func (e Goose) Extract(link string) (data data.ArticleExtract, err error) {
defer func() {
if r := recover(); r != nil {
err = errors.New(fmt.Sprintf("%v", r))
}
}()
g := goose.New()
/* TODO: preserve links */
formatted, err := g.ExtractFromURL(link)
content := formatted.CleanedText
buf := util.BufferPool.GetBuffer()
defer util.BufferPool.Put(buf)
paragraphs := strings.Split(content, "\n")
var html []template.HTML
for _, p := range paragraphs {
if strings.TrimSpace(p) != "" {
html = append(html, template.HTML(p))
}
}
e.renderer.Render(buf,
renderer.RenderData{"paragraphs": html, "topImage": formatted.TopImage},
nil, "goose-format-result.tmpl")
data.Content = buf.String()
data.Title = formatted.Title
data.TopImage = formatted.TopImage
data.Language = formatted.MetaLang
return
}
func (this *Crawler) Analyze(url string) *goose.Article {
g := goose.New()
article, err := g.ExtractFromURL(url)
if err != nil {
// TODO Probably want to handle this error...
panic(err)
}
return article
}
func (this *Crawler) Analyze(url string) *goose.Article {
g := goose.New()
article := g.ExtractFromUrl(url)
return article
}