prdt_test

Command line tool invoking BigQuery queries.

Implementation notes

• Tiers are calculated based on favorite_count ranking and the first one can contain up to 3 elements and seconds one up to 7. That means that if several questions have the same favorite_count they can land in different tiers. This behaviour can be easily changed by using RANK() function instead of ROW_NUMBER() in SAVE_FAVORITE_SHARES_IN_TIERS
• For simplicity tables are deleted and created again with each run. It means that in case of data corruption or other problems there wouldn't be a copy of the data to fall back on. In a real system I could use timestamp column or move the old data to different, archive table depending on requirements.
• Since whole logic depends on Google APIs and mocks are not available only the integration test is provided. It just performs dry run of the query jobs to check for errors.