[mythtv-users] Unicode characters showing up as question marks

Mr. Demeanour mrdemeanour at jackpot.uk.net
Tue Jan 16 14:45:46 UTC 2007

Dave M G wrote:
> I make bilingual Japanese/English web sites, so this is something I
> have a little experience with.

I see; I apologise for trying to sell you lessons in egg-sucking.

> When dealing with utf-8, it's important to make sure every step of
> the process is utf-8. In MySQL, there are "collation" settings which
> describe the character set used for fields, tables, and the whole
> database.

Are you sure? "Collation" normally refers only to the way a database
decides whether one character precedes another - for purposes of sorting
and comparison. It shouldn't be directly related to how data is encoded.
> Whoever designed the MySQL database left the collation settings at
> the default - which is (for some reason) "latin_swedish_ci".

That might be a default for the MySQL product itself.
> I changed all the text fields on the "program" table to use 
> "utf8_unicode_ci" as their collation. Then I set the collation for
> the table and the collation for the whole database to be
> "utf8_unicode_ci" as well.
> I wiped the table clean, and used mythfilldatabase to populate it
> again.
> And presto - the data was stored in the database in its proper utf8 
> encoded Japanese.

I can't claim to understand why that worked; but there ya go.
> Okay, well, but what about in the MythTV interface?
> When I go to MythTV, and call up the onscreen menu ("m" button on the
>  keyboard), then go to "Programme Guide", it comes up in Japanese!
> Great.

> But we're not out of the woods yet.
> If I press up or down keys on the keyboard to move channels, then a 
> brief summary of the channel guide appears in a band at the bottom of
>  the screen.

That's the OSD.

The channel number is in a red tab, and the summary text is
> in a blue bar.
> That text shows up as little boxes. Like I said, I work a lot with 
> encoding issues, and I know those boxes mean that the font can't find
>  the right character. In other words, it's not a unicode enabled
> font.

The OSD font is distinct from the font used for the rest of the UI. The
UI font may be taken from the QT settings (I don't know - I'm sure
someone else here will be able to clarify). I think the OSD font is
specified explicitly in Frontend Settings. So you might have some luck
tweaking the frontend Settings to use a Unicode font that contains
Japanese characters.


More information about the mythtv-users mailing list