jacktengg commented on code in PR #43624:
URL: https://github.com/apache/doris/pull/43624#discussion_r1838038970
##########
be/src/vec/columns/column_string.cpp:
##########
@@ -134,34 +133,41 @@ void ColumnStr<T>::insert_range_from(const IColumn& src,
size_t start, size_t le
if (length == 0) {
return;
}
+ auto do_insert = [&](const auto& src_concrete) {
+ const auto& src_offsets = src_concrete.get_offsets();
+ const auto& src_chars = src_concrete.get_chars();
+ if (start + length > src_offsets.size()) {
+ throw doris::Exception(
+ doris::ErrorCode::INTERNAL_ERROR,
+ "Parameter out of bound in
IColumnStr<T>::insert_range_from method.");
+ }
+ size_t nested_offset = src_offsets[static_cast<ssize_t>(start) - 1];
+ size_t nested_length = src_offsets[start + length - 1] - nested_offset;
- const auto& src_concrete = assert_cast<const ColumnStr<T>&>(src);
-
- if (start + length > src_concrete.offsets.size()) {
- throw doris::Exception(
- doris::ErrorCode::INTERNAL_ERROR,
- "Parameter out of bound in IColumnStr<T>::insert_range_from
method.");
- }
-
- size_t nested_offset = src_concrete.offset_at(start);
- size_t nested_length = src_concrete.offsets[start + length - 1] -
nested_offset;
-
- size_t old_chars_size = chars.size();
- check_chars_length(old_chars_size + nested_length, offsets.size() +
length);
- chars.resize(old_chars_size + nested_length);
- memcpy(&chars[old_chars_size], &src_concrete.chars[nested_offset],
nested_length);
+ size_t old_chars_size = chars.size();
+ check_chars_length(old_chars_size + nested_length, offsets.size() +
length);
+ chars.resize(old_chars_size + nested_length);
+ memcpy(&chars[old_chars_size], &src_chars[nested_offset],
nested_length);
- if (start == 0 && offsets.empty()) {
- offsets.assign(src_concrete.offsets.begin(),
src_concrete.offsets.begin() + length);
- } else {
- size_t old_size = offsets.size();
- size_t prev_max_offset = offsets.back(); /// -1th index is Ok, see
PaddedPODArray
- offsets.resize(old_size + length);
+ using OffsetsType = std::decay_t<decltype(src_offsets)>;
+ if (std::is_same_v<T, typename OffsetsType::value_type> && start == 0
&& offsets.empty()) {
+ offsets.assign(src_offsets.begin(), src_offsets.begin() + length);
+ } else {
+ size_t old_size = offsets.size();
+ size_t prev_max_offset = offsets.back(); /// -1th index is Ok, see
PaddedPODArray
+ offsets.resize(old_size + length);
- for (size_t i = 0; i < length; ++i) {
- offsets[old_size + i] =
- src_concrete.offsets[start + i] - nested_offset +
prev_max_offset;
+ for (size_t i = 0; i < length; ++i) {
+ offsets[old_size + i] = src_offsets[start + i] - nested_offset
+ prev_max_offset;
+ }
}
+ };
+ // insert_range_from maybe called by
ColumnArray::insert_indices_from(which is used by hash join operator),
+ // so we need to support both ColumnStr<uint32_t> and ColumnStr<uint64_t>
+ if (src.is_column_string64()) {
+ do_insert(assert_cast<const ColumnStr<uint64_t>&>(src));
Review Comment:
offset64 will be converted to offset32, and check_chars_length is called to
check if total len exceed 4G
--
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]
