xx

Author: zuochunwei

velox/dwio/common/SelectiveColumnReader.h

@@ -631,6 +631,10 @@ namespace facebook::velox::dwio::common {
 // Template parameter to indicate no hook in fast scan path. This is
 // referenced in decoders, thus needs to be declared in a header.
 struct NoHook : public ValueHook {
+  std::string toString() const override {
+    return "NoHook";
+  }
+
   void addValue(vector_size_t /*row*/, const void* FOLLY_NULLABLE /*value*/)
       override {}
 };

velox/dwio/dwrf/reader/SelectiveLongDecimalColumnReader.h

@@ -115,7 +115,29 @@ class SelectiveLongDecimalColumnReader
  private:
   template <bool dense>
   void processValueHook(RowSet rows, ValueHook* hook) {
-    VELOX_FAIL("TODO: orc long decimal process ValueHook");
+    // std::cout << "long decimal hook: %s" << hook->toString() << std::endl;
+
+    switch (hook->kind()) {
+      case aggregate::AggregationHook::kLongDecimalMax:
+        readHelper<dense, velox::common::AlwaysTrue>(
+            &dwio::common::alwaysTrue(),
+            rows,
+            dwio::common::ExtractToHook<
+                aggregate::MinMaxHook<UnscaledLongDecimal, false>>(hook));
+        break;
+      case aggregate::AggregationHook::kLongDecimalMin:
+        readHelper<dense, velox::common::AlwaysTrue>(
+            &dwio::common::alwaysTrue(),
+            rows,
+            dwio::common::ExtractToHook<
+                aggregate::MinMaxHook<UnscaledLongDecimal, true>>(hook));
+        break;
+      default:
+        readHelper<dense, velox::common::AlwaysTrue>(
+            &dwio::common::alwaysTrue(),
+            rows,
+            dwio::common::ExtractToGenericHook(hook));
+    }
   }
 
   template <bool dense, typename ExtractValues>

velox/dwio/dwrf/reader/SelectiveShortDecimalColumnReader.h

@@ -121,7 +121,29 @@ class SelectiveShortDecimalColumnReader
  private:
   template <bool dense>
   void processValueHook(RowSet rows, ValueHook* hook) {
-    VELOX_FAIL("TODO: orc short decimal process ValueHook");
+    // std::cout << "short decimal hook: %s" << hook->toString() << std::endl;
+
+    switch (hook->kind()) {
+      case aggregate::AggregationHook::kShortDecimalMax:
+        readHelper<dense, velox::common::AlwaysTrue>(
+            &dwio::common::alwaysTrue(),
+            rows,
+            dwio::common::ExtractToHook<
+                aggregate::MinMaxHook<UnscaledShortDecimal, false>>(hook));
+        break;
+      case aggregate::AggregationHook::kShortDecimalMin:
+        readHelper<dense, velox::common::AlwaysTrue>(
+            &dwio::common::alwaysTrue(),
+            rows,
+            dwio::common::ExtractToHook<
+                aggregate::MinMaxHook<UnscaledShortDecimal, true>>(hook));
+        break;
+      default:
+        readHelper<dense, velox::common::AlwaysTrue>(
+            &dwio::common::alwaysTrue(),
+            rows,
+            dwio::common::ExtractToGenericHook(hook));
+    }
   }
 
   template <bool dense, typename ExtractValues>

velox/exec/AggregationHook.h

@@ -35,6 +35,10 @@ class AggregationHook : public ValueHook {
   static constexpr Kind kFloatMin = 8;
   static constexpr Kind kDoubleMax = 9;
   static constexpr Kind kDoubleMin = 10;
+  static constexpr Kind kShortDecimalMax = 11;
+  static constexpr Kind kShortDecimalMin = 12;
+  static constexpr Kind kLongDecimalMax = 13;
+  static constexpr Kind kLongDecimalMin = 14;
 
   // Make null behavior known at compile time. This is useful when
   // templating a column decoding loop with a hook.
@@ -53,6 +57,12 @@ class AggregationHook : public ValueHook {
         groups_(groups),
         numNulls_(numNulls) {}
 
+  std::string toString() const override {
+    char buf[256];
+    sprintf(buf, "AggregationHook kind:%d", (int)kind());
+    return buf;
+  }
+
   bool acceptsNulls() const override final {
     return false;
   }
@@ -119,6 +129,17 @@ class SumHook final : public AggregationHook {
       uint64_t* numNulls)
       : AggregationHook(offset, nullByte, nullMask, groups, numNulls) {}
 
+  std::string toString() const override {
+    char buf[256];
+    sprintf(
+        buf,
+        "SumHook kind:%d TValue:%s TAggregate:%s",
+        (int)kind(),
+        typeid(TValue).name(),
+        typeid(TAggregate).name());
+    return buf;
+  }
+
   Kind kind() const override {
     if (std::is_same_v<TAggregate, double>) {
       if (std::is_same_v<TValue, double>) {
@@ -160,6 +181,18 @@ class SimpleCallableHook final : public AggregationHook {
       : AggregationHook(offset, nullByte, nullMask, groups, numNulls),
         updateSingleValue_(updateSingleValue) {}
 
+  std::string toString() const override {
+    char buf[256];
+    sprintf(
+        buf,
+        "SimpleCallableHook kind:%d TValue:%s TAggregate:%s UpdateSingleValue:%s",
+        (int)kind(),
+        typeid(TValue).name(),
+        typeid(TAggregate).name(),
+        typeid(UpdateSingleValue).name());
+    return buf;
+  }
+
   Kind kind() const override {
     return kGeneric;
   }
@@ -187,6 +220,17 @@ class MinMaxHook final : public AggregationHook {
       uint64_t* numNulls)
       : AggregationHook(offset, nullByte, nullMask, groups, numNulls) {}
 
+  std::string toString() const override {
+    char buf[256];
+    sprintf(
+        buf,
+        "MinMaxHook kind:%d T:%s isMin:%d",
+        (int)kind(),
+        typeid(T).name(),
+        (int)isMin);
+    return buf;
+  }
+
   Kind kind() const override {
     if (isMin) {
       if (std::is_same_v<T, int64_t>) {
@@ -198,6 +242,12 @@ class MinMaxHook final : public AggregationHook {
       if (std::is_same_v<T, double>) {
         return kDoubleMin;
       }
+      if (std::is_same_v<T, UnscaledShortDecimal>) {
+        return kShortDecimalMin;
+      }
+      if (std::is_same_v<T, UnscaledLongDecimal>) {
+        return kLongDecimalMin;
+      }
     } else {
       if (std::is_same_v<T, int64_t>) {
         return kBigintMax;
@@ -208,6 +258,12 @@ class MinMaxHook final : public AggregationHook {
       if (std::is_same_v<T, double>) {
         return kDoubleMax;
       }
+      if (std::is_same_v<T, UnscaledShortDecimal>) {
+        return kShortDecimalMax;
+      }
+      if (std::is_same_v<T, UnscaledLongDecimal>) {
+        return kLongDecimalMax;
+      }
     }
     return kGeneric;
   }

velox/vector/LazyVector.h

@@ -45,6 +45,10 @@ class ValueHook {
 
   virtual ~ValueHook() = default;
 
+  virtual std::string toString() const {
+    return "ValueHook";
+  }
+
   virtual bool acceptsNulls() const {
     return false;
   }