[Python] assertEqual tests ok when numpy.ndarray vs str, is that expected? or what have I done...

My unittest returns ok, but when running my code in production, I found that my value is 'wrapped' with square brackets. Further investigation shows that, it lies under the df.loc[].values . I am expecting a single str value.

Using the sample by cs95 and doing some slight modification, I am able to reproduce it to illustrate the idea.

This is my first time deploying python unittest and sorry for the lengthy post/info.

# test class and test code
import pandas as pd
import numpy as np
import unittest

class myclass:
def __init__(self):
mux = pd.MultiIndex.from_arrays([
list('aaaabbbbbccddddd'),
list('tuvwtuvwtuvwtuvw')
], names=['one','two'])
temp_a = np.arange(len(mux))
str_a = [ str(a) for a in temp_a ]
self.df = pd.DataFrame({'col': str_a}, mux)

def get_aw(self):
a = self.df.loc[('a','w')].values
return a

class TestAssert(unittest.TestCase):
def test_assert(self):
myclass_obj = myclass()
result = myclass_obj.get_aw()
expect = '3'
print(type(result))
print(type(expect))
print(f'result:{result}')
print(f'expect:{expect}')
self.assertEqual(result,expect)


if __name__ == '__main__':
unittest.main(verbosity=2)


Results: (I expected it to FAIL)

test_assert (__main__.TestAssert) ... test_code.py:18: PerformanceWarning: indexing past lexsort depth may impact performance.
a = self.df.loc[('a','w')].values
<class 'numpy.ndarray'>
<class 'str'>
result:[['3']] <-- square brackets
expect:3
ok

----------------------------------------------------------------------
Ran 1 test in 0.002s

OK


I changed the code with .values[0][0] so I get the str and I added assertTrue(isinstance(...)

Changed code:

import pandas as pd
import numpy as np
import unittest

class myclass:
def __init__(self):
mux = pd.MultiIndex.from_arrays([
list('aaaabbbbbccddddd'),
list('tuvwtuvwtuvwtuvw')
], names=['one','two'])
#self.df = pd.DataFrame({'col': np.arange(len(mux))}, mux)
temp_a = np.arange(len(mux))
str_a = [ str(a) for a in temp_a ]
self.df = pd.DataFrame({'col': str_a}, mux)

def get_aw(self):
a = self.df.loc[('a','w')].values[0][0] # updated
return a

class TestAssert(unittest.TestCase):
def test_assert(self):
myclass_obj = myclass()
result = myclass_obj.get_aw()
expect = '3'
print(type(result))
print(type(expect))
print(f'result:{result}')
print(f'expect:{expect}')
self.assertTrue(isinstance(result, str)) # added
self.assertEqual(result,expect)


if __name__ == '__main__':
unittest.main(verbosity=2)


Results:

test_assert (__main__.TestAssert) ... test_code.py:18: PerformanceWarning: indexing past lexsort depth may impact performance.
a = self.df.loc[('a','w')].values[0][0]
<class 'str'>
<class 'str'>
result:3
expect:3
ok

----------------------------------------------------------------------
Ran 1 test in 0.002s

OK

Continue reading...