fix: Some fixes to make ingest-replay-recordings usable (#135)

Co-authored-by: getsantry[bot] <66042841+getsantry[bot]@users.noreply.github.com>

Author: untitaker

examples/ingest-replay-recordings/1/recording-chunk.msgpack

@@ -20,8 +20,6 @@ def run(target_folder: str = "python/sentry_kafka_schemas/schema_types/") -> Non
             schema_data = sentry_kafka_schemas.sentry_kafka_schemas._get_schema(
                 topic_name, version
             )
-            if schema_data["type"] != "json":
-                continue
 
             schema_tmp_typename_base = f"{topic_name.replace('-', '_')}_v{version}"
             schema_tmp_module_name = schema_tmp_typename_base.lower()

examples/ingest-replay-recordings/1/recording.msgpack

@@ -1 +1,2 @@
+# Cannot upgrade to 2.2.0 due to issues described in https://github.com/sbrunner/jsonschema-gentypes/issues/469
 jsonschema-gentypes==1.5.0

python/generate_python_types.py

@@ -41,6 +41,18 @@ def _validate_title(obj: Any) -> None:
 
             used_titles.add(title)
 
+        if (
+            "properties" in obj
+            or "additionalProperites" in obj
+            or "patternProperties" in obj
+            or "required" in obj
+        ) and obj.get("type") != "object":
+            # Impose restriction so that types will be good:
+            # https://github.com/sbrunner/jsonschema-gentypes/issues/469
+            raise AssertionError(
+                "type=object needs to be specified explicitly on all schemas, if properties are defined"
+            )
+
         for value in obj.get("properties", {}).values():
             _validate_title(value)
 

python/requirements-build.txt

@@ -2,70 +2,24 @@
   "$schema": "http://json-schema.org/draft-07/schema#",
   "title": "replay_recording",
   "description": "A replay recording, or a chunk thereof",
-  "anyOf": [
-    {
-      "properties": {
-        "type": { "const": "replay_recording" },
-        "replay_id": { "type": "string" },
-        "key_id": { "type": ["null", "integer"] },
-        "org_id": { "type": "integer", "minimum": 0 },
-        "project_id": { "type": "integer", "minimum": 0 },
-        "received": { "type": "number", "minimum": 0 },
-        "retention_days": { "type": "integer" },
-        "replay_recording": { "$ref": "#/definitions/ChunkMeta" }
-      },
-      "required": [
-        "type",
-        "replay_id",
-        "org_id",
-        "project_id",
-        "received",
-        "retention_days",
-        "replay_recording"
-      ]
-    },
-    {
-      "properties": {
-        "type": { "const": "replay_recording_chunk" },
-        "payload": { "description": "msgpack bytes" },
-        "replay_id": { "type": "string" },
-        "project_id": { "type": "integer", "minimum": 0 },
-        "id": { "type": "string" },
-        "chunk_index": { "type": "integer", "minimum": 0 }
-      },
-      "required": ["type", "replay_id", "project_id", "id", "chunk_index"]
-    },
-    {
-      "properties": {
-        "type": { "const": "replay_recording_not_chunked" },
-        "replay_id": { "type": "string" },
-        "key_id": { "type": ["null", "integer"] },
-        "org_id": { "type": "integer", "minimum": 0 },
-        "project_id": { "type": "integer", "minimum": 0 },
-        "received": { "type": "number", "minimum": 0 },
-        "retention_days": { "type": "integer" },
-        "payload": { "description": "msgpack bytes" }
-      },
-      "required": [
-        "type",
-        "replay_id",
-        "org_id",
-        "project_id",
-        "received",
-        "retention_days",
-        "payload"
-      ]
-    }
-  ],
-  "definitions": {
-    "ChunkMeta": {
-      "title": "chunk_meta",
-      "properties": {
-        "id": { "type": "string" },
-        "chunks": { "type": "integer" },
-        "size": { "type": "integer" }
-      },
-      "required": ["id", "chunks"]
-    }
-  }
+  "type": "object",
+  "properties": {
+    "type": { "const": "replay_recording_not_chunked" },
+    "replay_id": { "type": "string" },
+    "key_id": { "type": ["null", "integer"] },
+    "org_id": { "type": "integer", "minimum": 0 },
+    "project_id": { "type": "integer", "minimum": 0 },
+    "received": { "type": "integer", "minimum": 0 },
+    "retention_days": { "type": "integer" },
+    "payload": { "description": "msgpack bytes" }
+  },
+  "required": [
+    "type",
+    "replay_id",
+    "org_id",
+    "project_id",
+    "received",
+    "retention_days",
+    "payload"
+  ]
 }

python/tests/test_valid_schemas.py

@@ -0,0 +1,5 @@
+#!/bin/sh
+set -xe
+./scripts/msgpack2py < "$1" > /tmp/edit.msgpack
+$EDITOR /tmp/edit.msgpack
+./scripts/py2msgpack < /tmp/edit.msgpack > "$1"