Closed JesperDramsch closed 6 years ago
Hi Jesper, thanks for reporting an issue. I suppose I have read the first one before, and it didn't work. It was due to the private content fetching API of pocket producing article HTML that the skill wouldn't understand.
I will look into the these articles and see what's the problem. If it is also the case, I'll do more guessing in the skill to improve the compatibility. I'll get back in a few hours.
The HTML structure returned by the Pocket text content API is kind of unreliable. Still figuring out a way to reliably get the text content for audio from the response.
Could this package help?
I wish. But the problem is actually not that trivial. After it gets the content html from pocket, it needs to parse it into paragraphs, and chunk it into sizes that fits inside the SSML response size limit. On a side note, the nodejs html-to-text utilities are unbelievably bad. I'd rather write a custom html-to-text function.
9a97642 should have fixed this. Thanks for motivating me to develop a proper fix to paragraphing.
I am deploying it to production. Stay tuned.
That is amazing. Thank you so much. Any way I can get you a beer?
The updated skill is now live!
A user sent me Ethereum last time, it was great. But most importantly make sure you leave a positive review for the skill. The rating is somewhat low now. Thanks!
Hi,
first thanks for the amazing work.
Most articles work really well. However, some that I had imported via pocket directly would not read at all or skip the first chunk.
This one breaks completely: https://getpocket.com/explore/item/how-to-work-40-hours-in-16-7-the-simple-technique-that-gave-me-my-life-back-1291711976
This one also doesn't work: https://medium.com/swlh/this-is-how-to-plan-a-day-94dc24135e1f