duckdb 0.3.5-dev692.0 → 0.3.5-dev705.0

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "duckdb",
   "main": "./lib/duckdb.js",
-  "version": "0.3.5-dev692.0",
+  "version": "0.3.5-dev705.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

package/src/connection.cpp

@@ -363,6 +363,13 @@ struct ExecTask : public Task {
 			}
 		}
 	}
+
+	void Callback() override {
+		auto env = object.Env();
+		Napi::HandleScope scope(env);
+		callback.Value().MakeCallback(object.Value(), {success ? env.Null() : Napi::String::New(env, error)});
+	};
+
 	std::string sql;
 	bool success;
 	std::string error;

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;
@@ -46260,7 +46260,6 @@ const SelectionVector *ConstantVector::ZeroSelectionVector(idx_t count, Selectio
 }
 
 void ConstantVector::Reference(Vector &vector, Vector &source, idx_t position, idx_t count) {
-	D_ASSERT(position < count);
 	auto &source_type = source.GetType();
 	switch (source_type.InternalType()) {
 	case PhysicalType::LIST: {
@@ -46308,7 +46307,7 @@ void ConstantVector::Reference(Vector &vector, Vector &source, idx_t position, i
 		auto &source_entries = StructVector::GetEntries(source);
 		auto &target_entries = StructVector::GetEntries(vector);
 		for (idx_t i = 0; i < source_entries.size(); i++) {
-			ConstantVector::Reference(*target_entries[i], *source_entries[i], position, count);
+			ConstantVector::Reference(*target_entries[i], *source_entries[i], struct_index, count);
 		}
 		vector.SetVectorType(VectorType::CONSTANT_VECTOR);
 		break;
@@ -51740,6 +51739,7 @@ static inline void TemplatedLoopHash(Vector &input, Vector &result, const Select
 
 template <bool HAS_RSEL, bool FIRST_HASH>
 static inline void StructLoopHash(Vector &input, Vector &hashes, const SelectionVector *rsel, idx_t count) {
+	input.Normalify(count);
 	auto &children = StructVector::GetEntries(input);
 
 	D_ASSERT(!children.empty());
@@ -114875,6 +114875,9 @@ void ClientContext::LogQueryInternal(ClientContextLock &, const string &query) {
 
 unique_ptr<QueryResult> ClientContext::Query(unique_ptr<SQLStatement> statement, bool allow_stream_result) {
 	auto pending_query = PendingQuery(move(statement), allow_stream_result);
+	if (!pending_query->success) {
+		return make_unique<MaterializedQueryResult>(pending_query->error);
+	}
 	return pending_query->Execute();
 }
 
@@ -181055,6 +181058,8 @@ void BaseStatistics::Verify(Vector &vector, idx_t count) const {
 
 
 
+#include <math.h>
+
 namespace duckdb {
 
 DistinctStatistics::DistinctStatistics()
@@ -181115,7 +181120,7 @@ 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 +181138,17 @@ 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