public override bool Generate_statemap()
{
//Long filepath related variables, longPathName is used as cache for long filepath readed.
bool hasLongFilepath = false;
string stored_Longname = "";
//PAX Extended filename
bool hasPAXFileName = false;
string stored_PAX_Filename = "";
//PAX Global extended filename
bool hasPAXGlobalFileName = false;
string stored_PAX_Global_Filename = "";
using (FileStream fs = File.OpenRead(filepath))
using (BinaryReader reader = new BinaryReader(fs, Encoding.UTF8))
{
while (!Skip_emprty_blocks(reader) &&
reader.BaseStream.Position < reader.BaseStream.Length)
{
long start_position = reader.BaseStream.Position;
//check header integrity
if (!verify_header(reader))
{
throw new FileCorruoptedException(
$"The provided tar file might be corrupted (Header position: 0x{reader.BaseStream.Position.ToString("X")})");
}
reader.BaseStream.Position = start_position;
current_state = new TarState();
current_state.fullFilepath = ByteUtil.To_readable_string(reader.ReadBytes(100)).Trim('\0');
if (states.ContainsKey(current_state.fullFilepath))
{
throw new ParseErrorException("Dupelicated filepath found in tar file, There is currently no plan to support hardlink/softlink");
}
//Skip metadata that has no use to this program
//File mode, Owner's numeric user ID and Group's numeric user ID
reader.BaseStream.Position += 24;
//File size
current_state.size = Read_file_size(reader);
//Skip metadata that has no use to this program
//Last modification time in numeric Unix time format (octal) and Checksum for header record
reader.BaseStream.Position += 20;
//Type flag
char type_flag = reader.ReadChar();
if (type_flag == '\0')
{
type_flag = TarType.File;
}
if (type_flag == TarType.File || //File
type_flag == TarType.Directory || //Directory
type_flag == TarType.GNU_Longname || //GNU_Longname
type_flag == TarType.PAX_Extended_header || //PAX Extended header
type_flag == TarType.PAX_Global_extended_header) //PAX Global extended header
{
current_state.type = type_flag;
}
else
{
throw new NotSupportedException("This Tar file is currently not supported, type:" + type_flag);
}
//Skip metadata that has no use to this program
//Name of linked file
reader.BaseStream.Position += 100;
//This is the end of v7 header
//Check for ustar marker
if (ByteUtil.To_readable_string(reader.ReadBytes(6)) == "ustar" &&
reader.ReadByte() == char.MinValue)
{
//Skip metadata that has no use to this program
//UStar version "00", Owner user name, Owner group name, Device major number and Device minor number
reader.BaseStream.Position += 82;
//filepath prefix
current_state.fullFilepath = ByteUtil.To_readable_string(reader.ReadBytes(155)).Trim('\0') + current_state.fullFilepath;
if (states.ContainsKey(current_state.fullFilepath))
{
throw new ParseErrorException("Dupelicated filepath found in tar file, There is currently no plan to support hardlink/softlink");
}
}
//Restore saved file name if there is any.
if (hasLongFilepath)
{
current_state.fullFilepath = stored_Longname;
hasLongFilepath = false;
}
else if (hasPAXFileName)
{
current_state.fullFilepath = stored_PAX_Filename;
hasPAXFileName = false;
}
else if (hasPAXGlobalFileName)
{
current_state.fullFilepath = stored_PAX_Global_Filename;
}
//final processing
current_state.fullFilepath = VFile.Unify_filepath(current_state.fullFilepath);
long roundup_filesize = current_state.size;
if (current_state.size % 512 > 0)
{
roundup_filesize += 512 - current_state.size % 512;
}
current_state.offset_in_tar_file = start_position + 512; //512 == size of tar header roundup
reader.BaseStream.Position = current_state.offset_in_tar_file;
if (type_flag == TarType.GNU_Longname)
{
//read longPathName before we skip it.
hasLongFilepath = true;
//TODO: support long path loger than 2147483647 bytes (aka 2147483647 UTF-8 characters) (I doubt I will do this in a decade)
stored_Longname = ByteUtil.To_readable_string(reader.ReadBytes((int)current_state.size)).Trim('\0');
}
else if (type_flag == TarType.PAX_Extended_header)
{
//PAX Extended header
(bool, string)result = read_PAX_header(reader, current_state.size);
hasPAXFileName = result.Item1;
stored_PAX_Filename = result.Item2;
}
else if (type_flag == TarType.PAX_Global_extended_header)
{
//PAX Global extended header
(bool, string)result = read_PAX_header(reader, current_state.size);
hasPAXGlobalFileName = result.Item1;
stored_PAX_Global_Filename = result.Item2;
}
else
{
states.Add(current_state.fullFilepath, current_state);
}
reader.BaseStream.Position = current_state.offset_in_tar_file + roundup_filesize;
//Register the folder/file we read.
if (type_flag == TarType.Directory)
{
//This is a folder
VFolder folder = new VFolder(current_state.fullFilepath, current_state.size);
folderList.Add(folder.AbsolutePath, folder);
}
else if (type_flag == TarType.File)
{
//This is a file
VFile file = new VFile(current_state.fullFilepath, current_state.size);
folderList[file.Prefix].Files.Add(file);
fileList.Add(file.AbsolutePath, file);
}
else if (type_flag != TarType.GNU_Longname &&
type_flag != TarType.PAX_Extended_header &&
type_flag != TarType.PAX_Global_extended_header)
{
//It is also not other supported flag???
throw new FileCorruoptedException($"Unknown type flag detected {(char)type_flag}, it's probably (99.9%) due to developer not taken care of the code.");
}
}
return(true);
}
}