AgentEvalTool/backend/agenteval/evaluation/engine.py
sinohqb c7f1dca49d v0.3-s2: 3 个新规则 + 组合逻辑 + 33 个测试
## 新规则(共 6 种,增加 3 种)

### semantic_similarity
- 调用 OpenAI 兼容 embedding API(asyncio.gather 并发两路请求)
- 余弦相似度与 reference 比对,min_score 可配置(默认 0.7)
- API 异常时明确返回失败原因,不隐藏错误

### json_schema
- 验证回复是否为合法 JSON(支持 markdown 代码块剥离)
- required_keys / forbidden_keys / key_types 三维校验
- dot-path 支持嵌套字段("data.id")
- strict_json=false 模式非阻断校验

### safety
- 双层检测:关键词黑名单(零延迟)+ 可选 moderation API
- API 不可用时自动降级黑名单,不中止评测
- 支持自定义 flagged_categories

## 规则组合逻辑(rule_logic + rule_pass_threshold)

- models.py: EvalRuleConfig 增加 weight 字段;Case 增加 rule_logic / rule_pass_threshold
- models.py: 新增 RuleLogic 枚举(all / any / weighted)
- engine._save_rule_results: 按 rule_logic 决定 case 通过/失败
  - ALL:全部通过才通过(原有行为,向下兼容)
  - ANY:至少一条通过即通过
  - WEIGHTED:加权平均分 >= rule_pass_threshold

