Starts improving the xls biff 8 codepage support, doesnt actually work yet

git-svn-id: https://svn.code.sf.net/p/lazarus-ccr/svn@2194 8e941d3f-bd1b-0410-a28a-d453659cc2b4

components/fpspreadsheet/laz_fpspreadsheet.lpk

@@ -1,10 +1,11 @@
 <?xml version="1.0"?>
 <CONFIG>
-  <Package Version="3">
+  <Package Version="4">
     <PathDelim Value="\"/>
     <Name Value="laz_fpspreadsheet"/>
+    <AddToProjectUsesSection Value="True"/>
     <CompilerOptions>
-      <Version Value="10"/>
+      <Version Value="11"/>
       <PathDelim Value="\"/>
       <SearchPaths>
         <UnitOutputDirectory Value="lib\$(TargetCPU)-$(TargetOS)"/>
@@ -16,7 +17,7 @@
       </Parsing>
       <Linking>
         <Debugging>
-          <GenerateDebugInfo Value="True"/>
+          <DebugInfoType Value="dsStabs"/>
         </Debugging>
       </Linking>
       <Other>
@@ -105,11 +106,14 @@
       </Item19>
     </Files>
     <Type Value="RunAndDesignTime"/>
-    <RequiredPkgs Count="1">
+    <RequiredPkgs Count="2">
       <Item1>
+        <PackageName Value="LazUtils"/>
+      </Item1>
+      <Item2>
         <PackageName Value="FCL"/>
         <MinVersion Major="1" Valid="True"/>
-      </Item1>
+      </Item2>
     </RequiredPkgs>
     <UsageOptions>
       <UnitPath Value="$(PkgOutDir)"/>

components/fpspreadsheet/xlsbiff8.pas

@@ -59,7 +59,7 @@ uses
   {$else}
   fpolestorage,
   {$endif}
-  fpsutils;
+  fpsutils, lazutf8;
 
 type
 
@@ -87,6 +87,7 @@ type
     procedure ReadRichString(const AStream: TStream);
     procedure ReadSST(const AStream: TStream);
     procedure ReadLabelSST(const AStream: TStream);
+    procedure ReadFont(const AStream: TStream);
   public
     { General reading methods }
     procedure ReadFromFile(AFileName: string; AData: TsWorkbook); override;
@@ -158,6 +159,7 @@ const
   INT_EXCEL_ID_CONTINUE   = $003C;
   INT_EXCEL_ID_LABELSST   = $00FD; //BIFF8 only
   INT_EXCEL_ID_PALETTE    = $0092;
+  INT_EXCEL_ID_CODEPAGE   = $0042;
 
   { Cell Addresses constants }
   MASK_EXCEL_ROW          = $3FFF;
@@ -1424,6 +1426,8 @@ begin
        INT_EXCEL_ID_BOUNDSHEET: ReadBoundSheet(AStream);
        INT_EXCEL_ID_EOF:        SectionEOF := True;
        INT_EXCEL_ID_SST:        ReadSST(AStream);
+       INT_EXCEL_ID_CODEPAGE:   ReadCodepage(AStream);
+       INT_EXCEL_ID_FONT:       ReadFont(AStream);
       else
         // nothing
       end;
@@ -1600,7 +1604,7 @@ end;
 function TsSpreadBIFF8Reader.ReadString(const AStream: TStream;
   const ALength: WORD): UTF8String;
 begin
-  Result:=UTF8Encode(ReadWideString(AStream, ALength));
+  Result:=UTF16ToUTF8(ReadWideString(AStream, ALength));
 end;
 
 procedure TsSpreadBIFF8Reader.ReadFromFile(AFileName: string; AData: TsWorkbook);
@@ -1697,7 +1701,7 @@ begin
   WideStrValue:=ReadWideString(AStream,L);
 
   { Save the data }
-  FWorksheet.WriteUTF8Text(ARow, ACol, UTF8Encode(WideStrValue));
+  FWorksheet.WriteUTF8Text(ARow, ACol, UTF16ToUTF8(WideStrValue));
 end;
 
 procedure TsSpreadBIFF8Reader.ReadNumber(AStream: TStream);
@@ -1789,13 +1793,16 @@ begin
           Raise Exception.Create('Expected CONTINUE not found.');
         end;
         PendingRecordSize:=WordLEtoN(AStream.ReadWord);
-        Dec(StringLength,Length(UTF8Decode(LString))); //Dec the used chars
+        Dec(StringLength,Length(UTF8ToUTF16(LString))); //Dec the used chars
         if StringLength=0 then break;
       end else begin
         break;
       end;
     end;
     FSharedStringTable.Add(LString);
+    {$ifdef XLSDEBUG}
+    WriteLn('Adding shared string: ' + LString);
+    {$endif}
     dec(Items);
   end;
 end;
@@ -1813,6 +1820,49 @@ begin
   FWorksheet.WriteUTF8Text(ARow, ACol, FSharedStringTable[SSTIndex]);
 end;
 
