Celes Renata
a72f336ad1
Multi-stage evidence-grounded inference architecture replacing the monolithic 9B model extraction pipeline. CPU-first specialist services handle routine extraction while the 9B vLLM model is preserved for semantic adjudication of ambiguous cases. Key components: - Capability-aware inference gateway (OpenAI-compatible + Ollama) - Endpoint registry with DB migrations and REST API - Sentence-aware document segmenter (property tests) - Deterministic financial parsing with offset integrity - Symbol resolution with ambiguity detection - Specialist service (GLiNER2, dynamic batching, K8s deployment) - Company-specific sentiment (FinBERT, calibration) - Retrieval-based novelty and duplicate detection - Confidence calibration pipeline - Deterministic routing engine (property tests) - 9B adjudication layer with VRAM gating - Stock-specific impact model (features, labels, baseline, trained) - Pipeline orchestrator (state machine, queues, leases, feature flags) - Bounded parallelism (async workers, semaphore, load shedding) - Observability (tracing, metrics, alerts) - Compatibility adapter (v3→v2 golden mapping tests) - Shadow/canary promotion framework - Active learning and fine-tuning pipeline Test results: 1,161 tests pass, ruff lint clean. All 282 spec tasks completed.
252 lines
7.7 KiB
Python
252 lines
7.7 KiB
Python
"""Pydantic request/response models for the inference registry API.
|
|
|
|
All response models EXCLUDE actual auth_secret_ref values.
|
|
Instead they show a status string: "configured" or "not_configured".
|
|
|
|
Requirements: 3.6, 3.7
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
from datetime import datetime
|
|
from typing import Any, Literal
|
|
from uuid import UUID
|
|
|
|
from pydantic import BaseModel, Field, field_validator
|
|
|
|
VALID_PROTOCOLS = ("ollama_native", "openai_chat", "specialist_http")
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Endpoint schemas
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
class EndpointCreate(BaseModel):
|
|
"""Request body for creating an inference endpoint."""
|
|
|
|
name: str = Field(..., min_length=1, max_length=255)
|
|
protocol: Literal["ollama_native", "openai_chat", "specialist_http"]
|
|
base_url: str = Field(..., min_length=1)
|
|
auth_secret_ref: str | None = None
|
|
auth_scheme: str = "bearer"
|
|
default_headers: dict[str, str] = Field(default_factory=dict)
|
|
health_path: str | None = None
|
|
enabled: bool = True
|
|
|
|
@field_validator("base_url")
|
|
@classmethod
|
|
def validate_url(cls, v: str) -> str:
|
|
"""Validate that base_url looks like a valid URL."""
|
|
if not v.startswith(("http://", "https://")):
|
|
raise ValueError("base_url must start with http:// or https://")
|
|
return v.rstrip("/")
|
|
|
|
@field_validator("protocol")
|
|
@classmethod
|
|
def validate_protocol(cls, v: str) -> str:
|
|
if v not in VALID_PROTOCOLS:
|
|
raise ValueError(f"protocol must be one of {VALID_PROTOCOLS}")
|
|
return v
|
|
|
|
|
|
class EndpointUpdate(BaseModel):
|
|
"""Request body for updating an inference endpoint."""
|
|
|
|
name: str | None = None
|
|
protocol: Literal["ollama_native", "openai_chat", "specialist_http"] | None = None
|
|
base_url: str | None = None
|
|
auth_secret_ref: str | None = Field(default=None)
|
|
auth_scheme: str | None = None
|
|
default_headers: dict[str, str] | None = None
|
|
health_path: str | None = None
|
|
enabled: bool | None = None
|
|
|
|
@field_validator("base_url")
|
|
@classmethod
|
|
def validate_url(cls, v: str | None) -> str | None:
|
|
if v is not None:
|
|
if not v.startswith(("http://", "https://")):
|
|
raise ValueError("base_url must start with http:// or https://")
|
|
return v.rstrip("/")
|
|
return v
|
|
|
|
@field_validator("protocol")
|
|
@classmethod
|
|
def validate_protocol(cls, v: str | None) -> str | None:
|
|
if v is not None and v not in VALID_PROTOCOLS:
|
|
raise ValueError(f"protocol must be one of {VALID_PROTOCOLS}")
|
|
return v
|
|
|
|
|
|
class EndpointResponse(BaseModel):
|
|
"""Response model for an inference endpoint. NEVER includes auth_secret_ref value."""
|
|
|
|
id: UUID
|
|
name: str
|
|
protocol: str
|
|
base_url: str
|
|
auth_secret_status: str = "not_configured" # "configured" or "not_configured"
|
|
auth_scheme: str = "bearer"
|
|
default_headers: dict[str, str] = Field(default_factory=dict)
|
|
health_path: str | None = None
|
|
enabled: bool = True
|
|
revision: int = 1
|
|
created_at: datetime | None = None
|
|
updated_at: datetime | None = None
|
|
last_probe: ProbeResponse | None = None
|
|
capabilities: dict[str, Any] | None = None
|
|
active_bindings: list[BindingResponse] | None = None
|
|
|
|
|
|
class EndpointListResponse(BaseModel):
|
|
"""Response model for listing endpoints."""
|
|
|
|
id: UUID
|
|
name: str
|
|
protocol: str
|
|
base_url: str
|
|
auth_secret_status: str = "not_configured"
|
|
enabled: bool = True
|
|
revision: int = 1
|
|
created_at: datetime | None = None
|
|
updated_at: datetime | None = None
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Deployment schemas
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
class DeploymentCreate(BaseModel):
|
|
"""Request body for creating a model deployment."""
|
|
|
|
endpoint_id: UUID
|
|
served_model_name: str = Field(..., min_length=1)
|
|
display_name: str = Field(..., min_length=1)
|
|
capabilities: dict[str, Any] = Field(default_factory=dict)
|
|
context_window: int | None = None
|
|
max_output_tokens: int | None = None
|
|
quantization: str | None = None
|
|
runtime_metadata: dict[str, Any] = Field(default_factory=dict)
|
|
enabled: bool = True
|
|
|
|
|
|
class DeploymentResponse(BaseModel):
|
|
"""Response model for a model deployment."""
|
|
|
|
id: UUID
|
|
endpoint_id: UUID
|
|
served_model_name: str
|
|
display_name: str
|
|
capabilities: dict[str, Any] = Field(default_factory=dict)
|
|
context_window: int | None = None
|
|
max_output_tokens: int | None = None
|
|
quantization: str | None = None
|
|
runtime_metadata: dict[str, Any] = Field(default_factory=dict)
|
|
enabled: bool = True
|
|
revision: int = 1
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Binding schemas
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
class BindingCreate(BaseModel):
|
|
"""Request body for creating an agent stage binding."""
|
|
|
|
agent_id: UUID
|
|
stage: str = Field(..., min_length=1)
|
|
model_deployment_id: UUID | None = None
|
|
route_order: int = 0
|
|
routing_config: dict[str, Any] = Field(default_factory=dict)
|
|
is_active: bool = True
|
|
|
|
|
|
class BindingResponse(BaseModel):
|
|
"""Response model for an agent stage binding."""
|
|
|
|
id: UUID
|
|
agent_id: UUID
|
|
stage: str
|
|
model_deployment_id: UUID | None = None
|
|
route_order: int = 0
|
|
routing_config: dict[str, Any] = Field(default_factory=dict)
|
|
is_active: bool = True
|
|
revision: int = 1
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Probe schemas
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
class ProbeResponse(BaseModel):
|
|
"""Response model for probe results."""
|
|
|
|
endpoint_id: UUID
|
|
timestamp: datetime | None = None
|
|
software_version: str | None = None
|
|
probe_duration_ms: int = 0
|
|
health_success: bool = False
|
|
health_detail: str = ""
|
|
model_listing_success: bool | None = None
|
|
json_schema_success: bool | None = None
|
|
usage_success: bool | None = None
|
|
seed_success: bool | None = None
|
|
output_token_field_success: bool | None = None
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Egress confirmation schema
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
class EgressConfirmation(BaseModel):
|
|
"""Request body for confirming external endpoint egress enablement.
|
|
|
|
Requires explicit confirmed=true flag.
|
|
"""
|
|
|
|
confirmed: bool = Field(
|
|
...,
|
|
description="Must be explicitly set to true to confirm external egress enablement",
|
|
)
|
|
|
|
@field_validator("confirmed")
|
|
@classmethod
|
|
def must_be_true(cls, v: bool) -> bool:
|
|
if not v:
|
|
raise ValueError("confirmed must be true to enable external egress")
|
|
return v
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Structured output test schemas
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
class StructuredOutputTestRequest(BaseModel):
|
|
"""Request body for testing structured output on an endpoint."""
|
|
|
|
json_schema: dict[str, Any] = Field(
|
|
default_factory=lambda: {
|
|
"type": "object",
|
|
"properties": {"status": {"type": "string"}},
|
|
"required": ["status"],
|
|
}
|
|
)
|
|
prompt: str = 'Respond with JSON: {"status": "ok"}'
|
|
|
|
|
|
class StructuredOutputTestResponse(BaseModel):
|
|
"""Response for structured output test."""
|
|
|
|
success: bool
|
|
structured_mode: str = ""
|
|
content: str = ""
|
|
parsed: dict[str, Any] | None = None
|
|
schema_valid: bool = False
|
|
latency_ms: int = 0
|
|
error: str | None = None
|