simonandClaude Opus 4.7
271a9343a5
7 Sprints 全部完成: Sprint 0: 基础设施 (DataManager + MariaDB) Sprint 1: 因子引擎 (34因子/12分类) Sprint 2: VectorBT 回测 (5策略+截面) Sprint 3: Optuna 优化 (+Walk-Forward) Sprint 4: ML 模型 (LightGBM+CatBoost) Sprint 5: Qwen 情绪因子 (三源新闻+日期对齐) Sprint 6: Agent 系统 (4Agent+日报.md/.html) 生产加固 (15项): Tushare双源fallback, SSH自动恢复, pool_pre_ping, save_daily先删后插, load_dotenv绝对路径, 日报5d/20d修复, RiskAgent改上证指数, 昨日对比+数据截止, mac_report utf8mb4, CLAUDE-*.md 9条已知Bug, demo全参数化, djapi数据源归一化, indexDatas API修正 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
394 lines
17 KiB
Python
394 lines
17 KiB
Python
import pandas as pd
|
|
# 将项目根目录添加到 sys.path
|
|
#project_root = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
|
|
#sys.path.append(project_root)
|
|
|
|
# 导入当站目录的config文件
|
|
try:
|
|
# 尝试相对导入（作为包的一部分）
|
|
from .config import TS_TOKEN, START_DATE, END_DATE, PRECISION_CONFIG
|
|
from .stock_utils import *
|
|
except (ImportError, SystemError):
|
|
# 失败则使用绝对导入（直接运行脚本）
|
|
from config import TS_TOKEN, START_DATE, END_DATE, PRECISION_CONFIG
|
|
from stock_utils import *
|
|
|
|
from .data_source import get_tushare_pro
|
|
pro = get_tushare_pro()
|
|
def getStockEp(TS_CODE,start_date=START_DATE,end_date=END_DATE):
|
|
"""
|
|
从tushare接口获取单只股票的财务数据，计算并填充每日的每股收益(EP)指标。
|
|
Parameters:
|
|
TS_CODE (str): 股票代码,格式为 '股票代码.SZ' 或 '股票代码.SH',例如 '000001.SZ'
|
|
START_DATE (str): 开始日期,格式为 'YYYYMMDD'
|
|
END_DATE (str): 结束日期,格式为 'YYYYMMDD'
|
|
|
|
Returns:
|
|
pd.DataFrame: 包含以下字段的DataFrame:
|
|
ts_code: 股票代码
|
|
ann_date: 财报公告日期
|
|
end_date: 财报结束日期/填充日期
|
|
basic_eps: 基本每股收益
|
|
diluted_eps: 稀释每股收益
|
|
trade_date: 实际交易日(来自日线数据)
|
|
close: 收盘价
|
|
basic_ep: 基本EP值(basic_eps/close)
|
|
diluted_ep: 稀释EP值(diluted_eps/close)
|
|
|
|
Raises:
|
|
Exception: 如果从Tushare接口获取数据时发生错误。
|
|
"""
|
|
|
|
'''
|
|
1. 调用get_trading_dates函数，填充完整df_income 变量内部end_date，规则为：以end_date为限，往前天从到上一个财报日期为止(不包含上一个财报日期)，期间填充basic_eps,diluted_eps值为报告日的相同值
|
|
2. 新增一个basic_ep列，数据为:basic_eps/当天交易日期的股价，当天交易日期的股价来自df_daily的close列
|
|
3. 新增一个diluted_ep列，数据为:diluted_eps/当天交易日期的股价，当天交易日期的股价来自df_daily的close列
|
|
'''
|
|
try:
|
|
# 获取股票财务数据
|
|
df_income = pro.income(ts_code=TS_CODE, start_date=start_date, end_date=end_date,
|
|
fields='ts_code,ann_date,end_date,basic_eps,diluted_eps')
|
|
|
|
# 填充df_income的end_date区间数据
|
|
if not df_income.empty:
|
|
# 按end_date分组并排序财报数据
|
|
df_income = df_income.sort_values('end_date')
|
|
grouped = df_income.groupby('end_date')
|
|
# 存储填充后的数据
|
|
filled_data = []
|
|
|
|
# 遍历每个财报期间
|
|
for end_date, group in grouped:
|
|
# 获取该日期到上一个财报日的所有交易日期
|
|
trading_dates = get_trading_dates(end_date)
|
|
# 填充数据
|
|
for date in trading_dates:
|
|
# 将日期对象转换为字符串格式（YYYYMMDD）
|
|
# 如果date是日期对象（有strftime方法），则调用strftime格式化
|
|
# 否则直接使用原值（假设已经是字符串格式）
|
|
