def consistency_impl(directory, pixel_totals, median):
# Test that project is consistent, e.g. size of arrays, common masks.
# This test works on all models regardless of elements, etc.
p = QACDProject()
p.set_filename(temp_filename)
p.import_raw_csv_files(
directory) #, [el+' K series.csv' for el in ['Mg', 'Ca', 'Na']])
p.filter(pixel_totals=pixel_totals, median=median)
p.normalise()
p.create_h_factor()
elements = p.elements
files = fnmatch.filter(os.listdir(directory), '* K series.csv')
assert elements == sorted([file_.split()[0] for file_ in files])
for i, element in enumerate(elements):
raw, raw_stats = p.get_raw(element, want_stats=True)
filtered, filtered_stats = p.get_filtered(element, want_stats=True)
normalised, normalised_stats = p.get_normalised(element,
want_stats=True)
if i == 0:
ny, nx = raw.shape
raw_cumulative = raw.copy()
raw_mask = raw.mask
filtered_cumulative = filtered.copy()
filtered_mask = filtered.mask
normalised_cumulative = normalised.copy()
normalised_mask = normalised.mask
else:
raw_cumulative += raw
filtered_cumulative += filtered
normalised_cumulative += normalised
check_masked_array(raw, (ny, nx), np.int32, raw_mask, -1)
check_stats(raw, raw_stats, 'min max mean median std invalid valid')
check_masked_array(filtered, (ny, nx), np.float64, filtered_mask,
np.nan)
check_stats(filtered, filtered_stats,
'min max mean median std invalid valid')
check_masked_array(normalised, (ny, nx), np.float64, normalised_mask,
np.nan)
check_stats(normalised, normalised_stats,
'min max mean median std invalid valid')
# Check raw masked carry across to filtered and normalised.
assert np.all(filtered.data[raw.mask] == np.nan)
assert np.all(normalised.data[raw.mask] == np.nan)
raw_total, raw_total_stats = p.get_raw_total(want_stats=True)
assert np.allclose(raw_total, raw_cumulative)
check_masked_array(raw_total, (ny, nx), np.int32, raw_mask, -1)
check_stats(raw_total, raw_total_stats,
'min max mean median std invalid valid')
filtered_total, filtered_total_stats = p.get_filtered_total(
want_stats=True)
assert np.allclose(filtered_total, filtered_cumulative)
check_masked_array(filtered_total, (ny, nx), np.float64, filtered_mask,
np.nan)
check_stats(filtered_total, filtered_total_stats,
'min max mean median std invalid valid')
assert np.allclose(normalised_cumulative, 1.0)
np.testing.assert_array_equal(normalised.mask, filtered.mask)
# h factor.
h_factor, h_factor_stats = p.get_h_factor(want_stats=True)
np.testing.assert_array_equal(h_factor.mask, filtered.mask)
check_masked_array(h_factor, (ny, nx), np.float64, h_factor.mask, np.nan)
check_stats(h_factor, h_factor_stats,
'min max mean median std invalid valid')
# Ratio.
ratio_elements = ['Ca', 'Mg']
formula = '{0}/({0}+{1})'.format(ratio_elements[0], ratio_elements[1])
for name, correction_model in zip(['ratioA', 'ratioB'], [None, 'garnet']):
p.create_ratio_map(name,
elements=ratio_elements,
correction_model=correction_model)
ratio_entry = p.ratios[name]
assert formula, name == ratio_entry
ratio, ratio_stats = p.get_ratio(name, want_stats=True)
ratio_stats.pop('formula')
ratio_stats.pop('correction_model')
ratio_stats.pop('preset')
check_masked_array(ratio, (ny, nx), np.float64, ratio.mask, np.nan)
check_stats(ratio, ratio_stats,
'min max mean median std invalid valid')
# k-means clustering
k_min = 5
k_max = 10
p.k_means_clustering(k_min, k_max, want_all_elements=True)
for k in range(k_min, k_max + 1):
cluster, cluster_stats = p.get_cluster_indices(k, want_stats=True)
np.testing.assert_array_equal(cluster.mask, filtered.mask)
check_masked_array(cluster, (ny, nx), np.int8, cluster.mask, -1)
check_stats(cluster, cluster_stats, 'min max invalid valid')
