1 files changed, 43 insertions, 21 deletions

diff --git a/lib/gitlab/background_migration/populate_untracked_uploads.rb b/lib/gitlab/background_migration/populate_untracked_uploads.rb
index ea04484c6a1..802b661886b 100644
--- a/lib/gitlab/background_migration/populate_untracked_uploads.rb
+++ b/lib/gitlab/background_migration/populate_untracked_uploads.rb
@@ -1,12 +1,18 @@
+# frozen_string_literal: true
+
 module Gitlab
   module BackgroundMigration
-    class PopulateUntrackedUploads
-      class UntrackedFile < ActiveRecord::Base
+    # This class processes a batch of rows in `untracked_files_for_uploads` by
+    # adding each file to the `uploads` table if it does not exist.
+    class PopulateUntrackedUploads # rubocop:disable Metrics/ClassLength
+      # This class is responsible for producing the attributes necessary to
+      # track an uploaded file in the `uploads` table.
+      class UntrackedFile < ActiveRecord::Base # rubocop:disable Metrics/ClassLength, Metrics/LineLength
         self.table_name = 'untracked_files_for_uploads'
 
         # Ends with /:random_hex/:filename
-        FILE_UPLOADER_PATH_PATTERN = %r{/\h+/[^/]+\z}
-        FILE_UPLOADER_CAPTURE_FULL_PATH_PATTERN = %r{\A(.+)#{FILE_UPLOADER_PATH_PATTERN}}
+        FILE_UPLOADER_PATH = %r{/\h+/[^/]+\z}
+        FULL_PATH_CAPTURE = %r{\A(.+)#{FILE_UPLOADER_PATH}}
 
         # These regex patterns are tested against a relative path, relative to
         # the upload directory.
@@ -44,7 +50,7 @@ module Gitlab
             model_type: 'Project'
           },
           {
-            pattern: FILE_UPLOADER_PATH_PATTERN,
+            pattern: FILE_UPLOADER_PATH,
             uploader: 'FileUploader',
             model_type: 'Project'
           }
@@ -63,13 +69,14 @@ module Gitlab
 
         def upload_path
           # UntrackedFile#path is absolute, but Upload#path depends on uploader
-          @upload_path ||= if uploader == 'FileUploader'
-                             # Path relative to project directory in uploads
-                             matchd = path_relative_to_upload_dir.match(FILE_UPLOADER_PATH_PATTERN)
-                             matchd[0].sub(%r{\A/}, '') # remove leading slash
-                           else
-                             path
-                           end
+          @upload_path ||=
+            if uploader == 'FileUploader'
+              # Path relative to project directory in uploads
+              matchd = path_relative_to_upload_dir.match(FILE_UPLOADER_PATH)
+              matchd[0].sub(%r{\A/}, '') # remove leading slash
+            else
+              path
+            end
         end
 
         def uploader
@@ -83,7 +90,8 @@ module Gitlab
         def model_id
           return @model_id if defined?(@model_id)
 
-          matchd = path_relative_to_upload_dir.match(matching_pattern_map[:pattern])
+          pattern = matching_pattern_map[:pattern]
+          matchd = path_relative_to_upload_dir.match(pattern)
 
           # If something is captured (matchd[1] is not nil), it is a model_id
           # Only the FileUploader pattern will not match an ID
@@ -105,14 +113,20 @@ module Gitlab
             path_relative_to_upload_dir.match(path_pattern_map[:pattern])
           end
 
-          raise "Unknown upload path pattern \"#{path}\"" unless @matching_pattern_map
+          unless @matching_pattern_map
+            raise "Unknown upload path pattern \"#{path}\""
+          end
 
           @matching_pattern_map
         end
 
         def file_uploader_model_id
-          matchd = path_relative_to_upload_dir.match(FILE_UPLOADER_CAPTURE_FULL_PATH_PATTERN)
-          raise "Could not capture project full_path from a FileUploader path: \"#{path_relative_to_upload_dir}\"" unless matchd
+          matchd = path_relative_to_upload_dir.match(FULL_PATH_CAPTURE)
+          not_found_msg = <<~MSG
+            Could not capture project full_path from a FileUploader path:
+              "#{path_relative_to_upload_dir}"
+          MSG
+          raise not_found_msg unless matchd
 
           full_path = matchd[1]
           project = Project.find_by_full_path(full_path)
@@ -123,7 +137,8 @@ module Gitlab
 
         # Not including a leading slash
         def path_relative_to_upload_dir
-          base = %r{\A#{Regexp.escape(Gitlab::BackgroundMigration::PrepareUntrackedUploads::RELATIVE_UPLOAD_DIR)}/}
+          upload_dir = Gitlab::BackgroundMigration::PrepareUntrackedUploads::RELATIVE_UPLOAD_DIR # rubocop:disable Metrics/LineLength
+          base = %r{\A#{Regexp.escape(upload_dir)}/}
           @path_relative_to_upload_dir ||= path.sub(base, '')
         end
 
@@ -132,6 +147,7 @@ module Gitlab
         end
       end
 
+      # This class is used to query the `uploads` table.
       class Upload < ActiveRecord::Base
         self.table_name = 'uploads'
       end
@@ -192,8 +208,10 @@ module Gitlab
         end
 
         ids.each do |model_type, model_ids|
-          found_ids = Object.const_get(model_type).where(id: model_ids.uniq).pluck(:id)
-          ids[model_type] = ids[model_type] - found_ids # replace with deleted ids
+          model_class = Object.const_get(model_type)
+          found_ids = model_class.where(id: model_ids.uniq).pluck(:id)
+          deleted_ids = ids[model_type] - found_ids
+          ids[model_type] = deleted_ids
         end
 
         ids
@@ -204,11 +222,15 @@ module Gitlab
           file.to_h.merge(created_at: 'NOW()')
         end
 
-        Gitlab::Database.bulk_insert('uploads', rows, disable_quote: :created_at)
+        Gitlab::Database.bulk_insert('uploads',
+                                     rows,
+                                     disable_quote: :created_at)
       end
 
       def drop_temp_table_if_finished
-        UntrackedFile.connection.drop_table(:untracked_files_for_uploads) if UntrackedFile.all.empty?
+        if UntrackedFile.all.empty?
+          UntrackedFile.connection.drop_table(:untracked_files_for_uploads)
+        end
       end
     end
   end