diff options
Diffstat (limited to 'vendor/gopkg.in/yaml.v3/decode.go')
-rw-r--r-- | vendor/gopkg.in/yaml.v3/decode.go | 931 |
1 files changed, 931 insertions, 0 deletions
diff --git a/vendor/gopkg.in/yaml.v3/decode.go b/vendor/gopkg.in/yaml.v3/decode.go new file mode 100644 index 00000000..be63169b --- /dev/null +++ b/vendor/gopkg.in/yaml.v3/decode.go @@ -0,0 +1,931 @@ +// +// Copyright (c) 2011-2019 Canonical Ltd +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package yaml + +import ( + "encoding" + "encoding/base64" + "fmt" + "io" + "math" + "reflect" + "strconv" + "time" +) + +// ---------------------------------------------------------------------------- +// Parser, produces a node tree out of a libyaml event stream. + +type parser struct { + parser yaml_parser_t + event yaml_event_t + doc *Node + anchors map[string]*Node + doneInit bool +} + +func newParser(b []byte) *parser { + p := parser{} + if !yaml_parser_initialize(&p.parser) { + panic("failed to initialize YAML emitter") + } + if len(b) == 0 { + b = []byte{'\n'} + } + yaml_parser_set_input_string(&p.parser, b) + return &p +} + +func newParserFromReader(r io.Reader) *parser { + p := parser{} + if !yaml_parser_initialize(&p.parser) { + panic("failed to initialize YAML emitter") + } + yaml_parser_set_input_reader(&p.parser, r) + return &p +} + +func (p *parser) init() { + if p.doneInit { + return + } + p.anchors = make(map[string]*Node) + p.expect(yaml_STREAM_START_EVENT) + p.doneInit = true +} + +func (p *parser) destroy() { + if p.event.typ != yaml_NO_EVENT { + yaml_event_delete(&p.event) + } + yaml_parser_delete(&p.parser) +} + +// expect consumes an event from the event stream and +// checks that it's of the expected type. +func (p *parser) expect(e yaml_event_type_t) { + if p.event.typ == yaml_NO_EVENT { + if !yaml_parser_parse(&p.parser, &p.event) { + p.fail() + } + } + if p.event.typ == yaml_STREAM_END_EVENT { + failf("attempted to go past the end of stream; corrupted value?") + } + if p.event.typ != e { + p.parser.problem = fmt.Sprintf("expected %s event but got %s", e, p.event.typ) + p.fail() + } + yaml_event_delete(&p.event) + p.event.typ = yaml_NO_EVENT +} + +// peek peeks at the next event in the event stream, +// puts the results into p.event and returns the event type. +func (p *parser) peek() yaml_event_type_t { + if p.event.typ != yaml_NO_EVENT { + return p.event.typ + } + if !yaml_parser_parse(&p.parser, &p.event) { + p.fail() + } + return p.event.typ +} + +func (p *parser) fail() { + var where string + var line int + if p.parser.problem_mark.line != 0 { + line = p.parser.problem_mark.line + // Scanner errors don't iterate line before returning error + if p.parser.error == yaml_SCANNER_ERROR { + line++ + } + } else if p.parser.context_mark.line != 0 { + line = p.parser.context_mark.line + } + if line != 0 { + where = "line " + strconv.Itoa(line) + ": " + } + var msg string + if len(p.parser.problem) > 0 { + msg = p.parser.problem + } else { + msg = "unknown problem parsing YAML content" + } + failf("%s%s", where, msg) +} + +func (p *parser) anchor(n *Node, anchor []byte) { + if anchor != nil { + n.Anchor = string(anchor) + p.anchors[n.Anchor] = n + } +} + +func (p *parser) parse() *Node { + p.init() + switch p.peek() { + case yaml_SCALAR_EVENT: + return p.scalar() + case yaml_ALIAS_EVENT: + return p.alias() + case yaml_MAPPING_START_EVENT: + return p.mapping() + case yaml_SEQUENCE_START_EVENT: + return p.sequence() + case yaml_DOCUMENT_START_EVENT: + return p.document() + case yaml_STREAM_END_EVENT: + // Happens when attempting to decode an empty buffer. + return nil + case yaml_TAIL_COMMENT_EVENT: + panic("internal error: unexpected tail comment event (please report)") + default: + panic("internal error: attempted to parse unknown event (please report): " + p.event.typ.String()) + } +} + +func (p *parser) node(kind Kind, defaultTag, tag, value