def testFindActionCrossDev(self):
"""Test that devices boundaries don't get crossed, also by default."""
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
request = rdf_client.FindSpec(pathspec=pathspec,
cross_devs=True,
path_regex=".")
request.iterator.number = 200
results = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in results if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 9)
request = rdf_client.FindSpec(pathspec=pathspec,
cross_devs=False,
path_regex=".")
request.iterator.number = 200
results = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in results if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 7)
request = rdf_client.FindSpec(pathspec=pathspec, path_regex=".")
request.iterator.number = 200
results = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in results if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 7)
def testCollectionOverwriting(self):
"""Test we overwrite the collection every time the flow is executed."""
client_mock = action_mocks.ActionMock(searching.Find)
# Prepare a findspec.
findspec = rdf_client.FindSpec()
findspec.path_regex = "bin"
findspec.pathspec.path = "/"
findspec.pathspec.pathtype = rdf_paths.PathSpec.PathType.OS
session_id = flow_test_lib.TestFlowHelper(find.FindFiles.__name__,
client_mock,
client_id=self.client_id,
token=self.token,
findspec=findspec)
# Check the results collection.
fd = flow.GRRFlow.ResultCollectionForFID(session_id)
self.assertEqual(len(fd), 2)
# Now find a new result, should overwrite the collection
findspec.path_regex = "dd"
session_id = flow_test_lib.TestFlowHelper(find.FindFiles.__name__,
client_mock,
client_id=self.client_id,
token=self.token,
findspec=findspec,
max_results=1)
# Check the results collection.
fd = flow.GRRFlow.ResultCollectionForFID(session_id)
self.assertEqual(len(fd), 1)
def testFindFilesWithGlob(self):
"""Test that the Find flow works with glob."""
client_mock = action_mocks.ActionMock(searching.Find)
# Prepare a findspec.
findspec = rdf_client.FindSpec(
path_glob="bash*",
pathspec=rdf_paths.PathSpec(
path="/", pathtype=rdf_paths.PathSpec.PathType.OS))
session_id = flow_test_lib.TestFlowHelper(find.FindFiles.__name__,
client_mock,
client_id=self.client_id,
token=self.token,
findspec=findspec)
# Check the results collection.
fd = flow.GRRFlow.ResultCollectionForFID(session_id)
# Make sure that bash is a file.
matches = set([x.AFF4Path(self.client_id).Basename() for x in fd])
self.assertEqual(sorted(matches), ["bash"])
self.assertEqual(len(fd), 2)
for child in fd:
path = utils.SmartStr(child.AFF4Path(self.client_id))
self.assertTrue(path.endswith("bash"))
self.assertEqual(child.__class__.__name__, "StatEntry")
def testFindDirectories(self):
"""Test that the Find flow works with directories."""
client_mock = action_mocks.ActionMock(searching.Find)
# Prepare a findspec.
findspec = rdf_client.FindSpec(
path_regex="bin",
pathspec=rdf_paths.PathSpec(
path="/", pathtype=rdf_paths.PathSpec.PathType.OS))
session_id = flow_test_lib.TestFlowHelper(find.FindFiles.__name__,
client_mock,
client_id=self.client_id,
token=self.token,
findspec=findspec)
# Check the results collection.
fd = flow.GRRFlow.ResultCollectionForFID(session_id)
# Make sure that bin is a directory
self.assertEqual(len(fd), 2)
for child in fd:
path = utils.SmartStr(child.AFF4Path(self.client_id))
self.assertTrue("bin" in path)
self.assertEqual(child.__class__.__name__, "StatEntry")
def testGIDFilter(self):
"""Test filtering based on gid happens correctly."""
