duckdb 0.7.2-dev16.0 → 0.7.2-dev314.0

package/src/duckdb/src/storage/table/table_statistics.cpp

@@ -6,10 +6,7 @@ namespace duckdb {
 void TableStatistics::Initialize(const vector<LogicalType> &types, PersistentTableData &data) {
 	D_ASSERT(Empty());
 
-	column_stats.reserve(data.column_stats.size());
-	for (auto &stats : data.column_stats) {
-		column_stats.push_back(make_shared<ColumnStatistics>(std::move(stats)));
-	}
+	column_stats = std::move(data.table_stats.column_stats);
 	if (column_stats.size() != types.size()) { // LCOV_EXCL_START
 		throw IOException("Table statistics column count is not aligned with table column count. Corrupt file?");
 	} // LCOV_EXCL_STOP
@@ -70,7 +67,7 @@ void TableStatistics::MergeStats(TableStatistics &other) {
 	auto l = GetLock();
 	D_ASSERT(column_stats.size() == other.column_stats.size());
 	for (idx_t i = 0; i < column_stats.size(); i++) {
-		column_stats[i]->stats->Merge(*other.column_stats[i]->stats);
+		column_stats[i]->Merge(*other.column_stats[i]);
 	}
 }
 
@@ -80,7 +77,7 @@ void TableStatistics::MergeStats(idx_t i, BaseStatistics &stats) {
 }
 
 void TableStatistics::MergeStats(TableStatisticsLock &lock, idx_t i, BaseStatistics &stats) {
-	column_stats[i]->stats->Merge(stats);
+	column_stats[i]->Statistics().Merge(stats);
 }
 
 ColumnStatistics &TableStatistics::GetStats(idx_t i) {
@@ -89,7 +86,30 @@ ColumnStatistics &TableStatistics::GetStats(idx_t i) {
 
 unique_ptr<BaseStatistics> TableStatistics::CopyStats(idx_t i) {
 	lock_guard<mutex> l(stats_lock);
-	return column_stats[i]->stats->Copy();
+	auto result = column_stats[i]->Statistics().Copy();
+	if (column_stats[i]->HasDistinctStats()) {
+		result.SetDistinctCount(column_stats[i]->DistinctStats().GetCount());
+	}
+	return result.ToUnique();
+}
+
+void TableStatistics::CopyStats(TableStatistics &other) {
+	for (auto &stats : column_stats) {
+		other.column_stats.push_back(stats->Copy());
+	}
+}
+
+void TableStatistics::Serialize(Serializer &serializer) {
+	for (auto &stats : column_stats) {
+		stats->Serialize(serializer);
+	}
+}
+
+void TableStatistics::Deserialize(Deserializer &source, ColumnList &columns) {
+	for (auto &col : columns.Physical()) {
+		auto stats = ColumnStatistics::Deserialize(source, col.GetType());
+		column_stats.push_back(std::move(stats));
+	}
 }
 
 unique_ptr<TableStatisticsLock> TableStatistics::GetLock() {

package/src/duckdb/src/storage/table/update_segment.cpp

@@ -1,9 +1,7 @@
 #include "duckdb/storage/table/update_segment.hpp"
 
 #include "duckdb/storage/statistics/distinct_statistics.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
-#include "duckdb/storage/statistics/string_statistics.hpp"
-#include "duckdb/storage/statistics/validity_statistics.hpp"
+
 #include "duckdb/storage/table/column_data.hpp"
 #include "duckdb/transaction/duck_transaction.hpp"
 #include "duckdb/transaction/update_info.hpp"
@@ -55,7 +53,7 @@ UpdateSegment::~UpdateSegment() {
 }
 
 void UpdateSegment::ClearUpdates() {
-	stats.Reset();
+	stats.statistics.Copy(BaseStatistics::CreateEmpty(stats.statistics.GetType()));
 	root.reset();
 	heap.Destroy();
 }
@@ -905,17 +903,17 @@ static UpdateSegment::merge_update_function_t GetMergeUpdateFunction(PhysicalTyp
 //===--------------------------------------------------------------------===//
 unique_ptr<BaseStatistics> UpdateSegment::GetStatistics() {
 	lock_guard<mutex> stats_guard(stats_lock);
-	return stats.statistics->Copy();
+	return stats.statistics.ToUnique();
 }
 
 idx_t UpdateValidityStatistics(UpdateSegment *segment, SegmentStatistics &stats, Vector &update, idx_t count,
                                SelectionVector &sel) {
 	auto &mask = FlatVector::Validity(update);
-	auto &validity = (ValidityStatistics &)*stats.statistics;
-	if (!mask.AllValid() && !validity.has_null) {
+	auto &validity = stats.statistics;
+	if (!mask.AllValid() && !validity.CanHaveNull()) {
 		for (idx_t i = 0; i < count; i++) {
 			if (!mask.RowIsValid(i)) {
-				validity.has_null = true;
+				validity.SetHasNull();
 				break;
 			}
 		}
@@ -932,7 +930,7 @@ idx_t TemplatedUpdateNumericStatistics(UpdateSegment *segment, SegmentStatistics
 
