func main() {
x, _ := goquery.Parse(example)
x.Find("a div").Val("lol")
fmt.Println(x.Html())
fmt.Println(x.Find("div").HasClass("yo"))
fmt.Println(x.Find("").Attrs("href"))
}
func getMostSimilarAltImage(bodyStr string, i *ItemObject) {
nodes, err := goquery.Parse(bodyStr)
if err != nil {
fmt.Println(err)
return
}
nodes = nodes.Find("img")
maxScore := 2
image := ""
for _, node := range nodes {
src, alt, width, _ := Attr(node)
if !strings.HasPrefix(src, "http") {
continue
}
score := GetSimilarityScore(i.Title, alt)
if score > maxScore {
image = src
maxScore = width
}
}
if image != "" {
i.ParsedImage = image
}
}
func getWidestImage(bodyStr string, i *ItemObject) {
nodes, err := goquery.Parse(bodyStr)
if err != nil {
fmt.Println(err)
return
}
nodes = nodes.Find("img")
maxWidth := 200
maxAlt := 100
image := ""
for _, node := range nodes {
src, alt, width, _ := Attr(node)
if !strings.HasPrefix(src, "http") {
continue
}
if width > maxWidth && isImageGood(src) {
image = src
maxWidth = width
}
if image == "" && len(alt) > maxAlt && width != 1 && isImageGood(src) {
image = src
maxAlt = len(alt)
}
}
if image != "" {
i.ParsedImage = image
}
}
func getImageFromClass(body, class string, i *ItemObject) {
nodes, err := goquery.Parse(body)
if err != nil {
fmt.Println(err)
return
}
nodes = nodes.Find(class)
getWidestImage(nodes.Html(), i)
}
func getSvg(svgBytes []byte) string {
svg := bytes.NewReader(svgBytes)
doc, err := goquery.Parse(svg)
check(err)
icon := doc.Find("symbol").OuterHtml()
if icon == "" {
logAndExit("I haven't found any <symbol> in the svg")
}
return icon
}
func getLinksFromBody(body io.Reader, host string) []string {
tree, _ := goquery.Parse(body)
links := tree.Find("a").Attrs("href")
normals := make([]string, 0, len(links))
for _, link := range links {
link, ok := normilizeLink(link, host)
if ok {
normals = append(normals, link)
}
}
return normals
}
func getOGImage(body string, i *ItemObject) {
nodes, err := goquery.Parse(body)
if err != nil {
fmt.Println(err)
return
}
nodes = nodes.Find("meta")
for _, node := range nodes {
property, content := MetaAttr(node)
if property == "og:image" && isImageGood(content) {
i.ParsedImage = content
return
}
}
}
func (ps *Parser) parseHtml() []string {
q, _ := goquery.Parse(strings.NewReader(ps.FileContent.Content))
ls := []string{}
var nodes goquery.Nodes
// 链接
nodes = q.Find("a")
ls = append(ls, ps.getAttr(nodes, "href", nil)...)
//css
nodes = q.Find("link")
ls = append(ls, ps.getAttr(nodes, "href", map[string]string{"type": "text/css"})...)
//js
nodes = q.Find("script")
ls = append(ls, ps.getAttr(nodes, "src", nil)...)
//图片
nodes = q.Find("img")
ls = append(ls, ps.getAttr(nodes, "src", nil)...)
return ls
}
func main() {
x, _ := goquery.Parse(example)
x.Find("a div").Print()
fmt.Println("---")
x.Find("a div.cow").Print()
}
