Song
commited on
Commit
·
a351846
1
Parent(s):
d1d8455
hi
Browse files- .gitignore +1 -0
- app.py +44 -49
.gitignore
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
app_origin.py
|
app.py
CHANGED
|
@@ -2,7 +2,6 @@
|
|
| 2 |
# -*- coding: utf-8 -*-
|
| 3 |
# ---------- 環境與快取設定 (應置於最前) ----------
|
| 4 |
import os
|
| 5 |
-
import json
|
| 6 |
import time
|
| 7 |
from typing import List, Dict, Any
|
| 8 |
from contextlib import asynccontextmanager
|
|
@@ -20,8 +19,8 @@ from linebot.v3.webhook import WebhookParser
|
|
| 20 |
from linebot.v3.exceptions import InvalidSignatureError
|
| 21 |
# --------------------------------------------------------------
|
| 22 |
from openai import OpenAI
|
| 23 |
-
from tavily import TavilyClient #
|
| 24 |
-
from sentence_transformers import SentenceTransformer, util #
|
| 25 |
|
| 26 |
# ==== CONFIG (從環境變數載入,或使用預設值) ====
|
| 27 |
def _require_env(var: str) -> str:
|
|
@@ -34,24 +33,24 @@ def _require_env(var: str) -> str:
|
|
| 34 |
CHANNEL_SECRET = _require_env("CHANNEL_SECRET")
|
| 35 |
CHANNEL_ACCESS_TOKEN = _require_env("CHANNEL_ACCESS_TOKEN")
|
| 36 |
|
| 37 |
-
# Tavily API Key (
|
| 38 |
-
TAVILY_API_KEY = "
|
| 39 |
|
| 40 |
-
# LLM API
|
| 41 |
LLM_API_CONFIG = {
|
| 42 |
-
"base_url": os.getenv("LLM_BASE_URL", "https://
|
| 43 |
-
"api_key":
|
| 44 |
}
|
| 45 |
|
| 46 |
-
# LLM 模型設定 (
|
| 47 |
LLM_MODEL_CONFIG = {
|
| 48 |
-
"model": os.getenv("LLM_MODEL", "
|
| 49 |
-
"max_tokens": int(os.getenv("MAX_TOKENS", 2000)),
|
| 50 |
"temperature": float(os.getenv("TEMPERATURE", 0.3)),
|
| 51 |
"seed": int(os.getenv("LLM_SEED", 42)),
|
| 52 |
}
|
| 53 |
|
| 54 |
-
#
|
| 55 |
SYSTEM_PROMPT = """你是一個友好的AI助手,請用簡單、親切的文字回覆用戶的問題。
|
| 56 |
回答複雜問題時,先給概念,再給詳細解釋。
|
| 57 |
使用條列式(如 - 或 1. 2. 3.)整理內容,讓它適合手機閱讀。
|
|
@@ -90,9 +89,9 @@ def estimate_tokens(messages: List[Dict[str, str]]) -> int:
|
|
| 90 |
total += len(msg["content"].split()) * 1.3 # 粗估 token
|
| 91 |
return total
|
| 92 |
|
| 93 |
-
# ----------
|
| 94 |
-
def perform_web_search(query: str, max_results: int = 5) -> str:
|
| 95 |
-
"""使用 Tavily
|
| 96 |
print(f"開始網路搜尋:查詢詞 = '{query}',最大結果數 = {max_results}")
|
| 97 |
try:
|
| 98 |
client = TavilyClient(api_key=TAVILY_API_KEY)
|
|
@@ -101,11 +100,10 @@ def perform_web_search(query: str, max_results: int = 5) -> str: # 改為 top 5
|
|
| 101 |
print("搜尋完成:沒有找到相關結果。")
|
| 102 |
return "沒有找到相關的網路搜尋結果。"
|
| 103 |
|
| 104 |
-
#
|
| 105 |
-
embedder =
|
| 106 |
query_emb = embedder.encode(query)
|
| 107 |
|
| 108 |
-
# 計算每個結果的相似度 (文字意義排序)
|
| 109 |
results_with_scores = []
|
| 110 |
for result in response['results']:
|
| 111 |
content = result['content']
|
|
@@ -113,7 +111,6 @@ def perform_web_search(query: str, max_results: int = 5) -> str: # 改為 top 5
|
|
| 113 |
score = util.cos_sim(query_emb, content_emb)[0][0].item()
|
| 114 |
results_with_scores.append((score, result))
|
| 115 |
|
| 116 |
-
# 排序並過濾相似度 > 0.3 的結果(確保相關性)
|
| 117 |
results_with_scores.sort(key=lambda x: x[0], reverse=True)
|
| 118 |
