From 94d34422ff937c9942f8016f873958b8661ac283 Mon Sep 17 00:00:00 2001
From: PHILO-HE <he_feilong@126.com>
Date: Fri, 24 Jan 2025 14:40:34 +0800
Subject: [PATCH] Initial

---
 velox/functions/lib/Re2Functions.cpp          |  9 +++++--
 .../functions/lib/tests/Re2FunctionsTest.cpp  | 25 +++++++++++++++++++
 2 files changed, 32 insertions(+), 2 deletions(-)

diff --git a/velox/functions/lib/Re2Functions.cpp b/velox/functions/lib/Re2Functions.cpp
index 78ce3e2f29b56..1ef1bb69a3539 100644
--- a/velox/functions/lib/Re2Functions.cpp
+++ b/velox/functions/lib/Re2Functions.cpp
@@ -149,8 +149,13 @@ bool re2Extract(
       result.setNoCopy(row, StringView(extracted.data(), extracted.size()));
       return !StringView::isInline(extracted.size());
     } else {
-      result.setNull(row, true);
-      return false;
+      if (emptyNoMatch) {
+        result.setNoCopy(row, StringView(nullptr, 0));
+        return true;
+      } else {
+        result.setNull(row, true);
+        return false;
+      }
     }
   }
 }
diff --git a/velox/functions/lib/tests/Re2FunctionsTest.cpp b/velox/functions/lib/tests/Re2FunctionsTest.cpp
index f27d9cb46453f..d7e79d5984fb4 100644
--- a/velox/functions/lib/tests/Re2FunctionsTest.cpp
+++ b/velox/functions/lib/tests/Re2FunctionsTest.cpp
@@ -44,6 +44,13 @@ std::shared_ptr<exec::VectorFunction> makeRegexExtract(
   return makeRe2Extract(name, inputArgs, config, /*emptyNoMatch=*/false);
 }
 
+std::shared_ptr<exec::VectorFunction> makeRegexExtractEmptyNoMatch(
+    const std::string& name,
+    const std::vector<exec::VectorFunctionArg>& inputArgs,
+    const core::QueryConfig& config) {
+  return makeRe2Extract(name, inputArgs, config, /*emptyNoMatch=*/true);
+}
+
 class Re2FunctionsTest : public test::FunctionBaseTest {
  public:
   static void SetUpTestCase() {
@@ -54,6 +61,8 @@ class Re2FunctionsTest : public test::FunctionBaseTest {
         "re2_search", re2SearchSignatures(), makeRe2Search);
     exec::registerStatefulVectorFunction(
         "re2_extract", re2ExtractSignatures(), makeRegexExtract);
+    exec::registerStatefulVectorFunction(
+        "re2_extract_empty_no_match", re2ExtractSignatures(), makeRegexExtractEmptyNoMatch);
     exec::registerStatefulVectorFunction(
         "re2_extract_all", re2ExtractAllSignatures(), makeRe2ExtractAll);
     exec::registerStatefulVectorFunction("like", likeSignatures(), makeLike);
@@ -383,6 +392,22 @@ TEST_F(Re2FunctionsTest, regexExtract) {
   });
 }
 
+template <typename F>
+void testRe2ExtractEmptyNoMatch(F&& regexExtract) {
+  // Group case that mismatch.
+  EXPECT_EQ(
+      regexExtract("rat cat\nbat dog", "ra(.)|blah(.)(.)", 2), "");
+}
+
+TEST_F(Re2FunctionsTest, regexExtract) {
+  testRe2ExtractEmptyNoMatch([&](std::optional<std::string> str,
+                     std::optional<std::string> pattern,
+                     std::optional<int> group) {
+    return evaluateOnce<std::string>(
+        "re2_extract_empty_no_match(c0, c1, c2)", str, pattern, group);
+  });
+}
+
 TEST_F(Re2FunctionsTest, regexExtractBigintGroupId) {
   testRe2Extract([&](std::optional<std::string> str,
                      std::optional<std::string> pattern,