def set_duration_console(du_date):
if cct.isMac():
cct.set_console(width, height)
else:
cct.set_console(width, height, title=str(du_date))
status = False
vol = ct.json_countVol
type = ct.json_countType
success = 0
top_all = pd.DataFrame()
time_s = time.time()
delay_time = 720000
# delay_time = cct.get_delay_time()
First = True
blkname = '062.blk'
block_path = tdd.get_tdx_dir_blocknew() + blkname
status_change = False
lastpTDX_DF = pd.DataFrame()
# dl=30
ptype = 'low'
# op, ra, duration_date, days = pct.get_linear_model_status('999999', filter='y', dl=dl, ptype=ptype, days=1)
# duration_date = int(ct.duration_date * 1.4)
duration_date = ct.duration_date_l
# duration_date = 120
# duration_date = 300
du_date = duration_date
resample = ct.resample_dtype
newdays = 5
end_date = None
ptype = 'low'
filter = 'y'
if cct.isMac():
cct.set_console(165, 16)
else:
cct.set_console(160, 15)
status = False
vol = '0'
type = '2'
# cut_num=10000
success = 0
top_all = pd.DataFrame()
time_s = time.time()
# delay_time = 3600
delay_time = cct.get_delay_time()
First = True
base_path = tdd.get_tdx_dir()
block_path = tdd.get_tdx_dir_blocknew() + '062.blk'
status_change = False
# all_diffpath = tdd.get_tdx_dir_blocknew() + '062.blk'
while 1:
try:
# df = sina_data.Sina().all
df = rl.get_sina_Market_json('all')
top_now = rl.get_market_price_sina_dd_realTime(df, vol, type)
# print top_now.loc['300208','name']
# top_now.to_hdf("testhdf5", 'marketDD', format='table', complevel=9)
df_count = len(df)
now_count = len(top_now)
del df
gc.collect()
radio_t = cct.get_work_time_ratio()
if cct.isMac():
cct.set_console(90, 16)
else:
cct.set_console(82, 16)
status = False
vol = "0"
type = "2"
cut_num = 20000
success = 0
top_all = pd.DataFrame()
time_s = time.time()
# delay_time = 7200
delay_time = cct.get_delay_time()
base_path = tdd.get_tdx_dir()
block_path = tdd.get_tdx_dir_blocknew() + "064.blk"
while 1:
try:
df = rl.get_sina_all_json_dd(vol, type)
if len(df) > cut_num:
df = df[:cut_num]
print len(df),
top_now = rl.get_sina_dd_count_price_realTime(df)
# print len(top_now)
time_d = time.time()
if time_d - time_s > delay_time:
status_change = True
time_s = time.time()
top_all = pd.DataFrame()
else:
# -*- coding:utf-8 -*-
import datetime
import time
import pandas as pd
import JohnsonUtil.commonTips as cct
# from JSONData import LineHistogram as lrh
from JSONData import powerCompute as pct
from JSONData import realdatajson as rl
from JSONData import tdx_data_Day as tdd
base_path = tdd.get_tdx_dir()
block_path = tdd.get_tdx_dir_blocknew() + '068.blk'
block_path_d = tdd.get_tdx_dir_blocknew() + '069.blk'
if __name__ == "__main__":
# df=sina_data.Sina().all.set_index('code')
today = datetime.date.today()
print today
def get_market_status(mk, type='m', start=None, end=None):
df = rl.get_sina_Market_json(mk).set_index('code')
time_s = time.time()
code_l = []
code_d = []
for code in df.index:
status, lenday, diff = pct.get_linear_model_status_LSH(code,
type=type,
start=start,
end=end)
# op, ra, status, dss = pct.get_linear_model_status(code, type=type, start=start, end=end)
# -*- coding:utf-8 -*-
import time
import datetime
import pandas as pd
import LineHistogram as lrh
from JSONData import realdatajson as rl
import JohhnsonUtil.commonTips as cct
from JSONData import tdx_data_Day as tdd
