push filter down traverse rule (#4987)

src/common/expression/PredicateExpression.cpp

@@ -19,7 +19,8 @@ std::unordered_map<std::string, PredicateExpression::Type> PredicateExpression::
 const Value& PredicateExpression::evalExists(ExpressionContext& ctx) {
   DCHECK(collection_->kind() == Expression::Kind::kAttribute ||
          collection_->kind() == Expression::Kind::kSubscript ||
-         collection_->kind() == Expression::Kind::kLabelTagProperty);
+         collection_->kind() == Expression::Kind::kLabelTagProperty)
+      << "actual kind: " << collection_->kind() << ", toString: " << toString();
 
   if (collection_->kind() == Expression::Kind::kLabelTagProperty) {
     result_ = !collection_->eval(ctx).isNull();

src/graph/optimizer/CMakeLists.txt

@@ -54,6 +54,7 @@ nebula_add_library(
     rule/EliminateAppendVerticesRule.cpp
     rule/PushLimitDownScanEdgesRule.cpp
     rule/RemoveProjectDedupBeforeGetDstBySrcRule.cpp
+    rule/PushFilterDownTraverseRule.cpp
 )
 
 nebula_add_subdirectory(test)

src/graph/optimizer/OptGroup.cpp

@@ -119,7 +119,9 @@ void OptGroup::addGroupNode(OptGroupNode *groupNode) {
   if (outputVar_.empty()) {
     outputVar_ = groupNode->node()->outputVar();
   } else {
-    DCHECK_EQ(outputVar_, groupNode->node()->outputVar());
+    DCHECK_EQ(outputVar_, groupNode->node()->outputVar())
+        << "outputVar_: " << outputVar_ << "groupNode plan: " << groupNode->node()->toString()
+        << ", groupNode->node()->outputVar(): " << groupNode->node()->outputVar();
   }
   groupNodes_.emplace_back(groupNode);
   groupNode->node()->updateSymbols();

