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 jsonpb
  5. import (
  6. "encoding/base64"
  7. "fmt"
  8. "math"
  9. "strconv"
  10. "strings"
  11. "github.com/golang/protobuf/v2/internal/encoding/json"
  12. "github.com/golang/protobuf/v2/internal/errors"
  13. "github.com/golang/protobuf/v2/internal/pragma"
  14. "github.com/golang/protobuf/v2/internal/set"
  15. "github.com/golang/protobuf/v2/proto"
  16. pref "github.com/golang/protobuf/v2/reflect/protoreflect"
  17. "github.com/golang/protobuf/v2/reflect/protoregistry"
  18. )
  19. // Unmarshal reads the given []byte into the given proto.Message.
  20. func Unmarshal(m proto.Message, b []byte) error {
  21. return UnmarshalOptions{}.Unmarshal(m, b)
  22. }
  23. // UnmarshalOptions is a configurable JSON format parser.
  24. type UnmarshalOptions struct {
  25. pragma.NoUnkeyedLiterals
  26. // Resolver is the registry used for type lookups when unmarshaling extensions
  27. // and processing Any. If Resolver is not set, unmarshaling will default to
  28. // using protoregistry.GlobalTypes.
  29. Resolver *protoregistry.Types
  30. }
  31. // Unmarshal reads the given []byte and populates the given proto.Message using
  32. // options in UnmarshalOptions object. It will clear the message first before
  33. // setting the fields. If it returns an error, the given message may be
  34. // partially set.
  35. func (o UnmarshalOptions) Unmarshal(m proto.Message, b []byte) error {
  36. mr := m.ProtoReflect()
  37. // TODO: Determine if we would like to have an option for merging or only
  38. // have merging behavior. We should at least be consistent with textproto
  39. // marshaling.
  40. resetMessage(mr)
  41. resolver := o.Resolver
  42. if resolver == nil {
  43. resolver = protoregistry.GlobalTypes
  44. }
  45. dec := decoder{
  46. Decoder: json.NewDecoder(b),
  47. resolver: resolver,
  48. }
  49. var nerr errors.NonFatal
  50. if err := dec.unmarshalMessage(mr); !nerr.Merge(err) {
  51. return err
  52. }
  53. // Check for EOF.
  54. val, err := dec.Read()
  55. if err != nil {
  56. return err
  57. }
  58. if val.Type() != json.EOF {
  59. return unexpectedJSONError{val}
  60. }
  61. return nerr.E
  62. }
  63. // resetMessage clears all fields of given protoreflect.Message.
  64. func resetMessage(m pref.Message) {
  65. knownFields := m.KnownFields()
  66. knownFields.Range(func(num pref.FieldNumber, _ pref.Value) bool {
  67. knownFields.Clear(num)
  68. return true
  69. })
  70. unknownFields := m.UnknownFields()
  71. unknownFields.Range(func(num pref.FieldNumber, _ pref.RawFields) bool {
  72. unknownFields.Set(num, nil)
  73. return true
  74. })
  75. extTypes := knownFields.ExtensionTypes()
  76. extTypes.Range(func(xt pref.ExtensionType) bool {
  77. extTypes.Remove(xt)
  78. return true
  79. })
  80. }
  81. // unexpectedJSONError is an error that contains the unexpected json.Value. This
  82. // is used by decoder methods to provide callers the read json.Value that it
  83. // did not expect.
  84. // TODO: Consider moving this to internal/encoding/json for consistency with
  85. // errors that package returns.
  86. type unexpectedJSONError struct {
  87. value json.Value
  88. }
  89. func (e unexpectedJSONError) Error() string {
  90. return newError("unexpected value %s", e.value).Error()
  91. }
  92. // newError returns an error object. If one of the values passed in is of
  93. // json.Value type, it produces an error with position info.
  94. func newError(f string, x ...interface{}) error {
  95. var hasValue bool
  96. var line, column int
  97. for i := 0; i < len(x); i++ {
  98. if val, ok := x[i].(json.Value); ok {
  99. line, column = val.Position()
  100. hasValue = true
  101. break
  102. }
  103. }
  104. e := errors.New(f, x...)
  105. if hasValue {
  106. return errors.New("(line %d:%d): %v", line, column, e)
  107. }
  108. return e
  109. }
  110. // decoder decodes JSON into protoreflect values.
  111. type decoder struct {
  112. *json.Decoder
  113. resolver *protoregistry.Types
  114. }
  115. // unmarshalMessage unmarshals a message into the given protoreflect.Message.