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
# Look for files that have gid of 500
request = rdf_client.FindSpec(pathspec=pathspec,
path_regex=".",
gid=500,
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 2)
self.assertEqual(all_files[0].pathspec.Dirname().Basename(),
"directory2")
self.assertEqual(all_files[0].pathspec.Basename(), "file.jpg")
self.assertEqual(all_files[1].pathspec.Dirname().Basename(),
"directory2")
self.assertEqual(all_files[1].pathspec.Basename(), "file.mp3")
# Look for files that have uid of 900
request = rdf_client.FindSpec(pathspec=pathspec,
path_regex=".",
gid=900,
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 2)
self.assertEqual(all_files[0].pathspec.Dirname().Basename(),
"directory1")
self.assertEqual(all_files[0].pathspec.Basename(), "file1.txt")
self.assertEqual(all_files[1].pathspec.Dirname().Basename(),
"directory1")
self.assertEqual(all_files[1].pathspec.Basename(), "file2.txt")
def RunFind():
pathspec = rdf_paths.PathSpec(
path=self.base_path, pathtype=rdf_paths.PathSpec.PathType.OS)
request = rdf_client.FindSpec(pathspec=pathspec)
request.iterator.number = 80
result = self.RunAction(searching.Find, request)
# 80 results plus one iterator.
self.assertEqual(len(result), 81)
def ProcessEntry(self, responses):
"""Process the responses from the client."""
if not responses.success:
return
# If we get a response with an unfinished iterator then we missed some
# files. Call Find on the client until we're done.
if (responses.iterator and
responses.iterator.state != responses.iterator.State.FINISHED):
findspec = rdf_client.FindSpec(responses.request.request.payload)
findspec.iterator = responses.iterator
self.CallClient(server_stubs.Find,
findspec,
next_state="ProcessEntry",
request_data=responses.request_data)
# The Find client action does not return a StatEntry but a
# FindSpec. Normalize to a StatEntry.
stat_responses = [
r.hit if isinstance(r, rdf_client.FindSpec) else r
for r in responses
]
# If this was a pure path matching call without any regex / recursion, we
# know exactly which node in the component tree we have to process next and
# get it from the component_path. If this was a regex match though, we
# sent the client a combined regex that matches all nodes in order to save
# round trips and client processing time. In that case we only get the
# base node and have to check for all subnodes if the response actually
# matches that subnode before we continue processing.
component_path = responses.request_data.get("component_path")
if component_path is not None:
for response in stat_responses:
self._ProcessResponse(response, [component_path])
else:
# This is a combined match.
base_path = responses.request_data["base_path"]
base_node = self.FindNode(base_path)
for response in stat_responses:
matching_components = []
for next_node in base_node:
pathspec = rdf_paths.PathSpec.FromSerializedString(
next_node)
if self._MatchPath(pathspec, response):
matching_path = base_path + [next_node]
matching_components.append(matching_path)
if matching_components:
self._ProcessResponse(response,
matching_components,
base_wildcard=True)
def testNoFilters(self):
"""Test the we get all files with no filters in place."""
# First get all the files at once
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
request = rdf_client.FindSpec(pathspec=pathspec, cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 9)
def testFindAction2(self):
"""Test the find action path regex."""
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
request = rdf_client.FindSpec(pathspec=pathspec, path_regex=".*mp3")
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 1)
self.assertEqual(all_files[0].pathspec.Basename(), "file.mp3")
def testFindAction(self):
"""Test the find action."""
# First get all the files at once
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
request = rdf_client.FindSpec(pathspec=pathspec, path_regex=".")
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
# Ask for the files one at the time
files = []
request = rdf_client.FindSpec(pathspec=pathspec, path_regex=".")
request.iterator.number = 1
while True:
result = self.RunAction(searching.Find, request)
if request.iterator.state == rdf_client.Iterator.State.FINISHED:
break
self.assertEqual(len(result), 2)
self.assertTrue(isinstance(result[0], rdf_client.FindSpec))
self.assertTrue(isinstance(result[1], rdf_client.Iterator))
files.append(result[0].hit)
request.iterator = result[1].Copy()
for x, y in zip(all_files, files):
self.assertRDFValuesEqual(x, y)
# Make sure the iterator is finished
self.assertEqual(request.iterator.state,
rdf_client.Iterator.State.FINISHED)
# Ensure we remove old states from client_state
self.assertEqual(len(request.iterator.client_state.dat), 0)
def testFindAction3(self):
"""Test the find action data regex."""
