blob: 5ea6fb85562d2d21f1014a9b6e986fad0521ac76 [file] [log] [blame]
Herbie Ongc96a79d2019-03-08 10:49:17 -08001// Copyright 2019 The Go Authors. All rights reserved.
2// Use of this source code is governed by a BSD-style
3// license that can be found in the LICENSE file.
4
5package jsonpb
6
7import (
8 "encoding/base64"
9 "fmt"
10 "math"
11 "strconv"
12 "strings"
13
14 "github.com/golang/protobuf/v2/internal/encoding/json"
15 "github.com/golang/protobuf/v2/internal/errors"
Herbie Onge52379a2019-03-15 18:00:19 -070016 "github.com/golang/protobuf/v2/internal/pragma"
Herbie Ongc96a79d2019-03-08 10:49:17 -080017 "github.com/golang/protobuf/v2/internal/set"
18 "github.com/golang/protobuf/v2/proto"
19 pref "github.com/golang/protobuf/v2/reflect/protoreflect"
Herbie Onge52379a2019-03-15 18:00:19 -070020 "github.com/golang/protobuf/v2/reflect/protoregistry"
Herbie Ongc96a79d2019-03-08 10:49:17 -080021)
22
23// Unmarshal reads the given []byte into the given proto.Message.
24func Unmarshal(m proto.Message, b []byte) error {
25 return UnmarshalOptions{}.Unmarshal(m, b)
26}
27
28// UnmarshalOptions is a configurable JSON format parser.
Herbie Onge52379a2019-03-15 18:00:19 -070029type UnmarshalOptions struct {
30 pragma.NoUnkeyedLiterals
31
32 // Resolver is the registry used for type lookups when unmarshaling extensions
33 // and processing Any. If Resolver is not set, unmarshaling will default to
34 // using protoregistry.GlobalTypes.
35 Resolver *protoregistry.Types
36}
Herbie Ongc96a79d2019-03-08 10:49:17 -080037
38// Unmarshal reads the given []byte and populates the given proto.Message using
39// options in UnmarshalOptions object. It will clear the message first before
40// setting the fields. If it returns an error, the given message may be
41// partially set.
42func (o UnmarshalOptions) Unmarshal(m proto.Message, b []byte) error {
43 mr := m.ProtoReflect()
44 // TODO: Determine if we would like to have an option for merging or only
45 // have merging behavior. We should at least be consistent with textproto
46 // marshaling.
47 resetMessage(mr)
48
Herbie Onge52379a2019-03-15 18:00:19 -070049 resolver := o.Resolver
50 if resolver == nil {
51 resolver = protoregistry.GlobalTypes
52 }
53
54 dec := decoder{
55 Decoder: json.NewDecoder(b),
56 resolver: resolver,
57 }
Herbie Ongc96a79d2019-03-08 10:49:17 -080058 var nerr errors.NonFatal
59 if err := dec.unmarshalMessage(mr); !nerr.Merge(err) {
60 return err
61 }
62
63 // Check for EOF.
64 val, err := dec.Read()
65 if err != nil {
66 return err
67 }
68 if val.Type() != json.EOF {
69 return unexpectedJSONError{val}
70 }
71 return nerr.E
72}
73
74// resetMessage clears all fields of given protoreflect.Message.
75func resetMessage(m pref.Message) {
76 knownFields := m.KnownFields()
77 knownFields.Range(func(num pref.FieldNumber, _ pref.Value) bool {
78 knownFields.Clear(num)
79 return true
80 })
81 unknownFields := m.UnknownFields()
82 unknownFields.Range(func(num pref.FieldNumber, _ pref.RawFields) bool {
83 unknownFields.Set(num, nil)
84 return true
85 })
86 extTypes := knownFields.ExtensionTypes()
87 extTypes.Range(func(xt pref.ExtensionType) bool {
88 extTypes.Remove(xt)
89 return true
90 })
91}
92
93// unexpectedJSONError is an error that contains the unexpected json.Value. This
94// is used by decoder methods to provide callers the read json.Value that it
95// did not expect.
