sync of changes

pkg/extractor/goose.go

@@ -0,0 +1,25 @@
+package extractor
+
+import (
+	"context"
+	goose "github.com/advancedlogic/GoOse"
+)
+
+type GooseExtractor struct {
+}
+
+func (GooseExtractor) Extract(ctx context.Context, url string) (Article, error) {
+	var res = Article{
+		URL: url,
+	}
+	g := goose.New()
+
+	article, err := g.ExtractFromURL(url)
+	if err != nil {
+		return res, err
+	}
+
+	res.Body = article.CleanedText
+	res.Title = article.Title
+	return res, nil
+}