fix(test): Move stray MultiThreadedHashJoinTest to HashJoinTest.cpp

hashJoinWithLazyProbeInputAndSpill was a MultiThreadedHashJoinTest test
that ended up in HashJoinTestSpill.cpp, which only defines HashJoinTest.
Move it back to HashJoinTest.cpp where MultiThreadedHashJoinTest is defined.

Author: kgpai

velox/exec/tests/HashJoinTest.cpp

@@ -3636,6 +3636,84 @@ TEST_P(HashJoinTest, semiProjectOverLazyVectors) {
       .run();
 }
 
+// Verifies that hash join with lazy probe input produces correct results
+// across multiple join types, with automatic spill injection via
+// HashJoinBuilder.
+TEST_P(MultiThreadedHashJoinTest, hashJoinWithLazyProbeInputAndSpill) {
+  VectorFuzzer::Options opts;
+  opts.vectorSize = 1'000;
+  VectorFuzzer fuzzer(opts, pool());
+
+  const int32_t numProbeVectors = 3;
+  std::vector<RowVectorPtr> probeVectors;
+  std::vector<RowVectorPtr> probeReference;
+  probeVectors.reserve(numProbeVectors);
+  probeReference.reserve(numProbeVectors);
+  for (int32_t i = 0; i < numProbeVectors; ++i) {
+    auto nonLazy = fuzzer.fuzzRow(probeType_);
+    probeReference.push_back(
+        std::dynamic_pointer_cast<RowVector>(
+            nonLazy->testingCopyPreserveEncodings()));
+    probeVectors.push_back(
+        VectorFuzzer(opts, pool()).fuzzRowChildrenToLazy(nonLazy));
+  }
+  createDuckDbTable("t", probeReference);
+
+  const int32_t numBuildVectors = 3;
+  std::vector<RowVectorPtr> buildVectors;
+  buildVectors.reserve(numBuildVectors);
+  for (int32_t i = 0; i < numBuildVectors; ++i) {
+    buildVectors.push_back(fuzzer.fuzzRow(buildType_));
+  }
+  createDuckDbTable("u", buildVectors);
+
+  struct {
+    core::JoinType joinType;
+    std::string filter;
+    std::vector<std::string> outputColumns;
+    std::string referenceQuery;
+  } testCases[] = {
+      {core::JoinType::kInner,
+       "",
+       concat(probeType_->names(), buildType_->names()),
+       "SELECT t_k1, t_k2, t_v1, u_k1, u_k2, u_v1 FROM t, u WHERE t.t_k1 = u.u_k1"},
+      {core::JoinType::kLeft,
+       "",
+       concat(probeType_->names(), buildType_->names()),
+       "SELECT t_k1, t_k2, t_v1, u_k1, u_k2, u_v1 FROM t LEFT JOIN u ON t.t_k1 = u.u_k1"},
+      {core::JoinType::kFull,
+       "",
+       concat(probeType_->names(), buildType_->names()),
+       "SELECT t_k1, t_k2, t_v1, u_k1, u_k2, u_v1 FROM t FULL OUTER JOIN u ON t.t_k1 = u.u_k1"},
+      {core::JoinType::kAnti,
+       "t_k2 <> u_k2",
+       probeType_->names(),
+       "SELECT t_k1, t_k2, t_v1 FROM t WHERE NOT EXISTS (SELECT 1 FROM u WHERE t.t_k1 = u.u_k1 AND t.t_k2 <> u.u_k2)"},
+      {core::JoinType::kLeftSemiProject,
+       "",
+       concat(probeType_->names(), {"match"}),
+       "SELECT t_k1, t_k2, t_v1, EXISTS (SELECT 1 FROM u WHERE t.t_k1 = u.u_k1) FROM t"},
+  };
+
+  for (const auto& testCase : testCases) {
+    SCOPED_TRACE(
+        fmt::format("joinType: {}", static_cast<int>(testCase.joinType)));
+
+    HashJoinBuilder(*pool_, duckDbQueryRunner_, driverExecutor_.get())
+        .numDrivers(numDrivers_)
+        .parallelizeJoinBuildRows(parallelBuildSideRowsEnabled_)
+        .probeKeys({"t_k1"})
+        .probeVectors(std::vector<RowVectorPtr>(probeVectors))
+        .buildKeys({"u_k1"})
+        .buildVectors(std::vector<RowVectorPtr>(buildVectors))
+        .joinType(testCase.joinType)
+        .joinFilter(testCase.filter)
+        .joinOutputLayout(std::vector<std::string>(testCase.outputColumns))
+        .referenceQuery(testCase.referenceQuery)
+        .run();
+  }
+}
+
 VELOX_INSTANTIATE_TEST_SUITE_P(
     MultiThreadedHashJoinTest,
     MultiThreadedHashJoinTest,

velox/exec/tests/HashJoinTestSpill.cpp

@@ -467,84 +467,6 @@ DEBUG_ONLY_TEST_P(HashJoinTest, hashProbeEnsureLazyInputLoaded) {
   }
 }
 
