manuwhat / similar-text

tests/Similar_textTest.php

Indentation +1 added lines, -1 removed lines

@@ -3,7 +3,7 @@
     require($DIR = (($dir = dirname(__DIR__)).DIRECTORY_SEPARATOR.'src'.DIRECTORY_SEPARATOR)).'similar_text.php';
     require $DIR.'simpleCommonTextSimilarities.php';
     require $DIR.'complexCommonTextSimilarities.php';
-     require $DIR.'distance.php';
+        require $DIR.'distance.php';
     require $dir.DIRECTORY_SEPARATOR.'similar_text.php';
 
     use PHPUnit\Framework\TestCase;

src/complexCommonTextSimilarities.php

Spacing +13 added lines, -13 removed lines

@@ -78,32 +78,32 @@ 
             if (!is_string($a) || !is_string($b)) {
                 return false;
             }
-            $filter = function ($v) {
+            $filter = function($v) {
                 return !(ctype_space($v));
             };
             self::filter($a, $b, $filter, true);
             return self::waorDiff($a, $b, count($a), count($b));
         }
         
-        private static function filter(&$a, &$b, $filter, $insensitive = true, $captureLength=false)
+        private static function filter(&$a, &$b, $filter, $insensitive = true, $captureLength = false)
         {
             if ($insensitive) {
                 $a = array_filter(self::getParts(self::strtolower($a), $c, $captureLength), $filter);
-                if ($c===1) {
-                    $a=self::strtolower($a);
+                if ($c === 1) {
+                    $a = self::strtolower($a);
                 }
                 $b = array_filter(self::getParts(self::strtolower($b), $c, $captureLength), $filter);
-                if ($c===1) {
-                    $b=self::strtolower($b);
+                if ($c === 1) {
+                    $b = self::strtolower($b);
                 }
             } else {
                 $a = array_filter(self::getParts(self::split($a), $c, $captureLength), $filter);
-                if ($c===1) {
-                    $a=self::strtolower($a);
+                if ($c === 1) {
+                    $a = self::strtolower($a);
                 }
                 $b = array_filter(self::getParts(self::split($b), $c, $captureLength), $filter);
-                if ($c===1) {
-                    $b=self::strtolower($b);
+                if ($c === 1) {
+                    $b = self::strtolower($b);
                 }
             }
         }
@@ -116,7 +116,7 @@ 
         
         public static function punctuationChangesOccured($a, $b, $insensitive = true, $considerSpace = true)
         {
-            $filter = function ($v) use ($considerSpace) {
+            $filter = function($v) use ($considerSpace) {
                 return $considerSpace ? !(ctype_space($v) || ctype_punct($v)) : !ctype_punct($v);
             };
             if (!is_string($a) || !is_string($b)) {
@@ -132,7 +132,7 @@ 
             if (!is_string($a) || !is_string($b)) {
                 return false;
             }
-            $filter = function ($v) {
+            $filter = function($v) {
                 return !(ctype_space($v[0]) || ctype_punct($v[0]));
             };
             
@@ -143,7 +143,7 @@ 
         private static function aoeStemming($a, $b)
         {
             foreach ($a as $index=>$word) {
-                if (!self::haveSameRoot($word[0], $b[$index][0]) || ($a[$index][1]>2 && $b[$index][1]>2)) {
+                if (!self::haveSameRoot($word[0], $b[$index][0]) || ($a[$index][1] > 2 && $b[$index][1] > 2)) {
                     return false;
                 }
             }

src/simpleCommonTextSimilarities.php

Spacing +1 added lines, -1 removed lines

@@ -36,7 +36,7 @@
         
         public static function wordsReorderOccured($a, $b, $considerPunctuation = true)
         {
-            $filter = function ($v) use ($considerPunctuation) {
+            $filter = function($v) use ($considerPunctuation) {
                 return $considerPunctuation ? !(ctype_space($v) || ctype_punct($v)) : !ctype_space($v);
             };
             return self::similarText($a, $b, 2, true, $check, true) && is_array($check) && self::wro_filter($check, $filter) ?true :false;

similar_text.php

Spacing +2 added lines, -2 removed lines

@@ -95,7 +95,7 @@ 
     }
     
     
-    function dice($a, $b, $round=2)
+    function dice($a, $b, $round = 2)
     {
         return Distance::dice($a, $b, $round);
     }
@@ -107,7 +107,7 @@ 
     }
     
