Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -27,28 +27,61 @@ flask_app = Flask(__name__)
|
|
| 27 |
class NewsApp:
|
| 28 |
def __init__(self):
|
| 29 |
self.db = NewsDatabase()
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
self.scheduler = NewsScheduler(self.db, self.crawler, self.sentiment_analyzer)
|
| 36 |
|
| 37 |
# 進度追蹤
|
| 38 |
-
self.current_progress = "
|
| 39 |
self.is_crawling = False
|
|
|
|
| 40 |
|
| 41 |
-
#
|
| 42 |
-
self.
|
| 43 |
|
| 44 |
-
|
| 45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
|
| 47 |
-
|
|
|
|
|
|
|
| 48 |
|
| 49 |
def update_progress(self, message: str):
|
| 50 |
"""更新進度信息"""
|
| 51 |
-
|
|
|
|
| 52 |
logger.info(f"進度更新: {message}")
|
| 53 |
|
| 54 |
def get_progress(self) -> str:
|
|
@@ -60,7 +93,7 @@ class NewsApp:
|
|
| 60 |
try:
|
| 61 |
news_data = self.db.get_recent_news(category=category, limit=limit)
|
| 62 |
if not news_data:
|
| 63 |
-
return "📰
|
| 64 |
|
| 65 |
return format_news_for_display(news_data)
|
| 66 |
|
|
@@ -70,6 +103,9 @@ class NewsApp:
|
|
| 70 |
|
| 71 |
def manual_crawl(self) -> str:
|
| 72 |
"""手動觸發爬蟲"""
|
|
|
|
|
|
|
|
|
|
| 73 |
if self.is_crawling:
|
| 74 |
return "⚠️ 爬蟲正在運行中,請稍後再試"
|
| 75 |
|
|
@@ -179,6 +215,12 @@ def api_get_stats():
|
|
| 179 |
def api_manual_crawl():
|
| 180 |
"""手動觸發爬蟲API"""
|
| 181 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 182 |
if app.is_crawling:
|
| 183 |
return jsonify({
|
| 184 |
'success': False,
|
|
@@ -206,9 +248,17 @@ def api_get_progress():
|
|
| 206 |
"""獲取爬蟲進度API"""
|
| 207 |
return jsonify({
|
| 208 |
'progress': app.get_progress(),
|
| 209 |
-
'is_crawling': app.is_crawling
|
|
|
|
| 210 |
})
|
| 211 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 212 |
# 創建 Gradio 介面
|
| 213 |
def create_interface():
|
| 214 |
with gr.Blocks(
|
|
@@ -223,7 +273,7 @@ def create_interface():
|
|
| 223 |
.positive-badge { background: #28a745; color: white; }
|
| 224 |
.negative-badge { background: #dc3545; color: white; }
|
| 225 |
.neutral-badge { background: #6c757d; color: white; }
|
| 226 |
-
.progress-box { background: #f8f9fa; border: 1px solid #dee2e6; border-radius: 5px; padding: 10px; font-family: monospace; }
|
| 227 |
"""
|
| 228 |
) as interface:
|
| 229 |
|
|
@@ -253,21 +303,20 @@ def create_interface():
|
|
| 253 |
|
| 254 |
# 進度顯示
|
| 255 |
progress_display = gr.Textbox(
|
| 256 |
-
label="📊
|
| 257 |
value=app.get_progress(),
|
| 258 |
interactive=False,
|
| 259 |
-
elem_classes=["progress-box"]
|
|
|
|
| 260 |
)
|
| 261 |
|
| 262 |
news_display = gr.HTML(label="新聞內容")
|
| 263 |
crawl_result = gr.Textbox(label="爬取結果", visible=False)
|
| 264 |
|
| 265 |
-
#
|
| 266 |
-
def
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
def auto_refresh():
|
| 270 |
-
return app.get_latest_news("all")
|
| 271 |
|
| 272 |
def refresh_news(category):
|
| 273 |
return app.get_latest_news(category)
|
|
@@ -278,11 +327,11 @@ def create_interface():
|
|
| 278 |
news = app.get_latest_news("all")
|
| 279 |
return result, news
|
| 280 |
|
| 281 |
-
#
|
| 282 |
-
|
| 283 |
-
|
| 284 |
-
|
| 285 |
-
|
| 286 |
)
|
| 287 |
|
| 288 |
# 綁定事件
|
|
@@ -295,6 +344,9 @@ def create_interface():
|
|
| 295 |
outputs=[crawl_result]
|
| 296 |
)
|
| 297 |
category_radio.change(refresh_news, inputs=[category_radio], outputs=[news_display])
|
|
|
|
|
|
|
|
|
|
| 298 |
|
| 299 |
with gr.Tab("📊 統計資訊"):
|
| 300 |
stats_display = gr.Markdown()
|
|
@@ -350,7 +402,7 @@ def create_interface():
