96 lines
2.8 KiB
YAML
96 lines
2.8 KiB
YAML
name: Kink Men
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- kinkmen.com
|
|
scraper: sceneScraper
|
|
|
|
sceneByFragment:
|
|
action: scrapeXPath
|
|
queryURL: https://www.kinkmen.com/shoot/{filename}
|
|
# constructs the scene URL from the filename, provided that the filename includes the scene id
|
|
queryURLReplace:
|
|
filename:
|
|
# the id in kink.com is a 1-6 digit number
|
|
- regex: ^(\d+)[^\d].* # support filenames in the form 12345_performer_other_data.mp4
|
|
with: $1
|
|
- regex: .*\((\d+)\)\.[a-zA-Z\d]+$ #support filenames in the form scene - date - performer (12345).mp4
|
|
with: $1
|
|
scraper: sceneScraper
|
|
performerByName:
|
|
action: scrapeXPath
|
|
queryURL: https://www.kinkmen.com/models?q={}
|
|
scraper: performerSearch
|
|
performerByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- kinkmen.com/model
|
|
- kink.com/model
|
|
scraper: performerScraper
|
|
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
common:
|
|
$shoot: //div[contains(@class, "shoot-detail-legend")]
|
|
scene:
|
|
Title: //h1
|
|
Date:
|
|
selector: ($shoot/span)[2]
|
|
postProcess:
|
|
- parseDate: 01/02/06
|
|
Details:
|
|
selector: //div[contains(@class, "shoot-detail-description")]/span/p/text()
|
|
concat: "\n"
|
|
Code: //div[contains(@class, "shootPage")]/@data-shootid
|
|
Director: //span[contains(@class, "director-name")]
|
|
Performers:
|
|
Name:
|
|
selector: //h1/following-sibling::*/a[contains(@href, "/model/")]
|
|
postProcess:
|
|
- replace:
|
|
- regex: \,
|
|
with: ""
|
|
Tags:
|
|
Name:
|
|
selector: //div[contains(@class, "shoot-detail-description")]//a[contains(@href, "/tag/")]
|
|
postProcess:
|
|
- replace:
|
|
- regex: \,
|
|
with: ""
|
|
Image: //video/@poster
|
|
Studio:
|
|
Name:
|
|
selector: ($shoot/span)[1]
|
|
performerSearch:
|
|
common:
|
|
$result: //a[contains(@class, "ratio-model")]
|
|
performer:
|
|
Name: $result/following-sibling::span
|
|
Image: $result/img/@src
|
|
URL:
|
|
selector: $result/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^
|
|
with: https://www.kinkmen.com
|
|
performerScraper:
|
|
performer:
|
|
Name: //h1
|
|
Twitter: //div/a[contains(@href, "twitter.com")]/@href
|
|
Image: //div[contains(@class, "kink-slider-images")]/img/@data-src
|
|
Tags:
|
|
Name: //a[contains(@href, "/tag/")]
|
|
Details:
|
|
selector: //p[@class="bio"]/following-sibling::p
|
|
concat: "\n\n"
|
|
postProcess:
|
|
- replace:
|
|
- regex: "<a[^>]*>"
|
|
with: ""
|
|
URL: //link[@rel="canonical"]/@href
|
|
driver:
|
|
headers:
|
|
- Key: User-Agent
|
|
Value: stash-scraper/1.0.0
|
|
# Last Updated November 06, 2023
|