From 8fe93946bc369873a7ffbb3a7403463aa80e3117 Mon Sep 17 00:00:00 2001
From: ziganshinmr <ziganshinmr@yandex-team.com>
Date: Mon, 17 Feb 2025 23:25:34 +0300
Subject: BlockMapJoinCore refactor
* Split storage and index parts from BlockMapJoinCore computation node into separate BlockStorage and BlockIndex nodes in order to allow multiple join nodes to reuse the same block data and index for the right table where possible
* Corresponding s-expressions changes
commit_hash:40e39fb0b22c2f929c184963b5bd901006122c14
---
yql/essentials/minikql/mkql_program_builder.h | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
(limited to 'yql/essentials/minikql/mkql_program_builder.h')
diff --git a/yql/essentials/minikql/mkql_program_builder.h b/yql/essentials/minikql/mkql_program_builder.h
index b67d80861d..01c0223293 100644
--- a/yql/essentials/minikql/mkql_program_builder.h
+++ b/yql/essentials/minikql/mkql_program_builder.h
@@ -17,6 +17,9 @@ class TBuiltinFunctionRegistry;
constexpr std::string_view RandomMTResource = "MTRand";
constexpr std::string_view ResourceQueuePrefix = "TResourceQueue:";
+constexpr std::string_view BlockStorageResourcePrefix = "TBlockStorage:";
+constexpr std::string_view BlockMapJoinIndexResourcePrefix = "TBlockMapJoinIndex:";
+constexpr std::string_view BlockMapJoinIndexResourceSeparator = ":$YqlKeyColumns:";
enum class EJoinKind {
Min = 1,
@@ -133,6 +136,7 @@ struct TAggInfo {
std::vector<ui32> ArgsColumns;
};
+std::vector<TType*> ValidateBlockType(const TType* type, bool unwrap = true);
std::vector<TType*> ValidateBlockStreamType(const TType* streamType, bool unwrap = true);
std::vector<TType*> ValidateBlockFlowType(const TType* flowType, bool unwrap = true);
@@ -258,9 +262,12 @@ public:
TRuntimeNode BlockFromPg(TRuntimeNode input, TType* returnType);
TRuntimeNode BlockPgResolvedCall(const std::string_view& name, ui32 id,
const TArrayRef<const TRuntimeNode>& args, TType* returnType);
- TRuntimeNode BlockMapJoinCore(TRuntimeNode leftStream, TRuntimeNode rightStream, EJoinKind joinKind,
+ TRuntimeNode BlockStorage(TRuntimeNode stream, TType* returnType);
+ TRuntimeNode BlockMapJoinIndex(TRuntimeNode blockStorage, TType* streamItemType, const TArrayRef<const ui32>& keyColumns, bool any, TType* returnType);
+ TRuntimeNode BlockMapJoinCore(TRuntimeNode leftStream, TRuntimeNode rightBlockStorage, TType* rightStreamItemType, EJoinKind joinKind,
const TArrayRef<const ui32>& leftKeyColumns, const TArrayRef<const ui32>& leftKeyDrops,
- const TArrayRef<const ui32>& rightKeyColumns, const TArrayRef<const ui32>& rightKeyDrops, bool rightAny, TType* returnType);
+ const TArrayRef<const ui32>& rightKeyColumns, const TArrayRef<const ui32>& rightKeyDrops, TType* returnType
+ );
//-- logical functions
TRuntimeNode BlockNot(TRuntimeNode data);
--
cgit v1.2.3