96// TODO: Consider moving this to internal/encoding/json for consistency with
97// errors that package returns.
98type unexpectedJSONError struct {
99 value json.Value
100}
101
102func (e unexpectedJSONError) Error() string {
103 return newError("unexpected value %s", e.value).Error()
104}
105
106// newError returns an error object. If one of the values passed in is of
107// json.Value type, it produces an error with position info.
108func newError(f string, x ...interface{}) error {
109 var hasValue bool
110 var line, column int
111 for i := 0; i < len(x); i++ {
112 if val, ok := x[i].(json.Value); ok {
113 line, column = val.Position()
114 hasValue = true
115 break
116 }
117 }
118 e := errors.New(f, x...)
119 if hasValue {
120 return errors.New("(line %d:%d): %v", line, column, e)
121 }
122 return e
123}
124
125// decoder decodes JSON into protoreflect values.
126type decoder struct {
127 *json.Decoder
Herbie Onge52379a2019-03-15 18:00:19 -0700128 resolver *protoregistry.Types
Herbie Ongc96a79d2019-03-08 10:49:17 -0800129}
130
131// unmarshalMessage unmarshals a message into the given protoreflect.Message.
132func (d decoder) unmarshalMessage(m pref.Message) error {
133 var nerr errors.NonFatal
134 var reqNums set.Ints
135 var seenNums set.Ints
136
137 msgType := m.Type()
138 knownFields := m.KnownFields()
139 fieldDescs := msgType.Fields()
Herbie Onge52379a2019-03-15 18:00:19 -0700140 xtTypes := knownFields.ExtensionTypes()
Herbie Ongc96a79d2019-03-08 10:49:17 -0800141
142 jval, err := d.Read()
143 if !nerr.Merge(err) {
144 return err
145 }
146 if jval.Type() != json.StartObject {
147 return unexpectedJSONError{jval}
148 }
149
150Loop:
151 for {
152 // Read field name.
153 jval, err := d.Read()
154 if !nerr.Merge(err) {
155 return err
156 }
157 switch jval.Type() {
158 default:
159 return unexpectedJSONError{jval}
160 case json.EndObject:
161 break Loop
162 case json.Name:
163 // Continue below.
164 }
165
166 name, err := jval.Name()
167 if !nerr.Merge(err) {
168 return err
169 }
170
Herbie Onge52379a2019-03-15 18:00:19 -0700171 // Get the FieldDescriptor.
172 var fd pref.FieldDescriptor
173 if strings.HasPrefix(name, "[") && strings.HasSuffix(name, "]") {
174 // Only extension names are in [name] format.
175 xtName := pref.FullName(name[1 : len(name)-1])
176 xt := xtTypes.ByName(xtName)
177 if xt == nil {
178 xt, err = d.findExtension(xtName)
179 if err != nil && err != protoregistry.NotFound {
180 return errors.New("unable to resolve [%v]: %v", xtName, err)
181 }
182 if xt != nil {
183 xtTypes.Register(xt)
184 }
185 }
186 fd = xt
187 } else {
188 // The name can either be the JSON name or the proto field name.
189 fd = fieldDescs.ByJSONName(name)
190 if fd == nil {
191 fd = fieldDescs.ByName(pref.Name(name))
192 }
Herbie Ongc96a79d2019-03-08 10:49:17 -0800193 }
194
195 if fd == nil {
196 // Field is unknown.
197 // TODO: Provide option to ignore unknown message fields.
198 return newError("%v contains unknown field %s", msgType.FullName(), jval)
199 }
200
201 // Do not allow duplicate fields.
202 num := uint64(fd.Number())
203 if seenNums.Has(num) {
204 return newError("%v contains repeated field %s", msgType.FullName(), jval)
205 }
206 seenNums.Set(num)
207
208 // No need to set values for JSON null.
