Участник:Maksim-e/zperev tit.py

Материал из Википедии — свободной энциклопедии

# -*- coding: UTF-8 -*-

__version__='$Id:'

import wikipedia
import re, sys
import codecs
import zperevod as perevod
from zbib_tekst import *

def main(slov,nssvoj,vhjaz,vihjaz,fvh,fvih,fipoln,finesl,fispit,flvsjo):

    otch.write(u'len(nssvoj)=%d\n'%len(nssvoj))
    otch.flush()

    kolotch=1500
    f0=codecs.open(fvh,'rb',encoding='utf-8')
    f1=codecs.open(fvih, 'w', 'utf-8')
    fpoln=codecs.open(fipoln, 'w', 'utf-8')
    fnesl=codecs.open(finesl, 'w', 'utf-8')
    fspit=codecs.open(fispit, 'w', 'utf-8')
    osnp={}
    osvap={}
    for s in f0.readlines():
#        wikipedia.output(u'%d\n' % ord(s[0]))
        if ord(s[0]) == 65279:
            s=s[1:]
#        wikipedia.output(u'%d %d  %s   %s\n' % (i,j,n,s))
        s=s.replace(u'\r',u'').replace(u'_',u' ')
#        if s[len(s)-1]!=u'\n':
#            s=s+u'\n'

        s=ubr_nk_prob(s)

#        (rez,fl_perev)=perevod(slov,vhjaz,vihjaz,s)
        (rez,stat)=perevod.perevod_stat(slov,vhjaz,vihjaz,s)


        if flvsjo or (stat.nnp==0 and stat.nvap==0):
            f1.write(s+u'|'+rez+u'\n')
            f1.flush()
            if nssvoj.has_key(rez):
                otch.write(rez+u'\n')
                otch.write(u' <- uzxe\n')
                otch.flush()
            else:
                fspit.write(s+u'\n')
                fspit.flush()

        fpoln.write(rez+u'\n')
        fpoln.flush()

        perevod.perev_uch_nezsl(osnp,osvap,stat)

    perevod.perev_pech_nezsl(slov,vhjaz,vihjaz,osnp,osvap,fnesl,kolotch)


if __name__ == "__main__":
    vhjaz = wikipedia.argHandler(sys.argv[1], 'perevod')
    vihjaz = wikipedia.argHandler(sys.argv[2], 'perevod')
    fslov = wikipedia.argHandler(sys.argv[3], 'perevod')
    fvh = wikipedia.argHandler(sys.argv[4], 'perevod')
    fi = wikipedia.argHandler(sys.argv[5], 'perevod')
    if fi.endswith(u'.txt'):
        fi=fi[:len(fi)-4]
    fvih= fi+u'.txt'
    fotch = fi+u'-ot.txt'
    fipoln = fi+u'-pp.txt'
    finesl = fi+u'-ns.txt'
    fispit = fi+u'-st.txt'

    flvsjo=0
    if len(sys.argv)>=7 and sys.argv[6]==u'2':
        flvsjo=1

    otch = codecs.open(fotch, 'w', 'utf-8')
    #otch2 = codecs.open(fotch2, 'w', 'utf-8')
    slov={}
    nssvoj={}
    slovdop={}
    perevod.zagruzslov(slov,slovdop,nssvoj,vhjaz,vihjaz,fslov)
    main(slov,nssvoj,vhjaz,vihjaz,fvh,fvih,fipoln,finesl,fispit,flvsjo)