The main code resides in emendis directory. Endpoints are in emendis/main.py.

Inside virtual environment:

pip install -r requirements.txt
pip install -e .

# migrate database
alembic upgrade head

# load some test data
emendis load-csv tests/test_data.csv

# start webserver
uvicorn emendis.main:app --reload

Then navigate to http://localhost:8000/docs to see this beatiful documentation:

See Makefile for some other useful commands. Also there is a cli (emendis).

The export endpoint supports basic filtering by sensor id and timestamp. Output json is somewhat compatible with plotting libraries that I used to work with. Please see swagger for details (http://localhost:8000/docs).

• Import endpoint accepts multiple pubsub messages at once. On a second thought it's probably not how google's pub/sub works.
• Authentication code in auth.py is incomplete: it doesn't check tokens.
• Tests are rudimentary and don't cover everything.
• Database indexes should help with range queries, but I didn't actually check the execution plans. Behaviour on large datasets wasn't tested (but "should just work" (TM)).