Scott Baker | eee8dd8 | 2019-09-24 12:52:34 -0700 | [diff] [blame^] | 1 | package yaml |
| 2 | |
| 3 | import ( |
| 4 | "encoding" |
| 5 | "fmt" |
| 6 | "io" |
| 7 | "reflect" |
| 8 | "regexp" |
| 9 | "sort" |
| 10 | "strconv" |
| 11 | "strings" |
| 12 | "time" |
| 13 | "unicode/utf8" |
| 14 | ) |
| 15 | |
| 16 | // jsonNumber is the interface of the encoding/json.Number datatype. |
| 17 | // Repeating the interface here avoids a dependency on encoding/json, and also |
| 18 | // supports other libraries like jsoniter, which use a similar datatype with |
| 19 | // the same interface. Detecting this interface is useful when dealing with |
| 20 | // structures containing json.Number, which is a string under the hood. The |
| 21 | // encoder should prefer the use of Int64(), Float64() and string(), in that |
| 22 | // order, when encoding this type. |
| 23 | type jsonNumber interface { |
| 24 | Float64() (float64, error) |
| 25 | Int64() (int64, error) |
| 26 | String() string |
| 27 | } |
| 28 | |
| 29 | type encoder struct { |
| 30 | emitter yaml_emitter_t |
| 31 | event yaml_event_t |
| 32 | out []byte |
| 33 | flow bool |
| 34 | // doneInit holds whether the initial stream_start_event has been |
| 35 | // emitted. |
| 36 | doneInit bool |
| 37 | } |
| 38 | |
| 39 | func newEncoder() *encoder { |
| 40 | e := &encoder{} |
| 41 | yaml_emitter_initialize(&e.emitter) |
| 42 | yaml_emitter_set_output_string(&e.emitter, &e.out) |
| 43 | yaml_emitter_set_unicode(&e.emitter, true) |
| 44 | return e |
| 45 | } |
| 46 | |
| 47 | func newEncoderWithWriter(w io.Writer) *encoder { |
| 48 | e := &encoder{} |
| 49 | yaml_emitter_initialize(&e.emitter) |
| 50 | yaml_emitter_set_output_writer(&e.emitter, w) |
| 51 | yaml_emitter_set_unicode(&e.emitter, true) |
| 52 | return e |
| 53 | } |
| 54 | |
| 55 | func (e *encoder) init() { |
| 56 | if e.doneInit { |
| 57 | return |
| 58 | } |
| 59 | yaml_stream_start_event_initialize(&e.event, yaml_UTF8_ENCODING) |
| 60 | e.emit() |
| 61 | e.doneInit = true |
| 62 | } |
| 63 | |
| 64 | func (e *encoder) finish() { |
| 65 | e.emitter.open_ended = false |
| 66 | yaml_stream_end_event_initialize(&e.event) |
| 67 | e.emit() |
| 68 | } |
| 69 | |
| 70 | func (e *encoder) destroy() { |
| 71 | yaml_emitter_delete(&e.emitter) |
| 72 | } |
| 73 | |
| 74 | func (e *encoder) emit() { |
| 75 | // This will internally delete the e.event value. |
| 76 | e.must(yaml_emitter_emit(&e.emitter, &e.event)) |
| 77 | } |
| 78 | |
| 79 | func (e *encoder) must(ok bool) { |
| 80 | if !ok { |
| 81 | msg := e.emitter.problem |
| 82 | if msg == "" { |
| 83 | msg = "unknown problem generating YAML content" |
| 84 | } |
| 85 | failf("%s", msg) |
| 86 | } |
| 87 | } |
| 88 | |
| 89 | func (e *encoder) marshalDoc(tag string, in reflect.Value) { |
| 90 | e.init() |
| 91 | yaml_document_start_event_initialize(&e.event, nil, nil, true) |
| 92 | e.emit() |
| 93 | e.marshal(tag, in) |
| 94 | yaml_document_end_event_initialize(&e.event, true) |
| 95 | e.emit() |
| 96 | } |
| 97 | |
| 98 | func (e *encoder) marshal(tag string, in reflect.Value) { |
| 99 | if !in.IsValid() || in.Kind() == reflect.Ptr && in.IsNil() { |
| 100 | e.nilv() |
| 101 | return |
| 102 | } |
| 103 | iface := in.Interface() |
| 104 | switch m := iface.(type) { |
