Add generic chunked fill and copy algorithms (#1564)

These are replacements for `std::ranges::fill` and `std::ranges::copy` that perform their respective loops in chunks of a dedicated size. After each chunk, a user-defined callback is invoked, which can be used e.g. for cancellation checks. The helpers are used within the `UNION` and `BIND` operations to add more cancellation checks there.
ad-freiburg · Oct 18, 2024 · bf36257 · bf36257
1 parent 264919e
commit bf36257
Show file tree

Hide file tree

Showing 5 changed files with 110 additions and 37 deletions.
diff --git a/src/engine/Bind.cpp b/src/engine/Bind.cpp
@@ -8,6 +8,7 @@
 #include "engine/QueryExecutionTree.h"
 #include "engine/sparqlExpressions/SparqlExpression.h"
 #include "engine/sparqlExpressions/SparqlExpressionGenerators.h"
+#include "util/ChunkedForLoop.h"
 #include "util/Exception.h"
 
 // BIND adds exactly one new column
@@ -170,9 +171,11 @@ IdTable Bind::computeExpressionBind(
           getInternallyVisibleVariableColumns().at(singleResult).columnIndex_;
       auto inputColumn = idTable.getColumn(columnIndex);
       AD_CORRECTNESS_CHECK(inputColumn.size() == outputColumn.size());
-      std::ranges::copy(inputColumn, outputColumn.begin());
+      ad_utility::chunkedCopy(inputColumn, outputColumn.begin(), CHUNK_SIZE,
+                              [this]() { checkCancellation(); });
     } else if constexpr (isStrongId) {
-      std::ranges::fill(outputColumn, singleResult);
+      ad_utility::chunkedFill(outputColumn, singleResult, CHUNK_SIZE,
+                              [this]() { checkCancellation(); });
     } else {
       constexpr bool isConstant = sparqlExpression::isConstantResult<T>;
 
@@ -187,7 +190,8 @@ IdTable Bind::computeExpressionBind(
               sparqlExpression::detail::constantExpressionResultToId(
                   std::move(*it), *outputLocalVocab);
           checkCancellation();
-          std::ranges::fill(outputColumn, constantId);
+          ad_utility::chunkedFill(outputColumn, constantId, CHUNK_SIZE,
+                                  [this]() { checkCancellation(); });
         }
       } else {
         size_t i = 0;

diff --git a/src/engine/Union.cpp b/src/engine/Union.cpp
@@ -6,6 +6,7 @@
 #include "Union.h"
 
 #include "engine/CallFixedSize.h"
+#include "util/ChunkedForLoop.h"
 #include "util/TransparentFunctors.h"
 
 const size_t Union::NO_COLUMN = std::numeric_limits<size_t>::max();
@@ -184,26 +185,6 @@ ProtoResult Union::computeResult(bool requestLaziness) {
           Result::getMergedLocalVocab(*subRes1, *subRes2)};
 }
 
-// _____________________________________________________________________________
-void Union::copyChunked(auto beg, auto end, auto target) const {
-  size_t total = end - beg;
-  for (size_t i = 0; i < total; i += chunkSize) {
-    checkCancellation();
-    size_t actualEnd = std::min(i + chunkSize, total);
-    std::copy(beg + i, beg + actualEnd, target + i);
-  }
-}
-
-// _____________________________________________________________________________
-void Union::fillChunked(auto beg, auto end, const auto& value) const {
-  size_t total = end - beg;
-  for (size_t i = 0; i < total; i += chunkSize) {
-    checkCancellation();
-    size_t actualEnd = std::min(i + chunkSize, total);
-    std::fill(beg + i, beg + actualEnd, value);
-  }
-};
-
 // _____________________________________________________________________________
 IdTable Union::computeUnion(
     const IdTable& left, const IdTable& right,
@@ -220,11 +201,14 @@ IdTable Union::computeUnion(
                             size_t inputColumnIndex, size_t offset) {
     if (inputColumnIndex != NO_COLUMN) {
       decltype(auto) input = inputTable.getColumn(inputColumnIndex);
-      copyChunked(input.begin(), input.end(), targetColumn.begin() + offset);
+      ad_utility::chunkedCopy(input, targetColumn.begin() + offset, chunkSize,
+                              [this]() { checkCancellation(); });
     } else {
-      fillChunked(targetColumn.begin() + offset,
-                  targetColumn.begin() + offset + inputTable.size(),
-                  Id::makeUndefined());
+      ad_utility::chunkedFill(
+          std::ranges::subrange{
+              targetColumn.begin() + offset,
+              targetColumn.begin() + offset + inputTable.size()},
+          Id::makeUndefined(), chunkSize, [this]() { checkCancellation(); });
     }
   };
 
@@ -263,8 +247,9 @@ IdTable Union::transformToCorrectColumnFormat(
     IdTable idTable, const std::vector<ColumnIndex>& permutation) const {
   while (idTable.numColumns() < getResultWidth()) {
     idTable.addEmptyColumn();
-    auto column = idTable.getColumn(idTable.numColumns() - 1);
-    fillChunked(column.begin(), column.end(), Id::makeUndefined());
+    ad_utility::chunkedFill(idTable.getColumn(idTable.numColumns() - 1),
+                            Id::makeUndefined(), chunkSize,
+                            [this]() { checkCancellation(); });
   }
 
   idTable.setColumnSubset(permutation);

diff --git a/src/engine/Union.h b/src/engine/Union.h
@@ -63,13 +63,6 @@ class Union : public Operation {
   }
 
  private:
-  // A drop-in replacement for `std::copy` that performs the copying in chunks
-  // of `chunkSize` and checks the timeout after each chunk.
-  void copyChunked(auto beg, auto end, auto target) const;
-
-  // A similar timeout-checking replacement for `std::fill`.
-  void fillChunked(auto beg, auto end, const auto& value) const;
-
   ProtoResult computeResult(bool requestLaziness) override;
 