relevant_results = [res for score, res in results_with_scores if score > 0.3]
|
| 119 |
|
|
@@ -123,8 +120,9 @@ def perform_web_search(query: str, max_results: int = 5) -> str: # 改為 top 5
|
|
| 123 |
|
| 124 |
search_summary = "以下是相關的網路搜尋結果摘要(已按文字相似度排序):\n"
|
| 125 |
search_summary += f"AI總結:{response.get('answer', '無總結可用')}\n\n"
|
| 126 |
-
for i, result in enumerate(relevant_results[:5], 1):
|
| 127 |
-
|
|
|
|
| 128 |
search_summary += f"{i}. {result['title']}: {result['content'][:200]}... (來源: {result['url']})\n"
|
| 129 |
print(f"搜尋完成:總結果數 = {len(response['results'])}, 相關結果數 = {len(relevant_results)}")
|
| 130 |
return search_summary
|
|
@@ -133,18 +131,29 @@ def perform_web_search(query: str, max_results: int = 5) -> str: # 改為 top 5
|
|
| 133 |
return f"搜尋時發生錯誤:{str(e)}。請稍後再試。"
|
| 134 |
|
| 135 |
# ---------- 聊天處理流程 (新增 retry 和 timeout) ----------
|
| 136 |
-
from tenacity import retry, stop_after_attempt, wait_exponential
|
| 137 |
|
| 138 |
class ChatPipeline:
|
| 139 |
def __init__(self):
|
| 140 |
if not LLM_API_CONFIG["api_key"] or not LLM_API_CONFIG["base_url"]:
|
| 141 |
raise ValueError("LLM API Key or Base URL is not configured.")
|
| 142 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 143 |
|
| 144 |
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=4, max=10))
|
| 145 |
def _llm_call(self, messages: List[Dict[str, str]]) -> str:
|
| 146 |
try:
|
| 147 |
-
# 估算 token 並 print 監控
|
| 148 |
token_est = estimate_tokens(messages)
|
| 149 |
print(f"LLM 呼叫:估計 token = {token_est}")
|
| 150 |
if token_est > 50000:
|
|
@@ -156,7 +165,7 @@ class ChatPipeline:
|
|
| 156 |
max_tokens=LLM_MODEL_CONFIG["max_tokens"],
|
| 157 |
temperature=LLM_MODEL_CONFIG["temperature"],
|
| 158 |
seed=LLM_MODEL_CONFIG["seed"],
|
| 159 |
-
timeout=30.0,
|
| 160 |
)
|
| 161 |
content = response.choices[0].message.content or ""
|
| 162 |
return content
|
|
@@ -168,7 +177,6 @@ class ChatPipeline:
|
|
| 168 |
return conversations.get(user_id, [])
|
| 169 |
|
| 170 |
def update_conversation_history(self, user_id: str, messages: List[Dict[str, str]]):
|
| 171 |
-
# 限制歷史:保留最近 20 條訊息 (約 10 輪)
|
| 172 |
recent = messages[-20:]
|
| 173 |
conversations[user_id] = recent
|
| 174 |
|
|
@@ -189,21 +197,19 @@ class ChatPipeline:
|
|
| 189 |
messages = [{"role": "system", "content": SYSTEM_PROMPT}]
|
| 190 |
messages.extend(history)
|
| 191 |
messages.append({"role": "user", "content": user_text})
|
| 192 |
-
if "沒有找到" not in search_results:
|
| 193 |
messages.append({"role": "system", "content": f"網路搜尋結果:{search_results}"})
|
| 194 |
|
| 195 |
response = self._llm_call(messages)
|
| 196 |
response = response.replace('*', '')
|
| 197 |
|
| 198 |
-
# 更新歷史紀錄
|
| 199 |
history.append({"role": "user", "content": user_text})
|
| 200 |
history.append({"role": "assistant", "content": response})
|
| 201 |
self.update_conversation_history(user_id, history)
|
| 202 |
|
| 203 |
-
# 如果回應過長,檢查 chunks 數量,如果超過5,生成摘要
|
| 204 |
chunks = split_text_for_line(response)
|
| 205 |
if len(chunks) > 5:
|
| 206 |
