app.py

import functools
import re
import gradio as gr

try:
    import pandas as pd
    USE_PANDAS = True
except ImportError:
    USE_PANDAS = False

from WordStoryAudioMaker import audioStoryMaker


def parse_phrases_text(phrases_text: str):
    parts = re.split(r'[,\n]+', phrases_text)
    return [p.strip() for p in parts if p.strip()]


def bold_to_html(text: str) -> str:
    if not text:
        return ""
    return re.sub(r"\*\*(.+?)\*\*", r"<b>\1</b>", text)


def dict_list_to_html(dict_list):
    if not dict_list:
        return "<p><i>（没有词典数据）</i></p>"
    col_map = {
        "dictKeyText": "单词或短语",
        "ukPhoneticSymbol": "英式音标",
        "usPhoneticSymbol": "美式音标",
        "definition": "释义",
        "example": "例句",
        "pos": "词性",
    }
    keys = [k for k in dict_list[0].keys() if k != "dictName"]
    for d in dict_list:
        if "definition" in d and isinstance(d["definition"], str):
            d["definition"] = d["definition"].replace("\n", "<br>")
    if USE_PANDAS:
        df = pd.DataFrame(dict_list)[keys]
        df.columns = [col_map.get(c, c) for c in df.columns]
        html = df.to_html(escape=False, index=False)
    else:
        header = "".join(f"<th>{col_map.get(k, k)}</th>" for k in keys)
        rows = []
        for d in dict_list:
            row = "".join(f"<td>{d.get(k, '')}</td>" for k in keys)
            rows.append(f"<tr>{row}</tr>")
        body = "\n".join(rows)
        html = f"""
        <table border='1' cellspacing='0' cellpadding='6' style='border-collapse:collapse;'>
            <thead style='background:#f5f5f5;'><tr>{header}</tr></thead>
            <tbody>{body}</tbody>
        </table>
        """
    return f"<div style='line-height:1.6;'>{html}</div>"


# ---------------------------
# 缓存实现部分
# ---------------------------
def _normalize_input(phrases_text: str) -> str:
    if phrases_text is None:
        return ""
    s = phrases_text.strip()
    s = re.sub(r'\r\n', '\n', s)
    s = re.sub(r'\n+', '\n', s)
    s = re.sub(r'[，,]+', '\n', s)
    return s


@functools.lru_cache(maxsize=128)
def _cached_audio_for_input(norm_phrases_text: str):
    phrases = parse_phrases_text(norm_phrases_text)
    phrasesListStr, audioFilePath, dictResultList, storyJson = audioStoryMaker(phrases)
    return phrasesListStr, audioFilePath, dictResultList, storyJson


def wrapper_run(phrases_text: str):
    phrases_norm = _normalize_input(phrases_text)
    if not phrases_norm:
        return "<p>（没有输入短语）</p>", None, "", "", ""
    phrasesListStr, audioFilePath, dictResultList, storyJson = _cached_audio_for_input(phrases_norm)
    story_en_html = bold_to_html(storyJson.get("en", "")) if isinstance(storyJson, dict) else ""
    story_zh_html = bold_to_html(storyJson.get("zh", "")) if isinstance(storyJson, dict) else ""
    dict_html = dict_list_to_html(dictResultList)
    story_en_html = f"<div style='line-height:1.6;'>{story_en_html}</div>"
    story_zh_html = f"<div style='line-height:1.6;'>{story_zh_html}</div>"
    phrases_html = f"<p><b>单词或短语：</b> {phrasesListStr}</p>"
    return phrases_html, audioFilePath, story_zh_html, story_en_html, dict_html


# ---------------------------
# Gradio 界面部分
# ---------------------------
default_phrases = """nobility
underperform
insane
liable
conspiracy
proprietary
ditch"""

with gr.Blocks(title="单词故事") as demo:
    gr.Markdown("""## 单词故事：听故事 = 复习 + 记忆单词！  
单词故事利用 AI 将用户提供的一组单词或短语编成一个有趣的小故事，然后合成音频，使用户可以通过听故事来复习之前记过的单词——将枯燥的单词记忆变得轻松有趣起来！""")
    inp = gr.Textbox(
        label="""输入单词或短语（每行一个），建议每次提交7个，这是由George A. Miller (1956) 的论文The Magical Number Seven, Plus or Minus Two所揭示的记忆规律。
        下面默认的内容是为了方便测试用，您可以删除后使用你的""",
        placeholder="在这里输入英文单词或短语",
        value=default_phrases,
        lines=7
    )
    btn = gr.Button("生成")
    out_phrasesListStr = gr.HTML(label="短语列表")
    out_audio = gr.Audio(label="单词故事音频（播放）", type="filepath", autoplay=True)
    out_storyZh = gr.HTML(label="中文故事")
    out_storyEn = gr.HTML(label="英文故事")
    out_dictTable = gr.HTML(label="词典释义表格")

    btn.click(
        fn=wrapper_run,
        inputs=[inp],
        outputs=[
            out_phrasesListStr,
            out_audio,
            out_storyZh,
            out_storyEn,
            out_dictTable
        ]
    )

if __name__ == "__main__":
    demo.launch()