kuzudb · andyfengHKU · Nov 10, 2023 · Nov 8, 2023
diff --git a/src/binder/bind/bind_graph_pattern.cpp b/src/binder/bind/bind_graph_pattern.cpp
@@ -266,8 +266,15 @@ std::shared_ptr<RelExpression> Binder::createNonRecursiveQueryRel(const std::str
     if (readVersion->getTableSchema(tableIDs[0])->getTableType() == TableType::RDF) {
         auto predicateID =
             expressionBinder.bindNodeOrRelPropertyExpression(*queryRel, std::string(rdf::PID));
-        auto resourceTableIDs = getNodeTableIDs(tableIDs);
-        auto resourceTableSchemas = readVersion->getTableSchemas(resourceTableIDs);
+        std::vector<common::table_id_t> resourceTableIDs;
+        std::vector<TableSchema*> resourceTableSchemas;
+        for (auto& tableID : tableIDs) {
+            auto rdfGraphSchema =
+                reinterpret_cast<RdfGraphSchema*>(readVersion->getTableSchema(tableID));
+            auto resourceTableID = rdfGraphSchema->getResourceTableID();
+            resourceTableIDs.push_back(resourceTableID);
+            resourceTableSchemas.push_back(readVersion->getTableSchema(resourceTableID));
+        }
         auto predicateIRI = createPropertyExpression(std::string(rdf::IRI),
             queryRel->getUniqueName(), queryRel->getVariableName(), resourceTableSchemas);
         auto rdfInfo =

diff --git a/src/binder/bind/copy/bind_copy_rdf_graph.cpp b/src/binder/bind/copy/bind_copy_rdf_graph.cpp
@@ -75,10 +75,11 @@ std::unique_ptr<BoundStatement> Binder::bindCopyRdfRelFrom(function::TableFuncti
         copyFunc->bindFunc(clientContext, tableFuncBindInput.get(), catalog.getReadOnlyVersion());
     auto boundFileScanInfo = std::make_unique<BoundFileScanInfo>(
         copyFunc, std::move(bindData), columns, relID, TableType::REL);
-    auto extraInfo = std::make_unique<ExtraBoundCopyRdfRelInfo>(columns[0], columns[1], columns[2]);
-    columns.push_back(std::move(relID));
-    auto boundCopyFromInfo = std::make_unique<BoundCopyFromInfo>(tableSchema,
-        std::move(boundFileScanInfo), containsSerial, std::move(columns), std::move(extraInfo));
+    auto extraInfo = std::make_unique<ExtraBoundCopyRdfRelInfo>(columns[0], columns[2]);
+    expression_vector columnsToCopy = {columns[0], columns[2], relID, columns[1]};
+    auto boundCopyFromInfo =
+        std::make_unique<BoundCopyFromInfo>(tableSchema, std::move(boundFileScanInfo),
+            containsSerial, std::move(columnsToCopy), std::move(extraInfo));
     return std::make_unique<BoundCopyFrom>(std::move(boundCopyFromInfo));
 }
 

diff --git a/src/binder/bind/ddl/bind_create_rdf_graph.cpp b/src/binder/bind/ddl/bind_create_rdf_graph.cpp
@@ -55,7 +55,7 @@ std::unique_ptr<BoundCreateTableInfo> Binder::bindCreateRdfGraphInfo(const Creat
     auto resourceTripleTableName = getRdfResourceTripleTableName(rdfGraphName);
     std::vector<std::unique_ptr<Property>> resourceTripleProperties;
     resourceTripleProperties.push_back(std::make_unique<Property>(
-        std::string(rdf::PID), std::make_unique<LogicalType>(LogicalTypeID::INT64)));
+        std::string(rdf::PID), std::make_unique<LogicalType>(LogicalTypeID::INTERNAL_ID)));
     auto boundResourceTripleExtraInfo =
         std::make_unique<BoundExtraCreateRelTableInfo>(RelMultiplicity::MANY_MANY, INVALID_TABLE_ID,
             INVALID_TABLE_ID, std::move(resourceTripleProperties));
@@ -65,7 +65,7 @@ std::unique_ptr<BoundCreateTableInfo> Binder::bindCreateRdfGraphInfo(const Creat
     auto literalTripleTableName = getRdfLiteralTripleTableName(rdfGraphName);
     std::vector<std::unique_ptr<Property>> literalTripleProperties;
     literalTripleProperties.push_back(std::make_unique<Property>(
-        std::string(rdf::PID), std::make_unique<LogicalType>(LogicalTypeID::INT64)));
+        std::string(rdf::PID), std::make_unique<LogicalType>(LogicalTypeID::INTERNAL_ID)));
     auto boundLiteralTripleExtraInfo =
         std::make_unique<BoundExtraCreateRelTableInfo>(RelMultiplicity::MANY_MANY, INVALID_TABLE_ID,
             INVALID_TABLE_ID, std::move(literalTripleProperties));