src/graph/optimizer/rule/PushFilterDownTraverseRule.cpp

@@ -0,0 +1,152 @@
+/* Copyright (c) 2022 vesoft inc. All rights reserved.
+ *
+ * This source code is licensed under Apache 2.0 License.
+ */
+
+#include "graph/optimizer/rule/PushFilterDownTraverseRule.h"
+
+#include "common/expression/ConstantExpression.h"
+#include "common/expression/Expression.h"
+#include "graph/optimizer/OptContext.h"
+#include "graph/optimizer/OptGroup.h"
+#include "graph/planner/plan/PlanNode.h"
+#include "graph/planner/plan/Query.h"
+#include "graph/util/ExpressionUtils.h"
+#include "graph/visitor/ExtractFilterExprVisitor.h"
+
+using nebula::Expression;
+using nebula::graph::Filter;
+using nebula::graph::PlanNode;
+using nebula::graph::QueryContext;
+using nebula::graph::Traverse;
+
+namespace nebula {
+namespace opt {
+
+std::unique_ptr<OptRule> PushFilterDownTraverseRule::kInstance =
+    std::unique_ptr<PushFilterDownTraverseRule>(new PushFilterDownTraverseRule());
+
+PushFilterDownTraverseRule::PushFilterDownTraverseRule() {
+  RuleSet::QueryRules().addRule(this);
+}
+
+const Pattern& PushFilterDownTraverseRule::pattern() const {
+  static Pattern pattern =
+      Pattern::create(PlanNode::Kind::kFilter,
+                      {Pattern::create(PlanNode::Kind::kAppendVertices,
+                                       {Pattern::create(PlanNode::Kind::kTraverse)})});
+  return pattern;
+}
+
+bool PushFilterDownTraverseRule::match(OptContext* ctx, const MatchedResult& matched) const {
+  if (!OptRule::match(ctx, matched)) {
+    return false;
+  }
+  DCHECK_EQ(matched.dependencies[0].dependencies[0].node->node()->kind(),
+            PlanNode::Kind::kTraverse);
+  auto traverse =
+      static_cast<const Traverse*>(matched.dependencies[0].dependencies[0].node->node());
+  return traverse->isOneStep();
+}
+
+StatusOr<OptRule::TransformResult> PushFilterDownTraverseRule::transform(
+    OptContext* ctx, const MatchedResult& matched) const {
+  auto* filterGroupNode = matched.node;
+  auto* filterGroup = filterGroupNode->group();
+  auto* filter = static_cast<graph::Filter*>(filterGroupNode->node());
+  auto* condition = filter->condition();
+
+  auto* avGroupNode = matched.dependencies[0].node;
+  auto* av = static_cast<graph::AppendVertices*>(avGroupNode->node());
+
+  auto* tvGroupNode = matched.dependencies[0].dependencies[0].node;
+  auto* tv = static_cast<graph::Traverse*>(tvGroupNode->node());
+  auto& edgeAlias = tv->edgeAlias();
+
+  auto qctx = ctx->qctx();
+  auto pool = qctx->objPool();
+
+  // Pick the expr looks like `$-.e[0].likeness
+  auto picker = [&edgeAlias](const Expression* e) -> bool {
+    // TODO(jie): Handle the strange exists expr. e.g. exists(e.likeness)
+    auto exprs = graph::ExpressionUtils::collectAll(e, {Expression::Kind::kPredicate});
+    for (auto* expr : exprs) {
+      if (static_cast<const PredicateExpression*>(expr)->name() == "exists") {
+        return false;
+      }
+    }
+
+    auto varProps = graph::ExpressionUtils::collectAll(
+        e, {Expression::Kind::kInputProperty, Expression::Kind::kVarProperty});
+    if (varProps.empty()) {
+      return false;
+    }
+    for (auto* expr : varProps) {
+      DCHECK(graph::ExpressionUtils::isPropertyExpr(expr));
+      auto& propName = static_cast<const PropertyExpression*>(expr)->prop();
+      if (propName != edgeAlias) return false;
+    }
+    return true;
+  };
+  Expression* filterPicked = nullptr;
+  Expression* filterUnpicked = nullptr;
+  graph::ExpressionUtils::splitFilter(condition, picker, &filterPicked, &filterUnpicked);
+
+  if (!filterPicked) {
+    return TransformResult::noTransform();
+  }
+  auto* newFilterPicked =
+      graph::ExpressionUtils::rewriteEdgePropertyFilter(pool, edgeAlias, filterPicked->clone());
+
+  Filter* newFilter = nullptr;
+  OptGroupNode* newFilterGroupNode = nullptr;
+  if (filterUnpicked) {
+    newFilter = Filter::make(qctx, nullptr, filterUnpicked);
+    newFilter->setOutputVar(filter->outputVar());
+    newFilter->setColNames(filter->colNames());
+    newFilterGroupNode = OptGroupNode::create(ctx, newFilter, filterGroup);
+  }
+
+  auto* newAv = static_cast<graph::AppendVertices*>(av->clone());
+
+  OptGroupNode* newAvGroupNode = nullptr;
+  if (newFilterGroupNode) {
+    auto* newAvGroup = OptGroup::create(ctx);
+    newAvGroupNode = newAvGroup->makeGroupNode(newAv);
+    newFilterGroupNode->dependsOn(newAvGroup);
+    newFilter->setInputVar(newAv->outputVar());
+  } else {
+    newAvGroupNode = OptGroupNode::create(ctx, newAv, filterGroup);
+    newAv->setOutputVar(filter->outputVar());
+  }
+
+  auto* eFilter = tv->eFilter();
+  Expression* newEFilter = eFilter
+                               ? LogicalExpression::makeAnd(pool, newFilterPicked, eFilter->clone())
+                               : newFilterPicked;
+
+  auto* newTv = static_cast<graph::Traverse*>(tv->clone());
+  newAv->setInputVar(newTv->outputVar());
+  newTv->setEdgeFilter(newEFilter);
+
+  auto* newTvGroup = OptGroup::create(ctx);
+  newAvGroupNode->dependsOn(newTvGroup);
+  auto* newTvGroupNode = newTvGroup->makeGroupNode(newTv);
+
+  for (auto dep : tvGroupNode->dependencies()) {
+    newTvGroupNode->dependsOn(dep);
+  }
+
+  TransformResult result;
+  result.eraseCurr = true;
+  result.newGroupNodes.emplace_back(newFilterGroupNode ? newFilterGroupNode : newAvGroupNode);
+
+  return result;
+}
+
+std::string PushFilterDownTraverseRule::toString() const {
+  return "PushFilterDownTraverseRule";
+}
+
+}  // namespace opt
+}  // namespace nebula

