copr: bit column support push down to TiKV

expression/builtin_cast.go

@@ -1955,9 +1955,16 @@ func WrapWithCastAsString(ctx sessionctx.Context, expr Expression) Expression {
 	if exprTp.Tp == mysql.TypeNewDecimal && argLen != types.UnspecifiedFsp {
 		argLen += 3
 	}
+
 	if exprTp.EvalType() == types.ETInt {
 		argLen = mysql.MaxIntWidth
+		// For TypeBit, castAsString will make length as int(( bit_len + 7 ) / 8) bytes due to
+		// TiKV needs the bit's real len during calculating, eg: ascii(bit).
+		if exprTp.Tp == mysql.TypeBit {
+			argLen = (exprTp.Flen + 7) / 8
+		}
 	}
+
 	// Because we can't control the length of cast(float as char) for now, we can't determine the argLen.
 	if exprTp.Tp == mysql.TypeFloat || exprTp.Tp == mysql.TypeDouble {
 		argLen = -1

expression/constant.go

@@ -227,6 +227,9 @@ func (c *Constant) EvalInt(ctx sessionctx.Context, row chunk.Row) (int64, bool,
 	} else if c.GetType().Hybrid() || dt.Kind() == types.KindString {
 		res, err := dt.ToInt64(ctx.GetSessionVars().StmtCtx)
 		return res, false, err
+	} else if dt.Kind() == types.KindMysqlBit {
+		uintVal, err := dt.GetBinaryLiteral().ToInt(ctx.GetSessionVars().StmtCtx)
+		return int64(uintVal), false, err
 	}
 	return dt.GetInt64(), false, nil
 }

expression/distsql_builtin.go

@@ -1124,7 +1124,7 @@ func PBToExpr(expr *tipb.Expr, tps []*types.FieldType, sc *stmtctx.StatementCont
 		return convertUint(expr.Val)
 	case tipb.ExprType_String:
 		return convertString(expr.Val, expr.FieldType)
-	case tipb.ExprType_Bytes:
+	case tipb.ExprType_Bytes, tipb.ExprType_MysqlBit:
 		return &Constant{Value: types.NewBytesDatum(expr.Val), RetType: types.NewFieldType(mysql.TypeString)}, nil
 	case tipb.ExprType_Float32:
 		return convertFloat(expr.Val, true)

expression/expr_to_pb.go

@@ -108,6 +108,9 @@ func (pc *PbConverter) encodeDatum(ft *types.FieldType, d types.Datum) (tipb.Exp
 	case types.KindString, types.KindBinaryLiteral:
 		tp = tipb.ExprType_String
 		val = d.GetBytes()
+	case types.KindMysqlBit:
+		tp = tipb.ExprType_MysqlBit
+		val = d.GetBytes()
 	case types.KindBytes:
 		tp = tipb.ExprType_Bytes
 		val = d.GetBytes()
@@ -179,7 +182,7 @@ func (pc PbConverter) columnToPBExpr(column *Column) *tipb.Expr {
 		return nil
 	}
 	switch column.GetType().Tp {
-	case mysql.TypeBit, mysql.TypeSet, mysql.TypeGeometry, mysql.TypeUnspecified:
+	case mysql.TypeSet, mysql.TypeGeometry, mysql.TypeUnspecified:
 		return nil
 	case mysql.TypeEnum:
 		if !IsPushDownEnabled("enum", kv.UnSpecified) {

expression/expr_to_pb_test.go

@@ -130,10 +130,9 @@ func TestColumn2Pb(t *testing.T) {
 	sc := new(stmtctx.StatementContext)
 	client := new(mock.Client)
 
-	colExprs = append(colExprs, genColumn(mysql.TypeBit, 1))
-	colExprs = append(colExprs, genColumn(mysql.TypeSet, 2))
-	colExprs = append(colExprs, genColumn(mysql.TypeGeometry, 4))
-	colExprs = append(colExprs, genColumn(mysql.TypeUnspecified, 5))
+	colExprs = append(colExprs, genColumn(mysql.TypeSet, 1))
+	colExprs = append(colExprs, genColumn(mysql.TypeGeometry, 2))
+	colExprs = append(colExprs, genColumn(mysql.TypeUnspecified, 3))
 
