duckdb 0.3.5-dev692.0 → 0.3.5-dev699.0

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "duckdb",
   "main": "./lib/duckdb.js",
-  "version": "0.3.5-dev692.0",
+  "version": "0.3.5-dev699.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

package/src/duckdb.cpp

@@ -41296,11 +41296,11 @@ inline uint64_t TemplatedHash(const string_t &elem) {
 	data_ptr_t data = (data_ptr_t)elem.GetDataUnsafe();
 	const auto &len = elem.GetSize();
 	uint64_t h = 0;
-	for (idx_t i = 0; i < len / 8; i += 8) {
+	for (idx_t i = 0; i + sizeof(uint64_t) <= len; i += sizeof(uint64_t)) {
 		h ^= TemplatedHash<uint64_t>(Load<uint64_t>(data));
-		data += 8;
+		data += sizeof(uint64_t);
 	}
-	switch (len & 7) {
+	switch (len & (sizeof(uint64_t) - 1)) {
 	case 4:
 		h ^= TemplatedHash<uint32_t>(Load<uint32_t>(data));
 		break;
@@ -181055,6 +181055,8 @@ void BaseStatistics::Verify(Vector &vector, idx_t count) const {
 
 
 
+#include <math.h>
+
 namespace duckdb {
 
 DistinctStatistics::DistinctStatistics()
@@ -181115,7 +181117,7 @@ void DistinctStatistics::Update(VectorData &vdata, const LogicalType &type, idx_
 		return;
 	}
 	total_count += count;
-	count = MaxValue<idx_t>(idx_t(SAMPLE_RATE * double(count)), 1);
+	count = MinValue<idx_t>(idx_t(SAMPLE_RATE * MaxValue<idx_t>(STANDARD_VECTOR_SIZE, count)), count);
 	sample_count += count;
 
 	uint64_t indices[STANDARD_VECTOR_SIZE];
@@ -181133,12 +181135,17 @@ idx_t DistinctStatistics::GetCount() const {
 	if (sample_count == 0 || total_count == 0) {
 		return 0;
 	}
-	// Estimate HLL count because we use sampling
-	double hll_count = log->Count();
-	double unique_proportion = hll_count / double(sample_count);
-	double actual_sample_rate = double(sample_count) / double(total_count);
-	double multiplier = double(1) + unique_proportion * (double(1) / actual_sample_rate - double(1));
-	return idx_t(multiplier * hll_count);
+
+	double u = MinValue<idx_t>(log->Count(), sample_count);
+	double s = sample_count;
+	double n = total_count;
+
+	// Assume this proportion of the the sampled values occurred only once
+	double u1 = pow(u / s, 2) * u;
+
+	// Estimate total uniques using Good Turing Estimation
+	idx_t estimate = u + u1 / s * (n - s);
+	return MinValue<idx_t>(estimate, total_count);
 }
 
 } // namespace duckdb

package/src/duckdb.hpp

@@ -11,8 +11,8 @@ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLI
 #pragma once
 #define DUCKDB_AMALGAMATION 1
 #define DUCKDB_AMALGAMATION_EXTENDED 1
-#define DUCKDB_SOURCE_ID "c2581dfeb"
-#define DUCKDB_VERSION "v0.3.5-dev692"
+#define DUCKDB_SOURCE_ID "360aedc4f"
+#define DUCKDB_VERSION "v0.3.5-dev699"
 //===----------------------------------------------------------------------===//
 //                         DuckDB
 //