class CorpusUtilsTestCase(TestCase):
def setUp(self):
self.corpus = Corpus()
def test_get_file_path(self):
"""
Test that a dotted path is properly converted to a file address.
"""
path = self.corpus.get_file_path('chatterbot.corpus.english')
self.assertIn(os.path.join('chatterbot_corpus', 'data', 'english'),
path)
def test_read_english_corpus(self):
corpus_path = os.path.join(self.corpus.data_directory, 'english',
'conversations.yml')
data = self.corpus.read_corpus(corpus_path)
self.assertIn('conversations', data)
def test_list_english_corpus_files(self):
data_files = self.corpus.list_corpus_files('chatterbot.corpus.english')
self.assertIn('.yml', data_files[0])
def test_load_english_corpus(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.english.greetings')
self.assertEqual(len(corpus), 1)
self.assertIn(['Hi', 'Hello'], corpus[0])
def test_load_english_corpus_categories(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.english.greetings')
self.assertEqual(len(corpus), 1)
self.assertIn('greetings', corpus.categories)
class CorpusFilePathTestCase(TestCase):
def setUp(self):
self.corpus = Corpus()
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 = self.corpus.load_corpus(file_path)
# Remove the test file
if os.path.exists(file_path):
os.remove(file_path)
self.assertEqual(len(corpus), 1)
self.assertEqual(len(corpus[0]), 2)
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 = self.corpus.load_corpus(file_path)
def test_load_corpus_english_greetings(self):
file_path = os.path.join(self.corpus.data_directory, 'english',
'greetings.yml')
corpus = self.corpus.load_corpus(file_path)
self.assertEqual(len(corpus), 1)
def test_load_corpus_english(self):
file_path = os.path.join(self.corpus.data_directory, 'english')
corpus = self.corpus.load_corpus(file_path)
self.assertGreater(len(corpus), 1)
def test_load_corpus_english_trailing_slash(self):
file_path = os.path.join(self.corpus.data_directory, 'english') + '/'
corpus = self.corpus.load_corpus(file_path)
self.assertGreater(len(corpus), 1)
class CorpusFilePathTestCase(TestCase):
def setUp(self):
self.corpus = Corpus()
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 = self.corpus.load_corpus(file_path)
# Remove the test file
if os.path.exists(file_path):
os.remove(file_path)
self.assertEqual(len(corpus), 1)
self.assertEqual(len(corpus[0]), 2)
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 = self.corpus.load_corpus(file_path)
def test_load_corpus_english_greetings(self):
file_path = os.path.join(self.corpus.data_directory, 'english', 'greetings.yml')
corpus = self.corpus.load_corpus(file_path)
self.assertEqual(len(corpus), 1)
def test_load_corpus_english(self):
file_path = os.path.join(self.corpus.data_directory, 'english')
corpus = self.corpus.load_corpus(file_path)
self.assertGreater(len(corpus), 1)
def test_load_corpus_english_trailing_slash(self):
file_path = os.path.join(self.corpus.data_directory, 'english') + '/'
corpus = self.corpus.load_corpus(file_path)
self.assertGreater(len(corpus), 1)
class CorpusUtilsTestCase(TestCase):
"""
This test case is designed to make sure that all
corpus data adheres to a few general rules.
"""
def setUp(self):
self.corpus = Corpus()
def test_character_count(self):
"""
Test that no line in the corpus excedes the
maximum number of characters.
"""
from chatterbot_corpus.corpus import DIALOG_MAXIMUM_CHARACTER_LENGTH
corpora = self.corpus.load_corpus('chatterbot.corpus')
for converstions in corpora:
for conversation in converstions:
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
)
)
class CorpusUtilsTestCase(TestCase):
def setUp(self):
self.corpus = Corpus()
def test_get_file_path(self):
"""
Test that a dotted path is properly converted to a file address.
