def validate_via_metapype(node):
errs = []
try:
validate.tree(node, errs)
except Exception as e:
print(f'validate_via_metapype: node={node.name} exception={e}')
return errs
def parse_xml_file(filename, filepath):
log_info(f"parse_xml_file: {filename}")
eml_version = ''
with open(filepath, "r") as f:
lines = f.readlines()
for line in lines:
if 'xmlns:eml' in line:
eml_version = line[-7:-2]
break
xml = "".join(lines)
eml_node = metapype_io.from_xml(xml,
clean=True,
literals=['literalLayout', 'markdown'])
assert isinstance(eml_node, Node) # TODO: error-handling
pruned_nodes = set()
errs = []
unknown_nodes = None
attr_errs = None
child_errs = None
other_errs = None
try:
validate.tree(eml_node, errs)
validate.tree(eml_node)
print(f'{filename} - {eml_version}: valid')
log_info(f'{filename} - {eml_version}: valid')
# return None
except Exception as e:
print(f'{filename} - {eml_version}: ', end='')
try:
pruned = validate.prune(eml_node, strict=False)
for x, _ in pruned:
pruned_nodes.add(x.name)
pruned_nodes = sorted(pruned_nodes)
unknown_nodes, attr_errs, child_errs, other_errs = extract_eml_errors(
errs)
if unknown_nodes:
print(f"Unknown nodes: {unknown_nodes}")
log_info(f"Unknown nodes: {unknown_nodes}")
if attr_errs:
print(f"Attribute errors: {attr_errs}")
log_info(f"Attribute errors: {attr_errs}")
if child_errs:
print(f"Child errors: {child_errs}")
log_info(f"Child errors: {child_errs}")
if other_errs:
print(f"Other errors: {other_errs}")
log_info(f"Other errors: {other_errs}")
if pruned:
print(f"Pruned nodes: {pruned_nodes}")
log_info(f"Pruned nodes: {pruned_nodes}")
else:
err_set = set()
for err in errs:
err_set.add(err[1])
print('***', sorted(err_set))
except Exception as e:
print(f'validate.prune FAILED: {e}')
log_info(f'validate.prune FAILED: {e}')
return eml_node, unknown_nodes, attr_errs, child_errs, other_errs, pruned_nodes
def test_validate_annotation():
annotation = Node(names.ANNOTATION)
property_uri = Node(names.PROPERTYURI)
property_uri.content = "http://purl.obolibrary.org/obo/IAO_0000136"
property_uri.add_attribute("label", "some property label")
annotation.add_child(property_uri)
value_uri = Node(names.VALUEURI)
value_uri.content = "http://purl.obolibrary.org/obo/IAO_0000136"
value_uri.add_attribute("label", "some value label")
annotation.add_child(value_uri)
validate.tree(annotation)
def validate_via_metapype(node):
errs = []
try:
start = time.perf_counter()
validate.tree(node, errs)
end = time.perf_counter()
elapsed = end - start
# if elapsed > 0.05:
# print(f"validate: {node.name} {elapsed}")
except Exception as e:
print(f'validate_via_metapype: node={node.name} exception={e}')
return errs
def test_validate_bad_tree(node):
dataset: Node = node.find_child(names.DATASET)
creator: Node = dataset.find_child(names.CREATOR)
dataset.remove_child(creator)
errs = list()
assert validate.tree(node, errs) is None
assert len(errs) != 0
def test_prune():
if "TEST_DATA" in os.environ:
test_data = os.environ["TEST_DATA"]
else:
test_data = tests.test_data_path
with open(f"{test_data}/eml.xml", "r") as f:
xml = "".join(f.readlines())
eml = metapype_io.from_xml(xml)
pruned = validate.prune(eml, strict=True)
for node in pruned:
print(f"pruned: {node[0].name} - {node[1]}")
errs = []
validate.tree(eml, errs)
for err in errs:
print(err)
def test_missing_numerical_unit():
unit = Node(names.UNIT, parent=None)
r = rule.get_rule(names.UNIT)
with pytest.raises(MetapypeRuleError):
r.validate_rule(unit)
# Check error
errs = []
validate.tree(unit, errs)
assert len(errs) == 1
err_code, msg, node, *args = errs[0]
assert err_code == ValidationError.MIN_CHOICE_UNMET
assert args[0] == 'unit'
# With a customUnit, it should be ok
custom_unit = Node(names.CUSTOMUNIT, parent=unit)
custom_unit.content = 'bushels per parsec'
unit.add_child(custom_unit)
validate.tree(unit)
def test_expand():
if "TEST_DATA" in os.environ:
xml_path = os.environ["TEST_DATA"]
else:
xml_path = tests.test_data_path
with open(f"{xml_path}/eml.xml", "r") as f:
xml = "".join(f.readlines())
eml = metapype_io.from_xml(xml)
validate.tree(eml)
references.expand(eml)
validate.tree(eml)
creator = eml.find_descendant(names.CREATOR)
creator.name = "node"
creator.attributes = dict()
metadata_provider = eml.find_descendant(names.METADATAPROVIDER)
metadata_provider.name = "node"
assert Node.is_equal(creator, metadata_provider)
def test_responsible_party():
creator = Node(names.CREATOR)
creator.add_attribute("id", "creator")
creator.add_namespace("eml", "https://eml.ecoinformatics.org/eml-2.2.0")
individual_name = Node(names.INDIVIDUALNAME)
creator.add_child(individual_name)
given_name = Node(names.GIVENNAME, content="Chase")
given_name.add_attribute("lang", "Spanish")
individual_name.add_child(given_name)
sur_name = Node(names.SURNAME, content="Gaucho")
sur_name.add_attribute("lang", "Spanish")
individual_name.add_child(sur_name)
individual_name = Node(names.INDIVIDUALNAME)
creator.add_child(individual_name)
given_name = Node(names.GIVENNAME, content="Cactus")
individual_name.add_child(given_name)
sur_name = Node(names.SURNAME, content="Jack")
individual_name.add_child(sur_name)
phone = Node(names.PHONE, content="999-999-9999")
creator.add_child(phone)
validate.tree(creator)
def test_responsible_party_with_role():
personnel = Node(names.PERSONNEL)
personnel.add_attribute("id", "personnel")
personnel.add_namespace("eml", "https://eml.ecoinformatics.org/eml-2.2.0")
individual_name = Node(names.INDIVIDUALNAME)
personnel.add_child(individual_name)
given_name = Node(names.GIVENNAME, content="Chase")
given_name.add_attribute("lang", "Spanish")
individual_name.add_child(given_name)
sur_name = Node(names.SURNAME, content="Gaucho")
sur_name.add_attribute("lang", "Spanish")
individual_name.add_child(sur_name)
individual_name = Node(names.INDIVIDUALNAME)
personnel.add_child(individual_name)
given_name = Node(names.GIVENNAME, content="Cactus")
individual_name.add_child(given_name)
sur_name = Node(names.SURNAME, content="Jack")
individual_name.add_child(sur_name)
phone = Node(names.PHONE, content="999-999-9999")
personnel.add_child(phone)
errs = []
# Without role, should get an error
with pytest.raises(MetapypeRuleError):
validate.tree(personnel)
# Error should name 'role' as the cause
validate.tree(personnel, errs)
for err_code, msg, node, *args in errs:
err_cause, min = args
assert err_cause == 'role'
# With role, it should be ok
role = Node(names.ROLE, content="drummer")
personnel.add_child(role)
validate.tree(personnel, errs)
def test_bounding_altitudes():
bounding_coordinates = Node(names.BOUNDINGCOORDINATES, parent=None)
bc_west = Node(names.WESTBOUNDINGCOORDINATE, parent=bounding_coordinates)
bc_east = Node(names.EASTBOUNDINGCOORDINATE, parent=bounding_coordinates)
bc_north = Node(names.NORTHBOUNDINGCOORDINATE, parent=bounding_coordinates)
bc_south = Node(names.SOUTHBOUNDINGCOORDINATE, parent=bounding_coordinates)
bc_west.content = "0.0"
bc_east.content = "0.0"
bc_north.content = "0.0"
bc_south.content = "0.0"
bounding_coordinates.add_child(bc_west)
bounding_coordinates.add_child(bc_east)
