Merge pull request #1815 from kuzudb/self-loop-edge

Add self-loop-edge
kuzudb · Jul 14, 2023 · 0662374 · 0662374
2 parents 9f7007a + 21c8c7d
commit 0662374
Show file tree

Hide file tree

Showing 9 changed files with 76 additions and 12 deletions.
diff --git a/src/binder/bind/bind_graph_pattern.cpp b/src/binder/bind/bind_graph_pattern.cpp
@@ -212,9 +212,6 @@ std::shared_ptr<RelExpression> Binder::bindQueryRel(const RelPattern& relPattern
     default:
         throw common::NotImplementedException("Binder::bindQueryRel");
     }
-    if (srcNode->getUniqueName() == dstNode->getUniqueName()) {
-        throw BinderException("Self-loop rel " + parsedName + " is not supported.");
-    }
     // bind variable length
     std::shared_ptr<RelExpression> queryRel;
     if (QueryRelTypeUtils::isRecursive(relPattern.getRelType())) {

diff --git a/src/binder/bind/bind_reading_clause.cpp b/src/binder/bind/bind_reading_clause.cpp
@@ -37,12 +37,37 @@ std::unique_ptr<BoundReadingClause> Binder::bindMatchClause(const ReadingClause&
     if (matchClause.hasWhereClause()) {
         whereExpression = bindWhereExpression(*matchClause.getWhereClause());
     }
+    // Rewrite self loop edge
+    // e.g. rewrite (a)-[e]->(a) as [a]-[e]->(b) WHERE id(a) = id(b)
+    expression_vector selfLoopEdgePredicates;
+    auto graphCollection = boundMatchClause->getQueryGraphCollection();
+    for (auto i = 0; i < graphCollection->getNumQueryGraphs(); ++i) {
+        auto queryGraph = graphCollection->getQueryGraph(i);
+        for (auto& queryRel : queryGraph->getQueryRels()) {
+            if (!queryRel->isSelfLoop()) {
+                continue;
+            }
+            auto src = queryRel->getSrcNode();
+            auto dst = queryRel->getDstNode();
+            auto newDst = createQueryNode(dst->getVariableName(), dst->getTableIDs());
+            queryGraph->addQueryNode(newDst);
+            queryRel->setDstNode(newDst);
+            auto predicate = expressionBinder.createEqualityComparisonExpression(
+                src->getInternalIDProperty(), newDst->getInternalIDProperty());
+            selfLoopEdgePredicates.push_back(std::move(predicate));
+        }
+    }
+    for (auto& predicate : selfLoopEdgePredicates) {
+        whereExpression =
+            expressionBinder.combineConjunctiveExpressions(predicate, whereExpression);
+    }
     // Rewrite key value pairs in MATCH clause as predicate
     for (auto& [key, val] : propertyCollection->getKeyVals()) {
         auto predicate = expressionBinder.createEqualityComparisonExpression(key, val);
         whereExpression =
             expressionBinder.combineConjunctiveExpressions(predicate, whereExpression);
     }
+
     boundMatchClause->setWhereExpression(std::move(whereExpression));
     return boundMatchClause;
 }

diff --git a/src/include/binder/expression/expression.h b/src/include/binder/expression/expression.h
@@ -81,6 +81,8 @@ class Expression : public std::enable_shared_from_this<Expression> {
 
     expression_vector splitOnAND();
 
+    inline bool operator==(const Expression& rhs) const { return uniqueName == rhs.uniqueName; }
+
     virtual std::string toString() const = 0;
 
     virtual std::unique_ptr<Expression> copy() const {

diff --git a/src/include/binder/expression/node_rel_expression.h b/src/include/binder/expression/node_rel_expression.h
@@ -15,6 +15,8 @@ class NodeOrRelExpression : public Expression {
           variableName(std::move(variableName)), tableIDs{std::move(tableIDs)} {}
     virtual ~NodeOrRelExpression() override = default;
 
+    inline std::string getVariableName() const { return variableName; }
+
     inline void addTableIDs(const std::vector<common::table_id_t>& tableIDsToAdd) {
         auto tableIDsSet = getTableIDsSet();
         for (auto tableID : tableIDsToAdd) {

diff --git a/src/include/binder/expression/rel_expression.h b/src/include/binder/expression/rel_expression.h
@@ -51,6 +51,7 @@ class RelExpression : public NodeOrRelExpression {
 
     inline std::shared_ptr<NodeExpression> getSrcNode() const { return srcNode; }
     inline std::string getSrcNodeName() const { return srcNode->getUniqueName(); }
+    inline void setDstNode(std::shared_ptr<NodeExpression> node) { dstNode = std::move(node); }
     inline std::shared_ptr<NodeExpression> getDstNode() const { return dstNode; }
     inline std::string getDstNodeName() const { return dstNode->getUniqueName(); }
 
@@ -72,6 +73,8 @@ class RelExpression : public NodeOrRelExpression {
         return recursiveInfo->lengthExpression;
     }
 
+    inline bool isSelfLoop() const { return *srcNode == *dstNode; }
+
 private:
     std::shared_ptr<NodeExpression> srcNode;
     std::shared_ptr<NodeExpression> dstNode;

diff --git a/test/test_files/exceptions/binder/binder_error.test b/test/test_files/exceptions/binder/binder_error.test
@@ -304,11 +304,6 @@ Binder exception: PropertyName: _id is an internal reserved propertyName.
 ---- error
 Binder exception: Node/Rel person1 does not exist.
 
