What I’m trying to do:
I am trying to receive text back to a Text Area from an API call to Openai’s Whisper Audio To Text Transcription API. I’m using the Material Design 3 Template.
What I’ve tried and what’s not working:
I’m getting text back there seems to a limit to the text length I’m able to get back into the text area. It occured to me that there could be something behind the scenes that sets the max length of a text area in Material Design 3?
I understand maxlength can be defined in HTML as below. I have checked the visible parts of the MD3 HTML/CSS, and the documentation and can’t see anything. Developer Tools might show the answer, but I can’t see it could use a pointer. Just checking to rule this out / work around it.
Thanks
e.g.
HTML
Have you seen any errors or checked the length of characters that do get added to the area? Are you sure that the server function is returning the value you think it is?
No, that’s a good point. I’m comparing non Anvil api calls from my laptop to calls over Anvil / server. Character length is shorter for the latter e.g. 14436 vs 6910 (printed to console & output to text area) for the same audio transcription. But its tricky as there could be other reasons for this, and it seems inconsistent for different audios. Was just wondering if you reach a limit, at some point, for a text area.
Ahh, well in that case the real culprit might have to do with server calls in anvil rather than the textarea if the textarea js getting whatever the server is returning.
Unfortunately I have even less understanding of that part of anvil
No, Okay thanks. Yes, I think thats a variable. There may not be a meaningful limit to the text area - 16820 characters can be achieved. Just trying to remove that as a variable and figure out the best way to test different implementations of the api call.
Do you definitely need the full characters or could you use a shortened response? Could you break up the response into several different calls and append it to the text areas text?
But if it is the anvil server limitation and you definitely need all those characters, you could definitely setup a flask http server somewhere else as well.
Will read up on that, thanks. I’m using a third party api to test at the moment but would need to chunk long audio and put it back together again for my own implementation, so that might make sense. Background Task works fine up to a certain point. Theres probably an art to the chunking algorithm etc.