decode.go 9.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457
  1. package goyaml
  2. import (
  3. "reflect"
  4. "strconv"
  5. )
  6. const (
  7. documentNode = 1 << iota
  8. mappingNode
  9. sequenceNode
  10. scalarNode
  11. aliasNode
  12. )
  13. type node struct {
  14. kind int
  15. line, column int
  16. tag string
  17. value string
  18. implicit bool
  19. children []*node
  20. anchors map[string]*node
  21. }
  22. // ----------------------------------------------------------------------------
  23. // Parser, produces a node tree out of a libyaml event stream.
  24. type parser struct {
  25. parser yaml_parser_t
  26. event yaml_event_t
  27. doc *node
  28. }
  29. func newParser(b []byte) *parser {
  30. p := parser{}
  31. if !yaml_parser_initialize(&p.parser) {
  32. panic("Failed to initialize YAML emitter")
  33. }
  34. if len(b) == 0 {
  35. b = []byte{'\n'}
  36. }
  37. yaml_parser_set_input_string(&p.parser, b)
  38. p.skip()
  39. if p.event.typ != yaml_STREAM_START_EVENT {
  40. panic("Expected stream start event, got " + strconv.Itoa(int(p.event.typ)))
  41. }
  42. p.skip()
  43. return &p
  44. }
  45. func (p *parser) destroy() {
  46. if p.event.typ != yaml_NO_EVENT {
  47. yaml_event_delete(&p.event)
  48. }
  49. yaml_parser_delete(&p.parser)
  50. }
  51. func (p *parser) skip() {
  52. if p.event.typ != yaml_NO_EVENT {
  53. if p.event.typ == yaml_STREAM_END_EVENT {
  54. panic("Attempted to go past the end of stream. Corrupted value?")
  55. }
  56. yaml_event_delete(&p.event)
  57. }
  58. if !yaml_parser_parse(&p.parser, &p.event) {
  59. p.fail()
  60. }
  61. }
  62. func (p *parser) fail() {
  63. var where string
  64. var line int
  65. if p.parser.problem_mark.line != 0 {
  66. line = p.parser.problem_mark.line
  67. } else if p.parser.context_mark.line != 0 {
  68. line = p.parser.context_mark.line
  69. }
  70. if line != 0 {
  71. where = "line " + strconv.Itoa(line) + ": "
  72. }
  73. var msg string
  74. if len(p.parser.problem) > 0 {
  75. msg = p.parser.problem
  76. } else {
  77. msg = "Unknown problem parsing YAML content"
  78. }
  79. panic(where + msg)
  80. }
  81. func (p *parser) anchor(n *node, anchor []byte) {
  82. if anchor != nil {
  83. p.doc.anchors[string(anchor)] = n
  84. }
  85. }
  86. func (p *parser) parse() *node {
  87. switch p.event.typ {
  88. case yaml_SCALAR_EVENT:
  89. return p.scalar()
  90. case yaml_ALIAS_EVENT:
  91. return p.alias()
  92. case yaml_MAPPING_START_EVENT:
  93. return p.mapping()
  94. case yaml_SEQUENCE_START_EVENT:
  95. return p.sequence()
  96. case yaml_DOCUMENT_START_EVENT:
  97. return p.document()
  98. case yaml_STREAM_END_EVENT:
  99. // Happens when attempting to decode an empty buffer.
  100. return nil
  101. default:
  102. panic("Attempted to parse unknown event: " +
  103. strconv.Itoa(int(p.event.typ)))
  104. }
  105. panic("Unreachable")
  106. }
  107. func (p *parser) node(kind int) *node {
  108. return &node{
  109. kind: kind,
  110. line: p.event.start_mark.line,
  111. column: p.event.start_mark.column,
  112. }
  113. }
  114. func (p *parser) document() *node {
  115. n := p.node(documentNode)
  116. n.anchors = make(map[string]*node)
  117. p.doc = n
  118. p.skip()
  119. n.children = append(n.children, p.parse())
  120. if p.event.typ != yaml_DOCUMENT_END_EVENT {
  121. panic("Expected end of document event but got " +
  122. strconv.Itoa(int(p.event.typ)))
  123. }
  124. p.skip()
  125. return n
  126. }
  127. func (p *parser) alias() *node {
  128. n := p.node(aliasNode)
  129. n.value = string(p.event.anchor)
  130. p.skip()
  131. return n
  132. }
  133. func (p *parser) scalar() *node {
  134. n := p.node(scalarNode)
  135. n.value = string(p.event.value)
  136. n.tag = string(p.event.tag)
  137. n.implicit = p.event.implicit
  138. p.anchor(n, p.event.anchor)
  139. p.skip()
  140. return n
  141. }
  142. func (p *parser) sequence() *node {
  143. n := p.node(sequenceNode)
  144. p.anchor(n, p.event.anchor)
  145. p.skip()
  146. for p.event.typ != yaml_SEQUENCE_END_EVENT {
  147. n.children = append(n.children, p.parse())
  148. }
  149. p.skip()
  150. return n
  151. }
  152. func (p *parser) mapping() *node {
  153. n := p.node(mappingNode)
  154. p.anchor(n, p.event.anchor)
  155. p.skip()
  156. for p.event.typ != yaml_MAPPING_END_EVENT {
  157. n.children = append(n.children, p.parse(), p.parse())
  158. }
  159. p.skip()
  160. return n
  161. }
  162. // ----------------------------------------------------------------------------
  163. // Decoder, unmarshals a node into a provided value.
