Spaces:
Sleeping
Sleeping
File size: 29,779 Bytes
8eeadb9 b76da84 8eeadb9 c66550a b76da84 c66550a 8eeadb9 47bec05 8eeadb9 47bec05 c66550a aa4568e c40d5cc 9e85da1 c66550a 4cc218a 8eeadb9 47bec05 8eeadb9 ab6561e 47bec05 4fcb155 1741c30 47bec05 4fcb155 f540a2a 4fcb155 f540a2a 4fcb155 f540a2a 4fcb155 d49cddf 4fcb155 d49cddf 4fcb155 c66550a 4fcb155 7c4588b 4fcb155 c66550a 4fcb155 d49cddf 4fcb155 c66550a 4fcb155 f540a2a 4fcb155 9e85da1 c66550a 4fcb155 c66550a 4fcb155 c66550a 4fcb155 4cc218a 4fcb155 4cc218a b76da84 4fcb155 b76da84 4fcb155 b76da84 4fcb155 4cc218a b76da84 4fcb155 4cc218a 5f83bd3 4fcb155 5f83bd3 4fcb155 5f83bd3 4fcb155 5f83bd3 65a3c09 4fcb155 65a3c09 4fcb155 47bec05 4fcb155 b4be37b 4fcb155 4cc218a 4fcb155 4cc218a 4fcb155 4368e1d 4cc218a 4368e1d 47bec05 4368e1d 4fcb155 4368e1d 4fcb155 4cc218a 4368e1d 4cc218a 4368e1d 2f5a866 4368e1d 2f5a866 4368e1d 47bec05 4368e1d 4fcb155 47bec05 4cc218a 4fcb155 47bec05 4fcb155 2f5a866 4fcb155 4cc218a 4fcb155 4cc218a 4fcb155 4cc218a 4fcb155 2f5a866 4fcb155 4cc218a 4fcb155 d49cddf 4fcb155 4cc218a 4fcb155 9853009 4fcb155 d49cddf 4fcb155 d49cddf 4fcb155 d49cddf 4fcb155 0f8b8aa 4fcb155 4cc218a 4fcb155 4cc218a 4fcb155 d49cddf 4cc218a 4fcb155 d49cddf 4fcb155 d49cddf 4fcb155 d49cddf 4cc218a 4fcb155 4cc218a 5f83bd3 4fcb155 47bec05 4cc218a 4fcb155 4cc218a 4fcb155 4cc218a 4fcb155 8eeadb9 47bec05 4fcb155 9e85da1 8eeadb9 c66550a ee31585 c66550a 4fcb155 c66550a ee31585 4fcb155 ee31585 8eeadb9 4fcb155 b4be37b 4fcb155 8eeadb9 4fcb155 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 |
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
DrugQA (ZH) — 優化版 FastAPI LINE Webhook (最終版)
整合 RAG 邏輯,包含 LLM 意圖偵測、子查詢分解、Intent-aware 檢索與 Rerank。
此版本專注於效能、可維護性、健壯性與使用者體驗。
"""
# ---------- 環境與快取設定 (應置於最前) ----------
import os
import pathlib
os.environ.setdefault("HF_HOME", "/tmp/hf")
os.environ.setdefault("SENTENCE_TRANSFORMERS_HOME", "/tmp/sentence_transformers")
os.environ.setdefault("XDG_CACHE_HOME", "/tmp/.cache")
for d in (os.getenv("HF_HOME"), os.getenv("SENTENCE_TRANSFORMERS_HOME"), os.getenv("XDG_CACHE_HOME")):
pathlib.Path(d).mkdir(parents=True, exist_ok=True)
# ---------- Python 標準函式庫 ----------
import re
import hmac
import base64
import hashlib
import pickle
import logging
import json
import textwrap
import time
import tenacity
from typing import List, Dict, Any, Optional, Tuple, Union
from functools import lru_cache
from dataclasses import dataclass, field
from contextlib import asynccontextmanager
# ---------- 第三方函式庫 ----------
import numpy as np
import pandas as pd
from fastapi import FastAPI, Request, Response, HTTPException, status, BackgroundTasks
import uvicorn
import jieba
