所需套件
簡繁轉換, 需先下載如下二個套件
langconv.py 及 zh_wiki.py
trans套件, 是將中文的五百三十五轉換成 535,請由如連結下載
zh_number.py
import os
import requests, re
import codecs
from bs4 import BeautifulSoup
from langconv import Converter
from zh_number import trans
page = requests.get("http://www.uuxs.tw/ls/22_22102/")
page.encoding="utf-8"
soup=BeautifulSoup(page.text, "html.parser")
print(soup.prettify())
pattern=re.compile('.*第.*章')
chapterList=soup.find_all('a', string=pattern)
urlno=[]
for l in chapterList:
urlno.append(int((l['href'].replace('.html',''))))
urlno.sort()
path = "d:/天降巨富"
if not os.path.isdir(path):
os.mkdir(path)
for no in urlno:
url="http://www.uuxs.tw/ls/22_22102/{0}.html".format(no)
page=requests.get(url)
page.encoding="utf-8"
soup = BeautifulSoup(page.text, 'html.parser')
cnChapter = soup.find(attrs={"class": "bookname"}).h1
content = soup.find(id='content')
zhChapter=Converter('zh-hant').convert(cnChapter.string).replace(' ','')
zh_num=zhChapter.split('章')[0].replace('第', '')
num=trans(zh_num)
zh_title=zhChapter.split('章')[1]
fileName="d:/天降巨富/No%03d.%s.txt" % (num, zh_title)
file = codecs.open(fileName, 'w', 'utf-8')
print(fileName)
file.write(zhChapter)
for s in content.strings:
zhContent=Converter('zh-hant').convert(s)
file.write(zhContent)
file.close()
