perf(taint): ask the caller check only of a parameter that starts a flow into the field

Asked of every parameter up front, 6ebbc89's caller check ran a repository-
wide flow query per parameter of every method in a file and cut PMPro's
completed injection questions from 272 to 67 in the same budget -- so its
48 -> 26 findings could not be read as precision. The flow into the field
is asked once as before; only a parameter at the head of an unsanitized
flow is checked against its callers. finding_dump now records completed
regions so runs compare only where both answered.

Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01EmM1j46qgKMh3zqD1QMwEo

benchmarks/push/finding_dump.py

@@ -81,11 +81,17 @@ def main() -> int:
         "seconds": round(time.monotonic() - started, 1),
         "questions": len(outcomes),
         "completed": sum(1 for o in outcomes if o.status == "completed"),
+        # Which questions answered, so two runs can be compared only where BOTH answered: a finding that
+        # vanishes because its question went unasked is lost coverage, not gained precision.
+        "completed_regions": sorted({f"{o.identity.path}:{o.identity.function or ''}" for o in outcomes if o.status == "completed"}),
         "degradations": sorted({str(d.get("reason")) for d in result.degradations}),
         "findings": findings,
     }
     args.out.write_text(json.dumps(record, indent=2) + "\n")
-    print(json.dumps({k: v for k, v in record.items() if k != "findings"} | {"findings": len(findings)}), flush=True)
+    print(
+        json.dumps({k: v for k, v in record.items() if k not in ("findings", "completed_regions")} | {"findings": len(findings)}),
+        flush=True,
+    )
     if not outcomes:
         raise SystemExit("no question was asked: the dump measured nothing")
     return 0

src/openultrasast/cpg/queries/taint.sc

@@ -306,10 +306,12 @@
         val methods = methodsIn(fileName)
         val framework: List[io.shiftleft.codepropertygraph.generated.nodes.CfgNode] =
           methods.flatMap(_.ast.isCall.filter(c => sourcePatterns.exists(p => c.code.contains(p))).l)
+        // Every parameter is a CANDIDATE start; whether its callers feed it is asked only of a parameter that
+        // actually begins an unsanitized flow into the field being read (`fieldIsTainted`). Asked of every
+        // parameter up front, the caller check ran a repository-wide flow query per parameter of every
+        // method in the file and cut PMPro's completed questions from 272 to 67 in the same budget.
         val params: List[io.shiftleft.codepropertygraph.generated.nodes.CfgNode] =
-          if (parameterSources == "true")
-            methods.flatMap(m => m.parameter.l.filter(p => p.name != "this" && fedByCaller(m, p)))
-          else Nil
+          if (parameterSources == "true") methods.flatMap(_.parameter.l.filter(_.name != "this")) else Nil
         framework ++ params
       }
     )
@@ -331,7 +333,13 @@
                 // queries that way: the fixed side of its pair went from 1 finding to 13 without this,
                 // which is the pair no longer separating at all.
                 val flows = args(1).start.reachableByFlows(seeds.iterator).l
-                flows.exists(f => !f.elements.map(_.code).l.exists(code => mentionsToken(code, sanitizerNames)))
+                flows.exists { f =>
+                  !f.elements.map(_.code).l.exists(code => mentionsToken(code, sanitizerNames)) && (f.elements.headOption match {
+                    case Some(param: io.shiftleft.codepropertygraph.generated.nodes.MethodParameterIn) =>
+                      fedByCaller(param.method, param)
+                    case _ => true
+                  })
+                }
               case _ => false
             })
           }