Update nat_datasets.py

nat_datasets.py

@@ -106,14 +106,15 @@ for tab in tqdm.tqdm(data):
     key1 = re.sub(rf'\b{re.escape(en_country)}\b', EN_NAT_PATTERN, f" {key} ", re.IGNORECASE)
     key1 = key1.strip()
     # ---
+    if EN_NAT_PATTERN in key1 and key1 in data_lists["categories_with_NAT_pattern"]:
+        data_lists["categories_with_NAT_pattern"][key1]["count"] += 1
+    # ---
     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:
+        if key1 not in data_lists["categories_with_NAT_pattern"]:
             data_lists["categories_with_NAT_pattern"][key1] = {"ar": value1, "count": 1}
         # ---
         # continue
@@ -124,18 +125,22 @@ for tab in tqdm.tqdm(data):
         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}
-            # ---
-            key2 = key1.replace(key_digits.group(), YEAR_PATTERN)
-            value2 = value1.replace(value_digits.group(), YEAR_PATTERN)
+        if not key_digits:
+            continue
+        # ----
+        key2 = key1.replace(key_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 not value_digits:
+            continue
+        # ----
+        value2 = value1.replace(value_digits.group(), YEAR_PATTERN)
+        # ----
+        if key_digits.group() == value_digits.group():
             # ---
-            if key2 in data_lists["categories_with_YEAR_NAT_pattern"]:
-                data_lists["categories_with_YEAR_NAT_pattern"][key2]["count"] += 1
-            else:
+            if key2 not in data_lists["categories_with_YEAR_NAT_pattern"]:
                 data_lists["categories_with_YEAR_NAT_pattern"][key2] = {"ar": value2, "count": 1}
             # ----
             continue