fix

englefly · Nov 30, 2023 · d9ff0fc · d9ff0fc
1 parent a2aa64b
commit d9ff0fc
Showing 1 changed file with 4 additions and 32 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
@@ -45,7 +45,7 @@
  */
 public class JoinEstimation {
     private static double DEFAULT_ANTI_JOIN_SELECTIVITY_COEFFICIENT = 0.3;
-
+    private static double UNKNOWN_COL_STATS_FILTER_SEL_LOWER_BOUND = 0.5;
     private static EqualPredicate normalizeHashJoinCondition(EqualPredicate equal, Statistics leftStats,
             Statistics rightStats) {
         boolean changeOrder = equal.left().getInputSlots().stream()
@@ -170,33 +170,7 @@ private static Statistics estimateNestLoopJoin(Statistics leftStats, Statistics
                 .build();
     }
 
-    private static double computeSelectivityForInnerJoinWhenColStatsUnknown(
-            Statistics leftStats, Statistics rightStats, Join join) {
-        double sel = 1.0;
-        for (Expression cond : join.getHashJoinConjuncts()) {
-            if (cond instanceof EqualTo) {
-                EqualTo equal = (EqualTo) cond;
-                if (equal.left() instanceof Slot && equal.right() instanceof Slot) {
-                    ColumnStatistic leftColStats = leftStats.findColumnStatistics(equal.left());
-                    ColumnStatistic rightColStats = rightStats.findColumnStatistics(equal.right());
-                    if (leftColStats == null) {
-                        leftColStats = leftStats.findColumnStatistics(equal.right());
-                        rightColStats = rightStats.findColumnStatistics(equal.left());
-                    }
-                    if (leftColStats != null && rightColStats != null) {
-                        double leftSel = Math.min(leftColStats.count / leftStats.getRowCount(), 1.0);
-                        leftSel = Math.max(leftSel, 0.5);
-                        double rightSel = Math.min(rightColStats.count / rightStats.getRowCount(), 1.0);
-                        rightSel = Math.max(rightSel, 0.5);
-                        sel = Math.min(sel, Math.min(leftSel, rightSel));
-                    }
-                }
-            }
-        }
-        return sel;
-    }
-
-    private static double computeSelectivityForSemiOrAntiJoinWhenColStatsUnknown(Statistics buildStats, Join join) {
+    private static double computeSelectivityForBuildSideWhenColStatsUnknown(Statistics buildStats, Join join) {
         double sel = 1.0;
         for (Expression cond : join.getHashJoinConjuncts()) {
             if (cond instanceof EqualTo) {
@@ -208,7 +182,7 @@ private static double computeSelectivityForSemiOrAntiJoinWhenColStatsUnknown(Sta
                     }
                     if (buildColStats != null) {
                         double buildSel = Math.min(buildStats.getRowCount() / buildColStats.count, 1.0);
-                        buildSel = Math.max(buildSel, 0.5);
+                        buildSel = Math.max(buildSel, UNKNOWN_COL_STATS_FILTER_SEL_LOWER_BOUND);
                         sel = Math.min(sel, buildSel);
                     }
                 }
@@ -220,8 +194,6 @@ private static double computeSelectivityForSemiOrAntiJoinWhenColStatsUnknown(Sta
     private static Statistics estimateInnerJoin(Statistics leftStats, Statistics rightStats, Join join) {
         if (hashJoinConditionContainsUnknownColumnStats(leftStats, rightStats, join)) {
             double rowCount = Math.max(leftStats.getRowCount(), rightStats.getRowCount());
-            double sel = computeSelectivityForInnerJoinWhenColStatsUnknown(leftStats, rightStats, join);
-            rowCount = rowCount * sel;
             rowCount = Math.max(1, rowCount);
             return new StatisticsBuilder()
                 .setRowCount(rowCount)
@@ -295,7 +267,7 @@ private static double estimateSemiOrAntiRowCountBySlotsEqual(Statistics leftStat
 
     private static Statistics estimateSemiOrAnti(Statistics leftStats, Statistics rightStats, Join join) {
         if (hashJoinConditionContainsUnknownColumnStats(leftStats, rightStats, join)) {
-            double sel = computeSelectivityForSemiOrAntiJoinWhenColStatsUnknown(rightStats, join);
+            double sel = computeSelectivityForBuildSideWhenColStatsUnknown(rightStats, join);
             if (join.getJoinType().isLeftSemiOrAntiJoin()) {
                 return new StatisticsBuilder().setRowCount(leftStats.getRowCount() * sel)
                         .putColumnStatistics(leftStats.columnStatistics())