feat: Allow schemas to reference other schemas

This allows us to reference external schemas to avoid duplication. They can be referenced like
```
"$ref": "file://uptime-results.v1.schema.json#/definitions/CheckResult"
```

The handler assumes all schemas will be located in out `schemas` directory.

Author: wedamija

python/sentry_kafka_schemas/codecs/json.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import pathlib
 from typing import Any, Optional, TypeVar, cast
 
 import fastjsonschema
@@ -8,14 +9,24 @@
 
 T = TypeVar("T")
 
+BASE_DIR = pathlib.Path(__file__).parent.parent / "schemas"
+
+
+def file_handler(uri):
+    absolute_path = BASE_DIR / uri[7:]
+    if not absolute_path.exists():
+        raise FileNotFoundError(f"Schema file not found: {absolute_path}")
+    with open(absolute_path) as f:
+        return rapidjson.load(f)
+
 
 class JsonCodec(Codec[T]):
     def __init__(
         self,
         json_schema: Optional[object],
     ) -> None:
         if json_schema is not None:
-            self.__validate = fastjsonschema.compile(json_schema)
+            self.__validate = fastjsonschema.compile(json_schema, handlers={"file": file_handler})
         else:
             self.__validate = lambda _: None
 

python/sentry_kafka_schemas/codecs/msgpack.py

@@ -3,6 +3,7 @@
 import fastjsonschema
 import msgpack
 from sentry_kafka_schemas.codecs import Codec, ValidationError
+from sentry_kafka_schemas.codecs.json import file_handler
 
 T = TypeVar("T")
 
@@ -14,7 +15,7 @@ class MsgpackCodec(Codec[T]):
 
     def __init__(self, json_schema: Optional[object]) -> None:
         if json_schema is not None:
-            self.__validate = fastjsonschema.compile(json_schema)
+            self.__validate = fastjsonschema.compile(json_schema, handlers={"file": file_handler})
         else:
             self.__validate = lambda _: None
 

python/tests/test_examples.py

@@ -5,8 +5,10 @@
 import pytest
 import rapidjson
 from sentry_kafka_schemas import iter_examples
+from sentry_kafka_schemas.codecs.json import file_handler, BASE_DIR
 from sentry_kafka_schemas.sentry_kafka_schemas import _get_schema, get_codec, get_topic, list_topics
 from sentry_kafka_schemas.types import Example
+from jsonschema import RefResolver
 
 
 def get_all_examples() -> Iterator[Tuple[str, int, Example]]:
@@ -65,11 +67,17 @@ def test_json_examples(
     example_data = example.load()
 
     if jsonschema_library == "fastjsonschema":
-        compiled = fastjsonschema.compile(schema)
+        compiled = fastjsonschema.compile(schema, handlers={"file": file_handler})
         compiled(example_data)
     elif jsonschema_library == "jsonschema":
         try:
-            jsonschema.validate(example_data, schema)
+            #TODO: Is this even necessary? Looks like we removed usage of jsonschema?
+            resolver = RefResolver(
+                base_uri=f"file:/{BASE_DIR}",
+                referrer=schema,
+                handlers={"file": file_handler}  # Use the custom_resolver function
+            )
+            jsonschema.validate(example_data, schema, resolver=resolver)
         except jsonschema.ValidationError as e:
             _get_most_specific_jsonschema_error(e)
     elif jsonschema_library == "rapidjson":