Initial import of code and dataset

Author: transactionalblog

.gitignore

@@ -0,0 +1,12 @@
+dataset/public.init
+dataset/public.result
+dataset/public.work
+dataset/r0
+dataset/r1
+dataset/r2
+dataset/r3
+dataset/r4
+dataset/r5
+dataset/r6
+dataset/r7
+dataset/r8

CMakeLists.txt

@@ -0,0 +1,37 @@
+cmake_minimum_required (VERSION 2.6)
+project (Sig18)
+
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
+# set(CMAKE_C_COMPILER /usr/local/bin/gcc)
+# set(CMAKE_CXX_COMPILER /usr/local/bin/g++)
+set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -g -fsanitize=address -Wall")
+
+include_directories(include)
+
+
+add_library(database Relation.cpp Operators.cpp Parser.cpp Utils.cpp Joiner.cpp)
+target_include_directories(database PUBLIC
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
+    $<INSTALL_INTERFACE:include>
+    PRIVATE src)
+
+OPTION(FORCE_TESTS "Build tests, regardless of build type." ON)
+if (CMAKE_BUILD_TYPE MATCHES "[Dd][Ee][Bb][Uu][Gg]" OR FORCE_TESTS)
+    add_subdirectory(test)
+endif()
+
+add_executable(Driver main.cpp)
+target_link_libraries(Driver database)
+
+# Interactive command line tool to translate our query format to SQL queries
+add_executable(Query2SQL Query2SQL.cpp)
+target_link_libraries(Query2SQL database)
+
+# Test harness
+add_executable(harness harness.cpp)
+
+ADD_CUSTOM_TARGET(link_target ALL
+  COMMAND ${CMAKE_COMMAND} -E create_symlink ${CMAKE_CURRENT_SOURCE_DIR}/workloads
+  ${CMAKE_CURRENT_BINARY_DIR}/workloads)

Joiner.cpp

@@ -0,0 +1,115 @@
+#include "Joiner.hpp"
+#include <cassert>
+#include <iostream>
+#include <string>
+#include <unordered_map>
+#include <utility>
+#include <set>
+#include <sstream>
+#include <vector>
+#include "Parser.hpp"
+//---------------------------------------------------------------------------
+using namespace std;
+//---------------------------------------------------------------------------
+void Joiner::addRelation(const char* fileName)
+// Loads a relation from disk
+{
+  relations.emplace_back(fileName);
+}
+//---------------------------------------------------------------------------
+Relation& Joiner::getRelation(unsigned relationId)
+// Loads a relation from disk
+{
+  if (relationId >= relations.size()) {
+    cerr << "Relation with id: " << relationId << " does not exist" << endl;
+    throw;
+  }
+  return relations[relationId];
+}
+//---------------------------------------------------------------------------
+unique_ptr<Operator> Joiner::addScan(set<unsigned>& usedRelations,SelectInfo& info,QueryInfo& query)
+  // Add scan to query
+{
+  usedRelations.emplace(info.binding);
+  vector<FilterInfo> filters;
+    for (auto& f : query.filters) {
+      if (f.filterColumn.binding==info.binding) {
+        filters.emplace_back(f);
+      }
+    }
+  return filters.size()?make_unique<FilterScan>(getRelation(info.relId),filters):make_unique<Scan>(getRelation(info.relId),info.binding);
+}
+//---------------------------------------------------------------------------
+enum QueryGraphProvides {  Left, Right, Both, None };
+//---------------------------------------------------------------------------
+static QueryGraphProvides analyzeInputOfJoin(set<unsigned>& usedRelations,SelectInfo& leftInfo,SelectInfo& rightInfo)
+  // Analyzes inputs of join
+{
+  bool usedLeft=usedRelations.count(leftInfo.binding);
+  bool usedRight=usedRelations.count(rightInfo.binding);
+
+  if (usedLeft^usedRight)
+    return usedLeft?QueryGraphProvides::Left:QueryGraphProvides::Right;
+  if (usedLeft&&usedRight)
+    return QueryGraphProvides::Both;
+  return QueryGraphProvides::None;
+}
+//---------------------------------------------------------------------------
+string Joiner::join(QueryInfo& query)
+  // Executes a join query
+{
+  //cerr << query.dumpText() << endl;
+  set<unsigned> usedRelations;
+
+  // We always start with the first join predicate and append the other joins to it (--> left-deep join trees)
+  // You might want to choose a smarter join ordering ...
+  auto& firstJoin=query.predicates[0];
+  auto left=addScan(usedRelations,firstJoin.left,query);
+  auto right=addScan(usedRelations,firstJoin.right,query);
+  unique_ptr<Operator> root=make_unique<Join>(move(left),move(right),firstJoin);
+
+
+
+  for (unsigned i=1;i<query.predicates.size();++i) {
+    auto& pInfo=query.predicates[i];
+    auto& leftInfo=pInfo.left; auto& rightInfo=pInfo.right;
+    unique_ptr<Operator> left, right;
+    switch(analyzeInputOfJoin(usedRelations,leftInfo,rightInfo)) {
+      case QueryGraphProvides::Left:
+        left=move(root);
+        right=addScan(usedRelations,rightInfo,query);
+        root=make_unique<Join>(move(left),move(right),pInfo);
+        break;
+      case QueryGraphProvides::Right:
+        left=addScan(usedRelations,leftInfo,query);
+        right=move(root);
+        root=make_unique<Join>(move(left),move(right),pInfo);
+        break;
+      case QueryGraphProvides::Both:
+        // All relations of this join are already used somewhere else in the query.
+        // Thus, we have either a cycle in our join graph or more than one join predicate per join.
+        root=make_unique<SelfJoin>(move(root),pInfo);
+        break;
+      case QueryGraphProvides::None:
+        // Process this predicate later when we can connect it to the other joins
+        // We never have cross products
+        query.predicates.push_back(pInfo);
+        break;
+    };
+  }
+
+  Checksum checkSum(move(root),query.selections);
+  checkSum.run();
+
+
+  stringstream out;
+  auto& results=checkSum.checkSums;
+  for (unsigned i=0;i<results.size();++i) {
+    out << (checkSum.resultSize==0?"NULL":to_string(results[i]));
+    if (i<results.size()-1)
+      out << " ";
+  }
+  out << "\n";
+  return out.str();
+}
+//---------------------------------------------------------------------------