[pdftex] Reflowing text

[Morten Høgholm]

On Fri, 28 Jan 2005 00:47:49 +0100, Petr Sojka <sojka at fi.muni.cz> wrote:

I'm adding Robin and Chris as they have interest in this subject too.

> On Fri, Jan 28, 2005 at 12:30:07AM +0100, Reinhard Kotucha wrote:
>> >>>>> "Morten" == Morten H�gholm <moho01ab at student.cbs.dk> writes:
>>
>>   > I've done a little testing now. It seems that when searching for a
>>   > text string, Acrobat disregards kerns less than app 1.13bp, thus
>>   > t\kern1.1bpest is still recognized as the word "test". The reflow
>>   > algorithm on the other hand considers any kern an artifact instead
>>   > of real content.
>>
>> Is this space (1.13bp) fixed or does it depend on the size of the
>> font?
>
> I vaguely remember that Acrobat uses algorithm heuristics
> for space/kern decision based on average/weighted skip sizes in the  
> given font,
> but Adobe hides the details of it so that nobody else can make
> search function exactly identical to that in Adobe's Acrobat.

Ah, that sound like Adobe. I've just signed up for a free seminar on  
Acrobat 7 although it's probably just a three hour commercial. Free food  
though! I should try to ask them if I get the chance.

The following testfile proves that Petr's memory isn't as vague as he  
claims :-)

\pdfcompresslevel0
\font\bigtenrm=cmr10 scaled 2000
\def\test#1{Some t\kern#1\fontdimen2\font ext \par}
\test{.34}% "text" recognized
\test{.341}% "text" not recognized
\bigtenrm
\test{.34}% "text" recognized
\test{.341}% "text" not recognized
\bye
-- 
Morten Høgholm