import pandas as pd import os from datetime import time as s_time from datetime import datetime import chardet import numpy as np # 日盘商品期货交易品种 commodity_day_dict = {'bb': s_time(15,00), 'jd': s_time(15,00), 'lh': s_time(15,00), 'l': s_time(15,00), 'fb': s_time(15,00), 'ec': s_time(15,00), 'AP': s_time(15,00), 'CJ': s_time(15,00), 'JR': s_time(15,00), 'LR': s_time(15,00), 'RS': s_time(15,00), 'PK': s_time(15,00), 'PM': s_time(15,00), 'PX': s_time(15,00), 'RI': s_time(15,00), 'SF': s_time(15,00), 'SM': s_time(15,00), 'UR': s_time(15,00), 'WH': s_time(15,00), 'ao': s_time(15,00), 'br': s_time(15,00), 'wr': s_time(15,00),} # 夜盘商品期货交易品种 commodity_night_dict = {'sc': s_time(2,30), 'bc': s_time(1,0), 'lu': s_time(23,0), 'nr': s_time(23,0),'au': s_time(2,30), 'ag': s_time(2,30), 'ss': s_time(1,0), 'sn': s_time(1,0), 'ni': s_time(1,0), 'pb': s_time(1,0),'zn': s_time(1,0), 'al': s_time(1,0), 'cu': s_time(1,0), 'ru': s_time(23,0), 'rb': s_time(23,0), 'hc': s_time(23,0), 'fu': s_time(23,0), 'bu': s_time(23,0), 'sp': s_time(23,0), 'PF': s_time(23,0), 'SR': s_time(23,0), 'CF': s_time(23,0), 'CY': s_time(23,0), 'RM': s_time(23,0), 'MA': s_time(23,0), 'TA': s_time(23,0), 'ZC': s_time(23,0), 'FG': s_time(23,0), 'OI': s_time(23,0), 'SA': s_time(23,0), 'p': s_time(23,0), 'j': s_time(23,0), 'jm': s_time(23,0), 'i': s_time(23,0), 'l': s_time(23,0), 'v': s_time(23,0), 'pp': s_time(23,0), 'eg': s_time(23,0), 'c': s_time(23,0), 'cs': s_time(23,0), 'y': s_time(23,0), 'm': s_time(23,0), 'a': s_time(23,0), 'b': s_time(23,0), 'rr': s_time(23,0), 'eb': s_time(23,0), 'pg': s_time(23,0), 'SH': s_time(23,00)} # 金融期货交易品种 financial_time_dict = {'IH': s_time(15,00), 'IF': s_time(15,00), 'IC': s_time(15,00), 'IM': s_time(15,00),'T': s_time(15,00), 'TS': s_time(15,00), 'TF': s_time(15,00), 'TL': s_time(15,00)} # 所有已列入的筛选品种 all_dict = {k: v for d in [commodity_day_dict, commodity_night_dict, financial_time_dict] for k, v in d.items()} def split_alpha_numeric(string): alpha_chars = "" numeric_chars = "" for char in string: if char.isalpha(): alpha_chars += char elif char.isdigit(): numeric_chars += char return alpha_chars, numeric_chars def merged_old_tickdata(merged_up_df, sp_char, alpha_chars, code_value): # merged_up_df = pd.DataFrame() # merged_up_df,alpha_chars,code_value = merged_old_unprocessed_tickdata(all_csv_files, sp_char) while alpha_chars not in all_dict.keys(): print("%s期货品种未列入所有筛选条件中!!!"%(code_value)) continue merged_df = pd.DataFrame() merged_df =pd.DataFrame({'main_contract':merged_up_df['统一代码'],'symbol':merged_up_df['合约代码'],'datetime':merged_up_df['时间'],'lastprice':merged_up_df['最新'],'volume':merged_up_df['成交量'], 'bid_p':merged_up_df['买一价'],'ask_p':merged_up_df['卖一价'],'bid_v':merged_up_df['买一量'],'ask_v':merged_up_df['卖一量']}) del merged_up_df merged_df['datetime'] = pd.to_datetime(merged_df['datetime']) merged_df['tmp_time'] = merged_df['datetime'].dt.strftime('%H:%M:%S.