evalengine: Add UUID functions (#13097)

This adds `BIN_TO_UUID`, `IS_UUID`, `UUID` & `UUID_TO_BIN`. This doesn't
implement `UUID_SHORT` though since that has specific requirements to
use the `server_id` which is not something that conceptually makes sense
for `vtgate` since it works across different backend MySQL servers.

So we avoid adding that here since there's no clear path for how to
implement that.

Signed-off-by: Dirkjan Bussink <d.bussink@gmail.com>

go/vt/vtgate/evalengine/compiler_asm.go

@@ -33,6 +33,8 @@ import (
 	"strconv"
 	"time"
 
+	"github.com/google/uuid"
+
 	"vitess.io/vitess/go/hack"
 	"vitess.io/vitess/go/mysql/collations"
 	"vitess.io/vitess/go/mysql/collations/charset"
@@ -4122,3 +4124,119 @@ func (asm *assembler) Fn_CONCAT_WS(tt querypb.Type, tc collations.TypedCollation
 		return 1
 	}, "FN CONCAT_WS VARCHAR(SP-1) VARCHAR(SP-2)...VARCHAR(SP-N)")
 }
+
+func (asm *assembler) Fn_BIN_TO_UUID0(col collations.TypedCollation) {
+	asm.emit(func(env *ExpressionEnv) int {
+		arg := env.vm.stack[env.vm.sp-1].(*evalBytes)
+
+		parsed, err := uuid.FromBytes(arg.bytes)
+		if err != nil {
+			env.vm.stack[env.vm.sp-1] = nil
+			env.vm.err = errIncorrectUUID(arg.bytes, "bin_to_uuid")
+			return 1
+		}
+		arg.bytes = hack.StringBytes(parsed.String())
+		arg.tt = int16(sqltypes.VarChar)
+		arg.col = col
+		return 1
+	}, "FN BIN_TO_UUID VARBINARY(SP-1)")
+}
+
+func (asm *assembler) Fn_BIN_TO_UUID1(col collations.TypedCollation) {
+	asm.adjustStack(-1)
+	asm.emit(func(env *ExpressionEnv) int {
+		arg := env.vm.stack[env.vm.sp-2].(*evalBytes)
+		b := arg.bytes
+
+		if env.vm.stack[env.vm.sp-1] != nil &&
+			env.vm.stack[env.vm.sp-1].(*evalInt64).i != 0 {
+			b = swapUUIDFrom(b)
+		}
+
+		parsed, err := uuid.FromBytes(b)
+		if err != nil {
+			env.vm.stack[env.vm.sp-2] = nil
+			env.vm.err = errIncorrectUUID(arg.bytes, "bin_to_uuid")
+			env.vm.sp--
+			return 1
+		}
+		arg.bytes = hack.StringBytes(parsed.String())
+		arg.tt = int16(sqltypes.VarChar)
+		arg.col = col
+		env.vm.sp--
+		return 1
+	}, "FN BIN_TO_UUID VARBINARY(SP-2) INT64(SP-1)")
+}
+
+func (asm *assembler) Fn_IS_UUID() {
+	asm.emit(func(env *ExpressionEnv) int {
+		arg := env.vm.stack[env.vm.sp-1].(*evalBytes)
+
+		_, err := uuid.ParseBytes(arg.bytes)
+		env.vm.stack[env.vm.sp-1] = env.vm.arena.newEvalBool(err == nil)
+		return 1
+	}, "FN IS_UUID VARBINARY(SP-1)")
+}
+
+func (asm *assembler) Fn_UUID() {
+	asm.adjustStack(1)
+	asm.emit(func(env *ExpressionEnv) int {
+		v, err := uuid.NewUUID()
+		if err != nil {
+			env.vm.err = err
+			env.vm.sp++
+			return 1
+		}
+		m, err := v.MarshalText()
+		if err != nil {
+			env.vm.err = err
+			env.vm.sp++
+			return 1
+		}
+
+		env.vm.stack[env.vm.sp] = env.vm.arena.newEvalText(m, collationUtf8mb3)
+		env.vm.sp++
+		return 1
+	}, "FN UUID")
+}
+
+func (asm *assembler) Fn_UUID_TO_BIN0() {
+	asm.emit(func(env *ExpressionEnv) int {
+		arg := env.vm.stack[env.vm.sp-1].(*evalBytes)
+
+		parsed, err := uuid.ParseBytes(arg.bytes)
+		if err != nil {
+			env.vm.stack[env.vm.sp-1] = nil
+			env.vm.err = errIncorrectUUID(arg.bytes, "uuid_to_bin")
+			return 1
+		}
+		arg.bytes = parsed[:]
+		arg.tt = int16(sqltypes.VarBinary)
+		arg.col = collationBinary
+		return 1
+	}, "FN UUID_TO_BIN VARBINARY(SP-1)")
+}
+
+func (asm *assembler) Fn_UUID_TO_BIN1() {
+	asm.adjustStack(-1)
+	asm.emit(func(env *ExpressionEnv) int {
+		arg := env.vm.stack[env.vm.sp-2].(*evalBytes)
+		parsed, err := uuid.ParseBytes(arg.bytes)
+		if err != nil {
+			env.vm.stack[env.vm.sp-2] = nil
+			env.vm.err = errIncorrectUUID(arg.bytes, "uuid_to_bin")
+			env.vm.sp--
+			return 1
+		}
+		b := parsed[:]
+		if env.vm.stack[env.vm.sp-1] != nil &&
+			env.vm.stack[env.vm.sp-1].(*evalInt64).i != 0 {
+			b = swapUUIDTo(b)
+		}
+		arg.bytes = b
+		arg.tt = int16(sqltypes.VarBinary)
+		arg.col = collationBinary
+		env.vm.sp--
+		return 1
+	}, "FN UUID_TO_BIN VARBINARY(SP-2) INT64(SP-1)")
+}

