# -*- coding: utf-8 -*-
#Luc Abate IRD URM224 MIVEGEC 05/02/2013
#fixFAS.py info sur l'auteur pour FSAconverter
#Plateforme Seq-Gen Erick Desmarais et Frédérique Cerqueira

from Tkinter import *
from tkFileDialog import *
from re import *
#from string import maketrans
import os

######fonctions de base
def selectfolder():
        global foldername
        foldername = askdirectory(parent=root, title='Please select a directory')
        if foldername:
                print "Folder selected : "
                print foldername
                return foldername

def selectfile():
        global filename
        filename = askopenfilename(parent=root, title='Please select a file')
        if filename:
                print "File selected : "
                print filename
                return filename

def saveas():
        global filenameout
        myFormats=[
                ('Text','*.txt')
                ]
        filenameout = asksaveasfilename(parent=root,filetypes=myFormats,title="Save output as...")
        if len(filenameout) > 0:
                print "Now saving as %s" % (filenameout)
                if search('.txt$',filenameout):
                        return filenameout
                else:
                        filenameout=filenameout+".txt"
                        return filenameout
        
######Pour menu sous fenetre

def help1():
   filewin = Toplevel(root)
   filewin.title('Help')
   tex = Label(filewin, text='******FSAconverter utility :******\nConvert .fsa files (from 3500 sequencer)\nfor compatibillty to genemapper previous to version 5\nNew files will be created in FSAconverter folder\nwithin the folder selected\n\n**********G2G utility :**********\nTo formate a genemapper exported genotype table\nin a genepop file\n\n**********G2S utility :**********\nTo formate a genepop file to a structure file\n')
   tex.pack()
   button = Button(filewin, text="Close window", command=filewin.destroy)
   button.pack()

def about():
   filewin = Toplevel(root)
   filewin.title('About')
   tex = Label(filewin, text='pyFSAconverter is a python program\nderived from fixFSA script\n\nMore info at :\nSeqGen platform - UMII\nBat 24 1°étage\ncc 63 - Pl. E Bataillon\n34095 Montpellier Cedex 5')
   tex.pack()
   button = Button(filewin, text="Close window", command=filewin.destroy)
   button.pack()

######pour FSAconverter
def findAddress(mystring):
	"""convertit les octets lus en decimal"""
	result = 0
	n=len(mystring)
	taille=range(n)
	taille.reverse()
	for i in taille:
		result=result+ord(mystring[i])*(16**(2*n-2*i-2))
	return result

######

def FSAconverter():
        print "                ************************************************"
        print "                                 FSAconverter"
        print "                              FSAconverter v.1 (2013-02)"
        print "                ************************************************"
        print "                       Beta version, l.Abate, MIVEGEC, IRD"
        print "                   based on fixFSA.py script provided by CeMEB"
        print "                ************************************************"

        selectfolder()
        if foldername:
                liste=os.listdir(foldername)
                print '%i file(s) in selected folder' % (len(liste))
                tbl_fsa=[]
                compteurfsa=0
                for s in liste:
                        if search('\.fsa', s):
                                tbl_fsa.append(s)
                                compteurfsa=compteurfsa+1

                if compteurfsa==0:
                        print "No fsa file found!"
                else:
                        if compteurfsa==1:
                                print "in which ", compteurfsa, "is a .fsa files!"
                        else:
                                print "in which ", compteurfsa, "are .fsa files!"
                        try:
                                os.mkdir(foldername+"\FSAconverted")
                        except OSError:
                                pass
                        i=0
                        while i<len(tbl_fsa):
                                fout=open(foldername+"\FSAconverted\patched_"+tbl_fsa[i], 'w+b')
                                f=open(foldername+"\\"+tbl_fsa[i],'r+b')
                                rawdata=str(f.read())
                                fout.write(rawdata)
                                myaddress=findAddress(rawdata[26:30])
                                while rawdata[myaddress:myaddress+4] != 'HCFG':
                                        myaddress=myaddress+28
                                fout.seek(findAddress(rawdata[myaddress+76:myaddress+80]))
                                fout.write('31')
                                print tbl_fsa[i], "patched!"
                                f.close()
                                fout.close()
                                i=i+1
                print "                ************************************************"
                print "                                    job fini"
                #print "                ************************************************" 

######Pour controle input format et fonctions elementaires

def gmtbl(entree):
        try:
                global testgmtbl
                testgmtbl=0
                f=open(entree,'r')

                for ligne in f :
                        if search('Sample Name.*Marker.*Allele 1.*Allele 2', ligne):
                                print "File recognized to be a genemapper table"
                                testgmtbl=1
                                break
                
                if (testgmtbl==0):
                        print "File not reconized to be a genemapper table"
                elif (testgmtbl==1):
                        return testgmtbl 

                f.close()
        except:
                print "erreur except gmtbl"

