5 Work with Files

You can store and organize your files in volumes in your AI Data Platform.

AI Data Platform supports multiple methods for accessing data stored in volumes:
  • POSIX-style paths: Allow users to provide access to data relative to the driver root (/). Users can read/write data to volumes or workspace folders.
  • URI-style paths: Allow users to provide access to data using a URI scheme. For example, if you want to read data in OCI Object Storage, you should provide a valid URI scheme to read/write that data.

Here are some examples:

Source Access Pattern Example
Volume POSIX Example 1
df_csv = spark.read.csv("/Volumes/<<catalog_name>>/<<schema_name>>/<<volume_name>>/<<file_name>>.csv",
    header=True,
    inferSchema=True,
    sep=",")
Example 2
import pandas as pd
df_panda_csv=pd.read_csv("/Volumes/<<catalog_name>>/<<schema_name>>/<<volume_name>>/<<file_name>>.csv", 
    header=0,
    sep=",")
Example 3
import os
os.listdir("/Volumes/<<catalog_name>>/<<schema_name>>/<<volume_name>>/")
URI 
df = spark.read.format("csv").option("header",True).load("file:///Volumes//<<catalog_name>>/<<schema_name>/<<volume_name>>/<<folder_path>>/<<file_name>>.csv")df.show()
Workspace POSIX Example 1
df_csv = spark.read.csv("/Workspace/<<folder_path>>/<<file_name>>.csv", header=True, inferSchema=True, sep=",")
df_csv.show()
Example 2
import pandas as pd
df_panda_csv=pd.read_csv("/Workspace/<<folder_path>>/<<file_name>>.csv", header=0, sep=",")
df_panda_csv.head()
Example 3
import osos.listdir("/Workspace/<<folder_path>>/")
URI 
spark.read.format("json").load("file:///Workspace/<<folder_path>>/<<file_name>>.json").show()
OCI Object Storage URI 
df_csv = spark.read.csv("oci://<<bucket_name>>@<<namespace>>/<<folder/file>>",
    header=True,
    inferSchema=True,   
    sep=",")