-
summary_prompt = [{"role": "system", "content": "請將以下內容生成一個簡潔但完整的中文摘要,保留關鍵事實和細節,長度控制在
|
| 207 |
summary_prompt.append({"role": "user", "content": response})
|
| 208 |
summary = self._llm_call(summary_prompt)
|
| 209 |
summary = summary.replace('*', '')
|
|
@@ -219,22 +225,17 @@ async def lifespan(app: FastAPI):
|
|
| 219 |
yield
|
| 220 |
|
| 221 |
app = FastAPI(lifespan=lifespan)
|
| 222 |
-
chat_pipeline = None
|
| 223 |
|
| 224 |
-
# ----------------- LINE Bot API v3
|
| 225 |
-
# 建立一個 Configuration 物件,並傳入你的 Access Token
|
| 226 |
configuration = Configuration(access_token=CHANNEL_ACCESS_TOKEN)
|
| 227 |
-
# 使用 Configuration 物件來初始化 AsyncApiClient 和 AsyncMessagingApi
|
| 228 |
async_api_client = AsyncApiClient(configuration)
|
| 229 |
line_bot_api = AsyncMessagingApi(async_api_client)
|
| 230 |
-
# 建立 WebhookParser 來解析請求
|
| 231 |
parser = WebhookParser(CHANNEL_SECRET)
|
| 232 |
-
# --------------------------------------------------------------
|
| 233 |
|
| 234 |
# ---------- LINE Webhook 處理 ----------
|
| 235 |
@app.post("/webhook")
|
| 236 |
async def line_webhook(request: Request):
|
| 237 |
-
# 驗證簽名
|
| 238 |
signature = request.headers['X-Line-Signature']
|
| 239 |
body = await request.body()
|
| 240 |
try:
|
|
@@ -243,7 +244,6 @@ async def line_webhook(request: Request):
|
|
| 243 |
raise HTTPException(status_code=400, detail="Invalid signature")
|
| 244 |
|
| 245 |
for event in events:
|
| 246 |
-
# 只處理文字訊息事件
|
| 247 |
if event.type != 'message' or event.message.type != 'text':
|
| 248 |
continue
|
| 249 |
|
|
@@ -256,7 +256,6 @@ async def line_webhook(request: Request):
|
|
| 256 |
|
| 257 |
try:
|
| 258 |
if user_text.lower() == "繼續" and user_id in pending_chunks:
|
| 259 |
-
# 處理繼續發送剩餘 chunks
|
| 260 |
remaining = pending_chunks[user_id]
|
| 261 |
if not remaining:
|
| 262 |
ai_response = "沒有更多內容了。"
|
|
@@ -276,22 +275,19 @@ async def line_webhook(request: Request):
|
|
| 276 |
messages=messages_to_send
|
| 277 |
)
|
| 278 |
)
|
| 279 |
-
continue
|
| 280 |
|
| 281 |
-
# 正常處理查詢
|
| 282 |
ai_response = chat_pipeline.answer_question(user_id, user_text)
|
| 283 |
chunks = split_text_for_line(ai_response)
|
| 284 |
|
| 285 |
if len(chunks) <= 5:
|
| 286 |
messages_to_send = [TextMessage(text=chunk) for chunk in chunks]
|
| 287 |
else:
|
| 288 |
-
# 發送前5個,並儲存剩餘
|
| 289 |
chunks_to_send = chunks[:5]
|
| 290 |
messages_to_send = [TextMessage(text=chunk) for chunk in chunks_to_send]
|
| 291 |
messages_to_send[-1].text += "\n\n內容過長,請回覆 '繼續' 以查看下一部分。"
|
| 292 |
pending_chunks[user_id] = chunks[5:]
|
| 293 |
|
| 294 |
-
# 發送訊息
|
| 295 |
await line_bot_api.reply_message(
|
| 296 |
ReplyMessageRequest(
|
| 297 |
reply_token=reply_token,
|
|
@@ -301,12 +297,11 @@ async def line_webhook(request: Request):
|
|
| 301 |
except Exception as e:
|
| 302 |
print(f"Error processing message: {e}")
|
| 303 |
error_message = "抱歉,系統發生錯誤,請稍後再試。"
|
| 304 |
-
# 使用 await 來呼叫非同步的 reply_message
|
| 305 |
await line_bot_api.reply_message(
|
| 306 |
-
|
| 307 |
-
|
