Mixed XML & URLl video NFO bug? - Printable Version
+- XBMC Community Forum (http://forum.xbmc.org)
+-- Forum: Development (/forumdisplay.php?fid=32)
+--- Forum: Scraper Development (/forumdisplay.php?fid=60)
+--- Thread: Mixed XML & URLl video NFO bug? (/showthread.php?tid=53135)
Mixed XML & URLl video NFO bug? - pletopia - 2009-06-18 20:41
this is something i noticed a while back but never got around to reporting it but since its still present in builds (i have r21082 at the moment running)
basically i REALLY like the idea of mixed url and xml nfo's .. it lets me have the abililty to specify certain fields in the db while having xbmc import from imdb.com the latest info for other fields when i upgrade or want to reload
anyway .. what i found is that some (not all tags) get pulled and entered into the db from both the nfo and imdb
i created a nfo file with MIP and added a imdb.com url to the movies page at the end of the file .. after i imported the movie into xbmc i open up the myvideos34.db file in SQLiteSpy and took a look at the movie in the movie table
looking through the columns i noticed some have dupes
genre (Action / Adventure / Thriller / Action / Adventure / Thriller)
director (Lee Tamahori / Lee Tamahori)
studio (Eon Productions / Eon Productions)
the only reason i bring this up is that i want to be able to set studio manually in some movies since some skins now have studio flags and not every default scrape from imdb gives a good studio (like in my example i rather set studio to United Artists
when i manually set studio in the nfo file <studio>United Artists</studio> and i looked into the db i had "Eon Productions / United Artists" in the column
hope this makes sense
copy of my test nfo file to fine duplicates
copy of my test nfo file to inject "United Artists" instead of "Eon Productions"
with this file .. i only get the studio column duplicated .. not the genre or director
- spiff - 2009-06-18 20:50
those are additive tags... come up with a sane way to handle it for all cases and i'll consider it.
- pletopia - 2009-06-19 00:37
umm .. what you mean sane way ??
basically it should be that if there is a xml tag present for specific column in the db .. it shouldn't be pulled from imdb
it works for all other tags (title, plot, rating, votes, etc) .. i don't see why it shouldn't pull data properly from both sources for most columns but randomly pull it for those three
i only bring this up cause these "additive tags" as you call them are actually quite useful
studio has now become important due to skins having studio flags and this will let you specify which ones will get displayed
genre recently i've noticed has also become more important because in my opinion imdb is too liberal with there genre categorization and i would like to limit my movies to specific genre's which might not coincide with what imdb has them listed as
- jmarshall - 2009-06-19 01:56
I believe he means that it's not necessarily an all or nothing thing. One could either completely ignore any genres scraped online if a genre is present in the XML, or one could append it to the list. This has implications for scrapers that are chained together as well. I presume that the code currently does the latter.
It thus would require modification, which may or may not be simple.
Perhaps a <genre append="true"> might be a solution?
- pletopia - 2009-06-19 07:52
ahh .. now i understand what spiff meant by additive .. i'm curious how feasible it would be to make a <xml append = false> setting for advancedsettings.xml
- spiff - 2009-06-19 17:57
hmm, i think the solution would be a <genre clear="true">. the specified xml information is applied after we have scraped. pletopia, you haven't quite grasped the problem yet if you think an advancedsetting will help
see, what i mean with additive is that when we encounter such a tag we add it to whatever is already around. this is necessary since you can have multiple genre's in the xml, and due to that genre's may in theory come from multiple scraper functions (the chaining jmarshall is refering to). the specified xml in the mixed nfo is treated just like a normal chain and loaded after the scraping process has finished. by specifying clear="true"/append="false" (whatever we choose) on the first genre in your mixed nfo you can clear those added earlier
- spiff - 2009-06-19 18:20
i'm gone for the weekend, feel free to test it
- pletopia - 2009-06-20 15:43
okay .. i got ya
so first the info is pulled from imdb.com THEN it gets pulled from xml nfo .. gotcha
- Paradise - 2009-06-24 00:22
Hm, i think my english is good, but i don't really understand that or how to include or handle this now.
Also see http://forum.xbmc.org/showthread.php?tid=53325 , cause therer are still other questions not answered.
- spiff - 2009-06-24 07:40
if your english is so good, you go read google if you do not understand the concept of a diff
- Paradise - 2009-06-24 13:49
There is no need for googleing for diff in english, cause its the same in german.
I mean where to i have to set clear="true"? Just one time or in <genre>, <credits>, <studio>...., everywhere.
Also i can't find a VideoInfoTag file on my windows.
- spiff - 2009-06-24 21:18
then what is the problem? i have already explained how it works AND that it is only a diff on trac at this point.
and VideoInfoTag is (shockingly) a SOURCE file. you know, what we apply those german diffs to ;P
- sho - 2009-06-25 12:29
The non programmer version:
This is an issue that has just been discovered.
The fix or workaround is being contemplated and is not part of the program yet.
If you have the skills you can get the diff that spiff graciously made and apply it to the source and compile your own XBMC version.
Otherwise you have to wait patiently until this or another approach will be put into the mainline code and a binary is released.
- Paradise - 2009-06-25 17:16
Ok, thanks sho.
For me a diff is a file comparison utility. And i didn't know that VideoInfoTag.cpp is something in the source. I tought its a file in XBMC. Also i thought "version changed from 9.04 "Babylon" to SVN" means it is in the latest SVN.
- Aenima99x - 2009-07-01 05:34
Spiff - I'm running svn ver. 21304 compiled today and I'm still getting duplicates in *some* fields when using nfo's with url's. This used to work before, I'm just not sure when it quit working. This is on Jaunty 32-bit.
I've also deleted the video db just to ensure there were no problems there.
Debug log here