feat: 添加索引模式开关 - 支持简单索引模式（不使用LLM）

新增配置:
- USE_LLM_INDEX = False (默认关闭LLM增强索引)

简单索引模式:
- 不调用LLM，速度快
- 使用词频统计提取关键词
- 适合快速建立索引

LLM增强模式(USE_LLM_INDEX=True):
- 使用LLM分析文档，提取摘要、关键词、实体
- 索引质量更高，但速度慢、需要LLM服务

.gitignore

@@ -2,5 +2,7 @@ instance/
 *.db
 __pycache__/
 *.pyc
-.envdocuments/
+.env
+documents/
 indexes/
+user_config.json

app.py

@@ -13,6 +13,38 @@ from config import *
 from models import db, Document, DocumentChunk, InvertedIndex, QueryLog, IndexStats
 from services import DocumentIndexer, SearchEngine, RAGGenerator
 
+# ==================== 配置文件路径 ====================
+CONFIG_FILE = os.path.join(os.path.dirname(__file__), 'config.py')
+USER_CONFIG_FILE = os.path.join(os.path.dirname(__file__), 'user_config.json')
+
+
+def load_user_config():
+    """加载用户配置"""
+    if os.path.exists(USER_CONFIG_FILE):
+        with open(USER_CONFIG_FILE, 'r', encoding='utf-8') as f:
+            return json.load(f)
+    return {}
+
+
+def save_user_config(config_type, config_data):
+    """保存用户配置"""
+    user_config = load_user_config()
+    user_config[config_type] = config_data
+    with open(USER_CONFIG_FILE, 'w', encoding='utf-8') as f:
+        json.dump(user_config, f, ensure_ascii=False, indent=2)
+
+
+def get_effective_config():
+    """获取有效配置（用户配置覆盖默认配置）"""
+    user_config = load_user_config()
+    
+    return {
+        'llm': {**LLM_CONFIG, **user_config.get('llm', {})},
+        'index': {**INDEX_CONFIG, **user_config.get('index', {})},
+        'doc': {**DOC_CONFIG, **user_config.get('doc', {})}
+    }
+
+
 # ==================== 创建应用 ====================
 app = Flask(__name__)
 app.config['SECRET_KEY'] = SECRET_KEY
@@ -66,6 +98,13 @@ def search_page():
     return render_template('search.html')
 
 
+@app.route('/settings')
+def settings_page():
+    """设置页"""
+    config = get_effective_config()
+    return render_template('settings.html', config=config)
+
+
 # ==================== API路由 ====================
 
 # === 文档管理 ===
@@ -138,7 +177,7 @@ def api_upload_document():
 
 @app.route('/api/documents/<int:doc_id>', methods=['GET'])
 def api_get_document(doc_id):
-    """获取文档详情"""
+    """获取文档详情API"""
     doc = Document.query.get_or_404(doc_id)
     
     chunks = DocumentChunk.query.filter_by(document_id=doc_id).all()
@@ -149,6 +188,14 @@ def api_get_document(doc_id):
     })
 
 
+@app.route('/documents/<int:doc_id>')
+def document_detail_page(doc_id):
+    """文档详情页面"""
+    doc = Document.query.get_or_404(doc_id)
+    chunks = DocumentChunk.query.filter_by(document_id=doc_id).order_by(DocumentChunk.chunk_index).all()
+    return render_template('document_detail.html', doc=doc, chunks=chunks)
+
+
 @app.route('/api/documents/<int:doc_id>', methods=['DELETE'])
 def api_delete_document(doc_id):
     """删除文档"""
@@ -315,6 +362,70 @@ def api_log_feedback(log_id):
     return jsonify({'success': True})
 
 
+# === 配置管理 ===
+
+@app.route('/api/config', methods=['GET'])
+def api_get_config():
+    """获取当前配置"""
+    return jsonify(get_effective_config())
+
+
+@app.route('/api/config/llm', methods=['POST'])
+def api_save_llm_config():
+    """保存LLM配置"""
+    data = request.json
+    save_user_config('llm', data)
+    return jsonify({'success': True})
+
+
+@app.route('/api/config/index', methods=['POST'])
+def api_save_index_config():
+    """保存索引配置"""
+    data = request.json
+    save_user_config('index', data)
+    return jsonify({'success': True})
+
+
+@app.route('/api/config/doc', methods=['POST'])
+def api_save_doc_config():
+    """保存文档配置"""
+    data = request.json
+    save_user_config('doc', data)
+    return jsonify({'success': True})
+
+
+@app.route('/api/config/test', methods=['POST'])
+def api_test_config():
+    """测试LLM连接"""
+    config = get_effective_config()
+    llm_config = config['llm']
+    
+    try:
+        from openai import OpenAI
+        client = OpenAI(
+            api_key=llm_config['api_key'],
+            base_url=llm_config['api_base'],
+        )
+        
+        # 发送简单测试请求
+        response = client.chat.completions.create(
+            model=llm_config['model'],
+            messages=[{"role": "user", "content": "Hello"}],
+            max_tokens=10
+        )
+        
+        return jsonify({
+            'success': True,
+            'model': llm_config['model'],
+            'response': response.choices[0].message.content
+        })
+    except Exception as e:
+        return jsonify({
+            'success': False,
+            'error': str(e)
+        })
+
+
 # ==================== 启动 ====================
 if __name__ == '__main__':
     init_app()

config.py

@@ -33,6 +33,12 @@ DOC_CONFIG = {
     "max_summary_length": 500,  # 摘要最大长度
 }
 
+# ==================== 索引模式 ====================
+# use_llm_index: 是否使用LLM增强索引
+# True - 使用LLM分析文档，提取关键词、摘要等（需要LLM服务，速度慢）
+# False - 使用简单分词和词频统计（速度快，无需LLM）
+USE_LLM_INDEX = False  # 默认关闭LLM索引
+
 # ==================== 索引配置 ====================
 INDEX_CONFIG = {
     # BM25参数

models.py

@@ -92,6 +92,8 @@ class Document(db.Model):
         return {
             'id': self.id,
             'filename': self.filename,
+            'file_type': self.file_type,
+            'file_size': self.file_size,
             'title': self.title,
             'status': self.status,
             'summary': self.summary,

services.py

@@ -12,21 +12,57 @@ from collections import Counter
 from openai import OpenAI
 from flask import current_app
 
