This paper describes a framework for retrospective document conversion in the library domain. Drawing on the experience and insight gained from the {\sc more} project launched over the present decade by the European Commission, it outlines the requirements for solving the problem of retroconversion of old catalogues in {\sc unimarc} format. Based on {\sc ocr} technique and automatic structure recognition, the system proposes a direct schema for the conversion of references in machine readable records. Furthermore, as the system is meant for a real production chain, the paper describes the industrial constraints and gives a complete benchmark realised on this chain for 11 volumes and 4568 references. Without any manual intervention, the recognition rate of the system is greater than 75\%