planner: introduce hashEquals for expression.Column/collationInfo/fie…

…ldType (#55691)

ref #51664

pkg/expression/BUILD.bazel

@@ -89,6 +89,7 @@ go_library(
         "//pkg/parser/opcode",
         "//pkg/parser/terror",
         "//pkg/parser/types",
+        "//pkg/planner/cascades/base",
         "//pkg/sessionctx/stmtctx",
         "//pkg/sessionctx/variable",
         "//pkg/types",
@@ -212,6 +213,7 @@ go_test(
         "//pkg/parser/model",
         "//pkg/parser/mysql",
         "//pkg/parser/terror",
+        "//pkg/planner/cascades/base",
         "//pkg/planner/core",
         "//pkg/planner/core/resolve",
         "//pkg/session",
@@ -246,6 +248,7 @@ go_test(
         "@com_github_tikv_client_go_v2//oracle",
         "@com_github_tikv_client_go_v2//tikv",
         "@io_opencensus_go//stats/view",
+        "@org_uber_go_atomic//:atomic",
         "@org_uber_go_goleak//:goleak",
     ],
 )

pkg/expression/collation.go

@@ -20,6 +20,7 @@ import (
 	"github.com/pingcap/tidb/pkg/parser/ast"
 	"github.com/pingcap/tidb/pkg/parser/charset"
 	"github.com/pingcap/tidb/pkg/parser/mysql"
+	"github.com/pingcap/tidb/pkg/planner/cascades/base"
 	"github.com/pingcap/tidb/pkg/types"
 	"github.com/pingcap/tidb/pkg/util/chunk"
 	"github.com/pingcap/tidb/pkg/util/collate"
@@ -28,6 +29,8 @@ import (
 	"go.uber.org/atomic"
 )
 
+var _ base.HashEquals = &collationInfo{}
+
 // ExprCollation is a struct that store the collation related information.
 type ExprCollation struct {
 	Coer      Coercibility
@@ -45,6 +48,37 @@ type collationInfo struct {
 	collation string
 }
 
+// Hash64 implements the base.Hasher.<0th> interface.
+func (c *collationInfo) Hash64(h base.Hasher) {
+	h.HashInt64(int64(c.coer))
+	h.HashBool(c.coerInit.Load())
+	h.HashInt(int(c.repertoire))
+	h.HashString(c.charset)
+	h.HashString(c.collation)
+}
+
+// Equals implements the base.Hasher.<1th> interface.
+func (c *collationInfo) Equals(other any) bool {
+	// the caller should care about c is nil or not.
+	if other == nil {
+		return false
+	}
+	var c2 *collationInfo
+	switch x := other.(type) {
+	case *collationInfo:
+		c2 = x
+	case collationInfo:
+		c2 = &x
+	default:
+		return false
+	}
+	return c.coer == c2.coer &&
+		c.coerInit.Load() == c2.coerInit.Load() &&
+		c.repertoire == c2.repertoire &&
+		c.charset == c2.charset &&
+		c.collation == c2.collation
+}
+
 func (c *collationInfo) HasCoercibility() bool {
 	return c.coerInit.Load()
 }

pkg/expression/collation_test.go

@@ -20,10 +20,12 @@ import (
 	"github.com/pingcap/tidb/pkg/parser/ast"
 	"github.com/pingcap/tidb/pkg/parser/charset"
 	"github.com/pingcap/tidb/pkg/parser/mysql"
+	"github.com/pingcap/tidb/pkg/planner/cascades/base"
 	"github.com/pingcap/tidb/pkg/types"
 	"github.com/pingcap/tidb/pkg/util/chunk"
 	"github.com/pingcap/tidb/pkg/util/mock"
 	"github.com/stretchr/testify/require"
+	"go.uber.org/atomic"
 )
 
 func newExpression(coercibility Coercibility, repertoire Repertoire, chs, coll string) Expression {
@@ -33,6 +35,71 @@ func newExpression(coercibility Coercibility, repertoire Repertoire, chs, coll s
 	return constant
 }
 
