Vincentran commited on
Commit
aee0892
·
1 Parent(s): d3157ae

Upload E-Commerce Product Intelligence Dashboard (frontend + backend)

Browse files
Files changed (1) hide show
  1. backend/scraper.py +7 -12
backend/scraper.py CHANGED
@@ -1,21 +1,21 @@
1
  # backend/scraper.py
2
 
3
  import os
4
- import json
5
- from kaggle.api.kaggle_api_extended import KaggleApi # ❌ XÃI import này
6
 
7
 
8
  def setup_kaggle_api():
9
- """Setup Kaggle API."""
10
- # Check if kaggle.json exists
 
 
 
11
  kaggle_json_path = os.path.join(os.environ.get('HOME', '/root'), '.config/kaggle/kaggle.json')
12
 
13
  if not os.path.exists(kaggle_json_path):
14
  print(f"⚠️ kaggle.json not found at {kaggle_json_path}")
15
- print("⚠️ Kaggle API will not work. Upload kaggle.json to HF Space secrets.")
16
  return None
17
 
18
- # Setup kaggle
19
  with open(kaggle_json_path) as f:
20
  kaggle_credentials = json.load(f)
21
 
@@ -33,7 +33,6 @@ def download_dataset():
33
  print("❌ Kaggle API not available. Skipping download.")
34
  return False
35
 
36
- # Download dataset
37
  dataset_name = "jackdaug/ecommerce-products-dataset"
38
  os.makedirs("data", exist_ok=True)
39
 
@@ -56,11 +55,7 @@ def run_scraper():
56
  return False
57
 
58
 
59
- # ✅ XÓA HOÀN TOÀN:
60
- # if __name__ == "__main__":
61
- # run_scraper()
62
-
63
- # ✅ Chỉ để:
64
  if __name__ == "__main__":
65
  print("📦 scraper.py imported (not running automatically)")
66
  print("📦 Call run_scraper() manually or via API")
 
1
  # backend/scraper.py
2
 
3
  import os
 
 
4
 
5
 
6
  def setup_kaggle_api():
7
+ """Setup Kaggle API (lazy import)."""
8
+ # Import kaggle CHỈ khi gọi hàm
9
+ import json
10
+ from kaggle.api.kaggle_api_extended import KaggleApi
11
+
12
  kaggle_json_path = os.path.join(os.environ.get('HOME', '/root'), '.config/kaggle/kaggle.json')
13
 
14
  if not os.path.exists(kaggle_json_path):
15
  print(f"⚠️ kaggle.json not found at {kaggle_json_path}")
16
+ print("⚠️ Upload kaggle.json to HF Space or set KAGGLE_USERNAME/KAGGLE_KEY env vars.")
17
  return None
18
 
 
19
  with open(kaggle_json_path) as f:
20
  kaggle_credentials = json.load(f)
21
 
 
33
  print("❌ Kaggle API not available. Skipping download.")
34
  return False
35
 
 
36
  dataset_name = "jackdaug/ecommerce-products-dataset"
37
  os.makedirs("data", exist_ok=True)
38
 
 
55
  return False
56
 
57
 
58
+ # ✅ KHÔNG chạy scraper khi khởi động
 
 
 
 
59
  if __name__ == "__main__":
60
  print("📦 scraper.py imported (not running automatically)")
61
  print("📦 Call run_scraper() manually or via API")