  116. func (d decoder) unmarshalMessage(m pref.Message) error {
  117. var nerr errors.NonFatal
  118. var reqNums set.Ints
  119. var seenNums set.Ints
  120. msgType := m.Type()
  121. knownFields := m.KnownFields()
  122. fieldDescs := msgType.Fields()
  123. xtTypes := knownFields.ExtensionTypes()
  124. jval, err := d.Read()
  125. if !nerr.Merge(err) {
  126. return err
  127. }
  128. if jval.Type() != json.StartObject {
  129. return unexpectedJSONError{jval}
  130. }
  131. Loop:
  132. for {
  133. // Read field name.
  134. jval, err := d.Read()
  135. if !nerr.Merge(err) {
  136. return err
  137. }
  138. switch jval.Type() {
  139. default:
  140. return unexpectedJSONError{jval}
  141. case json.EndObject:
  142. break Loop
  143. case json.Name:
  144. // Continue below.
  145. }
  146. name, err := jval.Name()
  147. if !nerr.Merge(err) {
  148. return err
  149. }
  150. // Get the FieldDescriptor.
  151. var fd pref.FieldDescriptor
  152. if strings.HasPrefix(name, "[") && strings.HasSuffix(name, "]") {
  153. // Only extension names are in [name] format.
  154. xtName := pref.FullName(name[1 : len(name)-1])
  155. xt := xtTypes.ByName(xtName)
  156. if xt == nil {
  157. xt, err = d.findExtension(xtName)
  158. if err != nil && err != protoregistry.NotFound {
  159. return errors.New("unable to resolve [%v]: %v", xtName, err)
  160. }
  161. if xt != nil {
  162. xtTypes.Register(xt)
  163. }
  164. }
  165. fd = xt
  166. } else {
  167. // The name can either be the JSON name or the proto field name.
  168. fd = fieldDescs.ByJSONName(name)
  169. if fd == nil {
  170. fd = fieldDescs.ByName(pref.Name(name))
  171. }
  172. }
  173. if fd == nil {
  174. // Field is unknown.
  175. // TODO: Provide option to ignore unknown message fields.
  176. return newError("%v contains unknown field %s", msgType.FullName(), jval)
  177. }
  178. // Do not allow duplicate fields.
  179. num := uint64(fd.Number())
  180. if seenNums.Has(num) {
  181. return newError("%v contains repeated field %s", msgType.FullName(), jval)
  182. }
  183. seenNums.Set(num)
  184. // No need to set values for JSON null.
  185. if d.Peek() == json.Null {
  186. d.Read()
  187. continue
  188. }
  189. if cardinality := fd.Cardinality(); cardinality == pref.Repeated {
  190. // Map or list fields have cardinality of repeated.
  191. if err := d.unmarshalRepeated(fd, knownFields); !nerr.Merge(err) {
  192. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  193. }
  194. } else {
  195. // Required or optional fields.
  196. if err := d.unmarshalSingular(fd, knownFields); !nerr.Merge(err) {
  197. return errors.New("%v|%q: %v", fd.FullName(), name, err)
  198. }
  199. if cardinality == pref.Required {
  200. reqNums.Set(num)
  201. }
  202. }
  203. }
  204. // Check for any missing required fields.
  205. allReqNums := msgType.RequiredNumbers()
  206. if reqNums.Len() != allReqNums.Len() {
  207. for i := 0; i < allReqNums.Len(); i++ {
  208. if num := allReqNums.Get(i); !reqNums.Has(uint64(num)) {
  209. nerr.AppendRequiredNotSet(string(fieldDescs.ByNumber(num).FullName()))
  210. }
  211. }
  212. }
  213. return nerr.E
  214. }
  215. // findExtension returns protoreflect.ExtensionType from the resolver if found.
  216. func (d decoder) findExtension(xtName pref.FullName) (pref.ExtensionType, error) {
  217. xt, err := d.resolver.FindExtensionByName(xtName)
  218. if err == nil {
  219. return xt, nil
  220. }
  221. // Check if this is a MessageSet extension field.
  222. xt, err = d.resolver.FindExtensionByName(xtName + ".message_set_extension")
  223. if err == nil && isMessageSetExtension(xt) {
  224. return xt, nil
  225. }
  226. return nil, protoregistry.NotFound
  227. }
  228. // unmarshalSingular unmarshals to the non-repeated field specified by the given
  229. // FieldDescriptor.