from rank_bm25 import BM25Okapi
from sentence_transformers import SentenceTransformer, CrossEncoder
import faiss
import torch
from openai import OpenAI
from tenacity import retry, stop_after_attempt, wait_fixed
import requests
# [MODIFIED] 限制 PyTorch 執行緒數量,避免在 CPU 環境下過度佔用資源
torch.set_num_threads(int(os.getenv("TORCH_NUM_THREADS", "1")))
# ==== CONFIG (從環境變數載入,或使用預設值) ====
# [MODIFIED] 新增環境變數健檢函式
def _require_env(var: str) -> str:
v = os.getenv(var)
if not v:
raise RuntimeError(f"FATAL: Missing required environment variable: {var}")
return v
# [MODIFIED] 檢查 LLM 相關環境變數
def _require_llm_config():
for k in ("LITELLM_BASE_URL", "LITELLM_API_KEY", "LM_MODEL"):
_require_env(k)
CSV_PATH = os.getenv("CSV_PATH", "cleaned_combined.csv")
FAISS_INDEX = os.getenv("FAISS_INDEX", "drug_sentences.index")
SENTENCES_PKL = os.getenv("SENTENCES_PKL", "drug_sentences.pkl")
BM25_PKL = os.getenv("BM25_PKL", "bm25.pkl")
TOP_K_SENTENCES = int(os.getenv("TOP_K_SENTENCES", 15))
PRE_RERANK_K = int(os.getenv("PRE_RERANK_K", 30))
MAX_RERANK_CANDIDATES = int(os.getenv("MAX_RERANK_CANDIDATES", 30))
EMBEDDING_MODEL = os.getenv("EMBEDDING_MODEL", "DMetaSoul/Dmeta-embedding-zh")
RERANKER_MODEL = os.getenv("RERANKER_MODEL", "BAAI/bge-reranker-v2-m3")
LLM_API_CONFIG = {
"base_url": os.getenv("LITELLM_BASE_URL"),
"api_key": os.getenv("LITELLM_API_KEY"),
"model": os.getenv("LM_MODEL")
}
LLM_MODEL_CONFIG = {
"max_context_chars": int(os.getenv("MAX_CONTEXT_CHARS", 10000)),
"max_tokens": int(os.getenv("MAX_TOKENS", 1024)),
"temperature": float(os.getenv("TEMPERATURE", 0.0)),
}
INTENT_CATEGORIES = [
"操作 (Administration)", "保存/攜帶 (Storage & Handling)", "副作用/異常 (Side Effects / Issues)",
"劑型相關 (Dosage Form Concerns)", "時間/併用 (Timing & Interaction)", "劑量調整 (Dosage Adjustment)",
"禁忌症/適應症 (Contraindications/Indications)"
]
DRUG_NAME_MAPPING = {
"fentanyl patch": "fentanyl", "spiriva respimat": "spiriva", "augmentin for syrup": "augmentin syrup",
"nitrostat": "nitroglycerin", "ozempic": "ozempic", "niflec": "niflec",
"fosamax": "fosamax", "humira": "humira", "premarin": "premarin", "smecta": "smecta",
}
DISCLAIMER = "本資訊僅供參考,若您對藥物使用有任何疑問,請務務必諮詢您的醫師或藥師。"
PROMPT_TEMPLATES = {
"analyze_query": """
請分析以下使用者問題,並完成以下兩個任務:
1. 將問題分解為1-3個核心的子問題。
2. 從清單中選擇所有相關的意圖分類。
請嚴格以 JSON 格式回覆,包含 'sub_queries' (字串陣列) 和 'intents' (字串陣列) 兩個鍵。
範例: {{"sub_queries": ["子問題一", "子問題二"], "intents": ["分類名稱一", "分類名稱二"]}}
意圖分類清單:
{options}
使用者問題:{query}
""",
"expand_query": """
請根據以下意圖:{intents},擴展這個查詢,加入相關同義詞或術語。
原始查詢:{query}
請僅輸出擴展後的查詢,不需任何額外的解釋或格式。
""",
"final_answer": """
你是一位專業且謹慎的台灣藥師。請嚴格根據「參考資料」回答使用者問題,使用繁體中文。
規則:
僅能依據參考資料,不得捏造或引用外部知識。
回覆內容需控制在 100 字以內,語氣專業且親切。
以清晰分行或簡短條列方式呈現,方便在 LINE 閱讀。
結尾需加一句提醒(如「如有不適請立即就醫」)。
