2022-08-23 09:50:02 +00:00
|
|
|
#include <Analyzer/UnionNode.h>
|
|
|
|
|
|
|
|
#include <Common/SipHash.h>
|
2022-08-31 15:21:17 +00:00
|
|
|
#include <Common/FieldVisitorToString.h>
|
2022-08-23 09:50:02 +00:00
|
|
|
|
|
|
|
#include <IO/WriteBuffer.h>
|
|
|
|
#include <IO/WriteHelpers.h>
|
|
|
|
#include <IO/Operators.h>
|
|
|
|
|
|
|
|
#include <Parsers/ASTExpressionList.h>
|
|
|
|
#include <Parsers/ASTTablesInSelectQuery.h>
|
|
|
|
#include <Parsers/ASTSubquery.h>
|
|
|
|
#include <Parsers/ASTSelectQuery.h>
|
|
|
|
#include <Parsers/ASTSelectWithUnionQuery.h>
|
|
|
|
#include <Parsers/ASTIdentifier.h>
|
|
|
|
#include <Parsers/ASTFunction.h>
|
|
|
|
|
|
|
|
#include <Core/ColumnWithTypeAndName.h>
|
2022-12-06 09:44:38 +00:00
|
|
|
#include <Core/NamesAndTypes.h>
|
2022-08-23 09:50:02 +00:00
|
|
|
|
|
|
|
#include <DataTypes/getLeastSupertype.h>
|
|
|
|
|
2022-12-06 09:44:38 +00:00
|
|
|
#include <Interpreters/Context.h>
|
|
|
|
|
2022-08-23 09:50:02 +00:00
|
|
|
#include <Analyzer/QueryNode.h>
|
|
|
|
#include <Analyzer/Utils.h>
|
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
|
|
|
|
namespace ErrorCodes
|
|
|
|
{
|
|
|
|
extern const int TYPE_MISMATCH;
|
2022-10-27 09:26:53 +00:00
|
|
|
extern const int BAD_ARGUMENTS;
|
2022-08-23 09:50:02 +00:00
|
|
|
}
|
|
|
|
|
2022-12-06 09:44:38 +00:00
|
|
|
UnionNode::UnionNode(ContextMutablePtr context_, SelectUnionMode union_mode_)
|
2022-10-07 10:44:28 +00:00
|
|
|
: IQueryTreeNode(children_size)
|
2022-12-06 09:44:38 +00:00
|
|
|
, context(std::move(context_))
|
2022-10-26 14:47:40 +00:00
|
|
|
, union_mode(union_mode_)
|
2022-08-23 09:50:02 +00:00
|
|
|
{
|
2022-10-26 14:47:40 +00:00
|
|
|
if (union_mode == SelectUnionMode::UNION_DEFAULT ||
|
|
|
|
union_mode == SelectUnionMode::EXCEPT_DEFAULT ||
|
|
|
|
union_mode == SelectUnionMode::INTERSECT_DEFAULT)
|
2022-10-31 10:45:12 +00:00
|
|
|
throw Exception(ErrorCodes::BAD_ARGUMENTS, "UNION mode {} must be normalized", toString(union_mode));
|
2022-10-26 14:47:40 +00:00
|
|
|
|
2022-08-23 09:50:02 +00:00
|
|
|
children[queries_child_index] = std::make_shared<ListNode>();
|
|
|
|
}
|
|
|
|
|
2022-08-31 15:21:17 +00:00
|
|
|
NamesAndTypes UnionNode::computeProjectionColumns() const
|
2022-08-23 09:50:02 +00:00
|
|
|
{
|
|
|
|
std::vector<NamesAndTypes> projections;
|
|
|
|
|
|
|
|
NamesAndTypes query_node_projection;
|
|
|
|
|
2022-10-17 17:46:30 +00:00
|
|
|
const auto & query_nodes = getQueries().getNodes();
|
|
|
|
projections.reserve(query_nodes.size());
|
|
|
|
|
2022-08-23 09:50:02 +00:00
|
|
|
for (const auto & query_node : query_nodes)
|
|
|
|
{
|
|
|
|
if (auto * query_node_typed = query_node->as<QueryNode>())
|
2022-10-19 10:25:27 +00:00
|
|
|
query_node_projection = query_node_typed->getProjectionColumns();
|
2022-08-23 09:50:02 +00:00
|
|
|
else if (auto * union_node_typed = query_node->as<UnionNode>())
|
2022-10-19 10:25:27 +00:00
|
|
|
query_node_projection = union_node_typed->computeProjectionColumns();
|
2022-08-23 09:50:02 +00:00
|
|
|
|
|
|
|
