decode.go 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566
  1. package yaml
  2. import (
  3. "encoding/base64"
  4. "fmt"
  5. "reflect"
  6. "strconv"
  7. "time"
  8. )
  9. const (
  10. documentNode = 1 << iota
  11. mappingNode
  12. sequenceNode
  13. scalarNode
  14. aliasNode
  15. )
  16. type node struct {
  17. kind int
  18. line, column int
  19. tag string
  20. value string
  21. implicit bool
  22. children []*node
  23. anchors map[string]*node
  24. }
  25. // ----------------------------------------------------------------------------
  26. // Parser, produces a node tree out of a libyaml event stream.
  27. type parser struct {
  28. parser yaml_parser_t
  29. event yaml_event_t
  30. doc *node
  31. }
  32. func newParser(b []byte) *parser {
  33. p := parser{}
  34. if !yaml_parser_initialize(&p.parser) {
  35. panic("Failed to initialize YAML emitter")
  36. }
  37. if len(b) == 0 {
  38. b = []byte{'\n'}
  39. }
  40. yaml_parser_set_input_string(&p.parser, b)
  41. p.skip()
  42. if p.event.typ != yaml_STREAM_START_EVENT {
  43. panic("Expected stream start event, got " + strconv.Itoa(int(p.event.typ)))
  44. }
  45. p.skip()
  46. return &p
  47. }
  48. func (p *parser) destroy() {
  49. if p.event.typ != yaml_NO_EVENT {
  50. yaml_event_delete(&p.event)
  51. }
  52. yaml_parser_delete(&p.parser)
  53. }
  54. func (p *parser) skip() {
  55. if p.event.typ != yaml_NO_EVENT {
  56. if p.event.typ == yaml_STREAM_END_EVENT {
  57. fail("Attempted to go past the end of stream. Corrupted value?")
  58. }
  59. yaml_event_delete(&p.event)
  60. }
  61. if !yaml_parser_parse(&p.parser, &p.event) {
  62. p.fail()
  63. }
  64. }
  65. func (p *parser) fail() {
  66. var where string
  67. var line int
  68. if p.parser.problem_mark.line != 0 {
  69. line = p.parser.problem_mark.line
  70. } else if p.parser.context_mark.line != 0 {
  71. line = p.parser.context_mark.line
  72. }
  73. if line != 0 {
  74. where = "line " + strconv.Itoa(line) + ": "
  75. }
  76. var msg string
  77. if len(p.parser.problem) > 0 {
  78. msg = p.parser.problem
  79. } else {
  80. msg = "Unknown problem parsing YAML content"
  81. }
  82. fail(where + msg)
  83. }
  84. func (p *parser) anchor(n *node, anchor []byte) {
  85. if anchor != nil {
  86. p.doc.anchors[string(anchor)] = n
  87. }
  88. }
  89. func (p *parser) parse() *node {
  90. switch p.event.typ {
  91. case yaml_SCALAR_EVENT:
  92. return p.scalar()
  93. case yaml_ALIAS_EVENT:
  94. return p.alias()
  95. case yaml_MAPPING_START_EVENT:
  96. return p.mapping()
  97. case yaml_SEQUENCE_START_EVENT:
  98. return p.sequence()
  99. case yaml_DOCUMENT_START_EVENT:
  100. return p.document()
  101. case yaml_STREAM_END_EVENT:
  102. // Happens when attempting to decode an empty buffer.
  103. return nil
  104. default:
  105. panic("Attempted to parse unknown event: " + strconv.Itoa(int(p.event.typ)))
  106. }
  107. panic("unreachable")
  108. }
  109. func (p *parser) node(kind int) *node {
  110. return &node{
  111. kind: kind,
  112. line: p.event.start_mark.line,
  113. column: p.event.start_mark.column,
  114. }
  115. }
  116. func (p *parser) document() *node {
  117. n := p.node(documentNode)
  118. n.anchors = make(map[string]*node)
  119. p.doc = n
  120. p.skip()
  121. n.children = append(n.children, p.parse())
  122. if p.event.typ != yaml_DOCUMENT_END_EVENT {
  123. panic("Expected end of document event but got " + strconv.Itoa(int(p.event.typ)))
  124. }
  125. p.skip()
  126. return n
  127. }
  128. func (p *parser) alias() *node {
  129. n := p.node(aliasNode)
  130. n.value = string(p.event.anchor)
  131. p.skip()
  132. return n
  133. }
  134. func (p *parser) scalar() *node {
  135. n := p.node(scalarNode)
  136. n.value = string(p.event.value)
  137. n.tag = string(p.event.tag)
  138. n.implicit = p.event.implicit
  139. p.anchor(n, p.event.anchor)
  140. p.skip()
  141. return n
  142. }
  143. func (p *parser) sequence() *node {
  144. n := p.node(sequenceNode)
  145. p.anchor(n, p.event.anchor)
  146. p.skip()
  147. for p.event.typ != yaml_SEQUENCE_END_EVENT {
  148. n.children = append(n.children, p.parse())
  149. }
  150. p.skip()
  151. return n
  152. }
  153. func (p *parser) mapping() *node {
  154. n := p.node(mappingNode)
  155. p.anchor(n, p.event.anchor)
  156. p.skip()
  157. for p.event.typ != yaml_MAPPING_END_EVENT {
  158. n.children = append(n.children, p.parse(), p.parse())
  159. }
  160. p.skip()
  161. return n
  162. }
  163. // ----------------------------------------------------------------------------
  164. // Decoder, unmarshals a node into a provided value.