| 105 | case jsonNumber: |
| 106 | integer, err := m.Int64() |
| 107 | if err == nil { |
| 108 | // In this case the json.Number is a valid int64 |
| 109 | in = reflect.ValueOf(integer) |
| 110 | break |
| 111 | } |
| 112 | float, err := m.Float64() |
| 113 | if err == nil { |
| 114 | // In this case the json.Number is a valid float64 |
| 115 | in = reflect.ValueOf(float) |
| 116 | break |
| 117 | } |
| 118 | // fallback case - no number could be obtained |
| 119 | in = reflect.ValueOf(m.String()) |
| 120 | case time.Time, *time.Time: |
| 121 | // Although time.Time implements TextMarshaler, |
| 122 | // we don't want to treat it as a string for YAML |
| 123 | // purposes because YAML has special support for |
| 124 | // timestamps. |
| 125 | case Marshaler: |
| 126 | v, err := m.MarshalYAML() |
| 127 | if err != nil { |
| 128 | fail(err) |
| 129 | } |
| 130 | if v == nil { |
| 131 | e.nilv() |
| 132 | return |
| 133 | } |
| 134 | in = reflect.ValueOf(v) |
| 135 | case encoding.TextMarshaler: |
| 136 | text, err := m.MarshalText() |
| 137 | if err != nil { |
| 138 | fail(err) |
| 139 | } |
| 140 | in = reflect.ValueOf(string(text)) |
| 141 | case nil: |
| 142 | e.nilv() |
| 143 | return |
| 144 | } |
| 145 | switch in.Kind() { |
| 146 | case reflect.Interface: |
| 147 | e.marshal(tag, in.Elem()) |
| 148 | case reflect.Map: |
| 149 | e.mapv(tag, in) |
| 150 | case reflect.Ptr: |
| 151 | if in.Type() == ptrTimeType { |
| 152 | e.timev(tag, in.Elem()) |
| 153 | } else { |
| 154 | e.marshal(tag, in.Elem()) |
| 155 | } |
| 156 | case reflect.Struct: |
| 157 | if in.Type() == timeType { |
| 158 | e.timev(tag, in) |
| 159 | } else { |
| 160 | e.structv(tag, in) |
| 161 | } |
| 162 | case reflect.Slice, reflect.Array: |
| 163 | if in.Type().Elem() == mapItemType { |
| 164 | e.itemsv(tag, in) |
| 165 | } else { |
| 166 | e.slicev(tag, in) |
| 167 | } |
| 168 | case reflect.String: |
| 169 | e.stringv(tag, in) |
| 170 | case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64: |
| 171 | if in.Type() == durationType { |
| 172 | e.stringv(tag, reflect.ValueOf(iface.(time.Duration).String())) |
| 173 | } else { |
| 174 | e.intv(tag, in) |
| 175 | } |
| 176 | case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr: |
| 177 | e.uintv(tag, in) |
| 178 | case reflect.Float32, reflect.Float64: |
| 179 | e.floatv(tag, in) |
| 180 | case reflect.Bool: |
| 181 | e.boolv(tag, in) |
| 182 | default: |
| 183 | panic("cannot marshal type: " + in.Type().String()) |
| 184 | } |
| 185 | } |
| 186 | |
| 187 | func (e *encoder) mapv(tag string, in reflect.Value) { |
| 188 | e.mappingv(tag, func() { |
| 189 | keys := keyList(in.MapKeys()) |
| 190 | sort.Sort(keys) |
| 191 | for _, k := range keys { |
| 192 | e.marshal("", k) |
| 193 | e.marshal("", in.MapIndex(k)) |
| 194 | } |
| 195 | }) |
| 196 | } |
| 197 | |
| 198 | func (e *encoder) itemsv(tag string, in reflect.Value) { |
| 199 | e.mappingv(tag, func() { |
| 200 | slice := in.Convert(reflect.TypeOf([]MapItem{})).Interface().([]MapItem) |
| 201 | for _, item := range slice { |
| 202 | e.marshal("", reflect.ValueOf(item.Key)) |
| 203 | e.marshal("", reflect.ValueOf(item.Value)) |
| 204 | } |
| 205 | }) |
| 206 | } |
| 207 | |
| 208 | func (e *encoder) structv(tag string, in reflect.Value) { |
