How Do I process a Large Document where the defining tags are located at the end of the document

I’m in the process of determining how to Handle Large Documents using IS and TN 4.6. I’ve read the Large Document Handling documentation. One thing that is mentioned, is that the tags that are used to recognize the document must be included in the first stream or bytes loaded. The documents that we get all have the identifying tags at the tail end of the documents. How should I proceed? Should I create a service with a nodeiterator that reads in the nodes that I need, create a new bizDoc and resubmit to TN for it to properly recognize the document Type? If I resubmit this new Doc to TN, how will I keep the reference to the original document on the hard drive so that the service that is invoked by the processing rule knows where to get the data? Also I assume I will have to rewrite any existing services to handle the streaming of a large document, correct? I’m new to this so excuse my ignorance.