Add thread lock to NameParserCache.
Add exception handling for add to cache.
Change increase NameParserCache to 1000 entries.
Change NameParserCache to OrderedDefaultdict.
Add move_to_end to OrderedDefaultdict class (backported from python 3).
Add first_key, last_key to OrderedDefaultdict.
Change use much faster first_key in NameParserCache.
Change remove import of python pre 2.7 OrderedDict.
Add the improved error handling to torrent providers.
Change cache_db to always include db upgrade statements when running SG for the first time.
Change split newznab limit logic into a reusable function, hit_limit_update() and use in BTN prov.
Change tweak CSS to make things a little neater with button spacings.
Add the actual time when current limit will expire to the UI.
Change instead of terminology "errors", use "failures".
Change improve BTN error handling.
Change ensure provider name is output to log at times where it wasn't.
Change ensure failed request URLs and POST params are output to log.
Add time of last failure + type to should_skip log message.
Change code DRY and simplification for improved readability.
Change occurrences of "error" to "fail" for consistency.
Add tmr limit handling to omg and change log_failure_url level to warning.
Change log the failed URL when an API hit limit is reached.
Change "hit" to the more universally generic "tmr" - Too Many Requests.
Change Db columns containing "hit_" are renamed "tmr_"
Show root dirs can not be set from general config menu.
Mass editing shows now has the ability to delete root dirs as well as edit them.
Daily search no longer is restricted to just 1 week of results for searching from which now allows for replacing lower quality downloads with higher quality ones if available.
RSS Cache is updated for each provider on demand now when performing manual, failed, backlog, or daily searches.
Provider getURL and downloadResult functions now removed and replaced with ones from helpers.py to help slim the code down plus allow more better control over request sessions.
Removed TVTumbler code.
Fixed HDBits provider.
Fixed config settings that were ment to be booleans but instead where set as str or int, should help resolve random html errors.
XEM Refresh check re-coded.
NameParser code for creating show object has been changed to only attempt at the very end once its found the bestMatch result, helps on resources and performance.
Fixed several providers and ssl cert verify issues.
Air-by-date shows now automatically get there respective season/episode numbers during release parsing.
Fixed charmap issues for anime show names.
Fixed issues with display show page and epCat key errors.
Fixed duplicate log messages for clearing provider caches.
Fixed issues with email notifier ep names not properly being encoded to UTF-8.
TVDB<->TVRAGE Indexer ID mapping is now performed on demand to be used when needed such as newznab providers can be searched with tvrage_id's and some will return tvrage_id's that later can be used to create show objects from for faster and more accurate name parsing, mapping is done via Trakt API calls.
Added stop event signals to schedualed tasks, SR now waits indefinate till task has been fully stopped before completing a restart or shutdown event.
NameParserCache is now persistent and stores 200 parsed results at any given time for quicker lookups and better performance, this helps maintain results between updates or shutdown/startup events.
Black and White lists for anime now only get used for anime shows as intended, performance gain for non-anime shows that dont need to load these lists.
Internal name cache now builds it self on demand when needed per show request plus checks if show is already in cache and if true exits routine to save time.
Schedualer and QueueItems classes are now a sub-class of threading.Thread and a stop threading event signal has been added to each.
If I forgot to list something it doesn't mean its not fixed so please test and report back if anything is wrong or has been corrected by this new release.
Search mode allows you to choose season packs only or episodes only which controls how backlogs search for data.
Search fallback if selected will attempt a 2nd search this time using the alternative search mode to see if it can get results if the previous attempt failed.
Provider settings are now read and stored dynamically instead of always having to retrieve them by there static names, this will allow for dropping in new provider modules easier to intergrate with existing options already present.
Fixed search code to not perform dupe searches by mistake.
Cache stores a list of shows we have rejected so that next time around we don't have to process them to know we don't like em.
