Use bit manipulation with the UniqueIDGenSvc and add a check for repe…

…ated numbers (#247)

Co-authored-by: Thomas Madlener <[email protected]>

k4FWCore/components/UniqueIDGenSvc.cpp

@@ -18,27 +18,53 @@
  */
 #include "UniqueIDGenSvc.h"
 
-DECLARE_COMPONENT(UniqueIDGenSvc)
+#include <cstdint>
+#include <stdexcept>
+#include <string>
 
 UniqueIDGenSvc::UniqueIDGenSvc(const std::string& name, ISvcLocator* svcLoc) : base_class(name, svcLoc) {}
 
-StatusCode UniqueIDGenSvc::initialize() {
-  StatusCode sc = Service::initialize();
-  return sc;
-}
-
-const size_t bits32    = std::numeric_limits<uint32_t>::digits;
-const size_t bits64    = std::numeric_limits<uint64_t>::digits;
-const size_t bitsSizeT = std::numeric_limits<size_t>::digits;
+constexpr size_t bits32    = std::numeric_limits<uint32_t>::digits;
+constexpr size_t bits64    = std::numeric_limits<uint64_t>::digits;
+constexpr size_t bitsSizeT = std::numeric_limits<size_t>::digits;
 
 size_t UniqueIDGenSvc::getUniqueID(uint32_t evt_num, uint32_t run_num, const std::string& name) const {
-  std::bitset<bits64>    seed_bits(this->m_seed);
-  std::bitset<bits32>    event_num_bits(evt_num), run_num_bits(run_num);
-  size_t                 str_hash = std::hash<std::string>{}(name);
-  std::bitset<bitsSizeT> name_bits(str_hash);
+  std::bitset<bits64>    seed_bits      = this->m_seed.value();
+  std::bitset<bits32>    event_num_bits = evt_num, run_num_bits = run_num;
+  size_t                 str_hash  = std::hash<std::string>{}(name);
+  std::bitset<bitsSizeT> name_bits = str_hash;
+
+  std::bitset<bits64 + bits32 + bits32 + bitsSizeT> combined_bits;
 
-  std::bitset<bits64 + bits32 + bits32 + bitsSizeT> combined_bits(seed_bits.to_string() + event_num_bits.to_string() +
-                                                                  run_num_bits.to_string() + name_bits.to_string());
+  for (size_t i = 0; i < bitsSizeT; i++) {
+    combined_bits[i] = name_bits[i];
+  }
+  for (size_t i = 0; i < bits32; i++) {
+    combined_bits[i + bitsSizeT] = run_num_bits[i];
+  }
+  for (size_t i = 0; i < bits32; i++) {
+    combined_bits[i + bits32 + bitsSizeT] = event_num_bits[i];
+  }
+  for (size_t i = 0; i < bits64; i++) {
+    combined_bits[i + bits32 + bits32 + bitsSizeT] = seed_bits[i];
+  }
 
-  return std::hash<std::bitset<bits64 + bits32 + bits32 + bitsSizeT>>{}(combined_bits);
+  auto hash     = std::hash<std::bitset<bits64 + bits32 + bits32 + bitsSizeT>>{}(combined_bits);
+  bool inserted = false;
+  {
+    std::lock_guard<std::mutex> lock(m_mutex);
+    std::tie(std::ignore, inserted) = m_uniqueIDs.insert(hash);
+  }
+  if (!inserted) {
+    error() << "Event number " << evt_num << ", run number " << run_num << " and algorithm name \"" << name
+            << "\" have already been used. Please check the uniqueness of the event number, run number and name."
+            << endmsg;
+    if (m_throwIfDuplicate) {
+      throw std::runtime_error("Duplicate event number, run number and algorithm name");
+    }
+  }
+
+  return hash;
 }
+
+DECLARE_COMPONENT(UniqueIDGenSvc)

k4FWCore/components/UniqueIDGenSvc.h

@@ -19,11 +19,13 @@
 #ifndef FWCORE_UNIQUEIDGENSVC_H
 #define FWCORE_UNIQUEIDGENSVC_H
 
+#include "GaudiKernel/Service.h"
+#include "k4Interface/IUniqueIDGenSvc.h"
+
 #include <cstdint>
+#include <functional>
 #include <string>
-
-#include <GaudiKernel/Service.h>
-#include "k4Interface/IUniqueIDGenSvc.h"
+#include <unordered_set>
 
