Update app.py

app.py

@@ -10,6 +10,7 @@ from os import getenv
 import requests
 from jinja2 import Environment, FileSystemLoader
 from backend.semantic_search import table, retriever
+
 # Bhashini API translation function
 api_key = getenv('API_KEY', '').strip()
 user_id = getenv('USER_ID', '').strip()
@@ -107,7 +108,7 @@ agent = Agent(
         "Use simple language and examples relevant to customs officers.",
         "Focus on topics like transhipment, AEO schemes, bonds, penalties, and CFS approvals.",
         "Structure responses with headings and bullet points when helpful.",
-        "If you don’t know the answer, say 'I don’t have enough information to answer that.'"
+        "If you don't know the answer, say 'I don't have enough information to answer that.'"
     ],
     model=Groq(id="llama3-70b-8192", api_key=api_key),
     markdown=True
@@ -196,7 +197,6 @@ def translate_text(selected_language, history):
 
 # Gradio interface
 with gr.Blocks(theme='gradio/soft') as CHATBOT:
-    history_state = gr.State([])
     with gr.Row():
         with gr.Column(scale=10):
             gr.HTML(value="""<div style="color: #FF4500;"><h1>ADWITIYA-</h1> <h1><span style="color: #008000">Custom Manual Chatbot </span></h1></div>""")
@@ -239,20 +239,22 @@ with gr.Blocks(theme='gradio/soft') as CHATBOT:
     prompt_html = gr.HTML()
     translated_textbox = gr.Textbox(label="Translated Response")
 
-    def update_chat_and_translate(txt, cross_encoder, history_state, language_dropdown):
-        history = history_state.value if history_state.value else []
-        history.append((txt, ""))
+    def update_chat_and_translate(txt, cross_encoder, history, language_dropdown):
+        # Fixed: history is now directly used instead of history_state.value
+        if not history:
+            history = []
         
         # Call simple_chat_function
-        msg, history, prompt_html_content = simple_chat_function(txt, history, cross_encoder)
+        msg, updated_history, prompt_html_content = simple_chat_function(txt, history, cross_encoder)
         
         # Translate text
-        translated_text = translate_text(language_dropdown, history)
+        translated_text = translate_text(language_dropdown, updated_history)
         
-        return history, prompt_html_content, translated_text
+        return updated_history, prompt_html_content, translated_text
 
-    txt_msg = txt_btn.click(update_chat_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
-    txt_msg = txt.submit(update_chat_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
+    # Fixed: Pass chatbot directly instead of history_state
+    txt_msg = txt_btn.click(update_chat_and_translate, [txt, cross_encoder, chatbot, language_dropdown], [chatbot, prompt_html, translated_textbox])
+    txt_msg = txt.submit(update_chat_and_translate, [txt, cross_encoder, chatbot, language_dropdown], [chatbot, prompt_html, translated_textbox])
 
     examples = [
         'My transhipment cargo is missing',
@@ -267,35 +269,29 @@ with gr.Blocks(theme='gradio/soft') as CHATBOT:
     gr.Examples(examples, txt)
 
 # Launch the Gradio application
-CHATBOT.launch(share=True, debug=True)# import requests
-# import gradio as gr
-# from ragatouille import RAGPretrainedModel
+CHATBOT.launch(share=True, debug=True)# import gradio as gr
+# from phi.agent import Agent
+# from phi.model.groq import Groq
 # import logging
 # from pathlib import Path
 # from time import perf_counter
 # from sentence_transformers import CrossEncoder
-# from huggingface_hub import InferenceClient
-# from jinja2 import Environment, FileSystemLoader
 # import numpy as np
 # from os import getenv
-# from backend.query_llm import generate_hf, generate_qwen
+# import requests
+# from jinja2 import Environment, FileSystemLoader
 # from backend.semantic_search import table, retriever
-# from huggingface_hub import InferenceClient
-
-
 # # Bhashini API translation function
-# api_key = getenv('API_KEY')
-# user_id = getenv('USER_ID')
+# api_key = getenv('API_KEY', '').strip()
+# user_id = getenv('USER_ID', '').strip()
 
 # def bhashini_translate(text: str, from_code: str = "en", to_code: str = "hi") -> dict:
 #     """Translates text from source language to target language using the Bhashini API."""
-    
 #     if not text.strip():
 #         print('Input text is empty. Please provide valid text for translation.')
 #         return {"status_code": 400, "message": "Input text is empty", "translated_content": None, "speech_content": None}
 #     else:
-#         print('Input text - ',text)
-#     print(f'Starting translation process from {from_code} to {to_code}...')
+#         print('Input text - ', text)
 #     print(f'Starting translation process from {from_code} to {to_code}...')
 #     gr.Warning(f'Translating to {to_code}...')
     
