Kobo Dictionary

[PeterT]

From looking at both a PC install of Kobo Desktop 2.02, and the zipped firmware 1.94 upgrade for a Kobo Touch it looks as if the dictionary is a series of files.

There is one binary file with an html extension for each pair of letters

28/05/2011  11:08 PM             2,027 aa.html
28/05/2011  11:08 PM            19,517 ab.html
28/05/2011  11:08 PM            34,131 ac.html
28/05/2011  11:08 PM            24,400 ad.html
28/05/2011  11:08 PM             6,675 ae.html
28/05/2011  11:08 PM            11,401 af.html
28/05/2011  11:08 PM            13,319 ag.html
28/05/2011  11:08 PM             1,324 ah.html
28/05/2011  11:08 PM             9,060 ai.html
28/05/2011  11:08 PM               331 aj.html
28/05/2011  11:08 PM             1,087 ak.html
28/05/2011  11:08 PM            36,961 al.html
28/05/2011  11:08 PM            24,794 am.html
28/05/2011  11:08 PM            54,143 an.html
28/05/2011  11:08 PM               845 ao.html
28/05/2011  11:08 PM            22,460 ap.html
28/05/2011  11:08 PM             2,070 aq.html

and then one large plain-text file that lists all the words

13/12/2010  10:21 AM         1,081,793 words

On the PC they are located in (assuming default locations) either c:\Program Files\Kobo\dicthtml OR c:\Program Files (x86)\Kobo\dicthtml.

[Jozsi01]

I can add that the files are compressed with gzip. They can be uncompressed to the original html format (eg.: mv ab.html ab.html.gz ; gzip -d ab.html.gz )
I don't know what format is the 'words' file - maybe some known dictionary format...

[miwie]

The words file probably is a Marisa file.
See: http://code.google.com/p/marisa-trie/