sources: factor out site_name method.

2022-03-11 23:16:36 -06:00
parent b4aea72d04
commit 28971fe103
23 changed files with 48 additions and 150 deletions
--- a/app/logical/source/url.rb
+++ b/app/logical/source/url.rb
@@ -3,6 +3,9 @@
 # A Source::URL is a URL from a source site, such as Twitter, Pixiv, etc. Each site has a
 # subclass responsible for parsing and extracting information from URLs for that site.
 #
+# Sources::Strategies are the main user of Source::URLs. Each Source::URL subclass usually
+# has a corresponding strategy for extracting data from that site.
+#
 # To add a new site, create a subclass of Source::URL and implement `#match?` to define
 # which URLs belong to the site, and `#parse` to parse and extract information from the URL.
 #
@@ -15,6 +18,7 @@
 #   url.status_id        # => "1496123903290314755"
 #   url.twitter_username # => "yasunavert"
 #
+# @see Danbooru::URL
 module Source
  class URL < Danbooru::URL
    SUBCLASSES = [
@@ -68,9 +72,43 @@ module Source
      raise NotImplementedError
    end

-    # @return [String, nil] The name of the site this URL belongs to, or possibly nil if unknown.
+    # The name of the site this URL belongs to.
+    #
+    # @return [String]
    def site_name
-      self.class.name.demodulize.titleize
+      # XXX should go in dedicated subclasses.
+      case host
+      when /ask\.fm\z/i
+        "Ask.fm"
+      when /bcy\.net\z/i
+        "BCY"
+      when /booth\.pm\z/i
+        "Booth.pm"
+      when /circle\.ms\z/i
+        "Circle.ms"
+      when /dlsite\.(com|net)\z/i
+        "DLSite"
+      when /doujinshi\.mugimugi\.org\z/i
+        "Doujinshi.org"
+      when /fc2\.com\z/i
+        "FC2"
+      when /ko-fi\.com\z/i
+        "Ko-fi"
+      when /mixi\.jp\z/i
+        "Mixi.jp"
+      when /piapro\.jp\z/i
+        "Piapro.jp"
+      when /sakura\.ne\.jp\z/i
+        "Sakura.ne.jp"
+      else
+        if self.class == Source::URL
+          # "www.melonbooks.co.jp" => "Melonbooks"
+          parsed_domain.sld.titleize
+        else
+          # "Source::URL::NicoSeiga" => "Nico Seiga"
+          self.class.name.demodulize.titleize
+        end
+      end
    end

    protected def initialize(...)
--- a/app/logical/sources/strategies/art_station.rb
+++ b/app/logical/sources/strategies/art_station.rb
@@ -7,10 +7,6 @@ module Sources::Strategies
      Source::URL::ArtStation === parsed_url
    end

-    def site_name
-      parsed_url.site_name
-    end
-
    def image_urls
      @image_urls ||= image_urls_sub.map { |asset| asset_url(asset, :largest) }
    end
--- a/app/logical/sources/strategies/base.rb
+++ b/app/logical/sources/strategies/base.rb
@@ -22,6 +22,7 @@ module Sources
      DOWNLOAD_TIMEOUT = 60

      attr_reader :url, :referer_url, :parsed_url, :parsed_referer
+      delegate :site_name, to: :parsed_url

      # Should return true if all prerequisites for using the strategy are met.
      # Return false if the strategy requires api keys that have not been configured.
@@ -53,74 +54,6 @@ module Sources
        false
      end

