LibDEFLATE: Add GZip support

This allows compressing and decompressing with data using GZip headers
and footers

LibDEFLATE/Compressor.cpp

@@ -580,6 +580,8 @@ namespace LibDEFLATE
 
 	BAN::ErrorOr<BAN::Vector<uint8_t>> Compressor::compress()
 	{
+		const uint32_t data_size = m_data.size();
+
 		uint32_t checksum = 0;
 		switch (m_type)
 		{
@@ -590,6 +592,21 @@ namespace LibDEFLATE
 				TRY(m_stream.write_bits(0x9C, 8)); // default compression
 				checksum = calculate_adler32(m_data);
 				break;
+			case StreamType::GZip:
+			{
+				const time_t current_time = time(nullptr);
+				TRY(m_stream.write_bits(0x1F, 8)); // ID1
+				TRY(m_stream.write_bits(0x8B, 8)); // ID2
+				TRY(m_stream.write_bits(8,    8)); // CM (deflate)
+				TRY(m_stream.write_bits(0,    8)); // FLG
+				TRY(m_stream.write_bits(current_time >>  0, 8)); // MTIME
+				TRY(m_stream.write_bits(current_time >>  8, 8));
+				TRY(m_stream.write_bits(current_time >> 16, 8));
+				TRY(m_stream.write_bits(current_time >> 24, 8));
+				TRY(m_stream.write_bits(0,    8)); // XFL
+				TRY(m_stream.write_bits(3,    8)); // OS (Unix)
+				checksum = calculate_crc32(m_data);
+			}
 		}
 
 		constexpr size_t max_block_size = 16 * 1024;
@@ -612,6 +629,16 @@ namespace LibDEFLATE
 				TRY(m_stream.write_bits(checksum >>  8, 8));
 				TRY(m_stream.write_bits(checksum >>  0, 8));
 				break;
+			case StreamType::GZip:
+				TRY(m_stream.write_bits(checksum >>  0, 8));
+				TRY(m_stream.write_bits(checksum >>  8, 8));
+				TRY(m_stream.write_bits(checksum >> 16, 8));
+				TRY(m_stream.write_bits(checksum >> 24, 8));
+				TRY(m_stream.write_bits(data_size >>  0, 8));
+				TRY(m_stream.write_bits(data_size >>  8, 8));
+				TRY(m_stream.write_bits(data_size >> 16, 8));
+				TRY(m_stream.write_bits(data_size >> 24, 8));
+				break;
 		}
 
 		return m_stream.take_buffer();

LibDEFLATE/Decompressor.cpp

@@ -127,6 +127,58 @@ namespace LibDEFLATE
 					TRY(m_stream.take_bits(16));
 				}
 
+				return {};
+			}
+			case StreamType::GZip:
+			{
+				const uint8_t id1 = TRY(m_stream.take_bits(8));
+				const uint8_t id2 = TRY(m_stream.take_bits(8));
+				if (id1 != 0x1F || id2 != 0x8B)
+				{
+					dwarnln("gzip header invalid identification");
+					return BAN::Error::from_errno(EINVAL);
+
+				}
+
+				const uint8_t cm = TRY(m_stream.take_bits(8));
+				if (cm != 8)
+				{
+					dwarnln("gzip does not use DEFLATE");
+					return BAN::Error::from_errno(EINVAL);
+				}
+
+				const uint8_t flg = TRY(m_stream.take_bits(8));
+				
+				TRY(m_stream.take_bits(16)); // mtime
+				TRY(m_stream.take_bits(16));
+
+				TRY(m_stream.take_bits(8)); // xfl
+
+				TRY(m_stream.take_bits(8)); // os
+
+				// extra fields
+				if (flg & (1 << 2))
+				{
+					const uint16_t xlen = TRY(m_stream.take_bits(16));
+					for (size_t i = 0; i < xlen; i++)
+						TRY(m_stream.take_bits(8));
+				}
+
+				// file name
+				if (flg & (1 << 3))
+					while (TRY(m_stream.take_bits(8)) != '\0')
+						continue;
+
+				// file comment
+				if (flg & (1 << 4))
+					while (TRY(m_stream.take_bits(8)) != '\0')
+						continue;
+
+				// crc16
+				// TODO: validate
+				if (flg & (1 << 1))
+					TRY(m_stream.take_bits(16));
+
 				return {};
 			}
 		}
@@ -154,6 +206,32 @@ namespace LibDEFLATE
 					return BAN::Error::from_errno(EINVAL);
 				}
 
+				return {};
+			}
+			case StreamType::GZip:
+			{
+				m_stream.skip_to_byte_boundary();
+
+				const uint32_t crc32 =
+					static_cast<uint32_t>(TRY(m_stream.take_bits(16))) |
+					static_cast<uint32_t>(TRY(m_stream.take_bits(16))) << 16;
+
+				if (crc32 != calculate_crc32(m_output.span()))
+				{
+					dwarnln("gzip final crc32 checksum failed");
+					return BAN::Error::from_errno(EINVAL);
+				}
+
+				const uint32_t isize =
+					static_cast<uint32_t>(TRY(m_stream.take_bits(16))) |
+					static_cast<uint32_t>(TRY(m_stream.take_bits(16))) << 16;
+
+				if (isize != m_output.size() % UINT32_MAX)
+				{
+					dwarnln("gzip final isize does not match {} vs {}", isize, m_output.size());
+					return BAN::Error::from_errno(EINVAL);
+				}
+
 				return {};
 			}
 		}

LibDEFLATE/include/LibDEFLATE/StreamType.h

@@ -7,6 +7,7 @@ namespace LibDEFLATE
 	{
 		Raw,
 		Zlib,
+		GZip,
 	};
 
 }

LibDEFLATE/include/LibDEFLATE/Utils.h

@@ -19,6 +19,25 @@ namespace LibDEFLATE
 		return (s2 << 16) | s1;
 	}
 
+	inline uint32_t calculate_crc32(BAN::ConstByteSpan data)
+	{
+		uint32_t crc32 = 0xFFFFFFFF;
+		uint32_t polynomial = 0xEDB88320;
+
+		for (size_t i = 0; i < data.size(); i++) {
+			crc32 ^= data[i];
+
+			for (size_t j = 0; j < 8; j++) {
+				if (crc32 & 1)
+					crc32 = (crc32 >> 1) ^ polynomial;
+				else
+					crc32 >>= 1;
+			}
+		}
+
+		return ~crc32;
+	}
+
 	inline constexpr uint16_t reverse_bits(uint16_t value, size_t count)
 	{
 		uint16_t reverse = 0;