fix_dates_dbpedia_person_dataset.py

import re
import datetime
import codecs

c = re.compile('\d{1,4}-\d{1,2}-\d{1,2}')

with codecs.open('persondata_wkd_uris_pt.ttl', 'r', 'utf-8') as f_in, codecs.open('persondata_wkd_uris_pt.ttl.out', 'w', 'utf-8') as f_out:

    for i, line in enumerate(f_in):

        m = c.search(line)

        if m:

            str_date = m.group()
            try:
                d = datetime.datetime.strptime(str_date, '%Y-%m-%d')
                str_date_out = d.strftime('%Y-%m-%d')
                line = line.replace(str_date, str_date_out)
                f_out.write(line)                
            except:
                print("[{i}] : {date}".format(i=i, date=str_date))
        else:
            f_out.write(line)