src/graph/optimizer/rule/PushFilterDownTraverseRule.h

@@ -0,0 +1,43 @@
+/* Copyright (c) 2022 vesoft inc. All rights reserved.
+ *
+ * This source code is licensed under Apache 2.0 License.
+ */
+
+#pragma once
+
+#include "graph/optimizer/OptRule.h"
+
+namespace nebula {
+namespace opt {
+
+/*
+ * Before:
+ * Filter(e.likeness > 78)
+ *        |
+ *   AppendVertices
+ *        |
+ *     Traverse
+ *
+ * After :
+ *   AppendVertices
+ *        |
+ *     Traverse(eFilter_: *.likeness > 78)
+ */
+class PushFilterDownTraverseRule final : public OptRule {
+ public:
+  const Pattern &pattern() const override;
+
+  bool match(OptContext *ctx, const MatchedResult &matched) const override;
+
+  StatusOr<TransformResult> transform(OptContext *ctx, const MatchedResult &matched) const override;
+
+  std::string toString() const override;
+
+ private:
+  PushFilterDownTraverseRule();
+
+  static std::unique_ptr<OptRule> kInstance;
+};
+
+}  // namespace opt
+}  // namespace nebula

src/graph/planner/plan/Query.h

@@ -1538,6 +1538,10 @@ class Traverse final : public GetNeighbors {
     return range_;
   }
 
+  bool isOneStep() const {
+    return !range_;
+  }
+
   // Contains zero step
   bool zeroStep() const {
     return range_ != nullptr && range_->min() == 0;
@@ -1555,6 +1559,17 @@ class Traverse final : public GetNeighbors {
     return trackPrevPath_;
   }
 
+  const std::string& nodeAlias() const {
+    auto& cols = this->colNames();
+    DCHECK_GE(cols.size(), 2);
+    return cols[cols.size() - 2];
+  }
+
+  const std::string& edgeAlias() const {
+    DCHECK(!this->colNames().empty());
+    return this->colNames().back();
+  }
+
   void setStepRange(MatchStepRange* range) {
     range_ = range;
   }
@@ -1618,6 +1633,11 @@ class AppendVertices final : public GetVertices {
     return trackPrevPath_;
   }
 
+  const std::string nodeAlias() const {
+    DCHECK(!this->colNames().empty());
+    return this->colNames().back();
+  }
+
   void setVertexFilter(Expression* vFilter) {
     vFilter_ = vFilter;
   }

src/graph/util/ExpressionUtils.cpp

@@ -350,6 +350,7 @@ Expression *ExpressionUtils::rewriteInExpr(const Expression *expr) {
       return const_cast<Expression *>(expr);
     }
     std::vector<Expression *> operands;
+    operands.reserve(values.size());
     for (const auto &v : values) {
       operands.emplace_back(
           RelationalExpression::makeEQ(pool, inExpr->left(), ConstantExpression::make(pool, v)));
@@ -1431,5 +1432,67 @@ bool ExpressionUtils::checkExprDepth(const Expression *expr) {
   return true;
 }
 
