yiguolei commented on code in PR #52676:
URL: https://github.com/apache/doris/pull/52676#discussion_r2179694411
##########
be/src/vec/data_types/serde/data_type_object_serde.cpp:
##########
@@ -210,27 +210,44 @@ Status DataTypeVariantSerDe::write_column_to_orc(const
std::string& timezone, co
std::vector<StringRef>&
buffer_list) const {
const auto* var = check_and_get_column<ColumnVariant>(column);
orc::StringVectorBatch* cur_batch =
dynamic_cast<orc::StringVectorBatch*>(orc_col_batch);
-
- INIT_MEMORY_FOR_ORC_WRITER()
-
+ // First pass: calculate total memory needed and collect serialized values
+ std::vector<std::string> serialized_values;
+ std::vector<size_t> valid_row_indices;
+ size_t total_size = 0;
for (size_t row_id = start; row_id < end; row_id++) {
if (cur_batch->notNull[row_id] == 1) {
auto serialized_value = std::make_unique<std::string>();
if (!var->serialize_one_row_to_string(row_id,
serialized_value.get())) {
throw doris::Exception(ErrorCode::INTERNAL_ERROR, "Failed to
serialize variant {}",
var->dump_structure());
}
- auto len = serialized_value->length();
-
- REALLOC_MEMORY_FOR_ORC_WRITER()
-
- memcpy(const_cast<char*>(bufferRef.data) + offset,
serialized_value->data(), len);
- cur_batch->data[row_id] = const_cast<char*>(bufferRef.data) +
offset;
- cur_batch->length[row_id] = len;
- offset += len;
+ size_t len = serialized_value->length();
+ total_size += len;
+ serialized_values.push_back(std::move(*serialized_value));
+ valid_row_indices.push_back(row_id);
}
}
-
+ // Allocate continues memory based on calculated size
+ char* ptr = (char*)malloc(total_size);
+ if (!ptr) {
+ return Status::InternalError(
+ "malloc memory error when write variant column data to orc
file.");
+ }
+ StringRef bufferRef;
+ bufferRef.data = ptr;
+ bufferRef.size = total_size;
+ buffer_list.emplace_back(bufferRef);
+ // Second pass: copy data to allocated memory
+ size_t offset = 0;
+ for (size_t i = 0; i < serialized_values.size(); i++) {
+ const auto& serialized_value = serialized_values[i];
+ size_t row_id = valid_row_indices[i];
+ size_t len = serialized_value.length();
+ memcpy(const_cast<char*>(bufferRef.data) + offset,
serialized_value.data(), len);
+ cur_batch->data[row_id] = const_cast<char*>(bufferRef.data) + offset;
Review Comment:
if (offset+len > total size) 应该报错
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
