86 lines
2.6 KiB
YAML
86 lines
2.6 KiB
YAML
name: Babes&Stars
|
|
performerByName:
|
|
action: scrapeXPath
|
|
queryURL: https://www.babesandstars.com/search/?t=models&q={}
|
|
scraper: performerSearch
|
|
performerByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- https://www.babesandstars.com
|
|
scraper: performerScraper
|
|
|
|
xPathScrapers:
|
|
performerSearch:
|
|
common:
|
|
$result: //div[@class="search models"]//div[@class="item"]/div[@class="thumb"]/a
|
|
performer:
|
|
Name: $result//div[@class="name"]/text()
|
|
URL:
|
|
selector: $result/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^
|
|
with: https://www.babesandstars.com
|
|
|
|
performerScraper:
|
|
common:
|
|
$result: //div[@class="features"]
|
|
performer:
|
|
Name: //div[@class="profile"]//h1/text()
|
|
URL:
|
|
selector: //link[@rel="canonical"]/@content
|
|
postProcess:
|
|
- replace:
|
|
- regex: "^http://"
|
|
with: "https://"
|
|
Twitter: //a[@class="twitter"]/@href
|
|
Birthdate:
|
|
selector: $result//span[contains(text(),'Birthday:')]/em/a/text()
|
|
concat: " "
|
|
postProcess:
|
|
- parseDate: 02 Jan 2006
|
|
Ethnicity:
|
|
selector: $result//span[contains(text(),'Ethnicity:')]/em/a/text()
|
|
Country:
|
|
selector: $result//span[contains(text(),'Country:')]/em/a/text()
|
|
HairColor:
|
|
selector: $result//span[contains(text(),'Hair:')]/em/a/text()
|
|
EyeColor:
|
|
selector: $result//span[contains(text(),'Eyes:')]/em/a/text()
|
|
Height:
|
|
selector: $result//span[contains(text(),'Height:')]/em/a/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^(\d+).+$
|
|
with: "$1"
|
|
Weight:
|
|
selector: $result//span[contains(text(),'Weight:')]/em/a/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: " kg"
|
|
with:
|
|
Measurements:
|
|
selector: $result//span[contains(text(),'Measurements:')]/em/a/text()|$result//span[contains(text(),'Cup:')]/em/a/text()
|
|
concat: "-"
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^([0-9]{2})-([0-9]{2})-([0-9]{2})-(AA?|[BC]|D{1,4}|[E-I])$
|
|
with: $1$4-$2-$3
|
|
Aliases:
|
|
selector: //div[@class="aliases"]/em/text()
|
|
Image:
|
|
selector: //div[@class="profile"]//div[@class="thumb"]/img/@src
|
|
Gender:
|
|
fixed: Female
|
|
Details:
|
|
selector: //div[@class="bio read_more"]//p/text()
|
|
concat: "\n\n"
|
|
Tags:
|
|
Name:
|
|
selector: $result//span[contains(text(),'Cup:')]/em/a/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^
|
|
with: "Cup "
|
|
# Last Updated November 19, 2021
|