internal XlsBiffLabelCell(byte[] bytes, uint offset, ExcelBinaryReader reader)
: base(bytes, offset, reader)
{
m_UseEncoding = reader.DefaultEncoding;
xlsString = XlsStringFactory.CreateXlsString(bytes, offset, reader);
}
internal XlsBiffBoundSheet(uint startOffset, SheetType type, SheetVisibility visibleState, string name)
: base(new byte[32])
{
StartOffset = startOffset;
Type = type;
VisibleState = visibleState;
_sheetName = new XlsInternalString(name);
}
internal XlsBiffHeaderFooterString(byte[] bytes, int biffVersion)
: base(bytes)
{
if (biffVersion < 8)
{
_xlsString = new XlsShortByteString(bytes, ContentOffset);
}
else if (biffVersion == 8)
{
_xlsString = new XlsUnicodeString(bytes, ContentOffset);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion, nameof(biffVersion));
}
}
internal XlsBiffBoundSheet(byte[] bytes, int biffVersion)
: base(bytes)
{
StartOffset = ReadUInt32(0x0);
Type = (SheetType)ReadByte(0x5);
VisibleState = (SheetVisibility)ReadByte(0x4);
if (biffVersion == 8)
{
_sheetName = new XlsShortUnicodeString(bytes, ContentOffset + 6);
}
else if (biffVersion == 5)
{
_sheetName = new XlsShortByteString(bytes, ContentOffset + 6);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion, nameof(biffVersion));
}
}
internal XlsBiffLabelCell(byte[] bytes, uint offset, int biffVersion)
: base(bytes, offset, biffVersion)
{
if (biffVersion == 2)
{
// BIFF2
_xlsString = new XlsShortByteString(bytes, offset + 4 + 7);
}
else if (biffVersion >= 3 && biffVersion <= 5)
{
// BIFF3-5
_xlsString = new XlsByteString(bytes, offset + 4 + 6);
}
else if (biffVersion == 8)
{
// BIFF8
_xlsString = new XlsUnicodeString(bytes, offset + 4 + 6);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion.ToString(), nameof(biffVersion));
}
}
internal XlsBiffFormatString(byte[] bytes, uint offset, int biffVersion)
: base(bytes, offset)
{
if (biffVersion == 2 || biffVersion == 3)
{
// BIFF2-3
_xlsString = new XlsShortByteString(bytes, offset + 4);
}
else if (biffVersion == 4 || biffVersion == 5)
{
// BIFF4-5
_xlsString = new XlsShortByteString(bytes, offset + 4 + 2);
}
else if (biffVersion == 8)
{
// BIFF8
_xlsString = new XlsUnicodeString(bytes, offset + 4 + 2);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion.ToString(), nameof(biffVersion));
}
}
internal XlsBiffLabelCell(byte[] bytes, uint offset, int biffVersion)
: base(bytes, offset, biffVersion)
{
if (Id == BIFFRECORDTYPE.LABEL_OLD)
{
// BIFF2
_xlsString = new XlsShortByteString(bytes, offset + 4 + 7);
}
else if (biffVersion >= 2 && biffVersion <= 5)
{
// BIFF3-5, or if there is a newer label record present in a BIFF2 stream
_xlsString = new XlsByteString(bytes, offset + 4 + 6);
}
else if (biffVersion == 8)
{
// BIFF8
_xlsString = new XlsUnicodeString(bytes, offset + 4 + 6);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion.ToString(), nameof(biffVersion));
}
}
internal XlsBiffFormulaString(byte[] bytes, int biffVersion)
: base(bytes)
{
if (biffVersion == 2)
{
// BIFF2
_xlsString = new XlsShortByteString(bytes, ContentOffset);
}
else if (biffVersion >= 3 && biffVersion <= 5)
{
// BIFF3-5
_xlsString = new XlsByteString(bytes, ContentOffset);
}
else if (biffVersion == 8)
{
// BIFF8
_xlsString = new XlsUnicodeString(bytes, ContentOffset);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion, nameof(biffVersion));
}
}
internal XlsBiffFormatString(byte[] bytes, int biffVersion)
: base(bytes)
{
if (Id == BIFFRECORDTYPE.FORMAT_V23)
{
// BIFF2-3
_xlsString = new XlsShortByteString(bytes, ContentOffset);
}
else if (biffVersion >= 2 && biffVersion <= 5)
{
// BIFF4-5, or if there is a newer format record in a BIFF2-3 stream
_xlsString = new XlsShortByteString(bytes, ContentOffset + 2);
}
else if (biffVersion == 8)
{
// BIFF8
_xlsString = new XlsUnicodeString(bytes, ContentOffset + 2);
}
else
{
throw new ArgumentException("Unexpected BIFF version " + biffVersion, nameof(biffVersion));
}
}
internal XlsBiffFont(byte[] bytes, uint offset, int biffVersion)
: base(bytes, offset)
{
if (Id == BIFFRECORDTYPE.FONT_V34)
{
_fontName = new XlsShortByteString(bytes, offset + 4 + 6);
}
else if (Id == BIFFRECORDTYPE.FONT && biffVersion == 2)
{
_fontName = new XlsShortByteString(bytes, offset + 4 + 4);
}
else if (Id == BIFFRECORDTYPE.FONT && biffVersion == 5)
{
_fontName = new XlsShortByteString(bytes, offset + 4 + 14);
}
else if (Id == BIFFRECORDTYPE.FONT && biffVersion == 8)
{
_fontName = new XlsShortUnicodeString(bytes, offset + 4 + 14);
}
else
{
_fontName = new XlsInternalString(string.Empty);
}
if (Id == BIFFRECORDTYPE.FONT && biffVersion >= 5)
{
// Encodings were mapped by correlating this:
// https://docs.microsoft.com/en-us/windows/desktop/intl/code-page-identifiers
// with the FONT record character set table here:
// https://www.openoffice.org/sc/excelfileformat.pdf
var byteStringCharacterSet = ReadByte(12);
switch (byteStringCharacterSet)
{
case 0: // ANSI Latin
case 1: // System default
ByteStringEncoding = EncodingHelper.GetEncoding(1252);
break;
case 77: // Apple roman
ByteStringEncoding = EncodingHelper.GetEncoding(10000);
break;
case 128: // ANSI Japanese Shift-JIS
ByteStringEncoding = EncodingHelper.GetEncoding(932);
break;
case 129: // ANSI Korean (Hangul)
ByteStringEncoding = EncodingHelper.GetEncoding(949);
break;
case 130: // ANSI Korean (Johab)
ByteStringEncoding = EncodingHelper.GetEncoding(1361);
break;
case 134: // ANSI Chinese Simplified GBK
ByteStringEncoding = EncodingHelper.GetEncoding(936);
break;
case 136: // ANSI Chinese Traditional BIG5
ByteStringEncoding = EncodingHelper.GetEncoding(950);
break;
case 161: // ANSI Greek
ByteStringEncoding = EncodingHelper.GetEncoding(1253);
break;
case 162: // ANSI Turkish
ByteStringEncoding = EncodingHelper.GetEncoding(1254);
break;
case 163: // ANSI Vietnamese
ByteStringEncoding = EncodingHelper.GetEncoding(1258);
break;
case 177: // ANSI Hebrew
ByteStringEncoding = EncodingHelper.GetEncoding(1255);
break;
case 178: // ANSI Arabic
ByteStringEncoding = EncodingHelper.GetEncoding(1256);
break;
case 186: // ANSI Baltic
ByteStringEncoding = EncodingHelper.GetEncoding(1257);
break;
case 204: // ANSI Cyrillic
ByteStringEncoding = EncodingHelper.GetEncoding(1251);
break;
case 222: // ANSI Thai
ByteStringEncoding = EncodingHelper.GetEncoding(874);
break;
case 238: // ANSI Latin II
ByteStringEncoding = EncodingHelper.GetEncoding(1250);
break;
case 255: // OEM Latin
ByteStringEncoding = EncodingHelper.GetEncoding(850);
break;
}
}
}
internal XlsBiffCodeName(byte[] bytes, uint offset)
: base(bytes, offset)
{
// BIFF8 only
_xlsString = new XlsUnicodeString(bytes, offset + 4 + 0);
}
internal XlsBiffCodeName(byte[] bytes)
: base(bytes)
{
// BIFF8 only
_xlsString = new XlsUnicodeString(bytes, ContentOffset);
}
private bool TryReadString(out IXlsString result)
{
if (CurrentState == SstState.StartStringHeader)
{
if (CurrentRecord.Size - CurrentRecordOffset == 0)
{
result = null;
return(false);
}
CurrentHeader = new XlsSSTStringHeader(CurrentRecord.Bytes, CurrentRecordOffset);
CurrentIsMultiByte = CurrentHeader.IsMultiByte;
CurrentHeaderBytes = (int)CurrentHeader.HeadSize;
CurrentRemainingCharacters = (int)CurrentHeader.CharacterCount;
const int XlsUnicodeStringHeaderSize = 3;
CurrentResult = new byte[XlsUnicodeStringHeaderSize + CurrentRemainingCharacters * 2];
CurrentResult[0] = (byte)(CurrentRemainingCharacters & 0x00FF);
CurrentResult[1] = (byte)((CurrentRemainingCharacters & 0xFF00) >> 8);
CurrentResult[2] = 1; // IsMultiByte = true
CurrentResultOffset = XlsUnicodeStringHeaderSize;
CurrentState = SstState.StringHeader;
}
if (CurrentState == SstState.StringHeader)
{
if (!Advance(CurrentHeaderBytes, out int advanceBytes))
{
CurrentHeaderBytes -= advanceBytes;
result = null;
return(false);
}
CurrentState = SstState.StringData;
if (CurrentRecord.Size - CurrentRecordOffset == 0)
{
// End of buffer before string data. Return false in StringData state to ensure reading the multibyte flag in the next record
result = null;
return(false);
}
}
if (CurrentState == SstState.StringData)
{
var bytesPerCharacter = CurrentIsMultiByte ? 2 : 1;
var maxRecordCharacters = (CurrentRecord.Size - CurrentRecordOffset) / bytesPerCharacter;
var readCharacters = Math.Min(maxRecordCharacters, CurrentRemainingCharacters);
ReadUnicodeBytes(CurrentResult, CurrentResultOffset, readCharacters, CurrentIsMultiByte);
CurrentResultOffset += readCharacters * 2; // The result is always multibyte
CurrentRemainingCharacters -= readCharacters;
if (CurrentIsMultiByte && CurrentRecord.Size - CurrentRecordOffset == 1)
{
// Skip leftover byte at the end of a multibyte Continue record
ReadByte();
}
if (CurrentRemainingCharacters > 0 && CurrentRecord.Size - CurrentRecordOffset == 0)
{
result = null;
return(false);
}
CurrentState = SstState.StringTail;
CurrentTailBytes = (int)CurrentHeader.TailSize;
}
if (CurrentState == SstState.StringTail)
{
// Skip formatting runs and phonetic/extended data. Can also span
// multiple Continue records
if (!Advance(CurrentTailBytes, out var advanceBytes))
{
result = null;
CurrentTailBytes -= advanceBytes;
return(false);
}
CurrentState = SstState.StartStringHeader;
result = new XlsUnicodeString(CurrentResult, 0);
return(true);
}
throw new InvalidOperationException("Unexpected state in SST reader");
}
