Refactor nicoseiga strategy

* Get rid of mechanize, fully switch to Danbooru::Http * Switch to mobile api, improving speed * Merge main and manga clients * Add full support for manga pages * Add support for anonymous and r-15 images * Don't fail when attempting to upload oekaki direct links * Various misc fixes
2020-06-02 12:29:04 +02:00
parent 025e09ff7f
commit 9f0e85e1b5
5 changed files with 278 additions and 285 deletions
--- a/app/logical/nico_seiga_api_client.rb
+++ b/app/logical/nico_seiga_api_client.rb
@@ -1,83 +1,101 @@
 class NicoSeigaApiClient
  extend Memoist
-  BASE_URL = "http://seiga.nicovideo.jp/api"
-  attr_reader :illust_id
+  XML_API = "https://seiga.nicovideo.jp/api"

-  def self.agent
-    mech = Mechanize.new
-    mech.redirect_ok = false
-    mech.keep_alive = false
+  def initialize(work_id:, type:)
+    @work_id = work_id
+    @work_type = type
+  end

-    session = Cache.get("nico-seiga-session")
-    if session
-      cookie = Mechanize::Cookie.new("user_session", session)
-      cookie.domain = ".nicovideo.jp"
-      cookie.path = "/"
-      mech.cookie_jar.add(cookie)
-    else
-      mech.get("https://account.nicovideo.jp/login") do |page|
-        page.form_with(:id => "login_form") do |form|
-          form["mail_tel"] = Danbooru.config.nico_seiga_login
-          form["password"] = Danbooru.config.nico_seiga_password
-        end.click_button
-      end
-      session = mech.cookie_jar.cookies.select {|c| c.name == "user_session"}.first
-      if session
-        Cache.put("nico-seiga-session", session.value, 1.week)
-      else
-        raise "Session not found"
+  def image_ids
+    if @work_type == "illust"
+      [api_response["id"]]
+    elsif @work_type == "manga"
+      manga_api_response.map do |x|
+        x["meta"]["source_url"].match(%r{/thumb/(\d+)\w}i).captures[0]
      end
    end
-
-    # This cookie needs to be set to allow viewing of adult works
-    cookie = Mechanize::Cookie.new("skip_fetish_warning", "1")
-    cookie.domain = "seiga.nicovideo.jp"
-    cookie.path = "/"
-    mech.cookie_jar.add(cookie)
-
-    mech.redirect_ok = true
-    mech
-  end
-
-  def initialize(illust_id:, user_id: nil)
-    @illust_id = illust_id
-    @user_id = user_id
-  end
-
-  def image_id
-    illust_xml["response"]["image"]["id"].to_i
-  end
-
-  def user_id
-    @user_id || illust_xml["response"]["image"]["user_id"].to_i
  end

  def title
-    illust_xml["response"]["image"]["title"]
+    api_response["title"]
  end

-  def desc
-    illust_xml["response"]["image"]["description"]
+  def description
+    api_response["description"]
  end

-  def moniker
-    artist_xml["response"]["user"]["nickname"]
+  def tags
+    api_response.dig("tag_list", "tag").to_a.map { |t| t["name"] }.compact
  end

-  def illust_xml
-    get("#{BASE_URL}/illust/info?id=#{illust_id}")
+  def user_id
+    api_response["user_id"]
  end

-  def artist_xml
-    get("#{BASE_URL}/user/info?id=#{user_id}")
+  def user_name
+    if @work_type == "illust"
+      api_response["nickname"]
+    elsif @work_type == "manga"
+      user_api_response(user_id)["nickname"]
+    end
+  end
+
+  def api_response
+    if @work_type == "illust"
+      resp = get("https://sp.seiga.nicovideo.jp/ajax/seiga/im#{@work_id}")
+      return {} if resp.blank? || resp.code.to_i == 404
+      api_response = JSON.parse(resp)["target_image"]
+
+    elsif @work_type == "manga"
+      resp = Danbooru::Http.cache(1.minute).get("#{XML_API}/theme/info?id=#{@work_id}")
+      return {} if resp.blank? || resp.code.to_i == 404
+      api_response = Hash.from_xml(resp.to_s)["response"]["theme"]
+    end
+
+    api_response || {}
+  rescue JSON::ParserError
+    {}
+  end
+
+  def manga_api_response
+    resp = get("https://ssl.seiga.nicovideo.jp/api/v1/app/manga/episodes/#{@work_id}/frames")
+    return {} if resp.blank? || resp.code.to_i == 404
+    JSON.parse(resp)["data"]["result"]
+  rescue JSON::ParserError
+    {}
+  end
+
+  def user_api_response(user_id)
+    resp = Danbooru::Http.cache(1.minute).get("#{XML_API}/user/info?id=#{user_id}")
+    return {} if resp.blank? || resp.code.to_i == 404
+    Hash.from_xml(resp.to_s)["response"]["user"]
  end

