decode.go 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683
  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.Proto1Legacy && 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. fd = extType
  161. } else {
  162. // The name can either be the JSON name or the proto field name.
  163. fd = fieldDescs.ByJSONName(name)
  164. if fd == nil {
  165. fd = fieldDescs.ByName(pref.Name(name))
  166. }
  167. if fd != nil && fd.IsWeak() && fd.Message().IsPlaceholder() {
  168. fd = nil // reset since the weak reference is not linked in
  169. }
  170. }
  171. if fd == nil {
  172. // Field is unknown.
  173. if o.DiscardUnknown {
  174. if err := skipJSONValue(o.decoder); err != nil {
  175. return err
  176. }
  177. continue
  178. }
  179. return newError("%v contains unknown field %s", messageDesc.FullName(), jval)
  180. }
  181. // Do not allow duplicate fields.
  182. num := uint64(fd.Number())
  183. if seenNums.Has(num) {
  184. return newError("%v contains repeated field %s", messageDesc.FullName(), jval)
  185. }
  186. seenNums.Set(num)
  187. // No need to set values for JSON null unless the field type is
  188. // google.protobuf.Value or google.protobuf.NullValue.
  189. if o.decoder.Peek() == json.Null && !isKnownValue(fd) && !isNullValue(fd) {
  190. o.decoder.Read()
  191. continue
  192. }
  193. switch {
  194. case fd.IsList():
  195. list := m.Mutable(fd).List()
  196. if err := o.unmarshalList(list, fd); err != nil {
  197. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  198. }
  199. case fd.IsMap():
  200. mmap := m.Mutable(fd).Map()
  201. if err := o.unmarshalMap(mmap, fd); err != nil {
  202. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  203. }
  204. default:
  205. // If field is a oneof, check if it has already been set.
  206. if od := fd.ContainingOneof(); od != nil {
  207. idx := uint64(od.Index())
  208. if seenOneofs.Has(idx) {
  209. return errors.New("%v: oneof is already set", od.FullName())
  210. }
  211. seenOneofs.Set(idx)
  212. }
  213. // Required or optional fields.
  214. if err := o.unmarshalSingular(m, fd); err != nil {
  215. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  216. }
  217. }
  218. }
  219. }
  220. // findExtension returns protoreflect.ExtensionType from the resolver if found.
  221. func (o UnmarshalOptions) findExtension(xtName pref.FullName) (pref.ExtensionType, error) {
  222. xt, err := o.Resolver.FindExtensionByName(xtName)
  223. if err == nil {
  224. return xt, nil
  225. }
  226. return messageset.FindMessageSetExtension(o.Resolver, xtName)
  227. }
  228. func isKnownValue(fd pref.FieldDescriptor) bool {
  229. md := fd.Message()
  230. return md != nil && md.FullName() == "google.protobuf.Value"
  231. }
  232. func isNullValue(fd pref.FieldDescriptor) bool {
  233. ed := fd.Enum()
  234. return ed != nil && ed.FullName() == "google.protobuf.NullValue"
  235. }
  236. // unmarshalSingular unmarshals to the non-repeated field specified by the given
  237. // FieldDescriptor.
  238. func (o UnmarshalOptions) unmarshalSingular(m pref.Message, fd pref.FieldDescriptor) error {
  239. var val pref.Value
  240. var err error
  241. switch fd.Kind() {
  242. case pref.MessageKind, pref.GroupKind:
  243. m2 := m.NewMessage(fd)
  244. err = o.unmarshalMessage(m2, false)
  245. val = pref.ValueOf(m2)
  246. default:
  247. val, err = o.unmarshalScalar(fd)
  248. }
  249. if err != nil {
  250. return err
  251. }
  252. m.Set(fd, val)
  253. return nil
  254. }
  255. // unmarshalScalar unmarshals to a scalar/enum protoreflect.Value specified by
  256. // the given FieldDescriptor.
