Merge pull request #355 from blinklabs-io/feat/cbor-value-structure

feat: generate JSON for arbitrary CBOR

Author: agaffney

cbor/cbor.go

@@ -30,6 +30,10 @@ const (
 
 	// Max value able to be stored in a single byte without type prefix
 	CBOR_MAX_UINT_SIMPLE uint8 = 0x17
+
+	// Useful tag numbers
+	CborTagSet = 258
+	CborTagMap = 259
 )
 
 // Create an alias for RawMessage for convenience

cbor/decode.go

@@ -64,7 +64,7 @@ func DecodeIdFromList(cborData []byte) (int, error) {
 		return 0, err
 	}
 	// Make sure that the value is actually numeric
-	switch v := tmp.Value.([]interface{})[0].(type) {
+	switch v := tmp.Value().([]interface{})[0].(type) {
 	// The upstream CBOR library uses uint64 by default for numeric values
 	case uint64:
 		return int(v), nil

cbor/value.go

@@ -15,88 +15,94 @@
 package cbor
 
 import (
+	"encoding/hex"
+	"encoding/json"
 	"fmt"
+	"sort"
+	"strings"
 )
 
 // Helpful wrapper for parsing arbitrary CBOR data which may contain types that
 // cannot be easily represented in Go (such as maps with bytestring keys)
 type Value struct {
-	Value interface{}
+	value interface{}
 	// We store this as a string so that the type is still hashable for use as map keys
 	cborData string
 }
 
-func (v *Value) UnmarshalCBOR(data []byte) (err error) {
+func (v *Value) UnmarshalCBOR(data []byte) error {
 	// Save the original CBOR
 	v.cborData = string(data[:])
 	cborType := data[0] & CBOR_TYPE_MASK
 	switch cborType {
 	case CBOR_TYPE_MAP:
-		// There are certain types that cannot be used as map keys in Go but are valid in CBOR. Trying to
-		// parse CBOR containing a map with keys of one of those types will cause a panic. We setup this
-		// deferred function to recover from a possible panic and return an error
-		defer func() {
-			if r := recover(); r != nil {
-				err = fmt.Errorf("decode failure, probably due to type unsupported by Go: %v", r)
-			}
-		}()
-		tmpValue := map[Value]Value{}
-		if _, err := Decode(data, &tmpValue); err != nil {
-			return err
-		}
-		// Extract actual value from each child value
-		newValue := map[interface{}]interface{}{}
-		for key, value := range tmpValue {
-			newValue[key.Value] = value.Value
-		}
-		v.Value = newValue
+		return v.processMap(data)
 	case CBOR_TYPE_ARRAY:
-		tmpValue := []Value{}
-		if _, err := Decode(data, &tmpValue); err != nil {
-			return err
-		}
-		// Extract actual value from each child value
-		newValue := []interface{}{}
-		for _, value := range tmpValue {
-			newValue = append(newValue, value.Value)
-		}
-		v.Value = newValue
+		return v.processArray(data)
