dbms: Server: queries with several replicas: development [#METR-14410]

dbms/include/DB/Storages/MergeTree/MergeTreeDataSelectExecutor.h

@@ -6,14 +6,11 @@
 namespace DB
 {
 
-class PartsWithRangesSplitter;
 
 /** Выполняет запросы SELECT на данных из merge-дерева.
   */
 class MergeTreeDataSelectExecutor
 {
-	friend class PartsWithRangesSplitter;
-
 public:
 	MergeTreeDataSelectExecutor(MergeTreeData & data_);
 
@@ -30,11 +27,7 @@ public:
 		unsigned threads = 1,
 		size_t * inout_part_index = nullptr);
 
-private:
-	MergeTreeData & data;
-
-	Logger * log;
-
+public:
 	struct RangesInDataPart
 	{
 		MergeTreeData::DataPartPtr data_part;
@@ -49,10 +42,13 @@ private:
 		}
 	};
 
-public:
 	typedef std::vector<RangesInDataPart> RangesInDataParts;
 
 private:
+	MergeTreeData & data;
+
+	Logger * log;
+
 	size_t min_marks_for_seek;
 	size_t min_marks_for_concurrent_read;
 	size_t max_marks_to_use_cache;

dbms/include/DB/Storages/MergeTree/PartsWithRangesSplitter.h

@@ -1,20 +1,21 @@
+#pragma once
+
 #include <DB/Storages/MergeTree/MergeTreeDataSelectExecutor.h>
 
 namespace DB
 {
 
-	/// Разбиваем parts_with_ranges на n частей.
-	/// Следующие условия должны быть выполнены:
-	/// - 1 <= n <= settings.max_clusters_count;
-	/// - каждая из n частей имеет >= min_cluster_size записей.
-	/// 3 levels: cluster / part / range
-
-	class PartsWithRangesSplitter
-	{
-	public:
+/// Разбиваем parts_with_ranges на n частей.
+/// Следующие условия должны быть выполнены:
+/// - 1 <= n <= settings.max_clusters_count;
+/// - каждая из n частей имеет >= min_cluster_size записей.
+/// 3 levels: cluster / part / range
+class PartsWithRangesSplitter
+{
+public:
 	typedef MergeTreeDataSelectExecutor::RangesInDataParts Cluster;
 
-	public:
+public:
 	PartsWithRangesSplitter(const Cluster & input_cluster_, 
 							size_t total_size_, size_t min_cluster_size_, size_t max_clusters_count_);
 
@@ -24,7 +25,7 @@ namespace DB
 
 	std::vector<Cluster> perform();
 
-	private:
+private:
 	void init();
 	bool emit();
 	bool updateCluster();
@@ -35,7 +36,7 @@ namespace DB
 	bool isRangeConsumed() const { return range_begin == range_end; }
 	bool isClusterConsumed() const { return cluster_begin == cluster_end; }
 
- 	private:
+private:
 	// Input data.
 	const Cluster & input_cluster;
 	Cluster::const_iterator input_part;
@@ -57,5 +58,6 @@ namespace DB
 
 	size_t cluster_begin;
 	size_t cluster_end;
-	};
+};
+
 }

dbms/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp

@@ -14,8 +14,9 @@
 
 namespace
 {
-	std::pair<UInt64, UInt64> computeHash(const DB::MergeTreeDataSelectExecutor::RangesInDataParts & cluster)
-	{
+
+std::pair<UInt64, UInt64> computeHash(const DB::MergeTreeDataSelectExecutor::RangesInDataParts & cluster)
+{
 	SipHash hash;
 	for (const auto & part_with_ranges : cluster)
 	{
@@ -34,7 +35,8 @@ namespace
 
 	hash.get128(lo, hi);
 	return std::make_pair(lo, hi);
-	}
+}
+
 }
 
 namespace DB

dbms/src/Storages/MergeTree/PartsWithRangesSplitter.cpp

@@ -3,25 +3,25 @@
 namespace DB
 {
 
-	PartsWithRangesSplitter::PartsWithRangesSplitter(const Cluster & input_cluster_, 
+PartsWithRangesSplitter::PartsWithRangesSplitter(const Cluster & input_cluster_, 
 						size_t total_size_, size_t min_cluster_size_, size_t max_clusters_count_) 
-	: input_cluster(input_cluster_),
-	total_size(total_size_),
-	remaining_size(total_size_),
-	min_cluster_size(min_cluster_size_),
-	max_clusters_count(max_clusters_count_)
-	{
-	}
+: input_cluster(input_cluster_),
+total_size(total_size_),
+remaining_size(total_size_),
+min_cluster_size(min_cluster_size_),
+max_clusters_count(max_clusters_count_)
+{
+}
 
-	std::vector<PartsWithRangesSplitter::Cluster> PartsWithRangesSplitter::perform()
-	{
+std::vector<PartsWithRangesSplitter::Cluster> PartsWithRangesSplitter::perform()
+{
 	init();
 	while (emit()) {}
 	return output_clusters;
-	}
+}
 
-	void PartsWithRangesSplitter::init()
-	{
+void PartsWithRangesSplitter::init()
+{
 	size_t clusters_count = max_clusters_count;
 	while ((clusters_count > 0) && (total_size < (min_cluster_size * clusters_count)))
 		--clusters_count;
@@ -39,10 +39,10 @@ namespace DB
 	current_output_cluster = output_clusters.begin();
 	addPart();
 	initClusterInfo();
-	}
+}
 
-	bool PartsWithRangesSplitter::emit()
-	{
+bool PartsWithRangesSplitter::emit()
+{
 	size_t new_size = std::min(range_end - range_begin, cluster_end - cluster_begin);
 	current_output_part->ranges.push_back(MarkRange(range_begin, range_begin + new_size));
 
@@ -55,10 +55,10 @@ namespace DB
 		return updateRange(true);
 	else
 		return false;
-	}
+}
 
-	bool PartsWithRangesSplitter::updateCluster()
-	{
+bool PartsWithRangesSplitter::updateCluster()
+{
 	++current_output_cluster;
 	if (current_output_cluster == output_clusters.end())
 		return false;
@@ -70,10 +70,10 @@ namespace DB
 	addPart();
 	initClusterInfo();
 	return true;
-	}
+}
 
-	bool PartsWithRangesSplitter::updateRange(bool add_part)
-	{
+bool PartsWithRangesSplitter::updateRange(bool add_part)
+{
 	++input_range;
 	if (input_range == input_part->ranges.end())
 	{
@@ -89,30 +89,30 @@ namespace DB
 
 	initRangeInfo();
 	return true;
-	}
+}
 
-	void PartsWithRangesSplitter::addPart()
-	{
+void PartsWithRangesSplitter::addPart()
+{
 	MergeTreeDataSelectExecutor::RangesInDataPart new_part;
 	new_part.data_part = input_part->data_part;
 	current_output_cluster->push_back(new_part);
 	current_output_part = &(current_output_cluster->back());
-	}
+}
 
-	void PartsWithRangesSplitter::initRangeInfo()
-	{
+void PartsWithRangesSplitter::initRangeInfo()
+{
 	range_begin = 0;
 	range_end = input_range->end - input_range->begin;
-	}
+}
 
-	void PartsWithRangesSplitter::initClusterInfo()
-	{
+void PartsWithRangesSplitter::initClusterInfo()
+{
 	cluster_begin = 0;
 	cluster_end = cluster_size;
 
 	remaining_size -= cluster_size;
 	if (remaining_size < cluster_size)
 		cluster_end += remaining_size;
-	}
+}
 
 }