Merge pull request #4801 from nonamethanks/feat-lofter

Add Lofter support

app/logical/sources/strategies.rb

@@ -16,7 +16,8 @@ module Sources
         Strategies::Mastodon,
         Strategies::Weibo,
         Strategies::Newgrounds,
-        Strategies::Skeb
+        Strategies::Skeb,
+        Strategies::Lofter
       ]
     end
 

app/logical/sources/strategies/lofter.rb

@@ -0,0 +1,100 @@
+# Image URLs
+#
+# # sample
+# * https://imglf3.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJSzFCWFlnUWgzb01DcUdpT1lreG5yQjJVMkhGS09HNGR3PT0.png?imageView&thumbnail=1680x0&quality=96&stripmeta=0
+#
+# # full size
+# * https://imglf3.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJSzFCWFlnUWgzb01DcUdpT1lreG5yQjJVMkhGS09HNGR3PT0.png
+# * http://imglf0.nosdn.127.net/img/cHl3bXNZdDRaaHBnNWJuN1Y4OXBqR01CeVBZSVNmU2FWZWtHc1h4ZTZiUGxlRzMwZnFDM1JnPT0.jpg (404)
+#
+# Page URLs
+#
+# * https://gengar563.lofter.com/post/1e82da8c_1c98dae1b
+#
+# Profile URLs
+#
+# * http://gengar563.lofter.com/
+
+module Sources
+  module Strategies
+    class Lofter < Base
+      PROFILE_URL = %r{\Ahttps?://(?<artist_name>[\w-]+).lofter.com}i
+      PAGE_URL =    %r{#{PROFILE_URL}/post/(?<illust_id>[\w-]+)}i
+      IMAGE_HOST =  %r{\Ahttps?://imglf\d\.(?:nosdn\d?\.12\d|lf127)\.net}i
+      IMAGE_URL =   %r{#{IMAGE_HOST}/img/\w+\.\w+}i
+
+      def domains
+        ["lofter.com", "lf127.net"]
+      end
+
+      def site_name
+        "Lofter"
+      end
+
+      def match?
+        return false if parsed_url.nil?
+        parsed_url.domain.in?(domains) || parsed_url.host =~ IMAGE_HOST
+      end
+
+      def image_url
+        if url =~ IMAGE_URL
+          get_full_version(url)
+        else
+          image_urls.first
+        end
+      end
+
+      def image_urls
+        images = page&.search(".pic img")
+        images.to_a.map { |img| get_full_version(img["src"]) }
+      end
+
+      def get_full_version(url)
+        parsed = URI.parse(url)
+        "https://#{parsed.host}#{parsed.path}"
+      end
+
+      def profile_url
+        return nil if artist_name.blank?
+        "https://#{artist_name}.lofter.com"
+      end
+
+      def page_url
+        return nil if illust_id.blank? || profile_url.blank?
+
+        "#{profile_url}/post/#{illust_id}"
+      end
+
+      def page
+        return nil if page_url.blank?
+
+        response = http.cache(1.minute).get(page_url)
+        response.parse if response.status == 200
+      end
+
+      def tags
+        tags = page&.search(".info .tag")
+
+        tags.to_a.map do |tag|
+          [tag.text, tag.attr("href")]
+        end
+      end
+
+      def artist_commentary_desc
+        page&.search(".ct .text")&.to_html
+      end
+
+      def normalize_for_source
+        page_url
+      end
+
+      def illust_id
+        urls.map { |u| u[PAGE_URL, :illust_id] }.compact.first
+      end
+
+      def artist_name
+        urls.map { |u| u[PROFILE_URL, :artist_name] || u[PAGE_URL, :artist_name] }.compact.first
+      end
+    end
+  end
+end

test/unit/sources/lofter_test.rb

@@ -0,0 +1,62 @@
+require 'test_helper'
+
+module Sources
+  class LofterTest < ActiveSupport::TestCase
+    context "A lofter post" do
+      setup do
+        @img = "https://imglf4.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJUFczb2RKSVlpMHJkNy9kc3BSQVQvQm5DNzB4eVhxay9nPT0.png?imageView&thumbnail=1680x0&quality=96&stripmeta=0"
+        @ref = "https://gengar563.lofter.com/post/1e82da8c_1c98dae1b"
+        @source = Sources::Strategies.find(@img, @ref)
+      end
+
+      should "get the artist name" do
+        assert_equal("gengar563", @source.artist_name)
+      end
+
+      should "get the commentary" do
+        assert_match(/发了三次发不出有毒…… \n.*\n失去耐心.jpg/, @source.dtext_artist_commentary_desc)
+      end
+      should "get profile url" do
+        assert_equal("https://gengar563.lofter.com", @source.profile_url)
+      end
+
+      should "get the image urls" do
+        images = %w[
+          https://imglf3.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJQ1RxY0lYaU1UUE9tQ0NvUE9rVXFpOFFEVzMwbnQ4aEFnPT0.jpg
+          https://imglf3.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJRWlXYTRVOEpXTU9TSGt3TjBDQ0JFZVpZMEJtWjFneVNBPT0.png
+          https://imglf6.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJR1d3Y2VvbTNTQlIvdFU1WWlqZHEzbjI4MFVNZVdoN3VBPT0.png
+          https://imglf6.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJTi83NDRDUjNvd3hySGxEZFovd2hwbi9oaG9NQ1hOUkZ3PT0.png
+          https://imglf4.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJUFczb2RKSVlpMHJkNy9kc3BSQVQvQm5DNzB4eVhxay9nPT0.png
+          https://imglf4.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJSStJZE9RYnJURktHazdIVHNNMjQ5eFJldHVTQy9XbDB3PT0.png
+          https://imglf3.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJSzFCWFlnUWgzb01DcUdpT1lreG5yQjJVMkhGS09HNGR3PT0.png
+        ]
+
+        assert_equal(images, @source.image_urls)
+      end
+
+      should "download the full-size image" do
+        assert_downloaded(2_739_443, @source.image_url)
+      end
+
+      should "find the correct artist" do
+        @artist = FactoryBot.create(:artist, name: "gengar563", url_string: "https://gengar563.lofter.com")
+        assert_equal([@artist], @source.artists)
+      end
+    end
+
+    context "A bad link" do
+      should "correctly get the full size" do
+        source = Sources::Strategies.find("https://imglf4.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJUFczb2RKSVlpMHJkNy9kc3BSQVQvQm5DNzB4eVhxay9nPT0.png?imageView&thumbnail=1680x0&quality=96&stripmeta=0")
+        assert_equal("https://imglf4.lf127.net/img/S1d2QlVsWkJhSW1qcnpIS0ZSa3ZJUFczb2RKSVlpMHJkNy9kc3BSQVQvQm5DNzB4eVhxay9nPT0.png", source.image_url)
+        assert_nothing_raised { source.to_h }
+      end
+    end
+
+    context "A dead link" do
+      should "not raise anything" do
+        source = Sources::Strategies.find("https://gxszdddd.lofter.com/post/322595b1_1ca5e6f66")
+        assert_nothing_raised { source.to_h }
+      end
+    end
+  end
+end