base_path = tdd.get_tdx_dir()
block_path = tdd.get_tdx_dir_blocknew() + '068.blk'
block_path_d = tdd.get_tdx_dir_blocknew() + '069.blk'
if __name__ == "__main__":
# df=sina_data.Sina().all.set_index('code')
today = datetime.date.today()
print today
def get_market_status(mk, type='m'):
df = rl.get_sina_Market_json(mk).set_index('code')
time_s = time.time()
code_l = []
code_d = []
for code in df.index:
status, lenday, diff = lrh.get_linear_model_status(code, type)
if status:
# print(df.loc[code,:])
# if df.loc[code, 'percent'] > 0 and lenday > 200:
if df.loc[code, 'trade'] != 0 and lenday > 200:
df.loc[code, 'diff'] = abs(diff)
code_l.append(df.loc[code, :])
elif df.loc[code, 'trade'] != 0 and lenday > 200:
def get_roll_mean_all(single=True,
tdx=False,
app=True,
duration=100,
ma_250_l=1.02,
ma_250_h=1.11):
# df = tdd.search_Tdx_multi_data_duration('tdx_all_df_300', 'all_300', df=None,code_l=code_list, start=start, end=None, freq=None, col=None, index='date')
block_path = tdd.get_tdx_dir_blocknew() + '060.blk'
if not app and cct.get_file_size(
block_path) > 100 and cct.creation_date_duration(block_path) == 0:
print "It's Today Update"
return True
code_list = sina_data.Sina().market('all').index.tolist()
print "all code:", len(code_list)
if duration < 300:
h5_fname = 'tdx_all_df' + '_' + str(300)
h5_table = 'all' + '_' + str(300)
else:
h5_fname = 'tdx_all_df' + '_' + str(900)
h5_table = 'all' + '_' + str(900)
# df = tdd.search_Tdx_multi_data_duration('tdx_all_df_300', 'all_300', df=None,code_l=code_list, start='20150501', end=None, freq=None, col=None, index='date')
df = tdd.search_Tdx_multi_data_duration(h5_fname,
h5_table,
df=None,
code_l=code_list,
start=None,
end=None,
freq=None,
col=None,
index='date')
# df = tdd.search_Tdx_multi_data_duration(h5_fname, h5_table, df=None,code_l=code_list, start=None, end=None, freq=None, col=None, index='date',tail=1)
code_uniquelist = df.index.get_level_values('code').unique()
code_select = code_uniquelist[random.randint(0, len(code_uniquelist) - 1)]
print round(time.time() - time_s, 2), df.index.get_level_values(
'code').unique().shape, code_select, df.loc[code_select].shape
# df.groupby(level=[0]),df.index.get_level_values(0)
# len(df.index.get_level_values('code').unique())
# df = df[~df.index.duplicated(keep='first')]
dfs = df
def get_groupby_mean_median_close(dfs):
groupd = dfs.groupby(level=[0])
df = groupd['close'].agg({'median': 'median', 'mean': 'mean'})
df['close'] = groupd.tail(1).reset_index().set_index(['code'])['close']
# dfs['mean'] = groupd['close'].agg('mean')
# dfs['median'] = groupd['close'].agg('median')
# dfs = dfs.fillna(0)
# idx = pd.IndexSlice
# mask = ( (dfs['mean'] > dfs['median'])
# & (dfs['close'] > dfs['mean'])
# )
# df=dfs.loc[idx[mask, :]]
df = df[(df['mean'] > df['median']) & (df['close'] > df['mean'])]
# dt_low = None
# if dl == 1:
# dfs = groupd.tail(1)
# print("dfs tail1")
# else:
# dl = 30
# dindex = tdd.get_tdx_Exp_day_to_df(
# '999999', dl=dl).sort_index(ascending=False)
# dt = tdd.get_duration_price_date('999999', df=dindex)
# dt = dindex[dindex.index >= dt].index.values
# dt_low = dt[-1]
# dtlen = len(dt) if len(dt) >0 else 1
# dfs = groupd.tail(dtlen)
# print("dfs tail:%s dt:%s"%(dtlen,dt))
# dfs = get_multi_date_duration(dfs,dt[-1])
return df
groupd = dfs.groupby(level=[0])
