adding CountryCountsByBatchcodeTablesMerger

src/CountriesByBatchcodeProvider.py

@@ -2,28 +2,15 @@ import pandas as pd
 from BatchCodeTableFactory import BatchCodeTableFactory
 from InternationalVaersCovid19Provider import getInternationalVaersCovid19
 from SummationTableFactory import SummationTableFactory
+from CountryCountsByBatchcodeTablesMerger import CountryCountsByBatchcodeTablesMerger
 
 
 def getCountryCountsByBatchcodeTable():
     return _combineCountryCountsByBatchcodeTables(
-        countryCountsByClickedBatchcode = _getCountryCountsByClickedBatchcode(),
+        countryCountsByClickedBatchcode = CountryCountsByBatchcodeTablesMerger.getCountryCountsByClickedBatchcodeTable(),
         countryCountsByBatchcodeBeforeDeletion = _getCountryCountsByBatchcodeBeforeDeletion())
 
 
-def _getCountryCountsByClickedBatchcode():
-    exploration = pd.read_csv('data/Country By Clicked Batchcode.csv', index_col = 0, skiprows = [0, 1, 2, 3, 4, 5, 7])
-    exploration.index.name = 'VAX_LOT'
-    exploration.rename(
-        columns =
-        {
-            'Country': 'COUNTRY',
-            'Event count': 'COUNTRY_COUNT_BY_VAX_LOT'
-        },
-        inplace = True)
-    exploration.set_index('COUNTRY',append = True, inplace = True)
-    return exploration
-
-
 def _getCountryCountsByBatchcodeBeforeDeletion():
     internationalVaersCovid19 = getInternationalVaersCovid19(dataDir = 'VAERS/VAERSBeforeDeletion', years = [2020, 2021, 2022])
     return (internationalVaersCovid19
@@ -47,7 +34,8 @@ def _combineCountryCountsByBatchcodeTables(countryCountsByClickedBatchcode, coun
 
 
 def getCountriesByClickedBatchcode():
-    return (_getCountryCountsByClickedBatchcode()
+    return (CountryCountsByBatchcodeTablesMerger
+        .getCountryCountsByClickedBatchcodeTable()
         .reset_index(level = 'COUNTRY')
         .groupby('VAX_LOT')
         .agg(

src/CountryCountsByBatchcodeTablesMerger.py

@@ -1,10 +1,21 @@
 import pandas as pd
+import glob
+from CountryCountsByClickedBatchcodeProvider import CountryCountsByClickedBatchcodeProvider
 
 class CountryCountsByBatchcodeTablesMerger:
 
     @staticmethod
-    def merge(countryCountsByBatchcodeTables):
+    def mergeCountryCountsByBatchcodeTables(countryCountsByBatchcodeTables):
         return (pd
                 .concat(countryCountsByBatchcodeTables)
                 .groupby(countryCountsByBatchcodeTables[0].index.names)
                 .sum())
+
+    @staticmethod
+    def getCountryCountsByClickedBatchcodeTable():
+        return CountryCountsByBatchcodeTablesMerger.mergeCountryCountsByBatchcodeTables(CountryCountsByBatchcodeTablesMerger._getTables())
+
+    @staticmethod
+    def _getTables():
+        files = glob.glob(r'data/*')
+        return [CountryCountsByClickedBatchcodeProvider.getCountryCountsByClickedBatchcode(file) for file in files]

src/CountryCountsByBatchcodeTablesMergerTest.py

@@ -6,7 +6,7 @@ from CountryCountsByBatchcodeTablesMerger import CountryCountsByBatchcodeTablesM
 
 class CountryCountsByBatchcodeTablesMergerTest(unittest.TestCase):
 
-    def test_merge(self):
+    def test_mergeCountryCountsByBatchcodeTables(self):
         # Given
         countryCountsByBatchcodeTable1 = TestHelper.createDataFrame(
             columns = ['COUNTRY_COUNT_BY_VAX_LOT'],
@@ -24,7 +24,7 @@ class CountryCountsByBatchcodeTablesMergerTest(unittest.TestCase):
                         tuples = [['12345',   'Germany']]))
             
         # When
-        mergedCountryCountsByBatchcodeTables = CountryCountsByBatchcodeTablesMerger.merge(
+        mergedCountryCountsByBatchcodeTables = CountryCountsByBatchcodeTablesMerger.mergeCountryCountsByBatchcodeTables(
             [
                 countryCountsByBatchcodeTable1,
                 countryCountsByBatchcodeTable2

src/CountryCountsByClickedBatchcodeProvider.py

@@ -0,0 +1,17 @@
+import pandas as pd
+
+class CountryCountsByClickedBatchcodeProvider:
+
+    @staticmethod
+    def getCountryCountsByClickedBatchcode(file):
+        exploration = pd.read_csv(file, index_col = 0, skiprows = [0, 1, 2, 3, 4, 5, 7])
+        exploration.index.name = 'VAX_LOT'
+        exploration.rename(
+            columns =
+            {
+                'Country': 'COUNTRY',
+                'Event count': 'COUNTRY_COUNT_BY_VAX_LOT'
+            },
+            inplace = True)
+        exploration.set_index('COUNTRY',append = True, inplace = True)
+        return exploration

src/HowBadIsMyBatch.ipynb

@@ -166,13 +166,26 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": null,
    "id": "dd9fb2b0",
    "metadata": {},
    "outputs": [],
    "source": [
     "countryCountsByBatchcode.to_excel('tmp/countryCountsByBatchcode.xlsx')"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "49ecce85",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from CountryCountsByBatchcodeTablesMerger import CountryCountsByBatchcodeTablesMerger\n",
+    "\n",
+    "mergedCountryCountsByClickedBatchcodeTables = CountryCountsByBatchcodeTablesMerger.getCountryCountsByClickedBatchcodeTable()\n",
+    "mergedCountryCountsByClickedBatchcodeTables"
+   ]
   }
  ],
  "metadata": {