Initial commit

.github/workflows/build-windows.yml

@@ -0,0 +1,58 @@
+# 工作流的名称
+name: Build Windows Executable
+
+# 触发条件：
+# 1. 当代码被推送到 main 分支时
+# 2. 当你创建一个新的 "Tag" (用于发布新版本)
+# 3. 允许你在GitHub页面手动触发
+on:
+  push:
+    branches: [ "main" ]
+  release:
+    types: [ created ]
+  workflow_dispatch:
+
+jobs:
+  build:
+    # 指定运行环境为最新的Windows服务器
+    runs-on: windows-latest
+
+    steps:
+    # 第1步：检出你的代码
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    # 第2步：设置Python环境
+    - name: Set up Python
+      uses: actions/setup-python@v5
+      with:
+        python-version: '3.11' # 你可以根据需要更改Python版本
+
+    # 第3步：安装依赖包
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install -r requirements.txt
+        pip install pyinstaller
+
+    # 第4步：使用PyInstaller打包
+    # --noconsole: 这是一个GUI应用，不显示控制台窗口
+    # --onefile: 打包成单个.exe文件
+    # --add-data: 将 config.json 文件包含到exe中
+    # --name: 指定生成的可执行文件名
+    - name: Build with PyInstaller
+      run: pyinstaller --noconsole --onefile --name "AI-Essay-Corrector" --add-data "config.json;." main.py
+
+    # 第5步：将打包好的.exe上传，以便在工作流页面下载 (适合测试)
+    - name: Upload artifact for testing
+      uses: actions/upload-artifact@v4
+      with:
+        name: AI-Essay-Corrector-Windows-exe
+        path: dist/AI-Essay-Corrector.exe
+
+    # 第6步 (专业级开源发布): 当你创建Tag时，自动创建Release并附上.exe
+    - name: Create Release and Upload Asset
+      if: startsWith(github.ref, 'refs/tags/') # 仅在创建Tag时运行此步骤
+      uses: softprops/action-gh-release@v2
+      with:
+        files: dist/AI-Essay-Corrector.exe

.gitignore

@@ -0,0 +1,29 @@
+# Python virtual environment
+venv/
+.venv/
+env/
+ENV/
+
+# Python cache files
+__pycache__/
+*.pyc
+*.pyo
+*.pyd
+
+# PyInstaller build artifacts
+dist/
+build/
+*.spec
+
+# IDE-specific files
+.vscode/
+.idea/
+
+# OS-specific files
+.DS_Store
+Thumbs.db
+
+# Local configuration file
+# This file is generated by the application and contains user-specific secrets.
+# It should not be committed to version control.
+config.json

api_services.py

@@ -0,0 +1,163 @@
+import base64
+from typing import Dict, Any
+from config_manager import ConfigManager
+import os
+import mimetypes
+import re
+from openai import OpenAI
+
+# 定义默认的LLM Prompt模板。使用`.format()`方法进行后续的动态填充。
+DEFAULT_LLM_PROMPT_TEMPLATE = """# ESSAY TOPIC
+{topic}
+
+# INSTRUCTIONS FOR AI (Process in English)
+## 1. ROLE & GOAL
+You are a highly experienced senior high school English teacher. Your task is to provide a detailed, constructive, and encouraging evaluation of a student's essay.
+## 2. INPUT DATA
+You will receive a quantitative `<wscore>` and the full `<text>` of the essay. The essay is based on the topic provided above.
+## 3. GRADING LOGIC (Total Score: 15 points)
+- **Content & Language (12 points):** Evaluate this based on grammar, vocabulary, sentence structure, etc., in relation to the essay topic.
+- **Handwriting & Presentation (3 points):** Calculate the score by first getting a raw score (`Raw Score = wscore * 3`), and then rounding the `Raw Score` **up** to the nearest half-point (0.5).
+    - *Rounding Logic Example:* A raw score of 2.49 becomes 2.5. A raw score of 2.51 becomes 3.0. A raw score of 2.50 remains 2.5. A score of 0 remains 0.
+## 4. FINAL TASK
+Analyze the text, calculate scores, and present your feedback in **Simplified Chinese** using the precise Markdown format specified below.
