Python获取A股上市公司财报数据

'''
import pandas as pd
import tushare as ts
from time import sleep
'''

下面是参数说明：

ts_code 股票代码

end_date 报告期

report_type 报表类型，1合并报表 2单季合并 3调整单季合并表 4调整合并报表 5调整前合并报表 6母公司报表 7母公司单季报 8 母公司调整单季表 9母公司调整表 10母公司调整前报表 11调整前合并报表 12母公司调整前报表

basic_eps 基本每股收益

diluted_eps 稀释每股收益

total_revenue 营业总收入

sell_exp 销售费用

admin_exp 管理费用

fin_exp 财务费用

total_profit 利润总额

income_tax 所得税费用

n_income 净利润(含少数股东损益)

'''

#这里是获取数据前的定义，方便修改，重复利用

#注册地址

https://tushare.pro/register?reg=229016

token='**这个需要大家去注册一个tushare，免费的，然后从后台获取**'

pro = ts.pro_api(token)

my_code='603858.SH'

fbegdate='20150101'

fenddate='20201231'

lrb_files='ts_code,end_date,report_type,basic_eps,diluted_eps,total_revenue,'

'sell_exp,admin_exp,fin_exp,total_profit,income_tax,n_income'

zcfzb_files='ts_code,end_date,report_type,money_cap,notes_receiv,accounts_receiv,oth_receiv,inventories,total_cur_assets,'

'cip,r_and_d,total_assets,total_cur_liab,total_liab,minority_int'

xjllb_files='ts_code,end_date,report_type,net_profit,finan_exp,c_fr_sale_sg,c_inf_fr_operate_a,c_paid_to_for_empl,'

'st_cash_out_act,n_cashflow_act,n_cashflow_inv_act,n_cash_flows_fnc_act'

cwzb_files='ts_code,end_date,'

'profit_dedt,gross_margin,inv_turn,ar_turn,ca_turn,fa_turn,assets_turn,fcff,'

'ocfps,netprofit_margin,grossprofit_margin,cogs_of_sales,expense_of_sales,roe,'

'roe_yearly,debt_to_assets,fixed_assets,rd_exp'

#下面是利用代码，获取网上数据

while True:

try:

#利润表

lr_df = pro.income(ts_code=my_code, start_date=fbegdate, end_date=fenddate,fields=lrb_files)

sleep(2)

#资产负债表

zcfz_df = pro.balancesheet(ts_code=my_code, start_date=fbegdate, end_date=fenddate,fields=zcfzb_files)

sleep(2)

#现金流量表

xjll_df = pro.cashflow(ts_code=my_code, start_date=fbegdate, end_date=fenddate, fields=xjllb_files)

sleep(2)

#财务主要指标表

cwzb_df = pro.fina_indicator(ts_code=my_code, start_date=fbegdate, end_date=fenddate,fields=cwzb_files)

sleep(2)

break

except Exception as e:

print(e)

continue

# 这里是利用年月日的时间生成，年、季度列；插入指定列,0表示第一列，即生成年度列和季度列

lr_df.insert(1, 'fyear',lr_df['end_date'])

lr_df.insert(2,'fperiod',lr_df['end_date'])

#利用drop_duplicates删除重复行

nlr_df=lr_df.drop_duplicates(subset=['ts_code','end_date','report_type'], keep='first')

nzcfz_df=zcfz_df.drop_duplicates(subset=['ts_code','end_date','report_type'], keep='first')

hz_df1=pd.merge(nlr_df,nzcfz_df,on=['ts_code','end_date'])

hz_df2=pd.merge(hz_df1,nxjll_df,on=['ts_code','end_date'])

hz_df3=pd.merge(hz_df2,ncwzb_df,on=['ts_code','end_date'])

#以下是我用来重命名，改成中文的主要为了方便，所以有点长，其实可以省略的

def renname(df):

i = 0 # 用i的目的是区别，列名重复的问题，确保重命名后不重复，为删除列做好准备

# 获取列名

for column in df.iloc[:,:]:

# 用i的目的是区别，列名重复的问题，确保重命名后不重复，为删除列做好准备

print(column)

i = i + 1

if column == 'ts_code':

newname = '股票代码'

elif column == 'end_date':

newname='时间'

elif column== 'fyear':

newname='年度'

elif column == 'fperiod':

newname='月份'

elif column == 'report_type_x':

newname='利润表类型'

elif column == 'basic_eps':

newname='基本每股收益'

elif column == 'diluted_eps':

newname='稀释每股收益'

elif column == 'total_revenue':

newname='营业总收入'

elif column == 'sell_exp':

newname='销售费用'

elif column == 'admin_exp':

newname='管理费用'

elif column == 'fin_exp':

newname='财务费用'

elif column == 'total_profit':

newname='利润总额'

elif column == 'income_tax':

newname='所得税费用'

elif column == 'n_income':

newname='净利润(含少数股东损益)'

elif column == 'report_type_y':

newname='资产负债表类型'

elif column == 'money_cap':

newname = '货币资金'

elif column == 'notes_receiv':

newname='应收票据'

elif column == 'accounts_receiv':

newname='应收账款'

elif column == 'oth_receiv':

newname='其他应收款'

elif column == 'inventories':

newname='存货'

elif column == 'total_cur_assets':

newname='流动资产合计'

elif column == 'cip':

newname='在建工程'

elif column == 'r_and_d':

newname='研发支出'

elif column == 'total_assets':

newname='资产总计'

elif column == 'total_cur_liab':

newname='流动负债合计'

elif column == 'total_liab':

newname='负债合计'

elif column == 'minority_int':

newname='少数股东权益'

elif column == 'report_type':

newname='现金流量表类型'

elif column == 'net_profit':

newname='净利润'

elif column == 'finan_exp':

newname='(现)财务费用'

elif column == 'c_fr_sale_sg':

newname='销售商品、提供劳务收到的现金'

elif column == 'c_inf_fr_operate_a':

newname='经营活动现金流入小计'

elif column == 'c_paid_to_for_empl':

newname='支付给职工以及为职工支付的现金'

elif column == 'st_cash_out_act':

newname='经营活动现金流出小计'

elif column == 'n_cashflow_act':

newname='经营活动产生的现金流量净额'

elif column == 'n_cashflow_inv_act':

newname='投资活动产生的现金流量净额'

elif column == 'n_cash_flows_fnc_act':

newname='筹资活动产生的现金流量净额'

elif column == 'profit_dedt':

newname='扣除非经常性损益后的净利润'

elif column == 'gross_margin':

newname='毛利'

elif column == 'inv_turn':

newname='存货周转率'

elif column == 'ar_turn':

newname='应收账款周转率'

elif column == 'ca_turn':

newname='流动资产周转率'

elif column == 'fa_turn':

newname='固定资产周转率'

elif column == 'assets_turn':

newname='总资产周转率'

elif column == 'fcff':

newname='企业自由现金流量'

elif column == 'ocfps':

newname='每股经营活动产生的现金流量净额'

elif column == 'netprofit_margin':

newname='销售净利率'

elif column == 'grossprofit_margin':

newname='销售毛利率'

elif column == 'cogs_of_sales':

newname='销售成本率'

elif column == 'expense_of_sales':

newname='销售期间费用率'

elif column == 'roe':

newname='净资产收益率'

elif column == 'roe_yearly':

newname='年化净资产收益率'

elif column == 'debt_to_assets':

newname='资产负债率'

elif column == 'fixed_assets':

newname='固定资产合计'

elif column == 'rd_exp':

newname='研发费用'

else:

newname=column

# df=df.drop([column],axis=1, inplace=True) #根据列名删除该列

df.rename(columns=newname}, inplace=True)

return df

#这里是调用了重命名函数，也可以不用的

renname(hz_df3)

#把结果保存的excel，备用

hz_df3.to_excel('C:/Test/%s_cbhz.xlsx'%my_code)

print('完成')