feat: Intelligence Pipeline v3 — full implementation

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.
2026-07-13 02:14:59 +00:00
parent 84634a365e
commit a72f336ad1
227 changed files with 50403 additions and 0 deletions
@@ -0,0 +1,89 @@
+"""Normalized error categories for the inference gateway.
+
+Maps provider-specific failures into a protocol-agnostic taxonomy
+so that retry logic, alerting, and metrics work uniformly across
+Ollama, OpenAI-compatible, and specialist endpoints.
+
+Requirements: 2.1, 2.9
+"""
+from __future__ import annotations
+
+from enum import Enum
+
+
+class InferenceErrorCategory(str, Enum):
+    """Normalized error categories for inference failures."""
+
+    # Network / transport
+    TIMEOUT = "timeout"
+    CONNECTION_REFUSED = "connection_refused"
+    CONNECTION_ERROR = "connection_error"
+
+    # Authentication / authorization
+    AUTH_FAILED = "auth_failed"
+    FORBIDDEN = "forbidden"
+
+    # Rate limiting
+    RATE_LIMITED = "rate_limited"
+
+    # Server errors
+    SERVER_ERROR = "server_error"
+    SERVICE_UNAVAILABLE = "service_unavailable"
+
+    # Client errors
+    BAD_REQUEST = "bad_request"
+    MODEL_NOT_FOUND = "model_not_found"
+    INVALID_REQUEST = "invalid_request"
+
+    # Response problems
+    INVALID_RESPONSE = "invalid_response"
+    EMPTY_RESPONSE = "empty_response"
+    SCHEMA_VIOLATION = "schema_violation"
+
+    # Capability / policy
+    CAPABILITY_UNAVAILABLE = "capability_unavailable"
+    POLICY_VIOLATION = "policy_violation"
+
+    # Ollama-specific
+    STALL_DETECTED = "stall_detected"
+
+    # Unknown
+    UNKNOWN = "unknown"
+
+    @property
+    def retryable(self) -> bool:
+        """Whether this error category should generally be retried."""
+        return self in _RETRYABLE_CATEGORIES
+
+
+_RETRYABLE_CATEGORIES = frozenset({
+    InferenceErrorCategory.TIMEOUT,
+    InferenceErrorCategory.CONNECTION_ERROR,
+    InferenceErrorCategory.CONNECTION_REFUSED,
+    InferenceErrorCategory.SERVER_ERROR,
+    InferenceErrorCategory.SERVICE_UNAVAILABLE,
+    InferenceErrorCategory.RATE_LIMITED,
+    InferenceErrorCategory.STALL_DETECTED,
+    InferenceErrorCategory.EMPTY_RESPONSE,
+})
+
+
+class InferenceError(Exception):
+    """Typed inference error with category and optional provider detail."""
+
+    def __init__(
+        self,
+        category: InferenceErrorCategory,
+        message: str = "",
+        *,
+        provider_detail: str | None = None,
+        status_code: int | None = None,
+    ) -> None:
+        self.category = category
+        self.provider_detail = provider_detail
+        self.status_code = status_code
+        super().__init__(message or category.value)
+
+    @property
+    def retryable(self) -> bool:
+        return self.category.retryable