Compare commits
21 Commits
f2af27245d
...
main
| Author | SHA1 | Date | |
|---|---|---|---|
| 8f6d8a43d3 | |||
| 6ec45b73ad | |||
| 73f1c2804f | |||
| 74d40f91c5 | |||
| d2e3c2db3e | |||
| be302839ee | |||
| 581e2b0ae0 | |||
| 975ebf536b | |||
| 38b0c7e62e | |||
| 8e46e635f1 | |||
| c2f50d3bd8 | |||
| 2adf9aef60 | |||
| 827371cb90 | |||
| e5d4724e82 | |||
| 9e7f9df384 | |||
| 47c89d888f | |||
| 6701df613b | |||
| ecad9ccd82 | |||
| 51350e3002 | |||
| 8e713be1ca | |||
| a9dc0d8b91 |
7
.claude/settings.local.json
Normal file
7
.claude/settings.local.json
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"permissions": {
|
||||
"allow": [
|
||||
"WebSearch"
|
||||
]
|
||||
}
|
||||
}
|
||||
35
.env.example
Normal file
35
.env.example
Normal file
@@ -0,0 +1,35 @@
|
||||
# ============================================================
|
||||
# FilesReadSystem 环境变量配置模板
|
||||
# 复制此文件为 .env 并填入实际值
|
||||
# ============================================================
|
||||
|
||||
# ==================== 应用配置 ====================
|
||||
DEBUG=false
|
||||
|
||||
# ==================== MongoDB ====================
|
||||
MONGO_ROOT_USER=admin
|
||||
MONGO_ROOT_PASSWORD=your_mongo_password
|
||||
MONGODB_DB_NAME=document_system
|
||||
|
||||
# ==================== MySQL ====================
|
||||
MYSQL_PASSWORD=your_mysql_password
|
||||
MYSQL_DATABASE=document
|
||||
|
||||
# ==================== Redis ====================
|
||||
REDIS_PASSWORD=your_redis_password
|
||||
|
||||
# ==================== LLM AI ====================
|
||||
LLM_API_KEY=your_llm_api_key
|
||||
LLM_BASE_URL=https://api.deepseek.com
|
||||
LLM_MODEL_NAME=deepseek-chat
|
||||
|
||||
# ==================== Supabase ====================
|
||||
SUPABASE_URL=https://your-project.supabase.co
|
||||
SUPABASE_ANON_KEY=your_anon_key
|
||||
SUPABASE_SERVICE_KEY=your_service_key
|
||||
|
||||
# ==================== Embedding / RAG ====================
|
||||
EMBEDDING_MODEL=all-MiniLM-L6-v2
|
||||
|
||||
# ==================== 前端配置 ====================
|
||||
VITE_APP_ID=your_app_id
|
||||
24
.gitignore
vendored
24
.gitignore
vendored
@@ -1,4 +1,5 @@
|
||||
/.git/
|
||||
/.gitignore
|
||||
/.idea/
|
||||
/.vscode/
|
||||
/backend/venv/
|
||||
@@ -18,11 +19,7 @@
|
||||
/frontend/.idea/
|
||||
/frontend/.env
|
||||
/frontend/*.log
|
||||
/技术路线.md
|
||||
/开发路径.md
|
||||
/开发日志_2026-03-16.md
|
||||
/frontendTest/
|
||||
/docs/
|
||||
|
||||
/frontend/src/api/
|
||||
/frontend/src/api/index.js
|
||||
/frontend/src/api/index.ts
|
||||
@@ -30,9 +27,22 @@
|
||||
/frontend/src/api/index.py
|
||||
/frontend/src/api/index.go
|
||||
/frontend/src/api/index.java
|
||||
|
||||
/frontend - 副本/
|
||||
|
||||
/docs/
|
||||
/frontend - 副本/*
|
||||
/frontendTest/
|
||||
/supabase.txt
|
||||
|
||||
**/__pycache__/*
|
||||
# 取消跟踪的文件 / Untracked files
|
||||
比赛备赛规划.md
|
||||
Q&A.xlsx
|
||||
package.json
|
||||
技术路线.md
|
||||
开发路径.md
|
||||
开发日志_2026-03-16.md
|
||||
/logs/
|
||||
|
||||
# Python cache
|
||||
**/__pycache__/**
|
||||
**.pyc
|
||||
|
||||
351
README.md
Normal file
351
README.md
Normal file
@@ -0,0 +1,351 @@
|
||||
# 智联文档
|
||||
|
||||
## 项目介绍 / Project Introduction
|
||||
|
||||
基于大语言模型的文档理解与多源数据融合系统，专为第十七届中国大学生服务外包创新创业大赛（A23赛题）开发。本系统利用大语言模型（LLM）解析、分析各类文档格式并提取结构化数据，支持通过自然语言指令自动填写模板表格。
|
||||
|
||||
A document understanding and multi-source data fusion system based on Large Language Models (LLM), developed for the 17th China University Student Service Outsourcing Innovation and Entrepreneurship Competition (Topic A23). This system uses LLMs to parse, analyze, and extract structured data from various document formats, supporting automatic template table filling through natural language instructions.
|
||||
|
||||
---
|
||||
|
||||
## 技术栈 / Technology Stack
|
||||
|
||||
| 层次 / Layer | 组件 / Component | 说明 / Description |
|
||||
|:---|:---|:---|
|
||||
| 后端 / Backend | FastAPI + Uvicorn | RESTful API，异步任务调度 / API & async task scheduling |
|
||||
| 前端 / Frontend | React + TypeScript + Vite | 文件上传、表格配置、聊天界面 / Upload, table config, chat UI |
|
||||
| 异步任务 / Async Tasks | Celery + Redis | 处理耗时的解析与AI提取 / Heavy parsing & AI extraction |
|
||||
| 文档数据库 / Document DB | MongoDB (Motor) | 元数据、提取结果、文档块存储 / Metadata, results, chunk storage |
|
||||
| 关系数据库 / Relational DB | MySQL (SQLAlchemy) | 结构化数据存储 / Structured data storage |
|
||||
| 缓存 / Cache | Redis | 缓存与任务队列 / Caching & task queue |
|
||||
| 向量检索 / Vector Search | FAISS | 高效相似性搜索 / Efficient similarity search |
|
||||
| AI集成 / AI Integration | LangChain-style + MiniMax API | RAG流水线、提示词管理 / RAG pipeline, prompt management |
|
||||
| 文档解析 / Document Parsing | python-docx, pandas, openpyxl, markdown-it | 多格式支持 / Multi-format support |
|
||||
|
||||
---
|
||||
|
||||
## 项目架构 / Project Architecture
|
||||
|
||||
```mermaid
|
||||
flowchart TB
|
||||
subgraph UI["用户界面 / User Interface"]
|
||||
Frontend["React + TypeScript + shadcn/ui"]
|
||||
end
|
||||
|
||||
subgraph Backend["FastAPI 后端 / Backend"]
|
||||
Upload["上传 API<br/>/upload"]
|
||||
Documents["文档管理<br/>/documents"]
|
||||
RAG["RAG 检索<br/>/rag/search"]
|
||||
AI["AI 分析<br/>/ai/analyze"]
|
||||
Template["模板填充<br/>/templates/fill"]
|
||||
Instruction["自然语言指令<br/>/instruction/execute"]
|
||||
Visual["可视化<br/>/visualization"]
|
||||
end
|
||||
|
||||
subgraph Data["数据层 / Data Layer"]
|
||||
MongoDB["MongoDB<br/>文档存储"]
|
||||
MySQL["MySQL<br/>结构化数据"]
|
||||
Redis["Redis<br/>缓存/队列"]
|
||||
FAISS["FAISS<br/>向量索引"]
|
||||
end
|
||||
|
||||
UI --> Backend
|
||||
Backend --> MongoDB
|
||||
Backend --> MySQL
|
||||
Backend --> Redis
|
||||
MongoDB --> FAISS
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 程序流程 / Program Flow
|
||||
|
||||
```mermaid
|
||||
flowchart TD
|
||||
Start([用户上传文档<br/>User Uploads Document]) --> Parse{解析文档格式<br/>Parse Document Format}
|
||||
|
||||
Parse -->|Excel| ParseXlsx["解析 Excel<br/>Parse XLSX"]
|
||||
Parse -->|Word| ParseDocx["解析 Word<br/>Parse DOCX"]
|
||||
Parse -->|Markdown| ParseMd["解析 Markdown<br/>Parse Markdown"]
|
||||
Parse -->|Text| ParseTxt["解析文本<br/>Parse Text"]
|
||||
|
||||
ParseXlsx --> Store1[(存储到<br/>MongoDB)]
|
||||
ParseDocx --> Store1
|
||||
ParseMd --> Store1
|
||||
ParseTxt --> Store1
|
||||
|
||||
Store1 --> Embed["Embedding 向量化<br/>Create Embeddings"]
|
||||
Embed --> Index[(索引到<br/>FAISS)]
|
||||
|
||||
Index --> TaskCreated{创建任务<br/>Create Task}
|
||||
|
||||
TaskCreated -->|同步| ProcessSync["同步处理<br/>Sync Process"]
|
||||
TaskCreated -->|异步| QueueTask["加入任务队列<br/>Queue to Celery"]
|
||||
|
||||
ProcessSync --> ReturnResult["返回结果<br/>Return Result"]
|
||||
|
||||
QueueTask --> CeleryWorker["Celery Worker<br/>异步处理"]
|
||||
CeleryWorker --> LLM["调用 LLM<br/>Call LLM API"]
|
||||
LLM --> StoreResult["存储结果<br/>Store Result"]
|
||||
StoreResult --> ReturnAsync["返回任务ID<br/>Return Task ID"]
|
||||
|
||||
ReturnResult --> End([完成<br/>Complete])
|
||||
ReturnAsync --> Poll{轮询任务状态<br/>Poll Task Status}
|
||||
Poll -->|进行中| Poll
|
||||
Poll -->|完成| GetResult["获取结果<br/>Get Result"]
|
||||
GetResult --> End
|
||||
|
||||
style Start fill:#e1f5fe
|
||||
style End fill:#c8e6c9
|
||||
style LLM fill:#fff3e0
|
||||
style CeleryWorker fill:#fff3e0
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 目录结构 / Directory Structure
|
||||
|
||||
```
|
||||
FilesReadSystem/
|
||||
├── backend/ # 后端服务（Python + FastAPI）
|
||||
│ ├── app/
|
||||
│ │ ├── api/endpoints/ # API路由层 / API endpoints
|
||||
│ │ │ ├── ai_analyze.py # AI分析接口 / AI analysis
|
||||
│ │ │ ├── documents.py # 文档管理 / Document management
|
||||
│ │ │ ├── instruction.py # 自然语言指令 / Natural language instruction
|
||||
│ │ │ ├── rag.py # RAG检索 / RAG retrieval
|
||||
│ │ │ ├── tasks.py # 任务管理 / Task management
|
||||
│ │ │ ├── templates.py # 模板管理 / Template management
|
||||
│ │ │ ├── upload.py # 文件上传 / File upload
|
||||
│ │ │ └── visualization.py # 可视化 / Visualization
|
||||
│ │ ├── core/
|
||||
│ │ │ ├── database/ # 数据库连接 / Database connections
|
||||
│ │ │ └── document_parser/ # 文档解析器 / Document parsers
|
||||
│ │ ├── services/ # 业务逻辑服务 / Business logic services
|
||||
│ │ │ ├── llm_service.py # LLM调用 / LLM service
|
||||
│ │ │ ├── rag_service.py # RAG流水线 / RAG pipeline
|
||||
│ │ │ ├── template_fill_service.py # 模板填充 / Template filling
|
||||
│ │ │ ├── excel_ai_service.py # Excel AI分析 / Excel AI analysis
|
||||
│ │ │ ├── word_ai_service.py # Word AI分析 / Word AI analysis
|
||||
│ │ │ └── table_rag_service.py # 表格RAG / Table RAG
|
||||
│ │ └── instruction/ # 指令解析与执行 / Instruction parsing & execution
|
||||
│ ├── requirements.txt # Python依赖 / Python dependencies
|
||||
│ └── README.md
|
||||
│
|
||||
├── frontend/ # 前端项目（React + TypeScript）
|
||||
│ ├── src/
|
||||
│ │ ├── pages/ # 页面组件 / Page components
|
||||
│ │ │ ├── Dashboard.tsx # 仪表板 / Dashboard
|
||||
│ │ │ ├── Documents.tsx # 文档管理 / Document management
|
||||
│ │ │ ├── TemplateFill.tsx # 模板填充 / Template fill
|
||||
│ │ │ └── InstructionChat.tsx # 指令聊天 / Instruction chat
|
||||
│ │ ├── components/ui/ # shadcn/ui组件库 / shadcn/ui components
|
||||
│ │ ├── contexts/ # React上下文 / React contexts
|
||||
│ │ ├── db/ # API调用封装 / API call wrappers
|
||||
│ │ └── supabase/functions/ # Edge函数 / Edge functions
|
||||
│ ├── package.json
|
||||
│ └── README.md
|
||||
│
|
||||
├── docs/ # 文档与测试数据 / Documentation & test data
|
||||
├── logs/ # 应用日志 / Application logs
|
||||
└── README.md # 本文件 / This file
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 主要功能 / Key Features
|
||||
|
||||
- **多格式文档解析** / Multi-format Document Parsing
|
||||
- Excel (.xlsx)
|
||||
- Word (.docx)
|
||||
- Markdown (.md)
|
||||
- Plain Text (.txt)
|
||||
|
||||
- **AI智能分析** / AI-Powered Analysis
|
||||
- 文档内容理解与摘要
|
||||
- 表格数据自动提取
|
||||
- 多文档联合推理
|
||||
|
||||
- **RAG检索增强** / RAG (Retrieval Augmented Generation)
|
||||
- 语义向量相似度搜索
|
||||
- 上下文感知的答案生成
|
||||
|
||||
- **模板自动填充** / Template Auto-fill
|
||||
- 智能表格模板识别
|
||||
- 自然语言指令驱动填写
|
||||
- 批量数据导入导出
|
||||
|
||||
- **自然语言指令** / Natural Language Instructions
|
||||
- 意图识别与解析
|
||||
- 多步骤任务自动执行
|
||||
|
||||
---
|
||||
|
||||
## API接口 / API Endpoints
|
||||
|
||||
| 方法 / Method | 路径 / Path | 说明 / Description |
|
||||
|:---|:---|:---|
|
||||
| GET | `/health` | 健康检查 / Health check |
|
||||
| POST | `/upload/document` | 单文件上传 / Single file upload |
|
||||
| POST | `/upload/documents` | 批量上传 / Batch upload |
|
||||
| GET | `/documents` | 文档库 / Document library |
|
||||
| GET | `/tasks/{task_id}` | 任务状态 / Task status |
|
||||
| POST | `/rag/search` | RAG语义搜索 / RAG search |
|
||||
| POST | `/templates/upload` | 模板上传 / Template upload |
|
||||
| POST | `/templates/fill` | 执行模板填充 / Execute template fill |
|
||||
| POST | `/ai/analyze/excel` | Excel AI分析 / Excel AI analysis |
|
||||
| POST | `/ai/analyze/word` | Word AI分析 / Word AI analysis |
|
||||
| POST | `/instruction/recognize` | 意图识别 / Intent recognition |
|
||||
| POST | `/instruction/execute` | 执行指令 / Execute instruction |
|
||||
| GET | `/visualization/statistics` | 统计图表 / Statistics charts |
|
||||
|
||||
---
|
||||
|
||||
## 环境配置 / Environment Setup
|
||||
|
||||
### 后端 / Backend
|
||||
|
||||
```bash
|
||||
cd backend
|
||||
|
||||
# 创建虚拟环境 / Create virtual environment
|
||||
python -m venv venv
|
||||
|
||||
# 激活虚拟环境 / Activate virtual environment
|
||||
# Windows PowerShell:
|
||||
.\venv\Scripts\Activate.ps1
|
||||
# Windows CMD:
|
||||
.\venv\Scripts\Activate.bat
|
||||
|
||||
# 安装依赖 / Install dependencies
|
||||
pip install -r requirements.txt
|
||||
|
||||
# 复制环境变量模板 / Copy environment template
|
||||
copy .env.example .env
|
||||
# 编辑 .env 填入API密钥 / Edit .env with your API keys
|
||||
```
|
||||
|
||||
### 前端 / Frontend
|
||||
|
||||
```bash
|
||||
cd frontend
|
||||
|
||||
# 安装依赖 / Install dependencies
|
||||
npm install
|
||||
|
||||
# 或使用 pnpm / Or using pnpm
|
||||
pnpm install
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 启动项目 / Starting the Project
|
||||
|
||||
### 后端启动 / Backend Startup
|
||||
|
||||
```bash
|
||||
cd backend
|
||||
./venv/Scripts/python.exe -m uvicorn app.main:app --host 127.0.0.1 --port 8000 --reload
|
||||
```
|
||||
|
||||
### 前端启动 / Frontend Startup
|
||||
|
||||
```bash
|
||||
cd frontend
|
||||
npm run dev
|
||||
# 或 / or
|
||||
pnpm dev
|
||||
```
|
||||
|
||||
前端地址 / Frontend URL: http://localhost:5173
|
||||
|
||||
---
|
||||
|
||||
## 配置说明 / Configuration
|
||||
|
||||
### 环境变量 / Environment Variables
|
||||
|
||||
| 变量 / Variable | 说明 / Description |
|
||||
|:---|:---|
|
||||
| `MONGODB_URL` | MongoDB连接地址 / MongoDB connection URL |
|
||||
| `MYSQL_HOST` | MySQL主机 / MySQL host |
|
||||
| `REDIS_URL` | Redis连接地址 / Redis connection URL |
|
||||
| `MINIMAX_API_KEY` | MiniMax API密钥 / MiniMax API key |
|
||||
| `MINIMAX_API_URL` | MiniMax API地址 / MiniMax API URL |
|
||||
|
||||
---
|
||||
|
||||
## Docker 部署 / Docker Deployment
|
||||
|
||||
### 快速启动 / Quick Start
|
||||
|
||||
```bash
|
||||
# 1. 复制环境变量模板并编辑
|
||||
cp .env.example .env
|
||||
# 编辑 .env 填入实际配置
|
||||
|
||||
# 2. 启动所有服务
|
||||
docker compose up -d
|
||||
|
||||
# 3. 查看日志
|
||||
docker compose logs -f
|
||||
|
||||
# 4. 检查服务状态
|
||||
docker compose ps
|
||||
|
||||
# 5. 更新部署
|
||||
docker compose up -d --build
|
||||
```
|
||||
|
||||
### 服务说明 / Services
|
||||
|
||||
| 服务 | 端口 | 说明 |
|
||||
|:---|:---|:---|
|
||||
| frontend | 80 | React 前端 (Nginx) |
|
||||
| backend | 8000 | FastAPI 后端 |
|
||||
| mongodb | 27017 | MongoDB 数据库 |
|
||||
| mysql | 3306 | MySQL 数据库 |
|
||||
| redis | 6379 | Redis 缓存/队列 |
|
||||
|
||||
### 环境变量 / Environment Variables
|
||||
|
||||
创建 `.env` 文件，参考 `.env.example`:
|
||||
|
||||
```bash
|
||||
# 数据库配置
|
||||
MONGO_ROOT_USER=admin
|
||||
MONGO_ROOT_PASSWORD=your_password
|
||||
MONGODB_DB_NAME=document_system
|
||||
MYSQL_PASSWORD=your_password
|
||||
MYSQL_DATABASE=document
|
||||
REDIS_PASSWORD=your_password
|
||||
|
||||
# LLM 配置
|
||||
LLM_API_KEY=your_api_key
|
||||
LLM_BASE_URL=https://api.deepseek.com
|
||||
LLM_MODEL_NAME=deepseek-chat
|
||||
|
||||
# Supabase 配置
|
||||
SUPABASE_URL=https://your-project.supabase.co
|
||||
SUPABASE_ANON_KEY=your_anon_key
|
||||
SUPABASE_SERVICE_KEY=your_service_key
|
||||
```
|
||||
|
||||
### 验证部署 / Verify Deployment
|
||||
|
||||
```bash
|
||||
# 检查所有服务状态
|
||||
docker compose ps
|
||||
|
||||
# 访问前端
|
||||
curl http://localhost
|
||||
|
||||
# 检查后端健康
|
||||
curl http://localhost:8000/health
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 许可证 / License
|
||||
|
||||
ISC
|
||||
@@ -34,9 +34,9 @@ REDIS_URL="redis://localhost:6379/0"
|
||||
# - 模型: glm-4-flash (快速文本模型), glm-4 (标准), glm-4-plus (高性能)
|
||||
# - API: https://open.bigmodel.cn
|
||||
# - API Key: https://open.bigmodel.cn/usercenter/apikeys
|
||||
LLM_API_KEY="ca79ad9f96524cd5afc3e43ca97f347d.cpiLLx2oyitGvTeU"
|
||||
LLM_BASE_URL="https://open.bigmodel.cn/api/paas/v4"
|
||||
LLM_MODEL_NAME="glm-4v-plus"
|
||||
LLM_API_KEY="your_llm_api_key_here"
|
||||
LLM_BASE_URL="https://api.deepseek.com"
|
||||
LLM_MODEL_NAME="deepseek-chat"
|
||||
|
||||
# ==================== Supabase 配置 ====================
|
||||
# Supabase 项目配置
|
||||
@@ -45,10 +45,14 @@ SUPABASE_ANON_KEY="your_supabase_anon_key_here"
|
||||
SUPABASE_SERVICE_KEY="your_supabase_service_key_here"
|
||||
|
||||
# ==================== 文件路径配置 ====================
|
||||
# 上传文件存储目录 (相对于项目根目录)
|
||||
# 上传文件存储目录
|
||||
# 本地开发: ./data/uploads
|
||||
# Docker部署: /app/data/uploads
|
||||
UPLOAD_DIR="./data/uploads"
|
||||
|
||||
# Faiss 向量数据库持久化目录 (LangChain + Faiss 实现)
|
||||
# Faiss 向量数据库持久化目录
|
||||
# 本地开发: ./data/faiss
|
||||
# Docker部署: /app/data/faiss
|
||||
FAISS_INDEX_DIR="./data/faiss"
|
||||
|
||||
# ==================== RAG 配置 ====================
|
||||
|
||||
38
backend/=3.0.0
Normal file
38
backend/=3.0.0
Normal file
@@ -0,0 +1,38 @@
|
||||
Requirement already satisfied: sentence-transformers in c:\python312\lib\site-packages (2.2.2)
|
||||
Requirement already satisfied: transformers<5.0.0,>=4.6.0 in c:\python312\lib\site-packages (from sentence-transformers) (4.57.6)
|
||||
Requirement already satisfied: tqdm in c:\python312\lib\site-packages (from sentence-transformers) (4.66.1)
|
||||
Requirement already satisfied: torch>=1.6.0 in c:\python312\lib\site-packages (from sentence-transformers) (2.10.0)
|
||||
Requirement already satisfied: torchvision in c:\python312\lib\site-packages (from sentence-transformers) (0.25.0)
|
||||
Requirement already satisfied: numpy in c:\python312\lib\site-packages (from sentence-transformers) (1.26.2)
|
||||
Requirement already satisfied: scikit-learn in c:\python312\lib\site-packages (from sentence-transformers) (1.8.0)
|
||||
Requirement already satisfied: scipy in c:\python312\lib\site-packages (from sentence-transformers) (1.16.3)
|
||||
Requirement already satisfied: nltk in c:\python312\lib\site-packages (from sentence-transformers) (3.9.3)
|
||||
Requirement already satisfied: sentencepiece in c:\python312\lib\site-packages (from sentence-transformers) (0.2.1)
|
||||
Requirement already satisfied: huggingface-hub>=0.4.0 in c:\python312\lib\site-packages (from sentence-transformers) (0.36.2)
|
||||
Requirement already satisfied: filelock in c:\python312\lib\site-packages (from huggingface-hub>=0.4.0->sentence-transformers) (3.25.2)
|
||||
Requirement already satisfied: fsspec>=2023.5.0 in c:\python312\lib\site-packages (from huggingface-hub>=0.4.0->sentence-transformers) (2026.2.0)
|
||||
Requirement already satisfied: packaging>=20.9 in c:\python312\lib\site-packages (from huggingface-hub>=0.4.0->sentence-transformers) (23.2)
|
||||
Requirement already satisfied: pyyaml>=5.1 in c:\python312\lib\site-packages (from huggingface-hub>=0.4.0->sentence-transformers) (6.0.1)
|
||||
Requirement already satisfied: requests in c:\python312\lib\site-packages (from huggingface-hub>=0.4.0->sentence-transformers) (2.31.0)
