Upgrade simdjson to 3.8.0 (from 3.2.0) (#9456)

Summary:

[simdjson] third-party library import

Reviewed By: Yuhta

Differential Revision: D56031937

CMake/resolve_dependency_modules/README.md

@@ -39,6 +39,7 @@ by Velox. See details on bundling below.
 | fbthrift          | v2024.04.01.00  | No       |
 | DuckDB (testing)  | 0.8.1           | Yes      |
 | cpr (testing)     | 1.10.15         | Yes      |
+| simdjson          | 3.8.0           | Yes      |
 
 # Bundled Dependency Management
 This module provides a dependency management system that allows us to automatically fetch and build dependencies from source if needed.
@@ -75,4 +76,3 @@ Ideally all patches should be upstream when possible and removed once merged.
 ## Specify a custom url/file path for an offline build
 
 Set environment variables `VELOX_<PACKAGE>_URL` to specify a custom dependency url or local tar file path, an optional sha256 checksum can be provided as `VELOX_<PACKAGE>_SHA256`.
-

CMake/resolve_dependency_modules/simdjson.cmake

@@ -13,7 +13,7 @@
 # limitations under the License.
 include_guard(GLOBAL)
 
-set(VELOX_SIMDJSON_VERSION 3.2.0)
+set(VELOX_SIMDJSON_VERSION 3.8.0)
 set(VELOX_SIMDJSON_BUILD_SHA256_CHECKSUM
     75a684dbbe38cf72b8b3bdbdc430764813f3615899a6029931c26ddd89812da4)
 set(VELOX_SIMDJSON_SOURCE_URL

CMakeLists.txt

@@ -437,7 +437,7 @@ if(${VELOX_BUILD_PYTHON_PACKAGE})
 endif()
 
 set_source(simdjson)
-resolve_dependency(simdjson 3.2.0)
+resolve_dependency(simdjson 3.8.0)
 
 # Locate or build folly.
 add_compile_definitions(FOLLY_HAVE_INT128_T=1)

velox/functions/prestosql/tests/JsonCastTest.cpp

@@ -828,7 +828,7 @@ TEST_F(JsonCastTest, toInteger) {
       JSON(),
       BIGINT(),
       {"233897314173811950000"_sv},
-      "Problem while parsing a number");
+      "BIGINT_ERROR: Big integer value that cannot be represented using 64 bits");
 }
 
 TEST_F(JsonCastTest, toDouble) {

velox/functions/prestosql/types/JsonType.cpp

@@ -699,6 +699,9 @@ struct CastFromJsonTypedImpl {
             case simdjson::ondemand::number_type::unsigned_integer:
               w = num.get_uint64() != 0;
               break;
+            case simdjson::ondemand::number_type::big_integer:
+              VELOX_UNREACHABLE(); // value.get_number() would have failed
+                                   // already.
           }
           break;
         }
@@ -919,6 +922,9 @@ struct CastFromJsonTypedImpl {
             return convertIfInRange<T>(num.get_int64(), writer);
           case simdjson::ondemand::number_type::unsigned_integer:
             return simdjson::NUMBER_OUT_OF_RANGE;
+          case simdjson::ondemand::number_type::big_integer:
+            VELOX_UNREACHABLE(); // value.get_number() would have failed
+                                 // already.
         }
         break;
       }