starting CountryColumnsMergerTest
This commit is contained in:
23
src/CountryColumnsMerger.py
Normal file
23
src/CountryColumnsMerger.py
Normal file
@@ -0,0 +1,23 @@
|
||||
import pandas as pd
|
||||
|
||||
|
||||
class CountryColumnsMerger:
|
||||
|
||||
@staticmethod
|
||||
def mergeCountryColumnOfSrcIntoDst(src: pd.DataFrame, dst: pd.DataFrame):
|
||||
def merge(series):
|
||||
if pd.isnull(series['COUNTRY_dst']):
|
||||
return series['COUNTRY_src']
|
||||
else:
|
||||
return series['COUNTRY_dst']
|
||||
|
||||
merged = pd.merge(
|
||||
dst,
|
||||
src,
|
||||
how = 'left',
|
||||
left_index = True,
|
||||
right_index = True,
|
||||
suffixes=('_dst', '_src'))
|
||||
merged['COUNTRY'] = merged.apply(merge, axis = 'columns').astype('string')
|
||||
return merged.drop(columns = ['COUNTRY_dst', 'COUNTRY_src'])
|
||||
|
||||
53
src/CountryColumnsMergerTest.py
Normal file
53
src/CountryColumnsMergerTest.py
Normal file
@@ -0,0 +1,53 @@
|
||||
import unittest
|
||||
from TestHelper import TestHelper
|
||||
from pandas.testing import assert_frame_equal
|
||||
import pandas as pd
|
||||
from CountryColumnsMerger import CountryColumnsMerger
|
||||
|
||||
|
||||
class CountryColumnsMergerTest(unittest.TestCase):
|
||||
|
||||
def test_mergeCountryColumns(self):
|
||||
# Given
|
||||
unknown = TestHelper.createDataFrame(
|
||||
columns = ['COUNTRY'],
|
||||
data = [ ['United Kingdom'],
|
||||
['France'],
|
||||
[None]],
|
||||
index = pd.Index(
|
||||
name = 'VAERS_ID',
|
||||
data = [
|
||||
'4711',
|
||||
'0815',
|
||||
'123']),
|
||||
dtypes = {'COUNTRY': 'string'})
|
||||
|
||||
known = TestHelper.createDataFrame(
|
||||
columns = ['COUNTRY'],
|
||||
data = [ ['United Kingdom'],
|
||||
['Germany']],
|
||||
index = pd.Index(
|
||||
name = 'VAERS_ID',
|
||||
data = [
|
||||
'4711',
|
||||
'123']),
|
||||
dtypes = {'COUNTRY': 'string'})
|
||||
|
||||
# When
|
||||
merged = CountryColumnsMerger.mergeCountryColumnOfSrcIntoDst(src = known, dst = unknown)
|
||||
|
||||
# Then
|
||||
assert_frame_equal(
|
||||
merged,
|
||||
TestHelper.createDataFrame(
|
||||
columns = ['COUNTRY'],
|
||||
data = [ ['United Kingdom'],
|
||||
['France'],
|
||||
['Germany']],
|
||||
index = pd.Index(
|
||||
name = 'VAERS_ID',
|
||||
data = [
|
||||
'4711',
|
||||
'0815',
|
||||
'123']),
|
||||
dtypes = {'COUNTRY': 'string'}))
|
||||
Reference in New Issue
Block a user