  def get(url)
-    response = Danbooru::Http.cache(1.minute).get(url)
-    raise "nico seiga api call failed (code=#{response.code}, body=#{response.body})" if response.code != 200
+    cookie_header = Cache.get("nicoseiga-cookie-header") || regenerate_cookie_header

-    Hash.from_xml(response.to_s)
+    resp = Danbooru::Http.headers({Cookie: cookie_header}).cache(1.minute).get(url)
+
+    if resp.headers["Location"] =~ %r{seiga\.nicovideo\.jp/login/}i
+      cookie_header = regenerate_cookie_header
+      resp = Danbooru::Http.headers({Cookie: cookie_header}).cache(1.minute).get(url)
+    end
+
+    resp
  end

-  memoize :artist_xml, :illust_xml
+  def regenerate_cookie_header
+    form = {
+      mail_tel: Danbooru.config.nico_seiga_login,
+      password: Danbooru.config.nico_seiga_password
+    }
+    resp = Danbooru::Http.post("https://account.nicovideo.jp/api/v1/login", form: form)
+    cookies = resp.cookies.map { |c| c.name + "=" + c.value }
+    cookies << "accept_fetish_warning=2"
+
+    Cache.put("nicoseiga-cookie-header", cookies.join(";"), 1.week)
+  end
+
+  memoize :api_response, :manga_api_response, :user_api_response
 end
--- a/app/logical/nico_seiga_manga_api_client.rb
+++ b/app/logical/nico_seiga_manga_api_client.rb
@@ -1,60 +0,0 @@
-class NicoSeigaMangaApiClient
-  extend Memoist
-  BASE_URL = "https://seiga.nicovideo.jp/api"
-  attr_reader :theme_id
-
-  def initialize(theme_id)
-    @theme_id = theme_id
-  end
-
-  def user_id
-    theme_info_xml["response"]["theme"]["user_id"].to_i
-  end
-
-  def title
-    theme_info_xml["response"]["theme"]["title"]
-  end
-
-  def desc
-    theme_info_xml["response"]["theme"]["description"]
-  end
-
-  def moniker
-    artist_xml["response"]["user"]["nickname"]
-  end
-
-  def image_ids
-    images = theme_data_xml["response"]["image_list"]["image"]
-    images = [images] unless images.is_a?(Array)
-    images.map {|x| x["id"]}
-  end
-
-  def tags
-    theme_info_xml["response"]["theme"]["tag_list"]["tag"].map {|x| x["name"]}
-  end
-
-  def theme_data_xml
-    uri = "#{BASE_URL}/theme/data?theme_id=#{theme_id}"
-    body = NicoSeigaApiClient.agent.get(uri).body
-    Hash.from_xml(body)
-  end
-
-  def theme_info_xml
-    uri = "#{BASE_URL}/theme/info?id=#{theme_id}"
-    body = NicoSeigaApiClient.agent.get(uri).body
-    Hash.from_xml(body)
-  end
-
-  def artist_xml
-    get("#{BASE_URL}/user/info?id=#{user_id}")
-  end
-
-  def get(url)
-    response = Danbooru::Http.cache(1.minute).get(url)
-    raise "nico seiga api call failed (code=#{response.code}, body=#{response.body})" if response.code != 200
-
-    Hash.from_xml(response.to_s)
-  end
-
-  memoize :theme_data_xml, :theme_info_xml, :artist_xml
-end
--- a/app/logical/sources/strategies/nico_seiga.rb
+++ b/app/logical/sources/strategies/nico_seiga.rb
@@ -1,25 +1,51 @@
-# Image Direct URL
+# Direct URL
 # * https://lohas.nicoseiga.jp/o/971eb8af9bbcde5c2e51d5ef3a2f62d6d9ff5552/1589933964/3583893
 # * http://lohas.nicoseiga.jp/priv/3521156?e=1382558156&h=f2e089256abd1d453a455ec8f317a6c703e2cedf
 # * http://lohas.nicoseiga.jp/priv/b80f86c0d8591b217e7513a9e175e94e00f3c7a1/1384936074/3583893
+#
+# * http://lohas.nicoseiga.jp/material/5746c5/4459092
+#
+# (Manga direct url)
+# * https://lohas.nicoseiga.jp/priv/f5b8966fd53bf7e06cccff9fbb2c4eef62877538/1590752727/8947170
+#
+# Samples
+# * http://lohas.nicoseiga.jp/thumb/2163478i?
+# * https://lohas.nicoseiga.jp/thumb/8947170p
+#
+## The direct urls and samples above can belong to both illust and manga.
+## There's two ways to tell them apart:
+## * visit the /source/ equivalent: illusts redirect to the /o/ intermediary page, manga redirect to /priv/ directly
+## * try an api call: illusts will succeed, manga will fail
+#
+# Source Link
 # * http://seiga.nicovideo.jp/image/source?id=3312222
 #
-# Image Page URL
+# Illust Page URL
 # * https://seiga.nicovideo.jp/seiga/im3521156
+# * https://seiga.nicovideo.jp/seiga/im520647 (anonymous artist)
 #
 # Manga Page URL
 # * http://seiga.nicovideo.jp/watch/mg316708
+#
+# Video Page URL (not supported)
+# * https://www.nicovideo.jp/watch/sm36465441
+#
+# Oekaki
+# * https://dic.nicovideo.jp/oekaki/52833.png

