Merge pull request #108 from bento-platform/features/anon-discovery-c…

…onfig-only

anonymous metadata queries should use discovery config only

bento_beacon/app.py

@@ -17,7 +17,7 @@
 from .utils.beacon_request import save_request_data, validate_request, verify_permissions
 from .utils.beacon_response import init_response_data
 from .utils.katsu_utils import katsu_censorship_settings
-from .utils.censorship import set_censorship_settings
+from .utils.censorship import set_censorship_settings, reject_query_if_not_permitted
 
 REQUEST_SPEC_RELATIVE_PATH = "beacon-v2/framework/json/requests/"
 BEACON_MODELS = ["analyses", "biosamples", "cohorts", "datasets", "individuals", "runs", "variants"]
@@ -96,6 +96,7 @@ def before_request():
         validate_request()
         verify_permissions()
         save_request_data()
+        reject_query_if_not_permitted()
         init_response_data()
 
 

bento_beacon/config_files/config.py

@@ -169,6 +169,9 @@ class Config:
     PHENOPACKETS_SCHEMA_REFERENCE = {"entityType": "individual", "schema": "phenopackets v1"}
 
     MAX_RETRIES_FOR_CENSORSHIP_PARAMS = 2
+
+    # don't let anonymous users query arbitrary phenopacket or experiment fields
+    ANONYMOUS_METADATA_QUERY_USES_DISCOVERY_CONFIG_ONLY = True
     # -------------------
     # gohan
 

bento_beacon/endpoints/individuals.py

@@ -2,13 +2,10 @@
     P_DOWNLOAD_DATA,
     P_QUERY_DATA,
 )
-from flask import Blueprint
+from flask import Blueprint, g
 from functools import reduce
 from ..authz.middleware import authz_middleware, check_permission
-from ..utils.beacon_request import (
-    query_parameters_from_request,
-    summary_stats_requested,
-)
+from ..utils.beacon_request import summary_stats_requested
 from ..utils.beacon_response import (
     add_info_to_response,
     add_stats_to_response,
@@ -32,7 +29,10 @@
 
 @individuals.route("/individuals", methods=["GET", "POST"])
 def get_individuals():
-    variants_query, phenopacket_filters, experiment_filters, config_filters = query_parameters_from_request()
+    variants_query = g.beacon_query_parameters["variants_query"]
+    phenopacket_filters = g.beacon_query_parameters["phenopacket_filters"]
+    experiment_filters = g.beacon_query_parameters["experiment_filters"]
+    config_filters = g.beacon_query_parameters["config_filters"]
 
     no_query = not (variants_query or phenopacket_filters or experiment_filters or config_filters)
     search_sample_ids = variants_query or experiment_filters

bento_beacon/endpoints/variants.py

@@ -1,6 +1,5 @@
-from flask import Blueprint
+from flask import Blueprint, g
 from ..authz.middleware import authz_middleware
-from ..utils.beacon_request import query_parameters_from_request
 from ..utils.beacon_response import build_query_response, add_info_to_response, zero_count_response
 from ..utils.gohan_utils import query_gohan, gohan_total_variants_count, gohan_totals_by_sample_id
 from ..utils.search import biosample_id_search
@@ -12,7 +11,10 @@
 @variants.route("/g_variants", methods=["GET", "POST"])
 @authz_middleware.deco_public_endpoint  # TODO: for now. eventually, return more depending on permissions
 def get_variants():
-    variants_query, phenopacket_filters, experiment_filters, config_filters = query_parameters_from_request()
+    variants_query = g.beacon_query_parameters["variants_query"]
+    phenopacket_filters = g.beacon_query_parameters["phenopacket_filters"]
+    experiment_filters = g.beacon_query_parameters["experiment_filters"]
+    config_filters = g.beacon_query_parameters["config_filters"]
     has_filters = phenopacket_filters or experiment_filters or config_filters
 
     # if no query, return total count of variants

bento_beacon/utils/beacon_request.py

@@ -21,9 +21,9 @@ def expand_path(id):
     return id.replace("/", ".[item].")
 
 
-def query_parameters_from_request():
-    variants_query = g.request_data.get("requestParameters", {}).get("g_variant") or {}
-    filters = g.request_data.get("filters") or []
+def parse_query_params(request_data):
+    variants_query = request_data.get("requestParameters", {}).get("g_variant") or {}
+    filters = request_data.get("filters") or []
     phenopacket_filters = list(filter(lambda f: f["id"].startswith("phenopacket."), filters))
     experiment_filters = list(filter(lambda f: f["id"].startswith("experiment."), filters))
     config_filters = [f for f in filters if f not in phenopacket_filters and f not in experiment_filters]
@@ -49,7 +49,12 @@ def query_parameters_from_request():
             experiment_filters,
         )
     )
-    return variants_query, phenopacket_filters, experiment_filters, config_filters
+    return {
+        "variants_query": variants_query,
+        "phenopacket_filters": phenopacket_filters,
+        "experiment_filters": experiment_filters,
+        "config_filters": config_filters,
+    }
 
 
 # structure GET params so they match the nested structure in POST
@@ -130,8 +135,12 @@ def save_request_data():
     if request_bento:
         request_data["bento"] = request_bento
 
+    # raw request data, this is echoed in response "meta" field
     g.request_data = request_data
 
+    # parsed query components
+    g.beacon_query_parameters = parse_query_params(request_data)
+
 
 def validate_request():
     if request.method == "POST":

bento_beacon/utils/censorship.py

@@ -70,3 +70,19 @@ def reject_if_too_many_filters(filters):
 def censored_chart_data(data):
     t = get_censorship_threshold()  # zero with correct permissions
     return [{"label": d["label"], "value": d["value"]} for d in data if d["value"] > t]
+
+
+def query_has_phenopacket_filter():
+    return bool(g.beacon_query_parameters["phenopacket_filters"])
+
+
+def query_has_experiment_filter():
+    return bool(g.beacon_query_parameters["experiment_filters"])
+
+
+# some anonymous queries are not permitted
+def reject_query_if_not_permitted():
+    if g.permission_query_data or not current_app.config["ANONYMOUS_METADATA_QUERY_USES_DISCOVERY_CONFIG_ONLY"]:
+        return
+    if query_has_phenopacket_filter() or query_has_experiment_filter():
+        raise InvalidQuery("anonymous queries should use filters from discovery config only")