@@ -305,6 +301,11 @@ CHATBOT.launch(share=True, debug=True)# import requests
 #         "userID": user_id,
 #         "ulcaApiKey": api_key
 #     }
+#     for key, value in headers.items():
+#         if not isinstance(value, str) or '\n' in value or '\r' in value:
+#             print(f"Invalid header value for {key}: {value}")
+#             return {"status_code": 400, "message": f"Invalid header value for {key}", "translated_content": None}
+
 #     payload = {
 #         "pipelineTasks": [{"taskType": "translation", "config": {"language": {"sourceLanguage": from_code, "targetLanguage": to_code}}}],
 #         "pipelineRequestConfig": {"pipelineId": "64392f96daac500b55c543cd"}
@@ -347,7 +348,6 @@ CHATBOT.launch(share=True, debug=True)# import requests
 #     print(f'Translation successful. Translated content: "{translated_content}"')
 #     return {"status_code": 200, "message": "Translation successful", "translated_content": translated_content}
 
-
 # # Existing chatbot functions
 # VECTOR_COLUMN_NAME = "vector"
 # TEXT_COLUMN_NAME = "text"
@@ -356,136 +356,114 @@ CHATBOT.launch(share=True, debug=True)# import requests
 
 # logging.basicConfig(level=logging.INFO)
 # logger = logging.getLogger(__name__)
-# client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1", token=HF_TOKEN)
-# proj_dir = Path(__file__).parent
-# env = Environment(loader=FileSystemLoader(proj_dir / 'templates'))
 
+# # Set up Jinja2 environment
+# env = Environment(loader=FileSystemLoader(proj_dir / 'templates'))
 # template = env.get_template('template.j2')
 # template_html = env.get_template('template_html.j2')
 
-# # def add_text(history, text):
-# #     history = [] if history is None else history
-# #     history = history + [(text, None)]
-# #     return history, gr.Textbox(value="", interactive=False)
-
-# def bot(history, cross_encoder):
-
+# # Initialize Grok Agent
+# api_key = getenv("GROQ_API_KEY")
+# if not api_key:
+#     gr.Warning("GROQ_API_KEY not found. Set it in 'Repository secrets'.")
+#     logger.error("GROQ_API_KEY not found.")
+#     api_key = ""  # Fallback, but will fail without a key
+
+# agent = Agent(
+#     name="Customs Assistant",
+#     role="You are a helpful assistant for CBIC officers, providing guidance on customs procedures and regulations.",
+#     instructions=[
+#         "You are an expert in customs regulations and CBIC procedures.",
+#         "Provide clear, accurate, and professional explanations.",
+#         "Use simple language and examples relevant to customs officers.",
+#         "Focus on topics like transhipment, AEO schemes, bonds, penalties, and CFS approvals.",
+#         "Structure responses with headings and bullet points when helpful.",
+#         "If you don’t know the answer, say 'I don’t have enough information to answer that.'"
+#     ],
+#     model=Groq(id="llama3-70b-8192", api_key=api_key),
+#     markdown=True
+# )
+
+# def simple_chat_function(message, history, cross_encoder_choice):
+#     """Chat function with semantic search and Grok agent integration"""
+#     if not message.strip():
+#         return "", history, ""
+    
 #     top_rerank = 25
 #     top_k_rank = 20
-#     query = history[-1][0] if history else ''
-#     print('\nQuery: ',query )
-#     print('\nHistory:',history)
-#     if not query:
-#         gr.Warning("Please submit a non-empty string as a prompt")
-#         raise ValueError("Empty string was submitted")
-
-#     logger.warning('Retrieving documents...')
-    
-#     if cross_encoder == '(HIGH ACCURATE) ColBERT':
-#         gr.Warning('Retrieving using ColBERT.. First time query will take a minute for model to load..pls wait')
-#         RAG = RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
-#         RAG_db = RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
-#         documents_full = RAG_db.search(query, k=top_k_rank)
-        
-#         documents = [item['content'] for item in documents_full]
-#         prompt = template.render(documents=documents, query=query)
-#         prompt_html = template_html.render(documents=documents, query=query)
-    
-#         generate_fn = generate_hf
-    
-#         history[-1][1] = ""
-#         for character in generate_fn(prompt, history[:-1]):
-#             history[-1][1] = character
-#             yield history, prompt_html
-#     else:
-#         document_start = perf_counter()
-    
-#         query_vec = retriever.encode(query)
-#         doc1 = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_k_rank)
     
