Update rentsense_former_tenants_to_refined.py

Author: daroclark

scripts/jobs/rentsense_former_tenants_to_refined.py

@@ -12,7 +12,7 @@
 from awsglue.utils import getResolvedOptions
 from pyspark.context import SparkContext
 from pyspark.sql.functions import (
-   col,
+    col,
     current_date,
     date_format,
     date_sub,
@@ -525,7 +525,8 @@ def export_dynamic_frame_as_xml_gzip(dynamic_frame: DynamicFrame, s3_bucket: str
     # patch = patch.distinct()
 
     # data loads - takes all records that have and end date and secure tenancies, intro and mesne tenancies are added after as there are some tenancy ids with both
-    accounts = df2.filter("endoftenuredate is not NULL and paymentreference<>'' and paymentreference<>'Blank - not in use'")
+    accounts = df2.filter(
+        "endoftenuredate is not NULL and paymentreference<>'' and paymentreference<>'Blank - not in use'")
     accounts_s = accounts.where(col("description").isin({"Secure", "Non-Secure", "Private Garage", "Tenant Garage"}))
     accounts_int = accounts.where(col("description").isin({"Introductory", "Mense Profit Ac"}))
     accounts_int = accounts_int.join(accounts_s, accounts_int.paymentreference == accounts_s.paymentreference,
@@ -570,9 +571,9 @@ def export_dynamic_frame_as_xml_gzip(dynamic_frame: DynamicFrame, s3_bucket: str
 
     # Arrangements
 
-    ten = accounts2.select('uh_ten_ref', 'paymentreference')\
+    ten = accounts2.select('uh_ten_ref', 'paymentreference') \
         .filter(col("previousweekbalance") > 0.00)
-    
+
     arr = df5.join(ten, df5.tenancy_ref == ten.uh_ten_ref, "inner")
     arr = arr.distinct()
     arr = arr.where(col("current_state").isin({"live", "breached"}))
@@ -802,14 +803,19 @@ def export_dynamic_frame_as_xml_gzip(dynamic_frame: DynamicFrame, s3_bucket: str
 
     accounts2 = accounts2.withColumn('days_since_termination', datediff(current_date(), F.col('TenancyEndDate')))
 
-    accounts3 = accounts2.withColumn('Patch',when(accounts2['days_since_termination'] < 7, 'New Terminations')
-                                          .when((accounts2['days_since_termination'] >= 7) & (accounts2['days_since_termination'] <= 31), '0-1 month')
-                                          .when((accounts2['days_since_termination'] > 31) & (accounts2['days_since_termination'] <= 92), '2-3 months')
-                                          .when((accounts2['days_since_termination'] > 92) & (accounts2['days_since_termination'] <= 182), '3-6 months')
-                                          .when((accounts2['days_since_termination'] > 182) & (accounts2['days_since_termination'] <= 365), '6-12 months')
-                                          .when((accounts2['days_since_termination'] > 365) & (accounts2['days_since_termination'] <= 2191), '1 year - 6 years')
-                                          .when(accounts2['days_since_termination'] > 2191, '6 years+')
-                                          .otherwise('Unknown'))
+    accounts3 = accounts2.withColumn('Patch', when(accounts2['days_since_termination'] < 7, 'New Terminations')
+                                     .when(
+        (accounts2['days_since_termination'] >= 7) & (accounts2['days_since_termination'] <= 31), '0-1 month')
+                                     .when(
+        (accounts2['days_since_termination'] > 31) & (accounts2['days_since_termination'] <= 92), '2-3 months')
+                                     .when(
+        (accounts2['days_since_termination'] > 92) & (accounts2['days_since_termination'] <= 182), '3-6 months')
+                                     .when(
+        (accounts2['days_since_termination'] > 182) & (accounts2['days_since_termination'] <= 365), '6-12 months')
+                                     .when(
+        (accounts2['days_since_termination'] > 365) & (accounts2['days_since_termination'] <= 2191), '1 year - 6 years')
+                                     .when(accounts2['days_since_termination'] > 2191, '6 years+')
+                                     .otherwise('Unknown'))
 
     accounts3 = accounts3.selectExpr("paymentreference as AccountReference",
                                      "description as TenureType",