app.py

import gradio as gr
import json
import tempfile
import os
import difflib

def compare_texts(correct_file, wrong_file):
    """讀取並返回文件內容"""
    correct_text = correct_file.decode('utf-8')
    wrong_text = wrong_file.decode('utf-8')

    # 比較兩個文本並找出不同的位置
    s = difflib.SequenceMatcher(None, wrong_text, correct_text)
    diff_positions = []
    correct_text_with_positions = ""
    wrong_text_with_positions = ""
    wrong_ids = []

    for i, (tag, i1, i2, j1, j2) in enumerate(s.get_opcodes()):
        if tag == "equal":
            correct_text_with_positions += correct_text[j1:j2]
            wrong_text_with_positions += wrong_text[i1:i2]
        else:
            correct_text_segment = correct_text[j1:j2]
            wrong_text_segment = wrong_text[i1:i2]
            # 對於不一緻的部分，添加標記
            wrong_ids.extend(range(i1, i2))
            correct_text_with_positions += f"[{correct_text_segment}]"
            wrong_text_with_positions += f"[{wrong_text_segment}]"

    wrong_ids_str = ",".join(map(str, wrong_ids))

    return correct_text_with_positions, wrong_text_with_positions, wrong_ids_str

def generate_json(file_id, correct_text, wrong_text, wrong_ids):
    """根據輸入生成 JSON 數據，並保存為文件"""
    wrong_ids_list = [int(x) for x in wrong_ids.split(',')]
    data = {
        "id": file_id,
        "original_text": wrong_text,
        "wrong_ids": wrong_ids_list,
        "correct_text": correct_text
    }
    json_data = json.dumps([data], ensure_ascii=False, indent=4)
    
    # 創建臨時文件來保存 JSON 數據
    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".json", mode='w', encoding='utf-8')
    temp_file.write(json_data)
    temp_file.close()
    
    return json_data, temp_file.name

with gr.Blocks() as demo:
    file_id_input = gr.Textbox(label="請輸入文件ID")
    # 使用 binary 作为文件类型，这样上传的文件将以二进制形式传递给函数
    correct_file = gr.File(label="上傳校正過的逐字稿文本文件", type="binary")
    wrong_file = gr.File(label="上傳未校正的ASR辨識文本文件", type="binary")
    
    compare_button = gr.Button("比較文本")
    correct_text_output = gr.TextArea(label="校正過的逐字稿文本內容")
    wrong_text_output = gr.TextArea(label="未校正的ASR辨識文本內容")
    wrong_ids_output = gr.Textbox(label="錯誤的文字位置")
    
    generate_button = gr.Button("生成 JSON 文件")
    json_output = gr.Text(label="JSON 輸出")
    json_download_link = gr.File(label="下載 JSON 文件")
    
    compare_button.click(
        compare_texts, 
        inputs=[correct_file, wrong_file], 
        outputs=[correct_text_output, wrong_text_output, wrong_ids_output]
    )
    generate_button.click(
        generate_json, 
        inputs=[file_id_input, correct_text_output, wrong_text_output, wrong_ids_output], 
        outputs=[json_output, json_download_link]
    )

demo.launch()