[question] Reading files with mmap for compression level "no compression"

[vadimkantorov]

Hi!

Am I right that for zip files with no compression, reading can be done simply via mmap after calculating offset and byte count?

Can miniz help in reading from such ZIPs: i.e. parsing the ZIP file tree and providing a requested file offset and byte size?

There is the mz_zip_archive_file_stat structure

miniz/miniz_zip.h

Lines 22 to 77 in 8714fd3

	typedef struct
	{
	/* Central directory file index. */
	mz_uint32 m_file_index;
	/* Byte offset of this entry in the archive's central directory. Note we currently only support up to UINT_MAX or less bytes in the central dir. */
	mz_uint64 m_central_dir_ofs;
	/* These fields are copied directly from the zip's central dir. */
	mz_uint16 m_version_made_by;
	mz_uint16 m_version_needed;
	mz_uint16 m_bit_flag;
	mz_uint16 m_method;
	/* CRC-32 of uncompressed data. */
	mz_uint32 m_crc32;
	/* File's compressed size. */
	mz_uint64 m_comp_size;
	/* File's uncompressed size. Note, I've seen some old archives where directory entries had 512 bytes for their uncompressed sizes, but when you try to unpack them you actually get 0 bytes. */
	mz_uint64 m_uncomp_size;
	/* Zip internal and external file attributes. */
	mz_uint16 m_internal_attr;
	mz_uint32 m_external_attr;
	/* Entry's local header file offset in bytes. */
	mz_uint64 m_local_header_ofs;
	/* Size of comment in bytes. */
	mz_uint32 m_comment_size;
	/* MZ_TRUE if the entry appears to be a directory. */
	mz_bool m_is_directory;
	/* MZ_TRUE if the entry uses encryption/strong encryption (which miniz_zip doesn't support) */
	mz_bool m_is_encrypted;
	/* MZ_TRUE if the file is not encrypted, a patch file, and if it uses a compression method we support. */
	mz_bool m_is_supported;
	/* Filename. If string ends in '/' it's a subdirectory entry. */
	/* Guaranteed to be zero terminated, may be truncated to fit. */
	char m_filename[MZ_ZIP_MAX_ARCHIVE_FILENAME_SIZE];
	/* Comment field. */
	/* Guaranteed to be zero terminated, may be truncated to fit. */
	char m_comment[MZ_ZIP_MAX_ARCHIVE_FILE_COMMENT_SIZE];
	#ifdef MINIZ_NO_TIME
	MZ_TIME_T m_padding;
	#else
	MZ_TIME_T m_time;
	#endif
	} mz_zip_archive_file_stat;

, but I need help figuring out how to get a byte offset from the ZIP file beginning from an instance of this structure (and if it's possible currently).

Or maybe this other mz_zip_reader_extract_iter_state structure could be used for this?

miniz/miniz_zip.h

Lines 186 to 208 in 8714fd3

	typedef struct
	{
	mz_zip_archive *pZip;
	mz_uint flags;
	int status;
	mz_uint64 read_buf_size, read_buf_ofs, read_buf_avail, comp_remaining, out_buf_ofs, cur_file_ofs;
	mz_zip_archive_file_stat file_stat;
	void *pRead_buf;
	void *pWrite_buf;
	size_t out_blk_remain;
	tinfl_decompressor inflator;
	#ifdef MINIZ_DISABLE_ZIP_READER_CRC32_CHECKS
	mz_uint padding;
	#else
	mz_uint file_crc32;
	#endif
	} mz_zip_reader_extract_iter_state;

Or is compression level "zero" not supported by miniz decompression routines? I unfortunately can't understand where in code the MZ_NO_COMPRESSION decompression case is treated :(

If it's not possible, adding such a field for "no compression" files would be nice for mmap-based file readers.

Thanks!