chore(repositories): When making a ProjectRepository link, upgrade the source if we have a stronger signal

src/sentry/api/endpoints/project_repo.py

@@ -36,10 +36,10 @@ def validate_repositoryId(self, value: int) -> int:
         return value
 
     def create(self, validated_data: dict[str, Any]) -> ProjectRepository:
-        project_repo, created = ProjectRepository.objects.get_or_create(
-            project=self.project,
+        project_repo, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
             repository_id=validated_data["repositoryId"],
-            defaults={"source": ProjectRepositorySource.SCM_ONBOARDING},
+            source=ProjectRepositorySource.SCM_ONBOARDING,
         )
         self._created = created
         return project_repo

src/sentry/integrations/api/endpoints/organization_code_mappings.py

@@ -123,10 +123,10 @@ def validate_default_branch(self, default_branch):
 
     def create(self, validated_data):
         with transaction.atomic(using=router.db_for_write(RepositoryProjectPathConfig)):
-            project_repo, _ = ProjectRepository.objects.get_or_create(
+            project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
                 project_id=validated_data.pop("project_id"),
                 repository_id=validated_data.pop("repository_id"),
-                defaults={"source": ProjectRepositorySource.MANUAL},
+                source=ProjectRepositorySource.MANUAL,
             )
             return RepositoryProjectPathConfig.objects.create(
                 organization_integration_id=self.org_integration.id,
@@ -141,14 +141,14 @@ def update(self, instance, validated_data):
             validated_data.pop("id")
         if self.instance:
             with transaction.atomic(using=router.db_for_write(RepositoryProjectPathConfig)):
-                project_repo, _ = ProjectRepository.objects.get_or_create(
+                project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
                     project_id=validated_data.pop(
                         "project_id", self.instance.project_repository.project_id
                     ),
                     repository_id=validated_data.pop(
                         "repository_id", self.instance.project_repository.repository_id
                     ),
-                    defaults={"source": ProjectRepositorySource.MANUAL},
+                    source=ProjectRepositorySource.MANUAL,
                 )
                 self.instance.project_repository = project_repo
                 for key, value in validated_data.items():

src/sentry/integrations/api/endpoints/organization_code_mappings_bulk.py

@@ -260,10 +260,10 @@ def post(self, request: Request, organization: Organization) -> Response:
         has_errors = False
         last_saved_config = None
 
-        project_repo, _ = ProjectRepository.objects.get_or_create(
-            project=project,
-            repository=repo,
-            defaults={"source": ProjectRepositorySource.MANUAL},
+        project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
+            project_id=project.id,
+            repository_id=repo.id,
+            source=ProjectRepositorySource.MANUAL,
         )
 
         defaults = {

src/sentry/issues/auto_source_code_config/code_mapping.py

@@ -371,10 +371,10 @@ def create_code_mapping(
         },
     )
     with transaction.atomic(using=router.db_for_write(RepositoryProjectPathConfig)):
-        project_repo, _ = ProjectRepository.objects.get_or_create(
-            project=project,
-            repository=repository,
-            defaults={"source": ProjectRepositorySource.MANUAL},
+        project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
+            project_id=project.id,
+            repository_id=repository.id,
+            source=ProjectRepositorySource.MANUAL,
         )
         new_code_mapping, _ = RepositoryProjectPathConfig.objects.update_or_create(
             project_repository=project_repo,

src/sentry/issues/auto_source_code_config/task.py

@@ -240,10 +240,10 @@ def create_code_mapping(
     created = False
     if not tags["dry_run"] and repository is not None:
         with transaction.atomic(using=router.db_for_write(RepositoryProjectPathConfig)):
-            project_repo, _ = ProjectRepository.objects.get_or_create(
-                project=project,
-                repository=repository,
-                defaults={"source": ProjectRepositorySource.AUTO_EVENT},
+            project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
+                project_id=project.id,
+                repository_id=repository.id,
+                source=ProjectRepositorySource.AUTO_EVENT,
             )
             _, created = RepositoryProjectPathConfig.objects.get_or_create(
                 project_repository=project_repo,

src/sentry/models/projectrepository.py

@@ -1,10 +1,13 @@
 from __future__ import annotations
 
-from django.db import models
+from typing import ClassVar
+
+from django.db import models, router, transaction
 
 from sentry.backup.scopes import RelocationScope
 from sentry.db.models import FlexibleForeignKey, cell_silo_model, sane_repr
 from sentry.db.models.base import DefaultFieldsModel
+from sentry.db.models.manager.base import BaseManager
 
 
 class ProjectRepositorySource(models.IntegerChoices):
@@ -15,6 +18,42 @@ class ProjectRepositorySource(models.IntegerChoices):
     SEER_PREFERENCE = 4, "seer_preference"
 
 
+# Indicates how strong of a signal each source is. A higher number indicates
+# more confidence that the project-repo linking is correct.
+SOURCE_PRIORITY: dict[ProjectRepositorySource, int] = {
+    ProjectRepositorySource.AUTO_NAME_MATCH: 100,
+    ProjectRepositorySource.AUTO_EVENT: 200,
+    ProjectRepositorySource.SCM_ONBOARDING: 500,
+    ProjectRepositorySource.MANUAL: 500,
+    ProjectRepositorySource.SEER_PREFERENCE: 500,
+}
+
+
+class ProjectRepositoryManager(BaseManager["ProjectRepository"]):
+    def get_or_create_with_source(
+        self,
+        project_id: int,
+        repository_id: int,
+        source: ProjectRepositorySource,
+    ) -> tuple[ProjectRepository, bool]:
+        """
+        Like get_or_create, but upgrades the source if the existing row
+        was created by a lower-priority mechanism.
+        """
+        project_repo, created = self.get_or_create(
+            project_id=project_id,
+            repository_id=repository_id,
+            defaults={"source": source},
+        )
+        new_priority = SOURCE_PRIORITY.get(source, 0)
+        current_priority = SOURCE_PRIORITY.get(ProjectRepositorySource(project_repo.source), 0)
+        if not created and new_priority > current_priority:
+            with transaction.atomic(router.db_for_write(type(project_repo))):
+                project_repo.source = source
+                project_repo.save(update_fields=["source", "date_updated"])
+        return project_repo, created
+
+
 @cell_silo_model
 class ProjectRepository(DefaultFieldsModel):
     __relocation_scope__ = RelocationScope.Global
@@ -26,6 +65,8 @@ class ProjectRepository(DefaultFieldsModel):
         default=ProjectRepositorySource.MANUAL,
     )
 
+    objects: ClassVar[ProjectRepositoryManager] = ProjectRepositoryManager()
+
     class Meta:
         app_label = "sentry"
         db_table = "sentry_projectrepository"

src/sentry/seer/autofix/utils.py

@@ -493,10 +493,10 @@ def _write_preferences_to_sentry_db(
 
         if project_repos_to_create:
             for project, repository_id, spr in project_repos_to_create:
-                spr.project_repository, _ = ProjectRepository.objects.get_or_create(
-                    project=project,
+                spr.project_repository, _ = ProjectRepository.objects.get_or_create_with_source(
+                    project_id=project.id,
                     repository_id=repository_id,
-                    defaults={"source": ProjectRepositorySource.SEER_PREFERENCE},
+                    source=ProjectRepositorySource.SEER_PREFERENCE,
                 )
 
             created_project_repos = SeerProjectRepository.objects.bulk_create(
@@ -838,10 +838,10 @@ def add_seer_project_repos(project: Project, repos_data: list[ProjectRepoCreateD
     result_ids = []
     with transaction.atomic(router.db_for_write(SeerProjectRepository)):
         for data in repos_data:
-            project_repo, _ = ProjectRepository.objects.get_or_create(
-                project=project,
+            project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
+                project_id=project.id,
                 repository_id=data["repository_id"],
-                defaults={"source": ProjectRepositorySource.SEER_PREFERENCE},
+                source=ProjectRepositorySource.SEER_PREFERENCE,
             )
             seer_project_repo, _ = SeerProjectRepository.objects.update_or_create(
                 project_repository=project_repo,
@@ -867,10 +867,10 @@ def replace_all_seer_project_repos(
         ).delete()
 
