1 files changed, 32 insertions, 0 deletions
diff --git a/lib/bulk_imports/path_normalization.rb b/lib/bulk_imports/path_normalization.rb
new file mode 100644
index 00000000000..dfeef330ff8
--- /dev/null
+++ b/lib/bulk_imports/path_normalization.rb
@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module BulkImports
+  module PathNormalization
+    private
+
+    def normalize_path(path)
+      path = path.parameterize.downcase
+      return path if path =~ Gitlab::Regex.oci_repository_path_regex
+
+      # remove invalid characters from end and start of path
+      delete_invalid_edge_characters(delete_invalid_edge_characters(path))
+      # remove invalid multiplied characters
+      delete_invalid_multiple_characters(path)
+    end
+
+    def delete_invalid_edge_characters(path)
+      path.reverse!
+      path.each_char do |char|
+        break path unless char.match(Gitlab::Regex.oci_repository_path_regex).nil?
+
+        path.delete_prefix!(char)
+      end
+    end
+
+    def delete_invalid_multiple_characters(path)
+      path.gsub!('-_', '-') if path.include?('-_')
+      path.gsub!('_-', '-') if path.include?('_-')
+      path
+    end
+  end
+end