fix: Use proper JSON contains function for PostgreSQL

Fix `json_array_contains_value` function to use the `@>` operator for
checking if a JSON array contains a value in PostgreSQL. This is
necessary because the `has_key` function only works for string values.

Also, remove `get_rom_collections` method, as it was doing the same
thing as `get_collections_by_rom_id`.

Fixes #1441.

Author: adamantike

backend/handler/database/collections_handler.py

@@ -1,7 +1,12 @@
+from collections.abc import Sequence
+from typing import Any
+
 from decorators.database import begin_session
 from models.collection import Collection
 from sqlalchemy import Select, delete, select, update
 from sqlalchemy.orm import Session
+from sqlalchemy.sql import ColumnExpressionArgument
+from utils.database import json_array_contains_value
 
 from .base_handler import DBBaseHandler
 
@@ -37,11 +42,19 @@ def get_collections(self, session: Session = None) -> Select[tuple[Collection]]:
 
     @begin_session
     def get_collections_by_rom_id(
-        self, rom_id: int, session: Session = None
+        self,
+        rom_id: int,
+        *,
+        order_by: Sequence[str | ColumnExpressionArgument[Any]] | None = None,
+        session: Session = None,
     ) -> list[Collection]:
-        return session.scalars(
-            select(Collection).filter(Collection.roms.contains(rom_id))
-        ).all()
+        query = select(Collection).filter(
+            json_array_contains_value(Collection.roms, rom_id, session=session)
+        )
+        if order_by is not None:
+            query = query.order_by(*order_by)
+
+        return session.scalars(query).all()
 
     @begin_session
     def update_collection(

backend/handler/database/roms_handler.py

@@ -6,7 +6,6 @@
 from models.rom import Rom, RomUser
 from sqlalchemy import and_, delete, func, or_, select, update
 from sqlalchemy.orm import Query, Session, selectinload
-from utils.database import json_array_contains_value
 
 from .base_handler import DBBaseHandler
 
@@ -181,24 +180,6 @@ def get_rom_by_filename_no_ext(
             query.filter_by(file_name_no_ext=file_name_no_ext).limit(1)
         )
 
-    @begin_session
-    def get_rom_collections(
-        self, rom: Rom, session: Session = None
-    ) -> list[Collection]:
-        return (
-            session.scalars(
-                select(Collection)
-                .filter(
-                    json_array_contains_value(
-                        Collection.roms, str(rom.id), session=session
-                    )
-                )
-                .order_by(Collection.name.asc())
-            )
-            .unique()
-            .all()
-        )
-
     @begin_session
     def update_rom(self, id: int, data: dict, session: Session = None) -> Rom:
         return session.execute(

backend/models/rom.py

@@ -17,6 +17,7 @@
     String,
     Text,
     UniqueConstraint,
+    func,
 )
 from sqlalchemy.orm import Mapped, mapped_column, relationship
 from utils.database import CustomJSON
@@ -141,9 +142,12 @@ def merged_screenshots(self) -> list[str]:
         return screenshots
 
     def get_collections(self) -> list[Collection]:
-        from handler.database import db_rom_handler
+        from handler.database import db_collection_handler
 
-        return db_rom_handler.get_rom_collections(self)
+        return db_collection_handler.get_collections_by_rom_id(
+            self.id,
+            order_by=[func.lower("name")],
+        )
 
     # Metadata fields
     @property

backend/utils/database.py

@@ -21,5 +21,11 @@ def json_array_contains_value(
     """Check if a JSON array column contains a single value."""
     conn = session.get_bind()
     if is_postgresql(conn):
-        return sa.type_coerce(column, sa_pg.JSONB()).has_key(value)
+        # In PostgreSQL, string values can be checked for containment using the `?` operator.
+        # For other types, we use the `@>` operator.
+        if isinstance(value, str):
+            return sa.type_coerce(column, sa_pg.JSONB).has_key(value)
+        return sa.type_coerce(column, sa_pg.JSONB).contains(
+            func.cast(value, sa_pg.JSONB)
+        )
     return func.json_contains(column, value)