Merge pull request #6941 from fjordllc/bug/save-rss-with-no-publish-date

記事ごとの登録日時を持たない場合も例外が発生しないよう修正

app/models/external_entry.rb

@@ -17,15 +17,15 @@ class << self
     def fetch_and_save_rss_feeds(users)
       threads = Concurrent::Promises.future do
         users.each do |user|
-          rss_items = ExternalEntry.parse_rss_feed(user.feed_url)
+          feed = ExternalEntry.parse_rss_feed(user.feed_url)
 
-          next unless rss_items
+          next unless feed
 
-          rss_items.each do |item|
+          feed.items.each do |item|
             case item.class.name
-            when 'RSS::Atom::Feed::Entry' then ExternalEntry.save_atom_feed(user, item)
-            when 'RSS::Rss::Channel::Item' then ExternalEntry.save_rss_feed(user, item)
-            when 'RSS::RDF::Item' then ExternalEntry.save_rdf_feed(user, item)
+            when 'RSS::Atom::Feed::Entry' then ExternalEntry.save_atom_feed(user, item, feed.updated&.content)
+            when 'RSS::Rss::Channel::Item' then ExternalEntry.save_rss_feed(user, item, feed.channel&.lastBuildDate)
+            when 'RSS::RDF::Item' then ExternalEntry.save_rdf_feed(user, item, feed.channel&.dc_date)
             end
           end
         end
@@ -38,50 +38,81 @@ def parse_rss_feed(feed_url)
       return nil if feed_url.blank?
 
       begin
-        RSS::Parser.parse(feed_url).items
+        RSS::Parser.parse(feed_url)
       rescue StandardError => e
         logger.warn("[RSS Feed] #{feed_url}: #{e.message}")
         nil
       end
     end
 
-    def save_rdf_feed(user, rdf_item)
+    def save_rdf_feed(user, rdf_item, feed_updated)
       return if ExternalEntry.find_by(url: rdf_item.link)
 
+      published_at = rdf_publish_date(rdf_item, feed_updated)
       ExternalEntry.create(
         title: rdf_item.title,
         url: rdf_item.link,
         summary: rdf_item.description,
         thumbnail_image_url: nil,
-        published_at: rdf_item.dc_date,
+        published_at: published_at.utc? ? Time.zone.parse(published_at.to_s) : published_at,
         user:
       )
     end
 
-    def save_rss_feed(user, rss_item)
+    def save_rss_feed(user, rss_item, feed_updated)
       return if ExternalEntry.find_by(url: rss_item.link)
 
+      published_at = rss_publish_date(rss_item, feed_updated)
       ExternalEntry.create(
         title: rss_item.title,
         url: rss_item.link,
         summary: rss_item.description,
         thumbnail_image_url: rss_item.enclosure&.url,
-        published_at: rss_item.pubDate,
+        published_at: published_at.utc? ? Time.zone.parse(published_at.to_s) : published_at,
         user:
       )
     end
 
-    def save_atom_feed(user, atom_item)
-      return if ExternalEntry.find_by(url: atom_item.link.href)
+    def save_atom_feed(user, atom_item, feed_updated)
+      return if ExternalEntry.find_by(url: atom_item.link&.href)
 
+      published_at = atom_publish_date(atom_item, feed_updated)
       ExternalEntry.create(
-        title: atom_item.title.content,
-        url: atom_item.link.href,
-        summary: atom_item.content.content,
-        thumbnail_image_url: atom_item.links.find { |link| !link.type.nil? && link.type.include?('image') }&.href,
-        published_at: atom_item.published.content,
+        title: atom_item.title&.content,
+        url: atom_item.link&.href,
+        summary: atom_item.content&.content,
+        thumbnail_image_url: atom_image_url(atom_item),
+        published_at: published_at.utc? ? Time.zone.parse(published_at.to_s) : published_at,
         user:
       )
     end
+
+    private
+
+    def rdf_publish_date(rdf_item, feed_updated)
+      return rdf_item.dc_date if rdf_item.dc_date
+      return feed_updated if feed_updated
+
+      Time.zone.now
+    end
+
+    def rss_publish_date(rss_item, feed_updated)
+      return rss_item.pubDate if rss_item.pubDate
+      return feed_updated if feed_updated
+
+      Time.zone.now
+    end
+
+    def atom_image_url(atom_item)
+      atom_item.links&.find { |link| !link.type.nil? && link.type.include?('image') }&.href
+    end
+
+    def atom_publish_date(atom_item, feed_updated)
+      return atom_item.published.content if atom_item.published
+      return atom_item.updated.content if atom_item.updated
+      return feed_updated if feed_updated
+
+      Time.zone.now
+    end
   end
 end

test/models/external_entry_test.rb

@@ -32,29 +32,84 @@ class ExternalEntryTest < ActiveSupport::TestCase
     rdf_item.expect(:link, 'https://test.com/index.rdf')
     rdf_item.expect(:description, 'article description')
     rdf_item.expect(:dc_date, Time.zone.local(2022, 1, 1, 0, 0, 0))
+    rdf_item.expect(:dc_date, Time.zone.local(2022, 1, 1, 0, 0, 0))
+
+    assert ExternalEntry.save_rdf_feed(user, rdf_item, nil)
+  end
+
+  test '.save_rdf_feed with no article publish date' do
+    user = users(:hatsuno)
+    rdf_item = Minitest::Mock.new
+    rdf_item.expect(:link, 'https://test.com/index.rdf')
+    rdf_item.expect(:title, 'test title')
+    rdf_item.expect(:link, 'https://test.com/index.rdf')
+    rdf_item.expect(:description, 'article description')
+    rdf_item.expect(:dc_date, nil)
+
+    assert ExternalEntry.save_rdf_feed(user, rdf_item, nil)
+  end
+
+  test '.save_rdf_feed no exception even if all rdf elements are nil' do
+    user = users(:hatsuno)
+    rdf_item = Minitest::Mock.new
+    rdf_item.expect(:link, nil)
+    rdf_item.expect(:title, nil)
+    rdf_item.expect(:link, nil)
+    rdf_item.expect(:description, nil)
+    rdf_item.expect(:dc_date, nil)
 
