def test_add_multiple_documents_to_corpus(self, mocked_post):
example_document_2 = self.example_document.copy()
example_document_2['url'] = "http://pypln.example.com/documents/2/"
mock1, mock2 = Mock(), Mock()
mock1.status_code = mock2.status_code = 201
mock1.json.return_value = self.example_document
mock2.json.return_value = example_document_2
mocked_post.side_effect = [mock1, mock2]
corpus = Corpus(session=self.session, **self.example_json)
result = corpus.add_documents(["content_1", "content_2"])
documents, errors = result
self.assertEqual(errors, [])
self.assertIs(type(documents[0]), Document)
self.assertIs(type(documents[1]), Document)
for key, value in self.example_document.items():
if key == 'properties':
key = 'properties_url'
self.assertEqual(value, getattr(documents[0], key))
for key, value in example_document_2.items():
if key == 'properties':
key = 'properties_url'
self.assertEqual(value, getattr(documents[1], key))
def test_add_multiple_documents_to_corpus(self, mocked_post):
example_document_2 = self.example_document.copy()
example_document_2['url'] = "http://pypln.example.com/documents/2/"
mock1, mock2 = Mock(), Mock()
mock1.status_code = mock2.status_code = 201
mock1.json.return_value = self.example_document
mock2.json.return_value = example_document_2
mocked_post.side_effect = [mock1, mock2]
corpus = Corpus(session=self.session, **self.example_json)
result = corpus.add_documents(["content_1", "content_2"])
documents, errors = result
self.assertEqual(errors, [])
self.assertIs(type(documents[0]), Document)
self.assertIs(type(documents[1]), Document)
for key, value in self.example_document.items():
if key == 'properties':
key = 'properties_url'
self.assertEqual(value, getattr(documents[0], key))
for key, value in example_document_2.items():
if key == 'properties':
key = 'properties_url'
self.assertEqual(value, getattr(documents[1], key))
def test_adding_document_to_corpus_fails(self, mocked_post):
mocked_post.return_value.status_code = 403
session = requests.Session()
session.auth = ('wrong_user', 'my_precious')
corpus = Corpus(session=session, **self.example_json)
with self.assertRaises(RuntimeError):
corpus.add_document("example.pdf")
def test_compare_corpora_with_different_urls(self):
corpus_1 = Corpus(session=None, **self.example_json)
json_2 = self.example_json.copy()
json_2['url'] = 'http://pypln.example.com.br/corpora/1/'
corpus_2 = Corpus(session=None, **json_2)
self.assertNotEqual(corpus_1, corpus_2)
def test_compare_corpora_with_different_owners(self):
corpus_1 = Corpus(session=None, **self.example_json)
json_2 = self.example_json.copy()
json_2['owner'] = 'admin'
corpus_2 = Corpus(session=None, **json_2)
self.assertNotEqual(corpus_1, corpus_2)
def test_compare_corpora_with_different_creation_dates(self):
corpus_1 = Corpus(session=None, **self.example_json)
json_2 = self.example_json.copy()
json_2['created_at'] = '2013-10-29T17:00:00.000Z'
corpus_2 = Corpus(session=None, **json_2)
self.assertNotEqual(corpus_1, corpus_2)
def test_compare_corpora_with_different_descriptions(self):
corpus_1 = Corpus(session=None, **self.example_json)
json_2 = self.example_json.copy()
json_2['description'] = 'Test Corpus 2'
corpus_2 = Corpus(session=None, **json_2)
self.assertNotEqual(corpus_1, corpus_2)
def test_compare_corpora_with_different_names(self):
corpus_1 = Corpus(session=None, **self.example_json)
json_2 = self.example_json.copy()
json_2['name'] = 'other_name'
corpus_2 = Corpus(session=None, **json_2)
self.assertNotEqual(corpus_1, corpus_2)
def test_adding_document_to_corpus_fails(self, mocked_post):
mocked_post.return_value.status_code = 403
session = requests.Session()
session.auth = ('wrong_user', 'my_precious')
corpus = Corpus(session=session, **self.example_json)
with self.assertRaises(RuntimeError):
corpus.add_document("example.pdf")
def test_instantiating_corpus_from_url_fails(self, mocked_get):
mocked_get.return_value.status_code = 403
mocked_get.return_value.json.return_value = self.example_json
url = self.example_json['url']
with self.assertRaises(RuntimeError):
corpus = Corpus.from_url(url, ('wrong_user', 'my_precious'))
def test_instantiating_corpus_from_url_fails(self, mocked_get):
mocked_get.return_value.status_code = 403
mocked_get.return_value.json.return_value = self.example_json
url = self.example_json['url']
with self.assertRaises(RuntimeError):
corpus = Corpus.from_url(url, ('wrong_user', 'my_precious'))
def test_adding_multiple_documents_returns_an_error(self, mocked_add_document):
results = [self.example_document, RuntimeError]
mocked_add_document.side_effect = results
corpus = Corpus(session=self.session, **self.example_json)
result = corpus.add_documents(["content_1", "content_2"])
expected_calls = [call("content_1"), call("content_2")]
mocked_add_document.assert_has_calls(expected_calls)
expected = ([self.example_document], [("content_2", RuntimeError())])
