def test_add_adds_to_track_variable_table(target_nested_dict_track,
nested_new_var_spec):
track: Track = target_nested_dict_track
track.add(nested_new_var_spec, VariableId("A"))
new_var: Variable = track[VariableId("A")]
new_var_dict: Dict = new_var.dump()
assert new_var_dict == nested_new_var_spec
def test_correct_variable_spec():
spec: Dict = {
"var1": {
"name": "name1",
"data_type": "Text",
"sort_order": 1
},
"var2": {
"name": "name2",
"data_type": "Integer",
"sort_order": 0,
"metadata": {
"notes": "notes2"
}
}
}
track = Track.build(spec, None, "")
var1 = track[VariableId("var1")]
assert var1.name == "name1"
assert var1.data_type == "Text"
assert var1.sort_order == 1
assert var1.metadata == {}
var2 = track[VariableId("var2")]
assert var2.name == "name2"
assert var2.data_type == "Integer"
assert var2.sort_order == 0
assert var2.metadata == {
"notes": "notes2"
}
def test_add_no_sort_order_change(target_nested_dict_track,
nested_new_var_spec):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
target_var_2: Variable = target_nested_dict_track[VariableId(
"target_var_2")]
assert target_var_2.sort_order == 0
def test_add_with_sources_alters_has_targets_for_source_after(
target_nested_dict_track, nested_new_var_spec):
target_track: Track = target_nested_dict_track
source_track: Track = target_track.source
nested_new_var_spec["sources"] = ["source_var_3"]
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
source_var: Variable = source_track[VariableId("source_var_3")]
assert source_var.has_targets
def test_add_pushes_sort_order_down(target_nested_dict_track,
nested_new_var_spec):
"""If the new variable has a sort order lower than the highest sort order, it pushes down all others at that
level."""
nested_new_var_spec["sort_order"] = 0
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
target_var_2: Variable = target_nested_dict_track[VariableId(
"target_var_2")]
assert target_var_2.sort_order == 1
def test_add_with_parent_alters_children_for_parent_after(
target_nested_dict_track, nested_new_var_spec):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
target_folder_2: Variable = target_nested_dict_track[VariableId(
"target_folder_2")]
expected: Set = {"target_var_2", "A"}
actual: Set = set(map(lambda child: child.var_id,
target_folder_2.children))
assert expected == actual
def test_add_with_sources_alters_source_of_for_source(target_nested_dict_track,
nested_new_var_spec):
target_track: Track = target_nested_dict_track
source_track: Track = target_track.source
nested_new_var_spec["sources"] = ["source_var_1"]
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
source_var: Variable = source_track[VariableId("source_var_1")]
actual: Set = set(source_var.targets())
expected: Set = {"A", "target_var_1"}
assert actual == expected
def test_add_alters_descendants_that(target_nested_dict_track,
nested_new_var_spec):
track: Track = target_nested_dict_track
track.add(nested_new_var_spec, VariableId("A"))
actual: Set[str] = {"A", "target_var_1", "target_var_2"}
expected: Set[str] = set(track.descendants_that(container=-1))
assert actual == expected
def test_add_changes_track_list(target_nested_dict_track, nested_new_var_spec):
track: Track = target_nested_dict_track
track_spec_old: Dict = track.dump()
expected: Dict = copy.deepcopy(track_spec_old)
expected["A"] = nested_new_var_spec
track.add(nested_new_var_spec, VariableId("A"))
actual: Dict = track.dump()
assert actual == expected
def __init__(self, specs: Dict, source: Optional["Track"], name: str):
"""Do not call directly; use Track.build()."""