 module Sources
  module Strategies
    class NicoSeiga < Base
-      URL = %r!\Ahttps?://(?:\w+\.)?nico(?:seiga|video)\.jp!
-      DIRECT1 = %r!\Ahttps?://lohas\.nicoseiga\.jp/priv/[0-9a-f]+!
-      DIRECT2 = %r!\Ahttps?://lohas\.nicoseiga\.jp/o/[0-9a-f]+/\d+/\d+!
-      DIRECT3 = %r!\Ahttps?://seiga\.nicovideo\.jp/images/source/\d+!
-      PAGE = %r!\Ahttps?://seiga\.nicovideo\.jp/seiga/im(\d+)!i
-      PROFILE = %r!\Ahttps?://seiga\.nicovideo\.jp/user/illust/(\d+)!i
-      MANGA_PAGE = %r!\Ahttps?://seiga\.nicovideo\.jp/watch/mg(\d+)!i
+      DIRECT       = %r{\Ahttps?://lohas\.nicoseiga\.jp/(priv|o)/(?:\w+/\d+/)?(?<image_id>\d+)(?:\?.+)?}i
+      SOURCE       = %r{\Ahttps?://seiga\.nicovideo\.jp/image/source(?:/|\?id=)(?<image_id>\d+)}i
+
+      ILLUST_THUMB = %r{\Ahttps?://lohas\.nicoseiga\.jp/thumb/(?<illust_id>\d+)i}i
+      MANGA_THUMB  = %r{\Ahttps?://lohas\.nicoseiga\.jp/thumb/(?<image_id>\d+)p}i
+
+      ILLUST_PAGE  = %r{\Ahttps?://(?:sp\.)?seiga\.nicovideo\.jp/seiga/im(?<illust_id>\d+)}i
+      MANGA_PAGE   = %r{\Ahttps?://(?:sp\.)?seiga\.nicovideo\.jp/watch/mg(?<manga_id>\d+)}i
+
+      PROFILE_PAGE = %r{\Ahttps?://seiga\.nicovideo\.jp/user/illust/(?<artist_id>\d+)}i

      def domains
        ["nicoseiga.jp", "nicovideo.jp"]
@@ -30,160 +56,125 @@ module Sources
      end

      def image_urls
-        if url =~ DIRECT1
-          return [url]
+        urls = []
+        return urls if api_client&.api_response.blank?
+
+        if image_id.present?
+          urls << "https://seiga.nicovideo.jp/image/source/#{image_id}"
+        elsif illust_id.present?
+          urls << "https://seiga.nicovideo.jp/image/source/#{illust_id}"
+        elsif manga_id.present? && api_client.image_ids.present?
+          urls += api_client.image_ids.map { |id| "https://seiga.nicovideo.jp/image/source/#{id}" }
+        end
+        urls
+      end
+
+      def image_url
+        return if image_urls.blank?
+        return url if api_client.blank?
+
+        img = case url
+        when DIRECT then "https://seiga.nicovideo.jp/image/source/#{image_id_from_url(url)}"
+        when SOURCE then url
+        else image_urls.first
        end

