Ticket #22363: 22363.12.patch

src/wp-includes/default-filters.php

@@ -197 +197 @@
 add_filter( 'teeny_mce_before_init',    '_mce_set_direction'                  );
 add_filter( 'pre_kses',                 'wp_pre_kses_less_than'               );
 add_filter( 'sanitize_title',           'sanitize_title_with_dashes',   10, 3 );
+add_filter( 'sanitize_file_name',       'remove_accents'                      );
 add_action( 'check_comment_flood',      'check_comment_flood_db',       10, 3 );
 add_filter( 'comment_flood_filter',     'wp_throttle_comment_flood',    10, 3 );
 add_filter( 'pre_comment_content',      'wp_rel_nofollow',              15    );

src/wp-includes/formatting.php

@@ -1764 +1764 @@
          * @param string $filename_raw  Filename as it was passed into sanitize_file_name().
          */
         $special_chars = apply_filters( 'sanitize_file_name_chars', $special_chars, $filename_raw );
-        $filename = preg_replace( "#\x{00a0}#siu", ' ', $filename );
         $filename = str_replace( $special_chars, '', $filename );
-        $filename = str_replace( array( '%20', '+' ), '-', $filename );
+        if ( ! seems_utf8( $filename ) ) {
+                // Reduce to ASCII.
+                $filename = preg_replace( '/[\x80-\xff]/', '-', $filename );
+        }
+        $filename = str_replace( array( '%20', '+', "\xc2\xa0" ), '-', $filename );
         $filename = preg_replace( '/[\r\n\t -]+/', '-', $filename );
         $filename = trim( $filename, '.-_' );
 

tests/phpunit/tests/formatting/SanitizeFileName.php

@@ -47 +47 @@
 
         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"));
         }
 
         /**
@@ -67 +67 @@
                 // Test a filenames that becomes extensionless.
                 $this->assertEquals( 'no-extension', sanitize_file_name( '_.no-extension' ) );
         }
+
+        /**
+         * @ticket 22363
+         */
+        function test_replaces_utf8_whitespace() {
+                $this->assertEquals("non-breaking", sanitize_file_name("non\xc2\xa0breaking"));
+        }
+
+        /**
+         * @ticket 22363
+         */
+        function test_replaces_accents() {
+                $in  = 'àáâãäåæçèéêëìíîïñòóôõöøùúûüýÿ';
+                $out = 'aaaaaaaeceeeeiiiinoooooouuuuyy';
+                $this->assertEquals( $out, sanitize_file_name( $in ) );
+        }
+
+        /**
+         * @ticket 22363
+         * @ticket 24661
+         */
+        /* Requires #24661 fix
+        function test_replaces_combining_accents() {
+                $in  = 'àáâãäåçèéêëìíîïñòóôõöùúûüýÿ';
+                $out = 'aaaaaaceeeeiiiinooooouuuuyy';
+                $this->assertEquals( $out, sanitize_file_name( $in ) );
+        }
+        */
+
+        /**
+         * @ticket 22363
+         */
+        function test_strips_non_alpha_html_entities() {
+                $this->assertEquals("start-copy-amp-euro-ndash-nbsp-end", sanitize_file_name( "start © & € –   end" ) );
+                $this->assertEquals("start-invalid-end", sanitize_file_name( "start &invalid; end" ) );
+        }
+
+        /**
+         * @ticket 22363
+         */
+        function test_converts_alpha_html_entities() {
+                $this->assertEquals("start-agrave-ecirc-ntilde-oslash-aelig-end", sanitize_file_name("start à ê ñ ø æ end" ) );
+        }
+
+        /**
+         * @ticket 22363
+         */
+        function test_removes_some_non_alphanum_characters() {
+                $test_input = "start ¿”©“? «±€—°» middle “√¼ = ♫” & „☺ + ☻ = ♥‟ end";
+                $this->assertEquals("start-¿”©“-«±E—°»-middle-“√¼-♫”-„☺-☻-♥‟-end", sanitize_file_name( $test_input ) );
+        }
+
+        /**
+         * @ticket 22363
+         */
+        function test_removes_nonascii_on_invalid_utf8() {
+                $this->assertEquals( 'Invalid-utf8', sanitize_file_name( "Invalid \xff utf8" ) );
+        }
 }