python-base--/tmp5.py

def exec(obj):
    from sqlalchemy import create_engine
    import pandas as pd
    from urllib.parse import quote_plus as urlquote
    import jieba
    from gensim import corpora, models, similarities

    def cReate_dAta_cOnn():
        engine = create_engine('mysql+pymysql://aimp_user:'+urlquote('vjeygLP76n7%UPx@')+'@rm-qsls3302.mysql.rds.aliyuncs.com:3302/bi_application')
        return engine

    def rEad_aNd_dAtaframe(sql_query):
        engine = cReate_dAta_cOnn()
        df = pd.read_sql(sql_query,engine)
        engine.dispose()
        return df

    def flag(x):
        if (x['sensoir'] == x['person']) and len(x['sensoir'])==3:
            return 1
        elif (x['sensoir'] == x['person']) and len(x['sensoir'])==2:
            return 2
        else:
            return 3

    def flag1(x):
        if x['f1'] == 1:
            return 1
        else:
            return 0

    def flag2(x):
        if (x['f1'] == 2) and x['sims']>0.8:
            return 1
        else:
            return 0

    def flag3(x):
        if x['sims']>0.95:
            return 1
        else:
            return 0

    def cal_similar(doc_goal,ssim):
        doc = rEad_aNd_dAtaframe('''select distinct credit_no,econ_reg_address from ext_anti_fraud_address ''')
        doc_list = [jieba.lcut(w) for w in doc['econ_reg_address']]
        target = [word for word in jieba.cut(doc_goal)]
        dictionary = corpora.Dictionary(doc_list)
        corpus = [dictionary.doc2bow(doc) for doc in doc_list]
        doc_goal_vec = dictionary.doc2bow(target)
        tfidf = models.TfidfModel(corpus)
        index = similarities.SparseMatrixSimilarity(tfidf[corpus], num_features = len(dictionary.keys()))
        sims = index[tfidf[doc_goal_vec]]
        similary = pd.DataFrame({"risk_address": list(doc['econ_reg_address']), "sims": list(sims)})
        similary["申请企业注册地址"] = doc_goal
        similary_data = similary[["申请企业注册地址", "risk_address", "sims"]].drop_duplicates()
        similary_data= similary_data[similary_data["sims"]>=ssim]
        return similary_data

    lcity = list(obj['city'])
    if len(lcity)>1 and len(lcity[0])>1:
        lcity = list(obj['city'])[0]
    elif len(lcity)==1 and len(lcity[0])>1:
        lcity = obj['city']
    else：
    obj['flag1'] = 0
    obj['flag2'] = 0
    obj['flag3'] = 0
    return obj
credit = []
address = []
sensoir = list(obj['oper_names'])
if len(sensoir)<1:
    obj['flag1'] = 0
    obj['flag2'] = 0
    credit.append(obj['credit_code'])
    address.append(obj['address_'])
    df_app = pd.DataFrame({"credit": credit,"address": address})
    df_add = rEad_aNd_dAtaframe("select distinct credit_no,econ_reg_address from ext_anti_fraud_address where city = {}".format(lcity))
    similary_data=cal_similar(df_app['address'].max(),0.95)
    if similary_data.shape[0]>0:
        obj['flag3'] = 1
    else:
        obj['flag3'] = 0
    obj['similary_data']=similary_data.to_json(orient='records')
    return obj
else:
    for i in sensoir:
        credit.append(obj['credit_code'])
        address.append(obj['address_'])
    df_app = pd.DataFrame({"credit": credit, "sensoir": sensoir,"address": address})
    df_add = rEad_aNd_dAtaframe("select distinct credit_no,econ_reg_address from ext_anti_fraud_address where city = {}".format(lcity))
    lcredit =str(list(df_add['credit_no'])).replace('[','').replace(']','')
    df_per = rEad_aNd_dAtaframe("select distinct credit_no,person from ext_anti_fraud_senior_person where credit_no in ({})".format(lcredit))
    df_dec = pd.merge(df_app,df_per,left_on = 'sensoir',right_on = 'person',how = 'inner')
    df_f = pd.merge(df_dec,df_add,on = 'credit_no',how = 'left')
    if df_f.shape[0]<1:
        obj['flag1'] = 0
        obj['flag2'] = 0
        obj['flag3'] = 0
        obj['df_f']=df_f.to_json(orient='records')
        return obj
    else:
        df_f['f1']=df_f.apply(flag,axis=1)
        similary_data=cal_similar(df_f['address'].max(),0)
        df = pd.merge(df_f,similary_data,left_on='econ_reg_address',right_on = 'risk_address',how = 'left')
        df['flag1']=df.apply(flag1,axis=1)
        df['flag2']=df.apply(flag2,axis=1)
        df['flag3']=df.apply(flag3,axis=1)#保存
        obj['df']=df.to_json(orient='records')
        obj['flag1']=df['flag1'].max()
        obj['flag2']=df['flag2'].max()
        obj['flag3']=df['flag3'].max()
        return obj