+#     try:
+#         start_time = perf_counter()
+        
+#         # Encode query and search documents
+#         query_vec = retriever.encode(message)
 #         documents = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_rerank).to_list()
 #         documents = [doc[TEXT_COLUMN_NAME] for doc in documents]
-    
-#         query_doc_pair = [[query, doc] for doc in documents]
-#         if cross_encoder == '(FAST) MiniLM-L6v2':
-#             cross_encoder1 = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
-#         elif cross_encoder == '(ACCURATE) BGE reranker':
-#             cross_encoder1 = CrossEncoder('BAAI/bge-reranker-base')
         
-#         cross_scores = cross_encoder1.predict(query_doc_pair)
-#         sim_scores_argsort = list(reversed(np.argsort(cross_scores)))
+#         # Re-rank documents using cross-encoder
+#         if cross_encoder_choice == '(FAST) MiniLM-L6v2':
+#             cross_encoder_model = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
+#         elif cross_encoder_choice == '(ACCURATE) BGE reranker':
+#             cross_encoder_model = CrossEncoder('BAAI/bge-reranker-base')
+#         elif cross_encoder_choice == '(HIGH ACCURATE) ColBERT':
+#             gr.Warning('Retrieving using ColBERT.. First time query may take a minute for model to load..pls wait')
+#             from ragatouille import RAGPretrainedModel
+#             RAG = RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
+#             RAG_db = RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
+#             documents = [item['content'] for item in RAG_db.search(message, k=top_k_rank)]
+#             cross_encoder_model = None  # No re-ranking needed for ColBERT
+
+#         if cross_encoder_model:
+#             query_doc_pair = [[message, doc] for doc in documents]
+#             cross_scores = cross_encoder_model.predict(query_doc_pair)
+#             sim_scores_argsort = list(reversed(np.argsort(cross_scores)))
+#             documents = [documents[idx] for idx in sim_scores_argsort[:top_k_rank]]
         
-#         documents = [documents[idx] for idx in sim_scores_argsort[:top_k_rank]]
-    
-#         document_time = perf_counter() - document_start
-    
-#         prompt = template.render(documents=documents, query=query)
-#         prompt_html = template_html.render(documents=documents, query=query)
-    
-#         #generate_fn = generate_hf
-#         generate_fn=generate_qwen
-#         # Create a new history entry instead of modifying the tuple directly
-#         new_history = history[:-1] + [ (prompt, "") ] # query replaced prompt
-#         output=''
-#         # for character in generate_fn(prompt, history[:-1]):
-#         #     #new_history[-1] = (query, character) 
-#         #     output+=character
-#         output=generate_fn(prompt, history[:-1])
+#         # Create context from top documents
+#         context = "\n\n".join(documents[:10]) if documents else ""
+#         context = f"Context information from customs materials:\n{context}\n\n"
         
-#         print('Output:',output)
-#         new_history[-1] = (prompt, output) #query replaced with prompt
-#         print('New History',new_history)
-#         #print('prompt html',prompt_html)# Update the last tuple with new text
+#         # Add conversation history for context
+#         history_context = ""
+#         if history and len(history) > 0:
+#             for user_msg, bot_msg in history[-2:]:  # Last 2 exchanges
+#                 if user_msg and bot_msg:
+#                     history_context += f"Previous Q: {user_msg}\nPrevious A: {bot_msg}\n"
         