+#--- End of English Instructions ---
+# OUTPUT SPECIFICATION (MUST BE IN SIMPLIFIED CHINESE)
+# 请严格使用以下Markdown格式，并用简体中文填充所有内容，优点可以两个到三个，问题建议要把全部问题找出来并且解析，都要遵循类似格式。
+
+
+###【作文内容】
+*   **作文文本:** [在此处粘贴完整的作文文本。]
+### 【综合评价】
+(在此处用一两句鼓励性的话，对本次作文进行总体概述。)
+### 【亮点与优点】
+*   **(优点1):** [具体描述作文内容或语言上的一个亮点。]
+*   **(优点2):** [具体描述另一个优点。]
+*   **(优点3):（以此类推，不限制数量，但建议控制在3个以内。）
+### 【问题与修改建议】
+*   **[问题1 - 语法/拼写错误]:**
+    *   **原文句子:** "[引用出现错误的原文句子]"
+    *   **问题分析:** [简要说明错误类型。]
+    *   **修改建议:** "[写出修改后的正确句子]"
+*   **[问题2 - 表达/逻辑]:**
+    *   **原文句子:** "[引用表达欠佳的原文句子]"
+    *   **问题分析:** [说明问题所在。]
+    *   **修改建议:** "[提供一个更好的表达方式。]
+*   **[问题3 - （以此类推，不限制数量，但建议控制在3个以内。）]:**
+    *   **原文句子:** "[说明问题所在。]"
+    *   **问题分析:** [说明问题所在。]
+### 【分数评估】
+*   **内容与语言分 (Content & Language):** [分数] / 12
+*   **卷面与书写分 (Handwriting & Presentation):** [分数] / 3
+*   ---
+*   **最终得分 (Final Score):** **[总分] / 15**
+
+# INPUT DATA FOR THIS TASK
+
+<wscore>{wscore}</wscore>
+<text>
+{essay_text}
+</text>
+"""
+
+class ApiService:
+    """封装了与外部API（VLM和LLM）交互的所有逻辑。"""
+    def __init__(self, config_manager: ConfigManager):
+        self.config = config_manager
+
+    def _encode_image_to_base64_url(self, image_path: str) -> str:
+        """将本地图片文件编码为Base64数据URL。"""
+        if not os.path.exists(image_path):
+            raise FileNotFoundError(f"Image file not found at: {image_path}")
+        mime_type, _ = mimetypes.guess_type(image_path)
+        if not mime_type or not mime_type.startswith('image'):
+            raise ValueError(f"File is not a recognizable image type: {mime_type}")
+        with open(image_path, "rb") as image_file:
+            encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
+        return f"data:{mime_type};base64,{encoded_string}"
+
+    def process_essay_image(self, file_path: str, topic: str) -> str:
+        """
+        执行完整的两步式作文批改流程：
+        1. VLM调用：分析作文图片，提取手写文本和书写质量分数。
+        2. LLM调用：基于VLM的输出和作文题目，生成详细的批改报告。
+        """
+        # --- 步骤 1: 调用VLM进行图像分析 ---
+        vlm_client = OpenAI(
+            api_key=self.config.get("VlmApiKey"),
+            base_url=self.config.get("VlmUrl")
+        )
+        base64_image_url = self._encode_image_to_base64_url(file_path)
+
+        vlm_prompt = """# ROLE
+You are a high-precision OCR (Optical Character Recognition) and handwriting analysis engine. Your only job is to analyze the provided image and output structured data. Do not add any conversational text or explanations.
+# TASK
+Analyze the handwriting quality and extract all text from the image.
+## 1. Handwriting Quality Analysis:
+- Critically evaluate the handwriting on a continuous scale from 0.0 to 1.0.
+- The scoring must be stringent. A score of 1.0 is reserved for flawless, machine-printed-like perfection, which is virtually unattainable.
+- **Score Tiers:**
+    - **0.90-0.99:** Near-perfect, professional calligrapher level. Extremely rare.
+    - **0.80-0.89:** Excellent, clear, consistent, and aesthetically pleasing. The best a top student can achieve.
+    - **0.70-0.79:** Good and very legible, but with minor inconsistencies in size or spacing.
+    - **0.60-0.69:** Clear and legible, but with noticeable inconsistencies.
+    - **Below 0.60:** Legibility is impacted.
+- Output this score enclosed in a single <wscore> XML tag.
+## 2. Full Text Extraction:
+- Perform a high-accuracy OCR on the entire image.
+- Preserve the original line breaks and paragraph structure as best as possible.
+- Output the full extracted text enclosed in a single <text> XML tag.
+# OUTPUT FORMAT
+Strictly adhere to the following format. Do not output anything else.
+<wscore>[Your calculated score, e.g., 0.85]</wscore>
+<text>
+[The full extracted text from the image goes here.]
