fakesnow

fakesnow lets you run and test code against Snowflake without a real Snowflake account, either by patching the Snowflake Connector for Python in the current process or by running fakesnow as a standalone HTTP server that any language's Snowflake-compatible client can connect to. The in-process approach patches the connector's connect function and its pandas helper automatically, and can be invoked as a wrapper around a script or pytest, or from inside your own code as a context manager; modules using from-import syntax must be patched explicitly by name.

By default, fake databases are in-memory and lost when the process ends, but a database path can be given to persist them between runs. The server mode is needed when patching does not apply, such as in a spawned subprocess or from a non-Python client, and can be started from the command line, from within a Python program, or via a published Docker image, listening on a configurable or random port.