compose-projects-arr/stash/config/scrapers/community/Baberotica/Baberotica.yml

name: Baberotica
sceneByURL:
  - action: scrapeXPath
    url: &urls
      - avidolz.com
      - baberotica.com 
      - baberoticavr.com
      - nucosplay.com
      - suckmevr.com
      - teenthais.com
      - vrpornpass.com
    scraper: sceneScraper
performerByURL:
  - action: scrapeXPath
    url: *urls
    scraper: performerScraper
xPathScrapers:
  sceneScraper:
    scene:
      Title:
        selector: //h5/a | //h1[@itemprop="name"] | //div[@class="rel"]/img/@alt
      Performers:
        Name: //p/span[@itemprop="actor"]/a/span | //div[@class="cat"][1]/a/text()
        URL: //div[@class="cat"][1]/a/@href | //span[@itemprop="actor"]/a/@href
      Date:
      # vrpornpass.com does not currently provide the scene published date on the non-paid site
        selector: //div[contains(@class, "pure-u-1")]//meta[@itemprop="datePublished"]/@content | //div[@class="video-date"]
        postProcess:
          - replace:
              - regex: .*Added:\s*(.*)\s*$
                with: $1
              - regex: "T.+$"
                with: ""
              - regex: (\d+)(st|nd|rd|th)
                with: "$1"
          - parseDate: 2006-01-02
          - parseDate: January 02, 2006
          - parseDate: 2006-01-02T15:04:05-07:00
      Tags:
        Name: //a[@itemprop="genre"] | //div[@class="cat"][2]/a
      Details:
        selector: //div[@itemprop="description"] | //p[span[@class="readmore"]]
        postProcess:
          - replace:
              - regex: ... Read More
                with:
      Image:
        selector: //div[contains(@class,"pure-u-1")]/meta[@itemprop="thumbnailUrl"]/@content | //div[@id="videohtml5tour"]//img[@class="pure-img"]/@src | //div[@id="videohtml5tour"]/video/@poster | //div[@class="rel"]//img/@src
        postProcess:
          - replace:
              - regex: ^//
                with: https://
      URL: 
        selector: //link[@rel="canonical"][1]/@href | //link[@rel="alternate"][1]/@href | //div[@class="title"]/h5/a/@href
        postProcess:
          - replace:
            - regex: \s*(.*)\s*feed/$
              with: $1
      Studio:
        Name:
          selector: //meta[@itemprop="url"]/@content
          postProcess:
            - map:
                https://avidolz.com/: AvIdolz
                https://baberotica.com/: Baberotica
                https://baberoticavr.com/: BaberoticaVR
                https://nucosplay.com/: Nu Cosplay
                https://suckmevr.com/: SuckMeVR
                https://teenthais.com/: TeenThais
                https://vrpornpass.com/: VR PornPass
  performerScraper:
    common:
      $profileBE: //div[@class="model-profile"]
      $profileTT: //div[@class="pure-u-1-4 pure-u-sm-1-2 pure-u-xs-1"]
      $profilePP: //div[@class="pure-u-1-4 pure-u-sm-1-2 pure-u-xs-1-3"]
    performer:
      Name:
        selector: $profileBE[contains(strong, "Name:")]//text() | //meta[@itemprop="title"]/@content | //div[@class="model-info"]//h1[@itemprop="name"]/text()
        postProcess:
          - replace:
              - regex: .*Name:\s*(.*)\s*$
                with: $1
              - regex: None
                with:
      Aliases:
        selector: $profileBE[contains(strong, "Alias name:")]//text()
        postProcess:
          - replace:
              - regex: .*Alias name:\s*(.*)\s*$
                with: $1
              - regex: None
                with:
      Gender:
      # These sites only have profiles for female performers, so using fixed: Female
        fixed: Female
      Birthdate:
        selector: $profileBE[contains(strong, "Birth date:")]//text() | $profilePP[contains(strong, "Birth date:")]/following-sibling::*[1]/text()
        postProcess:
          - replace:
              - regex: .*Birth date:\s*(.*)\s*$
                with: $1
          - parseDate: January 2, 2006
          - parseDate: 2006-01-02
      Country:
        selector: $profileBE[contains(strong, "Country:")]//text() | $profileTT[contains(strong, "Ethnicity:")]/following-sibling::*[1]/text() | $profilePP[contains(strong, "Country:")]/following-sibling::*[1]/text() | //div[@class="video-info"]/meta[@content="Japanese idol"]/@content
