func recurse(p *xml.Parser, name string, m TagMap) os.Error {
// depth++ // debugging
for {
// if depth > maxDepth { // debugging
// return nil
// }
tok, err := p.Token()
if err != nil {
return err
}
switch t := tok.(type) {
case xml.StartElement:
if m[t.Name.Local] == nil {
m[t.Name.Local] = TagMap{}
}
err = recurse(p, t.Name.Local, m[t.Name.Local])
if err != nil {
return err
}
case xml.EndElement:
// If ending the element we entered recurse for, return
if t.Name.Local == name {
return nil
}
}
}
return nil
}
// Scan XML token stream to find next StartElement.
func nextStart(p *xml.Parser) (xml.StartElement, os.Error) {
for {
t, err := p.Token()
if err != nil {
log.Fatal("token", err)
}
switch t := t.(type) {
case xml.StartElement:
return t, nil
}
}
panic("unreachable")
}
func getTagContents(parser *xml.Parser) string {
tag, err := parser.Token()
if err != nil {
return ""
}
switch dtype := tag.(type) {
case xml.CharData:
bytes := xml.CharData(dtype)
text := string([]byte(bytes))
return text
}
return ""
}
func start(p *xml.Parser, name string, m TagMap) os.Error {
for {
// if depth > maxDepth { // debugging
// return nil
// }
tok, err := p.Token()
if err != nil {
return err
}
switch t := tok.(type) {
case xml.StartElement:
if t.Name.Local == name {
if _, found := m[name]; !found {
m[name] = TagMap{}
}
err = recurse(p, name, m[name])
}
}
if err != nil {
return err
}
}
return nil
}
func parse(tok xml.Token, parser *xml.Parser, parent *Node) (node *Node, err os.Error) {
node = new(Node)
node.Parent = parent
st, _ := tok.(xml.StartElement)
node.Name = st.Name.Local
trace("parsing tag %s", node.Name)
node.Attr = []html.Attribute{}
for _, attr := range st.Attr {
a := html.Attribute{Key: attr.Name.Local, Val: attr.Value}
node.Attr = append(node.Attr, a)
}
// var childs vector.Vector
// var chld []*Node
for done := false; !done; {
var tok xml.Token
tok, err = parser.Token()
if err != nil {
if err == os.EOF {
err = nil
break
}
if node.Name == "script" {
err = os.ErrorString("Javascript: " + err.String())
}
return
}
switch t := tok.(type) {
case xml.StartElement:
var ch *Node
ch, err = parse(t, parser, node)
if err != nil {
return
}
node.Child = append(node.Child, ch)
node.subs = append(node.subs, ch)
if node.Full != "" {
node.Full += " "
}
node.Full += ch.Full
case xml.EndElement:
if t.Name.Local != node.Name {
fmt.Printf("Tag " + node.Name + " closed by " + t.Name.Local + "\n")
}
done = true
case xml.CharData:
b := bytes.NewBuffer([]byte(t))
s := b.String()
ct := " " + cleanText(s)
node.Text += ct
node.Full += ct
node.subs = append(node.subs, &Node{Parent: node, Name: TEXT_NODE, Text: s})
case xml.Comment, xml.Directive, xml.ProcInst:
// skip
default:
fmt.Printf("Very strange:\nType = %t\n Value = %#v\n", tok, tok)
}
}
node.Text = strings.Trim(node.Text, " \n\t\r")
node.Full = cleanText(node.Full)
prepareClasses(node)
trace("Made Node: " + node.String() + "\n")
return
}