| 308 |
-
|
| 309 |
-
|
| 310 |
)
|
| 311 |
|
| 312 |
return {"status": "ok"}
|
|
@@ -316,7 +311,7 @@ async def line_webhook(request: Request):
|
|
| 316 |
async def health_check():
|
| 317 |
return {"status": "ok"}
|
| 318 |
|
| 319 |
-
#
|
| 320 |
@app.get("/")
|
| 321 |
async def root():
|
| 322 |
return {"message": "LINE Bot is running"}
|
|
|
|
| 2 |
# -*- coding: utf-8 -*-
|
| 3 |
# ---------- 環境與快取設定 (應置於最前) ----------
|
| 4 |
import os
|
|
|
|
| 5 |
import time
|
| 6 |
from typing import List, Dict, Any
|
| 7 |
from contextlib import asynccontextmanager
|
|
|
|
| 19 |
from linebot.v3.exceptions import InvalidSignatureError
|
| 20 |
# --------------------------------------------------------------
|
| 21 |
from openai import OpenAI
|
| 22 |
+
from tavily import TavilyClient # Tavily 客戶端
|
| 23 |
+
from sentence_transformers import SentenceTransformer, util # 用於向量相似度排序(CPU友好)
|
| 24 |
|
| 25 |
# ==== CONFIG (從環境變數載入,或使用預設值) ====
|
| 26 |
def _require_env(var: str) -> str:
|
|
|
|
| 33 |
CHANNEL_SECRET = _require_env("CHANNEL_SECRET")
|
| 34 |
CHANNEL_ACCESS_TOKEN = _require_env("CHANNEL_ACCESS_TOKEN")
|
| 35 |
|
| 36 |
+
# Tavily API Key (強制從環境變數讀取,移除硬編碼)
|
| 37 |
+
TAVILY_API_KEY = _require_env("TAVILY_API_KEY")
|
| 38 |
|
| 39 |
+
# LLM API 設定(改用 OpenRouter)
|
| 40 |
LLM_API_CONFIG = {
|
| 41 |
+
"base_url": os.getenv("LLM_BASE_URL", "https://openrouter.ai/api/v1"),
|
| 42 |
+
"api_key": _require_env("OPENROUTER_API_KEY"), # 強制要求 OpenRouter API Key
|
| 43 |
}
|
| 44 |
|
| 45 |
+
# LLM 模型設定 (預設改用 gpt-4o,性價比高)
|
| 46 |
LLM_MODEL_CONFIG = {
|
| 47 |
+
"model": os.getenv("LLM_MODEL", "xiaomi/mimo-v2-flash:free"),
|
| 48 |
+
"max_tokens": int(os.getenv("MAX_TOKENS", 2000)),
|
| 49 |
"temperature": float(os.getenv("TEMPERATURE", 0.3)),
|
| 50 |
"seed": int(os.getenv("LLM_SEED", 42)),
|
| 51 |
}
|
| 52 |
|
| 53 |
+
# 系統提示詞(保持原樣)
|
| 54 |
SYSTEM_PROMPT = """你是一個友好的AI助手,請用簡單、親切的文字回覆用戶的問題。
|
| 55 |
回答複雜問題時,先給概念,再給詳細解釋。
|
| 56 |
使用條列式(如 - 或 1. 2. 3.)整理內容,讓它適合手機閱讀。
|
|
|
|
| 89 |
total += len(msg["content"].split()) * 1.3 # 粗估 token
|
| 90 |
return total
|
| 91 |
|
| 92 |
+
# ---------- 網路搜尋函數(優化:嵌入模型由 ChatPipeline 預載) ----------
|
| 93 |
+
def perform_web_search(query: str, max_results: int = 5) -> str:
|
| 94 |
+
"""使用 Tavily 進行網路搜尋,計算向量相似度排序結果,並返回摘要。"""
|
| 95 |
print(f"開始網路搜尋:查詢詞 = '{query}',最大結果數 = {max_results}")
|
| 96 |
try:
|
| 97 |
client = TavilyClient(api_key=TAVILY_API_KEY)
|
|
|
|
| 100 |
print("搜尋完成:沒有找到相關結果。")
|
| 101 |
return "沒有找到相關的網路搜尋結果。"
|
| 102 |
|
| 103 |
+
# 使用 ChatPipeline 中預載的 embedder
|
| 104 |
+
embedder = chat_pipeline.embedder
|
| 105 |
query_emb = embedder.encode(query)
|
| 106 |
|
|
|
|
| 107 |
results_with_scores = []
|
| 108 |
for result in response['results']:
|
| 109 |
content = result['content']
|
|
|
|
| 111 |
score = util.cos_sim(query_emb, content_emb)[0][0].item()
|
| 112 |
results_with_scores.append((score, result))
|
| 113 |
|
|
|
|
| 114 |
results_with_scores.sort(key=lambda x: x[0], reverse=True)
|
| 115 |
