finder.go 2.6 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697
  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
  5. import (
  6. "bytes"
  7. "fmt"
  8. "io"
  9. "time"
  10. "github.com/miniflux/miniflux/errors"
  11. "github.com/miniflux/miniflux/http"
  12. "github.com/miniflux/miniflux/logger"
  13. "github.com/miniflux/miniflux/reader/feed"
  14. "github.com/miniflux/miniflux/timer"
  15. "github.com/miniflux/miniflux/url"
  16. "github.com/PuerkitoBio/goquery"
  17. )
  18. var (
  19. errConnectionFailure = "Unable to open this link: %v"
  20. errUnreadableDoc = "Unable to analyze this page: %v"
  21. )
  22. // FindSubscriptions downloads and try to find one or more subscriptions from an URL.
  23. func FindSubscriptions(websiteURL string) (Subscriptions, error) {
  24. defer timer.ExecutionTime(time.Now(), fmt.Sprintf("[FindSubscriptions] url=%s", websiteURL))
  25. client := http.NewClient(websiteURL)
  26. response, err := client.Get()
  27. if err != nil {
  28. return nil, errors.NewLocalizedError(errConnectionFailure, err)
  29. }
  30. var buffer bytes.Buffer
  31. io.Copy(&buffer, response.Body)
  32. reader := bytes.NewReader(buffer.Bytes())
  33. if format := feed.DetectFeedFormat(reader); format != feed.FormatUnknown {
  34. var subscriptions Subscriptions
  35. subscriptions = append(subscriptions, &Subscription{
  36. Title: response.EffectiveURL,
  37. URL: response.EffectiveURL,
  38. Type: format,
  39. })
  40. return subscriptions, nil
  41. }
  42. reader.Seek(0, io.SeekStart)
  43. return parseDocument(response.EffectiveURL, bytes.NewReader(buffer.Bytes()))
  44. }
  45. func parseDocument(websiteURL string, data io.Reader) (Subscriptions, error) {
  46. var subscriptions Subscriptions
  47. queries := map[string]string{
  48. "link[type='application/rss+xml']": "rss",
  49. "link[type='application/atom+xml']": "atom",
  50. "link[type='application/json']": "json",
  51. }
  52. doc, err := goquery.NewDocumentFromReader(data)
  53. if err != nil {
  54. return nil, errors.NewLocalizedError(errUnreadableDoc, err)
  55. }
  56. for query, kind := range queries {
  57. doc.Find(query).Each(func(i int, s *goquery.Selection) {
  58. subscription := new(Subscription)
  59. subscription.Type = kind
  60. if title, exists := s.Attr("title"); exists {
  61. subscription.Title = title
  62. } else {
  63. subscription.Title = "Feed"
  64. }
  65. if feedURL, exists := s.Attr("href"); exists {
  66. subscription.URL, _ = url.AbsoluteURL(websiteURL, feedURL)
  67. }
  68. if subscription.Title == "" {
  69. subscription.Title = subscription.URL
  70. }
  71. if subscription.URL != "" {
  72. logger.Debug("[FindSubscriptions] %s", subscription)
  73. subscriptions = append(subscriptions, subscription)
  74. }
  75. })
  76. }
  77. return subscriptions, nil
  78. }