add scan rel ID blindly

kuzudb · Nov 28, 2022 · fee58f9 · fee58f9
1 parent 6ce87e0
commit fee58f9
Show file tree

Hide file tree

Showing 27 changed files with 169 additions and 100 deletions.
diff --git a/src/binder/bind/bind_graph_pattern.cpp b/src/binder/bind/bind_graph_pattern.cpp
@@ -97,6 +97,9 @@ void Binder::bindQueryRel(const RelPattern& relPattern, const shared_ptr<NodeExp
     }
     auto queryRel = make_shared<RelExpression>(
         getUniqueExpressionName(parsedName), tableID, srcNode, dstNode, lowerBound, upperBound);
+    if (!queryRel->isVariableLength()) {
+        queryRel->setInternalIDProperty(expressionBinder.bindInternalIDExpression(queryRel));
+    }
     queryRel->setAlias(parsedName);
     queryRel->setRawName(parsedName);
     if (!parsedName.empty()) {
@@ -141,6 +144,7 @@ shared_ptr<NodeExpression> Binder::createQueryNode(const NodePattern& nodePatter
     auto parsedName = nodePattern.getVariableName();
     auto tableIDs = bindNodeTableIDs(nodePattern.getTableNames());
     auto queryNode = make_shared<NodeExpression>(getUniqueExpressionName(parsedName), tableIDs);
+    queryNode->setInternalIDProperty(expressionBinder.bindInternalIDExpression(queryNode));
     queryNode->setAlias(parsedName);
     queryNode->setRawName(parsedName);
     if (!parsedName.empty()) {

diff --git a/src/binder/expression/existential_subquery_expression.cpp b/src/binder/expression/existential_subquery_expression.cpp
@@ -17,8 +17,8 @@ unordered_set<string> ExistentialSubqueryExpression::getDependentVariableNames()
 // expressions from predicates and return clause. Plus nodeID expressions from query graph.
 expression_vector ExistentialSubqueryExpression::getChildren() const {
     expression_vector result;
-    for (auto& nodeIDExpression : queryGraphCollection->getNodeIDExpressions()) {
-        result.push_back(nodeIDExpression);
+    for (auto& node : queryGraphCollection->getQueryNodes()) {
+        result.push_back(node->getInternalIDProperty());
     }
     if (hasWhereExpression()) {
         result.push_back(whereExpression);

diff --git a/src/binder/expression_binder.cpp b/src/binder/expression_binder.cpp
@@ -307,16 +307,30 @@ shared_ptr<Expression> ExpressionBinder::bindInternalIDExpression(
 shared_ptr<Expression> ExpressionBinder::bindInternalIDExpression(
     shared_ptr<Expression> nodeOrRel) {
     if (nodeOrRel->dataType.typeID == NODE) {
-        return ((NodeExpression*)nodeOrRel.get())->getNodeIDPropertyExpression();
+        return bindInternalNodeIDExpression(nodeOrRel);
     } else {
         assert(nodeOrRel->dataType.typeID == REL);
-        auto rel = (RelExpression*)nodeOrRel.get();
-        auto relTableSchema =
-            binder->catalog.getReadOnlyVersion()->getRelTableSchema(rel->getTableID());
-        return bindRelPropertyExpression(nodeOrRel, relTableSchema->getRelIDDefinition());
+        return bindInternalRelIDExpression(nodeOrRel);
     }
 }
 
+shared_ptr<Expression> ExpressionBinder::bindInternalNodeIDExpression(shared_ptr<Expression> node) {
+    auto nodeExpression = (NodeExpression*)node.get();
+    unordered_map<table_id_t, property_id_t> propertyIDPerTable;
+    for (auto tableID : nodeExpression->getTableIDs()) {
+        propertyIDPerTable.insert({tableID, INVALID_PROPERTY_ID});
+    }
+    return make_unique<PropertyExpression>(
+        DataType(NODE_ID), INTERNAL_ID_SUFFIX, std::move(propertyIDPerTable), node);
+}
+
+shared_ptr<Expression> ExpressionBinder::bindInternalRelIDExpression(shared_ptr<Expression> rel) {
+    auto relExpression = (RelExpression*)rel.get();
+    auto relTableSchema =
+        binder->catalog.getReadOnlyVersion()->getRelTableSchema(relExpression->getTableID());
+    return bindRelPropertyExpression(rel, relTableSchema->getRelIDDefinition());
+}
+
 shared_ptr<Expression> ExpressionBinder::bindParameterExpression(
     const ParsedExpression& parsedExpression) {
     auto& parsedParameterExpression = (ParsedParameterExpression&)parsedExpression;