|
||||
Requirement already satisfied: typing-extensions>=3.7.4.3 in c:\python312\lib\site-packages (from huggingface-hub>=0.4.0->sentence-transformers) (4.15.0)
|
||||
Requirement already satisfied: sympy>=1.13.3 in c:\python312\lib\site-packages (from torch>=1.6.0->sentence-transformers) (1.14.0)
|
||||
Requirement already satisfied: networkx>=2.5.1 in c:\python312\lib\site-packages (from torch>=1.6.0->sentence-transformers) (3.6.1)
|
||||
Requirement already satisfied: jinja2 in c:\python312\lib\site-packages (from torch>=1.6.0->sentence-transformers) (3.1.6)
|
||||
Requirement already satisfied: setuptools in c:\python312\lib\site-packages (from torch>=1.6.0->sentence-transformers) (82.0.1)
|
||||
Requirement already satisfied: colorama in c:\python312\lib\site-packages (from tqdm->sentence-transformers) (0.4.6)
|
||||
Requirement already satisfied: regex!=2019.12.17 in c:\python312\lib\site-packages (from transformers<5.0.0,>=4.6.0->sentence-transformers) (2026.2.28)
|
||||
Requirement already satisfied: tokenizers<=0.23.0,>=0.22.0 in c:\python312\lib\site-packages (from transformers<5.0.0,>=4.6.0->sentence-transformers) (0.22.2)
|
||||
Requirement already satisfied: safetensors>=0.4.3 in c:\python312\lib\site-packages (from transformers<5.0.0,>=4.6.0->sentence-transformers) (0.7.0)
|
||||
Requirement already satisfied: click in c:\python312\lib\site-packages (from nltk->sentence-transformers) (8.3.1)
|
||||
Requirement already satisfied: joblib in c:\python312\lib\site-packages (from nltk->sentence-transformers) (1.5.3)
|
||||
Requirement already satisfied: threadpoolctl>=3.2.0 in c:\python312\lib\site-packages (from scikit-learn->sentence-transformers) (3.6.0)
|
||||
Requirement already satisfied: pillow!=8.3.*,>=5.3.0 in c:\python312\lib\site-packages (from torchvision->sentence-transformers) (12.1.1)
|
||||
Requirement already satisfied: mpmath<1.4,>=1.1.0 in c:\python312\lib\site-packages (from sympy>=1.13.3->torch>=1.6.0->sentence-transformers) (1.3.0)
|
||||
Requirement already satisfied: MarkupSafe>=2.0 in c:\python312\lib\site-packages (from jinja2->torch>=1.6.0->sentence-transformers) (3.0.3)
|
||||
Requirement already satisfied: charset-normalizer<4,>=2 in c:\python312\lib\site-packages (from requests->huggingface-hub>=0.4.0->sentence-transformers) (3.4.6)
|
||||
Requirement already satisfied: idna<4,>=2.5 in c:\python312\lib\site-packages (from requests->huggingface-hub>=0.4.0->sentence-transformers) (3.11)
|
||||
Requirement already satisfied: urllib3<3,>=1.21.1 in c:\python312\lib\site-packages (from requests->huggingface-hub>=0.4.0->sentence-transformers) (2.6.3)
|
||||
Requirement already satisfied: certifi>=2017.4.17 in c:\python312\lib\site-packages (from requests->huggingface-hub>=0.4.0->sentence-transformers) (2026.2.25)
|
||||
|
||||
[notice] A new release of pip is available: 24.2 -> 26.0.1
|
||||
[notice] To update, run: python.exe -m pip install --upgrade pip
|
||||
7
backend/=4.0.0
Normal file
7
backend/=4.0.0
Normal file
@@ -0,0 +1,7 @@
|
||||
Collecting reportlab
|
||||
Using cached reportlab-4.4.10-py3-none-any.whl.metadata (1.7 kB)
|
||||
Requirement already satisfied: pillow>=9.0.0 in d:\code\filesreadsystem\backend\venv\lib\site-packages (from reportlab) (12.1.1)
|
||||
Requirement already satisfied: charset-normalizer in d:\code\filesreadsystem\backend\venv\lib\site-packages (from reportlab) (3.4.6)
|
||||
Using cached reportlab-4.4.10-py3-none-any.whl (2.0 MB)
|
||||
Installing collected packages: reportlab
|
||||
Successfully installed reportlab-4.4.10
|
||||
40
backend/Dockerfile
Normal file
40
backend/Dockerfile
Normal file
@@ -0,0 +1,40 @@
|
||||
# ============================================================
|
||||
# FilesReadSystem Backend Docker Image
|
||||
# ============================================================
|
||||
FROM python:3.12-slim
|
||||
|
||||
ENV PYTHONDONTWRITEBYTECODE=1
|
||||
ENV PYTHONUNBUFFERED=1
|
||||
|
||||
# 安装系统依赖 (FAISS, Pillow, tesseract 等)
|
||||
RUN apt-get update && apt-get install -y --no-install-recommends \
|
||||
gcc \
|
||||
g++ \
|
||||
libgl1-mesa-glx \
|
||||
libglib2.0-0 \
|
||||
tesseract-ocr \
|
||||
&& rm -rf /var/lib/apt/lists/*
|
||||
|
||||
WORKDIR /app
|
||||
|
||||
# 先复制依赖文件，再安装（利用 Docker 缓存）
|
||||
COPY requirements.txt .
|
||||
|
||||
# 安装 Python 依赖
|
||||
RUN pip install --no-cache-dir -r requirements.txt
|
||||
|
||||
# 复制应用代码
|
||||
COPY app/ ./app/
|
||||
|
||||
# 创建数据目录
|
||||
RUN mkdir -p /app/data/uploads /app/data/faiss /app/data/logs
|
||||
|
||||
# 暴露端口
|
||||
EXPOSE 8000
|
||||
|
||||
# 健康检查
|
||||
HEALTHCHECK --interval=30s --timeout=10s --start-period=10s --retries=3 \
|
||||
CMD python -c "import httpx; httpx.get('http://localhost:8000/health')" || exit 1
|
||||
|
||||
# 启动命令
|
||||
CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]
|
||||
@@ -13,6 +13,9 @@ from app.api.endpoints import (
|
||||
visualization,
|
||||
analysis_charts,
|
||||
health,
|
||||
instruction, # 智能指令
|
||||
conversation, # 对话历史
|
||||
pdf_converter, # PDF转换
|
||||
)
|
||||
|
||||
# 创建主路由
|
||||
@@ -29,3 +32,6 @@ api_router.include_router(templates.router) # 表格模板
|
||||
api_router.include_router(ai_analyze.router) # AI分析
|
||||
api_router.include_router(visualization.router) # 可视化
|
||||
api_router.include_router(analysis_charts.router) # 分析图表
|
||||
api_router.include_router(instruction.router) # 智能指令
|
||||
api_router.include_router(conversation.router) # 对话历史
|
||||
api_router.include_router(pdf_converter.router) # PDF转换
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
"""
|
||||
AI 分析 API 接口
|
||||
"""
|
||||
from fastapi import APIRouter, UploadFile, File, HTTPException, Query, Body
|
||||
from fastapi import APIRouter, UploadFile, File, HTTPException, Query, Body, Form
|
||||
from fastapi.responses import StreamingResponse
|
||||
from typing import Optional
|
||||
import logging
|
||||
@@ -12,6 +12,7 @@ from app.services.excel_ai_service import excel_ai_service
|
||||
from app.services.markdown_ai_service import markdown_ai_service
|
||||
from app.services.template_fill_service import template_fill_service
|
||||
from app.services.word_ai_service import word_ai_service
|
||||
from app.services.txt_ai_service import txt_ai_service
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@@ -20,7 +21,8 @@ router = APIRouter(prefix="/ai", tags=["AI 分析"])
|
||||
|
||||
@router.post("/analyze/excel")
|
||||
async def analyze_excel(
|
||||
file: UploadFile = File(...),
|
||||
file: Optional[UploadFile] = File(None),
|
||||
doc_id: Optional[str] = Form(None, description="文档ID（从数据库读取）"),
|
||||
user_prompt: str = Query("", description="用户自定义提示词"),
|
||||
analysis_type: str = Query("general", description="分析类型: general, summary, statistics, insights"),
|
||||
parse_all_sheets: bool = Query(False, description="是否分析所有工作表")
|
||||
@@ -29,7 +31,8 @@ async def analyze_excel(
|
||||
上传并使用 AI 分析 Excel 文件
|
||||
|
||||
Args:
|
||||
file: 上传的 Excel 文件
|
||||
file: 上传的 Excel 文件（与 doc_id 二选一）
|
||||
doc_id: 文档ID（从数据库读取）
|
||||
user_prompt: 用户自定义提示词
|
||||
analysis_type: 分析类型
|
||||
parse_all_sheets: 是否分析所有工作表
|
||||
@@ -37,7 +40,57 @@ async def analyze_excel(
|
||||
Returns:
|
||||
dict: 分析结果，包含 Excel 数据和 AI 分析结果
|
||||
"""
|
||||
# 检查文件类型
|
||||
filename = None
|
||||
|
||||
# 从数据库读取模式
|
||||
if doc_id:
|
||||
try:
|
||||
from app.core.database.mongodb import mongodb
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
if not doc:
|
||||
raise HTTPException(status_code=404, detail=f"文档不存在: {doc_id}")
|
||||
|
||||
filename = doc.get("metadata", {}).get("original_filename", "unknown.xlsx")
|
||||
file_ext = filename.split('.')[-1].lower()
|
||||
|
||||
if file_ext not in ['xlsx', 'xls']:
|
||||
raise HTTPException(status_code=400, detail=f"文档类型不是 Excel: {file_ext}")
|
||||
|
||||
file_path = doc.get("metadata", {}).get("file_path")
|
||||
if not file_path:
|
||||
raise HTTPException(status_code=400, detail="文档没有存储文件路径，请重新上传")
|
||||
|
||||
# 使用文件路径进行 AI 分析
|
||||
if parse_all_sheets:
|
||||
result = await excel_ai_service.batch_analyze_sheets_from_path(
|
||||
file_path=file_path,
|
||||
filename=filename,
|
||||
user_prompt=user_prompt,
|
||||
analysis_type=analysis_type
|
||||
)
|
||||
else:
|
||||
result = await excel_ai_service.analyze_excel_file_from_path(
|
||||
file_path=file_path,
|
||||
filename=filename,
|
||||
user_prompt=user_prompt,
|
||||
analysis_type=analysis_type
|
||||
)
|
||||
|
||||
if result.get("success"):
|
||||
return result
|
||||
else:
|
||||
return result
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"从数据库读取 Excel 文档失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"读取文档失败: {str(e)}")
|
||||
|
||||
# 文件上传模式
|
||||
if not file:
|
||||
raise HTTPException(status_code=400, detail="请提供文件或文档ID")
|
||||
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
@@ -60,7 +113,11 @@ async def analyze_excel(
|
||||
# 读取文件内容
|
||||
content = await file.read()
|
||||
|
||||
logger.info(f"开始分析文件: {file.filename}, 分析类型: {analysis_type}")
|
||||
# 验证文件内容不为空
|
||||
if not content:
|
||||
raise HTTPException(status_code=400, detail="文件内容为空，请确保文件已正确上传")
|
||||
|
||||
logger.info(f"开始分析文件: {file.filename}, 分析类型: {analysis_type}, 文件大小: {len(content)} bytes")
|
||||
|
||||
# 调用 AI 分析服务
|
||||
if parse_all_sheets:
|
||||
@@ -153,8 +210,9 @@ async def analyze_text(
|
||||
|
||||
@router.post("/analyze/md")
|
||||
async def analyze_markdown(
|
||||
file: UploadFile = File(...),
|
||||
analysis_type: str = Query("summary", description="分析类型: summary, outline, key_points, questions, tags, qa, statistics, section"),
|
||||
file: Optional[UploadFile] = File(None),
|
||||
doc_id: Optional[str] = Form(None, description="文档ID（从数据库读取）"),
|
||||
analysis_type: str = Query("summary", description="分析类型: summary, outline, key_points, questions, tags, qa, statistics, section, charts"),
|
||||
user_prompt: str = Query("", description="用户自定义提示词"),
|
||||
section_number: Optional[str] = Query(None, description="指定章节编号，如 '一' 或 '（一）'")
|
||||
):
|
||||
@@ -162,7 +220,8 @@ async def analyze_markdown(
|
||||
上传并使用 AI 分析 Markdown 文件
|
||||
|
||||
Args:
|
||||
file: 上传的 Markdown 文件
|
||||
file: 上传的 Markdown 文件（与 doc_id 二选一）
|
||||
doc_id: 文档ID（从数据库读取）
|
||||
analysis_type: 分析类型
|
||||
user_prompt: 用户自定义提示词
|
||||
section_number: 指定分析的章节编号
|
||||
@@ -170,16 +229,8 @@ async def analyze_markdown(
|
||||
Returns:
|
||||
dict: 分析结果
|
||||
"""
|
||||
# 检查文件类型
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
file_ext = file.filename.split('.')[-1].lower()
|
||||
if file_ext not in ['md', 'markdown']:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的文件类型: {file_ext}，仅支持 .md 和 .markdown"
|
||||
)
|
||||
filename = None
|
||||
tmp_path = None
|
||||
|
||||
# 验证分析类型
|
||||
supported_types = markdown_ai_service.get_supported_analysis_types()
|
||||
@@ -189,46 +240,96 @@ async def analyze_markdown(
|
||||
detail=f"不支持的分析类型: {analysis_type}，支持的类型: {', '.join(supported_types)}"
|
||||
)
|
||||
|
||||
try:
|
||||
# 读取文件内容
|
||||
content = await file.read()
|
||||
|
||||
# 保存到临时文件
|
||||
with tempfile.NamedTemporaryFile(mode='wb', suffix='.md', delete=False) as tmp:
|
||||
tmp.write(content)
|
||||
tmp_path = tmp.name
|
||||
|
||||
if doc_id:
|
||||
# 从数据库读取文档
|
||||
try:
|
||||
logger.info(f"开始分析 Markdown 文件: {file.filename}, 分析类型: {analysis_type}, 章节: {section_number}")
|
||||
from app.core.database.mongodb import mongodb
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
if not doc:
|
||||
raise HTTPException(status_code=404, detail=f"文档不存在: {doc_id}")
|
||||
|
||||
# 调用 AI 分析服务
|
||||
result = await markdown_ai_service.analyze_markdown(
|
||||
file_path=tmp_path,
|
||||
analysis_type=analysis_type,
|
||||
user_prompt=user_prompt,
|
||||
section_number=section_number
|
||||
filename = doc.get("metadata", {}).get("original_filename", "unknown.md")
|
||||
file_ext = filename.split('.')[-1].lower()
|
||||
|
||||
if file_ext not in ['md', 'markdown']:
|
||||
raise HTTPException(status_code=400, detail=f"文档类型不是 Markdown: {file_ext}")
|
||||
|
||||
content = doc.get("content") or ""
|
||||
if not content:
|
||||
raise HTTPException(status_code=400, detail="文档内容为空")
|
||||
|
||||
# 保存到临时文件
|
||||
with tempfile.NamedTemporaryFile(mode='wb', suffix='.md', delete=False) as tmp:
|
||||
tmp.write(content.encode('utf-8'))
|
||||
tmp_path = tmp.name
|
||||
|
||||
logger.info(f"从数据库加载 Markdown 文档: {filename}, 长度: {len(content)}")
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"从数据库读取 Markdown 文档失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"读取文档失败: {str(e)}")
|
||||
else:
|
||||
# 文件上传模式
|
||||
if not file:
|
||||
raise HTTPException(status_code=400, detail="请提供文件或文档ID")
|
||||
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
file_ext = file.filename.split('.')[-1].lower()
|
||||
if file_ext not in ['md', 'markdown']:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的文件类型: {file_ext}，仅支持 .md 和 .markdown"
|
||||
)
|
||||
|
||||
logger.info(f"Markdown 分析完成: {file.filename}, 成功: {result['success']}")
|
||||
try:
|
||||
# 读取文件内容
|
||||
content = await file.read()
|
||||
|
||||
if not result['success']:
|
||||
raise HTTPException(status_code=500, detail=result.get('error', '分析失败'))
|
||||
# 保存到临时文件
|
||||
with tempfile.NamedTemporaryFile(mode='wb', suffix='.md', delete=False) as tmp:
|
||||
tmp.write(content)
|
||||
tmp_path = tmp.name
|
||||
|
||||
return result
|
||||
filename = file.filename
|
||||
|
||||
finally:
|
||||
# 清理临时文件，确保在所有情况下都能清理
|
||||
try:
|
||||
if tmp_path and os.path.exists(tmp_path):
|
||||
os.unlink(tmp_path)
|
||||
except Exception as cleanup_error:
|
||||
logger.warning(f"临时文件清理失败: {tmp_path}, error: {cleanup_error}")
|
||||
except Exception as e:
|
||||
logger.error(f"读取 Markdown 文件失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"读取文件失败: {str(e)}")
|
||||
|
||||
try:
|
||||
logger.info(f"开始分析 Markdown 文件: {filename}, 分析类型: {analysis_type}, 章节: {section_number}")
|
||||
|
||||
# 调用 AI 分析服务
|
||||
result = await markdown_ai_service.analyze_markdown(
|
||||
file_path=tmp_path,
|
||||
analysis_type=analysis_type,
|
||||
user_prompt=user_prompt,
|
||||
section_number=section_number
|
||||
)
|
||||
|
||||
logger.info(f"Markdown 分析完成: {filename}, 成功: {result['success']}")
|
||||
|
||||
if not result['success']:
|
||||
raise HTTPException(status_code=500, detail=result.get('error', '分析失败'))
|
||||
|
||||
return result
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"Markdown AI 分析过程中出错: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"分析失败: {str(e)}")
|
||||
finally:
|
||||
# 清理临时文件
|
||||
if tmp_path and os.path.exists(tmp_path):
|
||||
try:
|
||||
os.unlink(tmp_path)
|
||||
except Exception as cleanup_error:
|
||||
logger.warning(f"临时文件清理失败: {tmp_path}, error: {cleanup_error}")
|
||||
|
||||
|
||||
@router.post("/analyze/md/stream")
|
||||
@@ -346,67 +447,100 @@ async def get_markdown_outline(
|
||||
|
||||
@router.post("/analyze/txt")
|
||||
async def analyze_txt(
|
||||
file: UploadFile = File(...),
|
||||
file: Optional[UploadFile] = File(None),
|
||||
doc_id: Optional[str] = Form(None, description="文档ID（从数据库读取）"),
|
||||
analysis_type: str = Query("structured", description="分析类型: structured, charts")
|
||||
):
|
||||
"""
|
||||
上传并使用 AI 分析 TXT 文本文件，提取结构化数据
|
||||
上传并使用 AI 分析 TXT 文本文件，提取结构化数据或生成图表
|
||||
|
||||
将非结构化文本转换为结构化表格数据，便于后续填表使用
|
||||
当 analysis_type=charts 时，可生成可视化图表
|
||||
|
||||
Args:
|
||||
file: 上传的 TXT 文件
|
||||
file: 上传的 TXT 文件（与 doc_id 二选一）
|
||||
doc_id: 文档ID（从数据库读取）
|
||||
analysis_type: 分析类型 - "structured"（默认，提取结构化数据）或 "charts"（生成图表）
|
||||
|
||||
Returns:
|
||||
dict: 分析结果，包含结构化表格数据
|
||||
dict: 分析结果，包含结构化表格数据或图表数据
|
||||
"""
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
file_ext = file.filename.split('.')[-1].lower()
|
||||
if file_ext not in ['txt', 'text']:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的文件类型: {file_ext}，仅支持 .txt"
|
||||
)
|
||||
|
||||
try:
|
||||
# 读取文件内容
|
||||
content = await file.read()
|
||||
|
||||
# 保存到临时文件
|
||||
with tempfile.NamedTemporaryFile(mode='wb', suffix='.txt', delete=False) as tmp:
|
||||
tmp.write(content)
|
||||
tmp_path = tmp.name
|
||||
filename = None
|
||||
text_content = None
|
||||
|
||||
if doc_id:
|
||||
# 从数据库读取文档
|
||||
try:
|
||||
logger.info(f"开始 AI 分析 TXT 文件: {file.filename}")
|
||||
from app.core.database.mongodb import mongodb
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
if not doc:
|
||||
raise HTTPException(status_code=404, detail=f"文档不存在: {doc_id}")
|
||||
|
||||
# 使用 template_fill_service 的 AI 分析方法
|
||||
result = await template_fill_service.analyze_txt_with_ai(
|
||||
content=content.decode('utf-8', errors='replace'),
|
||||
filename=file.filename
|
||||
filename = doc.get("metadata", {}).get("original_filename", "unknown.txt")
|
||||
file_ext = filename.split('.')[-1].lower()
|
||||
|
||||
if file_ext not in ['txt', 'text']:
|
||||
raise HTTPException(status_code=400, detail=f"文档类型不是 TXT: {file_ext}")
|
||||
|
||||
# 使用数据库中的 content
|
||||
text_content = doc.get("content") or ""
|
||||
|
||||
if not text_content:
|
||||
raise HTTPException(status_code=400, detail="文档内容为空")
|
||||
|
||||
logger.info(f"从数据库加载 TXT 文档: {filename}, 长度: {len(text_content)}")
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"从数据库读取 TXT 文档失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"读取文档失败: {str(e)}")
|
||||
else:
|
||||
# 文件上传模式
|
||||
if not file:
|
||||
raise HTTPException(status_code=400, detail="请提供文件或文档ID")
|
||||
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
file_ext = file.filename.split('.')[-1].lower()
|
||||
if file_ext not in ['txt', 'text']:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的文件类型: {file_ext}，仅支持 .txt"
|
||||
)
|
||||
|
||||
if result:
|
||||
logger.info(f"TXT AI 分析成功: {file.filename}")
|
||||
return {
|
||||
"success": True,
|
||||
"filename": file.filename,
|
||||
"structured_data": result
|
||||
}
|
||||
else:
|
||||
logger.warning(f"TXT AI 分析返回空结果: {file.filename}")
|
||||
return {
|
||||
"success": False,
|
||||
"filename": file.filename,
|
||||
"error": "AI 分析未能提取到结构化数据",
|
||||
"structured_data": None
|
||||
}
|
||||
# 读取文件内容
|
||||
content = await file.read()
|
||||
text_content = content.decode('utf-8', errors='replace')
|
||||
filename = file.filename
|
||||
|
||||
finally:
|
||||
# 清理临时文件
|
||||
if os.path.exists(tmp_path):
|
||||
os.unlink(tmp_path)
|
||||
try:
|
||||
logger.info(f"开始 AI 分析 TXT 文件: {filename}, analysis_type={analysis_type}")
|
||||
|
||||
# 使用 txt_ai_service 的 AI 分析方法
|
||||
result = await txt_ai_service.analyze_txt_with_ai(
|
||||
content=text_content,
|
||||
filename=filename,
|
||||
analysis_type=analysis_type
|
||||
)
|
||||
|
||||
if result:
|
||||
logger.info(f"TXT AI 分析成功: {filename}")
|
||||
return {
|
||||
"success": result.get("success", True),
|
||||
"filename": filename,
|
||||
"analysis_type": analysis_type,
|
||||
"result": result
|
||||
}
|
||||
else:
|
||||
logger.warning(f"TXT AI 分析返回空结果: {filename}")
|
||||
return {
|
||||
"success": False,
|
||||
"filename": filename,
|
||||
"error": "AI 分析未能提取到结构化数据",
|
||||
"result": None
|
||||
}
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
@@ -419,21 +553,90 @@ async def analyze_txt(
|
||||
|
||||
@router.post("/analyze/word")
|
||||
async def analyze_word(
|
||||
file: UploadFile = File(...),
|
||||
user_hint: str = Query("", description="用户提示词，如'请提取表格数据'")
|
||||
file: Optional[UploadFile] = File(None),
|
||||
doc_id: Optional[str] = Form(None, description="文档ID（从数据库读取）"),
|
||||
user_hint: str = Form("", description="用户提示词，如'请提取表格数据'"),
|
||||
analysis_type: str = Query("structured", description="分析类型: structured, charts")
|
||||
):
|
||||
"""
|
||||
使用 AI 解析 Word 文档，提取结构化数据
|
||||
使用 AI 解析 Word 文档，提取结构化数据或生成图表
|
||||
|
||||
适用于从非结构化的 Word 文档中提取表格数据、键值对等信息
|
||||
当 analysis_type=charts 时，可生成可视化图表
|
||||
|
||||
Args:
|
||||
file: 上传的 Word 文件
|
||||
file: 上传的 Word 文件（与 doc_id 二选一）
|
||||
doc_id: 文档ID（从数据库读取）
|
||||
user_hint: 用户提示词
|
||||
analysis_type: 分析类型 - "structured"（默认，提取结构化数据）或 "charts"（生成图表）
|
||||
|
||||
Returns:
|
||||
dict: 包含结构化数据的解析结果
|
||||
dict: 包含结构化数据的解析结果或图表数据
|
||||
"""
|
||||
# 获取文件名和扩展名
|
||||
filename = None
|
||||
file_ext = None
|
||||
|
||||
if doc_id:
|
||||
# 从数据库读取文档
|
||||
try:
|
||||
from app.core.database.mongodb import mongodb
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
if not doc:
|
||||
raise HTTPException(status_code=404, detail=f"文档不存在: {doc_id}")
|
||||
|
||||
filename = doc.get("metadata", {}).get("original_filename", "unknown.docx")
|
||||
file_ext = filename.split('.')[-1].lower()
|
||||
|
||||
if file_ext not in ['docx']:
|
||||
raise HTTPException(status_code=400, detail=f"文档类型不是 Word: {file_ext}")
|
||||
|
||||
# 使用数据库中的 content 进行分析
|
||||
content = doc.get("content", "") or ""
|
||||
structured_data = doc.get("structured_data") or {}
|
||||
tables = structured_data.get("tables", [])
|
||||
|
||||
# 调用 AI 分析服务，传入数据库内容
|
||||
if analysis_type == "charts":
|
||||
result = await word_ai_service.generate_charts_from_db(
|
||||
content=content,
|
||||
tables=tables,
|
||||
filename=filename,
|
||||
user_hint=user_hint
|
||||
)
|
||||
else:
|
||||
result = await word_ai_service.parse_word_with_ai_from_db(
|
||||
content=content,
|
||||
tables=tables,
|
||||
filename=filename,
|
||||
user_hint=user_hint or "请提取文档中的所有结构化数据，包括表格、键值对等"
|
||||
)
|
||||
|
||||
if result.get("success"):
|
||||
return {
|
||||
"success": True,
|
||||
"filename": filename,
|
||||
"analysis_type": analysis_type,
|
||||
"result": result
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"filename": filename,
|
||||
"error": result.get("error", "AI 解析失败"),
|
||||
"result": None
|
||||
}
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"从数据库读取 Word 文档失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"读取文档失败: {str(e)}")
|
||||
|
||||
# 文件上传模式
|
||||
if not file:
|
||||
raise HTTPException(status_code=400, detail="请提供文件或文档ID")
|
||||
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
@@ -453,16 +656,25 @@ async def analyze_word(
|
||||
tmp_path = tmp.name
|
||||
|
||||
try:
|
||||
# 使用 AI 解析 Word 文档
|
||||
result = await word_ai_service.parse_word_with_ai(
|
||||
file_path=tmp_path,
|
||||
user_hint=user_hint or "请提取文档中的所有结构化数据，包括表格、键值对等"
|
||||
)
|
||||
# 根据 analysis_type 选择处理方式
|
||||
if analysis_type == "charts":
|
||||
# 生成图表
|
||||
result = await word_ai_service.generate_charts(
|
||||
file_path=tmp_path,
|
||||
user_hint=user_hint
|
||||
)
|
||||
else:
|
||||
# 提取结构化数据
|
||||
result = await word_ai_service.parse_word_with_ai(
|
||||
file_path=tmp_path,
|
||||
user_hint=user_hint or "请提取文档中的所有结构化数据，包括表格、键值对等"
|
||||
)
|
||||
|
||||
if result.get("success"):
|
||||
return {
|
||||
"success": True,
|
||||
"filename": file.filename,
|
||||
"analysis_type": analysis_type,
|
||||
"result": result
|
||||
}
|
||||
else:
|
||||
|
||||
98
backend/app/api/endpoints/conversation.py
Normal file
98
backend/app/api/endpoints/conversation.py
Normal file
@@ -0,0 +1,98 @@
|
||||
"""
|
||||
对话历史 API 接口
|
||||
|
||||
提供对话历史的存储和查询功能
|
||||
"""
|
||||
import logging
|
||||
from typing import Optional
|
||||
|
||||
from fastapi import APIRouter, HTTPException
|
||||
from pydantic import BaseModel
|
||||
|
||||
from app.core.database import mongodb
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(prefix="/conversation", tags=["对话历史"])
|
||||
|
||||
|
||||
# ==================== 请求/响应模型 ====================
|
||||
|
||||
class ConversationMessage(BaseModel):
|
||||
role: str
|
||||
content: str
|
||||
intent: Optional[str] = None
|
||||
|
||||
|
||||
class ConversationHistoryResponse(BaseModel):
|
||||
success: bool
|
||||
messages: list
|
||||
|
||||
|
||||
class ConversationListResponse(BaseModel):
|
||||
success: bool
|
||||
conversations: list
|
||||
|
||||
|
||||
# ==================== 接口 ====================
|
||||
|
||||
@router.get("/{conversation_id}/history", response_model=ConversationHistoryResponse)
|
||||
async def get_conversation_history(conversation_id: str, limit: int = 20):
|
||||
"""
|
||||
获取对话历史
|
||||
|
||||
Args:
|
||||
conversation_id: 对话会话ID
|
||||
limit: 返回消息数量（默认20条）
|
||||
"""
|
||||
try:
|
||||
messages = await mongodb.get_conversation_history(conversation_id, limit=limit)
|
||||
return ConversationHistoryResponse(
|
||||
success=True,
|
||||
messages=messages
|
||||
)
|
||||
except Exception as e:
|
||||
logger.error(f"获取对话历史失败: {e}")
|
||||
return ConversationHistoryResponse(
|
||||
success=False,
|
||||
messages=[]
|
||||
)
|
||||
|
||||
|
||||
@router.delete("/{conversation_id}")
|
||||
async def delete_conversation(conversation_id: str):
|
||||
"""
|
||||
删除对话会话
|
||||
|
||||
Args:
|
||||
conversation_id: 对话会话ID
|
||||
"""
|
||||
try:
|
||||
success = await mongodb.delete_conversation(conversation_id)
|
||||
return {"success": success}
|
||||
except Exception as e:
|
||||
logger.error(f"删除对话失败: {e}")
|
||||
return {"success": False, "error": str(e)}
|
||||
|
||||
|
||||
@router.get("/all", response_model=ConversationListResponse)
|
||||
async def list_conversations(limit: int = 50, skip: int = 0):
|
||||
"""
|
||||
获取会话列表
|
||||
|
||||
Args:
|
||||
limit: 返回数量
|
||||
skip: 跳过数量
|
||||
"""
|
||||
try:
|
||||
conversations = await mongodb.list_conversations(limit=limit, skip=skip)
|
||||
return ConversationListResponse(
|
||||
success=True,
|
||||
conversations=conversations
|
||||
)
|
||||
except Exception as e:
|
||||
logger.error(f"获取会话列表失败: {e}")
|
||||
return ConversationListResponse(
|
||||
success=False,
|
||||
conversations=[]
|
||||
)
|
||||
@@ -4,6 +4,7 @@
|
||||
支持多格式文档(docx/xlsx/md/txt)上传、解析、存储和RAG索引
|
||||
集成 Excel 存储和 AI 生成字段描述
|
||||
"""
|
||||
import asyncio
|
||||
import logging
|
||||
import uuid
|
||||
from typing import List, Optional
|
||||
@@ -257,51 +258,8 @@ async def process_document(
|
||||
structured_data=result.data.get("structured_data")
|
||||
)
|
||||
|
||||
# 如果是 Word 文档，使用 AI 深度解析
|
||||
if doc_type == "docx":
|
||||
await redis_db.set_task_status(
|
||||
task_id, status="processing",
|
||||
meta={"progress": 40, "message": "正在使用 AI 解析 Word 文档"}
|
||||
)
|
||||
|
||||
try:
|
||||
from app.services.word_ai_service import word_ai_service
|
||||
|
||||
logger.info(f"开始 AI 解析 Word 文档: {original_filename}")
|
||||
ai_result = await word_ai_service.parse_word_with_ai(
|
||||
file_path=file_path,
|
||||
user_hint="请提取文档中的所有结构化数据，包括表格、键值对、列表项等"
|
||||
)
|
||||
|
||||
if ai_result.get("success"):
|
||||
# 更新 MongoDB 文档，添加 AI 解析结果
|
||||
ai_data = {
|
||||
"ai_parsed": True,
|
||||
"parse_type": ai_result.get("type", "unknown"),
|
||||
"headers": ai_result.get("headers", []),
|
||||
"rows": ai_result.get("rows", []),
|
||||
"tables": ai_result.get("tables", []),
|
||||
"key_values": ai_result.get("key_values", {}),
|
||||
"list_items": ai_result.get("list_items", []),
|
||||
"summary": ai_result.get("summary", ""),
|
||||
"description": ai_result.get("description", "")
|
||||
}
|
||||
|
||||
await mongodb.update_document(doc_id, {
|
||||
"ai_analysis": ai_data,
|
||||
"structured_data": {
|
||||
**result.data.get("structured_data", {}),
|
||||
**ai_data
|
||||
}
|
||||
})
|
||||
logger.info(f"Word AI 解析成功: {original_filename}, type={ai_result.get('type')}")
|
||||
else:
|
||||
logger.warning(f"Word AI 解析返回失败: {ai_result.get('error')}")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Word AI 解析异常: {str(e)}", exc_info=True)
|
||||
|
||||
# 如果是 Excel，存储到 MySQL + AI生成描述 + RAG索引
|
||||
mysql_table_name = None
|
||||
if doc_type in ["xlsx", "xls"]:
|
||||
await update_task_status(
|
||||
task_id, status="processing",
|
||||
@@ -309,17 +267,29 @@ async def process_document(
|
||||
)
|
||||
|
||||
try:
|
||||
# 使用 TableRAG 服务完成建表和RAG索引
|
||||
# 使用 TableRAG 服务存储到 MySQL（跳过 RAG 索引以提升速度）
|
||||
logger.info(f"开始存储Excel到MySQL: {original_filename}, file_path: {file_path}")
|
||||
rag_result = await table_rag_service.build_table_rag_index(
|
||||
file_path=file_path,
|
||||
filename=original_filename,
|
||||
sheet_name=parse_options.get("sheet_name"),
|
||||
header_row=parse_options.get("header_row", 0)
|
||||
header_row=parse_options.get("header_row", 0),
|
||||
skip_rag_index=True # 跳过 AI 字段描述生成和索引
|
||||
)
|
||||
|
||||
if rag_result.get("success"):
|
||||
logger.info(f"Excel存储到MySQL成功: {original_filename}, table: {rag_result.get('table_name')}")
|
||||
mysql_table_name = rag_result.get('table_name')
|
||||
logger.info(f"Excel存储到MySQL成功: {original_filename}, table: {mysql_table_name}")
|
||||
# 更新 MongoDB 中的 metadata，记录 MySQL 表名
|
||||
try:
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
if doc:
|
||||
metadata = doc.get("metadata", {})
|
||||
metadata["mysql_table_name"] = mysql_table_name
|
||||
await mongodb.update_document_metadata(doc_id, metadata)
|
||||
logger.info(f"已更新 MongoDB 文档的 mysql_table_name: {mysql_table_name}")
|
||||
except Exception as update_err:
|
||||
logger.warning(f"更新 MongoDB mysql_table_name 失败: {update_err}")
|
||||
else:
|
||||
logger.error(f"RAG索引构建失败: {rag_result.get('error')}")
|
||||
except Exception as e:
|
||||
@@ -327,17 +297,16 @@ async def process_document(
|
||||
|
||||
else:
|
||||
# 非结构化文档
|
||||
await update_task_status(
|
||||
task_id, status="processing",
|
||||
progress=60, message="正在建立索引"
|
||||
)
|
||||
|
||||
# 如果文档中有表格数据，提取并存储到 MySQL + RAG
|
||||
structured_data = result.data.get("structured_data", {})
|
||||
tables = structured_data.get("tables", [])
|
||||
|
||||
# 如果文档中有表格数据，提取并存储到 MySQL（不需要 RAG 索引）
|
||||
if tables:
|
||||
# 对每个表格建立 MySQL 表和 RAG 索引
|
||||
await update_task_status(
|
||||
task_id, status="processing",
|
||||
progress=60, message="正在存储表格数据"
|
||||
)
|
||||
# 对每个表格建立 MySQL 表（跳过 RAG 索引，速度更快）
|
||||
for table_info in tables:
|
||||
await table_rag_service.index_document_table(
|
||||
doc_id=doc_id,
|
||||
@@ -346,8 +315,14 @@ async def process_document(
|
||||
source_doc_type=doc_type
|
||||
)
|
||||
|
||||
# 同时对文档内容建立 RAG 索引
|
||||
await index_document_to_rag(doc_id, original_filename, result, doc_type)
|
||||
# 对文档内容建立 RAG 索引（非结构化文本需要语义搜索）
|
||||
content = result.data.get("content", "")
|
||||
if content and len(content) > 50: # 只有内容足够长才建立索引
|
||||
await update_task_status(
|
||||
task_id, status="processing",
|
||||
progress=80, message="正在建立语义索引"
|
||||
)
|
||||
await index_document_to_rag(doc_id, original_filename, result, doc_type)
|
||||
|
||||
# 完成
|
||||
await update_task_status(
|
||||
@@ -372,72 +347,95 @@ async def process_document(
|
||||
|
||||
|
||||
async def process_documents_batch(task_id: str, files: List[dict]):
|
||||
"""批量处理文档"""
|
||||
"""批量并行处理文档"""
|
||||
try:
|
||||
await update_task_status(
|
||||
task_id, status="processing",
|
||||
progress=0, message="开始批量处理"
|
||||
progress=0, message=f"开始批量处理 {len(files)} 个文档",
|
||||
result={"total": len(files), "files": []}
|
||||
)
|
||||
|
||||
results = []
|
||||
for i, file_info in enumerate(files):
|
||||
async def process_single_file(file_info: dict, index: int) -> dict:
|
||||
"""处理单个文件"""
|
||||
filename = file_info["filename"]
|
||||
try:
|
||||
# 解析文档
|
||||
parser = ParserFactory.get_parser(file_info["path"])
|
||||
result = parser.parse(file_info["path"])
|
||||
|
||||
if result.success:
|
||||
doc_id = await mongodb.insert_document(
|
||||
doc_type=file_info["ext"],
|
||||
content=result.data.get("content", ""),
|
||||
metadata={
|
||||
**result.metadata,
|
||||
"original_filename": file_info["filename"],
|
||||
"file_path": file_info["path"]
|
||||
},
|
||||
structured_data=result.data.get("structured_data")
|
||||
if not result.success:
|
||||
return {"index": index, "filename": filename, "success": False, "error": result.error or "解析失败"}
|
||||
|
||||
# 存储到 MongoDB
|
||||
doc_id = await mongodb.insert_document(
|
||||
doc_type=file_info["ext"],
|
||||
content=result.data.get("content", ""),
|
||||
metadata={
|
||||
**result.metadata,
|
||||
"original_filename": filename,
|
||||
"file_path": file_info["path"]
|
||||
},
|
||||
structured_data=result.data.get("structured_data")
|
||||
)
|
||||
|
||||
# Excel 处理
|
||||
if file_info["ext"] in ["xlsx", "xls"]:
|
||||
await table_rag_service.build_table_rag_index(
|
||||
file_path=file_info["path"],
|
||||
filename=filename,
|
||||
skip_rag_index=True # 跳过 AI 字段描述生成和索引
|
||||
)
|
||||
|
||||
# Excel 处理
|
||||
if file_info["ext"] in ["xlsx", "xls"]:
|
||||
await table_rag_service.build_table_rag_index(
|
||||
file_path=file_info["path"],
|
||||
filename=file_info["filename"]
|
||||
)
|
||||
else:
|
||||
# 非结构化文档：处理其中的表格 + 内容索引
|
||||
structured_data = result.data.get("structured_data", {})
|
||||
tables = structured_data.get("tables", [])
|
||||
|
||||
if tables:
|
||||
for table_info in tables:
|
||||
await table_rag_service.index_document_table(
|
||||
doc_id=doc_id,
|
||||
filename=file_info["filename"],
|
||||
table_data=table_info,
|
||||
source_doc_type=file_info["ext"]
|
||||
)
|
||||
|
||||
await index_document_to_rag(doc_id, file_info["filename"], result, file_info["ext"])
|
||||
|
||||
results.append({"filename": file_info["filename"], "doc_id": doc_id, "success": True})
|
||||
else:
|
||||
results.append({"filename": file_info["filename"], "success": False, "error": result.error})
|
||||
# 非结构化文档
|
||||
structured_data = result.data.get("structured_data", {})
|
||||
tables = structured_data.get("tables", [])
|
||||
|
||||
# 表格数据直接存 MySQL（跳过 RAG 索引）
|
||||
if tables:
|
||||
for table_info in tables:
|
||||
await table_rag_service.index_document_table(
|
||||
doc_id=doc_id,
|
||||
filename=filename,
|
||||
table_data=table_info,
|
||||
source_doc_type=file_info["ext"]
|
||||
)
|
||||
|
||||
# 只有内容足够长才建立语义索引
|
||||
content = result.data.get("content", "")
|
||||
if content and len(content) > 50:
|
||||
await index_document_to_rag(doc_id, filename, result, file_info["ext"])
|
||||
|
||||
return {"index": index, "filename": filename, "doc_id": doc_id, "file_path": file_info["path"], "success": True}
|
||||
|
||||
except Exception as e:
|
||||
results.append({"filename": file_info["filename"], "success": False, "error": str(e)})
|
||||
logger.error(f"处理文件 {filename} 失败: {e}")
|
||||
return {"index": index, "filename": filename, "success": False, "error": str(e)}
|
||||
|
||||
progress = int((i + 1) / len(files) * 100)
|
||||
await update_task_status(
|
||||
task_id, status="processing",
|
||||
progress=progress, message=f"已处理 {i+1}/{len(files)}"
|
||||
)
|
||||
# 并行处理所有文档
|
||||
tasks = [process_single_file(f, i) for i, f in enumerate(files)]
|
||||
results = await asyncio.gather(*tasks)
|
||||
|
||||
# 按原始顺序排序
|
||||
results.sort(key=lambda x: x["index"])
|
||||
|
||||
# 统计成功/失败数量
|
||||
success_count = sum(1 for r in results if r["success"])
|
||||
fail_count = len(results) - success_count
|
||||
|
||||
# 更新最终状态
|
||||
await update_task_status(
|
||||
task_id, status="success",
|
||||
progress=100, message="批量处理完成",
|
||||
result={"results": results}
|
||||
progress=100, message=f"批量处理完成: {success_count} 成功, {fail_count} 失败",
|
||||
result={
|
||||
"total": len(files),
|
||||
"success": success_count,
|
||||
"failure": fail_count,
|
||||
"results": results
|
||||
}
|
||||
)
|
||||
|
||||
logger.info(f"批量处理完成: {success_count}/{len(files)} 成功")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"批量处理失败: {str(e)}")
|
||||
await update_task_status(
|
||||
@@ -448,18 +446,22 @@ async def process_documents_batch(task_id: str, files: List[dict]):
|
||||
|
||||
|
||||
async def index_document_to_rag(doc_id: str, filename: str, result: ParseResult, doc_type: str):
|
||||
"""将非结构化文档索引到 RAG"""
|
||||
"""将非结构化文档索引到 RAG（使用分块索引，异步执行）"""
|
||||
try:
|
||||
content = result.data.get("content", "")
|
||||
if content:
|
||||
rag_service.index_document_content(
|
||||
# 使用异步方法索引，避免阻塞事件循环
|
||||
await rag_service.index_document_content_async(
|
||||
doc_id=doc_id,
|
||||
content=content[:5000],
|
||||
content=content,
|
||||
metadata={
|
||||
"filename": filename,
|
||||
"doc_type": doc_type
|
||||
}
|
||||
},
|
||||
chunk_size=1000, # 每块 1000 字符，提升速度
|
||||
chunk_overlap=100 # 块之间 100 字符重叠
|
||||
)
|
||||
logger.info(f"RAG 索引完成: {filename}, doc_id={doc_id}")
|
||||
except Exception as e:
|
||||
logger.warning(f"RAG 索引失败: {str(e)}")
|
||||
|
||||
|
||||
472
backend/app/api/endpoints/instruction.py
Normal file
472
backend/app/api/endpoints/instruction.py
Normal file
@@ -0,0 +1,472 @@
|
||||
"""
|
||||
智能指令 API 接口
|
||||
|
||||
支持自然语言指令解析和执行
|
||||
"""
|
||||
import logging
|
||||
import uuid
|
||||
from typing import Any, Dict, List, Optional
|
||||
|
||||
from fastapi import APIRouter, HTTPException, Query, BackgroundTasks
|
||||
from pydantic import BaseModel
|
||||
|
||||
from app.instruction.intent_parser import intent_parser
|
||||
from app.instruction.executor import instruction_executor
|
||||
from app.core.database import mongodb
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(prefix="/instruction", tags=["智能指令"])
|
||||
|
||||
|
||||
# ==================== 请求/响应模型 ====================
|
||||
|
||||
class InstructionRequest(BaseModel):
|
||||
instruction: str
|
||||
doc_ids: Optional[List[str]] = None # 关联的文档 ID 列表
|
||||
context: Optional[Dict[str, Any]] = None # 额外上下文
|
||||
conversation_id: Optional[str] = None # 对话会话ID，用于关联历史记录
|
||||
|
||||
|
||||
class IntentRecognitionResponse(BaseModel):
|
||||
success: bool
|
||||
intent: str
|
||||
params: Dict[str, Any]
|
||||
message: str
|
||||
|
||||
|
||||
class InstructionExecutionResponse(BaseModel):
|
||||
success: bool
|
||||
intent: str
|
||||
result: Dict[str, Any]
|
||||
message: str
|
||||
|
||||
|
||||
# ==================== 接口 ====================
|
||||
|
||||
@router.post("/recognize", response_model=IntentRecognitionResponse)
|
||||
async def recognize_intent(request: InstructionRequest):
|
||||
"""
|
||||
意图识别接口
|
||||
|
||||
将自然语言指令解析为结构化的意图和参数
|
||||
|
||||
示例指令:
|
||||
- "提取文档中的医院数量和床位数"
|
||||
- "根据这些数据填表"
|
||||
- "总结一下这份文档"
|
||||
- "对比这两个文档的差异"
|
||||
"""
|
||||
try:
|
||||
intent, params = await intent_parser.parse(request.instruction)
|
||||
|
||||
# 添加文档关联信息
|
||||
if request.doc_ids:
|
||||
params["document_refs"] = [f"doc_{doc_id}" for doc_id in request.doc_ids]
|
||||
|
||||
intent_names = {
|
||||
"extract": "信息提取",
|
||||
"fill_table": "表格填写",
|
||||
"summarize": "摘要总结",
|
||||
"question": "智能问答",
|
||||
"search": "文档搜索",
|
||||
"compare": "对比分析",
|
||||
"transform": "格式转换",
|
||||
"edit": "文档编辑",
|
||||
"unknown": "未知"
|
||||
}
|
||||
|
||||
return IntentRecognitionResponse(
|
||||
success=True,
|
||||
intent=intent,
|
||||
params=params,
|
||||
message=f"识别到意图: {intent_names.get(intent, intent)}"
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"意图识别失败: {e}")
|
||||
return IntentRecognitionResponse(
|
||||
success=False,
|
||||
intent="error",
|
||||
params={},
|
||||
message=f"意图识别失败: {str(e)}"
|
||||
)
|
||||
|
||||
|
||||
@router.post("/execute")
|
||||
async def execute_instruction(
|
||||
background_tasks: BackgroundTasks,
|
||||
request: InstructionRequest,
|
||||
async_execute: bool = Query(False, description="是否异步执行（仅返回任务ID）")
|
||||
):
|
||||
"""
|
||||
指令执行接口
|
||||
|
||||
解析并执行自然语言指令
|
||||
|
||||
示例:
|
||||
- 指令: "提取文档1中的医院数量"
|
||||
返回: {"extracted_data": {"医院数量": ["38710个"]}}
|
||||
|
||||
- 指令: "填表"
|
||||
返回: {"filled_data": {...}}
|
||||
|
||||
设置 async_execute=true 可异步执行，返回任务ID用于查询进度
|
||||
"""
|
||||
task_id = str(uuid.uuid4())
|
||||
|
||||
if async_execute:
|
||||
# 异步模式：立即返回任务ID，后台执行
|
||||
background_tasks.add_task(
|
||||
_execute_instruction_task,
|
||||
task_id=task_id,
|
||||
instruction=request.instruction,
|
||||
doc_ids=request.doc_ids,
|
||||
context=request.context
|
||||
)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"task_id": task_id,
|
||||
"message": "指令已提交执行",
|
||||
"status_url": f"/api/v1/tasks/{task_id}"
|
||||
}
|
||||
|
||||
# 同步模式：等待执行完成
|
||||
return await _execute_instruction_task(task_id, request.instruction, request.doc_ids, request.context)
|
||||
|
||||
|
||||
async def _execute_instruction_task(
|
||||
task_id: str,
|
||||
instruction: str,
|
||||
doc_ids: Optional[List[str]],
|
||||
context: Optional[Dict[str, Any]]
|
||||
) -> InstructionExecutionResponse:
|
||||
"""执行指令的后台任务"""
|
||||
from app.core.database import redis_db, mongodb as mongo_client
|
||||
|
||||
try:
|
||||
# 记录任务
|
||||
try:
|
||||
await mongo_client.insert_task(
|
||||
task_id=task_id,
|
||||
task_type="instruction_execute",
|
||||
status="processing",
|
||||
message="正在执行指令"
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
# 构建执行上下文
|
||||
ctx: Dict[str, Any] = context or {}
|
||||
|
||||
# 如果提供了文档 ID，获取文档内容
|
||||
if doc_ids:
|
||||
docs = []
|
||||
for doc_id in doc_ids:
|
||||
doc = await mongo_client.get_document(doc_id)
|
||||
if doc:
|
||||
docs.append(doc)
|
||||
|
||||
if docs:
|
||||
ctx["source_docs"] = docs
|
||||
logger.info(f"指令执行上下文: 关联了 {len(docs)} 个文档")
|
||||
|
||||
# 执行指令
|
||||
result = await instruction_executor.execute(instruction, ctx)
|
||||
|
||||
# 更新任务状态
|
||||
try:
|
||||
await mongo_client.update_task(
|
||||
task_id=task_id,
|
||||
status="success",
|
||||
message="执行完成",
|
||||
result=result
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
return InstructionExecutionResponse(
|
||||
success=result.get("success", False),
|
||||
intent=result.get("intent", "unknown"),
|
||||
result=result,
|
||||
message=result.get("message", "执行完成")
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"指令执行失败: {e}")
|
||||
try:
|
||||
await mongo_client.update_task(
|
||||
task_id=task_id,
|
||||
status="failure",
|
||||
message="执行失败",
|
||||
error=str(e)
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
return InstructionExecutionResponse(
|
||||
success=False,
|
||||
intent="error",
|
||||
result={"error": str(e)},
|
||||
message=f"指令执行失败: {str(e)}"
|
||||
)
|
||||
|
||||
|
||||
@router.post("/chat")
|
||||
async def instruction_chat(
|
||||
background_tasks: BackgroundTasks,
|
||||
request: InstructionRequest,
|
||||
async_execute: bool = Query(False, description="是否异步执行（仅返回任务ID）")
|
||||
):
|
||||
"""
|
||||
指令对话接口
|
||||
|
||||
支持多轮对话的指令执行
|
||||
|
||||
示例对话流程:
|
||||
1. 用户: "上传一些文档"
|
||||
2. 系统: "请上传文档"
|
||||
3. 用户: "提取其中的医院数量"
|
||||
4. 系统: 返回提取结果
|
||||
|
||||
设置 async_execute=true 可异步执行，返回任务ID用于查询进度
|
||||
"""
|
||||
task_id = str(uuid.uuid4())
|
||||
|
||||
if async_execute:
|
||||
# 异步模式：立即返回任务ID，后台执行
|
||||
background_tasks.add_task(
|
||||
_execute_chat_task,
|
||||
task_id=task_id,
|
||||
instruction=request.instruction,
|
||||
doc_ids=request.doc_ids,
|
||||
context=request.context,
|
||||
conversation_id=request.conversation_id
|
||||
)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"task_id": task_id,
|
||||
"message": "指令已提交执行",
|
||||
"status_url": f"/api/v1/tasks/{task_id}"
|
||||
}
|
||||
|
||||
# 同步模式：等待执行完成
|
||||
return await _execute_chat_task(task_id, request.instruction, request.doc_ids, request.context, request.conversation_id)
|
||||
|
||||
|
||||
async def _execute_chat_task(
|
||||
task_id: str,
|
||||
instruction: str,
|
||||
doc_ids: Optional[List[str]],
|
||||
context: Optional[Dict[str, Any]],
|
||||
conversation_id: Optional[str] = None
|
||||
):
|
||||
"""执行指令对话的后台任务"""
|
||||
from app.core.database import mongodb as mongo_client
|
||||
|
||||
try:
|
||||
# 记录任务
|
||||
try:
|
||||
await mongo_client.insert_task(
|
||||
task_id=task_id,
|
||||
task_type="instruction_chat",
|
||||
status="processing",
|
||||
message="正在处理对话"
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
# 构建上下文
|
||||
ctx: Dict[str, Any] = context or {}
|
||||
|
||||
# 获取对话历史
|
||||
if conversation_id:
|
||||
history = await mongo_client.get_conversation_history(conversation_id, limit=20)
|
||||
if history:
|
||||
ctx["conversation_history"] = history
|
||||
logger.info(f"加载对话历史: conversation_id={conversation_id}, 消息数={len(history)}")
|
||||
|
||||
# 获取关联文档
|
||||
if doc_ids:
|
||||
docs = []
|
||||
for doc_id in doc_ids:
|
||||
doc = await mongo_client.get_document(doc_id)
|
||||
if doc:
|
||||
docs.append(doc)
|
||||
if docs:
|
||||
ctx["source_docs"] = docs
|
||||
|
||||
# 执行指令
|
||||
result = await instruction_executor.execute(instruction, ctx)
|
||||
|
||||
# 存储对话历史
|
||||
if conversation_id:
|
||||
try:
|
||||
# 存储用户消息
|
||||
await mongo_client.insert_conversation(
|
||||
conversation_id=conversation_id,
|
||||
role="user",
|
||||
content=instruction,
|
||||
intent=result.get("intent", "unknown")
|
||||
)
|
||||
# 存储助手回复
|
||||
response_content = result.get("message", "")
|
||||
if response_content:
|
||||
await mongo_client.insert_conversation(
|
||||
conversation_id=conversation_id,
|
||||
role="assistant",
|
||||
content=response_content,
|
||||
intent=result.get("intent", "unknown")
|
||||
)
|
||||
logger.info(f"已存储对话历史: conversation_id={conversation_id}")
|
||||
except Exception as e:
|
||||
logger.error(f"存储对话历史失败: {e}")
|
||||
|
||||
# 根据意图类型添加友好的响应消息
|
||||
response_messages = {
|
||||
"extract": f"已提取 {len(result.get('extracted_data', {}))} 个字段的数据",
|
||||
"fill_table": f"填表完成，填写了 {len(result.get('result', {}).get('filled_data', {}))} 个字段",
|
||||
"summarize": "已生成文档摘要",
|
||||
"question": "已找到相关答案",
|
||||
"search": f"找到 {len(result.get('results', []))} 条相关内容",
|
||||
"compare": f"对比了 {len(result.get('comparison', []))} 个文档",
|
||||
"edit": "编辑操作已完成",
|
||||
"transform": "格式转换已完成",
|
||||
"unknown": "无法理解该指令，请尝试更明确的描述"
|
||||
}
|
||||
|
||||
response = {
|
||||
"success": result.get("success", False),
|
||||
"intent": result.get("intent", "unknown"),
|
||||
"result": result,
|
||||
"message": response_messages.get(result.get("intent", ""), result.get("message", "")),
|
||||
"hint": _get_intent_hint(result.get("intent", ""))
|
||||
}
|
||||
|
||||
# 更新任务状态
|
||||
try:
|
||||
await mongo_client.update_task(
|
||||
task_id=task_id,
|
||||
status="success",
|
||||
message="处理完成",
|
||||
result=response
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
return response
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"指令对话失败: {e}")
|
||||
try:
|
||||
await mongo_client.update_task(
|
||||
task_id=task_id,
|
||||
status="failure",
|
||||
message="处理失败",
|
||||
error=str(e)
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"处理失败: {str(e)}"
|
||||
}
|
||||
|
||||
|
||||
def _get_intent_hint(intent: str) -> Optional[str]:
|
||||
"""根据意图返回下一步提示"""
|
||||
hints = {
|
||||
"extract": "您可以继续说 '提取更多字段' 或 '将数据填入表格'",
|
||||
"fill_table": "您可以提供表格模板或说 '帮我创建一个表格'",
|
||||
"question": "您可以继续提问或说 '总结一下这些内容'",
|
||||
"search": "您可以查看搜索结果或说 '对比这些内容'",
|
||||
"unknown": "您可以尝试: '提取数据'、'填表'、'总结'、'问答' 等指令"
|
||||
}
|
||||
return hints.get(intent)
|
||||
|
||||
|
||||
@router.get("/intents")
|
||||
async def list_supported_intents():
|
||||
"""
|
||||
获取支持的意图类型列表
|
||||
|
||||
返回所有可用的自然语言指令类型
|
||||
"""
|
||||
return {
|
||||
"intents": [
|
||||
{
|
||||
"intent": "extract",
|
||||
"name": "信息提取",
|
||||
"examples": [
|
||||
"提取文档中的医院数量",
|
||||
"抽取所有机构的名称",
|
||||
"找出表格中的数据"
|
||||
],
|
||||
"params": ["field_refs", "document_refs"]
|
||||
},
|
||||
{
|
||||
"intent": "fill_table",
|
||||
"name": "表格填写",
|
||||
"examples": [
|
||||
"填表",
|
||||
"根据这些数据填写表格",
|
||||
"帮我填到Excel里"
|
||||
],
|
||||
"params": ["template", "document_refs"]
|
||||
},
|
||||
{
|
||||
"intent": "summarize",
|
||||
"name": "摘要总结",
|
||||
"examples": [
|
||||
"总结一下这份文档",
|
||||
"生成摘要",
|
||||
"概括主要内容"
|
||||
],
|
||||
"params": ["document_refs"]
|
||||
},
|
||||
{
|
||||
"intent": "question",
|
||||
"name": "智能问答",
|
||||
"examples": [
|
||||
"这段话说的是什么？",
|
||||
"有多少家医院？",
|
||||
"解释一下这个概念"
|
||||
],
|
||||
"params": ["question", "focus"]
|
||||
},
|
||||
{
|
||||
"intent": "search",
|
||||
"name": "文档搜索",
|
||||
"examples": [
|
||||
"搜索相关内容",
|
||||
"找找看有哪些机构",
|
||||
"查询医院相关的数据"
|
||||
],
|
||||
"params": ["field_refs", "question"]
|
||||
},
|
||||
{
|
||||
"intent": "compare",
|
||||
"name": "对比分析",
|
||||
"examples": [
|
||||
"对比这两个文档",
|
||||
"比较一下差异",
|
||||
"找出不同点"
|
||||
],
|
||||
"params": ["document_refs"]
|
||||
},
|
||||
{
|
||||
"intent": "edit",
|
||||
"name": "文档编辑",
|
||||
"examples": [
|
||||
"润色这段文字",
|
||||
"修改格式",
|
||||
"添加注释"
|
||||
],
|
||||
"params": []
|
||||
}
|
||||
]
|
||||
}
|
||||
208
backend/app/api/endpoints/pdf_converter.py
Normal file
208
backend/app/api/endpoints/pdf_converter.py
Normal file
@@ -0,0 +1,208 @@
|
||||
"""
|
||||
PDF 转换 API 接口
|
||||
|
||||
提供将 Word、Excel、Txt、Markdown 转换为 PDF 的功能
|
||||
"""
|
||||
import logging
|
||||
import uuid
|
||||
from typing import Optional
|
||||
|
||||
from fastapi import APIRouter, UploadFile, File, Form, HTTPException
|
||||
from fastapi.responses import StreamingResponse
|
||||
|
||||
from app.services.pdf_converter_service import pdf_converter_service
|
||||
from app.services.file_service import file_service
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(prefix="/pdf", tags=["PDF转换"])
|
||||
|
||||
# 临时存储转换后的 PDF（key: download_id, value: (pdf_content, original_filename)）
|
||||
_pdf_cache: dict = {}
|
||||
|
||||
|
||||
# ==================== 请求/响应模型 ====================
|
||||
|
||||
class ConvertResponse:
|
||||
"""转换响应"""
|
||||
def __init__(self, success: bool, message: str = "", filename: str = ""):
|
||||
self.success = success
|
||||
self.message = message
|
||||
self.filename = filename
|
||||
|
||||
|
||||
# ==================== 接口 ====================
|
||||
|
||||
@router.post("/convert")
|
||||
async def convert_to_pdf(
|
||||
file: UploadFile = File(...),
|
||||
):
|
||||
"""
|
||||
将上传的文件转换为 PDF
|
||||
|
||||
支持格式: docx, xlsx, txt, md
|
||||
|
||||
Args:
|
||||
file: 上传的文件
|
||||
|
||||
Returns:
|
||||
PDF 文件流
|
||||
"""
|
||||
try:
|
||||
# 检查文件格式
|
||||
filename = file.filename or "document"
|
||||
file_ext = filename.rsplit('.', 1)[-1].lower() if '.' in filename else ''
|
||||
|
||||
if file_ext not in pdf_converter_service.supported_formats:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的格式: {file_ext}，支持的格式: {', '.join(pdf_converter_service.supported_formats)}"
|
||||
)
|
||||
|
||||
# 读取文件内容
|
||||
content = await file.read()
|
||||
if not content:
|
||||
raise HTTPException(status_code=400, detail="文件内容为空")
|
||||
|
||||
logger.info(f"开始转换文件: {filename} ({file_ext})")
|
||||
|
||||
# 转换为 PDF
|
||||
pdf_content, error = await pdf_converter_service.convert_to_pdf(
|
||||
file_content=content,
|
||||
source_format=file_ext,
|
||||
filename=filename.rsplit('.', 1)[0] if '.' in filename else filename
|
||||
)
|
||||
|
||||
if error:
|
||||
raise HTTPException(status_code=500, detail=error)
|
||||
|
||||
# 直接返回 PDF 文件流
|
||||
return StreamingResponse(
|
||||
iter([pdf_content]),
|
||||
media_type="application/pdf",
|
||||
headers={
|
||||
"Content-Disposition": f"attachment; filename*=UTF-8''converted.pdf"
|
||||
}
|
||||
)
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"PDF转换失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"转换失败: {str(e)}")
|
||||
|
||||
|
||||
@router.get("/download/{download_id}")
|
||||
async def download_pdf(download_id: str):
|
||||
"""
|
||||
通过下载 ID 下载 PDF（支持 IDM 拦截）
|
||||
"""
|
||||
if download_id not in _pdf_cache:
|
||||
raise HTTPException(status_code=404, detail="下载链接已过期或不存在")
|
||||
|
||||
pdf_content, filename = _pdf_cache.pop(download_id) # 下载后删除
|
||||
|
||||
# 使用 RFC 5987 编码支持中文文件名
|
||||
from starlette.responses import StreamingResponse
|
||||
import urllib.parse
|
||||
|
||||
# URL 编码中文文件名
|
||||
encoded_filename = urllib.parse.quote(f"{filename}.pdf")
|
||||
|
||||
return StreamingResponse(
|
||||
iter([pdf_content]),
|
||||
media_type="application/pdf",
|
||||
headers={
|
||||
"Content-Disposition": f"attachment; filename*=UTF-8''{encoded_filename}"
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
@router.get("/formats")
|
||||
async def get_supported_formats():
|
||||
"""
|
||||
获取支持的源文件格式
|
||||
|
||||
Returns:
|
||||
支持的格式列表
|
||||
"""
|
||||
return {
|
||||
"success": True,
|
||||
"formats": pdf_converter_service.get_supported_formats()
|
||||
}
|
||||
|
||||
|
||||
@router.post("/convert/batch")
|
||||
async def batch_convert_to_pdf(
|
||||
files: list[UploadFile] = File(...),
|
||||
):
|
||||
"""
|
||||
批量将多个文件转换为 PDF
|
||||
|
||||
注意: 批量转换会返回多个 PDF 文件打包的 zip
|
||||
|
||||
Args:
|
||||
files: 上传的文件列表
|
||||
|
||||
Returns:
|
||||
ZIP 压缩包（包含所有PDF）
|
||||
"""
|
||||
try:
|
||||
import io
|
||||
import zipfile
|
||||
|
||||
results = []
|
||||
errors = []
|
||||
|
||||
for file in files:
|
||||
try:
|
||||
filename = file.filename or "document"
|
||||
file_ext = filename.rsplit('.', 1)[-1].lower() if '.' in filename else ''
|
||||
|
||||
if file_ext not in pdf_converter_service.supported_formats:
|
||||
errors.append(f"{filename}: 不支持的格式")
|
||||
continue
|
||||
|
||||
content = await file.read()
|
||||
pdf_content, error = await pdf_converter_service.convert_to_pdf(
|
||||
file_content=content,
|
||||
source_format=file_ext,
|
||||
filename=filename.rsplit('.', 1)[0] if '.' in filename else filename
|
||||
)
|
||||
|
||||
if error:
|
||||
errors.append(f"{filename}: {error}")
|
||||
else:
|
||||
results.append((filename, pdf_content))
|
||||
|
||||
except Exception as e:
|
||||
errors.append(f"{file.filename}: {str(e)}")
|
||||
|
||||
if not results:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"没有可转换的文件。错误: {'; '.join(errors)}"
|
||||
)
|
||||
|
||||
# 创建 ZIP 包
|
||||
zip_buffer = io.BytesIO()
|
||||
with zipfile.ZipFile(zip_buffer, 'w', zipfile.ZIP_DEFLATED) as zip_file:
|
||||
for original_name, pdf_content in results:
|
||||
pdf_name = f"{original_name.rsplit('.', 1)[0] if '.' in original_name else original_name}.pdf"
|
||||
zip_file.writestr(pdf_name, pdf_content)
|
||||
|
||||
zip_buffer.seek(0)
|
||||
|
||||
return StreamingResponse(
|
||||
iter([zip_buffer.getvalue()]),
|
||||
media_type="application/zip",
|
||||
headers={
|
||||
"Content-Disposition": "attachment; filename*=UTF-8''converted_pdfs.zip"
|
||||
}
|
||||
)
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"批量PDF转换失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"批量转换失败: {str(e)}")
|
||||
@@ -87,13 +87,7 @@ class ExportRequest(BaseModel):
|
||||
template_id: str
|
||||
filled_data: dict
|
||||
format: str = "xlsx" # xlsx 或 docx
|
||||
|
||||
|
||||
class FillAndExportRequest(BaseModel):
|
||||
"""填充并导出请求 - 直接填充原始模板"""
|
||||
template_path: str # 模板文件路径
|
||||
filled_data: dict # 填写数据，格式: {字段名: [值1, 值2, ...]} 或 {字段名: 单个值}
|
||||
format: str = "xlsx" # xlsx 或 docx
|
||||
filled_file_path: Optional[str] = None # 已填写的 Word 文件路径（可选）
|
||||
|
||||
|
||||
# ==================== 接口实现 ====================
|
||||
@@ -548,7 +542,7 @@ async def export_filled_template(
|
||||
if request.format == "xlsx":
|
||||
return await _export_to_excel(request.filled_data, request.template_id)
|
||||
elif request.format == "docx":
|
||||
return await _export_to_word(request.filled_data, request.template_id)
|
||||
return await _export_to_word(request.filled_data, request.template_id, request.filled_file_path)
|
||||
else:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
@@ -615,53 +609,106 @@ async def _export_to_excel(filled_data: dict, template_id: str) -> StreamingResp
|
||||
)
|
||||
|
||||
|
||||
async def _export_to_word(filled_data: dict, template_id: str) -> StreamingResponse:
|
||||
async def _export_to_word(filled_data: dict, template_id: str, filled_file_path: Optional[str] = None) -> StreamingResponse:
|
||||
"""导出为 Word 格式"""
|
||||
import re
|
||||
import tempfile
|
||||
import os
|
||||
import urllib.parse
|
||||
from docx import Document
|
||||
from docx.shared import Pt, RGBColor
|
||||
from docx.enum.text import WD_ALIGN_PARAGRAPH
|
||||
|
||||
doc = Document()
|
||||
def clean_text(text: str) -> str:
|
||||
"""清理文本，移除可能导致Word问题的非法字符"""
|
||||
if not text:
|
||||
return ""
|
||||
# 移除控制字符
|
||||
text = re.sub(r'[\x00-\x08\x0b\x0c\x0e-\x1f\x7f]', '', text)
|
||||
# 转义 XML 特殊字符以防破坏文档结构
|
||||
text = text.replace('&', '&').replace('<', '<').replace('>', '>')
|
||||
return text.strip()
|
||||
|
||||
# 添加标题
|
||||
title = doc.add_heading('填写结果', level=1)
|
||||
title.alignment = WD_ALIGN_PARAGRAPH.CENTER
|
||||
tmp_path = None
|
||||
try:
|
||||
# 如果有已填写的文件（通过 _fill_docx 填写了模板单元格），直接返回该文件
|
||||
if filled_file_path and os.path.exists(filled_file_path):
|
||||
filename = os.path.basename(filled_file_path)
|
||||
with open(filled_file_path, 'rb') as f:
|
||||
file_content = f.read()
|
||||
output = io.BytesIO(file_content)
|
||||
encoded_filename = urllib.parse.quote(filename)
|
||||
return StreamingResponse(
|
||||
output,
|
||||
media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
|
||||
headers={
|
||||
"Content-Disposition": f"attachment; filename*=UTF-8''{encoded_filename}",
|
||||
"Content-Length": str(len(file_content))
|
||||
}
|
||||
)
|
||||
|
||||
# 添加填写时间和模板信息
|
||||
from datetime import datetime
|
||||
info_para = doc.add_paragraph()
|
||||
info_para.add_run(f"模板ID: {template_id}\n").bold = True
|
||||
info_para.add_run(f"导出时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
|
||||
# 没有已填写文件，创建新的 Word 文档（表格形式）
|
||||
# 创建临时文件（立即关闭句柄，避免 Windows 文件锁问题）
|
||||
tmp_fd, tmp_path = tempfile.mkstemp(suffix='.docx')
|
||||
os.close(tmp_fd) # 关闭立即得到的 fd，让 docx 可以写入
|
||||
|
||||
doc.add_paragraph() # 空行
|
||||
doc = Document()
|
||||
doc.add_heading('填写结果', level=1)
|
||||
|
||||
# 添加字段表格
|
||||
table = doc.add_table(rows=1, cols=3)
|
||||
table.style = 'Light Grid Accent 1'
|
||||
from datetime import datetime
|
||||
info_para = doc.add_paragraph()
|
||||
template_filename = template_id.split('/')[-1].split('\\')[-1] if template_id else '未知'
|
||||
info_para.add_run(f"模板文件: {clean_text(template_filename)}\n").bold = True
|
||||
info_para.add_run(f"导出时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
|
||||
doc.add_paragraph()
|
||||
|
||||
# 表头
|
||||
header_cells = table.rows[0].cells
|
||||
header_cells[0].text = '字段名'
|
||||
header_cells[1].text = '填写值'
|
||||
header_cells[2].text = '状态'
|
||||
table = doc.add_table(rows=1, cols=3)
|
||||
table.style = 'Table Grid'
|
||||
|
||||
for field_name, field_value in filled_data.items():
|
||||
row_cells = table.add_row().cells
|
||||
row_cells[0].text = field_name
|
||||
row_cells[1].text = str(field_value) if field_value else ''
|
||||
row_cells[2].text = '已填写' if field_value else '为空'
|
||||
header_cells = table.rows[0].cells
|
||||
header_cells[0].text = '字段名'
|
||||
header_cells[1].text = '填写值'
|
||||
header_cells[2].text = '状态'
|
||||
|
||||
# 保存到 BytesIO
|
||||
output = io.BytesIO()
|
||||
doc.save(output)
|
||||
output.seek(0)
|
||||
for field_name, field_value in filled_data.items():
|
||||
row_cells = table.add_row().cells
|
||||
row_cells[0].text = clean_text(str(field_name))
|
||||
|
||||
filename = f"filled_template.docx"
|
||||
if isinstance(field_value, list):
|
||||
clean_values = [clean_text(str(v)) for v in field_value if v]
|
||||
display_value = ', '.join(clean_values) if clean_values else ''
|
||||
else:
|
||||
display_value = clean_text(str(field_value)) if field_value else ''
|
||||
|
||||
row_cells[1].text = display_value
|
||||
row_cells[2].text = '已填写' if display_value else '为空'
|
||||
|
||||
# 保存到临时文件
|
||||
doc.save(tmp_path)
|
||||
|
||||
# 读取文件内容
|
||||
with open(tmp_path, 'rb') as f:
|
||||
file_content = f.read()
|
||||
|
||||
finally:
|
||||
# 清理临时文件
|
||||
if tmp_path and os.path.exists(tmp_path):
|
||||
try:
|
||||
os.unlink(tmp_path)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
output = io.BytesIO(file_content)
|
||||
filename = "filled_template.docx"
|
||||
encoded_filename = urllib.parse.quote(filename)
|
||||
|
||||
return StreamingResponse(
|
||||
io.BytesIO(output.getvalue()),
|
||||
output,
|
||||
media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
|
||||
headers={"Content-Disposition": f"attachment; filename={filename}"}
|
||||
headers={
|
||||
"Content-Disposition": f"attachment; filename*=UTF-8''{encoded_filename}",
|
||||
"Content-Length": str(len(file_content))
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
@@ -699,427 +746,3 @@ async def export_to_word(
|
||||
Word 文件流
|
||||
"""
|
||||
return await _export_to_word(filled_data, template_id)
|
||||
|
||||
|
||||
@router.post("/fill-and-export")
|
||||
async def fill_and_export_template(
|
||||
request: FillAndExportRequest,
|
||||
):
|
||||
"""
|
||||
填充原始模板并导出
|
||||
|
||||
直接打开原始模板文件，将数据填入模板的表格中，然后导出
|
||||
|
||||
Args:
|
||||
request: 填充并导出请求
|
||||
|
||||
Returns:
|
||||
填充后的模板文件流
|
||||
"""
|
||||
import os
|
||||
|
||||
logger.info(f"=== fill-and-export 请求 ===")
|
||||
logger.info(f"template_path: {request.template_path}")
|
||||
logger.info(f"format: {request.format}")
|
||||
logger.info(f"filled_data: {request.filled_data}")
|
||||
logger.info(f"filled_data 类型: {type(request.filled_data)}")
|
||||
logger.info(f"filled_data 键数量: {len(request.filled_data) if request.filled_data else 0}")
|
||||
logger.info(f"=========================")
|
||||
|
||||
template_path = request.template_path
|
||||
|
||||
# 检查模板文件是否存在
|
||||
if not os.path.exists(template_path):
|
||||
raise HTTPException(status_code=404, detail=f"模板文件不存在: {template_path}")
|
||||
|
||||
file_ext = os.path.splitext(template_path)[1].lower()
|
||||
|
||||
try:
|
||||
if file_ext in ['.xlsx', '.xls']:
|
||||
return await _fill_and_export_excel(template_path, request.filled_data)
|
||||
elif file_ext == '.docx':
|
||||
return await _fill_and_export_word(template_path, request.filled_data)
|
||||
else:
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的模板格式: {file_ext}，仅支持 xlsx/xls/docx"
|
||||
)
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"填充模板失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"填充模板失败: {str(e)}")
|
||||
|
||||
|
||||
async def _fill_and_export_word(template_path: str, filled_data: dict) -> StreamingResponse:
|
||||
"""
|
||||
填充原始 Word 模板
|
||||
|
||||
打开原始 Word 模板，找到表格，将数据填入对应单元格
|
||||
|
||||
Args:
|
||||
template_path: 模板文件路径
|
||||
filled_data: 填写数据 {字段名: [值1, 值2, ...]}
|
||||
|
||||
Returns:
|
||||
填充后的 Word 文件流
|
||||
"""
|
||||
from docx import Document
|
||||
|
||||
logger.info(f"填充 Word 模板: {template_path}")
|
||||
logger.info(f"填写数据字段: {list(filled_data.keys())}")
|
||||
|
||||
# 打开原始模板
|
||||
doc = Document(template_path)
|
||||
|
||||
# 找到第一个表格（比赛模板通常是第一个表格）
|
||||
if not doc.tables:
|
||||
logger.warning("Word 模板中没有表格，创建新表格")
|
||||
# 如果没有表格，创建一个
|
||||
table = doc.add_table(rows=len(filled_data) + 1, cols=2)
|
||||
|
||||
# 表头
|
||||
header_cells = table.rows[0].cells
|
||||
header_cells[0].text = '字段名'
|
||||
header_cells[1].text = '填写值'
|
||||
|
||||
# 数据行
|
||||
for idx, (field_name, values) in enumerate(filled_data.items()):
|
||||
row_cells = table.rows[idx + 1].cells
|
||||
row_cells[0].text = field_name
|
||||
if isinstance(values, list):
|
||||
row_cells[1].text = '; '.join(str(v) for v in values if v)
|
||||
else:
|
||||
row_cells[1].text = str(values) if values else ''
|
||||
else:
|
||||
# 填充第一个表格
|
||||
table = doc.tables[0]
|
||||
logger.info(f"找到表格，行数: {len(table.rows)}, 列数: {len(table.columns)}")
|
||||
|
||||
# 打印表格内容（调试用）
|
||||
logger.info("=== 表格内容预览 ===")
|
||||
for row_idx, row in enumerate(table.rows[:5]): # 只打印前5行
|
||||
row_texts = [cell.text.strip() for cell in row.cells]
|
||||
logger.info(f" 行{row_idx}: {row_texts}")
|
||||
logger.info("========================")
|
||||
|
||||
# 构建字段名到列索引的映射
|
||||
field_to_col = {}
|
||||
if table.rows:
|
||||
# 假设第一行是表头
|
||||
header_row = table.rows[0]
|
||||
for col_idx, cell in enumerate(header_row.cells):
|
||||
field_name = cell.text.strip()
|
||||
if field_name:
|
||||
field_to_col[field_name] = col_idx
|
||||
field_to_col[field_name.lower()] = col_idx # 忽略大小写
|
||||
|
||||
logger.info(f"表头字段映射: {field_to_col}")
|
||||
logger.info(f"待填充数据字段: {list(filled_data.keys())}")
|
||||
|
||||
# 填充数据
|
||||
filled_count = 0
|
||||
for field_name, values in filled_data.items():
|
||||
# 查找匹配的列
|
||||
col_idx = field_to_col.get(field_name)
|
||||
|
||||
if col_idx is None:
|
||||
# 尝试模糊匹配
|
||||
for c_idx in range(len(table.columns)):
|
||||
header_text = table.rows[0].cells[c_idx].text.strip().lower()
|
||||
if field_name.lower() in header_text or header_text in field_name.lower():
|
||||
col_idx = c_idx
|
||||
logger.info(f"模糊匹配成功: '{field_name}' -> 列 {col_idx}")
|
||||
break
|
||||
else:
|
||||
col_idx = None
|
||||
|
||||
if col_idx is not None and col_idx < len(table.columns):
|
||||
# 填充该列的所有数据行
|
||||
if isinstance(values, list):
|
||||
value_str = '; '.join(str(v) for v in values if v)
|
||||
else:
|
||||
value_str = str(values) if values else ''
|
||||
|
||||
# 填充每一行（从第二行开始，跳过表头）
|
||||
for row_idx in range(1, min(len(table.rows), len(values) + 1) if isinstance(values, list) else 2):
|
||||
try:
|
||||
cell = table.rows[row_idx].cells[col_idx]
|
||||
if isinstance(values, list) and row_idx - 1 < len(values):
|
||||
cell.text = str(values[row_idx - 1]) if values[row_idx - 1] else ''
|
||||
elif not isinstance(values, list):
|
||||
if row_idx == 1:
|
||||
cell.text = str(values) if values else ''
|
||||
except Exception as e:
|
||||
logger.warning(f"填充单元格失败 [{row_idx}][{col_idx}]: {e}")
|
||||
|
||||
filled_count += 1
|
||||
logger.info(f"✅ 字段 '{field_name}' -> 列 {col_idx}, 值: {value_str[:50]}")
|
||||
else:
|
||||
logger.warning(f"❌ 未找到字段 '{field_name}' 对应的列")
|
||||
|
||||
logger.info(f"填充完成: {filled_count}/{len(filled_data)} 个字段")
|
||||
|
||||
# 保存到 BytesIO
|
||||
output = io.BytesIO()
|
||||
doc.save(output)
|
||||
output.seek(0)
|
||||
|
||||
filename = f"filled_template.docx"
|
||||
|
||||
logger.info(f"Word 模板填充完成")
|
||||
|
||||
return StreamingResponse(
|
||||
io.BytesIO(output.getvalue()),
|
||||
media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
|
||||
headers={"Content-Disposition": f"attachment; filename={filename}"}
|
||||
)
|
||||
|
||||
|
||||
async def _fill_and_export_excel(template_path: str, filled_data: dict) -> StreamingResponse:
|
||||
"""
|
||||
填充原始 Excel 模板
|
||||
|
||||
打开原始 Excel 模板，找到对应列，将数据填入
|
||||
|
||||
Args:
|
||||
template_path: 模板文件路径
|
||||
filled_data: 填写数据 {字段名: [值1, 值2, ...]}
|
||||
|
||||
Returns:
|
||||
填充后的 Excel 文件流
|
||||
"""
|
||||
from openpyxl import load_workbook
|
||||
import os
|
||||
|
||||
logger.info(f"填充 Excel 模板: {template_path}")
|
||||
logger.info(f"填写数据: {list(filled_data.keys())}")
|
||||
|
||||
# 检查文件是否存在
|
||||
if not os.path.exists(template_path):
|
||||
raise HTTPException(status_code=404, detail=f"模板文件不存在: {template_path}")
|
||||
|
||||
# 打开原始模板
|
||||
wb = load_workbook(template_path)
|
||||
ws = wb.active # 获取当前工作表
|
||||
|
||||
logger.info(f"工作表: {ws.title}, 行数: {ws.max_row}, 列数: {ws.max_column}")
|
||||
|
||||
# 读取表头行（假设第一行是表头）
|
||||
header_row = 1
|
||||
field_to_col = {}
|
||||
|
||||
for col_idx in range(1, ws.max_column + 1):
|
||||
cell_value = ws.cell(row=header_row, column=col_idx).value
|
||||
if cell_value:
|
||||
field_name = str(cell_value).strip()
|
||||
field_to_col[field_name] = col_idx
|
||||
field_to_col[field_name.lower()] = col_idx # 忽略大小写
|
||||
|
||||
logger.info(f"表头字段映射: {field_to_col}")
|
||||
|
||||
# 计算最大行数
|
||||
max_rows = 1
|
||||
for values in filled_data.values():
|
||||
if isinstance(values, list):
|
||||
max_rows = max(max_rows, len(values))
|
||||
|
||||
# 填充数据
|
||||
for field_name, values in filled_data.items():
|
||||
# 查找匹配的列
|
||||
col_idx = field_to_col.get(field_name)
|
||||
|
||||
if col_idx is None:
|
||||
# 尝试模糊匹配
|
||||
for col_idx in range(1, ws.max_column + 1):
|
||||
header_text = str(ws.cell(row=header_row, column=col_idx).value or '').strip().lower()
|
||||
if field_name.lower() in header_text or header_text in field_name.lower():
|
||||
break
|
||||
else:
|
||||
col_idx = None
|
||||
|
||||
if col_idx is not None:
|
||||
# 填充数据（从第二行开始）
|
||||
if isinstance(values, list):
|
||||
for row_idx, value in enumerate(values, start=2):
|
||||
ws.cell(row=row_idx, column=col_idx, value=value if value else '')
|
||||
else:
|
||||
ws.cell(row=2, column=col_idx, value=values if values else '')
|
||||
|
||||
logger.info(f"字段 {field_name} -> 列 {col_idx}, 值数量: {len(values) if isinstance(values, list) else 1}")
|
||||
else:
|
||||
logger.warning(f"未找到字段 {field_name} 对应的列")
|
||||
|
||||
# 如果需要扩展行数
|
||||
current_max_row = ws.max_row
|
||||
if max_rows > current_max_row - 1: # -1 是因为表头占一行
|
||||
# 扩展样式（简单复制最后一行）
|
||||
for row_idx in range(current_max_row + 1, max_rows + 2):
|
||||
for col_idx in range(1, ws.max_column + 1):
|
||||
source_cell = ws.cell(row=current_max_row, column=col_idx)
|
||||
target_cell = ws.cell(row=row_idx, column=col_idx)
|
||||
# 复制值（如果有对应数据）
|
||||
if isinstance(filled_data.get(str(ws.cell(row=1, column=col_idx).value), []), list):
|
||||
data_idx = row_idx - 2
|
||||
data_list = filled_data.get(str(ws.cell(row=1, column=col_idx).value), [])
|
||||
if data_idx < len(data_list):
|
||||
target_cell.value = data_list[data_idx]
|
||||
|
||||
# 保存到 BytesIO
|
||||
output = io.BytesIO()
|
||||
wb.save(output)
|
||||
output.seek(0)
|
||||
|
||||
# 关闭工作簿
|
||||
wb.close()
|
||||
|
||||
filename = f"filled_template.xlsx"
|
||||
|
||||
logger.info(f"Excel 模板填充完成")
|
||||
|
||||
return StreamingResponse(
|
||||
io.BytesIO(output.getvalue()),
|
||||
media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
|
||||
headers={"Content-Disposition": f"attachment; filename={filename}"}
|
||||
)
|
||||
|
||||
|
||||
# ==================== Word 文档结构化字段提取接口 ====================
|
||||
|
||||
@router.post("/parse-word-structure")
|
||||
async def parse_word_structure(
|
||||
file: UploadFile = File(...),
|
||||
):
|
||||
"""
|
||||
上传 Word 文档，提取结构化字段并存入数据库
|
||||
|
||||
专门用于比赛场景：从 Word 表格模板中提取字段定义
|
||||
（字段名、提示词、字段类型等）并存入 MongoDB
|
||||
|
||||
Args:
|
||||
file: 上传的 Word 文件
|
||||
|
||||
Returns:
|
||||
提取的结构化字段信息
|
||||
"""
|
||||
if not file.filename:
|
||||
raise HTTPException(status_code=400, detail="文件名为空")
|
||||
|
||||
file_ext = file.filename.split('.')[-1].lower()
|
||||
if file_ext != 'docx':
|
||||
raise HTTPException(
|
||||
status_code=400,
|
||||
detail=f"不支持的文件类型: {file_ext}，仅支持 docx"
|
||||
)
|
||||
|
||||
try:
|
||||
# 1. 保存文件
|
||||
content = await file.read()
|
||||
saved_path = file_service.save_uploaded_file(
|
||||
content,
|
||||
file.filename,
|
||||
subfolder="word_templates"
|
||||
)
|
||||
logger.info(f"Word 文件已保存: {saved_path}")
|
||||
|
||||
# 2. 解析文档，提取结构化数据
|
||||
parser = ParserFactory.get_parser(saved_path)
|
||||
parse_result = parser.parse(saved_path)
|
||||
|
||||
if not parse_result.success:
|
||||
raise HTTPException(status_code=400, detail=parse_result.error)
|
||||
|
||||
# 3. 提取表格模板字段
|
||||
from app.core.document_parser.docx_parser import DocxParser
|
||||
docx_parser = DocxParser()
|
||||
template_fields = docx_parser.extract_template_fields_from_docx(saved_path)
|
||||
|
||||
logger.info(f"从 Word 文档提取到 {len(template_fields)} 个字段")
|
||||
|
||||
# 4. 提取完整的结构化信息
|
||||
template_info = docx_parser.parse_tables_for_template(saved_path)
|
||||
|
||||
# 5. 存储到 MongoDB
|
||||
doc_id = await mongodb.insert_document(
|
||||
doc_type="docx",
|
||||
content=parse_result.data.get("content", ""),
|
||||
metadata={
|
||||
**parse_result.metadata,
|
||||
"original_filename": file.filename,
|
||||
"file_path": saved_path,
|
||||
"template_fields": template_fields,
|
||||
"table_count": len(template_info.get("tables", [])),
|
||||
"field_count": len(template_fields)
|
||||
},
|
||||
structured_data={
|
||||
**parse_result.data.get("structured_data", {}),
|
||||
"template_fields": template_fields,
|
||||
"template_info": template_info
|
||||
}
|
||||
)
|
||||
|
||||
logger.info(f"Word 文档结构化信息已存入 MongoDB, doc_id: {doc_id}")
|
||||
|
||||
# 6. 返回结果
|
||||
return {
|
||||
"success": True,
|
||||
"doc_id": doc_id,
|
||||
"filename": file.filename,
|
||||
"file_path": saved_path,
|
||||
"field_count": len(template_fields),
|
||||
"fields": template_fields,
|
||||
"tables": template_info.get("tables", []),
|
||||
"metadata": {
|
||||
"paragraph_count": parse_result.metadata.get("paragraph_count", 0),
|
||||
"table_count": parse_result.metadata.get("table_count", 0),
|
||||
"word_count": parse_result.metadata.get("word_count", 0),
|
||||
"has_tables": parse_result.metadata.get("has_tables", False)
|
||||
}
|
||||
}
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"解析 Word 文档结构失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"解析失败: {str(e)}")
|
||||
|
||||
|
||||
@router.get("/word-fields/{doc_id}")
|
||||
async def get_word_template_fields(
|
||||
doc_id: str,
|
||||
):
|
||||
"""
|
||||
根据 doc_id 获取 Word 文档的模板字段信息
|
||||
|
||||
Args:
|
||||
doc_id: MongoDB 文档 ID
|
||||
|
||||
Returns:
|
||||
模板字段信息
|
||||
"""
|
||||
try:
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
|
||||
if not doc:
|
||||
raise HTTPException(status_code=404, detail=f"文档不存在: {doc_id}")
|
||||
|
||||
# 从 structured_data 中提取模板字段信息
|
||||
structured_data = doc.get("structured_data", {})
|
||||
template_fields = structured_data.get("template_fields", [])
|
||||
template_info = structured_data.get("template_info", {})
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"doc_id": doc_id,
|
||||
"filename": doc.get("metadata", {}).get("original_filename", ""),
|
||||
"fields": template_fields,
|
||||
"tables": template_info.get("tables", []),
|
||||
"field_count": len(template_fields),
|
||||
"metadata": doc.get("metadata", {})
|
||||
}
|
||||
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"获取 Word 模板字段失败: {str(e)}")
|
||||
raise HTTPException(status_code=500, detail=f"获取失败: {str(e)}")
|
||||
|
||||
27
backend/app/celery_app.py
Normal file
27
backend/app/celery_app.py
Normal file
@@ -0,0 +1,27 @@
|
||||
# ============================================================
|
||||
# Celery 应用配置
|
||||
# ============================================================
|
||||
from celery import Celery
|
||||
|
||||
# 优先使用环境变量，否则使用默认值
|
||||
import os
|
||||
|
||||
CELERY_BROKER_URL = os.getenv("CELERY_BROKER_URL", "redis://localhost:6379/1")
|
||||
CELERY_RESULT_BACKEND = os.getenv("CELERY_RESULT_BACKEND", "redis://localhost:6379/2")
|
||||
|
||||
celery_app = Celery(
|
||||
"filesread",
|
||||
broker=CELERY_BROKER_URL,
|
||||
backend=CELERY_RESULT_BACKEND,
|
||||
)
|
||||
|
||||
celery_app.conf.update(
|
||||
task_serializer="json",
|
||||
accept_content=["json"],
|
||||
result_serializer="json",
|
||||
timezone="Asia/Shanghai",
|
||||
enable_utc=True,
|
||||
task_track_started=True,
|
||||
task_time_limit=3600, # 1小时超时
|
||||
worker_prefetch_multiplier=1,
|
||||
)
|
||||
@@ -64,6 +64,11 @@ class MongoDB:
|
||||
"""任务集合 - 存储任务历史记录"""
|
||||
return self.db["tasks"]
|
||||
|
||||
@property
|
||||
def conversations(self):
|
||||
"""对话集合 - 存储对话历史记录"""
|
||||
return self.db["conversations"]
|
||||
|
||||
# ==================== 文档操作 ====================
|
||||
|
||||
async def insert_document(
|
||||
@@ -99,28 +104,6 @@ class MongoDB:
|
||||
logger.info(f"✓ 文档已存入MongoDB: [{doc_type}] {filename} | ID: {doc_id}")
|
||||
return doc_id
|
||||
|
||||
async def update_document(self, doc_id: str, updates: Dict[str, Any]) -> bool:
|
||||
"""
|
||||
更新文档
|
||||
|
||||
Args:
|
||||
doc_id: 文档ID
|
||||
updates: 要更新的字段字典
|
||||
|
||||
Returns:
|
||||
是否更新成功
|
||||
"""
|
||||
from bson import ObjectId
|
||||
try:
|
||||
result = await self.documents.update_one(
|
||||
{"_id": ObjectId(doc_id)},
|
||||
{"$set": updates}
|
||||
)
|
||||
return result.modified_count > 0
|
||||
except Exception as e:
|
||||
logger.error(f"更新文档失败 {doc_id}: {str(e)}")
|
||||
return False
|
||||
|
||||
async def get_document(self, doc_id: str) -> Optional[Dict[str, Any]]:
|
||||
"""根据ID获取文档"""
|
||||
from bson import ObjectId
|
||||
@@ -139,14 +122,20 @@ class MongoDB:
|
||||
搜索文档
|
||||
|
||||
Args:
|
||||
query: 搜索关键词
|
||||
query: 搜索关键词（支持文件名和内容搜索）
|
||||
doc_type: 文档类型过滤
|
||||
limit: 返回数量
|
||||
|
||||
Returns:
|
||||
文档列表
|
||||
"""
|
||||
filter_query = {"content": {"$regex": query}}
|
||||
filter_query = {
|
||||
"$or": [
|
||||
{"content": {"$regex": query, "$options": "i"}},
|
||||
{"metadata.original_filename": {"$regex": query, "$options": "i"}},
|
||||
{"metadata.filename": {"$regex": query, "$options": "i"}},
|
||||
]
|
||||
}
|
||||
if doc_type:
|
||||
filter_query["doc_type"] = doc_type
|
||||
|
||||
@@ -163,6 +152,15 @@ class MongoDB:
|
||||
result = await self.documents.delete_one({"_id": ObjectId(doc_id)})
|
||||
return result.deleted_count > 0
|
||||
|
||||
async def update_document_metadata(self, doc_id: str, metadata: Dict[str, Any]) -> bool:
|
||||
"""更新文档 metadata 字段"""
|
||||
from bson import ObjectId
|
||||
result = await self.documents.update_one(
|
||||
{"_id": ObjectId(doc_id)},
|
||||
{"$set": {"metadata": metadata}}
|
||||
)
|
||||
return result.modified_count > 0
|
||||
|
||||
# ==================== RAG 索引操作 ====================
|
||||
|
||||
async def insert_rag_entry(
|
||||
@@ -273,6 +271,10 @@ class MongoDB:
|
||||
await self.tasks.create_index("task_id", unique=True)
|
||||
await self.tasks.create_index("created_at")
|
||||
|
||||
# 对话集合索引
|
||||
await self.conversations.create_index("conversation_id")
|
||||
await self.conversations.create_index("created_at")
|
||||
|
||||
logger.info("MongoDB 索引创建完成")
|
||||
|
||||
# ==================== 任务历史操作 ====================
|
||||
@@ -391,6 +393,108 @@ class MongoDB:
|
||||
result = await self.tasks.delete_one({"task_id": task_id})
|
||||
return result.deleted_count > 0
|
||||
|
||||
# ==================== 对话历史操作 ====================
|
||||
|
||||
async def insert_conversation(
|
||||
self,
|
||||
conversation_id: str,
|
||||
role: str,
|
||||
content: str,
|
||||
intent: Optional[str] = None,
|
||||
metadata: Optional[Dict[str, Any]] = None,
|
||||
) -> str:
|
||||
"""
|
||||
插入对话记录
|
||||
|
||||
Args:
|
||||
conversation_id: 对话会话ID
|
||||
role: 角色 (user/assistant)
|
||||
content: 对话内容
|
||||
intent: 意图类型
|
||||
metadata: 额外元数据
|
||||
|
||||
Returns:
|
||||
插入文档的ID
|
||||
"""
|
||||
message = {
|
||||
"conversation_id": conversation_id,
|
||||
"role": role,
|
||||
"content": content,
|
||||
"intent": intent,
|
||||
"metadata": metadata or {},
|
||||
"created_at": datetime.utcnow(),
|
||||
}
|
||||
result = await self.conversations.insert_one(message)
|
||||
return str(result.inserted_id)
|
||||
|
||||
async def get_conversation_history(
|
||||
self,
|
||||
conversation_id: str,
|
||||
limit: int = 20,
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
获取对话历史
|
||||
|
||||
Args:
|
||||
conversation_id: 对话会话ID
|
||||
limit: 返回消息数量
|
||||
|
||||
Returns:
|
||||
对话消息列表
|
||||
"""
|
||||
cursor = self.conversations.find(
|
||||
{"conversation_id": conversation_id}
|
||||
).sort("created_at", 1).limit(limit)
|
||||
|
||||
messages = []
|
||||
async for msg in cursor:
|
||||
msg["_id"] = str(msg["_id"])
|
||||
if msg.get("created_at"):
|
||||
msg["created_at"] = msg["created_at"].isoformat()
|
||||
messages.append(msg)
|
||||
return messages
|
||||
|
||||
async def delete_conversation(self, conversation_id: str) -> bool:
|
||||
"""删除对话会话"""
|
||||
result = await self.conversations.delete_many({"conversation_id": conversation_id})
|
||||
return result.deleted_count > 0
|
||||
|
||||
async def list_conversations(
|
||||
self,
|
||||
limit: int = 50,
|
||||
skip: int = 0,
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
获取会话列表（按最近一条消息排序）
|
||||
|
||||
Args:
|
||||
limit: 返回数量
|
||||
skip: 跳过数量
|
||||
|
||||
Returns:
|
||||
会话列表
|
||||
"""
|
||||
# 使用 aggregation 获取每个会话的最新一条消息
|
||||
pipeline = [
|
||||
{"$sort": {"created_at": -1}},
|
||||
{"$group": {
|
||||
"_id": "$conversation_id",
|
||||
"last_message": {"$first": "$$ROOT"},
|
||||
}},
|
||||
{"$replaceRoot": {"newRoot": "$last_message"}},
|
||||
{"$sort": {"created_at": -1}},
|
||||
{"$skip": skip},
|
||||
{"$limit": limit},
|
||||
]
|
||||
|
||||
conversations = []
|
||||
async for doc in self.conversations.aggregate(pipeline):
|
||||
doc["_id"] = str(doc["_id"])
|
||||
if doc.get("created_at"):
|
||||
doc["created_at"] = doc["created_at"].isoformat()
|
||||
conversations.append(doc)
|
||||
return conversations
|
||||
|
||||
|
||||
# ==================== 全局单例 ====================
|
||||
|
||||
|
||||
@@ -44,6 +44,22 @@ class DocxParser(BaseParser):
|
||||
error=f"文件不存在: {file_path}"
|
||||
)
|
||||
|
||||
# 尝试使用 python-docx 解析，失败则使用备用方法
|
||||
try:
|
||||
return self._parse_with_docx(path)
|
||||
except Exception as e:
|
||||
logger.warning(f"python-docx 解析失败，使用备用方法: {e}")
|
||||
try:
|
||||
return self._parse_fallback(path)
|
||||
except Exception as fallback_error:
|
||||
logger.error(f"备用解析方法也失败: {fallback_error}")
|
||||
return ParseResult(
|
||||
success=False,
|
||||
error=f"解析 Word 文档失败: {str(e)}"
|
||||
)
|
||||
|
||||
def _parse_with_docx(self, path: Path) -> ParseResult:
|
||||
"""使用 python-docx 解析文档"""
|
||||
# 检查文件扩展名
|
||||
if path.suffix.lower() not in self.supported_extensions:
|
||||
return ParseResult(
|
||||
@@ -51,98 +67,181 @@ class DocxParser(BaseParser):
|
||||
error=f"不支持的文件类型: {path.suffix}"
|
||||
)
|
||||
|
||||
# 读取 Word 文档
|
||||
doc = Document(path)
|
||||
|
||||
# 提取文本内容
|
||||
paragraphs = []
|
||||
for para in doc.paragraphs:
|
||||
if para.text.strip():
|
||||
paragraphs.append({
|
||||
"text": para.text,
|
||||
"style": str(para.style.name) if para.style else "Normal"
|
||||
})
|
||||
|
||||
# 提取段落纯文本（用于 AI 解析）
|
||||
paragraphs_text = [p["text"] for p in paragraphs if p["text"].strip()]
|
||||
|
||||
# 提取表格内容
|
||||
tables_data = []
|
||||
for i, table in enumerate(doc.tables):
|
||||
table_rows = []
|
||||
for row in table.rows:
|
||||
row_data = [cell.text.strip() for cell in row.cells]
|
||||
table_rows.append(row_data)
|
||||
|
||||
if table_rows:
|
||||
# 第一行作为表头，其余行作为数据
|
||||
headers = table_rows[0] if table_rows else []
|
||||
data_rows = table_rows[1:] if len(table_rows) > 1 else []
|
||||
tables_data.append({
|
||||
"table_index": i,
|
||||
"headers": headers, # 添加 headers 字段
|
||||
"rows": data_rows, # 数据行（不含表头）
|
||||
"row_count": len(data_rows),
|
||||
"column_count": len(headers) if headers else 0
|
||||
})
|
||||
|
||||
# 提取图片/嵌入式对象信息
|
||||
images_info = self._extract_images_info(doc, path)
|
||||
|
||||
# 合并所有文本（包括图片描述）
|
||||
full_text_parts = []
|
||||
full_text_parts.append("【文档正文】")
|
||||
full_text_parts.extend(paragraphs_text)
|
||||
|
||||
if tables_data:
|
||||
full_text_parts.append("\n【文档表格】")
|
||||
for idx, table in enumerate(tables_data):
|
||||
full_text_parts.append(f"--- 表格 {idx + 1} ---")
|
||||
for row in table["rows"]:
|
||||
full_text_parts.append(" | ".join(str(cell) for cell in row))
|
||||
|
||||
if images_info.get("image_count", 0) > 0:
|
||||
full_text_parts.append(f"\n【文档图片】文档包含 {images_info['image_count']} 张图片/图表")
|
||||
|
||||
full_text = "\n".join(full_text_parts)
|
||||
|
||||
# 构建元数据
|
||||
metadata = {
|
||||
"filename": path.name,
|
||||
"extension": path.suffix.lower(),
|
||||
"paragraph_count": len(paragraphs),
|
||||
"table_count": len(tables_data),
|
||||
"image_count": images_info.get("image_count", 0)
|
||||
}
|
||||
|
||||
return ParseResult(
|
||||
success=True,
|
||||
data={
|
||||
"content": full_text,
|
||||
"paragraphs": paragraphs,
|
||||
"paragraphs_with_style": paragraphs,
|
||||
"tables": tables_data,
|
||||
"images": images_info
|
||||
},
|
||||
metadata=metadata
|
||||
)
|
||||
|
||||
def _parse_fallback(self, path: Path) -> ParseResult:
|
||||
"""备用解析方法：直接解析 docx 的 XML 结构"""
|
||||
import zipfile
|
||||
from xml.etree import ElementTree as ET
|
||||
|
||||
try:
|
||||
# 读取 Word 文档
|
||||
doc = Document(file_path)
|
||||
with zipfile.ZipFile(path, 'r') as zf:
|
||||
# 读取 document.xml
|
||||
if 'word/document.xml' not in zf.namelist():
|
||||
return ParseResult(success=False, error="无效的 docx 文件格式")
|
||||
|
||||
# 提取文本内容
|
||||
paragraphs = []
|
||||
for para in doc.paragraphs:
|
||||
if para.text.strip():
|
||||
paragraphs.append({
|
||||
"text": para.text,
|
||||
"style": str(para.style.name) if para.style else "Normal"
|
||||
xml_content = zf.read('word/document.xml')
|
||||
root = ET.fromstring(xml_content)
|
||||
|
||||
# 命名空间
|
||||
namespaces = {
|
||||
'w': 'http://schemas.openxmlformats.org/wordprocessingml/2006/main'
|
||||
}
|
||||
|
||||
paragraphs = []
|
||||
tables = []
|
||||
current_table = []
|
||||
|
||||
for elem in root.iter():
|
||||
if elem.tag.endswith('}p'): # 段落
|
||||
text_parts = []
|
||||
for t in elem.iter():
|
||||
if t.tag.endswith('}t') and t.text:
|
||||
text_parts.append(t.text)
|
||||
text = ''.join(text_parts).strip()
|
||||
if text:
|
||||
paragraphs.append({'text': text, 'style': 'Normal'})
|
||||
elif elem.tag.endswith('}tr'): # 表格行
|
||||
row_data = []
|
||||
for tc in elem.iter():
|
||||
if tc.tag.endswith('}tc'): # 单元格
|
||||
cell_text = []
|
||||
for t in tc.iter():
|
||||
if t.tag.endswith('}t') and t.text:
|
||||
cell_text.append(t.text)
|
||||
row_data.append(''.join(cell_text).strip())
|
||||
if row_data:
|
||||
current_table.append(row_data)
|
||||
else:
|
||||
# 表格结束，保存
|
||||
if current_table:
|
||||
tables.append({
|
||||
'table_index': len(tables),
|
||||
'rows': current_table,
|
||||
'row_count': len(current_table),
|
||||
'column_count': len(current_table[0]) if current_table else 0
|
||||
})
|
||||
current_table = []
|
||||
|
||||
# 保存最后一张表格
|
||||
if current_table:
|
||||
tables.append({
|
||||
'table_index': len(tables),
|
||||
'rows': current_table,
|
||||
'row_count': len(current_table),
|
||||
'column_count': len(current_table[0]) if current_table else 0
|
||||
})
|
||||
|
||||
# 提取段落纯文本（用于 AI 解析）
|
||||
paragraphs_text = [p["text"] for p in paragraphs if p["text"].strip()]
|
||||
# 构建文本
|
||||
paragraphs_text = [p["text"] for p in paragraphs]
|
||||
full_text_parts = ["【文档正文】"] + paragraphs_text
|
||||
|
||||
# 提取表格内容
|
||||
tables_data = []
|
||||
for i, table in enumerate(doc.tables):
|
||||
table_rows = []
|
||||
for row in table.rows:
|
||||
row_data = [cell.text.strip() for cell in row.cells]
|
||||
table_rows.append(row_data)
|
||||
if tables:
|
||||
full_text_parts.append("\n【文档表格】")
|
||||
for idx, table in enumerate(tables):
|
||||
full_text_parts.append(f"--- 表格 {idx + 1} ---")
|
||||
for row in table["rows"]:
|
||||
full_text_parts.append(" | ".join(str(cell) for cell in row))
|
||||
|
||||
if table_rows:
|
||||
tables_data.append({
|
||||
"table_index": i,
|
||||
"rows": table_rows,
|
||||
"row_count": len(table_rows),
|
||||
"column_count": len(table_rows[0]) if table_rows else 0
|
||||
})
|
||||
full_text = "\n".join(full_text_parts)
|
||||
|
||||
# 提取图片/嵌入式对象信息
|
||||
images_info = self._extract_images_info(doc, path)
|
||||
|
||||
# 合并所有文本（包括图片描述）
|
||||
full_text_parts = []
|
||||
full_text_parts.append("【文档正文】")
|
||||
full_text_parts.extend(paragraphs_text)
|
||||
|
||||
if tables_data:
|
||||
full_text_parts.append("\n【文档表格】")
|
||||
for idx, table in enumerate(tables_data):
|
||||
full_text_parts.append(f"--- 表格 {idx + 1} ---")
|
||||
for row in table["rows"]:
|
||||
full_text_parts.append(" | ".join(str(cell) for cell in row))
|
||||
|
||||
if images_info.get("image_count", 0) > 0:
|
||||
full_text_parts.append(f"\n【文档图片】文档包含 {images_info['image_count']} 张图片/图表")
|
||||
|
||||
full_text = "\n".join(full_text_parts)
|
||||
|
||||
# 构建元数据
|
||||
metadata = {
|
||||
"filename": path.name,
|
||||
"extension": path.suffix.lower(),
|
||||
"file_size": path.stat().st_size,
|
||||
"paragraph_count": len(paragraphs),
|
||||
"table_count": len(tables_data),
|
||||
"word_count": len(full_text),
|
||||
"char_count": len(full_text.replace("\n", "")),
|
||||
"has_tables": len(tables_data) > 0,
|
||||
"has_images": images_info.get("image_count", 0) > 0,
|
||||
"image_count": images_info.get("image_count", 0)
|
||||
}
|
||||
|
||||
# 返回结果
|
||||
return ParseResult(
|
||||
success=True,
|
||||
data={
|
||||
"content": full_text,
|
||||
"paragraphs": paragraphs_text,
|
||||
"paragraphs_with_style": paragraphs,
|
||||
"tables": tables_data,
|
||||
"images": images_info,
|
||||
"word_count": len(full_text),
|
||||
"structured_data": {
|
||||
return ParseResult(
|
||||
success=True,
|
||||
data={
|
||||
"content": full_text,
|
||||
"paragraphs": paragraphs,
|
||||
"paragraphs_text": paragraphs_text,
|
||||
"tables": tables_data,
|
||||
"images": images_info
|
||||
"paragraphs_with_style": paragraphs,
|
||||
"tables": tables,
|
||||
"images": {"image_count": 0, "descriptions": []}
|
||||
},
|
||||
metadata={
|
||||
"filename": path.name,
|
||||
"extension": path.suffix.lower(),
|
||||
"paragraph_count": len(paragraphs),
|
||||
"table_count": len(tables),
|
||||
"image_count": 0,
|
||||
"parse_method": "fallback_xml"
|
||||
}
|
||||
},
|
||||
metadata=metadata
|
||||
)
|
||||
)
|
||||
|
||||
except zipfile.BadZipFile:
|
||||
return ParseResult(success=False, error="无效的 ZIP/文档文件")
|
||||
except Exception as e:
|
||||
logger.error(f"解析 Word 文档失败: {str(e)}")
|
||||
return ParseResult(
|
||||
success=False,
|
||||
error=f"解析 Word 文档失败: {str(e)}"
|
||||
)
|
||||
return ParseResult(success=False, error=f"备用解析失败: {str(e)}")
|
||||
|
||||
def extract_images_as_base64(self, file_path: str) -> List[Dict[str, str]]:
|
||||
"""
|
||||
@@ -197,6 +296,83 @@ class DocxParser(BaseParser):
|
||||
logger.info(f"共提取 {len(images)} 张图片")
|
||||
return images
|
||||
|
||||
def extract_text_from_images(self, file_path: str, lang: str = 'chi_sim+eng') -> Dict[str, Any]:
|
||||
"""
|
||||
对 Word 文档中的图片进行 OCR 文字识别
|
||||
|
||||
Args:
|
||||
file_path: Word 文件路径
|
||||
lang: Tesseract 语言代码，默认简体中文+英文 (chi_sim+eng)
|
||||
|
||||
Returns:
|
||||
包含识别结果的字典
|
||||
"""
|
||||
import zipfile
|
||||
from io import BytesIO
|
||||
from PIL import Image
|
||||
|
||||
try:
|
||||
import pytesseract
|
||||
except ImportError:
|
||||
logger.warning("pytesseract 未安装，OCR 功能不可用")
|
||||
return {
|
||||
"success": False,
|
||||
"error": "pytesseract 未安装，请运行: pip install pytesseract",
|
||||
"image_count": 0,
|
||||
"extracted_text": []
|
||||
}
|
||||
|
||||
results = {
|
||||
"success": True,
|
||||
"image_count": 0,
|
||||
"extracted_text": [],
|
||||
"total_chars": 0
|
||||
}
|
||||
|
||||
try:
|
||||
with zipfile.ZipFile(file_path, 'r') as zf:
|
||||
# 查找 word/media 目录下的图片文件
|
||||
media_files = [f for f in zf.namelist() if f.startswith('word/media/')]
|
||||
|
||||
for idx, filename in enumerate(media_files):
|
||||
ext = filename.split('.')[-1].lower()
|
||||
if ext not in ['png', 'jpg', 'jpeg', 'gif', 'bmp']:
|
||||
continue
|
||||
|
||||
try:
|
||||
# 读取图片数据
|
||||
image_data = zf.read(filename)
|
||||
image = Image.open(BytesIO(image_data))
|
||||
|
||||
# 使用 Tesseract OCR 提取文字
|
||||
text = pytesseract.image_to_string(image, lang=lang)
|
||||
text = text.strip()
|
||||
|
||||
if text:
|
||||
results["extracted_text"].append({
|
||||
"image_index": idx,
|
||||
"filename": filename,
|
||||
"text": text,
|
||||
"char_count": len(text)
|
||||
})
|
||||
results["total_chars"] += len(text)
|
||||
|
||||
logger.info(f"图片 {filename} OCR 识别完成，提取 {len(text)} 字符")
|
||||
|
||||
except Exception as e:
|
||||
logger.warning(f"图片 {filename} OCR 识别失败: {str(e)}")
|
||||
|
||||
results["image_count"] = len(results["extracted_text"])
|
||||
|
||||
except zipfile.BadZipFile:
|
||||
results["success"] = False
|
||||
results["error"] = "无效的 Word 文档文件"
|
||||
except Exception as e:
|
||||
results["success"] = False
|
||||
results["error"] = f"OCR 处理失败: {str(e)}"
|
||||
|
||||
return results
|
||||
|
||||
def extract_key_sentences(self, text: str, max_sentences: int = 10) -> List[str]:
|
||||
"""
|
||||
从文本中提取关键句子
|
||||
|
||||
@@ -1,15 +1,14 @@
|
||||
"""
|
||||
指令执行模块
|
||||
|
||||
注意: 此模块为可选功能，当前尚未实现。
|
||||
如需启用，请实现 intent_parser.py 和 executor.py
|
||||
支持文档智能操作交互，包括意图解析和指令执行
|
||||
"""
|
||||
from .intent_parser import IntentParser, DefaultIntentParser
|
||||
from .executor import InstructionExecutor, DefaultInstructionExecutor
|
||||
from .intent_parser import IntentParser, intent_parser
|
||||
from .executor import InstructionExecutor, instruction_executor
|
||||
|
||||
__all__ = [
|
||||
"IntentParser",
|
||||
"DefaultIntentParser",
|
||||
"intent_parser",
|
||||
"InstructionExecutor",
|
||||
"DefaultInstructionExecutor",
|
||||
"instruction_executor",
|
||||
]
|
||||
|
||||
@@ -2,34 +2,804 @@
|
||||
指令执行器模块
|
||||
|
||||
将自然语言指令转换为可执行操作
|
||||
|
||||
注意: 此模块为可选功能，当前尚未实现。
|
||||
"""
|
||||
from abc import ABC, abstractmethod
|
||||
from typing import Any, Dict
|
||||
import logging
|
||||
import json
|
||||
import re
|
||||
from typing import Any, Dict, List, Optional
|
||||
|
||||
from app.services.template_fill_service import template_fill_service, TemplateField
|
||||
from app.services.rag_service import rag_service
|
||||
from app.services.markdown_ai_service import markdown_ai_service
|
||||
from app.core.database import mongodb
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class InstructionExecutor(ABC):
|
||||
"""指令执行器抽象基类"""
|
||||
def _extract_filenames_from_text(text: str) -> List[str]:
|
||||
"""
|
||||
从指令文本中提取文件名列表。
|
||||
|
||||
@abstractmethod
|
||||
async def execute(self, instruction: str, context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
智能处理用'和'/'与'/'、分隔的多个文件名（尤其是带年号的统计公报）。
|
||||
"""
|
||||
# 先去掉"对比这两个文档"等引导语，只保留文件名部分
|
||||
text = re.sub(r'^(?:对比|比较)这两个?文档[的差异]?[：:]?', '', text).strip()
|
||||
text = re.sub(r'两个文档.*$', '', text).strip()
|
||||
if not text:
|
||||
return []
|
||||
|
||||
# 直接查找所有带扩展名的文件名模式
|
||||
results = []
|
||||
for m in re.finditer(r'[^\s，。！？、和与]+(?=\.(?:docx|xlsx|md|txt))', text):
|
||||
start = m.start()
|
||||
ext_match = re.search(r'\.(?:docx|xlsx|md|txt)', text[m.end():])
|
||||
if ext_match:
|
||||
fn = text[start:m.end() + ext_match.end()]
|
||||
if fn:
|
||||
results.append(fn)
|
||||
|
||||
return results
|
||||
|
||||
|
||||
class InstructionExecutor:
|
||||
"""指令执行器"""
|
||||
|
||||
def __init__(self):
|
||||
self.intent_parser = None # 将通过 set_intent_parser 设置
|
||||
|
||||
def set_intent_parser(self, intent_parser):
|
||||
"""设置意图解析器"""
|
||||
self.intent_parser = intent_parser
|
||||
|
||||
async def execute(self, instruction: str, context: Dict[str, Any] = None) -> Dict[str, Any]:
|
||||
"""
|
||||
执行指令
|
||||
|
||||
Args:
|
||||
instruction: 解析后的指令
|
||||
context: 执行上下文
|
||||
instruction: 自然语言指令
|
||||
context: 执行上下文（包含文档信息等）
|
||||
|
||||
Returns:
|
||||
执行结果
|
||||
"""
|
||||
pass
|
||||
if self.intent_parser is None:
|
||||
from app.instruction.intent_parser import intent_parser
|
||||
self.intent_parser = intent_parser
|
||||
|
||||
context = context or {}
|
||||
context["instruction"] = instruction # 保存原始指令以便后续使用
|
||||
|
||||
# 解析意图（传递对话历史上下文）
|
||||
intent, params = await self.intent_parser.parse(instruction, context)
|
||||
|
||||
# 根据意图类型执行相应操作
|
||||
if intent == "extract":
|
||||
return await self._execute_extract(params, context)
|
||||
elif intent == "fill_table":
|
||||
return await self._execute_fill_table(params, context)
|
||||
elif intent == "summarize":
|
||||
return await self._execute_summarize(params, context)
|
||||
elif intent == "question":
|
||||
return await self._execute_question(params, context)
|
||||
elif intent == "search":
|
||||
return await self._execute_search(params, context)
|
||||
elif intent == "compare":
|
||||
return await self._execute_compare(params, context)
|
||||
elif intent == "edit":
|
||||
return await self._execute_edit(params, context)
|
||||
elif intent == "transform":
|
||||
return await self._execute_transform(params, context)
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": f"未知意图类型: {intent}",
|
||||
"message": "无法理解该指令，请尝试更明确的描述"
|
||||
}
|
||||
|
||||
async def _execute_extract(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行信息提取"""
|
||||
try:
|
||||
# target_fields 来自意图解析，field_refs 来自引号/字段关键词匹配
|
||||
target_fields = params.get("target_fields", []) or params.get("field_refs", [])
|
||||
doc_ids = params.get("document_refs", [])
|
||||
instruction_text = context.get("instruction", "")
|
||||
|
||||
# 如果没有指定文档，尝试按文件名精确搜索
|
||||
if not doc_ids or "all_docs" in doc_ids:
|
||||
if instruction_text:
|
||||
import re
|
||||
# 提取引号内的内容或文件名
|
||||
filename_match = re.search(r'["""]([^"""]+)["""]', instruction_text)
|
||||
if filename_match:
|
||||
search_term = filename_match.group(1)
|
||||
else:
|
||||
match = re.search(r'([^\s]+\.(?:docx|xlsx|md|txt))', instruction_text)
|
||||
search_term = match.group(1) if match else None
|
||||
|
||||
if search_term:
|
||||
logger.info(f"提取时搜索文档: {search_term}")
|
||||
searched_docs = await mongodb.search_documents(search_term, limit=5)
|
||||
if searched_docs:
|
||||
# 优先选择文件名完全匹配的文档
|
||||
best_docs = [
|
||||
d for d in searched_docs
|
||||
if search_term.lower() in d.get("metadata", {}).get("original_filename", "").lower()
|
||||
]
|
||||
if not best_docs:
|
||||
best_docs = [searched_docs[0]]
|
||||
context["source_docs"] = best_docs
|
||||
doc_ids = [doc.get("_id", "") for doc in best_docs]
|
||||
logger.info(f"找到 {len(best_docs)} 个文档用于提取，最佳: {best_docs[0].get('metadata', {}).get('original_filename', '?')}")
|
||||
|
||||
if not target_fields:
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "extract",
|
||||
"error": "未指定要提取的字段",
|
||||
"message": "请明确说明要提取哪些字段，如：'提取医院数量和床位数'"
|
||||
}
|
||||
|
||||
# 如果指定了文档且还没有加载 source_docs，则验证并加载
|
||||
if doc_ids and "all_docs" not in doc_ids and not context.get("source_docs"):
|
||||
valid_docs = []
|
||||
for doc_ref in doc_ids:
|
||||
doc_id = doc_ref.replace("doc_", "")
|
||||
doc = await mongodb.get_document(doc_id)
|
||||
if doc:
|
||||
valid_docs.append(doc)
|
||||
if not valid_docs:
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "extract",
|
||||
"error": "指定的文档不存在",
|
||||
"message": "请检查文档编号是否正确"
|
||||
}
|
||||
context["source_docs"] = valid_docs
|
||||
|
||||
# 构建字段列表（使用 TemplateField dataclass）
|
||||
fields = [
|
||||
TemplateField(
|
||||
name=field_name,
|
||||
cell=f"A{i+1}",
|
||||
field_type="text",
|
||||
required=False
|
||||
)
|
||||
for i, field_name in enumerate(target_fields)
|
||||
]
|
||||
|
||||
# 调用填表服务
|
||||
result = await template_fill_service.fill_template(
|
||||
template_fields=fields,
|
||||
source_doc_ids=[doc.get("_id") for doc in context.get("source_docs", [])] if context.get("source_docs") else None,
|
||||
user_hint=f"请提取字段: {', '.join(target_fields)}"
|
||||
)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "extract",
|
||||
"extracted_data": result.get("filled_data", {}),
|
||||
"fields": target_fields,
|
||||
"message": f"成功提取 {len(result.get('filled_data', {}))} 个字段"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"提取执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"提取失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_fill_table(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行填表操作"""
|
||||
try:
|
||||
template_file = context.get("template_file")
|
||||
if not template_file:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "未提供表格模板",
|
||||
"message": "请先上传要填写的表格模板"
|
||||
}
|
||||
|
||||
# 获取源文档
|
||||
source_docs = context.get("source_docs", []) or []
|
||||
source_doc_ids = [doc.get("_id") for doc in source_docs if doc.get("_id")]
|
||||
|
||||
# 获取字段
|
||||
fields = context.get("template_fields", [])
|
||||
|
||||
# 调用填表服务
|
||||
result = await template_fill_service.fill_template(
|
||||
template_fields=fields,
|
||||
source_doc_ids=source_doc_ids if source_doc_ids else None,
|
||||
source_file_paths=context.get("source_file_paths"),
|
||||
user_hint=params.get("user_hint"),
|
||||
template_id=template_file if isinstance(template_file, str) else None,
|
||||
template_file_type=params.get("template", {}).get("type", "xlsx")
|
||||
)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "fill_table",
|
||||
"result": result,
|
||||
"message": f"填表完成，成功填写 {len(result.get('filled_data', {}))} 个字段"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"填表执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"填表失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_summarize(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行摘要总结 - 使用 LLM 生成真实摘要"""
|
||||
try:
|
||||
import re
|
||||
docs = context.get("source_docs", []) or []
|
||||
instruction_text = context.get("instruction", "")
|
||||
|
||||
# 从指令中提取文件名/关键词，优先搜索精确文档
|
||||
search_term = None
|
||||
if instruction_text:
|
||||
filename_match = re.search(r'["""]([^"""]+)["""]', instruction_text)
|
||||
if filename_match:
|
||||
search_term = filename_match.group(1)
|
||||
else:
|
||||
file_match = re.search(r'([^\s，。！？,]+\.(?:docx|xlsx|md|txt))', instruction_text)
|
||||
if file_match:
|
||||
search_term = file_match.group(1)
|
||||
|
||||
# 如果没有文档或有更精确的搜索词，尝试重新搜索
|
||||
if not docs or search_term:
|
||||
if search_term:
|
||||
logger.info(f"按关键词搜索文档: {search_term}")
|
||||
searched_docs = await mongodb.search_documents(search_term, limit=5)
|
||||
if searched_docs:
|
||||
# 优先使用文件名最匹配的文档
|
||||
docs = sorted(
|
||||
searched_docs,
|
||||
key=lambda d: 1 if search_term.lower() in d.get("metadata", {}).get("original_filename", "").lower() else 0,
|
||||
reverse=True
|
||||
)
|
||||
logger.info(f"找到 {len(docs)} 个文档，最佳匹配: {docs[0].get('metadata', {}).get('original_filename', '?')}")
|
||||
|
||||
if not docs:
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "summarize",
|
||||
"action_needed": "provide_document",
|
||||
"message": "我理解了，您想分析文档内容。",
|
||||
"suggestion": "请提供已上传文档的名称（可以是文件名或部分名称），或者上传您想要分析的文档。\n\n支持的格式：docx、xlsx、md、txt\n\n例如：'分析2021年民政事业发展统计公报' 或 '总结卫生健康数据'"
|
||||
}
|
||||
|
||||
# 对第一个（最佳匹配）文档生成 AI 摘要
|
||||
primary_doc = docs[0]
|
||||
content = primary_doc.get("content", "")
|
||||
filename = primary_doc.get("metadata", {}).get("original_filename", "未知文档")
|
||||
|
||||
if not content:
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "summarize",
|
||||
"error": "文档内容为空",
|
||||
"message": f"文档 {filename} 没有可供分析的文本内容"
|
||||
}
|
||||
|
||||
# 使用 LLM 生成摘要
|
||||
content_for_summary = content[:12000] # 最多取前 12000 字
|
||||
user_request = instruction_text or "请总结这份文档"
|
||||
|
||||
prompt = f"""请对以下文档进行全面、有条理的摘要分析。
|
||||
|
||||
文档名称：{filename}
|
||||
用户要求：{user_request}
|
||||
|
||||
文档内容：
|
||||
{content_for_summary}
|
||||
|
||||
请按以下格式输出摘要：
|
||||
1. **文档概述**：简述文档主题和背景（2-3句）
|
||||
2. **主要内容**：列出文档的核心数据和关键信息（用要点列出）
|
||||
3. **重要数据**：提取文档中的重要数字、统计数据
|
||||
4. **主要结论**：归纳文档的主要结论或趋势
|
||||
|
||||
要求：条理清晰，数据准确，不要遗漏关键信息。"""
|
||||
|
||||
from app.services.llm_service import llm_service
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个专业的文档分析助手，擅长提取关键信息并生成结构化摘要。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
response = await llm_service.chat(messages=messages, temperature=0.3, max_tokens=2000)
|
||||
ai_summary = llm_service.extract_message_content(response)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "summarize",
|
||||
"ai_summary": ai_summary,
|
||||
"filename": filename,
|
||||
"doc_id": primary_doc.get("_id", ""),
|
||||
"total_docs_found": len(docs),
|
||||
"message": f"已生成文档摘要"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"摘要执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "summarize",
|
||||
"error": str(e),
|
||||
"message": f"摘要生成失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_question(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行问答"""
|
||||
try:
|
||||
question = params.get("question", "")
|
||||
instruction_text = context.get("instruction", "")
|
||||
|
||||
if not question:
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "question",
|
||||
"error": "未提供问题",
|
||||
"message": "请输入要回答的问题"
|
||||
}
|
||||
|
||||
docs = context.get("source_docs", []) or []
|
||||
|
||||
# 如果没有文档，尝试从指令中提取文件名搜索
|
||||
if not docs:
|
||||
filename_match = re.search(r'["""]([^"""]+\.(?:docx|xlsx|md|txt))["""]', instruction_text)
|
||||
if not filename_match:
|
||||
filename_match = re.search(r'([^\s，。！？]+\.(?:docx|xlsx|md|txt))', instruction_text)
|
||||
if filename_match:
|
||||
found = await mongodb.search_documents(filename_match.group(1), limit=5)
|
||||
if found:
|
||||
docs = found
|
||||
|
||||
if not docs:
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "question",
|
||||
"question": question,
|
||||
"answer": None,
|
||||
"message": "请先上传文档，我才能回答您的问题"
|
||||
}
|
||||
|
||||
# 使用 RAG 检索相关文档
|
||||
rag_results = []
|
||||
for doc in docs:
|
||||
doc_id = doc.get("_id", "")
|
||||
if doc_id:
|
||||
results = rag_service.retrieve_by_doc_id(doc_id, top_k=3)
|
||||
rag_results.extend(results)
|
||||
|
||||
# 构建上下文
|
||||
context_text = "\n\n".join([
|
||||
r.get("content", "") for r in rag_results[:5]
|
||||
]) if rag_results else ""
|
||||
|
||||
# 如果没有 RAG 结果，使用文档内容
|
||||
if not context_text:
|
||||
context_text = "\n\n".join([
|
||||
doc.get("content", "")[:3000] for doc in docs[:3] if doc.get("content")
|
||||
])
|
||||
|
||||
if not context_text:
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "question",
|
||||
"question": question,
|
||||
"answer": None,
|
||||
"message": "文档内容为空，无法回答问题"
|
||||
}
|
||||
|
||||
# 使用 LLM 生成答案
|
||||
filename = docs[0].get("metadata", {}).get("original_filename", "文档")
|
||||
prompt = f"""基于以下文档内容，回答用户的问题。
|
||||
|
||||
文档名称：{filename}
|
||||
用户问题：{question}
|
||||
|
||||
文档内容：
|
||||
{context_text[:8000]}
|
||||
|
||||
请根据文档内容准确回答问题。如果文档中没有相关信息，请明确说明。"""
|
||||
|
||||
from app.services.llm_service import llm_service
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个专业的文档问答助手，根据提供的内容准确回答用户问题。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
response = await llm_service.chat(messages=messages, temperature=0.3, max_tokens=1500)
|
||||
answer = llm_service.extract_message_content(response)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "question",
|
||||
"question": question,
|
||||
"answer": answer,
|
||||
"filename": filename,
|
||||
"message": "已生成回答"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"问答执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"问答处理失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_search(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行搜索"""
|
||||
try:
|
||||
field_refs = params.get("field_refs", [])
|
||||
query = " ".join(field_refs) if field_refs else params.get("question", "")
|
||||
|
||||
if not query:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "未提供搜索关键词",
|
||||
"message": "请输入要搜索的关键词"
|
||||
}
|
||||
|
||||
# 使用 RAG 检索
|
||||
results = rag_service.retrieve(query, top_k=10, min_score=0.3)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "search",
|
||||
"query": query,
|
||||
"results": [
|
||||
{
|
||||
"content": r.get("content", "")[:200],
|
||||
"score": r.get("score", 0),
