comparison.py 5.15 KB
import re
from datetime import datetime
from .rmb_lower import rmb_handler


class Comparison:

    def __init__(self):
        self.CSIBM = 'CSIBM'
        self.CSSME = 'CSSME'
        self.CSOTH = 'CSOTH'

        self.TYPE_MAPPING = (
            (r'个体工商户', self.CSIBM),
            (r'有限责任公司', self.CSSME),
            (r'个人独资企业', self.CSSME),
            (r'有限合伙企业', self.CSSME),
            (r'股份合作制', self.CSSME),
        )

        self.RESULT_Y = 'Y'
        self.RESULT_N = 'N'
        self.RESULT_NA = 'NA'

        self.TRANS_MAP = {
            ' ': '',
            '·': '',
        }
        self.TRANS = str.maketrans(self.TRANS_MAP)
        self.re_obj = r'[(\(].*?[\))]'

    def build_res(self, result):
        if result:
            return self.RESULT_Y
        else:
            return self.RESULT_N

    def common_compare(self, input_str, ocr_str, idx, **kwargs):
        if not isinstance(ocr_str, str) or not isinstance(input_str, str):
            return self.RESULT_NA, ocr_str
        if ocr_str == '' or ocr_str.strip() == '':
            return self.RESULT_NA, None
        return self.build_res(input_str == ocr_str), ocr_str

    def company_compare(self, input_str, ocr_str, idx, **kwargs):
        if not isinstance(ocr_str, str) or not isinstance(input_str, str):
            return self.RESULT_NA, ocr_str
        if ocr_str == '' or ocr_str.strip() == '':
            return self.RESULT_NA, None
        input_tmp = re.sub(self.re_obj, '', input_str).strip()
        ocr_tmp = re.sub(self.re_obj, '', ocr_str).strip()
        return self.build_res(input_tmp == ocr_tmp), ocr_str

    def name_compare(self, input_str, ocr_str, idx, **kwargs):
        if not isinstance(ocr_str, str) or not isinstance(input_str, str):
            return self.RESULT_NA, ocr_str
        if ocr_str == '' or ocr_str.strip() == '':
            return self.RESULT_NA, None
        if kwargs.get('is_passport'):
            input_tmp = input_str.upper().replace(' ', '')
            ocr_tmp = ocr_str.upper().replace(' ', '')
            if input_tmp.find(ocr_tmp) == -1:
                return self.RESULT_N, ocr_str
            else:
                return self.RESULT_Y, ocr_str
        else:
            # if re.search(r'[a-zA-Z]]', input_str):
            #     return self.RESULT_NA, ocr_str
            input_s = input_str.translate(self.TRANS)
            ocr_s = ocr_str.translate(self.TRANS)
            return self.build_res(input_s == ocr_s), ocr_str

    def date_compare(self, input_str, ocr_str, idx, **kwargs):
        if not isinstance(ocr_str, str) or not isinstance(input_str, str):
            return self.RESULT_NA, ocr_str
        if ocr_str == '' or ocr_str.strip() == '':
            return self.RESULT_NA, None
        if kwargs.get('long', False):
            if '长期' in ocr_str or '永久' in ocr_str:
                if input_str == '2099-12-31' or input_str == '2099-01-01':
                    return self.RESULT_Y, '2099-12-31'
                else:
                    return self.RESULT_N, '2099-12-31'
        if kwargs.get('ocr_split', False):
            if '至' in ocr_str:
                ocr_str = ocr_str.split('至')[-1]
            elif '-' in ocr_str:
                ocr_str = ocr_str.split('-')[-1]
        if kwargs.get('ocr_replace', False):
            ocr_str = ocr_str.replace('年', '-').replace('月', '-').replace('日', '')
        if kwargs.get('input_replace') is not None:
            input_str = input_str.replace('-', kwargs.get('input_replace'))
            try:
                ocr_output = datetime.strptime(ocr_str, '%Y{0}%m{0}%d'.format(
                    kwargs.get('input_replace'))).strftime('%Y-%m-%d')
            except Exception as e:
                ocr_output = None
        else:
            try:
                ocr_output = datetime.strptime(ocr_str, '%Y-%m-%d').strftime('%Y-%m-%d')
            except Exception as e:
                ocr_output = None
        return self.build_res(input_str == ocr_str), ocr_output

    def rmb_compare(self, input_str, ocr_str, idx, **kwargs):
        if not isinstance(ocr_str, str) or not isinstance(input_str, str):
            return self.RESULT_NA, None
        if ocr_str == '' or ocr_str.strip() == '':
            return self.RESULT_NA, None
        try:
            ocr_lower = rmb_handler.to_rmb_lower()
            res = self.build_res(float(input_str) == ocr_lower)
        except Exception as e:
            return self.RESULT_N, None
        else:
            if res == self.RESULT_Y:
                return res, input_str
            else:
                return res, ocr_lower

    def type_compare(self, input_str, ocr_str, idx, **kwargs):
        if not isinstance(ocr_str, str) or not isinstance(input_str, str):
            return self.RESULT_NA, ocr_str
        if ocr_str == '' or ocr_str.strip() == '':
            return self.RESULT_NA, None
        for map_tuple in self.TYPE_MAPPING:
            if re.search(map_tuple[0], ocr_str) is not None:
                compare_str = map_tuple[1]
                break
        else:
            compare_str = self.CSOTH

        return self.build_res(input_str == compare_str), compare_str


cp = Comparison()