// Reads the long path found in the data section of a GNU entry of type 'K' or 'L'
// and replaces Name or LinkName, respectively, with the found string.
// Throws if end of stream is reached.
private void ReadGnuLongPathDataBlock(Stream archiveStream)
{
Debug.Assert(_typeFlag is TarEntryType.LongLink or TarEntryType.LongPath);
if (_size == 0)
{
return;
}
byte[] buffer = new byte[(int)_size];
if (archiveStream.Read(buffer.AsSpan()) != _size)
{
throw new EndOfStreamException();
}
string longPath = TarHelpers.GetTrimmedUtf8String(buffer);
if (_typeFlag == TarEntryType.LongLink)
{
_linkName = longPath;
}
else if (_typeFlag == TarEntryType.LongPath)
{
_name = longPath;
}
}
// Reads the ustar prefix attribute.
// Throws if a conversion to an expected data type fails.
private void ReadUstarAttributes(Span <byte> buffer)
{
_prefix = TarHelpers.GetTrimmedUtf8String(buffer.Slice(FieldLocations.Prefix, FieldLengths.Prefix));
// In ustar, Prefix is used to store the *leading* path segments of
// Name, if the full path did not fit in the Name byte array.
if (!string.IsNullOrEmpty(_prefix))
{
// Prefix never has a leading separator, so we add it
// it should always be a forward slash for compatibility
_name = string.Format(UstarPrefixFormat, _prefix, _name);
}
}
// Collects the GNU long path info from the buffer and sets it in the right field depending on the type flag.
private void ReadGnuLongPathDataFromBuffer(ReadOnlySpan <byte> buffer)
{
string longPath = TarHelpers.GetTrimmedUtf8String(buffer);
if (_typeFlag == TarEntryType.LongLink)
{
_linkName = longPath;
}
else if (_typeFlag == TarEntryType.LongPath)
{
_name = longPath;
}
}
// Returns a dictionary containing the extended attributes collected from the provided byte buffer.
private void ReadExtendedAttributesFromBuffer(ReadOnlySpan <byte> buffer, string name)
{
string dataAsString = TarHelpers.GetTrimmedUtf8String(buffer);
using StringReader reader = new(dataAsString);
while (TryGetNextExtendedAttribute(reader, out string?key, out string?value))
{
if (ExtendedAttributes.ContainsKey(key))
{
throw new FormatException(string.Format(SR.TarDuplicateExtendedAttribute, name));
}
ExtendedAttributes.Add(key, value);
}
}
// Collects the extended attributes found in the data section of a PAX entry of type 'x' or 'g'.
// Throws if end of stream is reached or if an attribute is malformed.
private void ReadExtendedAttributesBlock(Stream archiveStream)
{
Debug.Assert(_typeFlag is TarEntryType.ExtendedAttributes or TarEntryType.GlobalExtendedAttributes);
// Regardless of the size, this entry should always have a valid dictionary object
_extendedAttributes ??= new Dictionary <string, string>();
if (_size == 0)
{
return;
}
// It is not expected that the extended attributes data section will be longer than int.MaxValue, considering
// 4096 is a common max path length, and also the size field is 12 bytes long, which is under int.MaxValue.
if (_size > int.MaxValue)
{
throw new InvalidOperationException(string.Format(SR.TarSizeFieldTooLargeForExtendedAttribute, _typeFlag.ToString()));
}
byte[] buffer = new byte[(int)_size];
if (archiveStream.Read(buffer.AsSpan()) != _size)
{
throw new EndOfStreamException();
}
string dataAsString = TarHelpers.GetTrimmedUtf8String(buffer);
using StringReader reader = new(dataAsString);
while (TryGetNextExtendedAttribute(reader, out string?key, out string?value))
{
_extendedAttributes ??= new Dictionary <string, string>();
if (_extendedAttributes.ContainsKey(key))
{
throw new FormatException(string.Format(SR.TarDuplicateExtendedAttribute, _name));
}
_extendedAttributes.Add(key, value);
}
}
// Reads the long path found in the data section of a GNU entry of type 'K' or 'L'
// and replaces Name or LinkName, respectively, with the found string.
// Throws if end of stream is reached.
private void ReadGnuLongPathDataBlock(Stream archiveStream)
{
Debug.Assert(_typeFlag is TarEntryType.LongLink or TarEntryType.LongPath);
if (_size == 0)
{
return;
}
byte[] buffer = new byte[(int)_size];
archiveStream.ReadExactly(buffer);
string longPath = TarHelpers.GetTrimmedUtf8String(buffer);
if (_typeFlag == TarEntryType.LongLink)
{
_linkName = longPath;
}
else if (_typeFlag == TarEntryType.LongPath)
{
_name = longPath;
}
}
// Attempts to read the fields shared by all formats and stores them in their expected data type.
// Throws if any data type conversion fails.