%f') merged_df['time'] = merged_df['tmp_time'].apply(lambda x: datetime.strptime(x, '%H:%M:%S.%f')).dt.time del merged_df['tmp_time'] merged_df = filter_tickdata_time(merged_df, alpha_chars) del merged_df['time'] merged_df['datetime'] = sorted(merged_df['datetime']) print("%s%s数据生成成功!"%(code_value,sp_char)) return merged_df def merged_new_tickdata(merged_up_df, sp_char, alpha_chars, code_value): # merged_up_df = pd.DataFrame() # merged_up_df,alpha_chars,code_value = merged_new_unprocessed_tickdata(all_csv_files, sp_char) while alpha_chars not in all_dict.keys(): print("%s期货品种未列入所有筛选条件中!!!"%(code_value)) continue #日期修正 # merged_df['业务日期'] = pd.to_datetime(merged_df['业务日期']) # merged_df['业务日期'] = merged_df['业务日期'].dt.strftime('%Y-%m-%d') # merged_df['最后修改时间'] = pd.to_datetime(merged_df['最后修改时间']) merged_up_df['datetime'] = merged_up_df['业务日期'].astype(str) + ' '+merged_up_df['最后修改时间'].astype(str) + '.' + merged_up_df['最后修改毫秒'].astype(str) # merged_df['最后修改时间'].dt.time.astype(str) # 将'datetime' 列的数据类型更改为 datetime 格式，如果数据转换少8个小时，可以用timedelta处理 merged_up_df['datetime'] = pd.to_datetime(merged_up_df['datetime'], errors='coerce', format='%Y-%m-%d %H:%M:%S.%f') #计算瞬时成交量 merged_up_df['volume'] = merged_up_df['数量'] - merged_up_df['数量'].shift(1) merged_up_df['volume'] = merged_up_df['volume'].fillna(0) merged_df = pd.DataFrame() merged_df =pd.DataFrame({'main_contract':merged_up_df['统一代码'],'symbol':merged_up_df['合约代码'],'datetime':merged_up_df['datetime'],'lastprice':merged_up_df['最新价'],'volume':merged_up_df['volume'], 'bid_p':merged_up_df['申买价一'],'ask_p':merged_up_df['申卖价一'],'bid_v':merged_up_df['申买量一'],'ask_v':merged_up_df['申卖量一']}) del merged_up_df # merged_df['datetime'] = pd.to_datetime(merged_df['datetime']) merged_df['tmp_time'] = merged_df['datetime'].dt.strftime('%H:%M:%S.%f') merged_df['time'] = merged_df['tmp_time'].apply(lambda x: datetime.strptime(x, '%H:%M:%S.%f')).dt.time del merged_df['tmp_time'] merged_df = filter_tickdata_time(merged_df, alpha_chars) del merged_df['time'] # merged_df['datetime'] = sorted(merged_df['datetime']) sorted_merged_df = merged_df.sort_values(by = ['datetime'], inplace=True) print("%s%s数据生成成功!"%(code_value,sp_char)) return merged_df def filter_tickdata_time(filter_df, alpha_chars): if alpha_chars in financial_time_dict.keys(): drop_index1 = pd.DataFrame().index drop_index2 = filter_df.loc[(filter_df['time'] > s_time(11, 30, 0, 000000)) & (filter_df['time'] < s_time(13, 0, 0, 000000))].index drop_index3 = filter_df.loc[(filter_df['time'] > s_time(15, 0, 0, 000000)) | (filter_df['time'] < s_time(9, 30, 0, 000000))].index drop_index4 = pd.DataFrame().index print("按照中金所交易时间筛选金融期货品种") elif alpha_chars in commodity_night_dict.keys(): if commodity_night_dict[alpha_chars] == s_time(23,00): drop_index1 = filter_df.loc[(filter_df['time'] > s_time(10, 15, 0, 000000)) & (filter_df['time'] < s_time(10, 30, 0, 000000))].index drop_index2 = filter_df.loc[(filter_df['time'] > s_time(11, 30, 0, 000000)) & (filter_df['time'] < s_time(13, 30, 0, 000000))].index drop_index3 = filter_df.loc[(filter_df['time'] > s_time(15, 0, 0, 000000)) & (filter_df['time'] < s_time(21, 0, 0, 000000))].index drop_index4 = filter_df.loc[(filter_df['time'] > s_time(23, 0, 0, 000000)) | (filter_df['time'] < s_time(9, 0, 0, 000000))].index print("按照夜盘截止交易时间为23:00筛选商品期货品种") elif commodity_night_dict[alpha_chars] == s_time(1,00): drop_index1 = filter_df.loc[(filter_df['time'] > s_time(10, 15, 0, 000000)) & (filter_df['time'] < s_time(10, 30, 0, 000000))].index drop_index2 = filter_df.loc[(filter_df['time'] > s_time(11, 30, 0, 000000)) & (filter_df['time'] < s_time(13, 30, 0, 000000))].index drop_index3 = filter_df.loc[(filter_df['time'] > s_time(15, 0, 0, 000000)) & (filter_df['time'] < s_time(21, 0, 0, 000000))].index drop_index4 = filter_df.loc[(filter_df['time'] > s_time(1, 0, 0, 000000)) & (filter_df['time'] < s_time(9, 0, 0, 000000))].index print("按照夜盘截止交易时间为1:00筛选商品期货品种") elif commodity_night_dict[alpha_chars] == s_time(2,30): drop_index1 = filter_df.loc[(filter_df['time'] > s_time(10, 15, 0, 000000)) & (filter_df['time'] < s_time(10, 30, 0, 000000))].index drop_index2 = filter_df.loc[(filter_df['time'] > s_time(11, 30, 0, 000000)) & (filter_df['time'] < s_time(13, 30, 0, 000000))].index drop_index3 = filter_df.loc[(filter_df['time'] > s_time(15, 0, 0, 000000)) & (filter_df['time'] < s_time(21, 0, 0, 000000))].index drop_index4 = filter_df.loc[(filter_df['time'] > s_time(2, 30, 0, 000000)) & (filter_df['time'] < s_time(9, 0, 0, 000000))].index print("按照夜盘截止交易时间为2:30筛选商品期货品种") else: print("夜盘截止交易时间未设置或者设置错误！！！") elif alpha_chars in commodity_day_dict.keys(): drop_index1 = filter_df.loc[(filter_df['time'] > s_time(10, 15, 0, 000000)) & (filter_df['time'] < s_time(10, 30, 0, 000000))].index drop_index2 = filter_df.loc[(filter_df['time'] > s_time(11, 30, 0, 000000)) & (filter_df['time'] < s_time(13, 30, 0, 000000))].index drop_index3 = filter_df.loc[(filter_df['time'] > s_time(15, 0, 0, 000000)) | (filter_df['time'] < s_time(9, 0, 0, 000000))].index drop_index4 = pd.DataFrame().index print("按照无夜盘筛选商品期货品种") else: print("%s期货品种未执行时间筛选中!!!"%(alpha_chars)) # 清理不在交易时间段的数据 # 数据清理 filter_df.drop(labels=drop_index1, axis=0, inplace=True) filter_df.drop(drop_index2, axis=0, inplace=True) filter_df.drop(drop_index3, axis=0, inplace=True) filter_df.drop(drop_index4, axis=0, inplace=True) return filter_df def insert_main_contract(df): # 添加主力连续的合约代码，主力连续为888，指数连续可以用999，次主力连续可以使用889，表头用“统一代码” alpha_chars, numeric_chars = split_alpha_numeric(df.loc[0,'合约代码']) code_value = alpha_chars + "889" print("code_value