marksonad={}
tekstiaastad={}
m=open("jaotus1000_peateemaga.txt", encoding="utf-8").readlines()
kogus=len(m)//4
for nr in range(kogus):
  ryhmanimi=m[nr*4+2].split()[0]+"-"+"-".join(m[nr*4+1].split()[:2])
  rm=m[nr*4+1].strip().split()
  for marksona in rm:
    marksonad[marksona]=ryhmanimi
  
for rida in open("tekstiaastad.txt", encoding="utf-8"):
  m=rida.strip().split()
  tekstiaastad[m[0]]=m[1]


fnimeloend="Eri_Klas_files.txt"
fkataloog="C:/jaagup/22/bfmkeel/tekstid_syntaks/"
fnimed=[fnimi.strip() for fnimi in open(fnimeloend, encoding="utf-8").readlines()]
seosed=[]
for fnimi in fnimed:
  print(fnimi)

  fread=open(fkataloog+"s"+fnimi, encoding="utf-8").readlines()
  for rida in fread:
    lemma=rida.split(";")[3].replace("_", "").replace("=", "")
    if lemma in marksonad:
      marksona=marksonad[lemma].replace('"', "")
      seosed.append([fnimi.split(".")[0], marksona])
f2=open("seosed5.csv", "w", encoding="utf-8")
print("Source,Target", file=f2)
print("\n".join([",".join(rida) for rida in seosed]), file=f2)
f2.close()
f2=open("failid5.csv", "w")
print("Id\tLabel\ttimeset", file=f2)
for fnimi in fnimed:
   fnimi=fnimi.split(".")[0]
   aasta=tekstiaastad[fnimi]
   print(fnimi, fnimi, f"<[{aasta},{aasta}]>",  sep="\t", file=f2)
f2.close()