Add Column Sketches to Bliss (#55)

We added Column Sketches as an optional, we can use: cmake .. -DCOMPILE_COLUMNSKETCHES=ON to turn on building Column Sketches

Note that Column Sketches is a private repo, and by default the option is turned off.

CMakeLists.txt

@@ -46,6 +46,9 @@ add_subdirectory(tests)
 # HEADER bliss
 #   Bliss lib files
 # =============================================================================
+
+option(COMPILE_COLUMNSKETCHES "compile_column_sketches" OFF)
+
 add_library(bliss OBJECT
     ${CMAKE_SOURCE_DIR}/src/bliss/util/timer.h
     ${CMAKE_SOURCE_DIR}/src/bliss/util/reader.h
@@ -55,6 +58,7 @@ add_library(bliss OBJECT
     ${CMAKE_SOURCE_DIR}/src/bliss/bench_lipp.h
     ${CMAKE_SOURCE_DIR}/src/bliss/bench_alex.h
     ${CMAKE_SOURCE_DIR}/src/bliss/bench_btree.h
+    ${CMAKE_SOURCE_DIR}/src/bliss/bench_columnsketches.h
     ${CMAKE_SOURCE_DIR}/src/bliss/bench_skiplist.h
     ${CMAKE_SOURCE_DIR}/src/bliss/bench_pgm.h
     ${CMAKE_SOURCE_DIR}/src/bliss/bench_art.h
@@ -76,6 +80,12 @@ target_link_libraries(bliss PUBLIC
     art
 )
 
+if(COMPILE_COLUMNSKETCHES)
+    target_link_libraries(bliss PUBLIC
+        columnsketches
+    )
+endif()
+
 target_include_directories(bliss PUBLIC
     ${CMAKE_SOURCE_DIR}/src
 )

external/CMakeLists.txt

@@ -72,6 +72,33 @@ endif()
 add_library(tlx INTERFACE)
 target_include_directories(tlx INTERFACE ${tlx_SOURCE_DIR}/)
 
+FetchContent_Declare(
+        columnsketches
+        GIT_REPOSITORY https://github.com/altramarine/ColumnSketches-cpp.git
+        GIT_TAG main
+)
+
+if(COMPILE_COLUMNSKETCHES)
+    FetchContent_GetProperties(columnsketches)
+    if (NOT columnsketches_POPULATED)
+        FetchContent_Populate(columnsketches)
+    endif()
+    file(GLOB_RECURSE columnsketches_SOURCES
+    "${columnsketches_SOURCE_DIR}/src/*.cpp"
+    )
+    add_library(columnsketches STATIC ${columnsketches_SOURCES})
+    target_include_directories(columnsketches PUBLIC ${columnsketches_SOURCE_DIR}/include)
+    target_link_libraries(columnsketches PUBLIC pthread)
+    target_compile_options(columnsketches PUBLIC 
+    "-pthread"
+    "-mavx"
+    "-mavx2"
+    "-msse"
+    "-msse2"
+    "-O3"
+    "-DCOMPILE_COLUMNSKETCHES=1"
+    )
+endif()
 
 FetchContent_Declare(
     skiplist
@@ -114,4 +141,4 @@ if (NOT pgm_POPULATED)
 endif()
 
 add_library(pgm INTERFACE)
-target_include_directories(pgm INTERFACE ${pgm_SOURCE_DIR})
+target_include_directories(pgm INTERFACE ${pgm_SOURCE_DIR})

src/bliss/bench_columnsketches.h

@@ -0,0 +1,56 @@
+#ifndef BLISS_BENCH_COLUMNSKETCHES
+#define BLISS_BENCH_COLUMNSKETCHES
+#ifdef COMPILE_COLUMNSKETCHES
+#include <vector>
+
+#include "bliss/bliss_index.h"
+#include "column_sketches.h"
+
+namespace bliss {
+
+template <typename KEY_TYPE, typename VALUE_TYPE>
+class BlissColumnSketchesIndex : public BlissIndex<KEY_TYPE, VALUE_TYPE> {
+   public:
+    BlissColumnSketchesIndex() {
+        columnsketches_ = new ColumnSketchesInteger();
+    };
+
+    ~BlissColumnSketchesIndex()  {
+        delete columnsketches_;
+    }
+
+    void bulkload(
+        std::vector<std::pair<KEY_TYPE, VALUE_TYPE>> values) override {
+        std::vector<int32_t> vals;
+        for(auto x: values) {
+            vals.push_back(x.second);
+        }
+        columnsketches_->create(&vals[0], vals.size(), vals.size());
+    }
+
+    bool get(VALUE_TYPE start, VALUE_TYPE end) {
+        uint32_t *res = columnsketches_->query_ge_lt(start, end);
+        bool succeed = res != nullptr;
+        if(succeed) delete res;
+        return succeed;
+    }
+
+    bool get(KEY_TYPE key) override { 
+        std::runtime_error("Column Imprints does not support get(key, value).");
+        return false; 
+    }
+
+    void put(KEY_TYPE key, VALUE_TYPE value) {
+        std::runtime_error("Column Imprints does not support put(key, value).");
+        return;
+    }
+
+    void end_routine() override {}
+  private:
+
+    ColumnSketchesInteger *columnsketches_;
+};
+
+}  // namespace bliss
+#endif
+#endif  // !BLISS_BENCH_BTREE

src/bliss/bliss_index.h

@@ -18,4 +18,4 @@ class BlissIndex {
 
 }  // namespace bliss
 
-#endif  // !BLISS_INDEX
+#endif  // !BLISS_INDEX

src/bliss_bench.cpp

@@ -11,6 +11,7 @@
 #include "bliss/bench_btree.h"
 #include "bliss/bench_leveldb.h"
 #include "bliss/bench_lipp.h"
+#include "bliss/bench_columnsketches.h"
 #include "bliss/bench_pgm.h"
 #include "bliss/bench_skiplist.h"
 #include "bliss/bliss_index.h"
@@ -174,6 +175,13 @@ int main(int argc, char *argv[]) {
         index.reset(new bliss::BlissLippIndex<key_type, value_type>());
     } else if (config.index == "btree") {
         index.reset(new bliss::BlissBTreeIndex<key_type, value_type>());
+    } else if (config.index == "columnskteches") {
+#ifdef COMPILE_COLUMNSKETCHES
+        index.reset(new bliss::BlissColumnSketchesIndex<key_type, value_type>());
+#else
+        std::runtime_error("Column Sketches is not Imported");
+        exit(0);
+#endif
     } else if (config.index == "skiplist") {
         index.reset(new bliss::BlissSkipListIndex<key_type, value_type>());
     } else if (config.index == "art") {