## 测试(43 → 76,新增 33)
- test_s2_rules_and_logic.py:3 个新规则的 pass/fail/边界/API 降级 + 5 个组合逻辑集成测试

Co-Authored-By: Claude <noreply@anthropic.com>
2026-07-17 11:23:22 +08:00

571 lines
20 KiB
Python

"""Evaluation execution engine.
Async-first implementation: channels and LLM calls are awaited cooperatively,
so multiple cases can run concurrently and a run can be cancelled mid-flight
via an ``asyncio.Event`` cancel token.
"""
import asyncio
import uuid
from dataclasses import dataclass
from typing import Any, Callable, Optional
import httpx
from agenteval.channels.base import EvalChannel
from agenteval.channels.factory import ChannelFactory
from agenteval.evaluation.rules import get_rule
from agenteval.models import (
Case,
CaseType,
EvalResult,
EvalRun,
EvalTarget,
RuleLogic,
RunStatus,
Scenario,
Turn,
)
from agenteval.storage.db import get_session, utc_now
from agenteval.storage.repository import ResultRepository, RunRepository
from agenteval.utils.llm import extract_content_from_llm_response, extract_reply_text, parse_json_from_llm_text
# Progress callbacks may be sync or async; the engine awaits the result if
# it is a coroutine, otherwise treats it as a plain function.
ProgressCallback = Callable[[str, dict[str, Any]], Any]
class CancelledError(RuntimeError):
"""Raised inside the engine when the cancel token fires."""
@dataclass
class TimeoutConfig:
"""Per-operation timeouts (all in seconds)."""
poll_reply: float = 30.0
llm_generate: float = 60.0
def _build_send_message(content: str) -> dict[str, Any]:
return {
"msgType": "text",
"msgBody": {"content": content},
}
class EvalEngine:
"""Execute evaluation scenarios against targets.
The engine is async so the (slow, network-bound) channel and LLM calls
can be awaited cooperatively. Database writes remain synchronous for now
(SQLite + StaticPool); they are fast enough not to block the event loop
in practice.
"""
def __init__(
self,
target: EvalTarget,
scenario: Scenario,
session=None,
run_repo: Optional[RunRepository] = None,
result_repo: Optional[ResultRepository] = None,
cancel_token: Optional[asyncio.Event] = None,
timeout_config: Optional[TimeoutConfig] = None,
max_concurrent_cases: int = 1,
):
self.target = target
self.scenario = scenario
self.channel: EvalChannel = ChannelFactory.create(target)
self.session = session or get_session()
self.run_repo = run_repo or RunRepository(self.session)
self.result_repo = result_repo or ResultRepository(self.session)
self.cancel_token = cancel_token or asyncio.Event()
self.timeout_config = timeout_config or TimeoutConfig()
self._case_semaphore = asyncio.Semaphore(max(1, max_concurrent_cases))
# ── public entry point ────────────────────────────────────────────
async def run(
self,
progress_callback: Optional[ProgressCallback] = None,
existing_run: Optional[EvalRun] = None,
) -> EvalRun:
"""Run the evaluation and return the completed run record.
Cancellation is cooperative: set ``cancel_token`` and the engine will
mark the run as FAILED with ``cancelled_by_user`` at the next checkpoint.
"""
if existing_run:
run = existing_run
run.status = RunStatus.RUNNING
run.started_at = utc_now()
run = self.run_repo.update(run) or run
else:
run = EvalRun(
id=str(uuid.uuid4()),
target_id=self.target.id or "",
scenario_id=self.scenario.id or "",
status=RunStatus.RUNNING,
started_at=utc_now(),
)
run = self.run_repo.create(run)
try:
total_cases = len(self.scenario.cases)
passed_cases = 0
failed_cases = 0
for idx, case in enumerate(self.scenario.cases, start=1):
self._check_cancel()
await self._emit(
progress_callback,
"case_start",
{
"index": idx,
"total": total_cases,
"case_id": case.id,
},
)
async with self._case_semaphore:
case_passed, rule_pass, rule_total = await self._run_case(
run,
case,
progress_callback,
)
if case_passed:
passed_cases += 1
else:
failed_cases += 1
await self._emit(
progress_callback,
"case_end",
{
"index": idx,
"total": total_cases,
"case_id": case.id,
"passed": case_passed,
"rule_pass_count": rule_pass,
"rule_total": rule_total,
},
)
results = self.run_repo.get_results(run.id)
total_rules = len(results)
passed_rules = sum(1 for r in results if r.passed)
summary = {
"total_cases": total_cases,
"passed_cases": passed_cases,
"failed_cases": failed_cases,
"total_rules": total_rules,
"passed_rules": passed_rules,
"pass_rate": round(passed_rules / total_rules, 4) if total_rules else 0.0,
}
run.status = RunStatus.COMPLETED
run.completed_at = utc_now()
