fix: EventSession.commit() wasted and N+1 reloads

docs/module/crud.md

@@ -491,6 +491,16 @@ The distinct values for each facet field are returned in the `filter_attributes`
 !!! info "Key format uses `__` as a separator for relationship chains."
     A direct column `User.status` produces `"status"`. A relationship tuple `(User.role, Role.name)` produces `"role__name"`. A deeper chain `(User.role, Role.permission, Permission.name)` produces `"role__permission__name"`. An unknown `filter_by` key raises [`InvalidFacetFilterError`](../reference/exceptions.md#fastapi_toolsets.exceptions.exceptions.InvalidFacetFilterError) (HTTP 422).
 
+#### Skipping facet queries
+
+!!! info "Added in `v5.1.0`"
+
+Facet values only change with the filters, not with the page. Pass `include_facets=False` to `offset_paginate_params()` / `cursor_paginate_params()` on pages 2..N to skip the facet queries entirely (`filter_attributes` will be `None`):
+
+```python
+params: Annotated[dict, Depends(UserCrud.offset_paginate_params(include_facets=False))]
+```
+
 ## Sorting
 
 !!! info "Added in `v1.3`"

src/fastapi_toolsets/crud/factory.py

@@ -44,6 +44,7 @@ from ..types import (
 )
 from .search import (
     SearchConfig,
+    apply_search_joins,
     build_facets,
     build_filter_by,
     build_search_filters,
@@ -51,7 +52,6 @@ from .search import (
     search_field_keys,
 )
 
-
 _ForUpdateMode: TypeAlias = bool | Literal["nowait", "skip_locked"]
 
 
@@ -128,17 +128,6 @@ def _apply_joins(q: Any, joins: JoinType | None, outer_join: bool) -> Any:
     return q
 
 
-def _apply_search_joins(q: Any, search_joins: list[Any]) -> Any:
-    """Apply relationship-based outer joins (from search/filter_by) to a query."""
-    seen: set[str] = set()
-    for join_rel in search_joins:
-        key = str(join_rel)
-        if key not in seen:
-            seen.add(key)
-            q = q.outerjoin(join_rel)
-    return q
-
-
 class AsyncCrud(Generic[ModelType]):
     """Generic async CRUD operations for SQLAlchemy models.
 
@@ -184,17 +173,30 @@ class AsyncCrud(Generic[ModelType]):
         return cls.default_load_options
 
     @classmethod
-    async def _reload_with_options(
-        cls: type[Self], session: AsyncSession, instance: ModelType
+    def _capture_pk_values(cls: type[Self], instance: ModelType) -> dict[str, Any]:
+        """Capture PK values off instance — call before commit expires attributes."""
+        return {
+            cast(str, col.key): getattr(instance, cast(str, col.key))
+            for col in cls.model.__mapper__.primary_key
+        }
+
+    @classmethod
+    async def _reload_with_options_by_pk(
+        cls: type[Self], session: AsyncSession, pk_values: dict[str, Any]
     ) -> ModelType:
-        """Re-query instance by PK with default_load_options applied."""
-        mapper = cls.model.__mapper__
+        """Re-query by previously captured PK values, with default_load_options applied."""
+        # Only called when cls.default_load_options is set (see call sites).
         pk_filters = [
-            getattr(cls.model, cast(str, col.key))
-            == getattr(instance, cast(str, col.key))
-            for col in mapper.primary_key
+            getattr(cls.model, key) == value for key, value in pk_values.items()
         ]
-        return await cls.get(session, filters=pk_filters)
+        q = select(cls.model).where(and_(*pk_filters))
+        q = q.execution_options(populate_existing=True)
+        q = q.options(*cast(Sequence[ExecutableOption], cls.default_load_options))
+        result = await session.execute(q)
+        item = result.unique().scalar_one_or_none()
+        if item is None:  # pragma: no cover — row was just flushed in this transaction
+            raise NotFoundError()
+        return cast(ModelType, item)
 
     @classmethod
     async def _resolve_m2m(
@@ -265,12 +267,12 @@ class AsyncCrud(Generic[ModelType]):
         cls: type[Self],
         filter_by: dict[str, Any] | BaseModel | None,
         facet_fields: Sequence[FacetFieldType] | None,
-    ) -> tuple[list[Any], list[Any]]:
-        """Normalize filter_by and return (filters, joins) to apply to the query."""
+    ) -> tuple[dict[str, Any], list[Any]]:
+        """Normalize filter_by and return ({facet_key: filter}, joins) to apply to the query."""
         if isinstance(filter_by, BaseModel):
             filter_by = filter_by.model_dump(exclude_none=True)
         if not filter_by:
-            return [], []
+            return {}, []
         resolved = cls._resolve_facet_fields(facet_fields)
         return build_filter_by(filter_by, resolved or [])
 
@@ -281,8 +283,13 @@ class AsyncCrud(Generic[ModelType]):
         facet_fields: Sequence[FacetFieldType] | None,
         filters: list[Any],
         search_joins: list[Any],
+        *,
+        include_facets: bool = True,
+        own_filters: dict[str, Any] | None = None,
     ) -> dict[str, list[Any]] | None:
-        """Build facet filter_attributes, or return None if no facet fields configured."""
+        """Build facet filter_attributes, or None if disabled/no facet fields configured."""
+        if not include_facets:
+            return None
         resolved = cls._resolve_facet_fields(facet_fields)
         if not resolved:
             return None
@@ -292,6 +299,7 @@ class AsyncCrud(Generic[ModelType]):
             resolved,
             base_filters=filters,
             base_joins=search_joins,
+            own_filters=own_filters,
         )
 
