| 1 | //===----------------------------------------------------------------------===// |
|---|---|
| 2 | // DuckDB |
| 3 | // |
| 4 | // duckdb/common/gzip_file_system.hpp |
| 5 | // |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | |
| 9 | #pragma once |
| 10 | |
| 11 | #include "duckdb/common/compressed_file_system.hpp" |
| 12 | |
| 13 | namespace duckdb { |
| 14 | |
| 15 | class GZipFileSystem : public CompressedFileSystem { |
| 16 | // 32 KB |
| 17 | static constexpr const idx_t BUFFER_SIZE = 1u << 15; |
| 18 | |
| 19 | public: |
| 20 | unique_ptr<FileHandle> OpenCompressedFile(unique_ptr<FileHandle> handle, bool write) override; |
| 21 | |
| 22 | std::string GetName() const override { |
| 23 | return "GZipFileSystem"; |
| 24 | } |
| 25 | |
| 26 | //! Verifies that a buffer contains a valid GZIP header |
| 27 | static void VerifyGZIPHeader(uint8_t gzip_hdr[], idx_t read_count); |
| 28 | //! Consumes a byte stream as a gzip string, returning the decompressed string |
| 29 | static string UncompressGZIPString(const string &in); |
| 30 | |
| 31 | unique_ptr<StreamWrapper> CreateStream() override; |
| 32 | idx_t InBufferSize() override; |
| 33 | idx_t OutBufferSize() override; |
| 34 | }; |
| 35 | |
| 36 | static constexpr const uint8_t GZIP_COMPRESSION_DEFLATE = 0x08; |
| 37 | |
| 38 | static constexpr const uint8_t GZIP_FLAG_ASCII = 0x1; |
| 39 | static constexpr const uint8_t GZIP_FLAG_MULTIPART = 0x2; |
| 40 | static constexpr const uint8_t GZIP_FLAG_EXTRA = 0x4; |
| 41 | static constexpr const uint8_t GZIP_FLAG_NAME = 0x8; |
| 42 | static constexpr const uint8_t GZIP_FLAG_COMMENT = 0x10; |
| 43 | static constexpr const uint8_t GZIP_FLAG_ENCRYPT = 0x20; |
| 44 | |
| 45 | static constexpr const uint8_t GZIP_HEADER_MINSIZE = 10; |
| 46 | // MAXSIZE should be the same as input buffer size |
| 47 | static constexpr const idx_t GZIP_HEADER_MAXSIZE = 1u << 15; |
| 48 | static constexpr const uint8_t GZIP_FOOTER_SIZE = 8; |
| 49 | |
| 50 | static constexpr const unsigned char GZIP_FLAG_UNSUPPORTED = |
| 51 | GZIP_FLAG_ASCII | GZIP_FLAG_MULTIPART | GZIP_FLAG_COMMENT | GZIP_FLAG_ENCRYPT; |
| 52 | |
| 53 | } // namespace duckdb |
| 54 |