func node_clear_children(frag *html.Node) {
for child := frag.FirstChild; child != nil; {
next := child.NextSibling
frag.RemoveChild(child)
child = next
}
}
func clean_element_before_header(body *html.Node, name string) {
child := body.FirstChild
for child != nil {
if child.Type == html.ElementNode && child.Data != name {
next := child.NextSibling
body.RemoveChild(child)
child = next
} else {
break
}
}
}
func remove_decentant(n *html.Node, tag string) {
child := n.FirstChild
for child != nil {
if child.Type == html.ElementNode && child.Data == tag {
next := child.NextSibling
n.RemoveChild(child)
child = next
} else {
remove_decentant(child, tag)
child = child.NextSibling
}
}
}
// flatten inlines text image a object video audio seq
// n is element-node
// inline node may have div element
func (this *html_cleaner) clean_inline_node(n *html.Node) {
inlines := this.flatten_inline_node(n)
for child := n.FirstChild; child != nil; child = n.FirstChild {
n.RemoveChild(child)
}
for _, inline := range inlines {
p := inline.Parent
if p != nil {
p.RemoveChild(inline) // this.article.RemoveChild(child)
}
n.AppendChild(inline)
}
}
//删除行前后空白
func (this *html_cleaner) trim_empty_spaces_func(n *html.Node, trim func(string) string) {
child := n.FirstChild
for child != nil {
if child.Type == html.TextNode {
child.Data = trim(child.Data)
} else {
this.trim_empty_spaces_func(child, trim)
}
if node_is_not_empty(child) {
break
}
next := child.NextSibling
n.RemoveChild(child)
child = next
}
}
// The splitting process may leave TextNodes with no Data, which we keep
// around to make the data manipulation simpler. This function removes
// them.
func cleanupTree(node *html.Node) {
var next *html.Node
for n := node.FirstChild; n != nil; n = next {
next = n.NextSibling
switch n.Type {
case html.TextNode:
if len(n.Data) == 0 {
node.RemoveChild(n)
}
case html.ElementNode:
cleanupTree(n)
default:
// ignore other node types.
}
}
}
func processNode(node *html.Node) (err error) {
var stackTags [16]openTag
tags := stackTags[:0]
n := node.FirstChild
for n != nil {
var next, newParent *html.Node
next = n.NextSibling
if l := len(tags); l != 0 {
newParent = tags[l-1].node
}
switch n.Type {
case html.TextNode:
if tags, next, err = processTextNode(n, tags); err != nil {
return
}
case html.ElementNode:
if err = processNode(n); err != nil {
return
}
default:
// Other node types are just ignored.
}
// reparent the active node if necessary
if newParent != nil {
node.RemoveChild(n)
newParent.AppendChild(n)
}
n = next
}
if len(tags) != 0 {
err = fmt.Errorf("shortcodes still open at end of surrounding HTML tag: %+v", tags)
}
return
}
// sanitizeUnwrap traverses pre-order over the nodes, reattaching
// the whitelisted children of any element nodes that are not
// whitelisted to the parent of the unwhitelisted node
func (w *Whitelist) sanitizeUnwrap(n *html.Node) error {
return w.sanitizeNode(n, func(n *html.Node) bool {
if w.HasElement(n.Data) || n.Parent == nil {
return true
}
insertBefore := n.NextSibling
firstChild := n.FirstChild
for c := n.FirstChild; c != nil; {
nodeToUnwrap := c
c = c.NextSibling
n.RemoveChild(nodeToUnwrap)
n.Parent.InsertBefore(nodeToUnwrap, insertBefore)
}
n.Parent.RemoveChild(n)
// reset next sibling to support continuation
// of linked-list style traversal of parent node's children
n.NextSibling = firstChild
return false
})
}
//整理html文档，将block-level/inline-level混合的节点改成只有block-level的节点
//对已只有inline-level的节点，删除行前后的空白符
//将包含inline-level的节点展开成更为简单的形式，去掉想<font><span><strong>等等格式节点
func (this *html_cleaner) clean_block_node(n *html.Node) {
blks := node_has_block_children(n)
inlines := node_has_inline_children(n)
// has bocks and inlines
if blks && inlines {
child := n.FirstChild
for child != nil {
if node_is_inline(child) {
p := child.PrevSibling
if p == nil || p.Data != "p" {
p = create_element("p")
n.InsertBefore(p, child)
}
n.RemoveChild(child)
p.AppendChild(child)
child = p.NextSibling
} else {
child = child.NextSibling
}
}
inlines = false
}
// only inlines
if blks == false && inlines {
this.clean_inline_node(n)
this.trim_empty_spaces(n)
}
// only blocks
if blks && !inlines {
for child := n.FirstChild; child != nil; child = child.NextSibling {
this.clean_block_node(child)
}
}
}
func removeChildren(n *html.Node) {
for c := n.FirstChild; c != nil; c = c.NextSibling {
defer n.RemoveChild(c)
}
}
func remove_children(a *html.Node) {
for a.FirstChild != nil {
a.RemoveChild(a.FirstChild)
}
}