Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat(fuzzer): Update makeAlternatePlans to generate multi-join plans #11942

Open
wants to merge 6 commits into
base: main
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions velox/core/PlanNode.h
Original file line number Diff line number Diff line change
Expand Up @@ -324,6 +324,8 @@ class ValuesNode : public PlanNode {
const size_t repeatTimes_;
};

using ValuesNodePtr = std::shared_ptr<const ValuesNode>;

class ArrowStreamNode : public PlanNode {
public:
ArrowStreamNode(
Expand Down
4 changes: 2 additions & 2 deletions velox/exec/fuzzer/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -12,8 +12,8 @@
# See the License for the specific language governing permissions and
# limitations under the License.

add_library(velox_fuzzer_util DuckQueryRunner.cpp PrestoQueryRunner.cpp
FuzzerUtil.cpp ToSQLUtil.cpp)
add_library(velox_fuzzer_util ReferenceQueryRunner.cpp DuckQueryRunner.cpp
PrestoQueryRunner.cpp FuzzerUtil.cpp ToSQLUtil.cpp)

target_link_libraries(
velox_fuzzer_util
Expand Down
160 changes: 19 additions & 141 deletions velox/exec/fuzzer/DuckQueryRunner.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,11 @@
* See the License for the specific language governing permissions and
* limitations under the License.
*/

#include <optional>
#include <set>
#include <unordered_map>

#include "velox/exec/fuzzer/DuckQueryRunner.h"
#include "velox/exec/fuzzer/ToSQLUtil.h"
#include "velox/exec/tests/utils/QueryAssertions.h"
Expand Down Expand Up @@ -104,21 +109,22 @@ DuckQueryRunner::aggregationFunctionDataSpecs() const {

std::multiset<std::vector<velox::variant>> DuckQueryRunner::execute(
const std::string& sql,
const std::vector<RowVectorPtr>& input,
const RowTypePtr& resultType) {
const core::PlanNodePtr& plan) {
DuckDbQueryRunner queryRunner;
queryRunner.createTable("tmp", input);
return queryRunner.execute(sql, resultType);
std::unordered_map<std::string, std::vector<RowVectorPtr>> inputMap =
getAllTables(plan);
for (const auto& [tableName, input] : inputMap) {
queryRunner.createTable(tableName, input);
}
return queryRunner.execute(sql, plan->outputType());
}

std::multiset<std::vector<velox::variant>> DuckQueryRunner::execute(
const std::string& sql,
const std::vector<RowVectorPtr>& probeInput,
const std::vector<RowVectorPtr>& buildInput,
const std::vector<RowVectorPtr>& input,
const RowTypePtr& resultType) {
DuckDbQueryRunner queryRunner;
queryRunner.createTable("t", probeInput);
queryRunner.createTable("u", buildInput);
queryRunner.createTable("tmp", input);
return queryRunner.execute(sql, resultType);
}

Expand Down Expand Up @@ -164,6 +170,11 @@ std::optional<std::string> DuckQueryRunner::toSql(
return toSql(joinNode);
}

if (const auto valuesNode =
std::dynamic_pointer_cast<const core::ValuesNode>(plan)) {
return toSql(valuesNode);
}

VELOX_NYI();
}

Expand Down Expand Up @@ -340,137 +351,4 @@ std::optional<std::string> DuckQueryRunner::toSql(

return sql.str();
}

std::optional<std::string> DuckQueryRunner::toSql(
const std::shared_ptr<const core::HashJoinNode>& joinNode) {
const auto& joinKeysToSql = [](auto keys) {
std::stringstream out;
for (auto i = 0; i < keys.size(); ++i) {
if (i > 0) {
out << ", ";
}
out << keys[i]->name();
}
return out.str();
};

const auto filterToSql = [](core::TypedExprPtr filter) {
auto call = std::dynamic_pointer_cast<const core::CallTypedExpr>(filter);
return toCallSql(call);
};

const auto& joinConditionAsSql = [&](auto joinNode) {
std::stringstream out;
for (auto i = 0; i < joinNode->leftKeys().size(); ++i) {
if (i > 0) {
out << " AND ";
}
out << joinNode->leftKeys()[i]->name() << " = "
<< joinNode->rightKeys()[i]->name();
}
if (joinNode->filter()) {
out << " AND " << filterToSql(joinNode->filter());
}
return out.str();
};

const auto& outputNames = joinNode->outputType()->names();

std::stringstream sql;
if (joinNode->isLeftSemiProjectJoin()) {
sql << "SELECT "
<< folly::join(", ", outputNames.begin(), --outputNames.end());
} else {
sql << "SELECT " << folly::join(", ", outputNames);
}

