Refactor sync script

.coderabbit.yaml

@@ -2,5 +2,7 @@ early_access: true
 reviews:
   path_filters:
     - "!content/blog/!*.md"
+    - "!content/blog/sync_status.yml"
+    - "!test/fixtures/"
     - "!wp-content/**"
     - "!wp-includes/**"

.github/workflows/sync-and-publish.yml

@@ -33,9 +33,9 @@ jobs:
         run: |
           if [ "${{ github.event.inputs.force }}" = "true" ]
           then
-            bin/from_devto -f
+            bin/sync_with_devto -f
           else
-            bin/from_devto
+            bin/sync_with_devto
           fi
 
           bin/upload_assets_to_github

bin/sync/article_cleaner.rb

@@ -0,0 +1,49 @@
+require 'fileutils'
+require 'yaml'
+
+module ArticleCleaner
+  SYNC_STATUS_FILE = 'sync_status.yml'.freeze
+  ARTICLE_FILE = 'index.md'.freeze
+
+  def cleanup_renamed_articles
+    raise ArgumentError, "Working directory doesn't exist" unless Dir.exist?(working_dir)
+
+    deleted_folders = []
+    slugs = load_slugs_from_yaml
+
+    Dir.glob("#{working_dir}/*").each do |folder_path|
+      next unless File.directory?(folder_path) && File.exist?("#{folder_path}/#{ARTICLE_FILE}")
+
+      folder_name = File.basename(folder_path)
+      unless slugs.include?(folder_name)
+        begin
+          FileUtils.rm_rf(folder_path)
+          deleted_folders << folder_name
+          puts "Deleted folder: #{folder_name}"
+        rescue StandardError => e
+          puts "Failed to delete folder #{folder_name}: #{e.message}"
+        end
+      end
+    end
+    deleted_folders
+  end
+
+  private
+
+  def load_slugs_from_yaml
+    yaml_path = File.join(working_dir, SYNC_STATUS_FILE)
+
+    begin
+      yaml_data = YAML.load_file(yaml_path)
+      raise "Invalid YAML structure" unless yaml_data.is_a?(Hash)
+
+      yaml_data.values.map do |article|
+        raise "Invalid article data structure" unless article.is_a?(Hash) && article[:slug]
+        article[:slug]
+      end
+    rescue StandardError => e
+      logger.error "Failed to load slugs from YAML: #{e.message}"
+      []
+    end
+  end
+end

bin/sync/article_sync_checker.rb

@@ -0,0 +1,61 @@
+require 'json'
+
+module ArticleSyncChecker
+  USERNAME = 'jetthoughts'.freeze
+  SYNC_STATUS_FILE = 'sync_status.yml'.freeze
+  USELESS_WORDS = %w[and the a but to is so].freeze
+
+  def update_sync_status
+    ensure_sync_status_file_exists
+    @sync_status = sync_status
+    update_status(fetch_articles)
+    save_sync_status
+  end
+
+  private
+
+  def ensure_sync_status_file_exists
+    sync_file_path = File.join(working_dir, SYNC_STATUS_FILE)
+
+    unless File.exist?(sync_file_path)
+      File.open(sync_file_path, 'w') { |file| file.write({}.to_yaml) }
+    end
+  end
+
+  def save_sync_status
+    File.open(File.join(working_dir, SYNC_STATUS_FILE), 'w') do |file|
+      file.write(@sync_status.to_yaml)
+    end
+  end
+
+  def fetch_articles
+    response = http_client.get_articles(USERNAME, 0)
+    JSON.parse(response.body)
+  end
+
+  def slug(article)
+    slug_parts = article['slug'].split('-')[0..-2]
+    tags = article['tags'] ? article['tags'].split(", ") : []
+    selected_tags = tags.first(2)
+    [slug_parts, selected_tags]
+      .flatten
+      .uniq
+      .reject { |segment| USELESS_WORDS.include?(segment) }
+      .compact
+      .join('-')
+  end
+
+  def update_status(articles)
+    articles.each do |article|
+      id = article['id']
+      edited_at = article["edited_at"] || article["created_at"]
+
+      @sync_status[id] ||= { edited_at: edited_at, slug: slug(article), synced: false }
+
+      if @sync_status[id][:edited_at] != edited_at
+        @sync_status[id][:edited_at] = edited_at
+        @sync_status[id][:synced] = false
+      end
+    end
+  end
+end