YQL-13966 const predicate, equijoin for cartesian part

ref:62ff302cfdf97a1b8d24a6e4f3aafc4647e4cded

1 files changed, 162 insertions, 111 deletions

diff --git a/ydb/library/yql/core/common_opt/yql_co_pgselect.cpp b/ydb/library/yql/core/common_opt/yql_co_pgselect.cpp
index c81feb38890..f509664b4a0 100644
--- a/ydb/library/yql/core/common_opt/yql_co_pgselect.cpp
+++ b/ydb/library/yql/core/common_opt/yql_co_pgselect.cpp
@@ -296,18 +296,18 @@ std::pair<TExprNode::TPtr, TExprNode::TPtr> RewriteSubLinks(TPositionHandle pos,
                         .Build();
                 }
             } else {
-                auto fullColList = ctx.Builder(node->Pos())

-                    .List()

-                        .Do([&](TExprNodeBuilder &parent) -> TExprNodeBuilder & {

-                            ui32 i = 0;

-                            for (const auto& c : extColumns) {

-                                parent.Atom(i++, c);

-                            }

-

-                            return parent;

-                        })

-                    .Seal()

-                    .Build();

+                auto fullColList = ctx.Builder(node->Pos())
+                    .List()
+                        .Do([&](TExprNodeBuilder &parent) -> TExprNodeBuilder & {
+                            ui32 i = 0;
+                            for (const auto& c : extColumns) {
+                                parent.Atom(i++, c);
+                            }
+
+                            return parent;
+                        })
+                    .Seal()
+                    .Build();
 
                 auto select = ExpandPgSelectSublink(node->TailPtr(), ctx, optCtx, it->second, cleanedInputs, inputAliases);
 
@@ -869,6 +869,72 @@ TExprNode::TListType BuildCleanedColumns(TPositionHandle pos, const TExprNode::T
     return cleanedInputs;
 }
 
+TExprNode::TPtr BuildMinus(TPositionHandle pos, const TExprNode::TPtr& left, const TExprNode::TPtr& right, const TExprNode::TPtr& predicate, TExprContext& ctx) {
+    return ctx.Builder(pos)
+        .Callable("Filter")
+            .Add(0, left)
+            .Lambda(1)
+                .Param("x")
+                .Callable(0, "Not")
+                    .Callable(0, "HasItems")
+                        .Callable(0, "Filter")
+                            .Add(0, right)
+                            .Lambda(1)
+                                .Param("y")
+                                .Callable("Coalesce")
+                                    .Callable(0, "FromPg")
+                                        .Apply(0, predicate)
+                                            .With(0)
+                                                .Callable("FlattenMembers")
+                                                    .List(0)
+                                                        .Atom(0, "")
+                                                        .Arg(1,"x")
+                                                    .Seal()
+                                                    .List(1)
+                                                        .Atom(0, "")
+                                                        .Arg(1, "y")
+                                                    .Seal()
+                                                .Seal()
+                                            .Done()
+                                        .Seal()
+                                    .Seal()
+                                    .Callable(1, "Bool")
+                                        .Atom(0, "0")
+                                    .Seal()
+                                .Seal()
+                            .Seal()
+                        .Seal()
+                    .Seal()
+                .Seal()
+            .Seal()
+        .Seal()
+        .Build();
+}
+
+TExprNode::TPtr BuildScalarMinus(TPositionHandle pos, const TExprNode::TPtr& left, const TExprNode::TPtr& right,
+    const TExprNode::TPtr& coalescedPredicate, TExprContext& ctx) {
+    return ctx.Builder(pos)
+        .Callable("If")
+            .Callable(0, "Or")
+                .Callable(0, "Not")
+                    .Add(0, coalescedPredicate)
+                .Seal()
+                .Callable(1, "==")
+                    .Callable(0, "Length")
+                        .Add(0, right)
+                    .Seal()
+                    .Callable(1, "Uint64")
+                        .Atom(0, "0")
+                    .Seal()
+                .Seal()
+            .Seal()
+            .Add(1, left)
+            .Callable(2, "EmptyList")
+            .Seal()
+        .Seal()
+        .Build();
+}
+
 std::tuple<TVector<ui32>, TExprNode::TListType> BuildJoinGroups(TPositionHandle pos, const TExprNode::TListType& cleanedInputs,
     const TExprNode::TPtr& joinOps, TExprContext& ctx, TOptimizeContext& optCtx) {
     TVector<ui32> groupForIndex;
@@ -890,94 +956,79 @@ std::tuple<TVector<ui32>, TExprNode::TListType> BuildJoinGroups(TPositionHandle
             // current = join current & with
             auto join = groupTuple->Child(i);
             auto joinType = join->Child(0)->Content();
-            auto cartesian = ctx.Builder(pos)
-                    .Callable("FlatMap")
-                        .Add(0, current)
-                        .Lambda(1)
-                            .Param("x")
-                            .Callable("Map")
-                                .Add(0, with)
-                                .Lambda(1)
-                                    .Param("y")
-                                    .Callable("FlattenMembers")
-                                        .List(0)
-                                            .Atom(0, "")
-                                            .Arg(1,"x")
-                                        .Seal()
-                                        .List(1)
-                                            .Atom(0, "")
-                                            .Arg(1, "y")
-                                        .Seal()
-                                    .Seal()
-                                .Seal()
-                            .Seal()
+            auto cartesian = JoinColumns(pos, current, with, nullptr, 0, ctx);
+            if (joinType == "cross") {
+                current = cartesian;
+                continue;
+            }
+
+            auto predicate = join->Tail().TailPtr();
+            if (!IsDepended(predicate->Tail(), predicate->Head().Head())) {
+                auto coalescedPredicate = ctx.Builder(pos)
+                    .Callable("Coalesce")
+                        .Callable(0, "FromPg")
+                            .Add(0, predicate->TailPtr())
+                        .Seal()
+                        .Callable(1, "Bool")
+                            .Atom(0, "0")
                         .Seal()
                     .Seal()
                     .Build();
 
-            auto buildMinus = [&](auto left, auto right) {
-                return ctx.Builder(pos)
-                    .Callable("Filter")
-                        .Add(0, left)
-                        .Lambda(1)
-                            .Param("x")
-                            .Callable(0, "Not")
-                                .Callable(0, "HasItems")
-                                    .Callable(0, "Filter")
-                                        .Add(0, right)
-                                        .Lambda(1)
-                                            .Param("y")
-                                            .Callable("Coalesce")
-                                                .Callable(0, "FromPg")
-                                                    .Apply(0, join->Tail().TailPtr())
-                                                        .With(0)
-                                                            .Callable("FlattenMembers")
-                                                                .List(0)
-                                                                    .Atom(0, "")
-                                                                    .Arg(1,"x")
-                                                                .Seal()
-                                                                .List(1)
-                                                                    .Atom(0, "")
-                                                                    .Arg(1, "y")
-                                                                .Seal()
-                                                            .Seal()
-                                                        .Done()
-                                                    .Seal()
-                                                .Seal()
-                                                .Callable(1, "Bool")
-                                                    .Atom(0, "0")
-                                                .Seal()
-                                            .Seal()
-                                        .Seal()
-                                    .Seal()
-                                .Seal()
-                            .Seal()
+                auto main = ctx.Builder(pos)
+                    .Callable("If")
+                        .Add(0, coalescedPredicate)
+                        .Add(1, cartesian)
+                        .Callable(2, "EmptyList")
                         .Seal()
                     .Seal()
                     .Build();
-            };
 
-            TExprNode::TPtr filteredCartesian;
-            if (joinType != "cross") {
-                filteredCartesian = BuildFilter(pos, cartesian, join->Tail().TailPtr(), {}, {}, ctx, optCtx);
+                if (joinType == "inner") {
+                    current = main;
+                } else if (joinType == "left") {
+                    current = ctx.Builder(pos)
+                        .Callable("UnionAll")
+                            .Add(0, main)
+                            .Add(1, BuildScalarMinus(pos, current, with, coalescedPredicate, ctx))
+                        .Seal()
+                        .Build();
+                } else if (joinType == "right") {
+                    current = ctx.Builder(pos)
+                        .Callable("UnionAll")
+                            .Add(0, main)
+                            .Add(1, BuildScalarMinus(pos, with, current, coalescedPredicate, ctx))
+                        .Seal()
+                        .Build();
+                } else {
+                    YQL_ENSURE(joinType == "full");
+                    current = ctx.Builder(pos)
+                        .Callable("UnionAll")
+                            .Add(0, main)
+                            .Add(1, BuildScalarMinus(pos, current, with, coalescedPredicate, ctx))
+                            .Add(2, BuildScalarMinus(pos, with, current, coalescedPredicate, ctx))
+                        .Seal()
+                        .Build();
+                }
+
+                continue;
             }
 
