[experiment] Use SQLiteCpp

irq0 · Oct 18, 2023 · 71bf8d6 · 71bf8d6
1 parent 73fafa8
commit 71bf8d6
Show file tree

Hide file tree

Showing 8 changed files with 76 additions and 64 deletions.
diff --git a/.gitmodules b/.gitmodules
@@ -75,3 +75,6 @@
 [submodule "src/rgw/driver/sfs/sqlite/sqlite_orm"]
 	path = src/rgw/driver/sfs/sqlite/sqlite_orm
 	url = https://github.com/fnc12/sqlite_orm.git
+[submodule "src/rgw/driver/sfs/SQLiteCpp"]
+	path = src/rgw/driver/sfs/SQLiteCpp
+	url = https://github.com/SRombauts/SQLiteCpp.git
diff --git a/src/rgw/CMakeLists.txt b/src/rgw/CMakeLists.txt
@@ -217,6 +217,7 @@ if(WITH_RADOSGW_DBSTORE)
 endif()
 if(WITH_RADOSGW_SFS)
   include_directories(SYSTEM "${CMAKE_SOURCE_DIR}/src/rgw/driver/sfs/sqlite/sqlite_orm/include")
+  include_directories(SYSTEM "${CMAKE_SOURCE_DIR}/src/rgw/driver/sfs/SQLiteCpp/include")
   add_subdirectory(driver/sfs)
   list(APPEND librgw_common_srcs rgw_sal_sfs.cc rgw_s3gw_telemetry.cc)
 endif()

diff --git a/src/rgw/driver/sfs/CMakeLists.txt b/src/rgw/driver/sfs/CMakeLists.txt
@@ -13,6 +13,11 @@ find_package(SQLite3 REQUIRED)
 set(SQLITE_COMPILE_FLAGS "-DSQLITE_THREADSAFE=1")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${SQLITE_COMPILE_FLAGS}")
 
+set(SQLITECPP_RUN_CPPCHECK OFF CACHE BOOL "" FORCE)
+set(SQLITECPP_RUN_CPPLINT OFF CACHE BOOL "" FORCE)
+set(SQLITECPP_USE_STATIC_RUNTIME OFF CACHE BOOL "" FORCE)
+add_subdirectory(SQLiteCpp)
+
 set(sfs_srcs
   sqlite/sqlite_users.cc
   sqlite/sqlite_buckets.cc
@@ -56,6 +61,7 @@ target_compile_options(sfs PRIVATE
 # Mark ceph includes system to not see "our" warnings there
 target_include_directories(sfs
   SYSTEM PRIVATE "${CMAKE_SOURCE_DIR}/src/fmt/include"
+  SYSTEM PRIVATE "${CMAKE_SOURCE_DIR}/src/rgw/driver/sfs/SQLiteCpp/include"
   SYSTEM PRIVATE "${CMAKE_SOURCE_DIR}/src/rgw"
   SYSTEM PRIVATE "${CMAKE_SOURCE_DIR}/src"
   SYSTEM PRIVATE "${CMAKE_SOURCE_DIR}/src/include")
@@ -65,5 +71,5 @@ if(WITH_JAEGER)
   list(APPEND link_targets jaeger_base)
 endif()
 
-set(CMAKE_LINK_LIBRARIES ${CMAKE_LINK_LIBRARIES} ${link_targets} rgw_common sqlite3 pthread)
+set(CMAKE_LINK_LIBRARIES ${CMAKE_LINK_LIBRARIES} ${link_targets} rgw_common sqlite3 pthread SQLiteCpp)
 target_link_libraries(sfs PRIVATE ${CMAKE_LINK_LIBRARIES})
diff --git a/src/rgw/driver/sfs/SQLiteCpp b/src/rgw/driver/sfs/SQLiteCpp
diff --git a/src/rgw/driver/sfs/sqlite/dbconn.h b/src/rgw/driver/sfs/sqlite/dbconn.h
@@ -31,6 +31,7 @@
 #include "sqlite_orm.h"
 #include "users/users_definitions.h"
 #include "versioned_object/versioned_object_definitions.h"
+#include "SQLiteCpp/SQLiteCpp.h"
 
