animeNFO, tvtome, geos and IAFD loopup
#16
i think for tv a separate database is possibly the way to go. keeps the impact reasonably low in case of failure. plus there'll likely be a bunch of changes as things stabilize etc.
Always read the XBMC online-manual, FAQ and search the forum before posting.
Do not e-mail XBMC-Team members directly asking for support. Read/follow the forum rules.
For troubleshooting and bug reporting please make sure you read this first.


Image
Reply
#17
@spiff:

the music nav window code is specifically built for traversing the music database, but it can be used as a model to build something similar for video. this is something i was going to look at next. (any ideas on what the filter order should be? the music one was stolen from the ipod.)

building a new top-level "my tv" window with its own database would be the easiest way to go. if everything was unified under videos, how would you determine the content-type and what scraper to use when scanning?
Always read the XBMC online-manual, FAQ and search the forum before posting.
Do not e-mail XBMC-Team members directly asking for support. Read/follow the forum rules.
For troubleshooting and bug reporting please make sure you read this first.
Reply
#18
i was just gonna suggest something similar for videos like for music.
i'll give it a good thought and post it tonight.
Reply
#19
hi,

first of all, personally i'd really prefer to keep everything video in one window, eg my videos.

actually i thought that scraper selection should be done like it is done in my local test-tree, by manual selection (when you select lookup you choose which scraper to use from a (dynamically generated) list. by doing this way the user gets to tell the program which type of content a file/files are. we could have a nav window, where the base is

movies
tv shows
anime
porn

maybe anime and tv shows could be one entry.

for movies and porn subscreens like year, director, genre. for tv shows/anime genre->show->season->episodes..

since you guys want to put tv show information into a new database, i'll simply provide xml that can be parsed into whatever data structure you'd like.. if nobody objects this is the way i have designed it thus far:

when you look up a tv show on tvtome/geos you get general show information and the url's to to each episode in the episode guide. that way individual episode lookups are easy to do, but we avoid grabbing lots of information before it is required.

one thing i haven't given any thought yet is how to tell which file correspond to which episode. the easiest way would be assuming that the user does the lookup on a folder and that every file within the folder and its subfolders are episodes. then one could quite easily find episode numbers from file names. however having files from one show in multiple locations would then be tricky.

my hangover is starting to loose its grip so maybe i'll get some work done tonight..
Reply
#20
manual lookups are one thing, but what about scanning? i know i wouldnt manually lookup each file in my collection. it's not worth the time. somehow files need to be categorized so that a scan gets the information from the correct source.

the easiest way is to seperate each category into different windows, with their own bookmarks.

another way could be to add a category tag to the bookmark:

Quote:<bookmark>
<name>tv shows</name>
<path>smb://mycomputer/media/tv_shows/</path>
<category>tv</category>
</bookmark>

for finer granularity, users could use a folder.xml file that sets the category for that folder, and any subfolders until it finds another folder.xml. (i personally wouldnt bother with this because i dont like to clutter up my media directories.)

Quote:<folder><category>tv</category></folder>

re: anime and porn... i dont see them as top level categories. they are kind of a genre when you think about it.

re: manual lookups: do you have multiple lookup buttons, or a popup dialog which prompts the user to select?



Always read the XBMC online-manual, FAQ and search the forum before posting.
Do not e-mail XBMC-Team members directly asking for support. Read/follow the forum rules.
For troubleshooting and bug reporting please make sure you read this first.
Reply
#21
hi,

i agree with you on the anime/porn issue. i just wrote from the top of my head, and i agree that anime is a tv show and porn is a movie subfolder.

however, i do not agree with having a per-bookmark setting or having its own window. the first one would seriously clutter up my root window (i have four-five pc's as sources, multiply that with eg 4 and you get a lot of shares).

the second one would be a major pain in the arse, and i really think it is an overkill solution. the only reason for this would be that one could assume more about shares (eg, this is a tv show and every file in this folder is an episode). the same information is provided by the user when he selects tvtome/geos/animenfo as lookup source imho...

yes, i have a popup-dialogue where you select your source, however that is just a test i use during the development.

