pawprint

pawprint allows you to quickly track events occurring in your application, and analyse them using pandas. For the full API, see the documentation. These are a work in progress.

Write data flexibly

tracker.write(event="server_booted")
tracker.write(event="logged_in", user_id="alice")
tracker.write(event="navigation", user_id="bob", metadata={"to": "dashboard"})
tracker.write(event="invoice", metadata={"details": {"amount": 1214, "from": "Ardbeg"}})
tracker.write(event="invoice", metadata={"details": {"amount": 123, "from": "Lagavulin"}})

Query data intuitively

Read the full dataset.

tracker.read()

    id                    timestamp   user_id           event                                            metadata  
0    1   2017-03-31 15:51:50.590018      None   server_booted                                                None
1    2   2017-03-31 15:51:50.599256     alice       logged_in                                                None
2    3   2017-03-31 15:51:50.610069       bob      navigation                                 {'to': 'dashboard'}
3    4   2017-03-31 15:51:50.620759      None         invoice     {'details': {'from': 'Ardbeg', 'amount': 1214}}
4    5   2017-03-31 15:51:50.629837      None         invoice   {'details': {'from': 'Lagavulin', 'amount': 123}}

List only events where the user was Alice.

tracker.read("event", user_id="alice")

        event
0   logged_in

Query unstructured data to find out who invoiced you and when.

tracker.read("timestamp", "metadata__details__from", event="invoice")

                     timestamp   json_field
0   2017-03-31 15:51:50.620759       Ardbeg
1   2017-03-31 15:51:50.629837    Lagavulin

Perform aggregates over time.

tracker.count("logged_in", resolution="week")

      datetime   count
0   2017-03-27       1

Aggregate JSON subfields.

tracker.sum(event="invoice", field="metadata__details__amount", resolution="year")

      datetime      sum
0   2017-01-01   1337.0

Documentation

For installation, dependencies, API details, and a quickstart, please RTFM !