Integration test suite for the snowplow-unified dbt package.
The ./scripts directory contains the following:
-
integration_tests.sh: This tests the standard modules of the snowplow-unified package: -
App error test: runs the app errors optional module without test data to see if it compiles correctly.
-
Late enabled contexts test: runs the package first without contexts enabled (excluding cwv due to the separate dummy dataset) then runs them with enabled. It checkes if a new incremental table that is based on events_this_run runs smoothly or not indicating datatype errors.
-
All contexts except for cwv test: runs the Snowplow unified package 4 times to replicate incremental loading of events, then performs an equality test between the actual vs expected output.
-
Web Vital test: Runs the cwv optional module once and checks the outputs.
-
Web (all web contexts except for cwv) test: runs only the web related contexts and checks if it fails by doing so.
-
Mobile (all mobile contexts) test: runs only the web related contexts and checks if it fails by doing so.
Run the scripts using:
bash integration_tests.sh -d {warehouse}Supported warehouses (should be the same as your target in your profile.yml):
- redshift
- bigquery
- snowflake
- databricks
- postgres
- all (iterates through all supported warehouses)
Good-to-knows:
There are certain exceptions to how different warehouses process data and in places we had to adjust the integration test to work around that. Here's a list of things to keep in mind:
- the non-deterministic nature of row_number() function for Redshift/Postgres/Databricks means that we had to hard-code actuals and expected models for cases where we are testing duplicate rows with exact same results / window
- postgres / redshift needing the array format of : (within sessions_expected)
- bigquery handling of snowplow_utils.timestamp_diff() - absolute_time_in_s changes
- rotating domain_userid per session is hard-coded in the integration test expectations, when run in one batch the user_identifier differs: 2e340eb6e94820ea8369c0174c612260d1cfe9d41f0fe46268994e28d9c0bbf17 0e9ab97b5d9d9a174112df13fe9c44788af3ac9088a8b41e0998d92a8b4b5a4fc
- same with the number of quarantined sessions