atom.go 4.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192
  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 atom
  5. import (
  6. "encoding/xml"
  7. "log"
  8. "strconv"
  9. "strings"
  10. "time"
  11. "github.com/miniflux/miniflux2/helper"
  12. "github.com/miniflux/miniflux2/model"
  13. "github.com/miniflux/miniflux2/reader/date"
  14. "github.com/miniflux/miniflux2/reader/processor"
  15. )
  16. type atomFeed struct {
  17. XMLName xml.Name `xml:"http://www.w3.org/2005/Atom feed"`
  18. ID string `xml:"id"`
  19. Title string `xml:"title"`
  20. Author atomAuthor `xml:"author"`
  21. Links []atomLink `xml:"link"`
  22. Entries []atomEntry `xml:"entry"`
  23. }
  24. type atomEntry struct {
  25. ID string `xml:"id"`
  26. Title string `xml:"title"`
  27. Updated string `xml:"updated"`
  28. Links []atomLink `xml:"link"`
  29. Summary string `xml:"summary"`
  30. Content atomContent `xml:"content"`
  31. MediaGroup atomMediaGroup `xml:"http://search.yahoo.com/mrss/ group"`
  32. Author atomAuthor `xml:"author"`
  33. }
  34. type atomAuthor struct {
  35. Name string `xml:"name"`
  36. Email string `xml:"email"`
  37. }
  38. type atomLink struct {
  39. URL string `xml:"href,attr"`
  40. Type string `xml:"type,attr"`
  41. Rel string `xml:"rel,attr"`
  42. Length string `xml:"length,attr"`
  43. }
  44. type atomContent struct {
  45. Type string `xml:"type,attr"`
  46. Data string `xml:",chardata"`
  47. XML string `xml:",innerxml"`
  48. }
  49. type atomMediaGroup struct {
  50. Description string `xml:"http://search.yahoo.com/mrss/ description"`
  51. }
  52. func (a *atomFeed) Transform() *model.Feed {
  53. feed := new(model.Feed)
  54. feed.FeedURL = getRelationURL(a.Links, "self")
  55. feed.SiteURL = getURL(a.Links)
  56. feed.Title = strings.TrimSpace(a.Title)
  57. if feed.Title == "" {
  58. feed.Title = feed.SiteURL
  59. }
  60. for _, entry := range a.Entries {
  61. item := entry.Transform()
  62. if item.Author == "" {
  63. item.Author = getAuthor(a.Author)
  64. }
  65. feed.Entries = append(feed.Entries, item)
  66. }
  67. return feed
  68. }
  69. func (a *atomEntry) Transform() *model.Entry {
  70. entry := new(model.Entry)
  71. entry.URL = getURL(a.Links)
  72. entry.Date = getDate(a)
  73. entry.Author = getAuthor(a.Author)
  74. entry.Hash = getHash(a)
  75. entry.Content = processor.ItemContentProcessor(entry.URL, getContent(a))
  76. entry.Title = strings.TrimSpace(a.Title)
  77. entry.Enclosures = getEnclosures(a)
  78. if entry.Title == "" {
  79. entry.Title = entry.URL
  80. }
  81. return entry
  82. }
  83. func getURL(links []atomLink) string {
  84. for _, link := range links {
  85. if strings.ToLower(link.Rel) == "alternate" {
  86. return strings.TrimSpace(link.URL)
  87. }
  88. if link.Rel == "" && link.Type == "" {
  89. return strings.TrimSpace(link.URL)
  90. }
  91. }
  92. return ""
  93. }
  94. func getRelationURL(links []atomLink, relation string) string {
  95. for _, link := range links {
  96. if strings.ToLower(link.Rel) == relation {
  97. return strings.TrimSpace(link.URL)
  98. }
  99. }
  100. return ""
  101. }
  102. func getDate(a *atomEntry) time.Time {
  103. if a.Updated != "" {
  104. result, err := date.Parse(a.Updated)
  105. if err != nil {
  106. log.Println(err)
  107. return time.Now()
  108. }
  109. return result
  110. }
  111. return time.Now()
  112. }
  113. func getContent(a *atomEntry) string {
  114. if a.Content.Type == "html" || a.Content.Type == "text" {
  115. return a.Content.Data
  116. }
  117. if a.Content.Type == "xhtml" {
  118. return a.Content.XML
  119. }
  120. if a.Summary != "" {
  121. return a.Summary
  122. }
  123. if a.MediaGroup.Description != "" {
  124. return a.MediaGroup.Description
  125. }
  126. return ""
  127. }
  128. func getHash(a *atomEntry) string {
  129. for _, value := range []string{a.ID, getURL(a.Links)} {
  130. if value != "" {
  131. return helper.Hash(value)
  132. }
  133. }
  134. return ""
  135. }
  136. func getEnclosures(a *atomEntry) model.EnclosureList {
  137. enclosures := make(model.EnclosureList, 0)
  138. for _, link := range a.Links {
  139. if strings.ToLower(link.Rel) == "enclosure" {
  140. length, _ := strconv.Atoi(link.Length)
  141. enclosures = append(enclosures, &model.Enclosure{URL: link.URL, MimeType: link.Type, Size: length})
  142. }
  143. }
  144. return enclosures
  145. }
  146. func getAuthor(author atomAuthor) string {
  147. if author.Name != "" {
  148. return strings.TrimSpace(author.Name)
  149. }
  150. if author.Email != "" {
  151. return strings.TrimSpace(author.Email)
  152. }
  153. return ""
  154. }