youtube.go 4.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165
  1. // SPDX-FileCopyrightText: Copyright The Miniflux Authors. All rights reserved.
  2. // SPDX-License-Identifier: Apache-2.0
  3. package processor
  4. import (
  5. "encoding/json"
  6. "errors"
  7. "fmt"
  8. "log/slog"
  9. "net/url"
  10. "regexp"
  11. "strconv"
  12. "time"
  13. "github.com/PuerkitoBio/goquery"
  14. "miniflux.app/v2/internal/config"
  15. "miniflux.app/v2/internal/model"
  16. "miniflux.app/v2/internal/reader/fetcher"
  17. )
  18. var (
  19. youtubeRegex = regexp.MustCompile(`youtube\.com/watch\?v=(.*)$`)
  20. iso8601Regex = regexp.MustCompile(`^P((?P<year>\d+)Y)?((?P<month>\d+)M)?((?P<week>\d+)W)?((?P<day>\d+)D)?(T((?P<hour>\d+)H)?((?P<minute>\d+)M)?((?P<second>\d+)S)?)?$`)
  21. )
  22. func shouldFetchYouTubeWatchTime(entry *model.Entry) bool {
  23. if !config.Opts.FetchYouTubeWatchTime() {
  24. return false
  25. }
  26. matches := youtubeRegex.FindStringSubmatch(entry.URL)
  27. urlMatchesYouTubePattern := len(matches) == 2
  28. return urlMatchesYouTubePattern
  29. }
  30. func fetchYouTubeWatchTime(websiteURL string) (int, error) {
  31. if config.Opts.YouTubeApiKey() == "" {
  32. return fetchYouTubeWatchTimeFromWebsite(websiteURL)
  33. } else {
  34. return fetchYouTubeWatchTimeFromApi(websiteURL)
  35. }
  36. }
  37. func fetchYouTubeWatchTimeFromWebsite(websiteURL string) (int, error) {
  38. requestBuilder := fetcher.NewRequestBuilder()
  39. requestBuilder.WithTimeout(config.Opts.HTTPClientTimeout())
  40. requestBuilder.WithProxy(config.Opts.HTTPClientProxy())
  41. responseHandler := fetcher.NewResponseHandler(requestBuilder.ExecuteRequest(websiteURL))
  42. defer responseHandler.Close()
  43. if localizedError := responseHandler.LocalizedError(); localizedError != nil {
  44. slog.Warn("Unable to fetch YouTube page", slog.String("website_url", websiteURL), slog.Any("error", localizedError.Error()))
  45. return 0, localizedError.Error()
  46. }
  47. doc, docErr := goquery.NewDocumentFromReader(responseHandler.Body(config.Opts.HTTPClientMaxBodySize()))
  48. if docErr != nil {
  49. return 0, docErr
  50. }
  51. durs, exists := doc.Find(`meta[itemprop="duration"]`).First().Attr("content")
  52. if !exists {
  53. return 0, errors.New("duration has not found")
  54. }
  55. dur, err := parseISO8601(durs)
  56. if err != nil {
  57. return 0, fmt.Errorf("unable to parse duration %s: %v", durs, err)
  58. }
  59. return int(dur.Minutes()), nil
  60. }
  61. func fetchYouTubeWatchTimeFromApi(websiteURL string) (int, error) {
  62. requestBuilder := fetcher.NewRequestBuilder()
  63. requestBuilder.WithTimeout(config.Opts.HTTPClientTimeout())
  64. requestBuilder.WithProxy(config.Opts.HTTPClientProxy())
  65. parsedWebsiteURL, err := url.Parse(websiteURL)
  66. if err != nil {
  67. return 0, fmt.Errorf("unable to parse URL: %v", err)
  68. }
  69. apiQuery := url.Values{}
  70. apiQuery.Set("id", parsedWebsiteURL.Query().Get("v"))
  71. apiQuery.Set("key", config.Opts.YouTubeApiKey())
  72. apiQuery.Set("part", "contentDetails")
  73. apiURL := url.URL{
  74. Scheme: "https",
  75. Host: "www.googleapis.com",
  76. Path: "youtube/v3/videos",
  77. RawQuery: apiQuery.Encode(),
  78. }
  79. responseHandler := fetcher.NewResponseHandler(requestBuilder.ExecuteRequest(apiURL.String()))
  80. defer responseHandler.Close()
  81. if localizedError := responseHandler.LocalizedError(); localizedError != nil {
  82. slog.Warn("Unable to fetch contentDetails from YouTube API", slog.String("website_url", websiteURL), slog.Any("error", localizedError.Error()))
  83. return 0, localizedError.Error()
  84. }
  85. var videos struct {
  86. Items []struct {
  87. ContentDetails struct {
  88. Duration string `json:"duration"`
  89. } `json:"contentDetails"`
  90. } `json:"items"`
  91. }
  92. if err := json.NewDecoder(responseHandler.Body(config.Opts.HTTPClientMaxBodySize())).Decode(&videos); err != nil {
  93. return 0, fmt.Errorf("unable to decode JSON: %v", err)
  94. }
  95. if n := len(videos.Items); n != 1 {
  96. return 0, fmt.Errorf("invalid items length: %d", n)
  97. }
  98. durs := videos.Items[0].ContentDetails.Duration
  99. dur, err := parseISO8601(durs)
  100. if err != nil {
  101. return 0, fmt.Errorf("unable to parse duration %s: %v", durs, err)
  102. }
  103. return int(dur.Minutes()), nil
  104. }
  105. func parseISO8601(from string) (time.Duration, error) {
  106. var match []string
  107. var d time.Duration
  108. if iso8601Regex.MatchString(from) {
  109. match = iso8601Regex.FindStringSubmatch(from)
  110. } else {
  111. return 0, errors.New("could not parse duration string")
  112. }
  113. for i, name := range iso8601Regex.SubexpNames() {
  114. part := match[i]
  115. if i == 0 || name == "" || part == "" {
  116. continue
  117. }
  118. val, err := strconv.ParseInt(part, 10, 64)
  119. if err != nil {
  120. return 0, err
  121. }
  122. switch name {
  123. case "hour":
  124. d += (time.Duration(val) * time.Hour)
  125. case "minute":
  126. d += (time.Duration(val) * time.Minute)
  127. case "second":
  128. d += (time.Duration(val) * time.Second)
  129. default:
  130. return 0, fmt.Errorf("unknown field %s", name)
  131. }
  132. }
  133. return d, nil
  134. }