1
2
3
4
5
6
7 package xml
8
9
10
11
12
13 import (
14 "bufio"
15 "bytes"
16 "errors"
17 "fmt"
18 "io"
19 "strconv"
20 "strings"
21 "unicode"
22 "unicode/utf8"
23 )
24
25
26 type SyntaxError struct {
27 Msg string
28 Line int
29 }
30
31 func (e *SyntaxError) Error() string {
32 return "XML syntax error on line " + strconv.Itoa(e.Line) + ": " + e.Msg
33 }
34
35
36
37
38
39
40 type Name struct {
41 Space, Local string
42 }
43
44
45 type Attr struct {
46 Name Name
47 Value string
48 }
49
50
51
52 type Token any
53
54
55 type StartElement struct {
56 Name Name
57 Attr []Attr
58 }
59
60
61 func (e StartElement) Copy() StartElement {
62 attrs := make([]Attr, len(e.Attr))
63 copy(attrs, e.Attr)
64 e.Attr = attrs
65 return e
66 }
67
68
69 func (e StartElement) End() EndElement {
70 return EndElement{e.Name}
71 }
72
73
74 type EndElement struct {
75 Name Name
76 }
77
78
79
80
81 type CharData []byte
82
83
84 func (c CharData) Copy() CharData { return CharData(bytes.Clone(c)) }
85
86
87
88 type Comment []byte
89
90
91 func (c Comment) Copy() Comment { return Comment(bytes.Clone(c)) }
92
93
94 type ProcInst struct {
95 Target string
96 Inst []byte
97 }
98
99
100 func (p ProcInst) Copy() ProcInst {
101 p.Inst = bytes.Clone(p.Inst)
102 return p
103 }
104
105
106
107 type Directive []byte
108
109
110 func (d Directive) Copy() Directive { return Directive(bytes.Clone(d)) }
111
112
113 func CopyToken(t Token) Token {
114 switch v := t.(type) {
115 case CharData:
116 return v.Copy()
117 case Comment:
118 return v.Copy()
119 case Directive:
120 return v.Copy()
121 case ProcInst:
122 return v.Copy()
123 case StartElement:
124 return v.Copy()
125 }
126 return t
127 }
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142 type TokenReader interface {
143 Token() (Token, error)
144 }
145
146
147
148 type Decoder struct {
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170 Strict bool
171
172
173
174
175 AutoClose []string
176
177
178
179
180
181
182
183
184
185
186 Entity map[string]string
187
188
189
190
191
192
193 CharsetReader func(charset string, input io.Reader) (io.Reader, error)
194
195
196
197
198 DefaultSpace string
199
200 r io.ByteReader
201 t TokenReader
202 buf bytes.Buffer
203 saved *bytes.Buffer
204 stk *stack
205 stkDepth int
206 free *stack
207 needClose bool
208 toClose Name
209 nextToken Token
210 nextByte int
211 ns map[string]string
212 err error
213 line int
214 linestart int64
215 offset int64
216 inUnmarshalXML bool
217 }
218
219
220
221
222 func NewDecoder(r io.Reader) *Decoder {
223 d := &Decoder{
224 ns: make(map[string]string),
225 nextByte: -1,
226 line: 1,
227 Strict: true,
228 }
229 d.switchToReader(r)
230 return d
231 }
232
233
234 func NewTokenDecoder(t TokenReader) *Decoder {
235
236 if d, ok := t.(*Decoder); ok {
237 return d
238 }
239 d := &Decoder{
240 ns: make(map[string]string),
241 t: t,
242 nextByte: -1,
243 line: 1,
244 Strict: true,
245 }
246 return d
247 }
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275 func (d *Decoder) Token() (Token, error) {
276 var t Token
277 var err error
278 if d.stk != nil && d.stk.kind == stkEOF {
279 return nil, io.EOF
280 }
281 if d.nextToken != nil {
282 t = d.nextToken
283 d.nextToken = nil
284 } else {
285 if t, err = d.rawToken(); t == nil && err != nil {
286 if err == io.EOF && d.stk != nil && d.stk.kind != stkEOF {
287 err = d.syntaxError("unexpected EOF")
288 }
289 return nil, err
290 }
291
292
293 err = nil
294 }
295 if !d.Strict {
296 if t1, ok := d.autoClose(t); ok {
297 d.nextToken = t
298 t = t1
299 }
300 }
301 switch t1 := t.(type) {
302 case StartElement:
303
304
305
306
307 for _, a := range t1.Attr {
308 if a.Name.Space == xmlnsPrefix {
309 v, ok := d.ns[a.Name.Local]
310 d.pushNs(a.Name.Local, v, ok)
311 d.ns[a.Name.Local] = a.Value
312 }
313 if a.Name.Space == "" && a.Name.Local == xmlnsPrefix {
314
315 v, ok := d.ns[""]
316 d.pushNs("", v, ok)
317 d.ns[""] = a.Value
318 }
319 }
320
321 d.pushElement(t1.Name)
322 d.translate(&t1.Name, true)
323 for i := range t1.Attr {
324 d.translate(&t1.Attr[i].Name, false)
325 }
326 t = t1
327
328 case EndElement:
329 if !d.popElement(&t1) {
330 return nil, d.err
331 }
332 t = t1
333 }
334 return t, err
335 }
336
337 const (
338 xmlURL = "http://www.w3.org/XML/1998/namespace"
339 xmlnsPrefix = "xmlns"
340 xmlPrefix = "xml"
341 )
342
343
344
345
346 func (d *Decoder) translate(n *Name, isElementName bool) {
347 switch {
348 case n.Space == xmlnsPrefix:
349 return
350 case n.Space == "" && !isElementName:
351 return
352 case n.Space == xmlPrefix:
353 n.Space = xmlURL
354 case n.Space == "" && n.Local == xmlnsPrefix:
355 return
356 }
357 if v, ok := d.ns[n.Space]; ok {
358 n.Space = v
359 } else if n.Space == "" {
360 n.Space = d.DefaultSpace
361 }
362 }
363
364 func (d *Decoder) switchToReader(r io.Reader) {
365
366
367
368
369 if rb, ok := r.(io.ByteReader); ok {
370 d.r = rb
371 } else {
372 d.r = bufio.NewReader(r)
373 }
374 }
375
376
377
378
379
380 type stack struct {
381 next *stack
