handler.go 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214
  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 feed
  5. import (
  6. "fmt"
  7. "time"
  8. "github.com/miniflux/miniflux/errors"
  9. "github.com/miniflux/miniflux/http"
  10. "github.com/miniflux/miniflux/logger"
  11. "github.com/miniflux/miniflux/model"
  12. "github.com/miniflux/miniflux/reader/icon"
  13. "github.com/miniflux/miniflux/reader/processor"
  14. "github.com/miniflux/miniflux/storage"
  15. "github.com/miniflux/miniflux/timer"
  16. )
  17. var (
  18. errRequestFailed = "Unable to execute request: %v"
  19. errServerFailure = "Unable to fetch feed (statusCode=%d)"
  20. errDuplicate = "This feed already exists (%s)"
  21. errNotFound = "Feed %d not found"
  22. errEncoding = "Unable to normalize encoding: %v"
  23. errCategoryNotFound = "Category not found for this user"
  24. errEmptyFeed = "This feed is empty"
  25. )
  26. // Handler contains all the logic to create and refresh feeds.
  27. type Handler struct {
  28. store *storage.Storage
  29. }
  30. // CreateFeed fetch, parse and store a new feed.
  31. func (h *Handler) CreateFeed(userID, categoryID int64, url string, crawler bool) (*model.Feed, error) {
  32. defer timer.ExecutionTime(time.Now(), fmt.Sprintf("[Handler:CreateFeed] feedUrl=%s", url))
  33. if !h.store.CategoryExists(userID, categoryID) {
  34. return nil, errors.NewLocalizedError(errCategoryNotFound)
  35. }
  36. client := http.NewClient(url)
  37. response, err := client.Get()
  38. if err != nil {
  39. if _, ok := err.(errors.LocalizedError); ok {
  40. return nil, err
  41. }
  42. return nil, errors.NewLocalizedError(errRequestFailed, err)
  43. }
  44. if response.HasServerFailure() {
  45. return nil, errors.NewLocalizedError(errServerFailure, response.StatusCode)
  46. }
  47. // Content-Length = -1 when no Content-Length header is sent
  48. if response.ContentLength == 0 {
  49. return nil, errors.NewLocalizedError(errEmptyFeed)
  50. }
  51. if h.store.FeedURLExists(userID, response.EffectiveURL) {
  52. return nil, errors.NewLocalizedError(errDuplicate, response.EffectiveURL)
  53. }
  54. body, err := response.NormalizeBodyEncoding()
  55. if err != nil {
  56. return nil, errors.NewLocalizedError(errEncoding, err)
  57. }
  58. subscription, err := parseFeed(body)
  59. if err != nil {
  60. return nil, err
  61. }
  62. feedProcessor := processor.NewFeedProcessor(userID, h.store, subscription)
  63. feedProcessor.WithCrawler(crawler)
  64. feedProcessor.Process()
  65. subscription.Category = &model.Category{ID: categoryID}
  66. subscription.EtagHeader = response.ETag
  67. subscription.LastModifiedHeader = response.LastModified
  68. subscription.FeedURL = response.EffectiveURL
  69. subscription.UserID = userID
  70. subscription.Crawler = crawler
  71. if subscription.SiteURL == "" {
  72. subscription.SiteURL = subscription.FeedURL
  73. }
  74. err = h.store.CreateFeed(subscription)
  75. if err != nil {
  76. return nil, err
  77. }
  78. logger.Debug("[Handler:CreateFeed] Feed saved with ID: %d", subscription.ID)
  79. icon, err := icon.FindIcon(subscription.SiteURL)
  80. if err != nil {
  81. logger.Error("[Handler:CreateFeed] %v", err)
  82. } else if icon == nil {
  83. logger.Info("No icon found for feedID=%d", subscription.ID)
  84. } else {
  85. h.store.CreateFeedIcon(subscription, icon)
  86. }
  87. return subscription, nil
  88. }
  89. // RefreshFeed fetch and update a feed if necessary.