  165. type decoder struct {
  166. doc *node
  167. aliases map[string]bool
  168. }
  169. func newDecoder() *decoder {
  170. d := &decoder{}
  171. d.aliases = make(map[string]bool)
  172. return d
  173. }
  174. // d.setter deals with setters and pointer dereferencing and initialization.
  175. //
  176. // It's a slightly convoluted case to handle properly:
  177. //
  178. // - nil pointers should be initialized, unless being set to nil
  179. // - we don't know at this point yet what's the value to SetYAML() with.
  180. // - we can't separate pointer deref/init and setter checking, because
  181. // a setter may be found while going down a pointer chain.
  182. //
  183. // Thus, here is how it takes care of it:
  184. //
  185. // - out is provided as a pointer, so that it can be replaced.
  186. // - when looking at a non-setter ptr, *out=ptr.Elem(), unless tag=!!null
  187. // - when a setter is found, *out=interface{}, and a set() function is
  188. // returned to call SetYAML() with the value of *out once it's defined.
  189. //
  190. func (d *decoder) setter(tag string, out *reflect.Value, good *bool) (set func()) {
  191. if (*out).Kind() != reflect.Ptr && (*out).CanAddr() {
  192. setter, _ := (*out).Addr().Interface().(Setter)
  193. if setter != nil {
  194. var arg interface{}
  195. *out = reflect.ValueOf(&arg).Elem()
  196. return func() {
  197. *good = setter.SetYAML(shortTag(tag), arg)
  198. }
  199. }
  200. }
  201. again := true
  202. for again {
  203. again = false
  204. setter, _ := (*out).Interface().(Setter)
  205. if tag != yaml_NULL_TAG || setter != nil {
  206. if pv := (*out); pv.Kind() == reflect.Ptr {
  207. if pv.IsNil() {
  208. *out = reflect.New(pv.Type().Elem()).Elem()
  209. pv.Set((*out).Addr())
  210. } else {
  211. *out = pv.Elem()
  212. }
  213. setter, _ = pv.Interface().(Setter)
  214. again = true
  215. }
  216. }
  217. if setter != nil {
  218. var arg interface{}
  219. *out = reflect.ValueOf(&arg).Elem()
  220. return func() {
  221. *good = setter.SetYAML(shortTag(tag), arg)
  222. }
  223. }
  224. }
  225. return nil
  226. }
  227. func (d *decoder) unmarshal(n *node, out reflect.Value) (good bool) {
  228. switch n.kind {
  229. case documentNode:
  230. good = d.document(n, out)
  231. case scalarNode:
  232. good = d.scalar(n, out)
  233. case aliasNode:
  234. good = d.alias(n, out)
  235. case mappingNode:
  236. good = d.mapping(n, out)
  237. case sequenceNode:
  238. good = d.sequence(n, out)
  239. default:
  240. panic("Internal error: unknown node kind: " + strconv.Itoa(n.kind))
  241. }
  242. return
  243. }
  244. func (d *decoder) document(n *node, out reflect.Value) (good bool) {
  245. if len(n.children) == 1 {
  246. d.doc = n
  247. d.unmarshal(n.children[0], out)
  248. return true
  249. }
  250. return false
  251. }
  252. func (d *decoder) alias(n *node, out reflect.Value) (good bool) {
  253. an, ok := d.doc.anchors[n.value]
  254. if !ok {
  255. fail("Unknown anchor '" + n.value + "' referenced")
  256. }
  257. if d.aliases[n.value] {
  258. fail("Anchor '" + n.value + "' value contains itself")
  259. }
  260. d.aliases[n.value] = true
  261. good = d.unmarshal(an, out)
  262. delete(d.aliases, n.value)
  263. return good
  264. }
  265. var zeroValue reflect.Value
  266. func resetMap(out reflect.Value) {
  267. for _, k := range out.MapKeys() {
  268. out.SetMapIndex(k, zeroValue)
  269. }
  270. }
  271. var durationType = reflect.TypeOf(time.Duration(0))
  272. func (d *decoder) scalar(n *node, out reflect.Value) (good bool) {
