Print number of variables in repr

[Illviljan]

Show the printed and total number of variables in the repr.

• Passes isort . && black . && mypy . && flake8

import numpy as np
import xarray as xr

a = np.arange(0, 15)
b = np.core.defchararray.add("long_variable_name", a.astype(str))
c = np.arange(0, 25)
d = np.core.defchararray.add("attr_", c.astype(str))
e = {k: 2 for k in d}
coords = dict(time=da.array([0, 1]))
data_vars = dict()
for v in b:
    data_vars[v] = xr.DataArray(
        name=v,
        data=np.array([0, 1]).astype(bool),
        dims=["time"],
        coords=coords,
    )
ds1 = xr.Dataset(data_vars)
ds1.attrs = e

# The repr now shows how many attributes in total there are:
print(ds1)
Out[10]: 
<xarray.Dataset>
Dimensions:               (time: 2)
Coordinates:
  * time                  (time) int32 0 1
Data variables: (12/15)
    long_variable_name0   (time) bool False True
    long_variable_name1   (time) bool False True
    long_variable_name2   (time) bool False True
    long_variable_name3   (time) bool False True
    long_variable_name4   (time) bool False True
    long_variable_name5   (time) bool False True
                   ...
    long_variable_name9   (time) bool False True
    long_variable_name10  (time) bool False True
    long_variable_name11  (time) bool False True
    long_variable_name12  (time) bool False True
    long_variable_name13  (time) bool False True
    long_variable_name14  (time) bool False True
Attributes: (12/25)
    attr_0:   2
    attr_1:   2
    attr_2:   2
    attr_3:   2
    attr_4:   2
    attr_5:   2
      ...
    attr_19:  2
    attr_20:  2
    attr_21:  2
    attr_22:  2
    attr_23:  2
    attr_24:  2

[Illviljan]

Ouch, forgot about the doctests... Going through them by hand is not happening. Is there any automatic way to do that?

[keewis]

unfortunately, I can't find a tool that does that. I guess either someone will have to write that tool or you will have to go through all the files and update by hand. For the latter, I would recommend to wait until the repr format has been reviewed so you don't have to do that more than once.

[Illviljan]

I worked around it by only showing it when the repr is limited. That's the most important case anyway I think.

[max-sixty]

I've used pytest-regtest quite a lot, which is decent. It would be another dependency. For tests like this — where the output is easier to review than to create, these "expect" / "snapshot" / "regression" tests are ideal. If anyone wanted to swap out these tests for that, I would strongly support the effort.

An aside — I tried to make a similar tool that worked for inline results — https://github.com/max-sixty/pytest-accept. Unfortunately it's not possible to make it work with bare assert statements, without swapping out a lot of pytest's internals. For a while i worked a lot on https://github.com/mitsuhiko/insta, an excellent project, would be great to have the same thing in python.

[max-sixty]

Does this no longer get the title though?

[max-sixty]

Should it be +=?

[Illviljan]

It gets the title via summary[0] in the f-string. I did this because I want the number to be displayed on the same row as title, Attributes: (12/25). If we do += the numbers would be shown on a new line.

[max-sixty]

Ah sorry I missed that

[max-sixty]

LGTM, any final comments before we merge?

[max-sixty]

Thanks @Illviljan !