From 3ad6e1dbac2339ff9d06e5545f00ddad2160e4d0 Mon Sep 17 00:00:00 2001
From: Eva Lott <evalott100@gmail.com>
Date: Tue, 1 Oct 2024 10:12:42 +0100
Subject: [PATCH] updated jsonschema validator and added test

---
 event_model/__init__.py                   |  8 ++--
 event_model/documents/event_descriptor.py | 15 +++----
 event_model/schemas/event_descriptor.json | 12 ++---
 event_model/tests/test_auth.py            | 55 ++++++++++++++++++-----
 4 files changed, 62 insertions(+), 28 deletions(-)

diff --git a/event_model/__init__.py b/event_model/__init__.py
index 78788561..784e33ee 100644
--- a/event_model/__init__.py
+++ b/event_model/__init__.py
@@ -1801,18 +1801,20 @@ class MismatchedDataKeys(InvalidData):
 
 def _is_array(checker, instance):
     return (
-        jsonschema.validators.Draft7Validator.TYPE_CHECKER.is_type(instance, "array")
+        jsonschema.validators.Draft202012Validator.TYPE_CHECKER.is_type(
+            instance, "array"
+        )
         or isinstance(instance, tuple)
         or hasattr(instance, "__array__")
     )
 
 
-_array_type_checker = jsonschema.validators.Draft7Validator.TYPE_CHECKER.redefine(
+_array_type_checker = jsonschema.validators.Draft202012Validator.TYPE_CHECKER.redefine(
     "array", _is_array
 )
 
 _Validator = jsonschema.validators.extend(
-    jsonschema.validators.Draft7Validator, type_checker=_array_type_checker
+    jsonschema.validators.Draft202012Validator, type_checker=_array_type_checker
 )
 
 schema_validators = {
diff --git a/event_model/documents/event_descriptor.py b/event_model/documents/event_descriptor.py
index 721bfe7d..b27937a2 100644
--- a/event_model/documents/event_descriptor.py
+++ b/event_model/documents/event_descriptor.py
@@ -23,16 +23,18 @@ class Limits(TypedDict):
     warning: NotRequired[Annotated[LimitsRange, Field(description="Warning limits.")]]
     alarm: NotRequired[Annotated[LimitsRange, Field(description="Alarm limits.")]]
 
+
 _ConstrainedDtype = Annotated[
     str,
     Field(
         description="A numpy dtype e.g `<U9`, `<f16`",
-        pattern="[|<>][tbiufcmMOSUV][0-9]+"
-    )
+        pattern="[|<>][tbiufcmMOSUV][0-9]+",
+    ),
 ]
 
 _ConstrainedDtypeNpStructure = Tuple[str, _ConstrainedDtype]
 
+
 class DataKey(TypedDict):
     """Describes the objects in the data property of Event documents"""
 
@@ -56,16 +58,11 @@ class DataKey(TypedDict):
                 "The type of the data in the event, given as a broad "
                 "JSON schema type."
             )
-        )
+        ),
     ]
     dtype_numpy: NotRequired[
         Annotated[
-            Union[
-                _ConstrainedDtype,
-                List[
-                    _ConstrainedDtypeNpStructure
-                ]
-            ],
+            Union[_ConstrainedDtype, List[_ConstrainedDtypeNpStructure]],
             Field(
                 description=(
                     "The type of the data in the event, given as a "
diff --git a/event_model/schemas/event_descriptor.json b/event_model/schemas/event_descriptor.json
index 91968ef1..8f972a74 100644
--- a/event_model/schemas/event_descriptor.json
+++ b/event_model/schemas/event_descriptor.json
@@ -62,7 +62,9 @@
                 },
                 "dtype_numpy": {
                     "title": "Dtype Numpy",
-                    "description": "The type of the data in the event, given as a numpy dtype string (or, for structured dtypes, array).",
+                    "description": [
+                        "The type of the data in the event, given as a numpy dtype string (or, for structured dtypes, array)."
+                    ],
                     "anyOf": [
                         {
                             "description": "A numpy dtype e.g `<U9`, `<f16`",
@@ -71,8 +73,6 @@
                         },
                         {
                             "items": {
-                                "maxItems": 2,
-                                "minItems": 2,
                                 "prefixItems": [
                                     {
                                         "type": "string"
@@ -83,9 +83,11 @@
                                         "type": "string"
                                     }
                                 ],
-                                "type": "array"
+                                "type": "array",
+                                "unevaluatedItems": false
                             },
-                            "type": "array"
+                            "type": "array",
+                            "unevaluatedItems": false
                         }
                     ]
                 },
diff --git a/event_model/tests/test_auth.py b/event_model/tests/test_auth.py
index 012744e3..881405e7 100644
--- a/event_model/tests/test_auth.py
+++ b/event_model/tests/test_auth.py
@@ -76,15 +76,17 @@ def test_dots_not_allowed_in_keys():
 
 
 @pytest.mark.parametrize(
-        "dtype_numpy", [
-            "Z",
-            "i",
-            "i4",
-            "4i",
-            "i>4",
-            ">i",
-            [("some_str", "Z")],
-        ]
+    "dtype_numpy",
+    [
+        "Z",
+        "i",
+        "i4",
+        "4i",
+        "i>4",
+        ">i",
+        ("some_str_1", "<u4"),
+        [("some_str_1", "<u4"), ("some_str", "Z")],
+    ],
 )
 def test_bad_numpy_datakeys(dtype_numpy):
     descriptor = {
@@ -92,10 +94,41 @@ def test_bad_numpy_datakeys(dtype_numpy):
         "uid": new_uid(),
         "data_keys": {
             "a": {
-                "source": "", "dtype": "number", "shape": [], "dtype_numpy": dtype_numpy
+                "source": "",
+                "dtype": "number",
+                "shape": [],
+                "dtype_numpy": dtype_numpy,
             }
         },
         "run_start": new_uid(),
     }
-    with pytest.raises(jsonschema.ValidationError):
+    with pytest.raises(
+        jsonschema.ValidationError,
+    ):
         schema_validators[DocumentNames.descriptor].validate(descriptor)
+
+
+@pytest.mark.parametrize(
+    "dtype_numpy",
+    [
+        ">u4",
+        "<u4",
+        [("some_str_1", "<u4"), ("some_str", "<u4")],
+        [("some_str_1", "<u4"), ("some_str", ">u4")],
+    ],
+)
+def test_good_numpy_datakeys(dtype_numpy):
+    descriptor = {
+        "time": 0,
+        "uid": new_uid(),
+        "data_keys": {
+            "a": {
+                "source": "",
+                "dtype": "number",
+                "shape": [],
+                "dtype_numpy": dtype_numpy,
+            }
+        },
+        "run_start": new_uid(),
+    }
+    schema_validators[DocumentNames.descriptor].validate(descriptor)