# Phase map by thresholding.
elements_and_thresholds = [['Ca', 100, 1000], ['Mg', 0, 10000]]
name = 'two'
p.create_phase_map_by_thresholding(name, elements_and_thresholds)
phase, phase_stats = p.get_phase(name, want_stats=True)
source = phase_stats.pop('source')
assert source == 'thresholding'
phase_stats.pop('elements_and_thresholds')
assert p.phases[name] == ('thresholding', elements_and_thresholds)
mask = phase == False
check_masked_array(phase, (ny, nx), np.bool, mask, False)
check_stats(phase, phase_stats, 'invalid valid')
# Phase map from cluster.
k = k_min
original_values = [0]
name = 'three'
phase_map = cluster == 0
p.create_phase_map_from_cluster(name, phase_map, k, original_values)
phase, phase_stats = p.get_phase(name, want_stats=True)
source = phase_stats.pop('source')
assert source == 'cluster'
phase_stats.pop('k')
phase_stats.pop('original_values')
assert p.phases[name] == ('cluster', k, original_values)
mask = phase == False
check_masked_array(phase, (ny, nx), np.bool, mask, False)
check_stats(phase, phase_stats, 'invalid valid')
# Region.
region = p.calculate_region_ellipse((100, 60), (40, 20))
name = 'ellipse1'
shape_string = 'ellipse'
p.create_region(name, shape_string, region)
region, region_stats = p.get_region(name, want_stats=True)
shape = region_stats.pop('shape')
assert shape == shape_string
mask = region == False
check_masked_array(region, (ny, nx), np.bool, mask, False)
check_stats(region, region_stats, 'invalid valid')
# Display options.
options = p.display_options
# Check defaults.
assert options.colourmap_name == 'rainbow'
assert options.show_ticks_and_labels == True
assert options.overall_title == ''
assert options.show_project_filename == False
assert options.show_date == False
assert options.use_scale == False
assert options.pixel_size == 1.0
assert options.units == '\u03BCm'
assert options.show_scale_bar == True
assert options.scale_bar_location == 'lower left'
assert options.scale_bar_colour == 'black'
assert options.histogram_bin_count == 100
# Set options and check can read them back OK.
options.set_colourmap_name('viridis')
assert options.colourmap_name == 'viridis'
options.set_labels_and_scale(False, 'Title', True, True, True, 23.2, 'mm',
False, 'upper right', 'white')
assert options.show_ticks_and_labels == False
assert options.overall_title == 'Title'
assert options.show_project_filename == True
assert options.show_date == True
assert options.use_scale == True
assert options.pixel_size == 23.2
assert options.units == 'mm'
assert options.show_scale_bar == False
assert options.scale_bar_location == 'upper right'
assert options.scale_bar_colour == 'white'
options.set_labels_and_scale(False, 'Title', True, True, True, 23.2, 'nm',
False, 'upper left', 'white')
assert options.units == 'nm'
assert options.scale_bar_location == 'upper left'
options.set_labels_and_scale(False, 'Title', True, True, True, 23.2, 'nm',
False, 'lower right', 'white')
assert options.scale_bar_location == 'lower right'
with pytest.raises(RuntimeError): # Invalid colourmap name.
options.set_colourmap_name('unknown colourmap')
with pytest.raises(RuntimeError): # Zero pixel size.
options.set_labels_and_scale(False, 'Title', True, True, True, 0.0,
'mm', False, 'lower left', 'white')
with pytest.raises(RuntimeError): # Negative pixel size.
