Re: PDFFontDescriptor Ascent Descent and FontBBox
Alexandar, on a completely different note: It may be interesting to also look into fontbox (part of pdfbox), which is now also an apache project, and therefore we could use source synergy. http://incubator.apache.org/pdfbox/ For the issue you've mentioned: This may be due so some issues with CID fonts not properly defining ascent and descent, but I don't have a reference for it. We would probably need some fonts to test and check. Max 2009/9/21 Alexander Kiel alexanderk...@gmx.net: Hi, as I'm currently on the way to redesign the TrueType/OpenType file reading, from time to time I stumble over some odd things. But instead of blindly refactoring, I think its better to ask you before. Fonts have among other things the properties ascent, descent and fontBBox which are defined in the PDF spec chapter 5.7 Font Descriptors. If I look into PDFFactory rev. 800217 line 1445 - 1483 method makeFontDescriptor, I see two constructor calls, one for CID fonts and one for the other fonts. For CID fonts the constructor of PDFCIDFontDescriptor is called and for the others the constructor of PDFFontDescriptor is called. PDFCIDFontDescriptor is a subclass of PDFFontDescriptor. The odd thing here is, that the call of PDFCIDFontDescriptor doesn't use the ascent nor the descent of the font descriptor. Instead it uses some quantities out of the fontBBox to feed the PDFFontDescriptor. This can be seen inside PDFCIDFontDescriptor rev. 679326. So as fontBBox, ascent and descent don't be the same thing, I would recommend to change this. But this would properly break some things, because we would get other values for ascent and descent in CID fonts. Best Regards Alex - e-mail: alexanderk...@gmx.net web: www.alexanderkiel.net
Re: PDFFontDescriptor Ascent Descent and FontBBox
Hi Max, thanks for pointing me to fontbox. As I did not find a repository with a trunk, I had a look into fontbox-0.8.0-incubating. They have quite clean code to parse TrueType files. But they are also not able to read any OpenType data. They are even not able to read kerning data as FOP can already. So I think, I will continue my OpenType effort inside FOP for now, but we should consider to merge the font parsing into fontbox. Lets see what I can do in FOP right now. Currently I have all the current features refactured and written some tests against real fonts. I'm on the way to get kerning info out of the OpenType GPOS table. I have already written 19 test classes and added/modified 129 production classes. Best Regards Alex - e-mail: alexanderk...@gmx.net web:www.alexanderkiel.net On Tue, 2009-09-22 at 11:22 +0200, Max Berger wrote: Alexandar, on a completely different note: It may be interesting to also look into fontbox (part of pdfbox), which is now also an apache project, and therefore we could use source synergy. http://incubator.apache.org/pdfbox/ For the issue you've mentioned: This may be due so some issues with CID fonts not properly defining ascent and descent, but I don't have a reference for it. We would probably need some fonts to test and check. Max 2009/9/21 Alexander Kiel alexanderk...@gmx.net: Hi, as I'm currently on the way to redesign the TrueType/OpenType file reading, from time to time I stumble over some odd things. But instead of blindly refactoring, I think its better to ask you before. Fonts have among other things the properties ascent, descent and fontBBox which are defined in the PDF spec chapter 5.7 Font Descriptors. If I look into PDFFactory rev. 800217 line 1445 - 1483 method makeFontDescriptor, I see two constructor calls, one for CID fonts and one for the other fonts. For CID fonts the constructor of PDFCIDFontDescriptor is called and for the others the constructor of PDFFontDescriptor is called. PDFCIDFontDescriptor is a subclass of PDFFontDescriptor. The odd thing here is, that the call of PDFCIDFontDescriptor doesn't use the ascent nor the descent of the font descriptor. Instead it uses some quantities out of the fontBBox to feed the PDFFontDescriptor. This can be seen inside PDFCIDFontDescriptor rev. 679326. So as fontBBox, ascent and descent don't be the same thing, I would recommend to change this. But this would properly break some things, because we would get other values for ascent and descent in CID fonts. Best Regards Alex - e-mail: alexanderk...@gmx.net web:www.alexanderkiel.net signature.asc Description: This is a digitally signed message part
Re: PDFFontDescriptor Ascent Descent and FontBBox
Hi Alexander, Alexander Kiel wrote: Hi Max, thanks for pointing me to fontbox. As I did not find a repository with a trunk, I had a look into fontbox-0.8.0-incubating. They have quite clean code to parse TrueType files. But they are also not able to read any OpenType data. They are even not able to read kerning data as FOP can already. So I think, I will continue my OpenType effort inside FOP for now, but we should consider to merge the font parsing into fontbox. It all depends on which code has the most potential. If FontBox offers a cleaner basis then it makes sense to implement missing features there, then use it as a replacement of FOP’s own font code. I think it would be unfortunate to duplicate efforts on that area. I’m not sure we can afford it given our limited resources. OTOH, you seem to have made quite some progress on the FOP code already, so it’s up to you. Just make sure the font library is totally independent of any FOP specificity. Lets see what I can do in FOP right now. Currently I have all the current features refactured and written some tests against real fonts. I'm on the way to get kerning info out of the OpenType GPOS table. I have already written 19 test classes and added/modified 129 production classes. Sounds like non-trivial changes, so an ICLA will probably be necessary to integrate your work in the code base. Since the process takes a bit of time, I’d suggest you to submit one as soon as possible: http://www.apache.org/licenses/#clas Thanks and good luck, Vincent Best Regards Alex - e-mail: alexanderk...@gmx.net web:www.alexanderkiel.net On Tue, 2009-09-22 at 11:22 +0200, Max Berger wrote: Alexandar, on a completely different note: It may be interesting to also look into fontbox (part of pdfbox), which is now also an apache project, and therefore we could use source synergy. http://incubator.apache.org/pdfbox/ For the issue you've mentioned: This may be due so some issues with CID fonts not properly defining ascent and descent, but I don't have a reference for it. We would probably need some fonts to test and check. Max