######

def gptbl (entree):
        try:
                global testgptbl
                testgptbl=0
                f=open(entree,'r')

                for ligne in f:
                        if search(' \, \d{6}',ligne):
                                print "File recognized to be a genepop format file"
                                testgptbl=1
                                break
                if (testgptbl==0):
                        print "File not reconized to be a genepop format file"
                elif (testgptbl==1):
                        return testgptbl 

                f.close()
        except:
                print "erreur except gptbl"

######G2G

def g2g():
        print "                ************************************************"
        print "                           GeneMapper 2 Genepop format"
        print "                            G2G formater v.2 (2012-05)"
        print "                ************************************************"
        print "                Le fichier entree doit etre une table genemapper"
        print "                 exportee (onglet genotype) en texte tabulation"
        print "                                  Attention"
        print "                   une nommination identique des echantillons"
        print "                         de chaque specimen est requise"
        print "                  A noter qu une seule population sera notifiee"
        print "                         pour l'ensemble des specimens"
        print "                ************************************************"
        print "                       Beta version, l.Abate, MIVEGEC, IRD"
        print "                ************************************************"
        selectfile()
        if filename:
                gmtbl(filename)
                if (testgmtbl==1):
                        saveas()
                        if filenameout:
                                fout=open(filenameout,'w')
                        else:
                                print 'error in saving as !'

                        f=open(filename,'r')

                        compteur=0
                        table1=[]

                        for ligne in f:
                                if ligne!='':
                                        compteur=compteur+1
                                        ligne=sub('\n$','',ligne)
                                        
                                        liste=ligne.split('\t')
                                
                                        if compteur==1:
                                                i=0
                                                while i<=(len(liste)-1):
                                                        #
                                                        if search('Sample Name',liste[i]):
                                                                nb1=i
                                                                print "Indice Sample Name : \t",
                                                                print nb1
                                                        elif search('Marker',liste[i]):
                                                                nb2=i
                                                                print "Indice Marker : \t",
                                                                print nb2
                                                        elif search('Allele 1',liste[i]):
                                                                nb3=i
                                                                print "Indice Allele 1 : \t",
                                                                print nb3
                                                        elif search('Allele 2',liste[i]):
                                                                nb4=i
                                                                print "Indice Allele 2 : \t",
                                                                print nb4
                                                        i=i+1

                                        else:                
                                                try:
                                                        #si vide
                                                        if liste[nb3]=='':
                                                                liste[nb3]="000"
                                                        if liste[nb4]=='':
                                                                liste[nb4]=liste[nb3]
                                                        #si 2 chiffres mettre a 3 chiffre
                                                        if search('^\d{2}$',liste[nb3]):
                                                                liste[nb3]= "0"+liste[nb3]
                                                        if search('^\d{2}$',liste[nb4]):
                                                                liste[nb4]= "0"+liste[nb4]
                                                        if search('^\?$',liste[nb3]):
                                                                liste[nb3]= "???"
                                                        if search('^\?$',liste[nb4]):
                                                                liste[nb4]= "???"
                                                except :
                                                        continue


                                                string=liste[nb1]+"\t"+liste[nb2]+"\t"+liste[nb3]+"\t"+liste[nb4]
                                                table1.append(string)                                                

                        print "\t%d lignes enregistrees" % (compteur-1)
                        f.close()


                        ###
                        samplename=[]

                        booli=0
                        for s in table1:
                                liste=s.split('\t')
                                #rappel liste[0]=samplename liste[1]=marker liste[2]=allele1 liste[3]=allele2
                                if booli==0:
                                        samplename.append(liste[0])
                                        booli=1
                                else:
                                        boollock=0
                                        for lockname in samplename:
                                                if liste[0]==lockname:
                                                        boollock=boollock+1
                                        if boollock==0:
                                                samplename.append(liste[0])
                                
                        samplename.sort()

                        ###
                        marker=[]

                        booli=0
                        for s in table1:
                                liste=s.split('\t')
                                #rappel liste[0]=samplename liste[1]=marker liste[2]=allele1 liste[3]=allele2
                                if booli==0:
                                        marker.append(liste[1])
                                        booli=1
                                else:
                                        boollock=0
                                        for lockmarker in marker:
                                                if liste[1]==lockmarker:
                                                        boollock=boollock+1
                                        if boollock==0:
                                                marker.append(liste[1])


                        marker.sort()

