From 05b75756c9e3d82566e7112cb72400fe0b613bb5 Mon Sep 17 00:00:00 2001
From: frankknoll <Knoll_Frank@web.de>
Date: Wed, 14 Jun 2023 01:35:03 +0200
Subject: [PATCH] refining CountryColumnsMergerTest

---
 src/CountryColumnsMerger.py     |  10 ++-
 src/CountryColumnsMergerTest.py | 108 +++++++++++++++++++++++++++++---
 2 files changed, 108 insertions(+), 10 deletions(-)

diff --git a/src/CountryColumnsMerger.py b/src/CountryColumnsMerger.py
index 15150fdc160..3d4a022cf9c 100644
--- a/src/CountryColumnsMerger.py
+++ b/src/CountryColumnsMerger.py
@@ -7,7 +7,7 @@ class CountryColumnsMerger:
     def mergeCountryColumnOfSrcIntoDst(src: pd.DataFrame, dst: pd.DataFrame):
         merged = pd.merge(
             dst,
-            src,
+            CountryColumnsMerger._drop_duplicates(src['COUNTRY']),
             how = 'left',
             left_index = True,
             right_index = True,
@@ -21,10 +21,14 @@ class CountryColumnsMerger:
                                 .astype('string'))
         return merged.drop(columns = ['COUNTRY_dst', 'COUNTRY_src'])
     
+    @staticmethod
+    def _drop_duplicates(df):
+        return df[~df.index.duplicated(keep = 'first')]
+
     @staticmethod
     def _mergeSrcIntoDst(src, dst):
-        if (CountryColumnsMerger._notEqual(src, dst)) or (pd.isnull(src) and pd.isnull(dst)):
-            raise Exception()        
+        if CountryColumnsMerger._notEqual(src, dst):
+            raise Exception(f'{src} != {dst}')        
         
         if pd.isnull(dst) and not pd.isnull(src):
             return src
diff --git a/src/CountryColumnsMergerTest.py b/src/CountryColumnsMergerTest.py
index c718bbaec27..1fe53f5eefd 100644
--- a/src/CountryColumnsMergerTest.py
+++ b/src/CountryColumnsMergerTest.py
@@ -55,14 +55,108 @@ class CountryColumnsMergerTest(unittest.TestCase):
                 dtypes = {'COUNTRY': 'string'}),
                 check_like = True)
 
-    def test_shouldNotMergeCountryColumnOfSrcIntoDst_non_unique(self):
-        self.shouldNotMergeCountryColumnOfSrcIntoDst(val_dst = 'United Kingdom', val_src = 'Germany')
-
-    def test_shouldNotMergeCountryColumnOfSrcIntoDst3(self):
-        self.shouldNotMergeCountryColumnOfSrcIntoDst(val_dst = None, val_src = None)
-
-    def shouldNotMergeCountryColumnOfSrcIntoDst(self, val_dst, val_src):
+    def test_mergeCountryColumnOfSrcIntoDst_non_unique_index(self):
         # Given
+        src_val = 'Germany'
+        dst_val = None
+        dst = TestHelper.createDataFrame(
+                columns = ['COUNTRY',        'TestColumn'],
+                data = [  ['United Kingdom', 'test1'],
+                          ['France',         'test2'],
+                          [dst_val,          'test3']],
+                index = pd.Index(
+                        name = 'VAERS_ID',
+                        data = [
+                            '4711',
+                            '0815',
+                            '123']),
+                dtypes = {'COUNTRY': 'string'})
+        
+        src = TestHelper.createDataFrame(
+                columns = ['COUNTRY'],
+                data = [  ['United Kingdom'],
+                          [src_val],
+                          [src_val]],
+                index = pd.Index(
+                        name = 'VAERS_ID',
+                        data = [
+                            '4711',
+                            '123',
+                            '123']),
+                dtypes = {'COUNTRY': 'string'})
+
+        # When
+        merged = CountryColumnsMerger.mergeCountryColumnOfSrcIntoDst(src, dst)
+
+        # Then
+        assert_frame_equal(
+            merged,
+            TestHelper.createDataFrame(
+                columns = ['COUNTRY',        'TestColumn'],
+                data = [  ['United Kingdom', 'test1'],
+                          ['France',         'test2'],
+                          [src_val,          'test3']],
+                index = pd.Index(
+                        name = 'VAERS_ID',
+                        data = [
+                            '4711',
+                            '0815',
+                            '123']),
+                dtypes = {'COUNTRY': 'string'}),
+                check_like = True)
+
+    def test_mergeCountryColumnOfSrcIntoDst_None_None(self):
+        # Given
+        src_val = None
+        dst_val = None
+        dst = TestHelper.createDataFrame(
+                columns = ['COUNTRY',        'TestColumn'],
+                data = [  ['United Kingdom', 'test1'],
+                          ['France',         'test2'],
+                          [dst_val,          'test3']],
+                index = pd.Index(
+                        name = 'VAERS_ID',
+                        data = [
+                            '4711',
+                            '0815',
+                            '123']),
+                dtypes = {'COUNTRY': 'string'})
+        
+        src = TestHelper.createDataFrame(
+                columns = ['COUNTRY'],
+                data = [  ['United Kingdom'],
+                          [src_val]],
+                index = pd.Index(
+                        name = 'VAERS_ID',
+                        data = [
+                            '4711',
+                            '123']),
+                dtypes = {'COUNTRY': 'string'})
+
+        # When
+        merged = CountryColumnsMerger.mergeCountryColumnOfSrcIntoDst(src, dst)
+
+        # Then
+        assert_frame_equal(
+            merged,
+            TestHelper.createDataFrame(
+                columns = ['COUNTRY',        'TestColumn'],
+                data = [  ['United Kingdom', 'test1'],
+                          ['France',         'test2'],
+                          [src_val,          'test3']],
+                index = pd.Index(
+                        name = 'VAERS_ID',
+                        data = [
+                            '4711',
+                            '0815',
+                            '123']),
+                dtypes = {'COUNTRY': 'string'}),
+                check_like = True)
+
+    def test_shouldNotMergeCountryColumnOfSrcIntoDst_non_unique(self):
+        # Given
+        val_dst = 'United Kingdom'
+        val_src = 'Germany'
         dst = TestHelper.createDataFrame(
                 columns = ['COUNTRY'],
                 data = [  [val_dst]],