xml: allow unquoted attribute values in non-Strict mode

HTML4 standard supports unquoted attibute values in certain cases
(http://www.w3.org/TR/REC-html40/intro/sgmltut.html#h-3.2.2).

R=rsc
CC=golang-dev
https://golang.org/cl/207095

diff --git a/src/pkg/xml/xml.go b/src/pkg/xml/xml.go
index 33a86a2557c4a0b01fcd539a8cd7b6a96b5a8e97..67cbb824f7c57c588d5ec2ec242972b14b5713c3 100644 (file)
--- a/src/pkg/xml/xml.go
+++ b/src/pkg/xml/xml.go
@@ -589,14 +589,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                        return nil, p.err
                }
                p.space()
-               if b, ok = p.mustgetc(); !ok {
-                       return nil, p.err
-               }
-               if b != '"' && b != '\'' {
-                       p.err = SyntaxError("unquoted or missing attribute value in element")
-                       return nil, p.err
-               }
-               data := p.text(int(b), false)
+               data := p.attrval()
                if data == nil {
                        return nil, p.err
                }
@@ -610,6 +603,40 @@ func (p *Parser) RawToken() (Token, os.Error) {
        return StartElement{name, attr}, nil
 }
 
+func (p *Parser) attrval() []byte {
+       b, ok := p.mustgetc()
+       if !ok {
+               return nil
+       }
+       // Handle quoted attribute values
+       if b == '"' || b == '\'' {
+               return p.text(int(b), false)
+       }
+       // Handle unquoted attribute values for strict parsers
+       if p.Strict {
+               p.err = SyntaxError("unquoted or missing attribute value in element")
+               return nil
+       }
+       // Handle unquoted attribute values for unstrict parsers
+       p.ungetc(b)
+       p.buf.Reset()
+       for {
+               b, ok = p.mustgetc()
+               if !ok {
+                       return nil
+               }
+               // http://www.w3.org/TR/REC-html40/intro/sgmltut.html#h-3.2.2
+               if 'a' <= b && b <= 'z' || 'A' <= b && b <= 'Z' ||
+                       '0' <= b && b <= '9' || b == '_' || b == ':' || b == '-' {
+                       p.buf.WriteByte(b)
+               } else {
+                       p.ungetc(b)
+                       break
+               }
+       }
+       return p.buf.Bytes()
+}
+
 // Skip spaces if any
 func (p *Parser) space() {
        for {

diff --git a/src/pkg/xml/xml_test.go b/src/pkg/xml/xml_test.go
index fa19495001ea11d646e06e6fa72db77cd60fab4c..a4c55b73dca2288b0a17965d7b0e56702a8e36de 100644 (file)
--- a/src/pkg/xml/xml_test.go
+++ b/src/pkg/xml/xml_test.go
@@ -298,3 +298,23 @@ func TestIssue569(t *testing.T) {
                t.Fatalf("Expecting abcd")
        }
 }
+
+func TestUnquotedAttrs(t *testing.T) {
+       data := "<tag attr=azAZ09:-_\t>"
+       p := NewParser(StringReader(data))
+       p.Strict = false
+       token, err := p.Token()
+       if _, ok := err.(SyntaxError); ok {
+               t.Errorf("Unexpected error: %v", err)
+       }
+       if token.(StartElement).Name.Local != "tag" {
+               t.Errorf("Unexpected tag name: %v", token.(StartElement).Name.Local)
+       }
+       attr := token.(StartElement).Attr[0]
+       if attr.Value != "azAZ09:-_" {
+               t.Errorf("Unexpected attribute value: %v", attr.Value)
+       }
+       if attr.Name.Local != "attr" {
+               t.Errorf("Unexpected attribute name: %v", attr.Name.Local)
+       }
+}