string) *Node { + var style Style + if tag != "" && tag != "!" { + tag = shortTag(tag) + style = TaggedStyle + } else if defaultTag != "" { + tag = defaultTag + } else if kind == ScalarNode { + tag, _ = resolve("", value) + } + return &Node{ + Kind: kind, + Tag: tag, + Value: value, + Style: style, + Line: p.event.start_mark.line + 1, + Column: p.event.start_mark.column + 1, + HeadComment: string(p.event.head_comment), + LineComment: string(p.event.line_comment), + FootComment: string(p.event.foot_comment), + } +} + +func (p *parser) parseChild(parent *Node) *Node { + child := p.parse() + parent.Content = append(parent.Content, child) + return child +} + +func (p *parser) document() *Node { + n := p.node(DocumentNode, "", "", "") + p.doc = n + p.expect(yaml_DOCUMENT_START_EVENT) + p.parseChild(n) + if p.peek() == yaml_DOCUMENT_END_EVENT { + n.FootComment = string(p.event.foot_comment) + } + p.expect(yaml_DOCUMENT_END_EVENT) + return n +} + +func (p *parser) alias() *Node { + n := p.node(AliasNode, "", "", string(p.event.anchor)) + n.Alias = p.anchors[n.Value] + if n.Alias == nil { + failf("unknown anchor '%s' referenced", n.Value) + } + p.expect(yaml_ALIAS_EVENT) + return n +} + +func (p *parser) scalar() *Node { + var parsedStyle = p.event.scalar_style() + var nodeStyle Style + switch { + case parsedStyle&yaml_DOUBLE_QUOTED_SCALAR_STYLE != 0: + nodeStyle = DoubleQuotedStyle + case parsedStyle&yaml_SINGLE_QUOTED_SCALAR_STYLE != 0: + nodeStyle = SingleQuotedStyle + case parsedStyle&yaml_LITERAL_SCALAR_STYLE != 0: + nodeStyle = LiteralStyle + case parsedStyle&yaml_FOLDED_SCALAR_STYLE != 0: + nodeStyle = FoldedStyle + } + var nodeValue = string(p.event.value) + var nodeTag = string(p.event.tag) + var defaultTag string + if nodeStyle == 0 { + if nodeValue == "<<" { + defaultTag = mergeTag + } + } else { + defaultTag = strTag + } + n := p.node(ScalarNode, defaultTag, nodeTag, nodeValue) + n.Style |= nodeStyle + p.anchor(n, p.event.anchor) + p.expect(yaml_SCALAR_EVENT) + return n +} + +func (p *parser) sequence() *Node { + n := p.node(SequenceNode, seqTag, string(p.event.tag), "") + if p.event.sequence_style()&yaml_FLOW_SEQUENCE_STYLE != 0 { + n.Style |= FlowStyle + } + p.anchor(n, p.event.anchor) + p.expect(yaml_SEQUENCE_START_EVENT) + for p.peek() != yaml_SEQUENCE_END_EVENT { + p.parseChild(n) + } + n.LineComment = string(p.event.line_comment) + n.FootComment = string(p.event.foot_comment) + p.expect(yaml_SEQUENCE_END_EVENT) + return n +} + +func (p *parser) mapping() *Node { + n := p.node(MappingNode, mapTag, string(p.event.tag), "") + block := true + if p.event.mapping_style()&yaml_FLOW_MAPPING_STYLE != 0 { + block = false + n.Style |= FlowStyle + } + p.anchor(n, p.event.anchor) + p.expect(yaml_MAPPING_START_EVENT) + for p.peek() != yaml_MAPPING_END_EVENT { + k := p.parseChild(n) + if block && k.FootComment != "" { + // Must be a foot comment for the prior value when being dedented. + if len(n.Content) > 2 { + n.Content[len(n.Content)-3].FootComment = k.FootComment + k.FootComment = "" + } + } + v := p.parseChild(n) + if k.FootComment == "" && v.FootComment != "" { + k.FootComment = v.FootComment + v.FootComment = "" + } + if p.peek() == yaml_TAIL_COMMENT_EVENT { + if k.FootComment == "" { + k.FootComment = string(p.event.foot_comment) + } + p.expect(yaml_TAIL_COMMENT_EVENT) + } + } + n.LineComment = string(p.event.line_comment) + n.FootComment = string(p.event.foot_comment) + if n.Style&FlowStyle == 0 && n.FootComment != "" && len(n.Content) > 1 { + n.Content[len(n.Content)-2].FootComment = n.FootComment + n.FootComment = "" + } + p.expect(yaml_MAPPING_END_EVENT) + return n +} + +// ---------------------------------------------------------------------------- +// Decoder, unmarshals a node into a provided value. + +type decoder struct { + doc *Node + aliases map[*Node]bool + terrors []string + + stringMapType reflect.Type + generalMapType reflect.Type + + knownFields bool + uniqueKeys bool + decodeCount int + aliasCount int + aliasDepth int +} + +var ( + nodeType = reflect.TypeOf(Node{}) + durationType = reflect.TypeOf(time.Duration(0)) + stringMapType = reflect.TypeOf(map[string]interface{}{}) + generalMapType = reflect.TypeOf(map[interface{}]interface{}{}) + ifaceType = generalMapType.Elem() + timeType = reflect.TypeOf(time.Time{}) + ptrTimeType = reflect.TypeOf(&time.Time{}) +) + +func newDecoder() *decoder { + d := &decoder{ + stringMapType: stringMapType, + generalMapType: generalMapType, + uniqueKeys: true, + } + d.aliases = make(map[*Node]bool) + return d +} + +func (d *decoder) terror(n *Node, tag string, out reflect.Value) { + if n.Tag != "" { + tag = n.Tag + } + value := n.Value + if tag != seqTag && tag != mapTag { + if len(value) > 10 { + value = " `" + value[:7] + "...`" + } else { + value = " `" + value + "`" + } + } + d.terrors = append(d.terrors, fmt.Sprintf("line %d: cannot unmarshal %s%s into %s", n.Line, shortTag(tag), value, out.Type())) +} + +func (d *decoder) callUnmarshaler(n *Node, u Unmarshaler) (good bool) { + err := u.UnmarshalYAML(n) + if e, ok := err.(*TypeError); ok { + d.terrors = append(d.terrors, e.Errors...) + return false + } + if err != nil { + fail(err) + } + return true +} + +func (d *decoder) callObsoleteUnmarshaler(n *Node, u obsoleteUnmarshaler) (good bool) { + terrlen := len(d.terrors) + err := u.UnmarshalYAML(func(v interface{}) (err error) { + defer handleErr(&err) + d.unmarshal(n, reflect.ValueOf(v)) + if len(d.terrors) > terrlen { + issues := d.terrors[terrlen:] + d.terrors = d.terrors[:terrlen] + return &TypeError{issues} + } + return nil + }) + if e, ok := err.(*TypeError); ok { + d.terrors = append(d.terrors, e.Errors...) + return false + } + if err != nil { + fail(err) + } + return true +} + +// d.prepare initializes and dereferences pointers and calls UnmarshalYAML +// if a value is found to implement it. +// It returns the initialized and dereferenced out value, whether +// unmarshalling was already done by UnmarshalYAML, and if so whether +// its types unmarshalled appropriately. +// +// If n holds a null value, prepare returns before doing anything. +func (d *decoder) prepare(n *Node, out reflect.Value) (newout reflect.Value, unmarshaled, good bool) { + if n.ShortTag() == nullTag { + return out, false, false + } + again := true + for again { + again = false + if out.Kind() == reflect.Ptr { + if out.IsNil() { + out.Set(reflect.New(out.Type().Elem())) + } + out = out.Elem() + again = true + } + if out.CanAddr() { + outi := out.Addr().Interface() + if u, ok := outi.(Unmarshaler); ok { + good = d.callUnmarshaler(n, u) + return out, true, good + } + if u, ok := outi.(obsoleteUnmarshaler); ok { + good = d.callObsoleteUnmarshaler(n, u) + return out, true, good + } + } + } + return out, false, false +} + +func (d *decoder) fieldByIndex(n *Node, v reflect.Value, index []int) (field reflect.Value) { + if n.ShortTag() == nullTag { + return reflect.Value{} + } + for _, num := range index { + for { + if v.Kind() == reflect.Ptr { + if v.IsNil() { + v.Set(reflect.New(v.Type().Elem())) + } + v = v.Elem() + continue + } + break + } + v = v.Field(num) + } + return v +} + +const ( + // 400,000 decode operations is ~500kb of dense object declarations, or + // ~5kb of dense object declarations with 10000% alias expansion + alias_ratio_range_low = 400000 + + // 4,000,000 decode operations is ~5MB of dense object declarations, or + // ~4.5MB of dense object declarations with 10% alias expansion + alias_ratio_range_high = 4000000 + + // alias_ratio_range is the range over which we scale allowed alias ratios + alias_ratio_range = float64(alias_ratio_range_high - alias_ratio_range_low) +) + +func allowedAliasRatio(decodeCount int) float64 { + switch { + case decodeCount <= alias_ratio_range_low: + // allow 99% to come