+func TestCollationHashEquals(t *testing.T) {
+	c1 := collationInfo{
+		coer:       1,
+		coerInit:   atomic.Bool{},
+		repertoire: 1,
+		charset:    "aa",
+		collation:  "bb",
+	}
+	c2 := collationInfo{
+		coer:       1,
+		coerInit:   atomic.Bool{},
+		repertoire: 1,
+		charset:    "aabb",
+		collation:  "",
+	}
+	hasher1 := base.NewHashEqualer()
+	hasher2 := base.NewHashEqualer()
+	c1.Hash64(hasher1)
+	c2.Hash64(hasher2)
+	require.NotEqual(t, hasher1.Sum64(), hasher2.Sum64())
+	require.False(t, c1.Equals(c2))
+
+	c2.charset = "aa"
+	c2.collation = "bb"
+	hasher2.Reset()
+	c2.Hash64(hasher2)
+	require.Equal(t, hasher1.Sum64(), hasher2.Sum64())
+	require.True(t, c1.Equals(c2))
+
+	c2.coer = 2
+	hasher2.Reset()
+	c2.Hash64(hasher2)
+	require.NotEqual(t, hasher1.Sum64(), hasher2.Sum64())
+	require.False(t, c1.Equals(c2))
+
+	c2.coer = 1
+	c2.coerInit.Store(true)
+	hasher2.Reset()
+	c2.Hash64(hasher2)
+	require.NotEqual(t, hasher1.Sum64(), hasher2.Sum64())
+	require.False(t, c1.Equals(c2))
+
+	c2.coerInit.Store(false)
+	c2.repertoire = 2
+	hasher2.Reset()
+	c2.Hash64(hasher2)
+	require.NotEqual(t, hasher1.Sum64(), hasher2.Sum64())
+	require.False(t, c1.Equals(c2))
+
+	c2.repertoire = 1
+	c2.charset = ""
+	c2.collation = "aabb"
+	hasher2.Reset()
+	c2.Hash64(hasher2)
+	require.NotEqual(t, hasher1.Sum64(), hasher2.Sum64())
+	require.False(t, c1.Equals(c2))
+
+	c2.charset = "aa"
+	c2.collation = "bb"
+	hasher2.Reset()
+	c2.Hash64(hasher2)
+	require.Equal(t, hasher1.Sum64(), hasher2.Sum64())
+	require.True(t, c1.Equals(c2))
+}
+
 func TestInferCollation(t *testing.T) {
 	tests := []struct {
 		exprs []Expression

pkg/expression/column.go

@@ -27,12 +27,17 @@ import (
 	"github.com/pingcap/tidb/pkg/parser/charset"
 	"github.com/pingcap/tidb/pkg/parser/model"
 	"github.com/pingcap/tidb/pkg/parser/mysql"
+	"github.com/pingcap/tidb/pkg/planner/cascades/base"
 	"github.com/pingcap/tidb/pkg/types"
 	"github.com/pingcap/tidb/pkg/util/chunk"
 	"github.com/pingcap/tidb/pkg/util/codec"
 	"github.com/pingcap/tidb/pkg/util/size"
 )
 
+var (
+	_ base.HashEquals = &Column{}
+)
+
 // CorrelatedColumn stands for a column in a correlated sub query.
 type CorrelatedColumn struct {
 	Column
@@ -442,6 +447,61 @@ func (col *Column) GetStaticType() *types.FieldType {
 	return col.RetType
 }
 
+// Hash64 implements HashEquals.<0th> interface.
+func (col *Column) Hash64(h base.Hasher) {
+	if col.RetType == nil {
+		h.HashByte(base.NilFlag)
+	} else {
+		h.HashByte(base.NotNilFlag)
+		col.RetType.Hash64(h)
+	}
+	h.HashInt64(col.ID)
+	h.HashInt64(col.UniqueID)
+	h.HashInt(col.Index)
+	if col.VirtualExpr != nil {
+		h.HashByte(base.NilFlag)
+	} else {
+		h.HashByte(base.NotNilFlag)
+		//col.VirtualExpr.Hash64(h)
+	}
+	h.HashString(col.OrigName)
+	h.HashBool(col.IsHidden)
+	h.HashBool(col.IsPrefix)
+	h.HashBool(col.InOperand)
+	col.collationInfo.Hash64(h)
+	h.HashInt64(col.CorrelatedColUniqueID)
+}
+
+// Equals implements HashEquals.<1st> interface.
+func (col *Column) Equals(other any) bool {
+	if other == nil {
+		return false
+	}
+	var col2 *Column
+	switch x := other.(type) {
+	case Column:
+		col2 = &x
+	case *Column:
+		col2 = x
+	default:
+		return false
+	}
+	// when step into here, we could ensure that col1.RetType and col2.RetType are same type.
+	// and we should ensure col1.RetType and col2.RetType is not nil ourselves.
+	ftEqual := col.RetType == nil && col2.RetType == nil || col.RetType != nil && col2.RetType != nil && col.RetType.Equal(col2.RetType)
+	return ftEqual &&
+		col.ID == col2.ID &&
+		col.UniqueID == col2.UniqueID &&
+		col.Index == col2.Index &&
+		//col.VirtualExpr.Equals(col2.VirtualExpr) &&
+		col.OrigName == col2.OrigName &&
+		col.IsHidden == col2.IsHidden &&
+		col.IsPrefix == col2.IsPrefix &&
+		col.InOperand == col2.InOperand &&
+		col.collationInfo.Equals(&col2.collationInfo) &&
+		col.CorrelatedColUniqueID == col2.CorrelatedColUniqueID
+}
+
 // Traverse implements the TraverseDown interface.
 func (col *Column) Traverse(action TraverseAction) Expression {
 	return action.Transform(col)