Merge pull request #1331 from carlfeberhard/CORS

Core, CORS: allow (non-preflighted) CORS requests by echoing back the…

config/galaxy.ini.sample

@@ -172,7 +172,7 @@ paste.app_factory = galaxy.web.buildapp:app_factory
 # from the Tool Shed will fail.
 #tool_dependency_dir = None
 
-# The dependency resolves config file specifies an ordering and options for how 
+# The dependency resolves config file specifies an ordering and options for how
 # Galaxy resolves tool dependencies (requirement tags in Tool XML). The default
 # ordering is to the use the tool shed for tools installed that way, use local
 # Galaxy packages, and then use conda if available.
@@ -675,6 +675,14 @@ nglims_config_file = tool-data/nglims.yaml
 # by setting the following option to True.
 #serve_xss_vulnerable_mimetypes = False
 
+# Return a Access-Control-Allow-Origin response header that matches the Origin
+# header of the request if that Origin hostname matches one of the strings or
+# regular expressions listed here. This is a comma separated list of hostname
+# strings or regular expressions beginning and ending with /.
+# E.g. mysite.com,google.com,usegalaxy.org,/^[\w\.]*example\.com/
+# See: https://developer.mozilla.org/en-US/docs/Web/HTTP/Access_control_CORS
+#allowed_origin_hostnames = None
+
 # Set the following to True to use IPython nbconvert to build HTML from IPython
 # notebooks in Galaxy histories.  This process may allow users to execute
 # arbitrary code or serve arbitrary HTML.  If enabled, IPython must be
@@ -891,8 +899,8 @@ use_interactive = True
 # Set the following to a number of threads greater than 1 to spawn
 # a Python task queue for dealing with large tool submissions (either
 # through the tool form or as part of an individual workflow step across
-# large collection). The size of a "large" tool request is controlled by 
-# the second parameter below and defaults to 10. This affects workflow 
+# large collection). The size of a "large" tool request is controlled by
+# the second parameter below and defaults to 10. This affects workflow
 # scheduling and web processes, not job handlers.
 #tool_submission_burst_threads = 1
 #tool_submission_burst_at = 10

lib/galaxy/config.py

@@ -263,6 +263,7 @@ class Configuration( object ):
         if kwargs.get('sanitize_whitelist_file', None) is not None:
             self.reload_sanitize_whitelist()
         self.serve_xss_vulnerable_mimetypes = string_as_bool( kwargs.get( 'serve_xss_vulnerable_mimetypes', False ) )
+        self.allowed_origin_hostnames = self._parse_allowed_origin_hostnames( kwargs )
         self.trust_ipython_notebook_conversion = string_as_bool( kwargs.get( 'trust_ipython_notebook_conversion', False ) )
         self.enable_old_display_applications = string_as_bool( kwargs.get( "enable_old_display_applications", "True" ) )
         self.brand = kwargs.get( 'brand', None )
@@ -672,6 +673,24 @@ class Configuration( object ):
             port = None
         return port
 
+    def _parse_allowed_origin_hostnames( self, kwargs ):
+        """
+        Parse a CSV list of strings/regexp of hostnames that should be allowed
+        to use CORS and will be sent the Access-Control-Allow-Origin header.
+        """
+        allowed_origin_hostnames = listify( kwargs.get( 'allowed_origin_hostnames', None ) )
+        if not allowed_origin_hostnames:
+            return None
+
+        def parse( string ):
+            # a string enclosed in fwd slashes will be parsed as a regexp: e.g. /<some val>/
+            if string[0] == '/' and string[-1] == '/':
+                string = string[1:-1]
+                return re.compile( string, flags=( re.UNICODE | re.LOCALE ) )
+            return string
+
+        return [ parse( v ) for v in allowed_origin_hostnames if v ]
+
 
 def get_database_engine_options( kwargs, model_prefix='' ):
     """

lib/galaxy/web/framework/webapp.py

@@ -8,6 +8,7 @@ import random
 import socket
 import string
 import time
+import urlparse
 from Cookie import CookieError
 
 from Cheetah.Template import Template
@@ -184,6 +185,9 @@ class GalaxyWebTransaction( base.DefaultWebTransaction,
         self.galaxy_session = None
         self.error_message = None
 
+        # set any cross origin resource sharing headers if configured to do so
+        self.set_cors_headers()
+
         if self.environ.get('is_api_request', False):
             # With API requests, if there's a key, use it and associate the
             # user with the transaction.
@@ -255,6 +259,50 @@ class GalaxyWebTransaction( base.DefaultWebTransaction,
         t = Translations.load( dirname='locale', locales=locales, domain='ginga' )
         self.template_context.update( dict( _=t.ugettext, n_=t.ugettext, N_=t.ungettext ) )
 
+    def set_cors_headers( self ):
+        """Allow CORS requests if configured to do so by echoing back the request's
+        'Origin' header (if any) as the response header 'Access-Control-Allow-Origin'
+        """
+        # TODO: in order to use these, we need preflight to work, and to do that we
+        # need the OPTIONS method on all api calls (or everywhere we can POST/PUT)
+        # ALLOWED_METHODS = ( 'POST', 'PUT' )
+
+        # do not set any access control headers if not configured for it (common case)
+        if not self.app.config.get( 'allowed_origin_hostnames', None ):
+            return
+        # do not set any access control headers if there's no origin header on the request
+        origin_header = self.request.headers.get( "Origin", None )
+        if not origin_header:
+            return
+
+        # singular match
+        def matches_allowed_origin( origin, allowed_origin ):
+            if isinstance( allowed_origin, str ):
+                return origin == allowed_origin
+            match = allowed_origin.match( origin )
+            return match and match.group() == origin
+
+        # check for '*' or compare to list of allowed
+        def is_allowed_origin( origin ):
+            # localhost uses no origin header (== null)
+            if not origin:
+                return False
+            for allowed_origin in self.app.config.allowed_origin_hostnames:
+                if allowed_origin == '*' or matches_allowed_origin( origin, allowed_origin ):
+                    return True
+            return False
+
+        # boil origin header down to hostname
+        origin = urlparse.urlparse( origin_header ).hostname
+        # check against the list of allowed strings/regexp hostnames, echo original if cleared
+        if is_allowed_origin( origin ):
+            self.response.headers[ 'Access-Control-Allow-Origin' ] = origin_header
+            # TODO: see the to do on ALLOWED_METHODS above
+            # self.response.headers[ 'Access-Control-Allow-Methods' ] = ', '.join( ALLOWED_METHODS )
+
+        # NOTE: raising some errors (such as httpexceptions), will remove the header
+        # (e.g. client will get both cors error and 404 inside that)
+
     def get_user( self ):
         """Return the current user if logged in or None."""
         if self.galaxy_session:

test/unit/unittest_utils/galaxy_mock.py

@@ -19,11 +19,40 @@ class OpenObject( object ):
     pass
 
 
+def buildMockEnviron( **kwargs ):
+    environ = {
+        'CONTENT_LENGTH': '0',
+        'CONTENT_TYPE': '',
+        'HTTP_ACCEPT': '*/*',
+        'HTTP_ACCEPT_ENCODING': 'gzip, deflate',
+        'HTTP_ACCEPT_LANGUAGE': 'en-US,en;q=0.8,zh;q=0.5,ja;q=0.3',
+        'HTTP_CACHE_CONTROL': 'no-cache',
+        'HTTP_CONNECTION': 'keep-alive',
+        'HTTP_DNT': '1',
+        'HTTP_HOST': 'localhost:8000',
+        'HTTP_ORIGIN': 'http://localhost:8000',
+        'HTTP_PRAGMA': 'no-cache',
+        'HTTP_REFERER': 'http://localhost:8000',
+        'HTTP_USER_AGENT': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:43.0) Gecko/20100101 Firefox/43.0',
+        'PATH_INFO': '/',
+        'QUERY_STRING': '',
+        'REMOTE_ADDR': '127.0.0.1',
+        'REQUEST_METHOD': 'GET',
+        'SCRIPT_NAME': '',
+        'SERVER_NAME': '127.0.0.1',
+        'SERVER_PORT': '8080',
+        'SERVER_PROTOCOL': 'HTTP/1.1'
+    }
+    environ.update( **kwargs )
+    return environ
+
+
 class MockApp( object ):
 
-    def __init__( self, **kwargs ):
-        self.config = MockAppConfig( **kwargs )
+    def __init__( self, config=None, **kwargs ):
+        self.config = config or MockAppConfig( **kwargs )
         self.security = self.config.security
+        self.name = kwargs.get( 'name', 'galaxy' )
         self.object_store = objectstore.build_object_store_from_config( self.config )
         self.model = mapping.init( "/tmp", "sqlite:///:memory:", create_tables=True, object_store=self.object_store )
         self.security_agent = self.model.security_agent
@@ -37,6 +66,7 @@ class MockAppConfig( Bunch ):
     def __init__( self, root=None, **kwargs ):
         Bunch.__init__( self, **kwargs )
         self.security = security.SecurityHelper( id_secret='bler' )
+        self.use_remote_user = kwargs.get( 'use_remote_user', False )
         self.file_path = '/tmp'
         self.job_working_directory = '/tmp'
         self.new_file_path = '/tmp'
@@ -60,6 +90,7 @@ class MockWebapp( object ):
 
     def __init__( self, **kwargs ):
         self.name = kwargs.get( 'name', 'galaxy' )
+        self.security = security.SecurityHelper( id_secret='bler' )
 
 
 class MockTrans( object ):
@@ -75,6 +106,9 @@ class MockTrans( object ):
         self.security = self.app.security
         self.history = history
 
+        self.request = Bunch( headers={} )
+        self.response = Bunch( headers={} )
+
     def get_user( self ):
         if self.galaxy_session:
             return self.galaxy_session.user

