stock_fundamentals/src/scripts/stock_daily_data_collector.py

271 lines
11 KiB
Python
Raw Normal View History

2025-04-02 13:52:34 +08:00
# coding:utf-8
import requests
import pandas as pd
from sqlalchemy import create_engine, text
2025-06-24 10:51:05 +08:00
from datetime import datetime, timedelta
2025-04-02 13:52:34 +08:00
from tqdm import tqdm
2025-05-19 17:02:52 +08:00
from src.scripts.config import XUEQIU_HEADERS
2025-05-14 08:54:56 +08:00
import gc
2025-04-02 13:52:34 +08:00
class StockDailyDataCollector:
"""股票日线数据采集器类"""
2025-05-14 08:54:56 +08:00
2025-04-02 13:52:34 +08:00
def __init__(self, db_url):
2025-05-14 08:54:56 +08:00
self.engine = create_engine(
db_url,
pool_size=5,
max_overflow=10,
pool_recycle=3600
)
2025-04-02 13:52:34 +08:00
self.headers = XUEQIU_HEADERS
def fetch_all_stock_codes(self):
2025-05-19 17:02:52 +08:00
# 从gp_code_all获取股票代码
query_all = "SELECT gp_code FROM gp_code_all"
df_all = pd.read_sql(query_all, self.engine)
codes_all = df_all['gp_code'].tolist()
# 从gp_code_zs获取股票代码
query_zs = "SELECT gp_code FROM gp_code_zs"
df_zs = pd.read_sql(query_zs, self.engine)
codes_zs = df_zs['gp_code'].tolist()
2025-06-24 10:51:05 +08:00
# 从gp_code_hk获取股票代码
query_hk = "SELECT gp_code FROM gp_code_hk"
df_hk = pd.read_sql(query_hk, self.engine)
codes_hk = df_hk['gp_code'].tolist()
2025-05-19 17:02:52 +08:00
# 合并去重
2025-06-24 10:51:05 +08:00
all_codes = list(set(codes_all + codes_zs + codes_hk))
print(f"获取到股票代码: {len(codes_all)}个来自gp_code_all, {len(codes_zs)}个来自gp_code_zs, {len(codes_hk)}个来自gp_code_hk, 去重后共{len(all_codes)}")
2025-05-19 17:02:52 +08:00
return all_codes
2025-04-02 13:52:34 +08:00
2025-06-24 10:51:05 +08:00
def fetch_daily_stock_data(self, symbol, begin, count=-1):
"""获取日线数据count=-1表示最新一天-2表示最近两天-1800表示最近1800天"""
url = f"https://stock.xueqiu.com/v5/stock/chart/kline.json?symbol={symbol}&begin={begin}&period=day&type=before&count={count}&indicator=kline,pe,pb,ps,pcf,market_capital,agt,ggt,balance"
2025-05-14 08:54:56 +08:00
try:
2025-06-24 10:51:05 +08:00
response = requests.get(url, headers=self.headers, timeout=20)
2025-05-14 08:54:56 +08:00
return response.json()
except Exception as e:
print(f"Request error for {symbol}: {e}")
return {'error_code': -1, 'error_description': str(e)}
def transform_data(self, data, symbol):
2025-04-02 13:52:34 +08:00
try:
items = data['data']['item']
columns = data['data']['column']
except KeyError as e:
print(f"KeyError for {symbol}: {e}")
2025-05-14 08:54:56 +08:00
return None
2025-04-02 13:52:34 +08:00
df = pd.DataFrame(items, columns=columns)
df['symbol'] = symbol
2025-05-14 08:54:56 +08:00
required_columns = ['timestamp', 'volume', 'open', 'high', 'low', 'close',
'chg', 'percent', 'turnoverrate', 'amount', 'symbol', 'pb', 'pe', 'ps']
2025-04-02 13:52:34 +08:00
existing_columns = [col for col in required_columns if col in df.columns]
df = df[existing_columns]
