app.py

import os
import pyperclip
import gradio as gr
import nltk
import pytesseract
import google.generativeai as genai
from nltk.tokenize import sent_tokenize
from transformers import *
import torch
from tqdm import tqdm  # Import tqdm
import time

# Download necessary data for nltk
nltk.download('punkt')

OCR_TR_DESCRIPTION = '''# OCR Translate and Summary GeminiPro
<div id="content_align">OCR system based on Tesseract</div>'''

# Getting the list of available languages for Tesseract
choices = os.popen('tesseract --list-langs').read().split('\n')[1:-1]

# tesseract语言列表转pytesseract语言
def ocr_lang(lang_list):
    lang_str = ""
    lang_len = len(lang_list)
    if lang_len == 1:
        return lang_list[0]
    else:
        for i in range(lang_len):
            lang_list.insert(lang_len - i, "+")

        lang_str = "".join(lang_list[:-1])
        return lang_str


# ocr tesseract
def ocr_tesseract(img, languages):
    ocr_str = pytesseract.image_to_string(img, lang=ocr_lang(languages))
    return ocr_str


# 清除
def clear_content():
    return None


import pyperclip

# 复制到剪贴板
def cp_text(input_text):
    try:
        pyperclip.copy(input_text)
    except Exception as e:
        print("Error occurred while copying to clipboard")
        print(e)

# 清除剪贴板
def cp_clear():
    pyperclip.clear()

# Split the text into 2000 character chunks
def process_text_input_text(input_text):
    # Split the text into 2000 character chunks
    chunks = [input_text[i:i+2000] for i in range(0, len(input_text), 2000)]
    return chunks

def process_and_translate(api_key, input_text, src_lang, tgt_lang):
    # Process the input text into chunks
    chunks = process_text_input_text(input_text)

    # Translate each chunk and collect the results
    translated_chunks = []
    for chunk in chunks:
        if chunk is None or chunk == "":
            translated_chunks.append("System prompt: There is no content to translate!")
        else:
            prompt = f"This is an {src_lang} to {tgt_lang} translation, please provide the {tgt_lang} translation for this paragraph. Do not provide any explanations or text apart from the translation.\n{src_lang}: "
            #prompt = f"This is an {src_lang} to {tgt_lang} translation, please provide the {tgt_lang} translation for this sentence. Do not provide any explanations or text apart from the translation.\n{src_lang}: "

            genai.configure(api_key=api_key)

            generation_config = {
                                    "candidateCount": 1,
                                    "maxOutputTokens": 2048,
                                    "temperature": 0.3,
                                    "topP": 1
            }

            safety_settings = [
                    {
                        "category": "HARM_CATEGORY_HARASSMENT",
                        "threshold": "BLOCK_NONE",
                    },
                    {
                        "category": "HARM_CATEGORY_HATE_SPEECH",
                        "threshold": "BLOCK_NONE",
                    },
                    {
                        "category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
                        "threshold": "BLOCK_NONE",
                    },
                    {
                        "category": "HARM_CATEGORY_DANGEROUS_CONTENT",
                        "threshold": "BLOCK_NONE",
                    },
            ]

            model = genai.GenerativeModel(model_name='gemini-pro')
            response = model.generate_content([prompt, chunk],
                        #generation_config=generation_config, 
                        safety_settings=safety_settings
                    )
            translated_chunks.append(response.text)

    # Join the translated chunks back together into a single string
    response = '\n\n'.join(translated_chunks)

    return response

def process_and_summary(api_key, input_text, src_lang, tgt_lang):
    # Process the input text into chunks
    chunks = process_text_input_text(input_text)
    time.sleep(30)

    # Translate each chunk and collect the results
    translated_chunks = []
    for chunk in chunks:
        if chunk is None or chunk == "":
            translated_chunks.append("System prompt: There is no content to translate!")
        else:
            prompt = f"This is an {src_lang} to {tgt_lang} summarization and knowledge key points, please provide the {tgt_lang} summarization and list the {tgt_lang} knowledge key points for this sentence. Do not provide any explanations or text apart from the summarization.\n{src_lang}: "
            genai.configure(api_key=api_key)

            generation_config = {
                                    "candidateCount": 1,
                                    "maxOutputTokens": 2048,
                                    "temperature": 0.3,
                                    "topP": 1
            }

            safety_settings = [
                    {
                        "category": "HARM_CATEGORY_HARASSMENT",
                        "threshold": "BLOCK_NONE",
                    },
                    {
                        "category": "HARM_CATEGORY_HATE_SPEECH",
                        "threshold": "BLOCK_NONE",
                    },
                    {
                        "category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
                        "threshold": "BLOCK_NONE",
                    },
                    {
                        "category": "HARM_CATEGORY_DANGEROUS_CONTENT",
                        "threshold": "BLOCK_NONE",
                    },
            ]

            model = genai.GenerativeModel(model_name='gemini-pro')
            response = model.generate_content([prompt, chunk],
                        #generation_config=generation_config, 
                        safety_settings=safety_settings
                    )
            translated_chunks.append(response.text)