switch (joinNode->joinType()) {
case core::JoinType::kInner:
sql << " FROM t INNER JOIN u ON " << joinConditionAsSql(joinNode);
break;
case core::JoinType::kLeft:
sql << " FROM t LEFT JOIN u ON " << joinConditionAsSql(joinNode);
break;
case core::JoinType::kFull:
sql << " FROM t FULL OUTER JOIN u ON " << joinConditionAsSql(joinNode);
break;
case core::JoinType::kLeftSemiFilter:
// Multiple columns returned by a scalar subquery is not supported in
// DuckDB. A scalar subquery expression is a subquery that returns one
// result row from exactly one column for every input row.
if (joinNode->leftKeys().size() > 1) {
return std::nullopt;
}
sql << " FROM t WHERE " << joinKeysToSql(joinNode->leftKeys())
<< " IN (SELECT " << joinKeysToSql(joinNode->rightKeys())
<< " FROM u";
if (joinNode->filter()) {
sql << " WHERE " << filterToSql(joinNode->filter());
}
sql << ")";
break;
case core::JoinType::kLeftSemiProject:
if (joinNode->isNullAware()) {
sql << ", " << joinKeysToSql(joinNode->leftKeys()) << " IN (SELECT "
<< joinKeysToSql(joinNode->rightKeys()) << " FROM u";
if (joinNode->filter()) {
sql << " WHERE " << filterToSql(joinNode->filter());
}
sql << ") FROM t";
} else {
sql << ", EXISTS (SELECT * FROM u WHERE "
<< joinConditionAsSql(joinNode);
sql << ") FROM t";
}
break;
case core::JoinType::kAnti:
if (joinNode->isNullAware()) {
sql << " FROM t WHERE " << joinKeysToSql(joinNode->leftKeys())
<< " NOT IN (SELECT " << joinKeysToSql(joinNode->rightKeys())
<< " FROM u";
if (joinNode->filter()) {
sql << " WHERE " << filterToSql(joinNode->filter());
}
sql << ")";
} else {
sql << " FROM t WHERE NOT EXISTS (SELECT * FROM u WHERE "
<< joinConditionAsSql(joinNode);
sql << ")";
}
break;
default:
VELOX_UNREACHABLE(
"Unknown join type: {}", static_cast<int>(joinNode->joinType()));
}

return sql.str();
}

std::optional<std::string> DuckQueryRunner::toSql(
const std::shared_ptr<const core::NestedLoopJoinNode>& joinNode) {
std::stringstream sql;
sql << "SELECT " << folly::join(", ", joinNode->outputType()->names());

// Nested loop join without filter.
VELOX_CHECK(
joinNode->joinCondition() == nullptr,
"This code path should be called only for nested loop join without filter");
const std::string joinCondition{"(1 = 1)"};
switch (joinNode->joinType()) {
case core::JoinType::kInner:
sql << " FROM t INNER JOIN u ON " << joinCondition;
break;
case core::JoinType::kLeft:
sql << " FROM t LEFT JOIN u ON " << joinCondition;
break;
case core::JoinType::kFull:
sql << " FROM t FULL OUTER JOIN u ON " << joinCondition;
break;
default:
VELOX_UNREACHABLE(
"Unknown join type: {}", static_cast<int>(joinNode->joinType()));
}

return sql.str();
}
} // namespace facebook::velox::exec::test
23 changes: 11 additions & 12 deletions velox/exec/fuzzer/DuckQueryRunner.h
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,10 @@
*/
#pragma once

#include <optional>
#include <set>
#include <unordered_map>

#include "velox/exec/fuzzer/ReferenceQueryRunner.h"

namespace facebook::velox::exec::test {
Expand Down Expand Up @@ -46,20 +50,21 @@ class DuckQueryRunner : public ReferenceQueryRunner {
/// Assumes that source of AggregationNode or Window Node is 'tmp' table.
std::optional<std::string> toSql(const core::PlanNodePtr& plan) override;

/// Creates 'tmp' table with 'input' data and runs 'sql' query. Returns
/// results according to 'resultType' schema.
/// Executes SQL query returned by the 'toSql' method based on the plan.
std::multiset<std::vector<velox::variant>> execute(
const std::string& sql,
const std::vector<RowVectorPtr>& input,
const RowTypePtr& resultType) override;
const core::PlanNodePtr& plan) override;

/// Creates 'tmp' table with 'input' data and runs 'sql' query. Returns
/// results according to 'resultType' schema.
std::multiset<std::vector<velox::variant>> execute(
const std::string& sql,
const std::vector<RowVectorPtr>& probeInput,
const std::vector<RowVectorPtr>& buildInput,
const std::vector<RowVectorPtr>& input,
const RowTypePtr& resultType) override;

private:
using ReferenceQueryRunner::toSql;

std::optional<std::string> toSql(
const std::shared_ptr<const core::AggregationNode>& aggregationNode);

Expand All @@ -72,12 +77,6 @@ class DuckQueryRunner : public ReferenceQueryRunner {
std::optional<std::string> toSql(
const std::shared_ptr<const core::RowNumberNode>& rowNumberNode);

std::optional<std::string> toSql(
const std::shared_ptr<const core::HashJoinNode>& joinNode);

std::optional<std::string> toSql(
const std::shared_ptr<const core::NestedLoopJoinNode>& joinNode);

std::unordered_set<std::string> aggregateFunctionNames_;
};

Expand Down
Loading
Loading