382 kind int
383 name Name
384 ok bool
385 }
386
387 const (
388 stkStart = iota
389 stkNs
390 stkEOF
391 )
392
393 func (d *Decoder) push(kind int) *stack {
394 s := d.free
395 if s != nil {
396 d.free = s.next
397 } else {
398 s = new(stack)
399 }
400 s.next = d.stk
401 s.kind = kind
402 if kind == stkStart {
403 d.stkDepth++
404 }
405 d.stk = s
406 return s
407 }
408
409 func (d *Decoder) pop() *stack {
410 s := d.stk
411 if s != nil {
412 if s.kind == stkStart {
413 d.stkDepth--
414 }
415 d.stk = s.next
416 s.next = d.free
417 d.free = s
418 }
419 return s
420 }
421
422
423
424
425 func (d *Decoder) pushEOF() {
426
427
428
429 start := d.stk
430 for start.kind != stkStart {
431 start = start.next
432 }
433
434
435 for start.next != nil && start.next.kind == stkNs {
436 start = start.next
437 }
438 s := d.free
439 if s != nil {
440 d.free = s.next
441 } else {
442 s = new(stack)
443 }
444 s.kind = stkEOF
445 s.next = start.next
446 start.next = s
447 }
448
449
450
451 func (d *Decoder) popEOF() bool {
452 if d.stk == nil || d.stk.kind != stkEOF {
453 return false
454 }
455 d.pop()
456 return true
457 }
458
459
460 func (d *Decoder) pushElement(name Name) {
461 s := d.push(stkStart)
462 s.name = name
463 }
464
465
466
467 func (d *Decoder) pushNs(local string, url string, ok bool) {
468 s := d.push(stkNs)
469 s.name.Local = local
470 s.name.Space = url
471 s.ok = ok
472 }
473
474
475 func (d *Decoder) syntaxError(msg string) error {
476 return &SyntaxError{Msg: msg, Line: d.line}
477 }
478
479
480
481
482
483
484
485 func (d *Decoder) popElement(t *EndElement) bool {
486 s := d.pop()
487 name := t.Name
488 switch {
489 case s == nil || s.kind != stkStart:
490 d.err = d.syntaxError("unexpected end element </" + name.Local + ">")
491 return false
492 case s.name.Local != name.Local:
493 if !d.Strict {
494 d.needClose = true
495 d.toClose = t.Name
496 t.Name = s.name
497 return true
498 }
499 d.err = d.syntaxError("element <" + s.name.Local + "> closed by </" + name.Local + ">")
500 return false
501 case s.name.Space != name.Space:
502 ns := name.Space
503 if name.Space == "" {
504 ns = `""`
505 }
506 d.err = d.syntaxError("element <" + s.name.Local + "> in space " + s.name.Space +
507 " closed by </" + name.Local + "> in space " + ns)
508 return false
509 }
510
511 d.translate(&t.Name, true)
512
513
514
515 for d.stk != nil && d.stk.kind != stkStart && d.stk.kind != stkEOF {
516 s := d.pop()
517 if s.ok {
518 d.ns[s.name.Local] = s.name.Space
519 } else {
520 delete(d.ns, s.name.Local)
521 }
522 }
523
524 return true
525 }
526
527
528
529 func (d *Decoder) autoClose(t Token) (Token, bool) {
530 if d.stk == nil || d.stk.kind != stkStart {
531 return nil, false
532 }
533 for _, s := range d.AutoClose {
534 if strings.EqualFold(s, d.stk.name.Local) {
535
536 et, ok := t.(EndElement)
537 if !ok || !strings.EqualFold(et.Name.Local, d.stk.name.Local) {
538 return EndElement{d.stk.name}, true
539 }
540 break
541 }
542 }
543 return nil, false
544 }
545
546 var errRawToken = errors.New("xml: cannot use RawToken from UnmarshalXML method")
547
548
549
550
551 func (d *Decoder) RawToken() (Token, error) {
552 if d.inUnmarshalXML {
553 return nil, errRawToken
554 }
555 return d.rawToken()
556 }
557
558 func (d *Decoder) rawToken() (Token, error) {
559 if d.t != nil {
560 return d.t.Token()
561 }
562 if d.err != nil {
563 return nil, d.err
564 }
565 if d.needClose {
566
567
568
569 d.needClose = false
570 return EndElement{d.toClose}, nil
571 }
572
573 b, ok := d.getc()
574 if !ok {
575 return nil, d.err
576 }
577
578 if b != '<' {
579
580 d.ungetc(b)
581 data := d.text(-1, false)
582 if data == nil {
583 return nil, d.err
584 }
585 return CharData(data), nil
586 }
587
588 if b, ok = d.mustgetc(); !ok {
589 return nil, d.err
590 }
591 switch b {
592 case '/':
593
594 var name Name
595 if name, ok = d.nsname(); !ok {
596 if d.err == nil {
597 d.err = d.syntaxError("expected element name after </")
598 }
599 return nil, d.err
600 }
601 d.space()
602 if b, ok = d.mustgetc(); !ok {
603 return nil, d.err
604 }
605 if b != '>' {
606 d.err = d.syntaxError("invalid characters between </" + name.Local + " and >")
607 return nil, d.err
608 }
609 return EndElement{name}, nil
610
611 case '?':
612
613 var target string
614 if target, ok = d.name(); !ok {
615 if d.err == nil {
616 d.err = d.syntaxError("expected target name after <?")
617 }
618 return nil, d.err
619 }
620 d.space()
621 d.buf.Reset()
622 var b0 byte
623 for {
624 if b, ok = d.mustgetc(); !ok {
625 return nil, d.err
626 }
627 d.buf.WriteByte(b)
628 if b0 == '?' && b == '>' {
629 break
630 }
631 b0 = b
632 }
633 data := d.buf.Bytes()
634 data = data[0 : len(data)-2]
635
636 if target == "xml" {
637 content := string(data)
638 ver := procInst("version", content)
639 if ver != "" && ver != "1.0" {
640 d.err = fmt.Errorf("xml: unsupported version %q; only version 1.0 is supported", ver)
641 return nil, d.err
642 }
643 enc := procInst("encoding", content)