 	case CBOR_TYPE_TEXT_STRING:
 		var tmpValue string
 		if _, err := Decode(data, &tmpValue); err != nil {
 			return err
 		}
-		v.Value = tmpValue
+		v.value = tmpValue
 	case CBOR_TYPE_BYTE_STRING:
 		// Use our custom type which stores the bytestring in a way that allows it to be used as a map key
 		var tmpValue ByteString
 		if _, err := Decode(data, &tmpValue); err != nil {
 			return err
 		}
-		v.Value = tmpValue
+		v.value = tmpValue
 	case CBOR_TYPE_TAG:
 		// Parse as a raw tag to get number and nested CBOR data
 		tmpTag := RawTag{}
 		if _, err := Decode(data, &tmpTag); err != nil {
 			return err
 		}
-		// Parse the tag value via our custom Value object to handle problem types
-		tmpValue := Value{}
-		if _, err := Decode(tmpTag.Content, &tmpValue); err != nil {
-			return err
-		}
-		// Create new tag object with decoded value
-		newValue := Tag{
-			Number:  tmpTag.Number,
-			Content: tmpValue.Value,
+		switch tmpTag.Number {
+		case CborTagSet:
+			return v.processArray(tmpTag.Content)
+		case CborTagMap:
+			return v.processMap(tmpTag.Content)
+		default:
+			// Parse the tag value via our custom Value object to handle problem types
+			tmpValue := Value{}
+			if _, err := Decode(tmpTag.Content, &tmpValue); err != nil {
+				return err
+			}
+			// These tag numbers correspond to the Enumerated Alternative Data Items notable CBOR tags. These
+			// are often used in Plutus script datums
+			// https://www.ietf.org/archive/id/draft-bormann-cbor-notable-tags-07.html#name-enumerated-alternative-data
+			if tmpTag.Number >= 121 && tmpTag.Number <= 127 {
+				// Alternatives 0-6
+				v.value = Constructor{
+					constructor: uint(tmpTag.Number - 121),
+					value:       &tmpValue,
+				}
+			} else if tmpTag.Number >= 1280 && tmpTag.Number <= 1400 {
+				// Alternatives 7-127
+				v.value = Constructor{
+					constructor: uint(tmpTag.Number - 1280 + 7),
+					value:       &tmpValue,
+				}
+			} else if tmpTag.Number == 101 {
+				// Alternatives 128+
+				newValue := Value{
+					value: tmpValue.Value().([]interface{})[1],
+				}
+				v.value = Constructor{
+					constructor: tmpValue.Value().([]interface{})[0].(uint),
+					value:       &newValue,
+				}
+			} else {
+				return fmt.Errorf("unsupported CBOR tag: %d", tmpTag.Number)
+			}
 		}
-		v.Value = newValue
 	default:
 		var tmpValue interface{}
 		if _, err := Decode(data, &tmpValue); err != nil {
 			return err
 		}
-		v.Value = tmpValue
+		v.value = tmpValue
 	}
 	return nil
 }
@@ -105,19 +111,179 @@ func (v Value) Cbor() []byte {
 	return []byte(v.cborData)
 }
 