+</text>"""
+        vlm_messages = [{"role": "user", "content": [{"type": "text", "text": vlm_prompt}, {"type": "image_url", "image_url": {"url": base64_image_url}}]}]
+        
+        vlm_model = self.config.get("VlmModel", "gemini-2.5-pro")
+        vlm_response = vlm_client.chat.completions.create(model=vlm_model, messages=vlm_messages, max_tokens=4096, temperature=1)
+        vlm_output = vlm_response.choices[0].message.content or ""
+        
+        # 解析VLM返回的XML格式输出，提取分数和文本
+        wscore_match = re.search(r'<wscore>(.*?)</wscore>', vlm_output, re.DOTALL)
+        text_match = re.search(r'<text>(.*?)</text>', vlm_output, re.DOTALL)
+        original_wscore = float(wscore_match.group(1).strip()) if wscore_match else 0.0
+        essay_text = text_match.group(1).strip() if text_match else "错误：无法从图片中提取文本。"
+        
+        try:
+            sensitivity_factor = float(self.config.get("SensitivityFactor", "1.0"))
+        except (ValueError, TypeError):
+            # 如果配置的敏感度因子无效，则使用默认值1.0
+            sensitivity_factor = 1.0 
+            
+        wscore = original_wscore ** sensitivity_factor
+
+        if not text_match:
+            raise ValueError(f"VLM未能按预期格式返回，无法解析文本。模型返回：\n{vlm_output}")
+
+        # --- 步骤 2: 调用LLM生成批改报告 ---
+        llm_client = OpenAI(
+            api_key=self.config.get("LlmApiKey"),
+            base_url=self.config.get("LlmUrl")
+        )
+        
+        # 从配置加载Prompt模板，若用户未定义则使用默认模板
+        prompt_template = self.config.get("LlmPromptTemplate")
+        if not prompt_template:
+            prompt_template = DEFAULT_LLM_PROMPT_TEMPLATE
+        
+        # 使用作文题目、书写分数和识别出的文本填充Prompt模板
+        final_llm_prompt = prompt_template.format(
+            topic=topic,
+            wscore=wscore,
+            essay_text=essay_text
+        )
+        
+        llm_messages = [{"role": "user", "content": final_llm_prompt}]
+
+        llm_model = self.config.get("LlmModel", "gemini-2.5-pro")
+        llm_response = llm_client.chat.completions.create(model=llm_model, messages=llm_messages, temperature=1, max_tokens=16384)
+        final_report = llm_response.choices[0].message.content or "错误：AI未能生成报告。"
+
+        return final_report

app_ui.py

@@ -0,0 +1,387 @@
+import tkinter as tk
+from tkinter import ttk, filedialog, messagebox
+import threading
+import queue
+import os
+from typing import List
+import concurrent.futures
+from config_manager import ConfigManager
+from api_services import ApiService, DEFAULT_LLM_PROMPT_TEMPLATE
+
+class AboutDialog(tk.Toplevel):
+    """“关于”对话框，展示应用信息，支持滚动查看。"""
+    def __init__(self, parent):
+        super().__init__(parent)
+        self.transient(parent)
+        self.title("关于 AI 作文批改助手")
+        # 设置一个适合滚动的默认窗口尺寸
+        self.geometry("450x400")
+
+        # 主框架，用于容纳文本和滚动条
+        main_frame = ttk.Frame(self)
+        main_frame.pack(fill=tk.BOTH, expand=True, padx=10, pady=10)
+        main_frame.grid_rowconfigure(0, weight=1)
+        main_frame.grid_columnconfigure(0, weight=1)
+
+        # 使用Text控件以支持长文本和滚动条
+        text_widget = tk.Text(main_frame, wrap="word", relief="flat", spacing1=5, spacing3=5)
+        text_widget.grid(row=0, column=0, sticky="nsew")
+
+        # 创建并关联垂直滚动条
+        scrollbar = ttk.Scrollbar(main_frame, orient="vertical", command=text_widget.yview)
+        scrollbar.grid(row=0, column=1, sticky="ns")
+        text_widget.config(yscrollcommand=scrollbar.set)
+
+        about_text = """
+欢迎使用 AI 作文批改助手！这是一款专为教育者和学生设计的智能工具，旨在利用前沿的人工智能技术，提供高效、精准、个性化的英文作文批改体验。
+
+核心亮点:
+- **双核AI引擎:** 采用先进的两步式处理流程。首先由专用的视觉语言模型(VLM)对作文图片进行高精度手写识别(OCR)与专业的书写质量评估；随后，强大的大语言模型(LLM)会结合识别出的文本、作文题目以及书写评分，进行深度、全面的分析与批改。
+
+- **极致的灵活性与可配置性:**
+  * **服务分离:** 您可以为VLM和LLM设置完全独立的API服务地址、密钥和模型名称，轻松适配不同的AI提供商（需要兼容OpenAI格式）或自建服务。
+  * **逻辑定制:** 书写评分的“敏感度因子”可在设置中调整，以适应不同年级或要求的评分标准。
+  * **模板开放:** 核心的LLM批改指令模板（Prompt）完全开放给用户。您可以在设置中自由修改，调整评分维度、总分、反馈风格等，实现高度个性化的批改要求。
+
+- **闪电般的并发处理:** 内置高效的多线程并发引擎，无论您选择一张还是上百张图片，程序都能同时处理，大幅缩短批量批改的等待时间。最大并发任务数亦可在设置中自由调整。
