Source code for Muscat.Helpers.Profiler

# -*- coding: utf-8 -*-
#
# This file is subject to the terms and conditions defined in
# file 'LICENSE.txt', which is part of this source code package.
#
import cProfile
import pstats
import io
from collections import OrderedDict

import numpy as np

from Muscat.Helpers.Interpolation import BinarySearch




class Profiler():
    """Class to help profiling code, this is a very simple approach. it uses
    cProfile and pstats to do all the work.

        myProfiler = Profiler(0.2)

        myProfiler.Start()
        #do some work
        time.sleep(0.02)
        print("Hello")
        a = 3 + 4

        myProfiler.Stop()
        print(myProfiler)

    """

    def __init__(self, discardedPortion: float = 0.2, timerSignificantDigits: int = 3):
        """_summary_

        Parameters
        ----------
        discardedPortion : float, optional
            discarded portion on the output , by default 0.2 (20%)
        timerSignificantDigits : int, optional
            Number of digits to use for the times printed, by default 3
        """
        self.pr = cProfile.Profile()
        self.s = io.StringIO()
        self.ps = None
        self.discardedPortion = discardedPortion
        self.timerSignificantDigits = timerSignificantDigits
        self.totTimes = None
        self.cumTimes = None
        self.sumTotTime = None
        self.sumCumTime = None



    def Start(self) -> None:
        """Start the profiling
        """
        self.pr.enable()




    def Stop(self) -> None:
        """Stop the profiling
        """

        self.pr.disable()
        self.ps = pstats.Stats(self.pr, stream=self.s).sort_stats('cumulative')
        self.ps.reverse_order()
        self.ps.print_stats()

        lines = self.s.getvalue().split('\n')[5:-3]
        functionNames = [l[46:][-60:] for l in lines]
        parsing = np.array([l.split()[:5] for l in lines])

        totalTimes = np.array(parsing[:, 1], dtype=float)
        cumulatedTimes = np.array(parsing[:, 3], dtype=float)
        self.sumTotTime = np.sum(totalTimes)
        self.sumCumTime = np.sum(cumulatedTimes)

        totalTimes = totalTimes / np.sum(totalTimes)
        cumulatedTimes = cumulatedTimes / np.sum(cumulatedTimes)

        totalTimesArgSort = np.argsort(totalTimes)
        cumulatedTimesArgSort = np.argsort(cumulatedTimes)

        cumulatedSumTotalTimes = np.cumsum(totalTimes[totalTimesArgSort])
        cumulatedSumCumulatedTimes = np.cumsum(cumulatedTimes[cumulatedTimesArgSort])

        totalTimesArgSortInv = totalTimesArgSort[BinarySearch(cumulatedSumTotalTimes, self.discardedPortion)+1:][::-1]
        cumTimesArgSortInv = cumulatedTimesArgSort[BinarySearch(cumulatedSumCumulatedTimes, self.discardedPortion)+1:][::-1]

        totalTimes = totalTimes[totalTimesArgSortInv]
        totalTimes = np.hstack((totalTimes, 1 - np.sum(totalTimes)))

        cumulatedTimes = cumulatedTimes[cumTimesArgSortInv]
        cumulatedTimes = np.hstack((cumulatedTimes, 1 - np.sum(cumulatedTimes)))

        functionNamesTotalTime = [functionNames[i] for i in totalTimesArgSortInv] + ["REMAINING"]
        functionNamesCumTime = [functionNames[i] for i in cumTimesArgSortInv] + ["REMAINING"]

        self.totTimes = OrderedDict(zip(totalTimes, functionNamesTotalTime))
        self.cumTimes = OrderedDict(zip(cumulatedTimes, functionNamesCumTime))


    def __str__(self) -> str:
        """Return a string with  all the collected information

        Returns
        -------
        str
            Teh description
        """

        from Muscat.Helpers.TextFormatHelper import TFormat
        string = TFormat.InBlue("Profiler")+" discarding "
        string += str(int(100.*self.discardedPortion))+"% of the smallest functions\n"
        string += TFormat.InRed("Total times:\n")

        for part, name in self.totTimes.items():
            string += TFormat.InGreen(TFormat.Center(name, width=70))+": "+str(round(100.*part, 2))
            string += "% ("+str(round(self.sumTotTime*part, self.timerSignificantDigits))+"s)\n"
        string += TFormat.InRed("Cumulated  times:\n")
        for part, name in self.cumTimes.items():
            string += TFormat.InGreen(TFormat.Center(name, width=70))+": "+str(round(100.*part, 2))
            string += "% ("+str(round(self.sumCumTime*part, self.timerSignificantDigits))+"s)\n"
        return string





def CheckIntegrity(GUI:bool=False):

    import time

    p = Profiler(0.2)

    p.Start()
    time.sleep(0.02)
    print("Hello")
    a = 3 + 4
    p.Stop()

    print(p)

    return "ok"



if __name__ == '__main__':
    print(CheckIntegrity(GUI=True)) #pragma: no cover