# First get all the files at once
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
request = rdf_client.FindSpec(pathspec=pathspec,
data_regex="Secret",
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 2)
self.assertEqual(all_files[0].pathspec.Basename(), "file1.txt")
self.assertEqual(all_files[1].pathspec.Basename(), "long_file.text")
def testFileArtifactParser(self):
"""Test parsing a fake file artifact with a file parser."""
processor = config_file.CronAtAllowDenyParser()
source = rdf_artifact.ArtifactSource(
type=rdf_artifact.ArtifactSource.SourceType.FILE,
attributes={
"paths": ["VFSFixture/etc/passwd", "numbers.txt"],
})
paths = []
for path in source.attributes["paths"]:
paths.append(os.path.join(self.base_path, path))
stat_cache = utils.StatCache()
expanded_paths = []
opts = globbing.PathOpts(follow_links=True)
for path in paths:
for expanded_path in globbing.ExpandPath(path, opts):
expanded_paths.append(expanded_path)
results = []
for path in expanded_paths:
stat = stat_cache.Get(path, follow_symlink=True)
pathspec = rdf_paths.PathSpec(
pathtype=rdf_paths.PathSpec.PathType.OS,
path=client_utils.LocalPathToCanonicalPath(stat.GetPath()),
path_options=rdf_paths.PathSpec.Options.CASE_LITERAL)
response = rdf_client.FindSpec(pathspec=pathspec)
for res in artifact_collector.ParseResponse(processor, response, {}):
results.append(res)
self.assertEqual(len(results), 3)
self.assertTrue(
results[0]["filename"].endswith("test_data/VFSFixture/etc/passwd"))
self.assertIsInstance(results[0], rdf_protodict.AttributedDict)
self.assertEqual(len(results[0]["users"]), 3)
self.assertIsInstance(results[1], rdf_anomaly.Anomaly)
self.assertEqual(len(results[2]["users"]), 1000)
def testFindSizeLimits(self):
"""Test the find action size limits."""
# First get all the files at once
request = rdf_client.FindSpec(min_file_size=4,
max_file_size=15,
cross_devs=True)
request.pathspec.Append(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
request.iterator.number = 200
results = self.RunAction(searching.Find, request)
all_files = []
for result in results:
if isinstance(result, rdf_client.FindSpec):
all_files.append(result.hit.pathspec.Basename())
self.assertEqual(len(all_files), 5)
for filename in all_files:
# Our mock filesize is the length of the base filename, check all the
# files we got match the size criteria
self.assertTrue(4 <= len(filename) <= 15)
def testFindWithMaxFiles(self):
"""Test that the Find flow works when specifying proto directly."""
client_mock = action_mocks.ActionMock(searching.Find)
# Prepare a findspec.
findspec = rdf_client.FindSpec(
path_regex=".*",
pathspec=rdf_paths.PathSpec(
path="/", pathtype=rdf_paths.PathSpec.PathType.OS))
session_id = flow_test_lib.TestFlowHelper(find.FindFiles.__name__,
client_mock,
client_id=self.client_id,
token=self.token,
findspec=findspec,
iteration_count=3,
max_results=7)
# Check the output file is created
collection = flow.GRRFlow.ResultCollectionForFID(session_id)
# Make sure we got the right number of results.
self.assertEqual(len(collection), 7)
def testExtAttrsCollection(self):
with test_lib.AutoTempDirPath(remove_non_empty=True) as temp_dirpath:
foo_filepath = test_lib.TempFilePath(dir=temp_dirpath)
client_test_lib.SetExtAttr(foo_filepath,
name="user.quux",
value="foo")
bar_filepath = test_lib.TempFilePath(dir=temp_dirpath)
client_test_lib.SetExtAttr(bar_filepath,
name="user.quux",
value="bar")
baz_filepath = test_lib.TempFilePath(dir=temp_dirpath)
client_test_lib.SetExtAttr(baz_filepath,
name="user.quux",
value="baz")
request = rdf_client.FindSpec(pathspec=rdf_paths.PathSpec(
path=temp_dirpath, pathtype=rdf_paths.PathSpec.PathType.OS),
path_glob="*",
collect_ext_attrs=True)
request.iterator.number = 100
hits = []
for response in self.RunAction(searching.Find, request):
if isinstance(response, rdf_client.FindSpec):
hits.append(response.hit)
self.assertEqual(len(hits), 3)
values = []
for hit in hits:
self.assertEqual(len(hit.ext_attrs), 1)
values.append(hit.ext_attrs[0].value)
self.assertItemsEqual(values, ["foo", "bar", "baz"])
def Iterate(self, request, client_state):
"""Restores its way through the directory using an Iterator."""