+
+- **企业级的安全保障:** 我们深知API密钥的敏感性。所有密钥信息在保存到本地配置文件时，均经过强大的加密算法处理，有效防止明文泄露，保障您的账户安全。
+
+- **人性化的评分策略:** 卷面书写分采用更符合教学直觉的“向上取整至0.5分”规则，确保评分结果既精确又公平。
+
+使用说明:
+1. **初次配置:** 点击“设置”，分别填入您的VLM和LLM服务提供商的URL、API密钥和模型名称。
+2. **输入题目:** 在主界面上方的文本框中，输入本次批改的“作文题目”。
+3. **选择文件:** 点击“选择图片”，一次性选择所有需要批改的学生作文图片。
+4. **开始批改:** 点击“开始批改”，程序将自动在后台进行并发处理，您可以在日志区看到实时进度。
+5. **获取报告:** 任务完成后，每一张图片对应的Markdown格式详细批改报告，都会自动生成在原图片所在的目录下。
+
+作者: Eric_Terminal
+https://github.com/Eric-Terminal
+版本: 2.4
+"""
+        
+        text_widget.insert("1.0", about_text)
+        # 将文本设置为只读，防止用户修改
+        text_widget.config(state="disabled")
+
+        # 放置“关闭”按钮的框架
+        btn_frame = ttk.Frame(main_frame)
+        btn_frame.grid(row=1, column=0, columnspan=2, sticky="e", pady=(10, 0))
+        close_button = ttk.Button(btn_frame, text="关闭", command=self.destroy)
+        close_button.pack()
+
+        self.protocol("WM_DELETE_WINDOW", self.destroy)
+        self.grab_set()
+        self.wait_window(self)
+
+
+class SettingsDialog(tk.Toplevel):
+    """“设置”对话框，允许用户配置VLM、LLM服务及其他应用参数。"""
+    def __init__(self, parent, current_config: dict):
+        super().__init__(parent)
+        self.transient(parent)
+        self.title("设置")
+        self.result = None
+
+        # 为VLM和LLM服务分别创建Tkinter字符串变量
+        self.vlm_url = tk.StringVar(value=current_config.get("VlmUrl", ""))
+        self.vlm_api_key = tk.StringVar(value=current_config.get("VlmApiKey", ""))
+        self.vlm_model = tk.StringVar(value=current_config.get("VlmModel", "gemini-2.5-pro"))
+        self.llm_url = tk.StringVar(value=current_config.get("LlmUrl", ""))
+        self.llm_api_key = tk.StringVar(value=current_config.get("LlmApiKey", ""))
+        self.llm_model = tk.StringVar(value=current_config.get("LlmModel", "gemini-2.5-pro"))
+        self.sensitivity_factor = tk.StringVar(value=current_config.get("SensitivityFactor", "1.5"))
+        self.max_workers = tk.StringVar(value=current_config.get("MaxWorkers", "4"))
+        
+        # 智能加载Prompt模板：优先使用用户自定义模板，否则使用默认模板
+        user_template = current_config.get("LlmPromptTemplate")
+        self.llm_prompt_template_str = user_template if user_template else DEFAULT_LLM_PROMPT_TEMPLATE
+
+
+        frame = ttk.Frame(self, padding="10")
+        frame.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S))
+        frame.grid_columnconfigure(0, weight=1)
+
+        # VLM服务设置区域
+        vlm_frame = ttk.LabelFrame(frame, text="VLM (视觉模型) 设置", padding="10")
+        vlm_frame.grid(row=0, column=0, sticky=(tk.W, tk.E), pady=5)
+        vlm_frame.grid_columnconfigure(1, weight=1)
+        ttk.Label(vlm_frame, text="VLM URL:").grid(column=0, row=0, sticky=tk.W, pady=2)
+        ttk.Entry(vlm_frame, textvariable=self.vlm_url, width=40).grid(column=1, row=0, sticky=(tk.W, tk.E))
+        ttk.Label(vlm_frame, text="VLM API Key:").grid(column=0, row=1, sticky=tk.W, pady=2)
+        ttk.Entry(vlm_frame, textvariable=self.vlm_api_key, width=40).grid(column=1, row=1, sticky=(tk.W, tk.E))
+        ttk.Label(vlm_frame, text="VLM 模型:").grid(column=0, row=2, sticky=tk.W, pady=2)
+        ttk.Entry(vlm_frame, textvariable=self.vlm_model, width=40).grid(column=1, row=2, sticky=(tk.W, tk.E))
+
+        # LLM服务设置区域
+        llm_frame = ttk.LabelFrame(frame, text="LLM (语言模型) 设置", padding="10")
+        llm_frame.grid(row=1, column=0, sticky=(tk.W, tk.E), pady=5)
+        llm_frame.grid_columnconfigure(1, weight=1)
+        ttk.Label(llm_frame, text="LLM URL:").grid(column=0, row=0, sticky=tk.W, pady=2)
+        ttk.Entry(llm_frame, textvariable=self.llm_url, width=40).grid(column=1, row=0, sticky=(tk.W, tk.E))
+        ttk.Label(llm_frame, text="LLM API Key:").grid(column=0, row=1, sticky=tk.W, pady=2)
+        ttk.Entry(llm_frame, textvariable=self.llm_api_key, width=40).grid(column=1, row=1, sticky=(tk.W, tk.E))
+        ttk.Label(llm_frame, text="LLM 模型:").grid(column=0, row=2, sticky=tk.W, pady=2)
+        ttk.Entry(llm_frame, textvariable=self.llm_model, width=40).grid(column=1, row=2, sticky=(tk.W, tk.E))
+
+        # 其他应用参数设置区域
+        other_frame = ttk.LabelFrame(frame, text="其他设置", padding="10")
+        other_frame.grid(row=2, column=0, sticky=(tk.W, tk.E), pady=5)
+        other_frame.grid_columnconfigure(1, weight=1)
+        ttk.Label(other_frame, text="手写打分敏感度:").grid(column=0, row=0, sticky=tk.W, pady=2)
+        ttk.Entry(other_frame, textvariable=self.sensitivity_factor, width=40).grid(column=1, row=0, sticky=(tk.W, tk.E))
+        ttk.Label(other_frame, text="最大并发数:").grid(column=0, row=1, sticky=tk.W, pady=2)
+        ttk.Entry(other_frame, textvariable=self.max_workers, width=40).grid(column=1, row=1, sticky=(tk.W, tk.E))
+        
+        # LLM Prompt模板编辑区域
+        prompt_frame = ttk.LabelFrame(frame, text="LLM Prompt 模板 (可在此修改，请勿修改{}占位符内容导致程序参数无法正常传递，通常情况下修改总分即可)", padding="10")
+        prompt_frame.grid(row=3, column=0, sticky=(tk.W, tk.E), pady=5)
+        prompt_frame.grid_columnconfigure(0, weight=1)
+        prompt_frame.grid_rowconfigure(0, weight=1)
+        
+        self.llm_prompt_text = tk.Text(prompt_frame, height=10, wrap="word")
+        self.llm_prompt_text.grid(row=0, column=0, sticky="nsew")
+        prompt_scrollbar = ttk.Scrollbar(prompt_frame, orient="vertical", command=self.llm_prompt_text.yview)
+        prompt_scrollbar.grid(row=0, column=1, sticky="ns")
+        self.llm_prompt_text.config(yscrollcommand=prompt_scrollbar.set)
+        self.llm_prompt_text.insert("1.0", self.llm_prompt_template_str)
+
+        btn_frame = ttk.Frame(frame)
+        btn_frame.grid(row=4, column=0, sticky=tk.E, pady=10)
+        ttk.Button(btn_frame, text="确定", command=self.on_ok).pack(side=tk.LEFT, padx=5)
+        ttk.Button(btn_frame, text="关闭", command=self.on_close).pack(side=tk.LEFT)
+
+        self.protocol("WM_DELETE_WINDOW", self.on_close)
+        self.grab_set()
+        self.wait_window(self)
+
+    def on_ok(self):
+        self.result = {
+            "VlmUrl": self.vlm_url.get(),
+            "VlmApiKey": self.vlm_api_key.get(),
+            "VlmModel": self.vlm_model.get(),
+            "LlmUrl": self.llm_url.get(),
+            "LlmApiKey": self.llm_api_key.get(),
+            "LlmModel": self.llm_model.get(),
+            "SensitivityFactor": self.sensitivity_factor.get(),
+            "MaxWorkers": self.max_workers.get(),
+            "LlmPromptTemplate": self.llm_prompt_text.get("1.0", "end-1c") # 从Text控件获取用户修改后的Prompt模板
+        }
+        self.destroy()
+
+    def on_close(self):
+        self.result = None
+        self.destroy()
+
+
+class MainApp:
+    """应用主窗口类，负责构建UI界面、处理用户交互和协调后台服务。"""
+    def __init__(self, root: tk.Tk, config_manager: ConfigManager, api_service: ApiService):
+        self.root = root
+        self.config_manager = config_manager
+        self.api_service = api_service
+        
+        self.file_paths: List[str] = []
+        self.is_file_selected = False
+        self.ui_queue = queue.Queue()
+        self.topic_input = None
+        self.processed_count = 0
+        self.lock = threading.Lock()
+
+        self._setup_ui()
+        self._initialize_config()
+        self.root.after(100, self._process_ui_queue)
+
+    def _setup_ui(self):
+        """初始化和布局主窗口的所有UI组件。"""
+        self.root.title("AI 作文批改助手")
+        self.root.geometry("550x450")
+        main_frame = ttk.Frame(self.root, padding="10")
+        main_frame.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S))
+        self.root.grid_columnconfigure(0, weight=1)
+        self.root.grid_rowconfigure(0, weight=1)
+
+        # 顶部进度条
+        top_frame = ttk.Frame(main_frame)
+        top_frame.grid(row=0, column=0, columnspan=2, sticky=(tk.W, tk.E), pady=5)
+        top_frame.grid_columnconfigure(0, weight=1)
+        self.progress_bar = ttk.Progressbar(top_frame, orient="horizontal", mode="determinate")
+        self.progress_bar.grid(row=0, column=0, sticky=(tk.W, tk.E))
