Add Fantia support

Also fixes a regression in 74fdeef10c
that stopped mastodon urls from being given the right priority.

app/logical/artist_finder.rb

@@ -46,6 +46,8 @@ module ArtistFinder
     "facebook.com", # https://www.facebook.com/LuutenantsLoot
     "fantia.jp", # http://fantia.jp/no100
     "fantia.jp/fanclubs", # https://fantia.jp/fanclubs/1711
+    "fantia.jp/posts", # https://fantia.jp/posts/20000
+    "fantia.jp/products", # https://fantia.jp/products/10000
     "fav.me", # http://fav.me/d9y1njg
     /blog-imgs-\d+(?:-origin)?\.fc2\.com/i,
     %r{blog\.fc2\.com(/\w)+/?}i, # http://blog71.fc2.com/a/abk00/file/20080220194219.jpg

app/logical/source/url.rb

@@ -23,6 +23,7 @@ module Source
       Source::URL::ArtStation,
       Source::URL::DeviantArt,
       Source::URL::Fanbox,
+      Source::URL::Fantia,
       Source::URL::Foundation,
       Source::URL::HentaiFoundry,
       Source::URL::Lofter,

app/logical/source/url/fantia.rb

@@ -0,0 +1,90 @@
+# frozen_string_literal: true
+
+# Unparsed:
+# https://fantia.jp/asanagi
+
+class Source::URL::Fantia < Source::URL
+  attr_reader :full_image_url
+
+  def self.match?(url)
+    url.domain == "fantia.jp"
+  end
+
+  def parse
+    case [host, *path_segments]
+
+    # posts:
+    # https://c.fantia.jp/uploads/post/file/1070093/main_16faf0b1-58d8-4aac-9e86-b243063eaaf1.jpeg (sample)
+    # https://c.fantia.jp/uploads/post/file/1070093/16faf0b1-58d8-4aac-9e86-b243063eaaf1.jpeg
+    # https://cc.fantia.jp/uploads/post_content_photo/file/4563389/main_a9763427-3ccd-4e51-bcde-ff5e1ce0aa56.jpg?Key-Pair-Id=APKAIOCKYZS7WKBB6G7A&Policy=eyJTdGF0ZW1lbnQiOlt7IlJlc291cmNlIjoiaHR0cHM6Ly9jYy5mYW50aWEuanAvdXBsb2Fkcy9wb3N0X2NvbnRlbnRfcGhvdG8vZmlsZS80NTYzMzg5L21haW5fYTk3NjM0MjctM2NjZC00ZTUxLWJjZGUtZmY1ZTFjZTBhYTU2LmpwZyIsIkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTY0NjkxMzk3OH19fV19&Signature=jyW5ankfO9uCHlKkozYU9RPpO3jzKTW2HuyXgS81i~cRgrXcI9orYU0IXuiit~0TznIyXbB7F~6Z790t7lX948PYAb9luYIREJC2u7pRMP3OBbsANbbFE0o4VR-6O3ZKbYQ4aG~ofVEZfiFVGoKoVtdJxj0bBNQV29eeFylGQATkFmywne1YMtJMqDirRBFMIatqNuunGsiWCQHqLYNHCeS4dZXlOnV8JQq0u1rPkeAQBmDCStFMA5ywjnWTfSZK7RN6RXKCAsMTXTl5X~I6EZASUPoGQy2vHUj5I-veffACg46jpvqTv6mLjQEw8JG~JLIOrZazKZR9O2kIoLNVGQ__
+    # from file download: https://cc.fantia.jp/uploads/post_content/file/1830956/cbcdfcbe_20220224_120_040_100.png?Key-Pair-Id=APKAIOCKYZS7WKBB6G7A&Policy=eyJTdGF0ZW1lbnQiOlt7IlJlc291cmNlIjoiaHR0cHM6Ly9jYy5mYW50aWEuanAvdXBsb2Fkcy9wb3N0X2NvbnRlbnQvZmlsZS8xODMwOTU2L2NiY2RmY2JlXzIwMjIwMjI0XzEyMF8wNDBfMTAwLnBuZyIsIkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTY0NjkxNDU4Nn19fV19&Signature=d1nw8gs9vcshIAeEH4oESm9-7z6y4A7MfoIRRvtUtV9iqTNA8KM0ORuCI7NwEoYc1VHsxy9ByeuSBpNaJoknnc3TOmHFhVRcLn~OWpnWqiHEPpMcSEG7uGlorysjEPmYYRGHjE7LJYcWiiJxjZ~fSBbYzxxwsjroPm-fyGUtNhdJWEMNp52vHe5P9KErb7M8tP01toekGdOqO-pkWm1t9xm2Tp5P7RWcbtQPOixgG4UgOhE0f3LVwHGHYJV~-lB5RjrDbTTO3ezVi7I7ybZjjHotVUK5MbHHmXzC1NqI-VN3vHddTwTbTK9xEnPMR27NHSlho3-O18WcNs1YgKD48w__
