refactor : Change C-style casts to C++-style (Part 4)

As per the security guideline in
https://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#es49-if-you-must-use-a-cast-use-a-named-cast

Covers the findings in velox/functions

velox/functions/CoverageUtil.cpp

@@ -204,7 +204,8 @@ void printCoverageMap(
   // Split scalar functions into 'numScalarColumns' columns. Put all aggregate
   // functions into one column.
   auto numRows = std::max(
-      {(size_t)std::ceil((double)scalarCnt / numScalarColumns),
+      {static_cast<size_t>(
+           std::ceil(static_cast<double>(scalarCnt) / numScalarColumns)),
        aggCnt,
        windowCnt});
 
@@ -426,8 +427,8 @@ void printVeloxFunctions(
   auto scalarCnt = scalarNames.size();
   auto aggCnt = aggNames.size();
   auto windowCnt = windowNames.size();
-  auto numRows =
-      std::max({(size_t)std::ceil(scalarCnt / 3.0), aggCnt, windowCnt});
+  auto numRows = std::max(
+      {static_cast<size_t>(std::ceil(scalarCnt / 3.0)), aggCnt, windowCnt});
 
   auto printName = [&](const std::string& name) {
     return linkBlockList.count(name) == 0 ? toFuncLink(name, domain) : name;

velox/functions/prestosql/Sequence.cpp

@@ -48,21 +48,24 @@ T add(T value, K step, int32_t sequence);
 
 template <>
 int64_t add(int64_t value, int64_t step, int32_t sequence) {
-  const auto delta = (int128_t)step * (int128_t)sequence;
+  const auto delta =
+      static_cast<int128_t>(step) * static_cast<int128_t>(sequence);
   // Since step is calcuated from start and stop,
   // the sum of 'value' and 'add' is within int64_t.
   return value + delta;
 }
 
 template <>
 int32_t add(int32_t value, int64_t step, int32_t sequence) {
-  const auto delta = (int128_t)step * (int128_t)sequence;
+  const auto delta =
+      static_cast<int128_t>(step) * static_cast<int128_t>(sequence);
   return value + delta;
 }
 
 template <>
 Timestamp add(Timestamp value, int64_t step, int32_t sequence) {
-  const auto delta = (int128_t)step * (int128_t)sequence;
+  const auto delta =
+      static_cast<int128_t>(step) * static_cast<int128_t>(sequence);
   return Timestamp::fromMillis(value.toMillis() + delta);
 }
 
@@ -181,9 +184,8 @@ class SequenceFunction : public exec::VectorFunction {
     if (isYearMonth) {
       sequenceCount = getStepCount(start, stop, step);
     } else {
-      sequenceCount =
-          ((int128_t)toInt64(stop) - (int128_t)toInt64(start)) / step +
-          1; // prevent overflow
+      sequenceCount = static_cast<int128_t>(toInt64(stop)) -
+          static_cast<int128_t>(toInt64(start)) / step + 1; // prevent overflow
     }
     VELOX_USER_CHECK_LE(
         sequenceCount,

velox/functions/prestosql/WidthBucketArray.cpp

@@ -124,7 +124,7 @@ class WidthBucketArrayFunctionConstantBins : public exec::VectorFunction {
     VELOX_USER_CHECK(!std::isnan(operand), "Operand cannot be NaN");
 
     int lower = 0;
-    int upper = (int)bins.size();
+    int upper = static_cast<int>(bins.size());
     while (lower < upper) {
       int index = (lower + upper) / 2;
       auto bin = bins.at(index);

velox/functions/prestosql/aggregates/ApproxMostFrequentAggregate.cpp

@@ -159,9 +159,17 @@ struct ApproxMostFrequentAggregate : exec::Aggregate {
     auto rowVec = (*result)->as<RowVector>();
     VELOX_CHECK(rowVec);
     rowVec->childAt(0) = std::make_shared<ConstantVector<int64_t>>(
-        rowVec->pool(), numGroups, false, BIGINT(), int64_t(buckets_));
+        rowVec->pool(),
+        numGroups,
+        false,
+        BIGINT(),
+        static_cast<int64_t>(buckets_));
     rowVec->childAt(1) = std::make_shared<ConstantVector<int64_t>>(
-        rowVec->pool(), numGroups, false, BIGINT(), int64_t(capacity_));
+        rowVec->pool(),
+        numGroups,
+        false,
+        BIGINT(),
+        static_cast<int64_t>(capacity_));
     auto values = rowVec->childAt(2)->as<ArrayVector>();
     auto counts = rowVec->childAt(3)->as<ArrayVector>();
     rowVec->resize(numGroups);

velox/functions/prestosql/aggregates/ApproxPercentileAggregate.cpp

@@ -306,13 +306,17 @@ class ApproxPercentileAggregate : public exec::Aggregate {
         BaseVector::wrapInConstant(numGroups, 0, std::move(array));
     rowResult->childAt(kPercentilesIsArray) =
         std::make_shared<ConstantVector<bool>>(
-            pool, numGroups, false, BOOLEAN(), bool(percentiles_->isArray));
+            pool,
+            numGroups,
+            false,
+            BOOLEAN(),
+            static_cast<bool>(percentiles_->isArray));
     rowResult->childAt(kAccuracy) = std::make_shared<ConstantVector<double>>(
         pool,
         numGroups,
         accuracy_ == kMissingNormalizedValue,
         DOUBLE(),
-        double(accuracy_));
+        static_cast<double>(accuracy_));
     auto k = rowResult->childAt(kK)->asFlatVector<int32_t>();
     auto n = rowResult->childAt(kN)->asFlatVector<int64_t>();
     auto minValue = rowResult->childAt(kMinValue)->asFlatVector<T>();

velox/functions/prestosql/aggregates/CountAggregate.cpp

@@ -137,7 +137,7 @@ class CountAggregate : public SimpleNumericAggregate<bool, int64_t, int64_t> {
       folly::Range<const vector_size_t*> indices) override {
     for (auto i : indices) {
       // result of count is never null
-      *value<int64_t>(groups[i]) = (int64_t)0;
+      *value<int64_t>(groups[i]) = static_cast<int64_t>(0);
     }
   }
 

velox/functions/prestosql/aggregates/CovarianceAggregates.cpp

@@ -94,9 +94,10 @@ struct CovarAccumulator {
     double deltaMeanX = meanXOther - meanX();
     double deltaMeanY = meanYOther - meanY();
     c2_ += c2Other +
-        deltaMeanX * deltaMeanY * count() * countOther / (double)newCount;
-    meanX_ += deltaMeanX * countOther / (double)newCount;
-    meanY_ += deltaMeanY * countOther / (double)newCount;
+        deltaMeanX * deltaMeanY * count() * countOther /
+            static_cast<double>(newCount);
+    meanX_ += deltaMeanX * countOther / static_cast<double>(newCount);
+    meanY_ += deltaMeanY * countOther / static_cast<double>(newCount);
     count_ = newCount;
   }
 
@@ -674,7 +675,7 @@ class CovarianceAggregate : public exec::Aggregate {
         auto* accumulator = this->accumulator(group);
         if (TResultAccessor::hasResult(*accumulator)) {
           clearNull(rawNulls, i);
-          rawValues[i] = (T)TResultAccessor::result(*accumulator);
+          rawValues[i] = static_cast<T>(TResultAccessor::result(*accumulator));
         } else {
           vector->setNull(i, true);
         }

velox/functions/prestosql/aggregates/EntropyAggregates.cpp

@@ -48,8 +48,8 @@ struct EntropyAccumulator {
     if (count == 0) {
       return;
     }
-    sumC_ += (double)count;
-    sumCLogC_ += (double)count * std::log(count);
+    sumC_ += static_cast<double>(count);
+    sumCLogC_ += static_cast<double>(count) * std::log(count);
   }
 
   void merge(double sumCOther, double sumCLogCOther) {
@@ -125,7 +125,8 @@ class EntropyAggregate : public exec::Aggregate {
         // The "sum" is the constant value times the number of rows.
         // Use double to prevent overflows (this is the same as what is done in
         // updateNonNullValue).
-        const auto sum = (double)numRows * (double)value;
+        const auto sum =
+            static_cast<double>(numRows) * static_cast<double>(value);
         EntropyAccumulator accData(sum, sum * std::log(value));
         updateNonNullValue(group, accData);
       }

velox/functions/prestosql/aggregates/MaxSizeForStatsAggregate.cpp

@@ -95,7 +95,7 @@ class MaxSizeForStatsAggregate
           result = value;
         },
         mayPushdown,
-        (int64_t)0);
+        static_cast<int64_t>(0));
   }
 
   void addSingleGroupRawInput(

velox/functions/prestosql/aggregates/SumDataSizeForStatsAggregate.cpp

@@ -85,7 +85,7 @@ class SumDataSizeForStatsAggregate
           result = checkedPlus(result, value);
         },
         mayPushdown,
-        (int64_t)0);
+        static_cast<int64_t>(0));
   }
 
   void addSingleGroupRawInput(

velox/functions/prestosql/aggregates/VarianceAggregates.cpp

@@ -74,7 +74,8 @@ struct VarianceAccumulator {
     int64_t newCount = countOther + count();
     double delta = meanOther - mean();
     double newMean = mean() + delta / newCount * countOther;
-    m2_ += m2Other + delta * delta * countOther * count() / (double)newCount;
+    m2_ += m2Other +
+        delta * delta * countOther * count() / static_cast<double>(newCount);
     count_ = newCount;
     mean_ = newMean;
   }
@@ -218,7 +219,7 @@ class VarianceAggregate : public exec::Aggregate {
       if (!decodedRaw_.isNullAt(0)) {
         const T value = decodedRaw_.valueAt<T>(0);
         const auto numRows = rows.countSelected();
-        VarianceAccumulator accData(numRows, (double)value);
+        VarianceAccumulator accData(numRows, static_cast<double>(value));
         updateNonNullValue(group, accData);
       }
     } else if (decodedRaw_.mayHaveNulls()) {
@@ -384,7 +385,7 @@ class VarianceAggregate : public exec::Aggregate {
       exec::Aggregate::clearNull(group);
     }
     VarianceAccumulator* thisAccData = accumulator(group);
-    thisAccData->update((double)value);
+    thisAccData->update(static_cast<double>(value));
   }
 
   template <bool tableHasNulls = true>

velox/functions/prestosql/benchmarks/ArrayWriterBenchmark.cpp

@@ -154,7 +154,7 @@ class ArrayWriterBenchmark : public functions::test::FunctionBenchmarkBase {
   }
 
   vector_size_t size = 1'000;
-  size_t totalItemsCount = (size) * (size + 1) / 2;
+  auto totalItemsCount = static_cast<size_t>((size) * (size + 1) / 2);
 
   auto makeInput() {
     std::vector<int64_t> inputData(size, 0);

velox/functions/prestosql/window/CumeDist.cpp

@@ -51,7 +51,7 @@ class CumeDistFunction : public exec::WindowFunction {
       if (peerStart != currentPeerGroupStart_) {
         currentPeerGroupStart_ = peerStart;
         runningTotal_ += peerGroupEndsVector[i] - peerStart + 1;
-        cumeDist_ = double(runningTotal_) / numPartitionRows_;
+        cumeDist_ = static_cast<double>(runningTotal_) / numPartitionRows_;
       }
       rawValues[resultOffset + i] = cumeDist_;
     }

velox/functions/sparksql/DecimalArithmetic.cpp

@@ -165,9 +165,9 @@ struct DecimalAddSubtractBase {
     // Adjust fractional parts to higher scale.
     const auto higherScale = std::max(aScale, bScale);
     const auto aFractionScaled =
-        increaseScale((int128_t)aFraction, higherScale - aScale);
+        increaseScale(static_cast<int128_t>(aFraction), higherScale - aScale);
     const auto bFractionScaled =
-        increaseScale((int128_t)bFraction, higherScale - bScale);
+        increaseScale(static_cast<int128_t>(bFraction), higherScale - bScale);
 
     int128_t fraction;
     bool carryToLeft = false;
@@ -205,12 +205,13 @@ struct DecimalAddSubtractBase {
     // Adjust fractional parts to higher scale.
     const auto higherScale = std::max(aScale, bScale);
     const auto aFractionScaled =
-        increaseScale((int128_t)aFraction, higherScale - aScale);
+        increaseScale(static_cast<int128_t>(aFraction), higherScale - aScale);
     const auto bFractionScaled =
-        increaseScale((int128_t)bFraction, higherScale - bScale);
+        increaseScale(static_cast<int128_t>(bFraction), higherScale - bScale);
 
     // No need to consider overflow because two inputs are opposite.
-    int128_t whole = (int128_t)aWhole + (int128_t)bWhole;
+    int128_t whole =
+        static_cast<int128_t>(aWhole) + static_cast<int128_t>(bWhole);
     int128_t fraction = aFractionScaled + bFractionScaled;
 
     // If the whole and fractional parts have different signs, adjust them to

velox/functions/sparksql/MakeTimestamp.cpp

@@ -61,7 +61,8 @@ std::optional<Timestamp> makeTimeStampFromDecodedArgs(
 
   // Micros has at most 8 digits (2 for seconds + 6 for microseconds),
   // thus it's safe to cast micros from int64_t to int32_t.
-  auto localMicros = util::fromTime(hour, minute, 0, (int32_t)micros);
+  auto localMicros =
+      util::fromTime(hour, minute, 0, static_cast<int32_t>(micros));
   return util::fromDatetime(daysSinceEpoch.value(), localMicros);
 }
 

velox/functions/sparksql/aggregates/AverageAggregate.cpp

@@ -68,7 +68,7 @@ class AverageAggregate
         vector->setNull(i, true);
       } else {
         this->clearNull(rawNulls, i);
-        rawValues[i] = (TResult)sumCount->sum / sumCount->count;
+        rawValues[i] = static_cast<TResult>(sumCount->sum) / sumCount->count;
       }
     }
   }

velox/functions/sparksql/specialforms/DecimalRound.cpp

@@ -31,8 +31,12 @@ class DecimalRoundFunction : public exec::VectorFunction {
       uint8_t resultScale)
       : scale_(
             scale >= 0
-                ? std::min(scale, (int32_t)LongDecimalType::kMaxPrecision)
-                : std::max(scale, -(int32_t)LongDecimalType::kMaxPrecision)),
+                ? std::min(
+                      scale,
+                      static_cast<int32_t>(LongDecimalType::kMaxPrecision))
+                : std::max(
+                      scale,
+                      -static_cast<int32_t>(LongDecimalType::kMaxPrecision))),
         inputPrecision_(inputPrecision),
         inputScale_(inputScale),
         resultPrecision_(resultPrecision),
@@ -54,7 +58,7 @@ class DecimalRoundFunction : public exec::VectorFunction {
       VELOX_USER_CHECK_GT(
           rescale, 0, "A non-negative rescale value is expected.");
       return DecimalUtil::kPowersOfTen[std::min(
-          rescale, (int32_t)LongDecimalType::kMaxPrecision)];
+          rescale, static_cast<int32_t>(LongDecimalType::kMaxPrecision))];
     };
     if (scale_ < 0) {
       divideFactor_ = rescaleFactor(inputScale_ - scale_);
@@ -179,18 +183,23 @@ DecimalRoundCallToSpecialForm::getResultPrecisionScale(
     // rounding, and the result scale is 0.
     const auto newPrecision = std::max(
         integralLeastNumDigits,
-        -std::max(roundScale, -(int32_t)LongDecimalType::kMaxPrecision) + 1);
+        -std::max(
+            roundScale, -static_cast<int32_t>(LongDecimalType::kMaxPrecision)) +
+            1);
     // We have to accept the risk of overflow as we can't exceed the max
     // precision.
-    return {std::min(newPrecision, (int32_t)LongDecimalType::kMaxPrecision), 0};
+    return {
+        std::min(
+            newPrecision, static_cast<int32_t>(LongDecimalType::kMaxPrecision)),
+        0};
   }
-  const uint8_t newScale = std::min((int32_t)scale, roundScale);
+  const uint8_t newScale = std::min(static_cast<int32_t>(scale), roundScale);
   // We have to accept the risk of overflow as we cannot exceed the max
   // precision.
   return {
       std::min(
           integralLeastNumDigits + newScale,
-          (int32_t)LongDecimalType::kMaxPrecision),
+          static_cast<int32_t>(LongDecimalType::kMaxPrecision)),
       newScale};
 }
 

velox/functions/sparksql/specialforms/MakeDecimal.cpp

@@ -74,8 +74,9 @@ class MakeDecimalFunction final : public exec::VectorFunction {
             [&](vector_size_t row) { rawResults[row] = constant; });
       }
     } else {
-      rows.applyToSelected(
-          [&](vector_size_t row) { rawResults[row] = (int128_t)constant; });
+      rows.applyToSelected([&](vector_size_t row) {
+        rawResults[row] = static_cast<int128_t>(constant);
+      });
     }
   }
 
@@ -105,7 +106,7 @@ class MakeDecimalFunction final : public exec::VectorFunction {
       });
     } else {
       rows.applyToSelected([&](vector_size_t row) {
-        rawResults[row] = (int128_t)rawValues[row];
+        rawResults[row] = static_cast<int128_t>(rawValues[row]);
       });
     }
   }