Create sql_generator.py

sql_generator.py

@@ -0,0 +1,60 @@
+import spacy
+
+# Load the spaCy English model
+nlp = spacy.load("en_core_web_sm")
+
+# Define the mapping of keywords to SQL column names based on the new schema
+column_mapping = {
+    "handle": "Handle",
+    "title": "Title",
+    "body": "Body (HTML)",
+    "vendor": "Vendor",
+    "type": "Type",
+    "tags": "Tags",
+    "published": "Published",
+    "option1_name": "Option1 Name",
+    "option1_value": "Option1 Value",
+    "option2_name": "Option2 Name",
+    "option2_value": "Option2 Value",
+    "option3_name": "Option3 Name",
+    "option3_value": "Option3 Value",
+    "variant_sku": "Variant SKU",
+    "variant_price": "Variant Price",
+    "variant_inventory_qty": "Variant Inventory Qty",
+    # Add more mappings as needed
+}
+
+# Function to generate SQL query based on natural language input
+def generate_sql_query(natural_language_input):
+    doc = nlp(natural_language_input)
+
+    # Initialize SQL query parts
+    sql_query = "SELECT * FROM products WHERE "
+    conditions = []
+
+    # Extract keywords and values for building the query
+    for token in doc:
+        if token.lemma_ in column_mapping:
+            column_name = column_mapping[token.lemma_]
+            # Look for the word 'is' and the value after it
+            if token.nbor(1).text.lower() == "is":
+                value_token = token.nbor(2)  # Get the token after 'is'
+                # Check if the value token is valid (e.g., if it's a string)
+                if value_token and value_token.text not in ["and", "or"]:
+                    # Format the value correctly for SQL
+                    conditions.append(f"{column_name} = '{value_token.text}'")
+
+    # Join conditions with AND
+    if conditions:
+        sql_query += " AND ".join(conditions)
+    else:
+        sql_query = sql_query[:-7]  # Remove the last " WHERE " if no conditions were added
+    
+    return sql_query
+
+# Example natural language input
+natural_language_input = "Find products where title is laptop and vendor is apple and published is active"
+
+# Generate SQL query
+sql_query = generate_sql_query(natural_language_input)
+print(sql_query)