+    #
+    # products:
+    # https://c.fantia.jp/uploads/product/image/249638/main_fd5aef8f-c217-49d0-83e8-289efb33dfc4.jpg
+    # https://c.fantia.jp/uploads/product_image/file/219407/main_bd7419c2-2450-4c53-a28a-90101fa466ab.jpg (sample)
+    # https://c.fantia.jp/uploads/product_image/file/219407/bd7419c2-2450-4c53-a28a-90101fa466ab.jpg
+    in _, "uploads", image_type, ("file" | "image"), image_id, /(?:\w+_)?([\w-]+\.\w+)/ => file
+      #  post_id/product_id == image_id only for the first image in a post/product
+      case image_type
+      when "post"
+        @post_id = image_id
+        @full_image_url = "https://c.fantia.jp/uploads/post/file/#{@post_id}/#{$1}"
+      when "product"
+        @product_id = image_id
+        @full_image_url = "https://c.fantia.jp/uploads/product/image/#{@product_id}/#{$1}"
+      when "product_image"
+        @full_image_url = "https://c.fantia.jp/uploads/product_image/file/#{image_id}/#{$1}"
+      else
+        @full_image_url = original_url
+      end
+
+    # https://fantia.jp/posts/1143951/download/1830956
+    in _, "posts", post_id, "download", image_id
+      @post_id = post_id
+      @download_id = image_id
+
+    # https://fantia.jp/posts/1148334
+    in _, "posts", /\d+/ => post_id
+      @post_id = post_id
+
+    # https://fantia.jp/products/249638
+    in _, "products", /\d+/ => product_id
+      @product_id = product_id
+
+    # https://fantia.jp/fanclubs/64496
+    # https://fantia.jp/fanclubs/1654/posts
+    in _, "fanclubs", /\d+/ => fanclub_id, *rest
+      @fanclub_id = fanclub_id
+
+    else
+    end
+  end
+
+  def image_url?
+    @full_image_url.present?
+  end
+
+  def downloadable?
+    @download_id.present?
+  end
+
+  def page_url
+    if @post_id.present?
+      "https://fantia.jp/posts/#{@post_id}"
+    elsif @product_id.present?
+      "https://fantia.jp/products/#{@product_id}"
+    end
+  end
+
+  def work_id
+    @post_id || @product_id
+  end
+
+  def work_type
+    if @post_id.present?
+      "post"
+    elsif @product_id.present?
+      "product"
+    end
+  end
+end

app/logical/sources/strategies.rb

@@ -24,6 +24,7 @@ module Sources
         Strategies::Foundation,
         Strategies::Plurk,
         Strategies::TwitPic,
+        Strategies::Fantia,
       ]
     end
 

