project_loader: handle invalid unicode chars

snapcraft/internal/project_loader/_config.py

@@ -22,6 +22,8 @@
 
 import jsonschema
 import yaml
+import yaml.reader
+
 
 import snapcraft
 from snapcraft.internal import common, deprecations, remote_parts, states
@@ -262,12 +264,21 @@ def _snapcraft_yaml_load(yaml_file):
     else:
         encoding = 'utf-8'
 
+    # Work-around for https://github.com/yaml/pyyaml/issues/25
+    yaml.reader.Reader.NON_PRINTABLE = re.compile(
+        u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-'
+        '\uD7FF\uE000-\uFFFD\U00010000-\U0010FFFF]')
+
     try:
         with open(yaml_file, encoding=encoding) as fp:
             return yaml.load(fp)
     except yaml.scanner.ScannerError as e:
         raise errors.YamlValidationError('{} on line {} of {}'.format(
             e.problem, e.problem_mark.line + 1, yaml_file)) from e
+    except yaml.reader.ReaderError as e:
+        raise errors.YamlValidationError(
+            'Invalid character {!r} at position {} of {}: {}'.format(
+                chr(e.character), e.position + 1, yaml_file, e.reason)) from e
 
 
 def _ensure_confinement_default(yaml_data, schema):

snapcraft/tests/unit/project_loader/test_config.py

@@ -479,6 +479,41 @@ def test_invalid_yaml_missing_icon(self):
         self.assertThat(raised.message,
                         Equals("Specified icon 'icon.png' does not exist"))
 
+    def test_invalid_yaml_invalid_unicode_chars(self):
+        fake_logger = fixtures.FakeLogger(level=logging.ERROR)
+        self.useFixture(fake_logger)
+
+        self.make_snapcraft_yaml("""name: foobar
+version: "1"
+summary: test\uffff
+description: nothing
+""")
+        raised = self.assertRaises(
+            errors.YamlValidationError,
+            _config.Config)
+
+        self.assertThat(raised.message, Equals(
+            "Invalid character '\\uffff' at position 40 "
+            "of snap/snapcraft.yaml: special characters are not allowed"))
+
+    def test_invalid_yaml_invalid_unicode_workaround(self):
+        fake_logger = fixtures.FakeLogger(level=logging.ERROR)
+        self.useFixture(fake_logger)
+
+        self.make_snapcraft_yaml("""name: foobar
+version: "1"
+summary: test💩
+description: nothing
+
+parts:
+  part1:
+    plugin: go
+    stage-packages: [fswebcam]
+""")
+        c = _config.Config()
+
+        self.assertThat(c.data['summary'], Equals('test💩'))
+
     def test_invalid_yaml_invalid_name_chars(self):
         fake_logger = fixtures.FakeLogger(level=logging.ERROR)
         self.useFixture(fake_logger)