0.7.x #404
0.7.x #404
Conversation
…al to work with python v2.4+
…impaired (HI) or non-HI
…t providers being downloaded. Just download 1 (the best matched); bugfix
…ing Impaired (HI) first before non-HI or vs versa. This is kind of an extension to the bestscore enhancment already added in a previous commit.
…bliminal. Provider now supports new page layout
|
Quote support isn't pushed into the master branch yet, but i'm sure if someone beats me to it; it's pretty straight forward what i did (feel free to cherry pick away). I incremented the DogPile caching version too just to ensure no one references a local copy of their old quoted cached content. |
…unnecessary entries when not debugging)
…tests for it were removed in an earlier commit)
|
Changes Unknown when pulling 3c7634f on caronc:0.7.x into * on Diaoul:0.7.x*. |
|
Changes Unknown when pulling 7fef3cf on caronc:0.7.x into * on Diaoul:0.7.x*. |
the flexibility for 3rd party apps that wrap the subliminal framework to do their own guess management. The default behaviour of this fuction will remain the same as it always has. This new feature will only kick in if the video is specified to be used instead.
|
Changes Unknown when pulling e6bd704 on caronc:0.7.x into * on Diaoul:0.7.x*. |
Hopefully the Addic7ed administrator can respond to my email inquiring as to why they are blocking us by the User-Agent string. Automation is the 21st century of the internet; No one wants to click 8 times past banners just to get a 1KB (in size) subtitle. Most people have Ad blocking software and don't even see these banners anyway. There are many other ways to get people on board with helping them out financially (if that's what this is about), and at the same time accomodate those who've automated their service. I will roll back this commit when we can come to a better resolution.
Hopefully the Addic7ed administrator can respond to my email inquiring as to why they are blocking us by the User-Agent string. Automation is the 21st century of the internet; No one wants to click 8 times past banners just to get a 1KB (in size) subtitle. Most people have Ad blocking software and don't even see these banners anyway. There are many other ways to get people on board with helping them out financially (if that's what this is about), and at the same time accomodate those who've automated their service. I will roll back this commit when we can come to a better resolution.
| logger.warning('Provider %r is not available, discarding it', provider_name) | ||
| logger.debug('ProviderNotAvailable error: %r', str(err)) |
There was a problem hiding this comment.
You can use exc_info=True as kwargs for logger.debug to add trace information
|
I just reviewed your changes. There are just a few points:
|
|
I realize Python v2.6 is old, but I'm old fashioned too. I'd trade stability over bleeding edge any day. I run Red Hat 6.6 on all my production systems at work and CentOS 6.6 at home. As a result, I'm stuck with it's shipped version of Python 2.6. I've even packaged all of your software in RPMs which makes it so easy to deploy your fantastic tool. :) I intend on updating to CentOS 7 soon which will bring me to Python v2.7. I'll still be old; but we'll at least be on the same page. :) I'll read your review notes, I'll reply to them individually and do my best to explain them; from there you can let me know how you want me to proceed. |
|
I don't mind the dict and set comprehension stuff for 2.6 compat, what bothers me is more unittests that are less readable (and outputs less readable failures) and the zipfile stuff. Maybe for 2.6 compat we can use 2.7+ syntax with unittest2? |
|
Changes Unknown when pulling 407cbe1 on caronc:0.7.x into * on Diaoul:0.7.x*. |
|
@sioban: just to avoid confusion, an option as you suggest would prove to be unnecessary (IMO). At this time, the proper User-Agent (Subliminal v0.7) is correctly applied to all providers except Addic7ed (as of very recently). If I were to revert back and apply the same User-Agent for Addic7ed too (as an option), then the provider simply won't work. |
|
That's the problem with the unittests. I think the best for unittest would be to have a mock proxy that forwards the request to the original website, caches the response and send it back to the client as-is. If the original website fails to respond, the cache is used. So instead of testing against real servers we test against real or last time servers. That should be enough. I can host this mock proxy locally or on free Heroku dyno I think. I'll merge this and eventually make a few changes so I can make a new bugfix release of subliminal this week. |
|
Sure! Thanks for accepting the pull request btw. I'm looking forward to dropping all my patch files! :) |
|
@Diaoul What you describe - a mock proxy - is handled in ruby by a gem called VCR. There are two python ports for it - https://github.com/sigmavirus24/betamax & https://github.com/kevin1024/vcrpy |
I've been maintaining a plugin (forum) to your fantastic subtitle fetcher you've written. The plugin is for NZBGet (a UseNet solution). You can see this blog if interested. I'm only utilizing the 0.7.x branch, but have added some stability to it along with back porting older code so that it works with people running older versions of python (v2.6+) (like myself). The backporting specifically addresses an old ticket closed. I adapted to the podnapsi's new page layout and added a few other minor features. I just thought I should share my changes back!
Just added support for quotes which should fix issue 378 and issue 412 for you.