控制节点
URL管理器
import pickle
import hashlib
class UrlManager():
def __init__(self):
self.new_urls=self.load_process('new_urls.txt')
self.old_urls=self.load_process('old_urls.txt')
def has_new_url(self):
return self.new_url_size()!=0
def get_new_url(self):
new_url=self.new_urls.pop()
m=hashlib.md5()
m.update(new_url)
self.old_urls.add(m.hexdigest()[8:-8])
return new_url
def add_new_url(self,url):
if url is None:
return
m=hashlib.md5()
m.update(url)
url_md5=m.hexdigest()[8:-8]
if url not in self.new_urls and url_md5 not in self.old_urls:
self.new_urls.add(url)
def add_new_urls(self,urls):
if urls is None or len(urls)==0:
return
for url in urls:
self.add_new_url(url)
def new_url_size(self):
return len(self.new_urls)
def old_url_size(self):
return len(self.old_urls)
def save_progress(self,path,data):
with open(path,'wb') as f:
cPickle.dump(data,f)
def load_progress(self,path):
print('[+]从文件中加载进度: %s'%path)
try:
with open(path,'rb') as f:
tmp=pickle.load(f)
return tmp
except:
print('[!]无进度文件,创建: %s'%path)
return set()
数据存储器
import codecs
import time
class DataOutput():
def __init__(self):
self.filepath='baike_%s.html'%(time.strftime('%Y_%m_%d_%H_%M_%S',time.localtimr()))
self.ouput_head(self.filepath)
self.datas=[]
def store_data(self,data):
if data is None:
return
self.datas.append(data)
id len(self.datas)>10:
self.output_html(self.filepath)
def output_head(self,path):
fout=codecs.open(path,'w',encoding='utf-8')
fout.write('<html>')
fout('<body>')
fout.write('<table>')
fout.close()
def output_html(self,path):
fout=codes.open(path,'a',encoding='utf-8')
for data in self.datas:
fout.write('<tr>')
fout.write('<td>%s<td>'%data['url'])
fout.write('<td>%s<td>'%data['title'])
fout.write('<td>%s<td>'%data['summary'])
fout.write('</tr>')
self.datas.remove(data)
fout.close()
def output_end(self,path):
fout=codes.open(path,'a',encoding='utf-8')
fout.write('</table>')
fout.write('</body>')
fout.write('</html>')
fout.close()
控制调度器
def start_Manager(self,url_q,result_q):
BaseManager.register('get_task_quue',callable=lambda:url_q)
BaseManager.register('get_result_queue',callable=lambda:result_q)
manager=BaseManager(address={"",8001},authkey='baike')
return manager
def url_manager_proc(self,url_q,conn_q,root_url):
url_manager-UrlManager()
url_manager.add_new_url(root_url)
while True:
while(url_manager.has_new_url()):
new_url=url_manager.get_new_url()
url_q.put(new_url)
print('old_url=',url_manager.old_url_size())
if (url_manager.old_url_size()>2000):
url_q.put('end')
print('控制节点发起结束通知!')
url_manager.save_progress('new_urls.txt',url_manager.new_urls)
url_manager.save_progress('old_urls.txt',url_manager.old_urls)
return
try:
if not conn_q.empty():
urls=conn_q.get()
url_manager.add_new_urls(urls)
except BaseException as e:
time.sleep(0.1)
def result_solve_proc(self,result_q,conn_q,store_q):
while(True):
try:
if not result_q.empty():
content=result_q.get(True)
if content['new_urls']=='end':
print('结果分析进程接收通知然后结束!')
store_q.put('end')
return
conn_q.put(content['new_urls'])
store_q.put(cntent['data'])
else:
time.sleep(0.1)
except BaseException as e:
time.sleep(0.1)
def store_proc(self,store_q):
output=DataOutput()
while True:
if not store_q.empty():
data=store_q..get()
if data=='end':
print('存储进程接收通知然后结束!')
output.output_end(output.filepath)
return
output.store_data(data)
else:
time.sleep(0.1)
if __name__=='__main__':
url_q=Queue()
result_q=Queue()
store_q=Queue()
conn_q=Queue()
node=NodeManager()
manager=node.start_Manager(url_q,result_q)
url_manager_proc=Process(target=node.url_manager_proc,args=(url_q,conn_q,'https://baike.baidu.com/item/%E7%BD%91%E7%BB%9C%E7%88%AC%E8%99%AB/5162711?fr=aladdin',))
result_solve_proc=Process(target=node.result_solve_proc,args=(result_q,conn_q,store_q,))
store_proc=Process(target=node.store_proc,args=(store_q,))
url_manager_proc.start()
result_solve_proc.start()
store_proc.start()
manager.get_server().serve_forever()
爬虫节点
HTML下载器
import requests
class HtmlDownloader():
def download(self,url):
if url is None:
return None
headers = {
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.90 Safari/537.36 2345Explorer/9.3.2.17331',
'Referer': r'https://baike.baidu.com',
'Connection': 'keep-alive'
}
r=requests.get(url,headers=headers)
if r.status_code==200:
f.encoding='utf-8'
return r.text
return None
HTML解析器
import re
from urllib.parse import urlparse
from bs4 import BeautifulSoup
class HtmlParser():
def parser(self,page_url,html_cont):
if page_url is None or html_cont is None:
return 'Empty !!!'
soup=BeautifulSoup(html_cont,'html.parser',from_encoding='utf-8')
new_urls=self._get_new_urls(page_url,soup)
new_data=self._get_new_data(page_url,soup)
return new_urls,new_data
def _get_new_urls(self,page_url,soup):
new_urls=set()
links=soup.find_all('a',href=re.compile(r'\b(/item/)'))
for link in links:
new_url=link['href']
new_full_url=urlparse.urljoin(page_url,new_url)
new_urls.add(new_full_url)
return new_urls
def _get_new_data(self,page_url,soup):
data={}
data['url']=page_url
title=soup.find('dd','lemmaWgt-lemmaTitle-title').find('h1')
data['title']=title.get_text()
summary=soup.find('div','lemma-summary')
data['summary']=summary.get_text()
return data
爬虫调度器
class SpiderWork():
def __init__(self):
BaseManager.register('get_task_queue')
BaseManager.register('get_result_queue')
server_addr='127.0.0.1'
print('Connect to server %s ...'%server_addr)
self.manager=UrlManager()
self.m=BaseManager(address=(server_addr,8001),authkey='baike')
self.m.connect()
self.task=self.m.get_task_queue()
self.result=self.m.get_result_queue()
self.downloader=HtmlDownloader()
self.parser=HtmlParser()
print('init finish')
def crawl(self):
while(True):
try:
if not self.task.empty():
url=self.task.get()
if url=='end':
print('控制节点通知爬虫节点停止工作...')
self.result.put({'new_urls':'end','data':'end'})
return
print('爬虫节点正在解析:%s'%url.encode('utf-8'))
content=self.downloader.download(url)
new_urls,data=self.parder.parser(url,content)
self.result.put({'new_urls':new_urls,'data':data})
except EOFError as e:
print('连接工作节点失败')
return
except Exception as e:
print(e,'\nCraw fail !')
self.manager.add_new_url(root_url)
while(self.manager.has_new_url() and self.manager.old_url_size()<100):
try:
new_url=self.manager.get_new_url()
html=self.downloader.download(new_url)
new_urls,data=self.parser.parser(new_url,html)
self.manager.add_new_urls(new_urls)
self.output.store_data(data)
print('已抓取 %s 个链接'%self.manager.old_url_size())
except Exception as e:
print('crawl failed')
if __name__=='__main__':
spider=SpiderWor()
spider.crawl()