def _type_query_from_web_query(self, db_query):
# Get the raw name of the statement type (we fix some variations).
act_raw = self._pop('type', None)
if act_raw is not None:
if isinstance(act_raw, list):
assert len(act_raw) == 1, \
f"Got multiple entries for statement type: {act_raw}."
act_raw = act_raw[0]
act = make_statement_camel(act_raw)
db_query &= HasType([act])
return db_query
def get_statements():
"""Get some statements constrained by query."""
logger.info("Got query for statements!")
query_dict = request.args.copy()
limit_behavior = query_dict.pop('on_limit', 'sample')
if limit_behavior not in ['sample', 'truncate', 'error']:
abort(
Response('Unrecognized value for on_limit: %s' % limit_behavior,
400))
logger.info("Getting query details.")
try:
# Get the agents without specified locations (subject or object).
free_agents = [
__process_agent(ag) for ag in query_dict.poplist('agent')
]
# Get the agents with specified roles.
roled_agents = {
role: __process_agent(query_dict.pop(role))
for role in ['subject', 'object']
if query_dict.get(role) is not None
}
except DbAPIError as e:
logger.exception(e)
abort(Response('Failed to make agents from names: %s\n' % str(e), 400))
return
# Get the raw name of the statement type (we allow for variation in case).
act_uncamelled = query_dict.pop('type', None)
# Fix the case, if we got a statement type.
if act_uncamelled is not None:
act = make_statement_camel(act_uncamelled)
else:
act = None
# If there was something else in the query, there shouldn't be, so someone's
# probably confused.
if query_dict:
abort(
Response(
"Unrecognized query options; %s." % list(query_dict.keys()),
400))
return
# Make sure we got SOME agents. We will not simply return all
# phosphorylations, or all activations.
if not any(roled_agents.values()) and not free_agents:
logger.error("No agents.")
abort(
Response(("No agents. Must have 'subject', 'object', or "
"'other'!\n"), 400))
# Check to make sure none of the agents are None.
assert None not in roled_agents.values() and None not in free_agents, \
"None agents found. No agents should be None."
# Now find the statements.
stmts = []
logger.info("Getting statements...")
# First look for statements matching the role'd agents.
for role, (ag_id, ns) in roled_agents.items():
stmts = _get_relevant_statements(stmts, ag_id, ns, act, role)
if not len(stmts):
break
else:
# Second (if we're still looking), look for statements from free agents.
for ag_id, ns in free_agents:
stmts = _get_relevant_statements(stmts, ag_id, ns, act)
if not len(stmts):
break
# Check to make sure we didn't get too many statements
hit_limit = (len(stmts) > MAX_STATEMENTS)
if hit_limit:
if limit_behavior == 'error':
# Divide the statements up by type, and get counts of each type.
stmt_counts = {}
for stmt in stmts:
stmt_type = type(stmt).__name__
if stmt_type not in stmt_counts:
stmt_counts[stmt_type] = {'count': 0}
stmt_counts[stmt_type]['count'] += 1
return jsonify({'limited': True, 'statements': stmt_counts}), 413
elif limit_behavior == 'truncate':
stmts = stmts[:MAX_STATEMENTS]
elif limit_behavior == 'sample':
from random import sample
stmts = sample(stmts, MAX_STATEMENTS)
# TODO: This is a temporary patch. Remove ASAP.
# Fix the names from BE to FPLX
for s in stmts:
for ag in s.agent_list():
if ag is not None:
if 'BE' in ag.db_refs.keys():
ag.db_refs['FPLX'] = ag.db_refs.pop('BE')
# Create the json response, and send off.
resp = jsonify({
'limited': hit_limit,
'statements': [stmt.to_json() for stmt in stmts]
})
logger.info("Exiting with %d statements of nominal size %f MB." %
(len(stmts), sys.getsizeof(resp.data) / 1e6))
return resp
def _db_query_from_web_query(query_dict, require=None, empty_web_query=False):
db_query = EmptyQuery()
num_agents = 0
filter_ev = query_dict.pop('filter_ev', 'false').lower() == 'true'
ev_filter = EvidenceFilter()
# Get the agents without specified locations (subject or object).
for raw_ag in iter_free_agents(query_dict):
ag, ns = process_agent(raw_ag)
db_query &= HasAgent(ag, namespace=ns)
