refactoring

src/HistogramFactory.py

@@ -0,0 +1,14 @@
+from SymptomHistogramByBatchcodeTableFactory import SymptomHistogramByBatchcodeTableFactory
+from HistogramTable2DictTableConverter import HistogramTable2DictTableConverter
+
+
+def createGlobalHistograms(symptomByBatchcodeTable):
+    symptomHistogramByBatchcodeTable = SymptomHistogramByBatchcodeTableFactory.createGlobalSymptomHistogramByBatchcodeTable(symptomByBatchcodeTable)
+    dictByBatchcodeTable = HistogramTable2DictTableConverter.convertGlobalHistogramTable2DictTable(symptomHistogramByBatchcodeTable)
+    return dictByBatchcodeTable
+
+
+def createHistograms(symptomByBatchcodeTable):
+    symptomHistogramByBatchcodeTable = SymptomHistogramByBatchcodeTableFactory.createSymptomHistogramByBatchcodeTable(symptomByBatchcodeTable)
+    dictByBatchcodeTable = HistogramTable2DictTableConverter.convertHistogramTable2DictTable(symptomHistogramByBatchcodeTable)
+    return dictByBatchcodeTable

src/HistogramFactoryAndPersister.py

@@ -0,0 +1,24 @@
+from HistogramFactory import createGlobalHistograms, createHistograms
+from HistogramPersister import saveHistograms
+
+
+def createAndSaveGlobalHistograms(symptomByBatchcodeTable):
+    dictByBatchcodeTable4Country = createGlobalHistograms(symptomByBatchcodeTable)
+    saveHistograms(dictByBatchcodeTable4Country, 'Global')
+
+
+def createAndSaveHistogramsForCountries(symptomByBatchcodeTable, countries):
+    dictByBatchcodeTable = createHistograms(symptomByBatchcodeTable)
+    for count, country in enumerate(countries, start = 1):
+        _createAndSaveHistogramsForCountry(
+            count = count,
+            numCountries = len(countries),
+            country = country,
+            dictByBatchcodeTable = dictByBatchcodeTable)
+
+
+def _createAndSaveHistogramsForCountry(count, numCountries, country, dictByBatchcodeTable):
+    # FK-TODO: use https://github.com/tqdm/tqdm
+    print(f'saving histograms for country {count}/{numCountries}: {country}')
+    dictByBatchcodeTable4Country = dictByBatchcodeTable[dictByBatchcodeTable['COUNTRY'] == country]
+    saveHistograms(dictByBatchcodeTable4Country, country)

src/HistogramPersister.py

@@ -0,0 +1,13 @@
+from MultiIndexValuesProvider import MultiIndexValuesProvider
+from HistogramDescriptionPersister import HistogramDescriptionPersister
+import shutil
+
+def saveHistograms(dictByBatchcodeTable4Country, country):
+    batchcodes = MultiIndexValuesProvider.getValues(dictByBatchcodeTable4Country.index)
+    batchcodes = {batchcode for batchcode in batchcodes if batchcode != 'nan'}
+    directory = f'../docs/data/histograms/{country}'
+    shutil.rmtree(directory, ignore_errors = True)
+    HistogramDescriptionPersister(directory).saveHistogramDescriptionsForBatchcodes(
+        batchcodes,
+        dictByBatchcodeTable4Country,
+        progress = lambda count, size, batchcode: print(f'{count}/{size}: {batchcode}'))

src/HowBadIsMyBatch.ipynb

@@ -18,7 +18,10 @@
     "from DateProvider import DateProvider\n",
     "from InternationalVaersCovid19Provider import getInternationalVaersCovid19, get_international_VAERSVAX_VAERSSYMPTOMS_Covid19\n",
     "from BatchCodeTableHtmlUpdater import updateBatchCodeTableHtmlFile\n",
-    "from BatchCodeTablePersister import createAndSaveBatchCodeTables"
+    "from BatchCodeTablePersister import createAndSaveBatchCodeTables\n",
+    "from SymptomByBatchcodeTableFactory import SymptomByBatchcodeTableFactory\n",
+    "from HistogramFactoryAndPersister import createAndSaveGlobalHistograms, createAndSaveHistogramsForCountries\n",
+    "from HtmlUtils import getCountries"
    ]
   },
   {
@@ -104,8 +107,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from SymptomByBatchcodeTableFactory import SymptomByBatchcodeTableFactory\n",
-    "\n",
     "symptomByBatchcodeTable = SymptomByBatchcodeTableFactory.createSymptomByBatchcodeTable(international_VAERSVAX_Covid19, international_VAERSSYMPTOMS)\n",
     "symptomByBatchcodeTable"
    ]
@@ -113,76 +114,21 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "5997ee59",
+   "id": "23731536",
    "metadata": {},
    "outputs": [],
    "source": [
-    "isGlobal = False"
+    "createAndSaveGlobalHistograms(symptomByBatchcodeTable)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "9268d60d",
+   "id": "f8e42955",
    "metadata": {},
    "outputs": [],
    "source": [
-    "from SymptomHistogramByBatchcodeTableFactory import SymptomHistogramByBatchcodeTableFactory\n",
-    "\n",
-    "symptomHistogramByBatchcodeTable = SymptomHistogramByBatchcodeTableFactory.createGlobalSymptomHistogramByBatchcodeTable(symptomByBatchcodeTable) if isGlobal else SymptomHistogramByBatchcodeTableFactory.createSymptomHistogramByBatchcodeTable(symptomByBatchcodeTable)\n",
-    "symptomHistogramByBatchcodeTable"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "5170efad",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from HistogramTable2DictTableConverter import HistogramTable2DictTableConverter\n",
-    "\n",
-    "dictByBatchcodeTable = HistogramTable2DictTableConverter.convertGlobalHistogramTable2DictTable(symptomHistogramByBatchcodeTable) if isGlobal else HistogramTable2DictTableConverter.convertHistogramTable2DictTable(symptomHistogramByBatchcodeTable)\n",
-    "dictByBatchcodeTable"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "a57e0fd5",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from HtmlUtils import getCountries\n",
-    "from MultiIndexValuesProvider import MultiIndexValuesProvider\n",
-    "from HistogramDescriptionPersister import HistogramDescriptionPersister\n",
-    "import shutil\n",
-    "\n",
-    "if isGlobal:\n",
-    "    batchcodes = MultiIndexValuesProvider.getValues(dictByBatchcodeTable.index)\n",
-    "    batchcodes = {batchcode for batchcode in batchcodes if batchcode != 'nan'}\n",
-    "    directory = f'../docs/data/histograms/Global'\n",
-    "    shutil.rmtree(directory, ignore_errors = True)\n",
-    "    histogramDescriptionPersister = HistogramDescriptionPersister(directory)\n",
-    "    histogramDescriptionPersister.saveHistogramDescriptionsForBatchcodes(\n",
-    "        batchcodes,\n",
-    "        dictByBatchcodeTable,\n",
-    "        progress = lambda count, size, batchcode: print(f'{count}/{size}: {batchcode}'))\n",
-    "else:\n",
-    "    countries = getCountries(international_VAERSVAX_Covid19)\n",
-    "    for count, country in enumerate(countries, start = 1):\n",
-    "        # FK-TODO: use https://github.com/tqdm/tqdm\n",
-    "        print(f'saving histograms for country {count}/{len(countries)}: {country}')\n",
-    "        dictByBatchcodeTable4Country = dictByBatchcodeTable[dictByBatchcodeTable['COUNTRY'] == country]\n",
-    "        batchcodes = MultiIndexValuesProvider.getValues(dictByBatchcodeTable4Country.index)\n",
-    "        batchcodes = {batchcode for batchcode in batchcodes if batchcode != 'nan'}\n",
-    "        directory = f'../docs/data/histograms/{country}'\n",
-    "        shutil.rmtree(directory, ignore_errors = True)\n",
-    "        histogramDescriptionPersister = HistogramDescriptionPersister(directory)\n",
-    "        histogramDescriptionPersister.saveHistogramDescriptionsForBatchcodes(\n",
-    "            batchcodes,\n",
-    "            dictByBatchcodeTable4Country,\n",
-    "            progress = lambda count, size, batchcode: print(f'{count}/{size}: {batchcode}'))"
+    "createAndSaveHistogramsForCountries(symptomByBatchcodeTable, getCountries(international_VAERSVAX_Covid19))"
    ]
   },
   {