|
|
| 350 |
POST /api/crawl
|
| 351 |
```
|
| 352 |
|
| 353 |
-
#### 4.
|
| 354 |
```
|
| 355 |
GET /api/progress
|
| 356 |
```
|
|
@@ -362,12 +414,16 @@ def create_interface():
|
|
| 362 |
import requests
|
| 363 |
|
| 364 |
# 獲取所有新聞
|
| 365 |
-
response = requests.get('http://localhost:
|
| 366 |
news_data = response.json()
|
| 367 |
|
| 368 |
# 獲取美股新聞
|
| 369 |
-
response = requests.get('http://localhost:
|
| 370 |
us_news = response.json()
|
|
|
|
|
|
|
|
|
|
|
|
|
| 371 |
```
|
| 372 |
|
| 373 |
**JavaScript:**
|
|
@@ -398,6 +454,7 @@ def create_interface():
|
|
| 398 |
- **模型**: `uer/roberta-base-finetuned-jd-binary-chinese`
|
| 399 |
- **分類**: 正面 (綠色) / 負面 (紅色) / 中性 (灰色)
|
| 400 |
- **準確性**: 針對中文金融新聞優化
|
|
|
|
| 401 |
|
| 402 |
### 🕷️ 新聞爬蟲
|
| 403 |
- **來源**: 鉅亨網 (cnyes.com)
|
|
@@ -419,8 +476,9 @@ def create_interface():
|
|
| 419 |
|
| 420 |
---
|
| 421 |
|
| 422 |
-
💡
|
| 423 |
-
-
|
|
|
|
| 424 |
- 建議使用SSD硬碟提升SQLite性能
|
| 425 |
- API接口可用於整合其他應用系統
|
| 426 |
""")
|
|
@@ -438,6 +496,11 @@ if __name__ == "__main__":
|
|
| 438 |
flask_thread = threading.Thread(target=run_flask, daemon=True)
|
| 439 |
flask_thread.start()
|
| 440 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 441 |
# 啟動Gradio介面
|
| 442 |
interface = create_interface()
|
| 443 |
interface.launch(
|
|
|
|
| 27 |
class NewsApp:
|
| 28 |
def __init__(self):
|
| 29 |
self.db = NewsDatabase()
|
| 30 |
+
|
| 31 |
+
# 延遲初始化情緒分析器
|
| 32 |
+
self.sentiment_analyzer = None
|
| 33 |
+
self.crawler = None
|
| 34 |
+
self.scheduler = None
|
|
|
|
| 35 |
|
| 36 |
# 進度追蹤
|
| 37 |
+
self.current_progress = "正在初始化系統..."
|
| 38 |
self.is_crawling = False
|
| 39 |
+
self.is_initialized = False
|
| 40 |
|
| 41 |
+
# 在背景初始化重型組件
|
| 42 |
+
self._initialize_components()
|
| 43 |
|
| 44 |
+
logger.info("新聞應用程式開始初始化")
|
| 45 |
+
|
| 46 |
+
def _initialize_components(self):
|
| 47 |
+
"""在背景初始化重型組件"""
|
| 48 |
+
def init_task():
|
| 49 |
+
try:
|
| 50 |
+
self.update_progress("正在載入情緒分析模型...")
|
| 51 |
+
self.sentiment_analyzer = SentimentAnalyzer()
|
| 52 |
+
|
| 53 |
+
self.update_progress("正在初始化爬蟲...")
|
| 54 |
+
self.crawler = CnYesNewsCrawler(
|
| 55 |
+
sentiment_analyzer=self.sentiment_analyzer,
|
| 56 |
+
database=self.db
|
| 57 |
+
)
|
| 58 |
+
|
| 59 |
+
self.update_progress("正在設置排程器...")
