def test_time_box_date_io(self):
file_name = 'date_io.npb'
tb = example_time_box(file_name)
tb._date_differentials = None
tb._dates = np.array([
np.datetime64('2018-01-01T00:00', 's'),
np.datetime64('2018-01-02T12:00', 's'),
np.datetime64('2018-01-03T05:00', 's'),
np.datetime64('2018-01-05T00:00', 's')
])
tb.write()
self.assertEqual(3, tb._date_differentials.size)
self.assertEqual(np.uint8, tb._date_differentials.dtype)
self.assertEqual(HOURS, tb._date_differential_units)
self.assertEqual(24 + 12, tb._date_differentials[0])
self.assertEqual(12 + 5, tb._date_differentials[1])
self.assertEqual(19 + 24, tb._date_differentials[2])
tb_new = TimeBox(file_name)
tb_new.read()
self.assertEqual(1, tb_new._timebox_version)
self.assertFalse(tb_new._tag_names_are_strings)
self.assertTrue(tb_new._date_differentials_stored)
self.assertEqual(1, tb_new._bytes_per_date_differential)
self.assertEqual(HOURS, tb_new._date_differential_units)
self.assertEqual(np.uint8, tb_new._date_differentials.dtype)
self.assertEqual(3, tb_new._date_differentials.size)
self.assertEqual(24 + 12, tb_new._date_differentials[0])
self.assertEqual(12 + 5, tb_new._date_differentials[1])
self.assertEqual(19 + 24, tb_new._date_differentials[2])
os.remove(file_name)
return
def test_unpack_options(self):
# we are looking at individual bits.
# so far only 2 options, so we'll test integer
# values 0, 1, 2, and 3
# 0000 0000
# 0000 0001
# 0000 0010
# 0000 0011
tb = TimeBox('')
tb._unpack_options(0)
self.assertFalse(tb._tag_names_are_strings)
self.assertFalse(tb._date_differentials_stored)
tb._unpack_options(1)
self.assertTrue(tb._tag_names_are_strings)
self.assertFalse(tb._date_differentials_stored)
tb._unpack_options(2)
self.assertFalse(tb._tag_names_are_strings)
self.assertTrue(tb._date_differentials_stored)
tb._unpack_options(3)
self.assertTrue(tb._tag_names_are_strings)
self.assertTrue(tb._date_differentials_stored)
return
def test_io_pandas(self):
file_name = 'save_pandas.npb'
df = pd.read_csv('timebox/tests/data/test1.csv').set_index('date')
logging.debug('Starting test_io_pandas with df\n{}'.format(df))
tb = TimeBox.save_pandas(df, file_name)
tb_read = TimeBox(file_name)
df2 = tb_read.to_pandas()
self.assertListEqual(list(df.columns.sort_values()),
list(df2.columns.sort_values()))
df = df.sort_index()
# ensure index is same
for i in range(0, len(df.index)):
self.assertEqual(pd.to_datetime(df.index[i]),
pd.to_datetime(df2.index[i]))
# ensure each value is the same
columns = df.columns
for c in columns:
logging.debug('Testing column: {}'.format(c))
logging.debug('Original frame:{}'.format(df[c]))
logging.debug('TB frame:{}'.format(df2[c]))
self.assertEqual(df[c].sum(), df2[c].sum())
os.remove(file_name)
return
def test_read_save_read_save(self):
tb_file_name = 'timebox/tests/data/test_tb_io.npb'
df = pd.read_csv('timebox/tests/data/ETH-USD_integers.csv',
index_col=0)
tb = TimeBox.save_pandas(df, tb_file_name)
tb2 = TimeBox(tb.file_path)
tb2.read()
tb2.write()
tb2.read()
tb.read()
os.remove(tb_file_name)
return
def test_read_write_data_with_tag_name_as_string(self):
file_name = 'test_io.npb'
tb = example_time_box(file_name)
tb._update_required_bytes_for_tag_identifier()
self.assertEqual(15 * 4, tb._num_bytes_for_tag_identifier)
tb.write()
tb_read = TimeBox(file_name)
tb_read.read()
self.assertTrue(tb_read._tag_names_are_strings)
self.assertEqual(3, len(tb_read._tags))
self.assertTrue('tag_0' in tb_read._tags)
self.assertTrue('tag_1' in tb_read._tags)
self.assertTrue('tag_2_long_name' in tb_read._tags)
os.remove(file_name)
return
def test_save_pandas(self):
file_name = 'save_pandas.npb'
df = pd.read_csv('timebox/tests/data/ETH-USD_combined_utc.csv',