 	pushed, remained := PushDownExprs(sc, colExprs, client, kv.UnSpecified)
 	require.Len(t, pushed, 0)
@@ -168,6 +167,7 @@ func TestColumn2Pb(t *testing.T) {
 	colExprs = append(colExprs, genColumn(mysql.TypeVarString, 22))
 	colExprs = append(colExprs, genColumn(mysql.TypeString, 23))
 	colExprs = append(colExprs, genColumn(mysql.TypeEnum, 24))
+	colExprs = append(colExprs, genColumn(mysql.TypeBit, 25))
 	pushed, remained = PushDownExprs(sc, colExprs, client, kv.UnSpecified)
 	require.Len(t, pushed, len(colExprs))
 	require.Len(t, remained, 0)
@@ -198,6 +198,7 @@ func TestColumn2Pb(t *testing.T) {
 		"{\"tp\":201,\"val\":\"gAAAAAAAABY=\",\"sig\":0,\"field_type\":{\"tp\":253,\"flag\":0,\"flen\":-1,\"decimal\":-1,\"collate\":-46,\"charset\":\"utf8mb4\"},\"has_distinct\":false}",
 		"{\"tp\":201,\"val\":\"gAAAAAAAABc=\",\"sig\":0,\"field_type\":{\"tp\":254,\"flag\":0,\"flen\":-1,\"decimal\":-1,\"collate\":-46,\"charset\":\"utf8mb4\"},\"has_distinct\":false}",
 		"{\"tp\":201,\"val\":\"gAAAAAAAABg=\",\"sig\":0,\"field_type\":{\"tp\":247,\"flag\":0,\"flen\":-1,\"decimal\":-1,\"collate\":-63,\"charset\":\"binary\"},\"has_distinct\":false}",
+		"{\"tp\":201,\"val\":\"gAAAAAAAABk=\",\"sig\":0,\"field_type\":{\"tp\":16,\"flag\":0,\"flen\":-1,\"decimal\":-1,\"collate\":-63,\"charset\":\"binary\"},\"has_distinct\":false}",
 	}
 	for i, pbExpr := range pbExprs {
 		require.NotNil(t, pbExprs)

expression/integration_test.go

@@ -7043,3 +7043,31 @@ func TestIssue22206(t *testing.T) {
 	unixTime = time.Unix(5000000000, 0).In(tz).String()[:19]
 	result.Check(testkit.Rows(unixTime))
 }
+
+func TestIssue30738(t *testing.T) {
+	store, clean := testkit.CreateMockStore(t)
+	defer clean()
+	tk := testkit.NewTestKit(t, store)
+	tk.MustExec("use test")
+	tk.MustExec("drop table if exists t1;")
+	tk.MustExec("CREATE TABLE t1 (a int, b bit(8), c bit(1), d bit(24));")
+	tk.MustExec("insert into t1 values(1, 65, 1, 65);") //A
+	tk.MustExec("insert into t1 values(2, 53, 0, 53);") //5
+	tk.MustExec("insert into t1 values(3, 54, 1, 54);") //6
+
+	tk.MustQuery("select a from t1 where b+1=54;").Check(testkit.Rows("2"))
+	tk.MustQuery("select a from t1 where c+1=2;").Check(testkit.Rows("1", "3"))
+	tk.MustQuery("select a from t1 where d-1=53;").Check(testkit.Rows("3"))
+
+	tk.MustQuery("select a from t1 where ascii(b)=65;").Check(testkit.Rows("1"))
+	tk.MustQuery("select a from t1 where ascii(c)=0;").Check(testkit.Rows("2"))
+	tk.MustQuery("select a from t1 where ascii(d)=54;").Check(testkit.Rows())
+
+	tk.MustQuery("select a from t1 where cast(b as char)='A';").Check(testkit.Rows("1"))
+	tk.MustQuery("select a from t1 where cast(c as char)='\\1';").Check(testkit.Rows())
+	tk.MustQuery("select a from t1 where cast(d as char)='\\0\\06';").Check(testkit.Rows("3"))
+
+	tk.MustQuery("select a from t1 where concat(b, 'kji')='Akji';").Check(testkit.Rows("1"))
+	tk.MustQuery("select a from t1 where concat(c, 'kji')='\\0kji';").Check(testkit.Rows("2"))
+	tk.MustQuery("select a from t1 where concat(d, 'kji')='\\0\\06kji';").Check(testkit.Rows("3"))
+}

kv/checker.go

@@ -41,7 +41,7 @@ func (d RequestTypeSupportedChecker) supportExpr(exprType tipb.ExprType) bool {
 	switch exprType {
 	case tipb.ExprType_Null, tipb.ExprType_Int64, tipb.ExprType_Uint64, tipb.ExprType_String, tipb.ExprType_Bytes,
 		tipb.ExprType_MysqlDuration, tipb.ExprType_MysqlTime, tipb.ExprType_MysqlDecimal,
-		tipb.ExprType_Float32, tipb.ExprType_Float64, tipb.ExprType_ColumnRef, tipb.ExprType_MysqlEnum:
+		tipb.ExprType_Float32, tipb.ExprType_Float64, tipb.ExprType_ColumnRef, tipb.ExprType_MysqlEnum, tipb.ExprType_MysqlBit:
 		return true
 	// aggregate functions.
 	// NOTE: tipb.ExprType_GroupConcat is only supported by TiFlash, So checking it for TiKV case outside.