209 if d.Peek() == json.Null {
210 d.Read()
211 continue
212 }
213
214 if cardinality := fd.Cardinality(); cardinality == pref.Repeated {
215 // Map or list fields have cardinality of repeated.
216 if err := d.unmarshalRepeated(fd, knownFields); !nerr.Merge(err) {
217 return errors.New("%v|%q: %v", fd.FullName(), name, err)
218 }
219 } else {
220 // Required or optional fields.
221 if err := d.unmarshalSingular(fd, knownFields); !nerr.Merge(err) {
222 return errors.New("%v|%q: %v", fd.FullName(), name, err)
223 }
224 if cardinality == pref.Required {
225 reqNums.Set(num)
226 }
227 }
228 }
229
230 // Check for any missing required fields.
231 allReqNums := msgType.RequiredNumbers()
232 if reqNums.Len() != allReqNums.Len() {
233 for i := 0; i < allReqNums.Len(); i++ {
234 if num := allReqNums.Get(i); !reqNums.Has(uint64(num)) {
235 nerr.AppendRequiredNotSet(string(fieldDescs.ByNumber(num).FullName()))
236 }
237 }
238 }
239
240 return nerr.E
241}
242
Herbie Onge52379a2019-03-15 18:00:19 -0700243// findExtension returns protoreflect.ExtensionType from the resolver if found.
244func (d decoder) findExtension(xtName pref.FullName) (pref.ExtensionType, error) {
245 xt, err := d.resolver.FindExtensionByName(xtName)
246 if err == nil {
247 return xt, nil
248 }
249
250 // Check if this is a MessageSet extension field.
251 xt, err = d.resolver.FindExtensionByName(xtName + ".message_set_extension")
252 if err == nil && isMessageSetExtension(xt) {
253 return xt, nil
254 }
255 return nil, protoregistry.NotFound
256}
257
Herbie Ongc96a79d2019-03-08 10:49:17 -0800258// unmarshalSingular unmarshals to the non-repeated field specified by the given
259// FieldDescriptor.
260func (d decoder) unmarshalSingular(fd pref.FieldDescriptor, knownFields pref.KnownFields) error {
261 var val pref.Value
262 var err error
263 num := fd.Number()
264
265 switch fd.Kind() {
266 case pref.MessageKind, pref.GroupKind:
267 m := knownFields.NewMessage(num)
268 err = d.unmarshalMessage(m)
269 val = pref.ValueOf(m)
270 default:
271 val, err = d.unmarshalScalar(fd)
272 }
273
274 var nerr errors.NonFatal
275 if !nerr.Merge(err) {
276 return err
277 }
278 knownFields.Set(num, val)
279 return nerr.E
280}
281
282// unmarshalScalar unmarshals to a scalar/enum protoreflect.Value specified by
283// the given FieldDescriptor.
284func (d decoder) unmarshalScalar(fd pref.FieldDescriptor) (pref.Value, error) {
285 const b32 int = 32
286 const b64 int = 64
287
288 var nerr errors.NonFatal
289 jval, err := d.Read()
290 if !nerr.Merge(err) {
291 return pref.Value{}, err
292 }
293
294 kind := fd.Kind()
295 switch kind {
296 case pref.BoolKind:
297 return unmarshalBool(jval)
298
299 case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
300 return unmarshalInt(jval, b32)
301
302 case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
303 return unmarshalInt(jval, b64)
304
305 case pref.Uint32Kind, pref.Fixed32Kind:
306 return unmarshalUint(jval, b32)
307
308 case pref.Uint64Kind, pref.Fixed64Kind:
309 return unmarshalUint(jval, b64)
310
311 case pref.FloatKind:
312 return unmarshalFloat(jval, b32)
313
314 case pref.DoubleKind:
315 return unmarshalFloat(jval, b64)
316
317 case pref.StringKind:
318 pval, err := unmarshalString(jval)
319 if !nerr.Merge(err) {
320 return pval, err
321 }
322 return pval, nerr.E
323
324 case pref.BytesKind:
325 return unmarshalBytes(jval)
326
327 case pref.EnumKind:
328 return unmarshalEnum(jval, fd)
329 }
330
331 panic(fmt.Sprintf("invalid scalar kind %v", kind))
332}
333
334func unmarshalBool(jval json.Value) (pref.Value, error) {
335 if jval.Type() != json.Bool {
336 return pref.Value{}, unexpectedJSONError{jval}
337 }
338 b, err := jval.Bool()
339 return pref.ValueOf(b), err
340}
341
342func unmarshalInt(jval json.Value, bitSize int) (pref.Value, error) {
343 switch jval.Type() {
344 case json.Number:
345 return getInt(jval, bitSize)
346
347 case json.String:
Herbie Onge52379a2019-03-15 18:00:19 -0700348 // Decode number from string.
