gistfile1.py

from disco.core import Job
from disco.worker.classic.func import chain_reader

import datetime

class Download(object):
    def __init__(self, id, doi, date, ip):
        self.id = id
        self.doi = doi
        self.date = datetime.date(int(date[0:4]), int(date[5:7]), int(date[8:10]))
        self.ip = ip

class ParseDownloads(Job):
    map_reader = staticmethod(chain_reader)

    @staticmethod
    def map(line, params):
        [id, doi, date, ip] = line.split('\t')
        yield id, Download(id, doi, date, ip)