File size: 6,307 Bytes
b2b6341 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 | """
backend/api/evaluation.py
REST endpoints for RAGAS Evaluation
POST /eval/start β kicks off evaluation as a background job
GET /eval/status/{job_id} β polls job progress & results
GET /eval/results/{job_id} β returns final results JSON
"""
from __future__ import annotations
import threading
import time
import uuid
from typing import Any
from fastapi import APIRouter, BackgroundTasks, HTTPException
from pydantic import BaseModel
router = APIRouter()
# ββ In-memory job store (survives for the backend session) βββββββββββββββββββ
# Maps job_id -> {"status": "running"|"done"|"error", "progress": {...}, "result": {...}}
_jobs: dict[str, dict[str, Any]] = {}
class EvalStartRequest(BaseModel):
n_questions: int = 15 # How many test questions to evaluate (5-30 recommended)
agentic_mode: bool = False # Whether to use the deep research pipeline
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# Background worker
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def _run_eval_job(job_id: str, n_questions: int, use_agentic: bool = False):
"""Runs in a background thread. Updates _jobs[job_id] progressively."""
_jobs[job_id]["status"] = "running"
_jobs[job_id]["progress"] = {"current": 0, "total": n_questions, "message": "Starting..."}
try:
# Import here to avoid circular imports at module load time
from backend.rag.evaluator import run_evaluation
from backend.rag.multi_retriever import MultiSourceResult
from backend.rag.generator import generate_answer, _build_citations
from backend.api.query import _safe_classify, _do_retrieve
from backend.rag.agent_workflow import run_agentic_workflow
def retriever_fn(question: str) -> MultiSourceResult:
analysis = _safe_classify(question)
# Normal retrieval function
def base_retriever(q, sids):
return _do_retrieve(q, sids, analysis)
if use_agentic:
# Use the agentic research workflow
multi_result, _ = run_agentic_workflow(
question=question,
retriever_fn=base_retriever,
source_ids=None,
is_legal=False # Default to false for generic eval
)
return multi_result
else:
# Direct retrieval
return base_retriever(question, None)
def generator_fn(question: str, multi_result: MultiSourceResult) -> str:
try:
result = generate_answer(question, multi_result, history=None)
return result.answer
except Exception as e:
return f"[Generation error: {e}]"
def progress_cb(current: int, total: int, msg: str):
mode_prefix = "[DEEP] " if use_agentic else "[NORMAL] "
_jobs[job_id]["progress"] = {
"current": current,
"total": total,
"message": mode_prefix + msg,
}
result = run_evaluation(
retriever_fn=retriever_fn,
generator_fn=generator_fn,
n_questions=n_questions,
progress_cb=progress_cb,
)
_jobs[job_id]["status"] = "done"
_jobs[job_id]["result"] = result
_jobs[job_id]["progress"]["message"] = "Evaluation complete!"
except Exception as e:
print(f"[EvalAPI] Job {job_id[:8]} failed: {e}")
_jobs[job_id]["status"] = "error"
_jobs[job_id]["error"] = str(e)
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# Endpoints
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
@router.post("/start")
def start_evaluation(req: EvalStartRequest):
"""
POST /eval/start
Kicks off a RAGAS evaluation run in the background.
Returns a job_id to poll with GET /eval/status/{job_id}
"""
# Limit concurrent jobs
running = [j for j in _jobs.values() if j.get("status") == "running"]
if running:
raise HTTPException(
status_code=409,
detail="An evaluation is already running. Wait for it to finish.",
)
n = max(5, min(req.n_questions, 25)) # clamp 5-25 for Groq limits
job_id = str(uuid.uuid4())
_jobs[job_id] = {"status": "queued", "progress": {}, "result": None}
t = threading.Thread(target=_run_eval_job, args=(job_id, n, req.agentic_mode), daemon=True)
t.start()
return {"job_id": job_id, "status": "queued", "n_questions": n}
@router.get("/status/{job_id}")
def get_eval_status(job_id: str):
"""
GET /eval/status/{job_id}
Returns current status, progress, and result (when done).
Frontend polls this every 2 seconds.
"""
if job_id not in _jobs:
raise HTTPException(status_code=404, detail="Job not found")
job = _jobs[job_id]
response = {
"job_id": job_id,
"status": job.get("status"),
"progress": job.get("progress", {}),
}
if job.get("status") == "done":
response["result"] = job.get("result")
if job.get("status") == "error":
response["error"] = job.get("error")
return response
@router.get("/jobs")
def list_jobs():
"""GET /eval/jobs β list all evaluation jobs (for debug)."""
return [
{
"job_id": jid,
"status": j.get("status"),
"n_evaluated": j.get("result", {}).get("n_evaluated") if j.get("result") else None,
}
for jid, j in _jobs.items()
]
|