diff --git a/src/binder/query/bound_delete_clause.cpp b/src/binder/query/bound_delete_clause.cpp
@@ -8,6 +8,11 @@ expression_vector BoundDeleteClause::getPropertiesToRead() const {
     for (auto& deleteNode : deleteNodes) {
         result.push_back(deleteNode->getPrimaryKeyExpression());
     }
+    for (auto& deleteRel : deleteRels) {
+        if (deleteRel->hasInternalIDProperty()) {
+            result.push_back(deleteRel->getInternalIDProperty());
+        }
+    }
     return result;
 }
 

diff --git a/src/binder/query/query_graph.cpp b/src/binder/query/query_graph.cpp
@@ -187,14 +187,6 @@ bool QueryGraph::isConnected(const QueryGraph& other) {
     return false;
 }
 
-vector<shared_ptr<Expression>> QueryGraph::getNodeIDExpressions() const {
-    vector<shared_ptr<Expression>> result;
-    for (auto& queryNode : queryNodes) {
-        result.push_back(queryNode->getNodeIDPropertyExpression());
-    }
-    return result;
-}
-
 void QueryGraphCollection::addAndMergeQueryGraphIfConnected(
     unique_ptr<QueryGraph> queryGraphToAdd) {
     bool isMerged = false;
@@ -209,11 +201,21 @@ void QueryGraphCollection::addAndMergeQueryGraphIfConnected(
     }
 }
 
-expression_vector QueryGraphCollection::getNodeIDExpressions() const {
-    expression_vector result;
+vector<shared_ptr<NodeExpression>> QueryGraphCollection::getQueryNodes() const {
+    vector<shared_ptr<NodeExpression>> result;
+    for (auto& queryGraph : queryGraphs) {
+        for (auto& node : queryGraph->getQueryNodes()) {
+            result.push_back(node);
+        }
+    }
+    return result;
+}
+
+vector<shared_ptr<RelExpression>> QueryGraphCollection::getQueryRels() const {
+    vector<shared_ptr<RelExpression>> result;
     for (auto& queryGraph : queryGraphs) {
-        for (auto& nodeID : queryGraph->getNodeIDExpressions()) {
-            result.push_back(nodeID);
+        for (auto& rel : queryGraph->getQueryRels()) {
+            result.push_back(rel);
         }
     }
     return result;

diff --git a/src/include/binder/expression/node_expression.h b/src/include/binder/expression/node_expression.h
@@ -20,19 +20,20 @@ class NodeExpression : public Expression {
         return *tableIDs.begin();
     }
 
-    inline string getIDProperty() const { return uniqueName + "." + INTERNAL_ID_SUFFIX; }
-
-    inline shared_ptr<Expression> getNodeIDPropertyExpression() {
-        unordered_map<table_id_t, property_id_t> propertyIDPerTable;
-        for (auto tableID : tableIDs) {
-            propertyIDPerTable.insert({tableID, INVALID_PROPERTY_ID});
-        }
-        return make_unique<PropertyExpression>(DataType(NODE_ID), INTERNAL_ID_SUFFIX,
-            std::move(propertyIDPerTable), shared_from_this());
+    inline void setInternalIDProperty(shared_ptr<Expression> expression) {
+        internalIDExpression = std::move(expression);
+    }
+    inline shared_ptr<Expression> getInternalIDProperty() const {
+        assert(internalIDExpression != nullptr);
+        return internalIDExpression;
+    }
+    inline string getInternalIDPropertyName() const {
+        return internalIDExpression->getUniqueName();
     }
 
 private:
     unordered_set<table_id_t> tableIDs;
+    shared_ptr<Expression> internalIDExpression;
 };
 
 } // namespace binder

diff --git a/src/include/binder/expression/rel_expression.h b/src/include/binder/expression/rel_expression.h
@@ -1,17 +1,17 @@
 #pragma once
 
+#include "common/exception.h"
 #include "node_expression.h"
 
 namespace kuzu {
 namespace binder {
 
 class RelExpression : public Expression {
-
 public:
     RelExpression(const string& uniqueName, table_id_t tableID, shared_ptr<NodeExpression> srcNode,
         shared_ptr<NodeExpression> dstNode, uint64_t lowerBound, uint64_t upperBound)
-        : Expression{VARIABLE, REL, uniqueName}, tableID{tableID}, srcNode{move(srcNode)},
-          dstNode{move(dstNode)}, lowerBound{lowerBound}, upperBound{upperBound} {}
+        : Expression{VARIABLE, REL, uniqueName}, tableID{tableID}, srcNode{std::move(srcNode)},
+          dstNode{std::move(dstNode)}, lowerBound{lowerBound}, upperBound{upperBound} {}
 
     inline table_id_t getTableID() const { return tableID; }
 
@@ -29,12 +29,25 @@ class RelExpression : public Expression {
 
     inline bool isVariableLength() const { return !(lowerBound == 1 && upperBound == 1); }
 
+    inline void setInternalIDProperty(shared_ptr<Expression> expression) {
+        internalIDExpression = std::move(expression);
+    }
+    inline bool hasInternalIDProperty() const { return internalIDExpression != nullptr; }
+    inline shared_ptr<Expression> getInternalIDProperty() const {
+        if (!hasInternalIDProperty()) {
+            throw NotImplementedException(
+                "Cannot read internal ID property for variable length rel " + getRawName());
+        }
+        return internalIDExpression;
+    }
+
 private:
     table_id_t tableID;
     shared_ptr<NodeExpression> srcNode;
     shared_ptr<NodeExpression> dstNode;
     uint64_t lowerBound;
     uint64_t upperBound;
+    shared_ptr<Expression> internalIDExpression;
 };
 
 } // namespace binder

diff --git a/src/include/binder/expression_binder.h b/src/include/binder/expression_binder.h
@@ -54,6 +54,8 @@ class ExpressionBinder {
 
     shared_ptr<Expression> bindInternalIDExpression(const ParsedExpression& parsedExpression);
     shared_ptr<Expression> bindInternalIDExpression(shared_ptr<Expression> nodeOrRel);
+    shared_ptr<Expression> bindInternalNodeIDExpression(shared_ptr<Expression> node);
+    shared_ptr<Expression> bindInternalRelIDExpression(shared_ptr<Expression> rel);
 
     shared_ptr<Expression> bindParameterExpression(const ParsedExpression& parsedExpression);
 

diff --git a/src/include/binder/query/reading_clause/bound_match_clause.h b/src/include/binder/query/reading_clause/bound_match_clause.h
@@ -41,6 +41,11 @@ class BoundMatchClause : public BoundReadingClause {
 
     inline expression_vector getSubPropertyExpressions() const override {
         expression_vector expressions;
+        for (auto& rel : queryGraphCollection->getQueryRels()) {
+            if (rel->hasInternalIDProperty()) {
+                expressions.push_back(rel->getInternalIDProperty());
+            }
+        }
         if (this->hasWhereExpression()) {
             for (auto& property : this->getWhereExpression()->getSubPropertyExpressions()) {
                 expressions.push_back(property);

diff --git a/src/include/binder/query/reading_clause/query_graph.h b/src/include/binder/query/reading_clause/query_graph.h
@@ -87,6 +87,7 @@ class QueryGraph {
     inline bool containsQueryNode(const string& queryNodeName) const {
         return queryNodeNameToPosMap.contains(queryNodeName);
     }
+    inline vector<shared_ptr<NodeExpression>> getQueryNodes() const { return queryNodes; }
     inline shared_ptr<NodeExpression> getQueryNode(const string& queryNodeName) const {
         return queryNodes[getQueryNodePos(queryNodeName)];
     }
@@ -112,6 +113,7 @@ class QueryGraph {
     inline bool containsQueryRel(const string& queryRelName) const {
         return queryRelNameToPosMap.contains(queryRelName);
     }
+    inline vector<shared_ptr<RelExpression>> getQueryRels() const { return queryRels; }
     inline shared_ptr<RelExpression> getQueryRel(const string& queryRelName) const {
         return queryRels.at(queryRelNameToPosMap.at(queryRelName));
     }
@@ -129,8 +131,6 @@ class QueryGraph {
 
     void merge(const QueryGraph& other);
 
-    vector<shared_ptr<Expression>> getNodeIDExpressions() const;
-
     inline unique_ptr<QueryGraph> copy() const { return make_unique<QueryGraph>(*this); }
 
 private:
@@ -150,7 +150,8 @@ class QueryGraphCollection {
     inline uint32_t getNumQueryGraphs() const { return queryGraphs.size(); }
     inline QueryGraph* getQueryGraph(uint32_t idx) const { return queryGraphs[idx].get(); }
 
