Add incremental file writing to diagnostics.py

[AOstenfeld]

Pull Request

Description

Adds ability to write data to the output files at given time intervals while the program is running.

This pull request addresses #145

Checklist

The following items have been checked for this PR:

• [x] Conforms to PEP8 style standards (check with pylint, flake8, or similar)
• [x] Code is documented with docstrings, or docstrings have been updated as appropriate
• [x] New unit test/integration test have been added to check new code
• [x] All existing tests still pass

[AOstenfeld]

Just for reference I've been testing this using the spring.py file in the block-on-spring repo. Add the "interval" keyword to the "Diagnostics" dictionary like so:

"Diagnostics": {

default values come first

        "directory": "output_leapfrog/",
        "output_type": "csv",
        "clock": {"filename": "time.csv"},
        "interval": 2,
        "BlockDiagnostic": [
            {'component': 'momentum', 'filename': 'block_p.csv'},
            {'component': 'position', 'filename': 'block_x.csv'}
        ]
    }

[AOstenfeld]

Alright, I think this is ready for review.

[arichar6]

@AOstenfeld sorry to spring this on you, but I've made some changes in main that affect this PR.

I’ve created a new class, IntervalHandler, which lets you specify a function to call, and an interval at which to call it. This separates out the logic for checking the time step, and should make the code easier to write/maintain. As an example of how to use this class, you can look at my HistoryDiagnostic class.

For the issue that you are working on for this, I’d like to see you use the new IntervalHandler class to extend the HistoryDiagnostic class.

• Add a new, optional, parameter write_interval
• If that parameter is specified, then create a new IntervalHandler whose action is to write the output to file. This will require creating a new method in the HistoryDiagnostic class which writes the data. Something like:

  self._write_data():
  self._traces.squeeze().to_netcdf(self._filename, 'w’)

• Modify the diagnose method to also call perform_action on the new interval handler

[AOstenfeld]

I'm moving this to another PR so that the diagnostics.py file is up to date.