atd-knack-services

Integration services for ATD's Knack applications.

Design

ATD Knack Services is comprised of a Python library (/services) and scripts (/scripts) which automate the flow of data from ATD's Knack applications to downstream systems.

These utilities are designed to:

• incrementally offload Knack application records and metadata as a JSON documents in a collection of S3 data stores
• incrementally fetch records and publish them to external systems such as Socrata and ArcGIS Online
• lay the groundwork for further integration with a data lake and/or a data warehouse
• be deployed in Airflow or similar task management frameworks

Configuration

S3 Data Store

Data is stored in an S3 bucket (s3://atd-knack-services), with one subdirectory per Knack application per environment. Each app subdirectory contains a subdirectory for each container, which holds invdividual records stored as JSON a file with its id serving as the filename. As such, each store follows the naming pattern s3://atd-knack-servies/<app-name>-<environment>/<container ID>.

Application metadata is also stored as a JSON file at the root of each S3 bucket.

. s3://atd-knack-services
|- data-tracker-prod
|   |-- 2x22pl1f7a63815efqx33p90.json   #  app metadata
|   |-- view_1
|       |-- 5f31673f7a63820015ef4c85.json
|       |-- 5b34fbc85295dx37f1402543.json
|       |-- 5b34fbc85295de37y1402337.json
|       |...

Scripts (/scripts)

Get the most recent successful DAG run

most_recent_dag_run.py is meant to be run as an initial Airflow task which fetches the most recent successful run of itself. The date can then be passed to subsequent tasks as a filter parameter to support incremental record processing.

$ python most_recent_dag_run.py --dag atd_signals_socrata  

CLI arguments

• --dag (str, required): the DAG ID of DAG run to be fetched.

Load App Metadata to S3

Use upload_metadata.py to load an application's metadata to S3.

$ python upload_metaddata.py \
    --app-name data-tracker \
    --env prod \

CLI arguments

• --app-name (str, required): the name of the source Knack application
• --env (str, required): The application environment. Must be prod or dev.

Load Knack Records to S3

Use knack_container_to_s3.py to incrementally load data from a Knack container (an object or view) to an S3 bucket.

$ python knack_container_to_s3.py \
    --app-name data-tracker \
    --container view_197 \
    --env prod \
    --date 1598387119 \

Publish Records to the Open Data Portal

Use upsert_knack_container_to_socrata.py to publish a Knack container to the Open Data Portal (aka, Socrata).

$ python upsert_knack_container_to_socrata.py \
    --app-name data-tracker \
    --container view_197 \
    --env prod \
    --date 1598387119 \

CLI arguments

• --app-name (str, required): the name of the source Knack application
• --container (str, required): the name of the object or view key of the source container
• --env (str, required): The application environment. Must be prod or dev.
• --date (int, required): a POSIX timestamp. only records which were modified at or after this date will be processed.

Services (/services)

The services package contains utilities for fetching and pushing data between Knack applications and AWS S3.

It is designed as a free-standing Python package can be installed with pip:

$ pip install atd-knack-services

and imported as services:

import services

services.s3.upload

Multi-threaded uploading of file-like objects to S3.

services.s3.download

Multi-threaded downloading of file objects from S3.

How To

• Create bucket(s)
• Add Knack app credentials to auth configuration file
• Add container configuration file to /services/config
• Create DAGs

An end-to-end ETL process will involve creating at least three Airflow tasks:

• Load app metadata to S3
• Load Knack records to S3
• Publish Knack records to destination system