View GitHub Repository UrlHarvester

Schedule Inspection
Subscribe

Helper::htmlToPlainText()   A
last analyzed

↳ Parent: Helper

Complexity

Conditions 2
Paths 2

Size

Total Lines 14
Code Lines 10

Duplication

Lines 0
Ratio 0 %

Code Coverage

Tests 9
CRAP Score 2.004

Importance

Changes 1
Bugs 0 Features 0
Metric Value
eloc 10
c 1
b 0
f 0
dl 0
loc 14
ccs 9
cts 10
cp 0.9
rs 9.9332
cc 2
nc 2
nop 2
crap 2.004
1 
<?php
2 

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    3
                

                                    
                                                    
                
                
                
namespace PiedWeb\UrlHarvester;

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    4
                

                                    
                                                    
                
                
                

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    5
                

                                    
                                                    
                
                
                
use ForceUTF8\Encoding;

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    6
                

                                    
                                                    
                
                
                

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    7
                

                                    
                                                    
                
                
                
class Helper

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    8
                

                                    
                                                    
                
                
                
{

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    9
                

                                    
                             30                         
                
                
                
    public static function clean(string $source)

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    10
                

                                    
                                                    
                
                
                
    {

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    11
                

                                    
                             30                         
                
                
                
        return trim(preg_replace('/\s{2,}/', ' ', Encoding::toUTF8($source)));

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    12
                

                                    
                                                    
                
                
                
    }

            


            
                                                                                                            
                                                                
            
                                    
            
            

                
                    13
                

                                    
                                                    
                
                
                

            


            
                                                                        
                            
            
                                    
            
            

                
                    14
                

                                    
                             6                         
                
                
                
    public static function htmlToPlainText($str, $keepN = false)

            


            
                                                                        
                            
            
                                    
            
            

                
                    15
                

                                    
                                                    
                
                
                
    {

            


            
                                                                        
                            
            
                                    
            
            

                
                    16
                

                                    
                             6                         
                
                
                
        $str = preg_replace('#<(style|script).*</(style|script)>#siU', ' ', $str);

            


            
                                                                        
                            
            
                                    
            
            

                
                    17
                

                                    
                             6                         
                
                
                
        $str = preg_replace('#</?(br|p|div)>#siU', "\n", $str);

            


            
                                                                        
                            
            
                                    
            
            

                
                    18
                

                                    
                             6                         
                
                
                
        $str = preg_replace('/<\/[a-z]+>/siU', ' ', $str);

            


            
                                                                        
                            
            
                                    
            
            

                
                    19
                

                                    
                             6                         
                
                
                
        $str = str_replace(["\r", "\t"], ' ', $str);

            


            
                                                                        
                            
            
                                    
            
            

                
                    20
                

                                    
                             6                         
                
                
                
        $str = strip_tags(preg_replace('/<[^<]+?>/', ' ', $str));

            


            
                                                                        
                            
            
                                    
            
            

                
                    21
                

                                    
                             6                         
                
                
                
        if ($keepN) {

            


            
                                                                        
                            
            
                                    
            
            

                
                    22
                

                                    
                                                    
                
                
                
            $str = preg_replace('/ {2,}/', ' ', $str);

            


            
                                                                        
                            
            
                                    
            
            

                
                    23
                

                                    
                                                    
                
                
                
        } else {

            


            
                                                                        
                            
            
                                    
            
            

                
                    24
                

                                    
                             6                         
                
                
                
            $str = preg_replace('/\s+/', ' ', $str);

            


            
                                                                        
                            
            
                                    
            
            

                
                    25
                

                                    
                                                    
                
                
                
        }

            


            
                                                                        
                            
            
                                    
            
            

                
                    26
                

                                    
                                                    
                
                
                

            


            
                                                                        
                            
            
                                    
            
            

                
                    27
                

                                    
                             6                         
                
                
                
        return $str;

            


            
                                                                                                            
                            
            
                                    
            
            

                
                    28
                

                                    
                                                    
                
                
                
    }

            


            
                                                                                                            
                                                                
            
                                    
            
            

                
                    29
                

                                    
                                                    
                
                
                
}

            


            
                                                        
            
                                    
            
            

                
                    30