darklyn/torrent-indexer
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Fix/IMDB and comando-filmes date detection (#26)

Browse Source 
* chg: fix: data and imdb detection

* chg: fix: improve IMDB detection

* chg: feat: add more audio strings
This commit is contained in:
Felipe Marinho
2024-12-13 12:45:17 -03:00
committed by GitHub
parent 88d6d506bf
commit dc3cb3be92
5 changed files with 174 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

57
api/comando_torrents.go
View File

@@ -163,19 +163,11 @@ func getTorrents(ctx context.Context, i *Indexer, link string) ([]schema.Indexed
// div itemprop="datePublished"
datePublished := strings.TrimSpace(article.Find("div[itemprop=\"datePublished\"]").Text())
// pattern: 10 de setembro de 2021
re := regexp.MustCompile(`(\d{2}) de (\w+) de (\d{4})`)
matches := re.FindStringSubmatch(datePublished)
var date time.Time
if len(matches) > 0 {
day := matches[1]
month := matches[2]
year := matches[3]
datePublished = fmt.Sprintf("%s-%s-%s", year, replacer.Replace(month), day)
date, err = time.Parse("2006-01-02", datePublished)
if err != nil {
return nil, err
}
date, err := parseLocalizedDate(datePublished)
if err != nil {
return nil, err
}
magnets := textContent.Find("a[href^=\"magnet\"]")
var magnetLinks []string
magnets.Each(func(i int, s *goquery.Selection) {
@@ -217,10 +209,9 @@ func getTorrents(ctx context.Context, i *Indexer, link string) ([]schema.Indexed
imdbLink := ""
article.Find("a").Each(func(i int, s *goquery.Selection) {
link, _ := s.Attr("href")
re := regexp.MustCompile(`https://www.imdb.com/title/(tt\d+)`)
matches := re.FindStringSubmatch(link)
if len(matches) > 0 {
imdbLink = matches[0]
_imdbLink, err := getIMDBLink(link)
if err == nil {
imdbLink = _imdbLink
}
})
@@ -293,6 +284,40 @@ func getTorrents(ctx context.Context, i *Indexer, link string) ([]schema.Indexed
return indexedTorrents, nil
}
func getIMDBLink(link string) (string, error) {
var imdbLink string
re := regexp.MustCompile(`https://www.imdb.com(/[a-z]{2})?/title/(tt\d+)/?`)
matches := re.FindStringSubmatch(link)
if len(matches) > 0 {
imdbLink = matches[0]
} else {
return "", fmt.Errorf("no imdb link found")
}
return imdbLink, nil
}
func parseLocalizedDate(datePublished string) (time.Time, error) {
re := regexp.MustCompile(`(\d{1,2}) de (\w+) de (\d{4})`)
matches := re.FindStringSubmatch(datePublished)
if len(matches) > 0 {
day := matches[1]
// append 0 to single digit day
if len(day) == 1 {
day = fmt.Sprintf("0%s", day)
}
month := matches[2]
year := matches[3]
datePublished = fmt.Sprintf("%s-%s-%s", year, replacer.Replace(month), day)
date, err := time.Parse("2006-01-02", datePublished)
if err != nil {
return time.Time{}, err
}
return date, nil
}
return time.Time{}, nil
}
func stableUniq(s []string) []string {
var uniq []map[string]interface{}
m := make(map[string]map[string]interface{})