Run Format

Source file src/pkg/encoding/xml/xml.go

     1	// Copyright 2009 The Go Authors. All rights reserved.
     2	// Use of this source code is governed by a BSD-style
     3	// license that can be found in the LICENSE file.
     4	
     5	// Package xml implements a simple XML 1.0 parser that
     6	// understands XML name spaces.
     7	package xml
     8	
     9	// References:
    10	//    Annotated XML spec: http://www.xml.com/axml/testaxml.htm
    11	//    XML name spaces: http://www.w3.org/TR/REC-xml-names/
    12	
    13	// TODO(rsc):
    14	//	Test error handling.
    15	
    16	import (
    17		"bufio"
    18		"bytes"
    19		"errors"
    20		"fmt"
    21		"io"
    22		"strconv"
    23		"strings"
    24		"unicode"
    25		"unicode/utf8"
    26	)
    27	
    28	// A SyntaxError represents a syntax error in the XML input stream.
    29	type SyntaxError struct {
    30		Msg  string
    31		Line int
    32	}
    33	
    34	func (e *SyntaxError) Error() string {
    35		return "XML syntax error on line " + strconv.Itoa(e.Line) + ": " + e.Msg
    36	}
    37	
    38	// A Name represents an XML name (Local) annotated
    39	// with a name space identifier (Space).
    40	// In tokens returned by Decoder.Token, the Space identifier
    41	// is given as a canonical URL, not the short prefix used
    42	// in the document being parsed.
    43	type Name struct {
    44		Space, Local string
    45	}
    46	
    47	// An Attr represents an attribute in an XML element (Name=Value).
    48	type Attr struct {
    49		Name  Name
    50		Value string
    51	}
    52	
    53	// A Token is an interface holding one of the token types:
    54	// StartElement, EndElement, CharData, Comment, ProcInst, or Directive.
    55	type Token interface{}
    56	
    57	// A StartElement represents an XML start element.
    58	type StartElement struct {
    59		Name Name
    60		Attr []Attr
    61	}
    62	
    63	func (e StartElement) Copy() StartElement {
    64		attrs := make([]Attr, len(e.Attr))
    65		copy(attrs, e.Attr)
    66		e.Attr = attrs
    67		return e
    68	}
    69	
    70	// End returns the corresponding XML end element.
    71	func (e StartElement) End() EndElement {
    72		return EndElement{e.Name}
    73	}
    74	
    75	// An EndElement represents an XML end element.
    76	type EndElement struct {
    77		Name Name
    78	}
    79	
    80	// A CharData represents XML character data (raw text),
    81	// in which XML escape sequences have been replaced by
    82	// the characters they represent.
    83	type CharData []byte
    84	
    85	func makeCopy(b []byte) []byte {
    86		b1 := make([]byte, len(b))
    87		copy(b1, b)
    88		return b1
    89	}
    90	
    91	func (c CharData) Copy() CharData { return CharData(makeCopy(c)) }
    92	
    93	// A Comment represents an XML comment of the form <!--comment-->.
    94	// The bytes do not include the <!-- and --> comment markers.
    95	type Comment []byte
    96	
    97	func (c Comment) Copy() Comment { return Comment(makeCopy(c)) }
    98	
    99	// A ProcInst represents an XML processing instruction of the form <?target inst?>
   100	type ProcInst struct {
   101		Target string
   102		Inst   []byte
   103	}
   104	
   105	func (p ProcInst) Copy() ProcInst {
   106		p.Inst = makeCopy(p.Inst)
   107		return p
   108	}
   109	
   110	// A Directive represents an XML directive of the form <!text>.
   111	// The bytes do not include the <! and > markers.
   112	type Directive []byte
   113	
   114	func (d Directive) Copy() Directive { return Directive(makeCopy(d)) }
   115	
   116	// CopyToken returns a copy of a Token.
   117	func CopyToken(t Token) Token {
   118		switch v := t.(type) {
   119		case CharData:
   120			return v.Copy()
   121		case Comment:
   122			return v.Copy()
   123		case Directive:
   124			return v.Copy()
   125		case ProcInst:
   126			return v.Copy()
   127		case StartElement:
   128			return v.Copy()
   129		}
   130		return t
   131	}
   132	
   133	// A Decoder represents an XML parser reading a particular input stream.
   134	// The parser assumes that its input is encoded in UTF-8.
   135	type Decoder struct {
   136		// Strict defaults to true, enforcing the requirements
   137		// of the XML specification.
   138		// If set to false, the parser allows input containing common
   139		// mistakes:
   140		//	* If an element is missing an end tag, the parser invents
   141		//	  end tags as necessary to keep the return values from Token
   142		//	  properly balanced.
   143		//	* In attribute values and character data, unknown or malformed
   144		//	  character entities (sequences beginning with &) are left alone.
   145		//
   146		// Setting:
   147		//
   148		//	d.Strict = false;
   149		//	d.AutoClose = HTMLAutoClose;
   150		//	d.Entity = HTMLEntity
   151		//
   152		// creates a parser that can handle typical HTML.
   153		//
   154		// Strict mode does not enforce the requirements of the XML name spaces TR.
   155		// In particular it does not reject name space tags using undefined prefixes.
   156		// Such tags are recorded with the unknown prefix as the name space URL.
   157		Strict bool
   158	
   159		// When Strict == false, AutoClose indicates a set of elements to
   160		// consider closed immediately after they are opened, regardless
   161		// of whether an end element is present.
   162		AutoClose []string
   163	
   164		// Entity can be used to map non-standard entity names to string replacements.
   165		// The parser behaves as if these standard mappings are present in the map,
   166		// regardless of the actual map content:
   167		//
   168		//	"lt": "<",
   169		//	"gt": ">",
   170		//	"amp": "&",
   171		//	"apos": "'",
   172		//	"quot": `"`,
   173		Entity map[string]string
   174	
   175		// CharsetReader, if non-nil, defines a function to generate
   176		// charset-conversion readers, converting from the provided
   177		// non-UTF-8 charset into UTF-8. If CharsetReader is nil or
   178		// returns an error, parsing stops with an error. One of the
   179		// the CharsetReader's result values must be non-nil.
   180		CharsetReader func(charset string, input io.Reader) (io.Reader, error)
   181	
   182		// DefaultSpace sets the default name space used for unadorned tags,
   183		// as if the entire XML stream were wrapped in an element containing
   184		// the attribute xmlns="DefaultSpace".
   185		DefaultSpace string
   186	
   187		r              io.ByteReader
   188		buf            bytes.Buffer
   189		saved          *bytes.Buffer
   190		stk            *stack
   191		free           *stack
   192		needClose      bool
   193		toClose        Name
   194		nextToken      Token
   195		nextByte       int
   196		ns             map[string]string
   197		err            error
   198		line           int
   199		unmarshalDepth int
   200	}
   201	
   202	// NewDecoder creates a new XML parser reading from r.
   203	func NewDecoder(r io.Reader) *Decoder {
   204		d := &Decoder{
   205			ns:       make(map[string]string),
   206			nextByte: -1,
   207			line:     1,
   208			Strict:   true,
   209		}
   210		d.switchToReader(r)
   211		return d
   212	}
   213	
   214	// Token returns the next XML token in the input stream.
   215	// At the end of the input stream, Token returns nil, io.EOF.
   216	//
   217	// Slices of bytes in the returned token data refer to the
   218	// parser's internal buffer and remain valid only until the next
   219	// call to Token.  To acquire a copy of the bytes, call CopyToken
   220	// or the token's Copy method.
   221	//
   222	// Token expands self-closing elements such as <br/>
   223	// into separate start and end elements returned by successive calls.
   224	//
   225	// Token guarantees that the StartElement and EndElement
   226	// tokens it returns are properly nested and matched:
   227	// if Token encounters an unexpected end element,
   228	// it will return an error.
   229	//
   230	// Token implements XML name spaces as described by
   231	// http://www.w3.org/TR/REC-xml-names/.  Each of the
   232	// Name structures contained in the Token has the Space
   233	// set to the URL identifying its name space when known.
   234	// If Token encounters an unrecognized name space prefix,
   235	// it uses the prefix as the Space rather than report an error.
   236	func (d *Decoder) Token() (t Token, err error) {
   237		if d.stk != nil && d.stk.kind == stkEOF {
   238			err = io.EOF
   239			return
   240		}
   241		if d.nextToken != nil {
   242			t = d.nextToken
   243			d.nextToken = nil
   244		} else if t, err = d.rawToken(); err != nil {
   245			return
   246		}
   247	
   248		if !d.Strict {
   249			if t1, ok := d.autoClose(t); ok {
   250				d.nextToken = t
   251				t = t1
   252			}
   253		}
   254		switch t1 := t.(type) {
   255		case StartElement:
   256			// In XML name spaces, the translations listed in the
   257			// attributes apply to the element name and
   258			// to the other attribute names, so process
   259			// the translations first.