  257. func (o UnmarshalOptions) unmarshalScalar(fd pref.FieldDescriptor) (pref.Value, error) {
  258. const b32 int = 32
  259. const b64 int = 64
  260. jval, err := o.decoder.Read()
  261. if err != nil {
  262. return pref.Value{}, err
  263. }
  264. kind := fd.Kind()
  265. switch kind {
  266. case pref.BoolKind:
  267. return unmarshalBool(jval)
  268. case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
  269. return unmarshalInt(jval, b32)
  270. case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
  271. return unmarshalInt(jval, b64)
  272. case pref.Uint32Kind, pref.Fixed32Kind:
  273. return unmarshalUint(jval, b32)
  274. case pref.Uint64Kind, pref.Fixed64Kind:
  275. return unmarshalUint(jval, b64)
  276. case pref.FloatKind:
  277. return unmarshalFloat(jval, b32)
  278. case pref.DoubleKind:
  279. return unmarshalFloat(jval, b64)
  280. case pref.StringKind:
  281. pval, err := unmarshalString(jval)
  282. if err != nil {
  283. return pval, err
  284. }
  285. return pval, nil
  286. case pref.BytesKind:
  287. return unmarshalBytes(jval)
  288. case pref.EnumKind:
  289. return unmarshalEnum(jval, fd)
  290. }
  291. panic(fmt.Sprintf("invalid scalar kind %v", kind))
  292. }
  293. func unmarshalBool(jval json.Value) (pref.Value, error) {
  294. if jval.Type() != json.Bool {
  295. return pref.Value{}, unexpectedJSONError{jval}
  296. }
  297. b, err := jval.Bool()
  298. return pref.ValueOf(b), err
  299. }
  300. func unmarshalInt(jval json.Value, bitSize int) (pref.Value, error) {
  301. switch jval.Type() {
  302. case json.Number:
  303. return getInt(jval, bitSize)
  304. case json.String:
  305. // Decode number from string.
  306. s := strings.TrimSpace(jval.String())
  307. if len(s) != len(jval.String()) {
  308. return pref.Value{}, errors.New("invalid number %v", jval.Raw())
  309. }
  310. dec := json.NewDecoder([]byte(s))
  311. jval, err := dec.Read()
  312. if err != nil {
  313. return pref.Value{}, err
  314. }
  315. return getInt(jval, bitSize)
  316. }
  317. return pref.Value{}, unexpectedJSONError{jval}
  318. }
  319. func getInt(jval json.Value, bitSize int) (pref.Value, error) {
  320. n, err := jval.Int(bitSize)
  321. if err != nil {
  322. return pref.Value{}, err
  323. }
  324. if bitSize == 32 {
  325. return pref.ValueOf(int32(n)), nil
  326. }
  327. return pref.ValueOf(n), nil
  328. }
  329. func unmarshalUint(jval json.Value, bitSize int) (pref.Value, error) {
  330. switch jval.Type() {
  331. case json.Number:
  332. return getUint(jval, bitSize)
  333. case json.String:
  334. // Decode number from string.
  335. s := strings.TrimSpace(jval.String())
  336. if len(s) != len(jval.String()) {
  337. return pref.Value{}, errors.New("invalid number %v", jval.Raw())
  338. }
  339. dec := json.NewDecoder([]byte(s))
  340. jval, err := dec.Read()
  341. if err != nil {
  342. return pref.Value{}, err
  343. }
  344. return getUint(jval, bitSize)
  345. }
  346. return pref.Value{}, unexpectedJSONError{jval}
  347. }
  348. func getUint(jval json.Value, bitSize int) (pref.Value, error) {
  349. n, err := jval.Uint(bitSize)
  350. if err != nil {
  351. return pref.Value{}, err
  352. }
  353. if bitSize == 32 {
  354. return pref.ValueOf(uint32(n)), nil
  355. }
  356. return pref.ValueOf(n), nil
  357. }
  358. func unmarshalFloat(jval json.Value, bitSize int) (pref.Value, error) {
  359. switch jval.Type() {
  360. case json.Number:
  361. return getFloat(jval, bitSize)
  362. case json.String:
  363. s := jval.String()
  364. switch s {
  365. case "NaN":
  366. if bitSize == 32 {
  367. return pref.ValueOf(float32(math.NaN())), nil
  368. }
  369. return pref.ValueOf(math.NaN()), nil
  370. case "Infinity":
  371. if bitSize == 32 {
  372. return pref.ValueOf(float32(math.Inf(+1))), nil
  373. }
  374. return pref.ValueOf(math.Inf(+1)), nil
  375. case "-Infinity":
  376. if bitSize == 32 {
  377. return pref.ValueOf(float32(math.Inf(-1))), nil
  378. }
  379. return pref.ValueOf(math.Inf(-1)), nil
  380. }
  381. // Decode number from string.
