def test_invalidate_cache(sdk, mocker):
mock_write = mocker.patch('toucan_data_sdk.sdk.ToucanDataSdk.write')
mock_write.return_value = {'df': DF, 'df2': DF2}
# Cache is empty -> fill it
dfs = sdk.dfs
assert sdk.client.sdk.post.call_count == 1
assert isinstance(dfs, dict)
assert 'df' in dfs
assert 'df2' in dfs
assert DF.equals(dfs['df'])
assert DF2.equals(dfs['df2'])
# Invalidate cache
sdk.invalidate_cache()
sdk.client.sdk.post.reset_mock()
dfs = sdk.dfs
assert sdk.client.sdk.post.call_count == 1
assert isinstance(dfs, dict)
assert 'df' in dfs
assert 'df2' in dfs
assert DF.equals(dfs['df'])
assert DF2.equals(dfs['df2'])
def test_read_from_cache(sdk):
with tempfile.TemporaryDirectory() as tmp_dir:
extraction_dir = os.path.join(tmp_dir, sdk.EXTRACTION_CACHE_PATH)
sdk.EXTRACTION_CACHE_PATH = extraction_dir
os.makedirs(extraction_dir)
joblib.dump(DF, os.path.join(extraction_dir, 'a'))
joblib.dump(DF2, os.path.join(extraction_dir, 'b'))
dfs = sdk.read_from_cache()
assert 'a' in dfs
assert 'b' in dfs
assert DF.equals(dfs['a'])
assert DF2.equals(dfs['b'])
dfs = sdk.read_from_cache(['a'])
assert 'a' in dfs
assert 'b' not in dfs
assert DF.equals(dfs['a'])
def test_extract_zip(zip_content):
with tempfile.NamedTemporaryFile() as tmp_file:
tmp_file.write(zip_content)
tmp_file.flush()
tmp_file.seek(0)
dfs = extract_zip(tmp_file.name)
assert 'df' in dfs
assert 'df2' in dfs
assert DF.equals(dfs['df'])
assert DF2.equals(dfs['df2'])
def test_cache(sdk, mocker):
mock_read = mocker.patch('toucan_data_sdk.sdk.ToucanDataSdk.read')
mock_write = mocker.patch('toucan_data_sdk.sdk.ToucanDataSdk.write')
mock_path_exists = mocker.patch('os.path.exists')
mock_is_dir = mocker.patch('os.path.isdir')
# Cache is empty -> fill it
mock_path_exists.return_value = False
mock_write.return_value = {'df': DF, 'df2': DF2}
dfs = sdk.dfs
assert sdk.client.sdk.post.call_count == 1
assert isinstance(dfs, dict)
assert 'df' in dfs
assert 'df2' in dfs
assert DF.equals(dfs['df'])
assert DF2.equals(dfs['df2'])
# Cache is filled, no request to the server
mock_path_exists.return_value = True
mock_is_dir.return_value = True
mock_read.return_value = {'df': DF, 'df2': DF2}
sdk.client.sdk.post.reset_mock()
_ = sdk.dfs
sdk.client.sdk.post.assert_not_called()
assert isinstance(dfs, dict)
assert 'df' in dfs
assert 'df2' in dfs
assert DF.equals(dfs['df'])
assert DF2.equals(dfs['df2'])
mocker.stopall()
def gen_df():
return DF.copy(deep=True)