fix(query-source): Fix query source relative filepath (#2717)

When generating the filename attribute for stack trace frames, the SDK uses the `filename_for_module` function. When generating the `code.filepath` attribute for query spans, the SDK does not use that function. Because of this inconsistency, code mappings that work with stack frames sometimes don't work with queries that come from the same files.

This change makes sure that query sources use `filename_for_module`, so the paths are consistent.

Author: gggritso

sentry_sdk/tracing_utils.py

@@ -7,6 +7,7 @@
 from sentry_sdk.consts import OP, SPANDATA
 from sentry_sdk.utils import (
     capture_internal_exceptions,
+    filename_for_module,
     Dsn,
     match_regex_list,
     to_string,
@@ -255,7 +256,9 @@ def add_query_source(hub, span):
         except Exception:
             filepath = None
         if filepath is not None:
-            if project_root is not None and filepath.startswith(project_root):
+            if namespace is not None and not PY2:
+                in_app_path = filename_for_module(namespace, filepath)
+            elif project_root is not None and filepath.startswith(project_root):
                 in_app_path = filepath.replace(project_root, "").lstrip(os.sep)
             else:
                 in_app_path = filepath

tests/integrations/asyncpg/__init__.py

@@ -1,4 +1,10 @@
+import os
+import sys
 import pytest
 
 pytest.importorskip("asyncpg")
 pytest.importorskip("pytest_asyncio")
+
+# Load `asyncpg_helpers` into the module search path to test query source path names relative to module. See
+# `test_query_source_with_module_in_search_path`
+sys.path.insert(0, os.path.join(os.path.dirname(__file__)))

tests/integrations/asyncpg/asyncpg_helpers/__init__.py

@@ -0,0 +1,2 @@
+async def execute_query_in_connection(query, connection):
+    await connection.execute(query)

tests/integrations/asyncpg/asyncpg_helpers/helpers.py

@@ -19,6 +19,7 @@
 PG_PORT = 5432
 
 
+from sentry_sdk._compat import PY2
 import datetime
 
 import asyncpg
@@ -592,6 +593,56 @@ async def test_query_source(sentry_init, capture_events):
     assert data.get(SPANDATA.CODE_FUNCTION) == "test_query_source"
 
 
+@pytest.mark.asyncio
+async def test_query_source_with_module_in_search_path(sentry_init, capture_events):
+    """
+    Test that query source is relative to the path of the module it ran in
+    """
+    sentry_init(
+        integrations=[AsyncPGIntegration()],
+        enable_tracing=True,
+        enable_db_query_source=True,
+        db_query_source_threshold_ms=0,
+    )
+
+    events = capture_events()
+
+    from asyncpg_helpers.helpers import execute_query_in_connection
+
+    with start_transaction(name="test_transaction", sampled=True):
+        conn: Connection = await connect(PG_CONNECTION_URI)
+
+        await execute_query_in_connection(
+            "INSERT INTO users(name, password, dob) VALUES ('Alice', 'secret', '1990-12-25')",
+            conn,
+        )
+
+        await conn.close()
+
+    (event,) = events
+
+    span = event["spans"][-1]
+    assert span["description"].startswith("INSERT INTO")
+
+    data = span.get("data", {})
+
+    assert SPANDATA.CODE_LINENO in data
+    assert SPANDATA.CODE_NAMESPACE in data
+    assert SPANDATA.CODE_FILEPATH in data
+    assert SPANDATA.CODE_FUNCTION in data
+
+    assert type(data.get(SPANDATA.CODE_LINENO)) == int
+    assert data.get(SPANDATA.CODE_LINENO) > 0
+    if not PY2:
+        assert data.get(SPANDATA.CODE_NAMESPACE) == "asyncpg_helpers.helpers"
+        assert data.get(SPANDATA.CODE_FILEPATH) == "asyncpg_helpers/helpers.py"
+
+    is_relative_path = data.get(SPANDATA.CODE_FILEPATH)[0] != os.sep
+    assert is_relative_path
+
+    assert data.get(SPANDATA.CODE_FUNCTION) == "execute_query_in_connection"
+
+
 @pytest.mark.asyncio
 async def test_no_query_source_if_duration_too_short(sentry_init, capture_events):
     sentry_init(

