add some tests, activate coverage report (hopefully)

.github/workflows/run-tests-push.yml

@@ -17,7 +17,7 @@ jobs:
 # - run: xvfb-run pytest tests/
  - run: tests/cmd_tests.sh
 
- run-tests-ubuntu-22_04-python-3_10:
+ run-tests-ubuntu-22_04-python-3_10-with-coverage:
  runs-on: ubuntu-22.04
  name: Ubuntu 22.04, Python 3.10
  steps:
@@ -28,8 +28,13 @@ jobs:
  - run: sudo apt install xvfb
  - run: pip install --upgrade pip
  - run: pip install .[dev]
- - run: xvfb-run pytest tests/
+ - run: xvfb-run coverage run --source countess -m pytest tests/
+ - run: coverage html
  - run: tests/cmd_tests.sh
+ - uses: actions/upload-artifact@v3
+ with:
+ name: test coverage report
+ path: htmlcov/index.html
 
 # run-tests-ubuntu-22_04-python-3_11_0rc2:
 # runs-on: ubuntu-22.04

pyproject.toml

@@ -33,6 +33,7 @@ dependencies = [
 dev = [
  'black<24',
  'build==0.10.0',
+ 'coverage==7.3.2',
  'mypy~=1.0.1',
  'pylint~=2.17',
  'types-psutil~=5.9.5',

tests/utils/test_pandas.py

@@ -0,0 +1,39 @@
+import pytest
+
+import pandas as pd
+
+from countess.utils.pandas import *
+
+df0 = pd.DataFrame([])
+df1 = pd.DataFrame([ {'a': n, 'b': n*100 } for n in range(0,100) ]).set_index('a')
+df2 = pd.DataFrame([ {'a': n, 'b': n*100 } for n in range(100,1000) ]).set_index('a')
+df3 = pd.DataFrame([ {'a': n, 'b': n*100 } for n in range(1000,10000) ]).set_index('a')
+df4 = pd.DataFrame([ {'a': n, 'b': n*100 } for n in range(10000,10100) ]).set_index('a')
+df5 = pd.DataFrame([ {'a': n, 'b': n*100 } for n in range(10100,10200) ]).set_index('a')
+df6 = pd.DataFrame([ {'a': n, 'b': n*100 } for n in range(10200,10300) ]).set_index('a')
+
+def assert_iterable_of_dataframes_equal(a,b):
+
+ assert pd.concat(a).sort_index().equals(
+ pd.concat(b).sort_index()
+ )
+
+def test_collect_dataframes_0():
+
+ x = list(collect_dataframes([df0,df0,None,df0].__iter__()))
+ assert len(x) == 0
+
+def test_collect_dataframes_1():
+ x = list(collect_dataframes([df1,df4,df5].__iter__(), 300))
+ assert_iterable_of_dataframes_equal([df1,df4,df5], x)
+ assert min(len(y) for y in x) > 100
+
+def test_collect_dataframes_2():
+ x = list(collect_dataframes([df1,df2,df4,df5].__iter__(), 300))
+ assert_iterable_of_dataframes_equal([df1,df2,df4,df5], x)
+ assert min(len(y) for y in x) > 100
+
+def test_collect_dataframes_3():
+ x = list(collect_dataframes([df1,df2,df4,df5,df6].__iter__(), 300))
+ assert_iterable_of_dataframes_equal([df1,df2,df4,df5,df6], x)
+ assert len(x) < 5