From 538b7fa167549dbe8ea6a50e585b92abe502a9ff Mon Sep 17 00:00:00 2001
From: Chris Lenk <clenk@mitre.org>
Date: Wed, 3 Apr 2024 13:26:34 -0400
Subject: [PATCH] Fix unicode errors when validating files

Previously undetected because testing performed on systems where utf-8
is default.
---
 stix2validator/test/v21/misc_tests.py                    | 7 +++++++
 .../test/v21/test_examples/identity_unicode.json         | 9 +++++++++
 stix2validator/validator.py                              | 2 +-
 3 files changed, 17 insertions(+), 1 deletion(-)
 create mode 100644 stix2validator/test/v21/test_examples/identity_unicode.json

diff --git a/stix2validator/test/v21/misc_tests.py b/stix2validator/test/v21/misc_tests.py
index 5451675c..33cc5d14 100644
--- a/stix2validator/test/v21/misc_tests.py
+++ b/stix2validator/test/v21/misc_tests.py
@@ -20,6 +20,8 @@
                         'test_examples', 'identity.json')
 IDENTITY_CUSTOM = os.path.join(os.path.dirname(os.path.realpath(__file__)),
                                'test_examples', 'identity_custom.json')
+IDENTITY_UNICODE = os.path.join(os.path.dirname(os.path.realpath(__file__)),
+                                'test_examples', 'identity_unicode.json')
 INVALID_BRACES = os.path.join(os.path.dirname(os.path.realpath(__file__)),
                               'test_examples', 'invalid_braces.json')
 INVALID_COMMA = os.path.join(os.path.dirname(os.path.realpath(__file__)),
@@ -71,6 +73,11 @@ def test_validate_file_warning(caplog):
     assert re.search("Custom property .+ should ", caplog.text)
 
 
+def test_validate_file_unicode(caplog):
+    results = validate_file(IDENTITY_UNICODE)
+    assert results.is_valid
+
+
 def test_validate_file_invalid_brace(caplog):
     results = validate_file(INVALID_BRACES)
     assert not results.is_valid
diff --git a/stix2validator/test/v21/test_examples/identity_unicode.json b/stix2validator/test/v21/test_examples/identity_unicode.json
new file mode 100644
index 00000000..ae3c9a29
--- /dev/null
+++ b/stix2validator/test/v21/test_examples/identity_unicode.json
@@ -0,0 +1,9 @@
+{
+  "type": "identity",
+  "spec_version": "2.1",
+  "id": "identity--8c6af861-7b20-41ef-9b59-6344fd872a8f",
+  "created": "2016-08-08T15:50:10.983Z",
+  "modified": "2016-08-08T15:50:10.983Z",
+  "name": "Heizölrückstoßabdämpfung",
+  "identity_class": "organization"
+}
diff --git a/stix2validator/validator.py b/stix2validator/validator.py
index f4983087..982f3562 100644
--- a/stix2validator/validator.py
+++ b/stix2validator/validator.py
@@ -453,7 +453,7 @@ def validate_file(fn, options=None):
         options = ValidationOptions(files=fn)
 
     try:
-        with open(fn) as instance_file:
+        with open(fn, encoding="utf-8") as instance_file:
             file_results.object_results = validate(instance_file, options)
 
     except Exception as ex: