func itemURL(s *goquery.Selection) (url string) {
s.Closest("tr").Next().Find("a").Each(func(_ int, s *goquery.Selection) {
href, _ := s.Attr("href")
if strings.HasPrefix(href, "item?id=") {
url = hnURL + href
}
})
return
}
func guessImgEnvironment(s *goquery.Selection) *goquery.Selection {
possibleEnvs := []string{
"figure",
".inline-image", // bbc
".inline-media__unlinked-image", // bloomberg
"header .inline-media", // bloomberg
"bigimgWrap", // bloomberg
".story-img", // technology review
".img-left", // technology review
".article-img", // technology review
".img-wrap", // the intercept
".overlap-img", // nautilus
".inline-content", // washington post
".image", // npr.org
".centered-figure-container", // npr.org
".slate-image", // slate.com
".wp-caption", // mcntyr.com and quantamagazine.com
".article__image", // aeon
}
var possibleParents []*goquery.Selection
for _, env := range possibleEnvs {
parent := s.Closest(env)
if parent.Nodes != nil {
Debug("Found img env: %v", env)
possibleParents = append(possibleParents, parent)
}
}
if len(possibleParents) == 0 {
Debug("No parents found")
return s
}
var htmlPossibleParents []string
for _, p := range possibleParents {
h, _ := p.Html()
htmlPossibleParents = append(htmlPossibleParents, h)
}
indexLongestElement, _ := longestElement(htmlPossibleParents)
Debug("longest Env: %v", htmlPossibleParents[indexLongestElement])
return possibleParents[indexLongestElement]
}
func parentAttr(s *goquery.Selection, attr string) string {
sel := "[" + attr + "]"
res, _ := s.Closest(sel).Attr(attr)
return res
}
