ManhattanDistance returns float, so bit vector is no longer limited by UINT16_MAX

Author: azevaykin

ydb/library/yql/udfs/common/knn/knn-distance.h

@@ -48,7 +48,7 @@ static std::optional<float> KnnManhattanDistance(const TStringRef& str1, const T
             if (Y_UNLIKELY(vector1.size() != vector2.size() || vector1.empty() || vector1.size() > UINT16_MAX))
                 return {};  
 
-            ui16 ret = 0;
+            ui64 ret = 0;
             for (size_t i = 0; i < vector1.size(); ++i)
                 ret += __builtin_popcountll(vector1[i] ^ vector2[i]);
             return ret;
@@ -91,10 +91,10 @@ static std::optional<float> KnnEuclideanDistance(const TStringRef& str1, const T
             if (Y_UNLIKELY(vector1.size() != vector2.size() || vector1.empty() || vector1.size() > UINT16_MAX))
                 return {};  
 
-            ui16 ret = 0;
+            ui64 ret = 0;
             for (size_t i = 0; i < vector1.size(); ++i)
                 ret += __builtin_popcountll(vector1[i] ^ vector2[i]);
-            return NPrivate::NL2Distance::L2DistanceSqrt<ui64>(ret);
+            return NPrivate::NL2Distance::L2DistanceSqrt(ret);
         }
         default:
             return {};

ydb/library/yql/udfs/common/knn/knn-serializer.h

@@ -80,14 +80,12 @@ class TKnnVectorSerializer {
 // Encode all positive floats as bit 1, negative floats as bit 0.
 // So 1024 float vector is serialized in 1024/8=128 bytes.
 // Place all bits in ui64. So, only vector sizes divisible by 64 are supported.
-// Max vector lenght is 32767.
 class TKnnBitVectorSerializer {
 public:
     static TUnboxedValue Serialize(const IValueBuilder* valueBuilder, const TUnboxedValue x) {
         auto serialize = [&x] (IOutputStream& outStream) {
             ui64 accumulator = 0;
             ui8 filledBits = 0;
-            ui64 lenght = 0;
 
             EnumerateVector(x,  [&] (float element) { 
                 if (element > 0)
@@ -96,7 +94,6 @@ class TKnnBitVectorSerializer {
                 ++filledBits;
                 if (filledBits == 64) {
                     outStream.Write(&accumulator, sizeof(ui64));
-                    lenght++;
                     accumulator = 0;
                     filledBits = 0;
                 }
@@ -106,10 +103,6 @@ class TKnnBitVectorSerializer {
             if (Y_UNLIKELY(filledBits))
                 return false;
 
-            // max vector lenght is 32767
-            if (Y_UNLIKELY(lenght > UINT16_MAX))
-                return false;
-
             const EFormat format = EFormat::BitVector;
             outStream.Write(&format, HeaderLen);