Add include/exclude model filters

Author: successbyfailure

alembic/versions/006_add_routing_groups.py

@@ -0,0 +1,73 @@
+"""Add routing group tables.
+
+Revision ID: 006
+Revises: 005
+Create Date: 2026-01-05 00:00:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = "006"
+down_revision: Union[str, None] = "005"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "routing_groups",
+        sa.Column("id", sa.Integer, primary_key=True),
+        sa.Column("name", sa.String, nullable=False, unique=True, index=True),
+        sa.Column("description", sa.Text, nullable=True),
+        sa.Column("capabilities", sa.Text, nullable=True),
+        sa.Column("created_at", sa.DateTime, nullable=False),
+        sa.Column("updated_at", sa.DateTime, nullable=False),
+    )
+
+    op.create_table(
+        "routing_targets",
+        sa.Column("id", sa.Integer, primary_key=True),
+        sa.Column("group_id", sa.Integer, sa.ForeignKey("routing_groups.id", ondelete="CASCADE"), nullable=False),
+        sa.Column("provider_id", sa.Integer, sa.ForeignKey("providers.id", ondelete="CASCADE"), nullable=False),
+        sa.Column("model_id", sa.String, nullable=False),
+        sa.Column("weight", sa.Integer, nullable=False, server_default="1"),
+        sa.Column("priority", sa.Integer, nullable=False, server_default="0"),
+        sa.Column("enabled", sa.Boolean, nullable=False, server_default="1"),
+        sa.Column("created_at", sa.DateTime, nullable=False),
+        sa.Column("updated_at", sa.DateTime, nullable=False),
+        sa.UniqueConstraint("group_id", "provider_id", "model_id", name="uq_routing_target"),
+    )
+    op.create_index("ix_routing_targets_group_id", "routing_targets", ["group_id"])
+    op.create_index("ix_routing_targets_provider_id", "routing_targets", ["provider_id"])
+    op.create_index("ix_routing_targets_priority", "routing_targets", ["priority"])
+
+    op.create_table(
+        "routing_provider_limits",
+        sa.Column("id", sa.Integer, primary_key=True),
+        sa.Column("group_id", sa.Integer, sa.ForeignKey("routing_groups.id", ondelete="CASCADE"), nullable=False),
+        sa.Column("provider_id", sa.Integer, sa.ForeignKey("providers.id", ondelete="CASCADE"), nullable=False),
+        sa.Column("max_requests_per_hour", sa.Integer, nullable=True),
+        sa.Column("created_at", sa.DateTime, nullable=False),
+        sa.Column("updated_at", sa.DateTime, nullable=False),
+        sa.UniqueConstraint("group_id", "provider_id", name="uq_routing_provider_limit"),
+    )
+    op.create_index("ix_routing_provider_limits_group_id", "routing_provider_limits", ["group_id"])
+    op.create_index("ix_routing_provider_limits_provider_id", "routing_provider_limits", ["provider_id"])
+
+
+def downgrade() -> None:
+    op.drop_index("ix_routing_provider_limits_provider_id", table_name="routing_provider_limits")
+    op.drop_index("ix_routing_provider_limits_group_id", table_name="routing_provider_limits")
+    op.drop_table("routing_provider_limits")
+
+    op.drop_index("ix_routing_targets_priority", table_name="routing_targets")
+    op.drop_index("ix_routing_targets_provider_id", table_name="routing_targets")
+    op.drop_index("ix_routing_targets_group_id", table_name="routing_targets")
+    op.drop_table("routing_targets")
+
+    op.drop_table("routing_groups")

backend/__init__.py

@@ -1,3 +1,3 @@
 """Backend sync worker service."""
 
-__version__ = "0.6.20"
+__version__ = "0.6.21"

backend/provider_sync.py

@@ -11,6 +11,23 @@
 logger = logging.getLogger(__name__)
 
 
+def _parse_filter_terms(raw: str | None) -> list[str]:
+    """Split comma-separated filter terms into normalized lowercase tokens."""
+    if not raw:
+        return []
+    return [term.strip().lower() for term in raw.split(",") if term.strip()]
+
+
+def _matches_include_exclude(model_id: str, include_terms: list[str], exclude_terms: list[str]) -> bool:
+    """Return True if model_id matches include terms and not excluded."""
+    model_id_lower = model_id.lower()
+    if exclude_terms and any(term in model_id_lower for term in exclude_terms):
+        return False
+    if include_terms:
+        return any(term in model_id_lower for term in include_terms)
+    return True
+
+
 def _has_user_overrides(model: Model) -> bool:
     """
     Check if a model has any user-applied configuration overrides.
@@ -71,16 +88,19 @@ async def sync_provider(session, config, provider, push_to_litellm: bool = True)
         logger.info("Found %d models from %s", len(source_models.models), provider.name)
 
         # Apply model filter if configured
-        if provider.model_filter:
+        include_terms = _parse_filter_terms(provider.model_filter)
+        exclude_terms = _parse_filter_terms(provider.model_filter_exclude)
+        if include_terms or exclude_terms:
             original_count = len(source_models.models)
             filtered_models = [
                 m for m in source_models.models
-                if provider.model_filter.lower() in m.id.lower()
+                if _matches_include_exclude(m.id, include_terms, exclude_terms)
             ]
             source_models.models = filtered_models
             logger.info(
-                "Applied filter '%s': %d models matched (filtered out %d)",
-                provider.model_filter,
+                "Applied filters include='%s' exclude='%s': %d models matched (filtered out %d)",
+                provider.model_filter or "",
+                provider.model_filter_exclude or "",
                 len(filtered_models),
                 original_count - len(filtered_models)
             )

frontend/__init__.py

@@ -1,3 +1,3 @@
 """Frontend API and UI service."""
 