     
-    function jaroWinkler($a, $b, $round=2)
+    function jaroWinkler($a, $b, $round = 2)
     {
         return Distance::jaroWinkler($a, $b, $round);
     }

src/distance.php

Spacing +73 added lines, -73 removed lines

@@ -14,17 +14,17 @@ 
 namespace EZAMA{
     class distance extends complexCommonTextSimilarities
     {
-        public static function jaroWinkler($a, $b, $round=2)
+        public static function jaroWinkler($a, $b, $round = 2)
         {
-            if (!is_string($a)||!is_string($b)) {
+            if (!is_string($a) || !is_string($b)) {
                 return false;
             }
-            static $distance=array();
-            static $previous=array();
-            if (array($a,$b)===$previous) {
+            static $distance = array();
+            static $previous = array();
+            if (array($a, $b) === $previous) {
                 return $distance;
             }
-            $previous=array($a,$b);
+            $previous = array($a, $b);
             return self::getJWDistance($a, $b, $distance, $round);
         }
         
@@ -33,15 +33,15 @@ 
         private static function getJWDistance(&$a, &$b, &$distance, $round)
         {
             extract(self::prepareJaroWinkler($a, $b));
-            for ($i=0,$min=min(count($a), count($b)),$t=0;$i<$min;$i++) {
-                if ($a[$i]!==$b[$i]) {
+            for ($i = 0, $min = min(count($a), count($b)), $t = 0; $i < $min; $i++) {
+                if ($a[$i] !== $b[$i]) {
                     $t++;
                 }
             }
-            $t/=2;
-            $distance['jaro']=1/3*($corresponding/$ca+$corresponding/$cb+($corresponding-$t)/$corresponding);
-            $distance['jaro-winkler']=$distance['jaro']+(min($longCommonSubstr, 4)*0.1*(1-$distance['jaro']));
-            $distance=array_map(function ($v) use ($round) {
+            $t /= 2;
+            $distance['jaro'] = 1 / 3 * ($corresponding / $ca + $corresponding / $cb + ($corresponding - $t) / $corresponding);
+            $distance['jaro-winkler'] = $distance['jaro'] + (min($longCommonSubstr, 4) * 0.1 * (1 - $distance['jaro']));
+            $distance = array_map(function($v) use ($round) {
                 return round($v, $round);
             }, $distance);
             
@@ -50,15 +50,15 @@ 
         
         private static function prepareJaroWinkler(&$a, &$b)
         {
-            $a=self::split($a);
-            $b=self::split($b);
-            $transpositions=array('a'=>array(),'b'=>array(),'corresponding'=>0,'longCommonSubstr'=>0,'ca'=>count($a),'cb'=>count($b));
-            $Δ=max($transpositions['ca'], $transpositions['cb'])/2-1;
+            $a = self::split($a);
+            $b = self::split($b);
+            $transpositions = array('a'=>array(), 'b'=>array(), 'corresponding'=>0, 'longCommonSubstr'=>0, 'ca'=>count($a), 'cb'=>count($b));
+            $Δ = max($transpositions['ca'], $transpositions['cb']) / 2 - 1;
             self::jwMatches($a, $b, $transpositions, $Δ);
             ksort($transpositions['a']);
             ksort($transpositions['b']);
-            $transpositions['a']=array_values($transpositions['a']);
-            $transpositions['b']=array_values($transpositions['b']);
+            $transpositions['a'] = array_values($transpositions['a']);
+            $transpositions['b'] = array_values($transpositions['b']);
             return $transpositions;
         }
         
@@ -73,12 +73,12 @@ 
         
         private static function _jwMatches($chr, $char, $index, $ind, &$transpositions, $Δ)
         {
-            if ($chr===$char&&(abs($index-$ind)<=$Δ)) {
-                if ($ind!==$index) {
-                    $transpositions['a'][$ind]=$chr;
-                    $transpositions['b'][$index]=$char;
+            if ($chr === $char && (abs($index - $ind) <= $Δ)) {
+                if ($ind !== $index) {
+                    $transpositions['a'][$ind] = $chr;
+                    $transpositions['b'][$index] = $char;
                 } else {
-                    if ($ind-1<=$transpositions['longCommonSubstr']) {
+                    if ($ind - 1 <= $transpositions['longCommonSubstr']) {
                         $transpositions['longCommonSubstr']++;
                     }
                 }
@@ -89,59 +89,59 @@ 
         
         public static function hamming($a, $b)
         {
-            if (!is_string($a)||!is_string($b)||(strlen($a)!==strlen($b))) {
+            if (!is_string($a) || !is_string($b) || (strlen($a) !== strlen($b))) {
                 return false;
             }
-            static $distance=0;
-            static $previous=array();
-            if (array($a,$b)===$previous) {
+            static $distance = 0;
+            static $previous = array();
+            if (array($a, $b) === $previous) {
                 return $distance;
             }
-            $previous=array($a,$b);
-            $a=self::split($a);
-            $b=self::split($b);
-            $distance=count(array_diff_assoc($a, $b));
+            $previous = array($a, $b);
+            $a = self::split($a);
+            $b = self::split($b);
+            $distance = count(array_diff_assoc($a, $b));
             return $distance;
         }
         
