Consistently replace std::ranges by ql::ranges (#1729)

When building with `-DUSE_CPP_17_BACKPORT=ON`, then `range-v3` is used instead of `std::ranges` in all places (types, functions, concepts).
ad-freiburg · Jan 24, 2025 · 432cd64 · 432cd64
1 parent d79cf60
commit 432cd64
Show file tree

Hide file tree

Showing 44 changed files with 219 additions and 189 deletions.
diff --git a/benchmark/GroupByHashMapBenchmark.cpp b/benchmark/GroupByHashMapBenchmark.cpp
@@ -364,7 +364,7 @@ class GroupByHashMapBenchmark : public BenchmarkInterface {
     } else {
       firstColumn = generateRandomGroupVec(numInputRows, numGroups);
     }
-    std::ranges::transform(
+    ql::ranges::transform(
         firstColumn.begin(), firstColumn.end(), groupValues.begin(),
         [](size_t value) {
           return ValueId::makeFromInt(static_cast<int64_t>(value));
@@ -375,7 +375,7 @@ class GroupByHashMapBenchmark : public BenchmarkInterface {
     auto localVocab = LocalVocab{};
     if (valueTypes != ValueIdType::Strings) {
       auto secondColumn = generateRandomDoubleVec(numInputRows);
-      std::ranges::transform(
+      ql::ranges::transform(
           secondColumn.begin(), secondColumn.end(), otherValues.begin(),
           [&](double value) {
             if (valueTypes == ValueIdType::OnlyDouble)
@@ -396,10 +396,10 @@ class GroupByHashMapBenchmark : public BenchmarkInterface {
           numInputRows, randomStringLength);
       localVocab = std::move(newLocalVocab);
 
-      std::ranges::transform(indices.begin(), indices.end(),
-                             otherValues.begin(), [&](LocalVocabIndex idx) {
-                               return ValueId::makeFromLocalVocabIndex(idx);
-                             });
+      ql::ranges::transform(indices.begin(), indices.end(), otherValues.begin(),
+                            [&](LocalVocabIndex idx) {
+                              return ValueId::makeFromLocalVocabIndex(idx);
+                            });
     }
 
     std::vector<std::optional<Variable>> variables = {Variable{"?a"},

diff --git a/benchmark/JoinAlgorithmBenchmark.cpp b/benchmark/JoinAlgorithmBenchmark.cpp
@@ -115,7 +115,7 @@ struct SetOfIdTableColumnElements {
   */
   explicit SetOfIdTableColumnElements(
       const std::span<const ValueId>& idTableColumnRef) {
-    std::ranges::for_each(idTableColumnRef, [this](const ValueId& id) {
+    ql::ranges::for_each(idTableColumnRef, [this](const ValueId& id) {
       if (auto numOccurrencesIterator = numOccurrences_.find(id);
           numOccurrencesIterator != numOccurrences_.end()) {
         (numOccurrencesIterator->second)++;
@@ -190,7 +190,7 @@ static size_t createOverlapRandomly(IdTableAndJoinColumn* const smallerTable,
   // Create the overlap.
   ad_utility::HashMap<ValueId, std::reference_wrapper<const ValueId>>
       smallerTableElementToNewElement{};
-  std::ranges::for_each(
+  ql::ranges::for_each(
       smallerTableJoinColumnRef,
       [&randomDouble, &probabilityToCreateOverlap,
        &smallerTableElementToNewElement, &randomBiggerTableElement,
@@ -295,7 +295,7 @@ static size_t createOverlapRandomly(IdTableAndJoinColumn* const smallerTable,
   size_t newOverlapMatches{0};
   ad_utility::HashMap<ValueId, std::reference_wrapper<const ValueId>>
       smallerTableElementToNewElement{};
-  std::ranges::for_each(
+  ql::ranges::for_each(
       smallerTableJoinColumnSet.uniqueElements_,
       [&randomBiggerTableElement, &wantedNumNewOverlapMatches,
        &newOverlapMatches, &smallerTableElementToNewElement,
@@ -326,7 +326,7 @@ static size_t createOverlapRandomly(IdTableAndJoinColumn* const smallerTable,
       });
 
   // Overwrite the designated values in the smaller table.
-  std::ranges::for_each(
+  ql::ranges::for_each(
       smallerTableJoinColumnRef, [&smallerTableElementToNewElement](auto& id) {
         if (auto newValueIterator = smallerTableElementToNewElement.find(id);
             newValueIterator != smallerTableElementToNewElement.end()) {
@@ -465,17 +465,17 @@ static std::vector<T> mergeSortedVectors(
   std::vector<T> mergedVector{};
 
   // Merge.
-  std::ranges::for_each(intervals, [&mergedVector](std::vector<T> elements) {
+  ql::ranges::for_each(intervals, [&mergedVector](std::vector<T> elements) {
     if (mergedVector.empty() || elements.empty()) {
-      std::ranges::copy(elements, std::back_inserter(mergedVector));
+      ql::ranges::copy(elements, std::back_inserter(mergedVector));
       return;
     }
     const size_t idxOldLastElem = mergedVector.size() - 1;
-    std::ranges::copy(elements, std::back_inserter(mergedVector));
+    ql::ranges::copy(elements, std::back_inserter(mergedVector));
     if (mergedVector.at(idxOldLastElem) > mergedVector.at(idxOldLastElem + 1)) {
-      std::ranges::inplace_merge(
+      ql::ranges::inplace_merge(
           mergedVector,
-          std::ranges::next(mergedVector.begin(), idxOldLastElem + 1));
+          ql::ranges::next(mergedVector.begin(), idxOldLastElem + 1));
     }
   });
 
@@ -935,7 +935,7 @@ class GeneralInterfaceImplementation : public BenchmarkInterface {
             "' must be bigger than, or equal to, 0.")};
     config.addValidator(
         [](const benchmarkSampleSizeRatiosValueType& vec) {
-          return std::ranges::all_of(
+          return ql::ranges::all_of(
               vec,
               [](const benchmarkSampleSizeRatiosValueType::value_type ratio) {
                 return ratio >= 0.f;
@@ -961,7 +961,7 @@ class GeneralInterfaceImplementation : public BenchmarkInterface {
             ".")};
     config.addValidator(
         [](const benchmarkSampleSizeRatiosValueType& vec) {
-          return std::ranges::max(vec) <=
+          return ql::ranges::max(vec) <=
                  getMaxValue<benchmarkSampleSizeRatiosValueType::value_type>() -
                      1.f;
         },
@@ -1056,9 +1056,9 @@ class GeneralInterfaceImplementation : public BenchmarkInterface {
               },
               descriptor, descriptor, option);
         };
-    std::ranges::for_each(std::vector{minBiggerTableRows, maxBiggerTableRows,
-                                      minSmallerTableRows},
-                          addCastableValidator);
+    ql::ranges::for_each(std::vector{minBiggerTableRows, maxBiggerTableRows,
+                                     minSmallerTableRows},
+                         addCastableValidator);
   }
 
   /*
@@ -1684,7 +1684,7 @@ class BmOnlyBiggerTableSizeChanges final
               static_cast<double>(getConfigVariables().minBiggerTableRows_) /
               static_cast<double>(smallerTableNumRows)))};
           auto growthFunction = createDefaultGrowthLambda<float>(
-              10.f, std::ranges::max(minRatio, 10.f),
+              10.f, ql::ranges::max(minRatio, 10.f),
               generateNaturalNumberSequenceInterval(minRatio, 9.f));
           ResultTable& table = makeGrowingBenchmarkTable(
               &results, tableName, "Row ratio", alwaysFalse,
@@ -1742,8 +1742,8 @@ class BmOnlySmallerTableSizeChanges final
         for (const float ratioRows : mergeSortedVectors<float>(
                  {generateNaturalNumberSequenceInterval(
                       getConfigVariables().minRatioRows_,
-                      std::ranges::min(getConfigVariables().maxRatioRows_,
-                                       10.f)),
+                      ql::ranges::min(getConfigVariables().maxRatioRows_,
+                                      10.f)),
                   generateExponentInterval(
                       10.f, getConfigVariables().minRatioRows_,
                       getConfigVariables().maxRatioRows_)})) {
@@ -1755,7 +1755,7 @@ class BmOnlySmallerTableSizeChanges final
           // Returns the amount of rows in the smaller `IdTable`, used for the
           // measurements in a given row.
           auto growthFunction = createDefaultGrowthLambda(
-              10UL, std::ranges::max(
+              10UL, ql::ranges::max(
                         static_cast<size_t>(
                             static_cast<double>(
                                 getConfigVariables().minBiggerTableRows_) /
@@ -1867,7 +1867,7 @@ class BmSampleSizeRatio final : public GeneralInterfaceImplementation {
   BenchmarkResults runAllBenchmarks() override {
     BenchmarkResults results{};
     const auto& ratios{getConfigVariables().benchmarkSampleSizeRatios_};
-    const float maxSampleSizeRatio{std::ranges::max(ratios)};
+    const float maxSampleSizeRatio{ql::ranges::max(ratios)};
 
     /*
     We work with the biggest possible smaller and bigger table. That should make
@@ -2097,17 +2097,17 @@ class BmSmallerTableGrowsBiggerTableRemainsSameSize final
               static_cast<double>(biggerTableNumRows) /
               static_cast<double>(getConfigVariables().minSmallerTableRows_))};
           std::vector<size_t> smallerTableRows;
-          std::ranges::transform(
+          ql::ranges::transform(
               mergeSortedVectors<float>(
                   {generateNaturalNumberSequenceInterval(
-                       1.f, std::ranges::min(10.f, biggestRowRatio)),
+                       1.f, ql::ranges::min(10.f, biggestRowRatio)),
                    generateExponentInterval(10.f, 10.f, biggestRowRatio)}),
               std::back_inserter(smallerTableRows),
               [&biggerTableNumRows](const float ratio) {
                 return static_cast<size_t>(
                     static_cast<double>(biggerTableNumRows) / ratio);
               });
-          std::ranges::reverse(smallerTableRows);
+          ql::ranges::reverse(smallerTableRows);
           const size_t lastSmallerTableRow{smallerTableRows.back()};
           auto growthFunction = createDefaultGrowthLambda(
               10UL, lastSmallerTableRow + 1UL, std::move(smallerTableRows));

diff --git a/benchmark/ParallelMergeBenchmark.cpp b/benchmark/ParallelMergeBenchmark.cpp
@@ -28,12 +28,12 @@ class IdTableCompressedWriterBenchmark : public BenchmarkInterface {
                ad_utility::integerRange(numInputRows)) {
             res.push_back(gen());
           }
-          std::ranges::sort(res);
+          ql::ranges::sort(res);
           return res;
         };
     std::vector<std::vector<size_t>> inputs;
     inputs.resize(numInputs);
-    std::ranges::generate(inputs, generateRandomVec);
+    ql::ranges::generate(inputs, generateRandomVec);
 
     auto run = [&inputs]() {
       auto merger = ad_utility::parallelMultiwayMerge<size_t, false>(

diff --git a/benchmark/infrastructure/BenchmarkMain.cpp b/benchmark/infrastructure/BenchmarkMain.cpp
@@ -90,7 +90,7 @@ static void writeBenchmarkClassAndBenchmarkResultsToJsonFile(
 Print the configuration documentation of all registered benchmarks.
 */
 static __attribute__((noreturn)) void printConfigurationOptionsAndExit() {
-  std::ranges::for_each(
+  ql::ranges::for_each(
       BenchmarkRegister::getAllRegisteredBenchmarks(),
       [](const BenchmarkInterface* bench) {
         std::cerr << createCategoryTitle(
@@ -211,13 +211,13 @@ int main(int argc, char** argv) {
   // Actually processing the arguments.
   if (vm.count("print")) {
     // Print the results and metadata.
-    std::ranges::for_each(benchmarkClassAndResults,
-                          [](const auto& pair) {
-                            std::cout << benchmarkResultsToString(pair.first,
-                                                                  pair.second)
-                                      << "\n\n";
-                          },
-                          {});
+    ql::ranges::for_each(benchmarkClassAndResults,
+                         [](const auto& pair) {
+                           std::cout << benchmarkResultsToString(pair.first,
+                                                                 pair.second)
+                                     << "\n\n";
+                         },
+                         {});
   }
 
   if (vm.count("write")) {

diff --git a/benchmark/infrastructure/BenchmarkMeasurementContainer.cpp b/benchmark/infrastructure/BenchmarkMeasurementContainer.cpp
@@ -129,7 +129,7 @@ void ResultTable::init(const std::string& descriptor,
   descriptorForLog_ = std::move(descriptorForLog);
   columnNames_ = columnNames;
   entries_.resize(rowNames.size());
-  std::ranges::fill(entries_, std::vector<EntryType>(columnNames.size()));
+  ql::ranges::fill(entries_, std::vector<EntryType>(columnNames.size()));
 
   // Setting the row names.
   for (size_t row = 0; row < rowNames.size(); row++) {
@@ -287,10 +287,10 @@ ResultTable::operator std::string() const {
         });
 
     // Which of the entries is the longest?
-    columnMaxStringWidth.at(column) = std::ranges::max(stringWidthOfRow);
+    columnMaxStringWidth.at(column) = ql::ranges::max(stringWidthOfRow);
 
     // Is the name of the column bigger?
-    columnMaxStringWidth.at(column) = std::ranges::max(
+    columnMaxStringWidth.at(column) = ql::ranges::max(
         columnMaxStringWidth.at(column), columnNames_.at(column).length());
   }
 
@@ -384,7 +384,7 @@ void ResultGroup::deleteEntryImpl(T& entry) {
   }();
 
   // Delete `entry`.
-  auto entryIterator{std::ranges::find(
+  auto entryIterator{ql::ranges::find(
       vec, &entry, [](const ad_utility::CopyableUniquePtr<T>& pointer) {
         return pointer.get();
       })};

diff --git a/benchmark/infrastructure/BenchmarkToJson.cpp b/benchmark/infrastructure/BenchmarkToJson.cpp
@@ -46,8 +46,8 @@ static nlohmann::json transformIntoJsonArray(
   */
   nlohmann::ordered_json jsonArray = nlohmann::ordered_json::array();
 
-  std::ranges::transform(vec, std::back_inserter(jsonArray),
-                         translationFunction);
+  ql::ranges::transform(vec, std::back_inserter(jsonArray),
+                        translationFunction);
 
   return jsonArray;
 }

diff --git a/benchmark/util/ResultTableColumnOperations.h b/benchmark/util/ResultTableColumnOperations.h
@@ -37,8 +37,8 @@ requires(sizeof...(ColumnInputTypes) > 0) void generateColumnWithColumnInput(
   // Using a column more than once is the sign of an error.
   std::array<size_t, sizeof...(ColumnInputTypes)> allColumnNums{
       {inputColumns.columnNum_...}};
-  std::ranges::sort(allColumnNums);
-  AD_CONTRACT_CHECK(std::ranges::adjacent_find(allColumnNums) ==
+  ql::ranges::sort(allColumnNums);
+  AD_CONTRACT_CHECK(ql::ranges::adjacent_find(allColumnNums) ==
                     allColumnNums.end());
 
   // Fill the result column.

diff --git a/src/backports/algorithm.h b/src/backports/algorithm.h
@@ -12,9 +12,9 @@
 #include "backports/concepts.h"
 
 // The following defines namespaces `ql::ranges` and `ql::views` that are almost
-// drop-in replacements for `std::ranges` and `std::views`. In C++20 mode (when
+// drop-in replacements for `ql::ranges` and `std::views`. In C++20 mode (when
 // the `QLEVER_CPP_17` macro is not used), these namespaces are simply aliases
-// for `std::ranges` and `std::views`. In C++17 mode they contain the ranges and
+// for `ql::ranges` and `std::views`. In C++17 mode they contain the ranges and
 // views from Erice Niebler's `range-v3` library. NOTE: `ql::ranges::unique`
 // currently doesn't work, because the interface to this function is different
 // in both implementations. NOTE: There might be other caveats which we are

diff --git a/src/engine/AddCombinedRowToTable.h b/src/engine/AddCombinedRowToTable.h
@@ -356,7 +356,7 @@ class AddCombinedRowToIdTable {
       // Only merge non-null vocabs.
       auto range = currentVocabs_ | ql::views::filter(toBool) |
                    ql::views::transform(dereference);
-      mergedVocab_.mergeWith(std::ranges::ref_view{range});
+      mergedVocab_.mergeWith(ql::ranges::ref_view{range});
     }
   }
   const IdTableView<0>& inputLeft() const {

diff --git a/src/engine/CartesianProductJoin.cpp b/src/engine/CartesianProductJoin.cpp
@@ -182,9 +182,9 @@ VariableToColumnMap CartesianProductJoin::computeVariableToColumnMap() const {
 }
 
 // _____________________________________________________________________________
-IdTable CartesianProductJoin::writeAllColumns(
-    std::ranges::random_access_range auto idTables, size_t offset, size_t limit,
-    size_t lastTableOffset) const {
+CPP_template_def(typename R)(requires ql::ranges::random_access_range<R>)
+    IdTable CartesianProductJoin::writeAllColumns(
+        R idTables, size_t offset, size_t limit, size_t lastTableOffset) const {
   AD_CORRECTNESS_CHECK(offset >= lastTableOffset);
   IdTable result{getResultWidth(), getExecutionContext()->getAllocator()};
   // TODO<joka921> Find a solution to cheaply handle the case, that only a
@@ -303,12 +303,14 @@ CartesianProductJoin::calculateSubResults(bool requestLaziness) {
 }
 
 // _____________________________________________________________________________
-Result::Generator CartesianProductJoin::produceTablesLazily(
-    LocalVocab mergedVocab, std::ranges::range auto idTables, size_t offset,
-    size_t limit, size_t lastTableOffset) const {
+CPP_template_def(typename R)(requires ql::ranges::range<R>) Result::Generator
+    CartesianProductJoin::produceTablesLazily(LocalVocab mergedVocab,
+                                              R idTables, size_t offset,
+                                              size_t limit,
+                                              size_t lastTableOffset) const {
   while (limit > 0) {
     uint64_t limitWithChunkSize = std::min(limit, chunkSize_);
-    IdTable idTable = writeAllColumns(std::ranges::ref_view(idTables), offset,
+    IdTable idTable = writeAllColumns(ql::ranges::ref_view(idTables), offset,
                                       limitWithChunkSize, lastTableOffset);
     size_t tableSize = idTable.size();
     AD_CORRECTNESS_CHECK(tableSize <= limit);

diff --git a/src/engine/CartesianProductJoin.h b/src/engine/CartesianProductJoin.h
@@ -94,9 +94,9 @@ class CartesianProductJoin : public Operation {
   // rows to write at most. `lastTableOffset` is the offset of the last table,
   // to account for cases where the last table does not cover the whole result
   // and so index 0 of a table does not correspond to row 0 of the result.
-  IdTable writeAllColumns(std::ranges::random_access_range auto idTables,
-                          size_t offset, size_t limit,
-                          size_t lastTableOffset = 0) const;
+  CPP_template(typename R)(requires ql::ranges::random_access_range<R>) IdTable
+      writeAllColumns(R idTables, size_t offset, size_t limit,
+                      size_t lastTableOffset = 0) const;
 
   // Calculate the subresults of the children and store them into a vector. If
   // the rightmost child can produce a lazy result, it will be stored outside of
@@ -114,10 +114,9 @@ class CartesianProductJoin : public Operation {
   // `lastTableOffset` is the offset of the last table in the range. This is
   // used to handle `IdTable`s yielded by generators where the range of indices
   // they represent do not cover the whole result.
-  Result::Generator produceTablesLazily(LocalVocab mergedVocab,
-                                        std::ranges::range auto idTables,
-                                        size_t offset, size_t limit,
-                                        size_t lastTableOffset = 0) const;
+  CPP_template(typename R)(requires ql::ranges::range<R>) Result::Generator
+      produceTablesLazily(LocalVocab mergedVocab, R idTables, size_t offset,
+                          size_t limit, size_t lastTableOffset = 0) const;
 
   // Similar to `produceTablesLazily` but can handle a single lazy result.
   Result::Generator createLazyConsumer(

diff --git a/src/engine/Describe.cpp b/src/engine/Describe.cpp
@@ -53,10 +53,9 @@ string Describe::getCacheKeyImpl() const {
   const auto& defaultGraphs = describe_.datasetClauses_.defaultGraphs_;
   if (defaultGraphs.has_value()) {
     std::vector<std::string> graphIdVec;
-    std::ranges::transform(defaultGraphs.value(),
-                           std::back_inserter(graphIdVec),
-                           &TripleComponent::toRdfLiteral);
-    std::ranges::sort(graphIdVec);
+    ql::ranges::transform(defaultGraphs.value(), std::back_inserter(graphIdVec),
+                          &TripleComponent::toRdfLiteral);
+    ql::ranges::sort(graphIdVec);
     absl::StrAppend(&result,
                     "\nFiltered by Graphs:", absl::StrJoin(graphIdVec, " "));
   }
@@ -218,7 +217,7 @@ IdTable Describe::getIdsToDescribe(const Result& result,
   // Copy the `Id`s from the hash set to an `IdTable`.
   IdTable idsAsTable{1, allocator()};
   idsAsTable.resize(idsToDescribe.size());
-  std::ranges::copy(idsToDescribe, idsAsTable.getColumn(0).begin());
+  ql::ranges::copy(idsToDescribe, idsAsTable.getColumn(0).begin());
   return idsAsTable;
 }
 

diff --git a/src/engine/Distinct.cpp b/src/engine/Distinct.cpp
@@ -97,7 +97,7 @@ IdTable Distinct::distinct(
   LOG(DEBUG) << "Distinct on " << dynInput.size() << " elements.\n";
   IdTableStatic<WIDTH> result = std::move(dynInput).toStatic<WIDTH>();
 
-  // Variant of `std::ranges::unique` that allows to skip the begin rows of
+  // Variant of `ql::ranges::unique` that allows to skip the begin rows of
   // elements found in the previous table.
   auto begin =
       ql::ranges::find_if(result, [this, &previousRow](const auto& row) {