uchardet-enhanced / README.txt

medoc c6b73a8 


This package implements improvements forr the mozilla universalchardet
module described in:

The interface uses the C wrapper described here:

Modifications / improvements affect the core universalchardet module, not
the C wrapper (which is useful and necessary anyway for integration and

Directory contents:

 - Wikipedia index pages in target languages, sometimes in multiple
   encodings. The pages were manually stripped of english and boilerplate
   content, in hope that the remaining is significant and typical text.

 - Used to check how the detection works.

 - Data and code used to produce the bigram frequencies for a
   language/encoding pair, used for the "Two char Distribution Method"
   from the reference article (neither the article nor the mozilla module
   publish the scripts used to generate the tables or the reference data).

 - The C API from the reference above, with the modified mozilla code
   inside libcharsetdetect/mozilla/extensions/universalchardet/src/base/