-        public static function dice($a, $b, $round=2)
+        public static function dice($a, $b, $round = 2)
         {
-            if (!is_string($a)||!is_string($b)) {
+            if (!is_string($a) || !is_string($b)) {
                 return false;
             }
-            if (empty($a)||empty($b)) {
+            if (empty($a) || empty($b)) {
                 return 0.0;
             }
-            if ($a===$b) {
+            if ($a === $b) {
                 return 1.0;
             }
             
-            static $distance=0;
-            static $previous=array();
-            if (array($a,$b)===$previous) {
+            static $distance = 0;
+            static $previous = array();
+            if (array($a, $b) === $previous) {
                 return $distance;
             }
-            $previous=array($a,$b);
-            $a=self::split($a, 2);
-            $b=self::split($b, 2);
+            $previous = array($a, $b);
+            $a = self::split($a, 2);
+            $b = self::split($b, 2);
             return self::getDiceDistance($distance, $a, $b, $round);
         }
         
         private static function getDiceDistance(&$distance, &$a, &$b, $round)
         {
-            $ca=($caGrams=count($a))*2-self::getEndStrLen($a);
-            $cb=($cbGrams=count($b))*2-self::getEndStrLen($b);
-            $distance=round(2*count($caGrams>$cbGrams?array_intersect($a, $b):array_intersect($b, $a))/($ca+$cb), $round);
+            $ca = ($caGrams = count($a)) * 2 - self::getEndStrLen($a);
+            $cb = ($cbGrams = count($b)) * 2 - self::getEndStrLen($b);
+            $distance = round(2 * count($caGrams > $cbGrams ?array_intersect($a, $b) : array_intersect($b, $a)) / ($ca + $cb), $round);
         }
         
         private static function getEndStrLen($a)
         {
             if (function_exists('array_key_last')) {
-                $end=array_key_last($a);
-                $end=(isset($end[1]))?0:1;
+                $end = array_key_last($a);
+                $end = (isset($end[1])) ? 0 : 1;
             } else {
-                $end=end($a);
-                $end=(isset($end[1]))?0:1;
+                $end = end($a);
+                $end = (isset($end[1])) ? 0 : 1;
                 reset($a);
             }
             return $end;
@@ -149,56 +149,56 @@ 
         
         public static function levenshtein($a, $b)
         {
-            if (!is_string($a)||!is_string($b)) {
+            if (!is_string($a) || !is_string($b)) {
                 return false;
             }
             
             
-            static $distance=0;
-            static $previous=array();
-            if (array($a,$b)===$previous) {
+            static $distance = 0;
+            static $previous = array();
+            if (array($a, $b) === $previous) {
                 return $distance;
             }
-            $previous=array($a,$b);
-            $a=self::split($a);
-            $b=self::split($b);
+            $previous = array($a, $b);
+            $a = self::split($a);
+            $b = self::split($b);
             $ca = count($a);
             $cb = count($b);
             $dis = range(0, $cb);
             self::BuildLevenshteinCostMatrix($a, $b, $ca, $cb, $dis);
 
-            return $distance=$dis[$cb];
+            return $distance = $dis[$cb];
         }
         
         
         public static function levenshteinDamerau($a, $b)
         {
-            if (!is_string($a)||!is_string($b)) {
+            if (!is_string($a) || !is_string($b)) {
                 return false;
             }
               
-            static $distance=0;
-            static $previous=array();
-            if (array($a,$b)===$previous) {
+            static $distance = 0;
+            static $previous = array();
+            if (array($a, $b) === $previous) {
                 return $distance;
             }
-            $previous=array($a,$b);
-            $a=self::split($a);
-            $b=self::split($b);
+            $previous = array($a, $b);
+            $a = self::split($a);
+            $b = self::split($b);
             $ca = count($a);
             $cb = count($b);
             $dis = range(0, $cb);
             self::BuildLevenshteinCostMatrix($a, $b, $ca, $cb, $dis, true);
         
-            return $distance=$dis[$cb];
+            return $distance = $dis[$cb];
         }
         
