Update utility.py (#3)

- Update utility.py (f02a873dd316beda860aaedc9efaf823adaebf19)


Co-authored-by: Raphael M <[email protected]>

utility.py

@@ -4,6 +4,15 @@ from datetime import datetime
 import openai
 from google.cloud import firestore
 from dotenv import load_dotenv
+from pandasai import SmartDatalake
+from pandasai import Agent
+from pandasai.responses.response_parser import  ResponseParser
+import pandas as pd
+from pandasai.llm import OpenAI
+#from langchain.llms.sambanova import Sambaverse
+#from langchain_community.llms.sambanova import Sambaverse
+from langchain_community.chat_models.sambanova import ChatSambaNovaCloud
+
 
 # Make API connection
 load_dotenv()
@@ -19,6 +28,49 @@ client = openai.OpenAI(
     base_url="https://api.sambanova.ai/v1",
 )
 
+
+
+
+sambaverse_api_key = os.environ.get("SAMBAVERSE_API_KEY")
+
+
+llm = ChatSambaNovaCloud(
+    model="Meta-Llama-3.1-70B-Instruct",
+    max_tokens=1024,
+    temperature=0.7,
+    top_k=1,
+    top_p=0.01,
+)
+
+
+print(llm.invoke("Why should I use open source models?"))
+
+
+
+class FlaskResponse(ResponseParser):
+  def __init__(self, context) -> None:
+      super().__init__(context)
+
+  def format_dataframe(self, result):
+      return result['value'].to_html()
+
+  def format_plot(self, result):
+        # Save the plot using savefig
+      try:
+          
+          img_path = result['value']
+      
+        
+      except ValueError:
+          img_path = str(result['value'])
+          print("value error!", img_path)
+        
+      print("response_class_path:", img_path)
+      return img_path
+
+  def format_other(self, result):
+      return str(result['value'])
+      
 # Generate AI response from user input
 def generateResponse(prompt,model='Meta-Llama-3.1-70B-Instruct'):
   #----- Call API to classify and extract relevant transaction information
@@ -218,6 +270,23 @@ def parse_multiple_transactions(response_text):
 
     return transactions
 
+def read_datalake(user_phone, user_question):
+    inventory_ref = db.collection("users").document(user_phone).collection("inventory")
+
+    sales_ref = db.collection("users").document(user_phone).collection('sales')
+
+    inventory_list = [doc.to_dict() for doc in inventory_ref.stream()]
+    sales_list = [doc.to_dict() for doc in sales_ref.stream()]
+
+    inventory_df = pd.DataFrame(inventory_list)
+    sales_df = pd.DataFrame(sales_list)
+    
+
+    lake = SmartDatalake([inventory_df, sales_df], config={"llm": llm, "response_parser": FlaskResponse, "enable_cache": False, "save_logs": False})
+    response = lake.chat(user_question)
+
+    return response
+
 def create_inventory(user_phone, transaction_data):
     for transaction in transaction_data:
         item_name = transaction['details']['item'] # assumes unique item name
@@ -289,6 +358,8 @@ def delete_transaction(user_phone, transaction_id):
     print("Transaction deleted successfully!")
 
 
+
+
 # Example usage
 # response_text = """
 # The information provided indicates that you want to **create/record** a new transaction.