update migration to restrict a one migration per project v1 id

components/renku_data_services/migrations/versions/32fab117ebec_add_project_migrations.py → components/renku_data_services/migrations/versions/6e4dfb1ca90f_add_project_migrations.py

@@ -1,16 +1,16 @@
 """add project migrations
 
-Revision ID: 32fab117ebec
+Revision ID: 6e4dfb1ca90f
 Revises: d71f0f795d30
-Create Date: 2025-02-06 15:20:45.667192
+Create Date: 2025-02-11 10:05:25.003625
 
 """
 
 import sqlalchemy as sa
 from alembic import op
 
 # revision identifiers, used by Alembic.
-revision = "32fab117ebec"
+revision = "6e4dfb1ca90f"
 down_revision = "d71f0f795d30"
 branch_labels = None
 depends_on = None
@@ -25,7 +25,8 @@ def upgrade() -> None:
         sa.Column("project_id", sa.String(length=26), nullable=False),
         sa.ForeignKeyConstraint(["project_id"], ["projects.projects.id"], ondelete="CASCADE"),
         sa.PrimaryKeyConstraint("id"),
-        sa.UniqueConstraint("project_id", "project_v1_id", name="uq_project_migrations"),
+        sa.UniqueConstraint("project_v1_id"),
+        sa.UniqueConstraint("project_v1_id", name="uq_project_v1_id"),
         schema="projects",
     )
     op.create_index(

components/renku_data_services/project/db.py

@@ -917,30 +917,30 @@ async def migrate_v1_project(
         """Migrate a v1 project by creating a new project and tracking the migration."""
         if not session:
             raise errors.ProgrammingError(message="A database session is required")
-        created_project = await self.project_repo.insert_project(user, project)
-        if not created_project:
-            raise errors.ValidationError(
-                message=f"Failed to create a project for migration from v1 (project_v1_id={project_v1_id})."
-            )
 
         result = await session.scalars(
-            select(schemas.ProjectMigrationsORM)
-            .where(schemas.ProjectMigrationsORM.project_id == created_project.id)
-            .where(schemas.ProjectMigrationsORM.project_v1_id == project_v1_id)
+            select(schemas.ProjectMigrationsORM).where(schemas.ProjectMigrationsORM.project_v1_id == project_v1_id)
         )
         project_migration = result.one_or_none()
         if project_migration is not None:
+            raise errors.ValidationError(message=f"Project V1 with id '{project_v1_id}' already exists.")
+        created_project = await self.project_repo.insert_project(user, project)
+        if not created_project:
             raise errors.ValidationError(
-                message=(
-                    f"Project with project_v1_id '{project_v1_id}' and "
-                    f"Project id '{created_project.id}' already exists."
-                )
+                message=f"Failed to create a project for migration from v1 (project_v1_id={project_v1_id})."
             )
 
-        migration_entry = models.UnsavedProjectMigration(project=created_project, project_v1_id=project_v1_id)
-        session.add(migration_entry)
+        if not created_project.id:
+            raise errors.ValidationError(message="Failed to create project with a valid ID.")
+
+        migration_orm = schemas.ProjectMigrationsORM(project_id=created_project.id, project_v1_id=project_v1_id)
+
+        if migration_orm.project_id is None:
+            raise errors.ValidationError(message="Project ID cannot be None for the migration entry.")
+
+        session.add(migration_orm)
         await session.flush()
-        await session.refresh(migration_entry)
+        await session.refresh(migration_orm)
 
         return created_project
 

components/renku_data_services/project/models.py

@@ -161,7 +161,7 @@ class SessionSecretPatchSecretValue:
 class UnsavedProjectMigration:
     """Model representing a migration from an old project version that has not been persisted."""
 
-    project: Project
+    project_id: ULID
     project_v1_id: int
 
 
@@ -170,11 +170,11 @@ class ProjectMigration:
     """Model representing a migration from an old project version."""
 
     id: ULID
-    project: Project
+    project_id: ULID
     project_v1_id: int
     migrated_at: datetime = field(default_factory=lambda: datetime.now(UTC).replace(microsecond=0))
 
     @property
     def etag(self) -> str:
         """Entity tag value for this project migration object."""
-        return compute_etag_from_fields(self.migrated_at, self.project.id)
+        return compute_etag_from_fields(self.migrated_at, self.project_v1_id)

components/renku_data_services/project/orm.py

@@ -199,17 +199,15 @@ class ProjectMigrationsORM(BaseORM):
     """Tracks project migrations from an old project (project_v1_id) to a new project (project_id)."""
 
     __tablename__ = "project_migrations"
-    __table_args__ = (UniqueConstraint("project_id", "project_v1_id", name="uq_project_migrations"),)
+    __table_args__ = (UniqueConstraint("project_v1_id", name="uq_project_v1_id"),)
 
     id: Mapped[ULID] = mapped_column("id", ULIDType, primary_key=True, default_factory=lambda: str(ULID()), init=False)
 
-    project_v1_id: Mapped[int] = mapped_column("project_v1_id", Integer, nullable=False)
-    """The old project being migrated."""
+    project_v1_id: Mapped[int] = mapped_column("project_v1_id", Integer, nullable=False, unique=True)
+    """The old project being migrated. Must be unique."""
 
-    project_id: Mapped[Optional[ULID]] = mapped_column(
-        ForeignKey("projects.id", ondelete="CASCADE"), nullable=False, index=True
-    )
-    """The new project that replaces the old one."""
+    project_id: Mapped[ULID] = mapped_column(ForeignKey("projects.id", ondelete="CASCADE"), nullable=False, index=True)
+    """The new project of the migration of the v1."""
 
     project: Mapped[Optional[ProjectORM]] = relationship("ProjectORM", foreign_keys=[project_id], default=None)
     """Relationship to the new project."""

test/bases/renku_data_services/data_api/test_projects.py

@@ -1560,3 +1560,58 @@ async def test_get_project_after_group_moved(
     assert response.json is not None
     assert response.json.get("id") == project_id
     assert response.json.get("documentation") == "Hello, World!"
+
+
+@pytest.mark.asyncio
+async def test_migrate_v1_project(sanic_client, app_config, user_headers, regular_user, create_project) -> None:
+    v1_id = 1122
+    v1_project = {
+        "name": "New Migrated Project",
+        "slug": "new-project-slug",
+        "namespace": regular_user.namespace.slug,
+        "description": "Old project for migration",
+        "repositories": ["http://old-repository.com"],
+        "visibility": "private",
+        "keywords": ["old", "project"],
+    }
+
+    await app_config.event_repo.delete_all_events()
+
+    _, response = await sanic_client.post(
+        f"/api/data/renku_v1_projects/{v1_id}/migrations", headers=user_headers, json=v1_project
+    )
+
+    assert response.status_code == 201, response.text
+    migrated_project = response.json
+    assert migrated_project["name"] == "New Migrated Project"
+    assert migrated_project["slug"] == "new-project-slug"
+    assert migrated_project["created_by"] == "user"
+    assert migrated_project["namespace"] == regular_user.namespace.slug
+    assert migrated_project["description"] == "Old project for migration"
+    assert migrated_project["visibility"] == "private"
+    assert migrated_project["keywords"] == ["old", "project"]
+    assert migrated_project["repositories"] == ["http://old-repository.com"]
+
+    # events = await app_config.event_repo.get_pending_events()
+    # assert len(events) == 2
+    # project_created_event = next(e for e in events if e.get_message_type() == "project.created")
+    # project_created = deserialize_event(project_created_event)
+    # assert project_created.name == v1_project["name"]
+    # assert project_created.slug == v1_project["slug"]
+    # project_auth_added_event = next(e for e in events if e.get_message_type() == "projectAuth.added")
+    # project_auth_added = deserialize_event(project_auth_added_event)
+    # assert project_auth_added.userId == "user"
+    # assert project_auth_added.role == MemberRole.OWNER
+
+    # migrated_project_id = migrated_project["id"]
+    # _, response = await sanic_client.get(f"/api/data/projects/{migrated_project_id}", headers=user_headers)
+    # assert response.status_code == 200, response.text
+    # migrated_project = response.json
+    # assert migrated_project["name"] == "New Migrated Project"
+    # assert migrated_project["slug"] == "new-project-slug"
+    # assert migrated_project["created_by"] == "user"
+    # assert migrated_project["namespace"] == regular_user.namespace.slug
+    # assert migrated_project["description"] == "Old project for migration"
+    # assert migrated_project["visibility"] == "private"
+    # assert migrated_project["keywords"] == ["old", "project"]
+    # assert migrated_project["repositories"] == ["http://old-repository.com"]