-    assert ExternalEntry.save_rdf_feed(user, rdf_item)
+    assert ExternalEntry.save_rdf_feed(user, rdf_item, Time.zone.local(2023, 1, 1, 0, 0, 0).utc)
   end
 
   test '.save_rss_feed' do
     user = users(:hatsuno)
     rss_item = Minitest::Mock.new
     rss_item.expect(:link, 'https://test.com/rss')
+    rss_item.expect(:pubDate, Time.zone.local(2022, 1, 1, 0, 0, 0))
+    rss_item.expect(:pubDate, Time.zone.local(2022, 1, 1, 0, 0, 0))
     rss_item.expect(:title, 'test title')
     rss_item.expect(:link, 'https://test.com/rss')
     rss_item.expect(:description, 'article description')
     rss_item.expect(:enclosure, rss_item)
     rss_item.expect(:url, 'https://test.com/image.png')
-    rss_item.expect(:pubDate, Time.zone.local(2022, 1, 1, 0, 0, 0))
 
-    assert ExternalEntry.save_rss_feed(user, rss_item)
+    assert ExternalEntry.save_rss_feed(user, rss_item, nil)
+  end
+
+  test '.save_rss_feed with no article publish date' do
+    user = users(:hatsuno)
+    rss_item = Minitest::Mock.new
+    rss_item.expect(:link, 'https://test.com/rss')
+    rss_item.expect(:pubDate, nil)
+    rss_item.expect(:title, 'test title')
+    rss_item.expect(:link, 'https://test.com/rss')
+    rss_item.expect(:description, 'article description')
+    rss_item.expect(:enclosure, nil)
+
+    assert ExternalEntry.save_rss_feed(user, rss_item, Time.zone.local(2023, 1, 1, 0, 0, 0).utc)
+  end
+
+  test '.save_rss_feed no exception even if all rss elements are nil' do
+    user = users(:hatsuno)
+    rss_item = Minitest::Mock.new
+    rss_item.expect(:link, nil)
+    rss_item.expect(:pubDate, nil)
+    rss_item.expect(:title, nil)
+    rss_item.expect(:link, nil)
+    rss_item.expect(:description, nil)
+    rss_item.expect(:enclosure, nil)
+
+    assert ExternalEntry.save_rss_feed(user, rss_item, nil)
   end
 
   test '.save_atom_feed' do
     user = users(:hatsuno)
     atom_item = Minitest::Mock.new
     atom_item.expect(:link, atom_item)
     atom_item.expect(:href, 'https://test.com/feed')
+    atom_item.expect(:published, atom_item)
+    atom_item.expect(:published, atom_item)
+    atom_item.expect(:content, Time.zone.local(2022, 1, 1, 0, 0, 0))
     atom_item.expect(:title, atom_item)
     atom_item.expect(:content, 'test title')
     atom_item.expect(:link, atom_item)
@@ -65,10 +120,44 @@ class ExternalEntryTest < ActiveSupport::TestCase
     atom_item.expect(:find, atom_item)
     atom_item.expect(:nil?, atom_item)
     atom_item.expect(:href, 'https://test.com/image.png')
-    atom_item.expect(:published, atom_item)
-    atom_item.expect(:content, Time.zone.local(2022, 1, 1, 0, 0, 0))
 
-    assert ExternalEntry.save_atom_feed(user, atom_item)
+    assert ExternalEntry.save_atom_feed(user, atom_item, nil)
+  end
+
+  test '.save_atom_feed with no article publish date' do
+    user = users(:hatsuno)
+    atom_item = Minitest::Mock.new
+    atom_item.expect(:link, atom_item)
+    atom_item.expect(:href, 'https://test.com/feed')
+    atom_item.expect(:published, nil)
+    atom_item.expect(:updated, nil)
+    atom_item.expect(:title, atom_item)
+    atom_item.expect(:content, 'test title')
+    atom_item.expect(:link, atom_item)
+    atom_item.expect(:href, 'https://test.com/feed')
+    atom_item.expect(:content, atom_item)
+    atom_item.expect(:content, 'article description')
+    atom_item.expect(:links, atom_item)
+    atom_item.expect(:find, atom_item)
+    atom_item.expect(:nil?, atom_item)
+    atom_item.expect(:href, 'https://test.com/image.png')
+
+    assert ExternalEntry.save_atom_feed(user, atom_item, Time.zone.local(2023, 1, 1, 0, 0, 0).utc)
+  end
+
+  test '.save_atom_feed no exception even if all atom elements are nil' do
+    user = users(:hatsuno)
+    atom_item = Minitest::Mock.new
+    atom_item.expect(:link, nil)
+    atom_item.expect(:published, nil)
+    atom_item.expect(:updated, nil)
+    atom_item.expect(:title, nil)
+    atom_item.expect(:content, nil)
+    atom_item.expect(:link, nil)
+    atom_item.expect(:content, nil)
+    atom_item.expect(:links, nil)
+
+    assert ExternalEntry.save_atom_feed(user, atom_item, nil)
   end
 
   test '.fetch_and_save_rss_feeds' do