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

name: BluMedia Network
sceneByURL:
  - action: scrapeXPath
    url:
      - boygusher.com/play/
    scraper: bgScraper
  - action: scrapeXPath
    url:
      - brokestraightboys.com/play/
    scraper: bsbScraper
  - action: scrapeXPath
    url:
      - collegeboyphysicals.com/play/
    scraper: cbpScraper
  - action: scrapeXPath
    url:
      - collegedudes.com/play/
    scraper: cdScraper
performerByURL:
  - action: scrapeXPath
    url:
      # Performers from before ~August 2016 have no details
      - boygusher.com/modelpage
    scraper: bgScraper
  - action: scrapeXPath
    url:
      - brokestraightboys.com/model/
    scraper: bsbScraper
  - action: scrapeXPath
    url:
      - collegeboyphysicals.com/modelpage
    scraper: cbpScraper
  - action: scrapeXPath
    url:
      - collegedudes.com/model/
    scraper: cdScraper

xPathScrapers:
  bgScraper:
    scene:
      Title: //div[@class="dettl"]
      # This is a bad selector and should be improved
      Details: //div[@class="desc"]
      Image: (//div[@class="player"]//img)[1]/@src
      Studio: &studioFromTitle
        Name:
          selector: //title/text()
          postProcess:
            - replace:
                - regex: ( - .*)$
                  with:
      Performers:
        Name: //div[contains(@class, "model-desc")]//a[contains(@href, "modelpage")]/text()
        URL:
          selector: //div[contains(@class, "model-desc")]//a[contains(@href, "modelpage")]/@href
          postProcess:
            - replace:
                - regex: ^
                  with: https://boygusher.com
    performer:
      Name: //div[@class="dettl"]
      Image: //div[@class="modeltn"]/img/@src
      Height:
        selector: //div[@class="model-dtls"]//li[contains(text(), "Height")]/strong
        postProcess: &extractMetric
          - replace:
              - regex: .*\((\d*).*
                with: $1
      Weight:
        selector: //div[@class="model-dtls"]//li[contains(text(), "Weight")]/strong
        postProcess:
          - lbToKg: true
      PenisLength:
        selector: //div[@class="model-dtls"]//li[contains(text(), "Cock")]/strong
        postProcess: *extractMetric
      Circumcised:
        selector: //div[@class="model-dtls"]//li[contains(text(), "Cock")]/strong
        postProcess:
          - replace:
              - regex: .*\)\s*(.*)
                with: $1
  bsbScraper:
    scene:
      Title: (//header/following-sibling::div/div/div/text())[1]
      # This is a bad selector and should be improved
      Details:
        selector: //div[@class="dtlp"]/p
        concat: "\n\n"
      Image: (//div[@id="player"]//img)[1]/@src
      Studio: *studioFromTitle
      Performers:
        Name: //span/dl/a/text()
        URL:
          selector: //span/dl/a/@href
          postProcess:
            - replace:
                - regex: ^
                  with: https://brokestraightboys.com
    performer:
      Name: (//div[@class="deTlt"]/text())[1]
      Image: (//div[@class="wdcnt"]//img)[1]/@src
      Height:
        selector: //div[@class="wdcnt"]//li[contains(., "Height")]/text()
        postProcess: &extractMetric
          - replace:
              - regex: .*\((\d*).*
                with: $1
      Weight:
        selector: //div[@class="wdcnt"]//li[contains(., "Weight")]/text()
        postProcess:
          - lbToKg: true
      PenisLength:
        selector: //div[@class="wdcnt"]//li[contains(., "Cock")]/text()
        postProcess: *extractMetric
      Circumcised:
        selector: //div[@class="wdcnt"]//li[contains(., "Cock")]/text()
        postProcess:
          - replace:
              - regex: .*\)\s*(.*)
                with: $1
  cbpScraper:
    scene:
      Title: //div[@class="dettl"]
      Image: (//div[@class="player"]//img)[1]/@src
      Details: //div[@class="desc"]
      Studio: *studioFromTitle
      Performers:
        Name: //div[contains(@class, "model-desc")]//a[contains(@href, "modelpage")]/text()
        URL:
          selector: //div[contains(@class, "model-desc")]//a[contains(@href, "modelpage")]/@href
          postProcess:
            - replace:
                - regex: ^
                  with: https://collegeboyphysicals.com
    performer:
      Name: //div[@class="dettl"]
      Image: //div[@class="modeltn"]//img[1]/@src
      Height:
        selector: //div[@class="model-dtls"]//li[contains(., "Height")]/strong
        postProcess: *extractMetric
      Weight:
        selector: //div[@class="model-dtls"]//li[contains(., "Weight")]/strong
        postProcess:
          - lbToKg: true
      PenisLength:
        selector: //div[@class="model-dtls"]//li[contains(., "Cock")]/strong
        postProcess: *extractMetric
      Circumcised:
        selector: //div[@class="model-dtls"]//li[contains(., "Cock")]/strong
        postProcess:
          - replace:
              - regex: .*\)\s*(.*)
                with: $1
  cdScraper:
    scene:
      Title: //div[contains(@class, "bluBar")]//strong
      # This is a bad selector and should be improved
      Details:
        selector: //div[@class="descD"]//p[string-length(text()) > 1]
        concat: "\n\n"
      Image: (//div[@class="player"]//img)[1]/@src
      Studio: *studioFromTitle
      Performers:
        Name: //div[@class="eimg"]//a/text()
        # This site does not link to their performers, but they DO have profiles
    performer:
      Name: //div[contains(@class, "bluBar")]//strong
      Image: //div[@class="modelPic"]//img[1]/@src
      Height:
        selector: //div[@class="modelDtls"]//li[contains(., "Height")]/text()
        postProcess: *extractMetric
      Weight:
        selector: //div[@class="modelDtls"]//li[contains(., "Weight")]/text()
        postProcess:
          - lbToKg: true
      PenisLength:
        selector: //div[@class="modelDtls"]//li[contains(., "Cock")]/text()
        postProcess: *extractMetric
      Circumcised:
        selector: //div[@class="modelDtls"]//li[contains(., "Cock")]/text()
        postProcess:
          - replace:
              - regex: .*\)\s*(.*)
                with: $1
# Last Updated October 02, 2023