-from config import LLM_CONFIG, DOC_CONFIG, INDEX_CONFIG
-from models import db, Document, DocumentChunk, InvertedIndex, IndexStats
+from config import LLM_CONFIG, DOC_CONFIG, INDEX_CONFIG, USE_LLM_INDEX
+from models import db, Document, DocumentChunk, InvertedIndex, IndexStats, QueryLog
+
+
+def get_llm_config():
+    """获取有效的LLM配置（支持动态更新）"""
+    user_config_file = os.path.join(os.path.dirname(__file__), 'user_config.json')
+    if os.path.exists(user_config_file):
+        with open(user_config_file, 'r', encoding='utf-8') as f:
+            user_config = json.load(f)
+            return {**LLM_CONFIG, **user_config.get('llm', {})}
+    return LLM_CONFIG
+
+
+def get_doc_config():
+    """获取有效的文档配置"""
+    user_config_file = os.path.join(os.path.dirname(__file__), 'user_config.json')
+    if os.path.exists(user_config_file):
+        with open(user_config_file, 'r', encoding='utf-8') as f:
+            user_config = json.load(f)
+            return {**DOC_CONFIG, **user_config.get('doc', {})}
+    return DOC_CONFIG
+
+
+def get_index_config():
+    """获取有效的索引配置"""
+    user_config_file = os.path.join(os.path.dirname(__file__), 'user_config.json')
+    if os.path.exists(user_config_file):
+        with open(user_config_file, 'r', encoding='utf-8') as f:
+            user_config = json.load(f)
+            return {**INDEX_CONFIG, **user_config.get('index', {})}
+    return INDEX_CONFIG
 
 
 class LLMService:
     """LLM服务封装"""
     
     def __init__(self):
-        self.client = OpenAI(
-            api_key=LLM_CONFIG['api_key'],
-            base_url=LLM_CONFIG['api_base'],
+        pass  # 不再在初始化时设置配置
+    
+    def _get_client(self):
+        """获取LLM客户端"""
+        config = get_llm_config()
+        return OpenAI(
+            api_key=config['api_key'],
+            base_url=config['api_base'],
         )
-        self.model = LLM_CONFIG['model']
-        self.max_tokens = LLM_CONFIG['max_tokens']
-        self.temperature = LLM_CONFIG['temperature']
+    
+    def _get_config(self):
+        """获取当前配置"""
+        return get_llm_config()
     
     def analyze_document(self, content, title=None):
         """
@@ -60,8 +96,10 @@ class LLMService:
 只返回JSON，不要其他内容。"""
 
         try:
-            response = self.client.chat.completions.create(
-                model=self.model,
+            config = self._get_config()
+            client = self._get_client()
+            response = client.chat.completions.create(
+                model=config['model'],
                 messages=[{"role": "user", "content": prompt}],
                 max_tokens=1000,
                 temperature=0.3,
@@ -106,8 +144,10 @@ class LLMService:
 只返回JSON。"""
 
         try:
-            response = self.client.chat.completions.create(
-                model=self.model,
+            config = self._get_config()
+            client = self._get_client()
+            response = client.chat.completions.create(
+                model=config['model'],
                 messages=[{"role": "user", "content": prompt}],
                 max_tokens=500,
                 temperature=0.3,
@@ -150,8 +190,10 @@ class LLMService:
 只返回JSON。"""
 
         try:
-            response = self.client.chat.completions.create(
-                model=self.model,
+            config = self._get_config()
+            client = self._get_client()
+            response = client.chat.completions.create(
+                model=config['model'],
                 messages=[{"role": "user", "content": prompt}],
                 max_tokens=500,
                 temperature=0.3,
@@ -177,8 +219,11 @@ class DocumentIndexer:
     
     def __init__(self):
         self.llm = LLMService()
-        self.chunk_size = DOC_CONFIG['chunk_size']
-        self.chunk_overlap = DOC_CONFIG['chunk_overlap']
+    
+    def _get_chunk_config(self):
+        """获取分块配置"""
+        config = get_doc_config()
+        return config['chunk_size'], config['chunk_overlap']
     
     def index_document(self, doc_id):
         """
@@ -207,15 +252,25 @@ class DocumentIndexer:
             doc.content = content
             doc.word_count = len(content)
             
-            # 使用LLM分析整个文档
-            print(f"  正在分析文档: {doc.filename}")
-            analysis = self.llm.analyze_document(content, doc.title)
+            # 检查是否使用LLM增强索引
+            use_llm = USE_LLM_INDEX
             
-            doc.summary = analysis.get('summary', '')
-            doc.set_keywords(analysis.get('keywords', []))
-            doc.set_topics(analysis.get('topics', []))
-            doc.category = analysis.get('category', '')
-            doc.set_entities(analysis.get('entities', {}))
+            if use_llm:
+                # 使用LLM分析整个文档
+                print(f"  正在使用LLM分析文档: {doc.filename}")
+                analysis = self.llm.analyze_document(content, doc.title)
+                
+                doc.summary = analysis.get('summary', '')
+                doc.set_keywords(analysis.get('keywords', []))
+                doc.set_topics(analysis.get('topics', []))
+                doc.category = analysis.get('category', '')
+                doc.set_entities(analysis.get('entities', {}))
+            else:
+                # 简单模式：不使用LLM
+                print(f"  正在索引文档(简单模式): {doc.filename}")
+                # 从内容中提取简单关键词
+                simple_keywords = self._extract_simple_keywords(content)
+                doc.set_keywords(simple_keywords[:20])
             
             # 分块处理
             chunks = self._split_content(content)
@@ -234,10 +289,14 @@ class DocumentIndexer:
                     end_char=len(chunk_content)
                 )
                 
-                # LLM分析分块
-                chunk_analysis = self.llm.analyze_chunk(chunk_content)
-                chunk.summary = chunk_analysis.get('summary', '')
-                chunk.set_keywords(chunk_analysis.get('keywords', []))
+                if use_llm:
+                    # LLM分析分块
+                    chunk_analysis = self.llm.analyze_chunk(chunk_content)
+                    chunk.summary = chunk_analysis.get('summary', '')
+                    chunk.set_keywords(chunk_analysis.get('keywords', []))
+                else:
+                    # 简单模式：从分块提取关键词
+                    chunk.set_keywords(self._extract_simple_keywords(chunk_content)[:10])
                 
                 # 计算词频
                 term_freq = self._compute_term_freq(chunk_content)
@@ -268,23 +327,70 @@ class DocumentIndexer:
             print(f"  ✗ 索引失败: {e}")
             return False
     
+    def _extract_simple_keywords(self, content):
+        """简单提取关键词（不使用LLM）"""
+        import re
+        from collections import Counter
+        
+        # 提取中文词组（2-4字）
+        chinese_words = re.findall(r'[\u4e00-\u9fff]{2,4}', content)
+        
+        # 过滤常见无意义词
+        stopwords = {'的', '是', '在', '了', '和', '与', '或', '等', '这', '那', 
+                     '有', '为', '对', '也', '被', '把', '给', '向', '从', '到',
+                     '一个', '一些', '这种', '那种', '什么', '怎么', '如何', '为什么'}
+        
+        # 统计词频
+        word_freq = Counter(w for w in chinese_words if w not in stopwords)
+        
+        # 返回高频词
+        return [w for w, _ in word_freq.most_common(30)]
+    
     def _read_document(self, filepath):
         """读取文档内容"""
         ext = os.path.splitext(filepath)[1].lower()
         
-        # 尝试读取文本文件（包括没有扩展名的）
-        try:
-            with open(filepath, 'r', encoding='utf-8', errors='ignore') as f:
-                content = f.read()
-                if content.strip():  # 如果能读取到内容
-                    return content
-        except:
-            pass
+        # 尝试多种编码读取文本文件
+        encodings = ['utf-8', 'gbk', 'gb2312', 'gb18030', 'big5', 'utf-16', 'latin-1']
         
-        # 按扩展名处理特定格式
-        if ext in ['.txt', '.md', '.json', '.html']:
-            with open(filepath, 'r', encoding='utf-8', errors='ignore') as f:
-                return f.read()
+        def try_read_with_encoding(encodings_list):
+            for enc in encodings_list:
+                try:
+                    with open(filepath, 'r', encoding=enc) as f:
+                        content = f.read()
+                        # 检查是否有有效的中文字符
+                        if content.strip() and len(content) > 0:
+                            # 简单验证：检查是否有乱码
+                            # 如果内容看起来合理，返回它
+                            return content, enc
+                except (UnicodeDecodeError, UnicodeError):
+                    continue
+            return None, None
+        
+        # 按扩展名处理
+        if ext in ['.txt', '.md', '.json', '.html', '']:
+            # 先尝试常见编码
+            content, used_enc = try_read_with_encoding(encodings)
+            if content:
+                print(f"  使用编码 {used_enc} 读取文件")
+                return content
+            
+            # 如果都失败，尝试二进制读取后解码
+            try:
+                with open(filepath, 'rb') as f:
+                    raw = f.read()
+                # 尝试chardet检测编码
+                try:
+                    import chardet
+                    detected = chardet.detect(raw)
+                    if detected['encoding']:
+                        return raw.decode(detected['encoding'])
+                except:
+                    pass
+                # 最后尝试
+                return raw.decode('utf-8', errors='replace')
+            except:
+                return ""
         
         elif ext == '.pdf':
             try:
@@ -330,6 +436,7 @@ class DocumentIndexer:
         Returns:
             list: 内容块列表
         """
+        chunk_size, _ = self._get_chunk_config()
         chunks = []
         
         # 按段落分割
@@ -337,7 +444,7 @@ class DocumentIndexer:
         
         current_chunk = ""
         for para in paragraphs:
-            if len(current_chunk) + len(para) < self.chunk_size:
+            if len(current_chunk) + len(para) < chunk_size:
                 current_chunk += para + '\n\n'
             else:
                 if current_chunk.strip():
@@ -347,7 +454,7 @@ class DocumentIndexer:
         if current_chunk.strip():
             chunks.append(current_chunk.strip())
         
-        return chunks if chunks else [content[:self.chunk_size]]
+        return chunks if chunks else [content[:chunk_size]]
     
     def _compute_term_freq(self, content):
         """计算词频"""
@@ -424,8 +531,11 @@ class SearchEngine:
     
     def __init__(self):
         self.llm = LLMService()
-        self.k1 = INDEX_CONFIG['bm25_k1']
-        self.b = INDEX_CONFIG['bm25_b']
+    
+    def _get_bm25_params(self):
+        """获取BM25参数"""
+        config = get_index_config()
+        return config['bm25_k1'], config['bm25_b']
     
     def search(self, query, top_k=10):
         """
@@ -542,6 +652,7 @@ class SearchEngine:
                 continue
             
             # BM25计算
+            k1, b = self._get_bm25_params()
             score = 0
             doc_len = doc.word_count or 1000
             
@@ -557,8 +668,8 @@ class SearchEngine:
                 tf = data['terms'].get(term, 0)
                 
                 # BM25公式
-                tf_component = (tf * (self.k1 + 1)) / (
-                    tf + self.k1 * (1 - self.b + self.b * doc_len / avg_doc_len)
+                tf_component = (tf * (k1 + 1)) / (
+                    tf + k1 * (1 - b + b * doc_len / avg_doc_len)
                 )
                 
                 score += idf * tf_component
@@ -653,13 +764,14 @@ class RAGGenerator:
 请给出准确、简洁的回答，并标注信息来源。"""
 
         try:
+            llm_config = get_llm_config()
             client = OpenAI(
-                api_key=LLM_CONFIG['api_key'],
-                base_url=LLM_CONFIG['api_base'],
+                api_key=llm_config['api_key'],
+                base_url=llm_config['api_base'],
             )
             
             response = client.chat.completions.create(
-                model=LLM_CONFIG['model'],
+                model=llm_config['model'],
                 messages=[{"role": "user", "content": prompt}],
                 max_tokens=1000,
                 temperature=0.5,

static/js/main.js

@@ -0,0 +1,247 @@
+/**
+ * LLM Index RAG - 前端交互脚本
+ */
+
+// 搜索表单处理
+document.getElementById('searchForm')?.addEventListener('submit', async function(e) {
+    e.preventDefault();
+    
+    const query = document.getElementById('queryInput').value.trim();
+    const mode = document.querySelector('input[name="mode"]:checked').value;
+    
+    if (!query) {
+        alert('请输入查询内容');
+        return;
+    }
+    
+    // 显示加载状态
+    const resultsSection = document.getElementById('resultsSection');
+    const ragSection = document.getElementById('ragSection');
+    const resultsContainer = document.getElementById('resultsContainer');
+    const ragAnswer = document.getElementById('ragAnswer');
+    const ragSources = document.getElementById('ragSources');
+    const resultCount = document.getElementById('resultCount');
+    
+    resultsSection.style.display = 'none';
+    ragSection.style.display = 'none';
+    resultsContainer.innerHTML = '<div class="text-center py-4"><div class="spinner-border text-primary"></div><p class="mt-2">正在检索...</p></div>';
+    resultsSection.style.display = 'block';
+    
+    try {
+        if (mode === 'search') {
+            // 文档检索模式
+            const response = await fetch('/api/search', {
+                method: 'POST',
+                headers: {'Content-Type': 'application/json'},
+                body: JSON.stringify({query: query, top_k: 10})
+            });
+            
+            const data = await response.json();
+            
+            if (data.error) {
+                resultsContainer.innerHTML = `<div class="alert alert-danger">${data.error}</div>`;
+                return;
+            }
+            
+            resultCount.textContent = data.total;
+            
+            if (data.results && data.results.length > 0) {
+                resultsContainer.innerHTML = data.results.map(r => `
+                    <div class="result-item">
+                        <h6 class="mb-1">
+                            <a href="/documents" class="text-decoration-none">${r.title || r.document_title || '文档'}</a>
+                        </h6>
+                        <p class="text-muted small mb-1">${r.summary || r.content?.substring(0, 200) + '...' || ''}</p>
+                        <div class="d-flex gap-2 align-items-center">
+                            <span class="source-tag">${r.source || '本地文档'}</span>
+                            <span class="badge bg-primary">${(r.score * 100).toFixed(1)}%</span>
+                        </div>
+                    </div>
+                `).join('');
+            } else {
+                resultsContainer.innerHTML = `
+                    <div class="text-center py-4 text-muted">
+                        <i class="bi bi-search display-4"></i>
+                        <p class="mt-2">未找到相关结果</p>
+                        <p class="small">请尝试其他关键词，或先上传并索引文档</p>
+                    </div>
+                `;
+            }
+        } else {
+            // RAG问答模式
+            resultsContainer.innerHTML = '<div class="text-center py-4"><div class="spinner-border text-primary"></div><p class="mt-2">正在生成回答...</p></div>';
+            
+            const response = await fetch('/api/rag/answer', {
+                method: 'POST',
+                headers: {'Content-Type': 'application/json'},
+                body: JSON.stringify({query: query, top_k: 5})
+            });
+            
+            const data = await response.json();
+            
+            if (data.error) {
+                resultsContainer.innerHTML = `<div class="alert alert-danger">${data.error}</div>`;
+                return;
+            }
+            
+            resultsSection.style.display = 'none';
