Spaces:
Sleeping
Sleeping
Delete check_dupes.py with huggingface_hub
Browse files- check_dupes.py +0 -24
check_dupes.py
DELETED
|
@@ -1,24 +0,0 @@
|
|
| 1 |
-
import json
|
| 2 |
-
|
| 3 |
-
with open(r'c:\react_projects\VortexCommerce\kaggle\extra_products_vortex.json', 'r', encoding='utf-8') as f:
|
| 4 |
-
data = json.load(f)
|
| 5 |
-
if isinstance(data, dict):
|
| 6 |
-
products = data.get('products', [])
|
| 7 |
-
else:
|
| 8 |
-
products = data
|
| 9 |
-
|
| 10 |
-
seen_ar = set()
|
| 11 |
-
seen_en = set()
|
| 12 |
-
dupes = 0
|
| 13 |
-
for p in products:
|
| 14 |
-
ar = p.get('name_ar', '').strip()
|
| 15 |
-
en = p.get('name_en', '').strip()
|
| 16 |
-
if ar in seen_ar or en in seen_en:
|
| 17 |
-
dupes += 1
|
| 18 |
-
else:
|
| 19 |
-
seen_ar.add(ar)
|
| 20 |
-
seen_en.add(en)
|
| 21 |
-
|
| 22 |
-
print(f"Total Products in JSON: {len(products)}")
|
| 23 |
-
print(f"Duplicates found: {dupes}")
|
| 24 |
-
print(f"Expected unique: {len(products) - dupes}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|