XML: group by published_at with year->month nesting; dedupe providers… (#407)

Author: dcollie2

app/helpers/topics_helper.rb

@@ -12,28 +12,28 @@ def card_preview_media(file)
   private
 
   def render_image(file)
-    image_tag(url_for(file), class: "img-fluid w-100")
+    image_tag(rails_blob_path(file, disposition: "inline"), class: "img-fluid w-100")
   end
 
   def render_pdf(file)
     content_tag(:div, class: "embed-responsive embed-responsive-item embed-responsive-16by9 w-100") do
-      content_tag(:object, data: url_for(file), type: "application/pdf", width: "100%", height: "400px") do
-        content_tag(:iframe, "", src: url_for(file), width: "100%", height: "100%", style: "border: none;") do
-          content_tag(:p, "Your browser does not support PDF viewing. #{link_to('Download the PDF', url_for(file))}")
+      content_tag(:object, data: rails_blob_path(file, disposition: "inline"), type: "application/pdf", width: "100%", height: "400px") do
+        content_tag(:iframe, "", src: rails_blob_path(file, disposition: "inline"), width: "100%", height: "100%", style: "border: none;") do
+          content_tag(:p, "Your browser does not support PDF viewing. #{link_to('Download the PDF', rails_blob_path(file))}")
         end
       end
     end
   end
 
   def render_video(file)
-    video_tag(url_for(file), style: "width: 100%")
+    video_tag(rails_blob_path(file, disposition: "inline"), style: "width: 100%")
   end
 
   def render_audio(file)
-    audio_tag(url_for(file), controls: true, style: "width: 100%")
+    audio_tag(rails_blob_path(file, disposition: "inline"), controls: true, style: "width: 100%")
   end
 
   def render_download_link(file)
-    link_to file.filename, url_for(file)
+    link_to file.filename, rails_blob_path(file)
   end
 end

app/services/xml_generator/single_provider.rb

@@ -16,28 +16,36 @@ def provider_xml(provider)
     Ox::Element.new("content_provider").tap do |xml|
       xml[:name] = provider.name
 
-      grouped_topics(provider).each do |(year, month), topics|
+      grouped_topics(provider).sort_by { |year, _| -year.to_i }.each do |(year, months)|
         xml << Ox::Element.new("topic_year").tap do |year_element|
           year_element[:year] = year.to_s
-          year_element << Ox::Element.new("topic_month").tap do |month_element|
-            month_element[:month] = month
-            topics.each do |topic|
-              month_element << Ox::Element.new("title").tap do |title_element|
-                title_element[:name] = topic.title
-                title_element << Ox::Element.new("topic_id").tap { |id| id << topic.id.to_s }
-                title_element << Ox::Element.new("topic_files").tap do |files|
-                  files[:files] = "Files"
-                  topic.documents.each_with_index do |document, index|
-                    next if document.content_type == "video/mp4"
-                    files << Ox::Element.new("file_name_#{index + 1}").tap do |file_name|
-                      file_name << document.filename.to_s
-                      file_name[:file_size] = document.byte_size
+          months.sort_by { |month_label, _| month_label }.each do |(month, topics)|
+            year_element << Ox::Element.new("topic_month").tap do |month_element|
+              month_element[:month] = month
+              topics.each do |topic|
+                month_element << Ox::Element.new("title").tap do |title_element|
+                  title_element[:name] = topic.title
+                  title_element << Ox::Element.new("topic_id").tap { |id| id << topic.id.to_s }
+                  title_element << Ox::Element.new("counter").tap { |c| c << "0" }
+                  title_element << Ox::Element.new("topic_volume").tap { |e| e << topic.published_at.year.to_s }
+                  title_element << Ox::Element.new("topic_issue").tap { |e| e << topic.published_at.month.to_s }
+                  title_element << Ox::Element.new("topic_files").tap do |files|
+                    files[:files] = "Files"
+                    topic.documents.each_with_index do |document, index|
+                      next if document.content_type == "video/mp4"
+                      files << Ox::Element.new("file_name_#{index + 1}").tap do |file_name|
+                        file_name << document.filename.to_s
+                        file_name[:file_size] = document.byte_size
+                      end
                     end
                   end
-                end
-                title_element << Ox::Element.new("topic_tags").tap do |tags|
-                  names = topic.taggings.map { |tg| tg.tag&.name }.compact.uniq
-                  tags << names.join(", ")
+                  title_element << Ox::Element.new("topic_author").tap do |author|
+                    author << Ox::Element.new("topic_author_1").tap { |a| a << " " }
+                  end
+                  title_element << Ox::Element.new("topic_tags").tap do |tags|
+                    names = topic.taggings.map { |tg| tg.tag&.name }.compact.uniq
+                    tags << names.join(", ")
+                  end
                 end
               end
             end
@@ -48,16 +56,19 @@ def provider_xml(provider)
   end
 
