perf(SPARQL): cache endpoint — 1 COUNT query per fonte invece di 1 per gruppo

dati_camera: 103 gruppi → 1 query (prima 103)
dati_cultura: 213 gruppi → 1 query (prima 213)
dati_senato: 80 gruppi → 1 query (prima 80)
Riduce il tempo SPARQL da minuti a secondi.

Author: Gabrymi93

scripts/collectors/sparql.py

@@ -11,6 +11,10 @@
 
 logger = logging.getLogger(__name__)
 
+# Cache endpoint→reachable per run. Evita COUNT query per ogni gruppo
+# della stessa fonte SPARQL (dati_camera: 103 gruppi → 1 query invece di 103).
+_endpoint_cache: dict[str, bool] = {}
+
 
 def _group_sparql_bindings(bindings: list[dict]) -> dict[str, dict]:
     grouped: dict[str, dict] = {}
@@ -114,23 +118,26 @@ def validate_items(items: list[dict]) -> dict[str, Any]:
         result["error"] = "No SPARQL endpoint"
         return result
 
-    # Probe: execute a COUNT query
-    try:
-        count_query = f"""
-        SELECT (COUNT(*) AS ?cnt) WHERE {{
-            GRAPH <{graph_uri}> {{ ?s ?p ?o }}
-        }}
-        """.strip()
-        bindings = execute_sparql(endpoint, count_query, timeout=15)
-        if bindings:
-            cnt = bindings[0].get("cnt", "0")
-            result["reachable"] = True
-            result["triple_count"] = int(cnt) if cnt and str(cnt).isdigit() else 0
-            result["readiness_score"] = 3  # SPARQL reachable e risponde
-        else:
-            result["error"] = "Empty response"
-    except Exception as exc:
-        result["error"] = str(exc)
-        result["readiness_score"] = 0
+    # Cache endpoint: se gia' verificato, usa risultato senza COUNT query
+    if endpoint not in _endpoint_cache:
+        try:
+            count_query = f"""
+            SELECT (COUNT(*) AS ?cnt) WHERE {{
+                GRAPH <{graph_uri}> {{ ?s ?p ?o }}
+            }}
+            """.strip()
+            bindings = execute_sparql(endpoint, count_query, timeout=10)
+            if bindings:
+                cnt = bindings[0].get("cnt", "0")
+                _endpoint_cache[endpoint] = True
+                result["triple_count"] = int(cnt) if cnt and str(cnt).isdigit() else 0
+            else:
+                _endpoint_cache[endpoint] = False
+        except Exception as exc:
+            _endpoint_cache[endpoint] = False
+            result["error"] = str(exc)
+
+    result["reachable"] = _endpoint_cache.get(endpoint, False)
+    result["readiness_score"] = 3 if result["reachable"] else 0
 
     return result