        postProcess:
          - replace:
              - regex: .*Country:\s*(.*)\s*$
                with: $1
              - regex: "None|Other Country"
                with:
              - regex: ^Thai$
                with: Thailand
              - regex: ^Japanese idol$
                with: Japan
      Ethnicity:
        selector: $profileBE[contains(strong, "Ethnicity:")]//text() | $profilePP[contains(strong, "Ethnicity:")]/following-sibling::*[1]/text() | $profileTT[contains(strong, "Ethnicity:")]/following-sibling::*[1]/text() | //div[@class="video-info"]/meta[@content="Japanese idol"]/@content
        postProcess:
          - replace:
              - regex: .*Ethnicity:\s*(.*)\s*$
                with: $1
              - regex: None
                with:
              - regex: ^Thai$
                with: Asian
              - regex: ^Japanese idol$
                with: Asian
      Weight:
        selector: $profileBE[contains(strong, "Weight:")]//text()
        postProcess:
          - replace:
              - regex: .*Weight:\s*(.*)\s*$
                with: $1
              - regex: (\d+)kg
                with: $1
              - regex: None
                with:
      Height:
        selector: $profileBE[contains(strong, "Height:")]//text() | $profileTT[contains(strong, "Height:")]/following-sibling::*[1]/text()
        postProcess:
          - replace:
              - regex: .*Height:\s*(.*)\s*$
                with: $1
              - regex: ^cm$
                with:
              - regex: (\d+)cm.*
                with: $1
              - regex: None
                with:
      HairColor:
        selector: $profileBE[contains(strong, "Hair color:")]//text() | $profileTT[contains(strong, "Hair color:")]/following-sibling::*[1]/text()  | $profilePP[contains(strong, "Hair color:")]/following-sibling::*[1]/text()
        postProcess:
          - replace:
              - regex: (?i).*hair color:?\s*(.*)\s*$
                with: $1
              - regex: None|N/A
                with:
      EyeColor:
        selector: $profileBE[contains(strong, "Eye color:")]//text() | $profileTT[contains(strong, "Eye color:")]/following-sibling::*[1]/text() | $profilePP[contains(strong, "Eye Color:")]/following-sibling::*[1]/text()
        postProcess:
          - replace:
              - regex: (?i).*eye color:?\s*(.*)\s*$
                with: $1
              - regex: None|N/A
                with:
      Piercings:
        selector: $profileBE[contains(strong, "Piercings:")]//text() | $profilePP[contains(strong, "Piercings:")]/following-sibling::*[1]/text()
        postProcess:
          - replace:
              - regex: .*Piercings:\s*(.*)\s*$
                with: $1
              - regex: "None|No Piercings|^No$"
                with:
      Tattoos:
        selector: $profileBE[contains(strong, "Tattoo:")]//text() | $profilePP[contains(strong, "Tattoos:")]/following-sibling::*[1]/text()
        postProcess:
          - replace:
              - regex: .*Tattoo:\s*(.*)\s*$
                with: $1
              - regex: "None$|^No$|(?i)^No tattoo$"
                with:
      Instagram:
        selector: $profileBE[contains(strong, "Instagram:")]/a/@href
      Twitter:
        selector: $profileBE[contains(strong, "Twitter:")]/a/@href
      Details:
        selector: //p[@itemprop="description"]//text() | //p[span[@class="readmore"]]
        postProcess:
          - replace:
              - regex: ... Read More
                with:
      URL: 
        selector: //link[@rel="canonical"][1]/@href | //link[@rel="alternate"][1]/@href | //div[@class="model-info"]/a/@href
        postProcess:
          - replace:
            - regex: \s*(.*)\s*feed/$
              with: $1
      Image:
        selector: //div[@class="model-photo"]/img[@class="rounded"]/@src | //div[@class="pure-u-1-4 pure-u-sm-1-3 pure-u-xs-1-2"]//img/@src | //div[@class="pure-u-1-4 pure-u-sm-1-3 pure-u-xs-1"]//img/@src
        postProcess:
          - replace:
              - regex: ^//
                with: https://
              - regex: 460x640
                with: 690x960
              - regex: 270x480
                with: 405x720
# Last Updated October 10, 2023