     @classmethod
@@ -475,6 +483,7 @@ class AsyncCrud(Generic[ModelType]):
         default_page_size: int = 20,
         max_page_size: int = 100,
         include_total: bool = True,
+        include_facets: bool = True,
         search: bool = True,
         filter: bool = True,
         order: bool = True,
@@ -490,6 +499,7 @@ class AsyncCrud(Generic[ModelType]):
             default_page_size: Default ``items_per_page`` value.
             max_page_size: Maximum ``items_per_page`` value.
             include_total: Whether to include total count (not a query param).
+            include_facets: Whether to run facet queries (not a query param).
             search: Enable search query parameters.
             filter: Enable facet filter query parameters.
             order: Enable order query parameters.
@@ -519,7 +529,10 @@ class AsyncCrud(Generic[ModelType]):
         ]
         return cls._build_paginate_params(
             pagination_params=pagination_params,
-            pagination_fixed={"include_total": include_total},
+            pagination_fixed={
+                "include_total": include_total,
+                "include_facets": include_facets,
+            },
             dep_name=f"{cls.model.__name__}OffsetPaginateParams",
             search=search,
             filter=filter,
@@ -537,6 +550,7 @@ class AsyncCrud(Generic[ModelType]):
         *,
         default_page_size: int = 20,
         max_page_size: int = 100,
+        include_facets: bool = True,
         search: bool = True,
         filter: bool = True,
         order: bool = True,
@@ -551,6 +565,7 @@ class AsyncCrud(Generic[ModelType]):
         Args:
             default_page_size: Default ``items_per_page`` value.
             max_page_size: Maximum ``items_per_page`` value.
+            include_facets: Whether to run facet queries (not a query param).
             search: Enable search query parameters.
             filter: Enable facet filter query parameters.
             order: Enable order query parameters.
@@ -582,7 +597,7 @@ class AsyncCrud(Generic[ModelType]):
         ]
         return cls._build_paginate_params(
             pagination_params=pagination_params,
-            pagination_fixed={},
+            pagination_fixed={"include_facets": include_facets},
             dep_name=f"{cls.model.__name__}CursorPaginateParams",
             search=search,
             filter=filter,
@@ -602,6 +617,7 @@ class AsyncCrud(Generic[ModelType]):
         max_page_size: int = 100,
         default_pagination_type: PaginationType = PaginationType.OFFSET,
         include_total: bool = True,
+        include_facets: bool = True,
         search: bool = True,
         filter: bool = True,
         order: bool = True,
@@ -618,6 +634,7 @@ class AsyncCrud(Generic[ModelType]):
             max_page_size: Maximum ``items_per_page`` value.
             default_pagination_type: Default pagination strategy.
             include_total: Whether to include total count (not a query param).
+            include_facets: Whether to run facet queries (not a query param).
             search: Enable search query parameters.
             filter: Enable facet filter query parameters.
             order: Enable order query parameters.
@@ -666,7 +683,10 @@ class AsyncCrud(Generic[ModelType]):
         ]
         return cls._build_paginate_params(
             pagination_params=pagination_params,
-            pagination_fixed={"include_total": include_total},
+            pagination_fixed={
+                "include_total": include_total,
+                "include_facets": include_facets,
+            },
             dep_name=f"{cls.model.__name__}PaginateParams",
             search=search,
             filter=filter,
@@ -730,9 +750,14 @@ class AsyncCrud(Generic[ModelType]):
                     setattr(db_model, rel_attr, related_instances)
 
             session.add(db_model)
-        await session.refresh(db_model)
+            pk_values: dict[str, Any] | None = None
             if cls.default_load_options:
-            db_model = await cls._reload_with_options(session, db_model)
+                await session.flush()
+                pk_values = cls._capture_pk_values(db_model)
+        if pk_values is not None:
+            db_model = await cls._reload_with_options_by_pk(session, pk_values)
+        else:
+            await session.refresh(db_model)
         result = cast(ModelType, db_model)
         if schema:
             return Response(data=schema.model_validate(result))
@@ -1097,9 +1122,15 @@ class AsyncCrud(Generic[ModelType]):
                 m2m_resolved = await cls._resolve_m2m(session, obj, only_set=True)
                 for rel_attr, related_instances in m2m_resolved.items():
                     setattr(db_model, rel_attr, related_instances)
-        await session.refresh(db_model)
+
+            pk_values: dict[str, Any] | None = None
             if cls.default_load_options:
-            db_model = await cls._reload_with_options(session, db_model)
+                await session.flush()
+                pk_values = cls._capture_pk_values(db_model)
+        if pk_values is not None:
+            db_model = await cls._reload_with_options_by_pk(session, pk_values)
+        else:
+            await session.refresh(db_model)
         if schema:
             return Response(data=schema.model_validate(db_model))
         return db_model
@@ -1271,6 +1302,7 @@ class AsyncCrud(Generic[ModelType]):
         search_column: str | None = None,
         order_fields: Sequence[OrderFieldType] | None = None,
         facet_fields: Sequence[FacetFieldType] | None = None,
+        include_facets: bool = True,
         filter_by: dict[str, Any] | BaseModel | None = None,
         schema: type[BaseModel],
     ) -> OffsetPaginatedResponse[Any]:
@@ -1292,6 +1324,9 @@ class AsyncCrud(Generic[ModelType]):
             search_column: Restrict search to a single column key.
             order_fields: Fields allowed for sorting (overrides class default).
             facet_fields: Columns to compute distinct values for (overrides class default)
+            include_facets: When ``False``, skip facet queries entirely;
+                ``filter_attributes`` will be ``None``. Useful on pages 2..N
+                where the facet counts were already fetched on page 1.
             filter_by: Dict of {column_key: value} to filter by declared facet fields.
                 Keys must match the column.key of a facet field. Scalar → equality,
                 list → IN clause. Raises InvalidFacetFilterError for unknown keys.
@@ -1304,7 +1339,6 @@ class AsyncCrud(Generic[ModelType]):
         offset = (page - 1) * items_per_page
 
         fb_filters, search_joins = cls._prepare_filter_by(filter_by, facet_fields)
-        filters.extend(fb_filters)
 
         # Build search filters
         if search:
@@ -1318,6 +1352,11 @@ class AsyncCrud(Generic[ModelType]):
             filters.extend(search_filters)
             search_joins.extend(new_search_joins)
 
+        # Facets combine these with each facet's own filter individually, so
+        # fb_filters is applied to the query below but excluded here.
+        facet_base_filters = list(filters)
+        filters.extend(fb_filters.values())
+
         # Build query with joins
         q = select(cls.model)
 
@@ -1325,11 +1364,11 @@ class AsyncCrud(Generic[ModelType]):
         q = _apply_joins(q, joins, outer_join)
 
         # Apply search joins (always outer joins for search)
-        q = _apply_search_joins(q, search_joins)
+        q = apply_search_joins(q, search_joins)
 
         # Apply order joins (relation joins required for order_by field)
         if order_joins:
-            q = _apply_search_joins(q, order_joins)
+            q = apply_search_joins(q, order_joins)
 
         if filters:
             q = q.where(and_(*filters))
@@ -1352,7 +1391,7 @@ class AsyncCrud(Generic[ModelType]):
             count_q = _apply_joins(count_q, joins, outer_join)
 