 /** @class UniqueIDGenSvc
  *  Generate unique, reproducible numbers using
@@ -34,11 +36,13 @@
 class UniqueIDGenSvc : public extends<Service, IUniqueIDGenSvc> {
 public:
   UniqueIDGenSvc(const std::string& name, ISvcLocator* svcLoc);
-  StatusCode initialize() override;
-  size_t     getUniqueID(uint32_t evt_num, uint32_t run_num, const std::string& name) const override;
+  size_t getUniqueID(uint32_t evt_num, uint32_t run_num, const std::string& name) const override;
 
 private:
-  Gaudi::Property<int64_t> m_seed{this, "Seed", {123456789}};
+  Gaudi::Property<uint64_t>                         m_seed{this, "Seed", {123456789}};
+  mutable std::unordered_set<size_t, std::identity> m_uniqueIDs;
+  mutable std::mutex                                m_mutex;
+  Gaudi::Property<bool>                             m_throwIfDuplicate{this, "ThrowIfDuplicate", {true}};
 };
 
 #endif

k4Interface/include/k4Interface/IUniqueIDGenSvc.h

@@ -19,10 +19,7 @@
 #ifndef FWCORE_IUNIQUEIDGENSVC_H
 #define FWCORE_IUNIQUEIDGENSVC_H
 
-#include <bitset>
 #include <cstdint>
-#include <functional>
-#include <limits>
 #include <string>
 
 #include <GaudiKernel/IInterface.h>

test/k4FWCoreTest/CMakeLists.txt

@@ -125,7 +125,8 @@ add_test(NAME checkKeepDropSwitch
          COMMAND python scripts/check_KeepDropSwitch.py ${PROJECT_BINARY_DIR}/test/k4FWCoreTest/output_k4test_exampledata_2.root)
 set_test_env(checkKeepDropSwitch)
 set_property(TEST checkKeepDropSwitch APPEND PROPERTY DEPENDS ReadExampleEventData)
-add_test_with_env(TestUniqueIDGenSvc options/TestUniqueIDGenSvc.py)
+add_test_with_env(TestUniqueIDGenSvc options/TestUniqueIDGenSvc.py -n 1)
+add_test_with_env(TestUniqueIDGenSvcRepeated options/TestUniqueIDGenSvc.py -n 2 PROPERTIES PASS_REGULAR_EXPRESSION "Duplicate event number, run number and algorithm name")
 add_test_with_env(TestEventHeaderFiller options/createEventHeader.py)
 add_test_with_env(EventHeaderCheck options/runEventHeaderCheck.py PROPERTIES DEPENDS TestEventHeaderFiller)
 add_test(NAME TestExec WORKING_DIRECTORY ${CMAKE_CURRENT_LIST_DIR} COMMAND python options/TestExec.py)

test/k4FWCoreTest/src/components/TestUniqueIDGenSvc.cpp

@@ -18,6 +18,8 @@
  */
 #include "TestUniqueIDGenSvc.h"
 
+#include <cstdint>
+
 DECLARE_COMPONENT(TestUniqueIDGenSvc)
 
 TestUniqueIDGenSvc::TestUniqueIDGenSvc(const std::string& aName, ISvcLocator* aSvcLoc)
@@ -33,16 +35,19 @@ StatusCode TestUniqueIDGenSvc::initialize() {
   return StatusCode::SUCCESS;
 }
 
+// This is meant to run up to two times
+// For the first event, check that when giving two different event numbers, the unique IDs are different
+// For the second event, the service throws when trying to get the same ID twice
 StatusCode TestUniqueIDGenSvc::execute(const EventContext&) const {
-  uint        evt_num = 4;
-  uint        run_num = 3;
+  ++m_counter;
+  uint32_t    evt_num = 4;
+  uint32_t    run_num = 3 + m_counter.sum();
   std::string name    = "Some algorithm name";
 
   auto uid       = m_service->getUniqueID(evt_num, run_num, name);
-  auto uid_again = m_service->getUniqueID(evt_num, run_num, name);
-
-  if (uid != uid_again) {
-    return StatusCode::FAILURE;
+  auto uid_again = m_service->getUniqueID(evt_num + (m_counter.sum() % 2), run_num, name);
+  if (uid == uid_again) {
+    throw std::runtime_error("Unique IDs are the same");
   }
 
   return StatusCode::SUCCESS;

test/k4FWCoreTest/src/components/TestUniqueIDGenSvc.h

@@ -20,7 +20,8 @@
 #define TEST_UNIQUEIDGENSVC_H
 
 // GAUDI
-#include <Gaudi/Algorithm.h>
+#include "Gaudi/Accumulators.h"
+#include "Gaudi/Algorithm.h"
 
 #include "k4Interface/IUniqueIDGenSvc.h"
 
@@ -37,7 +38,8 @@ class TestUniqueIDGenSvc : public Gaudi::Algorithm {
   StatusCode execute(const EventContext&) const final;
 
 private:
-  SmartIF<IUniqueIDGenSvc> m_service;
+  SmartIF<IUniqueIDGenSvc>               m_service;
+  mutable Gaudi::Accumulators::Counter<> m_counter{this, "EventCounter"};
 };
 
 #endif  // TEST_UNIQUEIDGENSVC_H