In [1]:

    

import os
import re
import pandas as pd
from collections import defaultdict


    

In [2]:

    

'''
用于过滤异常编码字符的类
'''

class char_filter():
    def __init__(self):
        self.filter_pattern = re.compile(r"\\x[a-z0-9]{2}|\W|^b'|\{[0-9]+\}")
    def filter(self, text):
        foo = self.filter_pattern.sub(' ', text)
        return foo


    

In [3]:

    

'''
根据输入路径读取txt文件，返回一个列表
'''

def load_txt(file_path):
    article = []
    illegal_filter = char_filter()
    with open(file_path, 'rb') as f:
        for line in f:
            foo = str(line)
            article.append(illegal_filter.filter(foo))
    return article


    

In [4]:

    

'''
定义输入文件路径并读取文件
'''

filePath = '/home/da/task.txt'
article = load_txt(filePath)
article


    

    
Out[4]:





[' The windows in the Wradisley drawing rooms were large  one of them  a vast  shallow 10  bow  which seemed to admit the outside into the interior  rather than to enlighten the interior with the view of what was outside  Mrs  Wradisley sat within reach of  but not too near  a large  very red fire   a fire which was like the turf outside  the growth of generations  or at least had not at all the air of having been lighted to day or any recent day  It did not flame  but glowed steadily  adding something to the color of the room  but not much to the light  Later in the season  when larger parties assembled  there was tea in the hall for the sportsmen and the ladies who waited for them  but Mrs  Wradisley thought the hall draughty  and much preferred the drawing room  which was over furnished after the present mode of drawing rooms  but at least warm  and free from draughts  She was working   knitting with white pins  or else making mysterious chains and bridges in white wool with a crochet hook  her eyes being supposed to be not very strong  and this kind of industry the best adapted for them  As to what Lucy was doing  that defies description  She was doing everything 11  and nothing  She had something of a modern young lady   s contempt for every kind of needlework  and  then  along with that  a great admiration for it as something still more superior than the superiority of idleness  A needle is one of the things that has this double effect  It is the scorn of a great number of highly advanced  very cultured and superior feminine people  but yet here and there will arise one  still more advanced and cultured  who loves the old fashioned weapon  and speaks of it as a sacred implement of life  Lucy followed first one opinion and then another  She had half a dozen pieces of work about  begun under the influence of one class of her friends  abandoned under that of another  She had a little studio  too  where she painted and carved  and executed various of the humbler decorative arts  which  perhaps  to tell the truth  she enjoyed more than art proper  but these details of the young lady   s life may be left to show themselves where there is no need of such vanities  Lucy was  at all events  whatever her other qualities might be  a most enthusiastic friend  12  ']

In [5]:

    

'''
按空格把句子分割为单词
'''

regex = re.compile('\s+')
words = regex.split(article[0])


    

In [6]:

    

'''
对单词进行计数
'''

counts = defaultdict(lambda: 0)  # 使用lambda来定义简单的函数

for word in words:
    counts[word] += 1


    

In [7]:

    

'''
生成结果数据框
'''

result = pd.DataFrame({'word':list(counts.keys()), 'value':list(counts.values())})
result.sort_values('value', ascending=False)


    

    
Out[7]:






  

    

      

      
value
      
word
    
  
  

    

      
190
      
27
      
the
    
    

      
162
      
23
      
of
    
    

      
55
      
14
      
and
    
    

      
67
      
11
      
a
    
    

      
31
      
9
      
to
    
    

      
214
      
8
      
was
    
    

      
6
      
7
      
but
    
    

      
34
      
5
      
not
    
    

      
24
      
5
      
for
    
    

      
83
      
5
      
one
    
    

      
43
      
5
      
She
    
    

      
220
      
4
      
at
    
    

      
47
      
4
      
which
    
    

      
221
      
4
      
that
    
    

      
84
      
4
      
had
    
    

      
58
      
4
      
with
    
    

      
70
      
4
      
in
    
    

      
46
      
3
      
or
    
    

      
157
      
3
      
something
    
    

      
2
      
3
      
drawing
    
    

      
167
      
3
      
than
    
    

      
108
      
3
      
is
    
    

      
222
      
3
      
outside
    
    

      
21
      
3
      
there
    
    

      
120
      
3
      
very
    
    

      
92
      
3
      
Lucy
    
    

      
117
      
3
      
more
    
    

      
54
      
3
      
be
    
    

      
197
      
3
      
them
    
    

      
199
      
3
      
her
    
    

      
...
      
...
      
...
    
    

      
38
      
1
      
vast
    
    

      
134
      
1
      
opinion
    
    

      
37
      
1
      
tea
    
    

      
136
      
1
      
loves
    
    

      
137
      
1
      
wool
    
    

      
138
      
1
      
class
    
    

      
139
      
1
      
highly
    
    

      
140
      
1
      
will
    
    

      
141
      
1
      
no
    
    

      
124
      
1
      
vanities
    
    

      
123
      
1
      
number
    
    

      
41
      
1
      
perhaps
    
    

      
1
      
1
      
friends
    
    

      
105
      
1
      
qualities
    
    

      
106
      
1
      
weapon
    
    

      
107
      
1
      
old
    
    

      
48
      
1
      
themselves
    
    

      
109
      
1
      
studio
    
    

      
110
      
1
      
warm
    
    

      
111
      
1
      
everything
    
    

      
113
      
1
      
has
    
    

      
121
      
1
      
arts
    
    

      
114
      
1
      
strong
    
    

      
115
      
1
      
need
    
    

      
45
      
1
      
waited
    
    

      
44
      
1
      
modern
    
    

      
118
      
1
      
influence
    
    

      
119
      
1
      
feminine
    
    

      
42
      
1
      
sacred
    
    

      
224
      
1
      
were
    
  

225 rows × 2 columns