Don't throw exceptions for empty cloud config (#1130)

Warn during boot when an empty config is provided. Likewise,
`cloud-init devel schema --annotate` should not throw exception, return
something meaningful instead.

3 files changed, 67 insertions, 7 deletions

diff --git a/cloudinit/config/schema.py b/cloudinit/config/schema.py
index 8ec4ab6a..3a77ca00 100644
--- a/cloudinit/config/schema.py
+++ b/cloudinit/config/schema.py
@@ -181,6 +181,7 @@ def validate_cloudconfig_schema(
 
     @raises: SchemaValidationError when provided config does not validate
         against the provided schema.
+    @raises: RuntimeError when provided config sourced from YAML is not a dict.
     """
     try:
         (cloudinitValidator, FormatChecker) = get_jsonschema_validator()
@@ -217,13 +218,21 @@ def annotated_cloudconfig_file(cloudconfig, original_content, schema_errors):
     if not schema_errors:
         return original_content
     schemapaths = {}
+    errors_by_line = defaultdict(list)
+    error_footer = []
+    error_header = "# Errors: -------------\n{0}\n\n"
+    annotated_content = []
+    lines = original_content.decode().split("\n")
+    if not isinstance(cloudconfig, dict):
+        # Return a meaningful message on empty cloud-config
+        return "\n".join(
+            lines
+            + [error_header.format("# E1: Cloud-config is not a YAML dict.")]
+        )
     if cloudconfig:
         schemapaths = _schemapath_for_cloudconfig(
             cloudconfig, original_content
         )
-    errors_by_line = defaultdict(list)
-    error_footer = []
-    annotated_content = []
     for path, msg in schema_errors:
         match = re.match(r"format-l(?P<line>\d+)\.c(?P<col>\d+).*", path)
         if match:
@@ -236,7 +245,6 @@ def annotated_cloudconfig_file(cloudconfig, original_content, schema_errors):
             msg = "Line {line} column {col}: {msg}".format(
                 line=line, col=col, msg=msg
             )
-    lines = original_content.decode().split("\n")
     error_index = 1
     for line_number, line in enumerate(lines, 1):
         errors = errors_by_line[line_number]
@@ -247,11 +255,10 @@ def annotated_cloudconfig_file(cloudconfig, original_content, schema_errors):
                 error_footer.append("# E{0}: {1}".format(error_index, error))
                 error_index += 1
             annotated_content.append(line + "\t\t# " + ",".join(error_label))
+
         else:
             annotated_content.append(line)
-    annotated_content.append(
-        "# Errors: -------------\n{0}\n\n".format("\n".join(error_footer))
-    )
+    annotated_content.append(error_header.format("\n".join(error_footer)))
     return "\n".join(annotated_content)
 
 
@@ -318,6 +325,10 @@ def validate_cloudconfig_file(config_path, schema, annotate=False):
         if annotate:
             print(annotated_cloudconfig_file({}, content, error.schema_errors))
         raise error from e
+    if not isinstance(cloudconfig, dict):
+        # Return a meaningful message on empty cloud-config
+        if not annotate:
+            raise RuntimeError("Cloud-config is not a YAML dict.")
     try:
         validate_cloudconfig_schema(cloudconfig, schema, strict=True)
     except SchemaValidationError as e:
@@ -662,6 +673,8 @@ def handle_schema_args(name, args):
     exclusive_args = [args.config_file, args.docs, args.system]
     if len([arg for arg in exclusive_args if arg]) != 1:
         error("Expected one of --config-file, --system or --docs arguments")
+    if args.annotate and args.docs:
+        error("Invalid flag combination. Cannot use --annotate with --docs")
     full_schema = get_schema()
     if args.config_file or args.system:
         try:
diff --git a/cloudinit/handlers/cloud_config.py b/cloudinit/handlers/cloud_config.py
index 2e694671..8070c6cb 100644
--- a/cloudinit/handlers/cloud_config.py
+++ b/cloudinit/handlers/cloud_config.py
@@ -89,6 +89,9 @@ class CloudConfigPartHandler(handlers.Handler):
         # or the merge type from the headers or default to our own set
         # if neither exists (or is empty) from the later.
         payload_yaml = util.load_yaml(payload)
+        if payload_yaml is None:
+            raise ValueError("empty cloud config")
+
         mergers_yaml = mergers.dict_extract_mergers(payload_yaml)
         mergers_header = mergers.string_extract_mergers(merge_header_headers)
         all_mergers = []
@@ -139,6 +142,12 @@ class CloudConfigPartHandler(handlers.Handler):
             for i in ("\n", "\r", "\t"):
                 filename = filename.replace(i, " ")
             self.file_names.append(filename.strip())
+        except ValueError as err:
+            LOG.warning(
+                "Failed at merging in cloud config part from %s: %s",
+                filename,
+                err,
+            )
         except Exception:
             util.logexc(
                 LOG, "Failed at merging in cloud config part from %s", filename
diff --git a/tests/unittests/config/test_schema.py b/tests/unittests/config/test_schema.py
index fb5b891d..822efe5a 100644
--- a/tests/unittests/config/test_schema.py
+++ b/tests/unittests/config/test_schema.py
@@ -546,6 +546,31 @@ class AnnotatedCloudconfigFileTest(CiTestCase):
             content, annotated_cloudconfig_file({}, content, schema_errors=[])
         )
 
+    def test_annotated_cloudconfig_file_with_non_dict_cloud_config(self):
+        """Error when empty non-dict cloud-config is provided.
+
+        OurJSON validation when user-data is None type generates a bunch
+        schema validation errors of the format:
+        ('', "None is not of type 'object'"). Ignore those symptoms and
+        report the general problem instead.
+        """
+        content = b"\n\n\n"
+        expected = "\n".join(
+            [
+                content.decode(),
+                "# Errors: -------------",
+                "# E1: Cloud-config is not a YAML dict.\n\n",
+            ]
+        )
+        self.assertEqual(
+            expected,
+            annotated_cloudconfig_file(
+                None,
+                content,
+                schema_errors=[("", "None is not of type 'object'")],
+            ),
+        )
+
     def test_annotated_cloudconfig_file_schema_annotates_and_adds_footer(self):
         """With schema_errors, error lines are annotated and a footer added."""
         content = dedent(
@@ -658,6 +683,19 @@ class TestMain:
         _out, err = capsys.readouterr()
         assert "Error:\nConfigfile NOT_A_FILE does not exist\n" == err
 
+    def test_main_invalid_flag_combo(self, capsys):
+        """Main exits non-zero when invalid flag combo used."""
+        myargs = ["mycmd", "--annotate", "--docs", "DOES_NOT_MATTER"]
+        with mock.patch("sys.argv", myargs):
+            with pytest.raises(SystemExit) as context_manager:
+                main()
+        assert 1 == context_manager.value.code
+        _, err = capsys.readouterr()
+        assert (
+            "Error:\nInvalid flag combination. "
+            "Cannot use --annotate with --docs\n" == err
+        )
+
     def test_main_prints_docs(self, capsys):
         """When --docs parameter is provided, main generates documentation."""
         myargs = ["mycmd", "--docs", "all"]