Copilot commented on code in PR #9340:
URL: https://github.com/apache/incubator-gluten/pull/9340#discussion_r2046693130
##########
cpp-ch/local-engine/Parser/RelParsers/MergeTreeRelParser.cpp:
##########
@@ -66,69 +119,160 @@ static Int64 findMinPosition(const NameSet &
condition_table_columns, const Name
return min_position;
}
-DB::QueryPlanPtr MergeTreeRelParser::parseReadRel(
- DB::QueryPlanPtr query_plan, const substrait::ReadRel & rel, const
substrait::ReadRel::ExtensionTable & extension_table)
-{
- MergeTreeTableInstance merge_tree_table(extension_table);
- // ignore snapshot id for a query
- merge_tree_table.snapshot_id = "";
- auto storage =
merge_tree_table.restoreStorage(QueryContext::globalMutableContext());
- DB::Block input;
- DB::Block original_input;
+DB::Block MergeTreeRelParser::parseMergeTreeOutput(const substrait::ReadRel &
rel, SparkStorageMergeTreePtr storage)
+{
if (rel.has_base_schema() && rel.base_schema().names_size())
+ return TypeParser::buildBlockFromNamedStruct(rel.base_schema());
+
+ NamesAndTypesList one_column_name_type;
+
one_column_name_type.push_back(storage->getInMemoryMetadataPtr()->getColumns().getAll().front());
+ LOG_DEBUG(getLogger("SerializedPlanParser"), "Try to read ({}) instead of
empty header", one_column_name_type.front().dump());
+ return toSampleBlock(one_column_name_type);
+}
+
+
+DB::Block MergeTreeRelParser::replaceDeltaNameIfNeeded(const DB::Block &
output)
+{
+ DB::ColumnsWithTypeAndName read_block;
+ for (const auto & column : output)
{
- input = TypeParser::buildBlockFromNamedStruct(rel.base_schema());
- if (input.findByName(FileMetaColumns::INPUT_FILE_NAME) != nullptr)
+ if (DELTA_META_COLUMN_MAP.contains(column.name))
+ {
+ if (auto tuple = DELTA_META_COLUMN_MAP.at(column.name);
std::get<0>(tuple).has_value())
+ {
+ auto a= std::get<1>(tuple);
+ read_block.emplace_back(ColumnWithTypeAndName(a,
std::get<0>(tuple).value()));
+ }
+
+ }
+ else
{
- // mergetree use concat(path, _part) instead of input_file_name
-
input.insert(ColumnWithTypeAndName(ColumnWithTypeAndName(std::make_shared<DataTypeString>(),
VIRTUAL_COLUMN_PART)));
+ read_block.emplace_back(column);
}
- // remove input_file_name, input_file_block_start,
input_file_block_size due to mergetree doesn't have them
- input = FileMetaColumns::removeVirtualColumns(input);
+ }
+ return DB::Block(std::move(read_block));
+}
- SparkSQLConfig sql_config = SparkSQLConfig::loadFromContext(context);
- // case_insensitive_matching
- if (!sql_config.caseSensitive)
+void MergeTreeRelParser::recoverDeltaNameIfNeeded(
+ DB::QueryPlan & plan, const DB::Block & output, const
MergeTreeTableInstance & merge_tree_table)
+{
+ const auto & header = plan.getCurrentHeader();
+ DB::ActionsDAG actions_dag(header.getNamesAndTypesList());
+ NameSet names;
+ bool need_recover = false;
+ for (const auto & column : output)
+ {
+ if (DELTA_META_COLUMN_MAP.contains(column.name))
{
- original_input = input;
- auto all =
storage->getInMemoryMetadataPtr()->getColumns().getNamesOfPhysical();
- std::ranges::for_each(
- input,
- [&all](ColumnWithTypeAndName & column)
- {
- const auto found
- = std::ranges::find_if(all, [&column](const auto &
name) -> bool { return boost::iequals(column.name, name); });
- if (found != all.end())
- column.name = *found;
- });
+ need_recover = true;
+ auto tuple = DELTA_META_COLUMN_MAP.at(column.name);
+ ReplaceDeltaNodeFunc func = std::get<2>(tuple);
+ func(actions_dag, merge_tree_table, context);
}
+
+ names.insert(column.name);
}
- else
+
+ if (!need_recover)
+ return;
+
+ actions_dag.removeUnusedActions(names);
+ auto step = std::make_unique<DB::ExpressionStep>(plan.getCurrentHeader(),
std::move(actions_dag));
+ step->setStepDescription("Recover virtual columns");
+ steps.emplace_back(step.get());
+ plan.addStep(std::move(step));
+}
+
+void MergeTreeRelParser::replaceNodeWithCaseSensitive(DB::Block & read_block,
SparkStorageMergeTreePtr storage)
+{
+ // case_insensitive_matching
+ if (spark_sql_config.caseSensitive)
+ return;
+
+ auto all =
storage->getInMemoryMetadataPtr()->getColumns().getNamesOfPhysical();
+ std::ranges::for_each(
+ read_block,
+ [&all](ColumnWithTypeAndName & column)
+ {
+ const auto found
+ = std::ranges::find_if(all, [&column](const auto & name) ->
bool { return boost::iequals(column.name, name); });
+ if (found != all.end())
+ column.name = *found;
+ });
+}
+
+
+void MergeTreeRelParser::recoverNodeWithCaseSensitive(DB::QueryPlan &
query_plan, const DB::Block & output)
+{
+ if (spark_sql_config.caseSensitive)
+ return;
+
+ auto read_Header = query_plan.getCurrentHeader();
+ NameToNameMap names;
+ names.reserve(output.columns());
+ for (const auto & elem : output.getColumnsWithTypeAndName())
+ names[Poco::toLower(elem.name)] = elem.name;
+
+ DB::NamesWithAliases aliases;
+ aliases.reserve(read_Header.columns());
+ bool need_alisa = false;
+ for (const auto & elem : read_Header)
{
- NamesAndTypesList one_column_name_type;
-
one_column_name_type.push_back(storage->getInMemoryMetadataPtr()->getColumns().getAll().front());
- input = toSampleBlock(one_column_name_type);
- LOG_DEBUG(getLogger("SerializedPlanParser"), "Try to read ({}) instead
of empty header", one_column_name_type.front().dump());
+ if (auto lower_name = Poco::toLower(elem.name);
names.contains(lower_name))
+ {
+ if (!need_alisa && !boost::equals(elem.name, names[lower_name]))
+ need_alisa = true;
Review Comment:
The variable name 'need_alisa' appears to have a spelling mistake; consider
renaming it to 'need_alias' for clarity.
```suggestion
bool need_alias = false;
for (const auto & elem : read_Header)
{
if (auto lower_name = Poco::toLower(elem.name);
names.contains(lower_name))
{
if (!need_alias && !boost::equals(elem.name, names[lower_name]))
need_alias = true;
```
--
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]
