Участник:Maksim-e/zperekod ngc.py

Материал из Википедии — свободной энциклопедии

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

__version__='$Id:'

import wikipedia
import re, sys
import codecs
from zbib_tekst import *

def vivod(b):
    wikipedia.output(b)
#    otch.write(b)
#    otch.flush()

def zapis(d):
    if len(d)==0:
        return
    b=u''
    for i in range(len(d)):
        b+=d[i]
        if i<len(d)-1:
            b+=u';'

    frez.write(b+u'\n')
    frez.flush()

def main():
    n=1
    dpr=[]
    for s in f.readlines():
        if ord(s[0]) == 65279:
            s=s[1:]
        s=s.replace(u'\r',u'')
        if s[len(s)-1]==u'\n':
            s=s[:len(s)-1]
        if n<3 or len(s)<4:
            n+=1
            continue

        d=[]
        td=[
            0,5,
            5,6, # C
            7,8, # D
            9,10, # X
            10,12, # S
            13,14, # P
            15,18, # CON
            20,22, # Rect
            23,25, # Rect
            26,30, # Rect
            31,34, # Decl
            35,37, # Decl
            38,40, # Decl
            43,47, # Bmag
            50,54, # Vmag
            56,60, # SB
            61,67, # a
            68,73, # b
            74,77, # PA
            78,87, # Type
            87,94, # PGC
            96,111, # ID1
            112,127, # ID2
            128,143, # ID3
            144,200] # Remarks

        i=0
        while i+i<len(td):
            t=s[td[i+i]:td[i+i+1]]
            if i==16 and t[len(t)-1]==u'0':
                t=t[:len(t)-1]
            d.append(ubr_nk_prob(t))
            if i+i+2<len(td) and ubr_nk_prob(s[td[i+i+1]:td[i+i+2]])!=u'':
                vivod(u'!!! %d %d'%(n,i))
            i+=1

        if d[3]!=u'' and d[3]!=u'1':
            dpr[21]+=u', '+d[21]
            dpr[22]+=u', '+d[22]
            dpr[23]+=u', '+d[23]
            dpr[24]+=u', '+d[24]
        else:
            zapis(dpr)
            dpr=d
        n+=1
    zapis(dpr)

#fi = wikipedia.argHandler(sys.argv[1], 'cht_cat')
#filename = fi+'.txt'
#fotch = fi+'_ot.txt'
#fotch2 = fi+'_op.txt'
#firez = fi+'_sk.txt'

filename = wikipedia.argHandler(sys.argv[1], 'cht_cat')
firez = wikipedia.argHandler(sys.argv[2], 'cht_cat')
#fotch2 = wikipedia.argHandler(sys.argv[3], 'cht_cat')

fliwi=1
if len(sys.argv)>=3 and sys.argv[2]==u'w':
    fliwi=0

f=codecs.open(filename,'rb',encoding='utf-8')
#otch = codecs.open(fotch, 'w', 'utf-8')
#otch2 = codecs.open(fotch2, 'w', 'utf-8')
frez = codecs.open(firez, 'w', 'utf-8')
#mysite = wikipedia.getSite()

try:
    main()
except:
    wikipedia.stopme()
    raise
else:
    wikipedia.stopme()