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

chg: fix: year detection algorithim

Browse Source
This commit is contained in:
Marinho
2024-03-10 12:48:48 +00:00
parent bfe26c91e3
commit b1cece2743
2 changed files with 24 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
api/bludv.go
View File

@@ -96,22 +96,7 @@ func getTorrentsBluDV(ctx context.Context, i *Indexer, link string) ([]IndexedTo
article := doc.Find(".post")
title := strings.Replace(article.Find(".title > h1").Text(), " - Download", "", -1)
textContent := article.Find("div.content")
// 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 := getPublishedDate(doc)
magnets := textContent.Find("a[href^=\"magnet\"]")
var magnetLinks []string
magnets.Each(func(i int, s *goquery.Selection) {
@@ -142,7 +127,10 @@ func getTorrentsBluDV(ctx context.Context, i *Indexer, link string) ([]IndexedTo
text := s.Text()
audio = append(audio, findAudioFromText(text)...)
year = findYearFromText(text, title)
y := findYearFromText(text, title)
if y != "" {
year = y
}
size = append(size, findSizesFromText(text)...)
})
@@ -225,3 +213,15 @@ func getTorrentsBluDV(ctx context.Context, i *Indexer, link string) ([]IndexedTo
return indexedTorrents, nil
}
func getPublishedDate(document *goquery.Document) time.Time {
var date time.Time
//<meta property="article:published_time" content="2019-08-23T13:20:57+00:00">
datePublished := strings.TrimSpace(document.Find("meta[property=\"article:published_time\"]").AttrOr("content", ""))
if datePublished != "" {
date, _ = time.Parse(time.RFC3339, datePublished)
}
return date
}

8
api/comando_torrents.go
View File

@@ -159,7 +159,10 @@ func getTorrents(ctx context.Context, i *Indexer, link string) ([]IndexedTorrent
text := s.Text()
audio = append(audio, findAudioFromText(text)...)
year = findYearFromText(text, title)
y := findYearFromText(text, title)
if y != "" {
year = y
}
size = append(size, findSizesFromText(text)...)
})
@@ -283,6 +286,9 @@ func findYearFromText(text string, title string) (year string) {
}
if year == "" {
fmt.Println("DEBUG: year not found in text, trying to find in title")
fmt.Println("DEBUG: title:", title)
fmt.Println("DEBUG: text:", text)
re = regexp.MustCompile(`\((\d{4})\)`)
yearMatch := re.FindStringSubmatch(title)
if len(yearMatch) > 0 {