FIX: Importers should create duplicate categories if the parent category is different

script/import_scripts/base.rb

@@ -47,7 +47,7 @@ def initialize
 
     puts "loading existing categories..."
     CategoryCustomField.where(name: 'import_id').pluck(:category_id, :value).each do |category_id, import_id|
-      @categories_lookup[import_id] = Category.find(category_id.to_i)
+      @categories_lookup[import_id] = category_id
     end
 
     puts "loading existing posts..."
@@ -156,7 +156,7 @@ def find_user_by_import_id(import_id)
   end
 
   # Get the Discourse Category id based on the id of the source category
-  def category_from_imported_category_id(import_id)
+  def category_id_from_imported_category_id(import_id)
     @categories_lookup[import_id] || @categories_lookup[import_id.to_s]
   end
 
@@ -330,7 +330,8 @@ def create_categories(results)
     results.each do |c|
       params = yield(c)
 
-      next if params.nil? # block returns nil to skip
+      # block returns nil to skip
+      next if params.nil? || category_id_from_imported_category_id(params[:id])
 
       # Basic massaging on the category name
       params[:name] = "Blank" if params[:name].blank?
@@ -347,13 +348,13 @@ def create_categories(results)
       end
 
       new_category = create_category(params, params[:id])
-      @categories_lookup[params[:id]] = new_category
+      @categories_lookup[params[:id]] = new_category.id
     end
   end
 
   def create_category(opts, import_id)
-    existing = category_from_imported_category_id(import_id) || Category.where("LOWER(name) = ?", opts[:name].downcase).first
-    return existing if existing
+    existing = Category.where("LOWER(name) = ?", opts[:name].downcase).first
+    return existing if existing && existing.parent_category.try(:id) == opts[:parent_category_id]
 
     post_create_action = opts.delete(:post_create_action)
 

script/import_scripts/bbpress.rb

@@ -87,7 +87,7 @@ def import_posts
         mapped[:custom_fields] = {import_id: post["id"]}
 
         if post["post_type"] == "topic"
-          mapped[:category] = category_from_imported_category_id(post["post_parent"]).try(:name)
+          mapped[:category] = category_id_from_imported_category_id(post["post_parent"])
           mapped[:title] = CGI.unescapeHTML post["post_title"]
         else
           parent = topic_lookup_from_imported_post_id(post["post_parent"])

script/import_scripts/bespoke_1.rb

@@ -199,7 +199,7 @@ def import_post_batch!(posts, topics, offset, total)
         topic = topics[post[:topic_id]]
 
         unless topic[:post_id]
-          mapped[:category] = category_from_imported_category_id(topic[:category_id]).try(:name)
+          mapped[:category] = category_id_from_imported_category_id(topic[:category_id])
           mapped[:title] = post[:title]
           topic[:post_id] = post[:id]
         else

script/import_scripts/discuz_x.rb

@@ -171,8 +171,7 @@ def import_categories
         position: row['position'].to_i + max_position
       }
       if row['parent_id'].to_i > 0
-        parent = category_from_imported_category_id(row['parent_id'])
-        h[:parent_category_id] = parent.id if parent
+        h[:parent_category_id] = category_id_from_imported_category_id(row['parent_id'])
       end
       h
     end
@@ -217,7 +216,7 @@ def import_posts
         mapped[:created_at] = Time.zone.at(m['post_time'])
 
         if m['is_first_post'] == 1
-          mapped[:category] = category_from_imported_category_id(m['category_id']).try(:name)
+          mapped[:category] = category_id_from_imported_category_id(m['category_id'])
           mapped[:title] = CGI.unescapeHTML(m['title'])
           @first_post_id_by_topic_id[m['topic_id']] = m['id']
         else

script/import_scripts/drupal.rb

@@ -122,7 +122,7 @@ def create_forum_topics
         {
           id: "nid:#{row['nid']}",
           user_id: user_id_from_imported_user_id(row['uid']) || -1,
-          category: category_from_imported_category_id(row['tid']).try(:name),
+          category: category_id_from_imported_category_id(row['tid']),
           raw: row['body'],
           created_at: Time.zone.at(row['created']),
           pinned_at: row['sticky'].to_i == 1 ? Time.zone.at(row['created']) : nil,