|
| 60 |
+
self.scheduler = NewsScheduler(self.db, self.crawler, self.sentiment_analyzer)
|
| 61 |
+
|
| 62 |
+
# 設置爬蟲進度回調
|
| 63 |
+
self.crawler.set_progress_callback(self.update_progress)
|
| 64 |
+
|
| 65 |
+
# 啟動背景排程器
|
| 66 |
+
self.scheduler.start()
|
| 67 |
+
|
| 68 |
+
self.is_initialized = True
|
| 69 |
+
self.update_progress("系統初始化完成,已就緒")
|
| 70 |
+
logger.info("所有組件初始化完成")
|
| 71 |
+
|
| 72 |
+
except Exception as e:
|
| 73 |
+
error_msg = f"初始化失敗: {str(e)}"
|
| 74 |
+
self.update_progress(error_msg)
|
| 75 |
+
logger.error(error_msg)
|
| 76 |
|
| 77 |
+
# 在背景線程中初始化
|
| 78 |
+
init_thread = threading.Thread(target=init_task, daemon=True)
|
| 79 |
+
init_thread.start()
|
| 80 |
|
| 81 |
def update_progress(self, message: str):
|
| 82 |
"""更新進度信息"""
|
| 83 |
+
timestamp = datetime.now().strftime('%H:%M:%S')
|
| 84 |
+
self.current_progress = f"[{timestamp}] {message}"
|
| 85 |
logger.info(f"進度更新: {message}")
|
| 86 |
|
| 87 |
def get_progress(self) -> str:
|
|
|
|
| 93 |
try:
|
| 94 |
news_data = self.db.get_recent_news(category=category, limit=limit)
|
| 95 |
if not news_data:
|
| 96 |
+
return "📰 暫無新聞資料,請先執行爬蟲任務"
|
| 97 |
|
| 98 |
return format_news_for_display(news_data)
|
| 99 |
|
|
|
|
| 103 |
|
| 104 |
def manual_crawl(self) -> str:
|
| 105 |
"""手動觸發爬蟲"""
|
| 106 |
+
if not self.is_initialized:
|
| 107 |
+
return "⚠️ 系統還在初始化中,請稍後再試"
|
| 108 |
+
|
| 109 |
if self.is_crawling:
|
| 110 |
return "⚠️ 爬蟲正在運行中,請稍後再試"
|
| 111 |
|
|
|
|
| 215 |
def api_manual_crawl():
|
| 216 |
"""手動觸發爬蟲API"""
|
| 217 |
try:
|
| 218 |
+
if not app.is_initialized:
|
| 219 |
+
return jsonify({
|
| 220 |
+
'success': False,
|
| 221 |
+
'message': '系統還在初始化中'
|
| 222 |
+
})
|
| 223 |
+
|
| 224 |
if app.is_crawling:
|
| 225 |
return jsonify({
|
| 226 |
'success': False,
|
|
|
|
| 248 |
"""獲取爬蟲進度API"""
|
| 249 |
return jsonify({
|
| 250 |
'progress': app.get_progress(),
|
| 251 |
+
'is_crawling': app.is_crawling,
|
| 252 |
+
'is_initialized': app.is_initialized
|
| 253 |
})
|
| 254 |
|
| 255 |
+
# 定期更新函數
|
| 256 |
+
def periodic_update():
|
| 257 |
+
"""定期更新進度和新聞"""
|
| 258 |
+
while True:
|
| 259 |
+
time.sleep(3) # 每3秒更新一次
|
| 260 |
+
yield [app.get_progress(), app.get_latest_news("all")]
|
| 261 |
+
|
| 262 |
# 創建 Gradio 介面
|
| 263 |
def create_interface():
|
| 264 |
with gr.Blocks(
|
|
|
|
| 273 |
.positive-badge { background: #28a745; color: white; }
|
| 274 |
.negative-badge { background: #dc3545; color: white; }
|
| 275 |
.neutral-badge { background: #6c757d; color: white; }
|
| 276 |
+
.progress-box { background: #f8f9fa; border: 1px solid #dee2e6; border-radius: 5px; padding: 10px; font-family: monospace; font-size: 14px; }
|
| 277 |
"""
|
| 278 |
) as interface:
|
| 279 |
|
|
|
|
| 303 |
|
| 304 |
# 進度顯示
|
| 305 |