-#         history_list = list(history[-1])
-#         history_list[1] = output  # Assuming `character` is what you want to assign
-#         # Update the history with the modified list converted back to a tuple
-#         history[-1] = tuple(history_list)
-
-#             #history[-1][1] = character
-#         # yield new_history, prompt_html
-#         yield history, prompt_html
-#          # new_history,prompt_html
-#         # history[-1][1] = ""
-#         # for character in generate_fn(prompt, history[:-1]):
-#         #     history[-1][1] = character
-#         #     yield history, prompt_html
-
-# #def translate_text(response_text, selected_language):
-    
-# def translate_text(selected_language,history):
+#         # Create full prompt
+#         full_prompt = f"{history_context}{context}Question: {message}\n\nPlease answer the question using the context provided above. If the context doesn't contain relevant information, use your general knowledge about CBIC customs procedures."
+        
+#         # Generate response
+#         response = agent.run(full_prompt)
+#         response_text = response.content if hasattr(response, 'content') else str(response)
+        
+#         # Add to history
+#         history.append([message, response_text])
+        
+#         # Render template with documents and query
+#         prompt_html = template_html.render(documents=documents, query=message)
+        
+#         logger.info(f"Response generation took {perf_counter() - start_time:.2f} seconds")
+#         return "", history, prompt_html
     
+#     except Exception as e:
+#         logger.error(f"Error in response generation: {e}")
+#         return "", history, f"Error generating response: {str(e)}"
+
+# def translate_text(selected_language, history):
+#     """Translate the last response in history to the selected language."""
 #     iso_language_codes = {
-#         "Hindi": "hi",
-#         "Gom": "gom",
-#         "Kannada": "kn",
-#         "Dogri": "doi",
-#         "Bodo": "brx",
-#         "Urdu": "ur",
-#         "Tamil": "ta",
-#         "Kashmiri": "ks",
-#         "Assamese": "as",
-#         "Bengali": "bn",
-#         "Marathi": "mr",
-#         "Sindhi": "sd",
-#         "Maithili": "mai",
-#         "Punjabi": "pa",
-#         "Malayalam": "ml",
-#         "Manipuri": "mni",
-#         "Telugu": "te",
-#         "Sanskrit": "sa",
-#         "Nepali": "ne",
-#         "Santali": "sat",
-#         "Gujarati": "gu",
-#         "Odia": "or"
+#         "Hindi": "hi", "Gom": "gom", "Kannada": "kn", "Dogri": "doi", "Bodo": "brx", "Urdu": "ur",
+#         "Tamil": "ta", "Kashmiri": "ks", "Assamese": "as", "Bengali": "bn", "Marathi": "mr",
+#         "Sindhi": "sd", "Maithili": "mai", "Punjabi": "pa", "Malayalam": "ml", "Manipuri": "mni",
+#         "Telugu": "te", "Sanskrit": "sa", "Nepali": "ne", "Santali": "sat", "Gujarati": "gu", "Odia": "or"
 #     }
     
 #     to_code = iso_language_codes[selected_language]
-#     response_text = history[-1][1] if history else ''
-#     print('response_text for translation',response_text)
+#     response_text = history[-1][1] if history and history[-1][1] else ''
+#     print('response_text for translation', response_text)
 #     translation = bhashini_translate(response_text, to_code=to_code)
-#     return translation['translated_content']
-   
+#     return translation.get('translated_content', 'Translation failed.')
 
 # # Gradio interface
 # with gr.Blocks(theme='gradio/soft') as CHATBOT:
@@ -525,43 +503,336 @@ CHATBOT.launch(share=True, debug=True)# import requests
 #             "Sindhi", "Maithili", "Punjabi", "Malayalam", "Manipuri", "Telugu", "Sanskrit", "Nepali", "Santali",
 #             "Gujarati", "Odia"
 #         ],
-#         value="Hindi",  # default to Hindi
+#         value="Hindi",
 #         label="Select Language for Translation"
 #     )
     
