Update nat_datasets.py

nat_datasets.py

@@ -65,7 +65,7 @@ def new_func(value, ar_tab):
         country2 = f"ال{country}".replace(" ", " ال")
         # ---
         if country2 in value:
-            return country2, tag
+            return country2, tag.replace("}", "_AL}")
         elif country in value:
             return country, tag
 
@@ -109,54 +109,54 @@ for tab in tqdm.tqdm(data):
     value1 = re.sub(rf'\b{re.escape(ar_country)}\b', NAT_PATTERN, f" {value} ", re.IGNORECASE)
     value1 = value1.strip()
     # ---
-    # if EN_NAT_PATTERN in key1 and NAT_PATTERN in value1:
-    # ---
-    if key1 in data_lists["categories_with_NAT_pattern"]:
-        data_lists["categories_with_NAT_pattern"][key1]["count"] += 1
-    else:
-        data_lists["categories_with_NAT_pattern"][key1] = {"ar": value1, "count": 1}
-    # ---
-    # continue
-    # ---
-    # Add if key and value has 4 digits and they are the same
-    reg_year = r"(\d+[–-]\d+|\d{4})"
-    # ---
-    key_digits = re.search(reg_year, key, re.IGNORECASE)
-    value_digits = re.search(reg_year, value1, re.IGNORECASE)
-    # ----
-    if key_digits and value_digits and key_digits.group() == value_digits.group():
-        # if key1 in data_lists["categories_with_years"]:
-        #     data_lists["categories_with_years"][key1]["count"] += 1
-        # else:
-        #     data_lists["categories_with_years"][key1] = {"ar": value1, "count": 1}
+    if EN_NAT_PATTERN in key1 and NAT_PATTERN in value1:
         # ---
-        key2 = key1.replace(key_digits.group(), YEAR_PATTERN)
-        value2 = value1.replace(value_digits.group(), YEAR_PATTERN)
-        # ---
-        if key2 in data_lists["categories_with_YEAR_NAT_pattern"]:
-            data_lists["categories_with_YEAR_NAT_pattern"][key2]["count"] += 1
+        if key1 in data_lists["categories_with_NAT_pattern"]:
+            data_lists["categories_with_NAT_pattern"][key1]["count"] += 1
         else:
-            data_lists["categories_with_YEAR_NAT_pattern"][key2] = {"ar": value2, "count": 1}
-        # ----
-        continue
-        # ----
-        # البحث عن اسم الدولة في key2
-        match = re.search(countries_pattern, key2, re.IGNORECASE)
+            data_lists["categories_with_NAT_pattern"][key1] = {"ar": value1, "count": 1}
+        # ---
+        # continue
+        # ---
+        # Add if key and value has 4 digits and they are the same
+        reg_year = r"(\d+[–-]\d+|\d{4})"
+        # ---
+        key_digits = re.search(reg_year, key, re.IGNORECASE)
+        value_digits = re.search(reg_year, value1, re.IGNORECASE)
         # ----
-        if match:
-            en_country = match.group(1)
-            ar_country = countries.get(en_country)
+        if key_digits and value_digits and key_digits.group() == value_digits.group():
+            # if key1 in data_lists["categories_with_years"]:
+            #     data_lists["categories_with_years"][key1]["count"] += 1
+            # else:
+            #     data_lists["categories_with_years"][key1] = {"ar": value1, "count": 1}
+            # ---
+            key2 = key1.replace(key_digits.group(), YEAR_PATTERN)
+            value2 = value1.replace(value_digits.group(), YEAR_PATTERN)
             # ---
-            if ar_country and ar_country in value2:
-                key3 = re.sub(rf'\b{re.escape(en_country)}\b', COUNTRY_PATTERN, key2, re.IGNORECASE)
-                value3 = re.sub(rf'\b{re.escape(ar_country)}\b', COUNTRY_PATTERN, value2, re.IGNORECASE)
+            if key2 in data_lists["categories_with_YEAR_NAT_pattern"]:
+                data_lists["categories_with_YEAR_NAT_pattern"][key2]["count"] += 1
+            else:
+                data_lists["categories_with_YEAR_NAT_pattern"][key2] = {"ar": value2, "count": 1}
+            # ----
+            continue
+            # ----
+            # البحث عن اسم الدولة في key2
+            match = re.search(countries_pattern, key2, re.IGNORECASE)
+            # ----
+            if match:
+                en_country = match.group(1)
+                ar_country = countries.get(en_country)
                 # ---
-                if COUNTRY_PATTERN in key3 and COUNTRY_PATTERN in value3:
+                if ar_country and ar_country in value2:
+                    key3 = re.sub(rf'\b{re.escape(en_country)}\b', COUNTRY_PATTERN, key2, re.IGNORECASE)
+                    value3 = re.sub(rf'\b{re.escape(ar_country)}\b', COUNTRY_PATTERN, value2, re.IGNORECASE)
                     # ---
-                    if key3 in data_lists["categories_with_YEAR_COUNTRY_pattern"]:
-                        data_lists["categories_with_YEAR_COUNTRY_pattern"][key3]["count"] += 1
-                    else:
-                        data_lists["categories_with_YEAR_COUNTRY_pattern"][key3] = {"ar": value3, "count": 1}
+                    if COUNTRY_PATTERN in key3 and COUNTRY_PATTERN in value3:
+                        # ---
+                        if key3 in data_lists["categories_with_YEAR_COUNTRY_pattern"]:
+                            data_lists["categories_with_YEAR_COUNTRY_pattern"][key3]["count"] += 1
+                        else:
+                            data_lists["categories_with_YEAR_COUNTRY_pattern"][key3] = {"ar": value3, "count": 1}
 # ----
 print(f"{match1_done=}")