projections.push_back(query_node_projection);
|
|
|
|
|
|
|
|
if (query_node_projection.size() != projections.front().size())
|
|
|
|
throw Exception(ErrorCodes::TYPE_MISMATCH, "UNION different number of columns in queries");
|
|
|
|
}
|
|
|
|
|
2022-08-31 15:21:17 +00:00
|
|
|
NamesAndTypes result_columns;
|
2022-08-23 09:50:02 +00:00
|
|
|
|
|
|
|
size_t projections_size = projections.size();
|
|
|
|
DataTypes projection_column_types;
|
|
|
|
projection_column_types.resize(projections_size);
|
|
|
|
|
|
|
|
size_t columns_size = query_node_projection.size();
|
|
|
|
for (size_t column_index = 0; column_index < columns_size; ++column_index)
|
|
|
|
{
|
|
|
|
for (size_t projection_index = 0; projection_index < projections_size; ++projection_index)
|
|
|
|
projection_column_types[projection_index] = projections[projection_index][column_index].type;
|
|
|
|
|
|
|
|
auto result_type = getLeastSupertype(projection_column_types);
|
|
|
|
result_columns.emplace_back(projections.front()[column_index].name, std::move(result_type));
|
|
|
|
}
|
|
|
|
|
|
|
|
return result_columns;
|
|
|
|
}
|
|
|
|
|
2023-11-09 09:44:32 +00:00
|
|
|
void UnionNode::removeUnusedProjectionColumns(const std::unordered_set<std::string> & used_projection_columns)
|
|
|
|
{
|
|
|
|
auto projection_columns = computeProjectionColumns();
|
|
|
|
size_t projection_columns_size = projection_columns.size();
|
|
|
|
std::unordered_set<size_t> used_projection_column_indexes;
|
|
|
|
|
|
|
|
for (size_t i = 0; i < projection_columns_size; ++i)
|
|
|
|
{
|
|
|
|
const auto & projection_column = projection_columns[i];
|
|
|
|
if (used_projection_columns.contains(projection_column.name))
|
|
|
|
used_projection_column_indexes.insert(i);
|
|
|
|
}
|
|
|
|
|
|
|
|
auto & query_nodes = getQueries().getNodes();
|
|
|
|
for (auto & query_node : query_nodes)
|
|
|
|
{
|
|
|
|
if (auto * query_node_typed = query_node->as<QueryNode>())
|
|
|
|
query_node_typed->removeUnusedProjectionColumns(used_projection_column_indexes);
|
|
|
|
else if (auto * union_node_typed = query_node->as<UnionNode>())
|
|
|
|
union_node_typed->removeUnusedProjectionColumns(used_projection_column_indexes);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void UnionNode::removeUnusedProjectionColumns(const std::unordered_set<size_t> & used_projection_columns_indexes)
|
|
|
|
{
|
|
|
|
auto & query_nodes = getQueries().getNodes();
|
|
|
|
for (auto & query_node : query_nodes)
|
|
|
|
{
|
|
|
|
if (auto * query_node_typed = query_node->as<QueryNode>())
|
|
|
|
query_node_typed->removeUnusedProjectionColumns(used_projection_columns_indexes);
|
|
|
|
else if (auto * union_node_typed = query_node->as<UnionNode>())
|
|
|
|
union_node_typed->removeUnusedProjectionColumns(used_projection_columns_indexes);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-23 09:50:02 +00:00
|
|
|
void UnionNode::dumpTreeImpl(WriteBuffer & buffer, FormatState & format_state, size_t indent) const
|
|
|
|
{
|
|
|
|
buffer << std::string(indent, ' ') << "UNION id: " << format_state.getNodeId(this);
|
|
|
|
|
|
|
|
if (hasAlias())
|
|
|
|
buffer << ", alias: " << getAlias();
|
|
|
|
|
2022-10-19 10:25:27 +00:00
|
|
|
if (is_subquery)
|
|
|
|
buffer << ", is_subquery: " << is_subquery;
|
|
|
|
|
|
|
|
if (is_cte)
|
|
|
|
buffer << ", is_cte: " << is_cte;
|
2022-08-23 09:50:02 +00:00
|
|
|
|
|
|
|
if (!cte_name.empty())
|
|
|
|
buffer << ", cte_name: " << cte_name;
|
|
|
|
|
2022-09-06 16:46:30 +00:00
|
|
|
buffer << ", union_mode: " << toString(union_mode);
|
2022-08-23 09:50:02 +00:00
|
|
|
|
|
|
|
buffer << '\n' << std::string(indent + 2, ' ') << "QUERIES\n";
|
|
|
|
getQueriesNode()->dumpTreeImpl(buffer, format_state, indent + 4);
|
|
|
|
}
|
|
|
|
|
|
|
|
bool UnionNode::isEqualImpl(const IQueryTreeNode & rhs) const
|
|
|
|
{
|
|
|
|
const auto & rhs_typed = assert_cast<const UnionNode &>(rhs);
|
2022-08-31 15:21:17 +00:00
|
|
|
|
2022-10-19 10:25:27 +00:00
|
|
|
return is_subquery == rhs_typed.is_subquery && is_cte == rhs_typed.is_cte && cte_name == rhs_typed.cte_name &&
|
2022-10-26 14:47:40 +00:00
|
|
|
union_mode == rhs_typed.union_mode;
|
2022-08-23 09:50:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void UnionNode::updateTreeHashImpl(HashState & state) const
|
|
|
|
{
|
|
|
|
state.update(is_subquery);
|
|
|
|
state.update(is_cte);
|
|
|
|
|
|
|
|
state.update(cte_name.size());
|
|
|
|
state.update(cte_name);
|
2022-08-31 15:21:17 +00:00
|
|
|
|
2022-10-19 10:25:27 +00:00
|
|
|
state.update(static_cast<size_t>(union_mode));
|
2022-08-23 09:50:02 +00:00
|
|
|
}
|
|
|
|
|
2022-10-10 10:25:58 +00:00
|
|
|
QueryTreeNodePtr UnionNode::cloneImpl() const
|
|
|
|
{
|
2022-12-06 09:44:38 +00:00
|
|
|
auto result_union_node = std::make_shared<UnionNode>(context, union_mode);
|
2022-10-19 10:25:27 +00:00
|
|
|
|
|
|
|
result_union_node->is_subquery = is_subquery;
|
|
|
|
result_union_node->is_cte = is_cte;
|
|
|
|
result_union_node->cte_name = cte_name;
|
|
|
|
|
|
|
|
return result_union_node;
|
2022-10-10 10:25:58 +00:00
|
|
|
}
|
|
|
|
|
2023-03-14 09:14:58 +00:00
|
|
|
ASTPtr UnionNode::toASTImpl(const ConvertToASTOptions & options) const
|
2022-08-23 09:50:02 +00:00
|
|
|
{
|
|
|
|
auto select_with_union_query = std::make_shared<ASTSelectWithUnionQuery>();
|
|
|
|
select_with_union_query->union_mode = union_mode;
|
2022-10-26 14:47:40 +00:00
|
|
|
select_with_union_query->is_normalized = true;
|
2023-03-07 20:39:26 +00:00
|
|
|
select_with_union_query->children.push_back(getQueriesNode()->toAST(options));
|
2022-08-23 09:50:02 +00:00
|
|
|
select_with_union_query->list_of_selects = select_with_union_query->children.back();
|
|
|
|
|
2023-02-10 13:29:44 +00:00
|
|
|
if (is_subquery)
|
|
|
|
{
|
2024-02-18 04:38:34 +00:00
|
|
|
auto subquery = std::make_shared<ASTSubquery>(std::move(select_with_union_query));
|
2023-02-10 13:29:44 +00:00
|
|
|
subquery->cte_name = cte_name;
|
|
|
|
return subquery;
|
|
|
|
}
|
|
|
|
|
2022-08-23 09:50:02 +00:00
|
|
|
return select_with_union_query;
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|