def test_fingerprint_chunk_integration(self):
fingerprints = []
for text_chunk in chunk.chunk_text('similarity/tests/test_data/seuss_test_book.txt'):
fingerprints.append(compute_fingerprint.fingerprint_text(text_chunk))
result_dict = {key: 0 for key in constants.CHUNK_MODEL_FINGERPRINT_FIELDS}
for fingerprint in fingerprints:
for key, value in fingerprint.items():
result_dict[key]+=value
self.assertEquals(result_dict, seuss_result_dictionary)
def test_fingerprint_text(self):
empty_result = {key: 0 for key in constants.CHUNK_MODEL_FINGERPRINT_FIELDS}
test_texts = [
({'chunk': ' '},
empty_result),
({'chunk': test_text},
seuss_result_dictionary)
]
for argument_dictionary, result_dictionary in test_texts:
result_list = []
fingerprint_list=[]
fingerprint = compute_fingerprint.fingerprint_text(util.tokenize_sentences(argument_dictionary['chunk']))
for field in constants.CHUNK_MODEL_FINGERPRINT_FIELDS:
result_list.append(result_dictionary[field])
fingerprint_list.append(fingerprint[field])
self.assertEquals(fingerprint_list, result_list)
