Process to interact with blob storage files from Databricks notebooks

Question

Within a Azure Databricks notebook, I am attempting to perform a transformation on some csv’s which are in blob storage using the following:

*import os
    import glob
    import pandas as pd
    os.chdir(r'wasbs://dalefactorystorage.blob.core.windows.net/dale')
    allFiles = glob.glob("*.csv") # match your csvs
    for file in allFiles:
       df = pd.read_csv(file)
       df = df.iloc[4:,] # read from row 4 onwards.
       df.to_csv(file)
       print(f"{file} has removed rows 0-3")*

Unfortunately I am getting the following error:

*FileNotFoundError: [Errno 2] No such file or directory: ‘wasbs://dalefactorystorage.blob.core.windows.net/dale’

Am I missing something? (I am completely new to this).

Cheers,

Dale

Asked By: dale

||

Source

Answer 1

If you want to use package pandas to read CSV file from Azure blob process it and write
this CSV file to Azure blob in Azure Databricks, I suggest you mount Azure blob storage as Databricks filesystem then do that. For more details, please refer to here.

For example

Mount Azure blob

dbutils.fs.mount(
  source = "wasbs://<container-name>@<storage-account-name>.blob.core.windows.net",
  mount_point = "/mnt/<mount-name>",
  extra_configs = {"fs.azure.account.key.<storage-account-name>.blob.core.windows.net":"<account access key>"})

process csv

import os
import glob
import pandas as pd

os.chdir(r'/dbfs/mnt/<mount-name>/<>')
allFiles = glob.glob("*.csv") # match your csvs
for file in allFiles:
    print(f" The old content of  file {file} : ")
    df= pd.read_csv(file, header=None)
    print(df)
    df = df.iloc[4:,]
    df.to_csv(file, index=False,header=False)
    print(f" The new content of  file {file} : ")
    df= pd.read_csv(file,header=None)
    print(df)
    break

Answered By: Jim Xu

Answer 2

A, alternative approach is to mount the dbfs file as a spark dataframe, and then just convert it from a sparkdf to a pandas df:

# mount blob storage
spark.conf.set("fs.azure.account.key.storageaccountname.blob.core.windows.net",
"storageaccesskey")

dfspark = spark.read.csv("wasbs://[email protected]
/filename.csv", header="true")

# convert from sparkdf to pandasdf 
df = dfspark.toPandas()

Answered By: Deepak

Process to interact with blob storage files from Databricks notebooks

Question:

Answers: