def getsub(a,b):
for number in range(a,b):
browser.get("http://www.shooter.cn/xml/sub/%u/%u.xml" % (number/1000,number)) # Load page
i = 0
while "Shooter.cn" not in browser.title and i < 2:
i = i + 1
browser.get("http://www.shooter.cn/xml/sub/%u/%u.xml" % (number/1000,number)) # Load page
if "Shooter.cn" not in browser.title:
print str(number) + "'s page can't be loaded."
continue
# if not "Shooter.cn" in browser.title:
# print str(number) + " seems to be broken."
# continue
try:
table = browser.find_element_by_tag_name("table")
except:
print str(number) + " seems to be broken."
continue
title = browser.find_element_by_id("movietitle1")
row = table.find_elements_by_tag_name("tr")
# j=0
# for i in row:
# print str(j) + ": ",
# print i.text
# j=j+1
# print len(row)
# exit()
cells = row[2].find_elements_by_tag_name("td")
fmt = cells[1].text
lang = cells[3].text
cells = row[len(row) - 1].find_elements_by_tag_name("td")
filename = cells[1].text
# print filename
# exit()
c1 = 'Subrip' in fmt # or 'SSA' in fmt or 'ASS' in fmt or 'MacroDVD' in fmt
c2 = (u'简' in lang or u'中' in lang) and (u'英' in lang)
# print table.text
if c1 and c2:
print title.text
cur.execute('INSERT into movies VALUES(?,?,?,?)',(number,title.text,filename,table.text))
elem = browser.find_element_by_id("downsubbtn") # Find the query box
elem.click()
db.commit()
if number % 1000 == 999:
movefiles.movefiles(number/1000)
# -*- coding: utf-8 -*- #!/usr/bin/python import sys, os, time, shutil import movefiles sys.exit() movefiles.movefiles(5)