[mythtv-users] Incorrect characters in the EPG and program name

[Karl Dietz]

On 13.01.2013 22:16, spst wrote:
> 2013. 01. 13, vasárnap keltezéssel 20.56-kor Karl Dietz ezt írta:
>> On 13.01.2013 19:58, spst wrote:
>>> Hi,
>>> I started to know MythTV a few months ago. I successfully installed on
>>> opensuse 12.1 64 bit from packman repo. I have many questions because
>>> Google has't helped me, yet but these aren't important.
>>>
>>> I first installed the mythtv 0.24, but now I'm trying 0.26. I have a
>>> problem what is same for each version what I tried. EPG and program name
>>> encoding. I use a DVB-C card to watch cable tv on Hungary but character
>>> encoding is wrong. Hungarian words with accented letters can't be read.
>>
>> Sometimes providers don't signal the character encoding properly.
>
> Yes, i maybe but my two Samsung tvs can show it correctly.

It appears that most TV sets take no signaling as something else then 
written in the standard. (which would be a variant of ISO 6937)

>>   >  How can I fix it?
>>
>> You have to look at the raw data and check what they are signaling
>> http://dvbsnoop.sourceforge.net/examples/example-eit.html
>> then look at which encoding they really use and add a fixup for it
>> to MythTV at
>> http://code.mythtv.org/cgit/mythtv/tree/mythtv/libs/libmythtv/eithelper.cpp#n991
>> (notice all the EITFixUp::kEFixForceISO8859_*)
>
> I tried to dump stream but it was unsuccesfull... Someone can help how
> can I do it? I read a lot of net (google) but it was unsuccessfull...

You have to manually tune your card to a transport to look at either
with MythTV or dvbtune, then you can look at the raw data with
dvbsnoop -adapter <the adapter number> 0x12

> I think my provider doesn't use ISO8859-2 because it not applicable in
> Hungary. Provider didn't specified this parameter.

I think your provider uses either ISO-8859-2 or -16 (guessing from
Wikipedia's hint on the supported languages) but does not signal
anything which is interpreted as the DVB variant of ISO/IEC_6937.
Maybe they are using UTF-8 or UCS-2, a Windows encoding or something
custom (unlikely).

>> The fixes are keyed off the original_network_id, transport_id and/or
>> service_id.
>> If *all* services have identical wrong signaling its just one simply
>> entry. It gets complicated if that is not the case
>> (see the link above :)
>
> I think it use an uniqe encoding for Hungarian character.

If you have the raw data (including the hex dump), and can specify the 
real characters that must be shown instead, we can narrow it down to one
character encoding. (might be easiest to refer to the correct characters
by their english wikipedia page, e.g. http://en.wikipedia.org/wiki/Á or
their name, a-acute in the example)

Regards,
Karl