BigQuery View Analyzer

Description

bigquery-view-analyzer is a command-line tool for visualizing dependencies and managing permissions between BigQuery views.

To authorize a view, permissions must be granted at a dataset level for every view/table referenced in the view definition. This requirement cascades down to every view that's referenced by the parent view, they too must have permissions granted for every view/table they reference - and so on. This can quickly become difficult to manage if you have many nested views across multiple datasets and/or projects.

bigquery-view-analyzer automatically resolves these dependencies and applies the relevant permissions to all views and datasets referenced by the parent view.

Installation

$ pip install bigquery-view-analyzer

Usage

$ bqva --help

Example: CLI

Given the above datasets and tables in BigQuery, to authorize bqva-demo:dataset_4.shared_view, the following views would need to be authorized with each of the following datasets:

• Authorized views for dataset_1
  • bqva-demo:dataset_3.view_a_b_c_d
• Authorized views for dataset_2
  • bqva-demo:dataset_3.view_a_b_c_d
  • bqva-demo:dataset_1.view_c
• Authorized views for dataset_3
  • bqva-demo:dataset_2.view_d
  • bqva-demo:dataset_4.shared_view

You can easily visualize the above view hierarchy using the bqva tree command.

# View dependency tree and authorization status for 'bqva-demo:dataset_4.shared_view'
$ bqva tree --status --no-key --view "bqva-demo:dataset_4.shared_view"
bqva-demo:dataset_4.shared_view
└── bqva-demo:dataset_3.view_a_b_c_d (⨯)
    ├── bqva-demo:dataset_1.table_a (⨯)
    ├── bqva-demo:dataset_1.table_b (⨯)
    ├── bqva-demo:dataset_1.view_c (⨯)
    │   └── bqva-demo:dataset_2.table_c (⨯)
    └── bqva-demo:dataset_2.view_d (⨯)
        └── bqva-demo:dataset_3.table_d (⨯)

Permissions can be applied automatically to all datasets referenced by the parent view using the bqva authorize command.

# Apply all permissions required by 'bqva-demo:dataset_4.shared_view'
$ bqva authorize --view "bqva-demo:dataset_4.shared_view"
bqva-demo:dataset_4.shared_view
└── bqva-demo:dataset_3.view_a_b_c_d (✓)
    ├── bqva-demo:dataset_1.table_a (✓)
    ├── bqva-demo:dataset_1.table_b (✓)
    ├── bqva-demo:dataset_1.view_c (✓)
    │   └── bqva-demo:dataset_2.table_c (✓)
    └── bqva-demo:dataset_2.view_d (✓)
        └── bqva-demo:dataset_3.table_d (✓)

If you want to revoke permissions for a view, you can do that too!

# Revoke all permissions granted to 'bqva-demo:dataset_4.shared_view'
$ bqva revoke --view "bqva-demo:dataset_4.shared_view"
bqva-demo:dataset_4.shared_view
└── bqva-demo:dataset_3.view_a_b_c_d (⨯)
    ├── bqva-demo:dataset_1.table_a (⨯)
    ├── bqva-demo:dataset_1.table_b (⨯)
    ├── bqva-demo:dataset_1.view_c (⨯)
    │   └── bqva-demo:dataset_2.table_c (⨯)
    └── bqva-demo:dataset_2.view_d (⨯)
        └── bqva-demo:dataset_3.table_d (⨯)

Example: Python library

You can import the library within a Python project to programatically apply permissions to multiple datasets.

from bqva import ViewAnalyzer
from google.cloud import bigquery

client = bigquery.Client()


def auth_views(datasets=[], **kwargs):
    # get all datasets by default if none provided
    if len(datasets) == 0:
        datasets = client.list_datasets(max_results=1)
    for dataset in datasets:
        dataset = client.dataset(dataset)
        tables = client.list_tables(dataset.dataset_id)
        for table in tables:
            if table.table_type == "VIEW":
                view = ViewAnalyzer(
                    project_id=table.project,
                    dataset_id=table.dataset_id,
                    view_id=table.table_id,
                )
                view.apply_permissions()
            print(
                f"Authorised view: {table.project}.{table.dataset_id}.{table.table_id}"
            )


auth_views(["dataset_a", "dataset_b"])