+            ragSection.style.display = 'block';
+            
+            // 显示回答
+            ragAnswer.innerHTML = `
+                <div class="d-flex align-items-start gap-3">
+                    <div class="bg-primary text-white rounded-circle p-2">
+                        <i class="bi bi-robot"></i>
+                    </div>
+                    <div class="flex-grow-1">
+                        <div class="markdown-content">${data.answer || '抱歉，无法生成回答。'}</div>
+                    </div>
+                </div>
+            `;
+            
+            // 显示来源
+            if (data.sources && data.sources.length > 0) {
+                ragSources.innerHTML = data.sources.map(s => `
+                    <div class="result-item">
+                        <h6 class="mb-1">${s.title || s.document_title || '参考文档'}</h6>
+                        <p class="text-muted small mb-0">${s.content?.substring(0, 150) + '...' || ''}</p>
+                    </div>
+                `).join('');
+            } else {
+                ragSources.innerHTML = '<p class="text-muted">无参考来源</p>';
+            }
+        }
+    } catch (err) {
+        resultsContainer.innerHTML = `<div class="alert alert-danger">请求失败: ${err.message}</div>`;
+    }
+});
+
+// 文档上传
+document.getElementById('uploadForm')?.addEventListener('submit', async function(e) {
+    e.preventDefault();
+    
+    const formData = new FormData(this);
+    const fileInput = document.getElementById('fileInput');
+    
+    if (!fileInput.files.length) {
+        alert('请选择文件');
+        return;
+    }
+    
+    const uploadBtn = document.getElementById('uploadBtn');
+    uploadBtn.disabled = true;
+    uploadBtn.innerHTML = '<span class="spinner-border spinner-border-sm"></span> 上传中...';
+    
+    try {
+        const response = await fetch('/api/documents', {
+            method: 'POST',
+            body: formData
+        });
+        
+        const data = await response.json();
+        
+        if (data.success) {
+            alert('上传成功！');
+            location.reload();
+        } else {
+            alert('上传失败: ' + (data.error || '未知错误'));
+        }
+    } catch (err) {
+        alert('上传失败: ' + err.message);
+    } finally {
+        uploadBtn.disabled = false;
+        uploadBtn.innerHTML = '<i class="bi bi-upload"></i> 上传';
+    }
+});
+
+// 索引文档
+async function indexDocument(docId) {
+    if (!confirm('确定要索引此文档吗？这可能需要一些时间。')) return;
+    
+    try {
+        const response = await fetch(`/api/index/${docId}`, {method: 'POST'});
+        const data = await response.json();
+        
+        if (data.success) {
+            alert('索引完成！');
+            location.reload();
+        } else {
+            alert('索引失败: ' + (data.error || '未知错误'));
+        }
+    } catch (err) {
+        alert('索引失败: ' + err.message);
+    }
+}
+
+// 删除文档
+async function deleteDocument(docId) {
+    if (!confirm('确定要删除此文档吗？此操作不可恢复。')) return;
+    
+    try {
+        const response = await fetch(`/api/documents/${docId}`, {method: 'DELETE'});
+        const data = await response.json();
+        
+        if (data.success) {
+            alert('删除成功！');
+            location.reload();
+        } else {
+            alert('删除失败: ' + (data.error || '未知错误'));
+        }
+    } catch (err) {
+        alert('删除失败: ' + err.message);
+    }
+}
+
+// 批量索引
+async function batchIndex() {
+    if (!confirm('确定要索引所有待处理文档吗？')) return;
+    
+    try {
+        const response = await fetch('/api/index/batch', {method: 'POST'});
+        const data = await response.json();
+        
+        alert(`索引完成！成功: ${data.success}, 失败: ${data.failed}`);
+        location.reload();
+    } catch (err) {
+        alert('批量索引失败: ' + err.message);
+    }
+}
+
+// 重建索引
+async function rebuildIndex() {
+    if (!confirm('重建索引将清除所有现有索引，确定继续吗？')) return;
+    
+    try {
+        const response = await fetch('/api/index/rebuild', {method: 'POST'});
+        const data = await response.json();
+        
+        alert(`重建完成！成功: ${data.success}, 失败: ${data.failed}`);
+        location.reload();
+    } catch (err) {
+        alert('重建索引失败: ' + err.message);
+    }
+}
+
+// 加载统计信息
+async function loadStats() {
+    try {
+        const response = await fetch('/api/stats');
+        const stats = await response.json();
+        
+        document.getElementById('statDocs').textContent = stats.total_documents || 0;
+        document.getElementById('statChunks').textContent = stats.total_chunks || 0;
+        document.getElementById('statTerms').textContent = stats.total_terms || 0;
+        document.getElementById('statWords').textContent = (stats.total_words || 0).toLocaleString();
+    } catch (err) {
+        console.error('加载统计失败:', err);
+    }
+}
+
+// 页面加载时刷新统计
+document.addEventListener('DOMContentLoaded', function() {
+    // 如果在首页，定时刷新统计
+    if (document.getElementById('statDocs')) {
+        loadStats();
+        setInterval(loadStats, 30000); // 每30秒刷新
+    }
+});