from alias expansion for small-to-medium documents + return 0.99 + case decodeCount >= alias_ratio_range_high: + // allow 10% to come from alias expansion for very large documents + return 0.10 + default: + // scale smoothly from 99% down to 10% over the range. + // this maps to 396,000 - 400,000 allowed alias-driven decodes over the range. + // 400,000 decode operations is ~100MB of allocations in worst-case scenarios (single-item maps). + return 0.99 - 0.89*(float64(decodeCount-alias_ratio_range_low)/alias_ratio_range) + } +} + +func (d *decoder) unmarshal(n *Node, out reflect.Value) (good bool) { + d.decodeCount++ + if d.aliasDepth > 0 { + d.aliasCount++ + } + if d.aliasCount > 100 && d.decodeCount > 1000 && float64(d.aliasCount)/float64(d.decodeCount) > allowedAliasRatio(d.decodeCount) { + failf("document contains excessive aliasing") + } + if out.Type() == nodeType { + out.Set(reflect.ValueOf(n).Elem()) + return true + } + switch n.Kind { + case DocumentNode: + return d.document(n, out) + case AliasNode: + return d.alias(n, out) + } + out, unmarshaled, good := d.prepare(n, out) + if unmarshaled { + return good + } + switch n.Kind { + case ScalarNode: + good = d.scalar(n, out) + case MappingNode: + good = d.mapping(n, out) + case SequenceNode: + good = d.sequence(n, out) + default: + panic("internal error: unknown node kind: " + strconv.Itoa(int(n.Kind))) + } + return good +} + +func (d *decoder) document(n *Node, out reflect.Value) (good bool) { + if len(n.Content) == 1 { + d.doc = n + d.unmarshal(n.Content[0], out) + return true + } + return false +} + +func (d *decoder) alias(n *Node, out reflect.Value) (good bool) { + if d.aliases[n] { + // TODO this could actually be allowed in some circumstances. + failf("anchor '%s' value contains itself", n.Value) + } + d.aliases[n] = true + d.aliasDepth++ + good = d.unmarshal(n.Alias, out) + d.aliasDepth-- + delete(d.aliases, n) + return good +} + +var zeroValue reflect.Value + +func resetMap(out reflect.Value) { + for _, k := range out.MapKeys() { + out.SetMapIndex(k, zeroValue) + } +} + +func (d *decoder) scalar(n *Node, out reflect.Value) bool { + var tag string + var resolved interface{} + if n.indicatedString() { + tag = strTag + resolved = n.Value + } else { + tag, resolved = resolve(n.Tag, n.Value) + if tag == binaryTag { + data, err := base64.StdEncoding.DecodeString(resolved.(string)) + if err != nil { + failf("!!binary value contains invalid base64 data") + } + resolved = string(data) + } + } + if resolved == nil { + if out.CanAddr() { + switch out.Kind() { + case reflect.Interface, reflect.Ptr, reflect.Map, reflect.Slice: + out.Set(reflect.Zero(out.Type())) + return true + } + } + return false + } + if resolvedv := reflect.ValueOf(resolved); out.Type() == resolvedv.Type() { + // We've resolved to exactly the type we want, so use that. + out.Set(resolvedv) + return true + } + // Perhaps we can use the value as a TextUnmarshaler to + // set its value. + if out.CanAddr() { + u, ok := out.Addr().Interface().(encoding.TextUnmarshaler) + if ok { + var text []byte + if tag == binaryTag { + text = []byte(resolved.(string)) + } else { + // We let any value be unmarshaled into TextUnmarshaler. + // That might be more lax than we'd like, but the + // TextUnmarshaler itself should bowl out any dubious values. + text = []byte(n.Value) + } + err := u.UnmarshalText(text) + if err != nil { + fail(err) + } + return true + } + } + switch out.Kind() { + case reflect.String: + if tag == binaryTag { + out.SetString(resolved.(string)) + return true + } + out.SetString(n.Value) + return true + case reflect.Interface: + out.Set(reflect.ValueOf(resolved)) + return true + case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64: + // This used to work in v2, but it's very unfriendly. + isDuration := out.Type() == durationType + + switch resolved := resolved.(type) { + case int: + if !isDuration && !out.OverflowInt(int64(resolved)) { + out.SetInt(int64(resolved)) + return true + } + case int64: + if !isDuration && !out.OverflowInt(resolved) { + out.SetInt(resolved) + return true + } + case uint64: + if !isDuration && resolved <= math.MaxInt64 && !out.OverflowInt(int64(resolved)) { + out.SetInt(int64(resolved)) + return true + } + case float64: + if !isDuration && resolved <= math.MaxInt64 && !out.OverflowInt(int64(resolved)) { + out.SetInt(int64(resolved)) + return true + } + case string: + if out.Type() == durationType { + d, err := time.ParseDuration(resolved) + if err == nil { + out.SetInt(int64(d)) + return true + } + } + } + case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr: + switch resolved := resolved.(type) { + case int: + if resolved >= 0 && !out.OverflowUint(uint64(resolved)) { + out.SetUint(uint64(resolved)) + return true + } + case int64: + if resolved >= 0 && !out.OverflowUint(uint64(resolved)) { + out.SetUint(uint64(resolved)) + return true + } + case uint64: + if !out.OverflowUint(uint64(resolved)) { + out.SetUint(uint64(resolved)) + return true + } + case float64: + if resolved <= math.MaxUint64 && !out.OverflowUint(uint64(resolved)) { + out.SetUint(uint64(resolved)) + return true + } + } + case reflect.Bool: + switch resolved := resolved.(type) { + case bool: + out.SetBool(resolved) + return true + case string: + // This offers some compatibility with the 1.1 spec (https://yaml.org/type/bool.html). + // It only works if explicitly attempting to unmarshal into a typed bool value. + switch resolved { + case "y", "Y", "yes", "Yes", "YES", "on", "On", "ON": + out.SetBool(true) + return true + case "n", "N", "no", "No", "NO", "off", "Off", "OFF": + out.SetBool(false) + return true + } + } + case reflect.Float32, reflect.Float64: + switch resolved := resolved.(type) { + case int: + out.SetFloat(float64(resolved)) + return true + case int64: + out.SetFloat(float64(resolved)) + return true + case uint64: + out.SetFloat(float64(resolved)) + return true + case float64: + out.SetFloat(resolved) + return true + } + case reflect.Struct: + if resolvedv := reflect.ValueOf(resolved); out.Type() == resolvedv.Type() { + out.Set(resolvedv) + return true + } + case reflect.Ptr: + panic("yaml internal error: please report the issue") + } + d.terror(n, tag, out) + return false +} + +func settableValueOf(i interface{}) reflect.Value { + v := reflect.ValueOf(i) + sv := reflect.New(v.Type()).Elem() + sv.Set(v) + return sv +} + +func (d *decoder) sequence(n *Node, out reflect.Value) (good bool) { + l := len(n.Content) + + var iface reflect.Value + switch out.Kind() { + case reflect.Slice: + out.Set(reflect.MakeSlice(out.Type(), l, l)) + case reflect.Array: + if l != out.Len() { + failf("invalid array: want %d elements but got %d", out.Len(), l) + } + case reflect.Interface: + // No type hints. Will have to use a generic sequence. + iface = out + out = settableValueOf(make([]interface{}, l)) + default: + d.terror(n, seqTag, out) + return false + } + et := out.Type().Elem() + + j := 0 + for i := 0; i < l; i++ { + e := reflect.New(et).Elem() + if ok := d.unmarshal(n.Content[i], e); ok { + out.Index(j).Set(e) + j++ + } + } + if out.Kind() != reflect.Array { + out.Set(out.Slice(0, j)) + } + if iface.IsValid() { + iface.Set(out) + } + return true +} + +func (d *decoder) mapping(n *Node, out reflect.Value) (good bool) { + l := len(n.Content) + if d.uniqueKeys { + nerrs := len(d.terrors) + for i := 0; i < l; i += 2 { + ni := n.Content[i] + for j := i + 2; j < l; j += 2 { + nj := n.Content[j] + if ni.Kind == nj.Kind && ni.Value == nj.Value { + d.terrors = append(d.terrors, fmt.Sprintf("line %d: mapping key %#v already defined at line %d", nj.Line, nj.Value, ni.Line)) + } + } + } + if len(d.terrors) > nerrs { + return false + } + } + switch out.Kind() { + case reflect.Struct: + return d.mappingStruct(n, out) + case