def build_ferpa(): pipeline = Pipeline() pipeline.add_pipe('name', Pipe(Lookup(read_names()), Drop())) pipeline.add_pipe('zip', Pipe(ZipCodeClassifier, ZipCodeFilter())) pipeline.add_pipe('address', Pipe(AddressClassifier(), AddressFilter())) pipeline.add_pipe('date', Pipe(DateClassifier(), DateFilter())) pipeline.add_pipe('phone_number', Pipe(PhoneNumberClassifier, Drop())) pipeline.add_pipe('email', Pipe(EmailClassifier, Drop())) pipeline.add_pipe('ssn', Pipe(SSNClassifier, Drop())) return pipeline
def build_hipaa(): pipeline = Pipeline() pipeline.add_pipe('name', Pipe(Lookup(read_names()), Drop())) pipeline.add_pipe('zip', Pipe(ZipCodeClassifier, ZipCodeFilter())) pipeline.add_pipe('address', Pipe(AddressClassifier(), AddressFilter())) pipeline.add_pipe('date', Pipe(DateClassifier(), DateFilter())) pipeline.add_pipe('phone_number', Pipe(PhoneNumberClassifier, Drop())) pipeline.add_pipe('email', Pipe(EmailClassifier, Drop())) pipeline.add_pipe('url', Pipe(URLClassifier, Drop())) pipeline.add_pipe('ssn', Pipe(SSNClassifier, Drop())) pipeline.add_pipe('ip_address', Pipe(IPAddressClassifier, Drop())) pipeline.add_pipe('mac_address', Pipe(MACAddressClassifier, Drop())) pipeline.add_pipe('face', Pipe(FaceClassifier(), Drop())) # TODO: This is far too sensitive # pipeline.add_pipe('number', Pipe(NumberClassifier, Drop())) return pipeline