// Returns true on success, false if checksum is zero.
private static TarHeader?TryReadCommonAttributes(Span <byte> buffer, TarEntryFormat initialFormat)
{
// Start by collecting fields that need special checks that return early when data is wrong
// Empty checksum means this is an invalid (all blank) entry, finish early
Span <byte> spanChecksum = buffer.Slice(FieldLocations.Checksum, FieldLengths.Checksum);
if (TarHelpers.IsAllNullBytes(spanChecksum))
{
return(null);
}
int checksum = (int)TarHelpers.ParseOctal <uint>(spanChecksum);
// Zero checksum means the whole header is empty
if (checksum == 0)
{
return(null);
}
long size = (int)TarHelpers.ParseOctal <uint>(buffer.Slice(FieldLocations.Size, FieldLengths.Size));
if (size < 0)
{
throw new FormatException(string.Format(SR.TarSizeFieldNegative));
}
// Continue with the rest of the fields that require no special checks
TarHeader header = new(initialFormat,
name : TarHelpers.GetTrimmedUtf8String(buffer.Slice(FieldLocations.Name, FieldLengths.Name)),
mode : (int)TarHelpers.ParseOctal <uint>(buffer.Slice(FieldLocations.Mode, FieldLengths.Mode)),
mTime : TarHelpers.GetDateTimeOffsetFromSecondsSinceEpoch((long)TarHelpers.ParseOctal <ulong>(buffer.Slice(FieldLocations.MTime, FieldLengths.MTime))),
typeFlag : (TarEntryType)buffer[FieldLocations.TypeFlag])
{
_checksum = checksum,
_size = size,
_uid = (int)TarHelpers.ParseOctal <uint>(buffer.Slice(FieldLocations.Uid, FieldLengths.Uid)),
_gid = (int)TarHelpers.ParseOctal <uint>(buffer.Slice(FieldLocations.Gid, FieldLengths.Gid)),
_linkName = TarHelpers.GetTrimmedUtf8String(buffer.Slice(FieldLocations.LinkName, FieldLengths.LinkName))
};
if (header._format == TarEntryFormat.Unknown)
{
header._format = header._typeFlag switch
{
TarEntryType.ExtendedAttributes or
TarEntryType.GlobalExtendedAttributes => TarEntryFormat.Pax,
TarEntryType.DirectoryList or
TarEntryType.LongLink or
TarEntryType.LongPath or
TarEntryType.MultiVolume or
TarEntryType.RenamedOrSymlinked or
TarEntryType.TapeVolume => TarEntryFormat.Gnu,
// V7 is the only one that uses 'V7RegularFile'.
TarEntryType.V7RegularFile => TarEntryFormat.V7,
TarEntryType.SparseFile => throw new NotSupportedException(string.Format(SR.TarEntryTypeNotSupported, header._typeFlag)),
// We can quickly determine the *minimum* possible format if the entry type
// is the POSIX 'RegularFile', although later we could upgrade it to PAX or GNU
_ => (header._typeFlag == TarEntryType.RegularFile) ? TarEntryFormat.Ustar : TarEntryFormat.V7
};
}
return(header);
}
// Attempts to read the fields shared by all formats and stores them in their expected data type.
// Throws if any data type conversion fails.
// Returns true on success, false if checksum is zero.
private bool TryReadCommonAttributes(Span <byte> buffer)
{
// Start by collecting fields that need special checks that return early when data is wrong
// Empty checksum means this is an invalid (all blank) entry, finish early
Span <byte> spanChecksum = buffer.Slice(FieldLocations.Checksum, FieldLengths.Checksum);
if (TarHelpers.IsAllNullBytes(spanChecksum))
{
return(false);
}
_checksum = TarHelpers.GetTenBaseNumberFromOctalAsciiChars(spanChecksum);
// Zero checksum means the whole header is empty
if (_checksum == 0)
{
return(false);
}
_size = TarHelpers.GetTenBaseNumberFromOctalAsciiChars(buffer.Slice(FieldLocations.Size, FieldLengths.Size));
if (_size < 0)
{
throw new FormatException(string.Format(SR.TarSizeFieldNegative, _name));
}
// Continue with the rest of the fields that require no special checks
_name = TarHelpers.GetTrimmedUtf8String(buffer.Slice(FieldLocations.Name, FieldLengths.Name));
_mode = TarHelpers.GetTenBaseNumberFromOctalAsciiChars(buffer.Slice(FieldLocations.Mode, FieldLengths.Mode));
_uid = TarHelpers.GetTenBaseNumberFromOctalAsciiChars(buffer.Slice(FieldLocations.Uid, FieldLengths.Uid));
_gid = TarHelpers.GetTenBaseNumberFromOctalAsciiChars(buffer.Slice(FieldLocations.Gid, FieldLengths.Gid));
int mTime = TarHelpers.GetTenBaseNumberFromOctalAsciiChars(buffer.Slice(FieldLocations.MTime, FieldLengths.MTime));
_mTime = TarHelpers.GetDateTimeFromSecondsSinceEpoch(mTime);
_typeFlag = (TarEntryType)buffer[FieldLocations.TypeFlag];
_linkName = TarHelpers.GetTrimmedUtf8String(buffer.Slice(FieldLocations.LinkName, FieldLengths.LinkName));
if (_format == TarFormat.Unknown)
{
_format = _typeFlag switch
{
TarEntryType.ExtendedAttributes or
TarEntryType.GlobalExtendedAttributes => TarFormat.Pax,
TarEntryType.DirectoryList or
TarEntryType.LongLink or
TarEntryType.LongPath or
TarEntryType.MultiVolume or
TarEntryType.RenamedOrSymlinked or
TarEntryType.SparseFile or
TarEntryType.TapeVolume => TarFormat.Gnu,
// V7 is the only one that uses 'V7RegularFile'.
TarEntryType.V7RegularFile => TarFormat.V7,
// We can quickly determine the *minimum* possible format if the entry type
// is the POSIX 'RegularFile', although later we could upgrade it to PAX or GNU
_ => (_typeFlag == TarEntryType.RegularFile) ? TarFormat.Ustar : TarFormat.V7
};
}
return(true);
}