Files
py_sandbox/misc/histogram.py
Andreas 8f7760b2aa reorganising projects
--HG--
branch : sandbox
2009-10-24 15:00:35 +00:00

98 lines
2.7 KiB
Python

""" walk histogram
"""
import sys
import re
import datetime
mol = re.compile("\S+\s(\S+?)\|.{5}(.{6}).\|(.*)\|(.*)")
def main(argv = None):
if argv == None:
argv = sys.argv
ifn = "e:/livefeed/20070829/GDAXI.csv"
ofn = "e:/eclipse/workspace/filter/var/GDAXI_histogram.csv"
tick_list = [ ]
histogram = { }
print "Converting [%s] to [%s]" % (ifn, ofn)
try:
ifh = open(ifn, "r")
ofh = open(ofn, "w")
except IOError:
print "open [%s] failed" % ifn
return 2
try:
for line in ifh:
tick_list.append(parseCsv(line))
for i in xrange(1, len(tick_list)):
diff = tick_list[i].last - tick_list[i-1].last
key = int( diff * 100 )
if key not in histogram:
histogram[key] = 1
else:
histogram[key] += 1
# for key in sorted(histogram.keys()):
for i in xrange(-1000, 1000):
if i in histogram:
print >> ofh, float(i)/100, ",", histogram[i]
else:
print >> ofh, float(i)/100, ",0"
finally:
ifh.close()
ofh.close()
print "Done."
time_pattern = re.compile("(..):(..):(..).(\d+)")
def parseCsv(line):
flds = line.split(',')
mot = re.match(time_pattern, flds[0])
(hh, mm, ss, ms) = mot.group(1, 2, 3, 4)
time = datetime.time(int(hh), int(mm), int(ss), int(ms)*1000)
# pid = flds[1]
bid = flds[2]
ask = flds[3]
last = flds[4]
return Tick(time, float(bid), float(ask), float(last))
class Tick:
""" VDO helper class """
def __init__(self, ti, bid, ask, last):
self.time = ti
self.bid = bid
self.ask = ask
self.last= last
def __cmp__(self, other):
return cmp(self.time, other.time)
class Roi:
""" VDO helper class """
def __init__(self, buy_time, sell_time, pnl):
self.buy_time = buy_time
self.sell_time = sell_time
self.pnl = pnl
def pnl_per_minute(self):
dt = time_in_minutes(self.sell_time) - time_in_minutes(self.buy_time)
try:
ppm = self.pnl / dt
except ZeroDivisionError:
return self.pnl
return ppm
def __cmp__(self, other):
return cmp(self.pnl_per_minute(), other.pnl_per_minute())
def __repr__(self):
return "%s,%s,%.2f,%.2f" % (self.buy_time, self.sell_time,
self.pnl, self.pnl_per_minute())
def time_in_minutes(ti):
return ti.hour * 60 + ti.minute + float(ti.second) / 60
if __name__ == '__main__':
sys.exit(main(sys.argv))