TA的每日心情 | 擦汗
2021-11-17 09:18 |
|---|
签到天数: 79 天 [LV.6]常住居民II
|
发表于 2021-10-6 09:52:48
|
显示全部楼层
1 ~: ?: _/ j( a6 x& N' E7 N
（3） 在哪里加入- current_path = os.path.dirname(__file__)；current_path+"/OALD4_azure.txt
3 E6 E; W7 v, J9 o, m' P
我的版本是 3.9.7 ，目前没有遇到 No such file or directory 报错。 然后是genMDX_ox4.py 文件有部分中文乱码
. Q; ^4 b) N6 a. u
* E0 w" m4 Z) f% N; k5 N$ r2 l, f- # -*- coding: utf-8 -*-
A& x5 J6 h" G) z+ i% A4 ^ - # encoding=utf8: C1 @" Q8 x0 X* ^
! w( h2 E4 o% Q. B0 l" C- from __future__ import unicode_literals,print_function, absolute_import, division6 W J$ y& i3 ~% {4 x4 {" ~
( S, X% P; W d( F- # t: ~( m$ o' j$ d- [2 w% _
- import re, b- P0 L v+ i8 w$ {6 y8 U+ T. a! H9 a
- import copy) i, R$ O0 {- L& w8 u$ S7 K
- import chardet% Y- b7 P) M7 E. y* A
- 4 s8 b0 p/ G8 q% ?, E. D6 F
- import os
( ?( ]" l4 h& @0 O% A2 ?6 S' Z - import io+ X* H) p! K- ^* n
- import sys
l+ Q+ r( j0 C [7 o/ A - # reload(sys). k8 J2 Z0 m9 S( ~# \" T9 U+ B
- # sys.setdefaultencoding('utf-8')3 i3 } c4 E' u0 @2 c; e6 W& o
' H5 r W( ~. Z, N: ]2 c0 i- import collections5 i3 x" U" Q" l2 A
- from collections import defaultdict
2 u! s- _ i: s v+ D - 1 q$ o# i7 R; w/ y W8 B
4 h) G1 Y+ V) e( b- from writemdict import MDictWriter, encrypt_key5 v1 n3 S* R. q' ]
- from ripemd128 import ripemd128" D" V6 a0 a7 f" r" `0 \
% Z0 l8 T8 g8 g7 M& T& y
2 Q$ q2 {( U9 Y" b1 o0 u5 P- head = 0. F Y3 f7 {- K1 h$ j; U1 m
- new_mean =[]/ w) M7 X, g# ]/ o( l6 g* J. |' R
- f=io.open('OALD4_azure.txt', 'r',encoding='utf-8')2 {9 n7 |0 r( U o# u& E) @0 s2 f
- #f=io.open('oxford2_original.txt', 'r',encoding='utf-8')6 c. I) V3 P" S" m8 p+ u3 T p
- d = defaultdict(list) #����һ�����ֵ䣬Ҳ��ʹ��{}������
% a! D' v3 ?1 J( k5 @3 w - for line in f: #ÿ�δ�f�ж���һ��5 |/ E! |! C0 f8 W6 W3 H
- line=line.rstrip('\n')#ȥ����β�Ļ��з�5 H& c! B' r! H; n' n
- if line == '</>':0 R9 T4 F0 S- ~) L8 f
- if head == 2:
. I/ _+ n" s4 x - new_mean[0:] = ["".join(new_mean[0:])]" M) F" H! ^) M- p/ K
- d[word].append(new_mean[0])
2 f) H; u# F! c+ ]( u - head = 1;+ [" N( U9 M) [
- new_mean =[]
' }' D$ f; Q" x" \, a; i7 x3 q - elif head == 1:
% V& Z$ L6 M4 j - word = line6 x1 S2 X+ o- f6 G
- head = 2+ M# u5 J$ c: F8 l2 |' G# u
- elif head == 2:
! q- J D7 q( u; I! `0 z - new_mean.append(line)
2 L+ s' s( R* j6 s% L& K - head = 2
( V8 _$ l; I7 P" w& a - f.close()
* j1 Y& Y, e; l, N- b - ' ?0 G! C, ~( f2 \0 y4 b2 W
- 5 ]" y, T% R0 `! `
- ff=io.open('about_OX4.txt', 'r',encoding='utf-8')#�ʵ�about��Ϣ��txt�ļ��뱣��Ϊutf-8/ d. M8 L3 Z( }6 m, P
- about=[]
- p4 G4 }4 |! _$ B9 z! l - for line in ff: #ÿ�δ�f�ж���һ��# k7 }$ t4 o" Q1 B; b A& ` I
- about.append(line)
, B9 _# O: ?0 l - about[0:] = ["".join(about[0:])]
% O# ^/ A; i9 ~- t
; _+ U( Q, m( I9 H
4 [* }& ?- U; G; l- #outfile = open("example_output/��ţ��Beta_V2.2.1.mdx", "wb")
. ]+ C& y0 _* y; j - #writer = MDictWriter(d, "��ţ��Beta_V2.2.1", about[0])% }& S0 |1 T" g
- outfile = open("output_ox4/OALD4_Ex.mdx", "wb")
@ S* q/ @" ^/ [% m - writer = MDictWriter(d, "ţ��߽�˫��(���İ�)", about[0])$ m) x7 n3 `/ c1 a3 w
- writer.write(outfile)! y0 j- P/ M( ?+ u
- outfile.close()
1 S5 w2 \3 j1 i! E% X: o - 5 Q- J! j1 O. O" a, q$ K3 v
4 J2 c% V. H% E5 `- B& n
8 {* y- y, O- r$ Y) B3 Z" x( R
2 e* F# A3 P' z2 X+ E5 \9 u6 V) I是否可以看看你的文档呢 乱码的部分中文写的是什么？ |
|
楼主: lgmcw
[复制链接]