+
+        # 左侧控制按钮
+        left_frame = ttk.Frame(main_frame)
+        left_frame.grid(row=1, column=0, sticky=(tk.N, tk.W), padx=(0, 10))
+        ttk.Button(left_frame, text="选择图片", command=self._open_file_dialog).pack(fill=tk.X, pady=5)
+        ttk.Button(left_frame, text="开始批改", command=self._start_processing).pack(fill=tk.X, pady=5)
+        ttk.Button(left_frame, text="设置", command=self._open_settings_dialog).pack(fill=tk.X, pady=5)
+        ttk.Button(left_frame, text="关于", command=self._open_about_dialog).pack(fill=tk.X, pady=5)
+
+        # 右侧输入和日志区域
+        right_frame = ttk.Frame(main_frame)
+        right_frame.grid(row=1, column=1, sticky=(tk.W, tk.E, tk.N, tk.S))
+        main_frame.grid_columnconfigure(1, weight=1)
+        main_frame.grid_rowconfigure(1, weight=1)
+
+        # 配置右侧框架的网格权重，使组件能自适应缩放
+        right_frame.grid_rowconfigure(0, weight=7)  # 作文题目框占7份
+        right_frame.grid_rowconfigure(1, weight=3)  # 日志框占3份
+        right_frame.grid_columnconfigure(0, weight=1)
+
+        # 作文题目输入框
+        self.topic_input = tk.Text(right_frame, wrap="word")
+        self.topic_input.grid(row=0, column=0, sticky="nsew", pady=(0, 5))
+        
+        self.topic_input.insert("1.0", "（在此输入作文题目）")
+        self.topic_input.config(fg="grey")
+
+        # 实现输入框的占位符（placeholder）效果
+        def on_focus_in(event):
+            if self.topic_input.get("1.0", "end-1c").strip() == "（在此输入作文题目）":
+                self.topic_input.delete("1.0", tk.END)
+                self.topic_input.config(fg="black")
+
+        def on_focus_out(event):
+            if not self.topic_input.get("1.0", "end-1c").strip():
+                self.topic_input.insert("1.0", "（在此输入作文题目）")
+                self.topic_input.config(fg="grey")
+
+        self.topic_input.bind("<FocusIn>", on_focus_in)
+        self.topic_input.bind("<FocusOut>", on_focus_out)
+
+        # 日志输出框（带滚动条）
+        listbox_frame = ttk.Frame(right_frame)
+        listbox_frame.grid(row=1, column=0, sticky="nsew")
+        listbox_frame.grid_rowconfigure(0, weight=1)
+        listbox_frame.grid_columnconfigure(0, weight=1)
+
+        self.listbox = tk.Listbox(listbox_frame)
+        self.listbox.grid(row=0, column=0, sticky="nsew")
+        
+        scrollbar = ttk.Scrollbar(listbox_frame, orient="vertical", command=self.listbox.yview)
+        scrollbar.grid(row=0, column=1, sticky="ns")
+        
+        self.listbox.config(yscrollcommand=scrollbar.set)
+
+    def _log(self, message: str):
+        """将消息添加到日志列表框并滚动到底部。"""
+        self.listbox.insert(tk.END, message)
+        self.listbox.see(tk.END)
+
+    def _initialize_config(self):
+        """应用启动时检查配置完整性，如果配置不完整则强制用户设置。"""
+        is_ok, _ = self.config_manager.check_settings()
+        if not is_ok:
+            self._show_config_dialog_until_valid()
+
+    def _show_config_dialog_until_valid(self):
+        """循环显示设置对话框，直到所有必需配置项都已填写。"""
+        while True:
+            self._open_settings_dialog()
+            is_ok, missing_item = self.config_manager.check_settings()
+            if is_ok:
+                return
+            if messagebox.askretrycancel("配置未完成", f"请配置: {missing_item}") == "cancel":
+                self.root.quit()
+                return
+
+    def _open_settings_dialog(self):
+        """打开设置对话框，并根据返回结果更新和保存配置。"""
+        dialog = SettingsDialog(self.root, self.config_manager.config)
+        if dialog.result:
+            # 清理旧的、统一的AI配置和OCR配置，以兼容新版分离的配置
+            self.config_manager.config.pop("AiUrl", None)
+            self.config_manager.config.pop("AiApiKey", None)
+            self.config_manager.config.pop("OcrApiKey", None)
+            self.config_manager.config.pop("OcrSecretKey", None)
+            for key, value in dialog.result.items():
+                self.config_manager.set(key, value)
+            self.config_manager.save()
+
+    def _open_about_dialog(self):
+        """创建并显示“关于”对话框。"""
