options: add FETCH_NEBULA_WATCH_TIME

This commit is contained in:
fin444 2024-05-01 19:28:59 -04:00 committed by Frédéric Guillot
parent ca62b0b36b
commit a631bd527d
5 changed files with 96 additions and 0 deletions

View File

@ -2021,6 +2021,24 @@ func TestAuthProxyUserCreationAdmin(t *testing.T) {
}
}
func TestFetchNebulaWatchTime(t *testing.T) {
os.Clearenv()
os.Setenv("FETCH_NEBULA_WATCH_TIME", "1")
parser := NewParser()
opts, err := parser.ParseEnvironmentVariables()
if err != nil {
t.Fatalf(`Parsing failure: %v`, err)
}
expected := true
result := opts.FetchNebulaWatchTime()
if result != expected {
t.Fatalf(`Unexpected FETCH_NEBULA_WATCH_TIME value, got %v instead of %v`, result, expected)
}
}
func TestFetchOdyseeWatchTime(t *testing.T) {
os.Clearenv()
os.Setenv("FETCH_ODYSEE_WATCH_TIME", "1")

View File

@ -56,6 +56,7 @@ const (
defaultMediaResourceTypes = "image"
defaultMediaProxyURL = ""
defaultFilterEntryMaxAgeDays = 0
defaultFetchNebulaWatchTime = false
defaultFetchOdyseeWatchTime = false
defaultFetchYouTubeWatchTime = false
defaultYouTubeEmbedUrlOverride = "https://www.youtube-nocookie.com/embed/"
@ -140,6 +141,7 @@ type Options struct {
mediaProxyMode string
mediaProxyResourceTypes []string
mediaProxyCustomURL string
fetchNebulaWatchTime bool
fetchOdyseeWatchTime bool
fetchYouTubeWatchTime bool
filterEntryMaxAgeDays int
@ -216,6 +218,7 @@ func NewOptions() *Options {
mediaProxyResourceTypes: []string{defaultMediaResourceTypes},
mediaProxyCustomURL: defaultMediaProxyURL,
filterEntryMaxAgeDays: defaultFilterEntryMaxAgeDays,
fetchNebulaWatchTime: defaultFetchNebulaWatchTime,
fetchOdyseeWatchTime: defaultFetchOdyseeWatchTime,
fetchYouTubeWatchTime: defaultFetchYouTubeWatchTime,
youTubeEmbedUrlOverride: defaultYouTubeEmbedUrlOverride,
@ -486,6 +489,12 @@ func (o *Options) YouTubeEmbedUrlOverride() string {
return o.youTubeEmbedUrlOverride
}
// FetchNebulaWatchTime returns true if the Nebula video duration
// should be fetched and used as a reading time.
func (o *Options) FetchNebulaWatchTime() bool {
return o.fetchNebulaWatchTime
}
// FetchOdyseeWatchTime returns true if the Odysee video duration
// should be fetched and used as a reading time.
func (o *Options) FetchOdyseeWatchTime() bool {
@ -647,6 +656,7 @@ func (o *Options) SortedOptions(redactSecret bool) []*Option {
"DISABLE_SCHEDULER_SERVICE": !o.schedulerService,
"FILTER_ENTRY_MAX_AGE_DAYS": o.filterEntryMaxAgeDays,
"FETCH_YOUTUBE_WATCH_TIME": o.fetchYouTubeWatchTime,
"FETCH_NEBULA_WATCH_TIME": o.fetchNebulaWatchTime,
"FETCH_ODYSEE_WATCH_TIME": o.fetchOdyseeWatchTime,
"HTTPS": o.HTTPS,
"HTTP_CLIENT_MAX_BODY_SIZE": o.httpClientMaxBodySize,

View File

@ -259,6 +259,8 @@ func (p *Parser) parseLines(lines []string) (err error) {
p.opts.metricsPassword = parseString(value, defaultMetricsPassword)
case "METRICS_PASSWORD_FILE":
p.opts.metricsPassword = readSecretFile(value, defaultMetricsPassword)
case "FETCH_NEBULA_WATCH_TIME":
p.opts.fetchNebulaWatchTime = parseBool(value, defaultFetchNebulaWatchTime)
case "FETCH_ODYSEE_WATCH_TIME":
p.opts.fetchOdyseeWatchTime = parseBool(value, defaultFetchOdyseeWatchTime)
case "FETCH_YOUTUBE_WATCH_TIME":

