Use YouTube video duration as read time
This feature works by scraping YouTube website. To enable it, set the FETCH_YOUTUBE_WATCH_TIME environment variable to 1. Resolves #972.
This commit is contained in:
parent
e5b2eab727
commit
c3f871b49b
6 changed files with 153 additions and 1 deletions
|
@ -1413,3 +1413,21 @@ func TestAuthProxyUserCreationAdmin(t *testing.T) {
|
|||
t.Fatalf(`Unexpected AUTH_PROXY_USER_CREATION value, got %v instead of %v`, result, expected)
|
||||
}
|
||||
}
|
||||
|
||||
func TestFetchYouTubeWatchTime(t *testing.T) {
|
||||
os.Clearenv()
|
||||
os.Setenv("FETCH_YOUTUBE_WATCH_TIME", "1")
|
||||
|
||||
parser := NewParser()
|
||||
opts, err := parser.ParseEnvironmentVariables()
|
||||
if err != nil {
|
||||
t.Fatalf(`Parsing failure: %v`, err)
|
||||
}
|
||||
|
||||
expected := true
|
||||
result := opts.FetchYouTubeWatchTime()
|
||||
|
||||
if result != expected {
|
||||
t.Fatalf(`Unexpected FETCH_YOUTUBE_WATCH_TIME value, got %v instead of %v`, result, expected)
|
||||
}
|
||||
}
|
||||
|
|
|
@ -43,6 +43,7 @@ const (
|
|||
defaultCleanupArchiveUnreadDays = 180
|
||||
defaultCleanupRemoveSessionsDays = 30
|
||||
defaultProxyImages = "http-only"
|
||||
defaultFetchYouTubeWatchTime = false
|
||||
defaultCreateAdmin = false
|
||||
defaultAdminUsername = ""
|
||||
defaultAdminPassword = ""
|
||||
|
@ -108,6 +109,7 @@ type Options struct {
|
|||
adminUsername string
|
||||
adminPassword string
|
||||
proxyImages string
|
||||
fetchYouTubeWatchTime bool
|
||||
oauth2UserCreationAllowed bool
|
||||
oauth2ClientID string
|
||||
oauth2ClientSecret string
|
||||
|
@ -162,6 +164,7 @@ func NewOptions() *Options {
|
|||
workerPoolSize: defaultWorkerPoolSize,
|
||||
createAdmin: defaultCreateAdmin,
|
||||
proxyImages: defaultProxyImages,
|
||||
fetchYouTubeWatchTime: defaultFetchYouTubeWatchTime,
|
||||
oauth2UserCreationAllowed: defaultOAuth2UserCreation,
|
||||
oauth2ClientID: defaultOAuth2ClientID,
|
||||
oauth2ClientSecret: defaultOAuth2ClientSecret,
|
||||
|
@ -373,6 +376,12 @@ func (o *Options) AdminPassword() string {
|
|||
return o.adminPassword
|
||||
}
|
||||
|
||||
// FetchYouTubeWatchTime returns true if the YouTube video duration
|
||||
// should be fetched and used as a reading time.
|
||||
func (o *Options) FetchYouTubeWatchTime() bool {
|
||||
return o.fetchYouTubeWatchTime
|
||||
}
|
||||
|
||||
// ProxyImages returns "none" to never proxy, "http-only" to proxy non-HTTPS, "all" to always proxy.
|
||||
func (o *Options) ProxyImages() string {
|
||||
return o.proxyImages
|
||||
|
@ -469,6 +478,7 @@ func (o *Options) SortedOptions() []*Option {
|
|||
"DATABASE_MIN_CONNS": o.databaseMinConns,
|
||||
"DATABASE_URL": o.databaseURL,
|
||||
"DEBUG": o.debug,
|
||||
"FETCH_YOUTUBE_WATCH_TIME": o.fetchYouTubeWatchTime,
|
||||
"HSTS": o.hsts,
|
||||
"HTTPS": o.HTTPS,
|
||||
"HTTP_CLIENT_MAX_BODY_SIZE": o.httpClientMaxBodySize,
|
||||
|
|
|
@ -187,6 +187,8 @@ func (p *Parser) parseLines(lines []string) (err error) {
|
|||
p.opts.metricsRefreshInterval = parseInt(value, defaultMetricsRefreshInterval)
|
||||
case "METRICS_ALLOWED_NETWORKS":
|
||||
p.opts.metricsAllowedNetworks = parseStringList(value, []string{defaultMetricsAllowedNetworks})
|
||||
case "FETCH_YOUTUBE_WATCH_TIME":
|
||||
p.opts.fetchYouTubeWatchTime = parseBool(value, defaultFetchYouTubeWatchTime)
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
@ -107,6 +107,12 @@ Set the value to 1 to enable debug logs\&.