  230. func (d decoder) unmarshalSingular(fd pref.FieldDescriptor, knownFields pref.KnownFields) error {
  231. var val pref.Value
  232. var err error
  233. num := fd.Number()
  234. switch fd.Kind() {
  235. case pref.MessageKind, pref.GroupKind:
  236. m := knownFields.NewMessage(num)
  237. err = d.unmarshalMessage(m)
  238. val = pref.ValueOf(m)
  239. default:
  240. val, err = d.unmarshalScalar(fd)
  241. }
  242. var nerr errors.NonFatal
  243. if !nerr.Merge(err) {
  244. return err
  245. }
  246. knownFields.Set(num, val)
  247. return nerr.E
  248. }
  249. // unmarshalScalar unmarshals to a scalar/enum protoreflect.Value specified by
  250. // the given FieldDescriptor.
  251. func (d decoder) unmarshalScalar(fd pref.FieldDescriptor) (pref.Value, error) {
  252. const b32 int = 32
  253. const b64 int = 64
  254. var nerr errors.NonFatal
  255. jval, err := d.Read()
  256. if !nerr.Merge(err) {
  257. return pref.Value{}, err
  258. }
  259. kind := fd.Kind()
  260. switch kind {
  261. case pref.BoolKind:
  262. return unmarshalBool(jval)
  263. case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
  264. return unmarshalInt(jval, b32)
  265. case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
  266. return unmarshalInt(jval, b64)
  267. case pref.Uint32Kind, pref.Fixed32Kind:
  268. return unmarshalUint(jval, b32)
  269. case pref.Uint64Kind, pref.Fixed64Kind:
  270. return unmarshalUint(jval, b64)
  271. case pref.FloatKind:
  272. return unmarshalFloat(jval, b32)
  273. case pref.DoubleKind:
  274. return unmarshalFloat(jval, b64)
  275. case pref.StringKind:
  276. pval, err := unmarshalString(jval)
  277. if !nerr.Merge(err) {
  278. return pval, err
  279. }
  280. return pval, nerr.E
  281. case pref.BytesKind:
  282. return unmarshalBytes(jval)
  283. case pref.EnumKind:
  284. return unmarshalEnum(jval, fd)
  285. }
  286. panic(fmt.Sprintf("invalid scalar kind %v", kind))
  287. }
  288. func unmarshalBool(jval json.Value) (pref.Value, error) {
  289. if jval.Type() != json.Bool {
  290. return pref.Value{}, unexpectedJSONError{jval}
  291. }
  292. b, err := jval.Bool()
  293. return pref.ValueOf(b), err
  294. }
  295. func unmarshalInt(jval json.Value, bitSize int) (pref.Value, error) {
  296. switch jval.Type() {
  297. case json.Number:
  298. return getInt(jval, bitSize)
  299. case json.String:
  300. // Decode number from string.
  301. dec := json.NewDecoder([]byte(jval.String()))
  302. var nerr errors.NonFatal
  303. jval, err := dec.Read()
  304. if !nerr.Merge(err) {
  305. return pref.Value{}, err
  306. }
  307. return getInt(jval, bitSize)
  308. }
  309. return pref.Value{}, unexpectedJSONError{jval}
  310. }
  311. func getInt(jval json.Value, bitSize int) (pref.Value, error) {
  312. n, err := jval.Int(bitSize)
  313. if err != nil {
  314. return pref.Value{}, err
  315. }
  316. if bitSize == 32 {
  317. return pref.ValueOf(int32(n)), nil
  318. }
  319. return pref.ValueOf(n), nil
  320. }
  321. func unmarshalUint(jval json.Value, bitSize int) (pref.Value, error) {
  322. switch jval.Type() {
  323. case json.Number:
  324. return getUint(jval, bitSize)
  325. case json.String:
  326. // Decode number from string.
  327. dec := json.NewDecoder([]byte(jval.String()))
  328. var nerr errors.NonFatal
  329. jval, err := dec.Read()
  330. if !nerr.Merge(err) {
  331. return pref.Value{}, err
  332. }
  333. return getUint(jval, bitSize)
  334. }
  335. return pref.Value{}, unexpectedJSONError{jval}
  336. }
  337. func getUint(jval json.Value, bitSize int) (pref.Value, error) {
  338. n, err := jval.Uint(bitSize)
  339. if err != nil {
  340. return pref.Value{}, err
  341. }
  342. if bitSize == 32 {
  343. return pref.ValueOf(uint32(n)), nil
  344. }
  345. return pref.ValueOf(n), nil
  346. }
  347. func unmarshalFloat(jval json.Value, bitSize int) (pref.Value, error) {
  348. switch jval.Type() {
  349. case json.Number:
  350. return getFloat(jval, bitSize)
  351. case json.String:
  352. s := jval.String()
  353. switch s {
  354. case "NaN":
  355. if bitSize == 32 {
  356. return pref.ValueOf(float32(math.NaN())), nil
  357. }
  358. return pref.ValueOf(math.NaN()), nil
  359. case "Infinity":
  360. if bitSize == 32 {
  361. return pref.ValueOf(float32(math.Inf(+1))), nil
  362. }
  363. return pref.ValueOf(math.Inf(+1)), nil
  364. case "-Infinity":
  365. if bitSize == 32 {
  366. return pref.ValueOf(float32(math.Inf(-1))), nil
  367. }
  368. return pref.ValueOf(math.Inf(-1)), nil
  369. }
  370. // Decode number from string.