self._variables: Dict["VariableId", "Variable"] = {}
self.name = name
self.source = source
self.target = None
self.schema: Optional["Schema"] = None
self._cache: Dict[str, Any] = {}
if source:
source.target = self
logging.info("Building variables for track '%s'." % name)
n: int = 0
for variable_id, variable_data in specs.items():
if variable_id == '':
# Invalid var id
raise ValueError
logging.debug('Building variable "%s".' % variable_id)
variable: "Variable" = self.build_variable(variable_data,
variable_id)
self._variables[VariableId(variable_id)] = variable
n += 1
if n % 100 == 0:
logging.info("Built %i variables." % n)
logging.info('Finished building all %i variables for track "%s".' %
(n, name))
logging.info(
'Performing post-load validation on variables for track "%s".' %
name)
n = 0
validation_errors: ListType[Tuple[Variable, Exception]] = []
for variable in self.values():
try:
Validator.validate(variable, init=True)
except Exception as ex:
validation_errors.append((variable, ex))
n += 1
if n % 100 == 0:
if len(validation_errors) == 0:
logging.info("Validated %i variables.", n)
else:
logging.info("Validated %i variables (%i invalid).", n,
len(validation_errors))
if len(validation_errors) == 0:
logging.info('All variables valid "%s".' % name)
else:
logging.error(
'%i variables valid, %i invalid "%s".' %
(n - len(validation_errors), len(validation_errors), name))
raise ValidationError(validation_errors)
def __init__(self, specs: Dict, source: Optional["Track"], name: str):
"""Do not call directly; use Track.build()."""
self._variables: Dict["VariableId", "Variable"] = {}
self.name = name
self.source = source
self.target = None
self.schema: Optional["Schema"] = None
self._cache: Dict[str, Any] = {}
if source:
source.target = self
logging.info("Building variables for track '%s'." % name)
n: int = 0
for variable_id, variable_data in specs.items():
if variable_id == '':
# Invalid var id
raise ValueError
logging.debug('Building variable "%s".' % variable_id)
variable: "Variable" = self.build_variable(variable_data, variable_id)
self._variables[VariableId(variable_id)] = variable
n += 1
if n % 100 == 0:
logging.info("Built %i variables." % n)
logging.info('Finished building all %i variables for track "%s".' % (n, name))
# we only validate after the whole thing is built to be able to
# accurately compute siblings, parents and children
self.invalidate_variables_cache()
n = 0
if name.startswith("nonprofit_origin"):
logging.warning("Skipping validation for nonprofit origin. THIS IS DANGEROUS DEBUG LOGIC--REMOVE LATER.")
else:
logging.info('Performing post-load validation on variables for track "%s".' % name)
for variable in self.values():
Validator.validate(variable, init=True)
n += 1
if n % 100 == 0:
logging.info("Validated %i variables." % n)
logging.info('All variables valid "%s".' % name)
def new_var_id(self) -> VariableId:
"""If no ID is supplied, use <stage name>_<temporal|invarant>_<n+1>,
where n is the number of variables."""
# Missing the temporal/immutable part for now
return VariableId('{}_{}'.format(self.name, len(self._variables) + 1))
def do_source_swap(track: Track) -> Variable:
var: Variable = track[VariableId("target_var_2")]
var.sources = [VariableId("source_var_3")]
return var
def test_alter_source_changes_sources_list(target_nested_dict_track):
track: Track = target_nested_dict_track
var: Variable = track[VariableId("target_var_2")]
var.sources = ["source_var_2", "source_var_3"]
assert var.sources == ["source_var_2", "source_var_3"]
def test_add_non_container_parent_raises(target_nested_dict_track,
nested_new_var_spec):
nested_new_var_spec["parent"] = "target_var_1"
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
def test_alter_source_raises(target_nested_dict_track):
track: Track = target_nested_dict_track
var: Variable = track[VariableId("target_var_2")]
with pytest.raises(AttributeError):
var.sources = ["source_var_2", "source_var_3"]
def test_add_empty_var_id_raises(target_nested_dict_track,
nested_new_var_spec):
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId(""))
def test_add_invalid_source_raises(target_nested_dict_track,
nested_new_var_spec):
nested_new_var_spec["sources"] = ["non-existent source"]
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
def test_add_illegal_name_raises(illegal_name, target_nested_dict_track,
nested_new_var_spec):
nested_new_var_spec["name"] = illegal_name
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
def test_add_locally_non_unique_name_raises(target_nested_dict_track,
nested_new_var_spec):
nested_new_var_spec["name"] = "second_target"
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
def test_add_track_non_unique_id_raises(target_nested_dict_track,
nested_new_var_spec):
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec,
VariableId("target_var_2"))
def test_add_non_existent_parent_raises(target_nested_dict_track,
nested_new_var_spec):
nested_new_var_spec["parent"] = "non-existent parent"
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))
def test_invalid_sort_order_raises(target_nested_dict_track,
nested_new_var_spec):
nested_new_var_spec["sort_order"] = 5
with pytest.raises(ValueError):
target_nested_dict_track.add(nested_new_var_spec, VariableId("A"))