decode.go 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685
  1. // 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. package protojson
  5. import (
  6. "encoding/base64"
  7. "fmt"
  8. "math"
  9. "strconv"
  10. "strings"
  11. "google.golang.org/protobuf/internal/encoding/json"
  12. "google.golang.org/protobuf/internal/encoding/messageset"
  13. "google.golang.org/protobuf/internal/errors"
  14. "google.golang.org/protobuf/internal/flags"
  15. "google.golang.org/protobuf/internal/pragma"
  16. "google.golang.org/protobuf/internal/set"
  17. "google.golang.org/protobuf/proto"
  18. pref "google.golang.org/protobuf/reflect/protoreflect"
  19. "google.golang.org/protobuf/reflect/protoregistry"
  20. )
  21. // Unmarshal reads the given []byte into the given proto.Message.
  22. func Unmarshal(b []byte, m proto.Message) error {
  23. return UnmarshalOptions{}.Unmarshal(b, m)
  24. }
  25. // UnmarshalOptions is a configurable JSON format parser.
  26. type UnmarshalOptions struct {
  27. pragma.NoUnkeyedLiterals
  28. // If AllowPartial is set, input for messages that will result in missing
  29. // required fields will not return an error.
  30. AllowPartial bool
  31. // If DiscardUnknown is set, unknown fields are ignored.
  32. DiscardUnknown bool
  33. // Resolver is used for looking up types when unmarshaling
  34. // google.protobuf.Any messages or extension fields.
  35. // If nil, this defaults to using protoregistry.GlobalTypes.
  36. Resolver interface {
  37. protoregistry.MessageTypeResolver
  38. protoregistry.ExtensionTypeResolver
  39. }
  40. decoder *json.Decoder
  41. }
  42. // Unmarshal reads the given []byte and populates the given proto.Message using
  43. // options in UnmarshalOptions object. It will clear the message first before
  44. // setting the fields. If it returns an error, the given message may be
  45. // partially set.
  46. func (o UnmarshalOptions) Unmarshal(b []byte, m proto.Message) error {
  47. // TODO: Determine if we would like to have an option for merging or only
  48. // have merging behavior. We should at least be consistent with textproto
  49. // marshaling.
  50. proto.Reset(m)
  51. if o.Resolver == nil {
  52. o.Resolver = protoregistry.GlobalTypes
  53. }
  54. o.decoder = json.NewDecoder(b)
  55. if err := o.unmarshalMessage(m.ProtoReflect(), false); err != nil {
  56. return err
  57. }
  58. // Check for EOF.
  59. val, err := o.decoder.Read()
  60. if err != nil {
  61. return err
  62. }
  63. if val.Type() != json.EOF {
  64. return unexpectedJSONError{val}
  65. }
  66. if o.AllowPartial {
  67. return nil
  68. }
  69. return proto.IsInitialized(m)
  70. }
  71. // unexpectedJSONError is an error that contains the unexpected json.Value. This
  72. // is returned by methods to provide callers the read json.Value that it did not
  73. // expect.
  74. // TODO: Consider moving this to internal/encoding/json for consistency with
  75. // errors that package returns.
  76. type unexpectedJSONError struct {
  77. value json.Value
  78. }
  79. func (e unexpectedJSONError) Error() string {
  80. return newError("unexpected value %s", e.value).Error()
  81. }
  82. // newError returns an error object. If one of the values passed in is of
  83. // json.Value type, it produces an error with position info.
  84. func newError(f string, x ...interface{}) error {
  85. var hasValue bool
  86. var line, column int
  87. for i := 0; i < len(x); i++ {
  88. if val, ok := x[i].(json.Value); ok {
  89. line, column = val.Position()
  90. hasValue = true
  91. break
  92. }
  93. }
  94. e := errors.New(f, x...)
  95. if hasValue {
  96. return errors.New("(line %d:%d): %v", line, column, e)
  97. }
  98. return e
  99. }
  100. // unmarshalMessage unmarshals a message into the given protoreflect.Message.
  101. func (o UnmarshalOptions) unmarshalMessage(m pref.Message, skipTypeURL bool) error {
  102. if isCustomType(m.Descriptor().FullName()) {
  103. return o.unmarshalCustomType(m)
  104. }
  105. jval, err := o.decoder.Read()
  106. if err != nil {
  107. return err
  108. }
  109. if jval.Type() != json.StartObject {
  110. return unexpectedJSONError{jval}
  111. }
  112. if err := o.unmarshalFields(m, skipTypeURL); err != nil {
  113. return err
  114. }
  115. return nil
  116. }
  117. // unmarshalFields unmarshals the fields into the given protoreflect.Message.
