progress.py

from time import time
import scipy
from scipy import stats
from math import sqrt
from multiprocessing import current_process

class Progress(object): 
    def __init__(self, tot=None, prefix=None,
                template='sp: %(speed).02fi/s, pr: %(progress).02f%%, rt: %(remaining)s, it: %(iteration_time)s, %(cnt)s of %(tot)s, pt: %(process_time)s', 
                 threaded_cnt= None, print_time=0.25, print_pid=False):
        self.print_pid= print_pid
        self.tot= tot
        self.prefix= prefix
        self.print_time= print_time
        self.is_threaded= threaded_cnt is not None
        if threaded_cnt is not None: self.cnt= threaded_cnt
        else: self.cnt= 0

        self.t0= None
        self.last_t= None
        self.template= template
        self.avg_inc_time= 0 
        self.var_inc_time= 0 
        self.remaining_time=0
        self.times= []
        self.started_at= time()
    
    def _format_time(self, t):
        units= 'smhd'
        coefs= [60,60,24]
        t= float(t)
        for i in xrange(1,4):
            if t < 60: break
            t/=coefs[i-1]
        return '%.02f%s' % (t, units[i-1])

    def inc(self, **extras):
        if self.is_threaded: 
            self.cnt.value+=1
            cnt= self.cnt.value
        else: 
            self.cnt+=1
            cnt= self.cnt

        t= time()
        if self.last_t is not None:
            if self.tot is not None: progress= cnt/float(self.tot)*100
            iteration_time= t-self.last_t
            delta= iteration_time-self.avg_inc_time
            self.avg_inc_time+= delta/cnt
            self.var_inc_time= float((cnt-2))/(cnt-1)*self.var_inc_time + \
                               (iteration_time- self.avg_inc_time)*delta/(cnt-1)
            


        self.last_t= t
        if self.t0 is None: self.t0= t
        if t- self.t0 > self.print_time:
            self.t0= time()
            tot= self.tot
            iteration_time= self._format_time(iteration_time)
            process_time= self._format_time(time()-self.started_at)
            if self.tot is not None:
                remaining= self._format_time((tot-cnt)*self.avg_inc_time)
                h= sqrt(self.var_inc_time) * stats.t._ppf(0.9, self.cnt-1)
                h= self._format_time(h*(tot-cnt))
            speed= 1.0/self.avg_inc_time
            d= dict(extras)
            d.update(locals())
            if self.prefix is not None:
                str= '%s%s' % (self.prefix, self.template % d)
            else:
                str= self.template % d
            if self.print_pid:
                str= '%s: %s' % (current_process().pid, str)
            print str