feat: Allow skipping validation when constructing workspaces

src/pyhf/pdf.py

@@ -635,14 +635,21 @@ class Model:
  """The main pyhf model class."""
 
  def __init__(
- self, spec, modifier_set=None, batch_size=None, validate=True, **config_kwargs
+ self,
+ spec,
+ modifier_set=None,
+ batch_size=None,
+ validate: bool = True,
+ **config_kwargs,
  ):
  """
  Construct a HistFactory Model.
 
  Args:
  spec (:obj:`jsonable`): The HistFactory JSON specification
- batch_size (:obj:`None` or :obj:`int`): Number of simultaneous (batched) Models to compute.
+ batch_size (:obj:`None` or :obj:`int`): Number of simultaneous (batched)
+ Models to compute.
+ validate (:obj:`bool`): Whether to validate against a JSON schema
  config_kwargs: Possible keyword arguments for the model configuration
 
  Returns:
@@ -657,8 +664,8 @@ def __init__(
  self.schema = config_kwargs.pop('schema', 'model.json')
  self.version = config_kwargs.pop('version', None)
  # run jsonschema validation of input specification against the (provided) schema
- log.info(f"Validating spec against schema: {self.schema:s}")
  if validate:
+ log.info(f"Validating spec against schema: {self.schema:s}")
  utils.validate(self.spec, self.schema, version=self.version)
  # build up our representation of the specification
  poi_name = config_kwargs.pop('poi_name', 'mu')

src/pyhf/workspace.py

@@ -286,16 +286,28 @@ class Workspace(_ChannelSummaryMixin, dict):
 
  valid_joins = ['none', 'outer', 'left outer', 'right outer']
 
- def __init__(self, spec, **config_kwargs):
- """Workspaces hold the model, data and measurements."""
+ def __init__(self, spec, validate: bool = True, **config_kwargs):
+ """
+ Workspaces hold the model, data and measurements.
+
+ Args:
+ spec (:obj:`jsonable`): The HistFactory JSON specification
+ validate (:obj:`bool`): Whether to validate against a JSON schema
+ config_kwargs: Possible keyword arguments for the workspace configuration
+
+ Returns:
+ model (:class:`~pyhf.workspace.Workspace`): The Workspace instance
+
+ """
  spec = copy.deepcopy(spec)
  super().__init__(spec, channels=spec['channels'])
  self.schema = config_kwargs.pop('schema', 'workspace.json')
  self.version = config_kwargs.pop('version', spec.get('version', None))
 
  # run jsonschema validation of input specification against the (provided) schema
- log.info(f"Validating spec against schema: {self.schema}")
- utils.validate(self, self.schema, version=self.version)
+ if validate:
+ log.info(f"Validating spec against schema: {self.schema}")
+ utils.validate(self, self.schema, version=self.version)
 
  self.measurement_names = []
  for measurement in self.get('measurements', []):
@@ -784,14 +796,15 @@ def sorted(cls, workspace):
  return cls(newspec)
 
  @classmethod
- def build(cls, model, data, name='measurement'):
+ def build(cls, model, data, name='measurement', validate: bool = True):
  """
  Build a workspace from model and data.
 
  Args:
  model (~pyhf.pdf.Model): A model to store into a workspace
  data (:obj:`tensor`): A array holding observations to store into a workspace
  name (:obj:`str`): The name of the workspace measurement
+ validate (:obj:`bool`): Whether to validate against a JSON schema
 
  Returns:
  ~pyhf.workspace.Workspace: A new workspace object
@@ -823,4 +836,4 @@ def build(cls, model, data, name='measurement'):
  {'name': k, 'data': list(data[model.config.channel_slices[k]])}
  for k in model.config.channels
  ]
- return cls(workspace)
+ return cls(workspace, validate=validate)

tests/test_workspace.py

@@ -898,3 +898,14 @@ def test_wspace_unexpected_keyword_argument(simplemodels_model_data):
 
  with pytest.raises(pyhf.exceptions.Unsupported):
  pyhf.Workspace(spec, abc=True)
+
+
+def test_workspace_without_validation(mocker, simplemodels_model_data):
+ model, data = simplemodels_model_data
+
+ mocker.patch('pyhf.utils.validate')
+ ws = pyhf.Workspace.build(model, data, validate=False)
+ assert pyhf.utils.validate.called is False
+
+ pyhf.Workspace(dict(ws), validate=False)
+ assert pyhf.utils.validate.called is False