  118. func (o UnmarshalOptions) unmarshalFields(m pref.Message, skipTypeURL bool) error {
  119. messageDesc := m.Descriptor()
  120. if !flags.ProtoLegacy && messageset.IsMessageSet(messageDesc) {
  121. return errors.New("no support for proto1 MessageSets")
  122. }
  123. var seenNums set.Ints
  124. var seenOneofs set.Ints
  125. fieldDescs := messageDesc.Fields()
  126. for {
  127. // Read field name.
  128. jval, err := o.decoder.Read()
  129. if err != nil {
  130. return err
  131. }
  132. switch jval.Type() {
  133. default:
  134. return unexpectedJSONError{jval}
  135. case json.EndObject:
  136. return nil
  137. case json.Name:
  138. // Continue below.
  139. }
  140. name, err := jval.Name()
  141. if err != nil {
  142. return err
  143. }
  144. // Unmarshaling a non-custom embedded message in Any will contain the
  145. // JSON field "@type" which should be skipped because it is not a field
  146. // of the embedded message, but simply an artifact of the Any format.
  147. if skipTypeURL && name == "@type" {
  148. o.decoder.Read()
  149. continue
  150. }
  151. // Get the FieldDescriptor.
  152. var fd pref.FieldDescriptor
  153. if strings.HasPrefix(name, "[") && strings.HasSuffix(name, "]") {
  154. // Only extension names are in [name] format.
  155. extName := pref.FullName(name[1 : len(name)-1])
  156. extType, err := o.findExtension(extName)
  157. if err != nil && err != protoregistry.NotFound {
  158. return errors.New("unable to resolve [%v]: %v", extName, err)
  159. }
  160. if extType != nil {
  161. fd = extType.Descriptor()
  162. }
  163. } else {
  164. // The name can either be the JSON name or the proto field name.
  165. fd = fieldDescs.ByJSONName(name)
  166. if fd == nil {
  167. fd = fieldDescs.ByName(pref.Name(name))
  168. }
  169. if fd != nil && fd.IsWeak() && fd.Message().IsPlaceholder() {
  170. fd = nil // reset since the weak reference is not linked in
  171. }
  172. }
  173. if fd == nil {
  174. // Field is unknown.
  175. if o.DiscardUnknown {
  176. if err := skipJSONValue(o.decoder); err != nil {
  177. return err
  178. }
  179. continue
  180. }
  181. return newError("%v contains unknown field %s", messageDesc.FullName(), jval)
  182. }
  183. // Do not allow duplicate fields.
  184. num := uint64(fd.Number())
  185. if seenNums.Has(num) {
  186. return newError("%v contains repeated field %s", messageDesc.FullName(), jval)
  187. }
  188. seenNums.Set(num)
  189. // No need to set values for JSON null unless the field type is
  190. // google.protobuf.Value or google.protobuf.NullValue.
  191. if o.decoder.Peek() == json.Null && !isKnownValue(fd) && !isNullValue(fd) {
  192. o.decoder.Read()
  193. continue
  194. }
  195. switch {
  196. case fd.IsList():
  197. list := m.Mutable(fd).List()
  198. if err := o.unmarshalList(list, fd); err != nil {
  199. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  200. }
  201. case fd.IsMap():
  202. mmap := m.Mutable(fd).Map()
  203. if err := o.unmarshalMap(mmap, fd); err != nil {
  204. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  205. }
  206. default:
  207. // If field is a oneof, check if it has already been set.
  208. if od := fd.ContainingOneof(); od != nil {
  209. idx := uint64(od.Index())
  210. if seenOneofs.Has(idx) {
  211. return errors.New("%v: oneof is already set", od.FullName())
  212. }
  213. seenOneofs.Set(idx)
  214. }
  215. // Required or optional fields.
  216. if err := o.unmarshalSingular(m, fd); err != nil {
  217. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  218. }
  219. }
  220. }
  221. }
  222. // findExtension returns protoreflect.ExtensionType from the resolver if found.
  223. func (o UnmarshalOptions) findExtension(xtName pref.FullName) (pref.ExtensionType, error) {
  224. xt, err := o.Resolver.FindExtensionByName(xtName)
  225. if err == nil {
  226. return xt, nil
  227. }
  228. return messageset.FindMessageSetExtension(o.Resolver, xtName)
  229. }
  230. func isKnownValue(fd pref.FieldDescriptor) bool {
  231. md := fd.Message()
  232. return md != nil && md.FullName() == "google.protobuf.Value"
  233. }
  234. func isNullValue(fd pref.FieldDescriptor) bool {
  235. ed := fd.Enum()
  236. return ed != nil && ed.FullName() == "google.protobuf.NullValue"
  237. }
  238. // unmarshalSingular unmarshals to the non-repeated field specified by the given
  239. // FieldDescriptor.
