def test_http_pass(self):
"""Does http authentication work correctly"""
responses.add(responses.GET, "http://127.0.0.1/index.html", body='line1\nline2')
_ = smart_open.HttpOpenRead(smart_open.ParseUri("http://127.0.0.1/index.html"), user='******', password='******')
self.assertEquals(len(responses.calls), 1)
actual_request = responses.calls[0].request
self.assert_('Authorization' in actual_request.headers)
self.assert_(actual_request.headers['Authorization'].startswith('Basic '))
def test_https_readline(self):
"""Does https readline method work correctly"""
responses.add(responses.GET,
"https://127.0.0.1/index.html",
body='line1\nline2')
smart_open_object = smart_open.HttpOpenRead(
smart_open.ParseUri("https://127.0.0.1/index.html"))
self.assertEqual(smart_open_object.readline().decode("utf-8"), "line1")
def test_http_gz(self):
"""Can open gzip via http?"""
fpath = os.path.join(CURR_DIR, 'test_data/crlf_at_1k_boundary.warc.gz')
data = open(fpath, 'rb').read()
responses.add(responses.GET, "http://127.0.0.1/data.gz", body=data)
smart_open_object = smart_open.HttpOpenRead(
smart_open.ParseUri("http://127.0.0.1/data.gz"))
m = hashlib.md5(smart_open_object.read())
# decompress the gzip and get the same md5 hash
self.assertEqual(m.hexdigest(), '18473e60f8c7c98d29d65bf805736a0d')
def test_http_bz2(self):
"""Can open bz2 via http?"""
test_string = b'Hello World Compressed.'
test_file = tempfile.NamedTemporaryFile('wb',
suffix='.bz2',
delete=False).name
with smart_open.smart_open(test_file, 'wb') as outfile:
outfile.write(test_string)
with open(test_file, 'rb') as infile:
compressed_data = infile.read()
if os.path.isfile(test_file):
os.unlink(test_file)
responses.add(responses.GET,
"http://127.0.0.1/data.bz2",
body=compressed_data)
smart_open_object = smart_open.HttpOpenRead(
smart_open.ParseUri("http://127.0.0.1/data.bz2"))
# decompress the gzip and get the same md5 hash
self.assertEqual(smart_open_object.read(), test_string)
