如何在pyspark中使用具有许多条件的join？

Question

I am able to use the dataframe join statement with single on condition ( in pyspark) But, if I try to add multiple conditions, then It is failing.

Code :

   summary2 = summary.join(county_prop, ["category_id", "bucket"], how = "leftouter").

The above code works. However If I add some other condition for list like, summary.bucket == 9 or something, it fails. Please help me fix this issue.

   The error for the statement 
   summary2 = summary.join(county_prop, ["category_id", (summary.bucket)==9], how = "leftouter")

   ERROR : TypeError: 'Column' object is not callable

Edit :

Adding full working example.

   schema = StructType([StructField("category", StringType()), StructField("category_id", StringType()), StructField("bucket", StringType()), StructField("prop_count", StringType()), StructField("event_count", StringType()), StructField("accum_prop_count",StringType())])
   bucket_summary = sqlContext.createDataFrame([],schema)

   temp_county_prop = sqlContext.createDataFrame([("nation","nation",1,222,444,555),("nation","state",2,222,444,555)],schema)
   bucket_summary = bucket_summary.unionAll(temp_county_prop)
   county_prop = sqlContext.createDataFrame([("nation","state",2,121,221,551)],schema)

Want to do a join on :

category_id and bucket columns, I want to replace the values of county_prop on bucket_summary.

   cond = [bucket_summary.bucket == county_prop.bucket, bucket_summary.bucket == 2]

bucket_summary2 = bucket_summary.join(county_prop, cond, how = "leftouter")

   1. It works if I mention the whole statement with cols, but if I list conditions like ["category_id", "bucket"]  --- THis too works.

   2. But, if I use a combination of both like cond =["bucket", bucket_summary.category_id == "state"] 

It is not working. What can go wrong with the 2 statement?

Answer 1

eg

df1.join(df2, on=[df1['age'] == df2['age'], df1['sex'] == df2['sex']], how='left_outer')

But in your case, (summary.bucket)==9 should not appear as join condition

UPDATE:

In join condition you can use a list of Column join expression or a list of Column / column_name