                        ###
                        fout.write("Genepop Format "+filenameout+"\n")
                        print "genepop format"
                        for s in marker:
                                print "%s" % (s)
                                fout.write(s+"\n")
                        fout.write("pop\n")
                        
                        for s1 in samplename:
                                fout.write("%s ," %(s1))
                                print "%s ," %(s1),
                                for s2 in marker:
                                        for s3 in table1:
                                                liste=s3.split('\t')
                                                if liste[0]==s1 and liste[1]==s2:
                                                        fout.write(" "+liste[2]+liste[3])
                                                        print " "+liste[2]+liste[3],
                                fout.write("\n")
                                print ''

                        fout.close()
                        print "                ************************************************"
                        print "                                    job fini"
                        #print "                ************************************************"
                        
######G2S

def g2s():
        print "                ************************************************"
        print "                           Genepop 2 Structure format"
        print "                           G2S formater v.2 (2012-05)"
        print "                ************************************************"
        print "                                    Attention"
        print "                  le fichier 'entree' (genepop) doit avoir une"
        print "                 notation a 6 chiffres par marqueur (3X2alleles)"
        print "                ************************************************"
        print "                exemple :"
        print "                  titre fichier genepop"
        print "                  marqueur1"
        print "                  marqueur2"
        print "                  pop"
        print "                  specimen1 099103 155155"
        print "                  specimen2 099099 143155"
        print "                  pop"
        print "                  specimen3 099106 175183"
        print "                  ..."
        print "                ************************************************"
        print "                       Beta version, l.Abate, MIVEGEC, IRD"
        print "                ************************************************"
        selectfile()
        if filename:
                gptbl(filename)
                if (testgptbl==1):
                        saveas()
                        if filenameout:
                                fout=open(filenameout,'w')
                        else:
                                print 'error in saving as !'

                        f=open(filename,'r')

                        compteur=0
                        nbpop=0
                        nbmarker=0
                        marker=[]
                        boolean=0
                        
                        for ligne in f:
                                if ligne!='':
                                        compteur=compteur+1
                                        if search(".+ , \d{6}", ligne):
                                                #pas top
                                                boolean=boolean+1
                                                if boolean==1:
                                                        j=0
                                                        while j<len(marker):
                                                                if j!=len(marker)-1:
                                                                        fout.write(marker[j]+"\t")
                                                                else:
                                                                        fout.write(marker[j]+"\n")
                                                                j=j+1
                                                        print ''
                                                nbpopstr=str(nbpop)

                                                liste=ligne.split(" ")
                                                #indice 0 = sample name
                                                #indice 1 = ,
                                                #indice 2 à nbmarker+1 = alleles 3x2

                                                print liste[0],
                                                print nbpopstr,
                                                fout.write(liste[0]+"\t"+nbpopstr+"\t")
                                                
                                                i=2#commence alleles
                                                while i<=len(liste)-1:
                                                        s=sub("\?", "0", liste[i])
                                                        s=sub("(^\d{3})", "\\1\t", s)
                                                        s=sub("000", "-9", s)

                                                        print s,
                                                        if i!=len(liste)-1:
                                                                fout.write(s+"\t")
                                                        else:
                                                                fout.write(s)#deja \n
                                                        i=i+1
                                                        
                                        elif search("^pop" or "^Pop" or "^POP", ligne):
                                                nbpop=nbpop+1
                                        else:
                                                if nbpop==0 and compteur!=1:
                                                        s=sub("\n$", "", ligne)
                                                        marker.append(s)
                                                        nbmarker=nbmarker+1
                                                        print s,


                        f.close()
                        fout.close()
                        print "                ************************************************"
                        print "                                    job fini"
                        #print "                ************************************************"

######

root = Tk()
root.title('pyFSAconverter (beta version)')

menubar = Menu(root)

filemenu = Menu(menubar, tearoff=0)
filemenu.add_command(label="Exit", command=root.quit)
menubar.add_cascade(label="File", menu=filemenu)

utilitiesmenu = Menu(menubar, tearoff=0)
utilitiesmenu.add_command(label="FSAconverter", command=FSAconverter)
utilitiesmenu.add_separator()
utilitiesmenu.add_command(label="G2G", command=g2g)
utilitiesmenu.add_command(label="G2S", command=g2s)
menubar.add_cascade(label="Utilities", menu=utilitiesmenu)

helpmenu = Menu(menubar, tearoff=0)
helpmenu.add_command(label="Help", command=help1)
helpmenu.add_command(label="About...", command=about)
menubar.add_cascade(label="Help", menu=helpmenu)

root.config(menu=menubar)
root.mainloop()
root.destroy()