oss/miniflux-v2
oss/miniflux-v2
1
0
Fork 0
mirror of https://github.com/miniflux/v2.git synced 2025-06-27 16:36:00 +00:00
Code Wiki Activity

refactor(youtube): Remove a regex and make use of fetchWatchTime

Browse source
This commit is contained in:
jvoisin 2025-06-18 22:32:30 +02:00 • committed by Frédéric Guillot
parent c818d5bbb8
commit b139ac4a2c
1 changed files with 2 additions and 40 deletions
Download patch file Download diff file Expand all files Collapse all files

42
internal/reader/processor/youtube.go
View file

@ -5,29 +5,20 @@ package processor // import "miniflux.app/v2/internal/reader/processor"
import (
"encoding/json"
"errors"
"fmt"
"log/slog"
"net/url"
"regexp"
"strconv"
"strings"
"time"
"github.com/PuerkitoBio/goquery"
"miniflux.app/v2/internal/config"
"miniflux.app/v2/internal/model"
"miniflux.app/v2/internal/proxyrotator"
"miniflux.app/v2/internal/reader/fetcher"
)
var (
youtubeRegex = regexp.MustCompile(`youtube\.com/watch\?v=(.*)$`)
)
func isYouTubeVideoURL(websiteURL string) bool {
return len(youtubeRegex.FindStringSubmatch(websiteURL)) == 2
return strings.Contains(websiteURL, "youtube.com/watch?v=")
}
func getVideoIDFromYouTubeURL(websiteURL string) string {
@ -48,36 +39,7 @@ func shouldFetchYouTubeWatchTimeInBulk() bool {
}
func fetchYouTubeWatchTimeForSingleEntry(websiteURL string) (int, error) {
slog.Debug("Fetching YouTube watch time for a single entry", slog.String("website_url", websiteURL))
requestBuilder := fetcher.NewRequestBuilder()
requestBuilder.WithTimeout(config.Opts.HTTPClientTimeout())
requestBuilder.WithProxyRotator(proxyrotator.ProxyRotatorInstance)
responseHandler := fetcher.NewResponseHandler(requestBuilder.ExecuteRequest(websiteURL))
defer responseHandler.Close()
if localizedError := responseHandler.LocalizedError(); localizedError != nil {
slog.Warn("Unable to fetch YouTube page", 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
}
htmlDuration, exists := doc.FindMatcher(goquery.Single(`meta[itemprop="duration"]`)).Attr("content")
if !exists {
return 0, errors.New("youtube: duration has not found")
}
parsedDuration, err := parseISO8601(htmlDuration)
if err != nil {
return 0, fmt.Errorf("youtube: unable to parse duration %s: %v", htmlDuration, err)
}
return int(parsedDuration.Minutes()), nil
return fetchWatchTime(websiteURL, `meta[itemprop="duration"]`, true)
}
func fetchYouTubeWatchTimeInBulk(entries []*model.Entry) {