import pandas as pd
import gradio as gr

def check_csv(file):
    if file is None:
        return "請上傳檔案！"
    
    try:
        df = pd.read_csv(file.name, encoding="utf-8-sig")
        df.columns = [col.strip() for col in df.columns]

        required_columns = {"Question", "Context", "Answer"}
        actual_columns = set(df.columns)

        if actual_columns != required_columns:
            return f"欄位錯誤：應包含欄位 {required_columns}，實際為 {actual_columns}"

        if df.shape[0] == 0:
            return "檔案中沒有資料列!"

        invalid_rows = df[df["Question"].notnull() & (df["Answer"].isnull() | df["Context"].isnull())]
        if len(invalid_rows) > 0:
            missing_questions = "\n".join(f"- 問題：{q}" for q in invalid_rows["Question"].tolist())
            return f"發現 {len(invalid_rows)} 筆資料中 Answer 或 Context 為空：\n{missing_questions}"

        return "檔案格式正確，資料內容完整"

    except Exception as e:
        return f"發生錯誤：{str(e)}"

demo = gr.Interface(
    fn=check_csv,
    inputs=gr.File(label="上傳 CSV 檔案"),
    outputs=gr.Textbox(label="檢查結果"),
    title="繳交檔案 (Evaluation_Dataset.csv) 檢查工具",
    description="請上傳一個包含`Question`, `Context`, `Answer`欄位的 CSV 檔案，以檢查您的檔案格式是否正確。",
    flagging_mode="never"
)

demo.launch()