I had a similiar problem, and since efficiency was not a concern (and the file was not modified too often), I ended up using OpenOfice Calc to convert the Excel file to HTML, and using HTML::TableExtract. The program started with something like

    ...
    system ('C:\Program Files (x86)\OpenOffice 4\program\scalc.exe', 
        '-headless', 
        'macro:///Standard.Module1.ConvertToHtml("' . $xlsx . '","' . 
+$file->canonpath . '")') if ! $file->exists;
    ...
[download]

Meaning of $xlsx and $file should be obvious, and I had a short macro that would do the conversion (open file and save as html).