You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

73 lines
2.5 KiB

import sys
import json
import pandas as pd
class Unity(object):
def __init__(self):
const_path = sys.path[0].replace("\\analyze_data", "")
f = open(const_path + "\\const.json", "r", encoding="utf8")
self.consts = json.loads(f.read())
def do(self):
f1 = open(sys.path[0] + "\\hly_count_res_max_4_group_by_date.json", "r", encoding="utf-8")
hly = json.loads(f1.read())
f2 = open(sys.path[0] + "\\zljc\\zljc_day_a.json", "r", encoding="utf-8")
zljc = json.loads(f2.read())
res = {}
for code in zljc:
for a in zljc[code]:
for day in zljc[code][a]:
for date in zljc[code][a][day]["dates_yes"]:
# 筛选 hly 中该日期的股票有没有当前这只
if date in hly.keys() and code in hly[date]:
if date not in res.keys(): res[date] = []
if code not in res[date]: res[date].append(code)
f3 = open(sys.path[0] + "\\hly_and_zljc_res.json", "w", encoding="utf-8")
f3.write(json.dumps(res, ensure_ascii=False))
f3.close()
f2.close()
f1.close()
def verify(self):
f = open(sys.path[0] + "\\hly_and_zljc_res.json", "r", encoding="utf-8")
res = json.loads(f.read())
transform_dic = {}
for date in res:
for code in res[date]:
if code not in transform_dic:
transform_dic[code] = []
transform_dic[code].append(date)
# f3 = open(sys.path[0] + "\\hly_and_zljc_res_transform.json", "w", encoding="utf-8")
# f3.write(json.dumps(transform_dic, ensure_ascii=False))
# f3.close()
count = {
"yes": 0,
"no": 0
}
for code in transform_dic:
print(code)
try:
df = pd.read_csv("%s%s.csv" % (self.consts['path']['result']['hly'], code), encoding="gbk")
except:
print("ERROR OPENING %s" % code)
continue
for index, row in df.iterrows():
if row['日期'] in transform_dic[code]:
if index - 5 > 0:
if df.loc[index-5, "收盘价"] > row["收盘价"]:
count["yes"] += 1
else:
count["no"] += 1
print(count)
if __name__ == "__main__":
u = Unity()
# u.do()
u.verify()