若資料不足,請直接回覆:「根據提供的資料,無法回答您的問題。」
{additional_instruction}
---
參考資料:
{context}
---
使用者問題:{query}
請直接輸出最終的答案:
"""
}
# ---------- 日誌設定 ----------
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
log = logging.getLogger(__name__)
@dataclass
class FusedCandidate:
idx: int
fused_score: float
sem_score: float
bm_score: float
@dataclass
class RerankResult:
idx: int
rerank_score: float
text: str
meta: Dict[str, Any] = field(default_factory=dict)
# ---------- 核心 RAG 邏輯 ----------
class RagPipeline:
def __init__(self):
# [MODIFIED] 不再傳入 AppConfig,直接引用
if not LLM_API_CONFIG["api_key"] or not LLM_API_CONFIG["base_url"]:
raise ValueError("LLM API Key or Base URL is not configured.")
self.llm_client = OpenAI(api_key=LLM_API_CONFIG["api_key"], base_url=LLM_API_CONFIG["base_url"])
self.embedding_model = self._load_model(SentenceTransformer, EMBEDDING_MODEL, "embedding")
self.reranker = self._load_model(CrossEncoder, RERANKER_MODEL, "reranker")
self.drug_name_to_ids: Dict[str, List[str]] = {}
self.drug_vocab: Dict[str, set] = {"zh": set(), "en": set()}
self.state = type('state', (), {})()
def _load_model(self, model_class, model_name: str, model_type: str):
device = "cuda" if torch.cuda.is_available() else "cpu"
log.info(f"載入 {model_type} 模型:{model_name} 至 {device}...")
try:
return model_class(model_name, device=device)
except Exception as e:
log.warning(f"載入模型至 {device} 失敗: {e}。嘗試切換至 CPU。")
try:
return model_class(model_name, device="cpu")
except Exception as e_cpu:
log.error(f"切換至 CPU 仍無法載入模型: {model_name}。請確認模型路徑或網路連線。錯誤訊息: {e_cpu}")
raise RuntimeError(f"模型載入失敗: {model_name}")
def load_data(self):
log.info("開始載入資料與模型...")
# [MODIFIED] 增加檔案存在性檢查
for path in [CSV_PATH, FAISS_INDEX, SENTENCES_PKL, BM25_PKL]:
if not pathlib.Path(path).exists():
raise FileNotFoundError(f"必要的資料檔案不存在: {path}")
try:
self.df_csv = pd.read_csv(CSV_PATH, dtype=str).fillna('')
# [MODIFIED] 增加必要欄位檢查
for col in ("drug_name_norm", "drug_id"):
if col not in self.df_csv.columns:
raise KeyError(f"CSV 檔案 '{CSV_PATH}' 中缺少必要欄位: {col}")
self.df_csv['drug_name_norm_normalized'] = (
self.df_csv['drug_name_norm'].str.lower().str.replace(r'[^\w\s]', '', regex=True).str.strip()
)
self.drug_name_to_ids = self.df_csv.groupby('drug_name_norm_normalized')['drug_id'].unique().apply(list).to_dict()
# [MODIFIED] 把別名也變成可查鍵
for alias, canonical in DRUG_NAME_MAPPING.items():
alias_key = re.sub(r'[^\w\s]', '', alias.lower()).strip()
canonical_key = re.sub(r'[^\w\s]', '', canonical.lower()).strip()
if canonical_key in self.drug_name_to_ids:
self.drug_name_to_ids[alias_key] = self.drug_name_to_ids[canonical_key]
self._load_drug_name_vocabulary()
log.info("載入 FAISS 索引與句子資料...")
