prettytable new version 3.12.0 is conflicting

[TheoLeg]

What happens?

prettytable package new version (3.12.0) is conflicting with jupysql current version.

To Reproduce

requirements :
jupysql==0.10.14
jupysql-plugin==0.4.5
prettytable==3.12.0
snowflake-sqlalchemy==1.6.1
SQLAlchemy==1.4.54

from sqlalchemy import create_engine
from snowflake.sqlalchemy import URL

engine = create_engine(URL(
    account = '<account>,
    user = <user>,
    authenticator='externalbrowser'
))
%load_ext sql
%sql engine
%sql SELECT * FROM DATABASE.SCHEMA.TABLE LIMIT 10

Error :

---

KeyError Traceback (most recent call last)
Cell In[7], line 1
----> 1 get_ipython().run_line_magic('sql', 'SELECT * FROM DATABASE.SCHEMA.TABLE LIMIT 10')

--> [466] _style = prettytable.dict[self._config.style.upper()]
[467] pretty.set_style(_style)
[469] return pretty

KeyError: 'DEFAULT'

To fix : Downgrade to previous version of prettytable (3.10.2) and it's working again

OS:

macOS

JupySQL Version:

0.10.14

Full Name:

Theo Legruel

Affiliation:

April

[dereksz]

Confirming I have the same issue; Linux, JupySQL 0.10.14.

Can be avoided by pip install prettytable<3.12.0.

Digging into the prettytable.__dict__, the "DEFAULT" key is gone, replaced with "_DEPRECATED_DEFAULT".

Can simply be recreated with:

%sql sqlite://
%sql SELECT 1

[rlshuhart]

To round out the OS's, can also confirm issue on Windows OS. Downgrading prettytable also fixes. Thanks @TheoLeg for taking the time to create issue.

[edublancas]

I've fixed this in 0.10.15, note that this update is no longer compatible with Python 3.8 since prettytable 3.12.0 deprecated support. so you'll have to use prettytable<3.12.0 and jupysql<0.10.15 if you are using python 3.8

[lukebarousse]

Out of curiosity, how long until this is updated on conda-forge? It's still on version 0.10.14 there.

Background: I'm making a course using this tool, and we work in a conda environment.

[edublancas]

@lukebarousse the auto-generated conda recipe is broken (conda-forge/jupysql-feedstock#34). @flaviomartins is kindly helping fix it. I'd appreciate if you can give it a check/test. I'll review it in the next few days so I'm hoping this to go live this week

[lukebarousse]

@edublancas, thanks for the update, and I appreciate the quick turnaround! This library is a life-saver 🙌

[flaviomartins]

Out of curiosity, how long until this is updated on conda-forge? It's still on version 0.10.14 there.

Background: I'm making a course using this tool, and we work in a conda environment.

Same. You might want to have a look at the environment I created for my course at https://github.com/bdist/bdist-workspace

[lukebarousse]

Same. You might want to have a look at the environment I created for my course at https://github.com/bdist/bdist-workspace

Sweet! Thanks for sharing this; I like how you used docker for this, looks like this solves a lot of headaches.