   260			for _, a := range t1.Attr {
   261				if a.Name.Space == "xmlns" {
   262					v, ok := d.ns[a.Name.Local]
   263					d.pushNs(a.Name.Local, v, ok)
   264					d.ns[a.Name.Local] = a.Value
   265				}
   266				if a.Name.Space == "" && a.Name.Local == "xmlns" {
   267					// Default space for untagged names
   268					v, ok := d.ns[""]
   269					d.pushNs("", v, ok)
   270					d.ns[""] = a.Value
   271				}
   272			}
   273	
   274			d.translate(&t1.Name, true)
   275			for i := range t1.Attr {
   276				d.translate(&t1.Attr[i].Name, false)
   277			}
   278			d.pushElement(t1.Name)
   279			t = t1
   280	
   281		case EndElement:
   282			d.translate(&t1.Name, true)
   283			if !d.popElement(&t1) {
   284				return nil, d.err
   285			}
   286			t = t1
   287		}
   288		return
   289	}
   290	
   291	const xmlURL = "http://www.w3.org/XML/1998/namespace"
   292	
   293	// Apply name space translation to name n.
   294	// The default name space (for Space=="")
   295	// applies only to element names, not to attribute names.
   296	func (d *Decoder) translate(n *Name, isElementName bool) {
   297		switch {
   298		case n.Space == "xmlns":
   299			return
   300		case n.Space == "" && !isElementName:
   301			return
   302		case n.Space == "xml":
   303			n.Space = xmlURL
   304		case n.Space == "" && n.Local == "xmlns":
   305			return
   306		}
   307		if v, ok := d.ns[n.Space]; ok {
   308			n.Space = v
   309		} else if n.Space == "" {
   310			n.Space = d.DefaultSpace
   311		}
   312	}
   313	
   314	func (d *Decoder) switchToReader(r io.Reader) {
   315		// Get efficient byte at a time reader.
   316		// Assume that if reader has its own
   317		// ReadByte, it's efficient enough.
   318		// Otherwise, use bufio.
   319		if rb, ok := r.(io.ByteReader); ok {
   320			d.r = rb
   321		} else {
   322			d.r = bufio.NewReader(r)
   323		}
   324	}
   325	
   326	// Parsing state - stack holds old name space translations
   327	// and the current set of open elements.  The translations to pop when
   328	// ending a given tag are *below* it on the stack, which is
   329	// more work but forced on us by XML.
   330	type stack struct {
   331		next *stack
   332		kind int
   333		name Name
   334		ok   bool
   335	}
   336	
   337	const (
   338		stkStart = iota
   339		stkNs
   340		stkEOF
   341	)
   342	
   343	func (d *Decoder) push(kind int) *stack {
   344		s := d.free
   345		if s != nil {
   346			d.free = s.next
   347		} else {
   348			s = new(stack)
   349		}
   350		s.next = d.stk
   351		s.kind = kind
   352		d.stk = s
   353		return s
   354	}
   355	
   356	func (d *Decoder) pop() *stack {
   357		s := d.stk
   358		if s != nil {
   359			d.stk = s.next
   360			s.next = d.free
   361			d.free = s
   362		}
   363		return s
   364	}
   365	
   366	// Record that after the current element is finished
   367	// (that element is already pushed on the stack)
   368	// Token should return EOF until popEOF is called.
   369	func (d *Decoder) pushEOF() {
   370		// Walk down stack to find Start.
   371		// It might not be the top, because there might be stkNs
   372		// entries above it.
   373		start := d.stk
   374		for start.kind != stkStart {
   375			start = start.next
   376		}
   377		// The stkNs entries below a start are associated with that
   378		// element too; skip over them.
   379		for start.next != nil && start.next.kind == stkNs {
   380			start = start.next
   381		}
   382		s := d.free
   383		if s != nil {
   384			d.free = s.next
   385		} else {
   386			s = new(stack)
   387		}
   388		s.kind = stkEOF
   389		s.next = start.next
   390		start.next = s
   391	}
   392	
   393	// Undo a pushEOF.
   394	// The element must have been finished, so the EOF should be at the top of the stack.
   395	func (d *Decoder) popEOF() bool {
   396		if d.stk == nil || d.stk.kind != stkEOF {
   397			return false
   398		}
   399		d.pop()
   400		return true
   401	}
   402	
   403	// Record that we are starting an element with the given name.
   404	func (d *Decoder) pushElement(name Name) {
   405		s := d.push(stkStart)
   406		s.name = name
   407	}
   408	
   409	// Record that we are changing the value of ns[local].
   410	// The old value is url, ok.
   411	func (d *Decoder) pushNs(local string, url string, ok bool) {
   412		s := d.push(stkNs)
   413		s.name.Local = local
   414		s.name.Space = url
   415		s.ok = ok
   416	}
   417	
   418	// Creates a SyntaxError with the current line number.
   419	func (d *Decoder) syntaxError(msg string) error {
   420		return &SyntaxError{Msg: msg, Line: d.line}
   421	}
   422	
   423	// Record that we are ending an element with the given name.
   424	// The name must match the record at the top of the stack,
   425	// which must be a pushElement record.
   426	// After popping the element, apply any undo records from
   427	// the stack to restore the name translations that existed
   428	// before we saw this element.
   429	func (d *Decoder) popElement(t *EndElement) bool {
   430		s := d.pop()
   431		name := t.Name
   432		switch {
   433		case s == nil || s.kind != stkStart:
   434			d.err = d.syntaxError("unexpected end element </" + name.Local + ">")
   435			return false
   436		case s.name.Local != name.Local:
   437			if !d.Strict {
   438				d.needClose = true
   439				d.toClose = t.Name
   440				t.Name = s.name
   441				return true
   442			}
   443			d.err = d.syntaxError("element <" + s.name.Local + "> closed by </" + name.Local + ">")
   444			return false
   445		case s.name.Space != name.Space:
   446			d.err = d.syntaxError("element <" + s.name.Local + "> in space " + s.name.Space +
   447				"closed by </" + name.Local + "> in space " + name.Space)
   448			return false
   449		}
   450	
   451		// Pop stack until a Start or EOF is on the top, undoing the
   452		// translations that were associated with the element we just closed.
   453		for d.stk != nil && d.stk.kind != stkStart && d.stk.kind != stkEOF {
   454			s := d.pop()
   455			if s.ok {
   456				d.ns[s.name.Local] = s.name.Space
   457			} else {
   458				delete(d.ns, s.name.Local)
   459			}
   460		}
   461	
   462		return true
   463	}
   464	
   465	// If the top element on the stack is autoclosing and
   466	// t is not the end tag, invent the end tag.
   467	func (d *Decoder) autoClose(t Token) (Token, bool) {
   468		if d.stk == nil || d.stk.kind != stkStart {
   469			return nil, false
   470		}
   471		name := strings.ToLower(d.stk.name.Local)
   472		for _, s := range d.AutoClose {
   473			if strings.ToLower(s) == name {
   474				// This one should be auto closed if t doesn't close it.
   475				et, ok := t.(EndElement)
   476				if !ok || et.Name.Local != name {
   477					return EndElement{d.stk.name}, true
   478				}
   479				break
   480			}
   481		}
   482		return nil, false
   483	}
   484	
   485	var errRawToken = errors.New("xml: cannot use RawToken from UnmarshalXML method")
   486	
   487	// RawToken is like Token but does not verify that
   488	// start and end elements match and does not translate
   489	// name space prefixes to their corresponding URLs.
   490	func (d *Decoder) RawToken() (Token, error) {
   491		if d.unmarshalDepth > 0 {
   492			return nil, errRawToken
   493		}
   494		return d.rawToken()
   495	}
   496	
   497	func (d *Decoder) rawToken() (Token, error) {
   498		if d.err != nil {
   499			return nil, d.err
   500		}
   501		if d.needClose {
   502			// The last element we read was self-closing and
   503			// we returned just the StartElement half.
   504			// Return the EndElement half now.
   505			d.needClose = false
   506			return EndElement{d.toClose}, nil
   507		}
   508	
   509		b, ok := d.getc()
   510		if !ok {
   511			return nil, d.err
   512		}
   513	
   514		if b != '<' {
   515			// Text section.
   516			d.ungetc(b)
   517			data := d.text(-1, false)
   518			if data == nil {
   519				return nil, d.err
   520			}
   521			return CharData(data), nil
   522		}
   523	
   524		if b, ok = d.mustgetc(); !ok {
   525			return nil, d.err
   526		}
   527		switch b {
   528		case '/':
   529			// </: End element
   530			var name Name
   531			if name, ok = d.nsname(); !ok {
   532				if d.err == nil {
   533					d.err = d.syntaxError("expected element name after </")
   534				}
   535				return nil, d.err
   536			}
   537			d.space()
   538			if b, ok = d.mustgetc(); !ok {
   539				return nil, d.err
   540			}
   541			if b != '>' {
   542				d.err = d.syntaxError("invalid characters between </" + name.Local + " and >")
   543				return nil, d.err
   544			}
   545			return EndElement{name}, nil
   546	
   547		case '?':
   548			// <?: Processing instruction.
   549			// TODO(rsc): Should parse the <?xml declaration to make sure the version is 1.0.
