RFC7: add support for validator plugins

docs/source/configuration.rst

@@ -286,6 +286,8 @@ default.
                 storage_crs: http://www.opengis.net/def/crs/OGC/1.3/CRS84  # optional CRS in which data is stored, default: as 'crs' field
                 storage_crs_coordinate_epoch: 2017.23  # optional, if storage_crs is a dynamic coordinate reference system
                 always_xy: false  # optional should CRS respect axis ordering
+                validator:
+                    name: path.to.validator  # Python path of validation definition
           formatters:  # list of 1..n formatter definitions
               - name: path.to.formatter  # Python path of formatter definition
                 attachment: true  # whether or not to provide as an attachment or normal response

docs/source/cql2.rst

@@ -33,7 +33,7 @@ Queries
 
 The PostgreSQL provider uses `pygeofilter <https://github.com/geopython/pygeofilter>`_ allowing a range of filter expressions, see examples for:
 
-* `Comparison predicates (`Advanced <https://docs.ogc.org/is/21-065r2/21-065r2.html#advanced-comparison-operators>`_, `Case-insensitive <https://docs.ogc.org/is/21-065r2/21-065r2.html#case-insensitive-comparison>`_)
+* Comparison predicates (`Advanced <https://docs.ogc.org/is/21-065r2/21-065r2.html#advanced-comparison-operators>`_, `Case-insensitive <https://docs.ogc.org/is/21-065r2/21-065r2.html#case-insensitive-comparison>`_)
 * `Spatial predicates <https://docs.ogc.org/is/21-065r2/21-065r2.html#spatial-functions>`_
 * `Temporal predicates <https://docs.ogc.org/is/21-065r2/21-065r2.html#temporal-functions>`_
 

docs/source/plugins.rst

@@ -464,7 +464,7 @@ Below is a sample process definition as a Python dictionary:
 
 .. note::
 
-   Additional processing plugins can also be found in ``pygeoapi/process``.
+   Additional processing plugins can be found in ``pygeoapi/process``.
 
 .. _example-custom-pygeoapi-formatter:
 
@@ -503,6 +503,42 @@ The below template provides a minimal example (let's call the file ``mycooljsonf
 
            return out_data
 
+Example: custom pygeoapi validator
+----------------------------------
+
+Python code
+^^^^^^^^^^^
+
+The below template provides a minimal example (let's call the file ``mycooldatavalidator.py``:
+
+.. code-block:: python
+
+   from typing import Any
+
+   from pygeoapi.validator.base import BaseValidator, ValidatorValidationError
+
+   class MyCoolDataValidator(BaseValidator):
+       def __init__(self, validator_def):
+           """Inherit from parent class"""
+
+           super().__init__(validator_def)
+
+       def validate(self, data: Any, partial: bool = False) -> None:
+           if partial:  # plugin does not support partial updates to a given item (PATCH)
+               msg = 'Partial validation not supported'
+               raise ValidatorValidationError(msg)
+
+           # data is a dict of incoming data, validate accordingly
+           if 'some_property' not in data:
+               msg = 'Invalid data payload!'  # to add more detailed messaging, pass user_msg="string of text" to ValidatorValidationError
+               raise ValidatorValidationError(msg)
+
+       def __repr__(self):
+           return '<MyCoolValidator>'
+
+.. note::
+
+   Additional validator plugins can be found in ``pygeoapi/validator``.
 
 Featured plugins
 ----------------

docs/source/transactions.rst

@@ -7,8 +7,8 @@ pygeoapi supports the `OGC API - Features - Part 4: Create, Replace, Update and
 for transactional capabilities against feature and record data.
 
 To enable transactions in pygeoapi, a given resource provider needs to be editable (via the configuration resource provider
-``editable: true`` property).  Note that the feature or record provider MUST support create/update/delete.  See the
-:ref:`ogcapi-features` and :ref:`ogcapi-records` documentation for transaction support status of pygeoapi backends.
+``editable: true`` property).  Note that the feature or record provider MUST support create/update/delete.  See
+:ref:`ogcapi-features` and :ref:`ogcapi-records` for transaction support status of pygeoapi backends.
 
 Access control
 ^^^^^^^^^^^^^^
@@ -17,3 +17,33 @@ It should be made clear that authentication and authorization is beyond the resp
 if a pygeoapi user enables transactions, they must provide access control explicitly via another service.
 
