create_crowdflower_input.py 文件源码

def write_input_spreadsheet(data_units, outfile):
    # Merge all the keys to prepare the CSV headers
    headers = set([k for d in data_units for k in d.keys()])
    # Specific field for test (gold) units
    headers.add('_golden')
    headers = list(headers)
    gold_columns = []
    for header in headers:
        # Add gold answer columns for each chunk
        if re.search('chunk_[0-9]{2}$', header):
            gold_columns.append(header + '_gold')
    headers += gold_columns
    headers.sort()
    logger.debug('CSV headers: %s' % headers)
    writer = DictWriter(outfile, headers)
    writer.writeheader()
    writer.writerows(data_units)
    return 0