Xls label fix (#183)

* Fix label (pre-biff8) reading * Add support for reading "Book" entry for xl95 support. * Fix reading label (String) values in biff5 (xl95) * Fix reading rich strings in xl95.
MarkPflug · Aug 14, 2024 · e458b57 · e458b57
1 parent 15d3f4b
commit e458b57
Show file tree

Hide file tree

Showing 6 changed files with 127 additions and 82 deletions.
diff --git a/docs/ReleaseNotes.md b/docs/ReleaseNotes.md
@@ -1,5 +1,8 @@
 # Sylvan.Data.Excel Release Notes
 
+_0.4.25_
+- Fix some issues with reading Excel 95 .xls files.
+
 _0.4.24_
 - Fix for reading certain .xls files.
 - Handle writing NaN and infinity values.

diff --git a/source/Sylvan.Data.Excel.Tests/ExternalDataTests.cs b/source/Sylvan.Data.Excel.Tests/ExternalDataTests.cs
@@ -1,5 +1,6 @@
 #if NETCOREAPP3_0_OR_GREATER
 
+using Sylvan.Data.Csv;
 using System;
 using System.Collections.Generic;
 using System.Diagnostics;
@@ -231,7 +232,6 @@ public void GetValue(string path)
 			GetErrorAsNull = true
 		};
 		var edr = ExcelDataReader.Create(path, opts);
-
 		do
 		{
 			while (edr.Read())
@@ -243,6 +243,32 @@ public void GetValue(string path)
 			}
 		} while (edr.NextResult());
 	}
+
+	[Theory]
+	[MemberData(nameof(GetExcelFiles))]
+	public void ToCsv(string filename)
+	{
+		if (filename == null) return;
+
+		var root = GetRootPath();
+		var path = Path.Combine(root, filename);
+
+		var opts = new ExcelDataReaderOptions
+		{
+			Schema = ExcelSchema.NoHeaders,
+			GetErrorAsNull = true
+		};
+		var edr = ExcelDataReader.Create(path, opts);
+
+		do
+		{
+			var outPath = $"{filename}-{edr.WorksheetName}.csv";
+			var dir = Path.GetDirectoryName(outPath);
+			Directory.CreateDirectory(dir);
+			using var w = CsvDataWriter.Create($"{filename}-{edr.WorksheetName}.csv");
+			w.Write(edr.AsVariableField(e => e.RowFieldCount));
+		} while (edr.NextResult());
+	}
 }
 
 #endif
diff --git a/source/Sylvan.Data.Excel/ExcelDataReader.cs b/source/Sylvan.Data.Excel/ExcelDataReader.cs
@@ -633,6 +633,7 @@ public sealed override object GetValue(int ordinal)
 							var kind = fmt?.Kind ?? FormatKind.Number;
 							switch (kind)
 							{
+								case FormatKind.String:
 								case FormatKind.Number:
 									var doubleValue = GetDouble(ordinal);
 									unchecked

diff --git a/source/Sylvan.Data.Excel/Sylvan.Data.Excel.csproj b/source/Sylvan.Data.Excel/Sylvan.Data.Excel.csproj
@@ -3,7 +3,7 @@
 	<PropertyGroup>
 		<TargetFrameworks>net6.0;netstandard2.1;netstandard2.0</TargetFrameworks>
 		<LangVersion>latest</LangVersion>
-		<VersionPrefix>0.4.24</VersionPrefix>
+		<VersionPrefix>0.4.25</VersionPrefix>
 		<Description>A cross-platform .NET library for reading Excel data files.</Description>
 		<PackageTags>excel;xls;xlsx;xlsb;datareader</PackageTags>
 		<Nullable>enable</Nullable>

diff --git a/source/Sylvan.Data.Excel/Xls/XlsWorkbookReader+RecordReader.cs b/source/Sylvan.Data.Excel/Xls/XlsWorkbookReader+RecordReader.cs
@@ -115,57 +115,7 @@ public int ReadInt32()
 		{
 			return ReadByte() | ReadByte() << 8 | ReadByte() << 16 | ReadByte() << 24;
 		}
-
-		public string ReadString16()
-		{
-			if (bufferPos >= recordOff + recordLen)
-			{
-				var next = NextRecord();
-				if (!next || Type != RecordType.Continue)
-					throw new InvalidDataException();
-			}
-
-			// the length of the string in *characters*
-			int len = ReadInt16();
-			if (len < 0)
-			{
-				throw new InvalidDataException();
-			}
-			byte options = ReadByte();
-
-			bool compressed = (options & 0x01) == 0;
-			bool asian = (options & 0x04) != 0;
-			bool rich = (options & 0x08) != 0;
-
-			int richCount = 0;
-			if (rich)
-				richCount = ReadInt16();
-
-			int asianCount = 0;
-			if (asian)
-				asianCount = ReadInt32();
-
-			var str = ReadStringBuffer(len, compressed);
-
-			var remain = richCount * 4 + asianCount;
-
-			while (remain > 0)
-			{
-				var avail = recordOff + recordLen - bufferPos;
-				var c = Math.Min(remain, avail);
-				remain -= c;
-				bufferPos += c;
-				Assert();
-				if (remain > 0)
-				{
-					var next = NextRecord();
-					if (!next || Type != RecordType.Continue)
-						throw new InvalidDataException();
-				}
-			}
-
-			return str;
-		}
+
 
 		static readonly Encoding Encoding1252 = Encoding.GetEncoding(1252);
 
