def convert_nsarg(
nsarg: str,
api_url: str = None,
namespace_targets: Mapping[str, List[str]] = None,
canonicalize: bool = False,
decanonicalize: bool = False,
) -> str:
"""[De]Canonicalize NSArg
Args:
nsarg (str): bel statement string or partial string (e.g. subject or object)
api_url (str): BEL.bio api url to use, e.g. https://api.bel.bio/v1
namespace_targets (Mapping[str, List[str]]): formatted as in configuration file example
canonicalize (bool): use canonicalize endpoint/namespace targets
decanonicalize (bool): use decanonicalize endpoint/namespace targets
Results:
str: converted NSArg
"""
if not api_url:
api_url = config["bel_api"]["servers"]["api_url"]
if not api_url:
log.error("Missing api url - cannot convert namespace")
return None
params = None
if namespace_targets:
namespace_targets_str = json.dumps(namespace_targets)
params = {"namespace_targets": namespace_targets_str}
if not namespace_targets:
if canonicalize:
api_url = api_url + "/terms/{}/canonicalized"
elif decanonicalize:
api_url = api_url + "/terms/{}/decanonicalized"
else:
log.warning(
"Missing (de)canonical flag - cannot convert namespaces")
return nsarg
else:
api_url = api_url + "/terms/{}/canonicalized" # overriding with namespace_targets
request_url = api_url.format(url_path_param_quoting(nsarg))
r = http_client.get(request_url, params=params, timeout=10)
if r and r.status_code == 200:
nsarg = r.json().get("term_id", nsarg)
elif not r or r.status_code == 404:
log.error(f"[de]Canonicalization endpoint missing: {request_url}")
return nsarg
def validate_arg_values(ast, bo):
"""Recursively validate arg (NSArg and StrArg) values
Check that NSArgs are found in BELbio API and match appropriate entity_type.
Check that StrArgs match their value - either default namespace or regex string
Generate a WARNING if not.
Args:
bo: bel object
Returns:
bel object
"""
if not bo.api_url:
log.info("No API endpoint defined")
return bo
log.debug(f"AST: {ast}")
# Test NSArg terms
if isinstance(ast, NSArg):
term_id = "{}:{}".format(ast.namespace, ast.value)
value_types = ast.value_types
log.debug(f"Value types: {value_types} AST value: {ast.value}")
# Default namespaces are defined in the bel_specification file
if ast.namespace == "DEFAULT": # may use the DEFAULT namespace or not
for value_type in value_types:
default_namespace = [
ns["name"]
for ns in bo.spec["namespaces"][value_type]["info"]
] + [
ns["abbreviation"]
for ns in bo.spec["namespaces"][value_type]["info"]
]
if ast.value in default_namespace:
log.debug(
"Default namespace valid term: {}".format(term_id))
break
else: # if for loop doesn't hit the break, run this else
log.debug("Default namespace invalid term: {}".format(term_id))
bo.validation_messages.append(
("WARNING", f"Default Term: {term_id} not found"))
# Process normal, non-default-namespace terms
else:
request_url = bo.api_url + "/terms/{}".format(
url_path_param_quoting(term_id))
log.info(f"Validate Arg Values url {request_url}")
r = get_url(request_url)
if r and r.status_code == 200:
result = r.json()
# function signature term value_types doesn't match up with API term entity_types
log.debug(
f'AST.value_types {ast.value_types} Entity types {result.get("entity_types", [])}'
)
# Check that entity types match
if len(
set(ast.value_types).intersection(
result.get("entity_types", []))) == 0:
log.debug(
"Invalid Term - statement term {} allowable entity types: {} do not match API term entity types: {}"
.format(term_id, ast.value_types,
result.get("entity_types", [])))
bo.validation_messages.append((
"WARNING",
"Invalid Term - statement term {} allowable entity types: {} do not match API term entity types: {}"
.format(term_id, ast.value_types,
result.get("entity_types", [])),
))
if term_id in result.get("obsolete_ids", []):
bo.validation_messages.append((
"WARNING",
f'Obsolete term: {term_id} Current term: {result["id"]}'
))
elif r.status_code == 404:
bo.validation_messages.append(
("WARNING", f"Term: {term_id} not found in namespace"))
else:
log.error(f"Status {r.status_code} - Bad URL: {request_url}")
# Process StrArgs
if isinstance(ast, StrArg):
log.debug(f" Check String Arg: {ast.value} {ast.value_types}")
for value_type in ast.value_types:
# Is this a regex to match against
if re.match("/", value_type):
value_type = re.sub("^/", "", value_type)
value_type = re.sub("/$", "", value_type)
match = re.match(value_type, ast.value)
if match:
break
if value_type in bo.spec["namespaces"]:
default_namespace = [
ns["name"]
for ns in bo.spec["namespaces"][value_type]["info"]
] + [
ns["abbreviation"]
for ns in bo.spec["namespaces"][value_type]["info"]
]
if ast.value in default_namespace:
break
else: # If for loop doesn't hit the break, no matches found, therefore for StrArg value is bad
bo.validation_messages.append((
"WARNING",
f"String value {ast.value} does not match default namespace value or regex pattern: {ast.value_types}",
))
# Recursively process every NSArg by processing BELAst and Functions
if hasattr(ast, "args"):
for arg in ast.args:
validate_arg_values(arg, bo)
return bo