name: JavHD
sceneByURL:
  - action: scrapeXPath
    url:
      - javhd.com/en/id/
      - javhd.com/en/studio/
    scraper: javhd
performerByURL:
  - action: scrapeXPath
    url:
      - javhd.com/en/model/
    scraper: javhd
xPathScrapers:
  javhd:
    common:
      $script: //script[@type="application/ld+json"]/text()
    scene:
      Details: //h2[@class="content__desc"]
      Performers:
        Name: //div[@class="content-info"]/a
      Image:
        selector: $script
        postProcess:
          - replace:
              - regex: .+thumbnail":"([^"]+)",.+
                with: $1
      Date: # not to be trusted, they update the dates
        selector: $script
        postProcess:
          - replace:
              - regex: .+datePublished":"([^"]+)",.+
                with: $1
          - parseDate: "2006-01-02T15:04:05-07:00"
      Title:
        selector: //h1[@class="content__title"]
      Tags:
        Name: //div[@class="content-desc"]//a
      Studio:
        Name:
          fixed: JavHD
      URL: //link[@rel="canonical"]/@href

    performer:
      Name: //h1
      Image: //div[@class="avatar"]/img/@src
      URL: //link[@rel="canonical"]/@href
      Details: //div[@class="info"]/p
      Birthdate:
        selector: //strong[contains(., "Birth date")]/following-sibling::*
        postProcess:
          - parseDate: 2 January 2006
      EyeColor: //strong[contains(., "Eye color")]/following-sibling::*
      HairColor: //strong[contains(., "Hair color")]/following-sibling::*
      Height:
        selector: //strong[contains(., "Height")]/following-sibling::*
        postProcess:
          - replace:
              - regex: (\d+).*
                with: $1
      Weight: //strong[contains(., "Weight")]/following-sibling::*
      Ethnicity: //strong[contains(., "Ethnicity")]/following-sibling::*
      FakeTits: //strong[contains(., "Breast factor")]/following-sibling::*
# Last Updated April 05, 2024