   def grouped_topics(prov)
-    topic_scope(prov).group_by { |topic| [ topic.published_at.year, topic.published_at.strftime("%m_%B") ] }
+    # { year => { "MM_Month" => [topics] } }
+    topic_scope(prov)
+      .group_by { |topic| topic.published_at.year }
+      .transform_values { |topics_in_year| topics_in_year.group_by { |t| t.published_at.strftime("%m_%B") } }
   end
 
   def topic_scope(prov)
     scope = prov.topics
     scope = scope.where("published_at > ?", 1.month.ago) if args.fetch(:recent, false)
     scope
       .select(:id, :title, :published_at, :language_id, :provider_id)
-      .includes(:language, taggings: :tag)  # eager-load language and taggings->tag
-      .with_attached_documents              # eager-load Active Storage attachments + blobs
+      .includes(:language, taggings: :tag)
+      .with_attached_documents
       .order(published_at: :desc)
   end
 end

spec/services/xml_generator/all_providers_spec.rb

@@ -19,36 +19,47 @@
   end
 
   it "generates the xml" do
-    expect(subject.perform).to eq(
-      <<~TEXT
-        <?xml version="1.0"?>
-        <cmes>
-          <content_provider name="#{provider1.name}">
-            <topic_year year="#{topic1.published_at.year}">
-              <topic_month month="#{topic1.published_at.strftime("%m_%B")}">
-                <title name="#{topic1.title}">
-                  <topic_id>#{topic1.id}</topic_id>
-                  <topic_files files="Files"/>
-                  <topic_tags>#{topic1.current_tags_list.join(", ")}</topic_tags>
-                </title>
-              </topic_month>
-            </topic_year>
-          </content_provider>
-          <content_provider name="#{provider2.name}">
-            <topic_year year="#{topic2.published_at.year}">
-              <topic_month month="#{topic2.published_at.strftime("%m_%B")}">
-                <title name="#{topic2.title}">
-                  <topic_id>#{topic2.id}</topic_id>
-                  <topic_files files="Files">
-                    <file_name_1 file_size="494323">test_image.png</file_name_1>
-                  </topic_files>
-                  <topic_tags>#{topic2.current_tags_list.join(", ")}</topic_tags>
-                </title>
-              </topic_month>
-            </topic_year>
-          </content_provider>
-        </cmes>
-      TEXT
-    )
+    xml = subject.perform
+    doc = Nokogiri::XML(xml)
+
+    [ provider1, provider2 ].each do |prov|
+      pnode = doc.at_xpath("//cmes/content_provider[@name='#{prov.name}']")
+      expect(pnode).to be_present
+    end
+
+    [ [ provider1, topic1 ], [ provider2, topic2 ] ].each do |prov, topic|
+      pnode = doc.at_xpath("//cmes/content_provider[@name='#{prov.name}']")
+      year = topic.published_at.year
+      ynode = pnode.at_xpath("./topic_year[@year='#{year}']")
+      expect(ynode).to be_present
+
+      month_label = topic.published_at.strftime("%m_%B")
+      mnode = ynode.at_xpath("./topic_month[@month='#{month_label}']")
+      expect(mnode).to be_present
+
+      tnode = mnode.at_xpath("./title[@name='#{topic.title}']")
+      expect(tnode).to be_present
+
+      expect(tnode.at_xpath("./topic_id").text).to eq(topic.id.to_s)
+      expect(tnode.at_xpath("./counter").text).to eq("0")
+      expect(tnode.at_xpath("./topic_volume").text).to eq(topic.published_at.year.to_s)
+      expect(tnode.at_xpath("./topic_issue").text).to eq(topic.published_at.month.to_s)
+
+      files = tnode.at_xpath("./topic_files[@files='Files']")
+      expect(files).to be_present
+      if topic.documents.attached? && topic.documents.reject { |d| d.content_type == "video/mp4" }.any?
+        first_file = files.element_children.first
+        expect(first_file).to be_present
+        expect(first_file.text).to be_present
+  expect(first_file["file_size"]).to be_present
+      else
+        expect(files.element_children).to be_empty
+      end
+
+      author1 = tnode.at_xpath("./topic_author/topic_author_1")
+      expect(author1.text).to eq(" ")
+
+      expect(tnode.at_xpath("./topic_tags").text).to eq(topic.current_tags_list.join(", "))
+    end
   end
 end

