130 lines
3.5 KiB
YAML
130 lines
3.5 KiB
YAML
name: "StasyQ"
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- stasyq.com/r/
|
|
scraper: sceneScraper
|
|
performerByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- stasyq.com/m/
|
|
scraper: performerScraper
|
|
performerByName:
|
|
action: scrapeXPath
|
|
queryURL: https://www.stasyq.com/search/{}
|
|
scraper: performerSearch
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
common:
|
|
$script: //script[contains(.,'VideoObject')]/text()
|
|
$performer: //div[@class="release-card"]//a[@class="custom-link"]
|
|
scene:
|
|
Title: //h1/text()
|
|
Details: //div[@class="about-section__text"]
|
|
Date:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:uploadDate":\s")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
- parseDate: 2006-01-02
|
|
Image:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:thumbnailUrl":\s")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Studio:
|
|
Name:
|
|
fixed: StasyQ
|
|
Tags:
|
|
Name: //a[contains(@href,'/search/')]//text()
|
|
|
|
Performers:
|
|
Name: $performer//text()
|
|
URL: $performer/@href
|
|
URL: //link[@rel='canonical']/@href
|
|
performerSearch:
|
|
performer:
|
|
Name: //div[@class="model-preview-card__info"]//a[contains(@href,'/m/')]/text()
|
|
URL:
|
|
selector: //div[@class="model-preview-card__info"]//a[contains(@href,'/m/')]/@href
|
|
performerScraper:
|
|
common:
|
|
$script: //script[contains(.,'schema')]/text()
|
|
performer:
|
|
Name:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:name":\s")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Aliases:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:alternateName":.")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Gender:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:gender"\s:\s")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Country:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:birthPlace":\s")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Height:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:height":\s")(\d+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Weight:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:weight":\s")(\d+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Image:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+(?:image":\s")([^"]+).+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
Details:
|
|
selector: $script
|
|
postProcess:
|
|
- replace:
|
|
- regex: (\\u2019|'|’)
|
|
with: "'"
|
|
- regex: \\
|
|
with:
|
|
- regex: .+(?:description":")(.+)",".+
|
|
with: $1
|
|
- regex: .+schema\.org.+
|
|
with:
|
|
URL: //link[@rel='canonical']/@href
|
|
# Last Updated July 17, 2022
|