+procedure TsSpreadBIFF8Reader.ReadFont(const AStream: TStream);
+var
+  lCodePage: Word;
+  lHeight: Word;
+  lOptions: Word;
+  Len: Byte;
+  lFontName: UTF8String;
+begin
+  { Height of the font in twips = 1/20 of a point }
+  lHeight := AStream.ReadWord(); // WordToLE(200)
+
+  { Option flags }
+  lOptions := AStream.ReadWord();
+
+  { Colour index }
+  AStream.ReadWord();
+
+  { Font weight }
+  AStream.ReadWord();
+
+  { Escapement type }
+  AStream.ReadWord();
+
+  { Underline type }
+  AStream.ReadByte();
+
+  { Font family }
+  AStream.ReadByte();
+
+  { Character set }
+  lCodepage := AStream.ReadByte();
+  {$ifdef XLSDEBUG}
+  WriteLn('Reading Font Codepage='+IntToStr(lCodepage));
+  {$endif}
+
+  { Not used }
+  AStream.ReadByte();
+
+  { Font name: Unicodestring, char count in 1 byte }
+  Len := AStream.ReadByte();
+  lFontName := ReadString(AStream, Len);
+end;
+
 {*******************************************************************
 *  Initialization section
 *

components/fpspreadsheet/xlscommon.pas

@@ -9,7 +9,7 @@ interface
 uses
   Classes, SysUtils,
   fpspreadsheet,
-  fpsutils;
+  fpsutils{, lconvencoding};
 
 const
   { Formula constants TokenID values }
@@ -85,6 +85,8 @@ type
 
   TsSpreadBIFFReader = class(TsCustomSpreadReader)
   protected
+    FCodepage: string;
+    procedure ReadCodePage(AStream: TStream);
   end;
 
   { TsSpreadBIFFWriter }
@@ -103,6 +105,58 @@ type
 
 implementation
 
+{ TsSpreadBIFFReader }
+
+procedure TsSpreadBIFFReader.ReadCodePage(AStream: TStream);
+var
+  lCodePage: Word;
+begin
+  { Codepage }
+  lCodePage := WordLEToN(AStream.ReadWord());
+
+  case lCodePage of
+  // 016FH = 367 = ASCII
+  // 01B5H = 437 = IBM PC CP-437 (US)
+  //02D0H = 720 = IBM PC CP-720 (OEM Arabic)
+  //02E1H = 737 = IBM PC CP-737 (Greek)
+  //0307H = 775 = IBM PC CP-775 (Baltic)
+  //0352H = 850 = IBM PC CP-850 (Latin I)
+  //0354H = 852 = IBM PC CP-852 (Latin II (Central European))
+  //0357H = 855 = IBM PC CP-855 (Cyrillic)
+  //0359H = 857 = IBM PC CP-857 (Turkish)
+  $0359: FCodepage := 'cp857';
+  {035AH = 858 = IBM PC CP-858 (Multilingual Latin I with Euro)
+  035CH = 860 = IBM PC CP-860 (Portuguese)
+  035DH = 861 = IBM PC CP-861 (Icelandic)
+  035EH = 862 = IBM PC CP-862 (Hebrew)
+  035FH = 863 = IBM PC CP-863 (Canadian (French))
+  0360H = 864 = IBM PC CP-864 (Arabic)
+  0361H = 865 = IBM PC CP-865 (Nordic)
+  0362H = 866 = IBM PC CP-866 (Cyrillic (Russian))
+  0365H = 869 = IBM PC CP-869 (Greek (Modern))
+  036AH = 874 = Windows CP-874 (Thai)
+  03A4H = 932 = Windows CP-932 (Japanese Shift-JIS)
+  03A8H = 936 = Windows CP-936 (Chinese Simplified GBK)
+  03B5H = 949 = Windows CP-949 (Korean (Wansung))
+  03B6H = 950 = Windows CP-950 (Chinese Traditional BIG5)
+  04B0H = 1200 = UTF-16 (BIFF8)
+  04E2H = 1250 = Windows CP-1250 (Latin II) (Central European)
+  04E3H = 1251 = Windows CP-1251 (Cyrillic)
+  04E4H = 1252 = Windows CP-1252 (Latin I) (BIFF4-BIFF5)
+  04E5H = 1253 = Windows CP-1253 (Greek)}
+  //04E6H = 1254 = Windows CP-1254 (Turkish)
+  $04E6: FCodepage := 'cp1254';
+  {04E7H = 1255 = Windows CP-1255 (Hebrew)
+  04E8H = 1256 = Windows CP-1256 (Arabic)
+  04E9H = 1257 = Windows CP-1257 (Baltic)
+  04EAH = 1258 = Windows CP-1258 (Vietnamese)
+  0551H = 1361 = Windows CP-1361 (Korean (Johab))
+  2710H = 10000 = Apple Roman
+  8000H = 32768 = Apple Roman
+  8001H = 32769 = Windows CP-1252 (Latin I) (BIFF2-BIFF3)}
+  end;
+end;
+
 function TsSpreadBIFFWriter.FPSColorToEXCELPallete(AColor: TsColor): Word;
 begin
   case AColor of