def test_model_correllation_2(self):
""" This test is to try combining Constraints from two different DataSets.
This uses the new merge_output keyword option.
"""
mock_file_1 = ["/red_echidna"]
mock_file_2 = ["/blue_echidna"]
input_pattern = "/%colour%_%animal%"
# Create our mock DataSets.
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_file_1
test_ds_1 = PatternDataSet(input_pattern)
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_file_2
test_ds_2 = PatternDataSet(input_pattern)
# A ProcessUnit which merges the Constraint on colour.
the_process = ProcessUnit([test_ds_1, test_ds_2], "/tmp/%animal%_%colour%.file",
"echo", merge_output=["colour"])
output_ds = the_process.execute(simulate=False)
outfiles = [metafile for metafile in output_ds.files]
self.assertEqual(len(outfiles), 1)
self.assertEqual(outfiles[0].full_path, "/tmp/echidna_red-blue.file")
def test_alias_constraints(self):
""" The PatternDataSet should be able to alias Constraints.
This means when asked to get files for the aliased Constraint, it should
return files from another Constraints.
"""
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs'
) as mock_glob:
mock_glob.return_value = self.mock_file_list
pattern_ds = PatternDataSet(self.mock_file_pattern)
# Apply the constraint alias - when asked for hue,
# it will give you colour.
pattern_ds.alias_constraint("colour", "hue")
found_files = pattern_ds.get_files({
'hue': 'red',
'animal': 'kangaroo'
})
self.assertEqual(1, len(found_files))
self.assertEqual("/fake/red_kangaroo.txt",
found_files[0].full_path)
def test_change_mapping(self):
""" Test using multiple input datasets, like if you were calculating a change. """
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
fake_file_1 = '/a/fake/file_1956_red.nc'
fake_file_2 = '/a/fake/file_1981_red.nc'
mock_glob.return_value = [fake_file_1, fake_file_2]
first_pattern_ds = PatternDataSet("/a/fake/file_%date%_%colour%.nc",
set([Constraint('date', ['1956'])]))
second_pattern_ds = PatternDataSet("/a/fake/file_%date%_%colour%.nc",
set([Constraint('date', ['1981'])]))
# Overwrite the valid combinations for these mock datasets.
first_pattern_ds.valid_combinations = set([frozenset([Constraint('colour', ['red']),
Constraint('date', ['1956'])])])
second_pattern_ds.valid_combinations = set([frozenset([Constraint('colour', ['red']),
Constraint('date', ['1981'])])])
the_process_unit = ProcessUnit([first_pattern_ds, second_pattern_ds],
"/a/final/output/file_%start_date%_%end_date%_%colour%.txt",
'echo', map_dict={'start_date': ('date', 0),
'end_date': ('date', 1)})
ds_result = the_process_unit.execute(simulate=True)
outfiles = [file_thing for file_thing in ds_result.files]
self.assertEqual(len(outfiles), 1)
expected_string = self.script_header + "mkdir -p /a/final/output\necho /a/fake/file_1956_red.nc /a/fake/file_1981_red.nc /a/final/output/file_1956_1981_red.txt\n"
self.assertEqual(expected_string, the_process_unit.scheduler.job.to_str())
def test_getfiles(self):
""" Ensure that files are correctly returned using 'get_files'. """
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
# Add the mock fake glob function.
mock_glob.return_value = self.mock_file_list
test_patternds = PatternDataSet(self.mock_file_pattern)
found_files = test_patternds.get_files({'colour': 'green',
'animal': 'echidna'})
expected_files = [MetaFile('green_echidna.txt', '/fake', {})]
self.assertEqual(found_files, expected_files)
mock_glob.assert_called_once_with()
def compute(self):
# Determine file path
patterns = self.get_filepath_patterns()
logger.debug('Using pattern %s' % patterns)
# Create constraints
constraints = [
Constraint(attribute, [values])
for attribute, values in self.constraints.iteritems()
]
# Add user contraints
user_constraints = self.getInputFromPort("added_constraints")
if user_constraints:
constraints.extend(user_constraints)
else:
raise ModuleError(
self,
"No constraints set on DataSet - you can not run a workflow on the entire DataSet"
)
# Create dataset based on file search path and contraints
dataset = PatternDataSet(patterns, constraints)
if not dataset.files:
error_string = "No files found for this dataset with constraints: {}".format(
constraints)
error_string = error_string.replace('],', '],\n')
logger.error(error_string)
raise ModuleError(self, error_string)
self.setResult('out_dataset', dataset)
def setUp(self):
""" Set up some basic DataSets."""
