npikkus=7
hoidla={}
nr=0
for rida in open("sonaliigid_koos.txt"):
  nr+=1
  if nr % 10000 == 0: print(nr)
  r=rida.strip().replace("Z", "")
  ngramid=[r[koht: koht+npikkus] for koht in range(len(r)-npikkus+1)]
  for ngram in ngramid:
    if ngram in hoidla: hoidla[ngram]+=1
    else: hoidla[ngram]=1

votmed=list(hoidla.keys())
votmed.sort()
with open(str(npikkus)+"gram_eest.txt", "w") as f:
 for voti in votmed:
    print(voti, hoidla[voti], sep=",", file=f)

votmed.sort(key=lambda voti: voti[::-1])
with open(str(npikkus)+"gram_tagant.txt", "w") as f:
 for voti in votmed:
    print(voti, hoidla[voti], sep=",", file=f)

#votmed.sort(key=lambda voti: voti[1:4]+voti[0]+voti[4])
#with open(str(npikkus)+"gram_keskelt.txt", "w") as f:
# for voti in votmed:
#    print(voti, hoidla[voti], sep=",", file=f)

votmed.sort(key=lambda voti: voti[2:5]+voti[1]+voti[0]+voti[5]+voti[6])
with open(str(npikkus)+"gram_keskelt.txt", "w") as f:
 for voti in votmed:
    print(voti, hoidla[voti], sep=",", file=f)

kogum=[[voti, hoidla[voti]] for voti in votmed]
kogum.sort(key=lambda rida: -rida[1])
with open(str(npikkus)+"gram_kahanev.txt", "w") as f:
  for rida in kogum:
    print(rida[0], rida[1], sep=",", file=f)

kokku=sum([rida[1] for rida in kogum])
with open(str(npikkus)+"gram_kokku.txt", "w") as f:
   print(kokku, file=f)