reflect.Map: + // okay + case reflect.Interface: + iface := out + if isStringMap(n) { + out = reflect.MakeMap(d.stringMapType) + } else { + out = reflect.MakeMap(d.generalMapType) + } + iface.Set(out) + default: + d.terror(n, mapTag, out) + return false + } + + outt := out.Type() + kt := outt.Key() + et := outt.Elem() + + stringMapType := d.stringMapType + generalMapType := d.generalMapType + if outt.Elem() == ifaceType { + if outt.Key().Kind() == reflect.String { + d.stringMapType = outt + } else if outt.Key() == ifaceType { + d.generalMapType = outt + } + } + + if out.IsNil() { + out.Set(reflect.MakeMap(outt)) + } + for i := 0; i < l; i += 2 { + if isMerge(n.Content[i]) { + d.merge(n.Content[i+1], out) + continue + } + k := reflect.New(kt).Elem() + if d.unmarshal(n.Content[i], k) { + kkind := k.Kind() + if kkind == reflect.Interface { + kkind = k.Elem().Kind() + } + if kkind == reflect.Map || kkind == reflect.Slice { + failf("invalid map key: %#v", k.Interface()) + } + e := reflect.New(et).Elem() + if d.unmarshal(n.Content[i+1], e) { + out.SetMapIndex(k, e) + } + } + } + d.stringMapType = stringMapType + d.generalMapType = generalMapType + return true +} + +func isStringMap(n *Node) bool { + if n.Kind != MappingNode { + return false + } + l := len(n.Content) + for i := 0; i < l; i += 2 { + if n.Content[i].ShortTag() != strTag { + return false + } + } + return true +} + +func (d *decoder) mappingStruct(n *Node, out reflect.Value) (good bool) { + sinfo, err := getStructInfo(out.Type()) + if err != nil { + panic(err) + } + + var inlineMap reflect.Value + var elemType reflect.Type + if sinfo.InlineMap != -1 { + inlineMap = out.Field(sinfo.InlineMap) + inlineMap.Set(reflect.New(inlineMap.Type()).Elem()) + elemType = inlineMap.Type().Elem() + } + + for _, index := range sinfo.InlineUnmarshalers { + field := d.fieldByIndex(n, out, index) + d.prepare(n, field) + } + + var doneFields []bool + if d.uniqueKeys { + doneFields = make([]bool, len(sinfo.FieldsList)) + } + name := settableValueOf("") + l := len(n.Content) + for i := 0; i < l; i += 2 { + ni := n.Content[i] + if isMerge(ni) { + d.merge(n.Content[i+1], out) + continue + } + if !d.unmarshal(ni, name) { + continue + } + if info, ok := sinfo.FieldsMap[name.String()]; ok { + if d.uniqueKeys { + if doneFields[info.Id] { + d.terrors = append(d.terrors, fmt.Sprintf("line %d: field %s already set in type %s", ni.Line, name.String(), out.Type())) + continue + } + doneFields[info.Id] = true + } + var field reflect.Value + if info.Inline == nil { + field = out.Field(info.Num) + } else { + field = d.fieldByIndex(n, out, info.Inline) + } + d.unmarshal(n.Content[i+1], field) + } else if sinfo.InlineMap != -1 { + if inlineMap.IsNil() { + inlineMap.Set(reflect.MakeMap(inlineMap.Type())) + } + value := reflect.New(elemType).Elem() + d.unmarshal(n.Content[i+1], value) + inlineMap.SetMapIndex(name, value) + } else if d.knownFields { + d.terrors = append(d.terrors, fmt.Sprintf("line %d: field %s not found in type %s", ni.Line, name.String(), out.Type())) + } + } + return true +} + +func failWantMap() { + failf("map merge requires map or sequence of maps as the value") +} + +func (d *decoder) merge(n *Node, out reflect.Value) { + switch n.Kind { + case MappingNode: + d.unmarshal(n, out) + case AliasNode: + if n.Alias != nil && n.Alias.Kind != MappingNode { + failWantMap() + } + d.unmarshal(n, out) + case SequenceNode: + // Step backwards as earlier nodes take precedence. + for i := len(n.Content) - 1; i >= 0; i-- { + ni := n.Content[i] + if ni.Kind == AliasNode { + if ni.Alias != nil && ni.Alias.Kind != MappingNode { + failWantMap() + } + } else if ni.Kind != MappingNode { + failWantMap() + } + d.unmarshal(ni, out) + } + default: + failWantMap() + } +} + +func isMerge(n *Node) bool { + return n.Kind == ScalarNode && n.Value == "<<" && (n.Tag == "" || n.Tag == "!" || shortTag(n.Tag) == mergeTag) +} |