def test_load_corpus_file_non_existent(self):
"""
Test that a file path can be specified for a corpus.
"""
file_path = './test_corpus.yml'
self.assertFalse(os.path.exists(file_path))
with self.assertRaises(IOError):
corpus.load_corpus(file_path)
def test_load_corpus(self):
"""
Test loading the entire corpus of languages.
"""
corpus_data = corpus.load_corpus('chatterbot.corpus')
self.assertTrue(len(corpus_data))
def test_load_corpus_english_greetings(self):
file_path = os.path.join(corpus.DATA_DIRECTORY, 'english',
'greetings.yml')
corpus_data = corpus.load_corpus(file_path)
self.assertEqual(len(corpus_data), 1)
def test_load_english_corpus_categories(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.english.greetings')
self.assertEqual(len(corpus_data), 1)
# Test that each conversation gets labeled with the correct category
for conversation in corpus_data:
self.assertIn('greetings', conversation.categories)
def test_conversation_format(self):
corpora = corpus.load_corpus('chatterbot.corpus')
for conversations in corpora:
for conversation in conversations:
for text in conversation:
if not isinstance(text, str):
self.fail('"{}" must be a string, not {}.'.format(
str(text), type(text)))
def test_character_count(self):
"""
Test that no line in the corpus exceeds the
maximum number of characters.
"""
from chatterbot_corpus.corpus import DIALOG_MAXIMUM_CHARACTER_LENGTH
corpora = corpus.load_corpus('chatterbot.corpus')
for conversations in corpora:
for conversation in conversations:
for statement in conversation:
if len(statement) > DIALOG_MAXIMUM_CHARACTER_LENGTH:
self.fail(
u'"{}" cannot be longer than {} characters'.format(
statement, DIALOG_MAXIMUM_CHARACTER_LENGTH))
def test_load_corpus_file(self):
"""
Test that a file path can be specified for a corpus.
"""
# Create a file for testing
file_path = './test_corpus.yml'
with io.open(file_path, 'w') as test_corpus:
yml_data = u'\n'.join([
'conversations:', '- - Hello', ' - Hi', '- - Hi', ' - Hello'
])
test_corpus.write(yml_data)
# Load the content from the corpus
corpus_data = corpus.load_corpus(file_path)
# Remove the test file
if os.path.exists(file_path):
os.remove(file_path)
self.assertEqual(len(corpus_data), 1)
self.assertEqual(len(corpus_data[0]), 2)
def test_load_corpus_russian(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.russian')
self.assertTrue(len(corpus_data))
def test_load_corpus_portuguese(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.portuguese')
self.assertTrue(len(corpus_data))
def test_load_corpus_marathi(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.marathi')
self.assertTrue(len(corpus_data))
def test_load_corpus_indonesia(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.indonesia')
self.assertTrue(len(corpus_data))
def test_load_corpus_hindi(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.hindi')
self.assertTrue(len(corpus_data))
def test_load_corpus_french(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.french')
self.assertTrue(len(corpus_data))
def test_load_corpus_english(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.english')
self.assertTrue(len(corpus_data))
def test_load_corpus_traditional_chinese(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.tchinese')
self.assertTrue(len(corpus_data))
def test_load_english_corpus(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.english.greetings')
self.assertEqual(len(corpus_data), 1)
self.assertIn(['Hi', 'Hello'], corpus_data[0])
def test_load_corpus_english_trailing_slash(self):
file_path = os.path.join(corpus.DATA_DIRECTORY, 'english') + '/'
corpus_data = corpus.load_corpus(file_path)
self.assertGreater(len(corpus_data), 1)
def test_load_corpus_telugu(self):
corpus_data = corpus.load_corpus('chatterbot.corpus.telugu')
self.assertTrue(len(corpus_data))