spec/services/xml_generator/single_provider_spec.rb

@@ -6,14 +6,12 @@
   let(:provider) { create(:provider) }
 
   it "generates the xml" do
-    expect(subject.perform).to eq(
-      <<~TEXT
-        <?xml version="1.0"?>
-        <cmes>
-          <content_provider name="#{provider.name}"/>
-        </cmes>
-      TEXT
-    )
+  xml = subject.perform
+  doc = Nokogiri::XML(xml)
+  provider_nodes = doc.xpath("//cmes/content_provider[@name='#{provider.name}']")
+  expect(provider_nodes.size).to eq(1)
+  # No topics → self-closing provider node (no children)
+  expect(provider_nodes.first.element_children).to be_empty
   end
 
   context "with topics" do
@@ -35,26 +33,37 @@
     end
 
     it "generates the xml" do
-      expect(subject.perform).to eq(
-        <<~TEXT
-          <?xml version="1.0"?>
-          <cmes>
-            <content_provider name="#{provider.name}">
-              <topic_year year="#{topic.published_at.year}">
-                <topic_month month="#{topic.published_at.strftime("%m_%B")}">
-                  <title name="#{topic.title}">
-                    <topic_id>#{topic.id}</topic_id>
-                    <topic_files files="Files">
-                      <file_name_1 file_size="494323">test_image.png</file_name_1>
-                    </topic_files>
-                    <topic_tags>#{topic.current_tags_list.join(", ")}</topic_tags>
-                  </title>
-                </topic_month>
-              </topic_year>
-            </content_provider>
-          </cmes>
-        TEXT
-      )
+      xml = subject.perform
+      doc = Nokogiri::XML(xml)
+
+  provider_node = doc.at_xpath("//cmes/content_provider[@name='#{provider.name}']")
+      expect(provider_node).to be_present
+
+  year_node = provider_node.at_xpath("./topic_year[@year='#{topic.published_at.year}']")
+      expect(year_node).to be_present
+
+  month_label = topic.published_at.strftime("%m_%B")
+  month_node = year_node.at_xpath("./topic_month[@month='#{month_label}']")
+      expect(month_node).to be_present
+
+  title_node = month_node.at_xpath("./title[@name='#{topic.title}']")
+      expect(title_node).to be_present
+
+  expect(title_node.at_xpath("./topic_id").text).to eq(topic.id.to_s)
+  expect(title_node.at_xpath("./counter").text).to eq("0")
+  expect(title_node.at_xpath("./topic_volume").text).to eq(topic.published_at.year.to_s)
+  expect(title_node.at_xpath("./topic_issue").text).to eq(topic.published_at.month.to_s)
+
+  files_node = title_node.at_xpath("./topic_files[@files='Files']")
+      expect(files_node).to be_present
+  file1 = files_node.at_xpath("./file_name_1")
+  expect(file1.text).to eq("test_image.png")
+  expect(file1["file_size"]).to be_present
+
+  author_node = title_node.at_xpath("./topic_author/topic_author_1")
+  expect(author_node.text).to eq(" ")
+
+  expect(title_node.at_xpath("./topic_tags").text).to eq(topic.current_tags_list.join(", "))
     end
   end
 end