Ticket #32279: 32279.2.diff

src/wp-includes/wp-db.php

@@ -2629 +2629 @@
 
                         if ( is_array( $value['length'] ) ) {
                                 $length = $value['length']['length'];
+                                $truncate_by_byte_length = 'byte' === $value['length']['type'];
                         } else {
                                 $length = false;
+                                // Since we have no length, we'll never truncate.
+                                // Initialize the variable to false. true would take us
+                                // through an unnecessary (for this case) codepath below.
+                                $truncate_by_byte_length = false;
                         }
 
                         // There's no charset to work with.
@@ -2643 +2648 @@
                                 continue;
                         }
 
-                        $truncate_by_byte_length = 'byte' === $value['length']['type'];
-
                         $needs_validation = true;
                         if (
                                 // latin1 can store any byte sequence
@@ -2714 +2717 @@
                                                 $queries[ $value['charset'] ] = array();
                                         }
 
-                                        // We're going to need to truncate by characters or bytes, depending on the length value we have.
-                                        if ( 'byte' === $value['length']['type'] ) {
-                                                // Split the CONVERT() calls by charset, so we can make sure the connection is right
-                                                $queries[ $value['charset'] ][ $col ] = $this->prepare( "CONVERT( LEFT( CONVERT( %s USING binary ), %d ) USING {$value['charset']} )", $value['value'], $value['length']['length'] );
+                                        if ( is_array( $value['length'] ) ) {
+                                                // We're going to need to truncate by characters or bytes, depending on the length value we have.
+                                                if ( 'byte' === $value['length']['type'] ) {
+                                                        // Split the CONVERT() calls by charset, so we can make sure the connection is right
+                                                        $queries[ $value['charset'] ][ $col ] = $this->prepare( "CONVERT( LEFT( CONVERT( %s USING binary ), %d ) USING {$value['charset']} )", $value['value'], $value['length']['length'] );
+                                                } else {
+                                                        $queries[ $value['charset'] ][ $col ] = $this->prepare( "LEFT( CONVERT( %s USING {$value['charset']} ), %d )", $value['value'], $value['length']['length'] );
+                                                }
                                         } else {
-                                                $queries[ $value['charset'] ][ $col ] = $this->prepare( "LEFT( CONVERT( %s USING {$value['charset']} ), %d )", $value['value'], $value['length']['length'] );
+                                                $queries[ $value['charset'] ][ $col ] = $this->prepare( "CONVERT( %s USING {$value['charset']} )", $value['value'] );
                                         }
 
                                         unset( $data[ $col ]['db'] );

tests/phpunit/tests/db/charset.php

@@ -214 +214 @@
                                 'expected' => "\xd8ord\xd0ress",
                                 'length'   => array( 'type' => 'char', 'length' => 100 ),
                         ),
+                        'cp1251_no_length' => array(
+                                'charset'  => 'cp1251',
+                                'value'    => "\xd8ord\xd0ress",
+                                'expected' => "\xd8ord\xd0ress",
+                                'length'   => false,
+                        ),
+                        'cp1251_no_length_ascii' => array(
+                                'charset'  => 'cp1251',
+                                'value'    => "WordPress",
+                                'expected' => "WordPress",
+                                'length'   => false,
+                                // Don't set 'ascii' => false.  That's a different codepath than having only ASCII in the value.
+                        ),
                         'cp1251_char_length' => array(
                                 'charset'  => 'cp1251',
                                 'value'    => str_repeat( "\xd8\xd0", 10 ),
@@ -774 +787 @@
                 $stripped = $wpdb->strip_invalid_text_for_column( $wpdb->comments, 'comment_agent', str_repeat( 'A', 256 ) );
                 $this->assertEquals( 255, strlen( $stripped ) );
         }
+
+        /**
+         * @ticket 32279
+         */
+        function test_strip_invalid_text_from_query_cp1251_is_safe() {
+                $tablename = 'test_cp1251_query_' . rand_str( 5 );
+                if ( ! self::$_wpdb->query( "CREATE TABLE $tablename ( a VARCHAR(50) ) DEFAULT CHARSET 'cp1251'" ) ) {
+                        $this->markTestSkipped( "Test requires the 'cp1251' charset" );
+                }
+
+                $safe_query = "INSERT INTO $tablename( `a` ) VALUES( 'safe data' )";
+                $stripped_query = self::$_wpdb->strip_invalid_text_from_query( $safe_query );
+
+                self::$_wpdb->query( "DROP TABLE $tablename" );
+
+                $this->assertEquals( $safe_query, $stripped_query );
+        }
 }