Spaces:
Running
Running
| import gradio as gr | |
| import os | |
| import time | |
| from langchain.document_loaders import OnlinePDFLoader | |
| from langchain.text_splitter import CharacterTextSplitter | |
| from langchain.llms import OpenAI | |
| from langchain.embeddings import OpenAIEmbeddings | |
| from langchain.vectorstores import Chroma | |
| from langchain.chains import ConversationalRetrievalChain | |
| from langchain import PromptTemplate | |
| from transformers import Pix2StructForConditionalGeneration, Pix2StructProcessor | |
| import requests | |
| from PIL import Image | |
| import torch | |
| # _template = """Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question. | |
| # Chat History: | |
| # {chat_history} | |
| # Follow Up Input: {question} | |
| # Standalone question:""" | |
| # CONDENSE_QUESTION_PROMPT = PromptTemplate.from_template(_template) | |
| # template = """ | |
| # You are given the following extracted parts of a long document and a question. Provide a short structured answer. | |
| # If you don't know the answer, look on the web. Don't try to make up an answer. | |
| # Question: {question} | |
| # ========= | |
| # {context} | |
| # ========= | |
| # Answer in Markdown:""" | |
| torch.hub.download_url_to_file('https://raw.githubusercontent.com/vis-nlp/ChartQA/main/ChartQA%20Dataset/val/png/20294671002019.png', 'chart_example.png') | |
| torch.hub.download_url_to_file('https://raw.githubusercontent.com/vis-nlp/ChartQA/main/ChartQA%20Dataset/test/png/multi_col_1081.png', 'chart_example_2.png') | |
| torch.hub.download_url_to_file('https://raw.githubusercontent.com/vis-nlp/ChartQA/main/ChartQA%20Dataset/test/png/18143564004789.png', 'chart_example_3.png') | |
| torch.hub.download_url_to_file('https://sharkcoder.com/files/article/matplotlib-bar-plot.png', 'chart_example_4.png') | |
| model_name = "google/matcha-chartqa" | |
| model = Pix2StructForConditionalGeneration.from_pretrained(model_name) | |
| processor = Pix2StructProcessor.from_pretrained(model_name) | |
| device = torch.device("cuda" if torch.cuda.is_available() else "cpu") | |
| model.to(device) | |
| def filter_output(output): | |
| return output.replace("<0x0A>", "") | |
| def chart_qa(image, question): | |
| inputs = processor(images=image, text=question, return_tensors="pt").to(device) | |
| predictions = model.generate(**inputs, max_new_tokens=512) | |
| return filter_output(processor.decode(predictions[0], skip_special_tokens=True)) | |
| def loading_pdf(): | |
| return "Loading..." | |
| def pdf_changes(pdf_doc, open_ai_key): | |
| if open_ai_key is not None: | |
| os.environ['OPENAI_API_KEY'] = open_ai_key | |
| loader = OnlinePDFLoader(pdf_doc.name) | |
| documents = loader.load() | |
| text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0) | |
| texts = text_splitter.split_documents(documents) | |
| embeddings = OpenAIEmbeddings() | |
| db = Chroma.from_documents(texts, embeddings) | |
| retriever = db.as_retriever() | |
| global qa | |
| qa = ConversationalRetrievalChain.from_llm( | |
| llm=OpenAI(temperature=0.5), | |
| retriever=retriever, | |
| return_source_documents=True) | |
| return "Ready" | |
| else: | |
| return "You forgot OpenAI API key" | |
| def add_text(history, text): | |
| history = history + [(text, None)] | |
| return history, "" | |
| def bot(history): | |
| response = infer(history[-1][0], history) | |
| history[-1][1] = "" | |
| for character in response: | |
| history[-1][1] += character | |
| time.sleep(0.05) | |
| yield history | |
| def infer(question, history): | |
| res = [] | |
| for human, ai in history[:-1]: | |
| pair = (human, ai) | |
| res.append(pair) | |
| chat_history = res | |
| #print(chat_history) | |
| query = question | |
| result = qa({"question": query, "chat_history": chat_history}) | |
| #print(result) | |
| return result["answer"] | |
| css=""" | |
| #col-container {max-width: 700px; margin-left: auto; margin-right: auto;} | |
| """ | |
| title = """ | |
| <div style="text-align: center;"> | |
| <h1>SpreadSight Demo</h1> | |
| <p style="text-align: center;">Please specify OpenAI Key before use</p> | |
| </div> | |
| """ | |
| # with gr.Blocks(css=css) as demo: | |
| # with gr.Column(elem_id="col-container"): | |
| # gr.HTML(title) | |
| # with gr.Column(): | |
| # openai_key = gr.Textbox(label="You OpenAI API key", type="password") | |
| # pdf_doc = gr.File(label="Load a pdf", file_types=['.pdf'], type="file") | |
| # with gr.Row(): | |
| # langchain_status = gr.Textbox(label="Status", placeholder="", interactive=False) | |
| # load_pdf = gr.Button("Load pdf to langchain") | |
| # chatbot = gr.Chatbot([], elem_id="chatbot").style(height=350) | |
| # question = gr.Textbox(label="Question", placeholder="Type your question and hit Enter ") | |