-        private static function BuildLevenshteinCostMatrix($a, $b, $ca, $cb, &$dis, $damerau=false)
+        private static function BuildLevenshteinCostMatrix($a, $b, $ca, $cb, &$dis, $damerau = false)
         {
-            $dis_new=array();
-            for ($x=1;$x<=$ca;$x++) {
-                $dis_new[0]=$x;
-                for ($y=1;$y<=$cb;$y++) {
+            $dis_new = array();
+            for ($x = 1; $x <= $ca; $x++) {
+                $dis_new[0] = $x;
+                for ($y = 1; $y <= $cb; $y++) {
                     self::costMatrix($a, $b, $dis_new, $dis, $damerau, $x, $y);
                 }
                 $dis = $dis_new;
@@ -207,11 +207,11 @@ 
         
         private static function costMatrix(&$a, &$b, &$dis_new, &$dis, $damerau, $x, $y)
         {
-            $c = ($a[$x-1] == $b[$y-1])?0:1;
-            $dis_new[$y] = min($dis[$y]+1, $dis_new[$y-1]+1, $dis[$y-1]+$c);
+            $c = ($a[$x - 1] == $b[$y - 1]) ? 0 : 1;
+            $dis_new[$y] = min($dis[$y] + 1, $dis_new[$y - 1] + 1, $dis[$y - 1] + $c);
             if ($damerau) {
-                if ($x > 1 && $y > 1 && $a[$x-1] == $b[$y-2] && $a[$x-2] == $b[$y-1]) {
-                    $dis_new[$y]= min($dis_new[$y-1], $dis[$y-3] + $c) ;
+                if ($x > 1 && $y > 1 && $a[$x - 1] == $b[$y - 2] && $a[$x - 2] == $b[$y - 1]) {
+                    $dis_new[$y] = min($dis_new[$y - 1], $dis[$y - 3] + $c);
                 }
             }
         }

src/similar_text.php

Spacing +14 added lines, -14 removed lines

@@ -76,13 +76,13 @@ 
             return $stats;
         }
 
-        protected static function getParts($b, &$c = 0, $lengthCapture=false)
+        protected static function getParts($b, &$c = 0, $lengthCapture = false)
         {
             $parts = array();
             $tmp = '';
             $c = 0;
-            $length=0;
-            $lengthCapture=(bool)$lengthCapture;
+            $length = 0;
+            $lengthCapture = (bool) $lengthCapture;
             if ($lengthCapture) {
                 self::capturePartsWithLength($b, $length, $tmp, $c, $parts);
             } else {
@@ -114,17 +114,17 @@ 
             foreach ($b as $k=>$v) {
                 $length++;
                 if (ctype_space($v) || ctype_punct($v)) {
-                    $parts[] =array($tmp,$length-1);
-                    $parts[] = array($v,1);
+                    $parts[] = array($tmp, $length - 1);
+                    $parts[] = array($v, 1);
                     $c += 2;
                     $tmp = '';
-                    $length=0;
+                    $length = 0;
                     continue;
                 }
                 $tmp .= $v;
             }
             if (!empty($tmp)) {
-                $parts[] = array($tmp,$length);
+                $parts[] = array($tmp, $length);
                 $c++;
             }
         }
@@ -146,7 +146,7 @@ 
             if (is_array($split)) {
                 return
                     array_map(
-                        function ($val) {
+                        function($val) {
                             if (self::is_ascii($val)) {
                                 return strtolower($val);
                             }
@@ -161,16 +161,16 @@ 
             }
         }
         
-        protected static function split($str, $grams=false)
+        protected static function split($str, $grams = false)
         {
             if (!is_string($str)) {
                 return array();
             }
             static $split = [];
             static $old = '';
-            static $oldGrams=1;
-            $grams=is_int($grams) && $grams >=1 && $grams <= strlen($str) ? $grams : false;
-            if ($old === $str && $oldGrams===$grams) {
+            static $oldGrams = 1;
+            $grams = is_int($grams) && $grams >= 1 && $grams <= strlen($str) ? $grams : false;
+            if ($old === $str && $oldGrams === $grams) {
                 return $split;
             } else {
                 return self::_split($str, $split, $old, $oldGrams, $grams);
@@ -180,8 +180,8 @@ 
         private static function _split(&$str, &$split, &$old, &$oldGrams, $grams)
         {
             $old = $str;
-            $oldGrams=$grams;
-            $split = !$grams ? preg_split('//u', $str, -1, PREG_SPLIT_NO_EMPTY):preg_split('/(.{'.$grams.'})/su', $str, -1, PREG_SPLIT_NO_EMPTY|PREG_SPLIT_DELIM_CAPTURE);
+            $oldGrams = $grams;
+            $split = !$grams ? preg_split('//u', $str, -1, PREG_SPLIT_NO_EMPTY) : preg_split('/(.{'.$grams.'})/su', $str, -1, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_DELIM_CAPTURE);
             return $split;
         }
     }