ref: Stop using Relay for event schema validation (getsentry#783)

Co-authored-by: sentry-bot <markus+ghbot@sentry.io>

Author: untitaker

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "checkouts/data-schemas"]
+	path = checkouts/data-schemas
+	url = https://github.com/getsentry/sentry-data-schemas

.travis.yml

@@ -57,7 +57,6 @@ install:
   - pip install tox
   - pip install codecov
   - make install-zeus-cli
-  - bash scripts/download-relay.sh
 
 script:
   - coverage erase

checkouts/data-schemas

@@ -82,11 +82,15 @@ def process_event(event, hint):
                     return event
 
                 event.setdefault("tags", {}).setdefault(
-                    "stageId", task_context.stageId()
+                    "stageId", str(task_context.stageId())
+                )
+                event["tags"].setdefault("partitionId", str(task_context.partitionId()))
+                event["tags"].setdefault(
+                    "attemptNumber", str(task_context.attemptNumber())
+                )
+                event["tags"].setdefault(
+                    "taskAttemptId", str(task_context.taskAttemptId())
                 )
-                event["tags"].setdefault("partitionId", task_context.partitionId())
-                event["tags"].setdefault("attemptNumber", task_context.attemptNumber())
-                event["tags"].setdefault("taskAttemptId", task_context.taskAttemptId())
 
                 if task_context._localProperties:
                     if "sentry_app_name" in task_context._localProperties:

scripts/download-relay.sh

@@ -312,7 +312,9 @@ def _drop(event, cause, ty):
             event["level"] = self._level
 
         if event.get("type") != "transaction":
-            event.setdefault("breadcrumbs", []).extend(self._breadcrumbs)
+            event.setdefault("breadcrumbs", {}).setdefault("values", []).extend(
+                self._breadcrumbs
+            )
 
         if event.get("user") is None and self._user is not None:
             event["user"] = self._user

sentry_sdk/integrations/spark/spark_worker.py

@@ -503,7 +503,7 @@ def single_exception_from_error_tuple(
         errno = None
 
     if errno is not None:
-        mechanism = mechanism or {}
+        mechanism = mechanism or {"type": "generic"}
         mechanism.setdefault("meta", {}).setdefault("errno", {}).setdefault(
             "number", errno
         )

sentry_sdk/scope.py

@@ -4,6 +4,7 @@ tox==3.7.0
 Werkzeug==0.15.5
 pytest-localserver==0.5.0
 pytest-cov==2.8.1
+jsonschema==3.2.0
 
 gevent
 eventlet

sentry_sdk/utils.py

@@ -1,9 +1,8 @@
 import os
-import subprocess
 import json
-import uuid
 
 import pytest
+import jsonschema
 
 import gevent
 import eventlet
@@ -16,11 +15,14 @@
 
 from tests import _warning_recorder, _warning_recorder_mgr
 
-SENTRY_RELAY = "./relay"
 
-if not os.path.isfile(SENTRY_RELAY):
-    SENTRY_RELAY = None
+SENTRY_EVENT_SCHEMA = "./checkouts/data-schemas/relay/event.schema.json"
 
+if not os.path.isfile(SENTRY_EVENT_SCHEMA):
+    SENTRY_EVENT_SCHEMA = None
+else:
+    with open(SENTRY_EVENT_SCHEMA) as f:
+        SENTRY_EVENT_SCHEMA = json.load(f)
 
 try:
     import pytest_benchmark
@@ -118,7 +120,7 @@ def _capture_internal_warnings():
 
 
 @pytest.fixture
-def monkeypatch_test_transport(monkeypatch, relay_normalize):
+def monkeypatch_test_transport(monkeypatch, validate_event_schema):
     def check_event(event):
         def check_string_keys(map):
             for key, value in iteritems(map):
@@ -128,54 +130,19 @@ def check_string_keys(map):
 
         with capture_internal_exceptions():
             check_string_keys(event)
-            relay_normalize(event)
+            validate_event_schema(event)
 
     def inner(client):
         monkeypatch.setattr(client, "transport", TestTransport(check_event))
 
