data-clean/data_clean/utils/case_utils.py

# -*- coding: utf-8 -*-
# @Time : 2023/8/15 9:43
# @Author : XuJiakai
# @File : case_utils
# @Software: PyCharm
import asyncio
import datetime
import json
import re

from data_clean.api.es_api import search
from data_clean.api.hbase_api import get as hbase_get
from data_clean.utils.json_utils import json_path, del_key
from data_clean.utils.party_name_verify_utils import person_name_list_verify

_case_no_year_pattern = re.compile("^[（(](\\d{4}?)[）)].+$")


def case_no_year_datetime(case_no, add_year: int = 0):
    if not case_no:
        return None

    match = _case_no_year_pattern.match(case_no)
    if match:
        res = match.group(1)
        return datetime.datetime(year=int(res) + add_year, month=1, day=1)
    else:
        return None
    pass


async def _general_query_case_party(case_no: str, index: str, doc_type: str = '_doc', case_no_key: str = 'case_no',
                                    hbase_table_name: str = None):
    if not case_no:
        return None
    result = await search(index=index, doc_type=doc_type, body={
        "query": {
            "bool": {
                "must": [
                    {
                        "term": {
                            case_no_key: {
                                "value": case_no
                            }
                        }
                    }, {
                        "terms": {
                            "deleted": [
                                "0",
                                "1"
                            ]
                        }
                    }
                ]
            }
        }
    })
    result = result['hits']
    total = result['total']
    if total != 1:
        return None
        pass

    record = result['hits'][0]
    plaintiff_info = json_path(record, '$._source.plaintiff_info')
    defendant_info = json_path(record, '$._source.defendant_info')
    litigant_info = json_path(record, '$._source.litigant_info')
    if not litigant_info:
        if hbase_table_name:
            rowkey = json_path(record, '$._id')
            re = await hbase_get(hbase_table_name, rowkey)
            if 'LITIGANT_INFO' in re:
                li_str = re['LITIGANT_INFO']
                litigant_info = json.loads(li_str)
            pass
        else:
            litigant_info = []
            if plaintiff_info:
                litigant_info = litigant_info + plaintiff_info
            if defendant_info:
                litigant_info = litigant_info + defendant_info

    names = [i['name'] for i in litigant_info]
    flag, error_name = person_name_list_verify(names)
    if not flag:
        return None
        pass

    if plaintiff_info is None or defendant_info is None or litigant_info is None or len(plaintiff_info) == 0 or len(
            defendant_info) == 0 or len(litigant_info) == 0:
        return None

    for i in plaintiff_info:
        del_key(i, 'lawyer_info')
        del_key(i, 'judge_tendency')
    for i in defendant_info:
        del_key(i, 'lawyer_info')
        del_key(i, 'judge_tendency')
    for i in litigant_info:
        del_key(i, 'lawyer_info')
        del_key(i, 'judge_tendency')

    return {
        "plaintiff_info": plaintiff_info,
        "defendant_info": defendant_info,
        "litigant_info": litigant_info,
        "plaintiff": ','.join([i['name'] for i in plaintiff_info]),
        "defendant": ','.join([i['name'] for i in defendant_info]),
        "litigant": ','.join([i['name'] for i in litigant_info]),
    }
    pass


async def get_open_court_case_party(case_no: str):
    return await _general_query_case_party(case_no, "winhc_index_rt_company_court_open_announcement",
                                           hbase_table_name="NG_RT_COMPANY_COURT_OPEN_ANNOUNCEMENT")
    pass


async def get_register_case_party(case_no: str):
    return await _general_query_case_party(case_no, "winhc_index_rt_company_court_register")
    pass


async def get_judgement_document_party(case_no: str):
    return await _general_query_case_party(case_no, "old.wenshu_detail2", doc_type='wenshu_detail_type',
                                           case_no_key='case_no.keyword')
    pass


async def get_case_party(case_no: str, source: str = None):
    if not case_no:
        return None

    if source == 'case_register':
        result = await get_open_court_case_party(case_no)
        if result:
            return result
        result = await get_judgement_document_party(case_no)
        if result:
            return result
    elif source == 'open_court':
        result = await get_register_case_party(case_no)
        if result:
            return result
        result = await get_judgement_document_party(case_no)
        if result:
            return result
    elif source == 'judgement_doc':
        result = await get_register_case_party(case_no)
        if result:
            return result
        result = await get_open_court_case_party(case_no)
        if result:
            return result
    else:
        result = await get_register_case_party(case_no)
        if result:
            return result
        result = await get_open_court_case_party(case_no)
        if result:
            return result
        result = await get_judgement_document_party(case_no)
        if result:
            return result

    return None
    pass


async def test():
    from base_utils import to_string
    # print(await get_case_party("（2021）粤05民初2014号"))
    # print(await get_case_party("(2020)冀0181民初1940号"))
    # print(await get_case_party("(2020)冀0181民初2477号"))
    # print(to_string(await get_case_party("(2020)冀09民终5209号"), is_format=False))
    print(to_string(await get_case_party("（2020）苏0481民初1925号", source='open_court')))
    # print(await get_register_case_party("（2017）津0119民初2784号"))
    # print(await get_open_court_case_party("（2017）津0119民初2784号"))
    # print(await get_judgement_document_party("（2017）津0119民初2784号"))
    pass


if __name__ == '__main__':
    asyncio.run(test())
    pass