-        if theme_id
-          return api_client.image_ids.map do |image_id|
-            "https://seiga.nicovideo.jp/image/source/#{image_id}"
-          end
-        end
-
-        link = page.search("a#illust_link")
-
-        if link.any?
-          image_url = "http://seiga.nicovideo.jp" + link[0]["href"]
-          page = agent.get(image_url) # need to follow this redirect while logged in or it won't work
-
-          if page.is_a?(Mechanize::Image)
-            return [page.uri.to_s]
-          end
-
-          images = page.search("div.illust_view_big").select {|x| x["data-src"] =~ /\/priv\//}
-
-          if images.any?
-            return ["http://lohas.nicoseiga.jp" + images[0]["data-src"]]
-          end
-        end
-
-        raise "image url not found for (#{url}, #{referer_url})"
+        resp = api_client.get(img)
+        resp.headers["Location"]&.gsub(%r{nicoseiga.jp/o/}i, 'nicoseiga.jp/priv/')
      end

      def page_url
-        [url, referer_url].each do |x|
-          if x =~ %r!\Ahttps?://lohas\.nicoseiga\.jp/o/[a-f0-9]+/\d+/(\d+)!
-            return "http://seiga.nicovideo.jp/seiga/im#{$1}"
-          end
-
-          if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/(\d+)\?e=\d+&h=[a-f0-9]+}i
-            return "http://seiga.nicovideo.jp/seiga/im#{$1}"
-          end
-
-          if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/[a-f0-9]+/\d+/(\d+)}i
-            return "http://seiga.nicovideo.jp/seiga/im#{$1}"
-          end
-
-          if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/(\d+)}i
-            return "http://seiga.nicovideo.jp/seiga/im#{$1}"
-          end
-
-          if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp//?thumb/(\d+)i?}i
-            return "http://seiga.nicovideo.jp/seiga/im#{$1}"
-          end
-
-          if x =~ %r{/seiga/im\d+}
-            return x
-          end
-
-          if x =~ %r{/watch/mg\d+}
-            return x
-          end
-
-          if x =~ %r{/image/source\?id=(\d+)}
-            return "http://seiga.nicovideo.jp/seiga/im#{$1}"
-          end
+        if illust_id.present?
+          "https://seiga.nicovideo.jp/seiga/im#{illust_id}"
+        elsif manga_id.present?
+          "https://seiga.nicovideo.jp/watch/mg#{manga_id}"
+        elsif image_id.present?
+          "https://seiga.nicovideo.jp/image/source/#{image_id}"
        end
-
-        return super
-      end
-
-      def canonical_url
-        image_url
      end

      def profile_url
-        if url =~ PROFILE
-          return url
-        end
+        user_id = api_client&.user_id
+        return if user_id.blank? # artists can be anonymous

        "http://seiga.nicovideo.jp/user/illust/#{api_client.user_id}"
      end

      def artist_name
-        api_client.moniker
+        return if api_client.blank?
+        api_client.user_name
      end

      def artist_commentary_title
+        return if api_client.blank?
        api_client.title
      end

      def artist_commentary_desc
-        api_client.desc
+        return if api_client.blank?
+        api_client.description
+      end
+
+      def dtext_artist_commentary_desc
+        DText.from_html(artist_commentary_desc).gsub(/[^\w]im(\d+)/, ' seiga #\1 ')
      end

      def normalize_for_source
-        if illust_id.present?
-          "https://seiga.nicovideo.jp/seiga/im#{illust_id}"
-        elsif theme_id.present?
-          "http://seiga.nicovideo.jp/watch/mg#{theme_id}"
+        # There's no way to tell apart illust from manga from the direct image url alone. What's worse,
+        # nicoseiga itself doesn't know how to normalize back to manga, so if it's not an illust type then
+        # it's impossible to get the original manga page back from the image url alone.
+        # /source/ links on the other hand correctly redirect, hence we use them to normalize saved direct sources.
+        if url =~ DIRECT
+          "https://seiga.nicovideo.jp/image/source/#{image_id}"
+        else
+          page_url
        end
      end

