Update fix langlinks.py

fix langlinks.py

@@ -13,15 +13,25 @@ json_url = "https://huggingface.co/Ibrahemqasim/enwiki_to_arwiki_categories/reso
 response = requests.get(json_url)
 data = response.json()
 
+# تحميل الملف JSON من الرابط مباشرة
+json_url2 = "https://huggingface.co/Ibrahemqasim/enwiki_to_arwiki_categories/resolve/main/countries.json"
+response2 = requests.get(json_url2)
+countries = response2.json()
+
 # تحويل القاموس إلى قائمة من القواميس [{ "en": "value", "ar": "value" }, ...]
 
 data_lists = {
     "langlinks" : {},
     "filtered_data" : {},
+    "cats_2000_contry" : {},
     "cats_2000" : {},
 }
 
-for key, value in tqdm.tqdm(data.items()):
+for tab in tqdm.tqdm(data):
+    # ---
+    key = tab["en"]
+    value = tab["ar"]
+    # ---
     # "Category:1. FC Köln non-playing staff"
     # remove " from start and end
     # ---
@@ -49,6 +59,13 @@ for key, value in tqdm.tqdm(data.items()):
         # ---
         # data_lists["cats_2000"].append({"en": key2, "ar": value2})
         data_lists["cats_2000"][key] = value
+        # ----
+        for en_c, ar_c in countries.items():
+            if en_c in key2 and ar_c in value2:
+                key3 = key2.replace(en_c, "country")
+                value3 = value2.replace(ar_c, "country")
+                # ---
+                data_lists["cats_2000_contry"][key3] = value3
 
 
 for x, data_list in data_lists.items():
@@ -65,8 +82,7 @@ for x, data_list in data_lists.items():
         # repo_type="dataset",  # نوع المستودع (نستخدم dataset للملفات)
     )
 
-    print(f"__________________")
+    print(f"______________")
     print(f"file: {x} uploaded successfully!")
-
     print(f"{len(data)=}.")
     print(f"{len(data_list)} rows uploaded.")