name: TheMovieDB

movieByURL:
  - action: scrapeXPath
    url:
      - themoviedb.org
    scraper: movieScraper
sceneByURL:
  - action: scrapeXPath
    url:
      - themoviedb.org
    scraper: sceneScraper

xPathScrapers:
  movieScraper:
    movie:
      Name: (//h2/a)[1]
      Director: //div[@class='header_info']/ol[@class='people no_image']/li[@class='profile']/p[1]/a
      Duration:
        selector: //span[@class='runtime']
        postProcess:
          - replace:
              - regex: "h "
                with: ":"
              - regex: "m"
                with: ":00"
      Date:
        selector: //span[@class='release']
        postProcess:
          - replace:
              - regex: \(.*$
                with:
          - parseDate: 01/02/2006
      Synopsis: //div[@class='header_info']/div[@class='overview']/p
      URL: //meta[@name='og:url']/@content
      FrontImage:
        selector: (//img[@class='poster']/@src)[1]
        postProcess:
          - replace:
              - regex: w300
                with: w600
          - replace:
              - regex: h450
                with: h900
  sceneScraper:
    scene:
      Title: (//h2/a)[1]
      Details: //div[@class='header_info']/div[@class='overview']/p
      Director: //div[@class='header_info']/ol[@class='people no_image']/li[@class='profile']/p[1]/a
      Date:
        selector: //span[@class='release']
        postProcess:
          - replace:
              - regex: \(.*$
                with:
          - parseDate: 01/02/2006
      Studio:
        Name: //a[@label="Studio"]/text()
      Movies:
        Name: (//h2/a)[1]
        URL: //link[@rel="canonical"]/@href
        Director: //div[@class='header_info']/ol[@class='people no_image']/li[@class='profile']/p[1]/a
        Duration:
          selector: //small[contains(text(), "Length")]/following-sibling::text()
          postProcess:
            - replace:
                - regex: " hrs. "
                  with: ":"
                - regex: " mins."
                  with: ":00"
        Date:
          selector: //span[@class='release']
          postProcess:
            - replace:
                - regex: \(.*$
                  with:
            - parseDate: 01/02/2006
        Synopsis: //div[@class='header_info']/div[@class='overview']/p
        FrontImage:
          selector: (//img[@class='poster']/@src)[1]
          postProcess:
            - replace:
                - regex: w300
                  with: w600
            - replace:
                - regex: h450
                  with: h900
      Performers:
        Name: //ol[@class='people scroller']/li[@class='card']/p[1]
      URL: //meta[@name='og:url']/@content

driver:
  headers:
    - Key: "User-Agent"
      Value: stash-scraper/1.0.0
# Last Updated March 11, 2024