test/unit/web/framework/test_webapp.py

@@ -0,0 +1,121 @@
+"""
+Unit tests for ``galaxy.web.framework.webapp``
+"""
+import os
+import imp
+import unittest
+
+import logging
+log = logging.getLogger( __name__ )
+
+test_utils = imp.load_source( 'test_utils',
+    os.path.join( os.path.dirname( __file__), '../../unittest_utils/utility.py' ) )
+import galaxy_mock
+
+import re
+from galaxy.web.framework import webapp as Webapp
+import galaxy.config
+
+
+class StubGalaxyWebTransaction( Webapp.GalaxyWebTransaction ):
+    def _ensure_valid_session( self, session_cookie, create=True ):
+        pass
+
+
+class CORSParsingMockConfig( galaxy_mock.MockAppConfig ):
+    # we can't use the actual Configuration for parsing*, so steal the parser for the mock instead
+    # *It causes problems when it's change to tempfile.tempdir persists across tests
+    _parse_allowed_origin_hostnames = galaxy.config.Configuration._parse_allowed_origin_hostnames.__func__
+
+    def __init__( self, **kwargs ):
+        super( CORSParsingMockConfig, self ).__init__( **kwargs )
+        self.allowed_origin_hostnames = self._parse_allowed_origin_hostnames( kwargs )
+
+
+class GalaxyWebTransaction_Headers_TestCase( test_utils.unittest.TestCase ):
+
+    def _new_trans( self, allowed_origin_hostnames=None ):
+        app = galaxy_mock.MockApp()
+        app.config = CORSParsingMockConfig(
+            allowed_origin_hostnames=allowed_origin_hostnames
+        )
+        webapp = galaxy_mock.MockWebapp()
+        environ = galaxy_mock.buildMockEnviron()
+        trans = StubGalaxyWebTransaction( environ, app, webapp )
+        return trans
+
+    def assert_cors_header_equals( self, headers, should_be ):
+        self.assertEqual( headers.get( 'access-control-allow-origin', None ), should_be )
+
+    def assert_cors_header_missing( self, headers ):
+        self.assertFalse( 'access-control-allow-origin' in headers )
+
+    def test_parse_allowed_origin_hostnames( self ):
+        """Should return a list of (possibly) mixed strings and regexps"""
+        config = CORSParsingMockConfig()
+
+        # falsy listify value should return None
+        self.assertEqual( config._parse_allowed_origin_hostnames({
+            "allowed_origin_hostnames" : ""
+        }), None )
+
+        # should parse regex if using fwd slashes, string otherwise
+        hostnames = config._parse_allowed_origin_hostnames({
+            "allowed_origin_hostnames" : "/host\d{2}/,geocities.com,miskatonic.edu"
+        })
+        self.assertTrue( isinstance( hostnames[0], re._pattern_type ) )
+        self.assertTrue( isinstance( hostnames[1], str ) )
+        self.assertTrue( isinstance( hostnames[2], str ) )
+
+    def test_default_set_cors_headers( self ):
+        """No CORS headers should be set (or even checked) by default"""
+        trans = self._new_trans( allowed_origin_hostnames=None )
+        self.assertTrue( isinstance( trans, Webapp.GalaxyWebTransaction ) )
+
+        trans.request.headers[ 'Origin' ] = 'http://lisaskelprecipes.pinterest.com?id=kelpcake'
+        trans.set_cors_headers()
+        self.assert_cors_header_missing( trans.response.headers )
+
+    def test_set_cors_headers( self ):
+        """Origin should be echo'd when it matches an allowed hostname"""
+        # an asterisk is a special 'allow all' string
+        trans = self._new_trans( allowed_origin_hostnames='*,beep.com' )
+        trans.request.headers[ 'Origin' ] = 'http://xxdarkhackerxx.disney.com'
+        trans.set_cors_headers()
+        self.assert_cors_header_equals( trans.response.headers, 'http://xxdarkhackerxx.disney.com' )
+
+        # subdomains should pass
+        trans = self._new_trans( allowed_origin_hostnames='something.com,/^[\w\.]*beep\.com/' )
+        trans.request.headers[ 'Origin' ] = 'http://boop.beep.com'
+        trans.set_cors_headers()
+        self.assert_cors_header_equals( trans.response.headers, 'http://boop.beep.com' )
+
+        # ports should work
+        trans = self._new_trans( allowed_origin_hostnames='somethingelse.com,/^[\w\.]*beep\.com/' )
+        trans.request.headers[ 'Origin' ] = 'http://boop.beep.com:8080'
+        trans.set_cors_headers()
+        self.assert_cors_header_equals( trans.response.headers, 'http://boop.beep.com:8080' )
+
+        # localhost should work
+        trans = self._new_trans( allowed_origin_hostnames='/localhost/' )
+        trans.request.headers[ 'Origin' ] = 'http://localhost:8080'
+        trans.set_cors_headers()
+        self.assert_cors_header_equals( trans.response.headers, 'http://localhost:8080' )
+
+        # spoofing shouldn't be easy
+        trans.response.headers = {}
+        trans.request.headers[ 'Origin' ] = 'http://localhost.badstuff.tv'
+        trans.set_cors_headers()
+        self.assert_cors_header_missing( trans.response.headers )
+
+        # unicode should work
+        trans = self._new_trans( allowed_origin_hostnames='/öbb\.at/' )
+        trans.request.headers[ 'Origin' ] = 'http://öbb.at'
+        trans.set_cors_headers()
+        self.assertEqual(
+            trans.response.headers[ 'access-control-allow-origin' ], 'http://öbb.at'
+        )
+
+
+if __name__ == '__main__':
+    unittest.main()