python自动写小说的原理_python写的小说抓取源码【原创】

遇到解码问题，搜索后找到解决方法。源码经过修改如下：

import re

from urllib.request import urlopen

myurl = 'http://www.purepen.com/hlm/'

myhost = myurl

f = open('d:/mytemp/红楼梦.txt','w+',encoding='gb18030')

smsg = urlopen(myurl).read()

#选择解码字符集

if re.search(b'[a-zA-Z0-9\-]*',smsg[smsg.find(b'charset=')+8:]).group() == b'GB2312' \

or re.search(b'[a-zA-Z0-9\-]*',smsg[smsg.find(b'charset=')+8:]).group() == b'gb2312':

charset = 'gb18030'

if re.search(b'[a-zA-Z0-9\-]*',smsg[smsg.find(b'charset=')+8:]).group() == b'GBK' \

or re.search(b'[a-zA-Z0-9\-]*',smsg[smsg.find(b'charset=')+8:]).group() == b'gbk':

charset = 'gbk'

if re.search(b'[a-zA-Z0-9\-]*',smsg[smsg.find(b'charset=')+8:]).group() == b'UTF-8' \

or re.search(b'[a-zA-Z0-9\-]*',smsg[smsg.find(b'charset=')+8:]).group() == b'utf-8':

charset = 'utf-8'

smsg = smsg.decode(charset) #解码

tmsg = smsg.find("

第 一 回")

t = smsg[tmsg:smsg.find("",tmsg)]

tmsg = t.find("

while tmsg > 0:

#smsg = smsg[smsg.find("

")

chapurl = myhost + t[tmsg + 9 : t.find("\"",tmsg + 12)]

#chapname = t[t.find("html\">")+6:t.find("

",t.find("html\">")+6)]+"\n"

tmsg = t.find("")

t = t[tmsg+6:]

tmsg = t.find("

temp = urlopen(chapurl).read().decode('gb18030')

chapname = temp[temp.find("")+3:temp.find("")] + '\n'

f.write(chapname)

temp = temp[temp.find("

")]

content = temp[temp.find("size=\"3\">")+9:temp.find("")] + "\n"

content = content.replace("\t","")

content = content.replace("
","")

content = content.replace(" ","")

temp = re.split(r'\n',content)

for i in range(len(temp)):

if len(temp[i])<33:

temp[i] = temp[i]+'\n'

f.write(temp[i])

f.close()