  90. func (h *Handler) RefreshFeed(userID, feedID int64) error {
  91. defer timer.ExecutionTime(time.Now(), fmt.Sprintf("[Handler:RefreshFeed] feedID=%d", feedID))
  92. originalFeed, err := h.store.FeedByID(userID, feedID)
  93. if err != nil {
  94. return err
  95. }
  96. if originalFeed == nil {
  97. return errors.NewLocalizedError(errNotFound, feedID)
  98. }
  99. client := http.NewClientWithCacheHeaders(originalFeed.FeedURL, originalFeed.EtagHeader, originalFeed.LastModifiedHeader)
  100. response, err := client.Get()
  101. if err != nil {
  102. var customErr errors.LocalizedError
  103. if lerr, ok := err.(errors.LocalizedError); ok {
  104. customErr = lerr
  105. } else {
  106. customErr = errors.NewLocalizedError(errRequestFailed, err)
  107. }
  108. originalFeed.ParsingErrorCount++
  109. originalFeed.ParsingErrorMsg = customErr.Error()
  110. h.store.UpdateFeed(originalFeed)
  111. return customErr
  112. }
  113. originalFeed.CheckedAt = time.Now()
  114. if response.HasServerFailure() {
  115. err := errors.NewLocalizedError(errServerFailure, response.StatusCode)
  116. originalFeed.ParsingErrorCount++
  117. originalFeed.ParsingErrorMsg = err.Error()
  118. h.store.UpdateFeed(originalFeed)
  119. return err
  120. }
  121. if response.IsModified(originalFeed.EtagHeader, originalFeed.LastModifiedHeader) {
  122. logger.Debug("[Handler:RefreshFeed] Feed #%d has been modified", feedID)
  123. // Content-Length = -1 when no Content-Length header is sent
  124. if response.ContentLength == 0 {
  125. err := errors.NewLocalizedError(errEmptyFeed)
  126. originalFeed.ParsingErrorCount++
  127. originalFeed.ParsingErrorMsg = err.Error()
  128. h.store.UpdateFeed(originalFeed)
  129. return err
  130. }
  131. body, err := response.NormalizeBodyEncoding()
  132. if err != nil {
  133. return errors.NewLocalizedError(errEncoding, err)
  134. }
  135. subscription, err := parseFeed(body)
  136. if err != nil {
  137. originalFeed.ParsingErrorCount++
  138. originalFeed.ParsingErrorMsg = err.Error()
  139. h.store.UpdateFeed(originalFeed)
  140. return err
  141. }
  142. feedProcessor := processor.NewFeedProcessor(userID, h.store, subscription)
  143. feedProcessor.WithScraperRules(originalFeed.ScraperRules)
  144. feedProcessor.WithRewriteRules(originalFeed.RewriteRules)
  145. feedProcessor.WithCrawler(originalFeed.Crawler)
  146. feedProcessor.Process()
  147. originalFeed.EtagHeader = response.ETag
  148. originalFeed.LastModifiedHeader = response.LastModified
  149. // Note: We don't update existing entries when the crawler is enabled (we crawl only inexisting entries).
  150. if err := h.store.UpdateEntries(originalFeed.UserID, originalFeed.ID, subscription.Entries, !originalFeed.Crawler); err != nil {
  151. return err
  152. }
  153. if !h.store.HasIcon(originalFeed.ID) {
  154. logger.Debug("[Handler:RefreshFeed] Looking for feed icon")
  155. icon, err := icon.FindIcon(originalFeed.SiteURL)
  156. if err != nil {
  157. logger.Debug("[Handler:RefreshFeed] %v", err)
  158. } else {
  159. h.store.CreateFeedIcon(originalFeed, icon)
  160. }
  161. }
  162. } else {
  163. logger.Debug("[Handler:RefreshFeed] Feed #%d not modified", feedID)
  164. }
  165. originalFeed.ParsingErrorCount = 0
  166. originalFeed.ParsingErrorMsg = ""
  167. if originalFeed.SiteURL == "" {
  168. originalFeed.SiteURL = originalFeed.FeedURL
  169. }
  170. return h.store.UpdateFeed(originalFeed)
  171. }
  172. // NewFeedHandler returns a feed handler.
  173. func NewFeedHandler(store *storage.Storage) *Handler {
  174. return &Handler{store: store}
  175. }