  273. var tag string
  274. var resolved interface{}
  275. if n.tag == "" && !n.implicit {
  276. tag = yaml_STR_TAG
  277. resolved = n.value
  278. } else {
  279. tag, resolved = resolve(n.tag, n.value)
  280. if tag == yaml_BINARY_TAG {
  281. data, err := base64.StdEncoding.DecodeString(resolved.(string))
  282. if err != nil {
  283. fail("!!binary value contains invalid base64 data")
  284. }
  285. resolved = string(data)
  286. }
  287. }
  288. if set := d.setter(tag, &out, &good); set != nil {
  289. defer set()
  290. }
  291. if resolved == nil {
  292. if out.Kind() == reflect.Map && !out.CanAddr() {
  293. resetMap(out)
  294. } else {
  295. out.Set(reflect.Zero(out.Type()))
  296. }
  297. good = true
  298. return
  299. }
  300. switch out.Kind() {
  301. case reflect.String:
  302. if tag == yaml_BINARY_TAG {
  303. out.SetString(resolved.(string))
  304. good = true
  305. } else if resolved != nil {
  306. out.SetString(n.value)
  307. good = true
  308. }
  309. case reflect.Interface:
  310. if resolved == nil {
  311. out.Set(reflect.Zero(out.Type()))
  312. } else {
  313. out.Set(reflect.ValueOf(resolved))
  314. }
  315. good = true
  316. case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
  317. switch resolved := resolved.(type) {
  318. case int:
  319. if !out.OverflowInt(int64(resolved)) {
  320. out.SetInt(int64(resolved))
  321. good = true
  322. }
  323. case int64:
  324. if !out.OverflowInt(resolved) {
  325. out.SetInt(resolved)
  326. good = true
  327. }
  328. case float64:
  329. if resolved < 1<<63-1 && !out.OverflowInt(int64(resolved)) {
  330. out.SetInt(int64(resolved))
  331. good = true
  332. }
  333. case string:
  334. if out.Type() == durationType {
  335. d, err := time.ParseDuration(resolved)
  336. if err == nil {
  337. out.SetInt(int64(d))
  338. good = true
  339. }
  340. }
  341. }
  342. case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
  343. switch resolved := resolved.(type) {
  344. case int:
  345. if resolved >= 0 {
  346. out.SetUint(uint64(resolved))
  347. good = true
  348. }
  349. case int64:
  350. if resolved >= 0 {
  351. out.SetUint(uint64(resolved))
  352. good = true
  353. }
  354. case float64:
  355. if resolved < 1<<64-1 && !out.OverflowUint(uint64(resolved)) {
  356. out.SetUint(uint64(resolved))
  357. good = true
  358. }
  359. }
  360. case reflect.Bool:
  361. switch resolved := resolved.(type) {
  362. case bool:
  363. out.SetBool(resolved)
  364. good = true
  365. }
  366. case reflect.Float32, reflect.Float64:
  367. switch resolved := resolved.(type) {
  368. case int:
  369. out.SetFloat(float64(resolved))
  370. good = true
  371. case int64:
  372. out.SetFloat(float64(resolved))
  373. good = true
  374. case float64:
  375. out.SetFloat(resolved)
  376. good = true
  377. }
  378. case reflect.Ptr:
  379. if out.Type().Elem() == reflect.TypeOf(resolved) {
  380. elem := reflect.New(out.Type().Elem())
  381. elem.Elem().Set(reflect.ValueOf(resolved))
  382. out.Set(elem)
  383. good = true
  384. }
  385. }
  386. return good
  387. }
  388. func settableValueOf(i interface{}) reflect.Value {
  389. v := reflect.ValueOf(i)
  390. sv := reflect.New(v.Type()).Elem()
  391. sv.Set(v)
  392. return sv
  393. }
  394. func (d *decoder) sequence(n *node, out reflect.Value) (good bool) {
  395. if set := d.setter(yaml_SEQ_TAG, &out, &good); set != nil {
  396. defer set()
  397. }
  398. var iface reflect.Value
  399. if out.Kind() == reflect.Interface {
  400. // No type hints. Will have to use a generic sequence.