index_col=0)
tb = TimeBox.save_pandas(df, file_name)
self.assertTrue(os.path.exists(file_name))
tb_read = TimeBox(file_name)
df2 = tb_read.to_pandas()
df_columns = list(df)
df_columns.sort()
df2_columns = list(df2)
df2_columns.sort()
self.assertListEqual(df_columns, df2_columns)
os.remove(file_name)
return
def example_time_box(file_name: str):
tb = TimeBox(file_name)
tb._timebox_version = 1
tb._tag_names_are_strings = True
tb._date_differentials_stored = False
tb._num_points = 4
tb._tags = {
'tag_0': TimeBoxTag('tag_0', 1, 'u'),
'tag_1': TimeBoxTag('tag_1', 2, 'i'),
'tag_2_long_name': TimeBoxTag('tag_2_long_name', 4, 'f')
}
tb._start_date = np.datetime64('2018-01-01', 's')
tb._seconds_between_points = 3600
tb._tags['tag_0'].data = np.array([1, 2, 3, 4], dtype=np.uint8)
tb._tags['tag_1'].data = np.array([-4, -2, 0, 2000], dtype=np.int16)
tb._tags['tag_2_long_name'].data = np.array([5.2, 0.8, 3.1415, 8],
dtype=np.float32)
return tb
def test_tag_definitions_to_from_bytes_integer(self):
first = TimeBox('')
first._tag_names_are_strings = False
first._tags = example_tag_definitions()
first._update_required_bytes_for_tag_identifier()
tags_bytes_results = TimeBoxTag.tag_list_to_bytes(
[first._tags[t] for t in first._tags],
first._num_bytes_for_tag_identifier, first._tag_names_are_strings)
self.assertEqual(5 * (2 + NUM_BYTES_PER_DEFINITION_WITHOUT_IDENTIFIER),
tags_bytes_results.num_bytes)
self.assertEqual(
813,
np.frombuffer(tags_bytes_results.byte_code, dtype=np.uint8).sum())
second = TimeBox('')
second._num_bytes_for_tag_identifier = 2
second._tag_names_are_strings = False
second._tag_definitions = TimeBoxTag.tag_definitions_from_bytes(
tags_bytes_results.byte_code, second._num_bytes_for_tag_identifier,
second._tag_names_are_strings)
second_tag_bytes_results = TimeBoxTag.tag_list_to_bytes(
[second._tag_definitions[t] for t in second._tag_definitions],
second._num_bytes_for_tag_identifier,
second._tag_names_are_strings)
self.assertEqual(tags_bytes_results.num_bytes,
second_tag_bytes_results.num_bytes)
self.assertEqual(tags_bytes_results.byte_code,
second_tag_bytes_results.byte_code)
second._num_bytes_for_tag_identifier = 4
bad_bytes = TimeBoxTag.tag_list_to_bytes(
[second._tag_definitions[t] for t in second._tag_definitions],
second._num_bytes_for_tag_identifier,
second._tag_names_are_strings)
self.assertNotEqual(second_tag_bytes_results.byte_code,
bad_bytes.byte_code)
second._tag_names_are_strings = True
bad_bytes = TimeBoxTag.tag_list_to_bytes(
[second._tag_definitions[t] for t in second._tag_definitions],
second._num_bytes_for_tag_identifier,
second._tag_names_are_strings)
self.assertNotEqual(second_tag_bytes_results.byte_code,
bad_bytes.byte_code)
return
def test_update_required_bytes(self):
tb = TimeBox('')
tb._tag_names_are_strings = False
tb._tags[0] = TimeBoxTag(0, 1, 'u')
tb._update_required_bytes_for_tag_identifier()
self.assertEqual(1, tb._num_bytes_for_tag_identifier)
tb._tags[256] = TimeBoxTag(256, 1, 'u')
tb._update_required_bytes_for_tag_identifier()
self.assertEqual(2, tb._num_bytes_for_tag_identifier)
tb._tag_names_are_strings = True
tb._tags = {
'a': TimeBoxTag('a', 1, 'u'),
'ab': TimeBoxTag('ab', 1, 'u'),
'abc': TimeBoxTag('abc', 1, 'u')
}
tb._update_required_bytes_for_tag_identifier()
self.assertEqual(12, tb._num_bytes_for_tag_identifier)
return
def test_init(self):
tb = TimeBox('test_file_path.txt')
self.assertEqual('test_file_path.txt', tb.file_path)
return
def test_read_write_file_info_date_deltas(self):
tb = TimeBox('')
tb._timebox_version = 1
tb._tag_names_are_strings = False
tb._date_differentials_stored = True
tb._num_points = 10
tb._tags = example_tag_definitions()