options.set_labels_and_scale(False, 'Title', True, True, True, -0.1,
'mm', False, 'lower left', 'white')
with pytest.raises(RuntimeError): # Invalid units.
options.set_labels_and_scale(False, 'Title', True, True, True, 23.2,
'pm', False, 'lower left', 'white')
with pytest.raises(RuntimeError): # Invalid scale bar location.
options.set_labels_and_scale(False, 'Title', True, True, True, 23.2,
'mm', False, 'centre', 'white')
with pytest.raises(RuntimeError): # Invalid scale bar colour.
options.set_labels_and_scale(False, 'Title', True, True, True, 23.2,
'mm', False, 'lower left', 'purple')
options.set_histogram(True, 20, 0.1, 1000, False)
assert options.histogram_bin_count == 20
# Cleanup.
if os.path.isfile(temp_filename):
os.remove(temp_filename)
elements=['Mg', 'Ca'],
correction_model='pyroxene')
project.create_ratio_map('Anorthite', preset='anorthite')
print(project.get_valid_preset_ratios())
print(project.ratios)
#project.write_debug()
kmin = 5
kmax = 10
project.k_means_clustering(kmin, kmax, want_all_elements=True)
if 1:
# Plot k-means clustering.
for k in range(kmin, kmax + 1):
plt.subplot(2, 3, k + 1 - kmin)
labels, stats = project.get_cluster_indices(k, want_stats=True)
print(k, stats)
if 1:
cmap = cm.get_cmap('rainbow', k) # k discrete levels
else:
# This does not work if k > 10.
colors = ['C{}'.format(i) for i in range(k)]
cmap = mcolors.ListedColormap(colors)
plt.imshow(labels, cmap=cmap, vmin=-0.5, vmax=k - 0.5)
plt.colorbar(ticks=range(0, k))
plt.title('k={}'.format(k))
if 1:
k = kmin
plt.figure()
labels = project.get_cluster_indices(k)
def test_sequence():
# Test that sequence of operations is correct, which is mostly controlled
# by project state.
p = QACDProject()
assert p.state == State.INVALID
# INVALID -> EMPTY.
p.set_filename(temp_filename)
assert p.state == State.EMPTY
assert p.elements is None
with pytest.raises(RuntimeError):
p.get_raw('Mg')
with pytest.raises(RuntimeError):
p.get_raw_total()
# EMPTY -> RAW.
p.import_raw_csv_files(test_dir,
[el + ' K series.csv' for el in ['Na', 'Mg', 'Ca']])
assert p.state == State.RAW
with pytest.raises(RuntimeError):
p.set_filename('')
assert sorted(p.elements) == p.elements
assert p.elements == ['Ca', 'Mg', 'Na']
raw = p.get_raw('Mg')
raw_total = p.get_raw_total()
with pytest.raises(RuntimeError):
p.get_raw('Aa')
with pytest.raises(RuntimeError):
p.get_filtered('Mg')
with pytest.raises(RuntimeError):
p.create_phase_map_by_thresholding('name', [['Ca', 100, 1000]])
# RAW -> FILTERED.
p.filter(pixel_totals=True, median=True)
assert p.state == State.FILTERED
with pytest.raises(RuntimeError):
p.import_raw_csv_files(test_dir)
p.get_filtered('Mg')
p.get_filtered_total()
with pytest.raises(RuntimeError):
p.get_filtered('Aa')
with pytest.raises(RuntimeError):
p.get_normalised('Mg')
# FILTERED -> NORMALISED.
p.normalise()
assert p.state == State.NORMALISED
with pytest.raises(RuntimeError):
p.filter(pixel_totals=True, median=True)
p.get_normalised('Mg')
with pytest.raises(RuntimeError):
p.get_normalised('Aa')
with pytest.raises(RuntimeError):
p.get_h_factor()
k_min = 5
k_max = 8
with pytest.raises(RuntimeError):
p.k_means_clustering(k_min, k_max, want_all_elements=True)
with pytest.raises(RuntimeError):
p.create_ratio_map('some_ratio', ['Mg', 'Ca'])