|
|||
.br
|
||||
Disabled by default\&.
|
||||
.TP
|
||||
.B FETCH_YOUTUBE_WATCH_TIME
|
||||
Set the value to 1 to scrape video duration from YouTube website and
|
||||
use it as a reading time\&.
|
||||
.br
|
||||
Disabled by default\&.
|
||||
.TP
|
||||
.B SERVER_TIMING_HEADER
|
||||
Set the value to 1 to enable server-timing headers\&.
|
||||
.br
|
||||
|
|
|
@ -5,24 +5,35 @@
|
|||
package processor
|
||||
|
||||
import (
|
||||
"errors"
|
||||
"fmt"
|
||||
"math"
|
||||
"regexp"
|
||||
"strconv"
|
||||
"strings"
|
||||
"time"
|
||||
"unicode/utf8"
|
||||
|
||||
"miniflux.app/config"
|
||||
"miniflux.app/http/client"
|
||||
"miniflux.app/logger"
|
||||
"miniflux.app/metric"
|
||||
"miniflux.app/model"
|
||||
"miniflux.app/reader/browser"
|
||||
"miniflux.app/reader/rewrite"
|
||||
"miniflux.app/reader/sanitizer"
|
||||
"miniflux.app/reader/scraper"
|
||||
"miniflux.app/storage"
|
||||
|
||||
"github.com/PuerkitoBio/goquery"
|
||||
"github.com/rylans/getlang"
|
||||
)
|
||||
|
||||
var (
|
||||
youtubeRegex = regexp.MustCompile(`youtube\.com/watch\?v=(.*)`)
|
||||
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)?)?$`)
|
||||
)
|
||||
|
||||
// ProcessFeedEntries downloads original web page for entries and apply filters.
|
||||
func ProcessFeedEntries(store *storage.Storage, feed *model.Feed) {
|
||||
var filteredEntries model.Entries
|
||||
|
@ -63,7 +74,20 @@ func ProcessFeedEntries(store *storage.Storage, feed *model.Feed) {
|
|||
// The sanitizer should always run at the end of the process to make sure unsafe HTML is filtered.
|
||||
entry.Content = sanitizer.Sanitize(entry.URL, entry.Content)
|
||||
|
||||
if config.Opts.FetchYouTubeWatchTime() {
|
||||
if matches := youtubeRegex.FindStringSubmatch(entry.URL); len(matches) == 2 {
|
||||
watchTime, err := fetchYouTubeWatchTime(entry.URL)
|
||||
if err != nil {
|
||||
logger.Error("[Processor] Unable to fetch YouTube watch time: %q => %v", entry.URL, err)
|
||||
}
|
||||
entry.ReadingTime = watchTime
|
||||
}
|
||||
}
|
||||
|
||||
if entry.ReadingTime == 0 {
|
||||
entry.ReadingTime = calculateReadingTime(entry.Content)
|
||||
}
|
||||
|
||||
filteredEntries = append(filteredEntries, entry)
|
||||
}
|
||||
|
||||
|
@ -120,6 +144,68 @@ func ProcessEntryWebPage(entry *model.Entry) error {
|
|||
return nil
|
||||
}
|
||||
|
||||
func fetchYouTubeWatchTime(url string) (int, error) {
|
||||
clt := client.NewClientWithConfig(url, config.Opts)
|
||||
response, browserErr := browser.Exec(clt)
|
||||
if browserErr != nil {
|
||||
return 0, browserErr
|
||||
}
|
||||
|
||||
doc, docErr := goquery.NewDocumentFromReader(response.Body)
|
||||
if docErr != nil {
|
||||
return 0, docErr
|
||||
}
|
||||
|
||||
durs, exists := doc.Find(`meta[itemprop="duration"]`).First().Attr("content")
|
||||
if !exists {
|
||||
return 0, errors.New("duration has not found")
|
||||
}
|
||||
|
||||
dur, err := parseISO8601(durs)
|
||||
if err != nil {
|
||||
return 0, fmt.Errorf("unable to parse duration %s: %v", durs, err)
|
||||
}
|
||||
|
||||
return int(dur.Minutes()), nil
|
||||
}
|
||||
|
||||
// parseISO8601 parses an ISO 8601 duration string.
