// Copyright (c) 2018 GeometryFactory Sarl (France). // All rights reserved. // // This file is part of CGAL (www.cgal.org). // You can redistribute it and/or modify it under the terms of the GNU // General Public License as published by the Free Software Foundation, // either version 3 of the License, or (at your option) any later version. // // Licensees holding a valid commercial license may use this file in // accordance with the commercial license agreement provided with the software. // // This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE // WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. // // $URL$ // $Id$ // SPDX-License-Identifier: GPL-3.0+ // // Author(s) : Simon Giraudot #ifndef CGAL_CLASSIFICATION_COMPRESSED_FLOAT_H #define CGAL_CLASSIFICATION_COMPRESSED_FLOAT_H #include namespace CGAL { namespace Classification { /// \cond SKIP_IN_MANUAL #if defined(CGAL_CLASSIFICATION_DO_NOT_COMPRESS_FLOATS) typedef float compressed_float; inline float compress_float (const float& f, const float&, const float&) { return f; } inline float decompress_float (const float& t, const float&, const float&) { return t; } #else # if defined(CGAL_CLASSIFICATION_COMPRESS_FLOATS_WITH_USHORT) typedef unsigned short compressed_float; # else // Default = compress with unsigned char typedef unsigned char compressed_float; # endif inline compressed_float compress_float (const float& f, const float& min = 0.f, const float& max = 1.f) { return static_cast (float(std::numeric_limits::max()) * (f - min) / (max - min)); } inline float decompress_float (const compressed_float& t, const float& min = 0.f, const float& max = 1.f) { return ((max - min) * (t / float(std::numeric_limits::max())) + min); } #endif /// \endcond } // namespace Classification } // namespace CGAL #endif // CGAL_CLASSIFICATION_COMPRESSED_FLOAT_H