  164. type decoder struct {
  165. doc *node
  166. aliases map[string]bool
  167. }
  168. func newDecoder() *decoder {
  169. d := &decoder{}
  170. d.aliases = make(map[string]bool)
  171. return d
  172. }
  173. // d.setter deals with setters and pointer dereferencing and initialization.
  174. //
  175. // It's a slightly convoluted case to handle properly:
  176. //
  177. // - nil pointers should be initialized, unless being set to nil
  178. // - we don't know at this point yet what's the value to SetYAML() with.
  179. // - we can't separate pointer deref/init and setter checking, because
  180. // a setter may be found while going down a pointer chain.
  181. //
  182. // Thus, here is how it takes care of it:
  183. //
  184. // - out is provided as a pointer, so that it can be replaced.
  185. // - when looking at a non-setter ptr, *out=ptr.Elem(), unless tag=!!null
  186. // - when a setter is found, *out=interface{}, and a set() function is
  187. // returned to call SetYAML() with the value of *out once it's defined.
  188. //
  189. func (d *decoder) setter(tag string, out *reflect.Value, good *bool) (set func()) {
  190. again := true
  191. for again {
  192. again = false
  193. setter, _ := (*out).Interface().(Setter)
  194. if tag != "!!null" || setter != nil {
  195. if pv := (*out); pv.Kind() == reflect.Ptr {
  196. if pv.IsNil() {
  197. *out = reflect.New(pv.Type().Elem()).Elem()
  198. pv.Set((*out).Addr())
  199. } else {
  200. *out = pv.Elem()
  201. }
  202. setter, _ = pv.Interface().(Setter)
  203. again = true
  204. }
  205. }
  206. if setter != nil {
  207. var arg interface{}
  208. *out = reflect.ValueOf(&arg).Elem()
  209. return func() {
  210. *good = setter.SetYAML(tag, arg)
  211. }
  212. }
  213. }
  214. return nil
  215. }
  216. func (d *decoder) unmarshal(n *node, out reflect.Value) (good bool) {
  217. switch n.kind {
  218. case documentNode:
  219. good = d.document(n, out)
  220. case scalarNode:
  221. good = d.scalar(n, out)
  222. case aliasNode:
  223. good = d.alias(n, out)
  224. case mappingNode:
  225. good = d.mapping(n, out)
  226. case sequenceNode:
  227. good = d.sequence(n, out)
  228. default:
  229. panic("Internal error: unknown node kind: " + strconv.Itoa(n.kind))
  230. }
  231. return
  232. }
  233. func (d *decoder) document(n *node, out reflect.Value) (good bool) {
  234. if len(n.children) == 1 {
  235. d.doc = n
  236. d.unmarshal(n.children[0], out)
  237. return true
  238. }
  239. return false
  240. }
  241. func (d *decoder) alias(n *node, out reflect.Value) (good bool) {
  242. an, ok := d.doc.anchors[n.value]
  243. if !ok {
  244. panic("Unknown anchor '" + n.value + "' referenced")
  245. }
  246. if d.aliases[n.value] {
  247. panic("Anchor '" + n.value + "' value contains itself")
  248. }
  249. d.aliases[n.value] = true
  250. good = d.unmarshal(an, out)
  251. delete(d.aliases, n.value)
  252. return good
  253. }
  254. func (d *decoder) scalar(n *node, out reflect.Value) (good bool) {
  255. var tag string
  256. var resolved interface{}
  257. if n.tag == "" && !n.implicit {
  258. resolved = n.value
  259. } else {
  260. tag, resolved = resolve(n.tag, n.value)
  261. if set := d.setter(tag, &out, &good); set != nil {
  262. defer set()
  263. }
  264. }
  265. switch out.Kind() {
  266. case reflect.String:
  267. if resolved != nil {
  268. out.SetString(n.value)
  269. good = true
  270. }
  271. case reflect.Interface:
  272. if resolved == nil {
  273. out.Set(reflect.Zero(out.Type()))
  274. } else {
  275. out.Set(reflect.ValueOf(resolved))
  276. }
  277. good = true
  278. case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
  279. switch resolved := resolved.(type) {
  280. case int:
  281. if !out.OverflowInt(int64(resolved)) {
  282. out.SetInt(int64(resolved))
  283. good = true
  284. }
  285. case int64:
  286. if !out.OverflowInt(resolved) {
  287. out.SetInt(resolved)
  288. good = true
  289. }
  290. }
  291. case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
  292. switch resolved := resolved.(type) {
  293. case int:
  294. if resolved >= 0 {
  295. out.SetUint(uint64(resolved))
  296. good = true
  297. }
  298. case int64:
  299. if resolved >= 0 {
  300. out.SetUint(uint64(resolved))
  301. good = true
  302. }
  303. }
  304. case reflect.Bool:
  305. switch resolved := resolved.(type) {
  306. case bool:
  307. out.SetBool(resolved)
  308. good = true
  309. }
  310. case reflect.Float32, reflect.Float64:
  311. switch resolved := resolved.(type) {
  312. case float64:
  313. out.SetFloat(resolved)
  314. good = true
  315. }
  316. case reflect.Ptr:
  317. switch resolved.(type) {
  318. case nil:
  319. out.Set(reflect.Zero(out.Type()))
  320. good = true
  321. default:
  322. if out.Type().Elem() == reflect.TypeOf(resolved) {
  323. elem := reflect.New(out.Type().Elem())
  324. elem.Elem().Set(reflect.ValueOf(resolved))
  325. out.Set(elem)
  326. good = true
  327. }
  328. }
  329. }
  330. return good
  331. }
  332. func settableValueOf(i interface{}) reflect.Value {
  333. v := reflect.ValueOf(i)
  334. sv := reflect.New(v.Type()).Elem()
  335. sv.Set(v)
  336. return sv
  337. }
  338. func (d *decoder) sequence(n *node, out reflect.Value) (good bool) {
  339. if set := d.setter("!!seq", &out, &good); set != nil {
  340. defer set()
  341. }
  342. var iface reflect.Value
  343. if out.Kind() == reflect.Interface {
  344. // No type hints. Will have to use a generic sequence.
  345. iface = out
  346. out = settableValueOf(make([]interface{}, 0))
  347. }
  348. if out.Kind() != reflect.Slice {
  349. return false
  350. }
  351. et := out.Type().Elem()
  352. l := len(n.children)
  353. for i := 0; i < l; i++ {
  354. e := reflect.New(et).Elem()
  355. if ok := d.unmarshal(n.children[i], e); ok {
  356. out.Set(reflect.Append(out, e))
  357. }
  358. }
  359. if iface.IsValid() {
  360. iface.Set(out)
  361. }
  362. return true
  363. }
  364. func (d *decoder) mapping(n *node, out reflect.Value) (good bool) {
  365. if set := d.setter("!!map", &out, &good); set != nil {
  366. defer set()
  367. }
  368. if out.Kind() == reflect.Struct {
  369. return d.mappingStruct(n, out)
  370. }
  371. if out.Kind() == reflect.Interface {
  372. // No type hints. Will have to use a generic map.
  373. iface := out
  374. out = settableValueOf(make(map[interface{}]interface{}))
  375. iface.Set(out)
  376. }
  377. if out.Kind() != reflect.Map {
  378. return false
  379. }
  380. outt := out.Type()
  381. kt := outt.Key()
  382. et := outt.Elem()
  383. if out.IsNil() {
  384. out.Set(reflect.MakeMap(outt))
  385. }
  386. l := len(n.children)
  387. for i := 0; i < l; i += 2 {
  388. k := reflect.New(kt).Elem()
  389. if d.unmarshal(n.children[i], k) {
  390. e := reflect.New(et).Elem()
  391. if d.unmarshal(n.children[i+1], e) {
  392. out.SetMapIndex(k, e)
  393. }
  394. }
  395. }
  396. return true
  397. }
  398. func (d *decoder) mappingStruct(n *node, out reflect.Value) (good bool) {
  399. sinfo, err := getStructInfo(out.Type())
  400. if err != nil {
  401. panic(err)
  402. }
  403. name := settableValueOf("")
  404. l := len(n.children)
  405. for i := 0; i < l; i += 2 {
  406. if !d.unmarshal(n.children[i], name) {
  407. continue
  408. }
  409. if info, ok := sinfo.FieldsMap[name.String()]; ok {
  410. var field reflect.Value
  411. if info.Inline == nil {
  412. field = out.Field(info.Num)
  413. } else {
  414. field = out.FieldByIndex(info.Inline)
  415. }
  416. d.unmarshal(n.children[i+1], field)
  417. }
  418. }
  419. return true
  420. }