| 209 | sinfo, err := getStructInfo(in.Type()) |
| 210 | if err != nil { |
| 211 | panic(err) |
| 212 | } |
| 213 | e.mappingv(tag, func() { |
| 214 | for _, info := range sinfo.FieldsList { |
| 215 | var value reflect.Value |
| 216 | if info.Inline == nil { |
| 217 | value = in.Field(info.Num) |
| 218 | } else { |
| 219 | value = in.FieldByIndex(info.Inline) |
| 220 | } |
| 221 | if info.OmitEmpty && isZero(value) { |
| 222 | continue |
| 223 | } |
| 224 | e.marshal("", reflect.ValueOf(info.Key)) |
| 225 | e.flow = info.Flow |
| 226 | e.marshal("", value) |
| 227 | } |
| 228 | if sinfo.InlineMap >= 0 { |
| 229 | m := in.Field(sinfo.InlineMap) |
| 230 | if m.Len() > 0 { |
| 231 | e.flow = false |
| 232 | keys := keyList(m.MapKeys()) |
| 233 | sort.Sort(keys) |
| 234 | for _, k := range keys { |
| 235 | if _, found := sinfo.FieldsMap[k.String()]; found { |
| 236 | panic(fmt.Sprintf("Can't have key %q in inlined map; conflicts with struct field", k.String())) |
| 237 | } |
| 238 | e.marshal("", k) |
| 239 | e.flow = false |
| 240 | e.marshal("", m.MapIndex(k)) |
| 241 | } |
| 242 | } |
| 243 | } |
| 244 | }) |
| 245 | } |
| 246 | |
| 247 | func (e *encoder) mappingv(tag string, f func()) { |
| 248 | implicit := tag == "" |
| 249 | style := yaml_BLOCK_MAPPING_STYLE |
| 250 | if e.flow { |
| 251 | e.flow = false |
| 252 | style = yaml_FLOW_MAPPING_STYLE |
| 253 | } |
| 254 | yaml_mapping_start_event_initialize(&e.event, nil, []byte(tag), implicit, style) |
| 255 | e.emit() |
| 256 | f() |
| 257 | yaml_mapping_end_event_initialize(&e.event) |
| 258 | e.emit() |
| 259 | } |
| 260 | |
| 261 | func (e *encoder) slicev(tag string, in reflect.Value) { |
| 262 | implicit := tag == "" |
| 263 | style := yaml_BLOCK_SEQUENCE_STYLE |
| 264 | if e.flow { |
| 265 | e.flow = false |
| 266 | style = yaml_FLOW_SEQUENCE_STYLE |
| 267 | } |
| 268 | e.must(yaml_sequence_start_event_initialize(&e.event, nil, []byte(tag), implicit, style)) |
| 269 | e.emit() |
| 270 | n := in.Len() |
| 271 | for i := 0; i < n; i++ { |
| 272 | e.marshal("", in.Index(i)) |
| 273 | } |
| 274 | e.must(yaml_sequence_end_event_initialize(&e.event)) |
| 275 | e.emit() |
| 276 | } |
| 277 | |
| 278 | // isBase60 returns whether s is in base 60 notation as defined in YAML 1.1. |
| 279 | // |
| 280 | // The base 60 float notation in YAML 1.1 is a terrible idea and is unsupported |
| 281 | // in YAML 1.2 and by this package, but these should be marshalled quoted for |
| 282 | // the time being for compatibility with other parsers. |
| 283 | func isBase60Float(s string) (result bool) { |
| 284 | // Fast path. |
| 285 | if s == "" { |
| 286 | return false |
| 287 | } |
| 288 | c := s[0] |
| 289 | if !(c == '+' || c == '-' || c >= '0' && c <= '9') || strings.IndexByte(s, ':') < 0 { |
| 290 | return false |
| 291 | } |
| 292 | // Do the full match. |
| 293 | return base60float.MatchString(s) |
| 294 | } |
| 295 | |
| 296 | // From http://yaml.org/type/float.html, except the regular expression there |
| 297 | // is bogus. In practice parsers do not enforce the "\.[0-9_]*" suffix. |
| 298 | var base60float = regexp.MustCompile(`^[-+]?[0-9][0-9_]*(?::[0-5]?[0-9])+(?:\.[0-9_]*)?$`) |
| 299 | |
| 300 | func (e *encoder) stringv(tag string, in reflect.Value) { |
