STT & Text processing APIs

[andrey18106]

Attempt to implement dynamic registration of Speech-To-Text and Text processing APIs.

Due to technical obstacles one unresolved task left as there is no possibility in PHP to override ::class string, which is used for retrieving classes, implementing API interfaces widely used in Nextcloud.

• Find out a way to properly resolve anonymous classes
• Extract migration to a new one (as the previous one already released)
• Add check for feature availability (since 29.0.0)

We can proceed with these APIs implementation after the following server PRs accepted and finished:

• enh(SpeechToText): Allow providers to declare a dynamic ID instead of using className server#41087
• enh(TextProcessing): Allow providers and task types to declare a dynamic ID instead of using className server#41088
• Machine translation PR to be created

[andrey18106]

Possible solution found with Julius, will update this PR later

[andrey18106]

Current Text processing APIs in core requires adjustment in provider ID to use getId() method to identify unique provider registration like in other APIs, e.g. DashboardWidget (IWidget). Therefore current implementation would work only with single registration as in core it relies on $provider::class which in our case is anonymous class pointing to the line of code where it was created - so that multiple instances will not be handled properly.

[bigcat88]

we will implement all this in each separate PRs

[andrey18106]

STT and Text-Processing are done and merged in separate PRs. Machine translation needs server side changes the same as for STT and Text-Processing, waiting for them and after that will add this API support here.