characters:", code_value) df.insert(loc=0,column="统一代码", value=code_value) return df, alpha_chars, code_value def merged_old_unprocessed_tickdata(all_csv_files, sp_char): csv_files = [sp_file for sp_file in all_csv_files if sp_char in sp_file] print("csv_files:", csv_files) merged_up_df = pd.DataFrame() dir = os.getcwd() fileNum_errors = 0 # 循环遍历每个csv文件 for file in csv_files: try: # 读取csv文件，并使用第一行为列标题，编译不通过可以改为gbk df = pd.read_csv(file, header=0, # usecols=[ 1, 2, 3, 7, 12, 13, 14, 15], # names=[ # "合约代码", # "时间", # "最新", # "成交量", # "买一价", # "卖一价", # "买一量", # "卖一量", # ], encoding='gbk', low_memory= False, # skiprows=0, # parse_dates=['时间'] # 注意此处增加的排序，为了后面按时间排序 ) except: file_path = os.path.join(dir, file) fileNum_errors += 1 with open(file_path, 'rb') as file: data = file.read() # 使用chardet检测编码 detected_encoding = chardet.detect(data)['encoding'] # print("%s当前文件不为gbk格式,其文件格式为%s,需要转换为gbk格式,错误总数为%s"%(file,detected_encoding,fileNum_errors)) print("%s:%s当前文件不为gbk格式,其文件格式为%s,需要转换为gbk格式,错误总数为%s"%(datetime.now().strftime('%Y-%m-%d %H:%M:%S'),file_path,detected_encoding,fileNum_errors)) with open('output_error.txt', 'a') as f: print("%s:%s当前文件不为gbk格式,其文件格式为%s,需要转换为gbk格式,错误总数为%s"%(datetime.now().strftime('%Y-%m-%d %H:%M:%S'),file_path,detected_encoding,fileNum_errors), file = f) # 删除重复行 df.drop_duplicates(inplace=True) # 将数据合并到新的DataFrame中 merged_up_df = pd.concat([merged_up_df, df], ignore_index=True) # 删除重复列 merged_up_df.drop_duplicates(subset=merged_up_df.columns.tolist(), inplace=True) # 重置行索引 merged_up_df.reset_index(inplace=True, drop=True) merged_up_df,alpha_chars,code_value = insert_main_contract(merged_up_df) # 打印提示信息 # print("按年份未处理的CSV文件合并成功！") return merged_up_df,alpha_chars,code_value def merged_new_unprocessed_tickdata(all_csv_files, sp_char): csv_files = [sp_file for sp_file in all_csv_files if sp_char in sp_file] print("csv_files:", csv_files) merged_up_df = pd.DataFrame() dir = os.getcwd() fileNum_errors = 0 # 循环遍历每个csv文件 for file in csv_files: try: # 读取csv文件，并使用第一行为列标题，编译不通过可以改为gbk df = pd.read_csv( file, header=0, # usecols=[0, 1, 4, 11, 20, 21, 22, 23, 24, 25, 43], # names=[ # "交易日", # "合约代码", # "最新价", # "数量", # "最后修改时间", # "最后修改毫秒", # "申买价一", # "申买量一", # "申卖价一", # "申卖量一", # "业务日期", # ], encoding='gbk', low_memory= False, # skiprows=0, # parse_dates=['业务日期','最后修改时间','最后修改毫秒'] # 注意此处增加的排序，为了后面按时间排序 ) except: file_path = os.path.join(dir, file) fileNum_errors += 1 with open(file_path, 'rb') as file: data = file.read() # 使用chardet检测编码 detected_encoding = chardet.detect(data)['encoding'] # print("%s当前文件不为gbk格式,其文件格式为%s,需要转换为gbk格式,错误总数为%s"%(file_path,detected_encoding,fileNum_errors)) print("%s:%s当前文件不为gbk格式,其文件格式为%s,需要转换为gbk格式,错误总数为%s"%(datetime.now().strftime('%Y-%m-%d %H:%M:%S'),file_path,detected_encoding,fileNum_errors)) with open('output_error.txt', 'a') as f: print("%s:%s当前文件不为gbk格式,其文件格式为%s,需要转换为gbk格式,错误总数为%s"%(datetime.now().strftime('%Y-%m-%d %H:%M:%S'),file_path,detected_encoding,fileNum_errors), file = f) # 删除重复行 df.drop_duplicates(inplace=True) # 将数据合并到新的DataFrame中 merged_up_df = pd.concat([merged_up_df, df], ignore_index=True) # 删除重复列 merged_up_df.drop_duplicates(subset=merged_up_df.columns.tolist(), inplace=True) # 重置行索引 merged_up_df.reset_index(inplace=True, drop=True) merged_up_df,alpha_chars,code_value = insert_main_contract(merged_up_df) # 打印提示信息 # print("按年份未处理的CSV文件合并成功！") return merged_up_df,alpha_chars,code_value def reinstatement_tickdata(merged_rs_df): merged_rs_df['main_contract'] = merged_rs_df['main_contract'].astype(str) merged_rs_df['symbol'] = merged_rs_df['symbol'].astype(str) merged_rs_df['datetime'] = pd.to_datetime(merged_rs_df['datetime'], errors='coerce', format='%Y-%m-%d %H:%M:%S.%f') # merged_rs_df['lastprice'] = merged_rs_df['lastprice'].astype(float) merged_rs_df['volume'] = merged_rs_df['volume'].astype(int) # merged_rs_df['bid_p'] = merged_rs_df['bid_p'].astype(float) # merged_rs_df['ask_p'] = merged_rs_df['ask_p'].astype(float) merged_rs_df['bid_v'] = merged_rs_df['bid_v'].astype(int) merged_rs_df['ask_v'] = merged_rs_df['ask_v'].astype(int) # 等比复权,先不考虑 # df['复权因子'] = df['卖一价'].shift() / df['买一价'] # df['复权因子'] = np.where(df['合约代码'] != df['合约代码'].shift(), df['卖一价'].shift() / df['买一价'], 1) # df['复权因子'] = df['复权因子'].fillna(1) # df['买一价_adj'] = df['买一价'] * df['复权因子'].cumprod() # df['卖一价_adj'] = df['卖一价'] * df['复权因子'].cumprod() # df['最新_adj'] = df['最新'] * df['复权因子'].cumprod() # 等差复权 merged_rs_df['复权因子'] = np.where(merged_rs_df['symbol'] != merged_rs_df['symbol'].shift(), merged_rs_df['ask_p'].shift() - merged_rs_df['bid_p'], 0) merged_rs_df['复权因子'] = merged_rs_df['复权因子'].fillna(0) merged_rs_df['bid_p_adj'] = merged_rs_df['bid_p'] + merged_rs_df['复权因子'].cumsum() merged_rs_df['ask_p_adj'] = merged_rs_df['ask_p'] + merged_rs_df['复权因子'].cumsum() merged_rs_df['lastprice_adj'] = merged_rs_df['lastprice'] + merged_rs_df['复权因子'].cumsum() # 将调整后的数值替换原来的值 merged_rs_df['bid_p'] = merged_rs_df['bid_p_adj'].round(4) merged_rs_df['ask_p'] = merged_rs_df['ask_p_adj'].round(4) merged_rs_df['lastprice'] = merged_rs_df['lastprice_adj'].round(4) # 删除多余的值 del merged_rs_df['复权因子'] del merged_rs_df['bid_p_adj'] del merged_rs_df['ask_p_adj'] del merged_rs_df['lastprice_adj'] return merged_rs_df # def find_files(all_csv_files): # all_csv_files = sorted(all_csv_files) # sp_old_chars = ['_2019','_2020','_2021'] # sp_old_chars = sorted(sp_old_chars) # sp_new_chars = ['_2022','_2023'] # sp_new_chars = sorted(sp_new_chars) # csv_old_files = [file for file in all_csv_files if any(sp_char in file for sp_char in sp_old_chars)] # csv_new_files = [file for file in all_csv_files if any(sp_char in file for sp_char in sp_new_chars)] # return csv_old_files, csv_new_files