Add skeb support

app/logical/sources/strategies.rb

@@ -15,7 +15,8 @@ module Sources
         Strategies::Moebooru,
         Strategies::HentaiFoundry,
         Strategies::Weibo,
-        Strategies::Newgrounds
+        Strategies::Newgrounds,
+        Strategies::Skeb
       ]
     end
 

app/logical/sources/strategies/base.rb

@@ -90,8 +90,6 @@ module Sources
           "Patreon"
         when /privatter\.net\z/i
           "Privatter"
-        when /skeb\.jp\z/i
-          "Skeb"
         when /tinami\.com\z/i
           "Tinami"
         when /toranoana\.(jp|shop)\z/i

app/logical/sources/strategies/skeb.rb

@@ -0,0 +1,107 @@
+# Image URLS
+## Non-watermarked:
+# * https://skeb.imgix.net/requests/199886_0?bg=%23fff&auto=format&w=800&s=5a6a908ab964fcdfc4713fad179fe715
+## Watermarked:
+# * https://skeb.imgix.net/requests/73290_0?bg=%23fff&auto=format&txtfont=bold&txtshad=70&txtclr=BFFFFFFF&txtalign=middle%2Ccenter&txtsize=150&txt=SAMPLE&w=800&s=4843435cff85d623b1f657209d131526
+## Full Size (found in commissioner_upload):
+# * https://skeb.imgix.net/requests/53269_1?bg=%23fff&fm=png&dl=53269.png&w=1.0&h=1.0&s=44588ea9c41881049e392adb1df21cce
+#
+# The signature is required and tied to the parameters. Doesn't seem like it's possible to reverse engineer it to remove the watermark, unfortunately.
+#
+# Page URLS
+# * https://skeb.jp/@OrvMZ/works/3 (non-watermarked)
+# * https://skeb.jp/@OrvMZ/works/1 (separated request and client's message after delivery. We can't get the latter)
+# * https://skeb.jp/@asanagi/works/16 (age-restricted, watermarked)
+# * https://skeb.jp/@asanagi/works/6 (private, returns 404)
+#
+# Profile URLS
+# Since skeb forces login through twitter, usernames are the same as twitter
+# * https://skeb.jp/@asanagi
+
+module Sources
+  module Strategies
+    class Skeb < Base
+      PROFILE_URL = %r{https?://(?:www\.)?skeb\.jp/@(?<artist_name>\w+)}i
+
+      PAGE_URL    = %r{#{PROFILE_URL}/works/(?<illust_id>\d+)}i
+
+      IMAGE_URL   = %r{https?://(?:www\.)?skeb\.imgix\.net/requests/[\d_]+\?.*}i
+
+      def domains
+        ["skeb.jp"]
+      end
+
+      def match?
+        return false if parsed_url.nil?
+        parsed_url.domain.in?(domains) || parsed_url.host == "skeb.imgix.net"
+      end
+
+      def site_name
+        "Skeb"
+      end
+
+      def image_urls
+        if url =~ IMAGE_URL
+          [url]
+        elsif page.present?
+          [page.text[/window\.__NUXT__=.*,preview_url:"(.*?)",/, 1].gsub("\\u002F", "/")]
+        else
+          []
+        end
+      end
+
+      def page_url
+        urls.map { |u| u if u =~ PAGE_URL }.compact.first
+      end
+
+      def normalize_for_source
+        page_url
+      end
+
+      def page
+        return if page_url.blank?
+        response = http.cache(1.minute).get(page_url)
+        return nil unless response.status == 200
+        # The status check is required for private commissions, which return 404
+
+        response.parse
+      end
+
+      def profile_url
+        return nil if artist_name.blank?
+        "https://skeb.jp/@#{artist_name}"
+      end
+
+      def artist_name
+        url[PROFILE_URL, :artist_name]
+      end
+
+      def display_name
+        page&.at("title")&.text&.match(/.*by (.*?) \| skeb/i).to_a[1]
+      end
+
+      def other_names
+        [display_name].compact.uniq
+      end
+
+      def artist_commentary_desc
+        # skeb "titles" are not needed: it's just the first few characters of the description
+        return if page.blank?
+        page.at("[property='og:description']")["content"]
+      end
+
+      def client_response
+        return if page.blank?
+        page.text[/window\.__NUXT__=.*,thanks:"(.*?)",/, 1]&.gsub(/\\n/, "\n")
+      end
+
+      def dtext_artist_commentary_desc
+        if client_response.present? && artist_commentary_desc.present?
+          "h5. Original Request:\n#{artist_commentary_desc}\n\nh5. Client Response:\n#{client_response}"
+        else
+          artist_commentary_desc
+        end
+      end
+    end
+  end
+end

test/unit/sources/skeb_test.rb

@@ -0,0 +1,75 @@
+require 'test_helper'
+
+module Sources
+  class SkebTest < ActiveSupport::TestCase
+    context "The source for a skeb picture" do
+      setup do
+        @site = Sources::Strategies.find("https://skeb.jp/@kai_chiisame/works/6")
+      end
+
+      should "get the artist name" do
+        assert_equal("kai_chiisame", @site.artist_name)
+      end
+
+      should "get the artist commentary" do
+        commentary = <<~COMM.chomp
+          初めまして、先日アピールを頂きましたのでリクエストさせて頂きます。
+
+          〇キャラ
+          　東方の東風谷早苗さん
+
+          〇内容
+          　・水着や薄着などの若干セクシーめ・肌色多めな方向性で、細部は絵師さんにお任せ
+          　・念のためNSFW指定にしましたがエロでなくていいです
+
+          ご検討お願いします。
+        COMM
+
+        assert_equal(commentary, @site.artist_commentary_desc)
+      end
+
+      should "get profile url" do
+        assert_equal("https://skeb.jp/@kai_chiisame", @site.profile_url)
+      end
+
+      should "get the image url" do
+        assert_equal("https://skeb.imgix.net/requests/229088_2?bg=%23fff&auto=format&w=800&s=9cac8b76c0838f2df4f19ebc41c1ae0a", @site.image_url)
+      end
+
+      should "get the canonical url" do
+        assert_equal("https://skeb.jp/@kai_chiisame/works/6", @site.canonical_url)
+      end
+
+      should "find the correct artist" do
+        artist = FactoryBot.create(:artist, name: "kai_chiisame", url_string: @site.url)
+        assert_equal([artist], @site.artists)
+      end
+
+      should "not fail" do
+        assert_nothing_raised { @site.to_h }
+      end
+    end
+
+    context "A private or non-existent skeb url" do
+      setup do
+        @site = Sources::Strategies.find("https://skeb.jp/@kai_chiisame/works/2")
+      end
+
+      should "not raise anything" do
+        assert_nothing_raised { @site.to_h }
+      end
+
+      should "still find the right artist" do
+        artist = FactoryBot.create(:artist, name: "kai_chiisame", url_string: @site.url)
+        assert_equal([artist], @site.artists)
+      end
+    end
+
+    context "normalizing for source" do
+      should "avoid normalizing unnormalizable urls" do
+        bad_source = "https://skeb.imgix.net/requests/229088_2?bg=%23fff&auto=format&w=800&s=9cac8b76c0838f2df4f19ebc41c1ae0a"
+        assert_equal(bad_source, Sources::Strategies.normalize_source(bad_source))
+      end
+    end
+  end
+end