i uploaded my code @ sf.net

if you guys would like to have a look. the code probably isn't rock solid yet, and i'm certainly not proud of all the lousy solutions i had to use in some of the scrapers. it features

imdb lookups (doh)
tvtome lookups (including preliminary parsing of individual episode guide pages)
geos lookups (including preliminary parsing of individual episode guide pages)
iafd lookups (not very useful since there is really very little information on iafd..)
animenfo lookups (not as a scrapers, i will probably turn it into one later so we get episode guide support)

also my cvideolookupfactory class and the test code i use are included. and my modified version of the http scraper test framework (not exactly complete yet, i know absolutely nothing of win32 gui programming...)

spiff
Reply
#22
@spiff

the manual lookups are great improvement. i cant wait to play with this later today.

the database is where all the cool functions in xbmc come from. but today's scanning function is very rudimentiary. it only checks imdb, and it sometimes picks the wrong movie yeilding incorrect information.

so now how can we fix the scanner so that it will use the correct scraper, thereby populating the database with more accurate information? obviously, the scanner needs to know or guess the content type of the media file in order to use the correct scraper.

those were just some ideas. i kind of like setting a category for a bookmark. it would definately be optional, where the default setting would be movies like it is today. it allows users the ability to make scanning work correctly out of the box by seperating media into bookmarks by content. (would i use it? maybe. my media directories arent clean enough today, but it may be a reason to clean them up.)
Always read the XBMC online-manual, FAQ and search the forum before posting.
Do not e-mail XBMC-Team members directly asking for support. Read/follow the forum rules.
For troubleshooting and bug reporting please make sure you read this first.
Reply
#23
had trouble compiling the patch from sourceforge

i put the 4 dll's into a folder called scraper in system/

and copied 4 .cpp files to xbmc/utils and the 2 gui cpp's to xbmc compiled and it threw me errors regarding vectorfiles?

is this because of the cleanup from bobbin the other day?
Reply
#24
probably, look in cvideolookupfactory::collectscrapers().

will fix it and upload tomorrow (if i remember correctly it is simply replacing vecfileitems with something else..)

oh, and the path is q:\system\scrapers\video btw. should probably clearify a bit more @ sf.net

spiff
Reply
#25
well as far as tv goes i run a frontend to myth so i would personally prefer a seprate root menu item. but if all your tv is pre-recorded / not in a seprate servers database (ala myth) then i understand wanting to keep it in my videos.

Quote:movies/
------directors/
------actors/
------year/
------rating/
------genre/
------------imdb entry 1/
------------imdb entry 2/
------------imdb entry $n/
------------porn/ (to seprate porn from other listings)
tv/
------show title/
------------seasons/
------episode title/ (almost a file view)
------genre/
------------show title/
------------------seasons/
------cast members/
------------show/
------------------seasons/
------crew members/
------------show/
------------------seasons/
------network/
------------show/
------------------seasons/


is how i would think would work best.
even if tv is in a seprate window, (perhaps a option could be made for people with the ability to use livetv but this would screw up some skinning stuff)

there are plenty of different tv and movie animes. it will have to be on a case by case basis (which means user interaction unless it can be determined from what source returns information)
Reply
#26
just uploaded a new version of the patch, syncronized with anon cvs as of 14.03.05

spiff
Reply
#27
hi,

just a little update..

i finally had some spare time, so i have written a simple class to associate file names with seasons and episodes (and shows) based on regular expressions and assuming that everything before the episode number is associated with the name of the show. it works splendid for my media directories, however i am a very tidy man when it comes to my file systems..

if people want 'their' naming conventions for tv show episodes to
be supported, please give examples here.

currently i have support for

24.s01e01*
24.101*
24.1x01*
24.s01.e01.*

spiff
Reply
#28
spiff,

those look good, just make sure that they all support two characters for the season and episode numbers.
Always read the XBMC online-manual, FAQ and search the forum before posting.
Do not e-mail XBMC-Team members directly asking for support. Read/follow the forum rules.
For troubleshooting and bug reporting please make sure you read this first.
Reply
#29
yep, already working.
Reply
#30
<- retarded del if you please
Reply

Logout Mark Read Team Forum Stats Members Help
animeNFO, tvtome, geos and IAFD loopup0