// Languages returns the given and guessed languages
// found in abstract and fulltext. Note: This is slow.
// Skip detection on too short strings.
func (article *Article) Languages() []string {
set := container.NewStringSet()
if article.Front.Article.Abstract.Lang != "" {
base, err := language.ParseBase(article.Front.Article.Abstract.Lang)
if err == nil {
set.Add(base.ISO3())
}
}
vals := []string{
article.Front.Article.Abstract.Value,
article.Front.Article.TranslatedAbstract.Title.Value,
article.Body.Section.Value,
}
for _, s := range vals {
if len(s) < 20 {
continue
}
lang, err := span.DetectLang3(s)
if err != nil || lang == "und" {
continue
}
if !acceptedLanguages.Contains(lang) {
continue
}
set.Add(lang)
}
return set.Values()
}
// Languages returns the given and guessed languages found in abstract and
// fulltext. Note: This is slow. Skip detection on too short strings.
func (doc Document) Languages() []string {
set := container.NewStringSet()
vals := []string{doc.Title, doc.Text}
for _, s := range vals {
if len(s) < 20 {
continue
}
lang, err := span.DetectLang3(s)
if err != nil {
continue
}
if !acceptedLanguages.Contains(lang) {
continue
}
if lang == "und" {
continue
}
set.Add(lang)
}
return set.Values()
}