  240. func (o UnmarshalOptions) unmarshalSingular(m pref.Message, fd pref.FieldDescriptor) error {
  241. var val pref.Value
  242. var err error
  243. switch fd.Kind() {
  244. case pref.MessageKind, pref.GroupKind:
  245. m2 := m.NewMessage(fd)
  246. err = o.unmarshalMessage(m2, false)
  247. val = pref.ValueOf(m2)
  248. default:
  249. val, err = o.unmarshalScalar(fd)
  250. }
  251. if err != nil {
  252. return err
  253. }
  254. m.Set(fd, val)
  255. return nil
  256. }
  257. // unmarshalScalar unmarshals to a scalar/enum protoreflect.Value specified by
  258. // the given FieldDescriptor.
  259. func (o UnmarshalOptions) unmarshalScalar(fd pref.FieldDescriptor) (pref.Value, error) {
  260. const b32 int = 32
  261. const b64 int = 64
  262. jval, err := o.decoder.Read()
  263. if err != nil {
  264. return pref.Value{}, err
  265. }
  266. kind := fd.Kind()
  267. switch kind {
  268. case pref.BoolKind:
  269. return unmarshalBool(jval)
  270. case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
  271. return unmarshalInt(jval, b32)
  272. case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
  273. return unmarshalInt(jval, b64)
  274. case pref.Uint32Kind, pref.Fixed32Kind:
  275. return unmarshalUint(jval, b32)
  276. case pref.Uint64Kind, pref.Fixed64Kind:
  277. return unmarshalUint(jval, b64)
  278. case pref.FloatKind:
  279. return unmarshalFloat(jval, b32)
  280. case pref.DoubleKind:
  281. return unmarshalFloat(jval, b64)
  282. case pref.StringKind:
  283. pval, err := unmarshalString(jval)
  284. if err != nil {
  285. return pval, err
  286. }
  287. return pval, nil
  288. case pref.BytesKind:
  289. return unmarshalBytes(jval)
  290. case pref.EnumKind:
  291. return unmarshalEnum(jval, fd)
  292. }
  293. panic(fmt.Sprintf("invalid scalar kind %v", kind))
  294. }
  295. func unmarshalBool(jval json.Value) (pref.Value, error) {
  296. if jval.Type() != json.Bool {
  297. return pref.Value{}, unexpectedJSONError{jval}
  298. }
  299. b, err := jval.Bool()
  300. return pref.ValueOf(b), err
  301. }
  302. func unmarshalInt(jval json.Value, bitSize int) (pref.Value, error) {
  303. switch jval.Type() {
  304. case json.Number:
  305. return getInt(jval, bitSize)
  306. case json.String:
  307. // Decode number from string.
  308. s := strings.TrimSpace(jval.String())
  309. if len(s) != len(jval.String()) {
  310. return pref.Value{}, errors.New("invalid number %v", jval.Raw())
  311. }
  312. dec := json.NewDecoder([]byte(s))
  313. jval, err := dec.Read()
  314. if err != nil {
  315. return pref.Value{}, err
  316. }
  317. return getInt(jval, bitSize)
  318. }
  319. return pref.Value{}, unexpectedJSONError{jval}
  320. }
  321. func getInt(jval json.Value, bitSize int) (pref.Value, error) {
  322. n, err := jval.Int(bitSize)
  323. if err != nil {
  324. return pref.Value{}, err
  325. }
  326. if bitSize == 32 {
  327. return pref.ValueOf(int32(n)), nil
  328. }
  329. return pref.ValueOf(n), nil
  330. }
  331. func unmarshalUint(jval json.Value, bitSize int) (pref.Value, error) {
  332. switch jval.Type() {
  333. case json.Number:
  334. return getUint(jval, bitSize)
  335. case json.String:
  336. // Decode number from string.