| 301 | var style yaml_scalar_style_t |
| 302 | s := in.String() |
| 303 | canUsePlain := true |
| 304 | switch { |
| 305 | case !utf8.ValidString(s): |
| 306 | if tag == yaml_BINARY_TAG { |
| 307 | failf("explicitly tagged !!binary data must be base64-encoded") |
| 308 | } |
| 309 | if tag != "" { |
| 310 | failf("cannot marshal invalid UTF-8 data as %s", shortTag(tag)) |
| 311 | } |
| 312 | // It can't be encoded directly as YAML so use a binary tag |
| 313 | // and encode it as base64. |
| 314 | tag = yaml_BINARY_TAG |
| 315 | s = encodeBase64(s) |
| 316 | case tag == "": |
| 317 | // Check to see if it would resolve to a specific |
| 318 | // tag when encoded unquoted. If it doesn't, |
| 319 | // there's no need to quote it. |
| 320 | rtag, _ := resolve("", s) |
| 321 | canUsePlain = rtag == yaml_STR_TAG && !isBase60Float(s) |
| 322 | } |
| 323 | // Note: it's possible for user code to emit invalid YAML |
| 324 | // if they explicitly specify a tag and a string containing |
| 325 | // text that's incompatible with that tag. |
| 326 | switch { |
| 327 | case strings.Contains(s, "\n"): |
| 328 | style = yaml_LITERAL_SCALAR_STYLE |
| 329 | case canUsePlain: |
| 330 | style = yaml_PLAIN_SCALAR_STYLE |
| 331 | default: |
| 332 | style = yaml_DOUBLE_QUOTED_SCALAR_STYLE |
| 333 | } |
| 334 | e.emitScalar(s, "", tag, style) |
| 335 | } |
| 336 | |
| 337 | func (e *encoder) boolv(tag string, in reflect.Value) { |
| 338 | var s string |
| 339 | if in.Bool() { |
| 340 | s = "true" |
| 341 | } else { |
| 342 | s = "false" |
| 343 | } |
| 344 | e.emitScalar(s, "", tag, yaml_PLAIN_SCALAR_STYLE) |
| 345 | } |
| 346 | |
| 347 | func (e *encoder) intv(tag string, in reflect.Value) { |
| 348 | s := strconv.FormatInt(in.Int(), 10) |
| 349 | e.emitScalar(s, "", tag, yaml_PLAIN_SCALAR_STYLE) |
| 350 | } |
| 351 | |
| 352 | func (e *encoder) uintv(tag string, in reflect.Value) { |
| 353 | s := strconv.FormatUint(in.Uint(), 10) |
| 354 | e.emitScalar(s, "", tag, yaml_PLAIN_SCALAR_STYLE) |
| 355 | } |
| 356 | |
| 357 | func (e *encoder) timev(tag string, in reflect.Value) { |
| 358 | t := in.Interface().(time.Time) |
| 359 | s := t.Format(time.RFC3339Nano) |
| 360 | e.emitScalar(s, "", tag, yaml_PLAIN_SCALAR_STYLE) |
| 361 | } |
| 362 | |
| 363 | func (e *encoder) floatv(tag string, in reflect.Value) { |
| 364 | // Issue #352: When formatting, use the precision of the underlying value |
| 365 | precision := 64 |
| 366 | if in.Kind() == reflect.Float32 { |
| 367 | precision = 32 |
| 368 | } |
| 369 | |
| 370 | s := strconv.FormatFloat(in.Float(), 'g', -1, precision) |
| 371 | switch s { |
| 372 | case "+Inf": |
| 373 | s = ".inf" |
| 374 | case "-Inf": |
| 375 | s = "-.inf" |
| 376 | case "NaN": |
| 377 | s = ".nan" |
| 378 | } |
| 379 | e.emitScalar(s, "", tag, yaml_PLAIN_SCALAR_STYLE) |
| 380 | } |
| 381 | |
| 382 | func (e *encoder) nilv() { |
| 383 | e.emitScalar("null", "", "", yaml_PLAIN_SCALAR_STYLE) |
| 384 | } |
| 385 | |
| 386 | func (e *encoder) emitScalar(value, anchor, tag string, style yaml_scalar_style_t) { |
| 387 | implicit := tag == "" |
| 388 | e.must(yaml_scalar_event_initialize(&e.event, []byte(anchor), []byte(tag), []byte(value), implicit, implicit, style)) |
| 389 | e.emit() |
| 390 | } |