How To Save a File as a Pickle Object to the Databricks File System

Rasputin312
Databricks Partner

‎03-02-2025 03:42 PM

I tried running this code:

```

def save_file(name, obj😞
   with open(name, 'wb') as
        pickle.dump(obj, f)

``` 

One file was saved in the local file system, but the second was too large and so I need to save in the dbfs file system.  Unfortunately, I don't see any method that allows me to do that.  They all refer to saving dataframes, and this is not a dataframe.  It's a python object.

0 Kudos

JissMathew
Valued Contributor

‎03-03-2025 02:58 AM

To save a Python object to the Databricks File System (DBFS), you can use the dbutils.fs module to write files to DBFS. Since you are dealing with a Python object and not a DataFrame, you can use the pickle module to serialize the object and then write it to DBFS. Here's how you can modify your code to achieve this:

First, ensure you have imported the necessary modules:
Python

import pickle
import os
Use the dbutils.fs module to write the serialized object to DBFS. You can use the open function with a DBFS path to write the file:
Python

def save_file_to_dbfs(dbfs_path, obj):
# Serialize the object to a byte stream
serialized_obj = pickle.dumps(obj)

# Write the serialized object to a file in DBFS
with open('/dbfs' + dbfs_path, 'wb') as f:
f.write(serialized_obj)

my_object = {'key': 'value'} # Replace with your actual object
dbfs_file_path = '/FileStore/my_object.pkl' # Path in DBFS
save_file_to_dbfs(dbfs_file_path, my_object)

Jiss Mathew
India .

View solution in original post