  401. iface = out
  402. out = settableValueOf(make([]interface{}, 0))
  403. }
  404. if out.Kind() != reflect.Slice {
  405. return false
  406. }
  407. et := out.Type().Elem()
  408. l := len(n.children)
  409. for i := 0; i < l; i++ {
  410. e := reflect.New(et).Elem()
  411. if ok := d.unmarshal(n.children[i], e); ok {
  412. out.Set(reflect.Append(out, e))
  413. }
  414. }
  415. if iface.IsValid() {
  416. iface.Set(out)
  417. }
  418. return true
  419. }
  420. func (d *decoder) mapping(n *node, out reflect.Value) (good bool) {
  421. if set := d.setter(yaml_MAP_TAG, &out, &good); set != nil {
  422. defer set()
  423. }
  424. if out.Kind() == reflect.Struct {
  425. return d.mappingStruct(n, out)
  426. }
  427. if out.Kind() == reflect.Interface {
  428. // No type hints. Will have to use a generic map.
  429. iface := out
  430. out = settableValueOf(make(map[interface{}]interface{}))
  431. iface.Set(out)
  432. }
  433. if out.Kind() != reflect.Map {
  434. return false
  435. }
  436. outt := out.Type()
  437. kt := outt.Key()
  438. et := outt.Elem()
  439. if out.IsNil() {
  440. out.Set(reflect.MakeMap(outt))
  441. }
  442. l := len(n.children)
  443. for i := 0; i < l; i += 2 {
  444. if isMerge(n.children[i]) {
  445. d.merge(n.children[i+1], out)
  446. continue
  447. }
  448. k := reflect.New(kt).Elem()
  449. if d.unmarshal(n.children[i], k) {
  450. kkind := k.Kind()
  451. if kkind == reflect.Interface {
  452. kkind = k.Elem().Kind()
  453. }
  454. if kkind == reflect.Map || kkind == reflect.Slice {
  455. fail(fmt.Sprintf("invalid map key: %#v", k.Interface()))
  456. }
  457. e := reflect.New(et).Elem()
  458. if d.unmarshal(n.children[i+1], e) {
  459. out.SetMapIndex(k, e)
  460. }
  461. }
  462. }
  463. return true
  464. }
  465. func (d *decoder) mappingStruct(n *node, out reflect.Value) (good bool) {
  466. sinfo, err := getStructInfo(out.Type())
  467. if err != nil {
  468. panic(err)
  469. }
  470. name := settableValueOf("")
  471. l := len(n.children)
  472. for i := 0; i < l; i += 2 {
  473. ni := n.children[i]
  474. if isMerge(ni) {
  475. d.merge(n.children[i+1], out)
  476. continue
  477. }
  478. if !d.unmarshal(ni, name) {
  479. continue
  480. }
  481. if info, ok := sinfo.FieldsMap[name.String()]; ok {
  482. var field reflect.Value
  483. if info.Inline == nil {
  484. field = out.Field(info.Num)
  485. } else {
  486. field = out.FieldByIndex(info.Inline)
  487. }
  488. d.unmarshal(n.children[i+1], field)
  489. }
  490. }
  491. return true
  492. }
  493. func (d *decoder) merge(n *node, out reflect.Value) {
  494. const wantMap = "map merge requires map or sequence of maps as the value"
  495. switch n.kind {
  496. case mappingNode:
  497. d.unmarshal(n, out)
  498. case aliasNode:
  499. an, ok := d.doc.anchors[n.value]
  500. if ok && an.kind != mappingNode {
  501. fail(wantMap)
  502. }
  503. d.unmarshal(n, out)
  504. case sequenceNode:
  505. // Step backwards as earlier nodes take precedence.
  506. for i := len(n.children) - 1; i >= 0; i-- {
  507. ni := n.children[i]
  508. if ni.kind == aliasNode {
  509. an, ok := d.doc.anchors[ni.value]
  510. if ok && an.kind != mappingNode {
  511. fail(wantMap)
  512. }
  513. } else if ni.kind != mappingNode {
  514. fail(wantMap)
  515. }
  516. d.unmarshal(ni, out)
  517. }
  518. default:
  519. fail(wantMap)
  520. }
  521. }
  522. func isMerge(n *node) bool {
  523. return n.kind == scalarNode && n.value == "<<" && (n.implicit == true || n.tag == yaml_MERGE_TAG)
  524. }