-      def site_name
-        host = parsed_url&.host
-
-        # XXX should go in dedicated strategies.
-        case host
-        when /amazon\.(com|jp|co\.jp)\z/i
-          "Amazon"
-        when /ask\.fm\z/i
-          "Ask.fm"
-        when /bcy\.net\z/i
-          "BCY"
-        when /booth\.pm\z/i
-          "Booth.pm"
-        when /circle\.ms\z/i
-          "Circle.ms"
-        when /dlsite\.(com|net)\z/i
-          "DLSite"
-        when /doujinshi\.mugimugi\.org\z/i, /doujinshi\.org\z/i
-          "Doujinshi.org"
-        when /erogamescape\.dyndns\.org\z/i
-          "Erogamescape"
-        when /facebook\.com\z/i
-          "Facebook"
-        when /fc2\.com\z/i
-          "FC2"
-        when /gumroad\.com\z/i
-          "Gumroad"
-        when /instagram\.com\z/i
-          "Instagram"
-        when /ko-fi\.com\z/i
-          "Ko-fi"
-        when /livedoor\.(jp|com)\z/i
-          "Livedoor"
-        when /mangaupdates\.com\z/i
-          "Mangaupdates"
-        when /melonbooks\.co\.jp\z/i
-          "Melonbooks"
-        when /mihuashi\.com\z/i
-          "Mihuashi"
-        when /mixi\.jp\z/i
-          "Mixi.jp"
-        when /patreon\.com\z/i
-          "Patreon"
-        when /piapro\.jp\z/i
-          "Piapro.jp"
-        when /picarto\.tv\z/i
-          "Picarto"
-        when /privatter\.net\z/i
-          "Privatter"
-        when /sakura\.ne\.jp\z/i
-          "Sakura.ne.jp"
-        when /stickam\.jp\z/i
-          "Stickam"
-        when /tinami\.com\z/i
-          "Tinami"
-        when /toranoana\.(jp|shop)\z/i
-          "Toranoana"
-        when /twitch\.tv\z/i
-          "Twitch"
-        when /wikipedia\.org\z/i
-          "Wikipedia"
-        when /youtube\.com\z/i
-          "Youtube"
-        else
-          host
-        end
-      end
-
      # Whatever <tt>url</tt> is, this method should return the direct links
      # to the canonical binary files. It should not be an HTML page. It should
      # be a list of JPEG, PNG, GIF, WEBM, MP4, ZIP, etc. It is what the
--- a/app/logical/sources/strategies/deviant_art.rb
+++ b/app/logical/sources/strategies/deviant_art.rb
@@ -11,10 +11,6 @@ module Sources
        Source::URL::DeviantArt === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        [image_url]
      end
--- a/app/logical/sources/strategies/fanbox.rb
+++ b/app/logical/sources/strategies/fanbox.rb
@@ -8,10 +8,6 @@ module Sources
        Source::URL::Fanbox === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if parsed_url.image_url?
          [parsed_url.full_image_url]
--- a/app/logical/sources/strategies/fantia.rb
+++ b/app/logical/sources/strategies/fantia.rb
@@ -10,10 +10,6 @@ module Sources::Strategies
      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?
--- a/app/logical/sources/strategies/foundation.rb
+++ b/app/logical/sources/strategies/foundation.rb
@@ -8,10 +8,6 @@ module Sources
        parsed_url&.site_name == "Foundation"
      end

-      def site_name
-        "Foundation"
-      end
-
      def image_urls
        if parsed_url.full_image_url.present?
          [parsed_url.full_image_url]
--- a/app/logical/sources/strategies/hentai_foundry.rb
+++ b/app/logical/sources/strategies/hentai_foundry.rb
@@ -8,10 +8,6 @@ module Sources
        parsed_url&.site_name == "Hentai Foundry"
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        image = page&.search("#picBox img")

--- a/app/logical/sources/strategies/lofter.rb
+++ b/app/logical/sources/strategies/lofter.rb
@@ -8,10 +8,6 @@ module Sources
        parsed_url&.site_name == "Lofter"
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if parsed_url.image_url?
          [parsed_url.full_image_url]
--- a/app/logical/sources/strategies/mastodon.rb
+++ b/app/logical/sources/strategies/mastodon.rb
@@ -7,10 +7,6 @@ module Sources::Strategies
      Source::URL::Mastodon === parsed_url
    end

-    def site_name
-      parsed_url.site_name
-    end
-
    def domain
      case site_name
      when "Pawoo" then "pawoo.net"
--- a/app/logical/sources/strategies/newgrounds.rb
+++ b/app/logical/sources/strategies/newgrounds.rb
@@ -8,10 +8,6 @@ module Sources
        parsed_url&.site_name == "Newgrounds"
      end

