python文件處理（對比和篩選）

#!/user/bin/python
#!coding=utf-8
# -*- coding: utf-8 -*-
# 2017-9-25
#author:jingwenshuai
import sys
import difflib
import re
import os


#--------------------------------比對兩文件，將結果存入Result.html-------------------------------------------#

# 讀取配置文件函數
def read_file(file_name):
    try:
        file_handle = open(file_name, 'r')
        text = file_handle.read().splitlines()         # 讀取后以行進行分割
        file_handle.close()
        return text
    except IOError as error:
        print 'Read file Error: {0}'.format(error)
        sys.exit()


# 比較兩個文件並輸出html格式的結果
def compare_file(file1_name, file2_name):
    if file1_name == "" or file2_name == "":
        print '文件路徑不能為空：file1_name的路徑為：{0}, file2_name的路徑為：{1} .'.format(file1_name, file2_name)
        sys.exit()
    text1_lines = read_file(file1_name)
    text2_lines = read_file(file2_name)
    diff = difflib.HtmlDiff()                             # 創建htmldiff 對象
    result = diff.make_file(text1_lines,text2_lines)     # 通過make_file 方法輸出 html 格式的對比結果
    #  將結果保存到result.html文件中並打開
    try:
        with open('result.html', 'w') as result_file:     #同 f = open('result.html', 'w') 打開或創建一個result.html文件
            result_file.write(result)                     #同 f.write(result)
    except IOError as error:
        print '寫入html文件錯誤：{0}'.format(error)


#------------------------------------取出不同部分存入Result.txt-------------------------------------------#

#取出不同部分存入Result.txt
def result(file1_name,file2_name):
    if file1_name == "" or file2_name == "":
        print '文件路徑不能為空：file1_name的路徑為：{0}, file2_name的路徑為：{1} .'.format(file1_name, file2_name)
        sys.exit()
    str1=[]
    str2=[]
    str_dump=[]
    #將A.txt的內容逐行讀到str1中
    with open(file1_name,'r') as fa:                #相當於fa=open(file1_name,'r')
        for line in fa.readlines():
            str1.append(line.replace("\n",''))      #line.replace("\n",'') 去掉換行符\n
    #將B.txt中的內容逐行讀到str2中
    with open(file2_name,'r') as fb:
        for line in fb.readlines():
            str2.append(line.replace("\n",''))
     
    #將兩個文件中重復的行，添加到str_dump中
    for i in str1:
        if i in str2:
            str_dump.append(i)
     
    #將兩個文件的行合並，並去重
    str_all=set(str1+str2)
     
    #將重復的行，在去重的合並行中，remove掉，剩下的就是不重復的行了
    for i in str_dump:
        if i in str_all:
            str_all.remove(i)
    #寫入文件中
    with open("Result.txt",'w+') as fc:
        for i in list(str_all):
            fc.write(i+'\n')
    fa.close()
    fb.close()
    fc.close()


if __name__ == "__main__":
    x = raw_input(u"請輸入第一個文件路徑：")
    y = raw_input(u"請輸入第二個文件路徑：")
    compare_file(x, y)    #傳入兩文件的路徑
    result(x,y)    #傳入兩文件的路徑