2025-05-14 08:54:56 +08:00
2025-04-02 13:52:34 +08:00
if 'timestamp' in df.columns:
df['timestamp'] = pd.to_datetime(df['timestamp'], unit='ms', utc=True).dt.tz_convert('Asia/Shanghai')
2025-05-14 08:54:56 +08:00
return df
def save_batch_to_database(self, batch):
if batch:
df_all = pd.concat(batch, ignore_index=True)
df_all.to_sql('gp_day_data', self.engine, if_exists='append', index=False)
2025-04-02 13:52:34 +08:00
def fetch_data_for_date(self, date=None):
if date is None:
start_date = datetime.now()
date_str = start_date.strftime('%Y-%m-%d')
else:
start_date = datetime.strptime(date, '%Y-%m-%d')
date_str = date
2025-05-14 08:54:56 +08:00
delete_query = text("DELETE FROM gp_day_data WHERE `timestamp` LIKE :date_str")
with self.engine.begin() as conn:
2025-04-02 13:52:34 +08:00
conn.execute(delete_query, {"date_str": f"{date_str}%"})
2025-05-14 08:54:56 +08:00
2025-04-02 13:52:34 +08:00
stock_codes = self.fetch_all_stock_codes()
2025-05-14 08:54:56 +08:00
begin = int(start_date.replace(hour=0, minute=0, second=0, microsecond=0).timestamp() * 1000)
2025-04-02 13:52:34 +08:00
2025-05-14 08:54:56 +08:00
batch_data = []
for idx, symbol in enumerate(tqdm(stock_codes, desc=f"Fetching and saving daily stock data for {date_str}")):
2025-04-02 13:52:34 +08:00
data = self.fetch_daily_stock_data(symbol, begin)
2025-05-14 08:54:56 +08:00
if data.get('error_code') == 0:
df = self.transform_data(data, symbol)
if df is not None:
batch_data.append(df)
2025-04-02 13:52:34 +08:00
else:
2025-05-14 08:54:56 +08:00
print(f"Error fetching data for {symbol} on {date_str}: {data.get('error_description')}")
if len(batch_data) >= 100:
self.save_batch_to_database(batch_data)
batch_data.clear()
gc.collect()
# Save remaining data
if batch_data:
self.save_batch_to_database(batch_data)
gc.collect()
2025-04-02 13:52:34 +08:00
2025-05-14 08:54:56 +08:00
self.engine.dispose()
2025-04-02 13:52:34 +08:00
print(f"Daily data fetching and saving completed for {date_str}.")
2025-06-24 10:51:05 +08:00
def delete_stock_history(self, symbol):
"""删除指定股票的全部历史数据"""
delete_query = text("DELETE FROM gp_day_data WHERE symbol = :symbol")
try:
with self.engine.begin() as conn:
conn.execute(delete_query, {"symbol": symbol})
print(f"Deleted history for {symbol}")
return True
except Exception as e:
print(f"Error deleting history for {symbol}: {e}")
return False
def refetch_and_save_history(self, symbol, days=1800):
"""重新获取并保存指定股票的长期历史数据"""
print(f"Refetching last {days} days for {symbol}...")
begin = int(datetime.now().timestamp() * 1000)
data = self.fetch_daily_stock_data(symbol, begin, count=-days)
if data.get('error_code') == 0:
df = self.transform_data(data, symbol)
if df is not None and not df.empty:
self.save_batch_to_database([df])
print(f"Successfully refetched and saved history for {symbol}.")
else:
print(f"No data transformed for {symbol} after refetch.")
