Processes for fingerprinting a document file and for preventing information leakage are disclosed. Computer apparatus for implementing said processes are also disclosed. For fingerprinting a document, the document is provided and may be normalized. A sequence of hash values are generated for the document. A window size is adaptively determined depending upon the document. Fingerprints for the document are selected from amongst the hash values using the adaptively-sized window. The fingerprints for the document are added to a fingerprint set for content being protected by the information leakage prevention system. For information leakage prevention, suspect documents are processed at the deployment points by extracting fingerprints from the suspect documents and matching the extracted fingerprints against the fingerprints in the fingerprint set. Different fingerprint extraction methods are used at the server and the deployment points. Other embodiments, aspects and features are also disclosed.