# rollma = ['5','10','60','100','200']
# rollma = ['5','10','250']
if duration < 300:
rollma = ['10']
else:
rollma = ['10', '250']
rollma.extend([str(duration)])
# import ipdb;ipdb.set_trace()
# df.loc['300130'][:2]
# dfs['mean'] = groupd['close'].agg('mean')
# dfs['median'] = groupd['close'].agg('median')
for da in rollma:
cumdays = int(da)
dfs['ma%d' % cumdays] = groupd['close'].apply(pd.rolling_mean, cumdays)
if cumdays == 10:
dfs['upper'] = dfs['ma%d' % cumdays].apply(lambda x: round(
(1 + 11.0 / 100) * x, 1))
dfs['lower'] = dfs['ma%d' % cumdays].apply(lambda x: round(
(1 - 9.0 / 100) * x, 1))
dfs['ene'] = map(lambda x, y: round((x + y) / 2, 1), dfs['upper'],
dfs['lower'])
# df['upper'] = map(lambda x: round((1 + 11.0 / 100) * x, 1), df.ma10d)
# df['lower'] = map(lambda x: round((1 - 9.0 / 100) * x, 1), df.ma10d)
# df['ene'] = map(lambda x, y: round((x + y) / 2, 1), df.upper, df.lower)
# dfs['amount%d'%cumdays] = groupd['amount'].apply(pd.rolling_mean, cumdays)
# df.ix[df.index.levels[0]]
#df.ix[df.index[len(df.index)-1][0]] #last row
# dfs = tdd.search_Tdx_multi_data_duration(df=dfs,code_l=code_list, start='20170918', end='20170918', freq=None, col=None, index='date')
# print dfs[:1],len(dfs)
# groupd.agg({'low': 'min'})
# '''idx mask filter'''
# '''
dt_low = None
df_idx = None
if single:
dfs = groupd.tail(1)
print("dfs tail1")
else:
dl = 30
dindex = tdd.get_tdx_Exp_day_to_df('999999',
dl=dl).sort_index(ascending=False)
dt = tdd.get_duration_price_date('999999', df=dindex)
dt = dindex[dindex.index >= dt].index.values
dt_low = dt[-1]
dtlen = len(dt) if len(dt) > 0 else 1
dfs = groupd.tail(dtlen)
# import ipdb;ipdb.set_trace()
df_idx = get_groupby_mean_median_close(dfs)
print("dfs tail:%s dt:%s" % (dtlen, dt))
dfs = get_multi_date_duration(dfs, dt[-1])
# groupd2 = dfs.groupby(level=[0])
# dfs['ma%d'%cumdays] = groupd['close'].apply(pd.rolling_mean, cumdays)
# dfs.reset_index().groupby(['code'])['date'].transform('count')
single = True
dfs = dfs.fillna(0)
idx = pd.IndexSlice
# mask = (dfs[('ma%s')%(rollma[0])] > dfs[('ma%s')%(rollma[1])]) & (dfs[('ma%s')%(rollma[-1])] > 0) & (dfs[('close')] > dfs[('ma%s')%(rollma[0])]) & (dfs[('close')] > dfs[('ma%s')%(rollma[-1])])
# mask = (dfs[('ma%s')%(rollma[0])] > dfs[('ma%s')%(rollma[1])]) & (dfs[('ma%s')%(rollma[-1])] > 0) & (dfs[('close')] > dfs[('ma%s')%(rollma[1])]) & (dfs[('close')] > dfs[('ma%s')%(rollma[-1])])
# mask = (dfs[('ma%s')%(rollma[0])] > dfs[('ma%s')%(rollma[1])]) & (dfs[('ma%s')%(rollma[-1])] > 0) & (dfs[('close')] > dfs[('ma%s')%(rollma[-1])])
# mask = ( (dfs[('ma%s')%(rollma[0])] > 0) & (dfs[('ma%s')%(rollma[-1])] > 0) & (dfs[('close')] > dfs[('ma%s')%(rollma[-1])]) & (dfs[('close')] > dfs[('ma%s')%(rollma[0])]))
# mask = ( (dfs[('ma%s')%(rollma[0])] > 0) & (dfs[('ma%s')%(rollma[-1])] > 0)
# & (dfs[('close')] > dfs[('ma%s')%(rollma[-1])]*ma_250_l)
# & (dfs[('close')] < dfs[('ma%s')%(rollma[-1])]*ma_250_h)
# & (dfs[('close')] > dfs[('ma%s')%(rollma[0])]))
# & (dfs['mean'] > dfs['median'])
# & (dfs['close'] > dfs['mean'])
if len(rollma) > 1:
mask = ((dfs[('ma%s') % (rollma[0])] > 0) & (dfs[('ma%s') %
(rollma[-1])] > 0)
& (dfs[('ma%s') % (rollma[0])] > dfs[('ma%s') % (rollma[-1])])
& (dfs[('close')] > dfs[('ma%s') % (rollma[0])])
& (dfs[('close')] > dfs[('ma%s') % (rollma[-1])] * ma_250_h)
