- fix related to buggy full-case-folding used in grapheme_stri*()

pborreli · Apr 25, 2012 · 152f700 · 152f700
1 parent f88c902
commit 152f700
Show file tree

Hide file tree

Showing 4 changed files with 43 additions and 3 deletions.
diff --git a/class/Patchwork/PHP/Override/Intl.php b/class/Patchwork/PHP/Override/Intl.php
@@ -124,6 +124,7 @@ static function grapheme_strstr  ($s, $needle, $before_needle = false) {return m
     protected static function grapheme_position($s, $needle, $offset, $mode)
     {
         if ($offset > 0) $s = (string) self::grapheme_substr($s, $offset);
+        else if ($offset < 0) $offset = 0;
         if ('' === (string) $needle) return false;
         if ('' === (string) $s) return false;
 

diff --git a/class/Patchwork/Utf8.php b/class/Patchwork/Utf8.php
@@ -141,11 +141,29 @@ static function substr($s, $start, $len = 2147483647)
 
     static function strlen($s) {return grapheme_strlen($s);}
     static function strpos  ($s, $needle, $offset = 0) {return grapheme_strpos  ($s, $needle, $offset);}
-    static function stripos ($s, $needle, $offset = 0) {return grapheme_stripos ($s, $needle, $offset);}
     static function strrpos ($s, $needle, $offset = 0) {return grapheme_strrpos ($s, $needle, $offset);}
-    static function strripos($s, $needle, $offset = 0) {return grapheme_strripos($s, $needle, $offset);}
+
+    static function stripos ($s, $needle, $offset = 0)
+    {
+        if ($offset < 0) $offset = 0;
+        if (!$needle = mb_stripos($s, $needle, $offset, 'UTF-8')) return $needle;
+        return grapheme_strlen(iconv_substr($s, 0, $needle, 'UTF-8'));
+    }
+
+    static function strripos($s, $needle, $offset = 0)
+    {
+        if ($offset < 0) $offset = 0;
+        if (!$needle = mb_strripos($s, $needle, $offset, 'UTF-8')) return $needle;
+        return grapheme_strlen(iconv_substr($s, 0, $needle, 'UTF-8'));
+    }
+
+    static function stristr ($s, $needle, $before_needle = false)
+    {
+        if ('' == (string) $needle) return false;
+        return mb_stristr($s, $needle, $before_needle, 'UTF-8');
+    }
+
     static function strstr  ($s, $needle, $before_needle = false) {return grapheme_strstr ($s, $needle, $before_needle);}
-    static function stristr ($s, $needle, $before_needle = false) {return grapheme_stristr($s, $needle, $before_needle);}
     static function strrchr ($s, $needle, $before_needle = false) {return mb_strrchr ($s, $needle, $before_needle, 'UTF-8');}
     static function strrichr($s, $needle, $before_needle = false) {return mb_strrichr($s, $needle, $before_needle, 'UTF-8');}
 

diff --git a/tests/Patchwork/Tests/PHP/Override/IntlTest.php b/tests/Patchwork/Tests/PHP/Override/IntlTest.php
@@ -89,6 +89,13 @@ function testGrapheme_substr()
             $this->assertSame( "", grapheme_substr($c, -1,  0) );
             $this->assertSame( false, grapheme_substr($c,  1, -4) );
         }
+        else
+        {
+            // Expected fail: this buggy behavior has been fixed in PHP5.4
+            $this->assertSame( false, grapheme_substr($c, -2,  3) );
+            $this->assertSame( false, grapheme_substr($c, -1,  0) );
+            $this->assertSame( 'éjà', grapheme_substr($c,  1, -4) );
+        }
         $this->assertSame( "j", grapheme_substr($c, -2, -1) );
         $this->assertSame( "", grapheme_substr($c, -2, -2) );
         $this->assertSame( false, grapheme_substr($c,  5,  0) );
@@ -143,6 +150,7 @@ function testGrapheme_strpos()
         $this->assertSame( false, p::grapheme_strrpos('한국어', '') );
         $this->assertSame( 1, p::grapheme_strrpos('한국어', '국') );
         $this->assertSame( 3, p::grapheme_strripos('DÉJÀ', 'à') );
+        $this->assertSame( 16, p::grapheme_stripos('der Straße nach Paris', 'Paris') );
     }
 
     /**
@@ -156,5 +164,12 @@ function testGrapheme_strstr()
 
         $this->assertSame( '국어', p::grapheme_strstr('한국어', '국') );
         $this->assertSame( 'ÉJÀ', p::grapheme_stristr('DÉJÀ', 'é') );
+        $this->assertSame( 'Paris', p::grapheme_stristr('der Straße nach Paris', 'Paris') );
+    }
+
+    function testGrapheme_bugs()
+    {
+        $this->assertSame( 17, grapheme_stripos('der Straße nach Paris', 'Paris') ); // Expected fail: the non-bugged result is 16
+        $this->assertSame( 'aris', grapheme_stristr('der Straße nach Paris', 'Paris') );  // Expected fail: the non-bugged result is Paris
     }
 }
diff --git a/tests/Patchwork/Tests/Utf8Test.php b/tests/Patchwork/Tests/Utf8Test.php
@@ -159,9 +159,12 @@ function testStrpos()
         $this->assertSame( 0, u::strpos('abc', 'a', -1) );
         $this->assertSame( 1, u::strpos('한국어', '국') );
         $this->assertSame( 3, u::stripos('DÉJÀ', 'à') );
+        $this->assertSame( 1, u::stripos('aςσb', 'ΣΣ') );
         $this->assertSame( false, u::strrpos('한국어', '') );
         $this->assertSame( 1, u::strrpos('한국어', '국') );
         $this->assertSame( 3, u::strripos('DÉJÀ', 'à') );
+        $this->assertSame( 1, u::strripos('aςσb', 'ΣΣ') );
+        $this->assertSame( 16, u::stripos('der Straße nach Paris', 'Paris') );
     }
 
     /**
@@ -174,13 +177,16 @@ function testStrstr()
     {
         $this->assertSame( 'éjàdéjà', u::strstr('déjàdéjà', 'é') );
         $this->assertSame( 'ÉJÀDÉJÀ', u::stristr('DÉJÀDÉJÀ', 'é') );
+        $this->assertSame( 'ςσb', u::stristr('aςσb', 'ΣΣ') );
         $this->assertSame( 'éjà', u::strrchr('déjàdéjà', 'é') );
         $this->assertSame( 'ÉJÀ', u::strrichr('DÉJÀDÉJÀ', 'é') );
 
         $this->assertSame( 'd', u::strstr('déjàdéjà', 'é', true) );
         $this->assertSame( 'D', u::stristr('DÉJÀDÉJÀ', 'é', true) );
+        $this->assertSame( 'a', u::stristr('aςσb', 'ΣΣ', true) );
         $this->assertSame( 'déjàd', u::strrchr('déjàdéjà', 'é', true) );
         $this->assertSame( 'DÉJÀD', u::strrichr('DÉJÀDÉJÀ', 'é', true) );
+        $this->assertSame( 'Paris', u::stristr('der Straße nach Paris', 'Paris') );
     }
 
     /**