-__version__ = "0.6.20"
+__version__ = "0.6.21"

frontend/api.py

@@ -16,7 +16,7 @@
 
 from shared.database import create_engine, init_session_maker, get_session, ensure_minimum_schema
 from shared.crud import get_all_providers, get_config, get_provider_by_id
-from frontend.routes import providers, models, admin, compat, litellm
+from frontend.routes import providers, models, admin, compat, litellm, routing_groups
 from backend import provider_sync
 from sqlalchemy import select, func, case
 from shared.db_models import Model, Provider
@@ -96,6 +96,7 @@ def _human_source_type(source_type: str) -> str:
     app.include_router(models.router, prefix="/api/models", tags=["models"])
     app.include_router(admin.router, prefix="/api/admin", tags=["admin"])
     app.include_router(compat.router, prefix="/api/compat", tags=["compat"])
+    app.include_router(routing_groups.router, prefix="/api/routing-groups", tags=["routing-groups"])
     app.include_router(litellm.router, prefix="/litellm", tags=["litellm"])
 
     # HTML Routes
@@ -359,6 +360,13 @@ async def settings_page(request: Request, session = Depends(get_session)):
             "config": config_dict
         })
 
+    @app.get("/routing", response_class=HTMLResponse)
+    async def routing_groups_page(request: Request):
+        """Routing group configuration page."""
+        return templates.TemplateResponse("routing_groups.html", {
+            "request": request
+        })
+
     @app.post("/sync")
     async def manual_sync(request: Request, session = Depends(get_session)):
         """
@@ -451,6 +459,7 @@ async def create_provider_legacy(
         prefix: str | None = Form(None),
         default_ollama_mode: str | None = Form(None),
         model_filter: str | None = Form(None),
+        model_filter_exclude: str | None = Form(None),
         sync_interval_seconds: int | None = Form(None),
         session: AsyncSession = Depends(get_session)
     ):
@@ -465,6 +474,7 @@ async def create_provider_legacy(
             prefix=prefix,
             default_ollama_mode=default_ollama_mode,
             model_filter=model_filter,
+            model_filter_exclude=model_filter_exclude,
             sync_interval_seconds=sync_interval_seconds
         )
         from fastapi.responses import RedirectResponse

frontend/routes/providers.py

@@ -78,6 +78,7 @@ async def list_providers(session: AsyncSession = Depends(get_session)):
             "default_ollama_mode": p.default_ollama_mode,
             "auto_detect_fim": p.auto_detect_fim,
             "model_filter": p.model_filter,
+            "model_filter_exclude": p.model_filter_exclude,
             "tags": p.tags_list,
             "access_groups": p.access_groups_list,
             "sync_enabled": p.sync_enabled,
@@ -282,6 +283,7 @@ async def get_provider(provider_id: int, session: AsyncSession = Depends(get_ses
         "default_ollama_mode": provider.default_ollama_mode,
         "auto_detect_fim": provider.auto_detect_fim,
         "model_filter": provider.model_filter,
+        "model_filter_exclude": provider.model_filter_exclude,
         "tags": provider.tags_list,
         "access_groups": provider.access_groups_list,
         "sync_enabled": provider.sync_enabled,
@@ -301,6 +303,7 @@ async def add_provider(
     prefix: str | None = Form(None),
     default_ollama_mode: str | None = Form(None),
     model_filter: str | None = Form(None),
+    model_filter_exclude: str | None = Form(None),
     tags: str | None = Form(None),
     access_groups: str | None = Form(None),
     sync_enabled: bool | None = Form(True),
@@ -327,6 +330,7 @@ async def add_provider(
         prefix=_normalize_optional_str(prefix),
         default_ollama_mode=_normalize_optional_str(default_ollama_mode),
         model_filter=_normalize_optional_str(model_filter),
+        model_filter_exclude=_normalize_optional_str(model_filter_exclude),
         tags=_parse_csv_list(tags),
         access_groups=_parse_csv_list(access_groups),
         sync_enabled=sync_enabled_val,
@@ -393,6 +397,7 @@ async def update_provider_endpoint(
     prefix: str | None = Form(None),
     default_ollama_mode: str | None = Form(None),
     model_filter: str | None = Form(None),
+    model_filter_exclude: str | None = Form(None),
     tags: str | None = Form(None),
     access_groups: str | None = Form(None),
     sync_enabled: bool | None = Form(None),
@@ -418,6 +423,7 @@ async def update_provider_endpoint(
         prefix=_normalize_optional_str(prefix),
         default_ollama_mode=_normalize_optional_str(default_ollama_mode),
         model_filter=_normalize_optional_str(model_filter),
+        model_filter_exclude=_normalize_optional_str(model_filter_exclude),
         tags=_parse_csv_list(tags),
         access_groups=_parse_csv_list(access_groups),
         sync_enabled=_parse_bool(sync_enabled),