@@ -238,20 +188,44 @@ internal string ReadStringBuffer(int charCount, bool compressed)
 
 		public string ReadByteString(int lenSize)
 		{
-			int len;
-			if (lenSize == 1)
-				len = ReadByte();
-			else
-				len = ReadInt16();
-
-			ReadStringBuffer(len, true);
-			var str = new string(strBuffer, 0, len);
-			return str;
+			int len =
+				lenSize == 1
+				? ReadByte()
+				: ReadInt16();
+
+			return ReadStringBuffer(len, true);
 		}
 
 		public string ReadString8()
 		{
-			int len = ReadByte();
+			MaybeContinueString();
+			var len = ReadByte();
+			return ReadString(len);
+		}
+
+		public string ReadString16()
+		{
+			MaybeContinueString();
+			var len = ReadInt16();
+			return ReadString(len);
+		}
+
+		void MaybeContinueString()
+		{
+			if (bufferPos >= recordOff + recordLen)
+			{
+				var next = NextRecord();
+				if (!next || Type != RecordType.Continue)
+					throw new InvalidDataException();
+			}
+		}
+
+		public string ReadString(int len)
+		{
+			if (len < 0)
+			{
+				throw new InvalidDataException();
+			}
 			byte options = ReadByte();
 
 			bool compressed = (options & 0x01) == 0;
@@ -268,14 +242,21 @@ public string ReadString8()
 
 			var str = ReadStringBuffer(len, compressed);
 
-			for (int i = 0; i < richCount; i++)
-			{
-				ReadInt32();
-			}
+			var remain = richCount * 4 + asianCount;
 
-			for (int i = 0; i < asianCount; i++)
+			while (remain > 0)
 			{
-				ReadByte();
+				var avail = recordOff + recordLen - bufferPos;
+				var c = Math.Min(remain, avail);
+				remain -= c;
+				bufferPos += c;
+				Assert();
+				if (remain > 0)
+				{
+					var next = NextRecord();
+					if (!next || Type != RecordType.Continue)
+						throw new InvalidDataException();
+				}
 			}
 
 			return str;

diff --git a/source/Sylvan.Data.Excel/Xls/XlsWorkbookReader.cs b/source/Sylvan.Data.Excel/Xls/XlsWorkbookReader.cs
@@ -32,7 +32,10 @@ public XlsSheetInfo(string name, int offset, bool hidden) : base(name, hidden)
 	internal XlsWorkbookReader(Stream stream, ExcelDataReaderOptions options) : base(stream, options)
 	{
 		var pkg = new Ole2Package(stream);
-		var part = pkg.GetEntry("Workbook\0");
+		var part = 
+			pkg.GetEntry("Workbook\0") ??
+			pkg.GetEntry("Book\0");
+
 		if (part == null)
 			throw new InvalidDataException();
 		var ps = part.Open();
@@ -254,15 +257,10 @@ int ParseXF()
 	void ParseFormat()
 	{
 		int ifmt = reader.ReadInt16();
-		string str;
-		if (biffVersion == 0x0500)
-		{
-			str = reader.ReadByteString(1);
-		}
-		else
-		{
-			str = reader.ReadString16();
-		}
+		string str =
+			biffVersion == 0x0500
+			? reader.ReadByteString(1)
+			: reader.ReadString16();
 
 		if (formats.ContainsKey(ifmt))
 		{
@@ -310,7 +308,38 @@ void ParseLabel()
 		int rowIdx = reader.ReadUInt16();
 		int colIdx = reader.ReadUInt16();
 		int xfIdx = reader.ReadUInt16();
-		string str = reader.ReadByteString(2);
+		int len = reader.ReadInt16();
+		if (len > 255) throw new InvalidDataException();
+		bool compressed = true;
+		if (biffVersion == 0x0500)
+		{
+			// apparently there are no flags in this version
+		}
+		else
+		{
+			byte flags = reader.ReadByte();
+			compressed = (flags & 1) == 0;
+		}
+
+		var str = reader.ReadStringBuffer(len, compressed);
+		SetRowData(colIdx, new FieldInfo(str));
+	}
+
+	void ParseRString()
+	{
+		int rowIdx = reader.ReadUInt16();
+		int colIdx = reader.ReadUInt16();
+		int xfIdx = reader.ReadUInt16();
+		var len = reader.ReadInt16();
+		var str = reader.ReadStringBuffer(len, true);
+
+		// consume the formatting info
+		var x = reader.ReadByte();
+		for (int i = 0; i < x; i++)
+		{
+			reader.ReadUInt16();
+		}
+
 		SetRowData(colIdx, new FieldInfo(str));
 	}
 
@@ -470,7 +499,10 @@ int NextRow()
 						}
 						else
 						{
-							throw new InvalidDataException();
+							peekRow = (ushort)(rowIndex + 1);
+							pendingRow = peekRow;
+							return 0;
+							//throw new InvalidDataException();
 						}
 					}
 					break;
@@ -516,10 +548,12 @@ int NextRow()
 				case RecordType.Formula:
 					ParseFormula();
 					break;
+				case RecordType.RString:
+					ParseRString();
+					break;
 				case RecordType.Blank:
 				case RecordType.BoolErr:
 				case RecordType.MulBlank:
-				case RecordType.RString:
 					break;
 				case RecordType.Array:
 				case RecordType.SharedFmla: