Sqlite3

Once you have a Union account, install union:

pip install union

Export the following environment variable to build and push images to your own container registry:

        
# replace with your registry name
export IMAGE_SPEC_REGISTRY="<your-container-registry>"

Then run the following commands to run the workflow:

        
    
git clone https://github.com/unionai/unionai-examples
cd unionai-examples
union run --remote tutorials/sentiment_classifier/sentiment_classifier.py main --model distilbert-base-uncased

The source code for this tutorial can be found here {octicon}mark-github.

The following example shows how you can write SQLite3 queries using the SQLite3Task, which is bundled as part of the core flytekit. Since SQL Queries are portable across workflows and Flyte installations (as long as the data exists), this task will always run with a pre-built container, specifically the flytekit container itself. Therefore, users are not required to build a container for SQLite3. You can simply implement the task - register and execute it immediately. In some cases local execution is not possible - e.g. Snowflake. But for SQLlite3 local execution is also supported.

        
    
import pandas
from flytekit import kwtypes, task, workflow
from flytekit.extras.sqlite3.task import SQLite3Config, SQLite3Task

SQLite3 queries in flyte produce a Schema output. The data in this example is taken from here.

        
from flytekit.types.schema import FlyteSchema

EXAMPLE_DB = "https://www.sqlitetutorial.net/wp-content/uploads/2018/03/chinook.zip"

the task is declared as a regular task. Alternatively it can be declared within a workflow just at the point of using it (example later)

        
    
sql_task = SQLite3Task(
    name="sqlite3.sample",
    query_template="select TrackId, Name from tracks limit {{.inputs.limit}}",
    inputs=kwtypes(limit=int),
    output_schema_type=FlyteSchema[kwtypes(TrackId=int, Name=str)],
    task_config=SQLite3Config(uri=EXAMPLE_DB, compressed=True),
)

As described elsewhere FlyteSchemas can be easily be received as pandas Dataframe and Flyte will autoconvert them

        
    
@task
def print_and_count_columns(df: pandas.DataFrame) -> int:
    return len(df[df.columns[0]])

The task can be used normally in the workflow, passing the declared inputs

        
    
@workflow
def wf() -> int:
    return print_and_count_columns(df=sql_task(limit=100))

It can also be executed locally.

        
    
if __name__ == "__main__":
    print(f"Running {__file__} main...")
    print(f"Running main {wf()}")

As mentioned earlier it is possible to also write the SQL Task inline as follows

        
    
@workflow
def query_wf() -> int:
    df = SQLite3Task(
        name="sqlite3.sample_inline",
        query_template="select TrackId, Name from tracks limit {{.inputs.limit}}",
        inputs=kwtypes(limit=int),
        output_schema_type=FlyteSchema[kwtypes(TrackId=int, Name=str)],
        task_config=SQLite3Config(uri=EXAMPLE_DB, compressed=True),
    )(limit=100)
    return print_and_count_columns(df=df)

It can also be executed locally.

        
    
if __name__ == "__main__":
    print(f"Running {__file__} main...")
    print(f"Running main {query_wf()}")