[mythtv-users] Schedules Direct guide data for AU / NZ

[Mark Perkins]

> I would recommend the tv_grab_zz_sdjson_sqlite grabber, it's quicker, more
> efficient, supports more of the XMLTV API and is still being actively
> developed, but it's not in the official release yet only the nightly
builds.
> 

Not sure if the following is a review of the Schedules Direct service or the
tv_grab_zz_sdjson_sqlite grabber, or a bit of both - but my notes so far
after setting up tv_grab_zz_sdjson_sqlite grabber. I am currently using a
slightly modified version of Shepherd as the comparison.

By and large the main information (channel, starttime, title, subtitle) are
the same between the two. So for example on Channel 10 Offspring episodes I
get no subtitle on either Shepherd or zz_sdjson_sqlite. Same with ABC2 The
Code.
There were some differences - ie Shepherd has "The Batman: Night of the
Batmen!" whereas zz_sdjson_sqlite shows "Batman: The Brave and the Bold:
Night of the Batmen!". By TheTVDB zz_sdjson_sqlite would be correct.
Categories were different between the two but I don't really ever use
categories so I largely ignored.
With zz_sdjson_sqlite I got pretty much a full 7 days worth of guide data.
Whereas with Shepherd day 5 is pretty sparse (maybe 5 channels only) and
nothing for days 6 and 7.
I won't comment on the program descriptions because I replace the Shepherd
program description data with the description from TheTVDB where-ever I can
programmatically so I can't easily tell whether the Shepherd data I have is
actually from Shepherd or TheTVDB.
zz_sdjson_sqlite doesn't seem to have stars or originalairdate info. I use
both so a little bit of an issue for me.

select distinct stars from program;
+-------+
| stars |
+-------+
|     0 |
+-------+

select distinct originalairdate from program;
+-----------------+
| originalairdate |
+-----------------+
| NULL            |
+-----------------+

select distinct airdate from program;
+---------+
| airdate |
+---------+
|    2016 |
+---------+

The ratings are different but not sure one is necessarily better or worse
than the other. Biggest difference is that mythweb seems to assume that a
program will only have one entry in the programrating table for each program
which is the way Shepherd does it, whereas zz_sdjson_sqlite has a separate
entry for each rating.

ie from zz_sdjson_sqlite (mythweb only displays "Drug use"):
select * from programrating where chanid=1001 and starttime="2016-09-10
01:30:00";
+--------+---------------------+----------+------------------+
| chanid | starttime           | system   | rating           |
+--------+---------------------+----------+------------------+
|   1001 | 2016-09-10 01:30:00 | advisory | Drug use         |
|   1001 | 2016-09-10 01:30:00 | advisory | Sexual reference |
|   1001 | 2016-09-10 01:30:00 | CB       | PG               |
+--------+---------------------+----------+------------------+

>From Shepherd (mythweb displays "PG d,s"):
select * from programrating where chanid=1001 and starttime="2016-09-10
01:30:00";
+--------+---------------------+--------+--------+
| chanid | starttime           | system | rating |
+--------+---------------------+--------+--------+
|   1001 | 2016-09-10 01:30:00 | ABA    | PG d,s |
+--------+---------------------+--------+--------+

It probably goes without saying that all the programid values are different
so moving to the zz_sdjson_sqlite will require battling through that
conversion period. I scrubbed the programid values using the sql code on the
wiki and that partially helped but a lot of my earlier recordings use
generic descriptions so they were showing to rerecord. For tomorrow, just as
an example, that pushed me up from about 13 recordings for the day to 35.

Note that I was using 0.27 for Shepherd and 0.28 for zz_sdjson_sqlite. I
assume that it makes no difference but will declare it up front in case it
does.

I will endeavour to test sd_json next and see if that materially changes any
of the above.