+/*static*/
+bool ExpressionUtils::isSingleLenExpandExpr(const std::string &edgeAlias, const Expression *expr) {
+  if (expr->kind() != Expression::Kind::kAttribute) {
+    return false;
+  }
+  auto attributeExpr = static_cast<const AttributeExpression *>(expr);
+  auto *left = attributeExpr->left();
+  auto *right = attributeExpr->right();
+
+  if (left->kind() != Expression::Kind::kSubscript) return false;
+  if (right->kind() != Expression::Kind::kConstant ||
+      !static_cast<const ConstantExpression *>(right)->value().isStr())
+    return false;
+
+  auto subscriptExpr = static_cast<const SubscriptExpression *>(left);
+  auto *listExpr = subscriptExpr->left();
+  auto *idxExpr = subscriptExpr->right();
+  if (listExpr->kind() != Expression::Kind::kInputProperty &&
+      listExpr->kind() != Expression::Kind::kVarProperty) {
+    return false;
+  }
+  if (static_cast<const PropertyExpression *>(listExpr)->prop() != edgeAlias) {
+    return false;
+  }
+
+  // NOTE(jie): Just handled `$-.e[0].likeness` for now, whileas the traverse is single length
+  // expand.
+  // TODO(jie): Handle `ALL(i IN e WHERE i.likeness > 78)`, whileas the traverse is var len
+  // expand.
+  if (idxExpr->kind() != Expression::Kind::kConstant ||
+      static_cast<const ConstantExpression *>(idxExpr)->value() != 0) {
+    return false;
+  }
+  return true;
+}
+
+// Transform expression `$-.e[0].likeness` to EdgePropertyExpression `like.likeness`
+// for more friendly to push down
+// \param pool object pool to hold ownership of objects alloacted
+// \param edgeAlias the name of edge. e.g. e in pattern -[e]->
+// \param expr the filter expression
+/*static*/ Expression *ExpressionUtils::rewriteEdgePropertyFilter(ObjectPool *pool,
+                                                                  const std::string &edgeAlias,
+                                                                  Expression *expr) {
+  graph::RewriteVisitor::Matcher matcher = [&edgeAlias](const Expression *e) -> bool {
+    return isSingleLenExpandExpr(edgeAlias, e);
+  };
+  graph::RewriteVisitor::Rewriter rewriter = [pool](const Expression *e) -> Expression * {
+    DCHECK_EQ(e->kind(), Expression::Kind::kAttribute);
+    auto attributeExpr = static_cast<const AttributeExpression *>(e);
+    auto *right = attributeExpr->right();
+    DCHECK_EQ(right->kind(), Expression::Kind::kConstant);
+
+    auto &prop = static_cast<const ConstantExpression *>(right)->value().getStr();
+
+    auto *edgePropExpr = EdgePropertyExpression::make(pool, "*", prop);
+    return edgePropExpr;
+  };
+
+  return graph::RewriteVisitor::transform(expr, matcher, rewriter);
+}
+
 }  // namespace graph
 }  // namespace nebula

src/graph/util/ExpressionUtils.h

@@ -228,6 +228,13 @@ class ExpressionUtils {
   // Whether the whole expression is vertex id predication
   // e.g. id(v) == 1, id(v) IN [...]
   static bool isVidPredication(const Expression* expr);
+
+  // Check if the expr looks like `$-.e[0].likeness`
+  static bool isSingleLenExpandExpr(const std::string& edgeAlias, const Expression* expr);
+
+  static Expression* rewriteEdgePropertyFilter(ObjectPool* pool,
+                                               const std::string& edgeAlias,
+                                               Expression* expr);
 };
 
 }  // namespace graph

src/storage/query/QueryBaseProcessor-inl.h

@@ -407,9 +407,11 @@ nebula::cpp2::ErrorCode QueryBaseProcessor<REQ, RESP>::checkExp(
       if (ret != nebula::cpp2::ErrorCode::SUCCEEDED) {
         return ret;
       }
-      ret = checkExp(predExp->filter(), returned, filtered, updated, allowNoexistentProp);
-      if (ret != nebula::cpp2::ErrorCode::SUCCEEDED) {
-        return ret;
+      if (predExp->hasFilter()) {
+        ret = checkExp(predExp->filter(), returned, filtered, updated, allowNoexistentProp);
+        if (ret != nebula::cpp2::ErrorCode::SUCCEEDED) {
+          return ret;
+        }
       }
       return nebula::cpp2::ErrorCode::SUCCEEDED;
     }