tests/integrations/asyncpg/test_asyncpg.py

@@ -1,3 +1,9 @@
+import os
+import sys
 import pytest
 
 pytest.importorskip("django")
+
+# Load `django_helpers` into the module search path to test query source path names relative to module. See
+# `test_query_source_with_module_in_search_path`
+sys.path.insert(0, os.path.join(os.path.dirname(__file__)))

tests/integrations/django/__init__.py

@@ -0,0 +1,9 @@
+from django.contrib.auth.models import User
+from django.http import HttpResponse
+from django.views.decorators.csrf import csrf_exempt
+
+
+@csrf_exempt
+def postgres_select_orm(request, *args, **kwargs):
+    user = User.objects.using("postgres").all().first()
+    return HttpResponse("ok {}".format(user))

tests/integrations/django/django_helpers/__init__.py

@@ -26,6 +26,7 @@ def path(path, *args, **kwargs):
 
 
 from . import views
+from django_helpers import views as helper_views
 
 urlpatterns = [
     path("view-exc", views.view_exc, name="view_exc"),
@@ -59,6 +60,11 @@ def path(path, *args, **kwargs):
     path("template-test3", views.template_test3, name="template_test3"),
     path("postgres-select", views.postgres_select, name="postgres_select"),
     path("postgres-select-slow", views.postgres_select_orm, name="postgres_select_orm"),
+    path(
+        "postgres-select-slow-from-supplement",
+        helper_views.postgres_select_orm,
+        name="postgres_select_slow_from_supplement",
+    ),
     path(
         "permission-denied-exc",
         views.permission_denied_exc,

tests/integrations/django/django_helpers/views.py

@@ -4,6 +4,7 @@
 import pytest
 from datetime import datetime
 
+from sentry_sdk._compat import PY2
 from django import VERSION as DJANGO_VERSION
 from django.db import connections
 
@@ -168,6 +169,62 @@ def test_query_source(sentry_init, client, capture_events):
         raise AssertionError("No db span found")
 
 
+@pytest.mark.forked
+@pytest_mark_django_db_decorator(transaction=True)
+def test_query_source_with_module_in_search_path(sentry_init, client, capture_events):
+    """
+    Test that query source is relative to the path of the module it ran in
+    """
+    client = Client(application)
+
+    sentry_init(
+        integrations=[DjangoIntegration()],
+        send_default_pii=True,
+        traces_sample_rate=1.0,
+        enable_db_query_source=True,
+        db_query_source_threshold_ms=0,
+    )
+
+    if "postgres" not in connections:
+        pytest.skip("postgres tests disabled")
+
+    # trigger Django to open a new connection by marking the existing one as None.
+    connections["postgres"].connection = None
+
+    events = capture_events()
+
+    _, status, _ = unpack_werkzeug_response(
+        client.get(reverse("postgres_select_slow_from_supplement"))
+    )
+    assert status == "200 OK"
+
+    (event,) = events
+    for span in event["spans"]:
+        if span.get("op") == "db" and "auth_user" in span.get("description"):
+            data = span.get("data", {})
+
+            assert SPANDATA.CODE_LINENO in data
+            assert SPANDATA.CODE_NAMESPACE in data
+            assert SPANDATA.CODE_FILEPATH in data
+            assert SPANDATA.CODE_FUNCTION in data
+
+            assert type(data.get(SPANDATA.CODE_LINENO)) == int
+            assert data.get(SPANDATA.CODE_LINENO) > 0
+
+            if not PY2:
+                assert data.get(SPANDATA.CODE_NAMESPACE) == "django_helpers.views"
+                assert data.get(SPANDATA.CODE_FILEPATH) == "django_helpers/views.py"
+
+            is_relative_path = data.get(SPANDATA.CODE_FILEPATH)[0] != os.sep
+            assert is_relative_path
+
+            assert data.get(SPANDATA.CODE_FUNCTION) == "postgres_select_orm"
+
+            break
+    else:
+        raise AssertionError("No db span found")
+
+
 @pytest.mark.forked
 @pytest_mark_django_db_decorator(transaction=True)
 def test_query_source_with_in_app_exclude(sentry_init, client, capture_events):