Use Original Airdate to determine between multiple title hits?

[evilpenguin]

An idea that just occurred to me when I was scanning though my missing metadata list: if a TV show title comes up with multiple hits if you do an air date search on all them prolly 99% of the time you'd be able to accuratly identify which series to use without the help of MediaTitles.xml.

[razrsharpe]

you're making the assumption that the original air date as set by the EPG is accurate though.... but ya i would think if you added logic around the OAD (and assumed it was correct) to figure out which series it was that solve the vast majority of the need for the mediatitles.xml

[stuckless]

Originally Posted by evilpenguin

An idea that just occurred to me when I was scanning though my missing metadata list: if a TV show title comes up with multiple hits if you do an air date search on all them prolly 99% of the time you'd be able to accuratly identify which series to use without the help of MediaTitles.xml.

You mean for the series? The media titles is used to identify a series... I'm still now sure how the air date is going to help me narrow down the series?

on recordings, I could probably use getSeriesInfo().getYear() to help narrow it down, but I'm not sure the air date will help... unless i'm missing something.

[razrsharpe]

Test Series: Parenthood

TVDB:
Parenthood - First Aired Date: 8/1/1990
Parenthood (2010) - First Aired Date: 3/2/2010

So you do something like this:

Get Original Air Date (OAD) from the Recording EPG metadata
Get Series List from TVDB, Get First Aired Date (FAD) for each series
Compare OAD to the FAD for each series.
Assume that the show belongs to the series with an OAD after the series FAD but before any other FAD of other series

Note: You are making the assumption that 2 shows of the same title do not originally air at the same time.

[evilpenguin]

What i'm thinking is this:

1. Do title search in theTVDB
2. Take all of the hits and do a check of the original airdate of the episode
3. Which ever returns a match is the correct show

Take last nights "The Biggest Loser" for example which the EPG comes down with the episode blank and the original air date set to "2011-01-25".

When BMT does a title search it returns 3 results...

• The Biggest Loser
• The Biggest Loser UK
• The Biggest Loser Australia

and it fails since without help from MediaTitles.xml it can't decide which to use.

But if you do an Original Airdate Check on each of the series...

• The Biggest Loser <-- Success
• The Biggest Loser UK <-- Fail
• The Biggest Loser Australia <-- Fail

The only one that returns data is the correct one.

As you say, the two issues that could come up are:

1. If the EPG has the incorrect OAD - If this is the case scraping will fail regardless so who cares?
2. If multiple titles returned have first run episode that air on the same day - The odds of this occurring are so small that its hardly worth considering. Edit: Actually I can think of a few cases where this is pretty possible, but mostly CBS and their army of CSI/NCIS spinoffs. While they don't currently air on the same night, I could see it happening. Although, I don't think this alone is enough of an exception to discount the method.

Looking at my MediaTitles.xml, a check like this would fix about half of my items (The Office, V, The Good Guys, Doctor Who, Human Target, Conan, Castle).

Edit: Actually this "The Biggest Loser" failure is pretty odd on its own. theTVDB has a direct title match with the EPG title, but for some reason BMT is choosing to check "The Biggest Loser UK"

2011-01-26 12:38:22,777 - ERROR; \\Secretivo\Data1\SageRecordings\TheBiggestLoser-8264267-0.mpg; The Biggest Loser; 8301668; Failed while parsing series: MediaSearchResult [extraArgs={CLEAN_TITLE: The Biggest Loser,YEAR: 0,RAW_TITLE: The Biggest Loser,EPISODE_DATE: 2011-01-18,}, id=92371, imdbId=null, metadata=null, providerId=tvdb, score=0.96, title=The Biggest Loser UK, type=TV, url=92371, year=0]
2011-01-26 12:38:23,405 - ERROR; \\Secretivo\Data1\SageRecordings\TheBiggestLoser-8285906-0.mpg; The Biggest Loser; 8324207; Failed while parsing series: MediaSearchResult [extraArgs={CLEAN_TITLE: The Biggest Loser,YEAR: 0,RAW_TITLE: The Biggest Loser,EPISODE_DATE: 2011-01-25,}, id=92371, imdbId=null, metadata=null, providerId=tvdb, score=0.96, title=The Biggest Loser UK, type=TV, url=92371, year=0]

[stuckless]

Thanks, I understand now. Although, I suspect that you've uncovered an issue, since if there is a direct match, then BMT should have chosen it, and not the UK version. The think that was interesting about the UK selection, is that the Year was 0 (in tvdb for some year), so I'm wondering if BMT chose it, because it would have match the year for your show (which would have been 0 as well)

I'll open a bug for this, and I'll consider your proposal. From a bmt perspective, the selecting of the series and the finding of the episode happen at 2 different stages of the lookup, and the original list is discard at the time that bmt tries to look for episodes

[ohpleaseno]

Originally Posted by razrsharpe

you're making the assumption that the original air date as set by the EPG is accurate though....

Yeah, This Old House is always OAD +2 from my EPG.

[evilpenguin]

Yeah, but it wouldn't really matter, without an episode title or correct OAD you'd have to manually scrape it regardless. The only issue would be if the incorrect OAD resulted in a false positive which it wouldn't.