             # Apply search joins to count query
-            count_q = _apply_search_joins(count_q, search_joins)
+            count_q = apply_search_joins(count_q, search_joins)
 
             if filters:
                 count_q = count_q.where(and_(*filters))
@@ -1372,7 +1411,12 @@ class AsyncCrud(Generic[ModelType]):
         items: list[Any] = [schema.model_validate(item) for item in raw_items]
 
         filter_attributes = await cls._build_filter_attributes(
-            session, facet_fields, filters, search_joins
+            session,
+            facet_fields,
+            facet_base_filters,
+            search_joins,
+            include_facets=include_facets,
+            own_filters=fb_filters,
         )
         search_columns = cls._resolve_search_columns(search_fields)
         order_columns = cls._resolve_order_columns(order_fields)
@@ -1408,6 +1452,7 @@ class AsyncCrud(Generic[ModelType]):
         search_column: str | None = None,
         order_fields: Sequence[OrderFieldType] | None = None,
         facet_fields: Sequence[FacetFieldType] | None = None,
+        include_facets: bool = True,
         filter_by: dict[str, Any] | BaseModel | None = None,
         schema: type[BaseModel],
     ) -> CursorPaginatedResponse[Any]:
@@ -1430,6 +1475,8 @@ class AsyncCrud(Generic[ModelType]):
             search_column: Restrict search to a single column key.
             order_fields: Fields allowed for sorting (overrides class default).
             facet_fields: Columns to compute distinct values for (overrides class default).
+            include_facets: When ``False``, skip facet queries entirely;
+                ``filter_attributes`` will be ``None``.
             filter_by: Dict of {column_key: value} to filter by declared facet fields.
                 Keys must match the column.key of a facet field. Scalar → equality,
                 list → IN clause. Raises InvalidFacetFilterError for unknown keys.
@@ -1441,7 +1488,6 @@ class AsyncCrud(Generic[ModelType]):
         filters = list(filters) if filters else []
 
         fb_filters, search_joins = cls._prepare_filter_by(filter_by, facet_fields)
-        filters.extend(fb_filters)
 
         if cls.cursor_column is None:
             raise ValueError(
@@ -1473,6 +1519,11 @@ class AsyncCrud(Generic[ModelType]):
             filters.extend(search_filters)
             search_joins.extend(new_search_joins)
 
+        # Facets combine these with each facet's own filter individually, so
+        # fb_filters is applied to the query below but excluded here.
+        facet_base_filters = list(filters)
+        filters.extend(fb_filters.values())
+
         # Build query
         q = select(cls.model)
 
@@ -1480,11 +1531,11 @@ class AsyncCrud(Generic[ModelType]):
         q = _apply_joins(q, joins, outer_join)
 
         # Apply search joins (always outer joins)
-        q = _apply_search_joins(q, search_joins)
+        q = apply_search_joins(q, search_joins)
 
         # Apply order joins (relation joins required for order_by field)
         if order_joins:
-            q = _apply_search_joins(q, order_joins)
+            q = apply_search_joins(q, order_joins)
 
         if filters:
             q = q.where(and_(*filters))
@@ -1541,7 +1592,12 @@ class AsyncCrud(Generic[ModelType]):
         items: list[Any] = [schema.model_validate(item) for item in items_page]
 
         filter_attributes = await cls._build_filter_attributes(
-            session, facet_fields, filters, search_joins
+            session,
+            facet_fields,
+            facet_base_filters,
+            search_joins,
+            include_facets=include_facets,
+            own_filters=fb_filters,
         )
         search_columns = cls._resolve_search_columns(search_fields)
         order_columns = cls._resolve_order_columns(order_fields)
@@ -1581,6 +1637,7 @@ class AsyncCrud(Generic[ModelType]):
         search_column: str | None = ...,
         order_fields: Sequence[OrderFieldType] | None = ...,
         facet_fields: Sequence[FacetFieldType] | None = ...,
+        include_facets: bool = ...,
         filter_by: dict[str, Any] | BaseModel | None = ...,
         schema: type[BaseModel],
     ) -> OffsetPaginatedResponse[Any]: ...
@@ -1607,6 +1664,7 @@ class AsyncCrud(Generic[ModelType]):
         search_column: str | None = ...,
         order_fields: Sequence[OrderFieldType] | None = ...,
         facet_fields: Sequence[FacetFieldType] | None = ...,
+        include_facets: bool = ...,
         filter_by: dict[str, Any] | BaseModel | None = ...,
         schema: type[BaseModel],
     ) -> CursorPaginatedResponse[Any]: ...
@@ -1632,6 +1690,7 @@ class AsyncCrud(Generic[ModelType]):
         search_column: str | None = None,
         order_fields: Sequence[OrderFieldType] | None = None,
         facet_fields: Sequence[FacetFieldType] | None = None,
+        include_facets: bool = True,
         filter_by: dict[str, Any] | BaseModel | None = None,
         schema: type[BaseModel],
     ) -> OffsetPaginatedResponse[Any] | CursorPaginatedResponse[Any]:
@@ -1662,6 +1721,8 @@ class AsyncCrud(Generic[ModelType]):
             order_fields: Fields allowed for sorting (overrides class default).
             facet_fields: Columns to compute distinct values for (overrides
                 class default).
+            include_facets: When ``False``, skip facet queries entirely;
+                ``filter_attributes`` will be ``None``.
             filter_by: Dict of ``{column_key: value}`` to filter by declared
                 facet fields.  Keys must match the ``column.key`` of a facet
                 field.  Scalar → equality, list → IN clause.  Raises
@@ -1692,6 +1753,7 @@ class AsyncCrud(Generic[ModelType]):
                     search_column=search_column,
                     order_fields=order_fields,
                     facet_fields=facet_fields,
+                    include_facets=include_facets,
                     filter_by=filter_by,
                     schema=schema,
                 )
@@ -1714,6 +1776,7 @@ class AsyncCrud(Generic[ModelType]):
                     search_column=search_column,
                     order_fields=order_fields,
                     facet_fields=facet_fields,
+                    include_facets=include_facets,
                     filter_by=filter_by,
                     schema=schema,
                 )

src/fastapi_toolsets/crud/search.py

@@ -1,12 +1,12 @@
 """Search utilities for AsyncCrud."""
 