+        AboutDialog(self.root)
+
+    def _open_file_dialog(self):
+        """打开文件选择对话框，让用户选择一个或多个图片文件。"""
+        paths = filedialog.askopenfilenames(title="选择作文图片", filetypes=[("图片文件", "*.jpg *.jpeg *.png *.bmp")])
+        if paths:
+            self.file_paths = paths
+            self.is_file_selected = True
+            self._log(f"已选择 {len(paths)} 个文件")
+        else:
+            self._log("取消选择")
+
+    def _start_processing(self):
+        """启动作文批改流程。"""
+        if not self.is_file_selected:
+            messagebox.showerror("操作错误", "请先选择文件")
+            return
+        
+        topic = self.topic_input.get("1.0", tk.END).strip()
+        if not topic or topic == "（在此输入作文题目）":
+            messagebox.showerror("操作错误", "请输入作文题目")
+            return
+
+        # 重置进度条和计数器
+        self.progress_bar['value'] = 0
+        self.progress_bar['maximum'] = len(self.file_paths)
+        self.processed_count = 0
+        
+        # 在后台线程中启动并发处理
+        thread = threading.Thread(target=self._concurrent_worker_manager, args=(self.file_paths, topic), daemon=True)
+        thread.start()
+
+    def _process_ui_queue(self):
+        """定期检查UI更新队列，并执行相应的UI操作（如记日志、更新进度条）。"""
+        try:
+            while True:
+                task, data = self.ui_queue.get_nowait()
+                if task == "log": 
+                    self._log(data)
+                elif task == "progress":
+                    with self.lock:
+                        self.processed_count += 1
+                        self.progress_bar['value'] = self.processed_count
+                elif task == "finish":
+                    messagebox.showinfo("完成", "所有文件处理完成")
+                    self.progress_bar['value'] = 0
+                    self.is_file_selected = False
+        except queue.Empty:
+            pass
+        finally:
+            # 持续轮询队列
+            self.root.after(100, self._process_ui_queue)
+
+    def _concurrent_worker_manager(self, file_paths: List[str], topic: str):
+        """使用线程池并发处理所有选定的文件。"""
+        try:
+            max_workers = int(self.config_manager.get("MaxWorkers", "4"))
+        except (ValueError, TypeError):
+            max_workers = 4
+
+        with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
+            for file_path in file_paths:
+                executor.submit(self._process_single_file, file_path, topic)
+        
+        # 所有任务完成后，向UI队列发送完成信号
+        self.ui_queue.put(("finish", None))
+
+    def _process_single_file(self, file_path: str, topic: str):
+        """处理单个图片文件的完整流程：调用API、保存报告、更新UI队列。"""
+        base_name = os.path.basename(file_path)
+        self.ui_queue.put(("log", f"开始处理: {base_name}"))
+        try:
+            final_report = self.api_service.process_essay_image(file_path, topic)
+            
+            report_filename = os.path.splitext(file_path)[0] + "_report.md"
+            with open(report_filename, 'w', encoding='utf-8') as f:
+                f.write(final_report)
+            
+            self.ui_queue.put(("log", f"完成批改: {base_name} -> {os.path.basename(report_filename)}"))
+
+        except Exception as e:
+            self.ui_queue.put(("log", f"文件: {base_name} 失败: {e}"))
+        
+        # 无论成功或失败，都更新进度
+        self.ui_queue.put(("progress", 1))

config_manager.py

@@ -0,0 +1,107 @@
+import json
+import os
+import base64
+import hashlib
+from typing import Dict, Optional, Tuple
+from cryptography.fernet import Fernet, InvalidToken
+
+class ConfigManager:
+    """
+    管理应用的配置（`config.json`），包括加载、保存以及对敏感信息的自动加解密。
+    """
+    # 定义需要进行加密处理的配置项
+    SENSITIVE_KEYS = ["VlmApiKey", "LlmApiKey"]
+    # 用于生成加密密钥的密码和盐。注意：修改这些值将导致旧的配置文件无法解密。
+    _ENCRYPTION_PASSWORD = b"a-strong-but-not-public-password-for-this-app"
+    _SALT = b'salt_for_llm_app_config'
+
+    def __init__(self, file_path: str = "config.json"):
+        self.file_path = file_path
+        self.config: Dict[str, str] = {}
+        self._fernet: Optional[Fernet] = None