else:
print(f"Error refetching history for {symbol}: {data.get('error_description')}")
def check_and_fix_ex_rights_data(self):
"""
检查所有股票是否发生除权如果发生则删除历史数据并重新获取
新逻辑直接用API返回的上个交易日的时间戳去数据库查询更稳妥
记录除权日期股票代码
"""
all_codes = self.fetch_all_stock_codes()
ex_rights_log_data = []
print("--- Step 1: Checking for ex-rights stocks ---")
for symbol in tqdm(all_codes, desc="Comparing prices"):
# 1. 从API获取最近两天的日线数据
begin = int(datetime.now().timestamp() * 1000)
data = self.fetch_daily_stock_data(symbol, begin, count=-2)
api_timestamp_str = None
api_close = None
if data.get('error_code') == 0 and data.get('data', {}).get('item') and len(data['data']['item']) >= 2:
try:
# API返回的数据是按时间升序的[-2]是上个交易日
prev_day_data = data['data']['item'][-2]
columns = data['data']['column']
timestamp_index = columns.index('timestamp')
close_index = columns.index('close')
api_timestamp_ms = prev_day_data[timestamp_index]
api_close = prev_day_data[close_index]
# 将毫秒时间戳转换为'YYYY-MM-DD'格式,用于数据库查询
api_timestamp_str = pd.to_datetime(api_timestamp_ms, unit='ms', utc=True).tz_convert('Asia/Shanghai').strftime('%Y-%m-%d')
except (ValueError, IndexError, TypeError) as e:
print(f"\nError parsing API data for {symbol}: {e}")
continue # 处理下一只股票
else:
# 获取API数据失败或数据不足跳过此股票
continue
# 如果未能从API解析出上个交易日的数据则跳过
if api_timestamp_str is None or api_close is None:
continue
# 2. 根据API返回的时间戳从数据库查询当天的收盘价
db_close = None
query = text("SELECT `close` FROM gp_day_data WHERE symbol = :symbol AND `timestamp` LIKE :date_str")
try:
with self.engine.connect() as conn:
result = conn.execute(query, {"symbol": symbol, "date_str": f"{api_timestamp_str}%"}).fetchone()
db_close = result[0] if result else None
except Exception as e:
print(f"\nError getting DB close for {symbol} on {api_timestamp_str}: {e}")
continue
# 3. 比较价格
if db_close is not None:
# 注意数据库中取出的db_close可能是Decimal类型需要转换
if not abs(float(db_close) - api_close) < 0.001:
print(f"\nEx-rights detected for {symbol} on {api_timestamp_str}: DB_close={db_close}, API_close={api_close}")
ex_rights_log_data.append({
'symbol': symbol,
'date': datetime.now().strftime('%Y-%m-%d'),
'db_price': float(db_close),
'api_price': api_close,
'log_time': datetime.now()
})
# 如果数据库当天没有数据,我们无法比较,所以不处理。
# 这可能是新股或之前采集失败,不属于除权范畴。
# 4. 对发生除权的股票进行记录和修复
if not ex_rights_log_data:
print("\n--- No ex-rights stocks found. Data is consistent. ---")
self.engine.dispose()
return
# 在修复前,先将日志保存到数据库
self.save_ex_rights_log(ex_rights_log_data)
# 从日志数据中提取出需要修复的股票代码列表
ex_rights_stocks = [item['symbol'] for item in ex_rights_log_data]
print(f"\n--- Step 2: Found {len(ex_rights_stocks)} stocks to fix: {ex_rights_stocks} ---")
for symbol in tqdm(ex_rights_stocks, desc="Fixing data"):
if self.delete_stock_history(symbol):
self.refetch_and_save_history(symbol, days=1800)
self.engine.dispose()
print("\n--- Ex-rights data fixing process completed. ---")
def save_ex_rights_log(self, log_data: list):
"""将除权日志保存到数据库"""
if not log_data:
return
print(f"--- Saving {len(log_data)} ex-rights events to log table... ---")
try:
df = pd.DataFrame(log_data)
# 确保列名与数据库字段匹配
df = df.rename(columns={
'symbol': 'stock_code',
'date': 'change_date',
'db_price': 'before_price',
'api_price': 'after_price',
'log_time': 'update_time'
})
df.to_sql('gp_ex_rights_log', self.engine, if_exists='append', index=False)
print("--- Ex-rights log saved successfully. ---")
except Exception as e:
print(f"!!! Error saving ex-rights log: {e}")
2025-04-02 13:52:34 +08:00
def collect_stock_daily_data(db_url, date=None):
collector = StockDailyDataCollector(db_url)
collector.fetch_data_for_date(date)
2025-06-24 10:51:05 +08:00
collector.check_and_fix_ex_rights_data()
2025-04-02 13:52:34 +08:00
if __name__ == "__main__":
2025-04-02 16:16:35 +08:00
db_url = 'mysql+pymysql://root:Chlry#$.8@192.168.18.199:3306/db_gp_cj'
2025-06-24 10:51:05 +08:00
# --- 使用方式 ---
# 1. 日常采集当天数据
# collect_stock_daily_data(db_url)
# 2. 手动执行除权检查和数据修复
# collector = StockDailyDataCollector(db_url)
# collector.check_and_fix_ex_rights_data()