[BUG] DataFrame column names including a dot (".") are not usable

[lucasisonline]

A DataFrame in the format of

x_x	y_y
1	3
2	4

will work fine.

The following column naming will result in visible but "unusable" data.

x.x	y.y
1	3
2	4

[ObservedObserver]

Graphic Walker does not parse the column id, which causes a parse error in vega-lite.

This can be fixed by both Graphic Walker and PgGWalker .

TODO:

For independent usage of GW as an app, GW should parse the field id for safety. But for embedding cases, it loses consistency of GW's columns' id and its host columns' id. I will implement this in GW, but leaving a param allows developers to decide whether to use auto column id parser. @Asm-Def

[Asm-Def]

@lucasisonline Could you have a try with pip install "pygwalker>=0.1.4a2" and give me some feedback?

BTW: Welcome to join our discord to keep track of the latest progress

[lucasisonline]

Hi @Asm-Def ,

thank you for the fast fix!

I tested it on a >7k dataset with >150 features including column names of the type very-much-autogenerated-including-special-characters.

Works like a charm! Nice and easy way to quickly explore the dataset. ♥

(To be very explicit in the characters used [a-zA-Z()._ ])