relevant_results = [res for score, res in results_with_scores if score > 0.3]
|
| 116 |
|
|
|
|
| 120 |
|
| 121 |
search_summary = "以下是相關的網路搜尋結果摘要(已按文字相似度排序):\n"
|
| 122 |
search_summary += f"AI總結:{response.get('answer', '無總結可用')}\n\n"
|
| 123 |
+
for i, result in enumerate(relevant_results[:5], 1):
|
| 124 |
+
score = results_with_scores[i-1][0]
|
| 125 |
+
print(f"結果 {i}: 標題 = '{result['title']}',內容 = '{result['content'][:200]}...',來源 = '{result['url']}',相似度 = {score:.2f}")
|
| 126 |
search_summary += f"{i}. {result['title']}: {result['content'][:200]}... (來源: {result['url']})\n"
|
| 127 |
print(f"搜尋完成:總結果數 = {len(response['results'])}, 相關結果數 = {len(relevant_results)}")
|
| 128 |
return search_summary
|
|
|
|
| 131 |
return f"搜尋時發生錯誤:{str(e)}。請稍後再試。"
|
| 132 |
|
| 133 |
# ---------- 聊天處理流程 (新增 retry 和 timeout) ----------
|
| 134 |
+
from tenacity import retry, stop_after_attempt, wait_exponential
|
| 135 |
|
| 136 |
class ChatPipeline:
|
| 137 |
def __init__(self):
|
| 138 |
if not LLM_API_CONFIG["api_key"] or not LLM_API_CONFIG["base_url"]:
|
| 139 |
raise ValueError("LLM API Key or Base URL is not configured.")
|
| 140 |
+
|
| 141 |
+
# 預載入嵌入模型(大幅提升搜尋速度)
|
| 142 |
+
self.embedder = SentenceTransformer('all-MiniLM-L6-v2')
|
| 143 |
+
|
| 144 |
+
# OpenAI client(相容 OpenRouter,並加入建議 headers)
|
| 145 |
+
self.llm_client = OpenAI(
|
| 146 |
+
api_key=LLM_API_CONFIG["api_key"],
|
| 147 |
+
base_url=LLM_API_CONFIG["base_url"],
|
| 148 |
+
default_headers={
|
| 149 |
+
"HTTP-Referer": os.getenv("SITE_URL", "https://your-line-bot.example.com"), # 建議設定你的網站域名
|
| 150 |
+
"X-Title": os.getenv("SITE_NAME", "My LINE Bot"), # 建議設定 Bot 名稱
|
| 151 |
+
}
|
| 152 |
+
)
|
| 153 |
|
| 154 |
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=4, max=10))
|
| 155 |
def _llm_call(self, messages: List[Dict[str, str]]) -> str:
|
| 156 |
try:
|
|
|
|
| 157 |
token_est = estimate_tokens(messages)
|
| 158 |
print(f"LLM 呼叫:估計 token = {token_est}")
|
| 159 |
if token_est > 50000:
|
|
|
|
| 165 |
max_tokens=LLM_MODEL_CONFIG["max_tokens"],
|
| 166 |
temperature=LLM_MODEL_CONFIG["temperature"],
|
| 167 |
seed=LLM_MODEL_CONFIG["seed"],
|
| 168 |
+
timeout=30.0,
|
| 169 |
)
|
| 170 |
content = response.choices[0].message.content or ""
|
| 171 |
return content
|
|
|
|
| 177 |
return conversations.get(user_id, [])
|
| 178 |
|
| 179 |
def update_conversation_history(self, user_id: str, messages: List[Dict[str, str]]):
|
|
|
|
| 180 |
recent = messages[-20:]
|
| 181 |
conversations[user_id] = recent
|
| 182 |
|
|
|
|
| 197 |
messages = [{"role": "system", "content": SYSTEM_PROMPT}]
|
| 198 |
messages.extend(history)
|
| 199 |
messages.append({"role": "user", "content": user_text})
|
| 200 |
+
if "沒有找到" not in search_results:
|
| 201 |
messages.append({"role": "system", "content": f"網路搜尋結果:{search_results}"})
|
| 202 |
|
| 203 |
response = self._llm_call(messages)
|
| 204 |
response = response.replace('*', '')
|
| 205 |
|
|
|
|
| 206 |
history.append({"role": "user", "content": user_text})
