name: R18.dev (JAV)

sceneByFragment:
  action: scrapeJson
  queryURL: https://r18.dev/videos/vod/movies/detail/-/dvd_id={filename}/json
  queryURLReplace:
    filename:
      # gets just the JAV ID out of the filename. This also removes the file extension which is pretty nice.
      # You can have your filename be something like "Something Something ABC123 Something Something.mp4" and it will scrape as ABC123.
      - regex: (.*[^a-zA-Z0-9])*([a-zA-Z-]+\d+)(.+)
        with: $2
  scraper: sceneSearchIndirect

sceneByURL:
  - action: scrapeJson
    url:
      - r18.dev/videos/vod/movies/detail/-/i
    scraper: sceneScraper
    queryURL: "https://r18.dev/videos/vod/movies/detail/-/combined={url}/json"
    queryURLReplace:
      url:
        - regex: ".+/id=(.+)/?$"
          with: "$1"

sceneByName:
  action: scrapeJson
  scraper: sceneSearch
  queryURL: "https://r18.dev/videos/vod/movies/detail/-/dvd_id={}/json"

sceneByQueryFragment:
  action: scrapeJson
  queryURL: "{url}"
  scraper: sceneScraper

movieByURL:
  - action: scrapeJson
    url:
      - r18.dev/videos/vod/movies/detail/-/i
    queryURL: "https://r18.dev/videos/vod/movies/detail/-/combined={url}/json"
    queryURLReplace:
      url:
        - regex: ".+/id=(.+)/?$"
          with: "$1"
    scraper: movieScraper

jsonScrapers:
  sceneScraper:
    scene:
      Title:
        selector: dvd_id
      Date:
        selector: release_date
      Performers:
        Name:
          selector: actresses.#.name_romaji
        Aliases:
          selector: actresses.#.name_kanji
      Image:
        selector: jacket_full_url
      Director:
        selector: directors.#.name_romaji
      Details:
        selector: title_en
        # All this uncensoring stuff happens automatically on the HTML version of the page due to some JS on the frontend there.
        # I put in a request to the site dev to add this to the backend json api as well so we don't have to do it everytime ourselves plus keep a ever growing list of this stuff
        postProcess: &ppUncensor
          - replace:
              - regex: A\*{3}ed\b
                with: "Abused"
              - regex: A\*{3}e\b
                with: "Abuse"
              - regex: A\*{3}es\b
                with: "Abuses"
              - regex: A\*{4}p\b
                with: "Asleep" # ZMEN-037
              - regex: A\*{5}t
                with: "Assault"
              - regex: A\*{5}ted\b
                with: "Assaulted"
              - regex: B\*{5}p\b
                with: "Bang Up" # Not Sure
              - regex: B\*{5}k\b # IPX-374
                with: "Berserk"
              - regex: B\*{3}d\b
                with: "Blood"
              - regex: B\*{3}dy\b
                with: "Bloody" # Not Sure
              - regex: B\*{6}y\b
                with: "Brutally"
              - regex: Chai\*{1}saw\b
                with: "Chainsaw"
              - regex: C\*{3}d\b
                with: "Child"
              - regex: C\*{3}dcare\b
                with: "Childcare"
              - regex: C\*{3}dhood\b
                with: "Childhood"
              - regex: C\*{3}dish\b
                with: "Childish" # MMYM-045
              - regex: C\*{3}dren\b
                with: "Children"
              - regex: C\*{1}ck\b
                with: "Cock"
              - regex: C\*{1}cks\b
                with: "Cocks"
              - regex: C\*{1}llegiate\b
                with: "Collegiate"
              - regex: C\*{5}y\b
                with: "Cruelty"
              - regex: CrumB\*{2}d\b
                with: "Crumbled"
              - regex: D\*{1}ck\b
                with: "Dick"
              - regex: D\*{6}e\b
                with: "Disgrace"
              - regex: D\*{6}ed\b
                with: "Disgraced"
              - regex: D\*{6}eful\b
                with: "Disgraceful"
              - regex: D\*{3}king\b
                with: "Drinking"
              - regex: D\*{3}ks\b
                with: "Drinks"
              - regex: D\*{2}g\b
                with: "Drug"
              - regex: D\*{2}gged\b
                with: "Drugged"
              - regex: D\*{2}gs\b
                with: "Drugs"
              - regex: D\*{3}k\b
                with: "Drunk"
              - regex: D\*{3}ken\b
                with: "Drunken"
              - regex: D\*{3}kest\b
                with: "Drunkest"
              - regex: EnS\*{3}ed\b
                with: "Enslaved"
              - regex: F\*{3}e\b
                with: "Force"
              - regex: F\*{3}eful\b
                with: "Forceful"
              - regex: F\*{3}efully\b
                with: "Forcefully"
              - regex: F\*{3}es\b
                with: "Forces" # Not Sure
              - regex: F\*{3}ed\b
                with: "Fucked"
              - regex: F\*{5}g\b
                with: "Fucking" # SSNI-391
              - regex: G\*{9}d\b
                with: "Gang-Banged"
              - regex: G\*{6}g\b
                with: "Gangbang" # STAR-976
              - regex: G\*{7}g\b
                with: "Gangbang"
              - regex: G\*{6}ged\b
                with: "Gangbanged" # SSNI-242
              - regex: G\*{7}ged\b
                with: "Gangbanged"
              - regex: G\*{7}gers\b
                with: "Gangbangers"
              - regex: G\*{6}ging\b
                with: "Gangbanging"
              - regex: G\*{7}ging\b
                with: "Gangbanging"
              - regex: G\*{7}gs\b
                with: "Gangbangs"
              - regex: Half-A\*{4}p\b
                with: "Half-Asleep" # ZMEN-037
              - regex: HumB\*{2}d\b
                with: "Humbled"
              - regex: H\*{9}n\b
                with: "Humiliation"
              - regex: H\*{2}t\b
                with: "Hurt"
              - regex: H\*{2}ts\b
                with: "Hurts"
              - regex: H\*{7}m\b
                with: "Hypnotism"
              - regex: H\*{7}ed\b
                with: "Hypnotized" # PPPD-376
              - regex: I\*{4}t\b
                with: "Incest"
              - regex: I\*{4}tuous\b
                with: "Incestuous"
              - regex: I\*{4}ts\b
                with: "Insults"
              - regex: J\*{1}\b
                with: "Jo" # Not Sure
              - regex: J\*{1}s\b
                with: "Jos" # Not Sure
              - regex: K\*{1}d\b
                with: "Kid"
              - regex: K\*{1}dding\b
                with: "Kidding"
              - regex: K\*{4}pped\b
                with: "Kidnapped"
              - regex: K\*{4}pper\b
                with: "Kidnapper"
              - regex: K\*{4}pping\b
                with: "Kidnapping"
              - regex: K\*{1}ds\b
                with: "Kids"
              - regex: K\*{2}l\b
                with: "Kill"
              - regex: K\*{2}led\b
                with: "Killed" # SNIS-036
              - regex: K\*{2}ler\b
                with: "Killer" # Not Sure
              - regex: K\*{2}ling\b
                with: "Killing"
              - regex: Lol\*{1}pop\b
                with: "Lolipop"
              - regex: Lo\*{2}ta\b
                with: "Lolita"
              - regex: Ma\*{1}ko\b
                with: "Maiko"
              - regex: M\*{4}t\b
                with: "Molest"
              - regex: M\*{4}tation\b
                with: "Molestation"
              - regex: M\*{4}ted\b
                with: "Molested"
              - regex: M\*{4}ter\b
                with: "Molester"
              - regex: M\*{4}ters\b
                with: "Molesters"
              - regex: M\*{4}ting\b
                with: "Molesting"
              - regex: M\*{4}tor\b
                with: "Molestor"
              - regex: P\*{4}h\b
                with: "Punish"
              - regex: P\*{4}hed\b
                with: "Punished"
              - regex: P\*{4}hment\b
                with: "Punishment"
              - regex: P\*{1}ssy\b
                with: "Pussy"
              - regex: R\*{2}e\b
                with: "Rape"
              #- regex: R\*{1}pe\b
              #  with: "Rape" # Can be Rope ? IPX-311
              - regex: R\*{2}ed\b
                with: "Raped"
              - regex: R\*{1}ped\b
                with: "Raped"
              - regex: R\*{2}es\b
                with: "Rapes"
              - regex: R\*{4}g\b
                with: "Raping"
              - regex: S\*{9}l\b
                with: "School Girl" # Not Sure
              - regex: S\*{9}ls\b
                with: "School Girls" # SSNI-296
              - regex: S\*{8}l\b
                with: "Schoolgirl"
              - regex: Sch\*{2}lgirl\b
                with: "Schoolgirl"
              - regex: S\*{9}ls\b
                with: "Schoolgirls" # Not Sure (PPPD-811)
              - regex: S\*{8}ls\b
                with: "Schoolgirls"
              - regex: S\*{9}s\b
                with: "Schoolgirls" # NFDM-234
              - regex: Sch\*{2}lgirls\b
                with: "Schoolgirls"
              - regex: SK\*{2}led\b
                with: "Skilled"
              - regex: SK\*{2}lful\b
                with: "Skillful"
              - regex: SK\*{2}lfully\b
                with: "Skillfully"
              - regex: SK\*{2}ls\b
                with: "Skills"
              - regex: S\*{3}e\b
                with: "Slave"
              - regex: S\*{3}ery\b
                with: "Slavery"
              - regex: S\*{3}es\b
                with: "Slaves"
              - regex: S\*{6}g\b
                with: "Sleeping"
              - regex: StepB\*{16}r\b
                with: "StepBrother And Sister" # Not Sure
              - regex: StepK\*{1}ds \b
                with: "StepKids"
              - regex: StepM\*{12}n\b
                with: "StepMother And Son" # GVG-299
              - regex: S\*{5}t\b
                with: "Student"
              - regex: S\*{5}ts\b
                with: "Students"
              - regex: S\*{8}n\b
                with: "Submission"
              - regex: T\*{6}e\b
                with: "Tentacle" #MIDD-648
              - regex: T\*{6}es\b
                with: "Tentacles"
              - regex: T\*{5}e\b
                with: "Torture"
              - regex: T\*{5}ed\b
                with: "Tortured"
              - regex: T\*{5}es\b
                with: "Tortures" #MIDD-648
              - regex: U\*{9}sly\b
                with: "Unconsciously"
              - regex: U\*{7}g\b
                with: "Unwilling"
              - regex: V\*{5}e\b
                with: "Violate"
              - regex: V\*{1}olated\b
                with: "Violated"
              - regex: V\*{5}ed\b
                with: "Violated"
              - regex: V\*{5}es\b
                with: "Violates"
              - regex: V\*{6}e\b
                with: "Violence"
              - regex: V\*{5}t\b
                with: "Violent"
              - regex: Y\*{8}l\b
                with: "Young Girl" # Not Sure
              - regex: Y\*{8}ls\b
                with: "Young Girls" # Not Sure
      Studio:
        Name: maker_name_en
      Code:
        selector: dvd_id
      Tags:
        Name:
          selector: categories.#.name_en
          postProcess: *ppUncensor
      URL:
        selector: content_id
        postProcess:
          - replace:
              - regex: ^
                with: https://r18.dev/videos/vod/movies/detail/-/id=

  movieScraper:
    movie:
      Name:
        selector: dvd_id
      Aliases: title_ja
      Duration:
        selector: runtime_mins
        postProcess:
          - replace:
              - regex: $
                with: ":00"
      Date:
        selector: release_date
      FrontImage:
        selector: jacket_full_url
      Director:
        selector: directors.#.name_romaji
      Synopsis:
        selector: title_en
        postProcess: *ppUncensor
      Studio:
        Name: maker_name_en
      URL:
        selector: content_id
        postProcess:
          - replace:
              - regex: ^
                with: https://r18.dev/videos/vod/movies/detail/-/id=

  # Used for fragment scraping - the api is a little messy if we don't have the exact ID and need to do a search because the search result api doesn't have all the info we need.
  # So we get the ID from the search result page and make another search for it
  # Note I am not getting aliases yet for performers here as it didn't seem worth it to make that many subqueries. You should hopefully be matching them up with StashDB versions of the performers
  # or doing a performer search anyways.
  # I've also sent in a request to the dev of the site to make their API a little easier to use, so hopefully that goes through and we can have our cake and eat it too. :)
  sceneSearchIndirect:
    scene:
      # Title not on this page, so we need to fetch another page to get it
      Title:
        selector: content_id
        postProcess:
          - replace:
              - regex: (.+)
                with: https://r18.dev/videos/vod/movies/detail/-/combined=$1/json
          - subScraper:
              selector: dvd_id
      # Code not on this page, so we need to fetch another page to get it. It's the same as title. Surely there must be a way to only query once?
      Code:
        selector: content_id
        postProcess:
          - replace:
              - regex: (.+)
                with: https://r18.dev/videos/vod/movies/detail/-/combined=$1/json
          - subScraper:
              selector: dvd_id
      Details:
        selector: title
        postProcess: *ppUncensor
      Image: images.jacket_image.large2
      Director: director
      Date: release_date
      Tags:
        Name:
          selector: categories.#.name
          postProcess: *ppUncensor
      Studio:
        Name: maker.name
      Performers:
        Name: actresses.#.name
      URL:
        selector: content_id
        postProcess:
          - replace:
              - regex: (.+)
                with: https://r18.dev/videos/vod/movies/detail/-/id=$1/

  sceneSearch:
    scene:
      Title:
        selector: title
      Image:
        selector: images.jacket_image.large2
      Date:
        selector: release_date
      URL:
        selector: content_id
        postProcess:
          - replace:
              - regex: (.+)
                with: https://r18.dev/videos/vod/movies/detail/-/combined=$1/json
# Last Updated August 22, 2023