finder.go 2.3 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788
  1. // Copyright 2017 Frédéric Guillot. All rights reserved.
  2. // Use of this source code is governed by the Apache 2.0
  3. // license that can be found in the LICENSE file.
  4. package subscription // import "miniflux.app/reader/subscription"
  5. import (
  6. "io"
  7. "strings"
  8. "miniflux.app/errors"
  9. "miniflux.app/http/client"
  10. "miniflux.app/reader/browser"
  11. "miniflux.app/reader/parser"
  12. "miniflux.app/url"
  13. "github.com/PuerkitoBio/goquery"
  14. )
  15. var (
  16. errUnreadableDoc = "Unable to analyze this page: %v"
  17. )
  18. // FindSubscriptions downloads and try to find one or more subscriptions from an URL.
  19. func FindSubscriptions(websiteURL, userAgent, username, password string) (Subscriptions, *errors.LocalizedError) {
  20. request := client.New(websiteURL)
  21. request.WithCredentials(username, password)
  22. request.WithUserAgent(userAgent)
  23. response, err := browser.Exec(request)
  24. if err != nil {
  25. return nil, err
  26. }
  27. body := response.String()
  28. if format := parser.DetectFeedFormat(body); format != parser.FormatUnknown {
  29. var subscriptions Subscriptions
  30. subscriptions = append(subscriptions, &Subscription{
  31. Title: response.EffectiveURL,
  32. URL: response.EffectiveURL,
  33. Type: format,
  34. })
  35. return subscriptions, nil
  36. }
  37. return parseDocument(response.EffectiveURL, strings.NewReader(body))
  38. }
  39. func parseDocument(websiteURL string, data io.Reader) (Subscriptions, *errors.LocalizedError) {
  40. var subscriptions Subscriptions
  41. queries := map[string]string{
  42. "link[type='application/rss+xml']": "rss",
  43. "link[type='application/atom+xml']": "atom",
  44. "link[type='application/json']": "json",
  45. }
  46. doc, err := goquery.NewDocumentFromReader(data)
  47. if err != nil {
  48. return nil, errors.NewLocalizedError(errUnreadableDoc, err)
  49. }
  50. for query, kind := range queries {
  51. doc.Find(query).Each(func(i int, s *goquery.Selection) {
  52. subscription := new(Subscription)
  53. subscription.Type = kind
  54. if title, exists := s.Attr("title"); exists {
  55. subscription.Title = title
  56. } else {
  57. subscription.Title = "Feed"
  58. }
  59. if feedURL, exists := s.Attr("href"); exists {
  60. subscription.URL, _ = url.AbsoluteURL(websiteURL, feedURL)
  61. }
  62. if subscription.Title == "" {
  63. subscription.Title = subscription.URL
  64. }
  65. if subscription.URL != "" {
  66. subscriptions = append(subscriptions, subscription)
  67. }
  68. })
  69. }
  70. return subscriptions, nil
  71. }