app.py

import gradio as gr
import os
import re
import pandas as pd
from pathlib import Path
from time import sleep
from tqdm import tqdm
from api_calls import *

ROOT_DIR = Path(__file__).resolve().parents[0]

def disable_btn():
    return gr.Button.update(interactive=False)

def enable_btn():
    return gr.Button.update(interactive=True)

def preview_uploaded_file(file_paths):
    if file_paths:
        return gr.update(value=file_paths[0])
    else:
        return gr.update(value=None)

def open_data_check(checked):
    if checked:
        return gr.update(visible=True)
    else:
        return gr.update(visible=False)

def uploaded_file_process(file_path, ocr_model_choice):
    name, filetype = Path(file_path).parts[-1].split(".")[0], Path(file_path).parts[-1].split(".")[-1]
    print(name)
    ocr_extracted_data = api_ocr(
        image_filepath=file_path, model_provider=ocr_model_choice)
    return ocr_extracted_data

def reference_from_file(file_paths, ocr_model_choice="Gemini Pro Vision"):

    data_array = []
    for file_path in tqdm(file_paths):
        data = uploaded_file_process(file_path, ocr_model_choice=ocr_model_choice)
        data_array.append(data)
        sleep(1)
    return data_array

def print_like_dislike(x: gr.LikeData):
    print(x.index, x.value, x.liked)

def bot(query, history, data_array, file_paths, qa_prompt_tmpl, checkbox_replace):
    if data_array:
        params = {"query": query, "filtered_data": data_array}
    else:
        params = {"query": query}
    if checkbox_replace:
        params.update({"prompt_template": qa_prompt_tmpl})
    
    if not file_paths or "大台北" in file_paths:
        func = api_qa_waterfee
    else:
        func = api_qa_normal

    response = func(**params)
    

    full_anwser = ""
    for chunk in response.iter_content(chunk_size=32):
        if chunk:
            try:
                _c = chunk.decode('utf-8')
            except UnicodeDecodeError:
                _c = " "
            full_anwser += _c
            yield full_anwser
            # print(_c, flush=True, end="")

            # for character in response:
            #     full_anwser += character
            #     yield full_anwser

def cat_report_explanation(data_array):
    response = api_qa_cat_report(data_array)
    full_anwser = ""
    for chunk in response.iter_content(chunk_size=32):
        if chunk:
            try:
                _c = chunk.decode('utf-8')
            except UnicodeDecodeError:
                _c = " "
            full_anwser += _c
            yield full_anwser

def draw_cat_pain_assessment_result(user_input_image):
    if user_input_image:
        json_result = api_model_cat_pain_assessment(user_input_image)
        print(json_result)
        total_score = sum(list(json_result.values()))
        df_result = pd.DataFrame(json_result, index=[0]).T.reset_index()
        df_result.columns = ["a", "b"]
        return gr.BarPlot(
            df_result,
            x="a",
            y="b",
            x_title="Aspects",
            y_title="Score",
            title="Cat Pain Assessment",
            vertical=False,
            height=400,
            width=800,
            tooltip=["a", "b"],
            y_lim=[0, 2],
            scale=1,
        ), gr.HTML(
            '<h3>Total Score</h3>'
            f'<span style="font-size: 50px;">{total_score}</span>'
            '<span style="font-size: 40px;">/10</span>'
        ), gr.HTML(
            '<h3>Explanation</h3>'
            '<p>Ear position: 0-2</p>'
            '<p>Orbital tightening: 0-2</p>'
            '<p>Muzzle tension: 0-2</p>'
            '<p>Whiskers change: 0-2</p>'
            '<p>Head position: 0-2</p>'
        )
    else:
        return gr.update(value=None)

chatbot = gr.Chatbot(
    [(None, "我是 ESG AI Chat\n有什麼能為您服務的嗎？")],
    elem_id="chatbot",
    scale=1,
    height=700,
    bubble_full_width=False
)
css = """
#examples_file_to_ocr {color: green !important}
#center {text-align: center}
footer {visibility: hidden}
a {color: rgb(255, 206, 10) !important}
"""
with gr.Blocks(css=css, theme=gr.themes.Monochrome(neutral_hue="green")) as demo:
    gr.HTML("<h1>GlobalModelAI AI Product Test</h1><p>Made by `GlobalModelAI Abao`</p>", elem_id="center")
    
    with gr.Tab("OCR + Text2SQL"):
        with gr.Row():
            with gr.Column():
                gr.Markdown("## OCR Processing", elem_id="center")
                ocr_model_choice = gr.Dropdown(label="Model", value="Gemini Pro Vision", choices=["GPT-4", "Gemini Pro Vision"])
                file_preview = gr.Image(type="filepath", image_mode="RGB", sources=None, label="File Preview")
                file_upload = gr.File(label="Upload File", file_types=["png", "jpg", "jpeg", "helc"], file_count='multiple')
                checkbox_open_data_check = gr.Checkbox(label="Open Data Check")
                text_data_from_file_check = gr.Textbox(label="File Upload Status", interactive=False, visible=False)
                gr.Examples(
                    examples=[
                        [[f"{ROOT_DIR}/data/image_for_test/screenshot_for_test-esg_report_table.png"]],
                        [[f"{ROOT_DIR}/data/image_for_test/screenshot_for_test-esg_report_table2.png"],
                        [f"{ROOT_DIR}/data/image_for_test/screenshot_for_test-esg_report_table3.png"]], 
                        [[f"{ROOT_DIR}/data/image_for_test/screenshot_for_test-medical_thesis_table.png"],
                        [f"{ROOT_DIR}/data/image_for_test/screenshot_for_test-medical_thesis_table2.jpg"]],
                    ],
                    inputs=file_upload,
                    outputs=text_data_from_file_check,
                    fn=reference_from_file,
                    cache_examples=True,
                    elem_id="examples_file_to_ocr"
                )
            with gr.Column():
                gr.Markdown("## Chat with your data", elem_id="center")
                with gr.Accordion("Revise Your Prompt", open=False):
                    checkbox_replace = gr.Checkbox(label="Replace with new prompt")
                    qa_prompt_tmpl = gr.Textbox(
                        label="希望用於本次問答的prompt",
                        info="必須使用到的變數：{filtered_data}、{query}",
                        value="",
                        interactive=True,
                    )
                