349 dec := json.NewDecoder([]byte(jval.String()))
Herbie Ongc96a79d2019-03-08 10:49:17 -0800350 var nerr errors.NonFatal
351 jval, err := dec.Read()
352 if !nerr.Merge(err) {
353 return pref.Value{}, err
354 }
355 return getInt(jval, bitSize)
356 }
357 return pref.Value{}, unexpectedJSONError{jval}
358}
359
360func getInt(jval json.Value, bitSize int) (pref.Value, error) {
361 n, err := jval.Int(bitSize)
362 if err != nil {
363 return pref.Value{}, err
364 }
365 if bitSize == 32 {
366 return pref.ValueOf(int32(n)), nil
367 }
368 return pref.ValueOf(n), nil
369}
370
371func unmarshalUint(jval json.Value, bitSize int) (pref.Value, error) {
372 switch jval.Type() {
373 case json.Number:
374 return getUint(jval, bitSize)
375
376 case json.String:
Herbie Onge52379a2019-03-15 18:00:19 -0700377 // Decode number from string.
378 dec := json.NewDecoder([]byte(jval.String()))
Herbie Ongc96a79d2019-03-08 10:49:17 -0800379 var nerr errors.NonFatal
380 jval, err := dec.Read()
381 if !nerr.Merge(err) {
382 return pref.Value{}, err
383 }
384 return getUint(jval, bitSize)
385 }
386 return pref.Value{}, unexpectedJSONError{jval}
387}
388
389func getUint(jval json.Value, bitSize int) (pref.Value, error) {
390 n, err := jval.Uint(bitSize)
391 if err != nil {
392 return pref.Value{}, err
393 }
394 if bitSize == 32 {
395 return pref.ValueOf(uint32(n)), nil
396 }
397 return pref.ValueOf(n), nil
398}
399
400func unmarshalFloat(jval json.Value, bitSize int) (pref.Value, error) {
401 switch jval.Type() {
402 case json.Number:
403 return getFloat(jval, bitSize)
404
405 case json.String:
406 s := jval.String()
407 switch s {
408 case "NaN":
409 if bitSize == 32 {
410 return pref.ValueOf(float32(math.NaN())), nil
411 }
412 return pref.ValueOf(math.NaN()), nil
413 case "Infinity":
414 if bitSize == 32 {
415 return pref.ValueOf(float32(math.Inf(+1))), nil
416 }
417 return pref.ValueOf(math.Inf(+1)), nil
418 case "-Infinity":
419 if bitSize == 32 {
420 return pref.ValueOf(float32(math.Inf(-1))), nil
421 }
422 return pref.ValueOf(math.Inf(-1)), nil
423 }
Herbie Onge52379a2019-03-15 18:00:19 -0700424 // Decode number from string.