 .. _`OGC API - Features - Part 4: Create, Replace, Update and Delete`: https://docs.ogc.org/DRAFTS/20-002.html
+
+Validation
+^^^^^^^^^^
+
+pygeoapi transaction support includes the option to implement custom validation when adding or updating features or records.
+
+To enable validation in transactions in pygeoapi, a given resource provider can specify a custom validator plugin to implement
+custom business rules as needed to ensure data is valid prior to adding or updating a given provider backend.
+
+Given the example below:
+
+.. code-block:: yaml
+
+   providers:
+       - type: feature
+         name: Elasticsearch
+         data /path/to/file
+         id_field: stn_id
+         editable: true
+         validator:
+             name: mycooldatapackage.mycooldatavalidator.MyCoolDataValidator
+             # name: GeoJSON  # shipped with pygeoapi, referred to be a shortname
+
+The ``validator.name`` element refers to one of the following:
+
+* a validator plugin that is shipped with pygeoapi (noting that the core
+  pygeoapi plugin registry can be found in ``pygeoapi.plugin.PLUGINS``) which can be referred to
+  via a shortname
+* a custom Python module/class that implements a pygeoapi validator plugin.  See :ref:`plugins`
+  for more information on implementing validator plugins.

pygeoapi/api/itemtypes.py

@@ -797,6 +797,28 @@ def manage_collection_item(
             HTTPStatus.BAD_REQUEST, headers, request.format,
             'InvalidParameterValue', msg)
 
+    if action in ['create', 'update']:
+        if p.validator is not None:
+            LOGGER.debug('Provider is configured for validation')
+            LOGGER.debug('Loading validator')
+            try:
+                v = load_plugin('validator', {'name': p.validator['name']})
+            except Exception:
+                msg = 'Invalid validator configured'
+                return api.get_exception(
+                    HTTPStatus.INTERNAL_SERVER_ERROR, headers, request.format,
+                    'NoApplicableCode', msg)
+
+            LOGGER.debug('Validating item')
+            try:
+                v.validate(request.data)
+            except Exception as err:
+                msg = err.user_msg or 'Item is not valid, please check and validate payload'  # noqa
+                LOGGER.error(f'Validation errors: {err.message}')
+                return api.get_exception(
+                   HTTPStatus.INTERNAL_SERVER_ERROR, headers, request.format,
+                   'InvalidParameterValue', msg)
+
     if action == 'create':
         LOGGER.debug('Creating item')
         try:

pygeoapi/plugin.py

@@ -90,6 +90,9 @@
         'HTTP': 'pygeoapi.pubsub.http.HTTPPubSubClient',
         'Kafka': 'pygeoapi.pubsub.kafka.KafkaPubSubClient',
         'MQTT': 'pygeoapi.pubsub.mqtt.MQTTPubSubClient'
+    },
+    'validator': {
+        'GeoJSON': 'pygeoapi.validator.geojson.GeoJSONValidator'
     }
 }
 

pygeoapi/provider/base.py

@@ -81,6 +81,7 @@ def __init__(self, provider_def):
         self.include_extra_query_parameters = provider_def.get('include_extra_query_parameters', False)  # noqa
         self._fields = {}
         self.filename = None
+        self.validator = provider_def.get('validator')
 
         # CRS properties
         storage_crs_uri = provider_def.get('storage_crs', DEFAULT_STORAGE_CRS)
@@ -337,7 +338,7 @@ class ProviderTypeError(ProviderGenericError):
 
 class ProviderInvalidQueryError(ProviderGenericError):
     """provider invalid query error"""
-    ogc_exception_code = 'InvalidQuery'
+    ogc_exception_code = 'InvalidParameterValue'
     http_status_code = HTTPStatus.BAD_REQUEST
     default_msg = "query error"
 

pygeoapi/resources/schemas/config/pygeoapi-config-0.x.yml

@@ -637,6 +637,15 @@ properties:
                                           type: boolean
                                           description: whether to include extra query parameters
                                           default: false
+                                      validator:
+                                          type: object
+                                          description: custom validator to apply on transactions
+                                          properties:
+                                              name:
+                                                  type: string
+                                                  description: name of validator
+                                          required:
+                                              - name
                                   required:
                                       - type
                                       - name
@@ -753,4 +762,4 @@ required:
     - server
     - logging
     - metadata
-    - resources
+    - resources