pyspark join with 2 lookup tables

Question

I have one sales data and product details in two lookup table

df_prod_lookup1

ID     product     description
1      cereal      Minipack
2      canola      bottle
4      rice        bag

df_prod_lookup2

ID     product     description
6      glass       bottle
8      plants      hibiscus
10     tree        banyan

sales_df

ID     product     
10     tree        
1      cereal      
4      rice        
8      plants 

Expected output:

ID     product     description
10     tree        banyan
1      cereal      Minipack
4      rice        bag
8      plants      hibiscus

I am supposed to use lookup table 1 and later lookup table 2 if ID is not available in lookup table 1

lookup table 1 and 2 are of different column names and can not be merged as one. Is tehre a way to check if ID is available in lookuptable 1 and do the join if not then lookup table 2 for every record in the sales? Thanks.

I could do only simple join with one lookup table.

df_final = sales_df.join(df_prod_lookup1 on=['ID'], how='left')

Regards

Answer 1

Left join first with lookup table 1, and then with lookup table 2.
The coalesce function allows you to merge the description fields.

df_prod_lookup1 = df_prod_lookup1.withColumnRenamed("product", "product1").withColumnRenamed("description", "description1")
df_prod_lookup2 = df_prod_lookup2.withColumnRenamed("product", "product2").withColumnRenamed("description", "description2")

from pyspark.sql.functions import coalesce

# Edit based on comments #
sales_df.join(df_prod_lookup1, on=['ID'], how='left')\
        .join(df_prod_lookup2, on=['ID'], how='left')\
        .withColumn('product', coalesce('product1', 'product2'))\
        .withColumn('description', coalesce('description1', 'description2'))\
        .drop('product1', 'product2', 'description1', 'description2').show()

+---+-------+-----------+
| ID|product|description|
+---+-------+-----------+
|  8| plants|   hibiscus|
|  1| cereal|   Minipack|
| 10|   tree|     banyan|
|  4|   rice|        bag|
+---+-------+-----------+