[mythtvnz] http://epg.org.nz/freeview.xml.gz

[David Moore]

On 26/03/13 14:27, Stephen Worthington wrote:
> On Tue, 26 Mar 2013 11:55:36 +1300, you wrote:
>
>> On 26/03/13 03:56, Stephen Worthington wrote:
>>> On Mon, 25 Mar 2013 20:50:20 +1300, you wrote:
>>>
>>>> On 25/03/13 15:48, Stephen Worthington wrote:
>>>>> On Mon, 25 Mar 2013 15:24:36 +1300, you wrote:
>>>>>
>>>>>> On 25/03/13 03:08, Stephen Worthington wrote:
>>>>>>> I have run mhegepgsnoop.py (0.5.1) with the -v (verbose) option, and
>>>>>>> it seems that my problem is caused by its fuzzy channel matching - it
>>>>>>> is picking up the wrong channels from my MythTV database when it
>>>>>>> matches them against my channels. Having all the Sky versions of the
>>>>>>> Freeview channels in my database is what is triggering the problem. I
>>>>>>> will have a look tomorrow to see if I can make a modification to the
>>>>>>> channel lookups so that it will only match against DVB-T channels.
>>>>>> I have just discovered a related problem with the fuzzy channel matching
>>>>>> on my system. I'm getting Canterbury Television matched against
>>>>>> fv-maoritelevision. And I'm in Auckland. :) Not sure if there is a bug
>>>>>> in how I use the fuzzy matching lib or I'm just unlucky with channel
>>>>>> names. I'm thinking about adding another option to exclude specific
>>>>>> channels discovered in the MHEG data. Or try and improve fuzzy matching.
>>>>>> Or both. Don't hold your breath waiting for a fix from me...
>>>>> It seems that the correct spelling of Maori with a macron over the o
>>>>> character in the MHEG5 data is confusing things.  I am working on a
>>>>> possible fix for that at the moment.  I think I already have a fix
>>>>> that makes the fuzzy matching only work on the DVB-T channels, which
>>>>> is to change the select query used to get the MythTV channel data to:
>>>>>
>>>>> select callsign, name, xmltvid from channel c, dtv_multiplex d where
>>>>> c.mplexid is not NULL and c.mplexid = d.mplexid and d.mod_sys !=
>>>>> 'DVB-S'
>>>>>
>>>>> (This is in two places in the mhegepgsnoop.py file).  This fix needs
>>>>> testing on databases other than mine that have satellite Freeview
>>>>> channels and/or Sky channels, to make sure the database fields all
>>>>> work the way I thing they do.
>>>>>
>>>>>
> [snip]
>>>> Well I decided to have a crack at it and found a bug in the fuzzy
>>>> channel matching. Should be fixed now. I made a mistake with sorting the
>>>> matching results. Seems to work properly now and has no problem
>>>> correctly matching the funny character in "Maori..".
>>>>
>>>> Also added a trap for the index error in the tuning function. Haven't
>>>> tested this on real data.
>>>>
>>>> Version 0.5.2 download is ready at
>>>> http://sourceforge.net/projects/mhegepgsnooppy/
>>> The trap for the index error is working, but the fuzzy matching is
>>> not.  I am now getting multiple channel id entries in the output file
>>> for the same channel:
>>>
>>> ....
>> Yes, I see what your problem is. I only have DVB-T so I missed the point
>> that you have many non-DVB-T channels in your mythtv setup. The fuzzy
>> matching does do correct matches now. This was the bug I fixed
>> yesterday. The problem is that it doesn't (yet) filter out non-DVB-T
>> channels. Maybe your modified select query is the way to go or maybe
>> filter on source id? I'm not sure and I can't test or inspect the
>> database because I don't have anything except DVB-T. I suppose the quick
>> fix is to add your query as a command line option so it can be turned
>> off if it doesn't work for someone else.
> Filtering on a sourceid specified on the command line should work if
> the setup has one source for DVB-T and DVB-T only, which is probably
> the way things should be set up.  But there is always someone out
> there who has something different.
>
> I have put my select statement back in and it now all works properly
> on my database.  Including Maori TV - your fix tidied up that problem
> nicely, thank you.  But the new select really needs testing on more
> than just my database.  I have tested it on my Mother's MythTV box,
> which has only DVB-T, and it works fine there.
>
> Here is a diff:
>
> root at mypvr:/usr/local/bin# diff mhegepgsnoop-0.5.2.py mhegepgsnoop.py
> 3c3
> < VERSION = '0.5.2'
> ---
>> VERSION = '0.5.2 JSW'
> 6c6
> < Version: 0.5.2
> ---
>> Version: 0.5.2 JSW modified version
> 95a96
>> 0.5.2 JSW Change selects used to get channels from MythTV to only get DVB-T channels.
> 870c871
> <                       dbconn.execute("select callsign, name, xmltvid
> from channel")
> ---
>>                        dbconn.execute("select callsign, name, xmltvid from channel c, dtv_multiplex d where c.mplexid is not NULL and c.mplexid = d.mplexid and d.mod_sys != 'DVB-S'")
> 883c884
> <                       f = os.popen("mysql -ss " + mysql_args + " -e
> 'select callsign, name, xmltvid from channel' mythconverg")
> ---
>>                        f = os.popen('mysql -ss ' + mysql_args + ' -e \'select callsign, name, xmltvid from channel c, dtv_multiplex d where c.mplexid is not NULL and c.mplexid = d.mplexid and d.mod_sys != "DVB-S"\' mythconverg')
> I only use the -p option as without it, mhegepgsnoop.py does not get
> access to my database properly.  So the non -p version  (the second
> select) has not been tested.  Given the complex quoting in that line,
> it needs testing by someone.
>
I can test the mysql string on my system. Well a limited test anyway 
because I don't have any DVB-S channels to filter out.