def test_composite_transducer(self):
'''Transducer should allow composite transducer. A composite transducer is a list of transducers that are applied
in a specific order. Now by making a composite transducer of a->b + aa->c we can get the 'bbbb' output from the length ordering test
'''
data = [{"word": "aaba"}]
transduced_data = [{"word": "bbbb"}]
data_df = DataFrame(data)
transduced_data_df = DataFrame(transduced_data)
t_path = os.path.join(self.test_transducers_path,
'test_composite.json')
transducer = Transducer(
transducers_needed=[{
'source': 'word',
'target': 'word',
'functions': ['test_composite']
}],
transducers_available_dir=self.test_transducers_path)
fns = transducer.load_composite(t_path)
f_in = data[0]['word']
for fn in fns:
f_in = fn(f_in)
self.assertEqual(f_in, 'bbbb')
self.assertTrue(
transduced_data_df.equals(transducer.apply_to_data_frame(data_df)))
def test_intermediate(self):
'''Transductions should not feed. a -> b & b -> c should turn a -> b, not a -> (b) -> c
'''
t_path = os.path.join(self.test_transducers_path,
'test_intermediate_transducer.csv')
transducer = Transducer()
transducer_fn = transducer.create_transducer_function(t_path)
self.assertEqual('bcb', transducer_fn('aba'))
def test_length_ordering(self):
'''Transductions should be reverse ordered by length. Here with a transducer that turns
a->b and aa->c we get 'cbb' instead of 'bbbb'
'''
t_path = os.path.join(self.test_transducers_path,
'test_length_transducer.csv')
transducer = Transducer()
transducer_fn = transducer.create_transducer_function(t_path)
self.assertEqual('cbb', transducer_fn('aaba'))
def transduce(self) -> List[Dict[str, Union[ResourceManifest, pd.DataFrame]]]:
'''Return transduced data objs. Also transduces self.data_objs
'''
transduced_data_objs = []
for data_obj in self.data_objs:
df = data_obj['data']
transducers = []
if "transducers" in data_obj['manifest']:
transducers = data_obj['manifest']['transducers']
transducer = Transducer(transducers)
data_obj['data'] = transducer.apply_to_data_frame(df)
return transduced_data_objs
def test_lambda_transducer(self):
'''Transducer should allow lambda transductions.
'''
data = [{"word": "test"}]
transduced_data = [{"word": "TEST"}]
data_df = DataFrame(data)
transduced_data_df = DataFrame(transduced_data)
transducer = Transducer([{
'source': 'word',
'target': 'word',
'functions': ['lambda x: x.upper()']
}])
transducer_fn = eval(transducer.transducers_needed[0]['functions'][0])
self.assertEqual('TEST', transducer_fn('test'))
self.assertTrue(
transduced_data_df.equals(transducer.apply_to_data_frame(data_df)))
def test_feeding(self):
'''Prevent feeding. I.e. if a rules a->b and b->c exist, aba should produce bcb, not ccc
'''
data = [{"word": "aba"}]
transduced_data = [{"word": "bcb"}]
t_path = os.path.join(self.test_transducers_path, 'test_feeding.csv')
transducer = Transducer([{
'source': 'word',
'target': 'word',
'functions': [t_path]
}])
transducer_fn = transducer.create_transducer_function(t_path)
self.assertEqual(transduced_data[0]["word"],
transducer_fn(data[0]['word']))
def test_incorrect_source_transducer(self):
'''Test incorrect source specified
'''
data = [{"word": "aaa"}]
transduced_data = [{"word": "bbb"}]
data_df = DataFrame(data)
t_path = os.path.join(self.test_transducers_path,
'test_transducer.csv')
transducer = Transducer([{
'source': 'foobar',
'target': 'word',
'functions': [t_path]
}])
transducer_fn = transducer.create_transducer_function(t_path)
with self.assertRaises(TransducerSourceNotFoundError):
transducer.apply_to_data_frame(data_df)
def return_formatted_config(self, form: str = "js") -> Union[str, dict]:
'''Return config for Dictionary as either obj, js, or json.
