Ticket #22363: test-SanitizeFileName.patch

tests/phpunit/tests/formatting/SanitizeFileName.php

@@ -16 +16 @@
                 foreach ( $special_chars as $char )
                         $string .= $char;
                 $string .= 'test';
-                $this->assertEquals( 'testtest', sanitize_file_name( $string ) );
+                $this->assertEquals( 'test-test', sanitize_file_name( $string ) );
         }
 
         function test_replaces_any_number_of_hyphens_with_one_hyphen() {
@@ -29 +29 @@
 
         function test_replaces_any_amount_of_whitespace_with_one_hyphen() {
                 $this->assertEquals("a-t", sanitize_file_name("a          t"));
-                $this->assertEquals("a-t", sanitize_file_name("a    \n\n\nt"));
+                $this->assertEquals("a-t", sanitize_file_name("a    \t\r\n\n\nt"));
         }
+        
+        function pcre_utf8_support() {
+                static $support = null;
+                if ( is_null($support) )
+                        $support = ( 1 === @preg_match( '`[\p{L}]`u', "\xc3\xa0" ) );
+                return $support;
+        }
+        
+        function test_replaces_utf8_whitespace() {
+                if ( ! $this->pcre_utf8_support() )
+                        $this->markTestSkipped();
+                $this->assertEquals("non-breaking", sanitize_file_name("non\xc2\xa0breaking"));
+        }
+        
+        function test_returns_lowercase() {
+                if ( ! function_exists( 'mb_strtolower' ) )
+                        $this->markTestSkipped();
+                $this->assertEquals( 'abcd', sanitize_file_name('ABCD') );
+        }
+        
+        function test_replaces_accents() {
+                $in  = 'àáâãäåæçèéêëìíîïñòóôõöøùúûüýÿ';
+                $out = 'aaaaaaaeceeeeiiiinoooooouuuuyy';
+                $this->assertEquals( $out, sanitize_file_name( $in ) );
+        }
+        
+        function test_strips_non_alpha_html_entities() {
+                if ( ! $this->pcre_utf8_support() )
+                        $this->markTestSkipped();
+                $this->assertEquals("start-end", sanitize_file_name( "start © & € –   end" ) );
+                $this->assertEquals("start-invalid-end", sanitize_file_name( "start &invalid; end" ) );
+        }
+        
+        function test_converts_alpha_html_entities() {
+                if ( ! $this->pcre_utf8_support() )
+                        $this->markTestSkipped();
+                $this->assertEquals("start-a-e-n-o-ae-end", sanitize_file_name("start à ê ñ ø æ end" ) );
+        }
+        
+        function test_removes_non_alphanum_characters() {
+                if ( ! $this->pcre_utf8_support() )
+                        $this->markTestSkipped();
+                $test_input = "start ¿”©“? «±€—°» middle “√¼ = ♫” & „☺ + ☻ = ♥‟ end";
+                $this->assertEquals("start-middle-end", sanitize_file_name( $test_input ) );
+        }
+        
+        function test_returns_empty_on_invalid_utf8() {
+                if ( ! $this->pcre_utf8_support() )
+                        $this->markTestSkipped();
+                $this->assertEquals( '', sanitize_file_name( "Invalid \xff utf8" ) );
+        }
+        
+        function test_returns_lowercase_utf8() {
+                if ( ! $this->pcre_utf8_support() || ! function_exists( 'mb_strtolower' ) )
+                        $this->markTestSkipped();
+                $this->assertEquals( 'βασιλ-правах', sanitize_file_name( 'ΒΑΣΙΛ ПРАВАХ' ) );
+        }
 }