Add hook for service registration by other addons

[wizzwizz4]

I would like to customise AwesomeTTS in a way that can't be done through the preferences menu without having my changes overwritten when AwesomeTTS updates. A relatively simple way of doing this would be to add a hook handler to awesometts/services/__init__.py:

def _service(name, obj):
    __dict__[name] = obj
    __all__.append(name)
_addHook("AwesomeTTS_register_service", _service)
_runHook("AwesomeTTS_loaded_services")

I am not very familiar with the code, so I don't know what else would need changing. From what I can tell, this and perhaps the service list "cache"(s) (if there is / are one(s)) would need changing (perhaps with an AwesomeTTS_flush_services hook). In order to use this, another addon would simply have to include:

def register_my_service():
    runHook("AwesomeTTS_register_service", "my_service", my_service_object)
addHook("AwesomeTTS_loaded_services", register_my_service)  # Iff I load before AwesomeTTS
register_my_service()  # Iff I load after AwesomeTTS

which would work assuming that nobody else wants to use the hook name AwesomeTTS_register_service, in which case their hook would run twice if they load before both AwesomeTTS and the target add-on. But who's going to do that?

[wizzwizz4]

I assume that Feature Requests are allowed.

[krassowski]

to customise AwesomeTTS in a way that can't be done through the preferences menu without having my changes overwritten when AwesomeTTS updates

What's the customization you would like to perform that causes such troubles?

[wizzwizz4]

I'd like to add support for breaking up spacedelimited phrases into separate words in order to pass them through Duden, then join those to make a final audio phrase. Then, as needs changed, I might add support for fallbacks (e.g. Wiktionary).

[krassowski]

Sounds like as an useful feature to me. Really, I thought about such a thing too, but I would like to separate words by other delimiter, pipe: |, just to get better pronunciation for two words I compare. Maybe you could share your changes and contribute a patch upstream, instead of running it with a hook/filter?

[wizzwizz4]

I might, if my code is sane. However, I think that adding support for some other service registration method (preferably a better one than my suggestion) would be preferable to in-place allow other add-ons to interface with AwesomeTTS - essentially allowing AwesomeTTS addons to exist. This would drastically increase the features available for AwesomeTTS as Anki addons an drastically increase the features available for Anki.

[luc-vocab]

@wizzwizz4 can you describe your use case in more details ? are you trying to generate audio for a full sentence from TTS services which only support individual words ? Can your needs be met by using a more advanced TTS service which supports full sentences ?

[wizzwizz4]

are you trying to generate audio for a full sentence from TTS services which only support individual words ?

Yes – or, at least, I was. I've dropped the project.

Can your needs be met by using a more advanced TTS service which supports full sentences ?

I vaguely remember that no such service was available at the time – but if one did, then yes. I wanted something not synthesised, because the synthesisers were getting it wrong and I didn't want to learn wrong.

In retrospect, I think I should've made an Anki feature request for allowing add-ons to "depend" on others, so they loaded after them.

[luc-vocab]

Understood. A lot of people have a separate field which contains pronounciation-only text, which is sometimes distorted to help the TTS system.