Anonymous avatar Anonymous committed 838b499

[1.3.X] Altered the behavior of URLField to avoid a potential DOS vector, and to avoid potential leakage of local filesystem data. A security announcement will be made shortly.

Backport of r16760 from trunk.

Comments (0)

Files changed (10)

django/core/validators.py

+import platform
 import re
 import urllib2
 import urlparse
         if not self.regex.search(smart_unicode(value)):
             raise ValidationError(self.message, code=self.code)
 
-class HeadRequest(urllib2.Request):
-    def get_method(self):
-        return "HEAD"
-
 class URLValidator(RegexValidator):
     regex = re.compile(
         r'^(?:http|ftp)s?://' # http:// or https://
         r'(?::\d+)?' # optional port
         r'(?:/?|[/?]\S+)$', re.IGNORECASE)
 
-    def __init__(self, verify_exists=False, validator_user_agent=URL_VALIDATOR_USER_AGENT):
+    def __init__(self, verify_exists=False,
+                 validator_user_agent=URL_VALIDATOR_USER_AGENT):
         super(URLValidator, self).__init__()
         self.verify_exists = verify_exists
         self.user_agent = validator_user_agent
         else:
             url = value
 
+        #This is deprecated and will be removed in a future release.
         if self.verify_exists:
             headers = {
                 "Accept": "text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5",
             broken_error = ValidationError(
                 _(u'This URL appears to be a broken link.'), code='invalid_link')
             try:
-                req = HeadRequest(url, None, headers)
-                u = urllib2.urlopen(req)
+                req = urllib2.Request(url, None, headers)
+                req.get_method = lambda: 'HEAD'
+                #Create an opener that does not support local file access
+                opener = urllib2.OpenerDirector()
+
+                #Don't follow redirects, but don't treat them as errors either
+                error_nop = lambda *args, **kwargs: True
+                http_error_processor = urllib2.HTTPErrorProcessor()
+                http_error_processor.http_error_301 = error_nop
+                http_error_processor.http_error_302 = error_nop
+                http_error_processor.http_error_307 = error_nop
+
+                handlers = [urllib2.UnknownHandler(),
+                            urllib2.HTTPHandler(),
+                            urllib2.HTTPDefaultErrorHandler(),
+                            urllib2.FTPHandler(),
+                            http_error_processor]
+                try:
+                    import ssl
+                    handlers.append(urllib2.HTTPSHandler())
+                except:
+                    #Python isn't compiled with SSL support
+                    pass
+                map(opener.add_handler, handlers)
+                if platform.python_version_tuple() >= (2, 6):
+                    opener.open(req, timeout=10)
+                else:
+                    opener.open(req)
             except ValueError:
                 raise ValidationError(_(u'Enter a valid URL.'), code='invalid')
-            except urllib2.HTTPError, e:
-                if e.code in (405, 501):
-                    # Try a GET request (HEAD refused)
-                    # See also: http://www.w3.org/Protocols/rfc2616/rfc2616.html
-                    try:
-                        req = urllib2.Request(url, None, headers)
-                        u = urllib2.urlopen(req)
-                    except:
-                        raise broken_error
-                else:
-                    raise broken_error
             except: # urllib2.URLError, httplib.InvalidURL, etc.
                 raise broken_error
 

django/db/models/fields/__init__.py

 class URLField(CharField):
     description = _("URL")
 
-    def __init__(self, verbose_name=None, name=None, verify_exists=True, **kwargs):
+    def __init__(self, verbose_name=None, name=None, verify_exists=False, **kwargs):
         kwargs['max_length'] = kwargs.get('max_length', 200)
         CharField.__init__(self, verbose_name, name, **kwargs)
         self.validators.append(validators.URLValidator(verify_exists=verify_exists))

docs/internals/deprecation.txt

           beyond that of a simple ``TextField`` since the removal of oldforms.
           All uses of ``XMLField`` can be replaced with ``TextField``.
 
