Newer
Older
Bartlomiej Koptyra
committed
"""Implementation of nlp_worker."""
import logging
import nlp_ws
from src.anonymizer import Anonymizer
from src.ccl_handler import Ccl_handler
_log = logging.getLogger(__name__)
class Worker(nlp_ws.NLPWorker):
"""Implements nlp_worker for anonymizer service."""
def process(self, input_file, task_options, output_file):
"""Anonymizes input text.
It is assumed input_file is encoded in UTF-8.
Options:
method - 'delete'/'tag'/'pseudo' - 'delete' deletes selected tokens,
'tag' replaces selected tokens with arbitrary tags, 'pseudo'
replaces selected tokens with a random token that
"""
anon = Anonymizer(task_options)
ccl_handler = Ccl_handler(input_file)
ccl_handler.process(output_file, anon.unmarshallers)