Wrote this crawler in Python, it dumps several parameters to JSON output file based on the input list of domains.
Have this question:
Do I need to close the HTTP connection in each thread? Input data is ca. 5 Million items. It process at the beginning at a rate ca. 50 iterations per second, but later after some time it drops to 1-2 per second and/or hangs (no kernel messages and no errors on stdout)? Can this be code or is network limiting related? I suspect software since when I restart it, it starts again with high rate (ca. 50 iteration per second)
Any tips how to improve the code below are also welcome, especially improve on speed and crawling throughput.
Code in questions:
import urllib2
import pprint
from tqdm import tqdm
import lxml.html
from Queue import Queue
from geoip import geolite2
import pycountry
from tld import get_tld
resfile = open("out.txt",'a')
concurrent = 200
def doWork():
while True:
url = q.get()
status = getStatus(url)
doSomethingWithResult(status)
q.task_done()
def getStatus(ourl):
try:
response = urllib2.urlopen("http://"+ourl)
peer = response.fp._sock.fp._sock.getpeername()
ip = peer[0]
header = response.info()
html = response.read()
html_element = lxml.html.fromstring(html)
generator = html_element.xpath("//meta[@name='generator']/@content")
try:
match = geolite2.lookup(ip)
if match is not None:
country= match.country
try:
c=pycountry.countries.lookup(country)
country=c.name
except:
country=""
except:
country=""
try:
res=get_tld("http://www"+ourl, as_object=True)
tld=res.suffix
except:
tld=""
try:
match = re.search(r'[\w\.-]+@[\w\.-]+', html)
email=match.group(0)
except:
email=""
try:
item= generator[0]
val = "{ \"Domain\":\"http://"+ourl.rstrip()+"\",\"IP:\""+ip+"\"," + "\"Server\":"+ "\""+str(header.getheader("Server")).replace("None","")+"\",\"PoweredBy\":" + "\""+str(header.getheader("X-Powered-By")).replace("None","")+"\""+",\"MetaGenerator\":\""+item+"\",\"Email\":\""+email+"\",\"Suffix\":\""+tld+"\",\"CountryHosted\":\""+country+"\" }"
except:
val = "{ \"Domain\":\"http://"+ourl.rstrip()+"\",\"IP:\""+ip+"\"," + "\"Server\":"+ "\""+str(header.getheader("Server")).replace("None","")+"\",\"PoweredBy\":" + "\""+str(header.getheader("X-Powered-By")).replace("None","")+"\""+",\"MetaGenerator\":\"\",\"Email\":\""+email+"\",\"Suffix\":\""+tld+"\",\"CountryHosted\":\""+country+"\" }"
return val
except Exception as e:
#print "error"+str(e)
pass
def doSomethingWithResult(status):
if status:
resfile.write(str(status)+"\n")
q = Queue(concurrent * 2)
for i in range(concurrent):
t = Thread(target=doWork)
t.daemon = True
t.start()
try:
for url in tqdm(open('list.txt')):
q.put(url.strip())
status = open("status.txt",'w')
status.write(str(url.strip()))
q.join()
except KeyboardInterrupt:
sys.exit(1)
Update 1:
Closing the Socket and FileDescriptor makes it work better, does not seem to hang anymore after some time. Performance is 50 reqs/sec on home laptop and ca 100 req/sec on a VPS
from threading import Thread
import httplib, sys
import urllib2
import pprint
from tqdm import tqdm
import lxml.html
from Queue import Queue
from geoip import geolite2
import pycountry
from tld import get_tld
import json
resfile = open("out.txt",'a')
concurrent = 200
def doWork():
while True:
url = q.get()
status = getStatus(url)
doSomethingWithResult(status)
q.task_done()
def getStatus(ourl):
try:
response = urllib2.urlopen("http://"+ourl)
realsock = response.fp._sock.fp._sock
peer = response.fp._sock.fp._sock.getpeername()
ip = peer[0]
header = response.info()
html = response.read()
realsock.close()
response.close()
html_element = lxml.html.fromstring(html)
generator = html_element.xpath("//meta[@name='generator']/@content")
try:
match = geolite2.lookup(ip)
if match is not None:
country= match.country
try:
c=pycountry.countries.lookup(country)
country=c.name
except:
country=""
except:
country=""
try:
res=get_tld("http://www"+ourl, as_object=True)
tld=res.suffix
except:
tld=""
try:
match = re.search(r'[\w\.-]+@[\w\.-]+', html)
email=match.group(0)
except:
email=""
try:
item= generator[0]
val = "{ \"Domain\":"+json.dumps("http://"+ourl.rstrip())+",\"IP\":\""+ip+"\",\"Server\":"+json.dumps(str(header.getheader("Server")).replace("None",""))+",\"PoweredBy\":" +json.dumps(str(header.getheader("X-Powered-By")).replace("None",""))+",\"MetaGenerator\":"+json.dumps(item)+",\"Email\":"+json.dumps(email)+",\"Suffix\":\""+tld+"\",\"CountryHosted\":\""+country+"\" }"
except:
val = "{ \"Domain\":"+json.dumps("http://"+ourl.rstrip())+",\"IP\":\""+ip+"\"," + "\"Server\":"+json.dumps(str(header.getheader("Server")).replace("None",""))+",\"PoweredBy\":" +json.dumps(str(header.getheader("X-Powered-By")).replace("None",""))+",\"MetaGenerator\":\"\",\"Email\":"+json.dumps(email)+",\"Suffix\":\""+tld+"\",\"CountryHosted\":\""+country+"\" }"
return val
except Exception as e:
print "error"+str(e)
pass
def doSomethingWithResult(status):
if status:
resfile.write(str(status)+"\n")
q = Queue(concurrent * 2)
for i in range(concurrent):
t = Thread(target=doWork)
t.daemon = True
t.start()
try:
for url in tqdm(open('list.txt')):
q.put(url.strip())
status = open("status.txt",'w')
status.write(str(url.strip()))
q.join()
except KeyboardInterrupt:
sys.exit(1)
The handles will be automatically garbage collected, but, you will be better off closing the handles yourself, especially as you are doing this in a tight loop.
You also asked for suggestions for improvement. A big one would be to stop using urllib2 and start using requests instead.