40 lines
1012 B
YAML
40 lines
1012 B
YAML
name: taboopov
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- taboopov.com/videos/
|
|
scraper: sceneScraper
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
scene:
|
|
Title:
|
|
selector: //div[@class='title']/text()
|
|
Performers:
|
|
Name:
|
|
selector: //a[starts-with(@href,'/models')]/text()
|
|
Details:
|
|
selector: //div[@class='title']//following-sibling::p/text()
|
|
concat: "\n\n"
|
|
Tags:
|
|
Name:
|
|
selector: //a[starts-with(@href,'/keywords')]/text()
|
|
Image:
|
|
selector: //img[@class='photo']/@src
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^
|
|
with: https://taboopov.com
|
|
Studio:
|
|
Name:
|
|
fixed: TabooPOV
|
|
Date:
|
|
selector: //td[contains(.,"Date Added:")]/text()
|
|
concat: " "
|
|
postProcess:
|
|
- replace:
|
|
- regex: .*(?:Date Added:)\s(\d{2}/\d{2}/\d{4}).*
|
|
with: $1
|
|
- parseDate: 01/02/2006
|
|
|
|
# Last Updated October 30, 2020
|