Merge pull request #61 from abhishekpnt/cbrelease-4.8.28

Issue #KB-0000: Fixing contentdf and course report

Author: shishirsuman092

dfutil/content/contentDFUtil.py

@@ -331,7 +331,12 @@ def preComputeContentWarehouseData(spark):
             .select(
                 col("courseID").alias("content_id"),
                 col("courseOrgID").alias("content_provider_id"),
-                when(col("courseOrgName").isNotNull & trim(col("courseOrgName")) != "", col("courseOrgName")).otherwise(col("contentCreator")).alias("content_provider_name"),
+                when(
+                    col("courseOrgName").isNotNull() & 
+                    (col("courseOrgName") != "") &
+                    (col("courseOrgName") != " "),
+                    col("courseOrgName")
+                ).otherwise(col("contentCreator")).alias("content_provider_name"),
                 col("courseName").alias("content_name"),
                 col("category").alias("content_type"),
                 col("batchID").alias("batch_id"),

jobs/stage-2/courseReport.py

@@ -334,7 +334,12 @@ def process_data(self, spark,config):
                 .select(
                     col("courseID").alias("content_id"),
                     col("courseOrgID").alias("content_provider_id"),
-                    when(col("courseOrgName").isNotNull & trim(col("courseOrgName")) != "", col("courseOrgName")).otherwise(col("contentCreator")).alias("content_provider_name"),
+                    when(
+                        col("courseOrgName").isNotNull() & 
+                        (col("courseOrgName") != "") &
+                        (col("courseOrgName") != " "),
+                        col("courseOrgName")
+                    ).otherwise(col("contentCreator")).alias("content_provider_name"),
                     col("courseName").alias("content_name"),
                     col("category").alias("content_type"),
                     col("batchID").alias("batch_id"),