 namespace rgw::sal::sfs::sqlite {
 
@@ -272,6 +273,8 @@ class DBConn {
 
   inline auto get_storage() const { return storage; }
 
+  auto get_sqlitecpp() const { return SQLite::Database(get_storage().filename()); }
+
   static std::string getDBPath(CephContext* cct) {
     auto rgw_sfs_path = cct->_conf.get_val<std::string>("rgw_sfs_data_path");
     auto db_path =

diff --git a/src/rgw/driver/sfs/sqlite/sqlite_list.cc b/src/rgw/driver/sfs/sqlite/sqlite_list.cc
@@ -13,6 +13,8 @@
 
 #include <limits>
 
+#include "SQLiteCpp/Database.h"
+#include "SQLiteCpp/Statement.h"
 #include "rgw/driver/sfs/sqlite/conversion_utils.h"
 #include "rgw/driver/sfs/sqlite/objects/object_definitions.h"
 #include "rgw/driver/sfs/sqlite/versioned_object/versioned_object_definitions.h"
@@ -101,76 +103,69 @@ bool SQLiteList::versions(
   // more available logic: request one more than max. if we get that
   // much set out_more_available, but return only up to max
   ceph_assert(max < std::numeric_limits<size_t>::max());
-  const size_t query_limit = max + 1;
+  const uint32_t query_limit = max + 1;
 
-  auto storage = conn->get_storage();
-  auto rows = storage.select(
-      columns(
-          &DBObject::name, &DBVersionedObject::version_id,
-          &DBVersionedObject::mtime, &DBVersionedObject::etag,
-          &DBVersionedObject::size, &DBVersionedObject::version_type,
-          is_equal(
-              // IsLatest logic
-              // - Use the id as secondary condition if multiple version
-              // with same max(commit_time) exists
-              sqlite_orm::select(
-                  &DBVersionedObject::id, from<DBVersionedObject>(),
-                  where(
-                      is_equal(
-                          &DBObject::uuid, &DBVersionedObject::object_id
-                      ) and
-                      is_equal(
-                          &DBVersionedObject::object_state,
-                          ObjectState::COMMITTED
-                      )
-                  ),
-                  multi_order_by(
-                      order_by(&DBVersionedObject::commit_time).desc(),
-                      order_by(&DBVersionedObject::id).desc()
-                  ),
-                  limit(1)
-              ),
-              &DBVersionedObject::id
-          )
-      ),
-      inner_join<DBVersionedObject>(
-          on(is_equal(&DBObject::uuid, &DBVersionedObject::object_id))
-      ),
-      where(
-          is_equal(&DBVersionedObject::object_state, ObjectState::COMMITTED) and
-          is_equal(&DBObject::bucket_id, bucket_id) and
-          greater_than(&DBObject::name, start_after_object_name) and
-          prefix_to_like(&DBObject::name, prefix)
-      ),
-      // Sort:
-      // names a-Z
-      // first delete markers, then versions - (See: LC CurrentExpiration)
-      // newest to oldest version
-      multi_order_by(
-          order_by(&DBObject::name).asc(),
-          order_by(&DBVersionedObject::commit_time).desc(),
-          order_by(&DBVersionedObject::id).desc()
-      ),
-      limit(query_limit)
-  );
+  SQLite::Database db = conn->get_sqlitecpp();
+  SQLite::Statement query{
+    db, R"sql(
+      SELECT
+         o.name, vo.version_id, vo.mtime, vo.etag, vo.size, vo.version_type,
+         (vo.id = ( SELECT id FROM versioned_objects
+           WHERE object_id = o.uuid
+           AND object_state = ?
+           ORDER BY commit_time desc, id desc
+           LIMIT 1
+         )) AS is_latest
+      FROM objects as o
+      INNER JOIN versioned_objects as vo
+      ON (o.uuid = vo.object_id)
+      WHERE vo.object_state = ?
+      AND o.bucket_id = ?
+      AND o.name > ?
+      AND o.name LIKE ? ESCAPE '\'
+      ORDER BY o.name ASC,
+        vo.commit_time DESC,
+        vo.id DESC
+      LIMIT ?)sql"};
 
