Update jsonschema validator (#226)

* Update jsonschema validator

Update jsonschema validator to support newer drafts. It now supports
drafts 2019-09 and 2020-12.

The logic for choosing the jsonschema validator class has changed so
that the following enhancements are available:

- When no draft is explicitly specified we now use use the validator
  draft that is specified in the "$schema" field of the criteria. This
  is done by the jsonschema module by default and should support
  possible future drafts without any changes to this code.

- Optionally allow to disable format checks in the code. As format
  checks are not required by the spec there might be situations where
  people want to disable them.

* Update requirements.txt

* Skip all tests which dependas on jsonschema 4.5

* jsonschema: Refactor code to support python 3.6

* Fix jsonschema requirements for python<3.7

* jsonschema: Update code for compatibility

* Better documentation and error handling for missing schema specifications

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Ashwini Mhatre <amhatre@redhat.com>
Co-authored-by: ashwini-mhatre <mashu97@gmail.com>

changelogs/fragments/validate_jsonschema_update.yaml

@@ -0,0 +1,5 @@
+---
+minor_changes:
+  - validate - Add support for JSON Schema draft 2019-09 and 2020-12 as well as automatically choosing the draft
+    from the `$schema` field of the criteria.
+  - validate - Add option `check_format` for the jsonschema engine to disable JSON Schema format checking.

galaxy.yml

@@ -13,4 +13,4 @@ issues: https://github.com/ansible-collections/ansible.utils/issues
 tags: [linux, networking, security, cloud, utilities, data, validation, utils]
 # NOTE(pabelanger): We create an empty version key to keep ansible-galaxy
 # happy. We dynamically inject version info based on git information.
-version: 2.9.1-dev
+version: 2.10.0-dev

plugins/sub_plugins/validate/jsonschema.py

@@ -22,17 +22,29 @@ DOCUMENTATION = """
         description:
         - This option provides the jsonschema specification that should be used
           for the validating the data. The I(criteria) option in the validate
-          plugin should follow the specification as mentioned by this option
+          plugin should follow the specification as mentioned by this option.
-        default: draft7
+          If this option is not specified, jsonschema will use the best validator
+          for the I($schema) field in the criteria. Specifications 2019-09 and
+          2020-12 are only available from jsonschema version 4.0 onwards.
         choices:
         - draft3
         - draft4
         - draft6
         - draft7
+        - 2019-09
+        - 2020-12
         env:
         - name: ANSIBLE_VALIDATE_JSONSCHEMA_DRAFT
         vars:
         - name: ansible_validate_jsonschema_draft
+      check_format:
+        description: If enabled, validate the I(format) specification in the criteria.
+        type: bool
+        default: true
+        env:
+        - name: ANSIBLE_VALIDATE_JSONSCHEMA_CHECK_FORMAT
+        vars:
+        - name: ansible_validate_jsonschema_check_format
     notes:
     - The value of I(data) option should be either a valid B(JSON) object or a B(JSON) string.
     - The value of I(criteria) should be B(list) of B(dict) or B(list) of B(strings) and each
@@ -45,11 +57,14 @@ from ansible.errors import AnsibleError
 from ansible.module_utils._text import to_text
 from ansible.module_utils.basic import missing_required_lib
 from ansible.module_utils.six import string_types
+from ansible.utils.display import Display
 
 from ansible_collections.ansible.utils.plugins.module_utils.common.utils import to_list
 from ansible_collections.ansible.utils.plugins.plugin_utils.base.validate import ValidateBase
 
 
+display = Display()
+
 # PY2 compatibility for JSONDecodeError
 try:
     from json.decoder import JSONDecodeError
@@ -58,6 +73,7 @@ except ImportError:
 
 try:
     import jsonschema
+    import jsonschema.validators
 
     HAS_JSONSCHEMA = True
 except ImportError:
@@ -79,6 +95,34 @@ def json_path(absolute_path):
 