      def tag_name
+        return if api_client&.user_id.blank?
        "nicoseiga#{api_client.user_id}"
      end

      def tags
-        string = page.at("meta[name=keywords]").try(:[], "content") || ""
-        string.split(/,/).map do |name|
-          [name, "https://seiga.nicovideo.jp/tag/#{CGI.escape(name)}"]
+        return [] if api_client.blank?
+
+        base_url = "https://seiga.nicovideo.jp/"
+        base_url += "manga/" if manga_id.present?
+        base_url += "tag/"
+
+        api_client.tags.map do |name|
+          [name, base_url + CGI.escape(name)]
        end
      end
-      memoize :tags
+
+      def image_id
+        image_id_from_url(url)
+      end
+
+      def image_id_from_url(url)
+        url[DIRECT, :image_id] || url[SOURCE, :image_id] || url[MANGA_THUMB, :image_id]
+      end
+
+      def illust_id
+        urls.map { |u| u[ILLUST_PAGE, :illust_id] || u[ILLUST_THUMB, :illust_id] }.compact.first
+      end
+
+      def manga_id
+        urls.compact.map { |u| u[MANGA_PAGE, :manga_id] }.compact.first
+      end

      def api_client
-        if illust_id
-          NicoSeigaApiClient.new(illust_id: illust_id)
-        elsif theme_id
-          NicoSeigaMangaApiClient.new(theme_id)
+        if illust_id.present?
+          NicoSeigaApiClient.new(work_id: illust_id, type: "illust")
+        elsif manga_id.present?
+          NicoSeigaApiClient.new(work_id: manga_id, type: "manga")
+        elsif image_id.present?
+          # We default to illust to attempt getting the api anyway
+          NicoSeigaApiClient.new(work_id: image_id, type: "illust")
        end
      end
      memoize :api_client
-
-      def illust_id
-        if page_url =~ PAGE
-          return $1.to_i
-        end
-
-        return nil
-      end
-
-      def theme_id
-        if page_url =~ MANGA_PAGE
-          return $1.to_i
-        end
-
-        return nil
-      end
-
-      def page
-        doc = agent.get(page_url)
-
-        if doc.search("a#link_btn_login").any?
-          # Session cache is invalid, clear it and log in normally.
-          Cache.delete("nico-seiga-session")
-          doc = agent.get(page_url)
-        end
-
-        doc
-      end
-      memoize :page
-
-      def agent
-        NicoSeigaApiClient.agent
-      end
-      memoize :agent
    end
  end
 end
--- a/test/unit/sources/nico_seiga_manga_test.rb
+++ b/test/unit/sources/nico_seiga_manga_test.rb
@@ -1,23 +0,0 @@
-require 'test_helper'
-
-module Sources
-  class NicoSeigaTest < ActiveSupport::TestCase
-    context "The source site for nico seiga" do
-      setup do
-        @site = Sources::Strategies.find("http://seiga.nicovideo.jp/watch/mg316708", "http://seiga.nicovideo.jp/watch/mg316708")
-      end
-
-      should "find the image urls" do
-        assert_equal(["https://seiga.nicovideo.jp/image/source/8100968", "https://seiga.nicovideo.jp/image/source/8100969", "https://seiga.nicovideo.jp/image/source/8100970", "https://seiga.nicovideo.jp/image/source/8100971", "https://seiga.nicovideo.jp/image/source/8100972", "https://seiga.nicovideo.jp/image/source/8100973", "https://seiga.nicovideo.jp/image/source/8100974", "https://seiga.nicovideo.jp/image/source/8100975"], @site.image_urls)
-      end
-
-      should "find the page url" do
-        assert_equal("http://seiga.nicovideo.jp/watch/mg316708", @site.page_url)
-      end
-
-      should "find the artist name" do
-        assert_not_nil(@site.artist_name)
-      end
-    end
-  end
-end
--- a/test/unit/sources/nico_seiga_test.rb
+++ b/test/unit/sources/nico_seiga_test.rb
@@ -6,32 +6,52 @@ module Sources
      setup do
        @site_1 = Sources::Strategies.find("http://lohas.nicoseiga.jp/o/910aecf08e542285862954017f8a33a8c32a8aec/1433298801/4937663")
        @site_2 = Sources::Strategies.find("http://seiga.nicovideo.jp/seiga/im4937663")
-        @site_3 = Sources::Strategies.find("http://seiga.nicovideo.jp/watch/mg376206")
+        @site_3 = Sources::Strategies.find("https://seiga.nicovideo.jp/watch/mg470189?track=ct_episode")
      end