planner/core/expression_rewriter.go

@@ -1132,6 +1132,11 @@ func (er *expressionRewriter) Leave(originInNode ast.Node) (retNode ast.Node, ok
 			} else {
 				castFunction.SetRepertoire(expression.UNICODE)
 			}
+
+			//See detailed comment in `expression/builtin_cast.go:WrapWithCastAsString()`; eg: "where CAST(bit as CAHR)"
+			if arg.GetType().Tp == mysql.TypeBit {
+				v.Tp.Flen = (arg.GetType().Flen + 7) / 8
+			}
 		} else {
 			castFunction.SetCoercibility(expression.CoercibilityNumeric)
 			castFunction.SetRepertoire(expression.ASCII)

planner/core/integration_test.go

@@ -2521,8 +2521,8 @@ func (s *testIntegrationSerialSuite) TestIssue18984(c *C) {
 func (s *testIntegrationSuite) TestScalarFunctionPushDown(c *C) {
 	tk := testkit.NewTestKit(c, s.store)
 	tk.MustExec("use test")
-	tk.MustExec("create table t(id int signed, id2 int unsigned ,c varchar(11), d datetime, b double)")
-	tk.MustExec("insert into t(id,c,d) values (1,'abc','2021-12-12')")
+	tk.MustExec("create table t(id int signed, id2 int unsigned, c varchar(11), d datetime, b double, bit_col bit(1))")
+	tk.MustExec("insert into t(id,c,d,bit_col) values (1, 'abc','2021-12-12', 1)")
 	rows := [][]interface{}{
 		{"TableReader_7", "root", "data:Selection_6"},
 		{"└─Selection_6", "cop[tikv]", "right(test.t.c, 1)"},
@@ -2576,6 +2576,10 @@ func (s *testIntegrationSuite) TestScalarFunctionPushDown(c *C) {
 	tk.MustQuery("explain analyze select /*+read_from_storage(tikv[t])*/ * from t where atan2(id,id);").
 		CheckAt([]int{0, 3, 6}, rows)
 
+	rows[1][2] = "ascii(cast(test.t.bit_col, var_string(1)))"
+	tk.MustQuery("explain analyze select /*+read_from_storage(tikv[t])*/ * from t where ascii(bit_col);").
+		CheckAt([]int{0, 3, 6}, rows)
+
 	rows[1][2] = "hour(cast(test.t.d, time))"
 	tk.MustQuery("explain analyze select /*+read_from_storage(tikv[t])*/ * from t where hour(d);").
 		CheckAt([]int{0, 3, 6}, rows)

planner/core/testdata/point_get_plan_out.json

@@ -191,9 +191,7 @@
       {
         "SQL": "select * from t5 where id in ('0')",
         "Plan": [
-          "Selection 8000.00 root  eq(test.t5.id, 0)",
-          "└─TableReader 10000.00 root  data:TableFullScan",
-          "  └─TableFullScan 10000.00 cop[tikv] table:t5 keep order:false, stats:pseudo"
+          "Point_Get 1.00 root table:t5, clustered index:PRIMARY(id) "
         ],
         "Res": null
       }

util/ranger/testdata/ranger_suite_out.json

@@ -408,9 +408,9 @@
       {
         "SQL": "select * from t where a = 0;",
         "Plan": [
-          "Selection_7 0.80 root  eq(test.t.a, 0)",
-          "└─TableReader_9 1.00 root  data:TableFullScan_8",
-          "  └─TableFullScan_8 1.00 cop[tikv] table:t, partition:p0 keep order:false"
+          "TableReader_9 1.00 root  data:Selection_8",
+          "└─Selection_8 1.00 cop[tikv]  eq(test.t.a, 0)",
+          "  └─TableFullScan_7 1.00 cop[tikv] table:t, partition:p0 keep order:false"
         ],
         "Result": [
           "\u0000 0"