Tvheadend

Ben Crundwell wrote:

I've searched through and can't appear to find any information, but apologise if the question has already been asked:

I am transferring over from a Windows based Argus setup to a new Ubuntu TvHeadend server. I have a working setup with a Schedules Direct XML EPG and am starting to transfer over my Series Recordings.

On the old Argus setup, it was possible to filter by Season/Episode number by adding the SxxExx data to the front of an episode's subtitle and then add "NOT S01 AND NOT S02" etc into the recording schedule. This is really useful if you only want to record the latest Season of say Family Guy, without filling up your DVR with 14 seasons of random episodes.

However with TvHeadend I cant find an easy way of adding the same functionality. What would be fantastic would be a RegEx filter in the AutoRec settings for Episode. Or failing that, one on the SubTitle so I can pre-process the XMLTV file to add SxxExx to that.

In 4.1 you can use regex in the title and optionally enable "Full text" which basically applies the regex to the subtitle, description and summary. I'm not sure it applies to season variables though.

I read that over time the database figures out what its already recorded and ignores it in future. Is it possible to add episodes to this list manually?

If your data has a repeat info you could set the "Broadcast type" option, e.g.

Can anyone offer some assistance in how to setup an AutoRec to only record a specific Season of something?

I've gotta say, REGEX in TVHeadend is absolutely the best feature for power users. I'm only just getting to grips with REGEX but I've replaced all my autorec entries with REGEX equivalents and am finding them extremely useful.

My recent use of this was to set a recording of the Premiership football highlights on Sky Sports but only where my team was playing either the home or away fixture (so where the title of the programme was MyTeam vs. SomeOtheTeam or SomeOtherTeam vs. MyTeam) - works beautifully.

Here's the REGEX I used for that little gem

PL( )?-( )?Burnley v(s.)?( )?(.*)|PL( )?-( )?(.*) v(s.)?( )?Burnley

Is is possible to preview the results of a RegEx? Would be nice to check what shows match to confirm its working right without having to trawl through the upcoming recordings and check.

This works nicely in the EPG view with a search filter, but i'm not sure you can do a RegEx in that view

Mark Clarkstone wrote:

If your data has a repeat info you could set the "Broadcast type" option, e.g.

How come my AutoRec dialog is missing a bunch of options compared with yours?

Sorry, my 4k monitor makes screen clipping ridiculous

Probably because you're not running a current 4.1 development build ...

Robert Cameron wrote:

Probably because you're not running a current 4.1 development build ...

Apologies for the stupid question, but is the dev build relatively 'stable' and if so how do I change build? I'm on 4.0.9-9~gd5af55c~trusty

Ben Crundwell wrote:

Robert Cameron wrote:

Probably because you're not running a current 4.1 development build ...

Apologies for the stupid question, but is the dev build relatively 'stable' and if so how do I change build? I'm on 4.0.9-9~gd5af55c~trusty

Yeah sorry about that I should have said, you can change your package sources to unstable if you want to give it a go but be warned anything could happen.

So backup your configuration for 4.0.x (something like sudo cp -ar /home/.hts/.hts ~/ should do it), that way if you're not happy or have issues you can uninstall 4.1x (use --purge), change your apt sources back to stable, reinstall 4.0 & copy your old config folder back.

I'm using the 4.1/development branch in my home environment which has to stand up to the scrutiny of my wife so it's fair to say it's fairly (very) reliable but of course it's a development build so can go from superb to shocking in one update, but I've been building and running from git-master for six months with no adverse effects. It's well worth updating for all the advances in 4.1 over 4.0.

I've actually just worked out the REGEX to grab a specific series of a programme;

TITLE.*SEASON.*

example:

NCIS.*S10.*

...returns only episodes from season 10 of NCIS. You need to enable the 'Fulltext' option for this to work.

Ben Crundwell wrote:

Is is possible to preview the results of a RegEx? Would be nice to check what shows match to confirm its working right without having to trawl through the upcoming recordings and check.

This works nicely in the EPG view with a search filter, but i'm not sure you can do a RegEx in that view

Just switch to the EPG tab, enable fulltext and paste your regex into the search field - it'll search through however much data you have cached in your EPG database (usually around 7-8 days worth).

Nice! I'll try that when I get home. I upgraded to the unstable branch successfully the other night too. Realised that the stable one is many many months old now! :S

Jonathan Thomson wrote:

...returns only episodes from season 10 of NCIS. You need to enable the 'Fulltext' option for this to work.

This doesn't work for me. Have you already appended the S10E01 to the Subtitle? As my RegEx appears only to search Title and SubTitle, not Episode.

I could not get this to work. I use the "New/Premiere" broadcast flag setting to only record new episodes, as TVH's interface and rules is not very good for interacting/limiting/searching based upon the EPG metadata.

Sorry folks, it turns out it was matching against the episode number that's contained within the body of the description - can't believe I didn't spot that

I finally got around to trying to fix this in my own weird way and thought I should update the forum, in case anyone else has the same problem.

I've modified the XMLTV JSON grabber script to append the episode and series number onto the front of the description and have uploaded my update here: https://github.com/bencrundwell/tv_grab_sd_json

However I'm still struggling to get the RegEx working I have Fulltext enabled, but can't appear to get it to match from both the title and the description fields simultaneously.

As an example.
Title:Big Bang Theory
Description: S09E06 - Leonard and Sheldon go to great lengths to secure the helium they need during a nation-wide shortage; Bernadette and Penny try to to find Amy a new man on a dating app.

I can find this episode with either:
Big Bang
or
S09E06

but can't find any RegEx that will combine the two to give a specific episode of a specific program.

Does anyone have any recommendations, or at least an idea of what the source text the RegEx is being applied to is, so I can figure it out?

Maybe ^Big Bang Theory*S09E06 ?

Nope, returns nothing when i test it on the EPG

I'm starting to think the search function parses the title and description as separate entries as any attempt to get it to match X AND Y fails.

Perhaps the only option is to also add the title into the description so it can search in a single field.

Yeah that's working now. I've updated GitHub. Try using the following formatting:
(Big Bang Theory).*(S09)

Enjoy

This thread is exactly what I am after but I cannot quite get it to work for me.

At the moment, I want to record every episode of Wheeler Dealers from Season 12 onwards.

What would I do to set up a timer for this?

Bump!

I got the same problem as AL.

Take this case as example:

Title         | Subtitle
----------------------------------------------
The Simpsons  | Episode title yada yada (1990)
The Simpsons  | Episode title yada yada (1991)
The Simpsons  | Episode title yada yada (2018)

Now I'd like to not record all 30 years of episodes, but only the current ones. But a search string like
(Simpsons).*(2018)
with fulltext checked turns up an empty result. Seems whatever I do my regexps can only match against title OR subtitle, but never a pattern in both combined.

Any ideas?

Regular expressions are applied per field so do not combine title, subtitle and description.

The reason it doesn't work for AL is that you use a different grabber. Ben said he has his own variant of tv_grab_sd (not the sqlite version you use).

You can either use his, or you could try my unsupported fork of tv_grab_zz_sdjson_sqlite on github [[https://github.com/azlm8t/tv_grab_zz_sdjson_sqlite]]

Just copy the tv_grab_az_sdjson_sqlite file (yes, "az" not "zz" in the filename to avoid conflict) in to the same directory as your other grabbers (maybe /usr/local/bin) ensuring it is executable (sudo chmod 755 /usr/local/bin/tv_grab_az_sdjson_sqlite).

If you use mine then you need to:

1. Backup your current .xmltv configuration from ~/.xmltv (since my version patches the DB to download artwork details);
2. Change the grabber to tv_grab_az_sdjson_sqlite (look for it in "path") in tvheadend's configuration->channel/epg->epg grabber modules, and add extra arguments of "--update-description-with-title" or "--update-description-with-all" (without the quotes) which has more description fixups such as adding year, age rating and stars. I'd also add "--artwork-max-width=720" to avoid slow artwork downloads so it becomes "--update-description-with-all --artwork-max-width=720 --merge-split=5";
3. ???
4. Profit.

With the "all" option, the description is changed to be something similar to:

NCIS - S07E07 - Endgame.

S7/15 E7/24. (2009-11-10). (TV-14). Vance's family...

So this can be searched by a fulltext regex such as "^NCIS.*S07" or "^NCIS.S07E0?". Using "^NCIS.*2009" does not work (since it is not on one line), but with pcre2 (in newer Tvheadend versions) something like "(*CR)NCIS.*2009" will work, as will, for example, "(*CR):NEW:LIVE:.*Cricket" to detect live cricket matches.

Due to merging of SD hierarchy of data, sometimes you get bad data such as "S3/2" (season 3 of 2 seasons). And, of course, descriptions can vary over time, so "S3/4" may become "S3/5" in a few years' time, so dup matching on descriptions will fail.

The date in the example is the first broadcast date of the episode, but for generic shows such as news it is often the date of the first ever episode.

Mine does a few things slightly different to the "zz" grabber, such as using ttvdb episode numbers (which sometimes differ from SD episode numbers, but does Kodi mismatches happen less frequently); flagging "premiere"; removing five minute "news" programmes (with --merge-split=5), and some other minor features.

It works for me. YMMV. If it doesn't work then you can revert back to your other version.

It is also possible to run this script on libreelec?
i put the "az" file to the others tv_grab files, "/storage/.kodi/addons/service.tvheadend44/bin/tv_grab_az_sdjson_sqlite"
It will not be recognized after a reboot..

To use the script, do I need an account at schedulesdirect.org?
or i can also attach my own xml file, from webgrabplus?

Yes, this only would only work with SchedulesDirect so needs an account with them.

I think LibreElec used to only have tv_grab_file and didn't (doesn't?) ship with perl (which this and most other grabbers use) so I'm not sure if it works on LibreElec.