SVGGraphics2D::drawString

[Kelsey Rider <ke...@hotmail.com>]

Hello,

I'm a developer working on a project to convert PDF files to SVG, using pdfbox and batik.  Pdfbox already contains methods to draw the PDF onto a Graphics2D object, with the goal to export to JPG, GIF, etc.  However, when I supply a SVGGraphics2D object, I observe two problems with the text (probably due to the same issue):
- the text is distorted - badly positioned, wrong size, etc.
- certain characters don't appear.  For example, the euro symbol.  This particular character actually causes SVGGraphics2D::drawString to put an invalid character (0x02) in the XML.  The string passed to drawString contains a single byte, 0x02; however, in the PDF this character is mapped to a type1 font and (I think) describes how to draw it.

Example code:
doc = PDDocument.load(url);DOMImplementation domImpl = GenericDOMImplementation.getDOMImplementation();Document document = domImpl.createDocument("http://www.w3.org/2000/svg", "svg", null);SVGGraphics2D graphics = new SVGGraphics2D(document);....PageDrawer.drawPage(graphics, page, pageDimension);File outFile = new File("out.svg");Writer out = new OutputStreamWriter(new FileOutputStream(outFile), "UTF-8");graphics.stream(out);out.close();

I realize that the problem seems like it may be with pdfbox; however, the output is fine when exporting to, say, JPG (in which case the graphics object is a SunGraphics2D).  I looked at the source code but I'm afraid it's a bit over my head; the biggest thing I can see is that the algorithms are completely different. :)  The output is correct when I manually set generatorCtx.svgFont=true, but of course this makes the output file bigger (10MB instead of 8MB).
Any help on this issue would be greatly appreciated.  If needed, I can send a PDF to duplicate the problem.
Thank you for your time,

Kelsey Rider 		 	   		  

Re: SVGGraphics2D::drawString

[th...@kodak.com]

Hi Kelsey,

Kelsey Rider <ke...@hotmail.com> wrote on 10/18/2010 12:03:52 PM:

> However, when I supply a SVGGraphics2D object, I observe two 
> problems with the text (probably due to the same issue):
> - the text is distorted - badly positioned, wrong size, etc.
> - certain characters don't appear.  For example, the euro symbol. 
> This particular character actually causes SVGGraphics2D::drawString 
> to put an invalid character (0x02) in the XML.  The string passed to
> drawString contains a single byte, 0x02; however, in the PDF this 
> character is mapped to a type1 font and (I think) describes how to draw 
it.
[...]
> The output is correct when I manually set generatorCtx.svgFont=true,
> but of course this makes the output file bigger (10MB instead of 8MB).

    I don't think there is a lot to be done in this case.  From your 
description of the Euro symbol
problem, it sounds like the PDF contains embedded fonts.  If the SVG 
doesn't contain a similar
set of embedded fonts it can't possibly correctly reproduce the PDF file. 
Setting 'svgFont' to
tells the SVGGraphics2D that it can't simply use the font-name, it needs 
to extract the used
characters from the font and put them in an SVG Font embedded in the 
output document 
(similar to the embedded font in the PDF document).

    It might be possible to only SVG Fontify some of the fonts used by the 
PDF file but I don't know
how the SVGGraphics2D could possibly differentiate between the fonts that 
needed to be
embedded and those that didn't.

    I think you simply need to accept the larger file size.
Thomas DeWeese | CDG Advanced Development | 
Eastman Kodak Company | 343 State Street | Rochester, NY 14650-0128 | 
Thomas.DeWeese@Kodak.com | 585 724-0294 | 
www.kodak.com