self.state.index = faiss.read_index(FAISS_INDEX)
self.state.faiss_metric = getattr(self.state.index, "metric_type", faiss.METRIC_L2)
if hasattr(self.state.index, "nprobe"):
self.state.index.nprobe = int(os.getenv("FAISS_NPROBE", "16"))
with open(SENTENCES_PKL, "rb") as f:
data = pickle.load(f)
self.state.sentences = data["sentences"]
self.state.meta = data["meta"]
log.info("載入 BM25 索引...")
with open(BM25_PKL, "rb") as f:
# 載入整個字典,然後取 'bm25' 這個鍵
bm25_data = pickle.load(f)
self.state.bm25 = bm25_data["bm25"]
if not isinstance(self.state.bm25, BM25Okapi):
raise ValueError("Loaded BM25 is not a BM25Okapi instance.")
except (FileNotFoundError, KeyError) as e:
log.exception(f"資料或索引檔案載入失敗: {e}")
raise RuntimeError(f"資料初始化失敗,請檢查檔案路徑與內容: {e}")
log.info("所有模型與資料載入完成。")
def _load_drug_name_vocabulary(self):
log.info("建立藥名詞庫...")
for norm_name in self.df_csv['drug_name_norm_normalized'].dropna().unique():
parts = norm_name.split()
for part in parts:
if re.search(r'[\u4e00-\u9fff]', part):
self.drug_vocab["zh"].add(part)
# [MODIFIED] 檢查詞彙是否已存在
if part not in jieba.dt.FREQ:
try:
jieba.add_word(part, freq=2_000_000)
except Exception:
pass
else:
self.drug_vocab["en"].add(part)
for alias in DRUG_NAME_MAPPING:
self.drug_vocab["en"].add(alias.lower())
if re.search(r'[\u4e00-\u9fff]', alias):
if alias not in jieba.dt.FREQ:
try:
jieba.add_word(alias, freq=2_000_000)
except Exception:
pass
@tenacity.retry(
wait=tenacity.wait_fixed(2),
stop=tenacity.stop_after_attempt(3),
retry=tenacity.retry_if_exception_type(ValueError),
before_sleep=tenacity.before_sleep_log(log, logging.WARNING),
after=tenacity.after_log(log, logging.INFO)
)
def _llm_call(self, messages: List[Dict[str, str]], max_tokens: Optional[int] = None, temperature: Optional[float] = None) -> str:
"""安全地呼叫 LLM API,並處理可能的回應內容為空錯誤。"""
log.info(f"LLM 呼叫開始. 模型: {self.model_name}, max_tokens: {max_tokens}, temperature: {temperature}")
# [DEBUG] 記錄完整的 LLM 提示內容,以便除錯
log.info(f"送出的 LLM 提示 (messages): {json.dumps(messages, ensure_ascii=False, indent=2)}")
start_time = time.time()
try:
response = self.client.chat.completions.create(
model=self.model_name,
messages=messages,
max_tokens=max_tokens,
temperature=temperature,
)
end_time = time.time()
# [DEBUG] 記錄 LLM 呼叫的完整 JSON 回應,即使內容為空
log.info(f"LLM 收到完整回應: {response.model_dump_json(indent=2)}")
if not response.choices or not response.choices[0].message.content:
# 即使狀態碼是 200 OK,若內容為空,也視為錯誤
log.error("LLM 呼叫成功 (200 OK),但回傳內容為空。")
raise ValueError("LLM response content is empty or not a string.")
content = response.choices[0].message.content
log.info(f"LLM 呼叫完成,耗時: {end_time - start_time:.2f} 秒。內容長度: {len(content)} 字。")
return content
except Exception as e:
# 捕獲所有其他可能的錯誤,並重新拋出以便 tenacity 處理
log.error(f"LLM API 呼叫失敗: {e}")
raise
# [MODIFIED] 實現動態流程,根據查詢複雜度決定是否使用 Reranker
def answer_question(self, q_orig: str) -> str:
start_time = time.time()
log.info(f"===== 處理新查詢: '{q_orig}' =====")
try:
drug_ids = self._find_drug_ids_from_name(q_orig)
if not drug_ids:
log.info("找不到藥品 ID,無法回答。")
return f"抱歉,資料庫中找不到該藥品。請確認藥品名稱,或直接諮詢醫師/藥師。{DISCLAIMER}"
log.info(f"步驟 1/5: 找到藥品 ID: {drug_ids},耗時: {time.time() - start_time:.2f} 秒")
step_start = time.time()
analysis = self._analyze_query(q_orig)
sub_queries, intents = analysis.get("sub_queries", [q_orig]), analysis.get("intents", [])
is_simple_query = self._is_simple_query(sub_queries, intents)
log.info(f"步驟 2/5: 意圖分析完成。子問題: {sub_queries}, 意圖: {intents}。判定為簡單查詢: {is_simple_query}。耗時: {time.time() - step_start:.2f} 秒")
step_start = time.time()
all_candidates = self._retrieve_candidates_for_all_queries(drug_ids, sub_queries, intents)
log.info(f"步驟 3/5: 檢索完成。所有子查詢共找到 {len(all_candidates)} 個不重複候選 chunks。耗時: {time.time() - step_start:.2f} 秒")
step_start = time.time()
if is_simple_query:
log.info("偵測到簡單查詢,跳過 Reranker 步驟。")
final_candidates = all_candidates[:TOP_K_SENTENCES]
reranked_results = [
RerankResult(idx=c.idx, rerank_score=c.fused_score, text=self.state.sentences[c.idx], meta=self.state.meta[c.idx])
for c in final_candidates
]
else:
log.info("偵測到複雜查詢,執行 Reranker。")
reranked_results = self._rerank_with_crossencoder(q_orig, all_candidates)
log.info(f"步驟 4/5: 最終選出 {len(reranked_results)} 個高品質候選。耗時: {time.time() - step_start:.2f} 秒")
step_start = time.time()
context = self._build_context(reranked_results)
if not context:
log.info("沒有足夠的上下文來回答問題。")
return f"根據您的問題,找不到相關的具體說明。建議您直接諮詢醫師或藥師以獲得最準確的資訊。{DISCLAIMER}"
prompt = self._make_final_prompt(q_orig, context, intents)
answer = self._llm_call([{"role": "user", "content": prompt}])
final_answer = f"{answer.strip()}\n\n{DISCLAIMER}"
log.info(f"步驟 5/5: 答案生成完成。答案長度: {len(answer.strip())} 字。耗時: {time.time() - step_start:.2f} 秒")
log.info(f"===== 查詢處理完成,總耗時: {time.time() - start_time:.2f} 秒 =====")
return final_answer
except Exception as e:
log.error(f"處理查詢 '{q_orig}' 時發生嚴重錯誤: {e}", exc_info=True)
return f"處理您的問題時發生內部錯誤,請稍後再試。{DISCLAIMER}"
def _is_simple_query(self, sub_queries: List[str], intents: List[str]) -> bool:
# 如果意圖分析回傳的子查詢數量 <= 1,且意圖分類數量也 <= 1,則判定為簡單問題
return len(sub_queries) <= 1 and len(intents) <= 1
@lru_cache(maxsize=128)
def _find_drug_ids_from_name(self, query: str) -> List[str]:
q = query.lower()
candidates = extract_drug_candidates_from_query(q, self.drug_vocab)
drug_ids = set()
# 英文:詞邊界;中文:也做子字串掃描
for k, ids in self.drug_name_to_ids.items():
if re.search(r'[\u4e00-\u9fff]', k):
if k in q:
drug_ids.update(ids)
else:
if re.search(rf"\b{re.escape(k)}\b", q):
drug_ids.update(ids)
# 仍保留舊的候選詞路徑(補強)
for alias in candidates:
# [MODIFIED] 英文藥名比對使用詞邊界,避免子字串誤判
is_english = not re.search(r'[\u4e00-\u9fff]', alias)
