twitter format

internal/twitter/format.go

@@ -2,30 +2,41 @@ package twitter
 
 import (
 	"fmt"
-	"regexp"
+	"strings"
 
+	text "github.com/MichaelMure/go-term-text"
+	"github.com/PuerkitoBio/goquery"
 	"jaytaylor.com/html2text"
 )
 
-func enrich(html string) (md string) {
+func Format(t *Embedded) (formatted string) {
+	formatted += fmt.Sprintf("URL: %s\n", t.Url)
+	formatted += "\n" + fitInScreen(plaintext(paragraph(t.Html)))
+	return
+}
+
+func fitInScreen(s string) string {
+	s, _ = text.WrapLeftPadded(s, 100, 3)
+	return s
+}
+
+func plaintext(html string) string {
 	md, err := html2text.FromString(html)
 	if err != nil {
 		panic(err)
 	}
-	return
+	return md
 }
 
-func extract(html string) (p string) {
-	r, err := regexp.Compile("<p.*>(.*)</p>")
+func paragraph(html string) string {
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
 	if err != nil {
 		panic(err)
 	}
-	return r.FindStringSubmatch(html)[0]
-}
-
-func Format(t *Embedded) (formatted string) {
-	formatted += fmt.Sprintf("URL: %s\n", t.Url)
-	formatted += enrich(extract(t.Html))
-	formatted += fmt.Sprintf("> %s\n", t.AuthorName)
-	return
+	doc.Find("p a").Remove()
+	msg, err := doc.Find("p").Html()
+	if err != nil {
+		panic(err)
+	}
+	return msg
 }