OT: need help installing FuzzyOCR

[Hauke de Vries]

Good work, but you will see that Jorge Valdes on 
http://www.joval.info/proj/FuzzyOcr.html
has developed further on FuzzyOcr.

At early stages ImageMagick was needed, later he 
abondoned it and Jorge Valdes reintroduced IM.

I had a tough time rebuilding PerlMagick. I'm on FC4 
so wanted the latest. If you don't like the hassle, 
then install ImageMagick(-perl) thru yum.

First I tried:

	cpan> install Magick::Perl

which failed miserably, because it needs the 
accompanying libraries.

So downloaded IM.tar.bz2 to /usr/src/redhat/SOURCES.
Created and edited an IM.spec from an older release 
and put it in /usr/src/redhat/SPECS and then

	rpmbuild -ba IM.spec

with a lot of warnings. Installed the files and 
tested FuzzyOcr: "Unknown symbol: InitializeMagick".

Weird, and I finally created some symlinks in 
/usr/lib (for lib{Magick,Wand}), then
ldconfig, cwd perl, and

	perl MakePerl.PL
	make
	make test
	make install

This worked and I can't tell you why it couldn't 
find the correct libs at build-time?!?

Anyway I'm running the latest FuzzyOcr in debug mode 
and so-far-so-good. Only problem seems to be that 
gocr can't handle text on colored background well.

Create an test.eml with graphic and

	spamassassin -D --lint \
		-p /etc/mail/spamassassin/mailscanner.cf \
		< test.eml 2&>1 | tee test.debug

Restarting MailScanner is not needed because when SA 
is invoked it seems to reread everything.

Hauke

On 19 Oct 2006 at 0:26 Dhawal Doshy wrote:

> Julian Field wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> > 
> > Please can you add it to the SpamAssassin section of the MailScanner
> > wiki!
> > 
> > It is important to keep and share stuff like this.
> 
> Done,
> http://wiki.mailscanner.info/doku.php?id=documentation:anti_spam:spamassassin:plugins:fuzzyocr
> 
> Now some who actually uses this could please verify/validate the
> instructions.
> 
> - dhawal