FontForge -- An Outline Font Editor

On Thu, 2006-01-05 at 08:06, Pablo Rodríguez wrote:
> Checking the feature to import fonts from a PDF file, I have realized 
> that fontforge is not able to import the polytonic Greek characters from 
> the fonts located at: http://www.magenta.gr/en/uc5/en_unicode_5.htm.
> 
> Fontforge should open all embedded glyphs in their correct possition, 
> shouldn't it?
Um yes. But most fonts in pdf files are subsets of the full font.
FontForge does what I expect it to do on the one file I looked at. Can
you be more specific in what you believe to be wrong?

George Williams wrote:
> On Thu, 2006-01-05 at 08:06, Pablo Rodríguez wrote:
> 
>>Checking the feature to import fonts from a PDF file, I have realized 
>>that fontforge is not able to import the polytonic Greek characters from 
>>the fonts located at: http://www.magenta.gr/en/uc5/en_unicode_5.htm.
>>
>>Fontforge should open all embedded glyphs in their correct possition, 
>>shouldn't it?
> 
> Um yes. But most fonts in pdf files are subsets of the full font.
> FontForge does what I expect it to do on the one file I looked at. Can
> you be more specific in what you believe to be wrong?

Probably my expectations are be wrong. I think that FontForge should 
import all (polytonic Greek) glyphs embedded in that files. If they are 
missing I don't quite see why this importing is right. If you want to 
discard accented glyphs, then it should be fine (although I don't 
understand why).

Thanks,


Pablo

On Thu, 2006-01-05 at 11:10, Pablo Rodríguez wrote:
> Probably my expectations are be wrong. I think that FontForge should 
> import all (polytonic Greek) glyphs embedded in that files. If they are 
> missing 
They are missing from the font in the pdf file.
> I don't quite see why this importing is right. If you want to 
> discard accented glyphs,
FontForge doesn't discard anything. The program which generates the pdf
file remakes the font and only includes a subset of the glyphs.

George Williams wrote:
> On Thu, 2006-01-05 at 11:10, Pablo Rodríguez wrote:
> 
>>Probably my expectations are be wrong. I think that FontForge should 
>>import all (polytonic Greek) glyphs embedded in that files. If they are 
>>missing 
> 
> They are missing from the font in the pdf file.

They are included in http://www.magenta.gr/font_samples/mgoptima.pdf and 
http://www.magenta.gr/font_samples/mgkingscript.pdf (on page 3), but the 
problem is that the extended Greek characters are wrong encoded. I don't 
know whether the encoding mismatch is due to the font itself or the PDF 
document generation process.

Thanks,


Pablo

On Sat, 2006-01-07 at 03:52, Pablo Rodríguez wrote:
> They are included in http://www.magenta.gr/font_samples/mgoptima.pdf and 
> http://www.magenta.gr/font_samples/mgkingscript.pdf (on page 3), but the 
> problem is that the extended Greek characters are wrong encoded. I don't 
> know whether the encoding mismatch is due to the font itself or the PDF 
> document generation process.
Then please give a very specific description:
   Which font in the pdf are you talking about?
   What glyphs are in the pdf which are not in the fontforge display?
   How are they wrongly encoded?

George Williams wrote:
> On Sat, 2006-01-07 at 03:52, Pablo Rodríguez wrote:
> 
>>They are included in http://www.magenta.gr/font_samples/mgoptima.pdf and 
>>http://www.magenta.gr/font_samples/mgkingscript.pdf (on page 3), but the 
>>problem is that the extended Greek characters are wrong encoded. I don't 
>>know whether the encoding mismatch is due to the font itself or the PDF 
>>document generation process.
> 
> Then please give a very specific description:
>    Which font in the pdf are you talking about?
>    What glyphs are in the pdf which are not in the fontforge display?
>    How are they wrongly encoded?

Sorry for the inaccuracies.

The font is AFAAJM+MgOptimaUCPol in mgoptima.pdf file and 
OOFDAN+MgKingScriptUCPol in mgkingscript.

The glyphs that aren't in the fontforge display are polytonic Greek ones 
   (such as ῦ), that is mainly all accented characters on page 3 from 
both documents.

I think that these are wrong encoded since acroread 7 is not able to 
copy the text of these glyphs.

I hope it is clear now.