+func (v Value) Value() interface{} {
+	return v.value
+}
+
+func (v Value) MarshalJSON() ([]byte, error) {
+	var tmpJson string
+	if v.value != nil {
+		astJson, err := generateAstJson(v.value)
+		if err != nil {
+			return nil, err
+		}
+		tmpJson = fmt.Sprintf(
+			`{"cbor":"%s","json":%s}`,
+			hex.EncodeToString([]byte(v.cborData)),
+			astJson,
+		)
+	} else {
+		tmpJson = fmt.Sprintf(
+			`{"cbor":"%s"}`,
+			hex.EncodeToString([]byte(v.cborData)),
+		)
+	}
+	return []byte(tmpJson), nil
+}
+
+func (v *Value) processMap(data []byte) (err error) {
+	// There are certain types that cannot be used as map keys in Go but are valid in CBOR. Trying to
+	// parse CBOR containing a map with keys of one of those types will cause a panic. We setup this
+	// deferred function to recover from a possible panic and return an error
+	defer func() {
+		if r := recover(); r != nil {
+			err = fmt.Errorf("decode failure, probably due to type unsupported by Go: %v", r)
+		}
+	}()
+	tmpValue := map[Value]Value{}
+	if _, err = Decode(data, &tmpValue); err != nil {
+		return err
+	}
+	// Extract actual value from each child value
+	newValue := map[interface{}]interface{}{}
+	for key, value := range tmpValue {
+		newValue[key.Value()] = value.Value()
+	}
+	v.value = newValue
+	return nil
+}
+
+func (v *Value) processArray(data []byte) error {
+	tmpValue := []Value{}
+	if _, err := Decode(data, &tmpValue); err != nil {
+		return err
+	}
+	// Extract actual value from each child value
+	newValue := []interface{}{}
+	for _, value := range tmpValue {
+		newValue = append(newValue, value.Value())
+	}
+	v.value = newValue
+	return nil
+}
+
+func generateAstJson(obj interface{}) ([]byte, error) {
+	tmpJsonObj := map[string]interface{}{}
+	switch v := obj.(type) {
+	case ByteString:
+		tmpJsonObj["bytes"] = hex.EncodeToString(v.Bytes())
+	case []interface{}:
+		tmpJson := `{"list":[`
+		for idx, val := range v {
+			tmpVal, err := generateAstJson(val)
+			if err != nil {
+				return nil, err
+			}
+			tmpJson += string(tmpVal)
+			if idx != (len(v) - 1) {
+				tmpJson += `,`
+			}
+		}
+		tmpJson += `]}`
+		return []byte(tmpJson), nil
+	case map[interface{}]interface{}:
+		tmpItems := []string{}
+		for key, val := range v {
+			keyAstJson, err := generateAstJson(key)
+			if err != nil {
+				return nil, err
+			}
+			valAstJson, err := generateAstJson(val)
+			if err != nil {
+				return nil, err
+			}
+			tmpJson := fmt.Sprintf(
+				`{"k":%s,"v":%s}`,
+				keyAstJson,
+				valAstJson,
+			)
+			tmpItems = append(tmpItems, string(tmpJson))
+		}
+		// We naively sort the rendered map items to give consistent ordering
+		sort.Strings(tmpItems)
+		tmpJson := fmt.Sprintf(
+			`{"map":[%s]}`,
+			strings.Join(tmpItems, ","),
+		)
+		return []byte(tmpJson), nil
+	case Constructor:
+		return json.Marshal(obj)
+	case int, uint, uint64, int64:
+		tmpJsonObj["int"] = v
+	case bool:
+		tmpJsonObj["bool"] = v
+	case string:
+		tmpJsonObj["string"] = v
+	default:
+		return nil, fmt.Errorf("unknown data type (%T) for value: %#v", obj, obj)
+	}
+	return json.Marshal(&tmpJsonObj)
+}
+
+type Constructor struct {
+	constructor uint
+	value       *Value
+}
+
+func (v Constructor) MarshalJSON() ([]byte, error) {
+	tmpJson := fmt.Sprintf(`{"constructor":%d,"fields":[`, v.constructor)
+	tmpList := [][]byte{}
+	for _, val := range v.value.Value().([]any) {
+		tmpVal, err := generateAstJson(val)
+		if err != nil {
+			return nil, err
+		}
+		tmpList = append(tmpList, tmpVal)
+	}
+	for idx, val := range tmpList {
+		tmpJson += string(val)
+		if idx != (len(tmpList) - 1) {
+			tmpJson += `,`
+		}
+	}
+	tmpJson += `]}`
+	return []byte(tmpJson), nil
+}
+
 type LazyValue struct {
-	*Value
+	value *Value
 }
 
 func (l *LazyValue) UnmarshalCBOR(data []byte) error {
-	if l.Value == nil {
-		l.Value = &Value{}
+	if l.value == nil {
+		l.value = &Value{}
 	}
-	l.cborData = string(data[:])
+	l.value.cborData = string(data[:])
 	return nil
 }
 
-func (l *LazyValue) Decode() (*Value, error) {
-	err := l.Value.UnmarshalCBOR([]byte(l.cborData))
-	return l.Value, err
+func (l *LazyValue) MarshalJSON() ([]byte, error) {
+	if l.Value() == nil {
+		// Try to decode if we can, but don't blow up if we can't
+		_, _ = l.Decode()
+	}
+	return l.value.MarshalJSON()
+}
+
+func (l *LazyValue) Decode() (interface{}, error) {
+	err := l.value.UnmarshalCBOR([]byte(l.value.cborData))
+	return l.Value(), err
+}
+
+func (l *LazyValue) Value() interface{} {
+	return l.value.Value()
+}
+
+func (l *LazyValue) Cbor() []byte {
+	return l.value.Cbor()
 }