date_str = date.strftime('%Y%m%d') if hasattr(date, 'strftime') else date
|
|
filled_row = {
|
|
'ts_code': TS_CODE,
|
|
'ann_date': group['ann_date'].iloc[0],
|
|
'end_date': date_str,
|
|
'basic_eps': group['basic_eps'].iloc[0],
|
|
'diluted_eps': group['diluted_eps'].iloc[0]
|
|
}
|
|
filled_data.append(filled_row)
|
|
|
|
# 合并填充后的数据
|
|
df_income = pd.DataFrame(filled_data)
|
|
# 获取股票日线数据以获取股价
|
|
min_end_date = df_income['end_date'].min()
|
|
df_daily = pro.daily(ts_code=TS_CODE, start_date=min_end_date, end_date=END_DATE)
|
|
|
|
# 2. 计算basic_ep和diluted_ep
|
|
if not df_income.empty and not df_daily.empty:
|
|
# 修改原因：原代码使用left_on='end_date'和right_on='trade_date'导致匹配失败
|
|
# 解决方案：将df_daily的trade_date转换为字符串格式再合并
|
|
df_daily['trade_date_str'] = df_daily['trade_date'].astype(str)
|
|
df_income = pd.merge(
|
|
df_income,
|
|
df_daily[['trade_date_str', 'close']],
|
|
left_on='end_date',
|
|
right_on='trade_date_str',
|
|
how='left'
|
|
)
|
|
# 将合并后的trade_date_str重命名为trade_date
|
|
df_income.rename(columns={'trade_date_str': 'trade_date'}, inplace=True)
|
|
|
|
# 计算basic_ep和diluted_ep
|
|
df_income['basic_ep'] = (df_income['basic_eps'] / df_income['close']).round(PRECISION_CONFIG)
|
|
df_income['diluted_ep'] = (df_income['diluted_eps'] / df_income['close']).round(PRECISION_CONFIG)
|
|
#print("df_income:",df_income)
|
|
# 按日期截断
|
|
#df_income = df_income[(df_income['trade_date'] >= START_DATE) & (df_income['trade_date'] <= END_DATE)]
|
|
|
|
# 数据修正
|
|
cols = ['basic_eps', 'diluted_eps', 'close', 'basic_ep', 'diluted_ep']
|
|
df_income = dataCorrect(df_income, cols).sort_values('trade_date')
|
|
|
|
return df_income
|
|
|
|
except Exception as e:
|
|
# 捕获并处理异常
|
|
print(f"错误: 获取股票 {TS_CODE} 的EP数据时发生错误: {e}")
|
|
return pd.DataFrame()
|
|
|
|
def getStockEp_ttm(TS_CODE,start_date=START_DATE,end_date=END_DATE):
|
|
"""
|
|
获取股票的TTM(最近12个月)每股收益与股价比率(EP)数据
|
|
|
|
Parameters:
|
|
TS_CODE (str): 股票代码,格式为 '股票代码.SZ' 或 '股票代码.SH',例如 '000001.SZ'
|
|
START_DATE (str): 开始日期,格式为 'YYYYMMDD'
|
|
END_DATE (str): 结束日期,格式为 'YYYYMMDD'
|
|
|
|
Returns:
|
|
pd.DataFrame: 包含以下字段的DataFrame:
|
|
ts_code: 股票代码
|
|
trade_date: 交易日
|
|
close: 收盘价
|
|
basic_ep_ttm: 基本EP_TTM值(basic_eps_q_ttm/close)
|
|
diluted_ep_ttm: 稀释EP_TTM值(diluted_eps_q_ttm/close)
|
|
basic_eps_q_ttm: 基本每股收益TTM值
|
|
diluted_eps_q_ttm: 稀释每股收益TTM值
|
|
|
|
Raises:
|
|
Exception: 如果从Tushare接口获取数据时发生错误
|
|
"""
|
|
try:
|
|
start_date=date_format_correction(start_date) #confirm date as yyyymmdd
|
|
end_date=date_format_correction(end_date) #confirm date as yyyymmdd
|
|
df_eps = get_quarterly_eps(tscodeCheck(TS_CODE), start_date, end_date)
|
|
df_eps_ttm = calculate_ttm_eps(df_eps)
|
|
df_eps_ttm=fill_trading_dates_with_eps(df_eps_ttm,start_date=start_date,end_date=end_date)
|
|
df_daily = pro.daily(ts_code=TS_CODE, start_date=start_date, end_date=end_date)
|
|
df_eps_ttm = pd.merge(
|
|
df_eps_ttm,
|
|
df_daily[['trade_date', 'close']],
|
|
on='trade_date',
|
|
how='left'
|
|
)
|
|
df_eps_ttm['basic_ep_ttm'] = (100*df_eps_ttm['basic_eps_q_ttm'] / df_eps_ttm['close']).round(PRECISION_CONFIG)
|
|
df_eps_ttm['diluted_ep_ttm'] = (100*df_eps_ttm['diluted_eps_q_ttm'] / df_eps_ttm['close']).round(PRECISION_CONFIG)
|
|
# 比较最大日期并补充数据, 解决当ep不存在，close值也不会显示的问题
|
|
if not df_eps_ttm.empty and not df_daily.empty:
|
|
max_eps_date = df_eps_ttm['trade_date'].max()
|
|
max_daily_date = df_daily['trade_date'].max()
|
|
print(f"max_eps_date:{max_eps_date}")
|
|
print(f"max_daily_date:{max_daily_date}")
|
|
if max_eps_date != max_daily_date:
|
|
# 获取需要补充的日期范围
|
|
mask = (df_daily['trade_date'] > max_eps_date) & (df_daily['trade_date'] <= max_daily_date)
|
|
additional_data = df_daily.loc[mask, ['ts_code', 'trade_date', 'close']].copy()
|
|
|
|
# 补充空列
|
|
for col in df_eps_ttm.columns:
|
|
if col not in ['ts_code', 'trade_date', 'close']:
|
|
additional_data[col] = ''
|
|
print(additional_data)
|
|
# 合并数据
|
|
df_eps_ttm = pd.concat([df_eps_ttm, additional_data], ignore_index=True)
|
|
|
|
cols=["close","basic_ep_ttm","diluted_ep_ttm","basic_eps_q_ttm","diluted_eps_q_ttm"]
|
|
df_eps_ttm = dataCorrect(df_eps_ttm,cols)
|
|
return df_eps_ttm[['ts_code', 'trade_date', 'close', 'basic_ep_ttm', 'diluted_ep_ttm', 'basic_eps_q_ttm', 'diluted_eps_q_ttm']]
|
|
except Exception as e:
|
|
# 捕获并处理异常
|
|
print(f"错误: 获取股票 {TS_CODE} 的EP_TTM数据时发生错误: {e}")
|
|
return pd.DataFrame()
|
|
|
|
def get_quarterly_eps(TS_CODE, start_date=START_DATE, end_date=END_DATE):
|
|
"""
|
|
获取单季度每股收益数据
|
|
Parameters:
|
|
TS_CODE (str): 股票代码
|
|
START_DATE (str): 开始日期(YYYYMMDD)
|
|
END_DATE (str): 结束日期(YYYYMMDD)
|
|
|
|
Returns:
|
|
pd.DataFrame: 包含以下字段的DataFrame:
|
|
ts_code: 股票代码
|
|
ann_date: 财报公告日期
|
|
end_date: 财报结束日期(YYYYMMDD格式)
|
|
basic_eps: 累计基本每股收益
|
|
diluted_eps: 累计稀释每股收益
|
|
basic_eps_q: 单季度基本每股收益
|
|
diluted_eps_q: 单季度稀释每股收益
|
|
"""
|
|
|
|
try:
|
|
# 扩展日期范围往前三个季度
|
|
extended_start = (pd.to_datetime(start_date) - pd.DateOffset(months=12)).strftime('%Y%m%d')
|
|
|
|
# 获取原始财务数据
|
|
df = pro.income(ts_code=TS_CODE, start_date=extended_start, end_date=end_date,
|
|
fields='ts_code,ann_date,end_date,basic_eps,diluted_eps')
|
|
|
|
if df.empty:
|
|
return pd.DataFrame()
|
|
|
|
# 按财报日期排序并去重（解决重复数据问题）
|
|
df = df.sort_values('end_date').drop_duplicates(subset=['end_date'], keep='last') # 修改原因：确保每个end_date只保留最新数据
|
|
# 计算单季度数据
|
|
df['basic_eps_q'] = df['basic_eps']
|
|
df['diluted_eps_q'] = df['diluted_eps']
|
|
|
|
# 非第一季度数据需要减去上季度数据
|
|
mask = ~df['end_date'].str.endswith('0331')
|
|
df.loc[mask, 'basic_eps_q'] = df['basic_eps'].diff()
|
|
df.loc[mask, 'diluted_eps_q'] = df['diluted_eps'].diff()