'''
config_template_object = {
"L1": {
"name": self.config['L1'],
"lettersInLanguage": self.config['alphabet']
},
"L2": {
"name": self.config['L2']
},
"build": datetime.datetime.today().strftime('%Y%m%d%H%M')
}
## Add transducer name that converts search queries
if 'L1_compare_transducer_name' in self.config:
config_template_object['L1']['compare'] = self.config[
'L1_compare_transducer_name']
if "audio_path" in self.config:
config_template_object['audio_path'] = self.config['audio_path']
if "img_path" in self.config:
config_template_object['img_path'] = self.config['img_path']
if form == 'obj':
return config_template_object
elif form == 'js':
## Add adhoc_vars
adhoc_vars = ''
if "adhoc_vars" in self.config:
adhoc_vars = []
for av in self.config['adhoc_vars']:
for k, v in av.items():
adhoc_vars.append(f"var {k} = {v};")
adhoc_vars = "\n".join(adhoc_vars)
## Add transducers
transducers_config = {}
for data_obj in self.data_objs:
transducers = []
if "transducers" in data_obj['manifest']:
transducers = data_obj['manifest']['transducers']
transducer_obj = Transducer(transducers)
configs = transducer_obj.return_mapping_configs()
config_template_object["L1"]["transducers"] = configs
return f"var config = {json.dumps(config_template_object)}" + adhoc_vars
elif form == 'json':
return json.dumps(config_template_object)
def test_javascript_transducer(self):
'''Test javascript transducer
'''
data = [{"word": "aaa"}]
transduced_data = [{"word": "bbb"}]
t_path = os.path.join(self.test_transducers_path,
'test_transducer.csv')
transducer = Transducer([{
'source': 'word',
'target': 'word',
'functions': [t_path]
}])
js = "var mtd = {'transducers': {}};"
js = js + transducer.return_js_template(t_path)
js = js + f"mtd.transducers.test_transducer('aaa');"
self.assertEqual(eval_js(js), transduced_data[0]['word'])
def test_find_path_to_transducers(self):
'''Find path to transducer. Raise NotFoundError if transducer doesn't exist.
'''
transducer = Transducer()
path = transducer.return_transducer_path('norm')
name = transducer.return_transducer_name(path)
self.assertEqual(name, 'norm')
self.assertTrue(os.path.exists(path))
with self.assertRaises(TransducerNotFoundError):
transducer.return_transducer_path('foobar')
with self.assertRaises(TransducerNotFoundError):
transducer.return_transducer_name(
os.path.join(self.transducers_path, 'foobar.csv'))
def test_chained_transducer(self):
'''Transducer should allow chained transductions with lambda transductions
'''
transducer = Transducer([{
'source':
'word',
'target':
'word',
'functions': [
'lambda x: x.upper()', 'lambda x: x.split("-")',
'lambda x: "".join(x)'
]
}])
f_in = 'test-test'
for t in transducer.transducers_needed:
for fn in t['functions']:
fn = eval(fn)
f_in = fn(f_in)
self.assertEqual('TESTTEST', f_in)
def test_normal_transducer(self):
'''Sanity check a->b transducer.
'''
data = [{"word": "aaa"}]
transduced_data = [{"word": "bbb"}]
data_df = DataFrame(data)
transduced_data_df = DataFrame(transduced_data)
t_path = os.path.join(self.test_transducers_path,
'test_transducer.csv')
t_path_json = os.path.join(self.test_transducers_path,
'test_transducer.csv')
transducer = Transducer([{
'source': 'word',
'target': 'word',
'functions': [t_path]
}])
transducer_json = Transducer([{
'source': 'word',
'target': 'word',
'functions': [t_path_json]
}])
transducer_fn = transducer.create_transducer_function(t_path)
self.assertEqual(transduced_data[0]["word"],
transducer_fn(data[0]['word']))
self.assertTrue(
transduced_data_df.equals(transducer.apply_to_data_frame(data_df)))
self.assertTrue(
transduced_data_df.equals(
transducer_json.apply_to_data_frame(data_df)))
