From e4d4e91b45763963980ab3f972758024f9830dbf Mon Sep 17 00:00:00 2001
From: "zhongjian.xzj" <zhongjian.xzj@zhongjianxzjdeMacBook-Pro.local>
Date: Tue, 24 Sep 2024 12:02:38 +0800
Subject: [PATCH] [opt](nereids) refine operator estimation

---
 .../org/apache/doris/nereids/stats/JoinEstimation.java     | 7 +++----
 .../main/java/org/apache/doris/statistics/Statistics.java  | 6 +++---
 2 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/fe/fe-core/src/main/java/org/apache/doris/nereids/stats/JoinEstimation.java b/fe/fe-core/src/main/java/org/apache/doris/nereids/stats/JoinEstimation.java
index e6e43700b656da3..99080566e848ca2 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/nereids/stats/JoinEstimation.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/nereids/stats/JoinEstimation.java
@@ -289,7 +289,6 @@ private static Statistics estimateSemiOrAnti(Statistics leftStats, Statistics ri
                         .putColumnStatistics(rightStats.columnStatistics())
                         .build();
             }
-            updateJoinConditionColumnStatistics(result, join);
             result.normalizeColumnStatistics();
             return result;
         }
@@ -306,7 +305,6 @@ private static Statistics estimateSemiOrAnti(Statistics leftStats, Statistics ri
             double baseRowCount =
                     join.getJoinType().isLeftSemiOrAntiJoin() ? leftStats.getRowCount() : rightStats.getRowCount();
             rowCount = Math.min(innerJoinStats.getRowCount(), baseRowCount);
-            updateJoinConditionColumnStatistics(innerJoinStats, join);
             return innerJoinStats.withRowCountAndEnforceValid(rowCount);
         } else {
             /*double crossRowCount = Math.max(1, leftStats.getRowCount()) * Math.max(1, rightStats.getRowCount());
@@ -336,7 +334,6 @@ private static Statistics estimateSemiOrAnti(Statistics leftStats, Statistics ri
             }
             builder.setRowCount(outputRowCount);
             Statistics outputStats = builder.build();
-            updateJoinConditionColumnStatistics(outputStats, join);
             outputStats.normalizeColumnStatistics();
             return outputStats;*/
             StatisticsBuilder builder;
@@ -366,7 +363,9 @@ public static Statistics estimate(Statistics leftStats, Statistics rightStats, J
                 .build();
         Statistics innerJoinStats = estimateInnerJoin(leftStats, rightStats, join);
         if (joinType.isSemiOrAntiJoin()) {
-            return estimateSemiOrAnti(leftStats, rightStats, innerJoinStats, join);
+            Statistics outputStats = estimateSemiOrAnti(leftStats, rightStats, innerJoinStats, join);
+            updateJoinConditionColumnStatistics(outputStats, join);
+            return outputStats;
         } else if (joinType == JoinType.INNER_JOIN) {
             updateJoinConditionColumnStatistics(innerJoinStats, join);
             return innerJoinStats;
diff --git a/fe/fe-core/src/main/java/org/apache/doris/statistics/Statistics.java b/fe/fe-core/src/main/java/org/apache/doris/statistics/Statistics.java
index 23b4a8e935295d3..4236993977aaa10 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/statistics/Statistics.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/statistics/Statistics.java
@@ -119,10 +119,10 @@ public void normalizeColumnStatistics(double inputRowCount, boolean isNumNullsDe
             // the following columnStatistic.isUnKnown() judgment is loop inside since current doris
             // supports partial stats deriving, i.e, allowing part of tables have stats and other parts don't,
             // or part of columns have stats but other parts don't, especially join and filter estimation.
-            if (!checkColumnStatsValid(columnStatistic, inputRowCount) && !columnStatistic.isUnKnown()) {
+            if (!checkColumnStatsValid(columnStatistic, rowCount) && !columnStatistic.isUnKnown()) {
                 ColumnStatisticBuilder columnStatisticBuilder = new ColumnStatisticBuilder(columnStatistic);
-                double ndv = Math.min(columnStatistic.ndv, inputRowCount);
-                double numNulls = Math.min(columnStatistic.numNulls * factor, inputRowCount - ndv);
+                double ndv = Math.min(columnStatistic.ndv, rowCount);
+                double numNulls = Math.min(columnStatistic.numNulls * factor, rowCount - ndv);
                 columnStatisticBuilder.setNumNulls(numNulls);
                 columnStatisticBuilder.setNdv(Math.min(ndv, rowCount - numNulls));
                 columnStatistic = columnStatisticBuilder.build();