   550			var target string
   551			if target, ok = d.name(); !ok {
   552				if d.err == nil {
   553					d.err = d.syntaxError("expected target name after <?")
   554				}
   555				return nil, d.err
   556			}
   557			d.space()
   558			d.buf.Reset()
   559			var b0 byte
   560			for {
   561				if b, ok = d.mustgetc(); !ok {
   562					return nil, d.err
   563				}
   564				d.buf.WriteByte(b)
   565				if b0 == '?' && b == '>' {
   566					break
   567				}
   568				b0 = b
   569			}
   570			data := d.buf.Bytes()
   571			data = data[0 : len(data)-2] // chop ?>
   572	
   573			if target == "xml" {
   574				enc := procInstEncoding(string(data))
   575				if enc != "" && enc != "utf-8" && enc != "UTF-8" {
   576					if d.CharsetReader == nil {
   577						d.err = fmt.Errorf("xml: encoding %q declared but Decoder.CharsetReader is nil", enc)
   578						return nil, d.err
   579					}
   580					newr, err := d.CharsetReader(enc, d.r.(io.Reader))
   581					if err != nil {
   582						d.err = fmt.Errorf("xml: opening charset %q: %v", enc, err)
   583						return nil, d.err
   584					}
   585					if newr == nil {
   586						panic("CharsetReader returned a nil Reader for charset " + enc)
   587					}
   588					d.switchToReader(newr)
   589				}
   590			}
   591			return ProcInst{target, data}, nil
   592	
   593		case '!':
   594			// <!: Maybe comment, maybe CDATA.
   595			if b, ok = d.mustgetc(); !ok {
   596				return nil, d.err
   597			}
   598			switch b {
   599			case '-': // <!-
   600				// Probably <!-- for a comment.
   601				if b, ok = d.mustgetc(); !ok {
   602					return nil, d.err
   603				}
   604				if b != '-' {
   605					d.err = d.syntaxError("invalid sequence <!- not part of <!--")
   606					return nil, d.err
   607				}
   608				// Look for terminator.
   609				d.buf.Reset()
   610				var b0, b1 byte
   611				for {
   612					if b, ok = d.mustgetc(); !ok {
   613						return nil, d.err
   614					}
   615					d.buf.WriteByte(b)
   616					if b0 == '-' && b1 == '-' && b == '>' {
   617						break
   618					}
   619					b0, b1 = b1, b
   620				}
   621				data := d.buf.Bytes()
   622				data = data[0 : len(data)-3] // chop -->
   623				return Comment(data), nil
   624	
   625			case '[': // <![
   626				// Probably <![CDATA[.
   627				for i := 0; i < 6; i++ {
   628					if b, ok = d.mustgetc(); !ok {
   629						return nil, d.err
   630					}
   631					if b != "CDATA["[i] {
   632						d.err = d.syntaxError("invalid <![ sequence")
   633						return nil, d.err
   634					}
   635				}
   636				// Have <![CDATA[.  Read text until ]]>.
   637				data := d.text(-1, true)
   638				if data == nil {
   639					return nil, d.err
   640				}
   641				return CharData(data), nil
   642			}
   643	
   644			// Probably a directive: <!DOCTYPE ...>, <!ENTITY ...>, etc.
   645			// We don't care, but accumulate for caller. Quoted angle
   646			// brackets do not count for nesting.
   647			d.buf.Reset()
   648			d.buf.WriteByte(b)
   649			inquote := uint8(0)
   650			depth := 0
   651			for {
   652				if b, ok = d.mustgetc(); !ok {
   653					return nil, d.err
   654				}
   655				if inquote == 0 && b == '>' && depth == 0 {
   656					break
   657				}
   658			HandleB:
   659				d.buf.WriteByte(b)
   660				switch {
   661				case b == inquote:
   662					inquote = 0
   663	
   664				case inquote != 0:
   665					// in quotes, no special action
   666	
   667				case b == '\'' || b == '"':
   668					inquote = b
   669	
   670				case b == '>' && inquote == 0:
   671					depth--
   672	
   673				case b == '<' && inquote == 0:
   674					// Look for <!-- to begin comment.
   675					s := "!--"
   676					for i := 0; i < len(s); i++ {
   677						if b, ok = d.mustgetc(); !ok {
   678							return nil, d.err
   679						}
   680						if b != s[i] {
   681							for j := 0; j < i; j++ {
   682								d.buf.WriteByte(s[j])
   683							}
   684							depth++
   685							goto HandleB
   686						}
   687					}
   688	
   689					// Remove < that was written above.
   690					d.buf.Truncate(d.buf.Len() - 1)
   691	
   692					// Look for terminator.
   693					var b0, b1 byte
   694					for {
   695						if b, ok = d.mustgetc(); !ok {
   696							return nil, d.err
   697						}
   698						if b0 == '-' && b1 == '-' && b == '>' {
   699							break
   700						}
   701						b0, b1 = b1, b
   702					}
   703				}
   704			}
   705			return Directive(d.buf.Bytes()), nil
   706		}
   707	
   708		// Must be an open element like <a href="foo">
   709		d.ungetc(b)
   710	
   711		var (
   712			name  Name
   713			empty bool
   714			attr  []Attr
   715		)
   716		if name, ok = d.nsname(); !ok {
   717			if d.err == nil {
   718				d.err = d.syntaxError("expected element name after <")
   719			}
   720			return nil, d.err
   721		}
   722	
   723		attr = make([]Attr, 0, 4)
   724		for {
   725			d.space()
   726			if b, ok = d.mustgetc(); !ok {
   727				return nil, d.err
   728			}
   729			if b == '/' {
   730				empty = true
   731				if b, ok = d.mustgetc(); !ok {
   732					return nil, d.err
   733				}
   734				if b != '>' {
   735					d.err = d.syntaxError("expected /> in element")
   736					return nil, d.err
   737				}
   738				break
   739			}
   740			if b == '>' {
   741				break
   742			}
   743			d.ungetc(b)
   744	
   745			n := len(attr)
   746			if n >= cap(attr) {
   747				nattr := make([]Attr, n, 2*cap(attr))
   748				copy(nattr, attr)
   749				attr = nattr
   750			}
   751			attr = attr[0 : n+1]
   752			a := &attr[n]
   753			if a.Name, ok = d.nsname(); !ok {
   754				if d.err == nil {
   755					d.err = d.syntaxError("expected attribute name in element")
   756				}
   757				return nil, d.err
   758			}
   759			d.space()
   760			if b, ok = d.mustgetc(); !ok {
   761				return nil, d.err
   762			}
   763			if b != '=' {
   764				if d.Strict {
   765					d.err = d.syntaxError("attribute name without = in element")
   766					return nil, d.err
   767				} else {
   768					d.ungetc(b)
   769					a.Value = a.Name.Local
   770				}
   771			} else {
   772				d.space()
   773				data := d.attrval()
   774				if data == nil {
   775					return nil, d.err
   776				}
   777				a.Value = string(data)
   778			}
   779		}
   780		if empty {
   781			d.needClose = true
   782			d.toClose = name
   783		}
   784		return StartElement{name, attr}, nil
   785	}
   786	
   787	func (d *Decoder) attrval() []byte {
   788		b, ok := d.mustgetc()
   789		if !ok {
   790			return nil
   791		}
   792		// Handle quoted attribute values
   793		if b == '"' || b == '\'' {
   794			return d.text(int(b), false)
   795		}
   796		// Handle unquoted attribute values for strict parsers
   797		if d.Strict {
   798			d.err = d.syntaxError("unquoted or missing attribute value in element")
   799			return nil
   800		}
   801		// Handle unquoted attribute values for unstrict parsers
   802		d.ungetc(b)
   803		d.buf.Reset()
   804		for {
   805			b, ok = d.mustgetc()
   806			if !ok {
   807				return nil
   808			}
   809			// http://www.w3.org/TR/REC-html40/intro/sgmltut.html#h-3.2.2
   810			if 'a' <= b && b <= 'z' || 'A' <= b && b <= 'Z' ||
   811				'0' <= b && b <= '9' || b == '_' || b == ':' || b == '-' {
   812				d.buf.WriteByte(b)
   813			} else {
   814				d.ungetc(b)
   815				break
   816			}
   817		}
   818		return d.buf.Bytes()
   819	}
   820	
   821	// Skip spaces if any
   822	func (d *Decoder) space() {
   823		for {
   824			b, ok := d.getc()
   825			if !ok {
   826				return
   827			}
   828			switch b {
   829			case ' ', '\r', '\n', '\t':
   830			default:
   831				d.ungetc(b)
   832				return
   833			}
   834		}
   835	}
   836	
   837	// Read a single byte.
   838	// If there is no byte to read, return ok==false
   839	// and leave the error in d.err.
   840	// Maintain line number.
   841	func (d *Decoder) getc() (b byte, ok bool) {
   842		if d.err != nil {
   843			return 0, false
   844		}
   845		if d.nextByte >= 0 {
   846			b = byte(d.nextByte)
   847			d.nextByte = -1
   848		} else {
   849			b, d.err = d.r.ReadByte()
   850			if d.err != nil {
   851				return 0, false
   852			}
   853			if d.saved != nil {
   854				d.saved.WriteByte(b)
   855			}
   856		}
   857		if b == '\n' {
   858			d.line++
   859		}
   860		return b, true
   861	}
   862	
   863	// Return saved offset.