644 if enc != "" && enc != "utf-8" && enc != "UTF-8" && !strings.EqualFold(enc, "utf-8") {
645 if d.CharsetReader == nil {
646 d.err = fmt.Errorf("xml: encoding %q declared but Decoder.CharsetReader is nil", enc)
647 return nil, d.err
648 }
649 newr, err := d.CharsetReader(enc, d.r.(io.Reader))
650 if err != nil {
651 d.err = fmt.Errorf("xml: opening charset %q: %w", enc, err)
652 return nil, d.err
653 }
654 if newr == nil {
655 panic("CharsetReader returned a nil Reader for charset " + enc)
656 }
657 d.switchToReader(newr)
658 }
659 }
660 return ProcInst{target, data}, nil
661
662 case '!':
663
664 if b, ok = d.mustgetc(); !ok {
665 return nil, d.err
666 }
667 switch b {
668 case '-':
669
670 if b, ok = d.mustgetc(); !ok {
671 return nil, d.err
672 }
673 if b != '-' {
674 d.err = d.syntaxError("invalid sequence <!- not part of <!--")
675 return nil, d.err
676 }
677
678 d.buf.Reset()
679 var b0, b1 byte
680 for {
681 if b, ok = d.mustgetc(); !ok {
682 return nil, d.err
683 }
684 d.buf.WriteByte(b)
685 if b0 == '-' && b1 == '-' {
686 if b != '>' {
687 d.err = d.syntaxError(
688 `invalid sequence "--" not allowed in comments`)
689 return nil, d.err
690 }
691 break
692 }
693 b0, b1 = b1, b
694 }
695 data := d.buf.Bytes()
696 data = data[0 : len(data)-3]
697 return Comment(data), nil
698
699 case '[':
700
701 for i := 0; i < 6; i++ {
702 if b, ok = d.mustgetc(); !ok {
703 return nil, d.err
704 }
705 if b != "CDATA["[i] {
706 d.err = d.syntaxError("invalid <![ sequence")
707 return nil, d.err
708 }
709 }
710
711 data := d.text(-1, true)
712 if data == nil {
713 return nil, d.err
714 }
715 return CharData(data), nil
716 }
717
718
719
720
721 d.buf.Reset()
722 d.buf.WriteByte(b)
723 inquote := uint8(0)
724 depth := 0
725 for {
726 if b, ok = d.mustgetc(); !ok {
727 return nil, d.err
728 }
729 if inquote == 0 && b == '>' && depth == 0 {
730 break
731 }
732 HandleB:
733 d.buf.WriteByte(b)
734 switch {
735 case b == inquote:
736 inquote = 0
737
738 case inquote != 0:
739
740
741 case b == '\'' || b == '"':
742 inquote = b
743
744 case b == '>' && inquote == 0:
745 depth--
746
747 case b == '<' && inquote == 0:
748
749 s := "!--"
750 for i := 0; i < len(s); i++ {
751 if b, ok = d.mustgetc(); !ok {
752 return nil, d.err
753 }
754 if b != s[i] {
755 for j := 0; j < i; j++ {
756 d.buf.WriteByte(s[j])
757 }
758 depth++
759 goto HandleB
760 }
761 }
762
763
764 d.buf.Truncate(d.buf.Len() - 1)
765
766
767 var b0, b1 byte
768 for {
769 if b, ok = d.mustgetc(); !ok {
770 return nil, d.err
771 }
772 if b0 == '-' && b1 == '-' && b == '>' {
773 break
774 }
775 b0, b1 = b1, b
776 }
777
778
779
780
781
782 d.buf.WriteByte(' ')
783 }
784 }
785 return Directive(d.buf.Bytes()), nil
786 }
787
788
789 d.ungetc(b)
790
791 var (
792 name Name
793 empty bool
794 attr []Attr
795 )
796 if name, ok = d.nsname(); !ok {
797 if d.err == nil {
798 d.err = d.syntaxError("expected element name after <")
799 }
800 return nil, d.err
801 }
802
803 attr = []Attr{}
804 for {
805 d.space()
806 if b, ok = d.mustgetc(); !ok {
807 return nil, d.err
808 }
809 if b == '/' {
810 empty = true
811 if b, ok = d.mustgetc(); !ok {
812 return nil, d.err
813 }
814 if b != '>' {
815 d.err = d.syntaxError("expected /> in element")
816 return nil, d.err
817 }
818 break
819 }
820 if b == '>' {
821 break
822 }
823 d.ungetc(b)
824
825 a := Attr{}
826 if a.Name, ok = d.nsname(); !ok {
827 if d.err == nil {
828 d.err = d.syntaxError("expected attribute name in element")
829 }
830 return nil, d.err
831 }
832 d.space()
833 if b, ok = d.mustgetc(); !ok {
834 return nil, d.err
835 }
836 if b != '=' {
837 if d.Strict {
838 d.err = d.syntaxError("attribute name without = in element")
839 return nil, d.err
840 }
841 d.ungetc(b)
842 a.Value = a.Name.Local
843 } else {
844 d.space()
845 data := d.attrval()
846 if data == nil {
847 return nil, d.err
848 }
849 a.Value = string(data)
850 }
851 attr = append(attr, a)
852 }
853 if empty {
854 d.needClose = true
855 d.toClose = name
856 }
857 return StartElement{name, attr}, nil
858 }
859
860 func (d *Decoder) attrval() []byte {
861 b, ok := d.mustgetc()
862 if !ok {
863 return nil
864 }
865
866 if b == '"' || b == '\'' {
867 return d.text(int(b), false)
868 }
869
870 if d.Strict {
871 d.err = d.syntaxError("unquoted or missing attribute value in element")
872 return nil
873 }
874
875 d.ungetc(b)
876 d.buf.Reset()
877 for {
878 b, ok = d.mustgetc()
879 if !ok {
880 return nil
881 }
882
883 if 'a' <= b && b <= 'z' || 'A' <= b && b <= 'Z' ||
884 '0' <= b && b <= '9' || b == '_' || b == ':' || b == '-' {
885 d.buf.WriteByte(b)
886 } else {
887 d.ungetc(b)
888 break
889 }
890 }
891 return d.buf.Bytes()
892 }
893
894
895 func (d *Decoder) space() {
896 for {
897 b, ok := d.getc()
898 if !ok {
899 return
900 }
901 switch b {
902 case ' ', '\r', '\n', '\t':
903 default:
904 d.ungetc(b)
905 return
906 }
907 }
908 }
909
910
911
912
913
914 func (d *Decoder) getc() (b byte, ok bool) {
915 if d.err != nil {
916 return 0, false
917 }
