def test_valid_templating(self):
template_string = self.template_string.replace('value: "CCO"',
'value: "$my_smiles$"')
template_string = template_string.replace('value: 75',
'value: $conversion$')
df = pd.DataFrame.from_dict({
'$my_smiles$': ['CCO', 'CCCO', 'CCCCO'],
'$conversion$': [75, 50, 30],
})
dataset = templating.generate_dataset(template_string, df)
expected_reactions = []
for smiles, conversion in zip(['CCO', 'CCCO', 'CCCCO'], [75, 50, 30]):
reaction = reaction_pb2.Reaction()
reaction.CopyFrom(self.valid_reaction)
reaction.inputs['in'].components[0].identifiers[0].value = smiles
reaction.outcomes[0].conversion.value = conversion
expected_reactions.append(reaction)
expected_dataset = dataset_pb2.Dataset(reactions=expected_reactions)
self.assertEqual(dataset, expected_dataset)
# Test without "$" in column names
df = pd.DataFrame.from_dict({
'my_smiles': ['CCO', 'CCCO', 'CCCCO'],
'conversion': [75, 50, 30],
})
dataset = templating.generate_dataset(template_string, df)
self.assertEqual(dataset, expected_dataset)
def test_bad_placeholders(self):
template_string = self.template_string.replace('value: "CCO"',
'value: "$my_smiles$"')
template_string = template_string.replace('value: 75',
'value: $conversion$')
df = pd.DataFrame.from_dict({
'$my_smiles$': ['CCO', 'CCCO', 'CCCCO'],
})
with self.assertRaisesRegex(ValueError, r'\$conversion\$ not found'):
templating.generate_dataset(template_string, df)
def enumerate_dataset():
"""Creates a new dataset based on a template reaction and a spreadsheet.
Three pieces of information are expected to be POSTed in a json object:
spreadsheet_name: the original filename of the uploaded spreadsheet.
spreadsheet_data: a base64-encoded string containing the contents of the
spreadsheet.
template_string: a string containing a text-formatted Reaction proto,
i.e., the contents of a pbtxt file.
A new dataset is created from the template and spreadsheet using
ord_schema.templating.generate_dataset.
"""
try:
data = flask.request.get_json(force=True)
basename, suffix = os.path.splitext(data['spreadsheet_name'])
if data['spreadsheet_data'].startswith('data:'):
# Remove the data URL prefix; see
# https://developer.mozilla.org/en-US/docs/Web/API/FileReader/readAsDataURL.
match = re.fullmatch('data:.*?;base64,(.*)',
data['spreadsheet_data'])
spreadsheet_data = match.group(1)
else:
spreadsheet_data = data['spreadsheet_data']
spreadsheet_data = io.BytesIO(base64.b64decode(spreadsheet_data))
dataframe = templating.read_spreadsheet(spreadsheet_data,
suffix=suffix)
dataset = templating.generate_dataset(data['template_string'],
dataframe,
validate=False)
put_dataset(f'{basename}_dataset', dataset)
return 'ok'
except Exception as error: # pylint: disable=broad-except
flask.abort(flask.make_response(str(error), 406))
def test_invalid_templating(self):
template_string = self.template_string.replace('value: "CCO"',
'value: "$my_smiles$"')
template_string = template_string.replace('precision: 99',
'precision: $precision$')
df = pd.DataFrame.from_dict({
'$my_smiles$': ['CCO', 'CCCO', 'CCCCO'],
'$precision$': [75, 50, -5],
})
expected_reactions = []
for smiles, precision in zip(['CCO', 'CCCO', 'CCCCO'], [75, 50, -5]):
reaction = reaction_pb2.Reaction()
reaction.CopyFrom(self.valid_reaction)
reaction.inputs['in'].components[0].identifiers[0].value = smiles
reaction.outcomes[0].conversion.precision = precision
expected_reactions.append(reaction)
expected_dataset = dataset_pb2.Dataset(reactions=expected_reactions)
with self.assertRaisesRegex(ValueError,
'Enumerated Reaction is not valid'):
templating.generate_dataset(template_string, df)
dataset = templating.generate_dataset(template_string,
df,
validate=False)
self.assertEqual(dataset, expected_dataset)
def main(argv):
del argv # Only used by app.run().
with open(FLAGS.template) as f:
template_string = f.read()
df = templating.read_spreadsheet(FLAGS.spreadsheet)
logging.info('generating new Dataset from %s and %s', FLAGS.template,
FLAGS.spreadsheet)
dataset = templating.generate_dataset(template_string,
df,
validate=FLAGS.validate)
if FLAGS.output:
output_filename = FLAGS.output
else:
basename, _ = os.path.splitext(FLAGS.spreadsheet)
output_filename = os.path.join(f'{basename}_dataset.pbtxt')
logging.info('writing new Dataset to %s', output_filename)
message_helpers.write_message(dataset, output_filename)
def test_missing_values(self):
# pylint: disable=too-many-locals
# Build a template reaction.
reaction = reaction_pb2.Reaction()
input1 = reaction.inputs['one']
input1_component1 = input1.components.add()
input1_component1.identifiers.add(value='CCO', type='SMILES')
input1_component1.amount.mass.value = 1.2
input1_component1.amount.mass.units = reaction_pb2.Mass.GRAM
input1_component2 = input1.components.add()
input1_component2.is_limiting = True
input1_component2.identifiers.add(value='c1ccccc1', type='SMILES')
input1_component2.amount.volume.value = 3.4
input1_component2.amount.volume.units = reaction_pb2.Volume.LITER
input2 = reaction.inputs['two']
input2_component1 = input2.components.add()
input2_component1.identifiers.add(value='COO', type='SMILES')
input2_component1.amount.mass.value = 5.6
input2_component1.amount.mass.units = reaction_pb2.Mass.GRAM
outcome = reaction.outcomes.add()
outcome.conversion.value = 75
template_string = text_format.MessageToString(reaction)
template_string = template_string.replace('value: "CCO"',
'value: "$smiles$"')
template_string = template_string.replace('value: 5.6', 'value: $mass$')
# Build a spreadsheet and test for proper edits.
filename = os.path.join(self.test_subdirectory, 'missing.csv')
with open(filename, 'w') as f:
f.write('smiles,mass\n')
f.write('CN,\n') # Missing mass.
f.write(',1.5\n') # Missing SMILES.
df = pd.read_csv(filename)
dataset = templating.generate_dataset(template_string, df)
expected_dataset = dataset_pb2.Dataset()
reaction1 = expected_dataset.reactions.add()
reaction1.CopyFrom(reaction)
reaction1.inputs['one'].components[0].identifiers[0].value = 'CN'
del reaction1.inputs['two'] # No components after amount removal.
reaction2 = expected_dataset.reactions.add()
reaction2.CopyFrom(reaction)
del reaction2.inputs['one'].components[0] # No indentifiers.
reaction2.inputs['two'].components[0].amount.mass.value = 1.5
self.assertEqual(dataset, expected_dataset)