-import asyncio
 import functools
 from collections.abc import Sequence
 from dataclasses import dataclass, replace
 from typing import TYPE_CHECKING, Any, Literal
 
-from sqlalchemy import String, and_, func, or_, select
+from sqlalchemy import String, and_, distinct, func, or_, select
+from sqlalchemy.dialects.postgresql import aggregate_order_by
 from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.orm import DeclarativeBase
 from sqlalchemy.orm.attributes import InstrumentedAttribute
@@ -159,8 +159,11 @@ def build_search_filters(
         else:
             column = field
 
-        # Build the filter (cast to String for non-text columns)
-        column_as_string = column.cast(String)
+        # Build the filter (cast to String only when needed, to preserve
+        # pg_trgm GIN index usability on already-String columns)
+        column_as_string = (
+            column if isinstance(column.type, String) else column.cast(String)
+        )
         if config.case_sensitive:
             filters.append(column_as_string.like(f"%{query}%"))
         else:
@@ -181,6 +184,21 @@ def search_field_keys(fields: Sequence[SearchFieldType]) -> list[str]:
     return facet_keys(fields)
 
 
+def apply_search_joins(q: Any, joins: Sequence[Any]) -> Any:
+    """Apply relationship-based outer joins (from search/filter_by/facets) to a query.
+
+    Deduplicates by relationship identity so a join used by several fields
+    (e.g. search + a facet on the same relation) is only applied once.
+    """
+    seen: set[str] = set()
+    for rel in joins:
+        rel_key = str(rel)
+        if rel_key not in seen:
+            seen.add(rel_key)
+            q = q.outerjoin(rel)
+    return q
+
+
 def facet_keys(facet_fields: Sequence[FacetFieldType]) -> list[str]:
     """Return a key for each facet field.
 
@@ -207,6 +225,7 @@ async def build_facets(
     *,
     base_filters: "list[ColumnElement[bool]] | None" = None,
     base_joins: list[InstrumentedAttribute[Any]] | None = None,
+    own_filters: "dict[str, ColumnElement[bool]] | None" = None,
 ) -> dict[str, list[Any]]:
     """Return distinct values for each facet field, respecting current filters.
 
@@ -216,15 +235,24 @@ async def build_facets(
         facet_fields: Columns or relationship tuples to facet on
         base_filters: Filter conditions already applied to the main query (search + caller filters)
         base_joins: Relationship joins already applied to the main query
+        own_filters: Map of facet key -> the ``filter_by`` condition for that
+            same key (if any). Excluded from that facet's own subquery so
+            filtering on a facet doesn't collapse its own value list down to
+            just the filtered value.
 
     Returns:
         Dict mapping column key to sorted list of distinct non-None values
     """
-    existing_join_keys: set[str] = {str(j) for j in (base_joins or [])}
+    if not facet_fields:
+        return {}
 
     keys = facet_keys(facet_fields)
+    own_filters = own_filters or {}
 
-    async def _query_facet(field: FacetFieldType, key: str) -> tuple[str, list[Any]]:
+    scalars: list[Any] = []
+    enum_classes: dict[str, Any] = {}
+
+    for field, key in zip(facet_fields, keys):
         if isinstance(field, tuple):
             # Relationship chain: (User.role, Role.name) — last element is the column
             rels = field[:-1]
@@ -235,51 +263,48 @@ async def build_facets(
 
         col_type = column.property.columns[0].type
         is_array = isinstance(col_type, ARRAY)
+        enum_classes[key] = getattr(col_type, "enum_class", None)
 
-        if is_array:
-            unnested = func.unnest(column).label(column.key)
-            q = select(unnested).select_from(model).distinct()
-        else:
-            q = select(column).select_from(model).distinct()
-
-        # Apply base joins (deduplicated) — needed here independently
-        seen_joins: set[str] = set()
-        for rel in base_joins or []:
-            rel_key = str(rel)
-            if rel_key not in seen_joins:
-                seen_joins.add(rel_key)
-                q = q.outerjoin(rel)
-
-        # Add any extra joins required by this facet field that aren't already applied
-        for rel in rels:
-            rel_key = str(rel)
-            if rel_key not in existing_join_keys and rel_key not in seen_joins:
-                seen_joins.add(rel_key)
-                q = q.outerjoin(rel)
-
-        if base_filters:
-            q = q.where(and_(*base_filters))
-
-        if is_array:
-            q = q.order_by(unnested)
-        else:
-            q = q.order_by(column)
-        result = await session.execute(q)
-        col_type = column.property.columns[0].type
-        enum_class = getattr(col_type, "enum_class", None)
-        values = [
-            row[0].name
-            if (enum_class is not None and isinstance(row[0], enum_class))
-            else row[0]
-            for row in result.all()
-            if row[0] is not None
+        filters = [
+            *(base_filters or []),
+            *(f for k, f in own_filters.items() if k != key),
         ]
-        return key, values
+        joins = [*(base_joins or []), *rels]
 
-    pairs = await asyncio.gather(
-        *[_query_facet(f, k) for f, k in zip(facet_fields, keys)]
+        if is_array:
+            unnested = apply_search_joins(
+                select(func.unnest(column).label("v")).select_from(model), joins
             )
-    return dict(pairs)
+            if filters:
+                unnested = unnested.where(and_(*filters))
+            unnested_sq = unnested.subquery()
+            v = unnested_sq.c.v
+            agg = (
+                select(func.array_agg(aggregate_order_by(distinct(v), v)))
+                .select_from(unnested_sq)
+                .where(v.isnot(None))
+            )
+        else:
+            agg = apply_search_joins(
+                select(
+                    func.array_agg(aggregate_order_by(distinct(column), column))
+                ).select_from(model),
+                joins,
+            )
+            agg = agg.where(and_(*filters, column.isnot(None)))
+
+        scalars.append(agg.scalar_subquery().label(key))
+
+    row = (await session.execute(select(*scalars))).one()
+
+    facets: dict[str, list[Any]] = {}
+    for key, values in zip(keys, row):
+        enum_class = enum_classes[key]
+        facets[key] = [
+            v.name if (enum_class is not None and isinstance(v, enum_class)) else v
+            for v in (values or [])
+        ]
+    return facets
 
 
 _EQUALITY_TYPES = (String, Integer, Numeric, Date, DateTime, Time, Enum, Uuid)
@@ -301,7 +326,7 @@ def _coerce_bool(value: Any) -> bool:
 def build_filter_by(
     filter_by: dict[str, Any],
     facet_fields: Sequence[FacetFieldType],
-) -> tuple["list[ColumnElement[bool]]", list[InstrumentedAttribute[Any]]]:
+) -> tuple["dict[str, ColumnElement[bool]]", list[InstrumentedAttribute[Any]]]:
     """Translate a {column_key: value} dict into SQLAlchemy filter conditions.
 
     Args:
@@ -309,7 +334,9 @@ def build_filter_by(
         facet_fields: Declared facet fields to validate keys against
 
     Returns:
-        Tuple of (filter_conditions, joins_needed)
+        Tuple of ({facet_key: filter_condition}, joins_needed). One filter
+        condition per key, so callers can identify (and exclude) a facet's
+        own filter when computing that facet's distinct values.
 
     Raises:
         InvalidFacetFilterError: If a key in filter_by is not a declared facet field
@@ -327,7 +354,7 @@ def build_filter_by(
         index[key] = (column, rels)
 
     valid_keys = set(index)
-    filters: list[ColumnElement[bool]] = []
+    filters: dict[str, ColumnElement[bool]] = {}
     joins: list[InstrumentedAttribute[Any]] = []
     added_join_keys: set[str] = set()
 
@@ -347,14 +374,14 @@ def build_filter_by(
         if isinstance(col_type, Boolean):
             coerce = _coerce_bool
             if isinstance(value, list):
-                filters.append(column.in_([coerce(v) for v in value]))
+                filters[key] = column.in_([coerce(v) for v in value])
             else:
-                filters.append(column == coerce(value))
+                filters[key] = column == coerce(value)
         elif isinstance(col_type, ARRAY):
             if isinstance(value, list):
-                filters.append(column.overlap(value))
+                filters[key] = column.overlap(value)
             else:
-                filters.append(column.any(value))
+                filters[key] = column.any(value)
         elif isinstance(col_type, Enum):
             enum_class = col_type.enum_class
             if enum_class is not None:
@@ -365,19 +392,19 @@ def build_filter_by(
                     return enum_class[v]  # lookup by name: "PENDING", "RED"
 
                 if isinstance(value, list):
-                    filters.append(column.in_([_coerce_enum(v) for v in value]))
+                    filters[key] = column.in_([_coerce_enum(v) for v in value])
                 else:
-                    filters.append(column == _coerce_enum(value))
+                    filters[key] = column == _coerce_enum(value)
             else:  # pragma: no cover
                 if isinstance(value, list):
-                    filters.append(column.in_(value))
+                    filters[key] = column.in_(value)
                 else:
-                    filters.append(column == value)
+                    filters[key] = column == value
         elif isinstance(col_type, _EQUALITY_TYPES):
             if isinstance(value, list):
-                filters.append(column.in_(value))
+                filters[key] = column.in_(value)
             else:
-                filters.append(column == value)
+                filters[key] = column == value
         else:
             raise UnsupportedFacetTypeError(key, type(col_type).__name__)
 

src/fastapi_toolsets/models/watched.py

@@ -5,7 +5,7 @@ from collections.abc import Callable
 from enum import Enum
 from typing import Any
 
-from sqlalchemy import event
+from sqlalchemy import event, select, tuple_
 from sqlalchemy import inspect as sa_inspect
 from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.orm.attributes import set_committed_value as _sa_set_committed_value
@@ -29,14 +29,11 @@ _SESSION_DELETES = "_ft_deletes"
 _SESSION_UPDATES = "_ft_updates"
 _DEFERRED_STRATEGY_KEY = (("deferred", True), ("instrument", True))
 _EVENT_HANDLERS: dict[tuple[type, ModelEvent], list[Callable[..., Any]]] = {}
-_WATCHED_MODELS: set[type] = set()
-_WATCHED_CACHE: dict[type, bool] = {}
 _HANDLER_CACHE: dict[tuple[type, ModelEvent], list[Callable[..., Any]]] = {}
 
 
 def _invalidate_caches() -> None:
     """Clear lookup caches after handler registration."""
-    _WATCHED_CACHE.clear()
     _HANDLER_CACHE.clear()
 
 
@@ -56,24 +53,12 @@ def listens_for(
     def decorator(fn: Callable[..., Any]) -> Callable[..., Any]:
         for ev in evs:
             _EVENT_HANDLERS.setdefault((model_class, ev), []).append(fn)
-        _WATCHED_MODELS.add(model_class)
         _invalidate_caches()
         return fn
 
     return decorator
 
 
-def _is_watched(obj: Any) -> bool:
-    """Return True if *obj*'s type (or any ancestor) has registered handlers."""
-    cls = type(obj)
-    try:
-        return _WATCHED_CACHE[cls]
-    except KeyError:
-        result = any(klass in _WATCHED_MODELS for klass in cls.__mro__)
-        _WATCHED_CACHE[cls] = result
-        return result
-
-
 def _get_handlers(cls: type, ev: ModelEvent) -> list[Callable[..., Any]]:
     """Return registered handlers for *cls* and *ev*, walking the MRO."""
     key = (cls, ev)
@@ -144,18 +129,18 @@ def _upsert_changes(
 def _after_flush(session: Any, flush_context: Any) -> None:
     # New objects: capture reference. Attributes will be refreshed after commit.
     for obj in session.new:
-        if _is_watched(obj):
+        if _get_handlers(type(obj), ModelEvent.CREATE):
             session.info.setdefault(_SESSION_CREATES, []).append(obj)
 
     # Deleted objects: snapshot now while attributes are still loaded.
     for obj in session.deleted:
-        if _is_watched(obj):
+        if _get_handlers(type(obj), ModelEvent.DELETE):
             snapshot = _snapshot_column_attrs(obj)
             session.info.setdefault(_SESSION_DELETES, []).append((obj, snapshot))
 
     # Dirty objects: read old/new from SQLAlchemy attribute history.
     for obj in session.dirty:
-        if not _is_watched(obj):
+        if not _get_handlers(type(obj), ModelEvent.UPDATE):
             continue
 
         watched = _get_watched_fields(type(obj))
@@ -204,9 +189,18 @@ async def _invoke_callback(
         await result
 
 
-async def _reload_if_present(session: AsyncSession, obj: Any, state: Any) -> None:
-    """Re-populate *obj* from the DB if its row still exists."""
-    await session.get(type(obj), state.key[1], populate_existing=True)
+async def _batch_reload(
+    session: AsyncSession, model: type, pk_tuples: list[tuple[Any, ...]]
+) -> None:
+    """Re-populate all rows of *model* identified by *pk_tuples* in one round trip."""
+    pk_cols = sa_inspect(model).primary_key
+    where = (
+        pk_cols[0].in_([pk[0] for pk in pk_tuples])
+        if len(pk_cols) == 1
+        else tuple_(*pk_cols).in_(pk_tuples)
+    )
+    q = select(model).where(where).execution_options(populate_existing=True)
+    await session.execute(q)
 
 
 class EventSession(AsyncSession):
@@ -250,15 +244,36 @@ class EventSession(AsyncSession):
                 k: v for k, v in field_changes.items() if k not in create_ids
             }
 