 
 class Validate(ValidateBase):
+    # All available schema versions with the format_check and validator class names.
+    _JSONSCHEMA_DRAFTS = {
+        "draft3": {
+            "validator_name": "Draft3Validator",
+            "format_checker_name": "draft3_format_checker",
+        },
+        "draft4": {
+            "validator_name": "Draft4Validator",
+            "format_checker_name": "draft4_format_checker",
+        },
+        "draft6": {
+            "validator_name": "Draft6Validator",
+            "format_checker_name": "draft6_format_checker",
+        },
+        "draft7": {
+            "validator_name": "Draft7Validator",
+            "format_checker_name": "draft7_format_checker",
+        },
+        "2019-09": {
+            "validator_name": "Draft201909Validator",
+            "format_checker_name": "draft201909_format_checker",
+        },
+        "2020-12": {
+            "validator_name": "Draft202012Validator",
+            "format_checker_name": "draft202012_format_checker",
+        },
+    }
+
     @staticmethod
     def _check_reqs():
         """Check the prerequisites are installed for jsonschema
@@ -126,6 +170,28 @@ class Validate(ValidateBase):
             )
             raise AnsibleError(msg)
 
+    def _check_drafts(self):
+        """For every possible draft check if our jsonschema version supports it and exchange the class names with
+        the actual classes. If it is not supported the draft is removed from the list.
+        """
+        for draft in list(self._JSONSCHEMA_DRAFTS.keys()):
+            draft_config = self._JSONSCHEMA_DRAFTS[draft]
+            try:
+                validator_class = getattr(jsonschema, draft_config["validator_name"])
+            except AttributeError:
+                display.vvv(
+                    'jsonschema draft "{draft}" not supported in this version'.format(draft=draft),
+                )
+                del self._JSONSCHEMA_DRAFTS[draft]
+                continue
+            draft_config["validator"] = validator_class
+            try:
+                format_checker_class = validator_class.FORMAT_CHECKER
+            except AttributeError:
+                # Older jsonschema version
+                format_checker_class = getattr(jsonschema, draft_config["format_checker_name"])
+            draft_config["format_checker"] = format_checker_class
+
     def validate(self):
         """Std entry point for a validate execution
 
@@ -141,6 +207,7 @@ class Validate(ValidateBase):
         """
         self._check_reqs()
         self._check_args()
+        self._check_drafts()
 
         try:
             self._validate_jsonschema()
@@ -153,30 +220,45 @@ class Validate(ValidateBase):
         error_messages = None
 
         draft = self._get_sub_plugin_options("draft")
+        check_format = self._get_sub_plugin_options("check_format")
         error_messages = []
 
         for criteria in self._criteria:
-            if draft == "draft3":
+            format_checker = None
-                validator = jsonschema.Draft3Validator(
+            validator_class = None
-                    criteria,
+            if draft is not None:
-                    format_checker=jsonschema.draft3_format_checker,
+                try:
-                )
+                    validator_class = self._JSONSCHEMA_DRAFTS[draft]["validator"]
-            elif draft == "draft4":
+                except KeyError:
-                validator = jsonschema.Draft4Validator(
+                    display.warning(
-                    criteria,
+                        'No validator available for "{draft}", falling back to autodetection. A newer version of jsonschema might support this draft.'.format(
-                    format_checker=jsonschema.draft4_format_checker,
+                            draft=draft,
-                )
+                        ),
-            elif draft == "draft6":
-                validator = jsonschema.Draft6Validator(
-                    criteria,
-                    format_checker=jsonschema.draft6_format_checker,
-                )
-            else:
-                validator = jsonschema.Draft7Validator(
-                    criteria,
-                    format_checker=jsonschema.draft7_format_checker,
                     )
+            if validator_class is None:
+                # Either no draft was specified or specified draft has no validator class
+                # in installed jsonschema version. Do autodetection instead.
+                validator_class = jsonschema.validators.validator_for(criteria)
 