View File

@ -29,6 +29,7 @@ import (
var (
youtubeRegex = regexp.MustCompile(`youtube\.com/watch\?v=(.*)$`)
nebulaRegex = regexp.MustCompile(`^https://nebula\.tv`)
odyseeRegex = regexp.MustCompile(`^https://odysee\.com`)
iso8601Regex = regexp.MustCompile(`^P((?P<year>\d+)Y)?((?P<month>\d+)M)?((?P<week>\d+)W)?((?P<day>\d+)D)?(T((?P<hour>\d+)H)?((?P<minute>\d+)M)?((?P<second>\d+)S)?)?$`)
customReplaceRuleRegex = regexp.MustCompile(`rewrite\("(.*)"\|"(.*)"\)`)
@ -277,6 +278,25 @@ func updateEntryReadingTime(store *storage.Storage, feed *model.Feed, entry *mod
}
}
if shouldFetchNebulaWatchTime(entry) {
if entryIsNew {
watchTime, err := fetchNebulaWatchTime(entry.URL)
if err != nil {
slog.Warn("Unable to fetch Nebula watch time",
slog.Int64("user_id", user.ID),
slog.Int64("entry_id", entry.ID),
slog.String("entry_url", entry.URL),
slog.Int64("feed_id", feed.ID),
slog.String("feed_url", feed.FeedURL),
slog.Any("error", err),
)
}
entry.ReadingTime = watchTime
} else {
entry.ReadingTime = store.GetReadTime(feed.ID, entry.Hash)
}
}
if shouldFetchOdyseeWatchTime(entry) {
if entryIsNew {
watchTime, err := fetchOdyseeWatchTime(entry.URL)
@ -311,6 +331,14 @@ func shouldFetchYouTubeWatchTime(entry *model.Entry) bool {
return urlMatchesYouTubePattern
}
func shouldFetchNebulaWatchTime(entry *model.Entry) bool {
if !config.Opts.FetchNebulaWatchTime() {
return false
}
matches := nebulaRegex.FindStringSubmatch(entry.URL)
return matches != nil
}
func shouldFetchOdyseeWatchTime(entry *model.Entry) bool {
if !config.Opts.FetchOdyseeWatchTime() {
return false
@ -350,6 +378,38 @@ func fetchYouTubeWatchTime(websiteURL string) (int, error) {
return int(dur.Minutes()), nil
}
func fetchNebulaWatchTime(websiteURL string) (int, error) {
requestBuilder := fetcher.NewRequestBuilder()
requestBuilder.WithTimeout(config.Opts.HTTPClientTimeout())
requestBuilder.WithProxy(config.Opts.HTTPClientProxy())
responseHandler := fetcher.NewResponseHandler(requestBuilder.ExecuteRequest(websiteURL))
defer responseHandler.Close()
if localizedError := responseHandler.LocalizedError(); localizedError != nil {
slog.Warn("Unable to fetch Nebula watch time", slog.String("website_url", websiteURL), slog.Any("error", localizedError.Error()))
return 0, localizedError.Error()
}
doc, docErr := goquery.NewDocumentFromReader(responseHandler.Body(config.Opts.HTTPClientMaxBodySize()))
if docErr != nil {
return 0, docErr
}
durs, exists := doc.Find(`meta[property="video:duration"]`).First().Attr("content")
// durs contains video watch time in seconds
if !exists {
return 0, errors.New("duration has not found")
}
dur, err := strconv.ParseInt(durs, 10, 64)
if err != nil {
return 0, fmt.Errorf("unable to parse duration %s: %v", durs, err)
}
return int(dur / 60), nil
}
func fetchOdyseeWatchTime(websiteURL string) (int, error) {
requestBuilder := fetcher.NewRequestBuilder()
requestBuilder.WithTimeout(config.Opts.HTTPClientTimeout())

View File

@ -244,6 +244,12 @@ Set the value to 1 to disable the internal scheduler service\&.
.br
Default is false (The internal scheduler service is enabled)\&.
.TP
.B FETCH_NEBULA_WATCH_TIME
Set the value to 1 to scrape video duration from Nebula website and
use it as a reading time\&.
.br
Disabled by default\&.
.TP
.B FETCH_ODYSEE_WATCH_TIME
Set the value to 1 to scrape video duration from Odysee website and
use it as a reading time\&.