Use the youtubei API for YouTube mixes. The corresponding has been updated because the new API breaks the tests of YoutubeMixPlaylistExtractorTest.
Remove some deprecated code (the old search code with the pbj JSON) and do some other improvements.
Use the Android mobile API to get the itag 22 (720p with audio), removed when the content is protected by signatureCiphers.
Also use this API when they are OTF streams, to get the itag 17 and 36, low 3GPP quality streams but also the itag 139.
Update the web client version.
Use again www.youtube.com and music.youtube.com domains instead of youtubei.googleapis.com domain because it spoofs more a web client of YouTube or YouTube Music and may reduce Google's detection of NewPipe Extractor users.
Add the origin and the referer headers with the https://www.youtube.com value for YouTube JSON POST requests.
Don't add the consent cookie header for the requests which use the youtubei/innertube API because it's uneeded.
Fix some tests and update YouTube mocks
Get the real name of the uploader (for autogenerated channels and music artist channels), like before the migration to the JSON pbj.
Do some other improvements, especially reformatting some code to be in the 100 characters line limit and use final where possible.
Change the domain from music.youtube.com to youtubei.googleapis.com.
Use a lightweight request to check if the hardcoded YouTubeMusic keys are valid. Increase the length of the response to 500 because if the key is invalid, the length of the response returned is higher than 250 and the response when the key is valid is higher than 1500.
Format the YoutubeMusicSearchExtractor file.
Update YouTube web client version and mocks
Add getSearchParameter, a new method in YoutubeSearchQueryHandlerFactory class which returns the params field for a search, or an empty string if there is no one.
Update mocks of YoutubeSearchExtractorTest.
- do not parse responseBody twice for continuation
instead try to get commentsTokenInside with the new pattern ("sectionListRenderer")
and try again with the old pattern ("commentSectionRenderer") on failure
- do not unescape responseBody multiple times
-> parse responseBody less times
Requires sending a POST request instead of GET.
clientName and clientVersion, which were required as headers previously now need to be part of the request payload.
continuation id also needs to be part of request body.
quick and dirty solution.
I searched for "StandardCharsets" to prevent possible bugs with Android KitKat. The import is unused and is therefor removed.
There are other usages of the StandardCharsets class in the test implementation, but those are not relevant.
This is needed so that a request is made for each test class when running multiple at once. This way RecordingDownloader records all necessary requests.
This works as long as tests are run sequentially and not in parallel.