918 if d.nextByte >= 0 {
919 b = byte(d.nextByte)
920 d.nextByte = -1
921 } else {
922 b, d.err = d.r.ReadByte()
923 if d.err != nil {
924 return 0, false
925 }
926 if d.saved != nil {
927 d.saved.WriteByte(b)
928 }
929 }
930 if b == '\n' {
931 d.line++
932 d.linestart = d.offset + 1
933 }
934 d.offset++
935 return b, true
936 }
937
938
939
940
941 func (d *Decoder) InputOffset() int64 {
942 return d.offset
943 }
944
945
946
947
948 func (d *Decoder) InputPos() (line, column int) {
949 return d.line, int(d.offset-d.linestart) + 1
950 }
951
952
953
954 func (d *Decoder) savedOffset() int {
955 n := d.saved.Len()
956 if d.nextByte >= 0 {
957 n--
958 }
959 return n
960 }
961
962
963
964
965
966 func (d *Decoder) mustgetc() (b byte, ok bool) {
967 if b, ok = d.getc(); !ok {
968 if d.err == io.EOF {
969 d.err = d.syntaxError("unexpected EOF")
970 }
971 }
972 return
973 }
974
975
976 func (d *Decoder) ungetc(b byte) {
977 if b == '\n' {
978 d.line--
979 }
980 d.nextByte = int(b)
981 d.offset--
982 }
983
984 var entity = map[string]rune{
985 "lt": '<',
986 "gt": '>',
987 "amp": '&',
988 "apos": '\'',
989 "quot": '"',
990 }
991
992
993
994
995
996 func (d *Decoder) text(quote int, cdata bool) []byte {
997 var b0, b1 byte
998 var trunc int
999 d.buf.Reset()
1000 Input:
1001 for {
1002 b, ok := d.getc()
1003 if !ok {
1004 if cdata {
1005 if d.err == io.EOF {
1006 d.err = d.syntaxError("unexpected EOF in CDATA section")
1007 }
1008 return nil
1009 }
1010 break Input
1011 }
1012
1013
1014
1015
1016 if quote < 0 && b0 == ']' && b1 == ']' && b == '>' {
1017 if cdata {
1018 trunc = 2
1019 break Input
1020 }
1021 d.err = d.syntaxError("unescaped ]]> not in CDATA section")
1022 return nil
1023 }
1024
1025
1026 if b == '<' && !cdata {
1027 if quote >= 0 {
1028 d.err = d.syntaxError("unescaped < inside quoted string")
1029 return nil
1030 }
1031 d.ungetc('<')
1032 break Input
1033 }
1034 if quote >= 0 && b == byte(quote) {
1035 break Input
1036 }
1037 if b == '&' && !cdata {
1038
1039
1040
1041
1042
1043 before := d.buf.Len()
1044 d.buf.WriteByte('&')
1045 var ok bool
1046 var text string
1047 var haveText bool
1048 if b, ok = d.mustgetc(); !ok {
1049 return nil
1050 }
1051 if b == '#' {
1052 d.buf.WriteByte(b)
1053 if b, ok = d.mustgetc(); !ok {
1054 return nil
1055 }
1056 base := 10
1057 if b == 'x' {
1058 base = 16
1059 d.buf.WriteByte(b)
1060 if b, ok = d.mustgetc(); !ok {
1061 return nil
1062 }
1063 }
1064 start := d.buf.Len()
1065 for '0' <= b && b <= '9' ||
1066 base == 16 && 'a' <= b && b <= 'f' ||
1067 base == 16 && 'A' <= b && b <= 'F' {
1068 d.buf.WriteByte(b)
1069 if b, ok = d.mustgetc(); !ok {
1070 return nil
1071 }
1072 }
1073 if b != ';' {
1074 d.ungetc(b)
1075 } else {
1076 s := string(d.buf.Bytes()[start:])
1077 d.buf.WriteByte(';')
1078 n, err := strconv.ParseUint(s, base, 64)
1079 if err == nil && n <= unicode.MaxRune {
1080 text = string(rune(n))
1081 haveText = true
1082 }
1083 }
1084 } else {
1085 d.ungetc(b)
1086 if !d.readName() {
1087 if d.err != nil {
1088 return nil
1089 }
1090 }
1091 if b, ok = d.mustgetc(); !ok {
1092 return nil
1093 }
1094 if b != ';' {
1095 d.ungetc(b)
1096 } else {
1097 name := d.buf.Bytes()[before+1:]
1098 d.buf.WriteByte(';')
1099 if isName(name) {
1100 s := string(name)
1101 if r, ok := entity[s]; ok {
1102 text = string(r)
1103 haveText = true
1104 } else if d.Entity != nil {
1105 text, haveText = d.Entity[s]
1106 }
1107 }
1108 }
1109 }
1110
1111 if haveText {
1112 d.buf.Truncate(before)
1113 d.buf.WriteString(text)
1114 b0, b1 = 0, 0
1115 continue Input
1116 }
1117 if !d.Strict {
1118 b0, b1 = 0, 0
1119 continue Input
1120 }
1121 ent := string(d.buf.Bytes()[before:])
1122 if ent[len(ent)-1] != ';' {
1123 ent += " (no semicolon)"
1124 }
1125 d.err = d.syntaxError("invalid character entity " + ent)
1126 return nil
1127 }
1128
1129
1130 if b == '\r' {
1131 d.buf.WriteByte('\n')
1132 } else if b1 == '\r' && b == '\n' {
1133
1134 } else {
1135 d.buf.WriteByte(b)
1136 }
1137
1138 b0, b1 = b1, b
1139 }
1140 data := d.buf.Bytes()
1141 data = data[0 : len(data)-trunc]
1142
1143
1144 buf := data
1145 for len(buf) > 0 {
1146 r, size := utf8.DecodeRune(buf)
1147 if r == utf8.RuneError && size == 1 {
1148 d.err = d.syntaxError("invalid UTF-8")
1149 return nil
1150 }
1151 buf = buf[size:]
1152 if !isInCharacterRange(r) {
1153 d.err = d.syntaxError(fmt.Sprintf("illegal character code %U", r))
1154 return nil
1155 }
1156 }
1157
1158 return data
1159 }
1160
1161
1162
1163
1164 func isInCharacterRange(r rune) (inrange bool) {
1165 return r == 0x09 ||
1166 r == 0x0A ||
1167 r == 0x0D ||
1168 r >= 0x20 && r <= 0xD7FF ||
1169 r >= 0xE000 && r <= 0xFFFD ||
1170 r >= 0x10000 && r <= 0x10FFFF
1171 }
1172
1173
1174
1175 func (d *Decoder) nsname() (name Name, ok bool) {
1176 s, ok := d.name()
1177 if !ok {
1178 return
1179 }
1180 if strings.Count(s, ":") > 1 {
1181 return name, false
1182 } else if space, local, ok := strings.Cut(s, ":"); !ok || space == "" || local == "" {
1183 name.Local = s
1184 } else {
