posts: automatically add the bad_link and bad_source tags.

Automatically add the bad_link tag when the source is an image url from
a known site, but it can't be converted to a page url (for example, a
Twitter or Tumblr direct image link).

Automatically add the bad_source tag when the source is from a known
site, but it's not an image or page url (for example, a Twitter or Pixiv
profile url)

app/logical/source/url.rb

@@ -9,6 +9,14 @@
 # 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.
 #
+# The following methods should be implemented by subclasses:
+#
+# * match?
+# * parse
+# * image_url?
+# * page_url
+# * profile_url
+#
 # Source::URL is a subclass of Danbooru::URL, so it inherits some common utility methods
 # from there.
 #
@@ -87,6 +95,13 @@ module Source
       self.class.name.demodulize.titleize
     end
 
+    # True if the URL is from a recognized site. False if the URL is from an unrecognized site.
+    #
+    # @return [Boolean]
+    def recognized?
+      true # overridden in Source::URL::Null to return false for unknown sites
+    end
+
     # True if the URL is a direct image URL.
     #
     # Examples:

app/logical/source/url/null.rb

@@ -38,6 +38,8 @@ class Source::URL::Null < Source::URL
   end
 
   def parse
+    @recognized = true
+
     case [subdomain, domain, *path_segments]
 
     # http://about.me/rig22
@@ -240,7 +242,12 @@ class Source::URL::Null < Source::URL
       @page_url = "https://www.zerochan.net/#{@work_id}#full"
 
     else
-      nil
+      @recognized = false
+
     end
   end
+
+  def recognized?
+    @recognized
+  end
 end

app/models/post.rb

@@ -430,6 +430,23 @@ class Post < ApplicationRecord
         tags << "non-web_source"
       end
 
+      source_url = parsed_source
+      if source_url.present? && source_url.recognized?
+        # A bad_link is an image URL from a recognized site that can't be converted to a page URL.
+        if source_url.image_url? && source_url.page_url.nil?
+          tags << "bad_link"
+        else
+          tags -= ["bad_link"]
+        end
+
+        # A bad_source is a source from a recognized site that isn't an image url or a page url.
+        if !source_url.image_url? && !source_url.page_url?
+          tags << "bad_source"
+        else
+          tags -= ["bad_source"]
+        end
+      end
+
       # Allow only Flash files to be manually tagged as `animated`; GIFs, PNGs, videos, and ugoiras are automatically tagged.
       tags -= ["animated"] unless is_flash?
       tags << "animated" if media_asset.is_animated?

test/unit/post_test.rb

@@ -1115,11 +1115,62 @@ class PostTest < ActiveSupport::TestCase
       context "a post with a non-web source" do
         should "automatically add the non-web_source tag" do
           @post.update!(source: "this was once revealed to me in a dream")
-          @post.save!
           assert_equal("non-web_source tag1 tag2", @post.tag_string)
         end
       end
 
+      context "a post with a bad_link source" do
+        should "add the bad_link tag for known bad sources" do
+          @post.update!(source: "https://pbs.twimg.com/media/FQjQA1mVgAMcHLv.jpg:orig")
+          assert_equal("bad_link tag1 tag2", @post.tag_string)
+
+          @post.update!(source: "https://media.tumblr.com/570edf684c7eb195d391115f8b18ca55/tumblr_pen2zwt3bK1uh1m9xo1_1280.png")
+          assert_equal("bad_link tag1 tag2", @post.tag_string)
+        end
+
+        should "remove the bad_link tag for known good sources" do
+          @post.update!(tag_string: "bad_link tag1 tag2")
+          @post.update!(source: "https://i.pximg.net/img-original/img/2022/04/25/08/03/14/97867015_p0.png")
+          assert_equal("tag1 tag2", @post.tag_string)
+        end
+
+        should "not add the bad_link tag for unknown sources" do
+          @post.update!(source: "https://www.example.com/image.jpg")
+          assert_equal("tag1 tag2", @post.tag_string)
+        end
+
+        should "not remove the bad_link tag for unknown sources" do
+          @post.update!(tag_string: "bad_link tag1 tag2", source: "https://www.example.com/image.jpg")
+          assert_equal("bad_link tag1 tag2", @post.tag_string)
+        end
+      end
+
+      context "a post with a bad source" do
+        should "add the bad_source tag for known bad sources" do
+          @post.update!(source: "https://twitter.com/danboorubot/")
+          assert_equal("bad_source tag1 tag2", @post.tag_string)
+
+          @post.update!(source: "https://www.pixiv.net/en/users/6210796")
+          assert_equal("bad_source tag1 tag2", @post.tag_string)
+        end
+
+        should "remove the bad_source tag for known good sources" do
+          @post.update!(tag_string: "bad_source tag1 tag2")
+          @post.update!(source: "https://twitter.com/kafun/status/1520766650907521024")
+          assert_equal("tag1 tag2", @post.tag_string)
+        end
+
+        should "not add the bad_source tag for unknown sources" do
+          @post.update!(source: "https://www.example.com/image.html")
+          assert_equal("tag1 tag2", @post.tag_string)
+        end
+
+        should "not remove the bad_source tag for unknown sources" do
+          @post.update!(tag_string: "bad_source tag1 tag2", source: "https://www.example.com/image.html")
+          assert_equal("bad_source tag1 tag2", @post.tag_string)
+        end
+      end
+
       context "a post with a blank source" do
         should "remove the non-web_source tag" do
           @post.update!(source: "", tag_string: "non-web_source")