WIP Upgrade to Schemathesis v4

annif/__init__.py

@@ -37,14 +37,19 @@ def create_cx_app(config_name: str | None = None) -> FlaskApp:
     import connexion
     from connexion.datastructures import MediaTypeDict
     from connexion.middleware import MiddlewarePosition
-    from connexion.validators import FormDataValidator, MultiPartFormDataValidator
+    from connexion.validators import MultiPartFormDataValidator
     from starlette.middleware.cors import CORSMiddleware
 
     import annif.registry
-    from annif.openapi.validation import CustomRequestBodyValidator
+    from annif.openapi.validation import (
+        CustomFormDataValidator,
+        CustomRequestBodyValidator,
+    )
 
     specdir = os.path.join(os.path.dirname(__file__), "openapi")
-    cxapp = connexion.FlaskApp(__name__, specification_dir=specdir)
+    cxapp = connexion.FlaskApp(
+        __name__, specification_dir=specdir, strict_validation=True
+    )
     config_name = _get_config_name(config_name)
     logger.debug(f"creating connexion app with configuration {config_name}")
     cxapp.app.config.from_object(config_name)
@@ -54,7 +59,7 @@ def create_cx_app(config_name: str | None = None) -> FlaskApp:
         "body": MediaTypeDict(
             {
                 "*/*json": CustomRequestBodyValidator,
-                "application/x-www-form-urlencoded": FormDataValidator,
+                "application/x-www-form-urlencoded": CustomFormDataValidator,
                 "multipart/form-data": MultiPartFormDataValidator,
             }
         ),

annif/openapi/annif.yaml

@@ -201,6 +201,8 @@ paths:
           content:
             application/json:
               {}
+        "400":
+          $ref: '#/components/responses/BadRequest'
         "403":
           $ref: '#/components/responses/NotAllowed'
         "404":
@@ -232,6 +234,7 @@ paths:
                   description: candidate languages as IETF BCP 47 codes
                   items:
                     type: string
+                    format: bcp47
                     maxLength: 3
                     minLength: 2
                     example: en

annif/openapi/validation.py

@@ -5,9 +5,9 @@
 import logging
 from typing import Any
 
-from connexion.exceptions import BadRequestProblem
+from connexion.exceptions import BadRequestProblem, ExtraParameterProblem
 from connexion.json_schema import format_error_with_path
-from connexion.validators import JSONRequestBodyValidator
+from connexion.validators import FormDataValidator, JSONRequestBodyValidator
 from jsonschema.exceptions import ValidationError
 
 logger = logging.getLogger("openapi.validation")
@@ -37,3 +37,23 @@ def _validate(self, body: Any) -> dict | None:
                 extra={"validator": "body"},
             )
             raise BadRequestProblem(detail=f"{exception.message}{error_path_msg}")
+
+
+class CustomFormDataValidator(FormDataValidator):
+    """Custom request body validator that allows additional metadata fields starting
+    with 'metadata_' in the request body while rejecting other fields."""
+
+    def __init__(self, *args, **kwargs) -> None:
+        super().__init__(*args, **kwargs)
+
+    def _validate_params_strictly(self, data: dict) -> None:
+        form_params = data.keys()
+        spec_params = self._schema.get("properties", {}).keys()
+
+        reduced_form_params = [
+            fp for fp in form_params if not fp.startswith("metadata_")
+        ]
+
+        errors = set(reduced_form_params).difference(set(spec_params))
+        if errors:
+            raise ExtraParameterProblem(param_type="formData", extra_params=errors)

pyproject.toml

@@ -69,7 +69,7 @@ flake8 = "*"
 bumpversion = "*"
 black = "25.*"
 isort = "*"
-schemathesis = "3.*.*"
+schemathesis = "4.*"
 
 [project.urls]
 homepage = "https://annif.org"

tests/test_openapi.py

@@ -2,12 +2,17 @@
 
 import pytest
 import schemathesis
-from hypothesis import settings
+from hypothesis import HealthCheck, settings, strategies
+from simplemma.strategies.dictionaries.dictionary_factory import SUPPORTED_LANGUAGES
 
 import annif
 
+bcp47_strategy = strategies.sampled_from(SUPPORTED_LANGUAGES)
+schemathesis.openapi.format("bcp47", bcp47_strategy)
+
+
 cxapp = annif.create_app(config_name="annif.default_config.TestingConfig")
-schema = schemathesis.from_path("annif/openapi/annif.yaml", app=cxapp)
+schema = schemathesis.openapi.from_asgi("/v1/openapi.json", app=cxapp)
 
 
 @schemathesis.hook("filter_path_parameters")
@@ -20,14 +25,14 @@ def filter_path_parameters(context, path_parameters):
 
 
 @schema.parametrize()
-@settings(max_examples=10)
+@settings(max_examples=10, suppress_health_check=[HealthCheck.filter_too_much])
 def test_openapi_fuzzy(case):
     case.call_and_validate()
 
 
 @pytest.mark.slow
-@schema.include(path_regex="/v1/projects/{project_id}").parametrize()
-@settings(max_examples=50)
+@schema.include(path_regex="projects/{project_id}").parametrize()
+@settings(max_examples=50, suppress_health_check=[HealthCheck.filter_too_much])
 def test_openapi_fuzzy_target_dummy_fi(case):
     case.path_parameters = {"project_id": "dummy-fi"}
     case.call_and_validate()