CODEWITHSUNDEEP
pythonpysparknested
Rasika
Rasika

Reputation: 507

Extract nested column using another column value

I have a df with rows as:

 account_number          | null                 
 additional_info         | null                 
 data                    | [,,,,,,,,,,,,,,,,... 
 dealer_id               | null                 
 device_id               | 0                    
 device_name             | null                 
 device_type             | panel                
 event                   | setting              
 event_class             | panel_settings_alarm 
 event_desc              | siren_disable is ... 
 event_timestamp         | 1621341705282        
 event_type              | siren_disable        
 imei                    | 99000000000000      
 is_event_desc_available | true                 
 message_uuid            | b32e6a3f-bced-4ba... 
 operation               | report               
 partition_id            | 0                    
 priority                | 2                    
 software_version        | 2.4.2                
 source                  | panel                
 setting_name            | siren_disable

the nested column data has multiple key value pairs. With Key as the setting_name column I need to make another column with value as data[setting_name] for eg: data['siren_disable'] The names in setting name keep changing. I tried:

df2 = df2.withColumn("setting_value",df2["data"].getItem(df2['setting_name']))

But getting error:

AnalysisException: Field name should be String Literal, but it's setting_name#3864;

Any suggestions?

Upvotes: 0

Views: 46

Answers (1)

mck
mck

Reputation: 42352

One possible way is to convert the struct type column to map type:

import pyspark.sql.functions as F

df2 = df.withColumn(
    'setting_value', 
    F.from_json(F.to_json('data'), 'map<string,string>')[F.col('setting_name')]
)

Upvotes: 1

Related Questions