From d7d9bea5e8f1bd67109dfa2859d42d73057d006f Mon Sep 17 00:00:00 2001
From: Mathias Vorreiter Pedersen <mathiasvp@github.com>
Date: Thu, 23 Feb 2023 14:27:39 +0000
Subject: [PATCH] QL: Add a query for computing the join order metric for
 non-recursive predicates.

---
 .../src/queries/performance/LargeJoinOrder.ql | 29 +++++++++++++++++++
 1 file changed, 29 insertions(+)
 create mode 100644 ql/ql/src/queries/performance/LargeJoinOrder.ql

diff --git a/ql/ql/src/queries/performance/LargeJoinOrder.ql b/ql/ql/src/queries/performance/LargeJoinOrder.ql
new file mode 100644
index 00000000000..3d51affd20f
--- /dev/null
+++ b/ql/ql/src/queries/performance/LargeJoinOrder.ql
@@ -0,0 +1,29 @@
+/**
+ * Shows a list of the non-recursive predicates with the worst join order as determined
+ * by the join order metric.
+ */
+
+import ql
+import codeql_ql.StructuredLogs
+import KindPredicatesLog
+
+/**
+ * Gets the badness of a non-recursive predicate evaluation.
+ *
+ * The badness is the maximum number of tuples in the pipeline divided by the
+ * maximum of two numbers: the size of the result and the size of the largest dependency.
+ */
+float getBadness(ComputeSimple simple) {
+  exists(float maxTupleCount, float resultSize, float largestDependency, float denom |
+    resultSize = simple.getResultSize() and
+    maxTupleCount = max(simple.getPipelineRun().getCount(_)) and
+    largestDependency = max(simple.getDependencies().getADependency().getResultSize()) and
+    denom = resultSize.maximum(largestDependency) and
+    denom > 0 and // avoid division by zero (which would create a NaN result).
+    result = maxTupleCount / denom
+  )
+}
+
+from ComputeSimple evt, float f
+where f = getBadness(evt) and f > 1.5
+select evt, f order by f desc