-    expression_vector getNodeIDExpressions() const;
+    vector<shared_ptr<NodeExpression>> getQueryNodes() const;
+    vector<shared_ptr<RelExpression>> getQueryRels() const;
 
     unique_ptr<QueryGraphCollection> copy() const;
 

diff --git a/src/include/planner/logical_plan/logical_operator/logical_extend.h b/src/include/planner/logical_plan/logical_operator/logical_extend.h
@@ -26,15 +26,15 @@ class LogicalExtend : public LogicalOperator {
     }
 
     inline void computeSchema(Schema& schema) {
-        auto boundGroupPos = schema.getGroupPos(boundNode->getIDProperty());
+        auto boundGroupPos = schema.getGroupPos(boundNode->getInternalIDPropertyName());
         uint32_t nbrGroupPos = 0u;
         if (isColumn) {
             nbrGroupPos = boundGroupPos;
         } else {
             assert(schema.getGroup(boundGroupPos)->getIsFlat());
             nbrGroupPos = schema.createGroup();
         }
-        schema.insertToGroupAndScope(nbrNode->getNodeIDPropertyExpression(), nbrGroupPos);
+        schema.insertToGroupAndScope(nbrNode->getInternalIDProperty(), nbrGroupPos);
     }
 
     inline shared_ptr<NodeExpression> getBoundNodeExpression() const { return boundNode; }

diff --git a/src/include/planner/logical_plan/logical_operator/logical_scan_node.h b/src/include/planner/logical_plan/logical_operator/logical_scan_node.h
@@ -19,7 +19,7 @@ class LogicalScanNode : public LogicalOperator {
 
     inline virtual void computeSchema(Schema& schema) {
         auto groupPos = schema.createGroup();
-        schema.insertToGroupAndScope(node->getNodeIDPropertyExpression(), groupPos);
+        schema.insertToGroupAndScope(node->getInternalIDProperty(), groupPos);
     }
 
     inline shared_ptr<NodeExpression> getNode() const { return node; }
@@ -43,7 +43,7 @@ class LogicalIndexScanNode : public LogicalScanNode {
 
     inline void computeSchema(Schema& schema) override {
         LogicalScanNode::computeSchema(schema);
-        auto groupPos = schema.getGroupPos(node->getIDProperty());
+        auto groupPos = schema.getGroupPos(node->getInternalIDPropertyName());
         schema.getGroup(groupPos)->setIsFlat(true);
     }
 

diff --git a/src/include/processor/operator/update/delete.h b/src/include/processor/operator/update/delete.h
@@ -56,15 +56,16 @@ struct DeleteRelInfo {
     table_id_t srcNodeTableID;
     DataPos dstNodePos;
     table_id_t dstNodeTableID;
+    DataPos relIDPos;
 
     DeleteRelInfo(RelTable* table, const DataPos& srcNodePos, table_id_t srcNodeTableID,
-        const DataPos& dstNodePos, table_id_t dstNodeTableID)
+        const DataPos& dstNodePos, table_id_t dstNodeTableID, const DataPos& relIDPos)
         : table{table}, srcNodePos{srcNodePos}, srcNodeTableID{srcNodeTableID},
-          dstNodePos{dstNodePos}, dstNodeTableID{dstNodeTableID} {}
+          dstNodePos{dstNodePos}, dstNodeTableID{dstNodeTableID}, relIDPos{relIDPos} {}
 
     inline unique_ptr<DeleteRelInfo> clone() {
         return make_unique<DeleteRelInfo>(
-            table, srcNodePos, srcNodeTableID, dstNodePos, dstNodeTableID);
+            table, srcNodePos, srcNodeTableID, dstNodePos, dstNodeTableID, relIDPos);
     }
 };
 
@@ -95,7 +96,9 @@ class DeleteRel : public PhysicalOperator {
 private:
     RelsStatistics& relsStatistics;
     vector<unique_ptr<DeleteRelInfo>> deleteRelInfos;
-    vector<pair<ValueVector*, ValueVector*>> srcDstNodeIDVectorPairs;
+    vector<ValueVector*> srcNodeVectors;
+    vector<ValueVector*> dstNodeVectors;
+    vector<ValueVector*> relIDVectors;
 };
 
 } // namespace processor