| # submit_btn = gr.Button("Send Message") | |
| # load_pdf.click(loading_pdf, None, langchain_status, queue=False) | |
| # load_pdf.click(pdf_changes, inputs=[pdf_doc, openai_key], outputs=[langchain_status], queue=False) | |
| # question.submit(add_text, [chatbot, question], [chatbot, question]).then( | |
| # bot, chatbot, chatbot | |
| # ) | |
| # submit_btn.click(add_text, [chatbot, question], [chatbot, question]).then( | |
| # bot, chatbot, chatbot) | |
| # demo.launch() | |
| """functions""" | |
| def load_file(): | |
| return "Loading..." | |
| def load_xlsx(name): | |
| import pandas as pd | |
| xls_file = rf'{name}' | |
| data = pd.read_excel(xls_file) | |
| return data | |
| def table_loader(table_file, open_ai_key): | |
| import os | |
| from langchain.llms import OpenAI | |
| from langchain.agents import create_pandas_dataframe_agent | |
| from pandas import read_csv | |
| global agent | |
| if open_ai_key is not None: | |
| os.environ['OPENAI_API_KEY'] = open_ai_key | |
| else: | |
| return "Enter API" | |
| if table_file.name.endswith('.xlsx') or table_file.name.endswith('.xls'): | |
| data = load_xlsx(table_file.name) | |
| agent = create_pandas_dataframe_agent(OpenAI(temperature=0), data) | |
| return "Ready!" | |
| elif table_file.name.endswith('.csv'): | |
| data = read_csv(table_file.name) | |
| agent = create_pandas_dataframe_agent(OpenAI(temperature=0), data) | |
| return "Ready!" | |
| else: | |
| return "Wrong file format! Upload excel file or csv!" | |
| def run(query): | |
| from langchain.callbacks import get_openai_callback | |
| with get_openai_callback() as cb: | |
| response = (agent.run(query)) | |
| costs = (f"Total Cost (USD): ${cb.total_cost}") | |
| output = f'{response} \n {costs}' | |
| return output | |
| def respond(message, chat_history): | |
| import time | |
| bot_message = run(message) | |
| chat_history.append((message, bot_message)) | |
| time.sleep(0.5) | |
| return "", chat_history | |
| with gr.Blocks() as demo: | |
| with gr.Column(elem_id="col-container"): | |
| gr.HTML(title) | |
| key = gr.Textbox( | |
| show_label=False, | |
| placeholder="Your OpenAI key", | |
| type = 'password', | |
| ).style(container=False) | |
| # PDF processing tab | |
| with gr.Tab("Files"): | |
| with gr.Row(): | |
| with gr.Column(scale=0.5): | |
| langchain_status = gr.Textbox(label="Status", placeholder="", interactive=False) | |
| load_pdf = gr.Button("Load pdf to Spreadsight") | |
| with gr.Column(scale=0.5): | |
| pdf_doc = gr.File(label="Load a pdf", file_types=['.pdf'], type="file") | |
| with gr.Row(): | |
| with gr.Column(scale=1): | |
| chatbot = gr.Chatbot([], elem_id="chatbot").style(height=350) | |
| with gr.Row(): | |
| with gr.Column(scale=0.85): | |
| question = gr.Textbox( | |
| show_label=False, | |
| placeholder="Enter text and press enter, or upload an image", | |
| ).style(container=False) | |
| with gr.Column(scale=0.15, min_width=0): | |
| clr_btn = gr.Button("Clear!") | |
| load_pdf.click(loading_pdf, None, langchain_status, queue=False) | |
| load_pdf.click(pdf_changes, inputs=[pdf_doc, key], outputs=[langchain_status], queue=True) | |
| question.submit(add_text, [chatbot, question], [chatbot, question]).then( | |
| bot, chatbot, chatbot | |
| ) | |
| # XLSX and CSV processing tab | |
| with gr.Tab("Spreadsheets"): | |
| with gr.Row(): | |
| with gr.Column(scale=0.5): | |
| status_sh = gr.Textbox(label="Status", placeholder="", interactive=False) | |
| load_table = gr.Button("Load csv|xlsx to langchain") | |
| with gr.Column(scale=0.5): | |
| raw_table = gr.File(label="Load a table file (xls or csv)", file_types=['.csv, xlsx, xls'], type="file") | |
| with gr.Row(): | |
| with gr.Column(scale=1): | |
| chatbot_sh = gr.Chatbot([], elem_id="chatbot").style(height=350) | |
| with gr.Row(): | |
| with gr.Column(scale=0.85): | |
| question_sh = gr.Textbox( | |
| show_label=False, | |
| placeholder="Enter text and press enter, or upload an image", | |
| ).style(container=False) | |
| with gr.Column(scale=0.15, min_width=0): | |
| clr_btn = gr.Button("Clear!") | |
| load_table.click(load_file, None, status_sh, queue=False) | |
| load_table.click(table_loader, inputs=[raw_table, key], outputs=[status_sh], queue=False) | |
| question_sh.submit(respond, [question_sh, chatbot_sh], [question_sh, chatbot_sh]) | |
| clr_btn.click(lambda: None, None, chatbot_sh, queue=False) | |
| with gr.Tab("Charts"): | |
| image = gr.Image(type="pil", label="Chart") | |
| question = gr.Textbox(label="Question") | |
| load_chart = gr.Button("Load chart and question!") | |
| answer = gr.Textbox(label="Model Output") | |
| load_chart.click(chart_qa, [image, question], answer) | |
| demo.queue(concurrency_count=3) | |
| demo.launch() |