   VariableToColumnMap computeVariableToColumnMap() const override;

diff --git a/src/util/ChunkedForLoop.h b/src/util/ChunkedForLoop.h
@@ -63,6 +63,56 @@ inline void chunkedForLoop(std::size_t start, std::size_t end,
     std::invoke(chunkOperation);
   }
 }
+
+// Helper concept that combines the sized range and input range concepts.
+template <typename R>
+concept SizedInputRange =
+    std::ranges::sized_range<R> && std::ranges::input_range<R>;
+
+// Similar to `std::ranges::copy`, but invokes `chunkOperation` every
+// `chunkSize` elements. (Round up to the next chunk size if the range size is
+// not a multiple of `chunkSize`.)
+template <SizedInputRange R, std::weakly_incrementable O>
+inline void chunkedCopy(R&& inputRange, O result,
+                        std::ranges::range_difference_t<R> chunkSize,
+                        const std::invocable auto& chunkOperation)
+    requires std::indirectly_copyable<std::ranges::iterator_t<R>, O> {
+  auto begin = std::ranges::begin(inputRange);
+  auto end = std::ranges::end(inputRange);
+  auto target = result;
+  while (std::ranges::distance(begin, end) >= chunkSize) {
+    auto start = begin;
+    std::ranges::advance(begin, chunkSize);
+    target = std::ranges::copy(start, begin, target).out;
+    chunkOperation();
+  }
+  std::ranges::copy(begin, end, target);
+  chunkOperation();
+}
+
+// Helper concept that combines the sized range and output range concepts.
+template <typename R, typename T>
+concept SizedOutputRange =
+    std::ranges::sized_range<R> && std::ranges::output_range<R, T>;
+
+// Similar to `std::ranges::fill`, but invokes `chunkOperation` every
+// `chunkSize` elements. (Round up to the next chunk size if the range size is
+// not a multiple of `chunkSize`.)
+template <typename T, SizedOutputRange<T> R>
+inline void chunkedFill(R&& outputRange, const T& value,
+                        std::ranges::range_difference_t<R> chunkSize,
+                        const std::invocable auto& chunkOperation) {
+  auto begin = std::ranges::begin(outputRange);
+  auto end = std::ranges::end(outputRange);
+  while (std::ranges::distance(begin, end) >= chunkSize) {
+    auto start = begin;
+    std::ranges::advance(begin, chunkSize);
+    std::ranges::fill(start, begin, value);
+    chunkOperation();
+  }
+  std::ranges::fill(begin, end, value);
+  chunkOperation();
+}
 }  // namespace ad_utility
 
 #endif  // QLEVER_CHUNKEDFORLOOP_H
diff --git a/test/ChunkedForLoopTest.cpp b/test/ChunkedForLoopTest.cpp
@@ -2,12 +2,14 @@
 //   Chair of Algorithms and Data Structures.
 //   Author: Robin Textor-Falconi <[email protected]>
 
-#include <gtest/gtest.h>
+#include <gmock/gmock.h>
 
 #include <atomic>
 
 #include "util/ChunkedForLoop.h"
 
+using ad_utility::chunkedCopy;
+using ad_utility::chunkedFill;
 using ad_utility::chunkedForLoop;
 
 TEST(ChunkedForLoop, testEmptyRange) {
@@ -103,3 +105,42 @@ TEST(ChunkedForLoop, verifyBreakWorksAsExpected) {
   EXPECT_EQ(counter, 4);
   EXPECT_EQ(chunkCounter, 1);
 }
+
+// _____________________________________________________________________________________________________________________
+TEST(ChunkedForLoop, chunkedFillHandlesEmptyRange) {
+  size_t chunkCounter = 0;
+  chunkedFill(std::array<int, 0>{}, 0, 10, [&]() { chunkCounter++; });
+
+  EXPECT_EQ(chunkCounter, 1);
+}
+
+// _____________________________________________________________________________________________________________________
+TEST(ChunkedForLoop, chunkedFillFillsCorrectly) {
+  size_t chunkCounter = 0;
+  std::array<int, 21> elements{};
+  chunkedFill(elements, 42, 10, [&]() { chunkCounter++; });
+
+  EXPECT_EQ(chunkCounter, 3);
+  EXPECT_THAT(elements, ::testing::Each(::testing::Eq(42)));
+}
+
+// _____________________________________________________________________________________________________________________
+TEST(ChunkedForLoop, chunkedCopyHandlesEmptyRange) {
+  size_t chunkCounter = 0;
+  std::array<int, 0> output{};
+  chunkedCopy(std::array<int, 0>{}, output.begin(), 2,
+              [&]() { chunkCounter++; });
+
+  EXPECT_EQ(chunkCounter, 1);
+}
+
+// _____________________________________________________________________________________________________________________
+TEST(ChunkedForLoop, chunkedCopyCopiesCorrectly) {
+  size_t chunkCounter = 0;
+  std::array<int, 5> input{5, 4, 3, 2, 1};
+  std::array<int, 5> output{};
+  chunkedCopy(input, output.begin(), 2, [&]() { chunkCounter++; });
+
+  EXPECT_EQ(chunkCounter, 3);
+  EXPECT_EQ(input, output);
+}