Use Pytest for `test_loaddata.py`

[bobleesj]

Problem

This is the only test file using unittest:

import unittest

import numpy
import pytest

from diffpy.utils.parsers import loadData


##############################################################################
class TestLoadData(unittest.TestCase):
    @pytest.fixture(autouse=True)
    def prepare_fixture(self, datafile):
        self.datafile = datafile

    def test_loadData_default(self):
        """check loadData() with default options"""
        loaddata01 = self.datafile("loaddata01.txt")
        d2c = numpy.array([[3, 31], [4, 32], [5, 33]])
        self.assertRaises(IOError, loadData, "doesnotexist")
        # the default minrows=10 makes it read from the third line
        d = loadData(loaddata01)
        self.assertTrue(numpy.array_equal(d2c, d))
        # the usecols=(0, 1) would make it read from the third line
        d = loadData(loaddata01, minrows=1, usecols=(0, 1))
        self.assertTrue(numpy.array_equal(d2c, d))
        # check the effect of usecols effect
        d = loadData(loaddata01, usecols=(0,))
        self.assertTrue(numpy.array_equal(d2c[:, 0], d))
        d = loadData(loaddata01, usecols=(1,))
        self.assertTrue(numpy.array_equal(d2c[:, 1], d))
        return

    def test_loadData_1column(self):
        """check loading of one-column data."""
        loaddata01 = self.datafile("loaddata01.txt")
        d1c = numpy.arange(1, 6)
        d = loadData(loaddata01, usecols=[0], minrows=1)
        self.assertTrue(numpy.array_equal(d1c, d))
        d = loadData(loaddata01, usecols=[0], minrows=2)
        self.assertTrue(numpy.array_equal(d1c, d))
        d = loadData(loaddata01, usecols=[0], minrows=3)
        self.assertFalse(numpy.array_equal(d1c, d))
        return

    def test_loadData_headers(self):
        """check loadData() with headers options enabled"""
        loaddatawithheaders = self.datafile("loaddatawithheaders.txt")
        hignore = ["# ", "// ", "["]  # ignore lines beginning with these strings
        delimiter = ": "  # what our data should be separated by
        hdata = loadData(loaddatawithheaders, headers=True, hdel=delimiter, hignore=hignore)
        # only fourteen lines of data are formatted properly
        assert len(hdata) == 14
        # check the following are floats
        vfloats = ["wavelength", "qmaxinst", "qmin", "qmax", "bgscale"]
        for name in vfloats:
            assert isinstance(hdata.get(name), float)
        # check the following are NOT floats
        vnfloats = ["composition", "rmax", "rmin", "rstep", "rpoly"]
        for name in vnfloats:
            assert not isinstance(hdata.get(name), float)


# End of class TestRoutines

if __name__ == "__main__":
    unittest.main()

Proposed solution

Since the rest of the test files are using pytest, It would be great to refactor test_loaddata.py using pytest for future maintenance purposes.

[sbillinge]

This is pretty low priority though. Better use of time is writing new and better tests for under-tested code in our base.....

[bobleesj]

noted, we could revisit later. I think it would be a quick fix something like

    def test_loadData_default(datafile):
        """check loadData() with default options"""
        loaddata01 = self.datafile("loaddata01.txt")
        d2c = numpy.array([[3, 31], [4, 32], [5, 33]])
        self.assertRaises(IOError, loadData, "doesnotexist")
        # the default minrows=10 makes it read from the third line
        d = loadData(loaddata01)
        self.assertTrue(numpy.array_equal(d2c, d))
        # the usecols=(0, 1) would make it read from the third line
        d = loadData(loaddata01, minrows=1, usecols=(0, 1))
        self.assertTrue(numpy.array_equal(d2c, d))
        # check the effect of usecols effect
        d = loadData(loaddata01, usecols=(0,))
        self.assertTrue(numpy.array_equal(d2c[:, 0], d))
        d = loadData(loaddata01, usecols=(1,))
        self.assertTrue(numpy.array_equal(d2c[:, 1], d))
        return

    def test_loadData_1column(datafile):
        """check loading of one-column data."""
        loaddata01 = self.datafile("loaddata01.txt")
        d1c = numpy.arange(1, 6)
        d = loadData(loaddata01, usecols=[0], minrows=1)
        self.assertTrue(numpy.array_equal(d1c, d))
        d = loadData(loaddata01, usecols=[0], minrows=2)
        self.assertTrue(numpy.array_equal(d1c, d))
        d = loadData(loaddata01, usecols=[0], minrows=3)
        self.assertFalse(numpy.array_equal(d1c, d))
        return

    def test_loadData_headers(datafile):
        """check loadData() with headers options enabled"""
        loaddatawithheaders = self.datafile("loaddatawithheaders.txt")
        hignore = ["# ", "// ", "["]  # ignore lines beginning with these strings
        delimiter = ": "  # what our data should be separated by
        hdata = loadData(loaddatawithheaders, headers=True, hdel=delimiter, hignore=hignore)
        # only fourteen lines of data are formatted properly
        assert len(hdata) == 14
        # check the following are floats
        vfloats = ["wavelength", "qmaxinst", "qmin", "qmax", "bgscale"]
        for name in vfloats:
            assert isinstance(hdata.get(name), float)
        # check the following are NOT floats
        vnfloats = ["composition", "rmax", "rmin", "rstep", "rpoly"]
        for name in vnfloats:
            assert not isinstance(hdata.get(name), float)

[sbillinge]

haha, if you already did it we can make the change.....but in general we are not in the business of moving unittest tests over toe pytest....be we do want all new tests, or tests that are fixed, to be pytest.