+       * ``django.db.models.fields.URLField.verify_exists`` has been
+         deprecated due to intractable security and performance
+         issues. Validation behavior has been removed in 1.4, and the
+         argument will be removed in 1.5.
+
+
     * 1.5
         * The ``mod_python`` request handler has been deprecated since the 1.3
           release. The ``mod_wsgi`` handler should be used instead.

docs/ref/forms/fields.txt

     If ``True``, the validator will attempt to load the given URL, raising
     ``ValidationError`` if the page gives a 404. Defaults to ``False``.
 
+.. deprecated:: 1.3.1
+
+   ``verify_exists`` was deprecated for security reasons and will be
+   removed in 1.4. This deprecation also removes ``validator_user_agent``.
+
 .. attribute:: URLField.validator_user_agent
 
     String used as the user-agent used when checking for a URL's existence.

docs/ref/models/fields.txt

 ``URLField``
 ------------
 
-.. class:: URLField([verify_exists=True, max_length=200, **options])
+.. class:: URLField([verify_exists=False, max_length=200, **options])
 
 A :class:`CharField` for a URL. Has one extra optional argument:
 
+.. deprecated:: 1.3.1
+
+   ``verify_exists`` is deprecated for security reasons as of 1.3.1
+   and will be removed in 1.4. Prior to 1.3.1, the default value was
+   ``True``.
+
 .. attribute:: URLField.verify_exists
 
-    If ``True`` (the default), the URL given will be checked for existence
-    (i.e., the URL actually loads and doesn't give a 404 response).
+    If ``True``, the URL given will be checked for existence (i.e.,
+    the URL actually loads and doesn't give a 404 response) using a
+    ``HEAD`` request. Redirects are allowed, but will not be followed.
 
     Note that when you're using the single-threaded development server,
     validating a URL being served by the same server will hang. This should not

docs/ref/settings.txt

 
 .. _See available choices: http://www.postgresql.org/docs/8.1/static/datetime-keywords.html#DATETIME-TIMEZONE-SET-TABLE
 
-.. setting:: URL_VALIDATOR_USER_AGENT
-
-URL_VALIDATOR_USER_AGENT
-------------------------
-
-Default: ``Django/<version> (http://www.djangoproject.com/)``
-
-The string to use as the ``User-Agent`` header when checking to see if URLs
-exist (see the ``verify_exists`` option on :class:`~django.db.models.URLField`).
-
 .. setting:: USE_ETAGS
 
 USE_ETAGS
    This setting has been replaced by :setting:`TEST_NAME` in
    :setting:`DATABASES`.
 
+
+
+URL_VALIDATOR_USER_AGENT
+------------------------
+
+.. deprecated:: 1.3.1
+   This setting has been removed due to intractable performance and
+   security problems.
+
+Default: ``Django/<version> (http://www.djangoproject.com/)``
+
+The string to use as the ``User-Agent`` header when checking to see if
+URLs exist (see the ``verify_exists`` option on
+:class:`~django.db.models.URLField`). This setting was deprecated in
+1.3.1 along with ``verify_exists`` and will be removed in 1.4.
+

tests/modeltests/validation/__init__.py

-from django.utils import unittest
+from django.test import TestCase
 
 from django.core.exceptions import ValidationError
 
-class ValidationTestCase(unittest.TestCase):
+class ValidationTestCase(TestCase):
     def assertFailsValidation(self, clean, failed_fields):
         self.assertRaises(ValidationError, clean)
         try:

tests/modeltests/validation/models.py

     parent = models.ForeignKey('self', blank=True, null=True, limit_choices_to={'number': 10})
     email = models.EmailField(blank=True)
     url = models.URLField(blank=True)
+    url_verify = models.URLField(blank=True, verify_exists=True)
     f_with_custom_validator = models.IntegerField(blank=True, null=True, validators=[validate_answer_to_universe])
 
     def clean(self):

tests/modeltests/validation/tests.py

         mtv = ModelToValidate(number=10, name='Some Name', url='not a url')
         self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url', [u'Enter a valid value.'])
 
