minifying histogram JSON files

docs/HistoDescrsProvider.js

@@ -1,13 +1,6 @@
 class HistoDescrsProvider {
 
     static getHistoDescrs(batchcode) {
-        return fetch(`data/histograms/Global/${batchcode}.json`)
+        return fetch(`data/histograms/Global/${batchcode}.json`).then(response => response.json())
-            .then(response => response.json())
-            .then(histoDescrs => {
-                histoDescrs.histograms.sort((histoDescr1, histoDescr2) => histoDescr1.batchcodes.length - histoDescr2.batchcodes.length);
-                histoDescrs.histogram = histoDescrs.histograms[0].histogram;
-                delete histoDescrs.histograms;
-                return histoDescrs;
-            });
     }
 }

src/BatchCodeTableIntoHistogramDescriptionTableMerger.py

@@ -28,9 +28,7 @@ class BatchCodeTableIntoHistogramDescriptionTableMerger:
                 'Deaths',
                 'Disabilities',
                 'Life Threatening Illnesses',
-                'Company',
+                'Company'
-                'Severe reports',
-                'Lethality'
             ]]
 
     def _merge_columns_into_HISTOGRAM_DESCRIPTION(self, table):

src/BatchCodeTableIntoHistogramDescriptionTableMergerTest.py

@@ -55,8 +55,6 @@ class BatchCodeTableIntoHistogramDescriptionTableMergerTest(unittest.TestCase):
                                 'Disabilities': 2,
                                 'Life Threatening Illnesses': 2,
                                 'Company': 'MODERNA',
-                                'Severe reports': 2/2 * 100,
-                                'Lethality': np.nan,
                                 'histograms': [
                                     {
                                         'batchcodes': ['1808982', 'EW0175', 'FD1921'],

src/HistogramDescriptionTableSelector.py

@@ -0,0 +1,25 @@
+import warnings
+
+class HistogramDescriptionTableSelector:
+
+    @staticmethod
+    def selectHistogramsWithShortestBatchcodeCombinations(histogramDescriptionTable):
+        histogramDescriptionTable['HISTOGRAM_DESCRIPTION'] = histogramDescriptionTable['HISTOGRAM_DESCRIPTION'].map(HistogramDescriptionTableSelector._selectHistogramWithShortestBatchcodeCombination)
+        return histogramDescriptionTable
+    
+    @staticmethod
+    def _selectHistogramWithShortestBatchcodeCombination(histoDescr):
+        return {
+            "batchcode": histoDescr["batchcode"],
+            "histogram": HistogramDescriptionTableSelector._getHistogramWithShortestBatchcodeCombination(histoDescr)
+            }
+    
+    @staticmethod
+    def _getHistogramWithShortestBatchcodeCombination(histoDescr):
+        histogramsSortedByShortestBatchcodeCombination = sorted(
+            histoDescr["histograms"],
+            key = lambda histogram: len(histogram["batchcodes"]))
+        histogramWithShortestBatchcodeCombination = histogramsSortedByShortestBatchcodeCombination[0]
+        if len(histogramWithShortestBatchcodeCombination["batchcodes"]) != 1:
+            warnings.warn(f"batchcode {histoDescr['batchcode']} has non unique batchcode combination {histogramWithShortestBatchcodeCombination['batchcodes']} for it's histogram")
+        return histogramWithShortestBatchcodeCombination["histogram"]

src/HistogramDescriptionTableSelectorTest.py

@@ -0,0 +1,51 @@
+import unittest
+from pandas.testing import assert_frame_equal
+from HistogramDescriptionTableSelector import HistogramDescriptionTableSelector
+from TestHelper import TestHelper
+import pandas as pd
+
+class HistogramDescriptionTableSelectorTest(unittest.TestCase):
+
+    def test_selectHistogramsWithShortestBatchcodeCombinations(self):
+        # Given
+        histogramDescriptionTable = TestHelper.createDataFrame(
+                columns = ['HISTOGRAM_DESCRIPTION'],
+                data = [  [
+                            {
+                                "batchcode": "1808982",
+                                "histograms": [
+                                    {
+                                        "batchcodes": ["1808982", "EW0175", "FD1921"],
+                                        "histogram": {"Blood pressure orthostatic abnormal": 5, "Chest discomfort": 1}
+                                    },
+                                    {
+                                        "batchcodes": ["015M20A", "1808982"],
+                                        "histogram": {"Chest discomfort": 2}
+                                    }
+                                ]
+                            }
+                          ]
+                        ],
+                index =  pd.Index(
+                    name = 'VAX_LOT',
+                    data = ['1808982']))
+
+        # When
+        histogramsWithShortestBatchcodeCombinationsTable = HistogramDescriptionTableSelector.selectHistogramsWithShortestBatchcodeCombinations(histogramDescriptionTable)
+
+        # Then
+        assert_frame_equal(
+            histogramsWithShortestBatchcodeCombinationsTable,
+            TestHelper.createDataFrame(
+                columns = ['HISTOGRAM_DESCRIPTION'],
+                data = [  [
+                            {
+                                "batchcode": "1808982",
+                                "histogram": {"Chest discomfort": 2}
+                            }
+                          ]
+                        ],
+                index =  pd.Index(
+                    name = 'VAX_LOT',
+                    data = ['1808982'])))
+        

src/HistogramFactoryAndPersister.py

@@ -2,6 +2,7 @@ from HistogramFactory import createHistograms
 from HistogramPersister import saveHistograms
 from MultiIndexExploder import MultiIndexExploder
 from HistogramDescriptionTableFactory import HistogramDescriptionTableFactory
+from HistogramDescriptionTableSelector import HistogramDescriptionTableSelector
 from BatchCodeTableIntoHistogramDescriptionTableMerger import BatchCodeTableIntoHistogramDescriptionTableMerger
 
 
@@ -10,6 +11,7 @@ def createAndSaveGlobalHistograms(symptomByBatchcodeTable, batchCodeTable):
     dictByBatchcodeTable = createHistograms(symptomByBatchcodeTable)
     explodedTable = MultiIndexExploder.explodeMultiIndexOfTable(dictByBatchcodeTable)
     histogramDescriptionTable = HistogramDescriptionTableFactory.createHistogramDescriptionTable(explodedTable)
+    histogramDescriptionTable = HistogramDescriptionTableSelector.selectHistogramsWithShortestBatchcodeCombinations(histogramDescriptionTable)
     histogramDescriptionTable = BatchCodeTableIntoHistogramDescriptionTableMerger().mergeBatchCodeTableIntoHistogramDescriptionTable(
         batchCodeTable = _rearrange(batchCodeTable),
         histogramDescriptionTable = histogramDescriptionTable)

src/HowBadIsMyBatch.ipynb

@@ -120,7 +120,7 @@
    "outputs": [],
    "source": [
     "batchCodeTable = createGlobalBatchCodeTable(\n",
-    "    deleteEntriesWithADRsLessThanOrEqual = 1,\n",
+    "    deleteEntriesWithADRsLessThanOrEqual = 2,\n",
     "    minADRsForLethality = 100,\n",
     "    batchCodeTableFactory = BatchCodeTableFactory(internationalVaersCovid19))\n",
     "batchCodeTable"

src/help.txt

@@ -4,7 +4,6 @@ FK-FIXME:
 
 FK-TODO:
 - add google captcha to batchCodeTable.html
-- aus den JSON-Dateien lethality und severity entfernen.
 
 anacron job:
 sudo cp src/intensivstationen_howbadismybatch.sh /etc/cron.daily/intensivstationen_howbadismybatch