# Get the agents with specified roles.
for role in ['subject', 'object']:
raw_ag = query_dict.pop(role, None)
if raw_ag is None:
continue
if isinstance(raw_ag, list):
assert len(raw_ag) == 1, f'Malformed agent for {role}: {raw_ag}'
raw_ag = raw_ag[0]
num_agents += 1
ag, ns = process_agent(raw_ag)
db_query &= HasAgent(ag, namespace=ns, role=role.upper())
# Get the raw name of the statement type (we allow for variation in case).
act_raw = query_dict.pop('type', None)
if act_raw is not None:
if isinstance(act_raw, list):
assert len(act_raw) == 1, \
f"Got multiple entries for statement type: {act_raw}."
act_raw = act_raw[0]
act = make_statement_camel(act_raw)
db_query &= HasType([act])
# Get whether the user wants a strict match
if query_dict.pop('strict', 'false').lower() == 'true':
db_query &= HasNumAgents((num_agents, ))
# Unpack hashes, if present.
hashes = query_dict.pop('hashes', None)
if hashes:
db_query &= HasHash(hashes)
# Unpack paper ids, if present:
id_tpls = set()
for id_dict in query_dict.pop('paper_ids', []):
val = id_dict['id']
typ = id_dict['type']
# Turn tcids and trids into integers.
id_val = int(val) if typ in ['tcid', 'trid'] else val
id_tpls.add((typ, id_val))
if id_tpls:
paper_q = FromPapers(id_tpls)
db_query &= paper_q
if filter_ev:
ev_filter &= paper_q.ev_filter()
# Unpack mesh ids.
for mesh_id in query_dict.pop('mesh_ids', []):
mesh_q = FromMeshId(mesh_id)
db_query &= mesh_q
if filter_ev:
ev_filter &= mesh_q.ev_filter()
# Check for health of the resulting query, and some other things.
if isinstance(db_query, EmptyQuery):
raise DbAPIError(f"No arguments from web query {query_dict} mapped to "
f"db query.")
assert isinstance(db_query,
QueryCore), "Somehow db_query is not QueryCore."
if require and (db_query is None
or set(require) > set(db_query.get_component_queries())):
raise DbAPIError(f"Required query elements not found: {require}")
if query_dict and empty_web_query:
raise DbAPIError(f"Invalid query options: {query_dict.keys()}.")
return db_query
def get_statements(query_dict, offs, max_stmts, ev_limit, best_first):
"""Get some statements constrained by query."""
logger.info("Getting query details.")
if ev_limit is None:
ev_limit = 10
try:
# Get the agents without specified locations (subject or object).
free_agents = [
__process_agent(ag) for ag in query_dict.poplist('agent')
]
ofaks = {k for k in query_dict.keys() if k.startswith('agent')}
free_agents += [__process_agent(query_dict.pop(k)) for k in ofaks]
# Get the agents with specified roles.
roled_agents = {
role: __process_agent(query_dict.pop(role))
for role in ['subject', 'object']
if query_dict.get(role) is not None
}
except DbAPIError as e:
logger.exception(e)
abort(Response('Failed to make agents from names: %s\n' % str(e), 400))
return
# Get the raw name of the statement type (we allow for variation in case).
act_raw = query_dict.pop('type', None)
# Fix the case, if we got a statement type.
act = None if act_raw is None else make_statement_camel(act_raw)
# If there was something else in the query, there shouldn't be, so
# someone's probably confused.
if query_dict:
abort(
Response(
"Unrecognized query options; %s." % list(query_dict.keys()),
400))
return
# Make sure we got SOME agents. We will not simply return all
# phosphorylations, or all activations.
if not any(roled_agents.values()) and not free_agents:
logger.error("No agents.")
abort(
Response(("No agents. Must have 'subject', 'object', or "
"'other'!\n"), 400))
# Check to make sure none of the agents are None.
assert None not in roled_agents.values() and None not in free_agents, \
"None agents found. No agents should be None."
# Now find the statements.
logger.info("Getting statements...")
agent_iter = [(role, ag_dbid, ns)
for role, (ag_dbid, ns) in roled_agents.items()]
agent_iter += [(None, ag_dbid, ns) for ag_dbid, ns in free_agents]
result = \
get_statement_jsons_from_agents(agent_iter, stmt_type=act, offset=offs,
max_stmts=max_stmts, ev_limit=ev_limit,
best_first=best_first)
return result