     return inner
 
 
-def _no_errors_in_relay_response(obj):
-    """Assert that relay didn't throw any errors when processing the
-    event."""
-
-    def inner(obj):
-        if not isinstance(obj, dict):
-            return
-
-        assert "err" not in obj
-
-        for value in obj.values():
-            inner(value)
-
-    try:
-        inner(obj.get("_meta"))
-        inner(obj.get(""))
-    except AssertionError:
-        raise AssertionError(obj)
-
-
 @pytest.fixture
-def relay_normalize(tmpdir):
+def validate_event_schema(tmpdir):
     def inner(event):
-        if not SENTRY_RELAY:
-            return
-
-        # Disable subprocess integration
-        with sentry_sdk.Hub(None):
-            # not dealing with the subprocess API right now
-            file = tmpdir.join("event-{}".format(uuid.uuid4().hex))
-            file.write(json.dumps(dict(event)))
-            with file.open() as f:
-                output = json.loads(
-                    subprocess.check_output(
-                        [SENTRY_RELAY, "process-event"], stdin=f
-                    ).decode("utf-8")
-                )
-            _no_errors_in_relay_response(output)
-            output.pop("_meta", None)
-            return output
+        if SENTRY_EVENT_SCHEMA:
+            jsonschema.validate(instance=event, schema=SENTRY_EVENT_SCHEMA)
 
     return inner
 

test-requirements.txt

@@ -16,7 +16,7 @@
 except ImportError:
     from django.core.urlresolvers import reverse
 
-from sentry_sdk import capture_message, capture_exception
+from sentry_sdk import capture_message, capture_exception, configure_scope
 from sentry_sdk.integrations.django import DjangoIntegration
 
 from tests.integrations.django.myapp.wsgi import application
@@ -182,16 +182,13 @@ def test_sql_queries(sentry_init, capture_events, with_integration):
 
     from django.db import connection
 
-    sentry_init(
-        integrations=[DjangoIntegration()],
-        send_default_pii=True,
-        _experiments={"record_sql_params": True},
-    )
-
     events = capture_events()
 
     sql = connection.cursor()
 
+    with configure_scope() as scope:
+        scope.clear_breadcrumbs()
+
     with pytest.raises(OperationalError):
         # table doesn't even exist
         sql.execute("""SELECT count(*) FROM people_person WHERE foo = %s""", [123])
@@ -201,7 +198,7 @@ def test_sql_queries(sentry_init, capture_events, with_integration):
     (event,) = events
 
     if with_integration:
-        crumb = event["breadcrumbs"][-1]
+        crumb = event["breadcrumbs"]["values"][-1]
 
         assert crumb["message"] == "SELECT count(*) FROM people_person WHERE foo = %s"
         assert crumb["data"]["db.params"] == [123]
@@ -224,6 +221,9 @@ def test_sql_dict_query_params(sentry_init, capture_events):
     sql = connections["postgres"].cursor()
 
     events = capture_events()
+    with configure_scope() as scope:
+        scope.clear_breadcrumbs()
+
     with pytest.raises(ProgrammingError):
         sql.execute(
             """SELECT count(*) FROM people_person WHERE foo = %(my_foo)s""",
@@ -233,7 +233,7 @@ def test_sql_dict_query_params(sentry_init, capture_events):
     capture_message("HI")
     (event,) = events
 
-    crumb = event["breadcrumbs"][-1]
+    crumb = event["breadcrumbs"]["values"][-1]
     assert crumb["message"] == (
         "SELECT count(*) FROM people_person WHERE foo = %(my_foo)s"
     )
@@ -266,14 +266,18 @@ def test_sql_psycopg2_string_composition(sentry_init, capture_events, query):
 
     sql = connections["postgres"].cursor()
 
+    with configure_scope() as scope:
+        scope.clear_breadcrumbs()
+
     events = capture_events()
+
     with pytest.raises(ProgrammingError):
         sql.execute(query(psycopg2.sql), {"my_param": 10})
 
     capture_message("HI")
 
     (event,) = events
-    crumb = event["breadcrumbs"][-1]
+    crumb = event["breadcrumbs"]["values"][-1]
     assert crumb["message"] == ('SELECT %(my_param)s FROM "foobar"')
     assert crumb["data"]["db.params"] == {"my_param": 10}
 
@@ -296,6 +300,9 @@ def test_sql_psycopg2_placeholders(sentry_init, capture_events):
     sql = connections["postgres"].cursor()
 
     events = capture_events()
+    with configure_scope() as scope:
+        scope.clear_breadcrumbs()
+
     with pytest.raises(DataError):
         names = ["foo", "bar"]
         identifiers = [psycopg2.sql.Identifier(name) for name in names]
@@ -313,10 +320,10 @@ def test_sql_psycopg2_placeholders(sentry_init, capture_events):
     capture_message("HI")
 
     (event,) = events
-    for crumb in event["breadcrumbs"]:
+    for crumb in event["breadcrumbs"]["values"]:
         del crumb["timestamp"]
 
-    assert event["breadcrumbs"][-2:] == [
+    assert event["breadcrumbs"]["values"][-2:] == [
         {
             "category": "query",
             "data": {"db.paramstyle": "format"},