mock_file_pattern_1 = "/fake/%food%_%animal%.file"
mock_file_list_1 = ["/fake/pizza_moose.file", "/fake/pizza_rabbit.file",
"/fake/chocolate_bilby.file","/fake/chocolate_rabbit.file"]
mock_file_pattern_2 = "/fake/%animal%.file"
mock_file_list_2 = ["/fake/moose.file", "/fake/rabbit.file",
"/fake/bilby.file"]
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_file_list_1
self.test_patternds_1 = PatternDataSet(mock_file_pattern_1)
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_file_list_2
self.test_patternds_2 = PatternDataSet(mock_file_pattern_2)
def test_build_glob_patterns(self):
""" When constraints are given in the constructor, restrict the patterns on the fs to glob. """
given_cons = set([Constraint('colour', ['pink', 'green'])])
pattern_ds = PatternDataSet(self.mock_file_pattern, given_cons)
expected_patterns = ['/fake/pink_*.txt', '/fake/green_*.txt']
self.assertItemsEqual(pattern_ds.glob_patterns, expected_patterns)
def test_noconstraints(self):
''' The PatternDataSet should glob the FS to find files. '''
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs'
) as mock_glob:
mock_glob.return_value = self.mock_file_list
test_patternds = PatternDataSet(self.mock_file_pattern)
self.assertEqual(test_patternds.constraints, self.fake_constraints)
# Check that we only try to glob the fs once.
mock_glob.assert_called_once_with()
def test_cons_from_pattern(self):
""" The PatternDataSet should build a complete set of constraints by globbing on the file system."""
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs'
) as mock_glob:
mock_glob.return_value = self.mock_file_list
pattern_ds = PatternDataSet(self.mock_file_pattern)
expected_cons = self.fake_constraints
self.assertEqual(pattern_ds.constraints, self.fake_constraints)
def test_model_correllation(self):
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = self.mock_obs_files
test_obsds = PatternDataSet(self.observational_pattern)
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = self.mock_model_files
test_model_ds = PatternDataSet(self.model_pattern)
output_pattern = "/%variable%_%obs_model%_%model%.nc"
our_process = ProcessUnit([test_obsds, test_model_ds],
output_pattern, "echo")
output = our_process.execute()
all_outs = [thing.full_path for thing in output.files]
good_names = ["/tas_HadISST_BadModel.nc", "/tas_AWAP_BadModel.nc",
"/tas_HadISST_GoodModel.nc", "/tas_AWAP_GoodModel.nc"]
self.assertItemsEqual(good_names, all_outs)
def setUp(self):
""" Makes a mock PatternDataSet. """
self.mock_file_list = ['/a/fake_1/file_1/pattern_1']
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs'
) as mock_glob:
#Add the mock fake glob function.
mock_glob.return_value = self.mock_file_list
self.a_pattern_ds = PatternDataSet('/a/%fake%/%file%/%pattern%')
self.script_header = "#!/bin/sh\nset -e\n\nmodule purge\nexport CWSL_CTOOLS={}\nexport PYTHONPATH=$PYTHONPATH:{}/pythonlib\n"\
.format(configuration.cwsl_ctools_path, configuration.cwsl_ctools_path)
def test_changefile_generation(self):
""" This test works to cover the common case when you want to calculate changes.
For example, comparing two dataset by time.
"""
mock_files_1 = ["/model1_1986_rain", "/model2_1986_rain",
"/model3_1986_rain", "/model4_1986_rain",
"/model1_1986_temp"]
mock_files_2 = ["/model1_2015_rain", "/model2_2015_rain",
"/model3_2015_rain", "/model4_2015_rain",
"/model1_2015_temp"]
input_pattern = "/%model%_%date%_%variable%"
# Create our mock DataSets.
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_files_1
test_ds_1 = PatternDataSet(input_pattern)
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_files_2
test_ds_2 = PatternDataSet(input_pattern)
# A ProcessUnit which merges the Constraint on colour.
the_process = ProcessUnit([test_ds_1, test_ds_2], "/tmp/%model%_%date%_%variable%",
"echo", merge_output=["date"])
output_ds = the_process.execute(simulate=True)
outfile_names = [metafile.full_path for metafile in output_ds.files]
expected_files = ["/tmp/model1_1986-2015_rain",
"/tmp/model2_1986-2015_rain",
"/tmp/model3_1986-2015_rain",
"/tmp/model4_1986-2015_rain",
"/tmp/model1_1986-2015_temp"]
self.assertItemsEqual(expected_files, outfile_names)
def compute(self):
cons_list = [
'model', 'experiment', 'variable', 'season_number', 'region'
]
in_cons = set([
Constraint(cons_name, [self.getInputFromPort(cons_name)])
for cons_name in cons_list if self.getInputFromPort(cons_name)
])
file_pattern = "/g/data/ua6/CAWCR_CVC_processed/staging/users/CWSL/SDM/COD/CMIP5_v2/%model%_%experiment%/%region%/%variable%/season_%season_number%/rawfield_analog_%season_number%"
output_ds = PatternDataSet(file_pattern, in_cons)
self.setResult('out_dataset', output_ds)
def test_alias_constraints(self):
""" The PatternDataSet should be able to alias Constraints.
