def test_nested_censoring(self):
example_handles = [
ZipHandle(
ZipSource(
SMBCHandle(
SMBCSource(
"//SERVER/Resource",
"username", driveletter="W"),
"Confidential Documents.zip")),
"doc/Personal Information.docx"),
FilteredHandle(
GzipSource(
SMBHandle(
SMBSource(
"//SERVER/usr", "username"),
"share/doc/coreutils"
"/changelog.Debian.gz")),
"changelog.Debian"),
]
for handle in example_handles:
with self.subTest(handle):
self.assertIsNotNone(handle.source.handle.source._user)
handle = handle.censor()
self.assertIsNone(handle.source.handle.source._user)
def test_smb_censoring(self):
example_handles = [
SMBHandle(
SMBSource(
"//SERVER/Resource", "username"),
"~ocument.docx"),
SMBCHandle(
SMBCSource(
"//SERVER/Resource",
"username", "topsecret", "WORKGROUP8"),
"~ocument.docx"),
]
for handle in example_handles:
with self.subTest(handle):
handle = handle.censor()
self.assertIsNone(handle.source._domain)
self.assertIsNone(handle.source._password)
self.assertIsNone(handle.source._user)
def make_engine2_source(self):
return SMBCSource(
self.url,
user=self.authentication.username,
password=self.authentication.get_password(),
domain=self.authentication.domain)
def test_sources(self):
sources_and_urls = [
(FilesystemSource("/usr"), "file:///usr"),
(
SMBSource("//10.0.0.30/Share$/Documents"),
"smb://10.0.0.30/Share%24/Documents",
),
(
SMBSource("//10.0.0.30/Share$/Documents", "FaithfullA"),
"smb://[email protected]/Share%24/Documents",
),
(
SMBSource(
"//10.0.0.30/Share$/Documents",
"FaithfullA",
"secretpassword",
),
"smb://*****:*****@10.0.0.30/Share%24/Documents",
),
(
SMBSource(
"//10.0.0.30/Share$/Documents",
"FaithfullA",
"secretpassword",
"SYSGRP",
),
"smb://SYSGRP;FaithfullA:[email protected]/Share%24"
"/Documents",
),
(
SMBSource(
"//10.0.0.30/Share$/Documents",
"FaithfullA",
None,
"SYSGRP",
),
"smb://SYSGRP;[email protected]/Share%24/Documents",
),
(
SMBCSource(
"//INT-SRV-01/Q$",
"FaithfullA",
None,
"SYSGRP",
),
"smbc://SYSGRP;FaithfullA@INT-SRV-01/Q%24",
),
(WebSource("http://www.example.com"), "http://www.example.com"),
(
SecureWebSource("https://www.example.com"),
"https://www.example.com",
),
(
DataSource(b"This is a test", "text/plain"),
"data:text/plain;base64,VGhpcyBpcyBhIHRlc3Q=",
),
]
for source, url in sources_and_urls:
with self.subTest(url):
generated_url = source.to_url()
self.assertEqual(url, generated_url)
def test_json_round_trip(self):
example_handles = [
FilesystemHandle(FilesystemSource("/usr/share/common-licenses"),
"GPL-3"),
DataHandle(DataSource(b"Test", "text/plain"), "file"),
FilteredHandle(
GzipSource(
FilesystemHandle(
FilesystemSource("/usr/share/doc/coreutils"),
"changelog.Debian.gz")), "changelog.Debian"),
SMBHandle(SMBSource("//SERVER/Resource", "username"),
"~ocument.docx"),
SMBCHandle(
SMBCSource("//SERVER/Resource", "username", "topsecret",
"WORKGROUP8"), "~ocument.docx"),
ZipHandle(
ZipSource(
SMBCHandle(
SMBCSource("//SERVER/Resource",
"username",
driveletter="W"),
"Confidential Documents.zip")),
"doc/Personal Information.docx"),
WebHandle(WebSource("https://secret.data.invalid/"),
"lottery-numbers-for-next-week.txt"),
TarHandle(
TarSource(
FilesystemHandle(FilesystemSource("/home/user"),
"Downloads/data.tar.gz")), "data0.txt"),
MailPartHandle(
MailSource(
EWSMailHandle(
EWSAccountSource(domain="cloudy.example",
server=CLOUD,
admin_user="******",
admin_password="******",
user="******"),
"SW5ib3hJRA==.TWVzc2dJRA==",
"Re: Castles in the sky")), "1/pictograph.jpeg",
"image/jpeg"),
PDFObjectHandle(
PDFPageSource(
PDFPageHandle(
PDFSource(
FilesystemHandle(
FilesystemSource("/home/kiddw"
"/Documents"),
"1699 Gardiners trip/"
"treasure_map.pdf")), "10")),
"X-marks-the-spot_000-0.png"),
LibreOfficeObjectHandle(
LibreOfficeSource(
FilesystemHandle(FilesystemSource("/media/user/USB STICK"),
"What I Did On My Holidays.doc")),
"What I Did On My Holidays.html")
]
for handle in example_handles:
with self.subTest(handle):
json = handle.to_json_object()
print(handle)
print(json)
self.assertEqual(handle, handle.from_json_object(json))
print("--")
def generate_sources(self):
yield SMBCSource(self.url,
user=self.authentication.username,
password=self.authentication.get_password(),
domain=self.authentication.domain,
driveletter=self.alias)
from os2datascanner.engine2.model.derived.pdf import (PDFSource, PDFPageHandle,
PDFPageSource,
PDFObjectHandle)
from os2datascanner.engine2.model.derived.tar import TarSource, TarHandle
from os2datascanner.engine2.model.derived.zip import ZipSource, ZipHandle
example_handles = [
FilesystemHandle(FilesystemSource("/usr/share/common-licenses"), "GPL-3"),
DataHandle(DataSource(b"Test", "text/plain"), "file"),
FilteredHandle(
GzipSource(
FilesystemHandle(FilesystemSource("/usr/share/doc/coreutils"),
"changelog.Debian.gz")), "changelog.Debian"),
SMBHandle(SMBSource("//SERVER/Resource", "username"), "~ocument.docx"),
SMBCHandle(
SMBCSource("//SERVER/Resource", "username", "topsecret", "WORKGROUP8"),
"~ocument.docx"),
ZipHandle(
ZipSource(
SMBCHandle(
SMBCSource("//SERVER/Resource", "username", driveletter="W"),
"Confidential Documents.zip")),
"doc/Personal Information.docx"),
WebHandle(WebSource("https://secret.data.invalid/"),
"lottery-numbers-for-next-week.txt"),
TarHandle(
TarSource(
FilesystemHandle(FilesystemSource("/home/user"),
"Downloads/data.tar.gz")), "data0.txt"),
MailPartHandle(
MailSource(