python 统计文件中的字符串数目示例

def getWords(filepath):
  file = open(filepath)
  wordOne=[]
  while(file):
    line = file.readline()
    word = line.split('/')
    wordOne.extend(word)
    if(not line):      #若读取结束了
      break 
  wordtwo=[]
  for i in wordOne:
    wordtwo.extend(i.split())
  return wordtwo

def getWordNum(words):
  dictWord={}
  for i in words:
    if(i not in dictWord):
      dictWord[i]=0
    dictWord[i]+=1
  return dictWord

filepath='data/new.txt'
words = getWords(filepath)
dictword = getWordNum(words)
print(dictword)

{'C4D': 9, 'max': 1, 'su': 1, 'maya': 1, 'AE': 3}

print(type(word)) 
print(type(splitData[0])) 

<class 'list'>
<class 'str'>

import os 
print(os.getcwd()) 

filepath='data/new.txt'
file = open(filepath)    #读取文件
wordOne=[]
symbol = '\n/'       #定义分隔符
symbol = "["+symbol+"]"   #拼接正则表达式
while(file):
  line = file.readline()
  word = re.split(symbol,line)
  wordOne.extend(word)
  if(not line):
    break
#通过上式得到的list中会含有很多的空字符串，所以要去空
wordOne = [x for x in wordOne if x]