def test_get_git_hash_returns_empty_if_git_not_found(mock_subprocess):
mock_subprocess.check_output.side_effect = OSError
with pytest.warns(UserWarning,
match="Error using git - is `git` installed?"):
git_hash = get_git_hash()
assert git_hash == ""
mock_subprocess.check_output.assert_called_with(
["git", "rev-parse", "HEAD"])
def test_get_git_hash_returns_empty_and_emits_warning_if_git_not_found(
mock_subprocess):
mock_subprocess.check_output.side_effect = subprocess.CalledProcessError(
128, cmd=["git", "rev-parse", "HEAD"])
with pytest.warns(
UserWarning,
match=
"Error using git - skipping git hash. Did you call `git init`?",
):
git_hash = get_git_hash()
assert git_hash == ""
def dump(self) -> dict:
"""
Creates a dictionary log of the model, including the serialized model,
path to the saved estimator and scores
Returns
-------
dict
Dictionary containing:
* model_name
* created_time
* versions
- ml_tooling -> version
- sklearn -> version
- pandas -> version
* git_hash
* metrics
- metric -> score
* estimator
- name -> name of pipeline step
- module -> name of module
- classname -> name of class
- params -> dict of params
* estimator_path
"""
from ml_tooling import __version__ as ml_tools_version
from sklearn import __version__ as sklearn_version
from pandas import __version__ as pandas_version
versions = {
"ml_tooling": ml_tools_version,
"sklearn": sklearn_version,
"pandas": pandas_version,
}
data = {
"model_name":
self.name,
"created_time":
datetime.now(),
"versions":
versions,
"git_hash":
get_git_hash(),
"metrics":
self.metrics.to_dict(),
"estimator":
self.estimator,
"estimator_path":
str(self.estimator_path) if self.estimator_path else None,
}
return data
def test_get_git_hash_returns_correctly():
git_hash = get_git_hash()
assert isinstance(git_hash, str)
assert 10 < len(git_hash)