Mime type detection in beagle [Was: IndexHelper eating cpu]

[Stephan Hegel <stephan hegel gmx de>]

Hi all,

Debajyoti Bera wrote:
> For update, the file above turned out to be a jpeg file which was causing a 
> loop in the jpeg handling routine. 
I've got the impression that the mime type detection with xgdmime depends
too much on the file extensions instead of the files' contents. A plain "file"
command or for images the "identify" utility from the ImageMagick package
deliver by far more accurate results.

Would it be easily possible to use these external commands for mime type
detection in beagle ? Lets say, like an external pre-filter before applying
the real filter ?

A bit funny: due to the current mime type detection I could identify quite a
lot of wrongly named images and files in my $HOME and a large static index
kind of "10 years Linux Magazine ..." as they caused an exception in beagled
when they were run through the wrong filter ;).

Kind regards,
   Stephan,