self.request = request
filters = self.BuildChecks(request)
limit = request.iterator.number
# TODO(user): What is a reasonable measure of work here?
for count, f in enumerate(
self.ListDirectory(request.pathspec, client_state)):
self.Progress()
# Ignore this file if any of the checks fail.
if not any((check(f) for check in filters)):
self.SendReply(rdf_client.FindSpec(hit=f))
# We only check a limited number of files in each iteration. This might
# result in returning an empty response - but the iterator is not yet
# complete. Flows must check the state of the iterator explicitly.
if count >= limit - 1:
logging.debug("Processed %s entries, quitting", count)
return
# End this iterator
request.iterator.state = rdf_client.Iterator.State.FINISHED
def testInvalidFindSpec(self):
"""Test that its impossible to produce an invalid findspec."""
# The regular expression is not valid.
with self.assertRaises(re.error):
rdf_client.FindSpec(path_regex="[")
def _ProcessResponse(self, response, component_paths, base_wildcard=False):
for component_path in component_paths:
regexes_to_get = []
recursions_to_get = {}
node = self.FindNode(component_path)
if not node:
# Node is empty representing a leaf node - we found a hit - report it.
self.GlobReportMatch(response)
return
# There are further components in the tree - iterate over them.
for component_str, next_node in iteritems(node):
component = rdf_paths.PathSpec.FromSerializedString(component_str)
next_component = component_path + [component_str]
# If we reach this point, we are instructed to go deeper into the
# directory structure. We only want to actually do this if
# - the last response was a proper directory,
# - or it was a file (an image) that was explicitly given meaning
# no wildcards or groupings,
# - or process_non_regular_files was set.
#
# This reduces the number of TSK opens on the client that may
# sometimes lead to instabilities due to bugs in the library.
if response and (not (stat.S_ISDIR(response.st_mode) or
not base_wildcard or
self.state.process_non_regular_files)):
continue
if component.path_options == component.Options.RECURSIVE:
recursions_to_get.setdefault(component.recursion_depth,
[]).append(component)
elif component.path_options == component.Options.REGEX:
regexes_to_get.append(component)
elif component.path_options == component.Options.CASE_INSENSITIVE:
# Here we need to create the next pathspec by appending the current
# component to what we already have. If we don't have anything yet, we
# fall back to the root path. If there is no root path either, the
# current component becomes the new base.
base_pathspec = self._GetBasePathspec(response)
if base_pathspec:
pathspec = base_pathspec.Append(component)
else:
pathspec = component
if not next_node:
# Check for the existence of the last node.
if (response is None or (response and
(response.st_mode == 0 or
not stat.S_ISREG(response.st_mode)))):
# If next node is empty, this node is a leaf node, we therefore
# must stat it to check that it is there. There is a special case
# here where this pathspec points to a file/directory in the root
# directory. In this case, response will be None but we still need
# to stat it.
# TODO(hanuszczak): Support for old clients ends on 2021-01-01.
# This conditional should be removed after that date.
if self.client_version >= 3221:
stub = server_stubs.GetFileStat
request = rdf_client.GetFileStatRequest(
pathspec=pathspec,
collect_ext_attrs=self.state.collect_ext_attrs)
else:
stub = server_stubs.StatFile
request = rdf_client.ListDirRequest(pathspec=pathspec)
self.CallClient(
stub,
request,
next_state="ProcessEntry",
request_data=dict(component_path=next_component))
else:
