manuwhat /
similar-text
@@ -151,7 +151,7 @@ |
||
| 151 | 151 | $split |
| 152 | 152 | ) |
| 153 | 153 | |
| 154 | - ; |
|
| 154 | + ; |
|
| 155 | 155 | } else { |
| 156 | 156 | return array(); |
| 157 | 157 | } |
@@ -12,10 +12,10 @@ discard block |
||
| 12 | 12 | function SimilarText( |
| 13 | 13 | $firstString, |
| 14 | 14 | $secondString, |
| 15 | - $round=2, |
|
| 16 | - $insensitive=true, |
|
| 17 | - &$stats=false, |
|
| 18 | - $getParts=false |
|
| 15 | + $round = 2, |
|
| 16 | + $insensitive = true, |
|
| 17 | + &$stats = false, |
|
| 18 | + $getParts = false |
|
| 19 | 19 | ) { |
| 20 | 20 | return EZAMA\similar_text::similarText( |
| 21 | 21 | $firstString, |
@@ -45,81 +45,81 @@ discard block |
||
| 45 | 45 | private function __construct() |
| 46 | 46 | { |
| 47 | 47 | } |
| 48 | - public static function similarText($a, $b, $round=2, $insensitive=true, &$stats=false, $getParts=false) |
|
| 48 | + public static function similarText($a, $b, $round = 2, $insensitive = true, &$stats = false, $getParts = false) |
|
| 49 | 49 | { |
| 50 | - if (!is_string($a)||!is_string($b)) { |
|
| 50 | + if (!is_string($a) || !is_string($b)) { |
|
| 51 | 51 | return false; |
| 52 | 52 | } |
| 53 | 53 | if ($insensitive) { |
| 54 | - $a=self::strtolower($a); |
|
| 55 | - $b=self::strtolower($b); |
|
| 54 | + $a = self::strtolower($a); |
|
| 55 | + $b = self::strtolower($b); |
|
| 56 | 56 | } else { |
| 57 | - $a=self::split($a); |
|
| 58 | - $b=self::split($b); |
|
| 57 | + $a = self::split($a); |
|
| 58 | + $b = self::split($b); |
|
| 59 | 59 | } |
| 60 | 60 | /* prevent bad types and useless memory usage due to for example array instead of simple boolean */ |
| 61 | 61 | unset($insensitive); |
| 62 | - $getParts=(bool)$getParts; |
|
| 62 | + $getParts = (bool) $getParts; |
|
| 63 | 63 | /* ******************************************************************************************** */ |
| 64 | - $ca=count($a); |
|
| 65 | - $cb=count($b); |
|
| 66 | - if ($ca<$cb) { |
|
| 67 | - $stats=self::getStats($cb, $a, self::_check($a, $b, $getParts, $round), $getParts, $round); |
|
| 64 | + $ca = count($a); |
|
| 65 | + $cb = count($b); |
|
| 66 | + if ($ca < $cb) { |
|
| 67 | + $stats = self::getStats($cb, $a, self::_check($a, $b, $getParts, $round), $getParts, $round); |
|
| 68 | 68 | } else { |
| 69 | - $stats=self::getStats($ca, $b, self::_check($b, $a, $getParts, $round), $getParts, $round); |
|
| 69 | + $stats = self::getStats($ca, $b, self::_check($b, $a, $getParts, $round), $getParts, $round); |
|
| 70 | 70 | } |
| 71 | 71 | return $stats['similar']; |
| 72 | 72 | } |
| 73 | 73 | |
| 74 | 74 | protected static function _check($a, $b, $getParts, $round) |
| 75 | 75 | { |
| 76 | - $diff=array(); |
|
| 76 | + $diff = array(); |
|
| 77 | 77 | if ($getParts) { |
| 78 | - $diff[]=array_diff($a, $b); |
|
| 79 | - $diff[]=array_diff($b, $a); |
|
| 78 | + $diff[] = array_diff($a, $b); |
|
| 79 | + $diff[] = array_diff($b, $a); |
|
| 80 | 80 | } |
| 81 | - $diff[]=array_intersect($a, $b); |
|
| 82 | - $diff[]=round(count(array_intersect(self::getParts($a, $c), self::getParts($b)))/$c*100, $round); |
|
| 83 | - $diff[]=$a===$b; |
|
| 81 | + $diff[] = array_intersect($a, $b); |
|
| 82 | + $diff[] = round(count(array_intersect(self::getParts($a, $c), self::getParts($b))) / $c * 100, $round); |
|
| 83 | + $diff[] = $a === $b; |
|
| 84 | 84 | return $diff; |
| 85 | 85 | } |
| 86 | 86 | |
| 87 | 87 | protected static function getStats($ca, $b, $diff, $getParts, $round) |
| 88 | 88 | { |
| 89 | - $stats=array(); |
|
| 89 | + $stats = array(); |
|
| 90 | 90 | if ($getParts) { |
| 91 | - $stats['similar']=round(count($diff[2])*100/$ca, $round); |
|
| 92 | - $stats['substr']=$diff[3]; |
|
| 93 | - $stats['contain']=($diff[2]===$b)?true:false; |
|
| 94 | - $stats['equal']=$diff[4]; |
|
| 95 | - $stats['a-b']=$diff[0]; |
|
| 96 | - $stats['b-a']=$diff[1]; |
|
| 97 | - $stats['a&b']=$diff[2]; |
|
| 91 | + $stats['similar'] = round(count($diff[2]) * 100 / $ca, $round); |
|
| 92 | + $stats['substr'] = $diff[3]; |
|
| 93 | + $stats['contain'] = ($diff[2] === $b) ?true:false; |
|
| 94 | + $stats['equal'] = $diff[4]; |
|
| 95 | + $stats['a-b'] = $diff[0]; |
|
| 96 | + $stats['b-a'] = $diff[1]; |
|
| 97 | + $stats['a&b'] = $diff[2]; |
|
| 98 | 98 | } else { |
| 99 | - $stats['similar']=round(count($diff[0])*100/$ca, $round); |
|
| 100 | - $stats['substr']=$diff[1]; |
|
| 101 | - $stats['contain']=($diff[0]===$b)?true:false; |
|
| 102 | - $stats['equal']=$diff[2]; |
|
| 99 | + $stats['similar'] = round(count($diff[0]) * 100 / $ca, $round); |
|
| 100 | + $stats['substr'] = $diff[1]; |
|
| 101 | + $stats['contain'] = ($diff[0] === $b) ?true:false; |
|
| 102 | + $stats['equal'] = $diff[2]; |
|
| 103 | 103 | } |
| 104 | 104 | return $stats; |
| 105 | 105 | } |
| 106 | 106 | |
| 107 | - protected static function getParts($b, &$c=0) |
|
| 107 | + protected static function getParts($b, &$c = 0) |
|
| 108 | 108 | { |
| 109 | - $parts=array(); |
|
| 110 | - $tmp=''; |
|
| 111 | - $c=0; |
|
| 109 | + $parts = array(); |
|
| 110 | + $tmp = ''; |
|
| 111 | + $c = 0; |
|
| 112 | 112 | foreach ($b as $k=>$v) { |
| 113 | - $tmp.=$v; |
|
| 113 | + $tmp .= $v; |
|
| 114 | 114 | if (ctype_space($v)) { |
| 115 | - $parts[]=$tmp; |
|
| 116 | - $parts[]=$v; |
|
| 117 | - $c+=2; |
|
| 118 | - $tmp=''; |
|
| 115 | + $parts[] = $tmp; |
|
| 116 | + $parts[] = $v; |
|
| 117 | + $c += 2; |
|
| 118 | + $tmp = ''; |
|
| 119 | 119 | } |
| 120 | 120 | } |
| 121 | 121 | if (!empty($tmp)) { |
| 122 | - $parts[]=$tmp; |
|
| 122 | + $parts[] = $tmp; |
|
| 123 | 123 | $c++; |
| 124 | 124 | } |
| 125 | 125 | return $parts; |
@@ -138,11 +138,11 @@ discard block |
||
| 138 | 138 | |
| 139 | 139 | protected static function strtolower($str) |
| 140 | 140 | { |
| 141 | - $split=self::split($str); |
|
| 141 | + $split = self::split($str); |
|
| 142 | 142 | if (is_array($split)) { |
| 143 | 143 | return |
| 144 | 144 | array_map( |
| 145 | - function ($val) { |
|
| 145 | + function($val) { |
|
| 146 | 146 | if (self::is_ascii($val)) { |
| 147 | 147 | return strtolower($val); |
| 148 | 148 | } |
@@ -162,25 +162,25 @@ discard block |
||
| 162 | 162 | if (!is_string($str)) { |
| 163 | 163 | return array(); |
| 164 | 164 | } |
| 165 | - static $split=[]; |
|
| 166 | - static $old=''; |
|
| 167 | - if ($old===$str) { |
|
| 165 | + static $split = []; |
|
| 166 | + static $old = ''; |
|
| 167 | + if ($old === $str) { |
|
| 168 | 168 | return $split; |
| 169 | 169 | } else { |
| 170 | - $old=$str; |
|
| 171 | - $split=preg_split('//u', $str, -1, PREG_SPLIT_NO_EMPTY); |
|
| 170 | + $old = $str; |
|
| 171 | + $split = preg_split('//u', $str, -1, PREG_SPLIT_NO_EMPTY); |
|
| 172 | 172 | return $split; |
| 173 | 173 | } |
| 174 | 174 | } |
| 175 | 175 | |
| 176 | 176 | public static function areAnagrams($a, $b) |
| 177 | 177 | { |
| 178 | - return self::similarText($a, $b, 2, true, $check)?$check['similar'] === 100.0&&$check['contain']===true:false; |
|
| 178 | + return self::similarText($a, $b, 2, true, $check) ? $check['similar'] === 100.0 && $check['contain'] === true : false; |
|
| 179 | 179 | } |
| 180 | 180 | |
| 181 | 181 | public static function similarButNotEqual($a, $b) |
| 182 | 182 | { |
| 183 | - return self::similarText($a, $b, 2, true, $check)&&is_array($check)&&$check['equal']===true?false:true; |
|
| 183 | + return self::similarText($a, $b, 2, true, $check) && is_array($check) && $check['equal'] === true ?false:true; |
|
| 184 | 184 | } |
| 185 | 185 | } |
| 186 | 186 | } |
