Fix Bytes stats - adjust according to number of columns used (S3) (#6910)

Hor911 · web-flow · commit fd2d747364fd · 2024-07-22T11:35:12.000+03:00
diff --git a/ydb/core/kqp/opt/kqp_statistics_transformer.cpp b/ydb/core/kqp/opt/kqp_statistics_transformer.cpp
@@ -316,6 +316,10 @@ void InferStatisticsForDqSourceWrap(const TExprNode::TPtr& input, TTypeAnnotatio
                         }
                     }
                     if (stats->ByteSize != 0.0) {
+                        if (stats->Ncols) {
+                            auto n = wrapBase.Cast().RowType().Ref().GetTypeAnn()->Cast<TTypeExprType>()->GetType()->Cast<TStructExprType>()->GetSize();
+                            stats->ByteSize = stats->ByteSize * n / stats->Ncols;
+                        }
                         YQL_CLOG(TRACE, CoreDq) << "Infer statistics for s3 data source " << path;
                         typeCtx->SetStats(input.Get(), stats);
                         typeCtx->SetStats(s3DataSource.Raw(), stats);

Original file line number	Diff line number	Diff line change
`@@ -316,6 +316,10 @@ void InferStatisticsForDqSourceWrap(const TExprNode::TPtr& input, TTypeAnnotatio`
`316`	`316`	`}`
`317`	`317`	`}`
`318`	`318`	`if (stats->ByteSize != 0.0) {`
	`319`	`+ if (stats->Ncols) {`
	`320`	`+ auto n = wrapBase.Cast().RowType().Ref().GetTypeAnn()->Cast<TTypeExprType>()->GetType()->Cast<TStructExprType>()->GetSize();`
	`321`	`+ stats->ByteSize = stats->ByteSize * n / stats->Ncols;`
	`322`	`+ }`
`319`	`323`	`YQL_CLOG(TRACE, CoreDq) << "Infer statistics for s3 data source " << path;`
`320`	`324`	`typeCtx->SetStats(input.Get(), stats);`
`321`	`325`	`typeCtx->SetStats(s3DataSource.Raw(), stats);`