  382. if len(s) != len(strings.TrimSpace(s)) {
  383. return pref.Value{}, errors.New("invalid number %v", jval.Raw())
  384. }
  385. dec := json.NewDecoder([]byte(s))
  386. jval, err := dec.Read()
  387. if err != nil {
  388. return pref.Value{}, err
  389. }
  390. return getFloat(jval, bitSize)
  391. }
  392. return pref.Value{}, unexpectedJSONError{jval}
  393. }
  394. func getFloat(jval json.Value, bitSize int) (pref.Value, error) {
  395. n, err := jval.Float(bitSize)
  396. if err != nil {
  397. return pref.Value{}, err
  398. }
  399. if bitSize == 32 {
  400. return pref.ValueOf(float32(n)), nil
  401. }
  402. return pref.ValueOf(n), nil
  403. }
  404. func unmarshalString(jval json.Value) (pref.Value, error) {
  405. if jval.Type() != json.String {
  406. return pref.Value{}, unexpectedJSONError{jval}
  407. }
  408. return pref.ValueOf(jval.String()), nil
  409. }
  410. func unmarshalBytes(jval json.Value) (pref.Value, error) {
  411. if jval.Type() != json.String {
  412. return pref.Value{}, unexpectedJSONError{jval}
  413. }
  414. s := jval.String()
  415. enc := base64.StdEncoding
  416. if strings.ContainsAny(s, "-_") {
  417. enc = base64.URLEncoding
  418. }
  419. if len(s)%4 != 0 {
  420. enc = enc.WithPadding(base64.NoPadding)
  421. }
  422. b, err := enc.DecodeString(s)
  423. if err != nil {
  424. return pref.Value{}, err
  425. }
  426. return pref.ValueOf(b), nil
  427. }
  428. func unmarshalEnum(jval json.Value, fd pref.FieldDescriptor) (pref.Value, error) {
  429. switch jval.Type() {
  430. case json.String:
  431. // Lookup EnumNumber based on name.
  432. s := jval.String()
  433. if enumVal := fd.Enum().Values().ByName(pref.Name(s)); enumVal != nil {
  434. return pref.ValueOf(enumVal.Number()), nil
  435. }
  436. return pref.Value{}, newError("invalid enum value %q", jval)
  437. case json.Number:
  438. n, err := jval.Int(32)
  439. if err != nil {
  440. return pref.Value{}, err
  441. }
  442. return pref.ValueOf(pref.EnumNumber(n)), nil
  443. case json.Null:
  444. // This is only valid for google.protobuf.NullValue.
  445. if isNullValue(fd) {
  446. return pref.ValueOf(pref.EnumNumber(0)), nil
  447. }
  448. }
  449. return pref.Value{}, unexpectedJSONError{jval}
  450. }
  451. func (o UnmarshalOptions) unmarshalList(list pref.List, fd pref.FieldDescriptor) error {
  452. jval, err := o.decoder.Read()
  453. if err != nil {
  454. return err
  455. }
  456. if jval.Type() != json.StartArray {
  457. return unexpectedJSONError{jval}
  458. }
  459. switch fd.Kind() {
  460. case pref.MessageKind, pref.GroupKind:
  461. for {
  462. m := list.NewMessage()
  463. err := o.unmarshalMessage(m, false)
  464. if err != nil {
  465. if e, ok := err.(unexpectedJSONError); ok {
  466. if e.value.Type() == json.EndArray {
  467. // Done with list.
  468. return nil
  469. }
  470. }
  471. return err
  472. }
  473. list.Append(pref.ValueOf(m))
  474. }
  475. default:
  476. for {
  477. val, err := o.unmarshalScalar(fd)
  478. if err != nil {
  479. if e, ok := err.(unexpectedJSONError); ok {
  480. if e.value.Type() == json.EndArray {
  481. // Done with list.
