From fd920aa077cd4214cef46c3b80775692cd48c22f Mon Sep 17 00:00:00 2001
From: Krishna Pai <kpai@fb.com>
Date: Tue, 9 Jan 2024 13:17:16 -0800
Subject: [PATCH] Add support for aggregation fuzzer to call Presto java and
 setup CI run. (#7947)

Summary:
This PR:
1. Sets up Aggregation Fuzzer to call Presto Java
2. Sets up an experimental scheduled run for the agg fuzzer.

Pull Request resolved: https://github.com/facebookincubator/velox/pull/7947

Reviewed By: mbasmanova

Differential Revision: D52040998

Pulled By: kgpai

fbshipit-source-id: 020282eef5f2c950c000cebed388fb160aba726b
---
 .github/workflows/experimental.yml            |  3 +-
 .../fuzzer/AggregationFuzzerTest.cpp          | 38 +++++++++++++------
 2 files changed, 28 insertions(+), 13 deletions(-)

diff --git a/.github/workflows/experimental.yml b/.github/workflows/experimental.yml
index cd330be6b1f9..a16b9d0f522e 100644
--- a/.github/workflows/experimental.yml
+++ b/.github/workflows/experimental.yml
@@ -116,7 +116,7 @@ jobs:
           name: aggregation
 
       - name: "Run Aggregate Fuzzer"
-        run: |  
+        run: |
           mkdir -p /tmp/aggregate_fuzzer_repro/
           rm -rfv /tmp/aggregate_fuzzer_repro/*
           chmod -R 777 /tmp/aggregate_fuzzer_repro
@@ -177,4 +177,3 @@ jobs:
           name: spark-agg-fuzzer-failure-artifacts
           path: |
             /tmp/spark_aggregate_fuzzer_repro
-
diff --git a/velox/functions/prestosql/fuzzer/AggregationFuzzerTest.cpp b/velox/functions/prestosql/fuzzer/AggregationFuzzerTest.cpp
index d41be51ad9d8..6fab78a4fd92 100644
--- a/velox/functions/prestosql/fuzzer/AggregationFuzzerTest.cpp
+++ b/velox/functions/prestosql/fuzzer/AggregationFuzzerTest.cpp
@@ -22,6 +22,7 @@
 #include "velox/exec/fuzzer/AggregationFuzzerOptions.h"
 #include "velox/exec/fuzzer/AggregationFuzzerRunner.h"
 #include "velox/exec/fuzzer/DuckQueryRunner.h"
+#include "velox/exec/fuzzer/PrestoQueryRunner.h"
 #include "velox/exec/fuzzer/TransformResultVerifier.h"
 #include "velox/functions/prestosql/aggregates/RegisterAggregateFunctions.h"
 #include "velox/functions/prestosql/fuzzer/ApproxDistinctInputGenerator.h"
@@ -45,6 +46,13 @@ DEFINE_string(
     "this comma separated list of function names "
     "(e.g: --only \"min\" or --only \"sum,avg\").");
 
+DEFINE_string(
+    presto_url,
+    "",
+    "Presto coordinator URI along with port. If set, we use Presto "
+    "source of truth. Otherwise, use DuckDB. Example: "
+    "--presto_url=http://127.0.0.1:8080");
+
 namespace facebook::velox::exec::test {
 namespace {
 
@@ -61,6 +69,23 @@ getCustomInputGenerators() {
   };
 }
 
+std::unique_ptr<ReferenceQueryRunner> setupReferenceQueryRunner() {
+  if (FLAGS_presto_url.empty()) {
+    auto duckQueryRunner = std::make_unique<DuckQueryRunner>();
+    duckQueryRunner->disableAggregateFunctions({
+        "skewness",
+        // DuckDB results on constant inputs are incorrect. Should be NaN,
+        // but DuckDB returns some random value.
+        "kurtosis",
+        "entropy",
+    });
+    return duckQueryRunner;
+  } else {
+    return std::make_unique<PrestoQueryRunner>(
+        FLAGS_presto_url, "aggregation_fuzzer");
+  }
+}
+
 } // namespace
 } // namespace facebook::velox::exec::test
 
@@ -81,16 +106,6 @@ int main(int argc, char** argv) {
 
   size_t initialSeed = FLAGS_seed == 0 ? std::time(nullptr) : FLAGS_seed;
 
-  auto duckQueryRunner =
-      std::make_unique<facebook::velox::exec::test::DuckQueryRunner>();
-  duckQueryRunner->disableAggregateFunctions({
-      "skewness",
-      // DuckDB results on constant inputs are incorrect. Should be NaN,
-      // but DuckDB returns some random value.
-      "kurtosis",
-      "entropy",
-  });
-
   // List of functions that have known bugs that cause crashes or failures.
   static const std::unordered_set<std::string> skipFunctions = {
       // https://github.com/facebookincubator/velox/issues/3493
@@ -101,6 +116,7 @@ int main(int argc, char** argv) {
 
   using facebook::velox::exec::test::ApproxDistinctResultVerifier;
   using facebook::velox::exec::test::ApproxPercentileResultVerifier;
+  using facebook::velox::exec::test::setupReferenceQueryRunner;
   using facebook::velox::exec::test::TransformResultVerifier;
 
   auto makeArrayVerifier = []() {
@@ -161,5 +177,5 @@ int main(int argc, char** argv) {
       facebook::velox::exec::test::getCustomInputGenerators();
   options.timestampPrecision =
       facebook::velox::VectorFuzzer::Options::TimestampPrecision::kMilliSeconds;
-  return Runner::run(initialSeed, std::move(duckQueryRunner), options);
+  return Runner::run(initialSeed, setupReferenceQueryRunner(), options);
 }