      should "get the profile" do
        assert_equal("http://seiga.nicovideo.jp/user/illust/7017777", @site_1.profile_url)
        assert_equal("http://seiga.nicovideo.jp/user/illust/7017777", @site_2.profile_url)
+        assert_equal("http://seiga.nicovideo.jp/user/illust/20797022", @site_3.profile_url)
      end

      should "get the artist name" do
        assert_equal("osamari", @site_1.artist_name)
        assert_equal("osamari", @site_2.artist_name)
+        assert_equal("風呂", @site_3.artist_name)
      end

      should "get the artist commentary" do
        assert_equal("コジコジ", @site_2.artist_commentary_title)
        assert_equal("コジコジのドット絵\nこんなかわいらしい容姿で毒を吐くコジコジが堪らん（切実）", @site_2.artist_commentary_desc)
+
+        assert_equal("ハコ女子 1ハコ目", @site_3.artist_commentary_title)
+        assert_equal("同じクラスの箱田さんはいつもハコを被っている。しかしてその素顔は…？　twitter(@hakojoshi1)にてだいたい毎日更新中。こっちだともうちょっと先まで読めるよ。", @site_3.artist_commentary_desc)
      end

-      should "get the image url" do
-        assert_match(/^http:\/\/lohas\.nicoseiga\.jp\/priv\//, @site_1.image_url)
-        assert_match(/^http:\/\/lohas\.nicoseiga\.jp\/priv\//, @site_2.image_url)
+      should "get the image url(s)" do
+        assert_match(%r{^https?://lohas\.nicoseiga\.jp/priv/}, @site_1.image_url)
+        assert_match(%r{^https?://lohas\.nicoseiga\.jp/priv/}, @site_2.image_url)
+
+        expected = %w[
+          https://seiga.nicovideo.jp/image/source/10315315
+          https://seiga.nicovideo.jp/image/source/10315318
+          https://seiga.nicovideo.jp/image/source/10315319
+          https://seiga.nicovideo.jp/image/source/10315320
+          https://seiga.nicovideo.jp/image/source/10315321
+          https://seiga.nicovideo.jp/image/source/10315322
+          https://seiga.nicovideo.jp/image/source/10315323
+          https://seiga.nicovideo.jp/image/source/10315324
+          https://seiga.nicovideo.jp/image/source/10315316
+        ]
+        assert_equal(expected.sort, @site_3.image_urls.sort)
+        assert_match(%r{^https?://lohas\.nicoseiga\.jp/priv/}, @site_3.image_url)
      end

      should "get the canonical url" do
-        assert_match(%r!\Ahttps?://lohas\.nicoseiga\.jp/priv/\h{40}/\d+/4937663!, @site_1.canonical_url)
-        assert_match(%r!\Ahttps?://lohas\.nicoseiga\.jp/priv/\h{40}/\d+/4937663!, @site_2.canonical_url)
+        assert_equal("https://seiga.nicovideo.jp/image/source/4937663", @site_1.canonical_url)
+        assert_equal("https://seiga.nicovideo.jp/seiga/im4937663", @site_2.canonical_url)
+        assert_equal("https://seiga.nicovideo.jp/watch/mg470189", @site_3.canonical_url)
      end

      should "get the tags" do
@@ -42,23 +62,70 @@ module Sources
        assert_not(@site_2.tags.empty?)
        first_tag = @site_2.tags.first
        assert_equal(["アニメ", "https://seiga.nicovideo.jp/tag/%E3%82%A2%E3%83%8B%E3%83%A1"], first_tag)
+
+        assert_not(@site_3.tags.empty?)
+        first_tag = @site_3.tags.first
+        assert_equal(["4コマ漫画", "https://seiga.nicovideo.jp/manga/tag/4%E3%82%B3%E3%83%9E%E6%BC%AB%E7%94%BB"], first_tag)
      end