 #     prompt_html = gr.HTML()
-    
 #     translated_textbox = gr.Textbox(label="Translated Response")
-#     def update_history_and_translate(txt, cross_encoder, history_state, language_dropdown):
-#         print('History state',history_state)
-#         history = history_state
+
+#     def update_chat_and_translate(txt, cross_encoder, history_state, language_dropdown):
+#         history = history_state.value if history_state.value else []
 #         history.append((txt, ""))
-#         #history_state.value=(history)
         
-#         # Call bot function
-#         # bot_output = list(bot(history, cross_encoder))
-#         bot_output = next(bot(history, cross_encoder))
-#         print('bot_output',bot_output)
-#         #history, prompt_html = bot_output[-1]
-#         history, prompt_html = bot_output
-#         print('History',history)
-#         # Update the history state
-#         history_state[:] = history
+#         # Call simple_chat_function
+#         msg, history, prompt_html_content = simple_chat_function(txt, history, cross_encoder)
         
 #         # Translate text
 #         translated_text = translate_text(language_dropdown, history)
-#         return history, prompt_html, translated_text
+        
+#         return history, prompt_html_content, translated_text
 
-#     txt_msg = txt_btn.click(update_history_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
-#     txt_msg = txt.submit(update_history_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
+#     txt_msg = txt_btn.click(update_chat_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
+#     txt_msg = txt.submit(update_chat_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
 
-#     examples = ['My transhipment cargo is missing','can u explain and tabulate difference between b 17 bond and a warehousing bond',
-#             'What are benefits of  the AEO Scheme and eligibility criteria?',
-#             'What are penalties for customs offences? ', 'what are penalties to customs officers misusing their powers under customs act?','What are eligibility criteria for exemption from cost recovery charges','list in detail what is procedure for obtaining new approval for openeing a CFS attached to an ICD']
+#     examples = [
+#         'My transhipment cargo is missing',
+#         'Can you explain and tabulate the difference between B-17 bond and a warehousing bond?',
+#         'What are the benefits of the AEO Scheme and eligibility criteria?',
+#         'What are penalties for customs offences?',
+#         'What are penalties for customs officers misusing their powers under the Customs Act?',
+#         'What are eligibility criteria for exemption from cost recovery charges?',
+#         'List in detail the procedure for obtaining new approval for opening a CFS attached to an ICD'
+#     ]
 
 #     gr.Examples(examples, txt)
 
-
 # # Launch the Gradio application
-# CHATBOT.launch(share=True,debug=True)
+# CHATBOT.launch(share=True, debug=True)# import requests
+# # import gradio as gr
+# # from ragatouille import RAGPretrainedModel
+# # import logging
+# # from pathlib import Path
+# # from time import perf_counter
+# # from sentence_transformers import CrossEncoder
+# # from huggingface_hub import InferenceClient
+# # from jinja2 import Environment, FileSystemLoader
+# # import numpy as np
+# # from os import getenv
+# # from backend.query_llm import generate_hf, generate_qwen
+# # from backend.semantic_search import table, retriever
+# # from huggingface_hub import InferenceClient
+
+
+# # # Bhashini API translation function
+# # api_key = getenv('API_KEY')
+# # user_id = getenv('USER_ID')
+
+# # def bhashini_translate(text: str, from_code: str = "en", to_code: str = "hi") -> dict:
+# #     """Translates text from source language to target language using the Bhashini API."""
+    
+# #     if not text.strip():
+# #         print('Input text is empty. Please provide valid text for translation.')
+# #         return {"status_code": 400, "message": "Input text is empty", "translated_content": None, "speech_content": None}
+# #     else:
+# #         print('Input text - ',text)
+# #     print(f'Starting translation process from {from_code} to {to_code}...')
+# #     print(f'Starting translation process from {from_code} to {to_code}...')
+# #     gr.Warning(f'Translating to {to_code}...')
+    
+# #     url = 'https://meity-auth.ulcacontrib.org/ulca/apis/v0/model/getModelsPipeline'
+# #     headers = {
+# #         "Content-Type": "application/json",
+# #         "userID": user_id,
+# #         "ulcaApiKey": api_key
+# #     }
+# #     payload = {
+# #         "pipelineTasks": [{"taskType": "translation", "config": {"language": {"sourceLanguage": from_code, "targetLanguage": to_code}}}],
+# #         "pipelineRequestConfig": {"pipelineId": "64392f96daac500b55c543cd"}
+# #     }
+    
+# #     print('Sending initial request to get the pipeline...')
+# #     response = requests.post(url, json=payload, headers=headers)
+    
+# #     if response.status_code != 200:
+# #         print(f'Error in initial request: {response.status_code}')
+# #         return {"status_code": response.status_code, "message": "Error in translation request", "translated_content": None}
+
+# #     print('Initial request successful, processing response...')
+# #     response_data = response.json()
+# #     service_id = response_data["pipelineResponseConfig"][0]["config"][0]["serviceId"]
+# #     callback_url = response_data["pipelineInferenceAPIEndPoint"]["callbackUrl"]
+    
+# #     print(f'Service ID: {service_id}, Callback URL: {callback_url}')
+    
+# #     headers2 = {
+# #         "Content-Type": "application/json",
+# #         response_data["pipelineInferenceAPIEndPoint"]["inferenceApiKey"]["name"]: response_data["pipelineInferenceAPIEndPoint"]["inferenceApiKey"]["value"]
+# #     }
+# #     compute_payload = {
+# #         "pipelineTasks": [{"taskType": "translation", "config": {"language": {"sourceLanguage": from_code, "targetLanguage": to_code}, "serviceId": service_id}}],
+# #         "inputData": {"input": [{"source": text}], "audio": [{"audioContent": None}]}
+# #     }
+    
+# #     print(f'Sending translation request with text: "{text}"')
+# #     compute_response = requests.post(callback_url, json=compute_payload, headers=headers2)
+    
+# #     if compute_response.status_code != 200:
+# #         print(f'Error in translation request: {compute_response.status_code}')
+# #         return {"status_code": compute_response.status_code, "message": "Error in translation", "translated_content": None}
+    
+# #     print('Translation request successful, processing translation...')
+# #     compute_response_data = compute_response.json()
+# #     translated_content = compute_response_data["pipelineResponse"][0]["output"][0]["target"]
+    
+# #     print(f'Translation successful. Translated content: "{translated_content}"')
+# #     return {"status_code": 200, "message": "Translation successful", "translated_content": translated_content}
+
+
+# # # Existing chatbot functions
+# # VECTOR_COLUMN_NAME = "vector"
+# # TEXT_COLUMN_NAME = "text"
+# # HF_TOKEN = getenv("HUGGING_FACE_HUB_TOKEN")
+# # proj_dir = Path(__file__).parent
+
+# # logging.basicConfig(level=logging.INFO)
+# # logger = logging.getLogger(__name__)
+# # client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1", token=HF_TOKEN)
+# # proj_dir = Path(__file__).parent
+# # env = Environment(loader=FileSystemLoader(proj_dir / 'templates'))
+
+# # template = env.get_template('template.j2')
+# # template_html = env.get_template('template_html.j2')
+
+# # # def add_text(history, text):
+# # #     history = [] if history is None else history
+# # #     history = history + [(text, None)]
+# # #     return history, gr.Textbox(value="", interactive=False)
+
+# # def bot(history, cross_encoder):
+
+# #     top_rerank = 25
+# #     top_k_rank = 20
+# #     query = history[-1][0] if history else ''
+# #     print('\nQuery: ',query )
+# #     print('\nHistory:',history)
+# #     if not query:
+# #         gr.Warning("Please submit a non-empty string as a prompt")
+# #         raise ValueError("Empty string was submitted")
+
+# #     logger.warning('Retrieving documents...')
+    
+# #     if cross_encoder == '(HIGH ACCURATE) ColBERT':
+# #         gr.Warning('Retrieving using ColBERT.. First time query will take a minute for model to load..pls wait')
+# #         RAG = RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")
+# #         RAG_db = RAG.from_index('.ragatouille/colbert/indexes/cbseclass10index')
+# #         documents_full = RAG_db.search(query, k=top_k_rank)
+        
+# #         documents = [item['content'] for item in documents_full]
+# #         prompt = template.render(documents=documents, query=query)
+# #         prompt_html = template_html.render(documents=documents, query=query)
+    
+# #         generate_fn = generate_hf
+    
+# #         history[-1][1] = ""
+# #         for character in generate_fn(prompt, history[:-1]):
+# #             history[-1][1] = character
+# #             yield history, prompt_html
+# #     else:
+# #         document_start = perf_counter()
+    
+# #         query_vec = retriever.encode(query)
+# #         doc1 = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_k_rank)
+    
+# #         documents = table.search(query_vec, vector_column_name=VECTOR_COLUMN_NAME).limit(top_rerank).to_list()
+# #         documents = [doc[TEXT_COLUMN_NAME] for doc in documents]
+    
+# #         query_doc_pair = [[query, doc] for doc in documents]
+# #         if cross_encoder == '(FAST) MiniLM-L6v2':
+# #             cross_encoder1 = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
+# #         elif cross_encoder == '(ACCURATE) BGE reranker':
+# #             cross_encoder1 = CrossEncoder('BAAI/bge-reranker-base')
+        
+# #         cross_scores = cross_encoder1.predict(query_doc_pair)
+# #         sim_scores_argsort = list(reversed(np.argsort(cross_scores)))
+        
+# #         documents = [documents[idx] for idx in sim_scores_argsort[:top_k_rank]]
+    
+# #         document_time = perf_counter() - document_start
+    
+# #         prompt = template.render(documents=documents, query=query)
+# #         prompt_html = template_html.render(documents=documents, query=query)
+    
+# #         #generate_fn = generate_hf
+# #         generate_fn=generate_qwen
+# #         # Create a new history entry instead of modifying the tuple directly
+# #         new_history = history[:-1] + [ (prompt, "") ] # query replaced prompt
+# #         output=''
+# #         # for character in generate_fn(prompt, history[:-1]):
+# #         #     #new_history[-1] = (query, character) 
+# #         #     output+=character
+# #         output=generate_fn(prompt, history[:-1])
+        
+# #         print('Output:',output)
+# #         new_history[-1] = (prompt, output) #query replaced with prompt
+# #         print('New History',new_history)
+# #         #print('prompt html',prompt_html)# Update the last tuple with new text
+        
+# #         history_list = list(history[-1])
+# #         history_list[1] = output  # Assuming `character` is what you want to assign
+# #         # Update the history with the modified list converted back to a tuple
+# #         history[-1] = tuple(history_list)
+
+# #             #history[-1][1] = character
+# #         # yield new_history, prompt_html
+# #         yield history, prompt_html
+# #          # new_history,prompt_html
+# #         # history[-1][1] = ""
+# #         # for character in generate_fn(prompt, history[:-1]):
+# #         #     history[-1][1] = character
+# #         #     yield history, prompt_html
+
+# # #def translate_text(response_text, selected_language):
+    
+# # def translate_text(selected_language,history):
+    
+# #     iso_language_codes = {
+# #         "Hindi": "hi",
+# #         "Gom": "gom",
+# #         "Kannada": "kn",
+# #         "Dogri": "doi",
+# #         "Bodo": "brx",
+# #         "Urdu": "ur",
+# #         "Tamil": "ta",
+# #         "Kashmiri": "ks",
+# #         "Assamese": "as",
+# #         "Bengali": "bn",
+# #         "Marathi": "mr",
+# #         "Sindhi": "sd",
+# #         "Maithili": "mai",
+# #         "Punjabi": "pa",
+# #         "Malayalam": "ml",
+# #         "Manipuri": "mni",
