So when I read it in, this is how I read it in. I'm changing the temporary view table name so it wont conflict with your other table.

Python:

df1 = spark.read.format("csv").option("header", "true").option("escape", "\"").option("multiLine",True).load("file_path_where_csv_file_is_located")
df1.createOrReplaceTempView("owner_final_delta1")
display(df1)

      select 
        from_json(contacts, """array<struct<address:struct<apartment:string,city:string,house:string,poBox:string,sources:array<string>,state:string,street:string,type:string,zip:string>,addresses:array<struct<apartment:string,city:string,house:string,lastSeen:string,poBox:string,sources:array<string>,state:string,street:string,type:string,zip:string>>,contactKey:string,emails:array<struct<emailId:string,lastSeen:string,sources:array<string>>>,lastModifiedDate:string,name:struct<firstNames:array<string>,lastNames:array<string>,middleNames:array<string>,salutations:array<string>,suffixes:array<string>>,phones:array<struct<extension:string,lastSeen:string,lineType:string,number:string,sources:array<string>,validSince:string>>,relationship:string,sources:array<string>>>""") contacts_parsed, 
        * 
      from owner_final_delta1

Then the last one is SQL.