  337. s := strings.TrimSpace(jval.String())
  338. if len(s) != len(jval.String()) {
  339. return pref.Value{}, errors.New("invalid number %v", jval.Raw())
  340. }
  341. dec := json.NewDecoder([]byte(s))
  342. jval, err := dec.Read()
  343. if err != nil {
  344. return pref.Value{}, err
  345. }
  346. return getUint(jval, bitSize)
  347. }
  348. return pref.Value{}, unexpectedJSONError{jval}
  349. }
  350. func getUint(jval json.Value, bitSize int) (pref.Value, error) {
  351. n, err := jval.Uint(bitSize)
  352. if err != nil {
  353. return pref.Value{}, err
  354. }
  355. if bitSize == 32 {
  356. return pref.ValueOf(uint32(n)), nil
  357. }
  358. return pref.ValueOf(n), nil
  359. }
  360. func unmarshalFloat(jval json.Value, bitSize int) (pref.Value, error) {
  361. switch jval.Type() {
  362. case json.Number:
  363. return getFloat(jval, bitSize)
  364. case json.String:
  365. s := jval.String()
  366. switch s {
  367. case "NaN":
  368. if bitSize == 32 {
  369. return pref.ValueOf(float32(math.NaN())), nil
  370. }
  371. return pref.ValueOf(math.NaN()), nil
  372. case "Infinity":
  373. if bitSize == 32 {
  374. return pref.ValueOf(float32(math.Inf(+1))), nil
  375. }
  376. return pref.ValueOf(math.Inf(+1)), nil
  377. case "-Infinity":
  378. if bitSize == 32 {
  379. return pref.ValueOf(float32(math.Inf(-1))), nil
  380. }
  381. return pref.ValueOf(math.Inf(-1)), nil
  382. }
  383. // Decode number from string.
  384. if len(s) != len(strings.TrimSpace(s)) {
  385. return pref.Value{}, errors.New("invalid number %v", jval.Raw())
  386. }
  387. dec := json.NewDecoder([]byte(s))
  388. jval, err := dec.Read()
  389. if err != nil {
  390. return pref.Value{}, err
  391. }
  392. return getFloat(jval, bitSize)
  393. }
  394. return pref.Value{}, unexpectedJSONError{jval}
  395. }
  396. func getFloat(jval json.Value, bitSize int) (pref.Value, error) {
  397. n, err := jval.Float(bitSize)
  398. if err != nil {
  399. return pref.Value{}, err
  400. }
  401. if bitSize == 32 {
  402. return pref.ValueOf(float32(n)), nil
  403. }
  404. return pref.ValueOf(n), nil
  405. }
  406. func unmarshalString(jval json.Value) (pref.Value, error) {
  407. if jval.Type() != json.String {
  408. return pref.Value{}, unexpectedJSONError{jval}
  409. }
  410. return pref.ValueOf(jval.String()), nil
  411. }
  412. func unmarshalBytes(jval json.Value) (pref.Value, error) {
  413. if jval.Type() != json.String {
  414. return pref.Value{}, unexpectedJSONError{jval}
  415. }
  416. s := jval.String()
  417. enc := base64.StdEncoding
  418. if strings.ContainsAny(s, "-_") {
  419. enc = base64.URLEncoding
  420. }
  421. if len(s)%4 != 0 {
  422. enc = enc.WithPadding(base64.NoPadding)
  423. }
  424. b, err := enc.DecodeString(s)
  425. if err != nil {
  426. return pref.Value{}, err
  427. }
  428. return pref.ValueOf(b), nil
  429. }
  430. func unmarshalEnum(jval json.Value, fd pref.FieldDescriptor) (pref.Value, error) {
  431. switch jval.Type() {
  432. case json.String:
  433. // Lookup EnumNumber based on name.
  434. s := jval.String()
  435. if enumVal := fd.Enum().Values().ByName(pref.Name(s)); enumVal != nil {
  436. return pref.ValueOf(enumVal.Number()), nil
  437. }
  438. return pref.Value{}, newError("invalid enum value %q", jval)
  439. case json.Number:
  440. n, err := jval.Int(32)
  441. if err != nil {
  442. return pref.Value{}, err
  443. }
  444. return pref.ValueOf(pref.EnumNumber(n)), nil
  445. case json.Null:
  446. // This is only valid for google.protobuf.NullValue.
  447. if isNullValue(fd) {
  448. return pref.ValueOf(pref.EnumNumber(0)), nil
  449. }
  450. }
  451. return pref.Value{}, unexpectedJSONError{jval}
  452. }
  453. func (o UnmarshalOptions) unmarshalList(list pref.List, fd pref.FieldDescriptor) error {
  454. jval, err := o.decoder.Read()
  455. if err != nil {
  456. return err
  457. }
  458. if jval.Type() != json.StartArray {
  459. return unexpectedJSONError{jval}
  460. }
  461. switch fd.Kind() {
  462. case pref.MessageKind, pref.GroupKind:
  463. for {
  464. m := list.NewMessage()
  465. err := o.unmarshalMessage(m, false)
  466. if err != nil {
  467. if e, ok := err.(unexpectedJSONError); ok {
  468. if e.value.Type() == json.EndArray {
  469. // Done with list.
