merge: chore/cleanup-remove-bloat-and-secrets into main

app/api/v1/rag/search.py

@@ -0,0 +1,81 @@
+"""V1 RAG route — thin HTTP layer over app.rag.
+
+The RAG system is the most coupled module (14 legacy files). This
+facade exposes the most-used operations: search, ingest, feedback.
+"""
+from __future__ import annotations
+
+from typing import Annotated, Any
+
+from fastapi import APIRouter, Depends, HTTPException
+from pydantic import BaseModel, Field
+
+from app.rag import (
+    FeedbackRecord,
+    IngestRequest,
+    IngestResult,
+    RAGService,
+    SearchRequest,
+    SearchResponse,
+)
+from app.rag.engine import bulk_ingest as engine_bulk_ingest
+from app.rag.engine import get_stats as engine_get_stats
+
+router = APIRouter(prefix="/api/v1/rag/v2", tags=["rag"])
+
+
+def _service() -> RAGService:
+    return RAGService()
+
+
+class BulkIngestRequest(BaseModel):
+    collection: str = "scam_intel"
+    items: list[dict[str, Any]] = Field(default_factory=list)
+
+
+@router.post("/search", response_model=SearchResponse)
+async def search(
+    req: SearchRequest,
+    svc: Annotated[RAGService, Depends(_service)],
+) -> SearchResponse:
+    """RAG search. Returns Pydantic response with hits + scores."""
+    return await svc.search(req)
+
+
+@router.post("/ingest", response_model=IngestResult)
+async def ingest(
+    req: IngestRequest,
+    svc: Annotated[RAGService, Depends(_service)],
+) -> IngestResult:
+    """Ingest a document into the RAG system."""
+    return await svc.ingest(req)
+
+
+@router.post("/feedback", response_model=IngestResult)
+async def feedback(
+    record: FeedbackRecord,
+    svc: Annotated[RAGService, Depends(_service)],
+) -> IngestResult:
+    """Record scanner → RAG feedback. Ingests known scam into known_scams collection."""
+    ok = await svc.record_feedback(record)
+    return IngestResult(
+        doc_id=record.token_address,
+        collection="known_scams",
+        status="ok" if ok else "failed",
+    )
+
+
+@router.get("/stats")
+async def stats() -> dict:
+    """Per-collection vector counts + active embedder backend."""
+    return engine_get_stats()
+
+
+@router.post("/bulk-ingest")
+async def bulk(req: BulkIngestRequest) -> dict:
+    """Ingest many items into a collection sequentially (max 500 per call)."""
+    if not req.items:
+        raise HTTPException(status_code=400, detail="items must be non-empty")
+    if len(req.items) > 500:
+        raise HTTPException(status_code=400, detail="bulk limit 500 per call")
+    return await engine_bulk_ingest(items=req.items, collection=req.collection)