425 dec := json.NewDecoder([]byte(s))
Herbie Ongc96a79d2019-03-08 10:49:17 -0800426 var nerr errors.NonFatal
427 jval, err := dec.Read()
428 if !nerr.Merge(err) {
429 return pref.Value{}, err
430 }
431 return getFloat(jval, bitSize)
432 }
433 return pref.Value{}, unexpectedJSONError{jval}
434}
435
436func getFloat(jval json.Value, bitSize int) (pref.Value, error) {
437 n, err := jval.Float(bitSize)
438 if err != nil {
439 return pref.Value{}, err
440 }
441 if bitSize == 32 {
442 return pref.ValueOf(float32(n)), nil
443 }
444 return pref.ValueOf(n), nil
445}
446
447func unmarshalString(jval json.Value) (pref.Value, error) {
448 if jval.Type() != json.String {
449 return pref.Value{}, unexpectedJSONError{jval}
450 }
451 return pref.ValueOf(jval.String()), nil
452}
453
454func unmarshalBytes(jval json.Value) (pref.Value, error) {
455 if jval.Type() != json.String {
456 return pref.Value{}, unexpectedJSONError{jval}
457 }
458
459 s := jval.String()
460 enc := base64.StdEncoding
461 if strings.ContainsAny(s, "-_") {
462 enc = base64.URLEncoding
463 }
464 if len(s)%4 != 0 {
465 enc = enc.WithPadding(base64.NoPadding)
466 }
467 b, err := enc.DecodeString(s)
468 if err != nil {
469 return pref.Value{}, err
470 }
471 return pref.ValueOf(b), nil
472}
473
474func unmarshalEnum(jval json.Value, fd pref.FieldDescriptor) (pref.Value, error) {
475 switch jval.Type() {
476 case json.String:
477 // Lookup EnumNumber based on name.
478 s := jval.String()
479 if enumVal := fd.EnumType().Values().ByName(pref.Name(s)); enumVal != nil {
480 return pref.ValueOf(enumVal.Number()), nil
481 }
482 return pref.Value{}, newError("invalid enum value %q", jval)
483
484 case json.Number:
485 n, err := jval.Int(32)
486 if err != nil {
487 return pref.Value{}, err
488 }
489 return pref.ValueOf(pref.EnumNumber(n)), nil
490 }
491
492 return pref.Value{}, unexpectedJSONError{jval}
493}
494
495// unmarshalRepeated unmarshals into a repeated field.
496func (d decoder) unmarshalRepeated(fd pref.FieldDescriptor, knownFields pref.KnownFields) error {
497 var nerr errors.NonFatal
498 num := fd.Number()
499 val := knownFields.Get(num)
500 if !fd.IsMap() {
501 if err := d.unmarshalList(fd, val.List()); !nerr.Merge(err) {
502 return err
503 }
504 } else {
505 if err := d.unmarshalMap(fd, val.Map()); !nerr.Merge(err) {
506 return err
507 }
508 }
509 return nerr.E
510}
511
512// unmarshalList unmarshals into given protoreflect.List.
513func (d decoder) unmarshalList(fd pref.FieldDescriptor, list pref.List) error {
514 var nerr errors.NonFatal
515 jval, err := d.Read()
516 if !nerr.Merge(err) {
517 return err
518 }
519 if jval.Type() != json.StartArray {
520 return unexpectedJSONError{jval}
521 }
522
523 switch fd.Kind() {
524 case pref.MessageKind, pref.GroupKind:
525 for {
526 m := list.NewMessage()
527 err := d.unmarshalMessage(m)
528 if !nerr.Merge(err) {
529 if e, ok := err.(unexpectedJSONError); ok {
530 if e.value.Type() == json.EndArray {
531 // Done with list.
532 return nerr.E
533 }
534 }
535 return err
536 }
537 list.Append(pref.ValueOf(m))
538 }
539 default:
540 for {
541 val, err := d.unmarshalScalar(fd)
542 if !nerr.Merge(err) {
543 if e, ok := err.(unexpectedJSONError); ok {
544 if e.value.Type() == json.EndArray {
545 // Done with list.
546 return nerr.E
547 }
548 }
549 return err
550 }
551 list.Append(val)
552 }
553 }
554 return nerr.E
555}
556
557// unmarshalMap unmarshals into given protoreflect.Map.
