79 lines
2.3 KiB
YAML
79 lines
2.3 KiB
YAML
name: "GroobyClub"
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
# checked for all sites https://www.grooby.club/tour/
|
|
- asianamericantgirls.com
|
|
- canada-tgirl.com
|
|
- euro-tgirls.com
|
|
- hazel-tucker.com
|
|
- krissy4u.com
|
|
- russian-tgirls.com
|
|
- tgirl40.com
|
|
- tgirlbbw.com
|
|
- tgirlpornstar.com
|
|
- tgirlpostop.com
|
|
- transexdomination.com
|
|
- ts-castingcouch.com
|
|
- uk-tgirls.com
|
|
scraper: substudioScraper
|
|
- action: scrapeXPath
|
|
url:
|
|
- grooby.club
|
|
# other grooby sites which work
|
|
- tgirljapan.com
|
|
- tgirljapanhardcore.com
|
|
scraper: clubScraper
|
|
|
|
xPathScrapers:
|
|
clubScraper:
|
|
scene:
|
|
Title: &title //div[@class="trailer_toptitle_left"]
|
|
Date: &date
|
|
selector: //b[contains(.,"Added")]/following-sibling::text()[1]
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^-
|
|
with: ""
|
|
- parseDate: January 2, 2006
|
|
Details: &details
|
|
selector: //div[@class="trailerpage_info"]/p/text()
|
|
concat: " "
|
|
Performers: &performers
|
|
Name: //div[@class="setdesc"]//a[contains(@href, "models")]/text()
|
|
Studio:
|
|
Name: //div[@class="sitename"]/a/text()
|
|
URL: &url //link[@rel="canonical"]/@href
|
|
Image: &image
|
|
selector: //meta[@property="og:image"]/@content
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^//
|
|
with: https://
|
|
Tags:
|
|
Name: &tagName //div[@class="set_tags"]/ul/li//a/text()
|
|
|
|
substudioScraper:
|
|
scene:
|
|
Title: *title
|
|
Date: *date
|
|
Details: *details
|
|
Performers: *performers
|
|
Studio:
|
|
Name: //meta[@name="author"]/@content
|
|
URL: *url
|
|
Image: *image
|
|
# Tags for these subsites only appear on grooby.club as of 2023-08-15
|
|
# but we have to extend the subScraper functionality in Stash
|
|
# if we want to be able to scrape more than just a single field
|
|
# TODO: write a python scraper, merge with GroobyNetwork-*.yml ?
|
|
Tags:
|
|
Name:
|
|
selector: //link[@rel="canonical"]/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^.+/tour
|
|
with: https://grooby.club/tour
|
|
- subScraper: *tagName
|
|
# Last Updated August 21, 2023
|