import re
from datetime import datetime

log_path = '/convert.out'


def process(_str=''):

    if not _str:
        with open(log_path, 'r') as f:
            line_list = f.readlines()
    else:
        line_list = _str.split('\n')

    md5_list = []
    md5_finished_list = []
    md5_larger_time_list = []
    for line in line_list:
        line_split = line.split(' ')
        if len(line_split) <= 8:
            continue
        _date = line_split[0]
        _time = line_split[1].split(',')[0]

        md5 = None
        for col in line_split:
            if len(col) == len('066ad08d38dd9fa8c37d03f7c67359c9') and re.search('^[0-9a-z]+$', col):
                md5 = col
                break
        if md5 is None:
            continue

        md5_list.append(md5)
        if 'is_success' in line:
            md5_finished_list.append(md5)

        try:
            _time = datetime.strptime(_date + ' ' + _time, "%Y-%m-%d %H:%M:%S")
            search_time = '2024-05-16 13:38:08'
            search_time = datetime.strptime(search_time, "%Y-%m-%d %H:%M:%S")
        except:
            continue
        if _time > search_time:
            md5_larger_time_list.append(md5)

    md5_list = list(set(md5_list))
    md5_not_list = list(set(md5_finished_list + md5_larger_time_list))
    for md5 in md5_list:
        if md5 not in md5_not_list:
            print('md5', md5)


if __name__ == '__main__':
    _str = '''
2024-05-16 17:08:25,765 - get_table - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - yolo detect cost: 0.07653164863586426
2024-05-16 17:08:25,765 - get_table - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - detect not b_table_list
2024-05-16 17:08:25,765 - image_process - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - botr process cost: 0.07787585258483887
2024-05-16 17:08:25,834 - from_atc_interface - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - into from_atc_interface
2024-05-16 17:08:25,834 - interface_pool_gunicorn - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - atc
2024-05-16 17:08:25,834 - interface_pool_gunicorn - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - http://192.168.0.115:18061
2024-05-16 17:08:25,919 - from_atc_interface - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - get interface return
2024-05-16 17:08:25,919 - from_atc_interface - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - from_atc_interface cost time 0.0849916934967041
2024-05-16 17:08:25,919 - cut_str - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - into cut_str
2024-05-16 17:08:25,920 - cut_str - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - into cut_str
2024-05-16 17:08:25,920 - _convert - INFO - 42aee7658c696ab9d44f49d632aa9239 - None - md5: 42aee7658c696ab9d44f49d632aa9239 finished result: ['光大水务（莒县）有限公司（城北厂）双 1034 is_success: 1 pdf 其他 0.8948788642883301
2024-05-16 12:38:41,339 - from_atc_interface - INFO - 65ba62da6b5f4f998133a5203f8f9e1f - None - from_atc_interface cost time 0.0882723331451416
2024-05-16 12:38:41,339 - cut_str - INFO - 65ba62da6b5f4f998133a5203f8f9e1f - None - into cut_str
2024-05-16 12:38:41,339 - cut_str - INFO - 65ba62da6b5f4f998133a5203f8f9e1f - None - into cut_str
98%2024-05-16 12:38:41,340 - _convert - INFO - 65ba62da6b5f4f998133a5203f8f9e1f - None - md5: 65ba62da6b5f4f998133a5203f8f9e1f finished result: ['合同编号：11N2020160092 2710 is_success: 1 pdf 其他 1.9107580184936523
2024-05-16 12:38:41,340 - _convert - INFO - 65ba62da6b5f4f998133a5203f8f9e1f - None - finally
'''
    process()