-  ceph_assert(rows.size() <= static_cast<size_t>(query_limit));
-  const size_t return_limit = std::min(max, rows.size());
-  out.reserve(return_limit);
-  for (size_t i = 0; i < return_limit; i++) {
-    const auto& row = rows[i];
+  query.bind(1, static_cast<int>(ObjectState::COMMITTED));
+  query.bind(2, static_cast<int>(ObjectState::COMMITTED));
+  query.bind(3, bucket_id);
+  query.bind(4, start_after_object_name);
+  query.bind(6, query_limit);
+  std::string like_expr;
+  like_expr.reserve(prefix.length() + 10);
+  for (const char c : prefix) {
+    switch (c) {
+      case '%':
+      case '_':
+        like_expr.push_back('\\');
+      default:
+        like_expr.push_back(c);
+    }
+  }
+  like_expr.push_back('%');
+  query.bind(5, like_expr);
+
+  out.reserve(max);
+  while (query.executeStep() && out.size() < max) {
     rgw_bucket_dir_entry e;
-    e.key.name = std::get<0>(row);
-    e.key.instance = std::get<1>(row);
-    e.meta.mtime = std::get<2>(row);
-    e.meta.etag = std::get<3>(row);
-    e.meta.size = std::get<4>(row);
+    e.key.name = query.getColumn(0).getString();
+    e.key.instance = query.getColumn(1).getString();
+    e.meta.mtime =
+        ceph::real_time(std::chrono::nanoseconds(query.getColumn(2).getInt64())
+        );
+    e.meta.etag = query.getColumn(3).getString();
+    e.meta.size = query.getColumn(4).getInt64();
     e.meta.accounted_size = e.meta.size;
-    e.flags = to_dentry_flag(std::get<5>(row), std::get<6>(row));
+    e.flags = to_dentry_flag(
+        static_cast<VersionType>(query.getColumn(5).getInt()),
+        query.getColumn(6).getInt()
+    );
     out.emplace_back(e);
   }
   if (out_more_available) {
-    *out_more_available = rows.size() == query_limit;
+    *out_more_available = !query.isDone();
   }
   return true;
 }

diff --git a/src/test/rgw/CMakeLists.txt b/src/test/rgw/CMakeLists.txt
@@ -18,6 +18,7 @@ endif()
 
 if(WITH_RADOSGW_SFS)
   include_directories(SYSTEM "${CMAKE_SOURCE_DIR}/src/rgw/driver/sfs/sqlite/sqlite_orm/include")
+  include_directories(SYSTEM "${CMAKE_SOURCE_DIR}/src/rgw/driver/sfs/SQLiteCpp/include")
 endif()
 
 if(WITH_JAEGER)

diff --git a/src/test/rgw/sfs/test_rgw_sfs_sqlite_list.cc b/src/test/rgw/sfs/test_rgw_sfs_sqlite_list.cc
@@ -47,6 +47,8 @@ class TestSFSList : public ::testing::Test {
         database_directory(create_database_directory()) {
     cct->_conf.set_val("rgw_sfs_data_path", database_directory);
     cct->_conf.set_val("rgw_sfs_sqlite_profile", "1");
+    cct->_conf.set_val("rgw_sfs_sqlite_profile_slowlog_time", "0");
+    cct->_log->set_stderr_level(5, 5);
     cct->_log->start();
     rgw_perf_start(cct.get());
   }