Create app/main.py

app/main.py

@@ -0,0 +1,148 @@
+import os
+import base64
+import requests
+import markdown
+import pdfkit
+import gradio as gr
+import re
+import random
+from datetime import datetime
+from jinja2 import Template
+import openai
+
+# —— 环境变量读取 —— #
+BAIDU_API_KEY    = os.getenv("BAIDU_API_KEY", "")
+BAIDU_SECRET_KEY = os.getenv("BAIDU_SECRET_KEY", "")
+OPENAI_API_KEY   = os.getenv("OPENAI_API_KEY", "")
+openai.api_key   = OPENAI_API_KEY
+
+# —— Baidu OCR —— #
+def get_access_token(api_key, secret_key):
+    resp = requests.post(
+        "https://aip.baidubce.com/oauth/2.0/token",
+        params={
+            "grant_type":"client_credentials",
+            "client_id": api_key,
+            "client_secret": secret_key
+        }
+    )
+    return resp.json().get("access_token")
+
+def ocr_image(image_bytes, token):
+    img_b64 = base64.b64encode(image_bytes).decode()
+    resp = requests.post(
+        f"https://aip.baidubce.com/rest/2.0/ocr/v1/handwriting?access_token={token}",
+        headers={"Content-Type":"application/x-www-form-urlencoded"},
+        data={"image": img_b64, "language_type":"ENG"}
+    )
+    return resp.json().get("words_result", [])
+
+# —— 文本高亮 —— #
+def highlight_brackets(text):
+    # 中括号绿色；小括号红色
+    text = re.sub(r'\[([^\[\]]+)\]', r'<span class="highlight-bracket-green">\1</span>', text)
+    text = re.sub(r'\(([^\(\)]+)\)', r'<span class="highlight-bracket">\1</span>', text)
+    return text.replace("\n", "<br>")
+
+# —— 主处理函数 —— #
+def process(image):
+    # 1. OCR
+    token = get_access_token(BAIDU_API_KEY, BAIDU_SECRET_KEY)
+    words = ocr_image(image.read(), token)
+    essay_text = "\n".join([w["words"] for w in words])
+
+    # 2. 格式化原文
+    fmt_prompt = (
+        "请帮我整理下面的英语作文文本格式，只整理英文正文部分，"
+        "保证原汁原味（明显错误空格换行、乱码、非常用字符改正），"
+        "拼写错误保留：\n\n" + essay_text
+    )
+    fm = openai.ChatCompletion.create(
+        model="gpt-4o-mini",
+        messages=[{"role":"user","content":fmt_prompt}]
+    )
+    revised = fm.choices[0].message.content
+
+    # 3. 批改
+    corr_prompt = (
+        "请帮我把下面的英语作文的语法错误改正，输出改正后的文章，"
+        "原文错误用()括起来，修改部分用[]括起来：\n\n" + revised
+    )
+    cm = openai.ChatCompletion.create(
+        model="gpt-4o-mini",
+        messages=[{"role":"user","content":corr_prompt}]
+    )
+    corrected = cm.choices[0].message.content
+
+    # 4. 批改意见
+    review_prompt = (
+        "下面是一份已经批改过的英语作文，请根据批注给出逐条批改意见：\n\n"
+        + corrected
+    )
+    rm = openai.ChatCompletion.create(
+        model="gpt-4o-mini",
+        messages=[{"role":"user","content":review_prompt}]
+    )
+    review = rm.choices[0].message.content
+
+    # 5. 评分
+    rate_prompt = (
+        "请按照 IELTS/CEFR 写作评价体系，从语言通顺度、连贯度、词汇与语法三维度打分并给出原因：\n\n"
+        + revised
+    )
+    rr = openai.ChatCompletion.create(
+        model="gpt-4o-mini",
+        messages=[{"role":"user","content":rate_prompt}]
+    )
+    rating = rr.choices[0].message.content
+
+    # 6. 优秀范文
+    rewrite_prompt = (
+        "请使用优秀表达重写下面这篇作文，加粗可供学习的部分：\n\n" + revised
+    )
+    wm = openai.ChatCompletion.create(
+        model="gpt-4o-mini",
+        messages=[{"role":"user","content":rewrite_prompt}]
+    )
+    perfect = wm.choices[0].message.content
+
+    # 7. 渲染 HTML
+    code = f"{random.randint(0,9999):04}-{datetime.now().strftime('%Y%m%d%H%M%S')}"
+    with open("app/templates/base.html", encoding="utf-8") as f:
+        tpl = Template(f.read())
+    html_content = (
+        "<h2>原文格式化</h2>"
+        + markdown.markdown(revised)
+        + "<h2>批改结果</h2>"
+        + highlight_brackets(corrected)
+        + "<h2>批改意见</h2>"
+        + markdown.markdown(review)
+        + "<h2>评分</h2>"
+        + markdown.markdown(rating)
+        + "<h2>优秀范文</h2>"
+        + markdown.markdown(perfect)
+    )
+    full_html = tpl.render(code=code, content=html_content)
+
+    # 写文件
+    html_path = f"/app/output/{code}.html"
+    pdf_path  = f"/app/output/{code}.pdf"
+    os.makedirs("/app/output", exist_ok=True)
+    with open(html_path, "w", encoding="utf-8") as f:
+        f.write(full_html)
+    pdfkit.from_string(full_html, pdf_path, options={"enable-local-file-access":""})
+
+    return full_html, html_path, pdf_path
+
+# —— Gradio 接口 —— #
+with gr.Blocks(title="英语作文批改") as demo:
+    gr.Markdown("## 上传英语作文照片，等待批改完成后下载 HTML 或 PDF")
+    image_in = gr.File(file_count="single", label="上传照片")
+    output_html = gr.HTML()
+    btn = gr.Button("开始批改")
+    file_html = gr.File(label="下载 HTML")
+    file_pdf  = gr.File(label="下载 PDF")
+    btn.click(fn=process, inputs=image_in, outputs=[output_html, file_html, file_pdf])
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)