InspectDB

[tarsil]

@rolandf Im opening the conversation here as we can discuss it. So, I've developed the inspect dB feature that is not yet merged and I'm running tests but you could also give it a try if you don't mind.

If you could branch of "feature/inspect_db" that would be great.

How to run it, it is similar to what you did. For example:

edgy inspectdb --database "sqlite:///db.sqlite" > models.py

This should trigger the process and save the results inside a models.py. I've tested for different DBs and SQLite is actually one I haven't fully.

It is important to understand, like Django, this generates based on existing information from the database and differently from Django, edgy generates a ReflectModel which is basically the same as a normal model but does not manage migrations.

https://edgy.tarsild.io/reflection/

If this goes well, which I strongly believe it will, we can even extend this to also generate ReflectModel for existing views 🙂 but that can come later.

ReflectModels have the same permissions of read, write, delete, update as a normal model would with again, no DB migrations being managed.

Can you give it a try? There is room for improvements here and it we can collaborate until it gets to the way we want before merging and releasing it.

[rolandf]

Hi Tiago,
I'll try to give it a look tomorrow if possible, if not during the week-end.
Frédéric

[tarsil]

Thank you @rolandf it will be truly appreciated. This will be merged with your collaboration 🙂

[rolandf]

As I don't expect my users to use sound field naming convention, I used a field name with a space in my sqlite DB and the space was kept in the model definition. I think it would be good to convert these spaces to underscores.

[tarsil]

Good to know @rolandf ! Great stuff! And about the rest? How do you feel?

[draew6]

@tarsil I received this error when running inspectDB

pydantic_core._pydantic_core.ValidationError: 1 validation error for Index
Value error, The max length of the index name must be 30. Got 31 [type=value_error, input_value=ArgsKwargs((), {'name': '...'match_id', 'user_id']}), input_type=ArgsKwargs]
For further information visit https://errors.pydantic.dev/2.5/v/value_error

[draew6]

@tarsil hi, i wasnt waiting for that all good (: you explained the issue and offered me opportunity to create PR. It's just that i didnt know if limit should be increased more than recommendation because i do not know much about this topic (:

[tarsil]

@draew6 as far as my research went, depends of the engines and for that reason this limit was increased and it should cover almost every scenario. I think SQL Server allows 128 so if we still have problems we can increase to 128 and that is it 👍🏼

This should now work for you

[tarsil]

Also @draew6 off topic. You you are interested, have a look at Esmerald. I think you will like it 😄

[draew6]

@tarsil thank you very much for new version and this change (: in my case i am using prisma for migrations so thats why i reached the limit. So probably it will solve issue for everybody coming from prisma.

regarding Esmerald. What is main advantage vs fastapi?

[tarsil]

Those are two different frameworks and use cases. Both built on top of Starlette which makes them very fast but Esmerald is more elegant, modular and more versatile in everything you can do. Nothing like playing around and let me know? 🙂.

Prisms is nice, I used it before

[tarsil]

@draew6 that that is the limit of the index name that is recommended but if you think we should increase that limit, by all means open a PR

https://github.com/tarsil/edgy/blob/main/edgy/core/db/datastructures.py

And let's make that change and I would be more than happy to accept that. The same for the unique constraint we also have in that file