Python 爬蟲

# -*- coding: UTF-8 -*-
from  bs4  import  BeautifulSoup
import urllib
import requests
import MySQLdb


train_link = []
train_news_title = []
links = ['http://www.railway.gov.tw/tw/news.aspx?n=6807']

for link in links:
    res = requests.get(link)
    soup = BeautifulSoup(res.text.encode("utf-8"), "html.parser")
    train_table = soup.findAll('tr',{'class':'gray01 text_12_1pt form01'})

conn = MySQLdb.connect(host='localhost',user='root',passwd='',db='crawl')
cur=conn.cursor()

for train_link in train_table:
    link = 'http://www.railway.gov.tw/tw/' + [tag['href'] for tag in train_link.findAll('a',{'href':True})][0]
    train_link.append(link)

for train_news_title in train_table:
    news_title = [tag['title'] for tag in train_news_title.findAll('span',{'title':True})][0]
    train_news_title.append(news_title)


if  len(train_link) == len(train_news_title):
        number1 = len(train_link)
        number2 = len(train_news_title)
for j in range(number1):
        a = train_link[j]
        b = train_news_title[j]
print a,b
sql = "INSERT INTO RAILWAY(URL,TITLE)  VALUES ('%s','%s')" %(a,b)
try:
    cur.execute(sql)
    conn.commit()
except Exception, e:
    print e
    conn.rollback()

cur.close()
conn.close()