# NORMALISED -> H_FACTOR.
p.create_h_factor()
assert p.state == State.H_FACTOR
with pytest.raises(RuntimeError):
p.normalise()
p.get_h_factor()
# H_FACTOR -> CLUSTERING.
assert not p.has_cluster()
p.k_means_clustering(k_min, k_max, want_all_elements=True)
assert p.state == State.H_FACTOR
assert p.has_cluster()
with pytest.raises(RuntimeError):
p.create_h_factor()
p.get_cluster_indices(k_min)
p.get_cluster_indices(k_max)
with pytest.raises(RuntimeError):
p.get_cluster_indices(k_min - 1)
with pytest.raises(RuntimeError):
p.get_cluster_indices(k_max + 1)
# Ratio maps - custom.
p.create_ratio_map('some_ratio', elements=['Mg', 'Ca'])
p.get_ratio('some_ratio')
assert len(p.ratios) == 1
assert 'some_ratio' in p.ratios
assert p.ratios['some_ratio'][0] == 'Mg / (Mg+Ca)'
with pytest.raises(RuntimeError):
p.create_ratio_map('some_ratio', elements=['Mg',
'Ca']) # Already exists.
with pytest.raises(RuntimeError):
p.create_ratio_map('ratio_no_such_element', elements=['Mg', 'Aa'])
p.create_ratio_map('corrected_ratio',
elements=['Ca', 'Na'],
correction_model='feldspar')
assert len(p.ratios) == 2
assert 'corrected_ratio' in p.ratios
assert p.ratios['corrected_ratio'][0] == 'Ca / (Ca+Na)'
with pytest.raises(RuntimeError):
p.create_ratio_map('invalid_correction',
elements=['Mg', 'Ca'],
correction_model='no_such_correction_name')
# Ratio maps - presets.
p.create_ratio_map('anorthite blah', preset='anorthite')
p.get_ratio('some_ratio')
assert len(p.ratios) == 3
assert p.ratios['anorthite blah'][0] == 'Ca / (Ca+Na)'
# Phase maps - by thresholding.
p.create_phase_map_by_thresholding('one', [['Ca', 100, 1000]])
assert len(p.phases) == 1
assert 'one' in p.phases
p.create_phase_map_by_thresholding('two',
[['Ca', 100, 1000], ['Mg', 0, 20000]])
assert len(p.phases) == 2
assert 'two' in p.phases
with pytest.raises(RuntimeError):
p.create_phase_map_by_thresholding(
'one', [['Ca', 100, 1000]]) # Already exists.
with pytest.raises(RuntimeError):
p.create_phase_map_by_thresholding('no elements', [[]])
with pytest.raises(RuntimeError):
p.create_phase_map_by_thresholding('no such element', [['Aa', 0, 1]])
# Phase maps - from cluster.
cluster = p.get_cluster_indices(k_min)
phase_map = cluster == 0
p.create_phase_map_from_cluster('from_cluster', phase_map, k_min, [0])
assert len(p.phases) == 3
assert 'from_cluster' in p.phases
with pytest.raises(RuntimeError):
p.create_phase_map_from_cluster('from_cluster', phase_map, k_min,
[0]) # Already exists.
with pytest.raises(RuntimeError):
p.create_phase_map_from_cluster('no original values', phase_map, k_min,
[])
# Regions.
region = p.calculate_region_ellipse((100, 60), (40, 20))
p.create_region('an_ellipse', 'ellipse', region)
assert len(p.regions) == 1
assert 'an_ellipse' in p.regions
with pytest.raises(RuntimeError):
p.create_region('an_ellipse', 'ellipse', region) # Same name
# Cleanup.
if os.path.isfile(temp_filename):
os.remove(temp_filename)