[rzz] create feed from the webpage instead

ansible/docker/rzz/rzz.rb

@@ -8,36 +8,32 @@ module Rzz
 
   class App < Roda
     plugin :caching
+    plugin :json
 
     route do |r|
       r.get do
         r.is "test-gym" do
-          url = "https://elemental.medium.com/feed"
+          url = "https://elemental.medium.com/test-gym/home"
-          xml = URI.open(url).read
+          html = URI.open(url).read
-          r.etag Digest::SHA1.hexdigest(xml)
+
-
+          doc = Nokogiri::HTML(html)
-          doc = Nokogiri::XML(xml)
+          items = doc.xpath("//a[@data-post-id]").map {|a|
-
+            attrs = a.attributes
-          self_url = "#{r.base_url}#{r.path}" # Should this be configured via the environment?
+            id = attrs.fetch("data-post-id").value
-          doc.at_xpath("/rss/channel/link/text()").content = self_url
+            title = a.xpath("./h3").text
-          doc.at_xpath("/rss/channel/atom:link[@href='#{url}']")["href"] = self_url
+            content = a.xpath("./div").text
-
+            href = URI(attrs.fetch("href"))
-          doc.at_xpath("/rss/channel/atom:link[@rel='hub']").remove
+            url = "#{href.scheme}://#{href.host}#{href.path}"
-
+            { id: id, title: title, content_text: content, url: url }
-          doc
+          }
-            .xpath("/rss/channel/item")
+
-            .select {|item| item.xpath("./category[text()='test-gym']").empty? }
+          {
-            .each(&:remove)
+            version: "https://jsonfeed.org/version/1",
-
+            title: "Test Gym",
-          doc
+            home_page_url: url,
-            .xpath("/rss/channel/item")
+            feed_url: "#{r.base_url}#{r.path}",
-            .each do |item|
+            items: items,
-              link = item.at_xpath("./link/text()").content
+          }
-              item.at_xpath("./description/text()").content = URI.open(link).read
-            end
-
-          response.headers["Content-Type"] = "text/xml"
-          doc.to_xml
         end
       end
     end