  470. return nil
  471. }
  472. }
  473. return err
  474. }
  475. list.Append(pref.ValueOf(m))
  476. }
  477. default:
  478. for {
  479. val, err := o.unmarshalScalar(fd)
  480. if err != nil {
  481. if e, ok := err.(unexpectedJSONError); ok {
  482. if e.value.Type() == json.EndArray {
  483. // Done with list.
  484. return nil
  485. }
  486. }
  487. return err
  488. }
  489. list.Append(val)
  490. }
  491. }
  492. return nil
  493. }
  494. func (o UnmarshalOptions) unmarshalMap(mmap pref.Map, fd pref.FieldDescriptor) error {
  495. jval, err := o.decoder.Read()
  496. if err != nil {
  497. return err
  498. }
  499. if jval.Type() != json.StartObject {
  500. return unexpectedJSONError{jval}
  501. }
  502. // Determine ahead whether map entry is a scalar type or a message type in
  503. // order to call the appropriate unmarshalMapValue func inside the for loop
  504. // below.
  505. var unmarshalMapValue func() (pref.Value, error)
  506. switch fd.MapValue().Kind() {
  507. case pref.MessageKind, pref.GroupKind:
  508. unmarshalMapValue = func() (pref.Value, error) {
  509. m := mmap.NewMessage()
  510. if err := o.unmarshalMessage(m, false); err != nil {
  511. return pref.Value{}, err
  512. }
  513. return pref.ValueOf(m), nil
  514. }
  515. default:
  516. unmarshalMapValue = func() (pref.Value, error) {
  517. return o.unmarshalScalar(fd.MapValue())
  518. }
  519. }
  520. Loop:
  521. for {
  522. // Read field name.
  523. jval, err := o.decoder.Read()
  524. if err != nil {
  525. return err
  526. }
  527. switch jval.Type() {
  528. default:
  529. return unexpectedJSONError{jval}
  530. case json.EndObject:
  531. break Loop
  532. case json.Name:
  533. // Continue.
  534. }
  535. name, err := jval.Name()
  536. if err != nil {
  537. return err
  538. }
  539. // Unmarshal field name.
  540. pkey, err := unmarshalMapKey(name, fd.MapKey())
  541. if err != nil {
  542. return err
  543. }
  544. // Check for duplicate field name.
  545. if mmap.Has(pkey) {
  546. return newError("duplicate map key %q", jval)
  547. }
  548. // Read and unmarshal field value.
  549. pval, err := unmarshalMapValue()
  550. if err != nil {
  551. return err
  552. }
  553. mmap.Set(pkey, pval)
  554. }
  555. return nil
  556. }
  557. // unmarshalMapKey converts given string into a protoreflect.MapKey. A map key type is any
  558. // integral or string type.
  559. func unmarshalMapKey(name string, fd pref.FieldDescriptor) (pref.MapKey, error) {
  560. const b32 = 32
  561. const b64 = 64
  562. const base10 = 10
  563. kind := fd.Kind()
  564. switch kind {
  565. case pref.StringKind:
  566. return pref.ValueOf(name).MapKey(), nil
  567. case pref.BoolKind:
  568. switch name {
  569. case "true":
  570. return pref.ValueOf(true).MapKey(), nil
  571. case "false":
  572. return pref.ValueOf(false).MapKey(), nil
  573. }
  574. return pref.MapKey{}, errors.New("invalid value for boolean key %q", name)
  575. case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
  576. n, err := strconv.ParseInt(name, base10, b32)
  577. if err != nil {
  578. return pref.MapKey{}, err
  579. }
  580. return pref.ValueOf(int32(n)).MapKey(), nil
  581. case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
  582. n, err := strconv.ParseInt(name, base10, b64)
  583. if err != nil {
  584. return pref.MapKey{}, err
  585. }
  586. return pref.ValueOf(int64(n)).MapKey(), nil
  587. case pref.Uint32Kind, pref.Fixed32Kind:
  588. n, err := strconv.ParseUint(name, base10, b32)
  589. if err != nil {
  590. return pref.MapKey{}, err
  591. }
  592. return pref.ValueOf(uint32(n)).MapKey(), nil
  593. case pref.Uint64Kind, pref.Fixed64Kind:
  594. n, err := strconv.ParseUint(name, base10, b64)
  595. if err != nil {
  596. return pref.MapKey{}, err
  597. }
  598. return pref.ValueOf(uint64(n)).MapKey(), nil
  599. }
  600. panic(fmt.Sprintf("%s: invalid kind %s for map key", fd.FullName(), kind))
  601. }