54 lines
1.8 KiB
YAML
54 lines
1.8 KiB
YAML
name: Inn of Sin
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- adreenawinters.com
|
|
- bbctitans.com
|
|
- dannyoceansadventures.com
|
|
- mydeepdarksecret.com
|
|
- puerto-rock.com
|
|
- rebelrhyder.com
|
|
- richardmannsworld.com
|
|
- tatsandtits.com
|
|
- trans4thefans.com
|
|
- wefuckin.com
|
|
scraper: sceneScraper
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
common:
|
|
$scene: //div[@class="vid-box" or @class="vidBox"]
|
|
scene:
|
|
Title: $scene//h1 | $scene//h2
|
|
Details:
|
|
selector: //*[contains(text(), "About") or contains(text(), "Description")]/following-sibling::p | //span[@itemprop="about"]/p
|
|
concat: "\n\n"
|
|
Image: //video/@poster
|
|
URL: //link[@rel="canonical"]/@href
|
|
Tags:
|
|
Name: $scene//span[@itemprop="keywords"]/a
|
|
Performers:
|
|
# Doesn't work on adreenawinters.com since they don't have model profiles
|
|
# we can fix this but let's wait to see if anyone needs it first
|
|
Name: $scene//a[contains(@href, "/pornstars/")]
|
|
Studio:
|
|
Name:
|
|
selector: //link[@rel="canonical"]/@href
|
|
postProcess:
|
|
- replace:
|
|
# Get the domain name without any subdomains
|
|
# https://regex101.com/r/JxFd9a/1
|
|
- regex: ^(?:https:\/\/[\w\.]*?)([^.]+)\.com.*$
|
|
with: $1
|
|
- map:
|
|
adreenawinters: Adreena Winters
|
|
bbctitans: BBC Titans
|
|
dannyoceansadventures: Danny Ocean's Adventures
|
|
mydeepdarksecret: My Deep Dark Secret
|
|
puerto-rock: Puerto Rock
|
|
rebelrhyder: Rebel Rhyder
|
|
richardmannsworld: Richard Mann's World
|
|
tatsandtits: Tats and Tits
|
|
trans4thefans: Trans 4 the Fans
|
|
wefuckin: We Fuckin
|
|
# Last Updated March 10, 2024
|