Avoid N+1 queries fetching Upload.errors

codecov/commands/executor.py

@@ -7,7 +7,6 @@
 from core.commands.flag import FlagCommands
 from core.commands.pull import PullCommands
 from core.commands.repository import RepositoryCommands
-from core.commands.upload import UploadCommands
 from utils.services import get_long_service_name
 
 mapping = {
@@ -17,7 +16,6 @@
     "branch": BranchCommands,
     "compare": CompareCommands,
     "pull": PullCommands,
-    "upload": UploadCommands,
     "flag": FlagCommands,
     "component": ComponentCommands,
 }

graphql_api/actions/commits.py

@@ -5,7 +5,7 @@
 
 from core.models import Commit, Pull, Repository
 from graphql_api.types.enums import CommitStatus
-from reports.models import CommitReport, ReportSession
+from reports.models import CommitReport
 
 
 def pull_commits(pull: Pull) -> QuerySet[Commit]:
@@ -23,29 +23,6 @@ def pull_commits(pull: Pull) -> QuerySet[Commit]:
     return Commit.objects.filter(id__in=subquery).defer("_report")
 
 
-def commit_uploads(commit: Commit) -> QuerySet[ReportSession]:
-    if not commit.commitreport:
-        return ReportSession.objects.none()
-
-    sessions = commit.commitreport.sessions.prefetch_related("flags")
-
-    # # sessions w/ flags and type 'uploaded'
-    # uploaded = sessions.filter(upload_type="uploaded")
-
-    # # carry forward flags that do not have an equivalent uploaded flag
-    # carried_forward = sessions.filter(upload_type="carriedforward").exclude(
-    #     uploadflagmembership__flag_id__in=uploaded.values_list( <------------ FIXME: looks like `flag_id__in` is causing a seq scan in prod
-    #         "uploadflagmembership__flag_id", flat=True
-    #     )
-    # )
-
-    # return (uploaded.prefetch_related("flags")).union(
-    #     carried_forward.prefetch_related("flags")
-    # )
-
-    return sessions
-
-
 def commit_status(
     commit: Commit, report_type: CommitReport.ReportType
 ) -> Optional[CommitStatus]:

graphql_api/types/commit/commit.py

@@ -16,7 +16,7 @@
 import services.path as path_service
 from codecov_auth.models import Owner
 from core.models import Commit
-from graphql_api.actions.commits import commit_status, commit_uploads
+from graphql_api.actions.commits import commit_status
 from graphql_api.actions.comparison import validate_commit_comparison
 from graphql_api.actions.path_contents import sort_path_contents
 from graphql_api.dataloader.bundle_analysis import (
@@ -30,6 +30,7 @@
     queryset_to_connection,
     queryset_to_connection_sync,
 )
+from graphql_api.helpers.requested_fields import selected_fields
 from graphql_api.types.comparison.comparison import (
     MissingBaseCommit,
     MissingHeadReport,
@@ -93,7 +94,19 @@ async def resolve_yaml_state(commit: Commit, info) -> YamlStates:
 @commit_bindable.field("uploads")
 @sync_to_async
 def resolve_list_uploads(commit: Commit, info, **kwargs):
-    queryset = commit_uploads(commit)
+    if not commit.commitreport:
+        return queryset_to_connection_sync([])
+
+    queryset = commit.commitreport.sessions
+
+    requested_fields = selected_fields(info)
+
+    # the `requested_fields` here are prefixed with `edges.node`, as this is a `Connection`
+    # and using `uploads { edges { node { ... } } }` is the way this is queried.
+    if "edges.node.flags" in requested_fields:
+        queryset = queryset.prefetch_related("flags")
+    if "edges.node.errors" in requested_fields:
+        queryset = queryset.prefetch_related("errors")
 
     if not kwargs:  # temp to override kwargs -> return all current uploads
         kwargs["first"] = queryset.count()

graphql_api/types/upload/upload.py

@@ -3,11 +3,11 @@
 from ariadne import ObjectType
 from asgiref.sync import sync_to_async
 from django.urls import reverse
+from graphql import GraphQLResolveInfo
 from shared.django_apps.utils.services import get_short_service_name
 
-from graphql_api.helpers.connection import queryset_to_connection
+from graphql_api.helpers.connection import queryset_to_connection_sync
 from graphql_api.types.enums import (
-    OrderingDirection,
     UploadErrorEnum,
     UploadState,
     UploadType,
@@ -26,43 +26,36 @@
 
 
 @upload_bindable.field("state")
-def resolve_state(upload, info):
+def resolve_state(upload: ReportSession, info: GraphQLResolveInfo) -> UploadState:
     if not upload.state:
         return UploadState.ERROR
     return UploadState(upload.state)
 
 
 @upload_bindable.field("id")
-def resolve_id(upload: ReportSession, info) -> Optional[int]:
+def resolve_id(upload: ReportSession, info: GraphQLResolveInfo) -> Optional[int]:
     return upload.order_number
 
 
 @upload_bindable.field("uploadType")
-def resolve_upload_type(upload, info) -> UploadType:
+def resolve_upload_type(upload: ReportSession, info: GraphQLResolveInfo) -> UploadType:
     return UploadType(upload.upload_type)
 
 
 @upload_bindable.field("errors")
-async def resolve_errors(report_session, info, **kwargs):
-    command = info.context["executor"].get_command("upload")
-    queryset = await command.get_upload_errors(report_session)
-    result = await queryset_to_connection(
-        queryset,
-        ordering=("updated_at",),
-        ordering_direction=OrderingDirection.ASC,
-        **kwargs,
-    )
-    return result
+@sync_to_async
+def resolve_errors(report_session: ReportSession, info: GraphQLResolveInfo, **kwargs):
+    return queryset_to_connection_sync(list(report_session.errors.all()))
 
 
 @upload_error_bindable.field("errorCode")
-def resolve_error_code(error, info):
+def resolve_error_code(error, info: GraphQLResolveInfo) -> UploadErrorEnum:
     return UploadErrorEnum(error.error_code)
 
 
 @upload_bindable.field("ciUrl")
 @sync_to_async
-def resolve_ci_url(upload, info):
+def resolve_ci_url(upload: ReportSession, info: GraphQLResolveInfo):
     return upload.ci_url
 
 

timeseries/tests/test_helpers.py

@@ -1178,6 +1178,7 @@ def test_no_dataset(self, trigger_backfill):
                 repository_id__in=[self.repo1.pk, self.repo2.pk],
             )
         )
+        datasets.reverse()
         assert len(datasets) == 2
         trigger_backfill.assert_called_once_with(datasets)