libspatialjoin integretion

CMakeLists.txt

@@ -105,7 +105,14 @@ FetchContent_Declare(
         GIT_TAG 1dc0b09abab1bdc7d085a78754abd5c6e37a5d0c # 0.12.0
 )
 
-
+#################################
+# libspatialjoin
+################################
+FetchContent_Declare(
+        spatialjoin
+        GIT_REPOSITORY https://github.com/ad-freiburg/spatialjoin
+        GIT_TAG 9cd62b173a766be8a5e1bc9b446c877da24cdc91
+)
 
 ################################
 # Threading
@@ -340,7 +347,7 @@ FetchContent_Declare(
 ################################
 # Apply FetchContent
 ################################
-FetchContent_MakeAvailable(googletest ctre abseil re2 stxxl fsst s2 nlohmann-json antlr range-v3)
+FetchContent_MakeAvailable(googletest ctre abseil re2 stxxl fsst s2 nlohmann-json antlr range-v3 spatialjoin)
 # Disable some warnings in RE2, STXXL, and GTEST
 target_compile_options(s2 PRIVATE -Wno-sign-compare -Wno-unused-parameter -Wno-class-memaccess -Wno-comment -Wno-redundant-move -Wno-unknown-warning-option -Wno-maybe-uninitialized -Wno-class-memaccess -Wno-unused-but-set-variable -Wno-unused-function)
 target_compile_options(re2 PRIVATE -Wno-unused-parameter)

src/engine/CMakeLists.txt

@@ -16,4 +16,4 @@ add_library(engine
         CountConnectedSubgraphs.cpp SpatialJoinAlgorithms.cpp PathSearch.cpp ExecuteUpdate.cpp
         Describe.cpp GraphStoreProtocol.cpp
         QueryExecutionContext.cpp)
-qlever_target_link_libraries(engine util index parser sparqlExpressions http SortPerformanceEstimator Boost::iostreams s2)
+qlever_target_link_libraries(engine util index parser sparqlExpressions http SortPerformanceEstimator Boost::iostreams s2 spatialjoin-dev pb_util)

src/engine/SpatialJoin.cpp

@@ -74,7 +74,11 @@ bool SpatialJoin::isConstructed() const { return childLeft_ && childRight_; }
 // ____________________________________________________________________________
 std::optional<size_t> SpatialJoin::getMaxDist() const {
   auto visitor = []<typename T>(const T& config) -> std::optional<size_t> {
-    return config.maxDist_;
+		if constexpr (std::is_same_v<T, SJConfig>) {
+			return std::nullopt;
+		} else {
+			return config.maxDist_;
+	}
   };
   return std::visit(visitor, config_.task_);
 }
@@ -84,6 +88,8 @@ std::optional<size_t> SpatialJoin::getMaxResults() const {
   auto visitor = []<typename T>(const T& config) -> std::optional<size_t> {
     if constexpr (std::is_same_v<T, MaxDistanceConfig>) {
       return std::nullopt;
+    } else if constexpr (std::is_same_v<T, SJConfig>) {
+      return std::nullopt;
     } else {
       static_assert(std::is_same_v<T, NearestNeighborsConfig>);
       return config.maxResults_;
@@ -129,6 +135,12 @@ string SpatialJoin::getCacheKeyImpl() const {
       os << "maxResults: " << maxResults.value() << "\n";
     }
 
+    // Uses distance variable?
+    auto algo = getAlgorithm();
+    if (algo == SpatialJoinAlgorithm::LIBSPATIALJOIN) {
+      os << "libspatialjoin on: " << (int)config_.joinType_ << "\n";
+    }
+
     // Uses distance variable?
     if (config_.distanceVariable_.has_value()) {
       os << "withDistanceVariable\n";
@@ -163,6 +175,9 @@ string SpatialJoin::getDescriptor() const {
     if constexpr (std::is_same_v<T, MaxDistanceConfig>) {
       return absl::StrCat("MaxDistJoin ", left, " to ", right, " of ",
                           config.maxDist_, " meter(s)");
+    } else if constexpr (std::is_same_v<T, SJConfig>) {
+      return absl::StrCat("Spatial Join ", left, " to ", right,
+                          " of type ", config.joinType_);
     } else {
       static_assert(std::is_same_v<T, NearestNeighborsConfig>);
       return absl::StrCat("NearestNeighborsJoin ", left, " to ", right,
@@ -225,7 +240,8 @@ size_t SpatialJoin::getCostEstimate() {
     } else {
       AD_CORRECTNESS_CHECK(
           config_.algo_ == SpatialJoinAlgorithm::S2_GEOMETRY ||
-              config_.algo_ == SpatialJoinAlgorithm::BOUNDING_BOX,
+              config_.algo_ == SpatialJoinAlgorithm::BOUNDING_BOX ||
+              config_.algo_ == SpatialJoinAlgorithm::LIBSPATIALJOIN,
           "Unknown SpatialJoin Algorithm.");
 
