move choosing bucket to separate function

Author: vladl2802

ydb/library/yql/minikql/comp_nodes/mkql_wide_combine.cpp

@@ -486,9 +486,7 @@ class TSpillingSupportState : public TComputationValue<TSpillingSupportState> {
             return isNew ? ETasteResult::Init : ETasteResult::Update;
         }
 
-        auto hash = Hasher(ViewForKeyAndState.data());
-        XXH64_hash_t hashed_hash = XXH64(&hash, sizeof(hash), 0);
-        auto bucketId = hashed_hash % SpilledBucketCount;
+        auto bucketId = ChooseBucket(ViewForKeyAndState.data());
         auto& bucket = SpilledBuckets[bucketId];
 
         if (bucket.BucketState == TSpilledBucket::EBucketState::InMemory) {
@@ -532,7 +530,15 @@ class TSpillingSupportState : public TComputationValue<TSpillingSupportState> {
 
         return value;
     }
+
 private:
+    ui64 ChooseBucket(const NUdf::TUnboxedValuePod *const key) {
+        auto provided_hash = Hasher(key);
+        XXH64_hash_t bucket = XXH64(&provided_hash, sizeof(provided_hash), 0) % SpilledBucketCount;
+        YQL_ENSURE(bucket < SpilledBucketCount, "bucket index is out of bounds");
+        return bucket;
+    }
+
     EUpdateResult FlushSpillingBuffersAndWait() {
         UpdateSpillingBuckets();
 
@@ -595,9 +601,7 @@ class TSpillingSupportState : public TComputationValue<TSpillingSupportState> {
             SplitStateSpillingBucket = -1;
         }
         while (const auto keyAndState = static_cast<NUdf::TUnboxedValue *>(InMemoryProcessingState.Extract())) {
-            auto hash = Hasher(keyAndState);  //Hasher uses only key for hashing
-            XXH64_hash_t hashed_hash = XXH64(&hash, sizeof(hash), 0);
-            auto bucketId = hashed_hash % SpilledBucketCount;
+            auto bucketId = ChooseBucket(keyAndState);  // This uses only key for hashing
             auto& bucket = SpilledBuckets[bucketId];
 
             bucket.LineCount++;
@@ -890,7 +894,8 @@ class TSpillingSupportState : public TComputationValue<TSpillingSupportState> {
     bool RecoverState; //sub mode for ProcessSpilledData
 
     TAsyncReadOperation AsyncReadOperation = std::nullopt;
-    static constexpr size_t SpilledBucketCount = 128;
+    static constexpr size_t SpilledBucketBits = 7;
+    static constexpr size_t SpilledBucketCount = 1 << SpilledBucketBits;
     std::deque<TSpilledBucket> SpilledBuckets;
     ui32 SpillingBucketsCount = 0;
     ui32 InMemoryBucketsCount = SpilledBucketCount;