[Tolriq]

Most of the ways to get non UTF-8 in Xbmc database have been resolved in recent Frodo builds (Sanitize NFO / Music Tags) but it seems that during database upgrade from Eden the old data is not sanitized and can still contain non UTF data.

Since JSON / JSON RPC specs are quite clear about data needing to be UTF-8 encoded this leads to most parser having troubles.

Earlier Montellese was quite clear that this sanitize job is not a JSON problem but an Xbmc internal problem is it possible to have some solution for the users who migrate from Eden ?

[jmarshall]

Eden databases should be entirely utf8 anyway. There was only a small window when taglib was initially put in place that screwed things up. i.e. nightly users.

[Tolriq]

Nope there was a problem with NFOs.

http://trac.xbmc.org/ticket/13247

And since the Wiki does give samples of NFO that do contain the utf-8 header lot's of users have bad NFO created with basic editor that did not generate correct UTF encoding.

I can assure you I have quite some users with this problems.

[jmarshall]

Right, but as pointed out in the ticket you link to, that problem has been fixed. We cannot fix the data internal to the database at this point, at least not realistically.

Since said users have their nfo files, they can just re-scan to fix the problem, right?

[Tolriq]

It seems in some case they need to change the nfo for the data to be regenerated but I don't know in witch case.

Is there a way to force a full rescrape and data refresh ? Since it's hard to ask a user to change a very large numbers of NFO.

[jmarshall]

Set content to None. Yes to clean out. Set content to Movies. Yes to refresh.

[Tolriq]

Thanks will add that to my Wiki