app/logical/sources/strategies/fantia.rb

@@ -0,0 +1,171 @@
+# frozen_string_literal: true
+
+module Sources::Strategies
+  class Fantia < Base
+    def self.enabled?
+      Danbooru.config.fantia_session_id.present?
+    end
+
+    def match?
+      Source::URL::Fantia === parsed_url
+    end
+
+    def site_name
+      parsed_url.site_name
+    end
+
+    def image_urls
+      return [parsed_url.full_image_url] if parsed_url.image_url?
+      return [image_from_downloadable(parsed_url)] if parsed_url.downloadable?
+
+      images = images_for_post.presence || images_for_product.presence || []
+
+      full_images = images.compact.map do |image|
+        parsed = Source::URL.parse(image)
+        if parsed&.image_url?
+          parsed.full_image_url
+        elsif parsed&.downloadable?
+          image_from_downloadable(parsed)
+        else
+          image
+        end
+      end
+      full_images.compact.uniq
+    end
+
+    def image_from_downloadable(url)
+      resp = http.head(url)
+      return url if resp.status != 200
+      resp.uri.to_s
+    end
+
+    def images_for_post
+      return [] unless api_response.present?
+      images = [api_response.dig("post", "thumb_micro")]
+      api_response.dig("post", "post_contents").to_a.map do |content|
+        next if content["visible_status"] != "visible"
+
+        case content["category"]
+        when "photo_gallery"
+          content["post_content_photos"].to_a.map { |i| images << i.dig("url", "original") }
+        when "file"
+          images << image_from_downloadable("https://www.fantia.jp/#{content["download_uri"]}")
+        when "blog"
+          begin
+            sub_json = JSON.parse(content["comment"])
+          rescue Json::ParserError
+            sub_json = {}
+          end
+          sub_json["ops"].to_a.map { |js| images << js.dig("insert", "fantiaImage", "url") }
+        end
+      end
+      images
+    end
+
+    def images_for_product
+      html_response&.css(".product-gallery-item .img-fluid").to_a.map do |element|
+        element["src"] unless element["src"] =~ %r{/fallback/}
+      end.compact
+    end
+
+    def page_url
+      parsed_url.page_url || parsed_referer&.page_url
+    end
+
+    def tags
+      case work_type
+      when "post"
+        api_response&.dig("post", "tags").to_a.map do |tag|
+          [tag["name"], "https://fantia.jp/posts?tag=#{tag["name"]}"]
+        end
+      when "product"
+        html_response&.css(".product-category a").to_a.map do |element|
+          tag_name = element.text.delete_prefix("#")
+          [tag_name, "https://fantia.jp/products?product_category=##{tag_name}"]
+        end
+      else
+        []
+      end
+    end
+
+    def other_names
+      case work_type
+      when "post"
+        [api_response&.dig("post", "fanclub", "creator_name")].compact
+      when "product"
+        [html_response&.at(".fanclub-name a")&.text].compact
+      end
+    end
+
+    def profile_url
+      case work_type
+      when "post"
+        fanclub_id = api_response&.dig("post", "fanclub", "id")
+        return unless fanclub_id.present?
+        "https://fantia.jp/fanclubs/#{fanclub_id}"
+      when "product"
+        href = html_response&.at(".fanclub-name a")&.[]("href")
+        return unless href.present?
+        URI.join("https://fantia.jp/", href).to_s
+      end
+    end
+
+    def artist_commentary_title
+      case work_type
+      when "post"
+        api_response&.dig("post", "title")
+      when "product"
+        html_response&.at(".product-title")&.text
+      end
+    end
+
+    def artist_commentary_desc
+      case work_type
+      when "post"
+        api_response&.dig("post", "comment")
+      when "product"
+        html_response&.at(".product-description")&.text
+      end
+    end
+
+    def dtext_artist_commentary_desc
+      DText.from_html(artist_commentary_desc)
+    end
+
+    def normalize_for_source
+      page_url
+    end
+
+    def work_type
+      parsed_url.work_type || parsed_referer&.work_type
+    end
+
+    def work_id
+      parsed_url.work_id || parsed_referer&.work_id
+    end
+
+    def api_response
+      return {} unless work_type == "post"
+      api_url = "https://fantia.jp/api/v1/posts/#{work_id}"
+
+      response = http.cache(1.minute).get(api_url)
+      return {} unless response.status == 200
+
+      JSON.parse(response)
+    rescue JSON::ParserError
+      {}
+    end
+
+    def html_response
+      return nil unless work_type == "product"
+      response = http.cache(1.minute).get("https://fantia.jp/products/#{work_id}")
+
+      return nil unless response.status == 200
+      response.parse
+    end
+
+    def http
+      Danbooru::Http.new.cookies(_session_id: Danbooru.config.fantia_session_id)
+    end
+  end
+end