bounding_coordinates.add_child(bc_north)
bounding_coordinates.add_child(bc_south)
# without boundingAltitudes should be ok
validate.node(bounding_coordinates)
# boundingAltitudes should fail if not all required children present
bounding_altitudes = Node(names.BOUNDINGALTITUDES, parent=bounding_coordinates)
bounding_coordinates.add_child(bounding_altitudes)
with pytest.raises(MetapypeRuleError):
validate.tree(bounding_coordinates)
altitude_minimum = Node(names.ALTITUDEMINIMUM, parent=bounding_altitudes)
bounding_altitudes.add_child(altitude_minimum)
with pytest.raises(MetapypeRuleError):
validate.tree(bounding_coordinates)
altitude_minimum.content = "0.0"
with pytest.raises(MetapypeRuleError):
validate.tree(bounding_coordinates)
# boundingAltitudes should fail if not all required children have content
altitude_maximum = Node(names.ALTITUDEMAXIMUM, parent=bounding_altitudes)
bounding_altitudes.add_child(altitude_maximum)
altitude_units = Node(names.ALTITUDEUNITS, parent=bounding_altitudes)
bounding_altitudes.add_child(altitude_units)
with pytest.raises(MetapypeRuleError):
validate.tree(bounding_coordinates)
# with content filled in, should pass
altitude_maximum.content = "1000.0"
altitude_units.content = "meter"
validate.tree(bounding_coordinates)
def test_validate_prune():
if "TEST_DATA" in os.environ:
xml_path = os.environ["TEST_DATA"]
else:
xml_path = tests.test_data_path
with open(f"{xml_path}/eml.xml", "r") as f:
xml = "".join(f.readlines())
eml = metapype_io.from_xml(xml)
assert isinstance(eml, Node)
referencePublication = Node("referencePublication")
usageCitation = Node("usageCitation")
dataset = eml.find_single_node_by_path([names.DATASET])
dataset.add_child(referencePublication)
dataset.add_child(usageCitation)
errs = list()
validate.tree(eml, errs)
assert len(errs) > 0
validate.prune(eml)
errs = list()
validate.tree(eml, errs)
assert len(errs) == 0
def test_copy():
creator = Node(names.CREATOR)
creator.add_attribute("id", "creator")
creator.add_namespace("eml", "https://eml.ecoinformatics.org/eml-2.2.0")
individual_name = Node(names.INDIVIDUALNAME)
creator.add_child(individual_name)
given_name = Node(names.GIVENNAME, content="Chase")
given_name.add_attribute("lang", "Spanish")
individual_name.add_child(given_name)
sur_name = Node(names.SURNAME, content="Gaucho")
sur_name.add_attribute("lang", "Spanish")
individual_name.add_child(sur_name)
individual_name = Node(names.INDIVIDUALNAME)
creator.add_child(individual_name)
given_name = Node(names.GIVENNAME, content="Cactus")
individual_name.add_child(given_name)
sur_name = Node(names.SURNAME, content="Jack")
individual_name.add_child(sur_name)
validate.tree(creator)
creator_copy = creator.copy()
validate.tree(creator_copy)
assert is_deep_copy(creator, creator_copy)
def construct_texttype_node(text, parent_name=None):
if parent_name and parent_name not in TEXTTYPE_NODES:
return text
try:
subtree = metapype_io.from_xml(remove_escapes(text),
clean=True,
literals=['literalLayout', 'markdown'])
validate.tree(subtree)
return subtree
except Exception as e:
if e.__class__.__name__ == 'XMLSyntaxError':
if str(e).startswith('Start tag expected'):
try:
# We may have a naked string. Try adding the root node.
text = f"<{parent_name}>{text}</{parent_name}>"
subtree = metapype_io.from_xml(
text,
clean=True,
literals=['literalLayout', 'markdown'])
validate.tree(subtree)
return subtree
except Exception as e2:
raise InvalidXMLError(str(e2))
raise InvalidXMLError(str(e))
def test_validate_tree(node):
assert validate.tree(node) is None