Feat/Rede torrents (#34)

* new: feat: add brand new redetorrent.com indexer

* chg: refactor: create common package

* chg: fix: comandohds formatting

* chg: fix: dual audio detection

* chg: fix: parsing issues

* chg: refactor: remove duplicated code

* chg: refactor: move test funcs to common file
This commit is contained in:
2025-07-22 14:57:05 -03:00
committed by GitHub
parent 0b21e3b1e7
commit 782f9d6b3f
13 changed files with 762 additions and 327 deletions

View File

@@ -230,19 +230,7 @@ func getTorrents(ctx context.Context, i *Indexer, link string) ([]schema.Indexed
releaseTitle := magnet.DisplayName
infoHash := magnet.InfoHash.String()
trackers := magnet.Trackers
magnetAudio := []schema.Audio{}
if strings.Contains(strings.ToLower(releaseTitle), "dual") || strings.Contains(strings.ToLower(releaseTitle), "dublado") {
magnetAudio = append(magnetAudio, audio...)
} else if len(audio) > 1 {
// remove portuguese audio, and append to magnetAudio
for _, a := range audio {
if a != schema.AudioPortuguese {
magnetAudio = append(magnetAudio, a)
}
}
} else {
magnetAudio = append(magnetAudio, audio...)
}
magnetAudio := getAudioFromTitle(releaseTitle, audio)
peer, seed, err := goscrape.GetLeechsAndSeeds(ctx, i.redis, i.metrics, infoHash, trackers)
if err != nil {
@@ -284,19 +272,6 @@ 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)
@@ -350,56 +325,6 @@ func stableUniq(s []string) []string {
return uniqValues
}
func findYearFromText(text string, title string) (year string) {
re := regexp.MustCompile(`Lançamento: (.*)`)
yearMatch := re.FindStringSubmatch(text)
if len(yearMatch) > 0 {
year = yearMatch[1]
}
if year == "" {
re = regexp.MustCompile(`\((\d{4})\)`)
yearMatch := re.FindStringSubmatch(title)
if len(yearMatch) > 0 {
year = yearMatch[1]
}
}
return strings.TrimSpace(year)
}
func findAudioFromText(text string) []schema.Audio {
var audio []schema.Audio
re := regexp.MustCompile(`(.udio|Idioma):.?(.*)`)
audioMatch := re.FindStringSubmatch(text)
if len(audioMatch) > 0 {
sep := getSeparator(audioMatch[2])
langs_raw := strings.Split(audioMatch[2], sep)
for _, lang := range langs_raw {
lang = strings.TrimSpace(lang)
a := schema.GetAudioFromString(lang)
if a != nil {
audio = append(audio, *a)
} else {
fmt.Println("unknown language:", lang)
}
}
}
return audio
}
func findSizesFromText(text string) []string {
var sizes []string
// everything that ends with GB or MB, using ',' or '.' as decimal separator
re := regexp.MustCompile(`(\d+[\.,]?\d+) ?(GB|MB)`)
sizesMatch := re.FindAllStringSubmatch(text, -1)
if len(sizesMatch) > 0 {
for _, size := range sizesMatch {
sizes = append(sizes, size[0])
}
}
return sizes
}
func processTitle(title string, a []schema.Audio) string {
// remove ' - Donwload' from title
title = strings.Replace(title, " Download", "", -1)
@@ -413,26 +338,6 @@ func processTitle(title string, a []schema.Audio) string {
return title
}
func appendAudioISO639_2Code(title string, a []schema.Audio) string {
if len(a) > 0 {
audio := []string{}
for _, lang := range a {
audio = append(audio, lang.String())
}
title = fmt.Sprintf("%s (%s)", title, strings.Join(audio, ", "))
}
return title
}
func getSeparator(s string) string {
if strings.Contains(s, "|") {
return "|"
} else if strings.Contains(s, ",") {
return ","
}
return " "
}
func getDocument(ctx context.Context, i *Indexer, link string) (*goquery.Document, error) {
// try to get from redis first
docCache, err := i.redis.Get(ctx, link)