# There is no need to go back to the client for intermediate
# paths in the prefix tree, just emulate this by recursively
# calling this state inline.
self.CallStateInline(
[rdf_client.StatEntry(pathspec=pathspec)],
next_state="ProcessEntry",
request_data=dict(component_path=next_component))
if recursions_to_get or regexes_to_get:
# Recursions or regexes need a base pathspec to operate on. If we
# have neither a response or a root path, we send a default pathspec
# that opens the root with pathtype "OS".
base_pathspec = self._GetBasePathspec(response)
if not base_pathspec:
base_pathspec = rdf_paths.PathSpec(path="/", pathtype="OS")
for depth, recursions in iteritems(recursions_to_get):
path_regex = "(?i)^" + "$|^".join(set([c.path for c in recursions
])) + "$"
findspec = rdf_client.FindSpec(
pathspec=base_pathspec,
cross_devs=True,
max_depth=depth,
path_regex=path_regex)
findspec.iterator.number = self.FILE_MAX_PER_DIR
self.CallClient(
server_stubs.Find,
findspec,
next_state="ProcessEntry",
request_data=dict(base_path=component_path))
if regexes_to_get:
path_regex = "(?i)^" + "$|^".join(
set([c.path for c in regexes_to_get])) + "$"
findspec = rdf_client.FindSpec(
pathspec=base_pathspec, max_depth=1, path_regex=path_regex)
findspec.iterator.number = self.FILE_MAX_PER_DIR
self.CallClient(
server_stubs.Find,
findspec,
next_state="ProcessEntry",
request_data=dict(base_path=component_path))
def testPermissionFilter(self):
"""Test filtering based on file/folder permission happens correctly."""
pathspec = rdf_paths.PathSpec(path="/mock2/",
pathtype=rdf_paths.PathSpec.PathType.OS)
# Look for files that match exact permissions
request = rdf_client.FindSpec(pathspec=pathspec,
path_regex=".",
perm_mode=0o644,
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 2)
self.assertEqual(all_files[0].pathspec.Dirname().Basename(),
"directory2")
self.assertEqual(all_files[0].pathspec.Basename(), "file.jpg")
self.assertEqual(all_files[1].pathspec.Dirname().Basename(),
"directory2")
self.assertEqual(all_files[1].pathspec.Basename(), "file.mp3")
# Look for files/folders where 'others' have 'write' permission. All other
# attributes don't matter. Setuid bit must also be set and guid or sticky
# bit must not be set.
request = rdf_client.FindSpec(pathspec=pathspec,
path_regex=".",
perm_mode=0o4002,
perm_mask=0o7002,
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 2)
self.assertEqual(all_files[0].pathspec.Dirname().Basename(),
"directory1")
self.assertEqual(all_files[0].pathspec.Basename(), "file1.txt")
self.assertEqual(all_files[1].pathspec.Dirname().Basename(),
"directory1")
self.assertEqual(all_files[1].pathspec.Basename(), "file2.txt")
# Look for files where 'others' have 'execute' permission. All other
# attributes don't matter. Only look for 'regular' files.
request = rdf_client.FindSpec(pathspec=pathspec,
path_regex=".",
perm_mode=0o0100001,
perm_mask=0o0100001,
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 2)
self.assertEqual(all_files[0].pathspec.Dirname().Basename(),
"directory3")
self.assertEqual(all_files[0].pathspec.Basename(), "file1.txt")
self.assertEqual(all_files[1].pathspec.Dirname().Basename(),
"directory3")
self.assertEqual(all_files[1].pathspec.Basename(), "long_file.text")
# Look for folders where 'group' have 'execute' permission. All other
# attributes don't matter. Only look for folders.
request = rdf_client.FindSpec(pathspec=pathspec,
path_regex=".",
perm_mode=0o0040010,
perm_mask=0o0040010,
cross_devs=True)
request.iterator.number = 200
result = self.RunAction(searching.Find, request)
all_files = [
x.hit for x in result if isinstance(x, rdf_client.FindSpec)
]
self.assertEqual(len(all_files), 3)
self.assertEqual(all_files[0].pathspec.Basename(), "directory2")
self.assertEqual(all_files[1].pathspec.Basename(), "directory1")
self.assertEqual(all_files[2].pathspec.Basename(), "directory3")
