Merge pull request #5125 from nonamethanks/booth-support

Add Booth support

app/logical/source/extractor.rb

@@ -27,6 +27,7 @@ module Source
     DOWNLOAD_TIMEOUT = 60
 
     attr_reader :url, :referer_url, :parsed_url, :parsed_referer
+
     delegate :site_name, to: :parsed_url
 
     SUBCLASSES = [
@@ -50,6 +51,7 @@ module Source
       Source::Extractor::Plurk,
       Source::Extractor::Tinami,
       Source::Extractor::Fantia,
+      Source::Extractor::Booth,
     ]
 
     # Should return true if the extractor is configured correctly. Return false

app/logical/source/extractor/booth.rb

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+# @see Source::URL::Booth
+class Source::Extractor
+  class Booth < Source::Extractor
+    def match?
+      Source::URL::Booth === parsed_url
+    end
+
+    def image_urls
+      if parsed_url.image_url?
+        if parsed_url.full_image_url?
+          [parsed_url.to_s]
+        else
+          [find_right_extension(parsed_url)]
+        end
+      else
+        page&.css(".market-item-detail-item-image")&.pluck("data-origin").to_a.compact
+      end
+    end
+
+    def profile_url
+      if page.present?
+        page.at(".summary [data-product-list*='shop_index']")&.[]("href")&.chomp("/")
+      else
+        parsed_url.profile_url || parsed_referer&.profile_url
+      end
+    end
+
+    def artist_name
+      return nil unless profile_url.present?
+      Source::URL.parse(profile_url)&.username
+    end
+
+    def display_name
+      page&.at(".summary .user-avatar")&.[]("alt")
+    end
+
+    def other_names
+      [display_name].compact
+    end
+
+    def artist_commentary_title
+      page&.at(".summary .u-tpg-title1")&.text
+    end
+
+    def artist_commentary_desc
+      page&.at(".autolink")&.to_html
+    end
+
+    def dtext_artist_commentary_desc
+      DText.from_html(artist_commentary_desc)
+    end
+
+    def tags
+      page&.css(".item-info-detail [data-product-list*='tag_category_search']").to_a.map do |element|
+        [element.text.gsub(/ x .*/, ""), element["href"]]
+      end
+    end
+
+    def page_url
+      parsed_url.page_url || parsed_referer&.page_url
+    end
+
+    def page
+      return nil if parsed_url.page_url.blank?
+
+      resp = http.cache(1.minute).cookies(adult: "t").get(page_url)
+      return nil if resp.code != 200
+
+      resp.parse
+    end
+    memoize :page
+
+    def find_right_extension(parsed_url)
+      extensions = %w[png jpg jpeg]
+      candidates = extensions.map { |ext| parsed_url.full_image_url_for(ext) }
+
+      chosen_url = candidates.find { |candidate| http_exists?(candidate) }
+      chosen_url || parsed_url.to_s
+    end
+  end
+end

app/logical/source/url/booth.rb

@@ -1,15 +1,13 @@
 # frozen_string_literal: true
 
-# Unhandled:
-#
-# https://booth.pximg.net/c/128x128/users/3193929/icon_image/5be9eff4-1d9e-4a79-b097-33c1cd4ad314_base_resized.jpg (profile icon)
-# https://s2.booth.pm/8bb9e4e3-d171-4027-88df-84480480f79d/3d70de06-8e7c-444e-b8eb-a8a95bf20638.png (profile cover image)
+# Unhandled
+# https://booth.pm/downloadables/1376468 (from https://booth.pm/en/items/2425521, requires pixiv login to download)
 
 module Source
   class URL::Booth < Source::URL
     RESERVED_SUBDOMAINS = ["www", "s", "s2", "asset", "accounts", nil]
 