+            if check_format:
+                try:
+                    format_checker = validator_class.FORMAT_CHECKER
+                except AttributeError:
+                    # TODO: Remove when Python 3.6 support is dropped.
+                    # On jsonschema<4.5, there is no connection between a validator and the correct format checker.
+                    # So we iterate through our known list of validators and if one matches the current class
+                    # we use the format_checker from that validator.
+                    for draft, draft_config in self._JSONSCHEMA_DRAFTS.items():
+                        if validator_class == draft_config["validator"]:
+                            display.vvv(
+                                "Using format_checker for {draft} validator".format(draft=draft),
+                            )
+                            format_checker = draft_config["format_checker"]
+                            break
+                    else:
+                        display.warning("jsonschema format checks not available")
+
+            validator = validator_class(criteria, format_checker=format_checker)
             validation_errors = sorted(validator.iter_errors(self._data), key=lambda e: e.path)
 
             if validation_errors:

tests/integration/requirements.txt

@@ -1,4 +1,4 @@
-jsonschema ; python_version >= '2.7'
+jsonschema
 netaddr
 ttp
 textfsm

tests/integration/targets/utils_validate/tests/jsonschema/filter.yaml

@@ -25,7 +25,7 @@
 
 - ansible.builtin.assert:
     that:
-      - "'value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0' in result.msg"
+      - "'value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0' in result.msg"
 
 - name: invalid engine value
   ansible.builtin.set_fact:

tests/integration/targets/utils_validate/tests/jsonschema/lookup.yaml

@@ -31,7 +31,7 @@
 
 - ansible.builtin.assert:
     that:
-      - "'value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0' in result.msg"
+      - "'value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0' in result.msg"
 
 - name: test invalid plugin configuration option, passed as task varaible
   ansible.builtin.set_fact:
@@ -43,7 +43,7 @@
 
 - ansible.builtin.assert:
     that:
-      - "'value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0' in result.msg"
+      - "'value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0' in result.msg"
 
 - name: invalid engine value
   ansible.builtin.set_fact:

tests/integration/targets/utils_validate/tests/jsonschema/module.yaml

@@ -52,7 +52,7 @@
     that:
       - "'errors' not in result"
       - "result['failed'] == true"
-      - "'value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0' in result.msg"
+      - "'value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0' in result.msg"
 
 - name: invalid engine value
   ansible.utils.validate:

tests/integration/targets/utils_validate/tests/jsonschema/test.yaml

@@ -23,7 +23,7 @@
 
 - ansible.builtin.assert:
     that:
-      - "'value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0' in result.msg"
+      - "'value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0' in result.msg"
 
 - name: invalid engine value
   ansible.builtin.set_fact:

tests/unit/plugins/action/test_validate.py

@@ -186,7 +186,7 @@ class TestValidate(unittest.TestCase):
 
         result = self._plugin.run(task_vars={"ansible_validate_jsonschema_draft": "draft0"})
         self.assertIn(
-            "value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0",
+            "value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0",
             result["msg"],
         )
 
@@ -285,3 +285,15 @@ class TestValidate(unittest.TestCase):
 
         result = self._plugin.run(task_vars=None)
         self.assertIn("Validation errors were found", result["msg"])
+
+    def test_support_for_disabled_format_with_invalid_data(self):
+        """Check passing valid data as per criteria"""
+
+        self._plugin._task.args = {
+            "engine": "ansible.utils.jsonschema",
+            "data": IN_VALID_DATA,
+            "criteria": CRITERIA_FORMAT_SUPPORT_CHECK,
+        }
+
+        result = self._plugin.run(task_vars=dict(ansible_validate_jsonschema_check_format=False))
+        self.assertIn("All checks passed", result["msg"])

tests/unit/plugins/filter/test_validate.py

@@ -150,7 +150,7 @@ class TestValidate(unittest.TestCase):
         with self.assertRaises(AnsibleFilterError) as error:
             validate(*args, **kwargs)
         self.assertIn(
-            "value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0",
+            "value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0",
             str(error.exception),
         )
 

tests/unit/plugins/test/test_validate.py

@@ -148,7 +148,7 @@ class TestValidate(unittest.TestCase):
         with self.assertRaises(AnsibleError) as error:
             validate(*args, **kwargs)
         self.assertIn(
-            "value of draft must be one of: draft3, draft4, draft6, draft7, got: draft0",
+            "value of draft must be one of: draft3, draft4, draft6, draft7, 2019-09, 2020-12, got: draft0",
             str(error.exception),
         )
 

tests/unit/requirements.txt

@@ -1,4 +1,4 @@
-jsonschema ; python_version >= '2.7'
+jsonschema
 netaddr
 textfsm
 ttp