go/vt/vtgate/evalengine/eval.go

@@ -19,6 +19,7 @@ package evalengine
 import (
 	"fmt"
 	"strconv"
+	"unicode/utf8"
 
 	"vitess.io/vitess/go/hack"
 	"vitess.io/vitess/go/mysql/collations"
@@ -361,3 +362,38 @@ func valueToEval(value sqltypes.Value, collation collations.TypedCollation) (eva
 		return nil, vterrors.Errorf(vtrpcpb.Code_INTERNAL, "Type is not supported: %q %s", value, value.Type())
 	}
 }
+
+const hexchars = "0123456789ABCDEF"
+
+func sanitizeErrorValue(s []byte) []byte {
+	var buf []byte
+	for width := 0; len(s) > 0; s = s[width:] {
+		r := rune(s[0])
+		width = 1
+		if r >= utf8.RuneSelf {
+			r, width = utf8.DecodeLastRune(s)
+		}
+		if width == 1 && r == utf8.RuneError {
+			buf = append(buf, `\x`...)
+			buf = append(buf, hexchars[s[0]>>4])
+			buf = append(buf, hexchars[s[0]&0xF])
+			continue
+		}
+
+		if strconv.IsPrint(r) {
+			if r < utf8.RuneSelf {
+				buf = append(buf, byte(r))
+			} else {
+				b := [utf8.UTFMax]byte{}
+				n := utf8.EncodeRune(b[:], r)
+				buf = append(buf, b[:n]...)
+			}
+			continue
+		}
+
+		buf = append(buf, `\x`...)
+		buf = append(buf, hexchars[s[0]>>4])
+		buf = append(buf, hexchars[s[0]&0xF])
+	}
+	return buf
+}

go/vt/vtgate/evalengine/eval_temporal.go

@@ -2,7 +2,6 @@ package evalengine
 
 import (
 	"time"
-	"unicode/utf8"
 
 	"vitess.io/vitess/go/hack"
 	"vitess.io/vitess/go/mysql/datetime"
@@ -156,35 +155,6 @@ func newEvalTime(time datetime.Time, l int) *evalTemporal {
 	return &evalTemporal{t: sqltypes.Time, dt: datetime.DateTime{Time: time.Round(l)}, prec: uint8(l)}
 }
 
-func sanitizeErrorValue(s []byte) []byte {
-	b := make([]byte, 0, len(s)+1)
-	invalid := false // previous byte was from an invalid UTF-8 sequence
-	for i := 0; i < len(s); {
-		c := s[i]
-		if c < utf8.RuneSelf {
-			i++
-			invalid = false
-			if c != 0 {
-				b = append(b, c)
-			}
-			continue
-		}
-		_, wid := utf8.DecodeRune(s[i:])
-		if wid == 1 {
-			i++
-			if !invalid {
-				invalid = true
-				b = append(b, '?')
-			}
-			continue
-		}
-		invalid = false
-		b = append(b, s[i:i+wid]...)
-		i += wid
-	}
-	return b
-}
-
 func errIncorrectTemporal(date string, in []byte) error {
 	return vterrors.NewErrorf(vtrpcpb.Code_INVALID_ARGUMENT, vterrors.WrongValue, "Incorrect %s value: '%s'", date, sanitizeErrorValue(in))
 }