Bump simdjson to 3.9.4

Signed-off-by: wyb <wybb86@gmail.com>

be/src/formats/json/numeric_column.cpp

@@ -87,6 +87,25 @@ static Status add_column_with_numeric_value(FixedLengthColumn<T>* column, const
         return Status::OK();
     }
 
+    case simdjson::ondemand::number_type::big_integer: {
+        auto s = value->raw_json_token();
+        StringParser::ParseResult r;
+        auto in = StringParser::string_to_int<int128_t>(s.data(), s.size(), &r);
+        if (r != StringParser::PARSE_SUCCESS) {
+            auto err_msg = strings::Substitute("Fail to convert big integer. column=$0, value=$1", name, s);
+            return Status::InvalidArgument(err_msg);
+        }
+
+        T out{};
+        if (!checked_cast(in, &out)) {
+            column->append_numbers(&out, sizeof(out));
+        } else {
+            auto err_msg = strings::Substitute("Value is overflow. column=$0, value=$1", name, in);
+            return Status::InvalidArgument(err_msg);
+        }
+        return Status::OK();
+    }
+
     case simdjson::ondemand::number_type::floating_point_number: {
         double in = value->get_double();
         T out{};

be/test/formats/json/numeric_column_test.cpp

@@ -103,7 +103,7 @@ TEST_F(AddNumericColumnTest, test_add_int64_overflow) {
     auto doc = parser.iterate(json);
     simdjson::ondemand::value val = doc.find_field("f_int64");
     auto st = add_numeric_column<int64_t>(column.get(), t, "f_int64", &val);
-    ASSERT_TRUE(st.is_data_quality_error());
+    ASSERT_TRUE(st.is_invalid_argument());
 }
 
 TEST_F(AddNumericColumnTest, test_add_int64_overflow2) {
@@ -119,6 +119,19 @@ TEST_F(AddNumericColumnTest, test_add_int64_overflow2) {
     ASSERT_TRUE(st.is_invalid_argument());
 }
 
+TEST_F(AddNumericColumnTest, test_add_int64_overflow3) {
+    auto column = FixedLengthColumn<int64_t>::create();
+    TypeDescriptor t(TYPE_BIGINT);
+
+    simdjson::ondemand::parser parser;
+    auto json = R"(  { "f_int64": 18446744073709551616} )"_padded;
+    auto doc = parser.iterate(json);
+    simdjson::ondemand::value val = doc.find_field("f_int64");
+
+    auto st = add_numeric_column<int64_t>(column.get(), t, "f_int64", &val);
+    ASSERT_TRUE(st.is_invalid_argument());
+}
+
 TEST_F(AddNumericColumnTest, test_add_int128) {
     auto column = FixedLengthColumn<int128_t>::create();
     TypeDescriptor t(TYPE_LARGEINT);
@@ -134,9 +147,7 @@ TEST_F(AddNumericColumnTest, test_add_int128) {
     ASSERT_EQ("[9223372036854775808]", column->debug_string());
 }
 
-// Currently simdjson can not parse number < -9223372036854775808 (lower bound of int64_t)
-// or > 18446744073709551615 (upper bound of uint64_t)
-TEST_F(AddNumericColumnTest, test_add_int128_invalid) {
+TEST_F(AddNumericColumnTest, test_add_int128_big_integer) {
     auto column = FixedLengthColumn<int128_t>::create();
     TypeDescriptor t(TYPE_LARGEINT);
 
@@ -146,14 +157,17 @@ TEST_F(AddNumericColumnTest, test_add_int128_invalid) {
     simdjson::ondemand::value val = doc.find_field("f_int128");
 
     auto st = add_numeric_column<int128_t>(column.get(), t, "f_int128", &val);
-    ASSERT_TRUE(st.is_data_quality_error());
+    ASSERT_TRUE(st.ok());
+    ASSERT_EQ("[-9223372036854775809]", column->debug_string());
 
+    column->reset_column();
     json = R"(  { "f_int128": 18446744073709551616} )"_padded;
     doc = parser.iterate(json);
     val = doc.find_field("f_int128");
 
     st = add_numeric_column<int128_t>(column.get(), t, "f_int128", &val);
-    ASSERT_TRUE(st.is_data_quality_error());
+    ASSERT_TRUE(st.ok());
+    ASSERT_EQ("[18446744073709551616]", column->debug_string());
 }
 
 } // namespace starrocks

be/test/formats/json/struct_column_test.cpp

@@ -55,6 +55,22 @@ TEST_F(AddStructColumnTest, test_bad_json) {
     EXPECT_EQ("{key1:'foo',key2:'bar',key3:NULL}", column->debug_string());
 }
 
+TEST_F(AddStructColumnTest, test_bad_json2) {
+    TypeDescriptor type_desc = TypeDescriptor::create_struct_type(
+            {"key1", "key2", "key3"},{TypeDescriptor::create_varchar_type(10), TypeDescriptor::create_json_type(),
+                                      TypeDescriptor::create_varchar_type(10)});
+    auto column = ColumnHelper::create_column(type_desc, false);
+
+    simdjson::ondemand::parser parser;
+    auto json = R"(  { "key1": "foo", "key2": {a:1,b:2}, "key3": "bar"}   )"_padded;
+    auto doc = parser.iterate(json);
+    simdjson::ondemand::value val = doc.get_value();
+
+    EXPECT_OK(add_struct_column(column.get(), type_desc, "root_key", &val));
+
+    EXPECT_EQ("{key1:'foo',key2:NULL,key3:NULL}", column->debug_string());
+}
+
 TEST_F(AddStructColumnTest, test_field_not_found) {
     TypeDescriptor type_desc = TypeDescriptor::create_struct_type(
             {"key1", "key2"}, {TypeDescriptor::create_varchar_type(10), TypeDescriptor::create_varchar_type(10)});

thirdparty/vars.sh

@@ -63,7 +63,7 @@ export TP_JAR_DIR=$TP_INSTALL_DIR/lib/jar
 # Definitions for architecture-related thirdparty
 MACHINE_TYPE=$(uname -m)
 # handle mac m1 platform, change arm64 to aarch64
-if [[ "${MACHINE_TYPE}" == "arm64" ]]; then 
+if [[ "${MACHINE_TYPE}" == "arm64" ]]; then
     MACHINE_TYPE="aarch64"
 fi
 
@@ -163,10 +163,10 @@ RAPIDJSON_SOURCE=rapidjson-1.1.0
 RAPIDJSON_MD5SUM="badd12c511e081fec6c89c43a7027bce"
 
 # simdjson
-SIMDJSON_DOWNLOAD="https://github.com/simdjson/simdjson/archive/refs/tags/v2.2.0.tar.gz"
-SIMDJSON_NAME=simdjson-v2.2.0.tar.gz
-SIMDJSON_SOURCE=simdjson-2.2.0
-SIMDJSON_MD5SUM="9bd0ced53281484d8842a9429065943d"
+SIMDJSON_DOWNLOAD="https://github.com/simdjson/simdjson/archive/refs/tags/v3.9.4.tar.gz"
+SIMDJSON_NAME=simdjson-v3.9.4.tar.gz
+SIMDJSON_SOURCE=simdjson-3.9.4
+SIMDJSON_MD5SUM="bdc1dfcb2a89dc0c09e8370808a946f5"
 
 # curl
 CURL_DOWNLOAD="https://curl.se/download/curl-8.4.0.tar.gz"