& ((dfs[('close')] > dfs['ene']) |
(dfs[('close')] > dfs['upper'])))
else:
mask = ((dfs[('ma%s') % (rollma[0])] > 0)
& (dfs[('close')] > dfs[('ma%s') % (rollma[0])])
& ((dfs[('close')] > dfs['ene']) |
(dfs[('close')] > dfs['upper'])))
# mask = ((dfs[('close')] > dfs[('ma%s')%(rollma[-1])]))
df = dfs.loc[idx[mask, :]]
df = get_multi_code_count(df)
print(df.couts[:5])
# import ipdb;ipdb.set_trace()
# df.sort_values(by='couts',ascending=0)
# groupd.first()[:2],groupd.last()[:2]
# groupd = df250.groupby(level=[0])
# '''
# groupd.transform(lambda x: x.iloc[-1])
# groupd.last()
# groupd.apply(lambda x: x.close > x.ma250)
# df.shape,df.sort_index(ascending=False)[:5]
# ?groupd.agg
# groupd = df.groupby(level=[0])
# groupd['close'].apply(pd.rolling_mean, 250, min_periods=1)
#ex:# Group df by df.platoon, then apply a rolling mean lambda function to df.casualties
# df.groupby('Platoon')['Casualties'].apply(lambda x:x.rolling(center=False,window=2).mean())
code_uniquelist = df.index.get_level_values('code').unique()
code_select = code_uniquelist[random.randint(0, len(code_uniquelist) - 1)]
if app:
print round(time.time() - time_s, 2), 's', df.index.get_level_values(
'code').unique().shape, code_select, df.loc[code_select][-1:]
if single:
# groupd = df.groupby(level=[0])
if tdx:
# block_path = tdd.get_tdx_dir_blocknew() + '060.blk'
# if cct.get_work_time():
# codew = df[df.date == cct.get_today()].index.tolist()
if dt_low is not None:
groupd2 = df.groupby(level=[0])
df = groupd2.tail(1)
df = df.reset_index().set_index('code')
# import ipdb;ipdb.set_trace()
# df = df[(df.date >= dt_low) & (df.date <= cct.get_today())]
dd = df[(df.date == dt_low)]
df = df[(df.date >= cct.last_tddate(1))]
# import ipdb;ipdb.set_trace()
print("df:%s df_idx:%s" % (len(df), len(df_idx)))
if df_idx is not None and len(df_idx) > 0:
df = df.loc[df_idx.index, :].dropna()
print("Main Down dd :%s MainUP df:%s couts std:%0.1f " %
(len(dd), len(df), df.couts.std()))
# print df.date.mode()[0]
df = df.sort_values(by='couts', ascending=1)
df = df[df.couts > df.couts.std()]
# df = df[(df.date >= df.date.mode()[0]) & (df.date <= cct.get_today())]
codew = df.index.tolist()
if app:
print round(time.time() - time_s, 2), 'groupd2', len(df)
else:
df = df.reset_index().set_index('code')
df = df[(df.date >= cct.last_tddate(days=10))
& (df.date <= cct.get_today())]
codew = df.index.tolist()
top_temp = tdd.get_sina_datadf_cnamedf(codew, df)
top_temp = top_temp[(~top_temp.index.str.contains('688'))
& (~top_temp.name.str.contains('ST'))]
codew = top_temp.index.tolist()
#clean st and 688
if app:
hdf5_wri = cct.cct_raw_input("rewrite code [Y] or append [N]:")
if hdf5_wri == 'y' or hdf5_wri == 'Y':
append_status = False
else:
append_status = True
else:
append_status = False
if len(codew) > 10:
cct.write_to_blocknew(block_path,
codew,
append_status,
doubleFile=False,
keep_last=0)
print "write:%s block_path:%s" % (len(codew), block_path)
else:
print "write error:%s block_path:%s" % (len(codew), block_path)
# df['date'] = df['date'].apply(lambda x:(x.replace('-','')))
# df['date'] = df['date'].astype(int)
# print df.loc[code_select].T,df.shape
MultiIndex = False
else:
MultiIndex = True
h5a.write_hdf_db('all300',
df,
table='roll200',
index=False,
baseCount=500,
append=False,
MultiIndex=MultiIndex)
return df