-      def site_name
-        "Newgrounds"
-      end
-
      def image_urls
        if parsed_url.image_url?
          [url]
--- a/app/logical/sources/strategies/nico_seiga.rb
+++ b/app/logical/sources/strategies/nico_seiga.rb
@@ -12,10 +12,6 @@ module Sources
        Source::URL::NicoSeiga === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if image_id.present?
          [image_url_for("https://seiga.nicovideo.jp/image/source/#{image_id}")]
--- a/app/logical/sources/strategies/nijie.rb
+++ b/app/logical/sources/strategies/nijie.rb
@@ -12,10 +12,6 @@ module Sources
        Source::URL::Nijie === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if parsed_url.image_url?
          [parsed_url.full_image_url]
--- a/app/logical/sources/strategies/pixiv.rb
+++ b/app/logical/sources/strategies/pixiv.rb
@@ -36,10 +36,6 @@ module Sources
        Source::URL::Pixiv === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if is_ugoira?
          [api_ugoira[:originalSrc]]
--- a/app/logical/sources/strategies/pixiv_sketch.rb
+++ b/app/logical/sources/strategies/pixiv_sketch.rb
@@ -8,10 +8,6 @@ module Sources
        Source::URL::PixivSketch === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if parsed_url.image_url?
          [parsed_url.full_image_url]
--- a/app/logical/sources/strategies/plurk.rb
+++ b/app/logical/sources/strategies/plurk.rb
@@ -8,10 +8,6 @@ module Sources
        Source::URL::Plurk === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        # * Posts can have up to 10 images.
        # * Artists commonly post extra images by replying to their own post.
--- a/app/logical/sources/strategies/skeb.rb
+++ b/app/logical/sources/strategies/skeb.rb
@@ -8,10 +8,6 @@ module Sources
        parsed_url&.site_name == "Skeb"
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if parsed_url.image_url?
          [url]
--- a/app/logical/sources/strategies/tumblr.rb
+++ b/app/logical/sources/strategies/tumblr.rb
@@ -11,10 +11,6 @@ module Sources::Strategies
      Source::URL::Tumblr === parsed_url
    end

-    def site_name
-      parsed_url.site_name
-    end
-
    def image_urls
      return [find_largest(parsed_url)].compact if parsed_url.asset_url?

--- a/app/logical/sources/strategies/twit_pic.rb
+++ b/app/logical/sources/strategies/twit_pic.rb
@@ -7,10 +7,6 @@ module Sources::Strategies
      parsed_url&.site_name == "TwitPic"
    end

-    def site_name
-      parsed_url.site_name
-    end
-
    def normalize_for_source
      parsed_url.page_url || url
    end
--- a/app/logical/sources/strategies/twitter.rb
+++ b/app/logical/sources/strategies/twitter.rb
@@ -27,10 +27,6 @@ module Sources::Strategies
      parsed_url&.site_name == "Twitter"
    end

-    def site_name
-      parsed_url.site_name
-    end
-
    def image_urls
      # https://pbs.twimg.com/media/EBGbJe_U8AA4Ekb.jpg:orig
      if parsed_url.image_url?
--- a/app/logical/sources/strategies/weibo.rb
+++ b/app/logical/sources/strategies/weibo.rb
@@ -9,10 +9,6 @@ module Sources
        Source::URL::Weibo === parsed_url
      end

-      def site_name
-        parsed_url.site_name
-      end
-
      def image_urls
        if parsed_url.image_url?
          [parsed_url.full_image_url]
--- a/app/models/artist_url.rb
+++ b/app/models/artist_url.rb
@@ -72,12 +72,11 @@ class ArtistURL < ApplicationRecord
  end

  def domain
-    Danbooru::URL.parse(normalized_url)&.domain.to_s
+    parsed_url&.domain.to_s
  end

  def site_name
-    source = Sources::Strategies.find(normalized_url)
-    source.site_name
+    parsed_url&.site_name.to_s
  end

  # A secondary URL is an artist URL that we don't normally want to display,
@@ -123,9 +122,14 @@ class ArtistURL < ApplicationRecord

  def url=(url)
    super(url)
+    @parsed_url = Source::URL.parse(url)
    self.normalized_url = self.class.normalize_normalized_url(self.url)
  end

+  def parsed_url
+    @parsed_url ||= Source::URL.parse(url)
+  end
+
  def to_s
    if is_active?
      url