Refactor

Author: mskapilks

velox/dwio/common/DirectDecoder.h

@@ -106,26 +106,14 @@ class DirectDecoder : public IntDecoder<isSigned> {
       } else if constexpr (std::is_same_v<
                                typename Visitor::DataType,
                                int128_t>) {
-        if (super::numBytes == 12) {
-          static constexpr int64_t kJulianToUnixEpochDays = 2440588LL;
-          static constexpr int64_t kSecondsPerDay = 86400LL;
-          static constexpr int64_t kNanosPerSecond =
-              Timestamp::kNanosecondsInMillisecond *
-              Timestamp::kMillisecondsInSecond;
+        if (super::numBytes == 12 /* INT96 */) {
+          int128_t encoded = super::template readInt<int128_t>();
+          int32_t days = encoded & ((1ULL << 32) - 1);
+          uint64_t nanos = static_cast<uint64_t>(encoded >> 32);
 
-          int128_t dat = super::template readInt<int128_t>();
-          int32_t days = dat & ((1ULL << 32) - 1);
-          dat = dat >> 32;
-          uint64_t nanos = static_cast<uint64_t>(dat);
-
-          int64_t seconds = (days - kJulianToUnixEpochDays) * kSecondsPerDay;
-          if (nanos > Timestamp::kMaxNanos) {
-            seconds += nanos / kNanosPerSecond;
-            nanos -= (nanos / kNanosPerSecond) * kNanosPerSecond;
-          }
-          Timestamp t(seconds, nanos);
-          auto val = *reinterpret_cast<int128_t*>(&t);
-          toSkip = visitor.process(val, atEnd);
+          auto timestamp = Timestamp::fromDaysAndNanos(days, nanos);
+          toSkip =
+              visitor.process(*reinterpret_cast<int128_t*>(&timestamp), atEnd);
         } else {
           toSkip = visitor.process(super::template readInt<int128_t>(), atEnd);
         }

velox/dwio/parquet/reader/PageReader.cpp

@@ -414,11 +414,7 @@ void PageReader::prepareDictionary(const PageHeader& pageHeader) {
       // We start from the end to allow in-place expansion.
       auto values = dictionary_.values->asMutable<Timestamp>();
       auto parquetValues = dictionary_.values->asMutable<char>();
-      static constexpr int64_t kJulianToUnixEpochDays = 2440588LL;
-      static constexpr int64_t kSecondsPerDay = 86400LL;
-      static constexpr int64_t kNanosPerSecond =
-          Timestamp::kNanosecondsInMillisecond *
-          Timestamp::kMillisecondsInSecond;
+
       for (auto i = dictionary_.numValues - 1; i >= 0; --i) {
         // Convert the timestamp into seconds and nanos since the Unix epoch,
         // 00:00:00.000000 on 1 January 1970.
@@ -432,12 +428,8 @@ void PageReader::prepareDictionary(const PageHeader& pageHeader) {
             &days,
             parquetValues + i * sizeof(Int96Timestamp) + sizeof(uint64_t),
             sizeof(int32_t));
-        int64_t seconds = (days - kJulianToUnixEpochDays) * kSecondsPerDay;
-        if (nanos > Timestamp::kMaxNanos) {
-          seconds += nanos / kNanosPerSecond;
-          nanos -= (nanos / kNanosPerSecond) * kNanosPerSecond;
-        }
-        values[i] = Timestamp(seconds, nanos);
+
+        values[i] = Timestamp::fromDaysAndNanos(days, nanos);
       }
       break;
     }

velox/type/Timestamp.h

@@ -86,6 +86,21 @@ struct Timestamp {
     VELOX_USER_DCHECK_LE(nanos, kMaxNanos, "Timestamp nanos out of range");
   }
 
+  static Timestamp fromDaysAndNanos(int32_t days, uint64_t nanos) {
+    static constexpr int64_t kJulianToUnixEpochDays = 2440588LL;
+    static constexpr int64_t kSecondsPerDay = 86400LL;
+    static constexpr int64_t kNanosPerSecond =
+        Timestamp::kNanosecondsInMillisecond * Timestamp::kMillisecondsInSecond;
+
+    int64_t seconds = (days - kJulianToUnixEpochDays) * kSecondsPerDay;
+    if (nanos > Timestamp::kMaxNanos) {
+      seconds += nanos / kNanosPerSecond;
+      nanos -= (nanos / kNanosPerSecond) * kNanosPerSecond;
+    }
+
+    return Timestamp(seconds, nanos);
+  }
+
   // Returns the current unix timestamp (ms precision).
   static Timestamp now();