  482. return nil
  483. }
  484. }
  485. return err
  486. }
  487. list.Append(val)
  488. }
  489. }
  490. return nil
  491. }
  492. func (o UnmarshalOptions) unmarshalMap(mmap pref.Map, fd pref.FieldDescriptor) error {
  493. jval, err := o.decoder.Read()
  494. if err != nil {
  495. return err
  496. }
  497. if jval.Type() != json.StartObject {
  498. return unexpectedJSONError{jval}
  499. }
  500. // Determine ahead whether map entry is a scalar type or a message type in
  501. // order to call the appropriate unmarshalMapValue func inside the for loop
  502. // below.
  503. var unmarshalMapValue func() (pref.Value, error)
  504. switch fd.MapValue().Kind() {
  505. case pref.MessageKind, pref.GroupKind:
  506. unmarshalMapValue = func() (pref.Value, error) {
  507. m := mmap.NewMessage()
  508. if err := o.unmarshalMessage(m, false); err != nil {
  509. return pref.Value{}, err
  510. }
  511. return pref.ValueOf(m), nil
  512. }
  513. default:
  514. unmarshalMapValue = func() (pref.Value, error) {
  515. return o.unmarshalScalar(fd.MapValue())
  516. }
  517. }
  518. Loop:
  519. for {
  520. // Read field name.
  521. jval, err := o.decoder.Read()
  522. if err != nil {
  523. return err
  524. }
  525. switch jval.Type() {
  526. default:
  527. return unexpectedJSONError{jval}
  528. case json.EndObject:
  529. break Loop
  530. case json.Name:
  531. // Continue.
  532. }
  533. name, err := jval.Name()
  534. if err != nil {
  535. return err
  536. }
  537. // Unmarshal field name.
  538. pkey, err := unmarshalMapKey(name, fd.MapKey())
  539. if err != nil {
  540. return err
  541. }
  542. // Check for duplicate field name.
  543. if mmap.Has(pkey) {
  544. return newError("duplicate map key %q", jval)
  545. }
  546. // Read and unmarshal field value.
  547. pval, err := unmarshalMapValue()
  548. if err != nil {
  549. return err
  550. }
  551. mmap.Set(pkey, pval)
  552. }
  553. return nil
  554. }
  555. // unmarshalMapKey converts given string into a protoreflect.MapKey. A map key type is any
  556. // integral or string type.
  557. func unmarshalMapKey(name string, fd pref.FieldDescriptor) (pref.MapKey, error) {
  558. const b32 = 32
  559. const b64 = 64
  560. const base10 = 10
  561. kind := fd.Kind()
  562. switch kind {
  563. case pref.StringKind:
  564. return pref.ValueOf(name).MapKey(), nil
  565. case pref.BoolKind:
  566. switch name {
  567. case "true":
  568. return pref.ValueOf(true).MapKey(), nil
  569. case "false":
  570. return pref.ValueOf(false).MapKey(), nil
  571. }
  572. return pref.MapKey{}, errors.New("invalid value for boolean key %q", name)
  573. case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
  574. n, err := strconv.ParseInt(name, base10, b32)
  575. if err != nil {
  576. return pref.MapKey{}, err
  577. }
  578. return pref.ValueOf(int32(n)).MapKey(), nil
  579. case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
  580. n, err := strconv.ParseInt(name, base10, b64)
  581. if err != nil {
  582. return pref.MapKey{}, err
  583. }
  584. return pref.ValueOf(int64(n)).MapKey(), nil
  585. case pref.Uint32Kind, pref.Fixed32Kind:
  586. n, err := strconv.ParseUint(name, base10, b32)
  587. if err != nil {
  588. return pref.MapKey{}, err
  589. }
  590. return pref.ValueOf(uint32(n)).MapKey(), nil
  591. case pref.Uint64Kind, pref.Fixed64Kind:
  592. n, err := strconv.ParseUint(name, base10, b64)
  593. if err != nil {
  594. return pref.MapKey{}, err
  595. }
  596. return pref.ValueOf(uint64(n)).MapKey(), nil
  597. }
  598. panic(fmt.Sprintf("%s: invalid kind %s for map key", fd.FullName(), kind))
  599. }