branch-4.0: [fix](parse) No longer throws exceptions when parse datetime failed in from_olap_string #63035

be/src/vec/data_types/serde/data_type_date_or_datetime_serde.cpp

@@ -408,7 +408,15 @@ Status DataTypeDateSerDe<T>::from_olap_string(const std::string& str, Field& fie
     // Exception!
     if (!CastToDateOrDatetime::from_string_non_strict_mode<IsDatetime>(
                 StringRef(str), res, options.timezone, params)) [[unlikely]] {
-        return Status::InvalidArgument("parse date or datetime fail, string: '{}'", str);
+        // Keep the same datelike Field-parser convention as DateV2/DateTimeV2: invalid OLAP
+        // strings fall back to the type-specific minimum. VecDateTimeValue stores DateV1 and
+        // DateTimeV1 values, whose minimum valid date is FIRST_DAY.
+        res = VecDateTimeValue::FIRST_DAY;
+        if constexpr (IsDatetime) {
+            res.to_datetime();
+        } else {
+            res.cast_to_date();
+        }
     }
     field = Field::create_field<T>(std::move(res));
     return Status::OK();

be/src/vec/data_types/serde/data_type_datetimev2_serde.cpp

@@ -130,6 +130,7 @@ Status DataTypeDateTimeV2SerDe::from_olap_string(const std::string& str, Field&
     std::string date_format = "%Y-%m-%d %H:%i:%s.%f";
 
     if (!res.from_date_format_str(date_format.data(), date_format.size(), str.data(), str.size())) {
+        // On parse failure, fall back to MIN_DATETIME_V2, the packed lower-bound DateTimeV2 value.
         res = DateV2Value<DateTimeV2ValueType>(MIN_DATETIME_V2);
     }
     field = Field::create_field<TYPE_DATETIMEV2>(std::move(res));

be/src/vec/data_types/serde/data_type_datev2_serde.cpp

@@ -242,6 +242,7 @@ Status DataTypeDateV2SerDe::from_olap_string(const std::string& str, Field& fiel
                 ((time_tm.tm_year + 1900) << 9) | ((time_tm.tm_mon + 1) << 5) | time_tm.tm_mday;
         res = DateV2Value<DateV2ValueType>(value);
     } else {
+        // On parse failure, fall back to MIN_DATE_V2, the packed lower-bound DateV2 value.
         res = DateV2Value<DateV2ValueType>(MIN_DATE_V2);
     }
     field = Field::create_field<TYPE_DATEV2>(std::move(res));

be/test/olap/olap_type_test.cpp

@@ -24,7 +24,11 @@
 #include "gtest/gtest_pred_impl.h"
 #include "olap/olap_common.h"
 #include "olap/types.h"
+#include "vec/core/field.h"
+#include "vec/data_types/data_type_factory.hpp"
+#include "vec/data_types/serde/data_type_serde.h"
 #include "vec/functions/cast/cast_to_string.h"
+#include "vec/runtime/vdatetime_value.h"
 
 namespace doris {
 
@@ -39,6 +43,16 @@ class OlapTypeTest : public testing::Test {
     }
 };
 
+static void expect_from_olap_string_default(const vectorized::DataTypePtr& data_type,
+                                            const std::string& input,
+                                            const vectorized::Field& expected_field) {
+    vectorized::Field field;
+    vectorized::DataTypeSerDe::FormatOptions options;
+    auto status = data_type->get_serde()->from_olap_string(input, field, options);
+    ASSERT_TRUE(status.ok()) << data_type->get_name() << " failed: " << status.to_string();
+    EXPECT_EQ(field, expected_field) << data_type->get_name();
+}
+
 // deserialize float string serialized by old version of Doris
 TEST_F(OlapTypeTest, deser_float_old) {
     std::vector<float> normal_input_values = {
@@ -576,4 +590,31 @@ TEST_F(OlapTypeTest, ser_deser_double) {
                 << ", diff_ratio: " << fmt::format("{:.17g}", diff_ratio);
     }
 }
-} // namespace doris
+
+TEST_F(OlapTypeTest, datelike_from_olap_string_parse_failure_defaults) {
+    const std::string invalid = "not-a-valid-datelike-value";
+
+    auto datev1_default = VecDateTimeValue::FIRST_DAY;
+    datev1_default.cast_to_date();
+    expect_from_olap_string_default(
+            vectorized::DataTypeFactory::instance().create_data_type(TYPE_DATE, false), invalid,
+            vectorized::Field::create_field<TYPE_DATE>(datev1_default));
+
+    auto datetimev1_default = VecDateTimeValue::FIRST_DAY;
+    datetimev1_default.to_datetime();
+    expect_from_olap_string_default(
+            vectorized::DataTypeFactory::instance().create_data_type(TYPE_DATETIME, false), invalid,
+            vectorized::Field::create_field<TYPE_DATETIME>(datetimev1_default));
+
+    expect_from_olap_string_default(
+            vectorized::DataTypeFactory::instance().create_data_type(TYPE_DATEV2, false), invalid,
+            vectorized::Field::create_field<TYPE_DATEV2>(
+                    DateV2Value<DateV2ValueType>(MIN_DATE_V2)));
+
+    expect_from_olap_string_default(
+            vectorized::DataTypeFactory::instance().create_data_type(TYPE_DATETIMEV2, false),
+            invalid,
+            vectorized::Field::create_field<TYPE_DATETIMEV2>(
+                    DateV2Value<DateTimeV2ValueType>(MIN_DATETIME_V2)));
+}
+} // namespace doris