-    attr_reader :work_id, :user_id, :username
+    attr_reader :work_id, :user_id, :user_uuid, :username
 
     def self.match?(url)
       url.domain == "booth.pm" || url.host == "booth.pximg.net"
@@ -21,6 +19,7 @@ module Source
       # https://booth.pximg.net/8bb9e4e3-d171-4027-88df-84480480f79d/i/2864768/00cdfef0-e8d5-454b-8554-4885a7e4827d_base_resized.jpg (full)
       # https://booth.pximg.net/c/300x300_a2_g5/8bb9e4e3-d171-4027-88df-84480480f79d/i/2864768/00cdfef0-e8d5-454b-8554-4885a7e4827d_base_resized.jpg (thumb)
       # https://booth.pximg.net/c/72x72_a2_g5/8bb9e4e3-d171-4027-88df-84480480f79d/i/2864768/00cdfef0-e8d5-454b-8554-4885a7e4827d_base_resized.jpg (thumb)
+      # https://booth.pximg.net/8bb9e4e3-d171-4027-88df-84480480f79d/i/2864768/00cdfef0-e8d5-454b-8554-4885a7e4827d.jpeg (full)
       #
       # https://s2.booth.pm/b242a7bd-0747-48c4-891d-9e8552edd5d7/i/3746752/52dbee27-7ad2-4048-9c1d-827eee36625c_base_resized.jpg (sample)
       # https://booth.pximg.net/b242a7bd-0747-48c4-891d-9e8552edd5d7/i/3746752/52dbee27-7ad2-4048-9c1d-827eee36625c.jpg (full)
@@ -32,6 +31,20 @@ module Source
         @work_id = work_id
         @file = file
 
+      # profile icons
+      # https://booth.pximg.net/c/128x128/users/3193929/icon_image/5be9eff4-1d9e-4a79-b097-33c1cd4ad314_base_resized.jpg (sample)
+      # https://booth.pximg.net/users/3193929/icon_image/5be9eff4-1d9e-4a79-b097-33c1cd4ad314.png (full)
+      in _, _, *, "users", user_id, "icon_image", file
+        @user_id = user_id
+        @file = file
+
+      # profile cover images
+      # https://s2.booth.pm/8bb9e4e3-d171-4027-88df-84480480f79d/3d70de06-8e7c-444e-b8eb-a8a95bf20638_base_resized.jpg (sample)
+      # https://s2.booth.pm/8bb9e4e3-d171-4027-88df-84480480f79d/3d70de06-8e7c-444e-b8eb-a8a95bf20638.png (full)
+      in _, _, *, /\h{8}-\h{4}-\h{4}-\h{4}-\h{12}/i => user_uuid, file
+        @user_uuid = user_uuid
+        @file = file
+
       # https://booth.pm/en/items/2864768
       # https://booth.pm/ja/items/2864768
       in _, "booth.pm", _, "items", work_id
@@ -53,7 +66,25 @@ module Source
     end
 
     def image_url?
-      url.host == "booth.pximg.net"
+      url.host.in?(["booth.pximg.net", "s2.booth.pm"])
+    end
+
+    def full_image_url?
+      image_url? && @file.exclude?("_base_resized")
+    end
+
+    def full_image_url_for(extension)
+      return unless @file.present?
+      full_file = @file.gsub(/_base_resized\.\w+$/, ".#{extension}")
+      if user_uuid
+        if work_id
+          "https://#{host}/#{user_uuid}/i/#{work_id}/#{full_file}"
+        else
+          "https://#{host}/#{user_uuid}/#{full_file}"
+        end
+      elsif user_id
+        "https://#{host}/users/#{user_id}/icon_image/#{full_file}"
+      end
     end
 
     def page_url

app/models/artist_url.rb

@@ -95,8 +95,8 @@ class ArtistURL < ApplicationRecord
   def priority
     sites = %w[
       Pixiv Twitter
-      ArtStation Baraag BCY Deviant\ Art Hentai\ Foundry Fantia Foundation Lofter Nico\ Seiga Nijie Pawoo Fanbox Pixiv\ Sketch Plurk Tinami Tumblr Weibo
-      Ask.fm Booth Facebook FC2 Gumroad Instagram Ko-fi Livedoor Mihuashi Mixi.jp Patreon Piapro.jp Picarto Privatter Sakura.ne.jp Stickam Skeb Twitch Youtube
+      ArtStation Baraag BCY Booth Deviant\ Art Hentai\ Foundry Fantia Foundation Lofter Nico\ Seiga Nijie Pawoo Fanbox Pixiv\ Sketch Plurk Tinami Tumblr Weibo
+      Ask.fm Facebook FC2 Gumroad Instagram Ko-fi Livedoor Mihuashi Mixi.jp Patreon Piapro.jp Picarto Privatter Sakura.ne.jp Stickam Skeb Twitch Youtube
       Amazon Circle.ms DLSite Doujinshi.org Erogamescape Mangaupdates Melonbooks Toranoana Wikipedia
     ]