This means when asked to get files for the aliased Constraint, it should
return files from another Constraints.
"""
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = self.mock_file_list
pattern_ds = PatternDataSet(self.mock_file_pattern)
# Apply the constraint alias - when asked for hue,
# it will give you colour.
pattern_ds.alias_constraint("colour", "hue")
found_files = pattern_ds.get_files({'hue': 'red',
'animal': 'kangaroo'})
self.assertEqual(1, len(found_files))
self.assertEqual("/fake/red_kangaroo.txt",
found_files[0].full_path)
def compute(self):
cons_list = [
'model', 'experiment', 'variable', 'season_number', 'region'
]
in_cons = set([
Constraint(cons_name, [self.getInputFromPort(cons_name)])
for cons_name in cons_list if self.getInputFromPort(cons_name)
])
file_pattern = "/home/548/teb548/cod/CMIP5_v2/%model%_%experiment%/%region%/%variable%/season_%season_number%/rawfield_analog_%season_number%"
output_ds = PatternDataSet(file_pattern, in_cons)
self.setResult('out_dataset', output_ds)
def test_getfiles(self):
""" Ensure that files are correctly returned using 'get_files'. """
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs'
) as mock_glob:
# Add the mock fake glob function.
mock_glob.return_value = self.mock_file_list
test_patternds = PatternDataSet(self.mock_file_pattern)
found_files = test_patternds.get_files({
'colour': 'green',
'animal': 'echidna'
})
expected_files = [
MetaFile('green_echidna.txt', '/fake', {
'colour': 'green',
'animal': 'echidna'
})
]
self.assertItemsEqual(found_files, expected_files)
mock_glob.assert_called_once_with()
def test_model_correllation_3(self):
""" This test is to try combining multiple DataSets, each with many files. """
mock_files_1 = ["/red_echidna", "/blue_echidna", "/green_echidna"]
mock_files_2 = ["/blue_echidna", "/red_echidna", "/green_echidna"]
input_pattern = "/%colour%_%animal%"
# Create our mock DataSets.
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_files_1
test_ds_1 = PatternDataSet(input_pattern)
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_files_2
test_ds_2 = PatternDataSet(input_pattern)
# A ProcessUnit which merges the Constraint on colour.
the_process = ProcessUnit([test_ds_1, test_ds_2], "/tmp/%animal%_%colour%.file",
"echo", merge_output=["colour"])
output_ds = the_process.execute(simulate=True)
outfile_names = [metafile.full_path for metafile in output_ds.files]
expected_outfiles = ["/tmp/echidna_red-red.file",
"/tmp/echidna_red-blue.file",
"/tmp/echidna_red-green.file",
"/tmp/echidna_blue-red.file",
"/tmp/echidna_blue-blue.file",
"/tmp/echidna_blue-green.file",
"/tmp/echidna_green-red.file",
"/tmp/echidna_green-blue.file",
"/tmp/echidna_green-green.file"]
self.assertItemsEqual(expected_outfiles, outfile_names)
def setUp(self):
# This creates a mock pattern dataset that returns a single file.
test_cons = set([Constraint('fake', ['fake_1']),
Constraint('file', ['file_1']),
Constraint('pattern', ['pattern_1'])])
self.a_pattern_ds = PatternDataSet('/a/%fake%/%file%/%pattern%',
constraint_set=test_cons)
# Mock the get_files method - we will only return a single, mock file object.
mock_file = mock.MagicMock()
mock_file.full_path = 'test_file1'
mock_file.__str__.return_value = 'test_file1'
mock_file.all_atts = {"fake": "fake_1",
"file": "file_1",
"pattern": "pattern_1"}
self.a_pattern_ds.get_files = mock.Mock(return_value=[mock_file])
# Create a valid set of contraints for the mock.
self.a_pattern_ds.valid_combinations = set([frozenset(test_cons)])
# Constant header for the job scripts.
self.script_header = "#!/bin/sh\nset -e\n\nmodule purge\nexport CWSL_CTOOLS={}\nexport PYTHONPATH=$PYTHONPATH:{}/pythonlib\n"\
.format(configuration.cwsl_ctools_path, configuration.cwsl_ctools_path)
def test_regex(self):
''' Given an input pattern, the PatternDataSet should create a regular expression. '''
test_patternds = PatternDataSet(self.mock_file_pattern)
self.assertEqual(test_patternds.regex_pattern, self.mock_regex)
class TestArgumentCreator(unittest.TestCase):
""" Tests to ensure that looping and grouping works correctly."""
