add price update function

data_collector/data_collector.py

@@ -1,10 +1,13 @@
+from operator import itemgetter
 from tkinter import E
 import FinanceDataReader as fdr
 import pandas_datareader as pdr
 from marcap import marcap_data
 from query_manager import QueryManager
 import pandas as pd
 from datetime import datetime
+import numpy as np
+import time
 
 class DataCollector(QueryManager):
 
@@ -45,6 +48,7 @@ def get_price_table(self):
 			except Exception as e:
 				print(f'{e} : {code}')
 
+
 	def get_finance_table(self, path1):
 		self.create_finance_table()
 		total = len(self.codes)
@@ -54,6 +58,7 @@ def get_finance_table(self, path1):
 			cnt = self.replace_finance_table(code, at, total, cnt, path1)
 			at+=1
 
+
 	def get_raw_price_info(self):
 		self.create_raw_price_info_table()
 		raw_price_info_df = marcap_data('1995-05-02', '2021-12-31')
@@ -63,6 +68,7 @@ def get_raw_price_info(self):
 			self.replace_raw_price_info_table(r, at, total)
 			at+=1
 
+
 	def add_main_sector_tocompany(self, path):
 		total = len(self.codes)
 		at = 0
@@ -73,6 +79,7 @@ def add_main_sector_tocompany(self, path):
 			except Exception as e:
 				print(e, code)
 
+
 	def get_price_average_info(self):
 		self.create_price_average_table()
 		total = len(self.codes)
@@ -86,12 +93,13 @@ def get_price_average_info(self):
 				at_code+=1
 			at+=1
 
+
 	def get_price_monthly_info(self):
 		self.create_price_monthly_info_table()
 		data = self.bring_additional_data()
 		total = len(self.codes)
 		error_list = []
-		at = 0;
+		at = 0
 		for code in self.codes.itertuples():
 			try:
 				np_adjclose = self.get_adjclose(code.Symbol)
@@ -104,10 +112,12 @@ def get_price_monthly_info(self):
 					self.replace_price_monthly_table(code, x, y, at, total, at_code, total_code)
 					at_code+=1
 				at+=1
+
 
 			except Exception as e:
 				error_list.append([e, code])
 				print(f'{e} : {code}')
+				pass
 
 		print(error_list)
 
@@ -118,7 +128,30 @@ def get_market_open_info(self):
 		at = 0
 		for r in df_open.itertuples():
 			self.replace_market_open_info_table(r, at, total)
-			at+=1	
+			at += 1
+
+
+	def update_new_price_info(self):
+		flag = 0
+		data = None
+		# ck = time.time()
+		for code in self.codes.itertuples():
+			new = self.get_new_price_info(code, '2022-01-03', '2022-01-03')
+			if new is None:
+				continue
+			if flag == 0:
+				flag = 1
+				data = new
+				continue
+			data = np.append(data, new, axis=0)
+			if flag == 4:
+				break
+		types = ['int64', 'object', 'int64', 'int64', 'int64', 'int64', 'int64', 'int64', 'float64', 'int64', 'int64', 'int64']
+		df = pd.DataFrame(data)
+		dic = { name:value for name, value in zip(df.columns, types) }
+		df = df.astype(dic)
+		# print(time.time() - ck) # 706 초
+		df.to_csv('/Users/choewonjun/Documents/coding/finance_data/quant/data/price_monthly.csv', index=False, header=False,mode='a')
 
 
 if __name__ == "__main__":
@@ -130,5 +163,5 @@ def get_market_open_info(self):
 	# dc.get_finance_table(path)
 
 	# dc.add_main_sector_tocompany(path)
-	dc.get_price_monthly_info()
+	dc.update_new_price_info()
 	# dc.get_price_average_info()

data_collector/data_preprocessor.py

@@ -7,6 +7,7 @@
 import math
 from operator import itemgetter
 from marcap import marcap_data
+from pykrx import stock
 
 
 class DataPreprocessor:
@@ -211,4 +212,43 @@ def check_dec(self, ymd_parm, at):
 			if self.ymd == 12:
 				return 1
 			else:
-				return 0
+				return 0
+
+	def get_new_price_only_info(self, code, start, end):
+		"""
+		start, end 에서는 2021-00-00 형태로 보내주면 된다.
+		Open, High, Low, Close, Volume, Change
+		"""
+		df = fdr.DataReader(f'{code}', int(start[:4]))
+		df = df.reset_index()
+		df = df[df["Date"] >= datetime(int(start[:4]), int(start[5:7]), int(start[8:10]))]
+		df = df[df["Date"] <= datetime(int(end[:4]), int(end[5:7]), int(end[8:10]))]
+		df_np = df.to_numpy()
+		return (df_np)
+
+	def get_new_price_info(self, code, start, end):
+		""" 
+		가격 정보 업데이트 함수
+		start 부터 end 날 까지의 일일 가격 정보를 리턴해준다. 그런데 price csv 파일이 월단위로 있어서 추후 어떻게 할지 정해야할듯 하다.
+		"""
+		try:
+			price_info = self.get_new_price_only_info(code.Symbol, start, end)
+			ids = [[code.ID]] * len(price_info) # ID
+			dates_to_volume = price_info[:,:6] # Date, Open, High, Low, Close, Volume
+			dates_to_volume[0][0] = dates_to_volume[0][0].date()
+			pvolume = [[0]] * len(price_info) # PVolume : 뭔지 몰라서 0으로 해주었다.
+			changes = price_info[:,6:] # Changes
+			marcap = stock.get_market_cap(start.replace('-',''), end.replace('-',''), code.Symbol) # 시가총액, 거래량, 거래대금, 상장주식수
+			marcap = marcap.to_numpy()
+			marcap = marcap[:, [0, 3]] # 시가총액, 상장주식수
+			ranks = [[0]] * len(price_info) # 어떻게 구할지 몰라서 일단 0으로 해두었다. 필요시 계산작업 추가하겠다.
+			tmp_a = np.append(ids, dates_to_volume, axis=1)
+			tmp_b = np.append(pvolume, changes, axis=1)
+			tmp_c = np.append(marcap, ranks, axis=1)
+			new_price = np.append(tmp_a, tmp_b, axis=1)
+			new_price = np.append(new_price, tmp_c, axis=1)
+			# return
+			return new_price
+		except:
+			print(f'Exception : {code.Symbol}')
+			return None