sqlmap/lib/utils/google.py

123 lines
3.5 KiB
Python
Raw Normal View History

2008-10-15 19:38:22 +04:00
#!/usr/bin/env python
"""
$Id: google.py 330 2008-08-28 21:25:48Z inquisb $
This file is part of the sqlmap project, http://sqlmap.sourceforge.net.
Copyright (c) 2006-2008 Bernardo Damele A. G. <bernardo.damele@gmail.com>
and Daniele Bellucci <daniele.bellucci@gmail.com>
sqlmap is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation version 2 of the License.
sqlmap is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
details.
You should have received a copy of the GNU General Public License along
with sqlmap; if not, write to the Free Software Foundation, Inc., 51
Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
"""
import cookielib
import re
import urllib2
from lib.core.convert import urlencode
from lib.core.data import conf
from lib.core.exception import sqlmapConnectionException
from lib.core.exception import sqlmapRegExprException
class Google:
"""
This class defines methods used to perform Google dorking (command
line option '-g <google dork>'
"""
def __init__(self, proxyHandler):
self.__googleCookie = None
self.__matches = []
self.__cj = cookielib.LWPCookieJar()
self.opener = urllib2.build_opener(proxyHandler, urllib2.HTTPCookieProcessor(self.__cj))
self.opener.addheaders = conf.httpHeaders
def __parsePage(self, page):
"""
Parse Google dork search results page to get the list of
HTTP addresses
"""
matches = []
regExpr = "class=r\076\074a href=\042(http[s]*://.+?)\042\sclass=l"
matches = re.findall(regExpr, page, re.I | re.M)
return matches
def getTargetUrls(self):
"""
This method returns the list of hosts with parameters out of
your Google dork search results
"""
targetUrls = set()
for match in self.__matches:
if re.search("(.*?)\?(.+)", match, re.I):
targetUrls.add(match)
return targetUrls
def getCookie(self):
"""
This method is the first to be called when initializing a
Google dorking object through this library. It is used to
retrieve the Google session cookie needed to perform the
further search
"""
try:
conn = self.opener.open("http://www.google.com/ncr")
headers = conn.info()
except urllib2.HTTPError, e:
headers = e.info()
except urllib2.URLError, e:
errMsg = "unable to connect to Google"
raise sqlmapConnectionException, errMsg
def search(self, googleDork):
"""
This method performs the effective search on Google providing
the google dork and the Google session cookie
"""
if not googleDork:
return None
url = "http://www.google.com/search?"
url += "q=%s&" % urlencode(googleDork)
url += "num=100&hl=en&safe=off&filter=0&btnG=Search"
try:
conn = self.opener.open(url)
page = conn.read()
except urllib2.HTTPError, e:
page = e.read()
except urllib2.URLError, e:
errMsg = "unable to connect to Google"
raise sqlmapConnectionException, errMsg
self.__matches = self.__parsePage(page)
return self.__matches