progress_display = gr.Textbox(
|
| 306 |
+
label="📊 系統狀態與進度",
|
| 307 |
value=app.get_progress(),
|
| 308 |
interactive=False,
|
| 309 |
+
elem_classes=["progress-box"],
|
| 310 |
+
lines=2
|
| 311 |
)
|
| 312 |
|
| 313 |
news_display = gr.HTML(label="新聞內容")
|
| 314 |
crawl_result = gr.Textbox(label="爬取結果", visible=False)
|
| 315 |
|
| 316 |
+
# 更新函數
|
| 317 |
+
def update_all():
|
| 318 |
+
"""更新所有信息"""
|
| 319 |
+
return app.get_progress(), app.get_latest_news("all")
|
|
|
|
|
|
|
| 320 |
|
| 321 |
def refresh_news(category):
|
| 322 |
return app.get_latest_news(category)
|
|
|
|
| 327 |
news = app.get_latest_news("all")
|
| 328 |
return result, news
|
| 329 |
|
| 330 |
+
# 使用定時器替代 every 參數
|
| 331 |
+
timer = gr.Timer(value=3) # 每3秒觸發一次
|
| 332 |
+
timer.tick(
|
| 333 |
+
fn=update_all,
|
| 334 |
+
outputs=[progress_display, news_display]
|
| 335 |
)
|
| 336 |
|
| 337 |
# 綁定事件
|
|
|
|
| 344 |
outputs=[crawl_result]
|
| 345 |
)
|
| 346 |
category_radio.change(refresh_news, inputs=[category_radio], outputs=[news_display])
|
| 347 |
+
|
| 348 |
+
# 初始載入
|
| 349 |
+
interface.load(update_all, outputs=[progress_display, news_display])
|
| 350 |
|
| 351 |
with gr.Tab("📊 統計資訊"):
|
| 352 |
stats_display = gr.Markdown()
|
|
|
|
| 402 |
POST /api/crawl
|
| 403 |
```
|
| 404 |
|
| 405 |
+
#### 4. 獲取系統狀態
|
| 406 |
```
|
| 407 |
GET /api/progress
|
| 408 |
```
|
|
|
|
| 414 |
import requests
|
| 415 |
|
| 416 |
# 獲取所有新聞
|
| 417 |
+
response = requests.get('http://localhost:5000/api/news')
|
| 418 |
news_data = response.json()
|
| 419 |
|
| 420 |
# 獲取美股新聞
|
| 421 |
+
response = requests.get('http://localhost:5000/api/news?category=us_stock&limit=10')
|
| 422 |
us_news = response.json()
|
| 423 |
+
|
| 424 |
+
# 檢查系統狀態
|
| 425 |
+
response = requests.get('http://localhost:5000/api/progress')
|
| 426 |
+
status = response.json()
|
| 427 |
```
|
| 428 |
|
| 429 |
**JavaScript:**
|
|
|
|
| 454 |
- **模型**: `uer/roberta-base-finetuned-jd-binary-chinese`
|
| 455 |
- **分類**: 正面 (綠色) / 負面 (紅色) / 中性 (灰色)
|
| 456 |
- **準確性**: 針對中文金融新聞優化
|
| 457 |
+
- **後台載入**: 模型在背景載入,不阻塞介面啟動
|
| 458 |
|
| 459 |
### 🕷️ 新聞爬蟲
|
| 460 |
- **來源**: 鉅亨網 (cnyes.com)
|
|
|
|
| 476 |
|
| 477 |
---
|
| 478 |
|
| 479 |
+
💡 **使用提示**:
|
| 480 |
+
- 首次啟動會在背景下載模型,請耐心等待
|
| 481 |
+
- 系統狀態會顯示初始化進度
|
| 482 |
- 建議使用SSD硬碟提升SQLite性能
|
| 483 |
- API接口可用於整合其他應用系統
|
| 484 |
""")
|
|
|
|
| 496 |
flask_thread = threading.Thread(target=run_flask, daemon=True)
|
| 497 |
flask_thread.start()
|
| 498 |
|
| 499 |
+
print("🚀 啟動股市新聞情緒分析器...")
|
| 500 |
+
print("📊 網頁介面: http://localhost:7860")
|
| 501 |
+
print("🔌 API接口: http://localhost:5000")
|
| 502 |
+
print("💡 首次啟動需要下載模型,請耐心等待...")
|
| 503 |
+
|
| 504 |
# 啟動Gradio介面
|
| 505 |
interface = create_interface()
|
| 506 |
interface.launch(
|