1185 name.Space = space
1186 name.Local = local
1187 }
1188 return name, true
1189 }
1190
1191
1192
1193
1194 func (d *Decoder) name() (s string, ok bool) {
1195 d.buf.Reset()
1196 if !d.readName() {
1197 return "", false
1198 }
1199
1200
1201 b := d.buf.Bytes()
1202 if !isName(b) {
1203 d.err = d.syntaxError("invalid XML name: " + string(b))
1204 return "", false
1205 }
1206 return string(b), true
1207 }
1208
1209
1210
1211
1212 func (d *Decoder) readName() (ok bool) {
1213 var b byte
1214 if b, ok = d.mustgetc(); !ok {
1215 return
1216 }
1217 if b < utf8.RuneSelf && !isNameByte(b) {
1218 d.ungetc(b)
1219 return false
1220 }
1221 d.buf.WriteByte(b)
1222
1223 for {
1224 if b, ok = d.mustgetc(); !ok {
1225 return
1226 }
1227 if b < utf8.RuneSelf && !isNameByte(b) {
1228 d.ungetc(b)
1229 break
1230 }
1231 d.buf.WriteByte(b)
1232 }
1233 return true
1234 }
1235
1236 func isNameByte(c byte) bool {
1237 return 'A' <= c && c <= 'Z' ||
1238 'a' <= c && c <= 'z' ||
1239 '0' <= c && c <= '9' ||
1240 c == '_' || c == ':' || c == '.' || c == '-'
1241 }
1242
1243 func isName(s []byte) bool {
1244 if len(s) == 0 {
1245 return false
1246 }
1247 c, n := utf8.DecodeRune(s)
1248 if c == utf8.RuneError && n == 1 {
1249 return false
1250 }
1251 if !unicode.Is(first, c) {
1252 return false
1253 }
1254 for n < len(s) {
1255 s = s[n:]
1256 c, n = utf8.DecodeRune(s)
1257 if c == utf8.RuneError && n == 1 {
1258 return false
1259 }
1260 if !unicode.Is(first, c) && !unicode.Is(second, c) {
1261 return false
1262 }
1263 }
1264 return true
1265 }
1266
1267 func isNameString(s string) bool {
1268 if len(s) == 0 {
1269 return false
1270 }
1271 c, n := utf8.DecodeRuneInString(s)
1272 if c == utf8.RuneError && n == 1 {
1273 return false
1274 }
1275 if !unicode.Is(first, c) {
1276 return false
1277 }
1278 for n < len(s) {
1279 s = s[n:]
1280 c, n = utf8.DecodeRuneInString(s)
1281 if c == utf8.RuneError && n == 1 {
1282 return false
1283 }
1284 if !unicode.Is(first, c) && !unicode.Is(second, c) {
1285 return false
1286 }
1287 }
1288 return true
1289 }
1290
1291
1292
1293
1294
1295
1296 var first = &unicode.RangeTable{
1297 R16: []unicode.Range16{
1298 {0x003A, 0x003A, 1},
1299 {0x0041, 0x005A, 1},
1300 {0x005F, 0x005F, 1},
1301 {0x0061, 0x007A, 1},
1302 {0x00C0, 0x00D6, 1},
1303 {0x00D8, 0x00F6, 1},
1304 {0x00F8, 0x00FF, 1},
1305 {0x0100, 0x0131, 1},
1306 {0x0134, 0x013E, 1},
1307 {0x0141, 0x0148, 1},
1308 {0x014A, 0x017E, 1},
1309 {0x0180, 0x01C3, 1},
1310 {0x01CD, 0x01F0, 1},
1311 {0x01F4, 0x01F5, 1},
1312 {0x01FA, 0x0217, 1},
1313 {0x0250, 0x02A8, 1},
1314 {0x02BB, 0x02C1, 1},
1315 {0x0386, 0x0386, 1},
1316 {0x0388, 0x038A, 1},
1317 {0x038C, 0x038C, 1},
1318 {0x038E, 0x03A1, 1},
1319 {0x03A3, 0x03CE, 1},
1320 {0x03D0, 0x03D6, 1},
1321 {0x03DA, 0x03E0, 2},
1322 {0x03E2, 0x03F3, 1},
1323 {0x0401, 0x040C, 1},
1324 {0x040E, 0x044F, 1},
1325 {0x0451, 0x045C, 1},
1326 {0x045E, 0x0481, 1},
1327 {0x0490, 0x04C4, 1},
1328 {0x04C7, 0x04C8, 1},
1329 {0x04CB, 0x04CC, 1},
1330 {0x04D0, 0x04EB, 1},
1331 {0x04EE, 0x04F5, 1},
1332 {0x04F8, 0x04F9, 1},
1333 {0x0531, 0x0556, 1},
1334 {0x0559, 0x0559, 1},
1335 {0x0561, 0x0586, 1},
1336 {0x05D0, 0x05EA, 1},
1337 {0x05F0, 0x05F2, 1},
1338 {0x0621, 0x063A, 1},
1339 {0x0641, 0x064A, 1},
1340 {0x0671, 0x06B7, 1},
1341 {0x06BA, 0x06BE, 1},
1342 {0x06C0, 0x06CE, 1},
1343 {0x06D0, 0x06D3, 1},
1344 {0x06D5, 0x06D5, 1},
1345 {0x06E5, 0x06E6, 1},
1346 {0x0905, 0x0939, 1},
1347 {0x093D, 0x093D, 1},
1348 {0x0958, 0x0961, 1},
1349 {0x0985, 0x098C, 1},
1350 {0x098F, 0x0990, 1},
1351 {0x0993, 0x09A8, 1},
1352 {0x09AA, 0x09B0, 1},
1353 {0x09B2, 0x09B2, 1},
1354 {0x09B6, 0x09B9, 1},
1355 {0x09DC, 0x09DD, 1},
1356 {0x09DF, 0x09E1, 1},
1357 {0x09F0, 0x09F1, 1},
1358 {0x0A05, 0x0A0A, 1},
1359 {0x0A0F, 0x0A10, 1},
1360 {0x0A13, 0x0A28, 1},
1361 {0x0A2A, 0x0A30, 1},
1362 {0x0A32, 0x0A33, 1},
1363 {0x0A35, 0x0A36, 1},
1364 {0x0A38, 0x0A39, 1},
1365 {0x0A59, 0x0A5C, 1},
1366 {0x0A5E, 0x0A5E, 1},
1367 {0x0A72, 0x0A74, 1},
1368 {0x0A85, 0x0A8B, 1},
1369 {0x0A8D, 0x0A8D, 1},
1370 {0x0A8F, 0x0A91, 1},
1371 {0x0A93, 0x0AA8, 1},
1372 {0x0AAA, 0x0AB0, 1},
1373 {0x0AB2, 0x0AB3, 1},
1374 {0x0AB5, 0x0AB9, 1},
1375 {0x0ABD, 0x0AE0, 0x23},
1376 {0x0B05, 0x0B0C, 1},
1377 {0x0B0F, 0x0B10, 1},
1378 {0x0B13, 0x0B28, 1},
1379 {0x0B2A, 0x0B30, 1},
1380 {0x0B32, 0x0B33, 1},
1381 {0x0B36, 0x0B39, 1},
1382 {0x0B3D, 0x0B3D, 1},
1383 {0x0B5C, 0x0B5D, 1},
1384 {0x0B5F, 0x0B61, 1},
1385 {0x0B85, 0x0B8A, 1},
1386 {0x0B8E, 0x0B90, 1},
1387 {0x0B92, 0x0B95, 1},
1388 {0x0B99, 0x0B9A, 1},
1389 {0x0B9C, 0x0B9C, 1},
1390 {0x0B9E, 0x0B9F, 1},
1391 {0x0BA3, 0x0BA4, 1},
1392 {0x0BA8, 0x0BAA, 1},
1393 {0x0BAE, 0x0BB5, 1},
1394 {0x0BB7, 0x0BB9, 1},
1395 {0x0C05, 0x0C0C, 1},
1396 {0x0C0E, 0x0C10, 1},
1397 {0x0C12, 0x0C28, 1},
1398 {0x0C2A, 0x0C33, 1},
1399 {0x0C35, 0x0C39, 1},
1400 {0x0C60, 0x0C61, 1},
1401 {0x0C85, 0x0C8C, 1},
1402 {0x0C8E, 0x0C90, 1},
1403 {0x0C92, 0x0CA8, 1},
1404 {0x0CAA, 0x0CB3, 1},
1405 {0x0CB5, 0x0CB9, 1},
1406 {0x0CDE, 0x0CDE, 1},
