Data Viewer/Variable Explorer support for non-array like objects (such as nested dicts)

[jmew]

Support for objects in a tree like view, like how you can view objects in debug mode

Ask from one of our users:

"VScode also has an interactive window that opens a Jupyter kernel (separate console/kernel) which displays a subset of variables without the need of debugger.
It turns out that you can directly execute code from the editor in the interactive window’s console (no copy/paste needed), but the variable explorer that it provides is not as powerful (doesn’t give info on objects/classes/methods).
For example, we can’t look at a model object and inspect its components with this method, which has become an important part of our development process.

I guess we could run VScode in debug mode whenever we desire that level of granularity, but it seems like an unnecessary concession when we can get the desired functionality from PyCharm."

Not sure if this is a dup

[paamand]

Suggestion to prioritize simple dict-support. I.e. for
a = {'b':42}
the variable overview shows

but inspecting the specific variable gives

For simple nested dicts, a={'b':{'c':42}} the variable inspector fails and simply displays

A nested, expandable viewer for nested dicts would be optimal.

[FPRM]

Some updates about this feature ?

[glennbarrett]

Any update on this?

[marcos379]

Any update on this? Spyder IDE Data Viewer is years ahead of VSCode in that particular aspect.

[rubrojas]

Has it been any update? I am also experiencing this issue

[lhg1992]

Even in July 2023, the data viewer still cannot display the keys in the dict, it can only show the values with the index numbers.

[bgupta]

Curious is there any workaround short of switching to Spyder?

[GitHunter0]

Hey folks, I think the ability to visually inspect nested objects (not only dictionaries but specially nested dataframes) is a critical feature for a data scientist, it saves a lot of time and gives quick insights.

Here's an example of RStudio Viewer for a nested dataframe:

library(tidyverse)

# create a data frame
df <- tibble(
  group = c("A", "A", "B", "B"),
  x = c(1, 2, 3, 4),
  y = c(5, 6, 7, 8)
)

# nest the data frame by group
nested_df <- df %>% 
  group_by(group) %>% 
  nest()

# view the nested data frame
nested_df

nested_df in the Viewer, it also shows the content when you hover on the cell:

And if you click on the cell with nested data, it opens another tab:

I believe it is a great paradigm to borrow from, what do you think, @DonJayamanne ?

Thank you

[Nightenom]

Just had a thought if it would be feasible to display at least result of __repr__ instead [object Object] which would tell way more for custom classes
(especially when the variables menu supports it, and data viewer menu not...)

[amunger]

duplicate of #1148

[GitHunter0]

@amunger , why is it not planned, could you elaborate?

[amunger]

Duplicating against another issues does not mean it isn't planned, but the majority of the request is covered in a different issue.

This functionality will very likely be implemented with the built-in variable viewer microsoft/vscode#165445