rdf.go 2.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119
  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 rdf // import "miniflux.app/reader/rdf"
  5. import (
  6. "encoding/xml"
  7. "strings"
  8. "time"
  9. "miniflux.app/crypto"
  10. "miniflux.app/logger"
  11. "miniflux.app/model"
  12. "miniflux.app/reader/date"
  13. "miniflux.app/reader/sanitizer"
  14. "miniflux.app/url"
  15. )
  16. type rdfFeed struct {
  17. XMLName xml.Name `xml:"RDF"`
  18. Title string `xml:"channel>title"`
  19. Link string `xml:"channel>link"`
  20. Items []rdfItem `xml:"item"`
  21. DublinCoreFeedElement
  22. }
  23. func (r *rdfFeed) Transform(baseURL string) *model.Feed {
  24. var err error
  25. feed := new(model.Feed)
  26. feed.Title = sanitizer.StripTags(r.Title)
  27. feed.FeedURL = baseURL
  28. feed.SiteURL, err = url.AbsoluteURL(baseURL, r.Link)
  29. if err != nil {
  30. feed.SiteURL = r.Link
  31. }
  32. for _, item := range r.Items {
  33. entry := item.Transform()
  34. if entry.Author == "" && r.DublinCoreCreator != "" {
  35. entry.Author = strings.TrimSpace(r.DublinCoreCreator)
  36. }
  37. if entry.URL == "" {
  38. entry.URL = feed.SiteURL
  39. } else {
  40. entryURL, err := url.AbsoluteURL(feed.SiteURL, entry.URL)
  41. if err == nil {
  42. entry.URL = entryURL
  43. }
  44. }
  45. feed.Entries = append(feed.Entries, entry)
  46. }
  47. return feed
  48. }
  49. type rdfItem struct {
  50. Title string `xml:"title"`
  51. Link string `xml:"link"`
  52. Description string `xml:"description"`
  53. DublinCoreEntryElement
  54. }
  55. func (r *rdfItem) Transform() *model.Entry {
  56. entry := new(model.Entry)
  57. entry.Title = r.entryTitle()
  58. entry.Author = r.entryAuthor()
  59. entry.URL = r.entryURL()
  60. entry.Content = r.entryContent()
  61. entry.Hash = r.entryHash()
  62. entry.Date = r.entryDate()
  63. return entry
  64. }
  65. func (r *rdfItem) entryTitle() string {
  66. return strings.TrimSpace(r.Title)
  67. }
  68. func (r *rdfItem) entryContent() string {
  69. switch {
  70. case r.DublinCoreContent != "":
  71. return r.DublinCoreContent
  72. default:
  73. return r.Description
  74. }
  75. }
  76. func (r *rdfItem) entryAuthor() string {
  77. return strings.TrimSpace(r.DublinCoreCreator)
  78. }
  79. func (r *rdfItem) entryURL() string {
  80. return strings.TrimSpace(r.Link)
  81. }
  82. func (r *rdfItem) entryDate() time.Time {
  83. if r.DublinCoreDate != "" {
  84. result, err := date.Parse(r.DublinCoreDate)
  85. if err != nil {
  86. logger.Error("rdf: %v (entry link = %s)", err, r.Link)
  87. return time.Now()
  88. }
  89. return result
  90. }
  91. return time.Now()
  92. }
  93. func (r *rdfItem) entryHash() string {
  94. value := r.Link
  95. if value == "" {
  96. value = r.Title + r.Description
  97. }
  98. return crypto.Hash(value)
  99. }