sources: factor out Source::URL::Newgrounds.

Also fix it so that the image URL is set as the source for Newgrounds posts, not the page URL. It's possible to generate the page URL from the image URL (except for images after the first in multi-image posts). * Page: https://www.newgrounds.com/art/view/natthelich/weaver * Image: https://art.ngfiles.com/images/1520000/1520217_natthelich_weaver.jpg?f1606365031
2022-02-25 22:00:42 -06:00
parent 64472a7b7e
commit f062f2d145
4 changed files with 123 additions and 27 deletions
--- a/app/logical/source/url.rb
+++ b/app/logical/source/url.rb
@@ -20,6 +20,7 @@ module Source
    SUBCLASSES = [
      Source::URL::Twitter,
      Source::URL::HentaiFoundry,
+      Source::URL::Newgrounds,
      Source::URL::Plurk,
      Source::URL::Skeb,
      Source::URL::TwitPic,
--- a/app/logical/source/url/newgrounds.rb
+++ b/app/logical/source/url/newgrounds.rb
@@ -0,0 +1,92 @@
+# frozen_string_literal: true
+
+# Image Urls
+#
+# * https://art.ngfiles.com/images/1543000/1543982_natthelich_pandora-2.jpg?f1607971817
+# * https://art.ngfiles.com/images/1033000/1033622_natthelich_fire-emblem-marth-plus-progress-pic.png?f1569487181
+#
+# * https://www.newgrounds.com/art/view/natthelich/weaver (page)
+# * https://art.ngfiles.com/images/1520000/1520217_natthelich_weaver.jpg?f1606365031
+# * https://art.ngfiles.com/comments/57000/iu_57615_7115981.jpg
+#
+# Thumbnail URLs
+#
+# * https://art.ngfiles.com/thumbnails/1543000/1543982_full.png?f1607971901
+# * https://art.ngfiles.com/thumbnails/1254000/1254985.png?f1588263349
+#
+# Page URLs
+#
+# * https://www.newgrounds.com/art/view/puddbytes/costanza-at-bat
+# * https://www.newgrounds.com/art/view/natthelich/weaver (multiple)
+#
+# Video URLs
+#
+# * https://www.newgrounds.com/portal/view/825916 (page)
+# * https://picon.ngfiles.com/825000/flash_825916_card.png?f1639666239 (poster)
+# * https://uploads.ungrounded.net/alternate/1801000/1801343_alternate_165104.1080p.mp4?1639666238
+# * https://uploads.ungrounded.net/alternate/1801000/1801343_alternate_165104.720p.mp4?1639666238
+# * https://uploads.ungrounded.net/alternate/1801000/1801343_alternate_165104.360p.mp4?1639666238
+#
+# Flash URLs
+#
+# * https://www.newgrounds.com/portal/view/225625 (page)
+# * https://uploads.ungrounded.net/225000/225625_colormedressup.swf?1111143751 (file)
+#
+# Other URLs
+#
+# * https://www.newgrounds.com/reviews/portal/1543982/4/
+# * https://www.newgrounds.com/reviews/portal/1543982/4/score/1
+# * https://www.newgrounds.com/content/share/1543982/4/
+# * https://www.newgrounds.com/favorites/content/who/1543982/4
+#
+# Profile URLs
+#
+# * https://natthelich.newgrounds.com
+# * https://natthelich.newgrounds.com/art
+#
+class Source::URL::Newgrounds < Source::URL
+  attr_reader :username, :work_id, :work_title
+
+  def self.match?(url)
+    url.domain.in?(["newgrounds.com", "ngfiles.com", "ungrounded.net"])
+  end
+
+  def parse
+    case [host, *path_segments]
+
+    # https://www.newgrounds.com/art/view/puddbytes/costanza-at-bat
+    # https://www.newgrounds.com/art/view/natthelich/fire-emblem-marth-plus-progress-pic
+    in "www.newgrounds.com", "art", "view", username, work_title
+      @username = username
+      @work_title = work_title
+
+    # https://art.ngfiles.com/images/1254000/1254722_natthelich_pandora.jpg
+    # https://art.ngfiles.com/images/1033000/1033622_natthelich_fire-emblem-marth-plus-progress-pic.png?f1569487181
+    in "art.ngfiles.com", "images", _, /^(\d+)_([^_]+)_(.*)\.\w+$/ => filename
+      @work_id = $1
+      @username = $2
+      @work_title = $3
+      @filename = filename
+
+    # https://art.ngfiles.com/thumbnails/1254000/1254985.png?f1588263349
+    in "art.ngfiles.com", "thumbnails", _, /^(\d+)\.\w+$/ => filename
+      @work_id = $1
+      @filename = filename
+
+    # https://art.ngfiles.com/comments/57000/iu_57615_7115981.jpg
+    in "art.ngfiles.com", "comments", _, /^iu/ => filename
+      @filename = filename
+
+    # https://natthelich.newgrounds.com
+    # https://natthelich.newgrounds.com/art/
+    in /^([a-z0-9-]+)\.newgrounds\.com$/, *rest if host != "www.newgrounds.com"
+      @username = $1
+
+    else
+    end
+  end
+
+  def image_url?
+    url.host == "art.ngfiles.com"
+  end
+end
--- a/app/logical/sources/strategies/newgrounds.rb
+++ b/app/logical/sources/strategies/newgrounds.rb
@@ -1,37 +1,19 @@
 # frozen_string_literal: true

