There's nothing more to do but using FPDI as before. It is possible to read PDF documents with a compressed crossreference as of now.
The FPDI PDF-Parser introduces a falback for reading corrupted documents. If a cross-reference is e.g. not found the object offsets will be extracted by walking through the complete document.
The parser also notice invalid crossreferences that were created by several scanners that start the crossreference table with 1 instead of 0.