From 8b753ed2ab7bcf9b24dab4c0e7df750709efc548 Mon Sep 17 00:00:00 2001
From: "menglingda.mld" <menglingda.mld@alibaba-inc.com>
Date: Thu, 28 May 2026 17:24:34 +0800
Subject: [PATCH] fix: date validation reject invalid dates instead of silent
 normalization

---
 src/paimon/common/utils/string_utils.cpp      | 8 ++++++--
 src/paimon/common/utils/string_utils_test.cpp | 4 ++++
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/src/paimon/common/utils/string_utils.cpp b/src/paimon/common/utils/string_utils.cpp
index 3b92209a0..eb327a94e 100644
--- a/src/paimon/common/utils/string_utils.cpp
+++ b/src/paimon/common/utils/string_utils.cpp
@@ -140,8 +140,10 @@ Result<int32_t> StringUtils::StringToDate(const std::string& str) {
     if (ss.fail()) {
         return Status::Invalid(fmt::format("failed to convert string '{}' to date", str));
     }
+    int32_t orig_mon = timeinfo.tm_mon;
+    int32_t orig_mday = timeinfo.tm_mday;
     std::time_t time = timegm(&timeinfo);
-    if (time == -1) {
+    if (time == -1 || timeinfo.tm_mon != orig_mon || timeinfo.tm_mday != orig_mday) {
         return Status::Invalid(fmt::format("failed to convert string '{}' to date", str));
     }
     static const int64_t SECONDS_PER_DAY = 86400l;  // = 24 * 60 * 60
@@ -207,8 +209,10 @@ Result<int64_t> StringUtils::StringToTimestampMillis(const std::string& str) {
                         str));
     }
 
+    int32_t orig_mon = timeinfo.tm_mon;
+    int32_t orig_mday = timeinfo.tm_mday;
     std::time_t time = mktime(&timeinfo);
-    if (time == -1) {
+    if (time == -1 || timeinfo.tm_mon != orig_mon || timeinfo.tm_mday != orig_mday) {
         return Status::Invalid(fmt::format("failed to convert string '{}' to timestamp", str));
     }
     return static_cast<int64_t>(time) * 1000 + millis_part;
diff --git a/src/paimon/common/utils/string_utils_test.cpp b/src/paimon/common/utils/string_utils_test.cpp
index b4190d309..ec5811151 100644
--- a/src/paimon/common/utils/string_utils_test.cpp
+++ b/src/paimon/common/utils/string_utils_test.cpp
@@ -406,6 +406,8 @@ TEST_F(StringUtilsTest, TestStringToDate) {
     ASSERT_NOK(StringUtils::StringToDate("-1970-01-02"));
     ASSERT_NOK(StringUtils::StringToDate(""));
     ASSERT_NOK(StringUtils::StringToDate("1970-XX-02"));
+    ASSERT_NOK(StringUtils::StringToDate("2023-02-29"));
+    ASSERT_NOK(StringUtils::StringToDate("2024-00-01"));
 }
 
 TEST_F(StringUtilsTest, TestStringToTimestampMillis) {
@@ -453,6 +455,8 @@ TEST_F(StringUtilsTest, TestStringToTimestampMillis) {
     ASSERT_NOK(StringUtils::StringToTimestampMillis("2023-06-01 00:00:00.12 "));
     // Trailing dot with no digits
     ASSERT_NOK(StringUtils::StringToTimestampMillis("2023-06-01 00:00:00."));
+    ASSERT_NOK(StringUtils::StringToTimestampMillis("2023-02-29 12:00:00"));
+    ASSERT_NOK(StringUtils::StringToTimestampMillis("2024-00-15 00:00:00"));
 }
 
 TEST_F(StringUtilsTest, TestVectorToString) {