filterByBatchcodes()

src/CountriesByBatchcodeProvider.py

@@ -49,3 +49,7 @@ def getCountriesByBatchcodeBeforeDeletion():
     internationalVaersCovid19 = getInternationalVaersCovid19(dataDir = 'VAERS/VAERSBeforeDeletion', years = [2020, 2021, 2022])
     batchCodeTable = BatchCodeTableFactory(internationalVaersCovid19).createGlobalBatchCodeTable(countriesAsList = True)
     return batchCodeTable[['Countries']]
+
+
+def filterByBatchcodes(countryCountsByBatchcode, batchcodes2Retain):
+    return countryCountsByBatchcode.loc[(batchcodes2Retain, slice(None)), :]

src/CountriesByBatchcodeProviderTest.py

@@ -0,0 +1,38 @@
+import unittest
+from TestHelper import TestHelper
+from pandas.testing import assert_frame_equal
+import pandas as pd
+from CountriesByBatchcodeProvider import filterByBatchcodes
+
+class CountriesByBatchcodeProviderTest(unittest.TestCase):
+    
+    def test_filterByBatchcodes(self):
+        # Given
+        countryCountsByBatchcode = TestHelper.createDataFrame(
+                columns = ['COUNTRY_COUNT_BY_VAX_LOT Clicked', 'COUNTRY_COUNT_BY_VAX_LOT Before Deletion'],
+                data = [  [10,                                 20],
+                          [90,                                 95],
+                          [15,                                 30],
+                          [70,                                 80]],
+                index = pd.MultiIndex.from_tuples(
+                            names =   ['VAX_LOT',   'COUNTRY'],
+                            tuples = [('!D0181',    'Germany'),
+                                      ('# 009C01A', 'United States'),
+                                      ('!D0181',    'Hungary'),
+                                      ('# 009C01A', 'Germany')]))
+
+        # When
+        countryCountsByBatchcodeFiltered = filterByBatchcodes(countryCountsByBatchcode, ['!D0181'])
+
+        # Then
+        assert_frame_equal(
+            countryCountsByBatchcodeFiltered,
+            TestHelper.createDataFrame(
+                columns = ['COUNTRY_COUNT_BY_VAX_LOT Clicked', 'COUNTRY_COUNT_BY_VAX_LOT Before Deletion'],
+                data = [  [10,                                 20],
+                          [15,                                 30]],
+                index = pd.MultiIndex.from_tuples(
+                            names =   ['VAX_LOT',   'COUNTRY'],
+                            tuples = [('!D0181',    'Germany'),
+                                      ('!D0181',    'Hungary')])))
+        

src/HowBadIsMyBatch.ipynb

@@ -24,7 +24,7 @@
     "from HistogramFactoryAndPersister import createAndSaveGlobalHistograms\n",
     "from BatchCodeTableFactory import BatchCodeTableFactory\n",
     "from Column2DataframeAdder import addColumn2Dataframe\n",
-    "from CountriesByBatchcodeProvider import getCountryCountsByBatchcodeTable\n",
+    "from CountriesByBatchcodeProvider import getCountryCountsByBatchcodeTable, filterByBatchcodes\n",
     "from CountryCountsByBatchcodeTable2BarChartDescriptionTableConverter import CountryCountsByBatchcodeTable2BarChartDescriptionTableConverter\n"
    ]
   },
@@ -178,22 +178,12 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "9588a10c",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "countryCountsByBatchcode = getCountryCountsByBatchcodeTable()\n",
-    "countryCountsByBatchcode"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 30,
    "id": "70fcc420",
    "metadata": {},
    "outputs": [],
    "source": [
-    "# FK-TODO: nur die Batchcodes von countryCountsByBatchcode beibehalten, die auch in batchCodeTable vorkommen"
+    "countryCountsByBatchcode = filterByBatchcodes(getCountryCountsByBatchcodeTable(), batchCodeTable['Batch'].values)\n",
+    "countryCountsByBatchcode"
    ]
   },
   {