From 28d39d8c4d5f697a7fec2fa0afadcde5f9d6b257 Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@users.noreply.github.com>
Date: Tue, 21 Jan 2025 14:23:32 -0800
Subject: [PATCH] Fix migration to avoid duplicate collection slugs and names
 (#2318)

Follow-up to #2301

Updates the 0039 migration to ensure collection slugs and names are
unique by:
- Removing all indexes
- Setting `slug` to random value
- Adding unique index to `slug` field.
- Attempting to set slug from name using `slug_from_name()`
- If rejected due to duplicate, append `-<counter>` at end of slug. Also
update name with ` <counter>`.
- Now that names should also be unique, add unique index on name field.

---------

Co-authored-by: Tessa Walsh <tessa@bitarchivist.net>
---
 backend/btrixcloud/migrations/__init__.py     |  2 +
 .../migrations/migration_0039_coll_slugs.py   | 62 +++++++++++++++++--
 2 files changed, 58 insertions(+), 6 deletions(-)

diff --git a/backend/btrixcloud/migrations/__init__.py b/backend/btrixcloud/migrations/__init__.py
index e983072b..b8c3b323 100644
--- a/backend/btrixcloud/migrations/__init__.py
+++ b/backend/btrixcloud/migrations/__init__.py
@@ -3,6 +3,7 @@ BaseMigration class to subclass in each migration module
 """
 
 import os
+import traceback
 from pymongo.errors import OperationFailure
 
 
@@ -73,6 +74,7 @@ class BaseMigration:
                 await self.set_db_version()
             except OperationFailure as err:
                 print(f"Error running migration {self.migration_version}: {err}")
+                traceback.print_exc()
                 return False
 
         else:
diff --git a/backend/btrixcloud/migrations/migration_0039_coll_slugs.py b/backend/btrixcloud/migrations/migration_0039_coll_slugs.py
index 6fa4cd01..8bd65ffc 100644
--- a/backend/btrixcloud/migrations/migration_0039_coll_slugs.py
+++ b/backend/btrixcloud/migrations/migration_0039_coll_slugs.py
@@ -2,10 +2,15 @@
 Migration 0039 -- collection slugs
 """
 
+from uuid import UUID
+
+from pymongo.errors import DuplicateKeyError
+from pymongo.collation import Collation
+import pymongo
+
 from btrixcloud.migrations import BaseMigration
 from btrixcloud.utils import slug_from_name
 
-
 MIGRATION_VERSION = "0039"
 
 
@@ -16,23 +21,68 @@ class Migration(BaseMigration):
     def __init__(self, mdb, **kwargs):
         super().__init__(mdb, migration_version=MIGRATION_VERSION)
 
+    async def dedup_slug(
+        self, name: str, slug_base: str, coll_id: UUID, colls_mdb
+    ) -> None:
+        """attempt to set slug, if duplicate, append suffix until a valid slug is found
+        also update original name with same suffix"""
+        slug = slug_base
+        count = 1
+
+        while True:
+            try:
+                await colls_mdb.find_one_and_update(
+                    {"_id": coll_id},
+                    {"$set": {"slug": slug}},
+                )
+                break
+            except DuplicateKeyError:
+                # pylint: disable=raise-missing-from
+                count += 1
+                slug = f"{slug_base}-{count}"
+
+        if count > 1:
+            print(f"Duplicate collection name '{name}' set to '{name} {count}'")
+            await colls_mdb.find_one_and_update(
+                {"_id": coll_id}, {"$set": {"name": f"{name} {count}"}}
+            )
+
     async def migrate_up(self):
         """Perform migration up.
 
         Add slug to collections that don't have one yet, based on name
         """
         colls_mdb = self.mdb["collections"]
+        case_insensitive_collation = Collation(locale="en", strength=1)
 
-        async for coll_raw in colls_mdb.find({"slug": None}):
+        await colls_mdb.drop_indexes()
+
+        # set slug to random value to ensure uniqueness
+        await colls_mdb.update_many(
+            {}, [{"$set": {"slug": {"$toString": {"$rand": {}}}}}]
+        )
+
+        await colls_mdb.create_index(
+            [("oid", pymongo.ASCENDING), ("slug", pymongo.ASCENDING)],
+            unique=True,
+            collation=case_insensitive_collation,
+        )
+
+        async for coll_raw in colls_mdb.find({}):
             coll_id = coll_raw["_id"]
             try:
-                await colls_mdb.find_one_and_update(
-                    {"_id": coll_id},
-                    {"$set": {"slug": slug_from_name(coll_raw.get("name", ""))}},
-                )
+                name = coll_raw.get("name", "")
+                slug = slug_from_name(name)
+                await self.dedup_slug(name, slug, coll_id, colls_mdb)
             # pylint: disable=broad-exception-caught
             except Exception as err:
                 print(
                     f"Error saving slug for collection {coll_id}: {err}",
                     flush=True,
                 )
+
+        await colls_mdb.create_index(
+            [("oid", pymongo.ASCENDING), ("name", pymongo.ASCENDING)],
+            unique=True,
+            collation=case_insensitive_collation,
+        )