-            if (joinType == "cross") {
-                current = cartesian;
-            } else if (joinType == "inner") {
+            auto filteredCartesian = BuildFilter(pos, cartesian, predicate, {}, {}, ctx, optCtx);
+            if (joinType == "inner") {
                 current = filteredCartesian;
             } else if (joinType == "left") {
                 current = ctx.Builder(pos)
                     .Callable("UnionAll")
                         .Add(0, filteredCartesian)
-                        .Add(1, buildMinus(current, with))
+                        .Add(1, BuildMinus(pos, current, with, predicate, ctx))
                     .Seal()
                     .Build();
             } else if (joinType == "right") {
                 current = ctx.Builder(pos)
                     .Callable("UnionAll")
                         .Add(0, filteredCartesian)
-                        .Add(1, buildMinus(with, current))
+                        .Add(1, BuildMinus(pos, with, current, predicate, ctx))
                     .Seal()
                     .Build();
             } else {
@@ -985,8 +1036,8 @@ std::tuple<TVector<ui32>, TExprNode::TListType> BuildJoinGroups(TPositionHandle
                 current = ctx.Builder(pos)
                     .Callable("UnionAll")
                         .Add(0, filteredCartesian)
-                        .Add(1, buildMinus(current, with))
-                        .Add(2, buildMinus(with, current))
+                        .Add(1, BuildMinus(pos, current, with, predicate, ctx))
+                        .Add(2, BuildMinus(pos, with, current, predicate, ctx))
                     .Seal()
                     .Build();
             }
@@ -1715,42 +1766,42 @@ TExprNode::TPtr JoinOuter(TPositionHandle pos, TExprNode::TPtr list,
         const auto& inputAlias = input->Head().Content();
         YQL_ENSURE(inputAlias == outerInputAliases[index]);
         inputAliases.push_back(TString(inputAlias));
-        cleanedInputs.push_back(outerInputs[index]);

+        cleanedInputs.push_back(outerInputs[index]);
 
         auto type = input->Tail().GetTypeAnn()->Cast<TTypeExprType>()->GetType()->Cast<TStructExprType>();
         if (type->GetSize() == 0) {
             continue;
         }
 
-        auto colList = ctx.Builder(pos)

-            .List()

-            .Do([&](TExprNodeBuilder &parent) -> TExprNodeBuilder & {

-                ui32 i = 0;

-                for (const auto& item : type->GetItems()) {

-                    parent.Atom(i++, NTypeAnnImpl::MakeAliasedColumn(inputAlias, item->GetName()));

-                }

-

-                return parent;

-            })

-            .Seal()

-            .Build();

-

-        auto outerInput = ctx.Builder(pos)

-            .Callable("ExtractMembers")

-            .Add(0, outerInputs[index])

-            .Add(1, colList)

-            .Seal()

-            .Build();

-

-        auto uniqueOuterInput = ctx.Builder(pos)

-            .Callable("Aggregate")

-            .Add(0, outerInput)

-            .Add(1, colList)

-            .List(2)

-            .Seal()

-            .Seal()

-            .Build();

-

+        auto colList = ctx.Builder(pos)
+            .List()
+            .Do([&](TExprNodeBuilder &parent) -> TExprNodeBuilder & {
+                ui32 i = 0;
+                for (const auto& item : type->GetItems()) {
+                    parent.Atom(i++, NTypeAnnImpl::MakeAliasedColumn(inputAlias, item->GetName()));
+                }
+
+                return parent;
+            })
+            .Seal()
+            .Build();
+
+        auto outerInput = ctx.Builder(pos)
+            .Callable("ExtractMembers")
+            .Add(0, outerInputs[index])
+            .Add(1, colList)
+            .Seal()
+            .Build();
+
+        auto uniqueOuterInput = ctx.Builder(pos)
+            .Callable("Aggregate")
+            .Add(0, outerInput)
+            .Add(1, colList)
+            .List(2)
+            .Seal()
+            .Seal()
+            .Build();
+
         list = JoinColumns(pos, list, uniqueOuterInput, nullptr, 0, ctx);
     }