I experience something really strange. If these PDF document are 
uncompressed (using pdftk [http://www.pdftk.com] or Multivalent 
[http://multivalent.sf.net]), fontforge is able to find more fonts on 
the document. And trying to open OOENOK+MgPolKingScriptM in the 
uncompressed version of mgkingscript.pdf or AEPKPH+MgPolOptimaM in the 
uncompressed version of mgoptima.pdf leads to segmentation fault. 
AFAAJM+MgOptimaUCPol or OOFDAN+MgKingScriptUCPol are opened fine in the 
uncompressed versions.

I hope this helps,


Pablo

On Sat, 2006-01-07 at 12:26, Pablo Rodríguez wrote:
> The glyphs that aren't in the fontforge display are polytonic Greek ones 
>    (such as ῦ), that is mainly all accented characters on page 3 from 
> both documents.
Unfortunately your greek character displays as an underscore in my mail
reader. Accented characters can be built by composing base glyphs with
the appropriate mark glyphs using GPOS features -- there is no guarantee
that they will be real glyphs.
> 
> I think that these are wrong encoded since acroread 7 is not able to 
> copy the text of these glyphs.
Then why are you blaming fontforge? If the glyphs are wrongly encoded in
the pdf, the ff will just read what's in the pdf.
> 
> I hope it is clear now.
No, I think I'm still not understanding.
> 
> I experience something really strange. If these PDF document are 
> uncompressed (using pdftk [http://www.pdftk.com] or Multivalent 
> [http://multivalent.sf.net]), fontforge is able to find more fonts on 
Ah. There's a very odd object cross reference table that ff could only
half handle. So it missed some objects. The attached patch should fix.

But remember: ff can't handle all the compression modes used in pdf. LZW
is protected by patents for example. The docs do say that ff can only
read "many" of the fonts out of pdf files.
> the document. And trying to open OOENOK+MgPolKingScriptM in the 
> uncompressed version of mgkingscript.pdf leads to segmentation fault. 
Ah. There's no encoding data stored in that font, and ff didn't like
that. The attached patch should fix.

George Williams wrote:

> But remember: ff can't handle all the compression modes used in pdf. LZW
> is protected by patents for example. 

	According to The League for Programming Freedom 
(http://lpf.ai.mit.edu/Patents/Gif/Gif.html), the LZW patents have all 
expired.  Are there some auxiliary patents still lurking about?

Geoffrey Alan Washburn wrote:
> George Williams wrote:
> 
>> But remember: ff can't handle all the compression modes used in pdf. LZW
>> is protected by patents for example. 
> 
>     According to The League for Programming Freedom 
> (http://lpf.ai.mit.edu/Patents/Gif/Gif.html), the LZW patents have all 
> expired.  Are there some auxiliary patents still lurking about?

According to Wikipedia 
(http://en.wikipedia.org/wiki/LZW_compression_algorithm#Patent_issues), 
IBM still has a patent on the LZW algorithm that will expire in the US 
on August 2006.

But xpdf includes a workaround to handle LZW compression without the patent.


Pablo

Hi,

I don't know why FontForge doesn't import right the characters contained 
in http://www.typophile.com/forums/messages/29/ORIGEN_PDF-42729.pdf.

Although the font contains some non-standard ligatures, acroread and 
evince seem to search/copy text right.

FontForge chooses the Adobe-Identity-0 cidmap by default, but I guess 
this is fine.

Am I missing something?

BTW, it would be very useful for the user to be able to select many/all 
fonts in the font selection dialog when opening a PDF document.

Thanks,


Pablo

On Sun, 2006-02-05 at 06:16, Pablo Rodríguez wrote:
> I don't know why FontForge doesn't import right the characters contained 
> in http://www.typophile.com/forums/messages/29/ORIGEN_PDF-42729.pdf.
A cursory examination did not reveal any glyphs imported incorrectly.
The fonts, as usual, are missing lots of glyphs, but that's nothing to
do with fontforge. FontForge just reports what's in the pdf file, and
applications always store font subsets there.

Please be very specific when reporting bugs.
What characters are not imported correctly? Which font?

> Am I missing something?
I don't know. I don't know what problem you are having.
> 
> BTW, it would be very useful for the user to be able to select many/all 
> fonts in the font selection dialog when opening a PDF document.
Um, no. 
You really should not make a practice of pulling fonts out of pdf
documents because the fonts are not complete.