templates/document_detail.html

@@ -0,0 +1,212 @@
+<!DOCTYPE html>
+<html lang="zh-CN">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>{{ doc.title or doc.filename }} - 文档详情</title>
+    <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/css/bootstrap.min.css" rel="stylesheet">
+    <link href="https://cdn.jsdelivr.net/npm/bootstrap-icons@1.10.0/font/bootstrap-icons.css" rel="stylesheet">
+    <style>
+        body { background-color: #f8f9fa; }
+        .chunk-card { border-left: 4px solid #667eea; }
+    </style>
+</head>
+<body>
+    <nav class="navbar navbar-expand-lg navbar-dark bg-dark">
+        <div class="container">
+            <a class="navbar-brand" href="/"><i class="bi bi-search"></i> LLM Index RAG</a>
+            <div class="navbar-nav ms-auto">
+                <a class="nav-link" href="/">首页</a>
+                <a class="nav-link" href="/documents">文档管理</a>
+                <a class="nav-link" href="/search">知识检索</a>
+                <a class="nav-link" href="/settings">系统设置</a>
+            </div>
+        </div>
+    </nav>
+
+    <div class="container py-4">
+        <nav aria-label="breadcrumb" class="mb-3">
+            <ol class="breadcrumb">
+                <li class="breadcrumb-item"><a href="/documents">文档管理</a></li>
+                <li class="breadcrumb-item active">{{ doc.title or doc.filename }}</li>
+            </ol>
+        </nav>
+
+        <!-- 文档信息 -->
+        <div class="card mb-4">
+            <div class="card-header d-flex justify-content-between align-items-center">
+                <h5 class="mb-0"><i class="bi bi-file-earmark-text"></i> 文档信息</h5>
+                <div>
+                    {% if doc.status == 'indexed' %}
+                    <span class="badge bg-success">已索引</span>
+                    {% elif doc.status == 'processing' %}
+                    <span class="badge bg-warning">处理中</span>
+                    {% elif doc.status == 'failed' %}
+                    <span class="badge bg-danger">失败</span>
+                    {% else %}
+                    <span class="badge bg-secondary">待索引</span>
+                    {% endif %}
+                </div>
+            </div>
+            <div class="card-body">
+                <div class="row">
+                    <div class="col-md-6">
+                        <table class="table table-sm">
+                            <tr>
+                                <th width="120">文件名</th>
+                                <td>{{ doc.filename }}</td>
+                            </tr>
+                            <tr>
+                                <th>标题</th>
+                                <td>{{ doc.title or '-' }}</td>
+                            </tr>
+                            <tr>
+                                <th>文件类型</th>
+                                <td>{{ doc.file_type }}</td>
+                            </tr>
+                            <tr>
+                                <th>文件大小</th>
+                                <td>{{ (doc.file_size / 1024)|round(1) }} KB</td>
+                            </tr>
+                        </table>
+                    </div>
+                    <div class="col-md-6">
+                        <table class="table table-sm">
+                            <tr>
+                                <th width="120">分块数</th>
+                                <td>{{ doc.chunk_count }}</td>
+                            </tr>
+                            <tr>
+                                <th>字数</th>
+                                <td>{{ doc.word_count|default(0) }}</td>
+                            </tr>
+                            <tr>
+                                <th>上传时间</th>
+                                <td>{{ doc.created_at.strftime('%Y-%m-%d %H:%M') if doc.created_at else '-' }}</td>
+                            </tr>
+                            <tr>
+                                <th>索引时间</th>
+                                <td>{{ doc.indexed_at.strftime('%Y-%m-%d %H:%M') if doc.indexed_at else '-' }}</td>
+                            </tr>
+                        </table>
+                    </div>
+                </div>
+
+                {% if doc.summary %}
+                <div class="mt-3">
+                    <h6><i class="bi bi-card-text"></i> 文档摘要</h6>
+                    <p class="text-muted">{{ doc.summary }}</p>
+                </div>
+                {% endif %}
+
+                {% if doc.keywords %}
+                <div class="mt-3">
+                    <h6><i class="bi bi-tags"></i> 关键词</h6>
+                    <div>
+                        {% for kw in doc.get_keywords() %}
+                        <span class="badge bg-primary me-1">{{ kw }}</span>
+                        {% endfor %}
+                    </div>
+                </div>
+                {% endif %}
+
+                {% if doc.category %}
+                <div class="mt-3">
+                    <h6><i class="bi bi-folder"></i> 分类</h6>
+                    <span class="badge bg-info">{{ doc.category }}</span>
+                </div>
+                {% endif %}
+            </div>
+        </div>
+
+        <!-- 文档分块 -->
+        <div class="card">
+            <div class="card-header">
+                <h5 class="mb-0"><i class="bi bi-puzzle"></i> 文档分块 ({{ chunks|length }} 个)</h5>
+            </div>
+            <div class="card-body">
+                {% if chunks %}
+                <div class="accordion" id="chunksAccordion">
+                    {% for chunk in chunks %}
+                    <div class="accordion-item">
+                        <h2 class="accordion-header">
+                            <button class="accordion-button collapsed" type="button" data-bs-toggle="collapse" data-bs-target="#chunk{{ loop.index }}">
+                                <strong>分块 {{ loop.index }}</strong>
+                                <span class="ms-3 text-muted small">
+                                    {{ chunk.content[:50] }}...
+                                </span>
+                            </button>
+                        </h2>
+                        <div id="chunk{{ loop.index }}" class="accordion-collapse collapse" data-bs-parent="#chunksAccordion">
+                            <div class="accordion-body">
+                                {% if chunk.summary %}
+                                <div class="alert alert-info small">
+                                    <strong>摘要:</strong> {{ chunk.summary }}
+                                </div>
+                                {% endif %}
+
+                                {% if chunk.keywords %}
+                                <div class="mb-2">
+                                    {% for kw in chunk.get_keywords() %}
+                                    <span class="badge bg-secondary me-1">{{ kw }}</span>
+                                    {% endfor %}
+                                </div>
+                                {% endif %}
+
+                                <pre class="bg-light p-3 rounded" style="white-space: pre-wrap; word-wrap: break-word;">{{ chunk.content }}</pre>
+                            </div>
+                        </div>
+                    </div>
+                    {% endfor %}
+                </div>
+                {% else %}
+                <p class="text-muted text-center py-4">暂无分块数据，请先索引文档</p>
+                {% endif %}
+            </div>
+        </div>
+
+        <div class="mt-4">
+            <a href="/documents" class="btn btn-secondary">
+                <i class="bi bi-arrow-left"></i> 返回列表
+            </a>
+            {% if doc.status == 'pending' %}
+            <button class="btn btn-primary" onclick="indexDoc({{ doc.id }})">
+                <i class="bi bi-arrow-repeat"></i> 索引文档
+            </button>
+            {% endif %}
+            <button class="btn btn-danger" onclick="deleteDoc({{ doc.id }})">
+                <i class="bi bi-trash"></i> 删除文档
+            </button>
+        </div>
+    </div>
+
+    <script src="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/js/bootstrap.bundle.min.js"></script>
+    <script>
+        function indexDoc(id) {
+            if (!confirm('确定要索引此文档吗？')) return;
+            
+            fetch(`/api/index/${id}`, { method: 'POST' })
+                .then(r => r.json())
+                .then(data => {
+                    alert(data.success ? '索引完成' : '索引失败: ' + (data.error || '未知错误'));
+                    if (data.success) location.reload();
+                });
+        }
+
+        function deleteDoc(id) {
+            if (!confirm('确定要删除此文档吗？此操作不可恢复！')) return;
+            
+            fetch(`/api/documents/${id}`, { method: 'DELETE' })
+                .then(r => r.json())
+                .then(data => {
+                    if (data.success) {
+                        alert('删除成功');
+                        window.location.href = '/documents';
+                    } else {
+                        alert('删除失败');
+                    }
+                });
+        }
+    </script>
+</body>
+</html>

