From 14196f36d89a5150c6c126fab792998891e6b421 Mon Sep 17 00:00:00 2001
From: evazion <noizave@gmail.com>
Date: Sun, 11 Jun 2017 00:00:25 -0500
Subject: [PATCH] pawoo: convert commentary to dtext.

---
 app/logical/sources/strategies/pawoo.rb | 35 ++++++++++++++++++++++++-
 test/unit/sources/pawoo_test.rb         | 10 +++++++
 2 files changed, 44 insertions(+), 1 deletion(-)

diff --git a/app/logical/sources/strategies/pawoo.rb b/app/logical/sources/strategies/pawoo.rb
index 33c478453..1605332e0 100644
--- a/app/logical/sources/strategies/pawoo.rb
+++ b/app/logical/sources/strategies/pawoo.rb
@@ -27,8 +27,12 @@ module Sources::Strategies
       "Pawoo"
     end
 
+    def api_response
+      @response ||= PawooApiClient.new.get_status(normalized_url)
+    end
+
     def get
-      response = PawooApiClient.new.get_status(normalized_url)
+      response = api_response
       @artist_name = response.account_name
       @profile_url = response.account_profile_url
       @image_url = response.image_urls.first
@@ -49,5 +53,34 @@ module Sources::Strategies
     def normalizable_for_artist_finder?
       true
     end
+
+    def dtext_artist_commentary_desc
+      to_dtext(artist_commentary_desc)
+    end
+
+    def to_dtext(text)
+      html = Nokogiri::HTML.fragment(text)
+
+      dtext = html.children.map do |element|
+        case element.name
+        when "text"
+          element.content
+        when "p"
+          to_dtext(element.inner_html) + "\n\n"
+        when "a"
+          # don't include links to the toot itself.
+          media_urls = api_response.json["media_attachments"].map { |attr| attr["text_url"] }
+          next if element.attribute("href").value.in?(media_urls)
+
+          title = to_dtext(element.inner_html)
+          url = element.attributes["href"].value
+          %("#{title}":[#{url}])
+        else
+          to_dtext(element.inner_html)
+        end
+      end.join.strip
+
+      dtext
+    end
   end
 end
diff --git a/test/unit/sources/pawoo_test.rb b/test/unit/sources/pawoo_test.rb
index 9e0338b32..bfa34a3cd 100644
--- a/test/unit/sources/pawoo_test.rb
+++ b/test/unit/sources/pawoo_test.rb
@@ -19,6 +19,16 @@ module Sources
       should "get the image url" do
         assert_equal("https://img.pawoo.net/media_attachments/files/000/128/953/original/4c0a06087b03343f.png", @site.image_url)
       end
+
+      should "get the commentary" do
+        desc = '<p>a mind forever voyaging through strange seas of thought alone <a href="https://pawoo.net/media/9hJzXvwxVl1CezW0ecM" rel="nofollow noopener" target="_blank"><span class="invisible">https://</span><span class="ellipsis">pawoo.net/media/9hJzXvwxVl1Cez</span><span class="invisible">W0ecM</span></a></p>'
+        assert_equal(desc, @site.artist_commentary_desc)
+      end
+
+      should "get the dtext-ified commentary" do
+        desc = 'a mind forever voyaging through strange seas of thought alone'
+        assert_equal(desc, @site.dtext_artist_commentary_desc)
+      end
     end
 
     context "The source site for a https://pawoo.net/$user/$id url"  do