calculating _daystoclose during ingest; nc code fixes

server/src/services/databaseOrm.py

@@ -25,9 +25,10 @@ class Ingest(Base, Mixin):
 
     # dates
     createddate = Column(DateTime, index=True)
+    closeddate = Column(DateTime)
+    _daystoclose = Column(Float(1))
     updateddate = Column(DateTime)
     servicedate = Column(DateTime)
-    closeddate = Column(DateTime)
 
     # about
     requesttype = Column(String, index=True)

server/src/services/sqlIngest.py

@@ -86,7 +86,7 @@ def dropDuplicates(table, report):
 
             report.append({
                 'description': 'dropped duplicate rows by srnumber',
-                'rows': rows.rowcount
+                'rowsAffected': rows.rowcount
             })
 
         def switchPrimaryKey(table, report):
@@ -97,7 +97,7 @@ def switchPrimaryKey(table, report):
 
             report.append({
                 'description': 'switched primary key column to srnumber',
-                'rows': 'N/A'
+                'rowsAffected': 'N/A'
             })
 
         def removeInvalidClosedDates(table, report):
@@ -112,13 +112,55 @@ def removeInvalidClosedDates(table, report):
                 'rowsAffected': result.rowcount
             })
 
+        def setDaysToClose(table, report):
+            result = exec_sql(f"""
+              UPDATE {table}
+              SET _daystoclose = EXTRACT (
+                  EPOCH FROM
+                  (closeddate::timestamp - createddate::timestamp) /
+                  (60 * 60 * 24)
+              );
+            """)
+
+            report.append({
+              'description': 'set _daystoclose column',
+              'rowsAffected': result.rowcount
+            })
+
+        def fixNorthWestwood(table, report):
+            result = exec_sql(f"""
+              UPDATE {table}
+              SET nc = 127
+              WHERE nc = 0 AND ncname = 'NORTH WESTWOOD NC'
+            """)
+
+            report.append({
+              'description': 'fix nc code for North Westwood NC',
+              'rowsAffected': result.rowcount
+            })
+
+        def fixHistoricCulturalNorth(table, report):
+            result = exec_sql(f"""
+              UPDATE {table}
+              SET nc = 128
+              WHERE nc = 0 AND ncname = 'HISTORIC CULTURAL NORTH NC'
+            """)
+
+            report.append({
+              'description': 'fix nc code for Historic Cultural North NC',
+              'rowsAffected': result.rowcount
+            })
+
         log('\nCleaning ingest table.')
         table = Ingest.__tablename__
         report = []
 
         dropDuplicates(table, report)
         switchPrimaryKey(table, report)
         removeInvalidClosedDates(table, report)
+        setDaysToClose(table, report)
+        fixNorthWestwood(table, report)
+        fixHistoricCulturalNorth(table, report)
 
         return report
 

server/src/services/timeToCloseService.py

@@ -53,25 +53,16 @@ def get_boxplot_stats(arr, C=1.5):
             }
 
         # grab the necessary data from the db
-        fields = [groupField, 'createddate', 'closeddate']
+        fields = [groupField, '_daystoclose']
         data = self.dataAccess.query(fields, filters)
 
         # read into a dataframe and drop the nulls
-        df = pd.DataFrame(data, columns=fields).dropna()
-
-        # generate a new dataframe that contains the number of days it
-        # takes to close each request, plus the type of request
-        df['closeddate'] = pd.to_datetime(df['closeddate'])
-        df['createddate'] = pd.to_datetime(df['createddate'])
-        df['time-to-close'] = df['closeddate'] - df['createddate']
-        df['hours-to-close'] = df['time-to-close'].astype('timedelta64[h]')
-        df['days-to-close'] = (df['hours-to-close'] / 24).round(2)
-        dtc_df = df[[groupField, 'days-to-close']]
+        dtc_df = pd.DataFrame(data, columns=fields).dropna()
 
         # group the requests by type and get box plot stats for each type
         data = dtc_df \
             .groupby(by=groupField) \
-            .apply(lambda df: get_boxplot_stats(df['days-to-close'].values)) \
+            .apply(lambda df: get_boxplot_stats(df['_daystoclose'].values)) \
             .to_dict()
 
         # if no rows exist for a particular item in the groupField,