46 lines
1.4 KiB
YAML
46 lines
1.4 KiB
YAML
name: "Bellesa"
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- bellesafilms.com/scene/
|
|
- bellesahouse.com/scene/
|
|
scraper: sceneScraper
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
scene:
|
|
Title: //h1/text()
|
|
Date:
|
|
selector: //span[div[text()="Release Date:"]]/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: "Release Date:"
|
|
with:
|
|
- parseDate: January 2, 2006
|
|
Details: //span/div[text()="Description:"]/following-sibling::text()
|
|
Tags:
|
|
Name:
|
|
selector: //span[div[text()='Categories:']]/a/text()[1]
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^\s*\,|\,\s*$
|
|
with:
|
|
Performers:
|
|
Name: //h1/following-sibling::div//a[contains(@href,"/model")]/text()
|
|
Studio:
|
|
Name:
|
|
selector: //link[@rel="canonical"]/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: .*www\.(.+)\.com.*
|
|
with: $1
|
|
- map:
|
|
bellesafilms: Bellesa Films
|
|
bellesahouse: Bellesa House
|
|
Image:
|
|
selector: //img[@alt=""][contains(@src,"poster")]/@src|//script[@type="application/ld+json"]/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: '.*"thumbnailUrl": "([^"]+)".*'
|
|
with: $1
|
|
# Last Updated February 03, 2024
|