|
|
|
|
# 过滤掉非季报数据（保留3/6/9/12月数据）
|
|
df = df[df['end_date'].str.endswith(('0331', '0630', '0930', '1231'))]
|
|
|
|
# 删除最小日期的数据行
|
|
if not df.empty:
|
|
df = df[df['end_date'] != df['end_date'].min()]
|
|
|
|
df['basic_eps_q'] = df['basic_eps_q'].round(PRECISION_CONFIG)
|
|
df['diluted_eps_q'] = df['diluted_eps_q'].round(PRECISION_CONFIG)
|
|
# 重置行号并保留原始EPS值
|
|
#return df[['ts_code', 'ann_date', 'end_date', 'basic_eps', 'diluted_eps', 'basic_eps_q', 'diluted_eps_q']].reset_index(drop=True)
|
|
#重置行号，去掉原始EPS值
|
|
return df[['ts_code', 'ann_date', 'end_date', 'basic_eps_q', 'diluted_eps_q']].reset_index(drop=True)
|
|
|
|
except Exception as e:
|
|
print(f"获取季度EPS数据出错: {e}")
|
|
return pd.DataFrame()
|
|
|
|
def calculate_ttm_eps(df):
|
|
"""
|
|
计算EPS指标的TTM(最近12个月)值
|
|
|
|
Parameters:
|
|
df (pd.DataFrame): get_quarterly_eps函数返回的DataFrame，包含以下列:
|
|
- ts_code: 股票代码
|
|
- ann_date: 公告日期
|
|
- end_date: 财报结束日期
|
|
- basic_eps: 基本每股收益(累计)
|
|
- diluted_eps: 稀释每股收益(累计)
|
|
- basic_eps_q: 单季度基本每股收益
|
|
- diluted_eps_q: 单季度稀释每股收益
|
|
|
|
Returns:
|
|
pd.DataFrame: 包含原始数据和TTM计算结果的DataFrame，新增以下列:
|
|
- basic_eps_ttm: 基本每股收益TTM值
|
|
- diluted_eps_ttm: 稀释每股收益TTM值
|
|
- basic_eps_q_ttm: 单季度基本每股收益TTM值
|
|
- diluted_eps_q_ttm: 单季度稀释每股收益TTM值
|
|
"""
|
|
if df.empty:
|
|
return df
|
|
|
|
try:
|
|
# 确保数据按end_date降序排列
|
|
df = df.sort_values('end_date', ascending=False).reset_index(drop=True)
|
|
|
|
# 初始化TTM结果列
|
|
#df['basic_eps_ttm'] = None
|
|
#df['diluted_eps_ttm'] = None
|
|
df['basic_eps_q_ttm'] = None
|
|
df['diluted_eps_q_ttm'] = None
|
|
|
|
# 遍历每一行数据计算TTM
|
|
for i in range(len(df)):
|
|
# 检查是否有足够的后续数据(至少3个季度)
|
|
if i + 3 >= len(df):
|
|
continue # 数据不足，跳过计算
|
|
|
|
# 计算TTM值(当前季度+后续3个季度)
|
|
#df.at[i, 'basic_eps_ttm'] = df.loc[i:i+3, 'basic_eps'].sum()
|
|
# df.at[i, 'diluted_eps_ttm'] = df.loc[i:i+3, 'diluted_eps'].sum()
|
|
df.at[i, 'basic_eps_q_ttm'] = df.loc[i:i+3, 'basic_eps_q'].sum()
|
|
df.at[i, 'diluted_eps_q_ttm'] = df.loc[i:i+3, 'diluted_eps_q'].sum()
|
|
|
|
# 四舍五入保留指定小数位数
|
|
ttm_cols = [ 'basic_eps_q_ttm', 'diluted_eps_q_ttm']
|
|
df[ttm_cols] = df[ttm_cols].round(PRECISION_CONFIG)
|
|
# 删除最后三行数据，因为最后三行ttm数据为None
|
|
df = df.iloc[:-3]
|
|
|
|
return df
|
|
|
|
except Exception as e:
|
|
print(f"计算TTM值时出错: {e}")
|
|
return pd.DataFrame()
|
|
def fill_trading_dates_with_eps(df_ttm,start_date=START_DATE,end_date=END_DATE):
|
|
"""
|
|
填充交易日期并保留EPS值
|
|
|
|
参数:
|
|
df_ttm (pd.DataFrame): calculate_ttm_eps函数返回的DataFrame，包含以下列:
|
|
- end_date: 财报结束日期(YYYYMMDD格式)
|
|
- basic_eps_ttm: 基本每股收益TTM值
|
|
- diluted_eps_ttm: 稀释每股收益TTM值
|
|
- basic_eps_q_ttm: 单季度基本每股收益TTM值
|
|
- diluted_eps_q_ttm: 单季度稀释每股收益TTM值
|
|
|
|
返回:
|
|
pd.DataFrame: 包含填充后的交易日期和对应EPS值的DataFrame
|
|
|
|
异常处理:
|
|
- 输入为空DataFrame时直接返回