# How should we test this? The second element of the 'errors' tuple is
# a different instance of RuntimeError, so it doesn't evaluate as equal
# to the one raise in the mock. For now I'll just check everything
# separatedly.
self.assertEqual(result[0], expected[0])
self.assertEqual(result[1][0][0], expected[1][0][0])
self.assertIsInstance(expected[1][0][1], RuntimeError)
def test_adding_multiple_documents_returns_an_error(
self, mocked_add_document):
results = [self.example_document, RuntimeError]
mocked_add_document.side_effect = results
corpus = Corpus(session=self.session, **self.example_json)
result = corpus.add_documents(["content_1", "content_2"])
expected_calls = [call("content_1"), call("content_2")]
mocked_add_document.assert_has_calls(expected_calls)
expected = ([self.example_document], [("content_2", RuntimeError())])
# How should we test this? The second element of the 'errors' tuple is
# a different instance of RuntimeError, so it doesn't evaluate as equal
# to the one raise in the mock. For now I'll just check everything
# separatedly.
self.assertEqual(result[0], expected[0])
self.assertEqual(result[1][0][0], expected[1][0][0])
self.assertIsInstance(expected[1][0][1], RuntimeError)
def test_add_document_to_corpus(self, mocked_post):
mocked_post.return_value.status_code = 201
mocked_post.return_value.json.return_value = self.example_document
corpus = Corpus(session=self.session, **self.example_json)
result = corpus.add_document("content.")
# requests takes either a file-like object or a string. Both should
# work.
files = {"blob": "content."}
data = {"corpus": corpus.url}
mocked_post.assert_called_with("http://pypln.example.com/documents/",
data=data, files=files)
self.assertIs(type(result), Document)
for key, value in self.example_document.items():
if key == 'properties':
key = 'properties_url'
self.assertEqual(value, getattr(result, key))
def test_add_document_to_corpus(self, mocked_post):
mocked_post.return_value.status_code = 201
mocked_post.return_value.json.return_value = self.example_document
corpus = Corpus(session=self.session, **self.example_json)
result = corpus.add_document("content.")
# requests takes either a file-like object or a string. Both should
# work.
files = {"blob": "content."}
data = {"corpus": corpus.url}
mocked_post.assert_called_with("http://pypln.example.com/documents/",
data=data,
files=files)
self.assertIs(type(result), Document)
for key, value in self.example_document.items():
if key == 'properties':
key = 'properties_url'
self.assertEqual(value, getattr(result, key))
def test_instantiate_corpus_from_url(self, mocked_get):
mocked_get.return_value.status_code = 200
mocked_get.return_value.json.return_value = self.example_json
url = self.example_json['url']
corpus = Corpus.from_url(url, self.auth)
mocked_get.assert_called_with(url)
self.assertIsInstance(corpus, Corpus)
for k, v in self.example_json.items():
self.assertEqual(getattr(corpus, k), v)
self.assertEqual(corpus.session.auth, self.auth)
def test_instantiate_corpus_from_url(self, mocked_get):
mocked_get.return_value.status_code = 200
mocked_get.return_value.json.return_value = self.example_json
url = self.example_json['url']
corpus = Corpus.from_url(url, self.auth)
mocked_get.assert_called_with(url)
self.assertIsInstance(corpus, Corpus)
for k, v in self.example_json.items():
self.assertEqual(getattr(corpus, k), v)
self.assertEqual(corpus.session.auth, self.auth)
def test_instantiate_corpus_from_json(self):
corpus = Corpus(session=None, **self.example_json)
for k, v in self.example_json.items():
self.assertEqual(getattr(corpus, k), v)
def test_compare_equal_corpora(self):
corpus_1 = Corpus(session=None, **self.example_json)
corpus_2 = Corpus(session=None, **self.example_json)
self.assertEqual(corpus_1, corpus_2)