   864	// If we did ungetc (nextByte >= 0), have to back up one.
   865	func (d *Decoder) savedOffset() int {
   866		n := d.saved.Len()
   867		if d.nextByte >= 0 {
   868			n--
   869		}
   870		return n
   871	}
   872	
   873	// Must read a single byte.
   874	// If there is no byte to read,
   875	// set d.err to SyntaxError("unexpected EOF")
   876	// and return ok==false
   877	func (d *Decoder) mustgetc() (b byte, ok bool) {
   878		if b, ok = d.getc(); !ok {
   879			if d.err == io.EOF {
   880				d.err = d.syntaxError("unexpected EOF")
   881			}
   882		}
   883		return
   884	}
   885	
   886	// Unread a single byte.
   887	func (d *Decoder) ungetc(b byte) {
   888		if b == '\n' {
   889			d.line--
   890		}
   891		d.nextByte = int(b)
   892	}
   893	
   894	var entity = map[string]int{
   895		"lt":   '<',
   896		"gt":   '>',
   897		"amp":  '&',
   898		"apos": '\'',
   899		"quot": '"',
   900	}
   901	
   902	// Read plain text section (XML calls it character data).
   903	// If quote >= 0, we are in a quoted string and need to find the matching quote.
   904	// If cdata == true, we are in a <![CDATA[ section and need to find ]]>.
   905	// On failure return nil and leave the error in d.err.
   906	func (d *Decoder) text(quote int, cdata bool) []byte {
   907		var b0, b1 byte
   908		var trunc int
   909		d.buf.Reset()
   910	Input:
   911		for {
   912			b, ok := d.getc()
   913			if !ok {
   914				if cdata {
   915					if d.err == io.EOF {
   916						d.err = d.syntaxError("unexpected EOF in CDATA section")
   917					}
   918					return nil
   919				}
   920				break Input
   921			}
   922	
   923			// <![CDATA[ section ends with ]]>.
   924			// It is an error for ]]> to appear in ordinary text.
   925			if b0 == ']' && b1 == ']' && b == '>' {
   926				if cdata {
   927					trunc = 2
   928					break Input
   929				}
   930				d.err = d.syntaxError("unescaped ]]> not in CDATA section")
   931				return nil
   932			}
   933	
   934			// Stop reading text if we see a <.
   935			if b == '<' && !cdata {
   936				if quote >= 0 {
   937					d.err = d.syntaxError("unescaped < inside quoted string")
   938					return nil
   939				}
   940				d.ungetc('<')
   941				break Input
   942			}
   943			if quote >= 0 && b == byte(quote) {
   944				break Input
   945			}
   946			if b == '&' && !cdata {
   947				// Read escaped character expression up to semicolon.
   948				// XML in all its glory allows a document to define and use
   949				// its own character names with <!ENTITY ...> directives.
   950				// Parsers are required to recognize lt, gt, amp, apos, and quot
   951				// even if they have not been declared.
   952				before := d.buf.Len()
   953				d.buf.WriteByte('&')
   954				var ok bool
   955				var text string
   956				var haveText bool
   957				if b, ok = d.mustgetc(); !ok {
   958					return nil
   959				}
   960				if b == '#' {
   961					d.buf.WriteByte(b)
   962					if b, ok = d.mustgetc(); !ok {
   963						return nil
   964					}
   965					base := 10
   966					if b == 'x' {
   967						base = 16
   968						d.buf.WriteByte(b)
   969						if b, ok = d.mustgetc(); !ok {
   970							return nil
   971						}
   972					}
   973					start := d.buf.Len()
   974					for '0' <= b && b <= '9' ||
   975						base == 16 && 'a' <= b && b <= 'f' ||
   976						base == 16 && 'A' <= b && b <= 'F' {
   977						d.buf.WriteByte(b)
   978						if b, ok = d.mustgetc(); !ok {
   979							return nil
   980						}
   981					}
   982					if b != ';' {
   983						d.ungetc(b)
   984					} else {
   985						s := string(d.buf.Bytes()[start:])
   986						d.buf.WriteByte(';')
   987						n, err := strconv.ParseUint(s, base, 64)
   988						if err == nil && n <= unicode.MaxRune {
   989							text = string(n)
   990							haveText = true
   991						}
   992					}
   993				} else {
   994					d.ungetc(b)
   995					if !d.readName() {
   996						if d.err != nil {
   997							return nil
   998						}
   999						ok = false
  1000					}
  1001					if b, ok = d.mustgetc(); !ok {
  1002						return nil
  1003					}
  1004					if b != ';' {
  1005						d.ungetc(b)
  1006					} else {
  1007						name := d.buf.Bytes()[before+1:]
  1008						d.buf.WriteByte(';')
  1009						if isName(name) {
  1010							s := string(name)
  1011							if r, ok := entity[s]; ok {
  1012								text = string(r)
  1013								haveText = true
  1014							} else if d.Entity != nil {
  1015								text, haveText = d.Entity[s]
  1016							}
  1017						}
  1018					}
  1019				}
  1020	
  1021				if haveText {
  1022					d.buf.Truncate(before)
  1023					d.buf.Write([]byte(text))
  1024					b0, b1 = 0, 0
  1025					continue Input
  1026				}
  1027				if !d.Strict {
  1028					b0, b1 = 0, 0
  1029					continue Input
  1030				}
  1031				ent := string(d.buf.Bytes()[before:])
  1032				if ent[len(ent)-1] != ';' {
  1033					ent += " (no semicolon)"
  1034				}
  1035				d.err = d.syntaxError("invalid character entity " + ent)
  1036				return nil
  1037			}
  1038	
  1039			// We must rewrite unescaped \r and \r\n into \n.
  1040			if b == '\r' {
  1041				d.buf.WriteByte('\n')
  1042			} else if b1 == '\r' && b == '\n' {
  1043				// Skip \r\n--we already wrote \n.
  1044			} else {
  1045				d.buf.WriteByte(b)
  1046			}
  1047	
  1048			b0, b1 = b1, b
  1049		}
  1050		data := d.buf.Bytes()
  1051		data = data[0 : len(data)-trunc]
  1052	
  1053		// Inspect each rune for being a disallowed character.
  1054		buf := data
  1055		for len(buf) > 0 {
  1056			r, size := utf8.DecodeRune(buf)
  1057			if r == utf8.RuneError && size == 1 {
  1058				d.err = d.syntaxError("invalid UTF-8")
  1059				return nil
  1060			}
  1061			buf = buf[size:]
  1062			if !isInCharacterRange(r) {
  1063				d.err = d.syntaxError(fmt.Sprintf("illegal character code %U", r))
  1064				return nil
  1065			}
  1066		}
  1067	
  1068		return data
  1069	}
  1070	
  1071	// Decide whether the given rune is in the XML Character Range, per
  1072	// the Char production of http://www.xml.com/axml/testaxml.htm,
  1073	// Section 2.2 Characters.
  1074	func isInCharacterRange(r rune) (inrange bool) {
  1075		return r == 0x09 ||
  1076			r == 0x0A ||
  1077			r == 0x0D ||
  1078			r >= 0x20 && r <= 0xDF77 ||
  1079			r >= 0xE000 && r <= 0xFFFD ||
  1080			r >= 0x10000 && r <= 0x10FFFF
  1081	}
  1082	
  1083	// Get name space name: name with a : stuck in the middle.
  1084	// The part before the : is the name space identifier.
  1085	func (d *Decoder) nsname() (name Name, ok bool) {
  1086		s, ok := d.name()
  1087		if !ok {
  1088			return
  1089		}
  1090		i := strings.Index(s, ":")
  1091		if i < 0 {
  1092			name.Local = s
  1093		} else {
  1094			name.Space = s[0:i]
  1095			name.Local = s[i+1:]
  1096		}
  1097		return name, true
  1098	}
  1099	
  1100	// Get name: /first(first|second)*/
  1101	// Do not set d.err if the name is missing (unless unexpected EOF is received):
  1102	// let the caller provide better context.
  1103	func (d *Decoder) name() (s string, ok bool) {
  1104		d.buf.Reset()
  1105		if !d.readName() {
  1106			return "", false
  1107		}
  1108	
  1109		// Now we check the characters.
  1110		s = d.buf.String()
  1111		if !isName([]byte(s)) {
  1112			d.err = d.syntaxError("invalid XML name: " + s)
  1113			return "", false
  1114		}
  1115		return s, true
  1116	}
  1117	
  1118	// Read a name and append its bytes to d.buf.
  1119	// The name is delimited by any single-byte character not valid in names.
  1120	// All multi-byte characters are accepted; the caller must check their validity.
