Azure Data Explorer (Kusto) dialect for SQLAlchemy

sqlalchemy-kusto implements a DBAPI (PEP-249) and SQLAlchemy dialect that enables SQL query execution via SQLAlchemy.

Current project includes support for two dialects: SQL dialect and KQL dialect.

SQL dialect

Current implementation has full support for SQL queries. But pay your attention that Kusto implementation of T-SQL has not full coverage; check the list of known issues.

KQL dialect

KQL dialect still in progress. Please, use it on your own risk for now.

Notice that implemented Kusto dialects don't support DDL statements and inserts, deletes, updates.

Installation

pip install sqlalchemy-kusto

Library usage

Using DBAPI

from sqlalchemy_kusto import connect

connection = connect(
        cluster=kusto_url,
        database=database_name,
        msi=False,
        user_msi=None,
        azure_ad_client_id=kusto_client_id,
        azure_ad_client_secret=kusto_client_secret,
        azure_ad_tenant_id=kusto_tenant_id,
        dev_mode=False
)

result = connection.execute(f"select 1").fetchall()

Using SQLAlchemy raw sql

from sqlalchemy.engine import create_engine

engine = create_engine(
    f"kustosql+{kusto_url}/{database_name}?"
    f"msi=False&azure_ad_client_id={kusto_client_id}&"
    f"azure_ad_client_secret={kusto_client_secret}&"
    f"azure_ad_tenant_id={kusto_tenant_id}&"
    f"dev_mode=False"
)
engine.connect()
cursor = engine.execute(f"select top 1")
data_rows = cursor.fetchall()

Using SQLAlchemy

from sqlalchemy import create_engine, MetaData, Table, Column, String, Integer

engine = create_engine(
    f"kustosql+{kusto_url}/{database_name}?"
    f"msi=False&azure_ad_client_id={kusto_client_id}&"
    f"azure_ad_client_secret={kusto_client_secret}&"
    f"azure_ad_tenant_id={kusto_tenant_id}"
)

my_table = Table(
        "MyTable",
        MetaData(),
        Column("Id", Integer),
        Column("Text", String),
)

query = my_table.select().limit(5)

engine.connect()
cursor = engine.execute(query)
print([row for row in cursor])

Using with Apache Superset

Apache Superset starting from version 1.5 also supports Kusto database engine spec.
When connecting to a new data source you may choose a data source type either KustoSQL or KustoKQL depending on the dialect you want to use.

There are following connection string formats:

# KustoSQL
kustosql+https://<CLUSTER_URL>/<DATABASE>?azure_ad_client_id=<CLIENT_ID>&azure_ad_client_secret=<CLIENT_SECRET>&azure_ad_tenant_id=<TENANT_ID>&msi=False

# KustoKQL
kustokql+https://<CLUSTER_URL>/<DATABASE>?azure_ad_client_id=<CLIENT_ID>&azure_ad_client_secret=<CLIENT_SECRET>&azure_ad_tenant_id=<TENANT_ID>&msi=False

Important notice on package version compatibility.
Apache Superset stable releases 1.5 and 2.0 dependent on sqlalchemy==1.3.24. If you want to use sqlalchemy-kusto with these versions you need to install version 1.* of the package.

Current master branch of the apache/superset dependent on sqlalchemy==1.4.36. If you want to use sqlalchemy-kusto with the latest unstable version of apache/superset, you need to install version 2.* of the package.

Running the tests

Unit tests need nothing but the package itself:

make unit

Integration tests run against a local Kusto emulator, so they need neither a cluster nor credentials:

make emulator      # starts the kustainer-linux container on http://localhost:8080
make integration
make emulator-stop

The emulator has no TLS and no authentication, hence the kustosql+http:// URL:

kustosql+http://localhost:8080/NetDefaultDB

To run the same tests against a real cluster instead, set KUSTO_URL, DATABASE and the AZURE_AD_* variables (see .env.sample). When neither the emulator nor a cluster is reachable, the integration tests are skipped rather than failed.

Contributing

Please see the CONTRIBUTING.md for development setup and contributing process guidelines.

---

Issue in Apache Superset repository that inspired current solution.