1+ #include " kqp_opt_cbo.h"
2+ #include " kqp_opt_log_impl.h"
3+
4+ #include < ydb/library/yql/core/yql_opt_utils.h>
5+ #include < ydb/library/yql/utils/log/log.h>
6+
7+
8+ namespace NKikimr ::NKqp::NOpt {
9+
10+ using namespace NYql ;
11+ using namespace NYql ::NCommon;
12+ using namespace NYql ::NDq;
13+ using namespace NYql ::NNodes;
14+
15+ namespace {
16+
17+ /* *
18+ * KQP specific rule to check if a LookupJoin is applicable
19+ */
20+ bool IsLookupJoinApplicableDetailed (const std::shared_ptr<NYql::TRelOptimizerNode>& node, const TVector<TString>& joinColumns, const TKqpProviderContext& ctx) {
21+
22+ auto rel = std::static_pointer_cast<TKqpRelOptimizerNode>(node);
23+ auto expr = TExprBase (rel->Node );
24+
25+ if (ctx.KqpCtx .IsScanQuery () && !ctx.KqpCtx .Config ->EnableKqpScanQueryStreamIdxLookupJoin ) {
26+ return false ;
27+ }
28+
29+ if (find_if (joinColumns.begin (), joinColumns.end (), [&] (const TString& s) { return node->Stats ->KeyColumns [0 ] == s;})) {
30+ return true ;
31+ }
32+
33+ auto readMatch = MatchRead<TKqlReadTable>(expr);
34+ TMaybeNode<TKqlKeyInc> maybeTablePrefix;
35+ size_t prefixSize;
36+
37+ if (readMatch) {
38+ if (readMatch->FlatMap && !IsPassthroughFlatMap (readMatch->FlatMap .Cast (), nullptr )){
39+ return false ;
40+ }
41+ auto read = readMatch->Read .Cast <TKqlReadTable>();
42+ maybeTablePrefix = GetRightTableKeyPrefix (read .Range ());
43+
44+ if (!maybeTablePrefix) {
45+ return false ;
46+ }
47+
48+ prefixSize = maybeTablePrefix.Cast ().ArgCount ();
49+
50+ if (!prefixSize) {
51+ return true ;
52+ }
53+ }
54+ else {
55+ readMatch = MatchRead<TKqlReadTableRangesBase>(expr);
56+ if (readMatch) {
57+ if (readMatch->FlatMap && !IsPassthroughFlatMap (readMatch->FlatMap .Cast (), nullptr )){
58+ return false ;
59+ }
60+ auto read = readMatch->Read .Cast <TKqlReadTableRangesBase>();
61+ if (TCoVoid::Match (read .Ranges ().Raw ())) {
62+ return true ;
63+ } else {
64+ auto prompt = TKqpReadTableExplainPrompt::Parse (read );
65+
66+ if (prompt.PointPrefixLen != prompt.UsedKeyColumns .size ()) {
67+ return false ;
68+ }
69+
70+ if (prompt.ExpectedMaxRanges != TMaybe<ui64>(1 )) {
71+ return false ;
72+ }
73+ prefixSize = prompt.PointPrefixLen ;
74+ }
75+ }
76+ }
77+ if (! readMatch) {
78+ return false ;
79+ }
80+
81+ if (prefixSize < node->Stats ->KeyColumns .size () && !(find_if (joinColumns.begin (), joinColumns.end (), [&] (const TString& s) {
82+ return node->Stats ->KeyColumns [prefixSize] == s;
83+ }))){
84+ return false ;
85+ }
86+
87+ return true ;
88+ }
89+
90+ bool IsLookupJoinApplicable (std::shared_ptr<IBaseOptimizerNode> left,
91+ std::shared_ptr<IBaseOptimizerNode> right,
92+ const std::set<std::pair<TJoinColumn, TJoinColumn>>& joinConditions,
93+ TKqpProviderContext& ctx) {
94+
95+ Y_UNUSED (left);
96+
97+ auto rightStats = right->Stats ;
98+
99+ if (rightStats->Type != EStatisticsType::BaseTable) {
100+ return false ;
101+ }
102+ if (joinConditions.size () > rightStats->KeyColumns .size ()) {
103+ return false ;
104+ }
105+
106+ for (auto [leftCol, rightCol] : joinConditions) {
107+ if (! find_if (rightStats->KeyColumns .begin (), rightStats->KeyColumns .end (),
108+ [rightCol] (const TString& s) {
109+ return rightCol.AttributeName == s;
110+ } )) {
111+ return false ;
112+ }
113+ }
114+
115+ TVector<TString> joinKeys;
116+ for ( auto [leftJc, rightJc] : joinConditions ) {
117+ joinKeys.emplace_back ( rightJc.AttributeName );
118+ }
119+
120+ return IsLookupJoinApplicableDetailed (std::static_pointer_cast<TRelOptimizerNode>(right), joinKeys, ctx);
121+ }
122+
123+ }
124+
125+ bool TKqpProviderContext::IsJoinApplicable (const std::shared_ptr<IBaseOptimizerNode>& left,
126+ const std::shared_ptr<IBaseOptimizerNode>& right,
127+ const std::set<std::pair<NDq::TJoinColumn, NDq::TJoinColumn>>& joinConditions,
128+ EJoinImplType joinImpl) {
129+
130+ switch ( joinImpl ) {
131+ case EJoinImplType::LookupJoin:
132+ return IsLookupJoinApplicable (left, right, joinConditions, *this );
133+ default :
134+ return true ;
135+ }
136+ }
137+
138+ double TKqpProviderContext::ComputeJoinCost (const TOptimizerStatistics& leftStats, const TOptimizerStatistics& rightStats, EJoinImplType joinImpl) const {
139+
140+ switch (joinImpl) {
141+ case EJoinImplType::LookupJoin:
142+ return -1 ;
143+ default :
144+ return leftStats.Nrows + 2.0 * rightStats.Nrows ;
145+ }
146+ }
147+
148+
149+ }
0 commit comments