Updated the dataloader to ignore case on product features. Note have deliberately not harmonised product feature capitalisation or similar features as this "messiness" is representative of real organisational data.

src/data_synthesis/data_loader.py

@@ -17,6 +17,9 @@ def db_file() -> str:
 
 
 def setup_db_tables() -> None:
+    """
+    Drop all the tables in the database and then re-build the structure empty
+    """
     con = sqlite3.connect(db_file())
     tables = ['reviews', 'product_features', 'features', 'products', 'categories']
     for t in tables:
@@ -42,11 +45,16 @@ def setup_db_tables() -> None:
     sql = "CREATE TABLE reviews (id INTEGER PRIMARY KEY AUTOINCREMENT, product_id INTEGER NOT NULL, rating INTEGER NOT NULL, review_text TEXT NOT NULL, FOREIGN KEY (product_id) REFERENCES products (id))"
     con.execute(sql)
 
+
 def insert_data() -> None:
+    """
+    Insert the data from the json data files into the data structure
+    """
     con = sqlite3.connect(db_file())
     cur = con.cursor()
 
     cats_and_features = get_categories_and_features()
+
     for cat, features in cats_and_features.items():
         sql = f"INSERT INTO categories('name') VALUES ('{cat}')"
         cat_id = con.execute(sql).lastrowid
@@ -64,7 +72,7 @@ def insert_data() -> None:
             con.commit()
 
             for feat in prod.features:
-                sql = f"SELECT id from features WHERE name='{feat}' AND category_id={cat_id}"
+                sql = f"SELECT id from features WHERE lower(name)='{feat.lower()}' AND category_id={cat_id}"
                 cur.execute(sql)
                 rows = cur.fetchall()
                 if len(rows) == 0: