Gios WORD .NET Library (using RTF specification)

18-Aug-05 9:50

Grazie Alessio,

Paolo.

good , but

Huisheng Chen8-Aug-05 20:25

Huisheng Chen

8-Aug-05 20:25

it does not support non-ASCII characters, such as Simplified Chinese(double-byte characters).

Regards,
unruledboy@hotmail.com

Rei Miyasaka8-Aug-05 21:05

8-Aug-05 21:05

Ouch... you're right, Japanese doesn't show up.

Paolo Gios8-Aug-05 23:29

8-Aug-05 23:29

Hello

I tried doing it, but I'm Italian and I really can't understand non-latin language. There is a simple way of doing it (take a look into the RTF specification)

That's why I let you use the methods:

myWordDocument.WriteRtf(string);
myWordDocument.WriteControlWord(string);

... and your help for supporting non latin characters will be appraciated!

Paolo.

Rei Miyasaka9-Aug-05 2:26

9-Aug-05 2:26

I gave it a try, but unfortunately I don't know enough about the RTF format to do too much Frown | :(

I managed to implement the \ansicpg and \uN described here[^] by changing the WordDocument.Header getter:

<br />
internal string Header<br />
{<br />
	get<br />
	{<br />
		string s= "{";<br />
		s+="\\rtf1\\ansi";<br />
		s+="\\ansicpg" + Encoding.Default.CodePage.ToString();<br />
		s+="\\uc" + Encoding.Default.CodePage.GetMaxByteCount(1);<br />
		s+=this.FontTable;<br />
		s+=this.ColorTable;<br />
		if (this.StartPage>-1) s+="\\pgnstart"+this.StartPage+"\n";<br />
		return s;<br />
	}<br />
}<br />

Your Utility.Encode method seems to do the hex encoding that it needs already, but somehow that's still not enough. Any idea what I'm missing?

Huisheng Chen9-Aug-05 2:44

Huisheng Chen

9-Aug-05 2:44

certainly not ansi, here is a sample rtf that contains non-ASCII characters:

{\rtf1\ansi\ansicpg936\deff0\deflang1033\deflangfe2052{\fonttbl{\f0\fmodern\fprq6\fcharset134 \'cb\'ce\'cc\'e5;}}
{\*\generator Msftedit 5.41.21.2500;}\viewkind4\uc1\pard\lang2052\f0\fs20 abc\'b2\'e2\'ca\'d4\par
}

notice that after abc, there is 4 hex, b2 e2 ca d4, as for double bytes, then there two pairs, then two words, combing b2 & e2, ca & d4, that's.

Regards,
unruledboy@hotmail.com

I think I got it

Rei Miyasaka9-Aug-05 16:32

9-Aug-05 16:32

I had to make several changes, and unfortunately there's a P/Invoke involved... here's what I've done.

Like I explained earlier, first WordDocument.Header had to be changed to emit \ansicpg and \uc:

<br />
internal string Header<br />
{<br />
get<br />
{<br />
string s= "{";<br />
s+="\\rtf1\\ansi";<br />
s+="\\ansicpg" + Encoding.Default.CodePage.ToString();<br />
s+="\\uc" + Encoding.Default.GetMaxByteCount(1);<br />
s+=this.FontTable;<br />
s+=this.ColorTable;<br />
if (this.StartPage>-1) s+="\\pgnstart"+this.StartPage+"\n";<br />
return s;<br />
}<br />
}<br />

Then I had to change Utility.Encode so that rather than encoding just the first byte, it encodes every byte of each character, and use the correct system encoding (this could be a potential problem for documents that use a language other than the system default codepage -- the only way to solve this would be to make a public Encoding property in WordDocument, and pass that to every call to Encoding.Utility):

<br />
if ((c>=48&&c<=122)||c=='\n'||c==' ') sb.Append(c);<br />
else<br />
{<br />
if (c=='€')<br />
sb.Append("\\'80");<br />
else<br />
{<br />
byte[] Bytes=System.Text.Encoding.Unicode.GetBytes(new char[]{c});<br />
foreach(byte b in Bytes)<br />
sb.Append("\\'" + Utility.ToHexString(b));<br />
}<br />
}<br />

Then I added a LOGFONT class, which is a Windows API class invoke:

<br />
using System.Runtime.InteropServices;<br />
<br />
[StructLayout(LayoutKind.Sequential, CharSet=CharSet.Auto)]<br />
class LOGFONT<br />
{<br />
public int lfHeight;<br />
public int lfWidth;<br />
public int lfEscapement;<br />
public int lfOrientation;<br />
public int lfWeight;<br />
public byte lfItalic;<br />
public byte lfUnderline;<br />
public byte lfStrikeOut;<br />
public byte lfCharSet;<br />
public byte lfOutPrecision;<br />
public byte lfClipPrecision;<br />
public byte lfQuality;<br />
public byte lfPitchAndFamily;<br />
[MarshalAs(UnmanagedType.ByValTStr, SizeConst=32)]<br />
public string lfFaceName;<br />
}<br />

Lastly, I changed WordDocument.AddFontAndGetID() to save the whole font rather than just the name (Font.Equals is implented; searching shouldn't be a problem), and had WordDocument.FontTable create a LOGFONT, specify the character set and the font name to the "real" one (apparently Japanese fonts have a Japanese name and an English name, Drawing.Font gives only the English one):

<br />
get<br />
{<br />
string s="{\\fonttbl\n";<br />
for (int index=0;index {<br />
Font f = FontList[index] as Font;<br />
LOGFONT logFont = new LOGFONT();<br />
f.ToLogFont(logFont);<br />
s+="{\\f"+index.ToString()+" \\fcharset" + logFont.lfCharSet.ToString() + " " + Utility.Encode(logFont.lfFaceName) + ";}\n";<br />
}<br />
s+="}\n";<br />
return s;<br />
}<br />

Real messy eh? According to this[^], Font.GdiCharSet isn't set unless you specify it. It just sets it to 1 (default)... which makes no sense. I hope they fix this in future versions, along with the font face name issue.

Hopefully I got everything right. I might have missed some things from the 1.5 specification, but it works in Word 2003 and Wordpad on XPSP2, so it should be better than nothing Smile | :)

Huisheng Chen10-Aug-05 5:22

Huisheng Chen

10-Aug-05 5:22

There is a solution out there:

sourceforge.net/projects/itextsharp

it HAS rtf builder.

Regards,
unruledboy@hotmail.com

Paolo Gios10-Aug-05 6:55

10-Aug-05 6:55

Let me just tell you that ITextSharp is something TOTALLY different from the Gios Libraries:

ITextSharp is a BIG project, more powerful (obviously), but much more difficult to use it and it is 2300 Kbytes against 36Kb of WORD .NET !!!

That means that I really appreciate helps for improving Gios in languages support (like Reinux-thank you very much) and ITextSharp IS NOT an alternative of my library.

Regards,
Paolo.

A little bit of sad news

Rei Miyasaka12-Aug-05 14:20

12-Aug-05 14:20

I sent suggestions to MSDN Product Feedback Center to fix the two issues with System.Drawing.Font, and they've decided that neither are worth fixing since they might introduce breaking changes.

I can understand since the framework is in itself a standard, and exceptions to the no-breaking-changes rule are expensive, but the two problems are critical to passive asian language support. Many developers are no doubt releasing products without noticing the problems until they get angry users. I'm sure almost everyone uses the GdiCharSet and Name properties expecting them to return the right values.

If you agree with my concern, I'd be glad if you could please submit votes to try to catch their attention.

Font.Name with non-English font names returns English name, misleading[^]

Font.GdiCharSet always returns 1[^]

Thanks,
Rei

percyboy16-Aug-05 23:02

percyboy

16-Aug-05 23:02

You may see my codes here:

http://www.percyboy.com/p/i/rtfsample.txt[^]

Note the DecodeChar method:

private void DecodeChar(char c)<br />
{<br />
   uint t = Convert.ToUInt32(c);<br />
   sb.Append(@"\u" + t.ToString());<br />
}

In RTF sources, it always uses the default CodePage of the local PC. But in RTF specification, it also supports Unicode coding. Yes, it's the "\u" control char.

you can try this simple way to solve your problem. just have a try!

please let me know, if you've solve it! Thanks!

Paolo Gios16-Aug-05 23:24

16-Aug-05 23:24

Thank you very much!
Maybe these were the few lines of code that we needed.

Paolo.

Simon Pham25-Aug-05 4:57

Simon Pham

25-Aug-05 4:57

I just want to say that for Vietnamese the code must be:
foreach (char c in s.ToCharArray())
{
if ((c>=48&&c<=122)||c=='\n'||c==' '||c=='\t' )
{
sb.Append(c);
}
else
{
if (c=='€')
sb.Append("\\'80");
else
{
if(c>271)
{
uint t = Convert.ToUInt32(c);
sb.Append(@"\u" + t.ToString());
}
if(c>255 && c<272)
{
byte[] Bytes=System.Text.Encoding.Unicode.GetBytes(new char[]{c});
sb.Append("\\'");
sb.Append(Utility.ToHexString(Bytes));
else
{
byte[] Bytes=ue.GetBytes(new char[]{c});
string tmp="\\'" + Utility.ToHexString(Bytes[0]);
sb.Append(tmp);
}
}
}

29718

dunglesi2-Jan-06 22:46

dunglesi

2-Jan-06 22:46

Can you show me more detail about your solution when working with Vietnamese, Simon Pham?

-- modified at 4:52 Tuesday 3rd January, 2006

Mausi23-Nov-05 0:34

Mausi

23-Nov-05 0:34

There is fix for ColorTable. It uses System.Reflection to get colors from:

- System.Drawing.Color
- System.Drawing.SystemColors
- System.Drawing.KnownColor

<br />
public WordDocument(WordDocumentFormat WordDocumentFormat)<br />
{<br />
    <br />
    this.WordDocument=this;<br />
    this.FontList=new ArrayList();<br />
    this.Colors=new ArrayList();<br />
<br />
    foreach( PropertyInfo pi in typeof(Color).GetProperties())<br />
    {<br />
        this.AddColorAndGetID(Color.FromName(pi.Name));<br />
    }<br />
<br />
    foreach( PropertyInfo pi in typeof(SystemColors).GetProperties())<br />
    {<br />
        this.AddColorAndGetID(Color.FromName(pi.Name));<br />
    }<br />
<br />
    foreach( PropertyInfo pi in typeof(KnownColor).GetProperties())<br />
    {<br />
        this.AddColorAndGetID(Color.FromName(pi.Name));<br />
    }<br />
<br />
    this.Objects=new ArrayList();<br />
    this.Write(WordDocumentFormat.ToLineStream(),false);<br />
}<br />

Test :

<br />
Font font = new Font("Courier",8.25f);<br />
WordDocument doc = new WordDocument(WordDocumentFormat.A4);<br />
doc.SetFont(font);<br />
doc.SetTextAlign(WordTextAlign.Left);<br />
WordTable tbl = doc.NewTable(font,Color.BlueViolet,2,2,5);<br />
tbl.Rows[0][0].SetForegroundColor(Color.Pink);<br />
tbl.Rows[0][0].Write("HALLO1");<br />
tbl.Rows[0][1].Write("HALLO2");<br />
tbl.Rows[1][0].SetForegroundColor(SystemColors.HotTrack);<br />
tbl.Rows[1][0].SetBackgroundColor(SystemColors.ControlDarkDark);<br />
tbl.Rows[1][0].Write("HALLO3");<br />
tbl.Rows[1][1].SetForegroundColor(Color.FromKnownColor(KnownColor.ActiveCaption));<br />
tbl.Rows[1][1].Write("HALLO4");<br />
tbl.SaveToDocument(2000,100);<br />
<br />
System.IO.MemoryStream ms = new System.IO.MemoryStream();<br />
doc.SaveToStream(ms);<br />
ms.Position=0;<br />
byte[] b= ms.ToArray();<br />
string str = System.Text.Encoding.UTF8.GetString(b);<br />
<br />
richTextBox1.Rtf=str;<br />

Test.gif

Member 95538034-Dec-12 22:03

Member 9553803

4-Dec-12 22:03

Lastly, I changed WordDocument.AddFontAndGetID() to save the whole font rather than just the name (Font.Equals is implented; searching shouldn't be a problem), and had WordDocument.FontTable create a LOGFONT, specify the character set and the font name to the "real" one (apparently Japanese fonts have a Japanese name and an English name, Drawing.Font gives only the English one):

get
{
string s="{\\fonttbl\n";
for (int index=0;index {
Font f = FontList[index] as Font;
LOGFONT logFont = new LOGFONT();
f.ToLogFont(logFont);
s+="{\\f"+index.ToString()+" \\fcharset" + logFont.lfCharSet.ToString() + " " + Utility.Encode(logFont.lfFaceName) + ";}\n";
}
s+="}\n";
return s;
}

I didnt follow this part can u explain this change.

meduza19-Oct-06 23:46

meduza

19-Oct-06 23:46

The answer to your question is that is necessary to change also the rtf charset value. After that it works.