class WhenTestingGetFileChecksumOperation(unittest.TestCase):
def setUp(self):
self.host = 'hostname'
self.port = '00000'
self.user_name = 'username'
self.webhdfs = PyWebHdfsClient(host=self.host, port=self.port,
user_name=self.user_name)
self.path = 'user/hdfs/old_dir'
self.response = MagicMock()
self.file_checksum = {
"FileChecksum": {
"algorithm": "MD5-of-1MD5-of-512CRC32",
"bytes": ("000002000000000000000000729a144ad5e9399f70c9bedd757"
"2e6bf00000000"),
"length": 28
}
}
self.response.json = MagicMock(return_value=self.file_checksum)
@patch.object(Session, 'get')
def test_get_status_throws_exception_for_not_ok(self, mock_get):
self.response.status_code = http_client.BAD_REQUEST
mock_get.return_value = self.response
with self.assertRaises(errors.PyWebHdfsException):
self.webhdfs.get_file_checksum(self.path)
@patch.object(Session, 'get')
def test_get_status_returns_true(self, mock_get):
self.response.status_code = http_client.OK
mock_get.return_value = self.response
result = self.webhdfs.get_file_checksum(self.path)
for key in result:
self.assertEqual(result[key], self.file_checksum[key])
class WhenTestingGetFileChecksumOperation(unittest.TestCase):
def setUp(self):
self.host = 'hostname'
self.port = '00000'
self.user_name = 'username'
self.webhdfs = PyWebHdfsClient(host=self.host, port=self.port,
user_name=self.user_name)
self.response = MagicMock()
self.requests = MagicMock(return_value=self.response)
self.path = 'user/hdfs/old_dir'
self.response = MagicMock()
self.file_checksum = {
"FileChecksum": {
"algorithm": "MD5-of-1MD5-of-512CRC32",
"bytes": ("000002000000000000000000729a144ad5e9399f70c9bedd757"
"2e6bf00000000"),
"length": 28
}
}
self.response.json = MagicMock(return_value=self.file_checksum)
def test_get_status_throws_exception_for_not_ok(self):
self.response.status_code = http_client.BAD_REQUEST
self.requests.get.return_value = self.response
with patch('pywebhdfs.webhdfs.requests', self.requests):
with self.assertRaises(errors.PyWebHdfsException):
self.webhdfs.get_file_checksum(self.path)
def test_get_status_returns_true(self):
self.response.status_code = http_client.OK
self.requests.get.return_value = self.response
with patch('pywebhdfs.webhdfs.requests', self.requests):
result = self.webhdfs.get_file_checksum(self.path)
for key in result:
self.assertEqual(result[key], self.file_checksum[key])
print('making new HDFS directory at: {0}\n'.format(example_dir))
hdfs.make_dir(example_dir)
# get a dictionary of the directory's status
dir_status = hdfs.get_file_dir_status(example_dir)
print(dir_status)
# create a new file on hdfs
print('making new file at: {0}\n'.format(example_file))
hdfs.create_file(example_file, example_data)
file_status = hdfs.get_file_dir_status(example_file)
print(file_status)
# get the checksum for the file
file_checksum = hdfs.get_file_checksum(example_file)
print(file_checksum)
# append to the file created in previous step
print('appending to file at: {0}\n'.format(example_file))
hdfs.append_file(example_file, example_data)
file_status = hdfs.get_file_dir_status(example_file)
print(file_status)
# checksum reflects file changes
file_checksum = hdfs.get_file_checksum(example_file)
print(file_checksum)
# read in the data for the file
print('reading data from file at: {0}\n'.format(example_file))
print('making new HDFS directory at: {0}\n'.format(example_dir))
hdfs.make_dir(example_dir)
# get a dictionary of the directory's status
dir_status = hdfs.get_file_dir_status(example_dir)
print(dir_status)
# create a new file on hdfs
print('making new file at: {0}\n'.format(example_file))
hdfs.create_file(example_file, example_data)
file_status = hdfs.get_file_dir_status(example_file)
print(file_status)
# get the checksum for the file
file_checksum = hdfs.get_file_checksum(example_file)
print(file_checksum)
# append to the file created in previous step
print('appending to file at: {0}\n'.format(example_file))
hdfs.append_file(example_file, example_data)
file_status = hdfs.get_file_dir_status(example_file)
print(file_status)
# checksum reflects file changes
file_checksum = hdfs.get_file_checksum(example_file)
print(file_checksum)
# read in the data for the file
print('reading data from file at: {0}\n'.format(example_file))