for drug_name_norm, ids in self.drug_name_to_ids.items():
match = False
if is_english:
if re.search(rf"\b{re.escape(alias)}\b", drug_name_norm):
match = True
elif alias in drug_name_norm:
match = True
if match:
drug_ids.update(ids)
return list(drug_ids)
def _analyze_query(self, query: str) -> Dict[str, Any]:
prompt = PROMPT_TEMPLATES["analyze_query"].format(
options="\n".join(f"- {c}" for c in INTENT_CATEGORIES),
query=query
)
response_str = self._llm_call([{"role": "user", "content": prompt}], temperature=0.1)
return self._safe_json_parse(response_str, default={"sub_queries": [query], "intents": []})
def _retrieve_candidates_for_all_queries(self, drug_ids: List[str], sub_queries: List[str], intents: List[str]) -> List[FusedCandidate]:
drug_ids_set = set(map(str, drug_ids))
relevant_indices = {i for i, m in enumerate(self.state.meta) if str(m.get("drug_id", "")) in drug_ids_set}
if not relevant_indices: return []
all_fused_candidates: Dict[int, FusedCandidate] = {}
for sub_q in sub_queries:
expanded_q = self._expand_query_with_llm(sub_q, tuple(intents))
q_emb = self.embedding_model.encode([expanded_q], convert_to_numpy=True).astype("float32")
if self.state.faiss_metric == faiss.METRIC_INNER_PRODUCT:
faiss.normalize_L2(q_emb)
distances, sim_indices = self.state.index.search(q_emb, PRE_RERANK_K)
tokenized_query = list(jieba.cut(expanded_q))
# [MODIFIED] 先過濾 relevant_indices 再取 TopK
bm25_scores = self.state.bm25.get_scores(tokenized_query)
rel_idx = np.fromiter(relevant_indices, dtype=int)
rel_scores = bm25_scores[rel_idx]
top_rel = rel_idx[np.argsort(rel_scores)[::-1][:PRE_RERANK_K]]
doc_to_bm25_score = {int(i): float(bm25_scores[i]) for i in top_rel}
candidate_scores: Dict[int, Dict[str, float]] = {}
# [MODIFIED] 把 distance 轉成「越大越好的相似度」
def to_similarity(d: float) -> float:
if self.state.faiss_metric == faiss.METRIC_INNER_PRODUCT:
return float(d) # IP 越大越好
else: # METRIC_L2(多半是平方 L2)
return 1.0 / (1.0 + float(d))
for i, dist in zip(sim_indices[0], distances[0]):
if i in relevant_indices:
similarity = to_similarity(dist)
candidate_scores[int(i)] = {"sem": float(similarity), "bm": 0.0}
for i, score in doc_to_bm25_score.items():
if i in relevant_indices:
candidate_scores.setdefault(i, {"sem": 0.0, "bm": 0.0})["bm"] = score
if not candidate_scores: continue
# [MODIFIED] 使用固定的鍵順序來確保分數對齊
keys = list(candidate_scores.keys())
sem_scores = np.array([candidate_scores[k]['sem'] for k in keys])
bm_scores = np.array([candidate_scores[k]['bm'] for k in keys])
def norm(x):
rng = x.max() - x.min()
return (x - x.min()) / (rng + 1e-8) if rng > 0 else np.zeros_like(x)
sem_n, bm_n = norm(sem_scores), norm(bm_scores)
for idx, k in enumerate(keys):
fused_score = sem_n[idx] * 0.6 + bm_n[idx] * 0.4
if k not in all_fused_candidates or fused_score > all_fused_candidates[k].fused_score:
all_fused_candidates[k] = FusedCandidate(
idx=k, fused_score=fused_score, sem_score=sem_scores[idx], bm_score=bm_scores[idx]
)
return sorted(all_fused_candidates.values(), key=lambda x: x.fused_score, reverse=True)