  1121	func (d *Decoder) readName() (ok bool) {
  1122		var b byte
  1123		if b, ok = d.mustgetc(); !ok {
  1124			return
  1125		}
  1126		if b < utf8.RuneSelf && !isNameByte(b) {
  1127			d.ungetc(b)
  1128			return false
  1129		}
  1130		d.buf.WriteByte(b)
  1131	
  1132		for {
  1133			if b, ok = d.mustgetc(); !ok {
  1134				return
  1135			}
  1136			if b < utf8.RuneSelf && !isNameByte(b) {
  1137				d.ungetc(b)
  1138				break
  1139			}
  1140			d.buf.WriteByte(b)
  1141		}
  1142		return true
  1143	}
  1144	
  1145	func isNameByte(c byte) bool {
  1146		return 'A' <= c && c <= 'Z' ||
  1147			'a' <= c && c <= 'z' ||
  1148			'0' <= c && c <= '9' ||
  1149			c == '_' || c == ':' || c == '.' || c == '-'
  1150	}
  1151	
  1152	func isName(s []byte) bool {
  1153		if len(s) == 0 {
  1154			return false
  1155		}
  1156		c, n := utf8.DecodeRune(s)
  1157		if c == utf8.RuneError && n == 1 {
  1158			return false
  1159		}
  1160		if !unicode.Is(first, c) {
  1161			return false
  1162		}
  1163		for n < len(s) {
  1164			s = s[n:]
  1165			c, n = utf8.DecodeRune(s)
  1166			if c == utf8.RuneError && n == 1 {
  1167				return false
  1168			}
  1169			if !unicode.Is(first, c) && !unicode.Is(second, c) {
  1170				return false
  1171			}
  1172		}
  1173		return true
  1174	}
  1175	
  1176	func isNameString(s string) bool {
  1177		if len(s) == 0 {
  1178			return false
  1179		}
  1180		c, n := utf8.DecodeRuneInString(s)
  1181		if c == utf8.RuneError && n == 1 {
  1182			return false
  1183		}
  1184		if !unicode.Is(first, c) {
  1185			return false
  1186		}
  1187		for n < len(s) {
  1188			s = s[n:]
  1189			c, n = utf8.DecodeRuneInString(s)
  1190			if c == utf8.RuneError && n == 1 {
  1191				return false
  1192			}
  1193			if !unicode.Is(first, c) && !unicode.Is(second, c) {
  1194				return false
  1195			}
  1196		}
  1197		return true
  1198	}
  1199	
  1200	// These tables were generated by cut and paste from Appendix B of
  1201	// the XML spec at http://www.xml.com/axml/testaxml.htm
  1202	// and then reformatting.  First corresponds to (Letter | '_' | ':')
  1203	// and second corresponds to NameChar.
  1204	
  1205	var first = &unicode.RangeTable{
  1206		R16: []unicode.Range16{
  1207			{0x003A, 0x003A, 1},
  1208			{0x0041, 0x005A, 1},
  1209			{0x005F, 0x005F, 1},
  1210			{0x0061, 0x007A, 1},
  1211			{0x00C0, 0x00D6, 1},
  1212			{0x00D8, 0x00F6, 1},
  1213			{0x00F8, 0x00FF, 1},
  1214			{0x0100, 0x0131, 1},
  1215			{0x0134, 0x013E, 1},
  1216			{0x0141, 0x0148, 1},
  1217			{0x014A, 0x017E, 1},
  1218			{0x0180, 0x01C3, 1},
  1219			{0x01CD, 0x01F0, 1},
  1220			{0x01F4, 0x01F5, 1},
  1221			{0x01FA, 0x0217, 1},
  1222			{0x0250, 0x02A8, 1},
  1223			{0x02BB, 0x02C1, 1},
  1224			{0x0386, 0x0386, 1},
  1225			{0x0388, 0x038A, 1},
  1226			{0x038C, 0x038C, 1},
  1227			{0x038E, 0x03A1, 1},
  1228			{0x03A3, 0x03CE, 1},
  1229			{0x03D0, 0x03D6, 1},
  1230			{0x03DA, 0x03E0, 2},
  1231			{0x03E2, 0x03F3, 1},
  1232			{0x0401, 0x040C, 1},
  1233			{0x040E, 0x044F, 1},
  1234			{0x0451, 0x045C, 1},
  1235			{0x045E, 0x0481, 1},
  1236			{0x0490, 0x04C4, 1},
  1237			{0x04C7, 0x04C8, 1},
  1238			{0x04CB, 0x04CC, 1},
  1239			{0x04D0, 0x04EB, 1},
  1240			{0x04EE, 0x04F5, 1},
  1241			{0x04F8, 0x04F9, 1},
  1242			{0x0531, 0x0556, 1},
  1243			{0x0559, 0x0559, 1},
  1244			{0x0561, 0x0586, 1},
  1245			{0x05D0, 0x05EA, 1},
  1246			{0x05F0, 0x05F2, 1},
  1247			{0x0621, 0x063A, 1},
  1248			{0x0641, 0x064A, 1},
  1249			{0x0671, 0x06B7, 1},
  1250			{0x06BA, 0x06BE, 1},
  1251			{0x06C0, 0x06CE, 1},
  1252			{0x06D0, 0x06D3, 1},
  1253			{0x06D5, 0x06D5, 1},
  1254			{0x06E5, 0x06E6, 1},
  1255			{0x0905, 0x0939, 1},
  1256			{0x093D, 0x093D, 1},
  1257			{0x0958, 0x0961, 1},
  1258			{0x0985, 0x098C, 1},
  1259			{0x098F, 0x0990, 1},
  1260			{0x0993, 0x09A8, 1},
  1261			{0x09AA, 0x09B0, 1},
  1262			{0x09B2, 0x09B2, 1},
  1263			{0x09B6, 0x09B9, 1},
  1264			{0x09DC, 0x09DD, 1},
  1265			{0x09DF, 0x09E1, 1},
  1266			{0x09F0, 0x09F1, 1},
  1267			{0x0A05, 0x0A0A, 1},
  1268			{0x0A0F, 0x0A10, 1},
  1269			{0x0A13, 0x0A28, 1},
  1270			{0x0A2A, 0x0A30, 1},
  1271			{0x0A32, 0x0A33, 1},
  1272			{0x0A35, 0x0A36, 1},
  1273			{0x0A38, 0x0A39, 1},
  1274			{0x0A59, 0x0A5C, 1},
  1275			{0x0A5E, 0x0A5E, 1},
  1276			{0x0A72, 0x0A74, 1},
  1277			{0x0A85, 0x0A8B, 1},
  1278			{0x0A8D, 0x0A8D, 1},
  1279			{0x0A8F, 0x0A91, 1},
  1280			{0x0A93, 0x0AA8, 1},
  1281			{0x0AAA, 0x0AB0, 1},
  1282			{0x0AB2, 0x0AB3, 1},
  1283			{0x0AB5, 0x0AB9, 1},
  1284			{0x0ABD, 0x0AE0, 0x23},
  1285			{0x0B05, 0x0B0C, 1},
  1286			{0x0B0F, 0x0B10, 1},
  1287			{0x0B13, 0x0B28, 1},
  1288			{0x0B2A, 0x0B30, 1},
  1289			{0x0B32, 0x0B33, 1},
  1290			{0x0B36, 0x0B39, 1},
  1291			{0x0B3D, 0x0B3D, 1},
  1292			{0x0B5C, 0x0B5D, 1},
  1293			{0x0B5F, 0x0B61, 1},
  1294			{0x0B85, 0x0B8A, 1},
  1295			{0x0B8E, 0x0B90, 1},
  1296			{0x0B92, 0x0B95, 1},
  1297			{0x0B99, 0x0B9A, 1},
  1298			{0x0B9C, 0x0B9C, 1},
  1299			{0x0B9E, 0x0B9F, 1},
  1300			{0x0BA3, 0x0BA4, 1},
  1301			{0x0BA8, 0x0BAA, 1},
  1302			{0x0BAE, 0x0BB5, 1},
  1303			{0x0BB7, 0x0BB9, 1},
  1304			{0x0C05, 0x0C0C, 1},
  1305			{0x0C0E, 0x0C10, 1},
  1306			{0x0C12, 0x0C28, 1},
  1307			{0x0C2A, 0x0C33, 1},
  1308			{0x0C35, 0x0C39, 1},
  1309			{0x0C60, 0x0C61, 1},
  1310			{0x0C85, 0x0C8C, 1},
  1311			{0x0C8E, 0x0C90, 1},
  1312			{0x0C92, 0x0CA8, 1},
  1313			{0x0CAA, 0x0CB3, 1},
  1314			{0x0CB5, 0x0CB9, 1},
  1315			{0x0CDE, 0x0CDE, 1},
  1316			{0x0CE0, 0x0CE1, 1},
  1317			{0x0D05, 0x0D0C, 1},
  1318			{0x0D0E, 0x0D10, 1},
  1319			{0x0D12, 0x0D28, 1},
  1320			{0x0D2A, 0x0D39, 1},
  1321			{0x0D60, 0x0D61, 1},
  1322			{0x0E01, 0x0E2E, 1},
  1323			{0x0E30, 0x0E30, 1},
  1324			{0x0E32, 0x0E33, 1},
  1325			{0x0E40, 0x0E45, 1},
  1326			{0x0E81, 0x0E82, 1},
  1327			{0x0E84, 0x0E84, 1},
  1328			{0x0E87, 0x0E88, 1},
  1329			{0x0E8A, 0x0E8D, 3},
  1330			{0x0E94, 0x0E97, 1},
  1331			{0x0E99, 0x0E9F, 1},
  1332			{0x0EA1, 0x0EA3, 1},
  1333			{0x0EA5, 0x0EA7, 2},
  1334			{0x0EAA, 0x0EAB, 1},
  1335			{0x0EAD, 0x0EAE, 1},
  1336			{0x0EB0, 0x0EB0, 1},
  1337			{0x0EB2, 0x0EB3, 1},
  1338			{0x0EBD, 0x0EBD, 1},
  1339			{0x0EC0, 0x0EC4, 1},
  1340			{0x0F40, 0x0F47, 1},
  1341			{0x0F49, 0x0F69, 1},
  1342			{0x10A0, 0x10C5, 1},
  1343			{0x10D0, 0x10F6, 1},
  1344			{0x1100, 0x1100, 1},
  1345			{0x1102, 0x1103, 1},
  1346			{0x1105, 0x1107, 1},
  1347			{0x1109, 0x1109, 1},
  1348			{0x110B, 0x110C, 1},
  1349			{0x110E, 0x1112, 1},
  1350			{0x113C, 0x1140, 2},
  1351			{0x114C, 0x1150, 2},
  1352			{0x1154, 0x1155, 1},
  1353			{0x1159, 0x1159, 1},
  1354			{0x115F, 0x1161, 1},
  1355			{0x1163, 0x1169, 2},
  1356			{0x116D, 0x116E, 1},
  1357			{0x1172, 0x1173, 1},
  1358			{0x1175, 0x119E, 0x119E - 0x1175},
  1359			{0x11A8, 0x11AB, 0x11AB - 0x11A8},
  1360			{0x11AE, 0x11AF, 1},
  1361			{0x11B7, 0x11B8, 1},
  1362			{0x11BA, 0x11BA, 1},
  1363			{0x11BC, 0x11C2, 1},
  1364			{0x11EB, 0x11F0, 0x11F0 - 0x11EB},
  1365			{0x11F9, 0x11F9, 1},
  1366			{0x1E00, 0x1E9B, 1},
  1367			{0x1EA0, 0x1EF9, 1},
  1368			{0x1F00, 0x1F15, 1},
  1369			{0x1F18, 0x1F1D, 1},
  1370			{0x1F20, 0x1F45, 1},
  1371			{0x1F48, 0x1F4D, 1},
  1372			{0x1F50, 0x1F57, 1},
  1373			{0x1F59, 0x1F5B, 0x1F5B - 0x1F59},
  1374			{0x1F5D, 0x1F5D, 1},
  1375			{0x1F5F, 0x1F7D, 1},
  1376			{0x1F80, 0x1FB4, 1},
  1377			{0x1FB6, 0x1FBC, 1},
  1378			{0x1FBE, 0x1FBE, 1},
  1379			{0x1FC2, 0x1FC4, 1},
  1380			{0x1FC6, 0x1FCC, 1},
  1381			{0x1FD0, 0x1FD3, 1},
  1382			{0x1FD6, 0x1FDB, 1},
  1383			{0x1FE0, 0x1FEC, 1},
  1384			{0x1FF2, 0x1FF4, 1},
  1385			{0x1FF6, 0x1FFC, 1},
  1386			{0x2126, 0x2126, 1},
  1387			{0x212A, 0x212B, 1},
  1388			{0x212E, 0x212E, 1},
  1389			{0x2180, 0x2182, 1},
  1390			{0x3007, 0x3007, 1},
  1391			{0x3021, 0x3029, 1},
  1392			{0x3041, 0x3094, 1},
  1393			{0x30A1, 0x30FA, 1},
  1394			{0x3105, 0x312C, 1},
  1395			{0x4E00, 0x9FA5, 1},
  1396			{0xAC00, 0xD7A3, 1},
  1397		},
  1398	}
  1399	
  1400	var second = &unicode.RangeTable{
  1401		R16: []unicode.Range16{
  1402			{0x002D, 0x002E, 1},
  1403			{0x0030, 0x0039, 1},
  1404			{0x00B7, 0x00B7, 1},
  1405			{0x02D0, 0x02D1, 1},
  1406			{0x0300, 0x0345, 1},
  1407			{0x0360, 0x0361, 1},
  1408			{0x0387, 0x0387, 1},
  1409			{0x0483, 0x0486, 1},
  1410			{0x0591, 0x05A1, 1},
  1411			{0x05A3, 0x05B9, 1},
  1412			{0x05BB, 0x05BD, 1},
  1413			{0x05BF, 0x05BF, 1},
  1414			{0x05C1, 0x05C2, 1},
  1415			{0x05C4, 0x0640, 0x0640 - 0x05C4},
  1416			{0x064B, 0x0652, 1},
  1417			{0x0660, 0x0669, 1},
  1418			{0x0670, 0x0670, 1},
  1419			{0x06D6, 0x06DC, 1},
  1420			{0x06DD, 0x06DF, 1},
  1421			{0x06E0, 0x06E4, 1},
  1422			{0x06E7, 0x06E8, 1},
  1423			{0x06EA, 0x06ED, 1},
  1424			{0x06F0, 0x06F9, 1},
  1425			{0x0901, 0x0903, 1},
  1426			{0x093C, 0x093C, 1},
  1427			{0x093E, 0x094C, 1},
  1428			{0x094D, 0x094D, 1},
  1429			{0x0951, 0x0954, 1},
  1430			{0x0962, 0x0963, 1},
  1431			{0x0966, 0x096F, 1},
  1432			{0x0981, 0x0983, 1},
  1433			{0x09BC, 0x09BC, 1},
  1434			{0x09BE, 0x09BF, 1},
  1435			{0x09C0, 0x09C4, 1},
  1436			{0x09C7, 0x09C8, 1},
  1437			{0x09CB, 0x09CD, 1},
  1438			{0x09D7, 0x09D7, 1},
  1439			{0x09E2, 0x09E3, 1},
  1440			{0x09E6, 0x09EF, 1},
  1441			{0x0A02, 0x0A3C, 0x3A},
  1442			{0x0A3E, 0x0A3F, 1},
  1443			{0x0A40, 0x0A42, 1},
  1444			{0x0A47, 0x0A48, 1},
  1445			{0x0A4B, 0x0A4D, 1},
  1446			{0x0A66, 0x0A6F, 1},
  1447			{0x0A70, 0x0A71, 1},
  1448			{0x0A81, 0x0A83, 1},
  1449			{0x0ABC, 0x0ABC, 1},
  1450			{0x0ABE, 0x0AC5, 1},
  1451			{0x0AC7, 0x0AC9, 1},
  1452			{0x0ACB, 0x0ACD, 1},
  1453			{0x0AE6, 0x0AEF, 1},
  1454			{0x0B01, 0x0B03, 1},
  1455			{0x0B3C, 0x0B3C, 1},
  1456			{0x0B3E, 0x0B43, 1},
  1457			{0x0B47, 0x0B48, 1},
  1458			{0x0B4B, 0x0B4D, 1},
  1459			{0x0B56, 0x0B57, 1},
  1460			{0x0B66, 0x0B6F, 1},
  1461			{0x0B82, 0x0B83, 1},
  1462			{0x0BBE, 0x0BC2, 1},
  1463			{0x0BC6, 0x0BC8, 1},
  1464			{0x0BCA, 0x0BCD, 1},
  1465			{0x0BD7, 0x0BD7, 1},
  1466			{0x0BE7, 0x0BEF, 1},
  1467			{0x0C01, 0x0C03, 1},
  1468			{0x0C3E, 0x0C44, 1},
  1469			{0x0C46, 0x0C48, 1},
  1470			{0x0C4A, 0x0C4D, 1},
  1471			{0x0C55, 0x0C56, 1},
  1472			{0x0C66, 0x0C6F, 1},
  1473			{0x0C82, 0x0C83, 1},
  1474			{0x0CBE, 0x0CC4, 1},
  1475			{0x0CC6, 0x0CC8, 1},
  1476			{0x0CCA, 0x0CCD, 1},
  1477			{0x0CD5, 0x0CD6, 1},
  1478			{0x0CE6, 0x0CEF, 1},
  1479			{0x0D02, 0x0D03, 1},
  1480			{0x0D3E, 0x0D43, 1},
  1481			{0x0D46, 0x0D48, 1},
  1482			{0x0D4A, 0x0D4D, 1},
  1483			{0x0D57, 0x0D57, 1},
  1484			{0x0D66, 0x0D6F, 1},
  1485			{0x0E31, 0x0E31, 1},
  1486			{0x0E34, 0x0E3A, 1},
  1487			{0x0E46, 0x0E46, 1},
  1488			{0x0E47, 0x0E4E, 1},
  1489			{0x0E50, 0x0E59, 1},
  1490			{0x0EB1, 0x0EB1, 1},
  1491			{0x0EB4, 0x0EB9, 1},
  1492			{0x0EBB, 0x0EBC, 1},
  1493			{0x0EC6, 0x0EC6, 1},
  1494			{0x0EC8, 0x0ECD, 1},
  1495			{0x0ED0, 0x0ED9, 1},
  1496			{0x0F18, 0x0F19, 1},
  1497			{0x0F20, 0x0F29, 1},
  1498			{0x0F35, 0x0F39, 2},
  1499			{0x0F3E, 0x0F3F, 1},
  1500			{0x0F71, 0x0F84, 1},
  1501			{0x0F86, 0x0F8B, 1},
  1502			{0x0F90, 0x0F95, 1},
  1503			{0x0F97, 0x0F97, 1},
  1504			{0x0F99, 0x0FAD, 1},
  1505			{0x0FB1, 0x0FB7, 1},
  1506			{0x0FB9, 0x0FB9, 1},
  1507			{0x20D0, 0x20DC, 1},
  1508			{0x20E1, 0x3005, 0x3005 - 0x20E1},
  1509			{0x302A, 0x302F, 1},
  1510			{0x3031, 0x3035, 1},
  1511			{0x3099, 0x309A, 1},
  1512			{0x309D, 0x309E, 1},
  1513			{0x30FC, 0x30FE, 1},
  1514		},
  1515	}
  1516	
  1517	// HTMLEntity is an entity map containing translations for the
  1518	// standard HTML entity characters.
  1519	var HTMLEntity = htmlEntity
  1520	
  1521	var htmlEntity = map[string]string{
  1522		/*
  1523			hget http://www.w3.org/TR/html4/sgml/entities.html |
  1524			ssam '
  1525				,y /\&gt;/ x/\&lt;(.|\n)+/ s/\n/ /g
  1526				,x v/^\&lt;!ENTITY/d
  1527				,s/\&lt;!ENTITY ([^ ]+) .*U\+([0-9A-F][0-9A-F][0-9A-F][0-9A-F]) .+/	"\1": "\\u\2",/g
  1528			'
  1529		*/
  1530		"nbsp":     "\u00A0",
  1531		"iexcl":    "\u00A1",
  1532		"cent":     "\u00A2",
  1533		"pound":    "\u00A3",
  1534		"curren":   "\u00A4",
  1535		"yen":      "\u00A5",
  1536		"brvbar":   "\u00A6",
  1537		"sect":     "\u00A7",
  1538		"uml":      "\u00A8",
  1539		"copy":     "\u00A9",
  1540		"ordf":     "\u00AA",
  1541		"laquo":    "\u00AB",
  1542		"not":      "\u00AC",
  1543		"shy":      "\u00AD",
  1544		"reg":      "\u00AE",
  1545		"macr":     "\u00AF",
  1546		"deg":      "\u00B0",
  1547		"plusmn":   "\u00B1",
  1548		"sup2":     "\u00B2",
  1549		"sup3":     "\u00B3",
  1550		"acute":    "\u00B4",
  1551		"micro":    "\u00B5",
  1552		"para":     "\u00B6",
  1553		"middot":   "\u00B7",
  1554		"cedil":    "\u00B8",
  1555		"sup1":     "\u00B9",
  1556		"ordm":     "\u00BA",
  1557		"raquo":    "\u00BB",
  1558		"frac14":   "\u00BC",
  1559		"frac12":   "\u00BD",
  1560		"frac34":   "\u00BE",
  1561		"iquest":   "\u00BF",
  1562		"Agrave":   "\u00C0",
  1563		"Aacute":   "\u00C1",
  1564		"Acirc":    "\u00C2",
  1565		"Atilde":   "\u00C3",
  1566		"Auml":     "\u00C4",
  1567		"Aring":    "\u00C5",
  1568		"AElig":    "\u00C6",
  1569		"Ccedil":   "\u00C7",
  1570		"Egrave":   "\u00C8",
  1571		"Eacute":   "\u00C9",
  1572		"Ecirc":    "\u00CA",
  1573		"Euml":     "\u00CB",
  1574		"Igrave":   "\u00CC",
  1575		"Iacute":   "\u00CD",
  1576		"Icirc":    "\u00CE",
  1577		"Iuml":     "\u00CF",
  1578		"ETH":      "\u00D0",
  1579		"Ntilde":   "\u00D1",
  1580		"Ograve":   "\u00D2",
  1581		"Oacute":   "\u00D3",
  1582		"Ocirc":    "\u00D4",
  1583		"Otilde":   "\u00D5",
  1584		"Ouml":     "\u00D6",
  1585		"times":    "\u00D7",
  1586		"Oslash":   "\u00D8",
  1587		"Ugrave":   "\u00D9",
  1588		"Uacute":   "\u00DA",
  1589		"Ucirc":    "\u00DB",
  1590		"Uuml":     "\u00DC",
  1591		"Yacute":   "\u00DD",
  1592		"THORN":    "\u00DE",
  1593		"szlig":    "\u00DF",
  1594		"agrave":   "\u00E0",
  1595		"aacute":   "\u00E1",
  1596		"acirc":    "\u00E2",
  1597		"atilde":   "\u00E3",
  1598		"auml":     "\u00E4",
  1599		"aring":    "\u00E5",
  1600		"aelig":    "\u00E6",
  1601		"ccedil":   "\u00E7",
  1602		"egrave":   "\u00E8",
  1603		"eacute":   "\u00E9",
  1604		"ecirc":    "\u00EA",
  1605		"euml":     "\u00EB",
  1606		"igrave":   "\u00EC",
  1607		"iacute":   "\u00ED",
  1608		"icirc":    "\u00EE",
  1609		"iuml":     "\u00EF",
  1610		"eth":      "\u00F0",
  1611		"ntilde":   "\u00F1",
  1612		"ograve":   "\u00F2",
  1613		"oacute":   "\u00F3",
  1614		"ocirc":    "\u00F4",
  1615		"otilde":   "\u00F5",
  1616		"ouml":     "\u00F6",
  1617		"divide":   "\u00F7",
  1618		"oslash":   "\u00F8",
  1619		"ugrave":   "\u00F9",
  1620		"uacute":   "\u00FA",
  1621		"ucirc":    "\u00FB",
  1622		"uuml":     "\u00FC",
  1623		"yacute":   "\u00FD",
  1624		"thorn":    "\u00FE",
  1625		"yuml":     "\u00FF",
  1626		"fnof":     "\u0192",
  1627		"Alpha":    "\u0391",
  1628		"Beta":     "\u0392",
  1629		"Gamma":    "\u0393",
  1630		"Delta":    "\u0394",
  1631		"Epsilon":  "\u0395",
  1632		"Zeta":     "\u0396",
  1633		"Eta":      "\u0397",
  1634		"Theta":    "\u0398",
  1635		"Iota":     "\u0399",
  1636		"Kappa":    "\u039A",
  1637		"Lambda":   "\u039B",
  1638		"Mu":       "\u039C",
  1639		"Nu":       "\u039D",
  1640		"Xi":       "\u039E",
  1641		"Omicron":  "\u039F",
  1642		"Pi":       "\u03A0",
  1643		"Rho":      "\u03A1",
  1644		"Sigma":    "\u03A3",
  1645		"Tau":      "\u03A4",
  1646		"Upsilon":  "\u03A5",
  1647		"Phi":      "\u03A6",
  1648		"Chi":      "\u03A7",
  1649		"Psi":      "\u03A8",
  1650		"Omega":    "\u03A9",
  1651		"alpha":    "\u03B1",
  1652		"beta":     "\u03B2",
  1653		"gamma":    "\u03B3",
  1654		"delta":    "\u03B4",
  1655		"epsilon":  "\u03B5",
  1656		"zeta":     "\u03B6",
  1657		"eta":      "\u03B7",
  1658		"theta":    "\u03B8",
  1659		"iota":     "\u03B9",
  1660		"kappa":    "\u03BA",
  1661		"lambda":   "\u03BB",
  1662		"mu":       "\u03BC",
  1663		"nu":       "\u03BD",
  1664		"xi":       "\u03BE",
  1665		"omicron":  "\u03BF",
  1666		"pi":       "\u03C0",
  1667		"rho":      "\u03C1",
  1668		"sigmaf":   "\u03C2",
  1669		"sigma":    "\u03C3",
  1670		"tau":      "\u03C4",
  1671		"upsilon":  "\u03C5",
  1672		"phi":      "\u03C6",
  1673		"chi":      "\u03C7",
  1674		"psi":      "\u03C8",
  1675		"omega":    "\u03C9",
  1676		"thetasym": "\u03D1",
  1677		"upsih":    "\u03D2",
  1678		"piv":      "\u03D6",
  1679		"bull":     "\u2022",
  1680		"hellip":   "\u2026",
  1681		"prime":    "\u2032",
  1682		"Prime":    "\u2033",
  1683		"oline":    "\u203E",
  1684		"frasl":    "\u2044",
  1685		"weierp":   "\u2118",
  1686		"image":    "\u2111",
  1687		"real":     "\u211C",
  1688		"trade":    "\u2122",
  1689		"alefsym":  "\u2135",
  1690		"larr":     "\u2190",
  1691		"uarr":     "\u2191",
  1692		"rarr":     "\u2192",
  1693		"darr":     "\u2193",
  1694		"harr":     "\u2194",
  1695		"crarr":    "\u21B5",
  1696		"lArr":     "\u21D0",
  1697		"uArr":     "\u21D1",
  1698		"rArr":     "\u21D2",
  1699		"dArr":     "\u21D3",
  1700		"hArr":     "\u21D4",
  1701		"forall":   "\u2200",
  1702		"part":     "\u2202",
  1703		"exist":    "\u2203",
  1704		"empty":    "\u2205",
  1705		"nabla":    "\u2207",
  1706		"isin":     "\u2208",
  1707		"notin":    "\u2209",
  1708		"ni":       "\u220B",
  1709		"prod":     "\u220F",
  1710		"sum":      "\u2211",
  1711		"minus":    "\u2212",
  1712		"lowast":   "\u2217",
  1713		"radic":    "\u221A",
  1714		"prop":     "\u221D",
  1715		"infin":    "\u221E",
  1716		"ang":      "\u2220",
  1717		"and":      "\u2227",
  1718		"or":       "\u2228",
  1719		"cap":      "\u2229",
  1720		"cup":      "\u222A",
  1721		"int":      "\u222B",
  1722		"there4":   "\u2234",
  1723		"sim":      "\u223C",
  1724		"cong":     "\u2245",
  1725		"asymp":    "\u2248",
  1726		"ne":       "\u2260",
  1727		"equiv":    "\u2261",
  1728		"le":       "\u2264",
  1729		"ge":       "\u2265",
  1730		"sub":      "\u2282",
  1731		"sup":      "\u2283",
  1732		"nsub":     "\u2284",
  1733		"sube":     "\u2286",
  1734		"supe":     "\u2287",
  1735		"oplus":    "\u2295",
  1736		"otimes":   "\u2297",
  1737		"perp":     "\u22A5",
  1738		"sdot":     "\u22C5",
  1739		"lceil":    "\u2308",
  1740		"rceil":    "\u2309",
  1741		"lfloor":   "\u230A",
  1742		"rfloor":   "\u230B",
  1743		"lang":     "\u2329",
  1744		"rang":     "\u232A",
  1745		"loz":      "\u25CA",
  1746		"spades":   "\u2660",
  1747		"clubs":    "\u2663",
  1748		"hearts":   "\u2665",
  1749		"diams":    "\u2666",
  1750		"quot":     "\u0022",
  1751		"amp":      "\u0026",
  1752		"lt":       "\u003C",
  1753		"gt":       "\u003E",
  1754		"OElig":    "\u0152",
  1755		"oelig":    "\u0153",
  1756		"Scaron":   "\u0160",
  1757		"scaron":   "\u0161",
  1758		"Yuml":     "\u0178",
  1759		"circ":     "\u02C6",
  1760		"tilde":    "\u02DC",
  1761		"ensp":     "\u2002",
  1762		"emsp":     "\u2003",
  1763		"thinsp":   "\u2009",
  1764		"zwnj":     "\u200C",
  1765		"zwj":      "\u200D",
  1766		"lrm":      "\u200E",
  1767		"rlm":      "\u200F",
  1768		"ndash":    "\u2013",
  1769		"mdash":    "\u2014",
  1770		"lsquo":    "\u2018",
  1771		"rsquo":    "\u2019",
  1772		"sbquo":    "\u201A",
  1773		"ldquo":    "\u201C",
  1774		"rdquo":    "\u201D",
  1775		"bdquo":    "\u201E",
  1776		"dagger":   "\u2020",
  1777		"Dagger":   "\u2021",
  1778		"permil":   "\u2030",
  1779		"lsaquo":   "\u2039",
  1780		"rsaquo":   "\u203A",
  1781		"euro":     "\u20AC",
  1782	}
  1783	
  1784	// HTMLAutoClose is the set of HTML elements that
  1785	// should be considered to close automatically.
  1786	var HTMLAutoClose = htmlAutoClose
  1787	
  1788	var htmlAutoClose = []string{
  1789		/*
  1790			hget http://www.w3.org/TR/html4/loose.dtd |
  1791			9 sed -n 's/<!ELEMENT ([^ ]*) +- O EMPTY.+/	"\1",/p' | tr A-Z a-z
  1792		*/
  1793		"basefont",
  1794		"br",
  1795		"area",
  1796		"link",
  1797		"img",
  1798		"param",
  1799		"hr",
  1800		"input",
  1801		"col",
  1802		"frame",
  1803		"isindex",
  1804		"base",
  1805		"meta",
  1806	}
  1807	
  1808	var (
  1809		esc_quot = []byte("&#34;") // shorter than "&quot;"
  1810		esc_apos = []byte("&#39;") // shorter than "&apos;"
  1811		esc_amp  = []byte("&amp;")
  1812		esc_lt   = []byte("&lt;")
  1813		esc_gt   = []byte("&gt;")
  1814		esc_tab  = []byte("&#x9;")
  1815		esc_nl   = []byte("&#xA;")
  1816		esc_cr   = []byte("&#xD;")
  1817		esc_fffd = []byte("\uFFFD") // Unicode replacement character
  1818	)
  1819	
  1820	// EscapeText writes to w the properly escaped XML equivalent
  1821	// of the plain text data s.
  1822	func EscapeText(w io.Writer, s []byte) error {
  1823		var esc []byte
  1824		last := 0
  1825		for i := 0; i < len(s); {
  1826			r, width := utf8.DecodeRune(s[i:])
  1827			i += width
  1828			switch r {
  1829			case '"':
  1830				esc = esc_quot
  1831			case '\'':
  1832				esc = esc_apos
  1833			case '&':
  1834				esc = esc_amp
  1835			case '<':
  1836				esc = esc_lt
  1837			case '>':
  1838				esc = esc_gt
  1839			case '\t':
  1840				esc = esc_tab
  1841			case '\n':
  1842				esc = esc_nl
  1843			case '\r':
  1844				esc = esc_cr
  1845			default:
  1846				if !isInCharacterRange(r) || (r == 0xFFFD && width == 1) {
  1847					esc = esc_fffd
  1848					break
  1849				}
  1850				continue
  1851			}
  1852			if _, err := w.Write(s[last : i-width]); err != nil {
  1853				return err
  1854			}
  1855			if _, err := w.Write(esc); err != nil {
  1856				return err
  1857			}
  1858			last = i
  1859		}
  1860		if _, err := w.Write(s[last:]); err != nil {
  1861			return err
  1862		}
  1863		return nil
  1864	}
  1865	
  1866	// EscapeString writes to p the properly escaped XML equivalent
  1867	// of the plain text data s.
  1868	func (p *printer) EscapeString(s string) {
  1869		var esc []byte
  1870		last := 0
  1871		for i := 0; i < len(s); {
  1872			r, width := utf8.DecodeRuneInString(s[i:])
  1873			i += width
  1874			switch r {
  1875			case '"':
  1876				esc = esc_quot
  1877			case '\'':
  1878				esc = esc_apos
  1879			case '&':
  1880				esc = esc_amp
  1881			case '<':
  1882				esc = esc_lt
  1883			case '>':
  1884				esc = esc_gt
  1885			case '\t':
  1886				esc = esc_tab
  1887			case '\n':
  1888				esc = esc_nl
  1889			case '\r':
  1890				esc = esc_cr
  1891			default:
  1892				if !isInCharacterRange(r) || (r == 0xFFFD && width == 1) {
  1893					esc = esc_fffd
  1894					break
  1895				}
  1896				continue
  1897			}
  1898			p.WriteString(s[last : i-width])
  1899			p.Write(esc)
  1900			last = i
  1901		}
  1902		p.WriteString(s[last:])
  1903	}
  1904	
  1905	// Escape is like EscapeText but omits the error return value.
  1906	// It is provided for backwards compatibility with Go 1.0.
  1907	// Code targeting Go 1.1 or later should use EscapeText.
  1908	func Escape(w io.Writer, s []byte) {
  1909		EscapeText(w, s)
  1910	}
  1911	
  1912	// procInstEncoding parses the `encoding="..."` or `encoding='...'`
  1913	// value out of the provided string, returning "" if not found.
  1914	func procInstEncoding(s string) string {
  1915		// TODO: this parsing is somewhat lame and not exact.
  1916		// It works for all actual cases, though.
  1917		idx := strings.Index(s, "encoding=")
  1918		if idx == -1 {
  1919			return ""
  1920		}
  1921		v := s[idx+len("encoding="):]
  1922		if v == "" {
  1923			return ""
  1924		}
  1925		if v[0] != '\'' && v[0] != '"' {
  1926			return ""
  1927		}
  1928		idx = strings.IndexRune(v[1:], rune(v[0]))
  1929		if idx == -1 {
  1930			return ""
  1931		}
  1932		return v[1 : idx+1]
  1933	}

View as plain text