-# Image Urls
-# * https://art.ngfiles.com/images/1254000/1254722_natthelich_pandora.jpg
-# * https://art.ngfiles.com/images/1033000/1033622_natthelich_fire-emblem-marth-plus-progress-pic.png?f1569487181
-# * https://art.ngfiles.com/comments/57000/iu_57615_7115981.jpg
-#
-# Page URLs
-# * https://www.newgrounds.com/art/view/puddbytes/costanza-at-bat
-# * https://www.newgrounds.com/art/view/natthelich/fire-emblem-marth-plus-progress-pic (multiple)
-#
-# Profile URLs
-# * https://natthelich.newgrounds.com/
-
+# @see Source::URL::Newgrounds
 module Sources
  module Strategies
    class Newgrounds < Base
-      IMAGE_URL   = %r{\Ahttps?://art\.ngfiles\.com/images/\d+/\d+_(?<user_name>[0-9a-z-]+)_(?<illust_title>[0-9a-z-]+)\.\w+}i
-      COMMENT_URL = %r{\Ahttps?://art\.ngfiles\.com/comments/\d+/\w+\.\w+}i
-
-      PAGE_URL    = %r{\Ahttps?://(?:www\.)?newgrounds\.com/art/view/(?<user_name>[0-9a-z-]+)/(?<illust_title>[0-9a-z-]+)(?:\?.*)?}i
-
-      PROFILE_URL = %r{\Ahttps?://(?<artist_name>(?!www)[0-9a-z-]+)\.newgrounds\.com(?:/.*)?}i
-
-      def domains
-        ["newgrounds.com", "ngfiles.com"]
+      def match?
+        parsed_url&.site_name == "Newgrounds"
      end

      def site_name
-        "NewGrounds"
+        "Newgrounds"
      end

      def image_urls
-        if url =~ COMMENT_URL || url =~ IMAGE_URL
+        if parsed_url.image_url?
          [url]
        else
          urls = []
@@ -97,16 +79,25 @@ module Sources
        DText.from_html(artist_commentary_desc)
      end

+      # The image url should be the post source, if we can generate the page url from the image url.
+      def canonical_url
+        if page_url.present?
+          url
+        else
+          page_url
+        end
+      end
+
      def normalize_for_source
        page_url
      end

      def user_name
-        urls.map { |u| url[PROFILE_URL, :artist_name] || u[IMAGE_URL, :user_name] || u[PAGE_URL, :user_name] }.compact.first
+        parsed_url.username || parsed_referer&.username
      end

      def illust_title
-        urls.map { |u| u[IMAGE_URL, :illust_title] || u[PAGE_URL, :illust_title] }.compact.first
+        parsed_url.work_title || parsed_referer&.work_title
      end
    end
  end