|
|
- Tushare接口调用失败时返回原始数据
|
|
"""
|
|
if df_ttm.empty:
|
|
return df_ttm
|
|
|
|
try:
|
|
# 1. 准备数据: 按end_date排序并转换为datetime格式
|
|
df_ttm = df_ttm.sort_values('end_date')
|
|
df_ttm['end_date_dt'] = pd.to_datetime(df_ttm['end_date'])
|
|
# 2. 获取所有需要填充的日期区间
|
|
date_ranges = []
|
|
for i in range(len(df_ttm)-1):
|
|
s_date = df_ttm['end_date_dt'].iloc[i]
|
|
e_date = df_ttm['end_date_dt'].iloc[i+1]
|
|
date_ranges.append((s_date, e_date))
|
|
# 检查是否需要添加最后一个区间
|
|
if df_ttm['end_date_dt'].max() < pd.to_datetime(end_date):
|
|
next_report_date = pd.to_datetime(get_next_report_date(df_ttm['end_date_dt'].max()))
|
|
next_report_date = min(next_report_date, pd.to_datetime(end_date))
|
|
|
|
date_ranges.append((df_ttm['end_date_dt'].max(), next_report_date))
|
|
# 3. 获取交易所交易日历
|
|
exchange = 'SZSE' if df_ttm['ts_code'].iloc[0].endswith('SZ') else 'SSE' # 可以不用考虑SH,SZ,BJ, 理论上日历应该是一样的
|
|
trade_cal = pro.trade_cal(exchange=exchange,
|
|
start_date=start_date,
|
|
end_date=end_date)
|
|
|
|
# 过滤出交易日
|
|
trade_cal = trade_cal[trade_cal['is_open'] == 1]
|
|
trade_cal['cal_date_dt'] = pd.to_datetime(trade_cal['cal_date'])
|
|
|
|
# 4. 填充每个区间内的交易日
|
|
filled_data = []
|
|
#eps_cols = ['basic_eps_ttm', 'diluted_eps_ttm', 'basic_eps_q_ttm', 'diluted_eps_q_ttm']
|
|
|
|
# 首先添加原始数据
|
|
for _, row in df_ttm.iterrows():
|
|
filled_data.append(row.to_dict())
|
|
|
|
# 然后填充区间数据
|
|
for start_date, end_date in date_ranges:
|
|
# 获取该区间内的所有交易日
|
|
mask = (trade_cal['cal_date_dt'] > start_date) & (trade_cal['cal_date_dt'] < end_date)
|
|
dates_in_range = trade_cal[mask]['cal_date_dt']
|
|
|
|
# 使用较小的日期(即start_date)的EPS值填充
|
|
ref_row = df_ttm[df_ttm['end_date_dt'] == start_date].iloc[0]
|
|
|
|
for date in dates_in_range:
|
|
new_row = ref_row.copy()
|
|
new_row['end_date'] = date.strftime('%Y%m%d')
|
|
new_row['end_date_dt'] = date
|
|
filled_data.append(new_row)
|
|
|
|
# 5. 转换为DataFrame并整理
|
|
result = pd.DataFrame(filled_data)
|
|
result = result.sort_values('end_date_dt')
|
|
|
|
# 删除临时列并重置索引
|
|
result = result.drop(columns=['end_date_dt']).reset_index(drop=True)
|
|
result = result.drop(columns=['ann_date']) #去掉ann_date列
|
|
result = result.rename(columns={'end_date': 'trade_date'}) #重命名 end_date 为trade_date
|
|
# 过滤掉非交易日
|
|
result = result[result['trade_date'].isin(trade_cal['cal_date'])]
|
|
return result
|
|
|
|
except Exception as e:
|
|
print(f"填充交易日期时出错: {e}, 返回原始数据")
|
|
return df_ttm
|
|
if __name__ == '__main__':
|
|
#df = getStockEp(tscodeCheck('688469'))
|
|
'''df = get_quarterly_eps(tscodeCheck('688556'), '20230101', END_DATE)
|
|
print(df)
|
|
df2 = calculate_ttm_eps(df)
|
|
print(df2)
|
|
df3=fill_trading_dates_with_eps(df2)
|
|
print(df3)'''
|
|
df4= getStockEp_ttm(tscodeCheck('300316'), '20230101', END_DATE)
|
|
print(df4)
|