tb._start_date = np.datetime64('2018-01-01', 's')
tb._bytes_per_date_differential = 4
tb._date_differential_units = SECONDS
file_name = 'test_delta.npb'
with open(file_name, 'wb') as f:
self.assertEqual(230, tb._write_file_info(f))
tb_read = TimeBox('')
with open(file_name, 'rb') as f:
self.assertEqual(230, tb_read._read_file_info(f))
self.assertEqual(tb._timebox_version, tb_read._timebox_version)
self.assertEqual(tb._tag_names_are_strings,
tb_read._tag_names_are_strings)
self.assertEqual(tb._date_differentials_stored,
tb_read._date_differentials_stored)
self.assertEqual(tb._num_points, tb_read._num_points)
self.assertEqual(tb._start_date, tb_read._start_date)
self.assertEqual(tb._bytes_per_date_differential,
tb_read._bytes_per_date_differential)
self.assertEqual(tb._date_differential_units,
tb_read._date_differential_units)
for t in tb._tags:
self.assertTrue(t in tb_read._tags)
self.assertEqual(tb._tags[t].identifier,
tb_read._tags[t].identifier)
self.assertEqual(tb._tags[t].type_char, tb_read._tags[t].type_char)
self.assertEqual(tb._tags[t].dtype, tb_read._tags[t].dtype)
self.assertEqual(tb._tags[t].bytes_per_value,
tb_read._tags[t].bytes_per_value)
os.remove(file_name)
return
def test_encode_options(self):
# we are looking at individual bits.
# so far only 2 options, so we'll test integer
# values 0, 1, 2, and 3
# 0000 0000
# 0000 0001
# 0000 0010
# 0000 0011
tb = TimeBox('')
tb._tag_names_are_strings = False
tb._date_differentials_stored = False
self.assertEqual(0, tb._encode_options())
tb._tag_names_are_strings = True
tb._date_differentials_stored = False
self.assertEqual(1, tb._encode_options())
tb._tag_names_are_strings = False
tb._date_differentials_stored = True
self.assertEqual(2, tb._encode_options())
tb._tag_names_are_strings = True
tb._date_differentials_stored = True
self.assertEqual(3, tb._encode_options())
return
def test_pandas_errors(self):
df = pd.DataFrame.from_dict(
{'value_1': np.array([0, 1, 2], dtype=np.uint8)}, orient='columns')
with self.assertRaises(InvalidPandasIndexError):
TimeBox.save_pandas(df, 'not_going_to_save.npb')
return
def example_time_box(file_name: str):
tb = TimeBox(file_name)
tb._timebox_version = 1
tb._tag_names_are_strings = False
tb._date_differentials_stored = True
tb._num_points = 4
tb._tags = {
0: TimeBoxTag(0, 1, 'u'),
1: TimeBoxTag(1, 2, 'i'),
2: TimeBoxTag(2, 4, 'f')
}
tb._start_date = np.datetime64('2018-01-01', 's')
tb._tags[0].data = np.array([1, 2, 3, 4], dtype=np.uint8)
tb._tags[1].data = np.array([-4, -2, 0, 2000], dtype=np.int16)
tb._tags[2].data = np.array([5.2, 0.8, 3.1415, 8], dtype=np.float32)
tb._date_differentials = np.array([1, 1, 1], dtype=np.uint8)
tb._date_differential_units = DAYS
tb._bytes_per_date_differential = 1
return tb
start = time()
df = pd.read_csv(reference_file, index_col=0)
time_to_read_csv = time() - start
copy_of_reference_file = 'timebox/tests/data/ETH-USD_combined_copy.csv'
start = time()
df.to_csv(copy_of_reference_file)
time_to_write_csv = time() - start
write_result('pandas csv', time_to_write_csv, time_to_read_csv,
os.path.getsize(reference_file))
os.remove(copy_of_reference_file)
timebox_file_name = 'timebox/tests/data/test_timebox_io.npb'
start = time()
TimeBox.save_pandas(df, timebox_file_name)
time_to_process_df_and_save_timebox = time() - start
new_tb = TimeBox(timebox_file_name)
start = time()
typed_df = new_tb.to_pandas()
time_to_read_and_convert_to_pandas = time() - start
write_result('file <-> timebox <-> pandas',
time_to_process_df_and_save_timebox,
time_to_read_and_convert_to_pandas,
os.path.getsize(timebox_file_name))
tb = TimeBox(timebox_file_name)
start = time()
tb.read()
