From c6dbbcbf277ac35f398637980bb57398a4434dbc Mon Sep 17 00:00:00 2001
From: Ilya Kuznetsov <ilya@manticoresearch.com>
Date: Sun, 20 Aug 2023 11:25:18 +0200
Subject: [PATCH] implemented better analyzer rewinding

---
 columnar/accessor/accessor.h       | 11 ++++++++
 columnar/accessor/accessortraits.h | 31 ++++++----------------
 columnar/columnar.cpp              | 42 +++++++++---------------------
 3 files changed, 32 insertions(+), 52 deletions(-)

diff --git a/columnar/accessor/accessor.h b/columnar/accessor/accessor.h
index 96e754b2..eff126c0 100644
--- a/columnar/accessor/accessor.h
+++ b/columnar/accessor/accessor.h
@@ -32,12 +32,23 @@ class MatchingBlocks_c
 	FORCE_INLINE void	Add ( int iBlock ) { m_dBlocks.push_back(iBlock); }
 	FORCE_INLINE int	GetBlock ( int iBlock ) const { return m_dBlocks[iBlock]; }
 	FORCE_INLINE int	GetNumBlocks() const { return (int)m_dBlocks.size(); }
+	FORCE_INLINE int	Find ( int iStartBlock, int iValue );
 
 private:
 	std::vector<int>	m_dBlocks;
 };
 
 
+int MatchingBlocks_c::Find ( int iStartBlock, int iValue )
+{
+	auto tFound = std::lower_bound ( m_dBlocks.begin()+iStartBlock, m_dBlocks.end(), iValue );
+	if ( tFound==m_dBlocks.end() )
+		return (int)m_dBlocks.size();
+
+	return tFound-m_dBlocks.begin();
+}
+
+
 using SharedBlocks_c = std::shared_ptr<MatchingBlocks_c>;
 
 class Analyzer_i : public common::BlockIterator_i
diff --git a/columnar/accessor/accessortraits.h b/columnar/accessor/accessortraits.h
index f72b16d7..b6545982 100644
--- a/columnar/accessor/accessortraits.h
+++ b/columnar/accessor/accessortraits.h
@@ -198,32 +198,17 @@ bool Analyzer_T<HAVE_MATCHING_BLOCKS>::MoveToSubblock ( int iSubblock )
 template <bool HAVE_MATCHING_BLOCKS>
 bool Analyzer_T<HAVE_MATCHING_BLOCKS>::HintRowID ( uint32_t tRowID )
 {
-	int iNextSubblock = m_iCurSubblock;
+	int iNextSubblock = m_tSubblockCalc.GetSubblockId(tRowID);
+	if constexpr ( HAVE_MATCHING_BLOCKS )
+		iNextSubblock = m_pMatchingSubblocks->Find ( m_iCurSubblock, iNextSubblock );
 
-	// we assume that we are only advancing forward
-	while ( iNextSubblock<m_iTotalSubblocks )
-	{
-		int iSubblockID;
-		if ( HAVE_MATCHING_BLOCKS )
-			iSubblockID = m_pMatchingSubblocks->GetBlock(iNextSubblock);
-		else
-			iSubblockID = iNextSubblock;
-
-		uint32_t tSubblockStart = m_tSubblockCalc.SubblockId2RowId(iSubblockID);
-		uint32_t tSubblockEnd = tSubblockStart + m_tSubblockCalc.m_iSubblockSize;
-
-		if ( tRowID<tSubblockStart || ( tRowID>=tSubblockStart && tRowID<tSubblockEnd ) )
-		{
-			if ( iNextSubblock!=m_iCurSubblock )
-				return MoveToSubblock(iNextSubblock);
-
-			return true;
-		}
+	if ( iNextSubblock>=m_iTotalSubblocks )
+		return false;
 
-		iNextSubblock++;
-	}
+	if ( iNextSubblock>m_iCurSubblock )
+		return MoveToSubblock(iNextSubblock);
 
-	return false;
+	return true;
 }
 
 template <bool HAVE_MATCHING_BLOCKS>
diff --git a/columnar/columnar.cpp b/columnar/columnar.cpp
index dfd05d6e..e7ed8acf 100644
--- a/columnar/columnar.cpp
+++ b/columnar/columnar.cpp
@@ -230,7 +230,7 @@ class BlockIterator_c : public BlockIterator_i
 	bool			WasCutoffHit() const final		{ return false; }
 
 private:
-	static const int MAX_COLLECTED = 128;
+	static const int MAX_COLLECTED = 1024;
 
 	std::shared_ptr<MatchingBlocks_c>	m_pMatchingBlocks;
 	std::array<uint32_t,MAX_COLLECTED>	m_dCollected;
@@ -250,9 +250,10 @@ class BlockIterator_c : public BlockIterator_i
 	int			m_iMinMaxLeafShift = 0;
 	int			m_iNumLevels = 0;
 
-	inline bool	SetCurBlock ( int iBlock );
-	FORCE_INLINE int GetNumDocs ( int iBlock ) const;
-	FORCE_INLINE uint32_t MinMaxBlockId2RowId ( int iBlockId ) const;
+	FORCE_INLINE bool		SetCurBlock ( int iBlock );
+	FORCE_INLINE int		GetNumDocs ( int iBlock ) const;
+	FORCE_INLINE uint32_t	MinMaxBlockId2RowId ( int iBlockId ) const		{ return iBlockId<<m_iMinMaxLeafShift; }
+	FORCE_INLINE int		RowId2MinMaxBlockId ( uint32_t uRowID ) const	{ return uRowID>>m_iMinMaxLeafShift; }
 };
 
 
@@ -284,35 +285,24 @@ bool BlockIterator_c::Setup ( const std::vector<HeaderWithLocator_t> & dHeaders,
 	return true;
 }
 
+
 void BlockIterator_c::AddDesc ( std::vector<IteratorDesc_t> & dDesc ) const
 {
 	for ( const auto & i : m_dAttrs )
 		dDesc.push_back ( { i, "prefilter" } );
 }
 
+
 bool BlockIterator_c::HintRowID ( uint32_t tRowID )
 {
-	int iNextBlock = m_iBlock;
-	int iNumBlocks = m_pMatchingBlocks->GetNumBlocks();
-
-	// we assume that we are only advancing forward
-	while ( iNextBlock<iNumBlocks )
-	{
-		uint32_t tSubblockStart = MinMaxBlockId2RowId ( m_pMatchingBlocks->GetBlock(iNextBlock) );
-		uint32_t tSubblockEnd = tSubblockStart + m_iDocsPerBlock;
-
-		if ( tRowID<tSubblockStart || ( tRowID>=tSubblockStart && tRowID<tSubblockEnd ) )
-		{
-			if ( iNextBlock!=m_iBlock )
-				SetCurBlock(iNextBlock);
-
-			return true;
-		}
+	int iNextBlock = m_pMatchingBlocks->Find ( m_iBlock, RowId2MinMaxBlockId(tRowID) );
+	if ( iNextBlock>=m_pMatchingBlocks->GetNumBlocks() )
+		return false;
 
-		iNextBlock++;
-	}
+	if ( iNextBlock>m_iBlock )
+		SetCurBlock(iNextBlock);
 
-	return false;
+	return true;
 }
 
 
@@ -377,12 +367,6 @@ int BlockIterator_c::GetNumDocs ( int iBlock ) const
 	return m_iDocsInLastBlock;
 }
 
-
-uint32_t BlockIterator_c::MinMaxBlockId2RowId ( int iBlockId ) const
-{
-	return iBlockId<<m_iMinMaxLeafShift;
-}
-
 //////////////////////////////////////////////////////////////////////////
 
 class Columnar_c final : public Columnar_i