name: PegasProductions

# This studio has two formats for scenes depending on URL.  The less useful format has '/abonnements' to 
# start the path  while the more useful one does not.  Scenes appear to be available in both formats, but 
# there is no map between the resources referenced in each style of URL.

sceneByURL:
  - action: scrapeXPath
    url:
      - pegasproductions.com/abonnements
    scraper: abonnementsScraper

  - action: scrapeXPath
    url:
      - pegasproductions.com
    scraper: sceneScraper

xPathScrapers:
  abonnementsScraper:
    scene:
      Title: //meta[@itemprop="name"]/@content
      Date:
        selector: //meta[@itemprop="uploadDate"]/@content
        postProcess:
          - replace:
              - regex: ^([^T]+).+
                with: $1
      Performers:
        Name:
          selector: //h1
          postProcess:
            - replace:
                - regex: ^.+?([^:]*)$
                  with: $1
                - regex: '[^a-zA-Z0-9\s]'
                  with: '' 
      Image: //meta[@itemprop="thumbnailUrl"]/@content
      Studio:
        Name:
          fixed: Pegas Productions

  sceneScraper:
    scene:
      Title:
        selector: //span[@itemprop="name"]
      
      # This format has options for English and French language versions of the scene.  Depending on 
      # the language the values we key off for Performers and Director will be in English or French.
      Director: //p[contains(b,"Director") or contains(b,"Réalisateur")]/text()
      Performers:
        Name: //div[@class="span10"]/p[contains(b,"Starring") or contains(b,"Distribution")]/a

      Date:
        selector: //div[@id="date-duree"]/div[1]/p[1]
        # English URLs display dates in the format of 02/01/2006.  
        # French URLs display dates in the format of 01 janvier 2006, using the full french name of the month.
        # This bit of hackery converts the dates used in the French version to format used on the English page.
        postProcess:
          - replace:
              - regex: janvier
                with: "01"
              - regex: f[eé]vrier 
                with: "02"
              - regex: mars
                with: "03"
              - regex: avril
                with: "04"
              - regex: mai
                with: "05"
              - regex: juin
                with: "06"
              - regex: juillet
                with: "07"
              - regex: aout
                with: "08"
              - regex: septembre
                with: "09"
              - regex: octobre
                with: "10"
              - regex: novembre
                with: "11"
              - regex: d[eé]cembre
                with: "12"
              - regex: \s
                with: /
          - parseDate: 02/01/2006

      Tags:
        Name:
          selector: //div[@class="span10"]/p[contains(b,"Tags")]
          postProcess:
            - replace:
                - regex: "Tags:"
                  with:
          split: ", "
      Details:
        selector: //h5[@itemprop="description"]/p
      Image:
        selector: //script[contains(text(),"flowplayer(container")]/text()
        postProcess:
          - replace:
              - regex: .*?(https?:\/\/\S*\.jpg).*
                with: $1
      Studio:
        Name:
          fixed: Pegas Productions

driver:
  headers:
    - Key: User-Agent
      Value: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:79.0) Gecko/20100101 Firefox/79.0)
  cookies:
    - CookieURL: "https://www.pegasproductions.com/"
      Cookies:
        - Name: "langue"
          Domain: ".pegasproductions.com"
          Value: "en"
          Path: "/"
# Last Updated March 29, 2024