# [MODIFIED] 移除 lru_cache,因對多變的長查詢效果不佳
def _expand_query_with_llm(self, query: str, intents: tuple) -> str:
if not intents:
return query
prompt = PROMPT_TEMPLATES["expand_query"].format(intents=list(intents), query=query)
try:
expanded_query = self._llm_call([{"role": "user", "content": prompt}])
if expanded_query and expanded_query.strip():
log.info(f"查詢擴展成功。原始: '{query}', 擴展後: '{expanded_query}'")
return expanded_query
else:
log.warning(f"查詢擴展回傳空內容。原始查詢: '{query}'。將使用原始查詢。")
return query
except Exception as e:
log.error(f"查詢擴展失敗: {e}。原始查詢: '{query}'。將使用原始查詢。")
return query
def _rerank_with_crossencoder(self, query: str, candidates: List[FusedCandidate]) -> List[RerankResult]:
if not candidates: return []
top_candidates = candidates[:MAX_RERANK_CANDIDATES]
pairs = [(query, self.state.sentences[c.idx]) for c in top_candidates]
scores = self.reranker.predict(pairs, show_progress_bar=False)
results = [
RerankResult(idx=c.idx, rerank_score=float(score), text=self.state.sentences[c.idx], meta=self.state.meta[c.idx])
for c, score in zip(top_candidates, scores)
]
return sorted(results, key=lambda x: x.rerank_score, reverse=True)[:TOP_K_SENTENCES]
def _build_context(self, reranked_results: List[RerankResult]) -> str:
context = ""
for res in reranked_results:
if len(context) + len(res.text) > LLM_MODEL_CONFIG["max_context_chars"]: break
context += res.text + "\n\n"
return context.strip()
def _make_final_prompt(self, query: str, context: str, intents: List[str]) -> str:
add_instr = ""
if any(i in intents for i in ["劑量調整 (Dosage Adjustment)", "時間/併用 (Timing & Interaction)"]):
add_instr = "在回答用藥劑量和時間時,務必提醒使用者,醫師開立的藥袋醫囑優先於仿單的一般建議。"
return PROMPT_TEMPLATES["final_answer"].format(
additional_instruction=add_instr, context=context, query=query
)
# [MODIFIED] 增強 JSON 解析的穩健性,從字串中提取 JSON 物件
def _safe_json_parse(self, s: str, default: Any = None) -> Any:
try:
# 嘗試解析完整字串
return json.loads(s)
except json.JSONDecodeError:
log.warning(f"無法解析完整 JSON。嘗試從字串中提取: {s[:200]}...")
# 如果失敗,嘗試用 regex 提取第一個 JSON 物件
m = re.search(r'\{.*?\}', s, re.DOTALL)
if m:
try:
return json.loads(m.group(0))
except json.JSONDecodeError:
log.warning(f"提取的 JSON 仍無法解析: {m.group(0)[:100]}...")
return default
# ---------- FastAPI 事件與路由 ----------
# [MODIFIED] 將 LINE 配置集中管理並進行啟動時檢查
class AppConfig:
CHANNEL_ACCESS_TOKEN = _require_env("CHANNEL_ACCESS_TOKEN")
CHANNEL_SECRET = _require_env("CHANNEL_SECRET")
rag_pipeline: Optional[RagPipeline] = None
# [MODIFIED] 使用 lifespan context manager
@asynccontextmanager
async def lifespan(app: FastAPI):
_require_llm_config()
global rag_pipeline
rag_pipeline = RagPipeline()
rag_pipeline.load_data()
log.info("啟動完成,服務準備就緒。")
yield
# 若有資源需要關閉可在這裡實作
log.info("服務關閉中。")
app = FastAPI(lifespan=lifespan)
@app.post("/webhook")
async def handle_webhook(request: Request, background_tasks: BackgroundTasks):
# [MODIFIED] 增強簽章驗證與環境變數檢查
signature = request.headers.get("X-Line-Signature")
if not signature:
raise HTTPException(status_code=400, detail="Missing X-Line-Signature")
if not AppConfig.CHANNEL_SECRET:
log.error("CHANNEL_SECRET is not configured.")
raise HTTPException(status_code=500, detail="Server configuration error")
body = await request.body()
try:
hash = hmac.new(AppConfig.CHANNEL_SECRET.encode('utf-8'), body, hashlib.sha256)
expected_signature = base64.b64encode(hash.digest()).decode('utf-8')
except Exception as e:
log.error(f"Failed to generate signature: {e}")
raise HTTPException(status_code=500, detail="Signature generation error")
if not hmac.compare_digest(expected_signature, signature):
raise HTTPException(status_code=403, detail="Invalid signature")
try:
data = json.loads(body.decode('utf-8'))
except json.JSONDecodeError:
raise HTTPException(status_code=400, detail="Invalid JSON body")
for event in data.get("events", []):
if event.get("type") == "message" and event.get("message", {}).get("type") == "text":
reply_token = event.get("replyToken")
user_text = event.get("message", {}).get("text", "").strip()
# [MODIFIED] 擷取 target
source = event.get("source", {})
stype = source.get("type") # "user" | "group" | "room"
target_id = source.get("userId") or source.get("groupId") or source.get("roomId")
if reply_token and user_text and target_id:
# [MODIFIED] 更改回覆策略:立即回覆處理中訊息,避免 replyToken 逾時
line_reply(reply_token, "收到您的問題,正在查詢資料庫,請稍候...")
# 將耗時的任務交給背景處理,使用 push message 回覆最終答案
background_tasks.add_task(process_user_query, stype, target_id, user_text)
return Response(status_code=status.HTTP_200_OK)
# [MODIFIED] 調整函式簽名,只接收 user_id 和 text,並使用 push message
def process_user_query(source_type: str, target_id: str, user_text: str):
try:
if rag_pipeline:
answer = rag_pipeline.answer_question(user_text)
else:
answer = "系統正在啟動中,請稍後再試。"
line_push_generic(source_type, target_id, answer)
except Exception as e:
log.error(f"背景處理 target_id={target_id} 發生錯誤: {e}", exc_info=True)
line_push_generic(source_type, target_id, f"抱歉,處理時發生未預期的錯誤。{DISCLAIMER}")
@retry(stop=stop_after_attempt(3), wait=wait_fixed(2))
def line_api_call(endpoint: str, data: Dict):
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {AppConfig.CHANNEL_ACCESS_TOKEN}"
}
try:
response = requests.post(f"https://api.line.me/v2/bot/message/{endpoint}", headers=headers, json=data, timeout=10)
response.raise_for_status()
except requests.exceptions.RequestException as e:
log.error(f"LINE API ({endpoint}) 呼叫失敗: {e} | Response: {e.response.text if e.response else 'N/A'}")
raise
def line_reply(reply_token: str, text: str):
messages = [{"type": "text", "text": chunk} for chunk in textwrap.wrap(text, 4800, replace_whitespace=False)[:5]]
line_api_call("reply", {"replyToken": reply_token, "messages": messages})
def line_push_generic(source_type: str, target_id: str, text: str):
messages = [{"type": "text", "text": chunk} for chunk in textwrap.wrap(text, 4800, replace_whitespace=False)[:5]]
endpoint = "push"
data = {"to": target_id, "messages": messages}
line_api_call(endpoint, data)
# [MODIFIED] 改善藥名提取的正則表達式
def extract_drug_candidates_from_query(query: str, drug_vocab: dict) -> list:
candidates = set()
q_lower = query.lower()
# 允許藥名中包含 -, /, . 等符號
words = re.findall(r"[a-z0-9][a-z0-9+\-/\.]*", q_lower)
for word in words:
if word in drug_vocab["en"]:
candidates.add(word)
for token in jieba.cut(q_lower):
if token in drug_vocab["zh"]:
candidates.add(token)
return list(candidates)
# ---------- 執行 ----------
if __name__ == "__main__":
port = int(os.getenv("PORT", 7860))
uvicorn.run(app, host="0.0.0.0", port=port) |