|
||||
func parseISO8601(from string) (time.Duration, error) {
|
||||
var match []string
|
||||
var d time.Duration
|
||||
|
||||
if iso8601Regex.MatchString(from) {
|
||||
match = iso8601Regex.FindStringSubmatch(from)
|
||||
} else {
|
||||
return 0, errors.New("could not parse duration string")
|
||||
}
|
||||
|
||||
for i, name := range iso8601Regex.SubexpNames() {
|
||||
part := match[i]
|
||||
if i == 0 || name == "" || part == "" {
|
||||
continue
|
||||
}
|
||||
|
||||
val, err := strconv.ParseInt(part, 10, 64)
|
||||
if err != nil {
|
||||
return 0, err
|
||||
}
|
||||
|
||||
switch name {
|
||||
case "hour":
|
||||
d = d + (time.Duration(val) * time.Hour)
|
||||
case "minute":
|
||||
d = d + (time.Duration(val) * time.Minute)
|
||||
case "second":
|
||||
d = d + (time.Duration(val) * time.Second)
|
||||
default:
|
||||
return 0, fmt.Errorf("unknown field %s", name)
|
||||
}
|
||||
}
|
||||
|
||||
return d, nil
|
||||
}
|
||||
|
||||
func calculateReadingTime(content string) int {
|
||||
sanitizedContent := sanitizer.StripTags(content)
|
||||
languageInfo := getlang.FromString(sanitizedContent)
|
||||
|
|
|
@ -6,6 +6,7 @@ package processor // import "miniflux.app/reader/processor"
|
|||
|
||||
import (
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"miniflux.app/model"
|
||||
)
|
||||
|
@ -47,3 +48,32 @@ func TestAllowEntries(t *testing.T) {
|
|||
}
|
||||
}
|
||||
}
|
||||
|
||||
func TestParseISO8601(t *testing.T) {
|
||||
var scenarios = []struct {
|
||||
duration string
|
||||
expected time.Duration
|
||||
}{
|
||||
// Live streams and radio.
|
||||
{"PT0M0S", 0},
|
||||
// https://www.youtube.com/watch?v=HLrqNhgdiC0
|
||||
{"PT6M20S", (6 * time.Minute) + (20 * time.Second)},
|
||||
// https://www.youtube.com/watch?v=LZa5KKfqHtA
|
||||
{"PT5M41S", (5 * time.Minute) + (41 * time.Second)},
|
||||
// https://www.youtube.com/watch?v=yIxEEgEuhT4
|
||||
{"PT51M52S", (51 * time.Minute) + (52 * time.Second)},
|
||||
// https://www.youtube.com/watch?v=bpHf1XcoiFs
|
||||
{"PT80M42S", (1 * time.Hour) + (20 * time.Minute) + (42 * time.Second)},
|
||||
}
|
||||
|
||||
for _, tc := range scenarios {
|
||||
result, err := parseISO8601(tc.duration)
|
||||
if err != nil {
|
||||
t.Errorf("Got an error when parsing %q: %v", tc.duration, err)
|
||||
}
|
||||
|
||||
if tc.expected != result {
|
||||
t.Errorf(`Unexpected result, got %v for duration %q`, result, tc.duration)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
Loading…
Reference in a new issue