Get categories content and check alphabetical order

scripts/validate/format.py

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 
 import re
+from typing import List, Tuple, Dict
 
 
 anchor = '###'
@@ -24,7 +25,53 @@ anchor_re = re.compile(anchor + '\s(.+)')
 section_title_re = re.compile('\*\s\[(.*)\]')
 link_re = re.compile('\[(.+)\]\((http.*)\)')
 
+# Type aliases
+APIList = List[str]
+Categories = Dict[str, APIList]
+CategoriesLineNumber = Dict[str, int]
+
 
 def error_message(line_number: int, message: str) -> str:
     line = line_number + 1
     return f'(L{line:03d}) {message}'
+
+
+def get_categories_content(contents: List[str]) -> Tuple[Categories, CategoriesLineNumber]:
+
+    categories = {}
+    category_line_num = {}
+
+    for line_num, line_content in enumerate(contents):
+
+        if line_content.startswith(anchor):
+            category = line_content.split(anchor)[1].strip()
+            categories[category] = []
+            category_line_num[category] = line_num
+            continue
+
+        if not line_content.startswith('|') or line_content.startswith('|---'):
+            continue
+
+        raw_title = [
+            raw_content.strip() for raw_content in line_content.split('|')[1:-1]
+        ][0]
+
+        title_match = link_re.match(raw_title)
+        if title_match:
+                title = title_match.group(1).upper()
+                categories[category].append(title)
+
+    return (categories, category_line_num)
+
+
+def check_alphabetical_order(lines: List[str]) -> None:
+
+    categories, category_line_num = get_categories_content(contents=lines)
+
+    for category, api_list in categories.items():
+        if sorted(api_list) != api_list:
+            message = error_message(
+                category_line_num[category], 
+                f'{category} category is not alphabetical order'
+            )
+            errors.append(message)