script/import_scripts/drupal_qa.rb

@@ -61,7 +61,7 @@ def create_forum_topics
         {
           id: "nid:#{row['nid']}",
           user_id: user_id_from_imported_user_id(row['uid']) || -1,
-          category: category_from_imported_category_id((row['tid'] || '').split(',')[0]).try(:name),
+          category: category_id_from_imported_category_id((row['tid'] || '').split(',')[0]),
           raw: row['body'],
           created_at: Time.zone.at(row['created']),
           pinned_at: nil,

script/import_scripts/kunena.rb

@@ -41,8 +41,7 @@ def execute
     create_categories(@client.query("SELECT id, parent, name, description, ordering FROM jos_kunena_categories ORDER BY parent, id;")) do |c|
       h = {id: c['id'], name: c['name'], description: c['description'], position: c['ordering'].to_i}
       if c['parent'].to_i > 0
-        parent = category_from_imported_category_id(c['parent'])
-        h[:parent_category_id] = parent.id if parent
+        h[:parent_category_id] = category_id_from_imported_category_id(c['parent'])
       end
       h
     end
@@ -121,7 +120,7 @@ def import_posts
         mapped[:created_at] = Time.zone.at(m['time'])
 
         if m['id'] == m['thread']
-          mapped[:category] = category_from_imported_category_id(m['catid']).try(:name)
+          mapped[:category] = category_id_from_imported_category_id(m['catid'])
           mapped[:title] = m['subject']
         else
           parent = topic_lookup_from_imported_post_id(m['parent'])

script/import_scripts/mybb.rb

@@ -69,8 +69,7 @@ def import_categories
     create_categories(results) do |row|
       h = {id: row['id'], name: CGI.unescapeHTML(row['name']), description: CGI.unescapeHTML(row['description'])}
       if row['parent_id'].to_i > 0
-        parent = category_from_imported_category_id(row['parent_id'])
-        h[:parent_category_id] = parent.id if parent
+        h[:parent_category_id] = category_id_from_imported_category_id(row['parent_id'])
       end
       h
     end
@@ -125,7 +124,7 @@ def import_posts
         mapped[:created_at] = Time.zone.at(m['post_time'])
 
         if m['id'] == m['first_post_id']
-          mapped[:category] = category_from_imported_category_id(m['category_id']).try(:name)
+          mapped[:category] = category_id_from_imported_category_id(m['category_id'])
           mapped[:title] = CGI.unescapeHTML(m['title'])
         else
           parent = topic_lookup_from_imported_post_id(m['first_post_id'])

script/import_scripts/ning.rb

@@ -200,7 +200,7 @@ def import_topics(topics_json, default_category=nil)
           mapped[:user_id] = user_id_from_imported_user_id(topic["contributorName"]) || -1
           mapped[:created_at] = Time.zone.parse(topic["createdDate"])
           unless topic["category"].nil? || topic["category"].downcase == "uncategorized"
-            mapped[:category] = category_from_imported_category_id(topic["category"]).try(:name)
+            mapped[:category] = category_id_from_imported_category_id(topic["category"])
           end
           if topic["category"].nil? && default_category
             mapped[:category] = default_category

script/import_scripts/phpbb3.rb

@@ -114,8 +114,7 @@ def import_categories
     create_categories(results) do |row|
       h = {id: row['id'], name: CGI.unescapeHTML(row['name']), description: CGI.unescapeHTML(row['description'])}
       if row['parent_id'].to_i > 0
-        parent = category_from_imported_category_id(row['parent_id'])
-        h[:parent_category_id] = parent.id if parent
+        h[:parent_category_id] = category_id_from_imported_category_id(row['parent_id'])
       end
       h
     end
@@ -156,7 +155,7 @@ def import_posts
         mapped[:created_at] = Time.zone.at(m['post_time'])
 
