add parse function

src/hardcoded_data.py

@@ -45,7 +45,6 @@ articles = [
         "category": "planetary science",
         "articles": planetary_science_articles
     },
-
 ]
 
 [
@@ -56,3 +55,35 @@ articles = [
         "tldr",
     },
 ]
+
+# [category1[articles[article1['title1', 'abstract1'], article2['title2', 'abstract2']]], category2[...], ...]
+
+def parse_list_of_lists(input_list, category_names):
+    articles_list = []
+    
+    for idx, item in enumerate(input_list):
+        category_name = category_names[idx]
+        category_articles = []
+        
+        for sub_item in item:
+            if isinstance(sub_item, list) and len(sub_item) == 2:
+                article_title, article_abstract = sub_item
+                category_articles.append({"title": article_title, "abstract": article_abstract})
+            else:
+                raise ValueError("Each sub-item should be a list containing [title, abstract].")
+        
+        articles_list.append({"category": category_name, "articles": category_articles})
+    
+    return articles_list
+
+
+input_list = [
+    [["article1 title", "article1 abstract"], ["article2 title", "article2 abstract"]],
+    [["article3 title", "article3 abstract"], ["article4 title", "article4 abstract"]]
+]
+
+category_names = ["machine learning", "planetary science"]
+
+output_list_of_dicts = parse_list_of_lists(input_list, category_names)
+print(output_list_of_dicts)
+