run.summary = summary
await self._emit(
progress_callback,
"run_completed",
{
"status": "completed",
"summary": summary,
},
)
except CancelledError:
run.status = RunStatus.FAILED
run.completed_at = utc_now()
run.summary = {
"error": {"code": "cancelled_by_user", "message": "评测已手动停止"},
}
await self._emit(
progress_callback,
"run_completed",
{
"status": "failed",
"reason": "cancelled",
"error": {"code": "cancelled_by_user", "message": "评测已手动停止"},
},
)
except Exception as exc:
run.status = RunStatus.FAILED
run.completed_at = utc_now()
run.summary = {"error": str(exc)}
await self._emit(progress_callback, "error", {"error": str(exc)})
await self._emit(
progress_callback,
"run_completed",
{
"status": "failed",
"error": str(exc),
},
)
raise
finally:
run = self.run_repo.update(run) or run
# Best-effort cleanup of the channel's HTTP client.
close = getattr(self.channel, "close", None)
if callable(close):
try:
result = close()
if asyncio.iscoroutine(result):
await result
except Exception:
pass
try:
self.session.close()
except Exception:
pass
return run
# ── case / turn execution ─────────────────────────────────────────
async def _run_case(
self,
run: EvalRun,
case: Case,
progress_callback: Optional[ProgressCallback],
) -> tuple[bool, int, int]:
"""Run a single case; returns (all_rules_passed, passed_rules, total_rules)."""
if case.type == CaseType.DYNAMIC:
generated = await self._generate_messages(case, progress_callback)
if not generated:
await self._emit(
progress_callback,
"error",
{
"error": "LLM 未能生成测试消息",
"case_id": case.id,
},
)
return False, 0, 0
case = case.model_copy(update={"messages": generated})
dialog: list[Turn] = []
for round_index, message in enumerate(case.messages, start=1):
self._check_cancel()
await self._emit(
progress_callback,
"turn_start",
{
"run_id": run.id,
"case_id": case.id,
"round": round_index,
"message": message,
},
)
sent_at = utc_now()
send_result = await self.channel.send(message)
if not send_result.ok:
turn = Turn(
id=str(uuid.uuid4()),
run_id=run.id,
case_id=case.id,
round_index=round_index,
sent_message=_build_send_message(message),
sent_at=sent_at,
)
self.result_repo.save_turn(turn)
await self._save_rule_results(run, case, turn, [], progress_callback)
await self._emit(
progress_callback,
"turn_error",
{
"case_id": case.id,
"round": round_index,
"error": send_result.error,
},
)
return False, 0, 0
try:
reply = await self.channel.poll_reply(
send_result.question_msg_id or "",
timeout=self.timeout_config.poll_reply,
)
except Exception as poll_exc:
received_at = utc_now()
turn = Turn(
id=str(uuid.uuid4()),
run_id=run.id,
case_id=case.id,
round_index=round_index,
sent_message=_build_send_message(message),
sent_at=sent_at,
question_msg_id=send_result.question_msg_id,
received_at=received_at,
)
self.result_repo.save_turn(turn)
await self._emit(
progress_callback,
"turn_error",
{
"case_id": case.id,
"round": round_index,
"error": f"poll_reply 异常: {poll_exc}",
},
)
return False, 0, 0
received_at = utc_now()
latency_ms = None
if sent_at and received_at:
latency_ms = int((received_at - sent_at).total_seconds() * 1000)
turn = Turn(
id=str(uuid.uuid4()),
run_id=run.id,
case_id=case.id,
round_index=round_index,
sent_message=_build_send_message(message),
sent_at=sent_at,
question_msg_id=send_result.question_msg_id,
reply=reply.raw_message if reply else None,
received_at=received_at,
latency_ms=latency_ms,
)
self.result_repo.save_turn(turn)
dialog.append(turn)
await self._emit(
progress_callback,
"turn_end",
{
"run_id": run.id,
"case_id": case.id,
"round": round_index,
"latency_ms": latency_ms,
"reply_text": extract_reply_text(reply.raw_message if reply else None),
},
)
if not dialog:
return False, 0, 0
return await self._save_rule_results(run, case, dialog[-1], dialog, progress_callback)
async def _save_rule_results(
self,
run: EvalRun,
case: Case,
turn: Turn,
dialog: list[Turn],
progress_callback: Optional[ProgressCallback],
) -> tuple[bool, int, int]:
"""Apply rules and save results; returns (case_passed, passed_count, total_count).
Combination logic (case.rule_logic):
ALL — all rules must pass (default)
ANY — at least one rule must pass
WEIGHTED — weighted average score >= case.rule_pass_threshold
"""
from agenteval.models import EvalRuleConfig
rules_config: list[EvalRuleConfig] = list(case.eval_rules)
# If no explicit rules, derive implicit rules from expectations.
if not rules_config:
if case.expectations.response_time_max_ms:
rules_config.append(
EvalRuleConfig(
type="response_time",
params={
"max_ms": case.expectations.response_time_max_ms,
},
)
)
if case.expectations.keywords_include or case.expectations.keywords_exclude:
rules_config.append(
EvalRuleConfig(
type="keyword_match",
params={
"keywords": case.expectations.keywords_include,
"exclude_keywords": case.expectations.keywords_exclude,
},
)
)
if not rules_config:
# No rules defined and no expectations → case passes with no checks
return True, 0, 0
passed_count = 0
total_count = 0
weighted_score = 0.0
total_weight = 0.0
for rule_config in rules_config:
rule = get_rule(rule_config.type, rule_config.params)
result = await rule.evaluate(case, dialog)
eval_result = EvalResult(
id=str(uuid.uuid4()),
run_id=run.id,
case_id=case.id,
turn_id=turn.id or "",
rule_type=rule_config.type,
passed=result.passed,
score=result.score,
reason=result.reason,
)
self.result_repo.save_result(eval_result)
total_count += 1
if result.passed:
passed_count += 1
# Weighted scoring: use rule score (default 1.0 if passed, 0.0 if failed)
score_val = result.score if result.score is not None else (1.0 if result.passed else 0.0)
weight = rule_config.weight
weighted_score += score_val * weight
total_weight += weight
await self._emit(
progress_callback,
"rule_result",
{
"run_id": run.id,
"case_id": case.id,
"rule_type": rule_config.type,
"passed": result.passed,
"score": result.score,
"reason": result.reason,
"weight": weight,
},
)
# Determine case pass/fail based on rule_logic
logic = case.rule_logic
if logic == RuleLogic.ALL:
case_passed = passed_count == total_count
elif logic == RuleLogic.ANY:
case_passed = passed_count > 0
elif logic == RuleLogic.WEIGHTED:
avg = weighted_score / total_weight if total_weight > 0 else 0.0
case_passed = avg >= case.rule_pass_threshold
else:
case_passed = passed_count == total_count
return case_passed, passed_count, total_count
async def _generate_messages(
self,
case: Case,
progress_callback: Optional[ProgressCallback],
) -> list[str]:
"""Use LLM to generate test messages for dynamic cases."""
llm_config = self.scenario.llm_config
if not llm_config:
await self._emit(
progress_callback,
"error",
{
"error": "动态用例需要配置 llm_config",
},
)
return []
api_url = llm_config.get("api_url")
api_key = llm_config.get("api_key")
model = llm_config.get("model", "doubao-seed-2.0-lite")
if not api_url:
await self._emit(progress_callback, "error", {"error": "llm_config 缺少 api_url"})
return []
turns = case.turns or 3
prompt = case.prompt or "请生成一些测试问题"
system_prompt = (
f"你需要扮演一个真实的用户/患者,根据以下要求生成 {turns} 条独立的测试问题。\n\n"
f"要求:{prompt}\n\n"
"输出格式要求:只输出一个 JSON 数组,包含 " + str(turns) + " 个字符串,每个字符串是一条消息。"
"不要输出任何解释、markdown 或其他内容。"
)
headers = {"Content-Type": "application/json"}
if api_key:
headers["Authorization"] = f"Bearer {api_key}"
payload = {
"model": model,
"messages": [
{"role": "system", "content": system_prompt},
{"role": "user", "content": f"请生成 {turns} 条测试消息"},
],
"temperature": 0.7,
}
try:
async with httpx.AsyncClient(timeout=self.timeout_config.llm_generate) as client:
resp = await client.post(api_url, headers=headers, json=payload)
resp.raise_for_status()
content = extract_content_from_llm_response(resp.json())
if not content:
await self._emit(
progress_callback,
"error",
{
"error": "LLM 返回内容为空或无法解析",
},
)
return []
try:
parsed = parse_json_from_llm_text(content)
except (ValueError, Exception) as parse_exc:
await self._emit(
progress_callback,
"error",
{
"error": f"LLM 返回无法解析为数组: {parse_exc}",
},
)
return []
if not isinstance(parsed, list):
await self._emit(progress_callback, "error", {"error": "LLM 返回的不是数组"})
return []
messages = [str(m) for m in parsed if isinstance(m, str) and m.strip()]
if not messages:
await self._emit(progress_callback, "error", {"error": "LLM 返回的消息为空"})
return []
await self._emit(
progress_callback,
"messages_generated",
{
"case_id": case.id,
"messages": messages,
},
)
return messages
except Exception as exc:
await self._emit(progress_callback, "error", {"error": f"LLM 生成消息失败: {exc}"})
return []
# ── helpers ────────────────────────────────────────────────────────
def _check_cancel(self) -> None:
if self.cancel_token.is_set():
raise CancelledError("run cancelled")
async def _emit(
self,
callback: Optional[ProgressCallback],
event: str,
data: dict[str, Any],
) -> None:
if not callback:
return
try:
result = callback(event, data)
if asyncio.iscoroutine(result):
await result
except Exception:
pass