       // Let n be the size of the left table and m the size of the right table.
@@ -370,6 +386,7 @@ PreparedSpatialJoinParams SpatialJoin::prepareJoin() const {
   ColumnIndex leftJoinCol = childLeft_->getVariableColumn(config_.left_);
   ColumnIndex rightJoinCol = childRight_->getVariableColumn(config_.right_);
 
+
   // Payload cols and join col
   auto varsAndColInfo = copySortedByColumnIndex(getVarColMapPayloadVars());
   std::vector<ColumnIndex> rightSelectedCols;
@@ -383,7 +400,7 @@ PreparedSpatialJoinParams SpatialJoin::prepareJoin() const {
                                    idTableRight,      std::move(resultRight),
                                    leftJoinCol,       rightJoinCol,
                                    rightSelectedCols, numColumns,
-                                   getMaxDist(),      getMaxResults()};
+                                   getMaxDist(),      getMaxResults(), config_.joinType_};
 }
 
 // ____________________________________________________________________________
@@ -397,6 +414,8 @@ Result SpatialJoin::computeResult([[maybe_unused]] bool requestLaziness) {
     return algorithms.BaselineAlgorithm();
   } else if (config_.algo_ == SpatialJoinAlgorithm::S2_GEOMETRY) {
     return algorithms.S2geometryAlgorithm();
+  } else if (config_.algo_ == SpatialJoinAlgorithm::LIBSPATIALJOIN) {
+    return algorithms.LibspatialjoinAlgorithm();
   } else {
     AD_CORRECTNESS_CHECK(config_.algo_ == SpatialJoinAlgorithm::BOUNDING_BOX,
                          "Unknown SpatialJoin Algorithm.");

src/engine/SpatialJoin.h

@@ -14,6 +14,9 @@
 #include "parser/PayloadVariables.h"
 #include "parser/data/Variable.h"
 
+// Selection of a SpatialJoin join type
+enum class SpatialJoinJoinType { INTERSECTS, CONTAINS, COVERS, CROSSES, TOUCHES, EQUALS, OVERLAPS };
+
 // A nearest neighbor search with optionally a maximum distance.
 struct NearestNeighborsConfig {
   size_t maxResults_;
@@ -25,11 +28,16 @@ struct MaxDistanceConfig {
   size_t maxDist_;
 };
 
+// Full spatial join on selected join type
+struct SJConfig {
+  SpatialJoinJoinType joinType_;
+};
+
 // Configuration to restrict the results provided by the SpatialJoin
-using SpatialJoinTask = std::variant<NearestNeighborsConfig, MaxDistanceConfig>;
+using SpatialJoinTask = std::variant<NearestNeighborsConfig, MaxDistanceConfig, SJConfig>;
 
 // Selection of a SpatialJoin algorithm
-enum class SpatialJoinAlgorithm { BASELINE, S2_GEOMETRY, BOUNDING_BOX };
+enum class SpatialJoinAlgorithm { BASELINE, S2_GEOMETRY, BOUNDING_BOX, LIBSPATIALJOIN };
 const SpatialJoinAlgorithm SPATIAL_JOIN_DEFAULT_ALGORITHM =
     SpatialJoinAlgorithm::S2_GEOMETRY;
 
@@ -54,6 +62,8 @@ struct SpatialJoinConfiguration {
   // Choice of algorithm. Both algorithms have equal results, but different
   // runtime characteristics.
   SpatialJoinAlgorithm algo_ = SPATIAL_JOIN_DEFAULT_ALGORITHM;
+
+  SpatialJoinJoinType joinType_ = SpatialJoinJoinType::INTERSECTS;
 };
 
 // helper struct to improve readability in prepareJoin()
@@ -68,6 +78,7 @@ struct PreparedSpatialJoinParams {
   size_t numColumns_;
   std::optional<size_t> maxDist_;
   std::optional<size_t> maxResults_;
+  std::optional<SpatialJoinJoinType> joinType_;
 };
 
 // The spatial join operation without a limit on the maximum number of results