+        self._initialize_encryption()
+        self.load()
+
+    def _initialize_encryption(self):
+        """使用预设的密码和盐生成加密密钥，并初始化Fernet加密/解密实例。"""
+        kdf = hashlib.pbkdf2_hmac('sha256', self._ENCRYPTION_PASSWORD, self._SALT, 100000)
+        key = base64.urlsafe_b64encode(kdf)
+        self._fernet = Fernet(key)
+
+    def _encrypt(self, value: str) -> str:
+        """使用初始化的Fernet实例加密字符串。"""
+        if not value or not self._fernet:
+            return ""
+        return self._fernet.encrypt(value.encode('utf-8')).decode('utf-8')
+
+    def _decrypt(self, encrypted_value: str) -> str:
+        """
+        使用初始化的Fernet实例解密字符串。
+        如果解密失败（例如，值是旧的明文或已损坏），则返回空字符串以避免程序崩溃。
+        """
+        if not encrypted_value or not self._fernet:
+            return ""
+        try:
+            return self._fernet.decrypt(encrypted_value.encode('utf-8')).decode('utf-8')
+        except InvalidToken:
+            # 如果解密失败，返回空字符串
+            return ""
+
+    def load(self) -> bool:
+        """从JSON文件加载配置。如果文件不存在，则创建一个空的配置文件。"""
+        if not os.path.exists(self.file_path):
+            # 如果配置文件不存在，则创建一个空的配置字典并保存
+            self.config = {}
+            self.save()
+            return True
+        try:
+            with open(self.file_path, 'r', encoding='utf-8') as f:
+                self.config = json.load(f)
+            return True
+        except (json.JSONDecodeError, IOError):
+            self.config = {}
+            return False
+
+    def save(self):
+        """将当前配置保存到JSON文件。敏感信息的加密在`set`方法中处理。"""
+        try:
+            with open(self.file_path, 'w', encoding='utf-8') as f:
+                json.dump(self.config, f, indent=4)
+        except IOError as e:
+            print(f"保存配置失败: {e}")
+
+    def get(self, key: str, default: Optional[str] = None) -> Optional[str]:
+        """获取指定键的配置值。如果键属于敏感信息，则自动解密后返回。"""
+        value = self.config.get(key)
+        if value is None:
+            return default
+        
+        if key in self.SENSITIVE_KEYS:
+            return self._decrypt(value)
+        return value
+
+    def set(self, key: str, value: str):
+        """设置指定键的配置值。如果键属于敏感信息，则自动加密后存储。"""
+        if key in self.SENSITIVE_KEYS:
+            self.config[key] = self._encrypt(value)
+        else:
+            self.config[key] = value
+
+    def check_settings(self) -> Tuple[bool, Optional[str]]:
+        """
+        检查所有必需的配置项是否都已设置。
+        返回一个元组，包含检查结果（布尔值）和第一个缺失的配置项名称（字符串）。
+        """
+        required_settings = {
+            "VlmUrl": "VLM服务地址",
+            "VlmApiKey": "VLM服务密钥",
+            "VlmModel": "VLM模型名称",
+            "LlmUrl": "LLM服务地址",
+            "LlmApiKey": "LLM服务密钥",
+            "LlmModel": "LLM模型名称",
+        }
+        for key, name in required_settings.items():
+            # 使用self.get()来确保我们检查的是解密后的值
+            if not self.get(key):
+                return False, name
+        return True, None

main.py

@@ -0,0 +1,37 @@
+import tkinter as tk
+from app_ui import MainApp
+from config_manager import ConfigManager
+from api_services import ApiService
+import sys
+import os
+
+def resource_path(relative_path):
+    """
+    获取资源的绝对路径，以支持PyInstaller打包后的单文件应用。
+
+    在开发环境中，返回基于当前工作目录的相对路径。
+    在PyInstaller打包的应用中，返回临时文件夹`_MEIPASS`中的路径。
+    """
+    try:
+        # 尝试获取PyInstaller在运行时创建的临时路径
+        base_path = sys._MEIPASS
+    except Exception:
+        # 如果`_MEIPASS`属性不存在，说明是在开发环境，使用当前目录
+        base_path = os.path.abspath(".")
+
+    return os.path.join(base_path, relative_path)
+
+if __name__ == "__main__":
+    # 1. 初始化核心服务
+    # 使用 resource_path 确保在打包后也能正确找到配置文件
+    config_manager = ConfigManager(resource_path("config.json"))
+    api_service = ApiService(config_manager)
+
+    # 2. 创建Tkinter主窗口
+    root = tk.Tk()
+    
+    # 3. 实例化主应用，将服务注入应用
+    app = MainApp(root, config_manager, api_service)
+
+    # 4. 启动Tkinter事件循环
+    root.mainloop()

requirements.txt

@@ -0,0 +1,2 @@
+cryptography
+openai