templates/index.html

@@ -28,6 +28,7 @@
                 <a class="nav-link" href="/">首页</a>
                 <a class="nav-link" href="/documents">文档管理</a>
                 <a class="nav-link" href="/search">知识检索</a>
+                <a class="nav-link" href="/settings">系统设置</a>
             </div>
         </div>
     </nav>

templates/settings.html

@@ -0,0 +1,284 @@
+<!DOCTYPE html>
+<html lang="zh-CN">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>系统设置 - LLM Index RAG</title>
+    <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/css/bootstrap.min.css" rel="stylesheet">
+    <link href="https://cdn.jsdelivr.net/npm/bootstrap-icons@1.10.0/font/bootstrap-icons.css" rel="stylesheet">
+    <style>
+        body { background-color: #f8f9fa; }
+        .config-card { border-radius: 10px; border: none; box-shadow: 0 2px 10px rgba(0,0,0,0.05); }
+    </style>
+</head>
+<body>
+    <!-- 导航栏 -->
+    <nav class="navbar navbar-expand-lg navbar-dark bg-dark">
+        <div class="container">
+            <a class="navbar-brand" href="/">
+                <i class="bi bi-search"></i> LLM Index RAG
+            </a>
+            <div class="navbar-nav ms-auto">
+                <a class="nav-link" href="/">首页</a>
+                <a class="nav-link" href="/documents">文档管理</a>
+                <a class="nav-link" href="/search">知识检索</a>
+                <a class="nav-link active" href="/settings">系统设置</a>
+            </div>
+        </div>
+    </nav>
+
+    <div class="container py-4">
+        <h4 class="mb-4"><i class="bi bi-gear"></i> 系统设置</h4>
+
+        <!-- LLM配置 -->
+        <div class="card config-card mb-4">
+            <div class="card-header bg-white">
+                <h5 class="mb-0"><i class="bi bi-cpu text-primary"></i> 大模型配置</h5>
+            </div>
+            <div class="card-body">
+                <form id="llmConfigForm">
+                    <div class="row">
+                        <div class="col-md-6 mb-3">
+                            <label class="form-label">API地址</label>
+                            <input type="text" class="form-control" id="apiBase"
+                                value="{{ config.llm.api_base }}" placeholder="http://localhost:1234/v1">
+                            <small class="text-muted">LLM API的基础URL</small>
+                        </div>
+                        <div class="col-md-6 mb-3">
+                            <label class="form-label">API Key</label>
+                            <input type="text" class="form-control" id="apiKey"
+                                value="{{ config.llm.api_key }}" placeholder="sk-xxx">
+                            <small class="text-muted">API密钥</small>
+                        </div>
+                    </div>
+                    <div class="row">
+                        <div class="col-md-6 mb-3">
+                            <label class="form-label">模型名称</label>
+                            <input type="text" class="form-control" id="model"
+                                value="{{ config.llm.model }}" placeholder="qwen/qwen3.5-35b-a3b">
+                        </div>
+                        <div class="col-md-3 mb-3">
+                            <label class="form-label">Max Tokens</label>
+                            <input type="number" class="form-control" id="maxTokens"
+                                value="{{ config.llm.max_tokens }}">
+                        </div>
+                        <div class="col-md-3 mb-3">
+                            <label class="form-label">Temperature</label>
+                            <input type="number" class="form-control" id="temperature"
+                                value="{{ config.llm.temperature }}" step="0.1" min="0" max="2">
+                        </div>
+                    </div>
+                    <div class="mb-3">
+                        <label class="form-label">超时时间（秒）</label>
+                        <input type="number" class="form-control" id="timeout"
+                            value="{{ config.llm.timeout }}">
+                    </div>
+                    <div class="d-flex gap-2">
+                        <button type="submit" class="btn btn-primary">
+                            <i class="bi bi-save"></i> 保存配置
+                        </button>
+                        <button type="button" class="btn btn-outline-secondary" onclick="testConnection()">
+                            <i class="bi bi-plug"></i> 测试连接
+                        </button>
+                    </div>
+                </form>
+                <div id="testResult" class="mt-3"></div>
+            </div>
+        </div>
+
+        <!-- 索引配置 -->
+        <div class="card config-card mb-4">
+            <div class="card-header bg-white">
+                <h5 class="mb-0"><i class="bi bi-list-columns text-success"></i> 索引配置</h5>
+            </div>
+            <div class="card-body">
+                <form id="indexConfigForm">
+                    <div class="row">
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">BM25 K1参数</label>
+                            <input type="number" class="form-control" id="bm25K1"
+                                value="{{ config.index.bm25_k1 }}" step="0.1">
+                            <small class="text-muted">词频饱和参数（推荐1.2-2.0）</small>
+                        </div>
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">BM25 B参数</label>
+                            <input type="number" class="form-control" id="bm25B"
+                                value="{{ config.index.bm25_b }}" step="0.05">
+                            <small class="text-muted">文档长度归一化（推荐0.75）</small>
+                        </div>
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">最大返回结果</label>
+                            <input type="number" class="form-control" id="maxResults"
+                                value="{{ config.index.max_results }}">
+                        </div>
+                    </div>
+                    <div class="row">
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">标题权重</label>
+                            <input type="number" class="form-control" id="titleWeight"
+                                value="{{ config.index.title_weight }}" step="0.5">
+                        </div>
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">关键词权重</label>
+                            <input type="number" class="form-control" id="keywordWeight"
+                                value="{{ config.index.keyword_weight }}" step="0.5">
+                        </div>
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">内容权重</label>
+                            <input type="number" class="form-control" id="contentWeight"
