chore(schemas): clean model schemas (provider and routers) (#783)

* chore(schemas): clean model schemas (provider and routers)

* chore(actions): exclude node mudule from secrets scan

Author: leoguillaume

.github/.trufflehog-exclude.txt

@@ -0,0 +1,2 @@
+docs/node_modules/
+node_modules/

.github/workflows/secrets_scan.yml

@@ -20,6 +20,7 @@ jobs:
         extra_args: |
           --results=verified,unknown
           --exclude-detectors=Postgres
+          --exclude-paths=./.github/.trufflehog-exclude.txt
     - name: Install git-secrets
       run: |
         git clone https://github.com/awslabs/git-secrets.git

api/domain/model/entities.py

@@ -1,21 +1,21 @@
-from enum import Enum
+from enum import StrEnum
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel
 
 
-class Metric(str, Enum):
+class Metric(StrEnum):
     TTFT = "ttft"  # time to first token
     LATENCY = "latency"  # requests latency
     INFLIGHT = "inflight"  # requests concurrency
     PERFORMANCE = "performance"  # custom performance metric
 
 
 class ModelCosts(BaseModel):
-    prompt_tokens: float = Field(default=0.0, ge=0.0, description="Cost of a million prompt tokens (decrease user budget)")
-    completion_tokens: float = Field(default=0.0, ge=0.0, description="Cost of a million completion tokens (decrease user budget)")
+    prompt_tokens: float = 0.0
+    completion_tokens: float = 0.0
 
 
-class ModelType(str, Enum):
+class ModelType(StrEnum):
     AUTOMATIC_SPEECH_RECOGNITION = "automatic-speech-recognition"
     IMAGE_TEXT_TO_TEXT = "image-text-to-text"
     IMAGE_TO_TEXT = "image-to-text"
@@ -25,10 +25,10 @@ class ModelType(str, Enum):
 
 
 class Model(BaseModel):
-    id: str = Field(..., description="The model identifier, which can be referenced in the API endpoints.")
-    type: ModelType = Field(..., description="The type of the model, which can be used to identify the model type.", examples=["text-generation"])  # fmt: off
-    aliases: list[str] | None = Field(default=None, description="Aliases of the model. It will be used to identify the model by users.", examples=[["model-alias", "model-alias-2"]])  # fmt: off
-    created: int = Field(..., description="Time of creation, as Unix timestamp.")
-    owned_by: str = Field(..., description="The organization that owns the model.")
-    max_context_length: int | None = Field(default=None, description="Maximum amount of tokens a context could contains. Makes sure it is the same for all models.")  # fmt: off
-    costs: ModelCosts = Field(..., description="Costs of the model.")
+    id: str
+    type: ModelType
+    aliases: list[str] = []
+    created: int
+    owned_by: str
+    max_context_length: int | None = None
+    costs: ModelCosts

api/domain/provider/_providerrepository.py

@@ -28,17 +28,11 @@ async def create_provider(
         pass
 
     @abstractmethod
-    async def delete_provider(
-        self,
-        provider_id: int,
-    ) -> Provider | None:
+    async def delete_provider(self, provider_id: int) -> Provider | None:
         pass
 
     @abstractmethod
-    async def get_one_provider(
-        self,
-        provider_id: int,
-    ) -> Provider | None:
+    async def get_one_provider(self, provider_id: int) -> Provider | None:
         pass
 
     @abstractmethod

api/domain/provider/entities.py

@@ -1,8 +1,6 @@
-from enum import Enum, StrEnum
-from typing import Literal
+from enum import StrEnum
 
 import pycountry
-from pydantic import Field, constr
 
 from api.domain import EntitiesPage
 from api.domain.model.entities import ModelType
@@ -12,8 +10,7 @@
 
 # Add world as a country code, default value of the carbon footprint computation framework
 country_codes = [country.alpha_3 for country in pycountry.countries] + ["WOR"]
-country_codes_dict = {str(code).upper(): str(code) for code in sorted(set(country_codes))}
-ProviderCarbonFootprintZone: type[Enum] = Enum("ProviderCarbonFootprintZone", country_codes_dict, type=str)
+ProviderCarbonFootprintZone = StrEnum("ProviderCarbonFootprintZone", {str(code).upper(): str(code) for code in sorted(set(country_codes))})
 
 
 class ProviderType(StrEnum):
@@ -72,22 +69,21 @@ class ProviderSortField(StrEnum):
 
 
 class Provider(BaseModel):
-    object: Literal["provider"] = "provider"
-    id: int = Field(..., description="Provider ID.")  # fmt: off
-    router_id: int = Field(..., description="ID of the router that owns the provider.")  # fmt: off
-    user_id: int = Field(..., description="ID of the user that owns the provider.")  # fmt: off
-    type: ProviderType = Field(..., description="Provider type.")  # fmt: off
-    url: constr(strip_whitespace=True, min_length=1, to_lower=True) | None = Field(default=None, description="Provider API url. The url must only contain the domain name (without `/v1` suffix for example).")  # fmt: off
-    key: str | None = Field(description="Provider API key.")  # fmt: off
-    timeout: int = Field(..., description="Timeout for the provider requests, after user receive an 500 error (model is too busy).")  # fmt: off
-    model_name: str = Field(..., description="Model name from the model provider.")  # fmt: off
-    model_hosting_zone: ProviderCarbonFootprintZone = Field(default=ProviderCarbonFootprintZone.WOR, description="Model hosting zone using ISO 3166-1 alpha-3 code format (e.g., `WOR` for World, `FRA` for France, `USA` for United States). This determines the electricity mix used for carbon intensity calculations. For more information, see https://ecologits.ai", examples=["WOR"])  # fmt: off
-    model_total_params: int = Field(default=0, ge=0, description="Total params of the model in billions of parameters for carbon footprint computation. If not provided, the active params will be used if provided, else carbon footprint will not be computed. For more information, see https://ecologits.ai")  # fmt: off
-    model_active_params: int = Field(default=0, ge=0, description="Active params of the model in billions of parameters for carbon footprint computation. If not provided, the total params will be used if provided, else carbon footprint will not be computed. For more information, see https://ecologits.ai")  # fmt: off
-    qos_metric: Metric | None = Field(description="The metric to use for the QoS policy. If not provided, no QoS policy is applied.")  # fmt: off
-    qos_limit: float | None = Field(default=None, ge=0.0, description="The value to use for the quality of service. Depends of the metric, the value can be a percentile, a threshold, etc.")  # fmt: off
-    created: int | None = Field(default=None, description="Time of creation, as Unix timestamp.")  # fmt: off
-    updated: int | None = Field(default=None, description="Time of last update, as Unix timestamp.")  # fmt: off
+    id: int
+    router_id: int
+    user_id: int
+    type: ProviderType
+    url: str
+    key: str | None = None
+    timeout: int
+    model_name: str
+    model_hosting_zone: ProviderCarbonFootprintZone = ProviderCarbonFootprintZone.WOR
+    model_total_params: int = 0
+    model_active_params: int = 0
+    qos_metric: Metric | None = None
+    qos_limit: float | None = None
+    created: int
+    updated: int
 
     def with_router_id(self, router_id: int) -> "Provider":
         return self.model_copy(update={"router_id": router_id})

api/domain/router/entities.py

@@ -1,6 +1,6 @@
 from enum import StrEnum
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel
 
 from api.domain import EntitiesPage
 from api.domain.model import ModelType as RouterType
@@ -15,19 +15,19 @@ class RouterLoadBalancingStrategy(StrEnum):
 
 
 class Router(BaseModel):
-    id: int = Field(..., description="ID of the router.")  # fmt: off
-    name: str = Field(..., description="Name of the router.")  # fmt: off
-    user_id: int = Field(..., description="ID of the user that owns the router.")  # fmt: off
-    type: RouterType = Field(..., description="Type of the model router. It will be used to identify the model router type.", examples=["text-generation"])  # fmt: off
-    aliases: list[str] | None = Field(default=None, description="Aliases of the model. It will be used to identify the model by users.", examples=[["model-alias", "model-alias-2"]])  # fmt: off
-    load_balancing_strategy: RouterLoadBalancingStrategy = Field(..., description="Routing strategy for load balancing between providers of the model. It will be used to identify the model type.", examples=["least_busy"])  # fmt: off
-    vector_size: int | None = Field(default=None, description="Dimension of the vectors, if the models are embeddings. Make sure it is the same for all models.")  # fmt: off
-    max_context_length: int | None = Field(default=None, description="Maximum amount of tokens a context could contains. Make sure it is the same for all models.")  # fmt: off
-    cost_prompt_tokens: float = Field(description="Cost of a million prompt tokens (decrease user budget)")
-    cost_completion_tokens: float = Field(description="Cost of a million completion tokens (decrease user budget)")
-    providers: int = Field(default=0, description="Number of providers in the router.")  # fmt: off
-    created: int = Field(..., description="Time of creation, as Unix timestamp.")  # fmt: off
-    updated: int = Field(..., description="Time of last update, as Unix timestamp.")  # fmt: off
+    id: int
+    name: str
+    user_id: int
+    type: RouterType
+    aliases: list[str] | None
+    load_balancing_strategy: RouterLoadBalancingStrategy
+    vector_size: int | None
+    max_context_length: int | None
+    cost_prompt_tokens: float
+    cost_completion_tokens: float
+    providers: int
+    created: int
+    updated: int
 
     def with_name(self, name: str) -> "Router":
         return self.model_copy(update={"name": name})