翻看自己以前寫的程序，發(fā)現(xiàn)寫過一個(gè)爬取盤多多百度云資源的東西，完全是當(dāng)時(shí)想看變形金剛才自己寫的，而且當(dāng)時(shí)第一次接觸python大概寫了有2天才搞出來這個(gè)程序，學(xué)習(xí)python語言，可以看得出來那時(shí)候的代碼寫的真的low。雖然現(xiàn)在也不怎么樣，哈哈，一直學(xué)習(xí)中，不做過多解釋，上圖上代碼，因?yàn)樽兞康穆暶魇鞘裁次易约阂餐耍ㄊ謩?dòng)傲嬌），連寫入文件當(dāng)時(shí)都不會(huì)哈哈哈哈哈哈哈哈，也不知道class中可以通過init初始化，唉學(xué)習(xí)python原來我學(xué)到了這么多東西，感謝python

cmd圖片

from bs4 import BeautifulSoup
import urllib
import requests
import re

adr =[]

''''對(duì)搜素資源名字進(jìn)行url編碼'''
search_text =raw_input('請(qǐng)輸入搜索資源名：')
search_text = search_text.decode('gbk')
search_text = search_text.encode('utf-8')
search_text = urllib.quote(search_text)


''''獲取文件地址'''
home = urllib.urlopen('http://www.panduoduo.net/s/name/'+search_text)


'''獲取百度云地址'''
def getbaidu(adr):
    for i in adr:
        url = urllib.urlopen('http://www.panduoduo.net'+i)
        bs = BeautifulSoup(url)
        bs1 = bs.select('.dbutton2')
        href = re.compile('http\%(\%|\d|\w|\/\/|\/|\.)*')
        b = href.search(str(bs1))
        name = str(bs.select('.center')).decode('utf-8')
        text1 = re.compile('\<h1\sclass\=\"center"\>[\d|\w|\D|\W]*\</h1\>')
        text2 = text1.search(name)
        rag1 = re.compile('\>[\d|\w|\D|\W]*\<')
        if text2:
            text3 = rag1.search(text2.group())
            if text3:
                print text3.group()
        if b:
            text = urllib.unquote(str(b.group())).decode('utf-8')
            print text

'''初始化'''
def init(adr):
    soup = BeautifulSoup(home)
    soup = soup.select('.row')
    pattern = re.compile('\/r\/\d+')
    for i in soup:
        i = str(i)
        adress = pattern.search(i)
        adress = adress.group()
        adr.append(adress)


print 'running---------'        
init(adr)
getbaidu(adr)