-        # Dispatch CREATE callbacks.
+        # Resolve reloadable state up front and group PKs by model type so
+        # the post-commit reload is one query per type instead of one
+        # session.get() per object.
+        create_items: list[Any] = []
+        update_items: list[tuple[Any, dict[str, dict[str, Any]]]] = []
+        pk_by_type: dict[type, list[tuple[Any, ...]]] = {}
+
         for obj in creates:
-            try:
             state = sa_inspect(obj, raiseerr=False)
-                if (
-                    state is None or state.detached or state.transient
-                ):  # pragma: no cover
+            if state is None or state.detached or state.transient:  # pragma: no cover
                 continue
-                await _reload_if_present(self, obj, state)
+            create_items.append(obj)
+            pk_by_type.setdefault(type(obj), []).append(state.key[1])
+
+        for obj, changes in field_changes.values():
+            state = sa_inspect(obj, raiseerr=False)
+            if state is None or state.detached or state.transient:  # pragma: no cover
+                continue
+            update_items.append((obj, changes))
+            pk_by_type.setdefault(type(obj), []).append(state.key[1])
+
+        for model, pk_tuples in pk_by_type.items():
+            try:
+                await _batch_reload(self, model, pk_tuples)
+            except Exception as exc:
+                _logger.error(_CALLBACK_ERROR_MSG, exc_info=exc)
+
+        # Dispatch CREATE callbacks.
+        for obj in create_items:
+            try:
                 for handler in _get_handlers(type(obj), ModelEvent.CREATE):
                     await _invoke_callback(handler, obj, ModelEvent.CREATE, None)
             except Exception as exc:
@@ -275,14 +290,8 @@ class EventSession(AsyncSession):
                 _logger.error(_CALLBACK_ERROR_MSG, exc_info=exc)
 
         # Dispatch UPDATE callbacks.
-        for obj, changes in field_changes.values():
+        for obj, changes in update_items:
             try:
-                state = sa_inspect(obj, raiseerr=False)
-                if (
-                    state is None or state.detached or state.transient
-                ):  # pragma: no cover
-                    continue
-                await _reload_if_present(self, obj, state)
                 for handler in _get_handlers(type(obj), ModelEvent.UPDATE):
                     await _invoke_callback(handler, obj, ModelEvent.UPDATE, changes)
             except Exception as exc:

tests/conftest.py

@@ -476,3 +476,26 @@ async def db_session(engine):
         # Drop tables after test
         async with engine.begin() as conn:
             await conn.run_sync(Base.metadata.drop_all)
+
+
+@pytest.fixture(scope="function")
+async def db_session_expire_on_commit(engine):
+    """Session with expire_on_commit=True (the SQLAlchemy default).
+
+    Attributes read off an instance after commit are expired and trigger an
+    implicit (sync) refresh under this setting — which fails under asyncio
+    with MissingGreenlet. The other ``db_session`` fixture uses
+    ``expire_on_commit=False`` and would not catch that class of bug.
+    """
+    async with engine.begin() as conn:
+        await conn.run_sync(Base.metadata.create_all)
+
+    session_factory = async_sessionmaker(engine, expire_on_commit=True)
+    session = session_factory()
+
+    try:
+        yield session
+    finally:
+        await session.close()
+        async with engine.begin() as conn:
+            await conn.run_sync(Base.metadata.drop_all)

tests/test_crud.py

@@ -417,6 +417,55 @@ class TestDefaultLoadOptionsIntegration:
         assert updated.role is not None
         assert updated.role.name == "admin"
 
+    @pytest.mark.anyio
+    async def test_default_load_options_applied_to_create_expire_on_commit(
+        self, db_session_expire_on_commit: AsyncSession
+    ):
+        """create()'s reload uses captured PK values, not an expired instance attribute.
+
+        Regression test for MissingGreenlet: reading a PK off `db_model` after
+        commit under expire_on_commit=True (the SQLAlchemy default) would
+        trigger an implicit sync refresh, which fails under asyncio.
+        """
+        UserWithDefaultLoad = CrudFactory(
+            User, default_load_options=[selectinload(User.role)]
+        )
+        role = await RoleCrud.create(
+            db_session_expire_on_commit, RoleCreate(name="admin")
+        )
+        user = await UserWithDefaultLoad.create(
+            db_session_expire_on_commit,
+            UserCreate(username="alice", email="alice@test.com", role_id=role.id),
+        )
+        assert user.role is not None
+        assert user.role.name == "admin"
+
+    @pytest.mark.anyio
+    async def test_default_load_options_applied_to_update_expire_on_commit(
+        self, db_session_expire_on_commit: AsyncSession
+    ):
+        """update()'s reload uses captured PK values, not an expired instance attribute.
+
+        Regression test for MissingGreenlet under expire_on_commit=True.
+        """
+        UserWithDefaultLoad = CrudFactory(
+            User, default_load_options=[selectinload(User.role)]
+        )
+        role = await RoleCrud.create(
+            db_session_expire_on_commit, RoleCreate(name="admin")
+        )
+        user = await UserCrud.create(
+            db_session_expire_on_commit,
+            UserCreate(username="alice", email="alice@test.com"),
+        )
+        updated = await UserWithDefaultLoad.update(
+            db_session_expire_on_commit,
+            UserUpdate(role_id=role.id),
+            filters=[User.id == user.id],
+        )
+        assert updated.role is not None
+        assert updated.role.name == "admin"
+
     @pytest.mark.anyio
     async def test_load_options_overrides_default_load_options(
         self, db_session: AsyncSession

tests/test_crud_search.py

@@ -372,6 +372,22 @@ class TestBuildSearchFilters:
 
         assert len(joins) == 1
 
+    def test_skips_cast_on_string_column(self):
+        """String columns are filtered directly, without a CAST (keeps pg_trgm indexable)."""
+        from fastapi_toolsets.crud.search import build_search_filters
+
+        filters, _ = build_search_filters(User, "john", search_fields=[User.username])
+
+        assert "CAST" not in str(filters[0])
+
+    def test_casts_non_string_column(self):
+        """Non-string columns (e.g. UUID) still get cast to String so ilike works."""
+        from fastapi_toolsets.crud.search import build_search_filters
+
+        filters, _ = build_search_filters(User, "123", search_fields=[User.id])
+
+        assert "CAST" in str(filters[0])
+
 
 class TestSearchConfig:
     """Tests for SearchConfig options."""
@@ -531,6 +547,15 @@ class TestFacetsNotSet:
 
         assert result.filter_attributes is None
 
+    @pytest.mark.anyio
+    async def test_build_facets_empty_field_list(self, db_session: AsyncSession):
+        """build_facets([]) is a no-op that returns {} without querying — the escape hatch."""
+        from fastapi_toolsets.crud.search import build_facets
+
+        result = await build_facets(db_session, User, [])
+
+        assert result == {}
+
 
 class TestFacetsDirectColumn:
     """Facets on direct model columns."""
@@ -606,6 +631,91 @@ class TestFacetsDirectColumn:
         assert "username" not in result.filter_attributes
 
 
+class TestFacetsMixedTypes:
+    """Facet values keep their native Python type through the batched query."""
+
+    @pytest.mark.anyio
+    async def test_enum_and_integer_facets_preserve_types(
+        self, db_session: AsyncSession
+    ):
+        """Enum facets return member names (not raw DB values); Integer facets return ints."""
+        OrderMixedCrud = CrudFactory(
+            Order, facet_fields=[Order.status, Order.priority, Order.color]
+        )
+        await OrderCrud.create(
+            db_session,
+            OrderCreate(
+                name="order-1", status=OrderStatus.PENDING, priority=1, color=Color.RED
+            ),
+        )
+        await OrderCrud.create(
+            db_session,
+            OrderCreate(
+                name="order-2", status=OrderStatus.SHIPPED, priority=3, color=Color.BLUE
+            ),
+        )
+
+        result = await OrderMixedCrud.offset_paginate(db_session, schema=OrderRead)
+
+        assert result.filter_attributes is not None
+        assert set(result.filter_attributes["status"]) == {"PENDING", "SHIPPED"}
+        assert all(isinstance(v, str) for v in result.filter_attributes["status"])
+        assert set(result.filter_attributes["priority"]) == {1, 3}
+        assert all(isinstance(v, int) for v in result.filter_attributes["priority"])
+        assert set(result.filter_attributes["color"]) == {"RED", "BLUE"}
+
+    @pytest.mark.anyio
+    async def test_bool_facet_keeps_python_bool(self, db_session: AsyncSession):
+        """A Boolean facet returns Python bool values, not stringified 'true'/'false'."""
+        UserBoolFacetCrud = CrudFactory(User, facet_fields=[User.is_active])
+        await UserCrud.create(
+            db_session, UserCreate(username="alice", email="a@test.com", is_active=True)
+        )
+        await UserCrud.create(
+            db_session, UserCreate(username="bob", email="b@test.com", is_active=False)
+        )
+
+        result = await UserBoolFacetCrud.offset_paginate(db_session, schema=UserRead)
+
+        assert result.filter_attributes is not None
+        assert set(result.filter_attributes["is_active"]) == {True, False}
+        assert all(isinstance(v, bool) for v in result.filter_attributes["is_active"])
+
+
+class TestIncludeFacets:
+    """include_facets=False skips facet queries entirely."""
+
+    @pytest.mark.anyio
+    async def test_offset_paginate_include_facets_false(self, db_session: AsyncSession):
+        """filter_attributes is None when include_facets=False, even with facet_fields set."""
+        UserFacetCrud = CrudFactory(User, facet_fields=[User.username])
+        await UserCrud.create(
+            db_session, UserCreate(username="alice", email="a@test.com")
+        )
+
+        result = await UserFacetCrud.offset_paginate(
+            db_session, include_facets=False, schema=UserRead
+        )
+
+        assert result.filter_attributes is None
+
+    @pytest.mark.anyio
+    async def test_cursor_paginate_include_facets_false(self, db_session: AsyncSession):
+        """filter_attributes is None when include_facets=False for cursor_paginate."""
+        UserFacetCursorCrud = CrudFactory(
+            User, cursor_column=User.id, facet_fields=[User.username]
+        )
+        await UserCrud.create(
+            db_session, UserCreate(username="alice", email="a@test.com")
+        )
+
+        result = await UserFacetCursorCrud.cursor_paginate(
+            db_session, include_facets=False, schema=UserRead
+        )
+
+        assert result.filter_attributes is None
+
+
 class TestFacetsRespectFilters:
     """Facets reflect the active filter conditions."""
 
@@ -630,6 +740,28 @@ class TestFacetsRespectFilters:
         assert result.filter_attributes is not None
         assert result.filter_attributes["username"] == ["alice"]
 
+    @pytest.mark.anyio
+    async def test_array_facet_respects_unrelated_filter(
+        self, db_session: AsyncSession
+    ):
+        """An ARRAY facet is scoped by a filter on a different column (not self-collapse)."""
+        ArticleFacetCrud = CrudFactory(Article, facet_fields=[Article.labels])
+        await ArticleCrud.create(
+            db_session, ArticleCreate(title="Post 1", labels=["python", "fastapi"])
+        )
+        await ArticleCrud.create(
+            db_session, ArticleCreate(title="Post 2", labels=["rust", "axum"])
+        )
+
+        result = await ArticleFacetCrud.offset_paginate(
+            db_session,
+            filters=[Article.title == "Post 1"],
+            schema=ArticleRead,
+        )
+
+        assert result.filter_attributes is not None
+        assert result.filter_attributes["labels"] == ["fastapi", "python"]
+
 
 class TestFacetsRelationship:
     """Facets on relationship columns via tuple syntax."""
@@ -785,8 +917,8 @@ class TestFilterBy:
 
         assert len(result.data) == 1
         assert result.data[0].username == "alice"
-        # facet also scoped to the filter
-        assert result.filter_attributes == {"username": ["alice"]}
+        # facet excludes its own filter_by condition, so it isn't collapsed
+        assert result.filter_attributes == {"username": ["alice", "bob"]}
 
     @pytest.mark.anyio
     async def test_list_filter_produces_in_clause(self, db_session: AsyncSession):
@@ -924,7 +1056,7 @@ class TestFilterBy:
 
         assert len(result.data) == 1
         assert result.data[0].username == "alice"
-        assert result.filter_attributes == {"username": ["alice"]}
+        assert result.filter_attributes == {"username": ["alice", "bob"]}
 
     @pytest.mark.anyio
     async def test_basemodel_filter_by_offset_paginate(self, db_session: AsyncSession):
@@ -1085,8 +1217,10 @@ class TestFilterBy:
         assert result.pagination.total_count == 2
         titles = {a.title for a in result.data}
         assert titles == {"Post 1", "Post 3"}
-        # facet returns individual unnested values, not whole arrays
-        assert result.filter_attributes == {"labels": ["django", "fastapi", "python"]}
+        # facet excludes its own filter_by condition (not collapsed to matching rows)
+        assert result.filter_attributes == {
+            "labels": ["axum", "django", "fastapi", "python", "rust"]
+        }
 
     @pytest.mark.anyio
     async def test_array_overlap_list_value(self, db_session: AsyncSession):
@@ -2179,7 +2313,12 @@ class TestOffsetPaginateParamsSchema:
             include_total=False, search=False, filter=False, order=False
         )
         result = await dep(page=2, items_per_page=10)
-        assert result == {"page": 2, "items_per_page": 10, "include_total": False}
+        assert result == {
+            "page": 2,
+            "items_per_page": 10,
+            "include_total": False,
+            "include_facets": True,
+        }
 
     @pytest.mark.anyio
     async def test_integrates_with_offset_paginate(self, db_session: AsyncSession):
@@ -2290,7 +2429,11 @@ class TestCursorPaginateParamsSchema:
             search=False, filter=False, order=False
         )
         result = await dep(cursor=None, items_per_page=5)
-        assert result == {"cursor": None, "items_per_page": 5}
+        assert result == {
+            "cursor": None,
+            "items_per_page": 5,
+            "include_facets": True,
+        }
 
     @pytest.mark.anyio
     async def test_integrates_with_cursor_paginate(self, db_session: AsyncSession):
@@ -2399,6 +2542,7 @@ class TestPaginateParamsSchema:
             "cursor": None,
             "items_per_page": 10,
             "include_total": True,
+            "include_facets": True,
         }
 
     @pytest.mark.anyio

tests/test_example_pagination_search.py

@@ -199,8 +199,9 @@ class TestOffsetPagination:
         resp = await client.get("/articles/offset?status=published")
 
         body = resp.json()
-        # draft is filtered out → should not appear in filter_attributes
-        assert "draft" not in body["filter_attributes"]["status"]
+        # a facet excludes its own filter_by condition, so filtering by
+        # status=published still shows every status the facet offers
+        assert "draft" in body["filter_attributes"]["status"]
 
     @pytest.mark.anyio
     async def test_search_and_filter_combined(self, client: AsyncClient, ex_db_session):

tests/test_models.py

@@ -25,13 +25,11 @@ from fastapi_toolsets.models.watched import (
     _SESSION_CREATES,
     _SESSION_DELETES,
     _SESSION_UPDATES,
-    _WATCHED_MODELS,
     EventSession,
     _after_flush,
     _after_rollback,
     _get_watched_fields,
     _invalidate_caches,
-    _is_watched,
     _snapshot_column_attrs,
     _upsert_changes,
 )
@@ -658,22 +656,6 @@ class TestWatchInheritance:
         assert "other" in _watch_inherit_events[0]["changes"]
 
 
-class TestIsWatched:
-    def test_watched_model_is_watched(self):
-        """_is_watched returns True for models with registered handlers."""
-        obj = WatchedModel(status="x", other="y")
-        assert _is_watched(obj) is True
-
-    def test_non_watched_model_is_not_watched(self):
-        """_is_watched returns False for models without registered handlers."""
-        assert _is_watched(object()) is False
-
-    def test_subclass_of_watched_model_is_watched(self):
-        """_is_watched returns True for subclasses of watched models (via MRO)."""
-        dog = PolyDog(name="Rex")
-        assert _is_watched(dog) is True
-
-
 class TestUpsertChanges:
     def test_inserts_new_entry(self):
         """New key is inserted with the full changes dict."""
@@ -715,7 +697,10 @@ class TestAfterFlush:
         """New watched objects are added to _SESSION_CREATES."""
         obj = object()
         session = SimpleNamespace(new=[obj], deleted=[], dirty=[], info={})
-        with patch("fastapi_toolsets.models.watched._is_watched", return_value=True):
+        with patch(
+            "fastapi_toolsets.models.watched._get_handlers",
+            return_value=[lambda *a: None],
+        ):
             _after_flush(session, None)
         assert session.info[_SESSION_CREATES] == [obj]
 
@@ -731,7 +716,10 @@ class TestAfterFlush:
         obj = object()
         session = SimpleNamespace(new=[], deleted=[obj], dirty=[], info={})
         with (
-            patch("fastapi_toolsets.models.watched._is_watched", return_value=True),
+            patch(
+                "fastapi_toolsets.models.watched._get_handlers",
+                return_value=[lambda *a: None],
+            ),
             patch(
                 "fastapi_toolsets.models.watched._snapshot_column_attrs",
                 return_value={"id": 1},
@@ -1023,28 +1011,19 @@ class TestEventCallbacks:
                 await other.commit()
             await engine.dispose()
 
-        real_get = mixin_session.get
-        real_refresh = mixin_session.refresh
+        real_batch_reload = _watched_module._batch_reload
 
-        def _matches_doomed(pk):
-            return pk == doomed_id or (isinstance(pk, tuple) and pk[0] == doomed_id)
-
-        async def racing_get(model, pk, *args, **kwargs):
-            if _matches_doomed(pk):
+        async def racing_batch_reload(session, model, pk_tuples):
+            if any(pk[0] == doomed_id for pk in pk_tuples):
                 await kill_doomed_row_once()
-            return await real_get(model, pk, *args, **kwargs)
+            return await real_batch_reload(session, model, pk_tuples)
 
-        async def racing_refresh(obj, *args, **kwargs):
-            if getattr(obj, "id", None) == doomed_id:
-                await kill_doomed_row_once()
-            return await real_refresh(obj, *args, **kwargs)
-
-        # Patch both possible reload mechanisms (session.get / session.refresh)
-        # so this test still exercises the race regardless of which one
-        # EventSession.commit() uses internally to pick up server defaults.
-        mixin_session.get = racing_get
-        mixin_session.refresh = racing_refresh
-        with patch.object(_watched_module._logger, "error") as mock_error:
+        # Patch the batched reload EventSession.commit() uses to pick up
+        # server defaults, so this test still exercises the race.
+        with (
+            patch.object(_watched_module, "_batch_reload", racing_batch_reload),
+            patch.object(_watched_module._logger, "error") as mock_error,
+        ):
             await mixin_session.commit()
             mock_error.assert_not_called()
 
@@ -1052,6 +1031,27 @@ class TestEventCallbacks:
         created_ids = {e["obj_id"] for e in _test_events if e["event"] == "create"}
         assert created_ids == {keep.id, doomed_id}
 
+    @pytest.mark.anyio
+    async def test_batch_reload_exception_is_logged_and_dispatch_continues(
+        self, mixin_session
+    ):
+        """A batched-reload failure is logged; CREATE handlers still fire."""
+        obj = WatchedModel(status="active", other="x")
+        mixin_session.add(obj)
+
+        async def failing_batch_reload(session, model, pk_tuples):
+            raise RuntimeError("reload failed")
+
+        with (
+            patch.object(_watched_module, "_batch_reload", failing_batch_reload),
+            patch.object(_watched_module._logger, "error") as mock_error,
+        ):
+            await mixin_session.commit()
+
+        mock_error.assert_called_once()
+        creates = [e for e in _test_events if e["event"] == "create"]
+        assert len(creates) == 1
+
 
 class TestTransientObject:
     """Create + delete within the same transaction should fire no events."""
@@ -1421,7 +1421,6 @@ class TestListensFor:
         for key in list(_EVENT_HANDLERS):
             if key[0] is ListenerModel:
                 del _EVENT_HANDLERS[key]
-        _WATCHED_MODELS.discard(ListenerModel)
         _invalidate_caches()
 
     @pytest.mark.anyio