 	if (mask.AllValid()) {
 		for (idx_t i = 0; i < count; i++) {
-			NumericStatistics::Update<T>(stats, update_data[i]);
+			NumericStats::Update<T>(stats.statistics, update_data[i]);
 		}
 		sel.Initialize(nullptr);
 		return count;
@@ -942,7 +940,7 @@ idx_t TemplatedUpdateNumericStatistics(UpdateSegment *segment, SegmentStatistics
 		for (idx_t i = 0; i < count; i++) {
 			if (mask.RowIsValid(i)) {
 				sel.set_index(not_null_count++, i);
-				NumericStatistics::Update<T>(stats, update_data[i]);
+				NumericStats::Update<T>(stats.statistics, update_data[i]);
 			}
 		}
 		return not_null_count;
@@ -955,7 +953,7 @@ idx_t UpdateStringStatistics(UpdateSegment *segment, SegmentStatistics &stats, V
 	auto &mask = FlatVector::Validity(update);
 	if (mask.AllValid()) {
 		for (idx_t i = 0; i < count; i++) {
-			((StringStatistics &)*stats.statistics).Update(update_data[i]);
+			StringStats::Update(stats.statistics, update_data[i]);
 			if (!update_data[i].IsInlined()) {
 				update_data[i] = segment->GetStringHeap().AddBlob(update_data[i]);
 			}
@@ -968,7 +966,7 @@ idx_t UpdateStringStatistics(UpdateSegment *segment, SegmentStatistics &stats, V
 		for (idx_t i = 0; i < count; i++) {
 			if (mask.RowIsValid(i)) {
 				sel.set_index(not_null_count++, i);
-				((StringStatistics &)*stats.statistics).Update(update_data[i]);
+				StringStats::Update(stats.statistics, update_data[i]);
 				if (!update_data[i].IsInlined()) {
 					update_data[i] = segment->GetStringHeap().AddBlob(update_data[i]);
 				}

package/src/duckdb/third_party/libpg_query/include/nodes/nodes.hpp

@@ -194,6 +194,9 @@ typedef enum PGNodeTag {
 	T_PGOnConflictExpr,
 	T_PGIntoClause,
 	T_PGLambdaFunction,
+	T_PGPivotExpr,
+	T_PGPivot,
+	T_PGPivotStmt,
 
 	/*
 	 * TAGS FOR EXPRESSION STATE NODES (execnodes.h)

package/src/duckdb/third_party/libpg_query/include/nodes/parsenodes.hpp

@@ -1162,6 +1162,37 @@ typedef struct PGUpdateStmt {
 	PGWithClause *withClause; /* WITH clause */
 } PGUpdateStmt;
 
+/* ----------------------
+ *		Pivot Expression
+ * ----------------------
+ */
+typedef struct PGPivot {
+	PGNodeTag type;
+	PGList *pivot_columns;  /* The column names to pivot on */
+	PGList *pivot_value;    /* The set of pivot values */
+	char *pivot_enum;       /* The enum to fetch the unique values from */
+} PGPivot;
+
+typedef struct PGPivotExpr {
+	PGNodeTag type;
+	PGNode *source;      /* the source subtree */
+	PGList *aggrs;       /* The aggregations to pivot over (PIVOT only) */
+	PGList *unpivots;    /* The names to unpivot over (UNPIVOT only) */
+	PGList *pivots;      /* The set of pivot values */
+	PGList *groups;      /* The set of groups to pivot over (if any) */
+	PGAlias *alias;      /* table alias & optional column aliases */
+	bool include_nulls;  /* Whether or not to include NULL values (UNPIVOT only */
+} PGPivotExpr;
+
+typedef struct PGPivotStmt {
+	PGNodeTag type;
+	PGNode *source;      /* The source to pivot */
+	PGList *aggrs;       /* The aggregations to pivot over (PIVOT only) */
+	PGList *unpivots;    /* The names to unpivot over (UNPIVOT only) */
+	PGList *columns;     /* The set of columns to pivot over */
+	PGList *groups;      /* The set of groups to pivot over (if any) */
+} PGPivotStmt;
+
 /* ----------------------
  *		Select Statement
  *
@@ -1204,6 +1235,9 @@ typedef struct PGSelectStmt {
 	 */
 	PGList *valuesLists; /* untransformed list of expression lists */
 
+	/* When representing a pivot statement, all values are NULL besides the pivot field */
+	PGPivotStmt *pivot;       /* PIVOT statement */
+
 	/*
 	 * These fields are used in both "leaf" SelectStmts and upper-level
 	 * SelectStmts.
@@ -2107,5 +2141,4 @@ typedef struct PGUseStmt {
 } PGUseStmt;
 
 
-
 }