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

name: LoveHerFilms
sceneByURL:
  - action: scrapeXPath
    url:
      - loveherboobs.com/tour/trailers/
      - loveherfeet.com/tour/trailers/
      - loveherfilms.com/tour/trailers/
      - shelovesblack.com/tour/trailers/
    scraper: sceneScraper
sceneByFragment:
  action: scrapeXPath
  queryURL: "{url}"
  scraper: sceneScraper
performerByName:
  action: scrapeXPath
  queryURL: "https://www.loveherfilms.com/tour/search.php?model_name={}"
  scraper: performerSearch
performerByURL:
  - action: scrapeXPath
    url:
      - loveherboobs.com/tour/models/
      - loveherfeet.com/tour/models/
      - loveherfilms.com/tour/models/
      - shelovesblack.com/tour/models/
    scraper: performerScraper
xPathScrapers:
  sceneScraper:
    scene:
      Title: //h1[@class='title']/text()
      Details: //p[@class='description']/text()
      Date:
        selector: //div[@class='date']/text()
        postProcess:
          - parseDate: January 2, 2006
      Code:
        selector: //*/div[@class='photos vide-section']/a[1]/img/@data-src
        postProcess:
          - replace:
              - regex: ^https?.+/(?:[a-zA-Z]+_(\d+)_.+|(\d+)_[a-zA-Z0-9]+(?:\.jpg)?.*)
                with: $1$2
          - replace:
              - regex: ^\D.*
                with: 
      Image:
        selector: //div[@class='video']//img[contains(@class,'mainThumb') or contains(@class,'update_thumb')]/@src0_3x|//video/@poster|//base/@href
        concat: "|"
        postProcess:
          - replace:
              - regex: ^(https[^|]+)\|.+ # mainThumb urls do not need anything
                with: $1
              - regex: ^/tour([^|]+)\|(.+) # video/@poster urls need a domain
                with: $2$1
      Studio:
        Name:
          selector: //meta[@name='author']/@content
          postProcess:
            - map:
                LoveHerFeet.com: "Love Her Feet"
      Tags:
        Name: //div[@class='video-tags']/a/text()
      Performers:
        Name: //div[@class='featured']/a/text()
        URL: //div[@class='featured']/a/@href
  performerSearch:
    performer:
      Name: //h3/../a/@title
      URL: //h3/../a/@href
  performerScraper:
    performer:
      Name: //div[@class="bio"]//h2/text()
      Details:
        selector: //div[@class="description"]/div[@class="about"]/p/text()
        postProcess:
          - replace:
              - regex: (?i)^This\smodel\shas\sno\sbio.*
                with:
      Image:
        selector: //div[@class="bio"]//img/@src0_3x
      Ethnicity:
        selector: &bio //script[contains(text(),'totalStats')]/text()
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Ethnicity:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*) # if above is not matched clear everything
                with: ""
      EyeColor:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Eye\sColor:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
      Birthdate:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Date\sof\sBirth:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
          - parseDate: January 2, 2006
      Measurements:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Measurements:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
      HairColor:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Hair\sColor:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
      Height:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.+<p><span>Height:\s</span>\s(\d+[^(<]+).*
                with: $1
              - regex: (^..document..ready.*)
                with: ""
          - feetToCm: true
      Weight:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push.+<p><span>Weight:\s</span>\s(\d+)\D[^<]+</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
          - lbToKg: true
      FakeTits:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>(?:Tits|Boob)\sType:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
              - regex: (?i).*(enhanced|fake).* # comment out the below regexes if you want to keep more info
                with: "Yes"
              - regex: (?i).*natural.*
                with: "No"
              - regex: (?i)unknown
                with: ""
      Tattoos:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Tattoos:\s</span>([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
      Piercings:
        selector: *bio
        postProcess:
          - replace:
              - regex: .*totalStats\.push\(.<p><span>Piercings:\s</span>\s([^)]+)</p>.+
                with: $1
              - regex: (^..document..ready.*)
                with: ""
      Gender:
        fixed: "Female"
# Last Updated April 13, 2023