aidp-snowflake

# `aidp-snowflake` — read Snowflake via AIDP `aidataplatform` Use the built-in AIDP Snowflake connector (`type=SNOWFLAKE`). It supports Basic and KeyPair authentication, direct ingestion reads, external-catalog reads, and SQL pushdown. **Do not generate write code for Snowflake in the 4.0 release.** ## When to use - Read from a Snowflake warehouse in an AIDP notebook. - Mentioned: "Snowflake", "warehouse", "private.key.file", or "KeyPair". ## When NOT to use - For a generic JDBC-only source that has no dedicated connector → [`aidp-jdbc-custom`](../aidp-jdbc-custom/SKILL.md). - For a Snowflake write request, explain that the 4.0 AIDP Snowflake connector is read-only. ## Basic authentication read ```python import os from oracle_ai_data_platform_connectors.aidataplatform import ( AIDP_FORMAT, aidataplatform_options, ) opts = aidataplatform_options( type="SNOWFLAKE", host=os.environ["SNOWFLAKE_HOST"], port=443, database_name=os.environ["SNOWFLAKE_DATABASE"], user=os.environ["SNOWFLAKE_USER"], password=os.environ["SNOWFLAKE_PASSWORD"], schema=os.environ["SNOWFLAKE_SCHEMA"], table=os.environ["SNOWFLAKE_TABLE"], extra={ "authentication.method": "Basic", "warehouse": os.environ["SNOWFLAKE_WAREHOUSE"], "role": os.environ.get("SNOWFLAKE_ROLE", ""), }, ) df = spark.read.format(AIDP_FORMAT).options(**opts).load() df.show(5) ``` ## Key-pair authentication read Set one of `private.key.file` or `private.key.content`. Use `private.key.pass.phrase` only for an encrypted key. ```python opts = aidataplatform_options( type="SNOWFLAKE", host=os.environ["SNOWFLAKE_HOST"], port=443, database_name=os.environ["SNOWFLAKE_DATABASE"], user=os.environ["SNOWFLAKE_USER"], schema=os.environ["SNOWFLAKE_SCHEMA"], table=os.environ["SNOWFLAKE_TABLE"], extra={ "authentication.method": "KeyPair", "warehouse": os.environ["SNOWFLAKE_WAREHOUSE"], "role": os.environ.get("SNOWFLAKE_ROLE", ""), "private.key.file": os.environ["SNOWFLAKE_PRIVATE_KEY_FILE"], "private.key.pass.phrase": os.environ.get("SNOWFLAKE_PRIVATE_KEY_PASSPHRASE", ""), }, ) df = spark.read.format(AIDP_FORMAT).options(**opts).load() df.show(5) ``` To supply the key inline, replace `private.key.file` with `private.key.content`; never hard-code a private key in a notebook. ## External-catalog read Create the Snowflake external catalog in **Master Catalogs** first. You can then use a three-part table name or reuse its connection with `catalog.id`. ```python catalog_df = spark.table("<CATALOG_NAME>.<SCHEMA>.<TABLE_NAME>") catalog_df.show(5) catalog_id_df = (spark.read.format(AIDP_FORMAT) .option("catalog.id", "<CATALOG_ID>") .option("schema", "<SCHEMA>") .option("table", "<TABLE_NAME>") .load()) catalog_id_df.show(5) ``` ## Pushdown SQL Use `pushdown.sql` to execute a source query in Snowflake. ```python pushdown_df = (spark.read.format(AIDP_FORMAT) .options(**opts) .option("pushdown.sql", "SELECT * FROM <SCHEMA>.<TABLE_NAME> LIMIT 10") .load()) pushdown_df.show(5) ``` ## Gotchas - **Read-only in 4.0.** Do not use `df.write`, `saveAsTable`, `insertInto`, or `write.mode` with Snowflake. - **Authentication values are case-sensitive.** Use `Basic` or `KeyPair`. - **Warehouse is required.** Set `warehouse` for both authentication modes; `role` is optional. - **Network reachability.** The AIDP cluster needs outbound HTTPS access to the Snowflake account endpoint. ## References - Official sample: [Snowflake notebook](https://github.com/oracle-samples/oracle-aidp-samples/blob/main/data-engineering/ingestion/Read_Only_Ingestion_Connectors/Snowflake.ipynb)