add_mailto_subject: New rewrite function

Dinosaur Comics (qwantz.com) likes to hide jokes in mailto: links, but
miniflux's sanitizer strips those out.
This commit is contained in:
Peter De Wachter 2019-08-13 17:44:23 +02:00 committed by Frédéric Guillot
parent 77125f45cc
commit b6f3160dbc
4 changed files with 45 additions and 0 deletions

View file

@ -7,6 +7,7 @@ package rewrite // import "miniflux.app/reader/rewrite"
import (
"fmt"
"html"
"net/url"
"regexp"
"strings"
@ -43,6 +44,38 @@ func addImageTitle(entryURL, entryContent string) string {
return entryContent
}
func addMailtoSubject(entryURL, entryContent string) string {
doc, err := goquery.NewDocumentFromReader(strings.NewReader(entryContent))
if err != nil {
return entryContent
}
matches := doc.Find(`a[href^="mailto:"]`)
if matches.Length() > 0 {
matches.Each(func(i int, a *goquery.Selection) {
hrefAttr, _ := a.Attr("href")
mailto, err := url.Parse(hrefAttr)
if err != nil {
return
}
subject := mailto.Query().Get("subject")
if subject == "" {
return
}
a.AppendHtml(" [" + html.EscapeString(subject) + "]")
})
output, _ := doc.Find("body").First().Html()
return output
}
return entryContent
}
func addDynamicImage(entryURL, entryContent string) string {
doc, err := goquery.NewDocumentFromReader(strings.NewReader(entryContent))
if err != nil {

View file

@ -27,6 +27,8 @@ func Rewriter(entryURL, entryContent, customRewriteRules string) string {
switch strings.TrimSpace(rule) {
case "add_image_title":
entryContent = addImageTitle(entryURL, entryContent)
case "add_mailto_subject":
entryContent = addMailtoSubject(entryURL, entryContent)
case "add_dynamic_image":
entryContent = addDynamicImage(entryURL, entryContent)
case "add_youtube_video":

View file

@ -98,6 +98,15 @@ func TestRewriteWithXkcdAndNoImage(t *testing.T) {
}
}
func TestRewriteMailtoLink(t *testing.T) {
description := `<a href="mailto:ryan@qwantz.com?subject=blah%20blah">contact</a>`
output := Rewriter("https://www.qwantz.com/", description, ``)
expected := `<a href="mailto:ryan@qwantz.com?subject=blah%20blah">contact [blah blah]</a>`
if expected != output {
t.Errorf(`Not expected output: got "%s" instead of "%s"`, output, expected)
}
}
func TestRewriteWithPDFLink(t *testing.T) {
description := "test"
output := Rewriter("https://example.org/document.pdf", description, ``)

View file

@ -22,6 +22,7 @@ var predefinedRules = map[string]string{
"oglaf.com": "add_image_title",
"optipess.com": "add_image_title",
"peebleslab.com": "add_image_title",
"www.qwantz.com": "add_image_title,add_mailto_subject",
"sentfromthemoon.com": "add_image_title",
"thedoghousediaries.com": "add_image_title",
"treelobsters.com": "add_image_title",