Auto complete for file/path name and pandas column name with bracket access

[noppGithub]

1. In Juypter notebook, when you are trying to read a file with command df = pd.read_csv('./myfile.csv')
   for the file name, you can just type './m then press TAB to get a list of your file that you can pick from.

2. In Jupyter notebook, when you are typing the name of a column, e.g. you are accessing df['date_start'], while you are typing df['d then you press TAB, it's will show the option or autocomplete for you,

I think it's would be nice if we have this feature in VSCODE

[jakebailey]

Both of these rely on runtime data, things we don't have access to as a static analyzer. Jupyter itself has this data in memory and should already be suggesting it. Are you not seeing these completions when using the Jupyter extension in a notebook?

[noppGithub]

Both of these rely on runtime data, things we don't have access to as a static analyzer. Jupyter itself has this data in memory and should already be suggesting it. Are you not seeing these completions when using the Jupyter extension in a notebook?

I can't see these completions when using Jupyter in VSCODE.
I'm not sure if I use VSCODE correctly, below are my steps for this issue.

• Create a new folder
• Create Python virtual environment in this folder
• Open VSCODE
• Select Python interpreter.
• Install pandas to the environment
• Create a CSV file
• Create a .ipynb file, click on it
• Get prompt to install ipykernel
• Press install
• Create some codes(as shown in the images below)
• Try to get autocompletion while pd.read_csv
• Try to get autocompletion while typing

Reference images.

I expected to see filename autocompletion here.

I expected to see column name autocompletion here.

[jakebailey]

Can you try manually triggering a completion by hitting Ctrl+Space inside of one of those strings? And is this the new or old notebook system in the Jupyter extension?

[gramster]

The pandas part is a dup? #102

[jakebailey]

Yes, it is.

As for the file-path one, I can't find it, but I believe that's a feature request we said we weren't going to do either (as it escapes the bounds of what's reasonable in a python analyzer). For notebooks, they already handle this because jupyter has this built-in and already accesses loads of runtime data.