  371. dec := json.NewDecoder([]byte(s))
  372. var nerr errors.NonFatal
  373. jval, err := dec.Read()
  374. if !nerr.Merge(err) {
  375. return pref.Value{}, err
  376. }
  377. return getFloat(jval, bitSize)
  378. }
  379. return pref.Value{}, unexpectedJSONError{jval}
  380. }
  381. func getFloat(jval json.Value, bitSize int) (pref.Value, error) {
  382. n, err := jval.Float(bitSize)
  383. if err != nil {
  384. return pref.Value{}, err
  385. }
  386. if bitSize == 32 {
  387. return pref.ValueOf(float32(n)), nil
  388. }
  389. return pref.ValueOf(n), nil
  390. }
  391. func unmarshalString(jval json.Value) (pref.Value, error) {
  392. if jval.Type() != json.String {
  393. return pref.Value{}, unexpectedJSONError{jval}
  394. }
  395. return pref.ValueOf(jval.String()), nil
  396. }
  397. func unmarshalBytes(jval json.Value) (pref.Value, error) {
  398. if jval.Type() != json.String {
  399. return pref.Value{}, unexpectedJSONError{jval}
  400. }
  401. s := jval.String()
  402. enc := base64.StdEncoding
  403. if strings.ContainsAny(s, "-_") {
  404. enc = base64.URLEncoding
  405. }
  406. if len(s)%4 != 0 {
  407. enc = enc.WithPadding(base64.NoPadding)
  408. }
  409. b, err := enc.DecodeString(s)
  410. if err != nil {
  411. return pref.Value{}, err
  412. }
  413. return pref.ValueOf(b), nil
  414. }
  415. func unmarshalEnum(jval json.Value, fd pref.FieldDescriptor) (pref.Value, error) {
  416. switch jval.Type() {
  417. case json.String:
  418. // Lookup EnumNumber based on name.
  419. s := jval.String()
  420. if enumVal := fd.EnumType().Values().ByName(pref.Name(s)); enumVal != nil {
  421. return pref.ValueOf(enumVal.Number()), nil
  422. }
  423. return pref.Value{}, newError("invalid enum value %q", jval)
  424. case json.Number:
  425. n, err := jval.Int(32)
  426. if err != nil {
  427. return pref.Value{}, err
  428. }
  429. return pref.ValueOf(pref.EnumNumber(n)), nil
  430. }
  431. return pref.Value{}, unexpectedJSONError{jval}
  432. }
  433. // unmarshalRepeated unmarshals into a repeated field.
  434. func (d decoder) unmarshalRepeated(fd pref.FieldDescriptor, knownFields pref.KnownFields) error {
  435. var nerr errors.NonFatal
  436. num := fd.Number()
  437. val := knownFields.Get(num)
  438. if !fd.IsMap() {
  439. if err := d.unmarshalList(fd, val.List()); !nerr.Merge(err) {
  440. return err
  441. }
  442. } else {
  443. if err := d.unmarshalMap(fd, val.Map()); !nerr.Merge(err) {
  444. return err
  445. }
  446. }
  447. return nerr.E
  448. }
  449. // unmarshalList unmarshals into given protoreflect.List.
  450. func (d decoder) unmarshalList(fd pref.FieldDescriptor, list pref.List) error {
  451. var nerr errors.NonFatal
  452. jval, err := d.Read()
  453. if !nerr.Merge(err) {
  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 := d.unmarshalMessage(m)
  464. if !nerr.Merge(err) {
  465. if e, ok := err.(unexpectedJSONError); ok {
  466. if e.value.Type() == json.EndArray {
  467. // Done with list.
  468. return nerr.E
  469. }
  470. }
  471. return err
  472. }
  473. list.Append(pref.ValueOf(m))
  474. }
  475. default:
  476. for {
  477. val, err := d.unmarshalScalar(fd)
  478. if !nerr.Merge(err) {
  479. if e, ok := err.(unexpectedJSONError); ok {
  480. if e.value.Type() == json.EndArray {
  481. // Done with list.
