[feature-wip](array-type) explode support more sub types (#10673)

1. explode support more sub types;
2. explode support nullable elements;

Co-authored-by: cambyzju <zhuxiaoli01@baidu.com>
This commit is contained in:
camby
2022-07-17 18:08:30 +08:00
committed by GitHub
parent 5793cb11d0
commit 09d19e3f0f
10 changed files with 341 additions and 40 deletions

View File

@ -27,24 +27,16 @@ VExplodeTableFunction::VExplodeTableFunction() {
Status VExplodeTableFunction::process_init(vectorized::Block* block) {
CHECK(_vexpr_context->root()->children().size() == 1)
<< "VExplodeTableFunction must be have 1 children but have "
<< "VExplodeTableFunction only support 1 child but has "
<< _vexpr_context->root()->children().size();
int value_column_idx = -1;
_vexpr_context->root()->children()[0]->execute(_vexpr_context, block, &value_column_idx);
if (block->get_by_position(value_column_idx).column->is_nullable()) {
auto array_nullable_column = check_and_get_column<ColumnNullable>(
*block->get_by_position(value_column_idx).column);
_array_null_map = array_nullable_column->get_null_map_column().get_data().data();
_array_column =
check_and_get_column<ColumnArray>(array_nullable_column->get_nested_column_ptr());
} else {
_array_null_map = nullptr;
_array_column =
check_and_get_column<ColumnArray>(*block->get_by_position(value_column_idx).column);
}
if (!_array_column) {
_array_column =
block->get_by_position(value_column_idx).column->convert_to_full_column_if_const();
if (!extract_column_array_info(*_array_column, _detail)) {
return Status::NotSupported("column type {} not supported now",
block->get_by_position(value_column_idx).column->get_name());
}
@ -56,26 +48,22 @@ Status VExplodeTableFunction::process_row(size_t row_idx) {
DCHECK(row_idx < _array_column->size());
_is_current_empty = false;
_eos = false;
_cur_offset = 0;
_array_offset = (*_detail.offsets_ptr)[row_idx - 1];
_cur_size = (*_detail.offsets_ptr)[row_idx] - _array_offset;
if (_array_null_map && _array_null_map[row_idx]) {
// array is NULL, or array is empty
if (_cur_size == 0 || (_detail.array_nullmap_data && _detail.array_nullmap_data[row_idx])) {
_is_current_empty = true;
_cur_size = 0;
_cur_offset = 0;
_pos = 0;
} else {
_cur_size =
_array_column->get_offsets()[row_idx] - _array_column->get_offsets()[row_idx - 1];
_cur_offset = 0;
_is_current_empty = (_cur_size == 0);
_pos = _array_column->get_offsets()[row_idx - 1];
}
return Status::OK();
}
Status VExplodeTableFunction::process_close() {
_array_column = nullptr;
_array_null_map = nullptr;
_pos = 0;
_detail.reset();
_array_offset = 0;
return Status::OK();
}
@ -91,7 +79,13 @@ Status VExplodeTableFunction::get_value(void** output) {
return Status::OK();
}
*output = const_cast<char*>(_array_column->get_data().get_data_at(_pos + _cur_offset).data);
size_t pos = _array_offset + _cur_offset;
if (_detail.nested_nullmap_data && _detail.nested_nullmap_data[pos]) {
*output = nullptr;
} else {
*output = const_cast<char*>(_detail.nested_col->get_data_at(pos).data);
}
return Status::OK();
}
@ -101,7 +95,13 @@ Status VExplodeTableFunction::get_value_length(int64_t* length) {
return Status::OK();
}
*length = _array_column->get_data().get_data_at(_pos + _cur_offset).size;
size_t pos = _array_offset + _cur_offset;
if (_detail.nested_nullmap_data && _detail.nested_nullmap_data[pos]) {
*length = 0;
} else {
*length = _detail.nested_col->get_data_at(pos).size;
}
return Status::OK();
}