1407 {0x0CE0, 0x0CE1, 1},
1408 {0x0D05, 0x0D0C, 1},
1409 {0x0D0E, 0x0D10, 1},
1410 {0x0D12, 0x0D28, 1},
1411 {0x0D2A, 0x0D39, 1},
1412 {0x0D60, 0x0D61, 1},
1413 {0x0E01, 0x0E2E, 1},
1414 {0x0E30, 0x0E30, 1},
1415 {0x0E32, 0x0E33, 1},
1416 {0x0E40, 0x0E45, 1},
1417 {0x0E81, 0x0E82, 1},
1418 {0x0E84, 0x0E84, 1},
1419 {0x0E87, 0x0E88, 1},
1420 {0x0E8A, 0x0E8D, 3},
1421 {0x0E94, 0x0E97, 1},
1422 {0x0E99, 0x0E9F, 1},
1423 {0x0EA1, 0x0EA3, 1},
1424 {0x0EA5, 0x0EA7, 2},
1425 {0x0EAA, 0x0EAB, 1},
1426 {0x0EAD, 0x0EAE, 1},
1427 {0x0EB0, 0x0EB0, 1},
1428 {0x0EB2, 0x0EB3, 1},
1429 {0x0EBD, 0x0EBD, 1},
1430 {0x0EC0, 0x0EC4, 1},
1431 {0x0F40, 0x0F47, 1},
1432 {0x0F49, 0x0F69, 1},
1433 {0x10A0, 0x10C5, 1},
1434 {0x10D0, 0x10F6, 1},
1435 {0x1100, 0x1100, 1},
1436 {0x1102, 0x1103, 1},
1437 {0x1105, 0x1107, 1},
1438 {0x1109, 0x1109, 1},
1439 {0x110B, 0x110C, 1},
1440 {0x110E, 0x1112, 1},
1441 {0x113C, 0x1140, 2},
1442 {0x114C, 0x1150, 2},
1443 {0x1154, 0x1155, 1},
1444 {0x1159, 0x1159, 1},
1445 {0x115F, 0x1161, 1},
1446 {0x1163, 0x1169, 2},
1447 {0x116D, 0x116E, 1},
1448 {0x1172, 0x1173, 1},
1449 {0x1175, 0x119E, 0x119E - 0x1175},
1450 {0x11A8, 0x11AB, 0x11AB - 0x11A8},
1451 {0x11AE, 0x11AF, 1},
1452 {0x11B7, 0x11B8, 1},
1453 {0x11BA, 0x11BA, 1},
1454 {0x11BC, 0x11C2, 1},
1455 {0x11EB, 0x11F0, 0x11F0 - 0x11EB},
1456 {0x11F9, 0x11F9, 1},
1457 {0x1E00, 0x1E9B, 1},
1458 {0x1EA0, 0x1EF9, 1},
1459 {0x1F00, 0x1F15, 1},
1460 {0x1F18, 0x1F1D, 1},
1461 {0x1F20, 0x1F45, 1},
1462 {0x1F48, 0x1F4D, 1},
1463 {0x1F50, 0x1F57, 1},
1464 {0x1F59, 0x1F5B, 0x1F5B - 0x1F59},
1465 {0x1F5D, 0x1F5D, 1},
1466 {0x1F5F, 0x1F7D, 1},
1467 {0x1F80, 0x1FB4, 1},
1468 {0x1FB6, 0x1FBC, 1},
1469 {0x1FBE, 0x1FBE, 1},
1470 {0x1FC2, 0x1FC4, 1},
1471 {0x1FC6, 0x1FCC, 1},
1472 {0x1FD0, 0x1FD3, 1},
1473 {0x1FD6, 0x1FDB, 1},
1474 {0x1FE0, 0x1FEC, 1},
1475 {0x1FF2, 0x1FF4, 1},
1476 {0x1FF6, 0x1FFC, 1},
1477 {0x2126, 0x2126, 1},
1478 {0x212A, 0x212B, 1},
1479 {0x212E, 0x212E, 1},
1480 {0x2180, 0x2182, 1},
1481 {0x3007, 0x3007, 1},
1482 {0x3021, 0x3029, 1},
1483 {0x3041, 0x3094, 1},
1484 {0x30A1, 0x30FA, 1},
1485 {0x3105, 0x312C, 1},
1486 {0x4E00, 0x9FA5, 1},
1487 {0xAC00, 0xD7A3, 1},
1488 },
1489 }
1490
1491 var second = &unicode.RangeTable{
1492 R16: []unicode.Range16{
1493 {0x002D, 0x002E, 1},
1494 {0x0030, 0x0039, 1},
1495 {0x00B7, 0x00B7, 1},
1496 {0x02D0, 0x02D1, 1},
1497 {0x0300, 0x0345, 1},
1498 {0x0360, 0x0361, 1},
1499 {0x0387, 0x0387, 1},
1500 {0x0483, 0x0486, 1},
1501 {0x0591, 0x05A1, 1},
1502 {0x05A3, 0x05B9, 1},
1503 {0x05BB, 0x05BD, 1},
1504 {0x05BF, 0x05BF, 1},
1505 {0x05C1, 0x05C2, 1},
1506 {0x05C4, 0x0640, 0x0640 - 0x05C4},
1507 {0x064B, 0x0652, 1},
1508 {0x0660, 0x0669, 1},
1509 {0x0670, 0x0670, 1},
1510 {0x06D6, 0x06DC, 1},
1511 {0x06DD, 0x06DF, 1},
1512 {0x06E0, 0x06E4, 1},
1513 {0x06E7, 0x06E8, 1},
1514 {0x06EA, 0x06ED, 1},
1515 {0x06F0, 0x06F9, 1},
1516 {0x0901, 0x0903, 1},
1517 {0x093C, 0x093C, 1},
1518 {0x093E, 0x094C, 1},
1519 {0x094D, 0x094D, 1},
1520 {0x0951, 0x0954, 1},
1521 {0x0962, 0x0963, 1},
1522 {0x0966, 0x096F, 1},
1523 {0x0981, 0x0983, 1},
1524 {0x09BC, 0x09BC, 1},
1525 {0x09BE, 0x09BF, 1},
1526 {0x09C0, 0x09C4, 1},
1527 {0x09C7, 0x09C8, 1},
1528 {0x09CB, 0x09CD, 1},
1529 {0x09D7, 0x09D7, 1},
1530 {0x09E2, 0x09E3, 1},
1531 {0x09E6, 0x09EF, 1},
1532 {0x0A02, 0x0A3C, 0x3A},
1533 {0x0A3E, 0x0A3F, 1},
1534 {0x0A40, 0x0A42, 1},
1535 {0x0A47, 0x0A48, 1},
1536 {0x0A4B, 0x0A4D, 1},
1537 {0x0A66, 0x0A6F, 1},
1538 {0x0A70, 0x0A71, 1},
1539 {0x0A81, 0x0A83, 1},
1540 {0x0ABC, 0x0ABC, 1},
1541 {0x0ABE, 0x0AC5, 1},
1542 {0x0AC7, 0x0AC9, 1},
1543 {0x0ACB, 0x0ACD, 1},
1544 {0x0AE6, 0x0AEF, 1},
1545 {0x0B01, 0x0B03, 1},
1546 {0x0B3C, 0x0B3C, 1},
1547 {0x0B3E, 0x0B43, 1},
1548 {0x0B47, 0x0B48, 1},
1549 {0x0B4B, 0x0B4D, 1},
1550 {0x0B56, 0x0B57, 1},
1551 {0x0B66, 0x0B6F, 1},
1552 {0x0B82, 0x0B83, 1},
1553 {0x0BBE, 0x0BC2, 1},
1554 {0x0BC6, 0x0BC8, 1},
1555 {0x0BCA, 0x0BCD, 1},
1556 {0x0BD7, 0x0BD7, 1},
1557 {0x0BE7, 0x0BEF, 1},
1558 {0x0C01, 0x0C03, 1},
1559 {0x0C3E, 0x0C44, 1},
1560 {0x0C46, 0x0C48, 1},
1561 {0x0C4A, 0x0C4D, 1},
1562 {0x0C55, 0x0C56, 1},
1563 {0x0C66, 0x0C6F, 1},
1564 {0x0C82, 0x0C83, 1},
1565 {0x0CBE, 0x0CC4, 1},
1566 {0x0CC6, 0x0CC8, 1},
1567 {0x0CCA, 0x0CCD, 1},
1568 {0x0CD5, 0x0CD6, 1},
1569 {0x0CE6, 0x0CEF, 1},
1570 {0x0D02, 0x0D03, 1},
1571 {0x0D3E, 0x0D43, 1},
1572 {0x0D46, 0x0D48, 1},
1573 {0x0D4A, 0x0D4D, 1},
1574 {0x0D57, 0x0D57, 1},
1575 {0x0D66, 0x0D6F, 1},
1576 {0x0E31, 0x0E31, 1},
1577 {0x0E34, 0x0E3A, 1},
1578 {0x0E46, 0x0E46, 1},
1579 {0x0E47, 0x0E4E, 1},
1580 {0x0E50, 0x0E59, 1},
1581 {0x0EB1, 0x0EB1, 1},
1582 {0x0EB4, 0x0EB9, 1},
1583 {0x0EBB, 0x0EBC, 1},
1584 {0x0EC6, 0x0EC6, 1},
1585 {0x0EC8, 0x0ECD, 1},
1586 {0x0ED0, 0x0ED9, 1},
1587 {0x0F18, 0x0F19, 1},
1588 {0x0F20, 0x0F29, 1},
1589 {0x0F35, 0x0F39, 2},
1590 {0x0F3E, 0x0F3F, 1},
1591 {0x0F71, 0x0F84, 1},
1592 {0x0F86, 0x0F8B, 1},
1593 {0x0F90, 0x0F95, 1},
1594 {0x0F97, 0x0F97, 1},
1595 {0x0F99, 0x0FAD, 1},
1596 {0x0FB1, 0x0FB7, 1},
1597 {0x0FB9, 0x0FB9, 1},
1598 {0x20D0, 0x20DC, 1},
1599 {0x20E1, 0x3005, 0x3005 - 0x20E1},
1600 {0x302A, 0x302F, 1},
1601 {0x3031, 0x3035, 1},
1602 {0x3099, 0x309A, 1},
1603 {0x309D, 0x309E, 1},
1604 {0x30FC, 0x30FE, 1},
1605 },
1606 }
1607
1608
1609
1610
1611
1612 var HTMLEntity map[string]string = htmlEntity
1613
1614 var htmlEntity = map[string]string{
1615
1623 "nbsp": "\u00A0",
1624 "iexcl": "\u00A1",
1625 "cent": "\u00A2",
1626 "pound": "\u00A3",
1627 "curren": "\u00A4",
1628 "yen": "\u00A5",
1629 "brvbar": "\u00A6",
1630 "sect": "\u00A7",
1631 "uml": "\u00A8",
1632 "copy": "\u00A9",
1633 "ordf": "\u00AA",
1634 "laquo": "\u00AB",
1635 "not": "\u00AC",
1636 "shy": "\u00AD",
1637 "reg": "\u00AE",
1638 "macr": "\u00AF",
1639 "deg": "\u00B0",
1640 "plusmn": "\u00B1",
1641 "sup2": "\u00B2",
1642 "sup3": "\u00B3",
1643 "acute": "\u00B4",
1644 "micro": "\u00B5",
1645 "para": "\u00B6",
1646 "middot": "\u00B7",
1647 "cedil": "\u00B8",
1648 "sup1": "\u00B9",
1649 "ordm": "\u00BA",
1650 "raquo": "\u00BB",
1651 "frac14": "\u00BC",
1652 "frac12": "\u00BD",
1653 "frac34": "\u00BE",
1654 "iquest": "\u00BF",
1655 "Agrave": "\u00C0",
1656 "Aacute": "\u00C1",
1657 "Acirc": "\u00C2",
1658 "Atilde": "\u00C3",
1659 "Auml": "\u00C4",
1660 "Aring": "\u00C5",
1661 "AElig": "\u00C6",
1662 "Ccedil": "\u00C7",
1663 "Egrave": "\u00C8",
1664 "Eacute": "\u00C9",
1665 "Ecirc": "\u00CA",
1666 "Euml": "\u00CB",
1667 "Igrave": "\u00CC",
1668 "Iacute": "\u00CD",
1669 "Icirc": "\u00CE",
1670 "Iuml": "\u00CF",
1671 "ETH": "\u00D0",
1672 "Ntilde": "\u00D1",
1673 "Ograve": "\u00D2",
1674 "Oacute": "\u00D3",
1675 "Ocirc": "\u00D4",
1676 "Otilde": "\u00D5",
1677 "Ouml": "\u00D6",
1678 "times": "\u00D7",
1679 "Oslash": "\u00D8",
1680 "Ugrave": "\u00D9",
1681 "Uacute": "\u00DA",
1682 "Ucirc": "\u00DB",
1683 "Uuml": "\u00DC",
1684 "Yacute": "\u00DD",
1685 "THORN": "\u00DE",
1686 "szlig": "\u00DF",
1687 "agrave": "\u00E0",
1688 "aacute": "\u00E1",
1689 "acirc": "\u00E2",
1690 "atilde": "\u00E3",
1691 "auml": "\u00E4",
1692 "aring": "\u00E5",
1693 "aelig": "\u00E6",
1694 "ccedil": "\u00E7",
1695 "egrave": "\u00E8",
1696 "eacute": "\u00E9",
1697 "ecirc": "\u00EA",
1698 "euml": "\u00EB",
1699 "igrave": "\u00EC",
1700 "iacute": "\u00ED",
1701 "icirc": "\u00EE",
1702 "iuml": "\u00EF",
1703 "eth": "\u00F0",
1704 "ntilde": "\u00F1",
1705 "ograve": "\u00F2",
1706 "oacute": "\u00F3",
1707 "ocirc": "\u00F4",
1708 "otilde": "\u00F5",
1709 "ouml": "\u00F6",
1710 "divide": "\u00F7",
1711 "oslash": "\u00F8",
1712 "ugrave": "\u00F9",
1713 "uacute": "\u00FA",
1714 "ucirc": "\u00FB",
1715 "uuml": "\u00FC",
1716 "yacute": "\u00FD",
1717 "thorn": "\u00FE",
1718 "yuml": "\u00FF",
1719 "fnof": "\u0192",
1720 "Alpha": "\u0391",
1721 "Beta": "\u0392",
1722 "Gamma": "\u0393",
1723 "Delta": "\u0394",
1724 "Epsilon": "\u0395",
1725 "Zeta": "\u0396",
1726 "Eta": "\u0397",
1727 "Theta": "\u0398",
1728 "Iota": "\u0399",
1729 "Kappa": "\u039A",
1730 "Lambda": "\u039B",
1731 "Mu": "\u039C",
1732 "Nu": "\u039D",
1733 "Xi": "\u039E",
1734 "Omicron": "\u039F",
1735 "Pi": "\u03A0",
1736 "Rho": "\u03A1",
1737 "Sigma": "\u03A3",
1738 "Tau": "\u03A4",
1739 "Upsilon": "\u03A5",
1740 "Phi": "\u03A6",
1741 "Chi": "\u03A7",
1742 "Psi": "\u03A8",
1743 "Omega": "\u03A9",
1744 "alpha": "\u03B1",
1745 "beta": "\u03B2",
1746 "gamma": "\u03B3",
1747 "delta": "\u03B4",
1748 "epsilon": "\u03B5",
1749 "zeta": "\u03B6",
1750 "eta": "\u03B7",
1751 "theta": "\u03B8",
1752 "iota": "\u03B9",
1753 "kappa": "\u03BA",
1754 "lambda": "\u03BB",
1755 "mu": "\u03BC",
1756 "nu": "\u03BD",
1757 "xi": "\u03BE",
1758 "omicron": "\u03BF",
1759 "pi": "\u03C0",
1760 "rho": "\u03C1",
1761 "sigmaf": "\u03C2",
1762 "sigma": "\u03C3",
1763 "tau": "\u03C4",
1764 "upsilon": "\u03C5",
1765 "phi": "\u03C6",
1766 "chi": "\u03C7",
1767 "psi": "\u03C8",
1768 "omega": "\u03C9",
1769 "thetasym": "\u03D1",
1770 "upsih": "\u03D2",
1771 "piv": "\u03D6",
1772 "bull": "\u2022",
1773 "hellip": "\u2026",
1774 "prime": "\u2032",
1775 "Prime": "\u2033",
1776 "oline": "\u203E",
1777 "frasl": "\u2044",
1778 "weierp": "\u2118",
1779 "image": "\u2111",
1780 "real": "\u211C",
1781 "trade": "\u2122",
1782 "alefsym": "\u2135",
1783 "larr": "\u2190",
1784 "uarr": "\u2191",
1785 "rarr": "\u2192",
1786 "darr": "\u2193",
1787 "harr": "\u2194",
1788 "crarr": "\u21B5",
1789 "lArr": "\u21D0",
1790 "uArr": "\u21D1",
1791 "rArr": "\u21D2",
1792 "dArr": "\u21D3",
1793 "hArr": "\u21D4",
1794 "forall": "\u2200",
1795 "part": "\u2202",
1796 "exist": "\u2203",
1797 "empty": "\u2205",
1798 "nabla": "\u2207",
1799 "isin": "\u2208",
1800 "notin": "\u2209",
1801 "ni": "\u220B",
1802 "prod": "\u220F",
1803 "sum": "\u2211",
1804 "minus": "\u2212",
1805 "lowast": "\u2217",
1806 "radic": "\u221A",
1807 "prop": "\u221D",
1808 "infin": "\u221E",
1809 "ang": "\u2220",
1810 "and": "\u2227",
1811 "or": "\u2228",
1812 "cap": "\u2229",
1813 "cup": "\u222A",
1814 "int": "\u222B",
1815 "there4": "\u2234",
1816 "sim": "\u223C",
1817 "cong": "\u2245",
1818 "asymp": "\u2248",
1819 "ne": "\u2260",
1820 "equiv": "\u2261",
1821 "le": "\u2264",
1822 "ge": "\u2265",
1823 "sub": "\u2282",
1824 "sup": "\u2283",
1825 "nsub": "\u2284",
1826 "sube": "\u2286",
1827 "supe": "\u2287",
1828 "oplus": "\u2295",
1829 "otimes": "\u2297",
1830 "perp": "\u22A5",
1831 "sdot": "\u22C5",
1832 "lceil": "\u2308",
1833 "rceil": "\u2309",
1834 "lfloor": "\u230A",
1835 "rfloor": "\u230B",
1836 "lang": "\u2329",
1837 "rang": "\u232A",
1838 "loz": "\u25CA",
1839 "spades": "\u2660",
1840 "clubs": "\u2663",
1841 "hearts": "\u2665",
1842 "diams": "\u2666",
1843 "quot": "\u0022",
1844 "amp": "\u0026",
1845 "lt": "\u003C",
1846 "gt": "\u003E",
1847 "OElig": "\u0152",
1848 "oelig": "\u0153",
1849 "Scaron": "\u0160",
1850 "scaron": "\u0161",
1851 "Yuml": "\u0178",
1852 "circ": "\u02C6",
1853 "tilde": "\u02DC",
1854 "ensp": "\u2002",
1855 "emsp": "\u2003",
1856 "thinsp": "\u2009",
1857 "zwnj": "\u200C",
1858 "zwj": "\u200D",
1859 "lrm": "\u200E",
1860 "rlm": "\u200F",
1861 "ndash": "\u2013",
1862 "mdash": "\u2014",
1863 "lsquo": "\u2018",
1864 "rsquo": "\u2019",
1865 "sbquo": "\u201A",
1866 "ldquo": "\u201C",
1867 "rdquo": "\u201D",
1868 "bdquo": "\u201E",
1869 "dagger": "\u2020",
1870 "Dagger": "\u2021",
1871 "permil": "\u2030",
1872 "lsaquo": "\u2039",
1873 "rsaquo": "\u203A",
1874 "euro": "\u20AC",
1875 }
1876
1877
1878
1879
1880
1881 var HTMLAutoClose []string = htmlAutoClose
1882
1883 var htmlAutoClose = []string{
1884
1888 "basefont",
1889 "br",
1890 "area",
1891 "link",
1892 "img",
1893 "param",
1894 "hr",
1895 "input",
1896 "col",
1897 "frame",
1898 "isindex",
1899 "base",
1900 "meta",
1901 }
1902
1903 var (
1904 escQuot = []byte(""")
1905 escApos = []byte("'")
1906 escAmp = []byte("&")
1907 escLT = []byte("<")
1908 escGT = []byte(">")
1909 escTab = []byte("	")
1910 escNL = []byte("
")
1911 escCR = []byte("
")
1912 escFFFD = []byte("\uFFFD")
1913 )
1914
1915
1916
1917 func EscapeText(w io.Writer, s []byte) error {
1918 return escapeText(w, s, true)
1919 }
1920
1921
1922
1923
1924 func escapeText(w io.Writer, s []byte, escapeNewline bool) error {
1925 var esc []byte
1926 last := 0
1927 for i := 0; i < len(s); {
1928 r, width := utf8.DecodeRune(s[i:])
1929 i += width
1930 switch r {
1931 case '"':
1932 esc = escQuot
1933 case '\'':
1934 esc = escApos
1935 case '&':
1936 esc = escAmp
1937 case '<':
1938 esc = escLT
1939 case '>':
1940 esc = escGT
1941 case '\t':
1942 esc = escTab
1943 case '\n':
1944 if !escapeNewline {
1945 continue
1946 }
1947 esc = escNL
1948 case '\r':
1949 esc = escCR
1950 default:
1951 if !isInCharacterRange(r) || (r == 0xFFFD && width == 1) {
1952 esc = escFFFD
1953 break
1954 }
1955 continue
1956 }
1957 if _, err := w.Write(s[last : i-width]); err != nil {
1958 return err
1959 }
1960 if _, err := w.Write(esc); err != nil {
1961 return err
1962 }
1963 last = i
1964 }
1965 _, err := w.Write(s[last:])
1966 return err
1967 }
1968
1969
1970
1971 func (p *printer) EscapeString(s string) {
1972 var esc []byte
1973 last := 0
1974 for i := 0; i < len(s); {
1975 r, width := utf8.DecodeRuneInString(s[i:])
1976 i += width
1977 switch r {
1978 case '"':
1979 esc = escQuot
1980 case '\'':
1981 esc = escApos
1982 case '&':
1983 esc = escAmp
1984 case '<':
1985 esc = escLT
1986 case '>':
1987 esc = escGT
1988 case '\t':
1989 esc = escTab
1990 case '\n':
1991 esc = escNL
1992 case '\r':
1993 esc = escCR
1994 default:
1995 if !isInCharacterRange(r) || (r == 0xFFFD && width == 1) {
1996 esc = escFFFD
1997 break
1998 }
1999 continue
2000 }
2001 p.WriteString(s[last : i-width])
2002 p.Write(esc)
2003 last = i
2004 }
2005 p.WriteString(s[last:])
2006 }
2007
2008
2009
2010
2011 func Escape(w io.Writer, s []byte) {
2012 EscapeText(w, s)
2013 }
2014
2015 var (
2016 cdataStart = []byte("<![CDATA[")
2017 cdataEnd = []byte("]]>")
2018 cdataEscape = []byte("]]]]><![CDATA[>")
2019 )
2020
2021
2022
2023 func emitCDATA(w io.Writer, s []byte) error {
2024 if len(s) == 0 {
2025 return nil
2026 }
2027 if _, err := w.Write(cdataStart); err != nil {
2028 return err
2029 }
2030
2031 for {
2032 before, after, ok := bytes.Cut(s, cdataEnd)
2033 if !ok {
2034 break
2035 }
2036
2037 if _, err := w.Write(before); err != nil {
2038 return err
2039 }
2040 if _, err := w.Write(cdataEscape); err != nil {
2041 return err
2042 }
2043 s = after
2044 }
2045
2046 if _, err := w.Write(s); err != nil {
2047 return err
2048 }
2049
2050 _, err := w.Write(cdataEnd)
2051 return err
2052 }
2053
2054
2055
2056 func procInst(param, s string) string {
2057
2058
2059 param = param + "="
2060 lenp := len(param)
2061 i := 0
2062 var sep byte
2063 for i < len(s) {
2064 sub := s[i:]
2065 k := strings.Index(sub, param)
2066 if k < 0 || lenp+k >= len(sub) {
2067 return ""
2068 }
2069 i += lenp + k + 1
2070 if c := sub[lenp+k]; c == '\'' || c == '"' {
2071 sep = c
2072 break
2073 }
2074 }
2075 if sep == 0 {
2076 return ""
2077 }
2078 j := strings.IndexByte(s[i:], sep)
2079 if j < 0 {
2080 return ""
2081 }
2082 return s[i : i+j]
2083 }
2084
View as plain text