         for data in repos_data:
-            project_repo, _ = ProjectRepository.objects.get_or_create(
-                project=project,
+            project_repo, _ = ProjectRepository.objects.get_or_create_with_source(
+                project_id=project.id,
                 repository_id=data["repository_id"],
-                defaults={"source": ProjectRepositorySource.SEER_PREFERENCE},
+                source=ProjectRepositorySource.SEER_PREFERENCE,
             )
             seer_project_repo, _ = SeerProjectRepository.objects.update_or_create(
                 project_repository=project_repo,

tests/sentry/models/test_projectrepository.py

@@ -0,0 +1,115 @@
+from sentry.models.projectrepository import ProjectRepository, ProjectRepositorySource
+from sentry.models.repository import Repository
+from sentry.testutils.cases import TestCase
+
+
+class ProjectRepositoryManagerTest(TestCase):
+    def setUp(self) -> None:
+        super().setUp()
+        self.org = self.create_organization()
+        self.project = self.create_project(organization=self.org)
+        self.repo = Repository.objects.create(
+            organization_id=self.org.id,
+            name="getsentry/sentry",
+            provider="integrations:github",
+            external_id="123",
+        )
+
+    def test_creates_with_source(self) -> None:
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.MANUAL,
+        )
+        assert created
+        assert pr.source == ProjectRepositorySource.MANUAL
+
+    def test_returns_existing_without_upgrade(self) -> None:
+        ProjectRepository.objects.create(
+            project=self.project,
+            repository=self.repo,
+            source=ProjectRepositorySource.MANUAL,
+        )
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.SCM_ONBOARDING,
+        )
+        assert not created
+        pr.refresh_from_db()
+        assert pr.source == ProjectRepositorySource.MANUAL
+
+    def test_upgrades_auto_name_match_to_manual(self) -> None:
+        ProjectRepository.objects.create(
+            project=self.project,
+            repository=self.repo,
+            source=ProjectRepositorySource.AUTO_NAME_MATCH,
+        )
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.MANUAL,
+        )
+        assert not created
+        assert pr.source == ProjectRepositorySource.MANUAL
+
+    def test_upgrades_auto_name_match_to_seer_preference(self) -> None:
+        ProjectRepository.objects.create(
+            project=self.project,
+            repository=self.repo,
+            source=ProjectRepositorySource.AUTO_NAME_MATCH,
+        )
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.SEER_PREFERENCE,
+        )
+        assert not created
+        pr.refresh_from_db()
+        assert pr.source == ProjectRepositorySource.SEER_PREFERENCE
+
+    def test_upgrades_auto_name_match_to_auto_event(self) -> None:
+        ProjectRepository.objects.create(
+            project=self.project,
+            repository=self.repo,
+            source=ProjectRepositorySource.AUTO_NAME_MATCH,
+        )
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.AUTO_EVENT,
+        )
+        assert not created
+        pr.refresh_from_db()
+        assert pr.source == ProjectRepositorySource.AUTO_EVENT
+
+    def test_upgrades_auto_event_to_manual(self) -> None:
+        ProjectRepository.objects.create(
+            project=self.project,
+            repository=self.repo,
+            source=ProjectRepositorySource.AUTO_EVENT,
+        )
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.MANUAL,
+        )
+        assert not created
+        pr.refresh_from_db()
+        assert pr.source == ProjectRepositorySource.MANUAL
+
+    def test_does_not_upgrade_auto_name_match_to_auto_name_match(self) -> None:
+        pr_original = ProjectRepository.objects.create(
+            project=self.project,
+            repository=self.repo,
+            source=ProjectRepositorySource.AUTO_NAME_MATCH,
+        )
+        pr, created = ProjectRepository.objects.get_or_create_with_source(
+            project_id=self.project.id,
+            repository_id=self.repo.id,
+            source=ProjectRepositorySource.AUTO_NAME_MATCH,
+        )
+        assert not created
+        pr.refresh_from_db()
+        assert pr.source == ProjectRepositorySource.AUTO_NAME_MATCH
+        assert pr.date_updated == pr_original.date_updated