#!/usr/bin/env python
# -*- coding: utf-8 -*-

import sys

class Csomopont(object):
    def __init__(self,b):
        self.betu = b
        self.balNulla=None
        self.jobbEgy=None

    def ujNullasGyermek (self,gy):
        self.balNulla = gy

    def ujEgyesGyermek (self,gy):
        self.jobbEgy = gy

class LZWBinFa(object):
    def __init__(self):
        self.gyoker = Csomopont ('/')
        self.fa = self.gyoker
        self.melyseg=0

    def beolvas (self,b):
        if b == '0':
            if self.fa.balNulla == None:
                uj = Csomopont ('0')
                self.fa.ujNullasGyermek (uj)
                self.fa = self.gyoker
            else:
                self.fa = self.fa.balNulla
        else:
            if self.fa.jobbEgy == None:
                uj = Csomopont ('1')
                self.fa.ujEgyesGyermek (uj)
                self.fa = self.gyoker
            else:
                self.fa = self.fa.jobbEgy

    def kiir (self,elem,fajl):
        if (elem != None):
            self.melyseg+=1
            self.kiir (elem.jobbEgy, fajl)
            for i in xrange(self.melyseg):
                fajl.write ("---")
            fajl.write (elem.betu + "(" + str(self.melyseg - 1) + ")" + "\n")
            self.kiir (elem.balNulla, fajl)
            self.melyseg-=1

    def rmelyseg (self,elem):
        if elem != None:
            self.melyseg+=1
            if self.melyseg > self.maxMelyseg:
                self.maxMelyseg = self.melyseg
            self.rmelyseg (elem.jobbEgy)
            self.rmelyseg (elem.balNulla)
            self.melyseg-=1
    def ratlag (self,elem):
        if elem != None:
            self.melyseg+=1
            self.ratlag (elem.jobbEgy)
            self.ratlag (elem.balNulla)
            self.melyseg-=1
            if elem.jobbEgy == None and elem.balNulla == None:
                self.atlagdb+=1
                self.atlagosszeg += self.melyseg

    def rszoras (self,elem):
        if elem != None:
            self.melyseg+=1
            self.rszoras (elem.jobbEgy)
            self.rszoras (elem.balNulla)
            self.melyseg-=1
            if (elem.jobbEgy == None and elem.balNulla == None):
                self.atlagdb+=1
                self.szorasosszeg += ((self.melyseg - self.atlag) * (self.melyseg - self.atlag))

    def getMelyseg (self):
        self.melyseg = self.maxMelyseg = 0
        self.rmelyseg (self.gyoker)
        return self.maxMelyseg - 1

    def getSzoras (self):
        self.atlag = self.getAtlag ()
        self.szorasosszeg = 0.0
        self.melyseg = self.atlagdb = 0
        self.rszoras (self.gyoker)
        if self.atlagdb - 1 > 0:
            self.szoras = pow((self.szorasosszeg / (self.atlagdb - 1)),1/2)
        else:
            self.szoras = pow(self.szorasosszeg,1/2)
        return self.szoras

    def getAtlag (self):
        self.melyseg = self.atlagosszeg = self.atlagdb = 0
        self.ratlag (self.gyoker)
        self.atlag = float(self.atlagosszeg) / self.atlagdb
        return self.atlag

class Humangenom(object):
    def usage (self):
        print "Usage: lzwtree in_file out_file"

    def main (self):
        if len(sys.argv)!=3:
            self.usage ()
            sys.exit(1)
        with open (sys.argv[1],"r") as beFile:
            with open (sys.argv[2],"w") as kiFile:
                self.binFa=LZWBinFa()
                li=[sor for sor in beFile]
                kommentben=False
                bekerul=False
                for sor in li:
                    if bekerul==False:
                        bekerul=True
                        continue
                    for b in sor:
                        if b == '>':
                            kommentben = True
                            continue
                        if b == '\n':
                            kommentben = False
                            continue
                        if kommentben:
                            continue
                        if b == 'N':
                            continue
                        szo = ord(b)
                        lii=[]
                        for i in xrange(8):
                            if int(szo / pow (2, 7 - i))==1:
                                lii.append(1)
                                szo -= pow (2, 7 - i)
                            else:
                                lii.append(0)
                        for i in lii:
                            self.binFa.beolvas (str(i))
                self.binFa.kiir (self.binFa.gyoker, kiFile)
                kiFile.write ("depth = " + str(self.binFa.getMelyseg ()) + "\n")
                kiFile.write ("mean = " + str(self.binFa.getAtlag ()) + "\n")
                kiFile.write ("var = " + str(self.binFa.getSzoras ()) + "\n")

if __name__=="__main__":
    a=Humangenom()
    a.main()