558func (d decoder) unmarshalMap(fd pref.FieldDescriptor, mmap pref.Map) error {
559 var nerr errors.NonFatal
560
561 jval, err := d.Read()
562 if !nerr.Merge(err) {
563 return err
564 }
565 if jval.Type() != json.StartObject {
566 return unexpectedJSONError{jval}
567 }
568
569 fields := fd.MessageType().Fields()
570 keyDesc := fields.ByNumber(1)
571 valDesc := fields.ByNumber(2)
572
573 // Determine ahead whether map entry is a scalar type or a message type in
574 // order to call the appropriate unmarshalMapValue func inside the for loop
575 // below.
576 unmarshalMapValue := func() (pref.Value, error) {
577 return d.unmarshalScalar(valDesc)
578 }
579 switch valDesc.Kind() {
580 case pref.MessageKind, pref.GroupKind:
581 unmarshalMapValue = func() (pref.Value, error) {
582 m := mmap.NewMessage()
583 if err := d.unmarshalMessage(m); err != nil {
584 return pref.Value{}, err
585 }
586 return pref.ValueOf(m), nil
587 }
588 }
589
590Loop:
591 for {
592 // Read field name.
593 jval, err := d.Read()
594 if !nerr.Merge(err) {
595 return err
596 }
597 switch jval.Type() {
598 default:
599 return unexpectedJSONError{jval}
600 case json.EndObject:
601 break Loop
602 case json.Name:
603 // Continue.
604 }
605
606 name, err := jval.Name()
607 if !nerr.Merge(err) {
608 return err
609 }
610
611 // Unmarshal field name.
612 pkey, err := unmarshalMapKey(name, keyDesc)
613 if !nerr.Merge(err) {
614 return err
615 }
616
617 // Check for duplicate field name.
618 if mmap.Has(pkey) {
619 return newError("duplicate map key %q", jval)
620 }
621
622 // Read and unmarshal field value.
623 pval, err := unmarshalMapValue()
624 if !nerr.Merge(err) {
625 return err
626 }
627
628 mmap.Set(pkey, pval)
629 }
630
631 return nerr.E
632}
633
634// unmarshalMapKey converts given string into a protoreflect.MapKey. A map key type is any
635// integral or string type.
636func unmarshalMapKey(name string, fd pref.FieldDescriptor) (pref.MapKey, error) {
637 const b32 = 32
638 const b64 = 64
639 const base10 = 10
640
641 kind := fd.Kind()
642 switch kind {
643 case pref.StringKind:
644 return pref.ValueOf(name).MapKey(), nil
645
646 case pref.BoolKind:
647 switch name {
648 case "true":
649 return pref.ValueOf(true).MapKey(), nil
650 case "false":
651 return pref.ValueOf(false).MapKey(), nil
652 }
653 return pref.MapKey{}, errors.New("invalid value for boolean key %q", name)
654
655 case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
656 n, err := strconv.ParseInt(name, base10, b32)
657 if err != nil {
658 return pref.MapKey{}, err
659 }
660 return pref.ValueOf(int32(n)).MapKey(), nil
661
662 case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
663 n, err := strconv.ParseInt(name, base10, b64)
664 if err != nil {
665 return pref.MapKey{}, err
666 }
667 return pref.ValueOf(int64(n)).MapKey(), nil
668
669 case pref.Uint32Kind, pref.Fixed32Kind:
670 n, err := strconv.ParseUint(name, base10, b32)
671 if err != nil {
672 return pref.MapKey{}, err
673 }
674 return pref.ValueOf(uint32(n)).MapKey(), nil
675
676 case pref.Uint64Kind, pref.Fixed64Kind:
677 n, err := strconv.ParseUint(name, base10, b64)
678 if err != nil {
679 return pref.MapKey{}, err
680 }
681 return pref.ValueOf(uint64(n)).MapKey(), nil
682 }
683
684 panic(fmt.Sprintf("%s: invalid kind %s for map key", fd.FullName(), kind))
685}