diff --git a/src/include/binder/copy/bound_copy_from.h b/src/include/binder/copy/bound_copy_from.h
@@ -69,16 +69,13 @@ struct ExtraBoundCopyRelInfo : public ExtraBoundCopyFromInfo {
 
 struct ExtraBoundCopyRdfRelInfo : public ExtraBoundCopyFromInfo {
     std::shared_ptr<Expression> subjectOffset;
-    std::shared_ptr<Expression> predicateOffset;
     std::shared_ptr<Expression> objectOffset;
 
-    ExtraBoundCopyRdfRelInfo(std::shared_ptr<Expression> subjectOffset,
-        std::shared_ptr<Expression> predicateOffset, std::shared_ptr<Expression> objectOffset)
-        : subjectOffset{std::move(subjectOffset)}, predicateOffset{std::move(predicateOffset)},
-          objectOffset{std::move(objectOffset)} {}
+    ExtraBoundCopyRdfRelInfo(
+        std::shared_ptr<Expression> subjectOffset, std::shared_ptr<Expression> objectOffset)
+        : subjectOffset{std::move(subjectOffset)}, objectOffset{std::move(objectOffset)} {}
     ExtraBoundCopyRdfRelInfo(const ExtraBoundCopyRdfRelInfo& other)
-        : subjectOffset{other.subjectOffset}, predicateOffset{other.predicateOffset},
-          objectOffset{other.objectOffset} {}
+        : subjectOffset{other.subjectOffset}, objectOffset{other.objectOffset} {}
 
     inline std::unique_ptr<ExtraBoundCopyFromInfo> copy() final {
         return std::make_unique<ExtraBoundCopyRdfRelInfo>(*this);

diff --git a/src/include/processor/operator/persistent/copy_rel.h b/src/include/processor/operator/persistent/copy_rel.h
@@ -18,6 +18,7 @@ struct CopyRelInfo {
     common::vector_idx_t partitioningIdx;
     common::RelDataDirection dataDirection;
     common::ColumnDataFormat dataFormat;
+    // TODO(Guodong): the following 3 fields are not being used.
     std::vector<DataPos> dataPoses;
     DataPos srcOffsetPos;
     DataPos relIDPos;

diff --git a/src/include/processor/operator/persistent/reader/rdf/rdf_reader.h b/src/include/processor/operator/persistent/reader/rdf/rdf_reader.h
@@ -48,10 +48,6 @@ class RDFReader {
     common::ValueVector* sVector; // subject
     common::ValueVector* pVector; // predicate
     common::ValueVector* oVector; // object
-
-    std::unique_ptr<common::ValueVector> sOffsetVector;
-    std::unique_ptr<common::ValueVector> pOffsetVector;
-    std::unique_ptr<common::ValueVector> oOffsetVector;
 };
 