      should "convert a page into a json representation" do
-        assert_nothing_raised do
-          @site_1.to_h
-        end
-        assert_nothing_raised do
-          @site_2.to_h
-        end
+        assert_nothing_raised { @site_1.to_h }
+        assert_nothing_raised { @site_2.to_h }
+        assert_nothing_raised { @site_3.to_h }
      end

      should "work for a https://lohas.nicoseiga.jp/thumb/${id}i url" do
        site = Sources::Strategies.find("https://lohas.nicoseiga.jp/thumb/6844226i")

-        full_image_url = %r!https?://lohas.nicoseiga.jp/priv/[a-f0-9]{40}/[0-9]+/6844226!
-        assert_match(full_image_url, site.image_url)
-        assert_match(full_image_url, site.canonical_url)
+        assert_match(%r!https?://lohas.nicoseiga.jp/priv/[a-f0-9]{40}/[0-9]+/6844226!, site.image_url)
+        assert_match("https://seiga.nicovideo.jp/seiga/im6844226", site.canonical_url)
+      end
+    end
+
+    context "A manga upload through bookmarklet" do
+      setup do
+        @url = "https://seiga.nicovideo.jp/image/source/9146749"
+        @ref = "https://seiga.nicovideo.jp/watch/mg389884"
+        @site = Sources::Strategies.find(@url, @ref)
+      end
+
+      should "get the correct pic" do
+        assert_match(%r!https?://lohas.nicoseiga.jp/priv/[a-f0-9]{40}/[0-9]+/9146749!, @site.image_url)
+      end
+
+      should "set the correct source" do
+        assert_equal(@ref, @site.canonical_url)
+      end
+    end
+
+    context "A nicoseiga video" do
+      should "not raise anything" do
+        site = Sources::Strategies.find("https://www.nicovideo.jp/watch/sm36465441")
+        assert_nothing_raised { site.to_h }
+      end
+    end
+
+    context "An anonymous picture" do
+      should "still work" do
+        site = Sources::Strategies.find("https://seiga.nicovideo.jp/seiga/im520647")
+
+        assert_nothing_raised { site.to_h }
+      end
+    end
+
+    context "An age-restricted picture" do
+      should "still work" do
+        site = Sources::Strategies.find("http://seiga.nicovideo.jp/seiga/im9208126")
+
+        assert_match(%r!https?://lohas.nicoseiga.jp/priv/[a-f0-9]{40}/[0-9]+/9208126!, site.image_url)
+        assert_nothing_raised { site.to_h }
+      end
+    end
+
+    context "An oekaki picture" do
+      should "still work" do
+        site = Sources::Strategies.find("https://dic.nicovideo.jp/oekaki/52833.png")
+        assert_nothing_raised { site.to_h }
      end
    end

@@ -69,10 +136,10 @@ module Sources
        source3 = "http://lohas.nicoseiga.jp/o/910aecf08e542285862954017f8a33a8c32a8aec/1433298801/4937663"
        source4 = "http://seiga.nicovideo.jp/image/source?id=3312222"

-        assert_equal("https://seiga.nicovideo.jp/seiga/im3521156", Sources::Strategies.normalize_source(source1))
-        assert_equal("https://seiga.nicovideo.jp/seiga/im3583893", Sources::Strategies.normalize_source(source2))
-        assert_equal("https://seiga.nicovideo.jp/seiga/im4937663", Sources::Strategies.normalize_source(source3))
-        assert_equal("https://seiga.nicovideo.jp/seiga/im3312222", Sources::Strategies.normalize_source(source4))
+        assert_equal("https://seiga.nicovideo.jp/image/source/3521156", Sources::Strategies.normalize_source(source1))
+        assert_equal("https://seiga.nicovideo.jp/image/source/3583893", Sources::Strategies.normalize_source(source2))
+        assert_equal("https://seiga.nicovideo.jp/image/source/4937663", Sources::Strategies.normalize_source(source3))
+        assert_equal("https://seiga.nicovideo.jp/image/source/3312222", Sources::Strategies.normalize_source(source4))
      end

      should "avoid normalizing unnormalizable urls" do