func (v *V2exCollector) CollectPage(uri string) (ret []Entry, err error) {
resp, err := Get(uri)
if err != nil {
return nil, v.Err("get %s error: %v", uri, err)
}
defer resp.Body.Close()
root, err := nw.Parse(resp.Body)
if err != nil {
return nil, v.Err("parse html %s: %v", uri, err)
}
var walkError error
root.Walk(nw.Css("div.cell span.item_title a", func(n *nw.Node) {
id, err := strconv.Atoi(v2exPidPattern.FindStringSubmatch(n.Attr["href"])[1])
if err != nil {
walkError = v.Err("no post id: %s", uri)
return
}
ret = append(ret, &V2exEntry{
Id: id,
Title: n.Text,
})
}))
if walkError != nil {
err = walkError
}
return
}
func (b *BilibiliCollector) CollectNewest(urlPattern string, page int) (ret []Entry, err error) {
// get content
url := s(urlPattern, page)
resp, err := GetWithCookie(url, b.cookie)
if err != nil {
return nil, b.Err("get newest page %s %v", url, err)
}
defer resp.Body.Close()
root, err := nw.Parse(resp.Body)
if err != nil {
return nil, b.Err("parse html %v", err)
}
var link, title, image string
var id int
var walkErr error
root.Walk(nw.Css("ul.vd_list li", nw.Multi(
nw.Css("a.title", func(n *nw.Node) {
link = "http://www.bilibili.com" + n.Attr["href"]
title = n.Text
}),
nw.Css("a.preview img", func(n *nw.Node) {
image = n.Attr["src"]
}),
func(node *nw.Node) {
id, err = strconv.Atoi(regexp.MustCompile(`av([0-9]+)`).FindStringSubmatch(link)[1])
if err != nil {
walkErr = b.Err("link without av id %s at %s", link, url)
return
}
ret = append(ret, &BilibiliEntry{
Id: id,
Link: link,
Title: title,
Image: image,
})
},
)))
if walkErr != nil {
return nil, walkErr
}
return
}