+    #The tests below which use url_verify are deprecated
     def test_correct_url_but_nonexisting_gives_404(self):
-        mtv = ModelToValidate(number=10, name='Some Name', url='http://google.com/we-love-microsoft.html')
-        self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url', [u'This URL appears to be a broken link.'])
+        mtv = ModelToValidate(number=10, name='Some Name', url_verify='http://qa-dev.w3.org/link-testsuite/http.php?code=404')
+        self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url_verify', [u'This URL appears to be a broken link.'])
 
     def test_correct_url_value_passes(self):
-        mtv = ModelToValidate(number=10, name='Some Name', url='http://www.example.com/')
+        mtv = ModelToValidate(number=10, name='Some Name', url_verify='http://www.google.com/')
+        self.assertEqual(None, mtv.full_clean()) # This will fail if there's no Internet connection
+
+    def test_correct_url_with_redirect(self):
+        mtv = ModelToValidate(number=10, name='Some Name', url_verify='http://qa-dev.w3.org/link-testsuite/http.php?code=301') #example.com is a redirect to iana.org now
         self.assertEqual(None, mtv.full_clean()) # This will fail if there's no Internet connection
 
     def test_correct_https_url_but_nonexisting(self):
-        mtv = ModelToValidate(number=10, name='Some Name', url='https://www.example.com/')
-        self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url', [u'This URL appears to be a broken link.'])
-
-    def test_correct_ftp_url_but_nonexisting(self):
-        mtv = ModelToValidate(number=10, name='Some Name', url='ftp://ftp.google.com/we-love-microsoft.html')
-        self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url', [u'This URL appears to be a broken link.'])
-
-    def test_correct_ftps_url_but_nonexisting(self):
-        mtv = ModelToValidate(number=10, name='Some Name', url='ftps://ftp.google.com/we-love-microsoft.html')
-        self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url', [u'This URL appears to be a broken link.'])
+        mtv = ModelToValidate(number=10, name='Some Name', url_verify='https://www.example.com/')
+        self.assertFieldFailsValidationWithMessage(mtv.full_clean, 'url_verify', [u'This URL appears to be a broken link.'])
 
     def test_text_greater_that_charfields_max_length_raises_erros(self):
         mtv = ModelToValidate(number=10, name='Some Name'*100)

tests/regressiontests/forms/tests/fields.py

             f.clean('http://www.broken.djangoproject.com') # bad domain
         except ValidationError, e:
             self.assertEqual("[u'This URL appears to be a broken link.']", str(e))
-        self.assertRaises(ValidationError, f.clean, 'http://google.com/we-love-microsoft.html') # good domain, bad page
+        self.assertRaises(ValidationError, f.clean, 'http://qa-dev.w3.org/link-testsuite/http.php?code=400') # good domain, bad page
         try:
             f.clean('http://google.com/we-love-microsoft.html') # good domain, bad page
         except ValidationError, e:
             self.assertEqual("[u'This URL appears to be a broken link.']", str(e))
 
     def test_urlfield_10(self):
-        # UTF-8 char in path, enclosed by a monkey-patch to make sure
-        # the encoding is passed to urllib2.urlopen
+        # UTF-8 in the domain.
         f = URLField(verify_exists=True)
-        try:
-            _orig_urlopen = urllib2.urlopen
-            urllib2.urlopen = lambda req: True
-            url = u'http://t\xfcr.djangoproject.com/'
-            self.assertEqual(url, f.clean(url))
-        finally:
-            urllib2.urlopen = _orig_urlopen
+        url = u'http://\u03b5\u03bb\u03bb\u03b7\u03bd\u03b9\u03ba\u03ac.idn.icann.org/\u0391\u03c1\u03c7\u03b9\u03ba\u03ae_\u03c3\u03b5\u03bb\u03af\u03b4\u03b1'
+        self.assertEqual(url, f.clean(url)) #This will fail without internet.
 
     # BooleanField ################################################################
 
Tip: Filter by directory path e.g. /media app.js to search for public/media/app.js.
Tip: Use camelCasing e.g. ProjME to search for ProjectModifiedEvent.java.
Tip: Filter by extension type e.g. /repo .js to search for all .js files in the /repo directory.
Tip: Separate your search with spaces e.g. /ssh pom.xml to search for src/ssh/pom.xml.
Tip: Use ↑ and ↓ arrow keys to navigate and return to view the file.
Tip: You can also navigate files with Ctrl+j (next) and Ctrl+k (previous) and view the file with Ctrl+o.
Tip: You can also navigate files with Alt+j (next) and Alt+k (previous) and view the file with Alt+o.