Stem Meter Events

This repo is my solution to Stem's telemetry event processing design challenge (as defined in prompt.md).

Installation and Usage

The repo is a Python3.7 package called meter-events. The package is defined in setup.py, and can be installed using pip install . from the root directory.

The application can then be run as a console script as process_meter_events. Use the -h flag for detailed console script usage.

The default usage is "service mode", where telemetry lines are read one-by-one from stdin. cat fixtures/telemetry.dat | process_meter_events

It can also be used in "file mode" for batch processing: process_meter_events -f fixtures/telemetry.dat

Design

The object-oriented design is straightforward.

sample.py: defines Sample, which encapsulates all functionality that's specific to an individual timeseries sample.

• Parsing / deserialization from input line format
• Serialization to output line format
• Validation of single-sample anomalies: kW, V, I

sample_buffer.py: defines SampleBuffer, which implements a 5-second windowed buffer of Samples.

• Appending new samples / expiring old samples
• Validation of chronological order
• Detection of time anomalies
• Calculating field averages of contiguous Samples

__main__.py: defines core logic of the console script.

• process(file): business logic of reading & processing lines
• main(): wrapper code for console script

Testing

Nose unit tests are defined in test/__init__.py - test_sample(): unit test Sample functionality - test_sample_buffer(): unit test SampleBuffer functionality

Unit tests are executed against the installed package: python3 setup.py test The package is also validated to comply against PEP8 style constraints. This is validated against the installed package: python3 setup.py flake8