Problems with json exporting about markovify HOT 11 CLOSED

Yes. I personally don't like the idea of "to_json" to be honest, because the user should be left with the task of serialising the stuff (I'm using pickle myself because I imagine it's faster) but I guess it's a good idea to have those methods for consistency with what there is in markovify.Chain. It can't hurt.

from markovify.

My apologies, and great catch. I've updated the README to fix that section and add a more explicit example. Does following that example work for you? (You should be able to swap markovify.Text for markovify.NewlineText.)

from markovify.

Why do you still need the corpus text when reconstituting the model? Isn't everything inside the json file?

from markovify.

The JSON file contains only the chain, rather than the full model (i.e., corpus + state_size + chain). But I could definitely imagine adding a feature to export/import full model. In fact, I'll try doing that right now.

from markovify.

Done! See #44, now merged into master and available on PyPi in v0.5.0.

from markovify.

Hm, that's fine, but there's a problem -- despite the names of the methods (from_json/to_json) those methods don't accept/return a str containing json, but a dict.

from markovify.

Ah, I see. That's a good point. I suppose it'd make sense to do this:

• .to_json/.from_json: accepts/returns JSON strings.
• .to_dict/.from_dict: accepts/returns dicts.

What do you think?

from markovify.

That sounds good.

from markovify.

Here's the proposed change: https://github.com/jsvine/markovify/pull/45/files

Along the lines of what you were thinking?

from markovify.

Makes sense, and thanks again for these suggestions. Much appreciated!

from markovify.

That's right, thanks for your work.

from markovify.