+# #         "Telugu": "te",
+# #         "Sanskrit": "sa",
+# #         "Nepali": "ne",
+# #         "Santali": "sat",
+# #         "Gujarati": "gu",
+# #         "Odia": "or"
+# #     }
+    
+# #     to_code = iso_language_codes[selected_language]
+# #     response_text = history[-1][1] if history else ''
+# #     print('response_text for translation',response_text)
+# #     translation = bhashini_translate(response_text, to_code=to_code)
+# #     return translation['translated_content']
+   
+
+# # # Gradio interface
+# # with gr.Blocks(theme='gradio/soft') as CHATBOT:
+# #     history_state = gr.State([])
+# #     with gr.Row():
+# #         with gr.Column(scale=10):
+# #             gr.HTML(value="""<div style="color: #FF4500;"><h1>ADWITIYA-</h1> <h1><span style="color: #008000">Custom Manual Chatbot </span></h1></div>""")
+# #             gr.HTML(value=f"""<p style="font-family: sans-serif; font-size: 16px;">Using GenAI for CBIC Capacity Building - A free chat bot developed by National Customs Targeting Center using Open source LLMs for CBIC Officers</p>""")
+# #             gr.HTML(value=f"""<p style="font-family: Arial, sans-serif; font-size: 14px;">Developed by NCTC,Mumbai. Suggestions may be sent to <a href="mailto:[email protected]" style="color: #00008B; font-style: italic;">[email protected]</a>.</p>""")
+
+# #         with gr.Column(scale=3):
+# #             gr.Image(value='logo.png', height=200, width=200)
+
+# #     chatbot = gr.Chatbot(
+# #         [],
+# #         elem_id="chatbot",
+# #         avatar_images=('https://aui.atlassian.com/aui/8.8/docs/images/avatar-person.svg',
+# #                        'https://huggingface.co/datasets/huggingface/brand-assets/resolve/main/hf-logo.svg'),
+# #         bubble_full_width=False,
+# #         show_copy_button=True,
+# #         show_share_button=True,
+# #     )
+
+# #     with gr.Row():
+# #         txt = gr.Textbox(
+# #             scale=3,
+# #             show_label=False,
+# #             placeholder="Enter text and press enter",
+# #             container=False,
+# #         )
+# #         txt_btn = gr.Button(value="Submit text", scale=1)
+    
+# #     cross_encoder = gr.Radio(choices=['(FAST) MiniLM-L6v2', '(ACCURATE) BGE reranker', '(HIGH ACCURATE) ColBERT'], value='(ACCURATE) BGE reranker', label="Embeddings", info="Only First query to Colbert may take little time)")
+# #     language_dropdown = gr.Dropdown(
+# #         choices=[
+# #             "Hindi", "Gom", "Kannada", "Dogri", "Bodo", "Urdu", "Tamil", "Kashmiri", "Assamese", "Bengali", "Marathi",
+# #             "Sindhi", "Maithili", "Punjabi", "Malayalam", "Manipuri", "Telugu", "Sanskrit", "Nepali", "Santali",
+# #             "Gujarati", "Odia"
+# #         ],
+# #         value="Hindi",  # default to Hindi
+# #         label="Select Language for Translation"
+# #     )
+    
+# #     prompt_html = gr.HTML()
+    
+# #     translated_textbox = gr.Textbox(label="Translated Response")
+# #     def update_history_and_translate(txt, cross_encoder, history_state, language_dropdown):
+# #         print('History state',history_state)
+# #         history = history_state
+# #         history.append((txt, ""))
+# #         #history_state.value=(history)
+        
+# #         # Call bot function
+# #         # bot_output = list(bot(history, cross_encoder))
+# #         bot_output = next(bot(history, cross_encoder))
+# #         print('bot_output',bot_output)
+# #         #history, prompt_html = bot_output[-1]
+# #         history, prompt_html = bot_output
+# #         print('History',history)
+# #         # Update the history state
+# #         history_state[:] = history
+        
+# #         # Translate text
+# #         translated_text = translate_text(language_dropdown, history)
+# #         return history, prompt_html, translated_text
+
+# #     txt_msg = txt_btn.click(update_history_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
+# #     txt_msg = txt.submit(update_history_and_translate, [txt, cross_encoder, history_state, language_dropdown], [chatbot, prompt_html, translated_textbox])
+
+# #     examples = ['My transhipment cargo is missing','can u explain and tabulate difference between b 17 bond and a warehousing bond',
+# #             'What are benefits of  the AEO Scheme and eligibility criteria?',
+# #             'What are penalties for customs offences? ', 'what are penalties to customs officers misusing their powers under customs act?','What are eligibility criteria for exemption from cost recovery charges','list in detail what is procedure for obtaining new approval for openeing a CFS attached to an ICD']
+
+# #     gr.Examples(examples, txt)
+
+
+# # # Launch the Gradio application
+# # CHATBOT.launch(share=True,debug=True)