func flattenSubtreeV3Inner(n, nClone *html.Node, lvl int) {
// log.Printf("fsbi\n")
for ch := n.FirstChild; ch != nil; ch = ch.NextSibling {
chClone := dom.CloneNode(ch)
switch {
case ch.Type == html.ElementNode && standard[ch.Data]:
nClone.AppendChild(chClone)
flattenSubtreeV3Inner(ch, chClone, lvl+1)
case ch.Type == html.ElementNode && ch.Data == "a":
nClone.AppendChild(chClone)
flattenSubtreeV3Inner(ch, chClone, lvl+1)
case ch.Type == html.ElementNode && ch.Data == "img":
nClone.AppendChild(chClone)
case ch.Data == "span":
// log.Printf(strings.Repeat(" ", lvl) + "span \n")
for cch := ch.FirstChild; cch != nil; cch = cch.NextSibling {
// log.Printf(strings.Repeat(" ", lvl)+"span child %v", cch.Data)
cchClone := dom.CloneNode(cch)
nClone.AppendChild(cchClone)
nClone.AppendChild(dom.Nd("text", " "))
flattenSubtreeV3Inner(cch, cchClone, lvl+1)
}
case ch.Type == html.TextNode && ch.Data != "":
chClone.Data = strings.TrimSpace(chClone.Data)
chClone.Data += " "
nClone.AppendChild(chClone)
default:
// nClone.AppendChild(chClone)
log.Printf("unhandled %s %s\n", dom.NodeTypeStr(ch.Type), ch.Data)
}
}
}
func flattenSubtreeV2(n *html.Node, b *bytes.Buffer, depth int, tpar *html.Node) (*bytes.Buffer, *html.Node) {
if b == nil {
b = new(bytes.Buffer)
}
if tpar == nil {
tpar = &html.Node{
Type: n.Type,
DataAtom: n.DataAtom,
Data: n.Data,
Attr: make([]html.Attribute, len(n.Attr)),
}
copy(tpar.Attr, n.Attr)
}
switch {
case n.Type == html.ElementNode && n.Data == "a":
n.Parent.RemoveChild(n)
tpar.AppendChild(n)
// wpf(b, "[a] ")
case n.Type == html.ElementNode && n.Data == "img":
// img2Link(n)
n.Parent.RemoveChild(n)
tpar.AppendChild(n)
case n.Data == "em" || n.Data == "strong":
wpf(b, "[%v l%v] ", n.Data, depth)
n.Parent.RemoveChild(n)
tpar.AppendChild(n)
case n.Data == "label" || n.Data == "input" || n.Data == "textarea":
n.Parent.RemoveChild(n)
tpar.AppendChild(n)
case n.Data == "p" || n.Data == "div" || n.Data == "li" || n.Data == "ol" || n.Data == "h1" || n.Data == "h2" || n.Data == "ul":
n.Parent.RemoveChild(n)
tpar.AppendChild(n)
case n.Data == "span":
for c := n.FirstChild; c != nil; c = c.NextSibling {
n.RemoveChild(c)
tpar.AppendChild(c)
}
n.Parent.RemoveChild(n)
case n.Type == html.TextNode && n.Data != "":
n.Data = strings.TrimSpace(n.Data)
n.Data += " "
wpf(b, n.Data)
n.Parent.RemoveChild(n)
tpar.AppendChild(n)
default:
log.Printf("unhandled %s %s\n", dom.NodeTypeStr(n.Type), n.Data)
n.Parent.RemoveChild(n)
}
//
//
children := []*html.Node{}
for c := n.FirstChild; c != nil; c = c.NextSibling {
// fmt.Printf("still has children %v\n", c.Data)
children = append(children, c) // assembling separately, before removing.
}
for _, c := range children {
flattenSubtreeV2(c, b, depth+1, tpar)
}
return b, tpar
}