def testGrep(self):
class MockCallFlow(object):
def CallFlow(self, *args, **kwargs):
self.args = args
self.kwargs = kwargs
mock_call_flow = MockCallFlow()
with utils.Stubber(collectors.ArtifactCollectorFlow, "CallFlow",
mock_call_flow.CallFlow):
collect_flow = collectors.ArtifactCollectorFlow(None,
token=self.token)
collect_flow.args = mock.Mock()
collect_flow.args.ignore_interpolation_errors = False
kb = rdf_client.KnowledgeBase()
kb.MergeOrAddUser(rdf_client.User(username="******"))
kb.MergeOrAddUser(rdf_client.User(username="******"))
collect_flow.state["knowledge_base"] = kb
collect_flow.current_artifact_name = "blah"
collector = rdf_artifacts.ArtifactSource(
type=rdf_artifacts.ArtifactSource.SourceType.GREP,
attributes={
"paths": ["/etc/passwd"],
"content_regex_list": [r"^a%%users.username%%b$"]
})
collect_flow.Grep(collector, rdf_paths.PathSpec.PathType.TSK)
conditions = mock_call_flow.kwargs["conditions"]
self.assertEqual(len(conditions), 1)
regexes = conditions[0].contents_regex_match.regex.SerializeToString()
self.assertItemsEqual(regexes.split("|"),
["(^atest1b$)", "(^atest2b$)"])
self.assertEqual(mock_call_flow.kwargs["paths"], ["/etc/passwd"])
def testGrepRegexCombination(self):
collect_flow = collectors.ArtifactCollectorFlow(None, token=self.token)
self.assertEqual(collect_flow._CombineRegex([r"simple"]), "simple")
self.assertEqual(collect_flow._CombineRegex(["a", "b"]), "(a)|(b)")
self.assertEqual(collect_flow._CombineRegex(["a", "b", "c"]),
"(a)|(b)|(c)")
self.assertEqual(collect_flow._CombineRegex(["a|b", "[^_]b", "c|d"]),
"(a|b)|([^_]b)|(c|d)")
def testNewArtifactLoaded(self):
"""Simulate a new artifact being loaded into the store via the UI."""
cmd_artifact = """name: "TestCmdArtifact"
doc: "Test command artifact for dpkg."
sources:
- type: "COMMAND"
attributes:
cmd: "/usr/bin/dpkg"
args: ["--list"]
labels: [ "Software" ]
supported_os: [ "Linux" ]
"""
no_datastore_artifact = """name: "NotInDatastore"
doc: "Test command artifact for dpkg."
sources:
- type: "COMMAND"
attributes:
cmd: "/usr/bin/dpkg"
args: ["--list"]
labels: [ "Software" ]
supported_os: [ "Linux" ]
"""
test_registry = artifact_registry.ArtifactRegistry()
test_registry.ClearRegistry()
test_registry.AddDatastoreSource(rdfvalue.RDFURN("aff4:/artifact_store"))
test_registry._dirty = False
with utils.Stubber(artifact_registry, "REGISTRY", test_registry):
collect_flow = collectors.ArtifactCollectorFlow(None, token=self.token)
with self.assertRaises(artifact_registry.ArtifactNotRegisteredError):
artifact_registry.REGISTRY.GetArtifact("TestCmdArtifact")
with self.assertRaises(artifact_registry.ArtifactNotRegisteredError):
artifact_registry.REGISTRY.GetArtifact("NotInDatastore")
# Add artifact to datastore but not registry
artifact_coll = artifact_registry.ArtifactCollection(
rdfvalue.RDFURN("aff4:/artifact_store"))
with data_store.DB.GetMutationPool() as pool:
for artifact_val in artifact_registry.REGISTRY.ArtifactsFromYaml(
cmd_artifact):
artifact_coll.Add(artifact_val, mutation_pool=pool)
# Add artifact to registry but not datastore
for artifact_val in artifact_registry.REGISTRY.ArtifactsFromYaml(
no_datastore_artifact):
artifact_registry.REGISTRY.RegisterArtifact(
artifact_val, source="datastore", overwrite_if_exists=False)
# This should succeeded because the artifacts will be reloaded from the
# datastore.
self.assertTrue(collect_flow._GetArtifactFromName("TestCmdArtifact"))
# We registered this artifact with datastore source but didn't
# write it into aff4. This simulates an artifact that was
# uploaded in the UI then later deleted. We expect it to get
# cleared when the artifacts are reloaded from the datastore.
with self.assertRaises(artifact_registry.ArtifactNotRegisteredError):
artifact_registry.REGISTRY.GetArtifact("NotInDatastore")
def testInterpolateArgs(self):
collect_flow = collectors.ArtifactCollectorFlow(None, token=self.token)
kb = rdf_client.KnowledgeBase()
kb.MergeOrAddUser(rdf_client.User(username="******"))
kb.MergeOrAddUser(rdf_client.User(username="******"))
collect_flow.state["knowledge_base"] = kb
collect_flow.current_artifact_name = "blah"
collect_flow.args = artifact_utils.ArtifactCollectorFlowArgs()
test_rdf = rdf_client.KnowledgeBase()
action_args = {
"usernames": ["%%users.username%%", "%%users.username%%"],
"nointerp": "asdfsdf",
"notastring": test_rdf
}
kwargs = collect_flow.InterpolateDict(action_args)
self.assertItemsEqual(kwargs["usernames"],
["test1", "test2", "test1", "test2"])
self.assertEqual(kwargs["nointerp"], "asdfsdf")
self.assertEqual(kwargs["notastring"], test_rdf)
# We should be using an array since users.username will expand to multiple
# values.
self.assertRaises(ValueError, collect_flow.InterpolateDict,
{"bad": "%%users.username%%"})
list_args = collect_flow.InterpolateList(
["%%users.username%%", r"%%users.username%%\aa"])
self.assertItemsEqual(list_args,
["test1", "test2", r"test1\aa", r"test2\aa"])
list_args = collect_flow.InterpolateList(["one"])
self.assertEqual(list_args, ["one"])
# Ignore the failure in users.desktop, report the others.
collect_flow.args.ignore_interpolation_errors = True
list_args = collect_flow.InterpolateList(
["%%users.desktop%%", r"%%users.username%%\aa"])
self.assertItemsEqual(list_args, [r"test1\aa", r"test2\aa"])
# Both fail.
list_args = collect_flow.InterpolateList(
[r"%%users.desktop%%\aa", r"%%users.sid%%\aa"])
self.assertItemsEqual(list_args, [])