Custom technical lineage

This repository contains examples and helper functions to help you develop your custom technical lineage files.

Convert single-file definition files to the new batch definition format

Usage: python3 -m tools.translate_to_batch_format <source_directory> <target_directory> [--migrate_source_code]

Where:

• <source_directory> is the existing directory with the single-file definition files that you want to convert.
• <target_directory> is the target directory for the resulting batch definition artifacts. If the target directory doesn't exist, it will be created.
• --migrate_source_code is an optional element that extracts the source code.

Convert CSV files to the new batch definition format

Usage: python3 -m tools.ingest_csv <source_directory> <target_directory> [--collibraInstance] [--username] [--password]

Where:

• <source_directory> is the existing directory with the CSV files that you want to convert.
• <target_directory> is the target directory for the resulting batch definition artifacts. If the target directory doesn't exist, it will be created.
• --collibraInstance is the Collibra instance name. If instance's URL is https://myinstance.collibra.com the instance name is myinstance
• --username is the Collibra username used to make API calls
• --password is the Collibra's account password

When collibraInstance, username and password are provided, the asset type uuids provided in the CSV files will be automatically fetched from your catalog instance. When not provided you need to update the function _get_default_asset_types in tools.ingest_csv.py so they return all the assets used.

The first row in every CSV file is the header. The following rows define the lineage relationships. If we take a look at an example header row, we find in order:

System,Database,Schema,Table,Column,fullname,domain_id,System,Database,Schema,Table,Column,fullname,domain_id,source_code,highlights,transformation_display_name

• System, Database, and Schema define the asset types for the nodes of the source. You can have as many nodes as you want, but you must have a least one.
• Table is the asset type of the parent asset of the source. This header is mandatory.
• Column is the asset type of the leaf asset of the source. This header is mandatory.
• fullname and domain_id are mandatory headers in cases where a custom fullname and domain_id must be provided for the source.
• System, Database, and Schema define the asset types for the nodes of the target. You can have as many nodes as you want, but you must have a least one.
• Table is the asset type of the parent asset of the target. This header is mandatory.
• Column is the asset type of the leaf asset of the target. This header is mandatory.
• fullname and domain_id are mandatory headers in cases where a custom fullname and domain_id must be provided for the target.
• source_code, highlights, and transformation_display_name are mandatory headers.

When providing the lineage relationships, keep the following in mind:

• Values for the node and parent asset types are mandatory for both the source and the target.
• A value for the leaf asset type is optional.
• Values for fullname and domain_id are optional.
• Values for source_code, highlights, and transformation_display_name are optional.
• source_code can either be a string or the full path to a file.

Headers define the asset types for which you define the lineage relationships; therefore, one file can only contain lineage relationships for the same type of assets in the source/target. You can, however, create as many CSV files as you want in the directory. The generated metadata.json file will contain the definition for System, Database, Schema, Table, and Column. If you are using any other asset types, you need to add these in the metadata.json file, with their respective uuid.

Let's have a look at two examples:

CSV Example 1

System,Database,Schema,Table,Column,fullname,domain_id,System,Database,Schema,Table,Column,fullname,domain_id,source_code,highlights,transformation_display_name
snowflake,KRISTOF,PUBLIC,T1,USERID,,,snowflake,KRISTOF,PUBLIC,V2,UI_2L,,,CREATE VIEW KRISTOF.PUBLIC.V2 AS select USERID from KRISTOF.PUBLIC.T1,"[0:70]",transformation
snowflake,KRISTOF,PUBLIC,T2,,domain1,,,KRISTOF,PUBLIC,V2,,,,,,

The first row defines the column-level lineage and provides details about the source_code, highlights and transformation_display_name. The second row defines the table-level lineage.

CSV Example 2

GCS File System,GCS Bucket,Directory,Directory,File,fullname,domain_id,System,Database,Schema,Table,Column,fullname,domain_id,source_code,highlights,transformation_display_name
gcs,catingestiontest,/,ingestion-test,mytest.csv,f13bf705-13a4-44c9-843e-f341feccfb6e > catingestiontest/ingestion-test/ingestion copy/mytest.csv/1611609340099809,fea1b0b0-705f-4e0d-b5eb-1f21132cc718,snowflake,KRISTOF,PUBLIC,V2,UI_2L,,,snowflake data pipline abc,,transformation

This example creates a lineage relationship between a file and a column. The custom fullname and domain_id are provided for the file because they are needed to obtain stitching.

Python batch definition custom technical lineage examples

tools.example.py and tools.example_with_props.py contain examples of how you can use the models and helper functions defined in src.models.py and src.helper.py.

Retrieve the fullname and domain ID of an asset, based on the domain ID, type ID or display name

Usage: python3 -m tools.collect_assets_fullname [--collibraInstance] [--username] [--password] [--domainId] [--typeId] [--name]

Where:

• --collibraInstance is the name of the Collibra environment. If, for example, the URL of the environment is https://myinstance.collibra.com, the environment name is myinstance.
• --username is the Collibra username used to make API calls.
• --password is the Collibra account password.
• --domainId is the domain ID of the relevant asset. This is optional.
• --typeId is the asset type ID of the relevant asset. This is optional.
• --name is the display name of the relevant asset. This is optional.

Retrieve asset type ID's based on asset type name

Usage: python3 -m tools.collect_assets_fullname [--collibraInstance] [--username] [--password] [--domainId] [--typeId] [--name]

Where:

• --collibraInstance is the Collibra instance name. If instance's URL is https://myinstance.collibra.com the instance name is myinstance
• --username is the Collibra username used to make API calls
• --password is the Collibra's account password
• --applicationNameis the type of data source for which you are creating a technical lineage
• --typeId optional: is the type ID of the assets details to be retrieved

License

Custom technical lineage examples are available under the Collibra Marketplace License agreement.