Example #1
0
func (b baiduNews) commonPrase(resp *context.Response) (infoStr string) {
	body := resp.GetDom().Find("body")

	var info *goquery.Selection

	if h1s := body.Find("h1"); len(h1s.Nodes) != 0 {
		for i := 0; i < len(h1s.Nodes); i++ {
			info = b.findP(h1s.Eq(i))
		}
	} else if h2s := body.Find("h2"); len(h2s.Nodes) != 0 {
		for i := 0; i < len(h2s.Nodes); i++ {
			info = b.findP(h2s.Eq(i))
		}
	} else if h3s := body.Find("h3"); len(h3s.Nodes) != 0 {
		for i := 0; i < len(h3s.Nodes); i++ {
			info = b.findP(h3s.Eq(i))
		}
	} else {
		info = body.Find("body")
	}
	// 去除标签
	// info.RemoveFiltered("script")
	// info.RemoveFiltered("style")
	infoStr, _ = info.Html()

	// 清洗HTML
	infoStr = CleanHtml(infoStr, 5)
	return
}