Change to ticketing redirect

[nklimchuk]

Current ticketing redirect works as follows (my understanding):

a. Searches "description" tag for a URL containing words like ticket, sale, buy, purchase. b. If it doesn't find one with those, it will check "tickets" tag and "website" tag with the same text search. c. If that fails, it uses the last.fm event link.

New steps:

1. It seems to me that the "website" URL is usually correct. And that the text search is dubious (lots of ticketing URLS do not have those words in them). So it might be easier to have it select that URL (if one exists in "website") without the check for those words.
2. If one doesn't exist, it should check "tickets" (no text analysis) and then "description". If only one URL exists in description, it should be selected. If not, then the text check can be used. If that fails, or no URLs exist in any of those tags:
3. Use the "venue" "website" tag. It's still better to send them to a venue website then to the last.fm website
4. If no venue link is available, only then send to the last.fm page.

[nklimchuk]

Couldn't use "<" ">" those brackets, so i replaced them with ""

[nklimchuk]

After looking at examples, the vast majority have the correct URL in the "website" tag on events. Sometimes it's a facebook URL and they put the actual ticketing URL in the description. But I don't think we have to overcomplicate it to get it perfectly right. Better to spend time on Seatgeek: looks better, will work most of the time and we earn money on it.

Below is the sequence I was thinking about originally. It might even be easier to just do steps (1), (2), (3 -with either no text analysis & just the first URL found or even skip 3 entirely), (4), (5)

Sequence

1. If "website" URL available, use (no text analysis).
2. Else, if "tickets" has URL, use. [haven't seen one in these tags at all, although slavik said 10% did when he checked]
3. Else, check "description": a. if only one URL exists, use. b. if more than 1 URL exists, use the text analysis script, use best match.
4. Else, use the URL in "venue""website" (not "venue""url")
5. Else, use the last.fm URL

Sequence with Seatgeek:

1. Check seatgeek.
2. Else, use above algorithm.

Examples:

Band of Horses 3751082 Would return with (1): http://www.ticketmaster.ca/event/10004B82B64C638D

Wavelength Music Festival 3761369 Would return with (1): http://www.wavelengthtoronto.com/wavelog/2013/12/wavelength-music-festival-fourteen

Justin Timberlake 3650152 Would return with (1): http://www.ticketmaster.ca/event/10004AC4F02EA4AA

Sam Roberts 3788977 Would return with (1): http://adelaidehallto.com/event/sam-roberts-band/

Robert Ellis 3758642 Would return with (1): http://www.collectiveconcerts.com/event/446067-robert-ellis-toronto/

Paint 3770209 Would return with (5): http://www.last.fm/event/3770209+Paint+at+Rancho+Relaxo+on+14+February+2014

Blue Rodeo 3757290 Would return (4): www.generalmotorscentre.com/

Graveface Roadshow: 3742671 Would return with (1): -> facebook, not the ticketing URL which is in the desc, but w.e. no point to try to tease out the difference. https://www.facebook.com/events/581540511899588/

Prok and Fitch: 3775584 Would return with (1): -> facebook, not the ticketing URL which is in the desc, but w.e.

Hunters: 3776610 Would return with (1): facebook, but no event url in desc or event website https://www.facebook.com/events/494672223984264/

Goo goo dolls 3755601 Would return with (4): www.generalmotorscentre.com/

Van She DJ Set <-- interesting example where text analysis would make sense, but again given the complexity and time, doesn't matter 3805464 Would return with (1): although it is a facebook event, and there are multiple URLs in the description, including the correct ticketing link.

Sarah Chang 3656109 Would return (4): http://www.markham.ca/Markham/Attractions/Theatre/