database: Alter teams.org_id column to be not null (#10572)

Out of the roughly 1100 teams in our database only 10 have `org_id: NULL`. These teams belong to GitHub organizations that don't exist anymore, so the corresponding teams are essentially useless. See #2796 (comment) for more context.

Since the nullability of the column makes our code more complicated than it needs to be, this commit proposes to remove these teams and turn the column into a `not null` column.

Author: Turbo87

crates/crates_io_database/src/schema.rs

@@ -761,7 +761,7 @@ diesel::table! {
         /// (Automatically generated by Diesel.)
         avatar -> Nullable<Varchar>,
         /// Unique organization ID on the GitHub API. When organizations are recreated with the same name then they will still get a different ID, so this allows us to avoid potential name reuse attacks.
-        org_id -> Nullable<Int4>,
+        org_id -> Int4,
     }
 }
 

migrations/2025-02-13-084853_not-null-org-id/down.sql

@@ -0,0 +1 @@
+alter table teams alter column org_id drop not null;

migrations/2025-02-13-084853_not-null-org-id/up.sql

@@ -0,0 +1,19 @@
+-- Delete crate owners that are associated with teams that have no `org_id`
+with broken_crate_owners as (
+    select crate_id, owner_id, owner_kind
+    from crate_owners
+      left join teams on teams.id = crate_owners.owner_id
+    where owner_kind = 1
+      and teams.org_id is null
+)
+delete from crate_owners
+  using broken_crate_owners
+where crate_owners.crate_id = broken_crate_owners.crate_id
+  and crate_owners.owner_id = broken_crate_owners.owner_id
+  and crate_owners.owner_kind = broken_crate_owners.owner_kind;
+
+-- Delete teams that have no `org_id`
+delete from teams where org_id is null;
+
+-- Make `org_id` not null
+alter table teams alter column org_id set not null;

src/models/team.rs

@@ -30,7 +30,7 @@ pub struct Team {
     pub name: Option<String>,
     pub avatar: Option<String>,
     /// The GitHub Organization ID this team sits under
-    pub org_id: Option<i32>,
+    pub org_id: i32,
 }
 
 #[derive(Insertable, AsChangeset, Debug, Builder)]
@@ -169,17 +169,7 @@ impl Team {
         gh_login: &str,
         token: &AccessToken,
     ) -> Result<bool, GitHubError> {
-        match self.org_id {
-            Some(org_id) => {
-                team_with_gh_id_contains_user(gh_client, org_id, self.github_id, gh_login, token)
-                    .await
-            }
-            // This means we don't have an org_id on file for the `self` team. It much
-            // probably was deleted from github by the time we backfilled the database.
-            // Short-circuiting to false since a non-existent team cannot contain any
-            // user
-            None => Ok(false),
-        }
+        team_with_gh_id_contains_user(gh_client, self.org_id, self.github_id, gh_login, token).await
     }
 
     pub async fn owning(krate: &Crate, conn: &mut AsyncPgConnection) -> QueryResult<Vec<Owner>> {