def setUp(self):
""" Set up some basic DataSets."""
mock_file_pattern_1 = "/fake/%food%_%animal%.file"
mock_file_list_1 = ["/fake/pizza_moose.file", "/fake/pizza_rabbit.file",
"/fake/chocolate_bilby.file","/fake/chocolate_rabbit.file"]
mock_file_pattern_2 = "/fake/%animal%.file"
mock_file_list_2 = ["/fake/moose.file", "/fake/rabbit.file",
"/fake/bilby.file"]
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_file_list_1
self.test_patternds_1 = PatternDataSet(mock_file_pattern_1)
with mock.patch('cwsl.core.pattern_dataset.PatternDataSet.glob_fs') as mock_glob:
mock_glob.return_value = mock_file_list_2
self.test_patternds_2 = PatternDataSet(mock_file_pattern_2)
def test_simple_looping(self):
""" Test that basic one-to-one looping works."""
one_to_one_creator = FileCreator("/output/%animal%.output",
[self.test_patternds_2.get_constraint("animal")])
looper = ArgumentCreator([self.test_patternds_2], one_to_one_creator)
all_outs = []
for thing in looper:
self.assertEqual(len(thing[0]), len(thing[1]))
all_outs.append(thing)
all_files = [thing for thing in one_to_one_creator.files]
self.assertEqual(len(all_outs), len(all_files))
def test_many_looping(self):
""" Test that simple many-to-one looping works."""
many_to_one_creator = FileCreator("/output/%animal%.output",
[self.test_patternds_1.get_constraint("animal")])
looper = ArgumentCreator([self.test_patternds_1], many_to_one_creator)
all_outs = []
for thing in looper:
all_outs.append(thing)
module_logger.debug("All outs are: {}".format(all_outs))
# There are three animals.
self.assertEqual(3, len(all_outs))
def test_multi_model(self):
""" Test for the case when there are groups on Constraints.
This seems to fail when people use FileCreators.
"""
institute_model_pattern = "/fake/%variable%_%model%_%institute%.file"
in_constraints = [Constraint('model', ['model_1', 'model_2']),
Constraint('variable', ['variable_1']),
Constraint('institute', ['institute_1', 'institute_2'])]
test_filecreator = FileCreator(institute_model_pattern, in_constraints)
# Set the valid combinations.
dummy_file_1 = test_filecreator.get_files({'model': 'model_1',
'institute': 'institute_1'},
update=True)
dummy_file_2 = test_filecreator.get_files({'model': 'model_2',
'institute': 'institute_2'},
update=True)
# Now create a FileCreator to use as output.
output_pattern = "/an/output/fake/%variable%_%model%_%institute%.file"
out_constraints = [Constraint('model', ['model_1', 'model_2']),
Constraint('variable', ['variable_1']),
Constraint('institute', ['institute_1', 'institute_2'])]
test_output_filecreator = FileCreator(output_pattern, out_constraints)
print("Valid input combinations are: {0}".format(test_filecreator.valid_combinations))
self.assertEqual(2, len(test_filecreator.valid_hashes))
test_argument_creator = ArgumentCreator([test_filecreator],
test_output_filecreator)
outputs = [combination for combination in test_argument_creator]
print("Output is: {0}".format(outputs))
# There should only be two outputs - not 4!
self.assertEqual(len(outputs), 2)
def test_two_inputs(self):
""" Test that the ArgumentCreator works with multiple input datasets."""
multi_ds_creator = FileCreator("/output/%animal%.output",
[self.test_patternds_1.get_constraint("animal")])
looper = ArgumentCreator([self.test_patternds_1, self.test_patternds_2],
multi_ds_creator)
all_outs = []
for thing in looper:
self.assertGreaterEqual(len(thing[0]), len(thing[1]))
all_outs.append(thing)
# There are three animals.
self.assertEqual(3, len(all_outs))
print(all_outs)
# The order is moose, then rabbit
# Moose: 2 ins, 1 out.
module_logger.debug("All outs[0]: {}".format(all_outs[0]))
self.assertEqual(len(all_outs[0][0]), 2)
self.assertEqual(len(all_outs[0][1]), 1)
# Rabbit: 3 in, 1 out
module_logger.debug("All outs[1]: {}".format(all_outs[1]))
self.assertEqual(len(all_outs[1][0]), 3)
self.assertEqual(len(all_outs[1][1]), 1)