    # Join the translated chunks back together into a single string
    response = '\n\n*Next Paragraph*\n\n'.join(translated_chunks)

    return response

# prompt = f"Display language is {tgt_lang}, do not display original text, As a Knowledge Video Content Analysis Expert, specialize in analyzing knowledge videos, identifying and clearly explaining key points in {tgt_lang}, ensuring accurate, easy-to-understand summaries suitable for diverse audiences, analyze, list key points, and explain detailedly below text: "


def main():

    with gr.Blocks(css='style.css') as ocr_tr:
        gr.Markdown(OCR_TR_DESCRIPTION)

        # -------------- OCR 文字提取 --------------
        with gr.Box():

            with gr.Row():
                gr.Markdown("### Step 01: Text Extraction")

            with gr.Row():
                with gr.Column():
                    with gr.Row():
                        inputs_img = gr.Image(image_mode="RGB", source="upload", type="pil", label="image")
                    with gr.Row():
                        inputs_lang = gr.CheckboxGroup(choices=["chi_sim", "eng"],
                                                       type="value",
                                                       value=['eng'],
                                                       label='language')

                    with gr.Row():
                        clear_img_btn = gr.Button('Clear')
                        ocr_btn = gr.Button(value='OCR Extraction', variant="primary")

                    with gr.Row():
                        # Use Markdown to display clickable URL
                        gr.Markdown("[Click here to get API key](https://makersuite.google.com/u/1/app/apikey)")

                    with gr.Row():
                        # Create a text input box for users to enter their API key
                        inputs_api_key = gr.Textbox(label="Please enter your API key here", type="password")

                with gr.Column():
                    with gr.Row():
                        outputs_text = gr.Textbox(label="Extract content", lines=20)
                    src_lang = gr.inputs.Dropdown(choices=["Chinese (Simplified)", "Chinese (Traditional)", "English", "Japanese", "Korean"], 
                                                           default="English", label='source language')
                    tgt_lang = gr.inputs.Dropdown(choices=["Chinese (Simplified)", "Chinese (Traditional)", "English", "Japanese", "Korean"], 
                                                           default="Chinese (Traditional)", label='target language')
                    with gr.Row():
                        clear_text_btn = gr.Button('Clear')
                        translate_btn = gr.Button(value='Translate', variant="primary")
                        summary_btn = gr.Button(value='Summary', variant="primary")


            with gr.Row():
                pass

        # -------------- 翻译 --------------
        with gr.Box():

            with gr.Row():
                gr.Markdown("### Step 02: Process")

            with gr.Row():
                outputs_tr_text = gr.Textbox(label="Process Content", lines=20)

            with gr.Row():
                cp_clear_btn = gr.Button(value='Clear Clipboard')
                cp_btn = gr.Button(value='Copy to clipboard', variant="primary")

        # ---------------------- OCR Tesseract ----------------------
        ocr_btn.click(fn=ocr_tesseract, inputs=[inputs_img, inputs_lang], outputs=[
            outputs_text,])
        clear_img_btn.click(fn=clear_content, inputs=[], outputs=[inputs_img])

        # ---------------------- 翻译 ----------------------
        translate_btn.click(fn=process_and_translate, inputs=[inputs_api_key, outputs_text, src_lang, tgt_lang], outputs=[outputs_tr_text])
        summary_btn.click(fn=process_and_summary, inputs=[inputs_api_key, outputs_text, src_lang, tgt_lang], outputs=[outputs_tr_text])
        clear_text_btn.click(fn=clear_content, inputs=[], outputs=[outputs_text])

        # ---------------------- 复制到剪贴板 ----------------------
        cp_btn.click(fn=cp_text, inputs=[outputs_tr_text], outputs=[])
        cp_clear_btn.click(fn=cp_clear, inputs=[], outputs=[])
        

    ocr_tr.launch(inbrowser=True)

if __name__ == '__main__':
    main()