 struct RdfScanLocalState final : public function::TableFuncLocalState {

diff --git a/src/planner/plan/append_extend.cpp b/src/planner/plan/append_extend.cpp
@@ -121,12 +121,12 @@ void QueryPlanner::appendNonRecursiveExtend(std::shared_ptr<NodeExpression> boun
         appendNodeLabelFilter(nbrNode->getInternalID(), nbrNode->getTableIDsSet(), plan);
     }
     if (iri) {
+        auto rdfInfo = rel->getRdfPredicateInfo();
+        appendFillTableID(rdfInfo->predicateID, rdfInfo->resourceTableIDs[0], plan);
         // Append hash join for remaining properties
         auto tmpPlan = std::make_unique<LogicalPlan>();
-        auto rdfInfo = rel->getRdfPredicateInfo();
         cardinalityEstimator->addNodeIDDom(*rdfInfo->predicateID, rdfInfo->resourceTableIDs);
         appendScanInternalID(rdfInfo->predicateID, rdfInfo->resourceTableIDs, *tmpPlan);
-        appendFillTableID(rdfInfo->predicateID, rel->getSingleTableID(), *tmpPlan);
         appendScanNodeProperties(
             rdfInfo->predicateID, rdfInfo->resourceTableIDs, expression_vector{iri}, *tmpPlan);
         appendHashJoin(expression_vector{rdfInfo->predicateID}, JoinType::INNER, plan, *tmpPlan);

diff --git a/src/processor/operator/persistent/reader/rdf/rdf_reader.cpp b/src/processor/operator/persistent/reader/rdf/rdf_reader.cpp
@@ -31,9 +31,6 @@ RDFReader::RDFReader(std::string filePath, std::unique_ptr<common::RdfReaderConf
     serd_reader_set_strict(reader, false /* strict */);
     serd_reader_set_error_sink(reader, errorHandle, this);
     serd_reader_start_stream(reader, fp, reinterpret_cast<const uint8_t*>(fileName.c_str()), true);
-    sOffsetVector = std::make_unique<ValueVector>(LogicalTypeID::INT64);
-    pOffsetVector = std::make_unique<ValueVector>(LogicalTypeID::INT64);
-    oOffsetVector = std::make_unique<ValueVector>(LogicalTypeID::INT64);
     counter = serd_reader_new(
         SERD_TURTLE, this, nullptr, nullptr, nullptr, counterStatementSink, nullptr);
     serd_reader_set_strict(counter, false /* strict */);
@@ -180,9 +177,9 @@ SerdStatus RDFReader::readerStatementSink(void* handle, SerdStatementFlags /*fla
         auto subjectOffset = lookupResourceNode(reader->config->index, subject);
         auto predicateOffset = lookupResourceNode(reader->config->index, predicate);
         auto objectOffset = lookupResourceNode(reader->config->index, object);
-        reader->sOffsetVector->setValue<int64_t>(reader->rowOffset, subjectOffset);
-        reader->pOffsetVector->setValue<int64_t>(reader->rowOffset, predicateOffset);
-        reader->oOffsetVector->setValue<int64_t>(reader->rowOffset, objectOffset);
+        reader->sVector->setValue<int64_t>(reader->rowOffset, subjectOffset);
+        reader->pVector->setValue<int64_t>(reader->rowOffset, predicateOffset);
+        reader->oVector->setValue<int64_t>(reader->rowOffset, objectOffset);
         reader->vectorSize++;
         reader->rowOffset++;
     } break;
@@ -193,9 +190,9 @@ SerdStatus RDFReader::readerStatementSink(void* handle, SerdStatementFlags /*fla
         auto subjectOffset = lookupResourceNode(reader->config->index, subject);
         auto predicateOffset = lookupResourceNode(reader->config->index, predicate);
         auto objectOffset = reader->rowOffset;
-        reader->sOffsetVector->setValue<int64_t>(reader->rowOffset, subjectOffset);
-        reader->pOffsetVector->setValue<int64_t>(reader->rowOffset, predicateOffset);
-        reader->oOffsetVector->setValue<int64_t>(reader->rowOffset, objectOffset);
+        reader->sVector->setValue<int64_t>(reader->rowOffset, subjectOffset);
+        reader->pVector->setValue<int64_t>(reader->rowOffset, predicateOffset);
+        reader->oVector->setValue<int64_t>(reader->rowOffset, objectOffset);
         reader->vectorSize++;
         reader->rowOffset++;
     } break;

diff --git a/test/test_files/copy/copy_rdf.test b/test/test_files/copy/copy_rdf.test
@@ -1,19 +1,18 @@
 -GROUP CopyRDFTest
 -DATASET CSV copy-test/rdf
 -BUFFER_POOL_SIZE 536870912
--SKIP
 
 --
 
 -CASE CopyRDFTest
 
 -LOG CountRDFNodeTable
--STATEMENT MATCH (s:taxonomy) RETURN COUNT(s.IRI)
+-STATEMENT MATCH (s:taxonomy_resource_t) RETURN COUNT(s.iri)
 ---- 1
 1138441
 