|
| 207 |
history.append({"role": "assistant", "content": response})
|
| 208 |
self.update_conversation_history(user_id, history)
|
| 209 |
|
|
|
|
| 210 |
chunks = split_text_for_line(response)
|
| 211 |
if len(chunks) > 5:
|
| 212 |
+
summary_prompt = [{"role": "system", "content": "請將以下內容生成一個簡潔但完整的中文摘要,保留關鍵事實和細節,長度控制在2000字元內。"}]
|
| 213 |
summary_prompt.append({"role": "user", "content": response})
|
| 214 |
summary = self._llm_call(summary_prompt)
|
| 215 |
summary = summary.replace('*', '')
|
|
|
|
| 225 |
yield
|
| 226 |
|
| 227 |
app = FastAPI(lifespan=lifespan)
|
| 228 |
+
chat_pipeline = None # 會在 lifespan 中初始化
|
| 229 |
|
| 230 |
+
# ----------------- LINE Bot API v3 初始化 -----------------
|
|
|
|
| 231 |
configuration = Configuration(access_token=CHANNEL_ACCESS_TOKEN)
|
|
|
|
| 232 |
async_api_client = AsyncApiClient(configuration)
|
| 233 |
line_bot_api = AsyncMessagingApi(async_api_client)
|
|
|
|
| 234 |
parser = WebhookParser(CHANNEL_SECRET)
|
|
|
|
| 235 |
|
| 236 |
# ---------- LINE Webhook 處理 ----------
|
| 237 |
@app.post("/webhook")
|
| 238 |
async def line_webhook(request: Request):
|
|
|
|
| 239 |
signature = request.headers['X-Line-Signature']
|
| 240 |
body = await request.body()
|
| 241 |
try:
|
|
|
|
| 244 |
raise HTTPException(status_code=400, detail="Invalid signature")
|
| 245 |
|
| 246 |
for event in events:
|
|
|
|
| 247 |
if event.type != 'message' or event.message.type != 'text':
|
| 248 |
continue
|
| 249 |
|
|
|
|
| 256 |
|
| 257 |
try:
|
| 258 |
if user_text.lower() == "繼續" and user_id in pending_chunks:
|
|
|
|
| 259 |
remaining = pending_chunks[user_id]
|
| 260 |
if not remaining:
|
| 261 |
ai_response = "沒有更多內容了。"
|
|
|
|
| 275 |
messages=messages_to_send
|
| 276 |
)
|
| 277 |
)
|
| 278 |
+
continue
|
| 279 |
|
|
|
|
| 280 |
ai_response = chat_pipeline.answer_question(user_id, user_text)
|
| 281 |
chunks = split_text_for_line(ai_response)
|
| 282 |
|
| 283 |
if len(chunks) <= 5:
|
| 284 |
messages_to_send = [TextMessage(text=chunk) for chunk in chunks]
|
| 285 |
else:
|
|
|
|
| 286 |
chunks_to_send = chunks[:5]
|
| 287 |
messages_to_send = [TextMessage(text=chunk) for chunk in chunks_to_send]
|
| 288 |
messages_to_send[-1].text += "\n\n內容過長,請回覆 '繼續' 以查看下一部分。"
|
| 289 |
pending_chunks[user_id] = chunks[5:]
|
| 290 |
|
|
|
|
| 291 |
await line_bot_api.reply_message(
|
| 292 |
ReplyMessageRequest(
|
| 293 |
reply_token=reply_token,
|
|
|
|
| 297 |
except Exception as e:
|
| 298 |
print(f"Error processing message: {e}")
|
| 299 |
error_message = "抱歉,系統發生錯誤,請稍後再試。"
|
|
|
|
| 300 |
await line_bot_api.reply_message(
|
| 301 |
+
ReplyMessageRequest(
|
| 302 |
+
reply_token=reply_token,
|
| 303 |
+
messages=[TextMessage(text=error_message)]
|
| 304 |
+
)
|
| 305 |
)
|
| 306 |
|
| 307 |
return {"status": "ok"}
|
|
|
|
| 311 |
async def health_check():
|
| 312 |
return {"status": "ok"}
|
| 313 |
|
| 314 |
+
# 根路由
|
| 315 |
@app.get("/")
|
| 316 |
async def root():
|
| 317 |
return {"message": "LINE Bot is running"}
|