cineblog: fix scraper

This commit is contained in:
ilmich
2024-08-24 17:00:31 +02:00
parent e93b58895c
commit e0cc7d9edb

View File

@@ -93,7 +93,7 @@ def search(item, text):
@support.scrape @support.scrape
def peliculas(item): def peliculas(item):
# debug = True #debug = True
# esclusione degli articoli 'di servizio' # esclusione degli articoli 'di servizio'
# curYear = datetime.date.today().year # curYear = datetime.date.today().year
# blacklist = ['BENVENUTI', 'Richieste Serie TV', 'CB01.UNO ▶ TROVA L’INDIRIZZO UFFICIALE ', # blacklist = ['BENVENUTI', 'Richieste Serie TV', 'CB01.UNO ▶ TROVA L’INDIRIZZO UFFICIALE ',
@@ -113,7 +113,7 @@ def peliculas(item):
action = 'episodios' action = 'episodios'
elif '/serietv/' not in item.url: elif '/serietv/' not in item.url:
patron = r'(?<!sticky )hentry.*?<div class="card-image">\s*<a[^>]+>\s*<img src="(?P<thumb>[^" ]+)" alt[^>]+>[^>]+>[^>]+>[^>]+>[^>]+>[^>]+><a href="?(?P<url>[^" >]+)(?:\/|"|\s+)>(?P<title>[^<[(]+)(?:\[(?P<quality>[a-zA-Z]+(?:[/]?3D)?)\]\s*)?(?:\[(?P<lang>Sub-ITA|ITA)\]\s*)?(?:\[(?P<quality2>[a-zA-Z/]+)\]\s*)? (?:\((?P<year>[0-9]{4})\))?[^>]+>[^>]+>[^>]+>[^>]+>[^>]+>(?P<genre>[^<>&‖“]+)(?:[^ ]+\s*DURATA\s*(?P<duration>[0-9]+)[^>]+>[^>]+>[^>]+>(?P<plot>[^<>]+))?' patron = r'(?<!sticky )hentry.*?<div class="card-image">\s*<a[^>]+>\s*<img src="(?P<thumb>[^" ]+)" alt[^>]+>[^>]+>[^>]+>[^>]+>[^>]+>[^>]+><a href="?(?P<url>[^" >]+)(?:\/|"|\s+)>(?P<title>[^<[(]+)(?:\[(?P<quality>[a-zA-Z]+(?:[/]?3D)?)\]\s*)?(?:\[(?P<lang>Sub-ITA|ITA)\]\s*)?(?:\[(?P<quality2>[a-zA-Z/]+)\]\s*)? (?:\((?P<year>[0-9]{4})\))?[^>]+>[^>]+>[^>]+>[^>]+>(?P<genre>[^<>&‖“]+)(?:[^ ]+\s*DURATA\s*(?P<duration>[0-9]+)[^>]+>[^>]+>[^>]+>(?P<plot>[^<>]+))?'
action = 'findvideos' action = 'findvideos'
else: else: