%s> <%s> <%s>.\n", $closestUri, $this->property, $uri);
100			}
101
102			}
103			fwrite($this->outputStream, "}}");
104			}
105			}

linkeddatacenter / BOTK-core

Push — master ( 51a98e...17386c )

Levenshtein::findClosestUriToToken() B

Complexity

Size

Duplication

Importance

How to fix Complexity

Long Method

Duplication Side-by-Side

Filter issues like