pipelinesds 0.0.8

Solution for DS Team

pipelinesds

Pipelinesds is a library that includes functions used in Kubeflow pipelines such as:

vertex_pipeline.py

get_data_from_bq()

Returns data from BigQuery as a DataFrame.

• Parameters:

  • bq_client: BigQuery client.
  • bq_storage_client: BigQuery Storage client.
  • table: Name of the table/view to get data from.
  • where_clause: Optional SQL WHERE clause to filter data.

• Returns:

  • pd.DataFrame: Data from the view/table.

call_procedure_and_get_data_from_bq()

Calls a stored procedure in BigQuery and returns the results as a DataFrame.

• Parameters:

  • bq_client: BigQuery client.
  • procedure_name: Name of the stored procedure to call.
  • parameters: Optional list of parameters to pass to the procedure. If no parameters are provided, an empty list is used.

• Returns:

  • pd.DataFrame: The result of the procedure call as a DataFrame.

delete_old_data()

Deletes old data from a BigQuery table.

• Parameters:
  • bq_client: BigQuery client.
  • table: Name of the table/view to delete data from.
  • where_clause: SQL WHERE clause to filter data for deletion.

write_dataframe_to_bq()

Writes a DataFrame to a BigQuery table.

• Parameters:
  • bq_client: BigQuery client.
  • df: DataFrame to write.
  • table_id: Table in BigQuery to write the DataFrame.
  • write_disposition: Type of write operation ('WRITE_APPEND', 'WRITE_TRUNCATE', or 'WRITE_EMPTY').
  • job_config: Configuration for the load job.

read_gcs_file()

Reads a file from a specific path on Google Cloud Storage.

• Parameters:

  • gcs_client: Google Cloud Storage client.
  • bucket_name: Name of the bucket on GCS where the file is stored.
  • destination_blob_name: Path in the bucket to read the file.

• Returns:

  • object: The object read from the file.

save_gcs_file()

Saves content to a specific path on Google Cloud Storage.

• Parameters:
  • gcs_client: Google Cloud Storage client.
  • bucket_name: Name of the bucket on GCS where the file will be saved.
  • destination_blob_name: Path in the bucket to save the file.
  • content: The content to be saved.
  • content_type: The MIME type of the content (e.g., 'text/html' or 'application/json').

monitoring.py

mapping()

Creates a column mapping from a configuration file.

• Parameters:

  • mapping_file: Dictionary containing mapping configuration with possible keys:
    • numerical_features
    • categorical_features
    • datetime
    • id

• Returns:

  • ColumnMapping: Evidently ColumnMapping object with configured mappings.

test_data()

Tests data for issues using a test suite.

• Parameters:

  • current_data: Current data to test.
  • reference_data: Reference data.
  • config_file: Tests configuration file.
  • stage: Stage of the pipeline ('test_input' or 'test_output').

• Returns:

  • pd.DataFrame: Test results.

check_data_drift()

Checks data for drift.

• Parameters:

  • current_data: Current data to check.
  • reference_data: Reference data.
  • config_file: Tests configuration file.

• Returns:

  • pd.DataFrame: Test results.

send_email_with_table()

Sends an email with an HTML table.

• Parameters:
  • credentials_frame: DataFrame with credentials.
  • subject: Subject of the email.
  • html_table: Data to send in the email.
  • receiver_email: Email address to send the email to.