81 lines
2.6 KiB
YAML
81 lines
2.6 KiB
YAML
name: Fetishnetwork
|
||
sceneByURL:
|
||
- action: scrapeXPath
|
||
url:
|
||
- fetishnetwork.com/t2/showgal
|
||
scraper: sceneScraper
|
||
xPathScrapers:
|
||
sceneScraper:
|
||
scene:
|
||
# Title: //div[contains(@class,"detail-area")]//h3
|
||
Title:
|
||
selector: //title
|
||
postProcess:
|
||
- replace:
|
||
- regex: ^(.+)(?:\s\|).+$
|
||
with: $1
|
||
Details:
|
||
selector: |
|
||
//div[contains(@class,"description-long-text")]//text()
|
||
|//span[@class="vg_description-scene"]/text()
|
||
|//span[@class="strong"][contains(text(), "Description")]/following-sibling::text()
|
||
postProcess:
|
||
- replace:
|
||
- regex: Description:\s?
|
||
with:
|
||
- regex: <EFBFBD>
|
||
with:
|
||
Date:
|
||
selector: //h4[@class="light-grey"]|//span[@class="vg_dateadded"]|//span[contains(text(), "Added")]/text()
|
||
postProcess:
|
||
- replace:
|
||
- regex: Added:\s
|
||
with:
|
||
- parseDate: January 2, 2006
|
||
Image:
|
||
selector: |
|
||
//video/@poster
|
||
|//div[@class="banner-img-single"]/img/@src
|
||
|//div[@class="banner-img banner-img-one"]/img/@src
|
||
|//div[@class="display-panel"]/img/@src
|
||
|(//img[contains(@src,"content/flash/")]/@src)[1]
|
||
postProcess:
|
||
- replace:
|
||
- regex: ^
|
||
with: http://www.fetishnetwork.com/t2/
|
||
Studio:
|
||
Name:
|
||
selector: |
|
||
//span[@class="white-text"]
|
||
|//div[@id="header"][1]/h1/a/text()
|
||
|//span[@class="vg_attribute-title"][contains(text(), "Site")]/following-sibling::a[1]
|
||
postProcess:
|
||
- replace:
|
||
- regex: \sVideos
|
||
with:
|
||
- regex: \.com
|
||
with:
|
||
- regex: BDSM
|
||
with: Bdsm
|
||
- regex: FemDom
|
||
with: Femdom
|
||
- regex: "[A-Z][^A-Z]+|[A-Z]+$"
|
||
with: "$0 "
|
||
Performers:
|
||
Name:
|
||
# If the genre contains lesbian, split the scene title in case it contains a comma,
|
||
# otherwise (title does not contain names) extract performer name from image url
|
||
selector: |
|
||
//a[contains(text(),"Lesbian")]/following::h3/text()[contains(., ", ")]
|
||
|//video/@poster
|
||
|//div[@class="display-panel"]/img/@src
|
||
split: ", "
|
||
postProcess:
|
||
- replace:
|
||
- regex: ^.+\d_(.*?)(?:_\d+)?(?:_video/thumbs|_1//thumbs).+$
|
||
with: $1
|
||
- regex: _
|
||
with: " "
|
||
|
||
# Last Updated December 19, 2020
|