 -LOG QueryRDFNodeTable
--STATEMENT MATCH (s:taxonomy) RETURN s.IRI ORDER BY s.IRI LIMIT 5
+-STATEMENT MATCH (s:taxonomy_resource_t) RETURN s.iri ORDER BY s.iri LIMIT 5
 ---- 5
 http://dbpedia.org/class/yago/'hood108641944
 http://dbpedia.org/class/yago/14July115200493
@@ -22,7 +21,7 @@ http://dbpedia.org/class/yago/16PF106475933
 http://dbpedia.org/class/yago/1750s115149933
 
 -LOG QueryRDFRelTable
--STATEMENT MATCH (s:taxonomy)-[p:taxonomy]->(o:taxonomy) RETURN s.IRI, o.IRI ORDER BY s.IRI LIMIT 2
+-STATEMENT MATCH (s:taxonomy_resource_t)-[p:taxonomy]->(o:taxonomy_resource_t) RETURN s.iri, o.iri ORDER BY s.iri LIMIT 2
 ---- 2
 http://dbpedia.org/class/yago/'hood108641944|http://dbpedia.org/class/yago/Vicinity108641113
 http://dbpedia.org/class/yago/14July115200493|http://dbpedia.org/class/yago/LegalHoliday115199592
diff --git a/test/test_files/rdf/rdfox_example.test b/test/test_files/rdf/rdfox_example.test
@@ -78,33 +78,36 @@ Peter||||||||||||||||Peter
 ||||Monday|1999-08-16a
 |||||Petera
 
--CASE tmp
--SKIP
-
--STATEMENT MATCH (a:example_resource_t)-[e:example_resource_triples_t]->(b:example_resource_t) RETURN a.iri, b.iri
+-STATEMENT MATCH (a:example_resource_t)-[e:example]->(b:example_resource_t) RETURN a.iri, e.iri, b.iri ORDER BY offset(id(e))
+-CHECK_ORDER
 ---- 10
-:peter|:Person
-:peter|:lois
-:lois|:Person
-:meg|:Person
-:meg|:lois
-:meg|:peter
-:chris|:Person
-:chris|:peter
-:stewie|:Person
-:stewie|:lois
+:peter|http://www.w3.org/1999/02/22-rdf-syntax-ns#type|:Person
+:peter|:marriedTo|:lois
+:lois|http://www.w3.org/1999/02/22-rdf-syntax-ns#type|:Person
+:meg|http://www.w3.org/1999/02/22-rdf-syntax-ns#type|:Person
+:meg|:hasParent|:lois
+:meg|:hasParent|:peter
+:chris|http://www.w3.org/1999/02/22-rdf-syntax-ns#type|:Person
+:chris|:hasParent|:peter
+:stewie|http://www.w3.org/1999/02/22-rdf-syntax-ns#type|:Person
+:stewie|:hasParent|:lois
 
--STATEMENT MATCH (a:example_resource_t)-[e:example_literal_triples_t]->(b:example_literal_t) RETURN a.iri, b.iri
----- 12
-:peter|Peter
-:peter|male
-:lois|Lois
-:lois|female
-:meg|Meg
-:meg|female
-:chris|Chris
-:chris|male
-:stewie|Stewie
-:stewie|male
-:brian|Brian
-:andy|12
+-STATEMENT MATCH (a:example_resource_t)-[e:example]->(b:example_literal_t) RETURN a.iri, e.iri, b.iri ORDER BY offset(id(e))
+-CHECK_ORDER
+---- 16
+:andy|:literal|12
+:andy|:literal|-14.900000
+:andy|:literal|True
+:andy|:literal|0.016630
+:andy|:literal|1999-08-16
+:peter|:forename|Peter
+:peter|:gender|male
+:lois|:forename|Lois
+:lois|:gender|female
+:meg|:forename|Meg
+:meg|:gender|female
+:chris|:forename|Chris
+:chris|:gender|male
+:stewie|:forename|Stewie
+:stewie|:gender|male
+:brian|:forename|Brian
diff --git a/test/test_files/rdf/tiny.test b/test/test_files/rdf/tiny.test