srmdumps.py 文件源码

def get_newest(base_url, url_pattern, links):
    '''
    Returns a tuple with the newest url in the `links` list matching the
    pattern `url_pattern` and a datetime object representing the creation
    date of the url.

    The creation date is extracted from the url using datetime.strptime().
    '''
    logger = logging.getLogger('auditor.srmdumps')
    times = []

    pattern_components = url_pattern.split('/')

    date_pattern = '{0}/{1}'.format(base_url, pattern_components[0])
    if len(pattern_components) > 1:
        postfix = '/' + '/'.join(pattern_components[1:])
    else:
        postfix = ''

    for link in links:
        try:
            time = datetime.datetime.strptime(link, date_pattern)
        except ValueError:
            pass
        else:
            times.append((str(link) + postfix, time))

    if not times:
        msg = 'No links found matching the pattern {0} in {1}'.format(date_pattern, links)
        logger.error(msg)
        raise Exception(msg)

    return max(times, key=operator.itemgetter(1))