         if m['id'] == m['first_post_id']
-          mapped[:category] = category_from_imported_category_id(m['category_id']).try(:name)
+          mapped[:category] = category_id_from_imported_category_id(m['category_id'])
           mapped[:title] = CGI.unescapeHTML(m['title'])
         else
           parent = topic_lookup_from_imported_post_id(m['first_post_id'])

script/import_scripts/smf2.rb

@@ -145,7 +145,7 @@ def import_categories
       FROM {prefix}boards
       ORDER BY id_parent ASC, id_board ASC
     SQL
-      parent_id = category_from_imported_category_id(board[:id_parent]).id if board[:id_parent] > 0
+      parent_id = category_id_from_imported_category_id(board[:id_parent]) if board[:id_parent] > 0
       groups = (board[:member_groups] || "").split(/,/).map(&:to_i)
       restricted = !groups.include?(GUEST_GROUP) && !groups.include?(MEMBER_GROUP)
       {
@@ -257,7 +257,7 @@ def guardian
         end
       }
       if message[:id_msg] == message[:id_first_msg]
-        post[:category] = category_from_imported_category_id(message[:id_board]).try(:name)
+        post[:category] = category_id_from_imported_category_id(message[:id_board])
         post[:title] = decode_entities(message[:subject])
       else
         parent = topic_lookup_from_imported_post_id(message[:id_first_msg])

script/import_scripts/vanilla.rb

@@ -144,8 +144,7 @@ def import_category(category)
         description: clean_up(category[:description]),
       }
       if category[:parent_category_id] != "-1"
-        parent_category = category_from_imported_category_id(category[:parent_category_id])
-        c[:parent_category_id] = parent_category.id if parent_category
+        c[:parent_category_id] = category_id_from_imported_category_id(category[:parent_category_id])
       end
       c
     end
@@ -162,7 +161,7 @@ def import_topics
           id: "discussion#" + discussion[:discussion_id],
           user_id: user_id_from_imported_user_id(discussion[:insert_user_id]) || Discourse::SYSTEM_USER_ID,
           title: discussion[:name],
-          category: category_from_imported_category_id(discussion[:category_id]).try(:name),
+          category: category_id_from_imported_category_id(discussion[:category_id]),
           raw: clean_up(discussion[:body]),
           created_at: parse_date(discussion[:date_inserted]),
         }

script/import_scripts/vbulletin.rb

@@ -181,7 +181,7 @@ def import_categories
         name: @htmlentities.decode(category["title"]).strip,
         position: category["displayorder"],
         description: @htmlentities.decode(category["description"]).strip,
-        parent_category_id: category_from_imported_category_id(category["parentid"]).try(:[], "id")
+        parent_category_id: category_id_from_imported_category_id(category["parentid"])
       }
     end
   end
@@ -216,7 +216,7 @@ def import_topics
           id: topic_id,
           user_id: user_id_from_imported_user_id(topic["postuserid"]) || Discourse::SYSTEM_USER_ID,
           title: @htmlentities.decode(topic["title"]).strip[0...255],
-          category: category_from_imported_category_id(topic["forumid"]).try(:name),
+          category: category_id_from_imported_category_id(topic["forumid"]),
           raw: raw,
           created_at: parse_timestamp(topic["dateline"]),
           visible: topic["visible"].to_i == 1,

script/import_scripts/vbulletin_old.rb

@@ -460,7 +460,7 @@ def import_topics
           id: id,
           user_id: user_id_from_imported_user_id(topic[:postuserid]) || Discourse::SYSTEM_USER_ID,
           title: CGI.unescapeHTML(topic[:title]).strip[0...255],
-          category: category_from_imported_category_id(topic[:forumid]).try(:name),
+          category: category_id_from_imported_category_id(topic[:forumid]),
           raw: post[:raw],
           created_at: Time.at(topic[:dateline].to_i),
           visible: topic[:visible].to_i == 1,