2016-12-24 17:09:28 +00:00
|
|
|
package telegraph
|
|
|
|
|
|
|
|
import (
|
|
|
|
"bytes"
|
2017-12-13 10:05:27 +00:00
|
|
|
"io"
|
2016-12-24 17:09:28 +00:00
|
|
|
"strings"
|
2017-01-10 10:28:23 +00:00
|
|
|
|
2017-10-03 13:46:11 +00:00
|
|
|
"golang.org/x/net/html"
|
2017-06-13 03:04:39 +00:00
|
|
|
)
|
2016-12-24 17:09:28 +00:00
|
|
|
|
2020-01-14 06:30:18 +00:00
|
|
|
// ContentFormat transforms data to a DOM-based format to represent the content of the page.
|
2018-07-27 11:51:14 +00:00
|
|
|
func ContentFormat(data interface{}) (n []Node, err error) {
|
2020-01-14 06:30:18 +00:00
|
|
|
var dst *html.Node
|
|
|
|
|
2018-07-27 11:51:14 +00:00
|
|
|
switch src := data.(type) {
|
2016-12-24 17:09:28 +00:00
|
|
|
case string:
|
2018-07-27 11:51:14 +00:00
|
|
|
dst, err = html.Parse(strings.NewReader(src))
|
2016-12-24 17:09:28 +00:00
|
|
|
case []byte:
|
2018-07-27 11:51:14 +00:00
|
|
|
dst, err = html.Parse(bytes.NewReader(src))
|
2017-12-13 10:05:27 +00:00
|
|
|
case io.Reader:
|
2018-07-27 11:51:14 +00:00
|
|
|
dst, err = html.Parse(src)
|
2016-12-24 17:09:28 +00:00
|
|
|
default:
|
2020-01-14 06:30:18 +00:00
|
|
|
return nil, ErrInvalidDataType
|
2019-07-24 08:03:29 +00:00
|
|
|
}
|
2020-01-14 06:30:18 +00:00
|
|
|
|
2019-07-24 08:03:29 +00:00
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
2016-12-24 17:09:28 +00:00
|
|
|
}
|
|
|
|
|
2018-07-27 11:51:14 +00:00
|
|
|
n = append(n, domToNode(dst.FirstChild))
|
2020-01-14 06:30:18 +00:00
|
|
|
|
|
|
|
return n, nil
|
2016-12-24 17:09:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
func domToNode(domNode *html.Node) interface{} {
|
|
|
|
if domNode.Type == html.TextNode {
|
|
|
|
return domNode.Data
|
|
|
|
}
|
|
|
|
|
|
|
|
if domNode.Type != html.ElementNode {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
2020-06-19 21:50:21 +00:00
|
|
|
nodeElement := new(NodeElement)
|
2020-01-14 06:30:18 +00:00
|
|
|
|
2017-10-03 13:46:11 +00:00
|
|
|
switch strings.ToLower(domNode.Data) {
|
2020-01-14 06:30:18 +00:00
|
|
|
case "a", "aside", "b", "blockquote", "br", "code", "em", "figcaption", "figure", "h3", "h4", "hr", "i",
|
|
|
|
"iframe", "img", "li", "ol", "p", "pre", "s", "strong", "u", "ul", "video":
|
2017-08-06 18:26:38 +00:00
|
|
|
nodeElement.Tag = domNode.Data
|
2017-09-04 21:09:59 +00:00
|
|
|
|
2017-10-03 13:46:11 +00:00
|
|
|
for i := range domNode.Attr {
|
|
|
|
switch strings.ToLower(domNode.Attr[i].Key) {
|
|
|
|
case "href", "src":
|
2020-01-14 06:30:18 +00:00
|
|
|
nodeElement.Attrs = map[string]string{domNode.Attr[i].Key: domNode.Attr[i].Val}
|
2017-10-03 13:46:11 +00:00
|
|
|
default:
|
|
|
|
continue
|
2016-12-24 17:09:28 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for child := domNode.FirstChild; child != nil; child = child.NextSibling {
|
|
|
|
nodeElement.Children = append(nodeElement.Children, domToNode(child))
|
|
|
|
}
|
|
|
|
|
|
|
|
return nodeElement
|
|
|
|
}
|