+                                value="{{ config.index.content_weight }}" step="0.5">
+                        </div>
+                    </div>
+                    <button type="submit" class="btn btn-success">
+                        <i class="bi bi-save"></i> 保存配置
+                    </button>
+                </form>
+            </div>
+        </div>
+
+        <!-- 文档配置 -->
+        <div class="card config-card">
+            <div class="card-header bg-white">
+                <h5 class="mb-0"><i class="bi bi-file-earmark-text text-warning"></i> 文档处理配置</h5>
+            </div>
+            <div class="card-body">
+                <form id="docConfigForm">
+                    <div class="row">
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">分块大小</label>
+                            <input type="number" class="form-control" id="chunkSize"
+                                value="{{ config.doc.chunk_size }}">
+                            <small class="text-muted">字符数</small>
+                        </div>
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">分块重叠</label>
+                            <input type="number" class="form-control" id="chunkOverlap"
+                                value="{{ config.doc.chunk_overlap }}">
+                        </div>
+                        <div class="col-md-4 mb-3">
+                            <label class="form-label">最大关键词数</label>
+                            <input type="number" class="form-control" id="maxKeywords"
+                                value="{{ config.doc.max_keywords }}">
+                        </div>
+                    </div>
+                    <button type="submit" class="btn btn-warning">
+                        <i class="bi bi-save"></i> 保存配置
+                    </button>
+                </form>
+            </div>
+        </div>
+    </div>
+
+    <script src="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/js/bootstrap.bundle.min.js"></script>
+    <script>
+        // 保存LLM配置
+        document.getElementById('llmConfigForm')?.addEventListener('submit', async function(e) {
+            e.preventDefault();
+            
+            const config = {
+                api_base: document.getElementById('apiBase').value,
+                api_key: document.getElementById('apiKey').value,
+                model: document.getElementById('model').value,
+                max_tokens: parseInt(document.getElementById('maxTokens').value),
+                temperature: parseFloat(document.getElementById('temperature').value),
+                timeout: parseInt(document.getElementById('timeout').value)
+            };
+            
+            try {
+                const res = await fetch('/api/config/llm', {
+                    method: 'POST',
+                    headers: {'Content-Type': 'application/json'},
+                    body: JSON.stringify(config)
+                });
+                const data = await res.json();
+                
+                if (data.success) {
+                    alert('LLM配置已保存！');
+                } else {
+                    alert('保存失败: ' + data.error);
+                }
+            } catch (err) {
+                alert('保存失败: ' + err.message);
+            }
+        });
+
+        // 测试连接
+        async function testConnection() {
+            const resultDiv = document.getElementById('testResult');
+            resultDiv.innerHTML = '<div class="alert alert-info"><span class="spinner-border spinner-border-sm"></span> 测试中...</div>';
+            
+            try {
+                const res = await fetch('/api/config/test', {method: 'POST'});
+                const data = await res.json();
+                
+                if (data.success) {
+                    resultDiv.innerHTML = '<div class="alert alert-success"><i class="bi bi-check-circle"></i> 连接成功！模型: ' + data.model + '</div>';
+                } else {
+                    resultDiv.innerHTML = '<div class="alert alert-danger"><i class="bi bi-x-circle"></i> 连接失败: ' + data.error + '</div>';
+                }
+            } catch (err) {
+                resultDiv.innerHTML = '<div class="alert alert-danger"><i class="bi bi-x-circle"></i> 连接失败: ' + err.message + '</div>';
+            }
+        }
+
+        // 保存索引配置
+        document.getElementById('indexConfigForm')?.addEventListener('submit', async function(e) {
+            e.preventDefault();
+            
+            const config = {
+                bm25_k1: parseFloat(document.getElementById('bm25K1').value),
+                bm25_b: parseFloat(document.getElementById('bm25B').value),
+                max_results: parseInt(document.getElementById('maxResults').value),
+                title_weight: parseFloat(document.getElementById('titleWeight').value),
+                keyword_weight: parseFloat(document.getElementById('keywordWeight').value),
+                content_weight: parseFloat(document.getElementById('contentWeight').value)
+            };
+            
+            try {
+                const res = await fetch('/api/config/index', {
+                    method: 'POST',
+                    headers: {'Content-Type': 'application/json'},
+                    body: JSON.stringify(config)
+                });
+                const data = await res.json();
+                
+                if (data.success) {
+                    alert('索引配置已保存！');
+                } else {
+                    alert('保存失败: ' + data.error);
+                }
+            } catch (err) {
+                alert('保存失败: ' + err.message);
+            }
+        });
+
+        // 保存文档配置
+        document.getElementById('docConfigForm')?.addEventListener('submit', async function(e) {
+            e.preventDefault();
+            
+            const config = {
+                chunk_size: parseInt(document.getElementById('chunkSize').value),
+                chunk_overlap: parseInt(document.getElementById('chunkOverlap').value),
+                max_keywords: parseInt(document.getElementById('maxKeywords').value)
+            };
+            
+            try {
+                const res = await fetch('/api/config/doc', {
+                    method: 'POST',
+                    headers: {'Content-Type': 'application/json'},
+                    body: JSON.stringify(config)
+                });
+                const data = await res.json();
+                
+                if (data.success) {
+                    alert('文档配置已保存！');
+                } else {
+                    alert('保存失败: ' + data.error);
+                }
+            } catch (err) {
+                alert('保存失败: ' + err.message);
+            }
+        });
+    </script>
+</body>
+</html>