I would recommend having a look at XML::Rules. Not only it can match the (assumed) simplicity of XML::Simple thanks to its ability to infer a default set of transformation rules from examples or a DTD, ensuring consistency of the datastructure built even with XMLs containing optionally repeated tags or optional attributes, but it allows you to tweak the datastructure built from your XML to better suit your needs and, should the files get too big, allows you to trim and/or process the datastructure as it's being built.