"""
path = self.corpus.get_file_path('chatterbot.corpus.english')
self.assertIn(
os.path.join('chatterbot_corpus', 'data', 'english'),
path
)
def test_read_english_corpus(self):
corpus_path = os.path.join(
self.corpus.data_directory,
'english', 'conversations.yml'
)
data = self.corpus.read_corpus(corpus_path)
self.assertIn('conversations', data)
def test_list_english_corpus_files(self):
data_files = self.corpus.list_corpus_files('chatterbot.corpus.english')
self.assertIn('.yml', data_files[0])
def test_load_english_corpus(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.english.greetings')
self.assertEqual(len(corpus), 1)
self.assertIn(['Hi', 'Hello'], corpus[0])
def test_load_english_corpus_categories(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.english.greetings')
self.assertEqual(len(corpus), 1)
# Test that each conversation gets labeled with the correct category
for conversation in corpus:
self.assertIn('greetings', conversation.categories)
class CorpusUtilsTestCase(TestCase):
"""
This test case is designed to make sure that all
corpus data adheres to a few general rules.
"""
def setUp(self):
self.corpus = Corpus()
def test_character_count(self):
"""
Test that no line in the corpus excedes the
maximum number of characters.
"""
from chatterbot_corpus.corpus import DIALOG_MAXIMUM_CHARACTER_LENGTH
corpora = self.corpus.load_corpus('chatterbot.corpus')
for converstions in corpora:
for conversation in converstions:
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 setUp(self):
self.corpus = Corpus()
def setUp(self):
self.corpus = Corpus()
class CorpusLoadingTestCase(TestCase):
def setUp(self):
self.corpus = Corpus()
def test_load_corpus_chinese(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.chinese')
self.assertTrue(len(corpus))
def test_load_corpus_traditional_chinese(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.tchinese')
self.assertTrue(len(corpus))
def test_load_corpus_english(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.english')
self.assertTrue(len(corpus))
def test_load_corpus_french(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.french')
self.assertTrue(len(corpus))
def test_load_corpus_german(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.german')
self.assertTrue(len(corpus))
def test_load_corpus_hindi(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.hindi')
self.assertTrue(len(corpus))
def test_load_corpus_indonesia(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.indonesia')
self.assertTrue(len(corpus))
def test_load_corpus_italian(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.italian')
self.assertTrue(len(corpus))
def test_load_corpus_marathi(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.marathi')
self.assertTrue(len(corpus))
def test_load_corpus_portuguese(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.portuguese')
self.assertTrue(len(corpus))
def test_load_corpus_russian(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.russian')
self.assertTrue(len(corpus))
def test_load_corpus_spanish(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.spanish')
self.assertTrue(len(corpus))
def test_load_corpus_telugu(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.telugu')
self.assertTrue(len(corpus))
def test_load_corpus(self):
"""
Test loading the entire corpus of languages.
"""
corpus = self.corpus.load_corpus('chatterbot.corpus')
self.assertTrue(len(corpus))
class CorpusLoadingTestCase(TestCase):
def setUp(self):
self.corpus = Corpus()
def test_load_corpus_chinese(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.chinese')
self.assertTrue(len(corpus))
def test_load_corpus_traditional_chinese(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.tchinese')
self.assertTrue(len(corpus))
def test_load_corpus_english(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.english')
self.assertTrue(len(corpus))
def test_load_corpus_french(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.french')
self.assertTrue(len(corpus))
def test_load_corpus_german(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.german')
self.assertTrue(len(corpus))
def test_load_corpus_hindi(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.hindi')
self.assertTrue(len(corpus))
def test_load_corpus_indonesia(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.indonesia')
self.assertTrue(len(corpus))
def test_load_corpus_italian(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.italian')
self.assertTrue(len(corpus))
def test_load_corpus_marathi(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.marathi')
self.assertTrue(len(corpus))
def test_load_corpus_portuguese(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.portuguese')
self.assertTrue(len(corpus))
def test_load_corpus_russian(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.russian')
self.assertTrue(len(corpus))
def test_load_corpus_spanish(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.spanish')
self.assertTrue(len(corpus))
def test_load_corpus_telugu(self):
corpus = self.corpus.load_corpus('chatterbot.corpus.telugu')
self.assertTrue(len(corpus))
def test_load_corpus(self):
"""
Test loading the entire corpus of languages.
"""
corpus = self.corpus.load_corpus('chatterbot.corpus')
self.assertTrue(len(corpus))