79 lines
2.2 KiB
YAML
79 lines
2.2 KiB
YAML
name: DorcelClub
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- dorcelclub.com/en/scene
|
|
- dorcelclub.com/scene
|
|
scraper: sceneScraper
|
|
|
|
movieByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- dorcelclub.com/en/porn-movie
|
|
scraper: movieScraper
|
|
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
scene:
|
|
Title: //meta[@property="og:title"]/@content
|
|
Details: //meta[@property="og:description"]/@content
|
|
Image: //img[@class="thumbnail lazyload"]/@data-src
|
|
Date:
|
|
selector: //span[@class="publish_date"]
|
|
postProcess:
|
|
- parseDate: January 02, 2006
|
|
Performers:
|
|
Name: //div[@class="actress"]/a
|
|
Studio:
|
|
Name:
|
|
fixed: Dorcel Club
|
|
Movies:
|
|
Name: //div[@class="left"]/span[@class="movie"]/a/text()
|
|
URL:
|
|
selector: //div[@class="left"]/span[@class="movie"]/a/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^/
|
|
with: https://dorcelclub.com/
|
|
Director:
|
|
selector: //div[@class="left"]/span[@class='director']/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: Director\s*:\s*(.*)
|
|
with: $1
|
|
|
|
movieScraper:
|
|
movie:
|
|
Name: //img[contains(@class, "cover")]/@alt
|
|
Duration:
|
|
selector: //span[@class='duration']/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: ([0-9]+)h\s*([0-9]+)?
|
|
with: "$1:$2:00"
|
|
- regex: ([0-9]+)m\s*([0-9]+)?
|
|
with: "0:$1:$2"
|
|
- regex: "^:"
|
|
with: "0:"
|
|
- regex: ":$"
|
|
with: ":00"
|
|
- regex: "::"
|
|
with: ":00:"
|
|
Studio:
|
|
Name:
|
|
fixed: Dorcel Club
|
|
Director:
|
|
selector: //span[@class='director']/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: Director\s*:\s*(.*)
|
|
with: $1
|
|
FrontImage:
|
|
selector: //img[contains(@class, "cover")]/@data-src
|
|
postProcess:
|
|
- replace:
|
|
- regex: ([^\s]*)\s1x
|
|
with: $1
|
|
Synopsis: //span[@class="full"]/p|/div[@class="content-text"]/p
|
|
# Last Updated March 22, 2023
|