                chat_interface = gr.ChatInterface(
                    fn=bot,
                    additional_inputs=[text_data_from_file_check, file_upload, qa_prompt_tmpl, checkbox_replace],
                    chatbot=chatbot,
                )
                chatbot.like(print_like_dislike, None, None)
    
    with gr.Tab("Cat Pain Assessment Model"):
        gr.Markdown("## Cat Pain Assessment Model", elem_id="center")
        with gr.Row():
            user_input_image = gr.Image(
                type="filepath", image_mode="RGB", 
                sources=["upload", "webcam", "clipboard"], 
                label="Upload a cat image")
            with gr.Column():
                cat_pain_assessment_barplot = gr.BarPlot(label="Cat Pain Assessment")
                cat_pain_assessment_score = gr.HTML(elem_id="center")
                cat_pain_assessment_explanation = gr.HTML()
        gr.Examples(
            examples=[
                [f"{ROOT_DIR}/data/cat_pain_detection/fgs_cat_examples/5f2afc_3c44de4afb8345a2a56828e3dd166f41~mv2.jpg"],
                [f"{ROOT_DIR}/data/cat_pain_detection/fgs_cat_examples/5f2afc_9d9838561cde41d3b2dc9ef079dc2303~mv2.jpg"],
                [f"{ROOT_DIR}/data/cat_pain_detection/fgs_cat_examples/5f2afc_da95c2a1a3294701a007d34ec02f62a5~mv2.jpg"],
            ],
            inputs=user_input_image,
            outputs=[cat_pain_assessment_barplot, cat_pain_assessment_score, cat_pain_assessment_explanation],
            fn=draw_cat_pain_assessment_result,
            cache_examples=True,
        )

    with gr.Tab("Cat Report Explanation"):
        gr.Markdown("## Cat Report Explanation", elem_id="center")
        with gr.Row():
            with gr.Column():
                gr.Markdown("## Report Processing", elem_id="center")
                catrep_ocr_model_choice = gr.Dropdown(label="Model", value="Gemini Pro Vision", choices=["GPT-4", "Gemini Pro Vision"])
                catrep_file_preview = gr.Image(type="filepath", image_mode="RGB", sources=None, label="File Preview")
                catrep_file_upload = gr.File(label="Upload File", file_types=["png", "jpg", "jpeg", "helc"], file_count='multiple')
                catrep_button_generation_explanation = gr.Button("Start Explanation")
                catrep_checkbox_open_data_check = gr.Checkbox(label="Open Data Check")
                catrep_text_data_from_file_check = gr.Textbox(label="File Upload Status", interactive=False, visible=False)
                gr.Examples(
                    examples=[
                        [[f"{ROOT_DIR}/data/image_for_test/screenshot_for_test-cat_report_12.png"]]
                    ],
                    inputs=catrep_file_upload,
                    outputs=catrep_text_data_from_file_check,
                    fn=reference_from_file,
                    cache_examples=True,
                    elem_id="examples_file_to_ocr"
                )
            with gr.Column():
                gr.Markdown("### View Explanation", elem_id="center")
                catrep_textbox_explanation = gr.Textbox(
                    label="Explanation",
                    placeholder="Explanation will show here after you upload image & click the button",
                    interactive=False,
                )

    # Callbacks
    ## OCR + Text2SQL
    file_upload.upload(
        reference_from_file, [file_upload, ocr_model_choice], [text_data_from_file_check]
    )
    file_upload.change(
        preview_uploaded_file, [file_upload], [file_preview]
    )
    ocr_model_choice.change(
        reference_from_file, [file_upload, ocr_model_choice], [text_data_from_file_check]
    )
    checkbox_open_data_check.select(
        open_data_check, [checkbox_open_data_check], [text_data_from_file_check]
    )

    ## Cat Pain Assessment Model
    user_input_image.change(
        draw_cat_pain_assessment_result, [user_input_image], 
        [cat_pain_assessment_barplot, cat_pain_assessment_score, cat_pain_assessment_explanation]
    )

    ## Cat Report Explanation
    catrep_file_upload.upload(
        reference_from_file, [catrep_file_upload, catrep_ocr_model_choice], [catrep_text_data_from_file_check]
    )
    catrep_file_upload.change(
        preview_uploaded_file, [catrep_file_upload], [catrep_file_preview]
    )
    catrep_ocr_model_choice.change(
        reference_from_file, [catrep_file_upload, catrep_ocr_model_choice], [catrep_text_data_from_file_check]
    )
    catrep_checkbox_open_data_check.select(
        open_data_check, [catrep_checkbox_open_data_check], [catrep_text_data_from_file_check]
    )
    catrep_button_generation_explanation.click(
        cat_report_explanation, [catrep_text_data_from_file_check], [catrep_textbox_explanation]
    )


if __name__ == "__main__":
    demo.queue().launch(max_threads=10)