  482. return nerr.E
  483. }
  484. }
  485. return err
  486. }
  487. list.Append(val)
  488. }
  489. }
  490. return nerr.E
  491. }
  492. // unmarshalMap unmarshals into given protoreflect.Map.
  493. func (d decoder) unmarshalMap(fd pref.FieldDescriptor, mmap pref.Map) error {
  494. var nerr errors.NonFatal
  495. jval, err := d.Read()
  496. if !nerr.Merge(err) {
  497. return err
  498. }
  499. if jval.Type() != json.StartObject {
  500. return unexpectedJSONError{jval}
  501. }
  502. fields := fd.MessageType().Fields()
  503. keyDesc := fields.ByNumber(1)
  504. valDesc := fields.ByNumber(2)
  505. // Determine ahead whether map entry is a scalar type or a message type in
  506. // order to call the appropriate unmarshalMapValue func inside the for loop
  507. // below.
  508. unmarshalMapValue := func() (pref.Value, error) {
  509. return d.unmarshalScalar(valDesc)
  510. }
  511. switch valDesc.Kind() {
  512. case pref.MessageKind, pref.GroupKind:
  513. unmarshalMapValue = func() (pref.Value, error) {
  514. m := mmap.NewMessage()
  515. if err := d.unmarshalMessage(m); err != nil {
  516. return pref.Value{}, err
  517. }
  518. return pref.ValueOf(m), nil
  519. }
  520. }
  521. Loop:
  522. for {
  523. // Read field name.
  524. jval, err := d.Read()
  525. if !nerr.Merge(err) {
  526. return err
  527. }
  528. switch jval.Type() {
  529. default:
  530. return unexpectedJSONError{jval}
  531. case json.EndObject:
  532. break Loop
  533. case json.Name:
  534. // Continue.
  535. }
  536. name, err := jval.Name()
  537. if !nerr.Merge(err) {
  538. return err
  539. }
  540. // Unmarshal field name.
  541. pkey, err := unmarshalMapKey(name, keyDesc)
  542. if !nerr.Merge(err) {
  543. return err
  544. }
  545. // Check for duplicate field name.
  546. if mmap.Has(pkey) {
  547. return newError("duplicate map key %q", jval)
  548. }
  549. // Read and unmarshal field value.
  550. pval, err := unmarshalMapValue()
  551. if !nerr.Merge(err) {
  552. return err
  553. }
  554. mmap.Set(pkey, pval)
  555. }
  556. return nerr.E
  557. }
  558. // unmarshalMapKey converts given string into a protoreflect.MapKey. A map key type is any
  559. // integral or string type.
  560. func unmarshalMapKey(name string, fd pref.FieldDescriptor) (pref.MapKey, error) {
  561. const b32 = 32
  562. const b64 = 64
  563. const base10 = 10
  564. kind := fd.Kind()
  565. switch kind {
  566. case pref.StringKind:
  567. return pref.ValueOf(name).MapKey(), nil
  568. case pref.BoolKind:
  569. switch name {
  570. case "true":
  571. return pref.ValueOf(true).MapKey(), nil
  572. case "false":
  573. return pref.ValueOf(false).MapKey(), nil
  574. }
  575. return pref.MapKey{}, errors.New("invalid value for boolean key %q", name)
  576. case pref.Int32Kind, pref.Sint32Kind, pref.Sfixed32Kind:
  577. n, err := strconv.ParseInt(name, base10, b32)
  578. if err != nil {
  579. return pref.MapKey{}, err
  580. }
  581. return pref.ValueOf(int32(n)).MapKey(), nil
  582. case pref.Int64Kind, pref.Sint64Kind, pref.Sfixed64Kind:
  583. n, err := strconv.ParseInt(name, base10, b64)
  584. if err != nil {
  585. return pref.MapKey{}, err
  586. }
  587. return pref.ValueOf(int64(n)).MapKey(), nil
  588. case pref.Uint32Kind, pref.Fixed32Kind:
  589. n, err := strconv.ParseUint(name, base10, b32)
  590. if err != nil {
  591. return pref.MapKey{}, err
  592. }
  593. return pref.ValueOf(uint32(n)).MapKey(), nil
  594. case pref.Uint64Kind, pref.Fixed64Kind:
  595. n, err := strconv.ParseUint(name, base10, b64)
  596. if err != nil {
  597. return pref.MapKey{}, err
  598. }
  599. return pref.ValueOf(uint64(n)).MapKey(), nil
  600. }
  601. panic(fmt.Sprintf("%s: invalid kind %s for map key", fd.FullName(), kind))
  602. }