competitor_product_analysis/handle/search_winhc_latest_date.py

# -*- coding: utf-8 -*-
# @Time : 2022/12/1 9:13
# @Author : XuJiakai
# @File : search_winhc_latest_date
# @Software: PyCharm
from sdk import get_es_sdk
from utils import map_2_json_str
from utils.dim_name_mapping import get_latest_date_map
from sdk.WinhcAllClient import get_all_client
from utils.base_utils import tuple_max
from utils.datetime_utils import datetime_format_transform
import re
from log import get_log

log = get_log("winhc_latest_date")
all_client = get_all_client()
hbase_client = all_client.get_hbase_client()

es_sdk = get_es_sdk("new")
old_es_sdk = get_es_sdk("old")

old_es_index = ['wenshu_detail2']
date_part = re.compile('\\d{4}年\\d{2}月\\d{2}日')


def get_latest_date(index: str, company_id_f: str, company_id: str, latest_date_f: str):
    dsl = {
        "size": 1,
        "_source": [latest_date_f],
        "query": {
            "bool": {
                "must": [
                    {
                        "term": {
                            company_id_f: {
                                "value": company_id
                            }
                        }
                    },
                    {
                        "term": {
                            "deleted": {
                                "value": 0
                            }
                        }
                    }
                ]
            }
        }
        , "sort": [
            {
                latest_date_f: {
                    "order": "desc"
                }
            }
        ]
    }
    if index in old_es_index:
        res = old_es_sdk.query(index=index, doc_type=None, dsl=dsl)
        pass
    else:
        res = es_sdk.query(index=index, doc_type='_doc', dsl=dsl)

    if len(res) == 0 or latest_date_f not in res[0]:
        return None

    latest_date = res[0][latest_date_f]

    # if date_part.match(latest_date)
    if date_part.match(latest_date):
        latest_date = datetime_format_transform(latest_date, '%Y年%m月%d日', "%Y-%m-%d %H:%M:%S")
        pass

    return latest_date
    pass


latest_date_map = get_latest_date_map()


def search_latest_date(company_id: str):
    result_data = {}
    for i in latest_date_map:
        str = latest_date_map[i]
        max_date = None
        for j in str.split(','):
            tmp_str = j.split(':')
            index = tmp_str[0]

            company_id_f = tmp_str[1]
            latest_date_f = tmp_str[2]
            tmp_date = get_latest_date(index=index, company_id_f=company_id_f, company_id=company_id,
                                       latest_date_f=latest_date_f)
            max_date = tuple_max(max_date, tmp_date)
            pass

        result_data[i] = max_date
        pass

    tmp_res = hbase_client.get_record('ng_rt_company', company_id)

    if tmp_res is not None and 'APPROVED_TIME' in tmp_res:
        result_data['基本信息'] = tmp_res['APPROVED_TIME']
        log.info('fetch hbase data: {}'.format(tmp_res))
    else:
        result_data['基本信息'] = None
    return result_data
    pass


if __name__ == '__main__':
    # d = get_latest_date(index='winhc_index_rt_company_punishment_info', company_id_f='company_id',
    #                     company_id='059f83641cc4df8b9577cb1e2d89939e', latest_date_f='decision_date')
    # print(d)

    d = search_latest_date(company_id='6e13b126ee0c5fcd8fe454693ab4bbda')
    print(map_2_json_str(d))


    pass