-// Verifies that hash join with lazy probe input produces correct results
-// across multiple join types, with automatic spill injection via
-// HashJoinBuilder.
-TEST_P(MultiThreadedHashJoinTest, hashJoinWithLazyProbeInputAndSpill) {
-  VectorFuzzer::Options opts;
-  opts.vectorSize = 1'000;
-  VectorFuzzer fuzzer(opts, pool());
-
-  const int32_t numProbeVectors = 3;
-  std::vector<RowVectorPtr> probeVectors;
-  std::vector<RowVectorPtr> probeReference;
-  probeVectors.reserve(numProbeVectors);
-  probeReference.reserve(numProbeVectors);
-  for (int32_t i = 0; i < numProbeVectors; ++i) {
-    auto nonLazy = fuzzer.fuzzRow(probeType_);
-    probeReference.push_back(
-        std::dynamic_pointer_cast<RowVector>(
-            nonLazy->testingCopyPreserveEncodings()));
-    probeVectors.push_back(
-        VectorFuzzer(opts, pool()).fuzzRowChildrenToLazy(nonLazy));
-  }
-  createDuckDbTable("t", probeReference);
-
-  const int32_t numBuildVectors = 3;
-  std::vector<RowVectorPtr> buildVectors;
-  buildVectors.reserve(numBuildVectors);
-  for (int32_t i = 0; i < numBuildVectors; ++i) {
-    buildVectors.push_back(fuzzer.fuzzRow(buildType_));
-  }
-  createDuckDbTable("u", buildVectors);
-
-  struct {
-    core::JoinType joinType;
-    std::string filter;
-    std::vector<std::string> outputColumns;
-    std::string referenceQuery;
-  } testCases[] = {
-      {core::JoinType::kInner,
-       "",
-       concat(probeType_->names(), buildType_->names()),
-       "SELECT t_k1, t_k2, t_v1, u_k1, u_k2, u_v1 FROM t, u WHERE t.t_k1 = u.u_k1"},
-      {core::JoinType::kLeft,
-       "",
-       concat(probeType_->names(), buildType_->names()),
-       "SELECT t_k1, t_k2, t_v1, u_k1, u_k2, u_v1 FROM t LEFT JOIN u ON t.t_k1 = u.u_k1"},
-      {core::JoinType::kFull,
-       "",
-       concat(probeType_->names(), buildType_->names()),
-       "SELECT t_k1, t_k2, t_v1, u_k1, u_k2, u_v1 FROM t FULL OUTER JOIN u ON t.t_k1 = u.u_k1"},
-      {core::JoinType::kAnti,
-       "t_k2 <> u_k2",
-       probeType_->names(),
-       "SELECT t_k1, t_k2, t_v1 FROM t WHERE NOT EXISTS (SELECT 1 FROM u WHERE t.t_k1 = u.u_k1 AND t.t_k2 <> u.u_k2)"},
-      {core::JoinType::kLeftSemiProject,
-       "",
-       concat(probeType_->names(), {"match"}),
-       "SELECT t_k1, t_k2, t_v1, EXISTS (SELECT 1 FROM u WHERE t.t_k1 = u.u_k1) FROM t"},
-  };
-
-  for (const auto& testCase : testCases) {
-    SCOPED_TRACE(
-        fmt::format("joinType: {}", static_cast<int>(testCase.joinType)));
-
-    HashJoinBuilder(*pool_, duckDbQueryRunner_, driverExecutor_.get())
-        .numDrivers(numDrivers_)
-        .parallelizeJoinBuildRows(parallelBuildSideRowsEnabled_)
-        .probeKeys({"t_k1"})
-        .probeVectors(std::vector<RowVectorPtr>(probeVectors))
-        .buildKeys({"u_k1"})
-        .buildVectors(std::vector<RowVectorPtr>(buildVectors))
-        .joinType(testCase.joinType)
-        .joinFilter(testCase.filter)
-        .joinOutputLayout(std::vector<std::string>(testCase.outputColumns))
-        .referenceQuery(testCase.referenceQuery)
-        .run();
-  }
-}
-
 TEST_P(HashJoinTest, dynamicFilters) {
   const int32_t numSplits = 10;
   const int32_t numRowsProbe = 333;