--LOG SelfLoopRel
--STATEMENT MATCH (a:person)-[e:knows]->(a) RETURN COUNT(*)
----- error
-Binder exception: Self-loop rel e is not supported.
-
 -LOG InvalidLimitNumberType
 -STATEMENT MATCH (a:person) RETURN a.age LIMIT "abc"
 ---- error

diff --git a/test/test_files/tck/match/match2.test b/test/test_files/tck/match/match2.test
@@ -33,3 +33,33 @@ Binder exception: No rel table exists in database.
 -STATEMENT MATCH (:A)-[r]->(:B) RETURN r;
 ---- 1
 (0:0)-{_LABEL: T1, _ID: 2:0}->(1:0)
+
+# Matching a self-loop with an undirected relationship pattern
+-CASE Scenario3&4
+-STATEMENT CREATE NODE TABLE A(name STRING, PRIMARY KEY(name));
+---- ok
+-STATEMENT CREATE REL TABLE T(FROM A TO A);
+---- ok
+-STATEMENT CREATE (a:A {name:'xx'}) CREATE (a)-[:T]->(a);
+---- ok
+# Note: tck test only returns 1 tuple. Though we shouldn't do so if the semantic of undirected is to match
+# both fwd and bwd adjList.
+-STATEMENT MATCH (a)-[:T]-(a) RETURN a;
+---- 2
+{_ID: 0:0, _LABEL: A, name: xx}
+{_ID: 0:0, _LABEL: A, name: xx}
+-STATEMENT MATCH (a)-[:T]->(a) RETURN a;
+---- 1
+{_ID: 0:0, _LABEL: A, name: xx}
+
+# Match relationship with inline property value
+-CASE Scenario5
+-STATEMENT CREATE NODE TABLE A(ID INT64, PRIMARY KEY(ID));
+---- ok
+-STATEMENT CREATE REL TABLE KNOWS(FROM A TO A, name STRING);
+---- ok
+-STATEMENT CREATE (:A {ID: 1})<-[:KNOWS {name: 'monkey'}]-(:A {ID: 3})-[:KNOWS {name: 'woot'}]->(:A {ID: 5});
+---- ok
+-STATEMENT MATCH (n)-[r:KNOWS {name: 'monkey'}]->(a) RETURN a;
+---- 1
+{_ID: 0:0, _LABEL: A, ID: 1}
diff --git a/test/test_files/tinysnb/var_length_extend/multi_label.test b/test/test_files/tinysnb/var_length_extend/multi_label.test
@@ -37,26 +37,31 @@
 4|[3:1,5:0]|[,2015]
 6|[3:2,5:1]|[,2010]
 
--LOG MixMultiLabelTest2
+-LOG MultiLabelTest1
 -STATEMENT MATCH (a:person)-[e:meets|:marries|:studyAt*2..2]->(b) WHERE a.fName = 'Alice' RETURN b.ID, properties(rels(e), '_id'), properties(nodes(e), 'fName')
 ---- 4
 1|[6:0,4:1]|[Bob]
 1|[7:0,4:1]|[Bob]
 5|[6:0,6:1]|[Bob]
 5|[7:0,6:1]|[Bob]
 
--LOG MixMultiLabelTest3
+-LOG MultiLabelTest2
 -STATEMENT MATCH (a:person)-[e:meets|:marries|:studyAt*2..2]->(b) WHERE a.fName = 'Alice' AND b.ID < 5 RETURN COUNT(*)
 -ENUMERATE
 ---- 1
 2
 
--LOG MixMultiLabelTest4
+-LOG MultiLabelTest3
 -STATEMENT MATCH (a:person)-[e*2..2]->(b:organisation) WHERE a.fName = 'Alice' RETURN COUNT(*)
 ---- 1
 5
 
--LOG MixMultiLabelTest5
+-LOG MultiLabelTest
 -STATEMENT MATCH (a:person)-[e*2..2 (r, _ | WHERE offset(id(r)) > 0)]->(b:organisation) WHERE a.fName = 'Alice' RETURN rels(e)
 ---- 1
 [(0:0)-{_LABEL: knows, _ID: 3:2, date: 2021-06-30, meetTime: 2012-12-11 20:07:22, validInterval: 10 days, comments: [ioji232,jifhe8w99u43434]}->(0:3),(0:3)-{_LABEL: workAt, _ID: 5:1, year: 2010, grading: [2.100000,4.400000], rating: 7.600000}->(1:2)]
+
+-LOG MultiLabelSelfLoopTest
+-STATEMENT MATCH (a)-[e:studyAt|:knows*2..3]-(a) WHERE a.ID = 1 RETURN COUNT(*)
+---- 1
+7
diff --git a/test/test_files/tinysnb/var_length_extend/n_n.test b/test/test_files/tinysnb/var_length_extend/n_n.test
@@ -72,3 +72,8 @@ Greg
 -STATEMENT MATCH (a:person)-[e:knows*1..2 (r,_ | WHERE list_contains(r.comments, 'rnme'))]->(b:person) WHERE a.fName='Alice' RETURN COUNT(*)
 ---- 1
 1
+
+-LOG SelfLoop
+-STATEMENT MATCH (a:person)-[e:knows*1..3]->(a) WHERE a.fName='Alice' RETURN COUNT(*)
+---- 1
+9