Reduce tests for video, fix video load, & remove binder icon

[xianyuanliu]

Fixes #175.

Description

• Reduce tests for video
• Fix video load
• Remove binder icon

Status

Ready

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• Breaking change (fix or new feature that would cause existing functionality to change).
• New tests added to cover the changes.
• In-line docstrings updated.
• Source for documentation at docs manually updated for new API.

[haipinglu]

@bobturneruk Forcing pandas>=1.3.0 is not working for 3.6
https://github.com/pykale/pykale/runs/2993309532
ERROR: Could not find a version that satisfies the requirement pandas>=1.3.0

@xianyuanliu You did not seem to have fixed pandas 1.3.0 compatibility problem yet.

[xianyuanliu]

Yes, I haven't solved it yet because it seems a general problem. I report it to pandas.
Dataframes saved with pandas<1.3.0 and "pickle" is no longer loadable with pandas 1.3.0 in pandas-dev/pandas#42345.

Btw, Pandas added this problem to the 1.3.1 milestone which due time is July 25.

[haipinglu]

Yes, I haven't solved it yet because it seems a general problem. I report it to pandas.
Dataframes saved with pandas<1.3.0 and "pickle" is no longer loadable with pandas 1.3.0 in pandas-dev/pandas#42345.

Btw, Pandas added this problem to the 1.3.1 milestone which due time is July 25.

Good finding! test_video_feature_extractor has several hundreds (?) tests, 60+% of all tests. Can they be reduced as well?

[mustafa1728]

I tried using the pandas function pd.read_pickle() instead of pickle.load(), and the tests pass on my local.

# with open(self.annotationfile_path, "rb") as input_file:
#     input_file = pickle.load(input_file)
 input_file = pd.read_pickle(self.annotationfile_path)

May I test on GitHub by committing in this PR?

It seems that pandas is handling this exception in pd.read_pickle() (source), as mentioned in this comment.

[haipinglu]

May I test on GitHub by committing in this PR?

Yes, please go ahead.

[xianyuanliu]

I have tested pickle.load and pd.read_pickle by the code below in the previous Pandas version.

with open(path, "rb") as input_file:
    input_file1 = pickle.load(input_file)
with open(path, "rb") as input_file:
    input_file2 = pd.read_pickle(input_file)
print(pd.testing.assert_frame_equal(input_file1, input_file2))

Output is None, which means no difference between input_file1 and input_file2.

Like what @mustafa1728 said, pandas uses pickle.load inside pd.read_pickle but updated this function this time. But I haven't tested whether they are the same in the newest version.

[codecov-commenter]

Codecov Report

Merging #178 (825d2d0) into main (04b2863) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #178      +/-   ##
==========================================
- Coverage   88.37%   88.36%   -0.01%     
==========================================
  Files          45       45              
  Lines        4197     4195       -2     
==========================================
- Hits         3709     3707       -2     
  Misses        488      488              

Impacted Files	Coverage Δ
kale/loaddata/video_datasets.py	86.44% <100.00%> (-0.45%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 04b2863...825d2d0. Read the comment docs.

[xianyuanliu]

@mustafa1728 Do we need this with ... open ... here to avoid forgetting to close the file? I am not sure if it is needed.

https://stackoverflow.com/questions/20101021/how-to-close-the-file-after-pickle-load-in-python

[mustafa1728]

Do we need this with ... open ... here to avoid forgetting to close the file? I am not sure if it is needed.

The read_pickle() function has a with ... open ... block inside it here, so its not needed, as it is closing the file anyways. However, this works both ways (path or file handler as in the docs), so it should not cause any errors inside the with ... open block either.

[xianyuanliu]

Oh, my mistake. I regarded it as pickle.load, but not the new one. Thanks.