import gradio as gr from typing import Any from langchain.vectorstores import Chroma from langchain.chat_models import ChatOpenAI from langchain.document_loaders import PyPDFLoader from langchain.embeddings.openai import OpenAIEmbeddings from langchain.chains import ConversationalRetrievalChain import re import uuid import chromadb title = 'DocTalker' description = """ PDFs Conversing Naturally. Extract insights, ask questions, get instant context. Transforming documents into dynamic dialogues. Engage intelligently with your content.""" # enable_box = gr.Textbox.update(value = None, placeholder = 'Upload your OpenAI API key',interactive = True) disable_box = gr.Textbox.update(value = 'OpenAI API key is Set', interactive = False) def set_apikey(api_key: str): app.OPENAI_API_KEY = api_key return disable_box # def enable_api_box(): # return enable_box def add_text(history, text: str): if not text: raise gr.Error('enter text') history = history + [(text,'')] return history class my_app: def __init__(self, OPENAI_API_KEY: str = None) -> None: self.OPENAI_API_KEY: str = OPENAI_API_KEY self.chain = None self.chat_history: list = [] self.N: int = 0 self.count: int = 0 self.model_type: str = '' self.max_token: int = 0 def __call__(self, file: str) -> Any: if self.count==0: self.chain = self.build_chain(file) self.count+=1 else: self.chain = None self.chat_history = [] # Clear chat history self.N = 0 self.count = 0 self.model_type = '' self.max_token = 0 self.chain = self.build_chain(file) return self.chain def chroma_client(self): #create a chroma client client = chromadb.Client() #create a collection collection = client.get_or_create_collection(name="my-collection") return client def process_file(self,file: str): loader = PyPDFLoader(file.name) documents = loader.load() pattern = r"/([^/]+)$" match = re.search(pattern, file.name) file_name = match.group(1) return documents, file_name def build_chain(self, file: str): if not self.model_type: raise ValueError("Model type must be set before building the chain.") documents, file_name = self.process_file(file) #Load embeddings model embeddings = OpenAIEmbeddings(openai_api_key=self.OPENAI_API_KEY) pdfsearch = Chroma.from_documents(documents, embeddings, collection_name= file_name,) # print(self.model_type) chain = ConversationalRetrievalChain.from_llm( ChatOpenAI( temperature = 0.0, openai_api_key = self.OPENAI_API_KEY, model_name = self.model_type, max_tokens = self.max_token ), retriever=pdfsearch.as_retriever(search_kwargs={"k": 1}), return_source_documents=True,) return chain def get_response(history, query, file): print(app.model_type) chain = app.chain result = chain({"question": query, 'chat_history':app.chat_history},return_only_outputs=True) app.chat_history += [(query, result["answer"])] app.N = list(result['source_documents'][0])[1][1]['page'] for char in result['answer']: history[-1][-1] += char yield history,'' def drop_value(file, typeM: str): app.model_type = typeM print(typeM) def slide_value(file, value: int): app.max_token = value # Assuming you have a max_token attribute in your app class print(value) def confirm_update(file): # Call app(file) once after updating attributes app(file) print('file loaded') return [gr.update(visible=True),gr.update(visible=False),gr.update(visible=False)] def update_textbox(selected_value): return selected_value input_box.value = selected_value app = my_app() with gr.Blocks() as demo: gr.Markdown(f'

{title}

') gr.Markdown(f'
{description}
') with gr.Column(): with gr.Row(): with gr.Column(scale=0.8): api_key = gr.Textbox(placeholder='Enter OpenAI API key and press enter', show_label=False, interactive=True).style(container=False) with gr.Column(scale=0.2): # change_api_key = gr.Button('Get OpenAI API Key', link="https://platform.openai.com/account/api-keys") gr.Markdown( '' ) with gr.Row(): with gr.Column(min_width=400): pdf_url = gr.Textbox(label='Enter PDF URL here') gr.Markdown("

OR

") btn = gr.File( label='Upload your PDF/ Research Paper / Book here', file_types=['.pdf'] ) # btn = gr.UploadButton("📁 upload a PDF", file_types=[".pdf"]).style() drop = gr.Dropdown(label="Select Model", info="GEN AI model", choices=["gpt-4-32k", "gpt-4", "gpt-3.5-turbo", "gpt-3.5-turbo-16k"]) slide = gr.Slider(0, 2048, step=20, label="Max Token", info="Change output Token size according",interactive=True) update_btn = gr.Button('Update App') confirm_btn = gr.Button("Confirm Update", variant="stop", visible=False) cancel_btn = gr.Button("Cancel", visible=False) drop_qus = gr.Dropdown( label="PRE-SET QUESTIONS", info="Click a question to automatically populate the input box, and then hit Enter!", choices=[ "Highlighted document equations or formulas?", "What specific areas does this document cover?", "Which theories are explored within this document?", "Are there any notable equations or formulas highlighted in the content?", "What is the central focus of this document or PDF?", "Could you offer a concise overview of the main points in this material?", "What limitations or shortcomings were identified in the study?", "Can you summarize the main contributions or advancements outlined?", "What conclusion or final remarks are provided in this paper?" ] ) with gr.Column(min_width=600): chatbot = gr.Chatbot(value=[], label='Doctalker', elem_id='chatbot').style(height=550) input_box = gr.Textbox( show_label=False, container=False, placeholder="Enter text and press enter", )#.style(container=False) # with gr.Row(): # with gr.Column(min_width=400): # dataset_selection = gr.Dataset( # components=[gr.Textbox(visible=False)], # label="PRE-SET QUESTIONS: Click a question to automatically populate the input box, and then hit Enter!", # samples=[ # ["Highlighted document equations or formulas?"], # ["What specific areas does this document cover?"], # ["Which theories are explored within this document?"], # ["Are there any notable equations or formulas highlighted in the content?"], # ["What is the central focus of this document or PDF?"], # ["Could you offer a concise overview of the main points in this material?"], # ["What limitations or shortcomings were identified in the study?"], # ["Can you summarize the main contributions or advancements outlined?"], # ["What conclusion or final remarks are provided in this paper?"] # ], # ) # with gr.Column(min_width=200): # submit_btn = gr.Button('submit') print(api_key) api_key.submit( fn=set_apikey, inputs=[api_key], outputs=[api_key,]) drop.change( fn=drop_value, inputs=[btn, drop], outputs=[]) slide.change( fn=slide_value, inputs=[btn, slide], outputs=[]) update_btn.click(lambda :[gr.update(visible=False), gr.update(visible=True), gr.update(visible=True)], None, [update_btn, confirm_btn, cancel_btn]) cancel_btn.click(lambda :[gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)], None, [update_btn, confirm_btn, cancel_btn]) confirm_btn.click( fn=confirm_update, inputs=[btn], outputs=[update_btn, confirm_btn, cancel_btn] ) drop_qus.change( fn=update_textbox, inputs=[drop_qus], outputs=[input_box] ) input_box.submit( fn=add_text, inputs=[chatbot,input_box], outputs=[chatbot, ], queue=False).success( fn=get_response, inputs = [chatbot, input_box, btn], outputs = [chatbot,input_box]) # submit_btn.click( # fn=add_text, # inputs=[chatbot,txt], # outputs=[chatbot, ], # queue=False).success( # fn=get_response, # inputs = [chatbot, txt, btn], # outputs = [chatbot,txt]) demo.queue() demo.launch()