duckdb 0.3.5-dev411.0 → 0.3.5-dev457.0

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 "43bb9d9ea"
-#define DUCKDB_VERSION "v0.3.5-dev411"
+#define DUCKDB_SOURCE_ID "e9ab0d515"
+#define DUCKDB_VERSION "v0.3.5-dev457"
 //===----------------------------------------------------------------------===//
 //                         DuckDB
 //
@@ -3978,6 +3978,9 @@ struct SequenceVector {
 extern "C" {
 #endif
 
+#ifndef ARROW_C_DATA_INTERFACE
+#define ARROW_C_DATA_INTERFACE
+
 #define ARROW_FLAG_DICTIONARY_ORDERED 1
 #define ARROW_FLAG_NULLABLE           2
 #define ARROW_FLAG_MAP_KEYS_SORTED    4
@@ -4014,7 +4017,10 @@ struct ArrowArray {
 	// Opaque producer-specific data
 	void *private_data;
 };
+#endif
 
+#ifndef ARROW_C_STREAM_INTERFACE
+#define ARROW_C_STREAM_INTERFACE
 // EXPERIMENTAL
 struct ArrowArrayStream {
 	// Callback to get the stream type
@@ -4039,6 +4045,7 @@ struct ArrowArrayStream {
 	// Opaque producer-specific data
 	void *private_data;
 };
+#endif
 
 #ifdef __cplusplus
 }
@@ -7161,32 +7168,46 @@ class FieldWriter;
 class FieldReader;
 class Vector;
 class ValidityStatistics;
+class DistinctStatistics;
+struct VectorData;
+
+enum StatisticsType { LOCAL_STATS = 0, GLOBAL_STATS = 1 };
 
 class BaseStatistics {
 public:
-	explicit BaseStatistics(LogicalType type);
+	BaseStatistics(LogicalType type, StatisticsType stats_type);
 	virtual ~BaseStatistics();
 
 	//! The type of the logical segment
 	LogicalType type;
 	//! The validity stats of the column (if any)
 	unique_ptr<BaseStatistics> validity_stats;
+	//! The approximate count distinct stats of the column (if any)
+	unique_ptr<BaseStatistics> distinct_stats;
+	//! Whether these are 'global' stats, i.e., over a whole table, or just over a segment
+	//! Some statistics are more expensive to keep, therefore we only keep them globally
+	StatisticsType stats_type;
 
 public:
+	static unique_ptr<BaseStatistics> CreateEmpty(LogicalType type, StatisticsType stats_type);
+
 	bool CanHaveNull() const;
 	bool CanHaveNoNull() const;
 
+	void UpdateDistinctStatistics(Vector &v, idx_t count);
+
 	virtual bool IsConstant() const {
 		return false;
 	}
 
-	static unique_ptr<BaseStatistics> CreateEmpty(LogicalType type);
-
 	virtual void Merge(const BaseStatistics &other);
 
 	virtual unique_ptr<BaseStatistics> Copy() const;
-	void Serialize(Serializer &serializer) const;
+	void CopyBase(const BaseStatistics &orig);
+
+	virtual void Serialize(Serializer &serializer) const;
 	virtual void Serialize(FieldWriter &writer) const;
+
 	static unique_ptr<BaseStatistics> Deserialize(Deserializer &source, LogicalType type);
 
 	//! Verify that a vector does not violate the statistics
@@ -7194,6 +7215,9 @@ public:
 	void Verify(Vector &vector, idx_t count) const;
 
 	virtual string ToString() const;
+
+protected:
+	void InitializeBase();
 };
 
 } // namespace duckdb
@@ -21653,7 +21677,7 @@ class Vector;
 
 class ValidityStatistics : public BaseStatistics {
 public:
-	ValidityStatistics(bool has_null = false, bool has_no_null = true);
+	explicit ValidityStatistics(bool has_null = false, bool has_no_null = true);
 
 	//! Whether or not the segment can contain NULL values
 	bool has_null;
@@ -21666,8 +21690,10 @@ public:
 	bool IsConstant() const override;
 
 	unique_ptr<BaseStatistics> Copy() const override;
+
 	void Serialize(FieldWriter &writer) const override;
-	static unique_ptr<BaseStatistics> Deserialize(FieldReader &reader);
+	static unique_ptr<ValidityStatistics> Deserialize(FieldReader &reader);
+
 	void Verify(Vector &vector, const SelectionVector &sel, idx_t count) const override;
 
 	static unique_ptr<BaseStatistics> Combine(const unique_ptr<BaseStatistics> &lstats,
@@ -21686,7 +21712,7 @@ public:
 	constexpr static uint32_t MAX_STRING_MINMAX_SIZE = 8;
 
 public:
-	explicit StringStatistics(LogicalType type);
+	explicit StringStatistics(LogicalType type, StatisticsType stats_type);
 
 	//! The minimum value of the segment, potentially truncated
 	data_t min[MAX_STRING_MINMAX_SIZE];
@@ -21780,8 +21806,8 @@ namespace duckdb {
 
 class NumericStatistics : public BaseStatistics {
 public:
-	explicit NumericStatistics(LogicalType type);
-	NumericStatistics(LogicalType type, Value min, Value max);
+	explicit NumericStatistics(LogicalType type, StatisticsType stats_type);
+	NumericStatistics(LogicalType type, Value min, Value max, StatisticsType stats_type);
 
 	//! The minimum value of the segment
 	Value min;