stats: fix selectivity estimation for primary key (#8134) (#8150)

pingcap · Nov 2, 2018 · e04bba1 · e04bba1
1 parent a168d43
commit e04bba1
Show file tree

Hide file tree

Showing 7 changed files with 29 additions and 8 deletions.
diff --git a/statistics/boostrap.go b/statistics/boostrap.go
@@ -111,7 +111,7 @@ func initStatsHistograms4Chunk(is infoschema.InfoSchema, tables statsCache, iter
 				continue
 			}
 			hist := NewHistogram(id, ndv, nullCount, version, &colInfo.FieldType, 0, totColSize)
-			table.Columns[hist.ID] = &Column{Histogram: *hist, Info: colInfo, Count: nullCount}
+			table.Columns[hist.ID] = &Column{Histogram: *hist, Info: colInfo, Count: nullCount, isHandle: tbl.Meta().PKIsHandle && mysql.HasPriKeyFlag(colInfo.Flag)}
 		}
 	}
 }

diff --git a/statistics/dump.go b/statistics/dump.go
@@ -126,6 +126,7 @@ func (h *Handle) LoadStatsFromJSON(tableInfo *model.TableInfo, jsonTbl *JSONTabl
 				CMSketch:  CMSketchFromProto(jsonCol.CMSketch),
 				Info:      colInfo,
 				Count:     count,
+				isHandle:  tableInfo.PKIsHandle && mysql.HasPriKeyFlag(colInfo.Flag),
 			}
 			tbl.Columns[col.ID] = col
 		}

diff --git a/statistics/handle.go b/statistics/handle.go
@@ -200,7 +200,7 @@ func (h *Handle) LoadNeededHistograms() error {
 		if err != nil {
 			return errors.Trace(err)
 		}
-		tbl.Columns[c.ID] = &Column{Histogram: *hg, Info: c.Info, CMSketch: cms, Count: int64(hg.totalRowCount())}
+		tbl.Columns[c.ID] = &Column{Histogram: *hg, Info: c.Info, CMSketch: cms, Count: int64(hg.totalRowCount()), isHandle: c.isHandle}
 		h.UpdateTableStats([]*Table{tbl}, nil)
 		histogramNeededColumns.delete(col)
 	}

diff --git a/statistics/histogram.go b/statistics/histogram.go
@@ -628,8 +628,9 @@ func (hg *Histogram) outOfRange(val types.Datum) bool {
 type Column struct {
 	Histogram
 	*CMSketch
-	Count int64
-	Info  *model.ColumnInfo
+	Count    int64
+	Info     *model.ColumnInfo
+	isHandle bool
 }
 
 func (c *Column) String() string {

diff --git a/statistics/selectivity.go b/statistics/selectivity.go
@@ -149,7 +149,7 @@ func (t *Table) Selectivity(ctx sessionctx.Context, exprs []expression.Expressio
 				return 0, errors.Trace(err)
 			}
 			sets = append(sets, &exprSet{tp: colType, ID: col.ID, mask: maskCovered, ranges: ranges})
-			if mysql.HasPriKeyFlag(colInfo.Info.Flag) {
+			if colInfo.isHandle {
 				sets[len(sets)-1].tp = pkType
 			}
 		}

diff --git a/statistics/selectivity_test.go b/statistics/selectivity_test.go
@@ -281,6 +281,23 @@ func (s *testSelectivitySuite) TestEstimationForUnknownValues(c *C) {
 	c.Assert(count, Equals, 0.0)
 }
 
+func (s *testSelectivitySuite) TestPrimaryKeySelectivity(c *C) {
+	testKit := testkit.NewTestKit(c, s.store)
+	testKit.MustExec("use test")
+	testKit.MustExec("drop table if exists t")
+	testKit.MustExec("create table t(a char(10) primary key, b int)")
+	testKit.MustQuery(`explain select * from t where a > "t"`).Check(testkit.Rows(
+		"IndexScan_8   cop table:t, index:a, range:(t,+inf], keep order:false 3333.33",
+		"TableScan_9   cop table:t, keep order:false 3333.33",
+		"IndexLookUp_10   root index:IndexScan_8, table:TableScan_9 3333.33"))
+
+	testKit.MustExec("drop table t")
+	testKit.MustExec("create table t(a int primary key, b int)")
+	testKit.MustQuery(`explain select * from t where a > 1`).Check(testkit.Rows(
+		"TableScan_5   cop table:t, range:(1,+inf], keep order:false 3333.33",
+		"TableReader_6   root data:TableScan_5 3333.33"))
+}
+
 func BenchmarkSelectivity(b *testing.B) {
 	c := &C{}
 	s := &testSelectivitySuite{}

diff --git a/statistics/table.go b/statistics/table.go
@@ -148,8 +148,9 @@ func (h *Handle) columnStatsFromStorage(row types.Row, table *Table, tableInfo *
 					LastUpdateVersion: histVer,
 					TotColSize:        totColSize,
 				},
-				Info:  colInfo,
-				Count: count + nullCount,
+				Info:     colInfo,
+				Count:    count + nullCount,
+				isHandle: tableInfo.PKIsHandle && mysql.HasPriKeyFlag(colInfo.Flag),
 			}
 			break
 		}
@@ -167,6 +168,7 @@ func (h *Handle) columnStatsFromStorage(row types.Row, table *Table, tableInfo *
 				Info:      colInfo,
 				CMSketch:  cms,
 				Count:     int64(hg.totalRowCount()),
+				isHandle:  tableInfo.PKIsHandle && mysql.HasPriKeyFlag(colInfo.Flag),
 			}
 		}
 		break
@@ -372,7 +374,7 @@ func PseudoTable(tblInfo *model.TableInfo) *Table {
 	}
 	for _, col := range tblInfo.Columns {
 		if col.State == model.StatePublic {
-			t.Columns[col.ID] = &Column{Info: col}
+			t.Columns[col.ID] = &Column{Info: col, isHandle: tblInfo.PKIsHandle && mysql.HasPriKeyFlag(col.Flag)}
 		}
 	}
 	for _, idx := range tblInfo.Indices {