* Use platform detection that works on mono 4.6+
* Move to use package reference for restoring nuget packages.
* DateTimeRoutines does not have Nuget packages that support .NET Standard (and therefore .NET Core). We will have to include them for now until we can get rid of this dependency.
* Start spliting some interfaces into their own files - this will help by allowing us to split them out in the future into a seperate project so the actual implementations can stay within their respective architectures when required
* Move out common libraries
* Few more tidy up tasks to get things working with .NET Standard
* Restructure the solution layout
* Encoding work to reduce rework later on platforms without Windows codepages (or require compliance with RFC1345)
* Move folder structure around to have more natural layout of the solutions
* DI server configuration to get rid of "temporary" hack and dependency circle for serverservice
* Make all encoding consistent to match the expected encoding casing for earlier versions of mono.
* Move to use package reference for restoring nuget packages.
* Return a task result for this async method.
* Update to a supported version of the .NET Framework. This also has the side effect of allowing us to automatically generate our binding redirects on build.
* Set the solution to target VS2017
* Update test solution csproj file to support being built through MSBuild 15
* Move to use package reference for restoring nuget packages.
* Return a task result for this async method.
* Update to a supported version of the .NET Framework. This also has the side effect of allowing us to automatically generate our binding redirects on build.
* Set the solution to target VS2017
* Update test solution csproj file to support being built through MSBuild 15
* DateTimeRoutines does not have Nuget packages that support .NET Standard (and therefore .NET Core). We will have to include them for now until we can get rid of this dependency.
* Move the interfaces into their own files. This will be useful when we share them between the .NET Core and .NET Framework WebAPI
* Stage services that need to point to the new interface namespace.
* Update CurlSharp to fix memory leak issue and support better runtime compatibility with OSX and Linux
* Start spliting some interfaces into their own files - this will help by allowing us to split them out in the future into a seperate project so the actual implementations can stay within their respective architectures when required
It's quite hard to encapsulate something this large. This refactor
contains multiple attacks on the current architecture and is changing
things that were probably created quite a while back then. Luckily this
was done in increments so it mustn't be that impossible to recall what
has been done. I just need to relax my memory a little bit.
So the basic idea was quite simple. Let's distingush metas and normal
indexers a little bit more. Both of them were originating from
BaseIndexer, however very little of the functionality was actually
shared between them. Actually quite a few things made it even harder to
implement a different kind of indexer, especially for a newcomer for
both Jackett and C#.
Then in order to further reduce whatever was encapsulated in
any kind of, a couple things had to be changed. Like CardigannIndexer,
which probably had quite a mindshift change. IndexerManager and the
configuration management were also encapsulated and refactored, and now
I have a feeling that although the code could be improved, at least the
responsibilities of services and what they actually do is now clearer.
Anyhow, it would be safe to assume that I will not be able to go
step-by-step and define everything that has been changed. I'm sorry.
* Line endings...
* Improve download handling of torrents in metas
Until now, downloads were handled by metas, however they had to "guess"
which indexer the result was originating from and resolve to that
indexer. While this has been working without an issue, it really
shouldn't be considered stable.
Therefore indexers now link themselves to the results they provide. In
order to keep my sanity and automate this as much as possible, I had to
slightly modify the interface (sorry, everyone).
* Line endings...
* Refactoring how MetaIndexers handle fallbacks
Originally this modification was part of a much larger refactoring,
however for the sake of reviewability I split it into smaller chunks.
Sadly it is still quite large.
I wanted to split it even more, however after a certain point there was
really no value in creating smaller chunks. The biggest part of this
modification would be still huge.
So all in all, there're 3 aspects of this modification
- It modifies BaseIndexer so that it now implements IIndexer (will be
very useful later on)
- Resolving most of the warnings currently in Jackett (the only ones
remaining are related to Autofac, however if I could I would just burn
Autofac altogether rather than fix the warnings. Will open discussion on
this.)
- Biggest part: refactoring how MetaIndexers handle fallbacks and how
they provide the final result set
MetaIndexers now accept any kind of fallback and filtering mechanism
that implements the necessary interface, so that in the future IMDB
fallback and filtering won't be the only one. I know there are not a lot
of unit tests around Jackett at the moment, however this renders the
class much more unittestable as well.
* Autofac started complaining... I don't understand...
extended refresh header handling to libcurl and safecurl
also some minor tweaks needed to have Curl 'behave' with certain servers by adding accept-language header to Curl
Added ability/template to allow the user to select their own SiteURL (see BitSoup implementation) with a minor update to ConfigurationData class to prevent the "display text" from disappearing on errors.
XSpeeds indexer - updated to handle a case where the returned XML occasionally is null terminated resulting in XML parsing exceptions
BitSoup indexer - added user customizable url selection, including some url validation
BaseIndexer - cleaned up some of my earlier implementation for accumulating cookies.
Nathan Holland