def _verify_reference(self,
reference,
original_request,
original_response,
possibly_broken,
be_recursive=True):
"""
The parameters are:
* Newly found URL
* The FuzzableRequest instance which generated the response where
the new URL was found
* The HTTPResponse generated by the FuzzableRequest
* Boolean indicating if we trust this reference or not
This method GET's every new link and parses it in order to get
new links and forms.
"""
#
# Remember that this "breaks" the cache=True in most cases!
# headers = { 'Referer': original_url }
#
# But this does not, and it is friendlier than simply ignoring the
# referer
#
# if self._fuzzy_browser.check_page( str(reference) ):
# self._fuzzy_browser.add_page( str(reference) )
# else:
# return
referer = original_response.get_url().base_url().url_string
headers = Headers([('Referer', referer)])
# Note: We're not grep'ing this HTTP request/response now because it
# has high probability of being a 404, and the grep plugins
# already got enough 404 responses to analyze (from is_404 for
# example). If it's not a 404 then we'll push it to the core
# and it will come back to this plugin's crawl() where it will
# be requested with grep=True
self._requests_count += 1
if self._requests_count > self._max_requests_count:
return
resp = self._uri_opener.GET(reference,
cache=True,
headers=headers,
grep=False)
if is_404(resp):
# Note: I WANT to follow links that are in the 404 page, but
# DO NOT return the 404 itself to the core.
#
# This will parse the 404 response and add the 404-links in the
# output queue, so that the core can get them
#
if be_recursive:
#
# Only follow one level of links in 404 pages, this limits the
# potential issue when this is found:
#
# http://foo.com/abc/ => 404
# Body: <a href="def/">link</a>
#
# Which would lead to this function to perform requests to:
# * http://foo.com/abc/
# * http://foo.com/abc/def/
# * http://foo.com/abc/def/def/
# * http://foo.com/abc/def/def/def/
# * ...
#
# Do not use threads here, it will dead-lock (for unknown
# reasons). This is tested in TestDeadLock unittest.
for args in self._urls_to_verify_generator(
resp, original_request):
self._verify_reference(*args, be_recursive=False)
# Store the broken links
if not possibly_broken and resp.get_code(
) not in self.UNAUTH_FORBID:
t = (resp.get_url(), original_request.get_uri())
self._broken_links.add(t)
else:
msg = '[web_spider] Sending link to w3af core: "%s"'
om.out.debug(msg % reference)
fuzz_req = FuzzableRequest(reference, headers=headers)
# These next steps are simple, but actually allows me to set the
# referer and cookie for the FuzzableRequest instances I'm sending
# to the core, which will then allow the fuzzer to create
# CookieMutant and HeadersMutant instances.
#
# Without setting the Cookie, the CookieMutant would never have any
# data to modify; remember that cookies are actually set by the
# urllib2 cookie handler when the request already exited the
# framework.
cookie = Cookie.from_http_response(original_response)
fuzz_req.set_referer(referer)
fuzz_req.set_cookie(cookie)
self.output_queue.put(fuzz_req)
def _verify_reference(self,
reference,
original_request,
original_response,
possibly_broken,
be_recursive=True):
"""
The parameters are:
* Newly found URL
* The FuzzableRequest instance which generated the response where
the new URL was found
* The HTTPResponse generated by the FuzzableRequest
* Boolean indicating if we trust this reference or not
This method GET's every new link and parses it in order to get
new links and forms.
"""
#
# Remember that this "breaks" the cache=True in most cases!
# headers = { 'Referer': original_url }
#
# But this does not, and it is friendlier than simply ignoring the
# referer
#
referer = original_response.get_url().base_url().url_string
headers = Headers([('Referer', referer)])
resp = self._uri_opener.GET(reference, cache=True, headers=headers)
if is_404(resp):
# Note: I WANT to follow links that are in the 404 page, but
# DO NOT return the 404 itself to the core.
#
# This will parse the 404 response and add the 404-links in the
# output queue, so that the core can get them
#
if be_recursive:
#
# Only follow one level of links in 404 pages, this limits the
# potential issue when this is found:
#
# http://foo.com/abc/ => 404
# Body: <a href="def/">link</a>
#
# Which would lead to this function to perform requests to:
# * http://foo.com/abc/
# * http://foo.com/abc/def/
# * http://foo.com/abc/def/def/
# * http://foo.com/abc/def/def/def/
# * ...
#
non_recursive_verify_ref = partial(self._verify_reference,
be_recursive=False)
self.worker_pool.map_multi_args(
non_recursive_verify_ref,
self._urls_to_verify_generator(resp, original_request))
# Store the broken links
if not possibly_broken and resp.get_code(
) not in self.UNAUTH_FORBID:
t = (resp.get_url(), original_request.get_uri())
self._broken_links.add(t)
else:
msg = 'Adding reference "%s" to the result.'
om.out.debug(msg % reference)
fuzz_req = FuzzableRequest(reference, headers=headers)
# These next steps are simple, but actually allows me to set the
# referer and cookie for the FuzzableRequest instances I'm sending
# to the core, which will then allow the fuzzer to create
# CookieMutant and HeadersMutant instances.
#
# Without setting the Cookie, the CookieMutant would never have any
# data to modify; remember that cookies are actually set by the
# urllib2 cookie handler when the request already exited the
# framework.
cookie = Cookie.from_http_response(original_response)
fuzz_req.set_referer(referer)
fuzz_req.set_cookie(cookie)
self.output_queue.put(fuzz_req)
def _verify_reference(self, reference, original_request,
original_response, possibly_broken,
be_recursive=True):
"""
The parameters are:
* Newly found URL
* The FuzzableRequest instance which generated the response where
the new URL was found
* The HTTPResponse generated by the FuzzableRequest
* Boolean indicating if we trust this reference or not
This method GET's every new link and parses it in order to get
new links and forms.
"""
#
# Remember that this "breaks" the cache=True in most cases!
# headers = { 'Referer': original_url }
#
# But this does not, and it is friendlier than simply ignoring the
# referer
#
referer = original_response.get_url().base_url().url_string
headers = Headers([('Referer', referer)])
# Note: We're not grep'ing this HTTP request/response now because it
# has high probability of being a 404, and the grep plugins
# already got enough 404 responses to analyze (from is_404 for
# example). If it's not a 404 then we'll push it to the core
# and it will come back to this plugin's crawl() where it will
# be requested with grep=True
resp = self._uri_opener.GET(reference, cache=True, headers=headers,
grep=False)
if is_404(resp):
# Note: I WANT to follow links that are in the 404 page, but
# DO NOT return the 404 itself to the core.
#
# This will parse the 404 response and add the 404-links in the
# output queue, so that the core can get them
#
if be_recursive:
#
# Only follow one level of links in 404 pages, this limits the
# potential issue when this is found:
#
# http://foo.com/abc/ => 404
# Body: <a href="def/">link</a>
#
# Which would lead to this function to perform requests to:
# * http://foo.com/abc/
# * http://foo.com/abc/def/
# * http://foo.com/abc/def/def/
# * http://foo.com/abc/def/def/def/
# * ...
#
# Do not use threads here, it will dead-lock (for unknown
# reasons). This is tested in TestDeadLock unittest.
for args in self._urls_to_verify_generator(resp, original_request):
self._verify_reference(*args, be_recursive=False)
# Store the broken links
if not possibly_broken and resp.get_code() not in self.UNAUTH_FORBID:
t = (resp.get_url(), original_request.get_uri())
self._broken_links.add(t)
else:
msg = '[web_spider] Sending link to w3af core: "%s"'
om.out.debug(msg % reference)
fuzz_req = FuzzableRequest(reference, headers=headers)
# These next steps are simple, but actually allows me to set the
# referer and cookie for the FuzzableRequest instances I'm sending
# to the core, which will then allow the fuzzer to create
# CookieMutant and HeadersMutant instances.
#
# Without setting the Cookie, the CookieMutant would never have any
# data to modify; remember that cookies are actually set by the
# urllib2 cookie handler when the request already exited the
# framework.
cookie = Cookie.from_http_response(original_response)
fuzz_req.set_referer(referer)
fuzz_req.set_cookie(cookie)
self.output_queue.put(fuzz_req)
def _verify_reference(self, reference, original_request,
original_response, possibly_broken,
be_recursive=True):
"""
The parameters are:
* Newly found URL
* The FuzzableRequest instance which generated the response where
the new URL was found
* The HTTPResponse generated by the FuzzableRequest
* Boolean indicating if we trust this reference or not
This method GET's every new link and parses it in order to get
new links and forms.
"""
#
# Remember that this "breaks" the cache=True in most cases!
# headers = { 'Referer': original_url }
#
# But this does not, and it is friendlier than simply ignoring the
# referer
#
referer = original_response.get_url().base_url().url_string
headers = Headers([('Referer', referer)])
resp = self._uri_opener.GET(reference, cache=True, headers=headers)
if is_404(resp):
# Note: I WANT to follow links that are in the 404 page, but
# DO NOT return the 404 itself to the core.
#
# This will parse the 404 response and add the 404-links in the
# output queue, so that the core can get them
#
if be_recursive:
#
# Only follow one level of links in 404 pages, this limits the
# potential issue when this is found:
#
# http://foo.com/abc/ => 404
# Body: <a href="def/">link</a>
#
# Which would lead to this function to perform requests to:
# * http://foo.com/abc/
# * http://foo.com/abc/def/
# * http://foo.com/abc/def/def/
# * http://foo.com/abc/def/def/def/
# * ...
#
non_recursive_verify_ref = partial(self._verify_reference,
be_recursive=False)
self.worker_pool.map_multi_args(
non_recursive_verify_ref,
self._urls_to_verify_generator(resp, original_request))
# Store the broken links
if not possibly_broken and resp.get_code() not in self.UNAUTH_FORBID:
t = (resp.get_url(), original_request.get_uri())
self._broken_links.add(t)
else:
msg = 'Adding reference "%s" to the result.'
om.out.debug(msg % reference)
fuzz_req = FuzzableRequest(reference, headers=headers)
# These next steps are simple, but actually allows me to set the
# referer and cookie for the FuzzableRequest instances I'm sending
# to the core, which will then allow the fuzzer to create
# CookieMutant and HeadersMutant instances.
#
# Without setting the Cookie, the CookieMutant would never have any
# data to modify; remember that cookies are actually set by the
# urllib2 cookie handler when the request already exited the
# framework.
cookie = Cookie.from_http_response(original_response)
fuzz_req.set_referer(referer)
fuzz_req.set_cookie(cookie)
self.output_queue.put(fuzz_req)