|
||||
"doc_id": r.get("doc_id", "")
|
||||
}
|
||||
for r in results[:10]
|
||||
],
|
||||
"message": f"找到 {len(results)} 条相关结果"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"搜索执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"搜索失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_compare(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行对比分析"""
|
||||
try:
|
||||
docs = context.get("source_docs", []) or []
|
||||
instruction_text = context.get("instruction", "")
|
||||
|
||||
# 优先从指令中提取具体的文件名
|
||||
filenames = _extract_filenames_from_text(instruction_text)
|
||||
|
||||
if filenames:
|
||||
# 只选择文件名匹配的那些文档
|
||||
matched_docs = []
|
||||
for doc in docs:
|
||||
fname = doc.get("metadata", {}).get("original_filename", "").lower()
|
||||
for fn in filenames:
|
||||
if fn.lower() in fname or fname in fn.lower():
|
||||
matched_docs.append(doc)
|
||||
break
|
||||
# 如果匹配到足够文档，用匹配的
|
||||
if len(matched_docs) >= 2:
|
||||
docs = matched_docs
|
||||
else:
|
||||
# 匹配不够，尝试按文件名搜索 MongoDB
|
||||
all_found = []
|
||||
for fn in filenames:
|
||||
found = await mongodb.search_documents(fn, limit=5)
|
||||
all_found.extend(found)
|
||||
seen = set()
|
||||
unique_docs = []
|
||||
for d in all_found:
|
||||
did = d.get("_id", "")
|
||||
if did and did not in seen:
|
||||
seen.add(did)
|
||||
unique_docs.append(d)
|
||||
if len(unique_docs) >= 2:
|
||||
docs = unique_docs
|
||||
elif len(unique_docs) == 1 and len(docs) >= 1:
|
||||
# 找到一个指定的 + 用一个通用的
|
||||
docs = unique_docs + docs[:1]
|
||||
elif docs and len(filenames) == 1:
|
||||
# 找到一个指定文件名但只有一个匹配，尝试补充
|
||||
docs = unique_docs + [d for d in docs if d not in unique_docs]
|
||||
docs = docs[:2]
|
||||
|
||||
if len(docs) < 2:
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "compare",
|
||||
"error": "对比需要至少2个文档",
|
||||
"message": "请上传至少2个文档进行对比，或明确说出要对比的文档名称"
|
||||
}
|
||||
|
||||
# 提取文档基本信息
|
||||
comparison = []
|
||||
for i, doc in enumerate(docs[:5]):
|
||||
comparison.append({
|
||||
"index": i + 1,
|
||||
"filename": doc.get("metadata", {}).get("original_filename", "未知"),
|
||||
"doc_type": doc.get("doc_type", "未知"),
|
||||
"content_length": len(doc.get("content", "")),
|
||||
"has_tables": bool(doc.get("structured_data", {}).get("tables")),
|
||||
})
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "compare",
|
||||
"comparison": comparison,
|
||||
"message": f"对比了 {len(comparison)} 个文档的基本信息"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"对比执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "compare",
|
||||
"error": str(e),
|
||||
"message": f"对比分析失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_edit(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""执行文档编辑操作"""
|
||||
try:
|
||||
docs = context.get("source_docs", []) or []
|
||||
instruction_text = context.get("instruction", "")
|
||||
|
||||
# 如果没有文档，尝试从指令中提取文件名搜索
|
||||
if not docs:
|
||||
filename_match = re.search(r'["""]([^"""]+\.(?:docx|xlsx|md|txt))["""]', instruction_text)
|
||||
if not filename_match:
|
||||
filename_match = re.search(r'([^\s，。！？]+\.(?:docx|xlsx|md|txt))', instruction_text)
|
||||
if filename_match:
|
||||
found = await mongodb.search_documents(filename_match.group(1), limit=3)
|
||||
if found:
|
||||
docs = found
|
||||
|
||||
if not docs:
|
||||
return {
|
||||
"success": False,
|
||||
"intent": "edit",
|
||||
"error": "没有可用的文档",
|
||||
"message": "请先上传要编辑的文档"
|
||||
}
|
||||
|
||||
doc = docs[0] # 默认编辑第一个文档
|
||||
content = doc.get("content", "")
|
||||
original_filename = doc.get("metadata", {}).get("original_filename", "未知文档")
|
||||
|
||||
if not content:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档内容为空",
|
||||
"message": "该文档没有可编辑的内容"
|
||||
}
|
||||
|
||||
# 使用 LLM 进行文本润色/编辑
|
||||
prompt = f"""请对以下文档内容进行编辑处理。
|
||||
|
||||
原文内容：
|
||||
{content[:8000]}
|
||||
|
||||
编辑要求：
|
||||
- 润色表述，使其更加专业流畅
|
||||
- 修正明显的语法错误
|
||||
- 保持原意不变
|
||||
- 只返回编辑后的内容，不要解释
|
||||
|
||||
请直接输出编辑后的内容："""
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个专业的文本编辑助手。请直接输出编辑后的内容。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
from app.services.llm_service import llm_service
|
||||
response = await llm_service.chat(messages=messages, temperature=0.3, max_tokens=8000)
|
||||
edited_content = llm_service.extract_message_content(response)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "edit",
|
||||
"edited_content": edited_content,
|
||||
"original_filename": original_filename,
|
||||
"message": "文档编辑完成，内容已返回"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"编辑执行失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"编辑处理失败: {str(e)}"
|
||||
}
|
||||
|
||||
async def _execute_transform(self, params: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""
|
||||
执行格式转换操作
|
||||
|
||||
支持：
|
||||
- Word -> Excel
|
||||
- Excel -> Word
|
||||
- Markdown -> Word
|
||||
- Word -> Markdown
|
||||
"""
|
||||
try:
|
||||
docs = context.get("source_docs", []) or []
|
||||
if not docs:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "没有可用的文档",
|
||||
"message": "请先上传要转换的文档"
|
||||
}
|
||||
|
||||
# 获取目标格式
|
||||
template_info = params.get("template", {})
|
||||
target_type = template_info.get("type", "")
|
||||
|
||||
if not target_type:
|
||||
# 尝试从指令中推断
|
||||
instruction = params.get("instruction", "")
|
||||
if "excel" in instruction.lower() or "xlsx" in instruction.lower():
|
||||
target_type = "xlsx"
|
||||
elif "word" in instruction.lower() or "docx" in instruction.lower():
|
||||
target_type = "docx"
|
||||
elif "markdown" in instruction.lower() or "md" in instruction.lower():
|
||||
target_type = "md"
|
||||
|
||||
if not target_type:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "未指定目标格式",
|
||||
"message": "请说明要转换成什么格式（如：转成Excel、转成Word）"
|
||||
}
|
||||
|
||||
doc = docs[0]
|
||||
content = doc.get("content", "")
|
||||
structured_data = doc.get("structured_data", {})
|
||||
original_filename = doc.get("metadata", {}).get("original_filename", "未知文档")
|
||||
|
||||
# 构建转换内容
|
||||
if structured_data.get("tables"):
|
||||
# 有表格数据，生成表格格式的内容
|
||||
tables = structured_data.get("tables", [])
|
||||
table_content = []
|
||||
for i, table in enumerate(tables[:3]): # 最多处理3个表格
|
||||
headers = table.get("headers", [])
|
||||
rows = table.get("rows", [])[:20] # 最多20行
|
||||
if headers:
|
||||
table_content.append(f"【表格 {i+1}】")
|
||||
table_content.append(" | ".join(str(h) for h in headers))
|
||||
table_content.append(" | ".join(["---"] * len(headers)))
|
||||
for row in rows:
|
||||
if isinstance(row, list):
|
||||
table_content.append(" | ".join(str(c) for c in row))
|
||||
elif isinstance(row, dict):
|
||||
table_content.append(" | ".join(str(row.get(h, "")) for h in headers))
|
||||
table_content.append("")
|
||||
|
||||
if target_type == "xlsx":
|
||||
# 生成 Excel 格式的数据（JSON）
|
||||
excel_data = []
|
||||
for table in tables[:1]: # 只处理第一个表格
|
||||
headers = table.get("headers", [])
|
||||
rows = table.get("rows", [])[:100]
|
||||
for row in rows:
|
||||
if isinstance(row, list):
|
||||
excel_data.append(dict(zip(headers, row)))
|
||||
elif isinstance(row, dict):
|
||||
excel_data.append(row)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "transform",
|
||||
"transform_type": "to_excel",
|
||||
"target_format": "xlsx",
|
||||
"excel_data": excel_data,
|
||||
"headers": headers,
|
||||
"message": f"已转换为 Excel 格式，包含 {len(excel_data)} 行数据"
|
||||
}
|
||||
elif target_type in ["docx", "word"]:
|
||||
# 生成 Word 格式的文本
|
||||
word_content = f"# {original_filename}\n\n"
|
||||
word_content += "\n".join(table_content)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "transform",
|
||||
"transform_type": "to_word",
|
||||
"target_format": "docx",
|
||||
"content": word_content,
|
||||
"message": "已转换为 Word 格式"
|
||||
}
|
||||
elif target_type == "md":
|
||||
# 生成 Markdown 格式
|
||||
md_content = f"# {original_filename}\n\n"
|
||||
md_content += "\n".join(table_content)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "transform",
|
||||
"transform_type": "to_markdown",
|
||||
"target_format": "md",
|
||||
"content": md_content,
|
||||
"message": "已转换为 Markdown 格式"
|
||||
}
|
||||
|
||||
# 无表格数据，使用纯文本内容转换
|
||||
if target_type == "xlsx":
|
||||
# 将文本内容转为 Excel 格式（每行作为一列）
|
||||
lines = [line.strip() for line in content.split("\n") if line.strip()][:100]
|
||||
excel_data = [{"行号": i+1, "内容": line} for i, line in enumerate(lines)]
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "transform",
|
||||
"transform_type": "to_excel",
|
||||
"target_format": "xlsx",
|
||||
"excel_data": excel_data,
|
||||
"headers": ["行号", "内容"],
|
||||
"message": f"已将文本内容转换为 Excel，包含 {len(excel_data)} 行"
|
||||
}
|
||||
elif target_type in ["docx", "word"]:
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "transform",
|
||||
"transform_type": "to_word",
|
||||
"target_format": "docx",
|
||||
"content": content,
|
||||
"message": "文档内容已准备好，可下载为 Word 格式"
|
||||
}
|
||||
elif target_type == "md":
|
||||
# 简单的文本转 Markdown
|
||||
md_lines = []
|
||||
for line in content.split("\n"):
|
||||
line = line.strip()
|
||||
if line:
|
||||
# 简单处理：如果行不长且不是列表格式，作为段落
|
||||
if len(line) < 100 and not line.startswith(("-", "*", "1.", "2.", "3.")):
|
||||
md_lines.append(line)
|
||||
else:
|
||||
md_lines.append(line)
|
||||
else:
|
||||
md_lines.append("")
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"intent": "transform",
|
||||
"transform_type": "to_markdown",
|
||||
"target_format": "md",
|
||||
"content": "\n".join(md_lines),
|
||||
"message": "已转换为 Markdown 格式"
|
||||
}
|
||||
|
||||
return {
|
||||
"success": False,
|
||||
"error": "不支持的目标格式",
|
||||
"message": f"暂不支持转换为 {target_type} 格式"
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"格式转换失败: {e}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"message": f"格式转换失败: {str(e)}"
|
||||
}
|
||||
|
||||
|
||||
class DefaultInstructionExecutor(InstructionExecutor):
|
||||
"""默认指令执行器"""
|
||||
|
||||
async def execute(self, instruction: str, context: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""暂未实现"""
|
||||
raise NotImplementedError("指令执行功能暂未实现")
|
||||
# 全局单例
|
||||
instruction_executor = InstructionExecutor()
|
||||
|
||||
@@ -2,33 +2,293 @@
|
||||
意图解析器模块
|
||||
|
||||
解析用户自然语言指令，识别意图和参数
|
||||
|
||||
注意: 此模块为可选功能，当前尚未实现。
|
||||
"""
|
||||
from abc import ABC, abstractmethod
|
||||
from typing import Any, Dict, Tuple
|
||||
import re
|
||||
import logging
|
||||
from typing import Any, Dict, List, Optional, Tuple
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class IntentParser(ABC):
|
||||
"""意图解析器抽象基类"""
|
||||
class IntentParser:
|
||||
"""意图解析器"""
|
||||
|
||||
@abstractmethod
|
||||
async def parse(self, text: str) -> Tuple[str, Dict[str, Any]]:
|
||||
# 意图类型定义
|
||||
INTENT_EXTRACT = "extract" # 信息提取
|
||||
INTENT_FILL_TABLE = "fill_table" # 填表
|
||||
INTENT_SUMMARIZE = "summarize" # 摘要总结
|
||||
INTENT_QUESTION = "question" # 问答
|
||||
INTENT_SEARCH = "search" # 搜索
|
||||
INTENT_COMPARE = "compare" # 对比分析
|
||||
INTENT_TRANSFORM = "transform" # 格式转换
|
||||
INTENT_EDIT = "edit" # 编辑文档
|
||||
INTENT_UNKNOWN = "unknown" # 未知
|
||||
|
||||
# 意图关键词映射
|
||||
INTENT_KEYWORDS = {
|
||||
INTENT_EXTRACT: ["提取", "抽取", "获取", "找出", "查找", "识别", "找到"],
|
||||
INTENT_FILL_TABLE: ["填表", "填写", "填充", "录入", "导入到表格", "填写到"],
|
||||
INTENT_SUMMARIZE: ["总结", "摘要", "概括", "概述", "归纳", "提炼", "分析", "聊聊"],
|
||||
INTENT_QUESTION: ["问答", "回答", "解释", "什么是", "为什么", "如何", "怎样", "多少", "几个"],
|
||||
INTENT_SEARCH: ["搜索", "查找", "检索", "查询", "找"],
|
||||
INTENT_COMPARE: ["对比", "比较", "差异", "区别", "不同"],
|
||||
INTENT_TRANSFORM: ["转换", "转化", "变成", "转为", "导出"],
|
||||
INTENT_EDIT: ["修改", "编辑", "调整", "改写", "润色", "优化"],
|
||||
}
|
||||
|
||||
# 实体模式定义
|
||||
ENTITY_PATTERNS = {
|
||||
"number": [r"\d+", r"[一二三四五六七八九十百千万]+"],
|
||||
"date": [r"\d{4}年", r"\d{1,2}月", r"\d{1,2}日"],
|
||||
"percentage": [r"\d+(\.\d+)?%", r"\d+(\.\d+)?‰"],
|
||||
"currency": [r"\d+(\.\d+)?万元", r"\d+(\.\d+)?亿元", r"\d+(\.\d+)?元"],
|
||||
}
|
||||
|
||||
def __init__(self):
|
||||
self.intent_history: List[Dict[str, Any]] = []
|
||||
|
||||
async def parse(self, text: str, context: Dict[str, Any] = None) -> Tuple[str, Dict[str, Any]]:
|
||||
"""
|
||||
解析自然语言指令
|
||||
|
||||
Args:
|
||||
text: 用户输入的自然语言
|
||||
context: 执行上下文（包含对话历史等）
|
||||
|
||||
Returns:
|
||||
(意图类型, 参数字典)
|
||||
"""
|
||||
pass
|
||||
text = text.strip()
|
||||
if not text:
|
||||
return self.INTENT_UNKNOWN, {}
|
||||
|
||||
# 检查对话历史中的上下文
|
||||
conversation_history = []
|
||||
if context and context.get("conversation_history"):
|
||||
conversation_history = context.get("conversation_history", [])
|
||||
logger.info(f"解析时使用对话历史: {len(conversation_history)} 条消息")
|
||||
|
||||
# 记录历史
|
||||
self.intent_history.append({"text": text, "intent": None})
|
||||
|
||||
# 识别意图（考虑对话上下文）
|
||||
intent = self._recognize_intent_with_context(text, conversation_history)
|
||||
|
||||
# 提取参数
|
||||
params = self._extract_params(text, intent)
|
||||
|
||||
# 更新历史
|
||||
if self.intent_history:
|
||||
self.intent_history[-1]["intent"] = intent
|
||||
|
||||
logger.info(f"意图解析: text={text[:50]}..., intent={intent}, params={params}")
|
||||
|
||||
return intent, params
|
||||
|
||||
def _recognize_intent_with_context(self, text: str, conversation_history: List[Dict[str, Any]]) -> str:
|
||||
"""
|
||||
基于对话历史识别意图
|
||||
|
||||
Args:
|
||||
text: 当前用户输入
|
||||
conversation_history: 对话历史
|
||||
|
||||
Returns:
|
||||
意图类型
|
||||
"""
|
||||
# 如果对话历史为空，使用基础意图识别
|
||||
if not conversation_history:
|
||||
return self._recognize_intent(text)
|
||||
|
||||
# 基于历史上下文进行意图识别
|
||||
# 分析最近的对话了解用户意图的延续性
|
||||
last_intent = None
|
||||
last_topic = None
|
||||
|
||||
for msg in conversation_history[-5:]: # 最多看最近5条消息
|
||||
if msg.get("role") == "assistant":
|
||||
last_intent = msg.get("intent")
|
||||
if msg.get("intent") and msg.get("intent") != "unknown":
|
||||
last_topic = msg.get("intent")
|
||||
|
||||
# 如果当前消息很短（如"继续"、"是的"），可能延续之前的意图
|
||||
short_confirmation = ["是", "是的", "好", "继续", "ok", "好", "接着", "然后", "还有吗"]
|
||||
if text.strip() in short_confirmation or len(text.strip()) <= 3:
|
||||
if last_topic:
|
||||
logger.info(f"简短确认，延续之前的意图: {last_topic}")
|
||||
return last_topic
|
||||
|
||||
# 否则使用标准意图识别
|
||||
return self._recognize_intent(text)
|
||||
|
||||
def _recognize_intent(self, text: str) -> str:
|
||||
"""识别意图类型"""
|
||||
intent_scores: Dict[str, float] = {}
|
||||
|
||||
for intent, keywords in self.INTENT_KEYWORDS.items():
|
||||
score = 0
|
||||
for keyword in keywords:
|
||||
if keyword in text:
|
||||
score += 1
|
||||
if score > 0:
|
||||
intent_scores[intent] = score
|
||||
|
||||
if not intent_scores:
|
||||
return self.INTENT_UNKNOWN
|
||||
|
||||
# 返回得分最高的意图
|
||||
return max(intent_scores, key=intent_scores.get)
|
||||
|
||||
def _extract_params(self, text: str, intent: str) -> Dict[str, Any]:
|
||||
"""提取参数"""
|
||||
params: Dict[str, Any] = {
|
||||
"entities": self._extract_entities(text),
|
||||
"document_refs": self._extract_document_refs(text),
|
||||
"field_refs": self._extract_field_refs(text),
|
||||
"template_refs": self._extract_template_refs(text),
|
||||
}
|
||||
|
||||
# 根据意图类型提取特定参数
|
||||
if intent == self.INTENT_QUESTION:
|
||||
params["question"] = text
|
||||
params["focus"] = self._extract_question_focus(text)
|
||||
elif intent == self.INTENT_FILL_TABLE:
|
||||
params["template"] = self._extract_template_info(text)
|
||||
elif intent == self.INTENT_EXTRACT:
|
||||
params["target_fields"] = self._extract_target_fields(text)
|
||||
|
||||
return params
|
||||
|
||||
def _extract_entities(self, text: str) -> Dict[str, List[str]]:
|
||||
"""提取实体"""
|
||||
entities: Dict[str, List[str]] = {}
|
||||
|
||||
for entity_type, patterns in self.ENTITY_PATTERNS.items():
|
||||
matches = []
|
||||
for pattern in patterns:
|
||||
found = re.findall(pattern, text)
|
||||
matches.extend(found)
|
||||
if matches:
|
||||
entities[entity_type] = list(set(matches))
|
||||
|
||||
return entities
|
||||
|
||||
def _extract_document_refs(self, text: str) -> List[str]:
|
||||
"""提取文档引用"""
|
||||
# 匹配 "文档1"、"doc1"、"第一个文档" 等
|
||||
refs = []
|
||||
|
||||
# 数字索引: 文档1, doc1, 第1个文档
|
||||
num_patterns = [
|
||||
r"[文档doc]+(\d+)",
|
||||
r"第(\d+)个文档",
|
||||
r"第(\d+)份",
|
||||
]
|
||||
for pattern in num_patterns:
|
||||
matches = re.findall(pattern, text.lower())
|
||||
refs.extend([f"doc_{m}" for m in matches])
|
||||
|
||||
# "所有文档"、"全部文档"
|
||||
if any(kw in text for kw in ["所有", "全部", "整个"]):
|
||||
refs.append("all_docs")
|
||||
|
||||
return refs
|
||||
|
||||
def _extract_field_refs(self, text: str) -> List[str]:
|
||||
"""提取字段引用"""
|
||||
fields = []
|
||||
|
||||
# 匹配引号内的字段名
|
||||
quoted = re.findall(r"['\"『「]([^'\"』」]+)['\"』」]", text)
|
||||
fields.extend(quoted)
|
||||
|
||||
# 匹配 "xxx字段"、"xxx列" 等
|
||||
field_patterns = [
|
||||
r"([^\s]+)字段",
|
||||
r"([^\s]+)列",
|
||||
r"([^\s]+)数据",
|
||||
]
|
||||
for pattern in field_patterns:
|
||||
matches = re.findall(pattern, text)
|
||||
fields.extend(matches)
|
||||
|
||||
return list(set(fields))
|
||||
|
||||
def _extract_template_refs(self, text: str) -> List[str]:
|
||||
"""提取模板引用"""
|
||||
templates = []
|
||||
|
||||
# 匹配 "表格模板"、"Excel模板"、"表1" 等
|
||||
template_patterns = [
|
||||
r"([^\s]+模板)",
|
||||
r"表(\d+)",
|
||||
r"([^\s]+表格)",
|
||||
]
|
||||
for pattern in template_patterns:
|
||||
matches = re.findall(pattern, text)
|
||||
templates.extend(matches)
|
||||
|
||||
return list(set(templates))
|
||||
|
||||
def _extract_question_focus(self, text: str) -> Optional[str]:
|
||||
"""提取问题焦点"""
|
||||
# "什么是XXX"、"XXX是什么"
|
||||
match = re.search(r"[什么是]([^?]+)", text)
|
||||
if match:
|
||||
return match.group(1).strip()
|
||||
|
||||
# "XXX有多少"
|
||||
match = re.search(r"([^?]+)有多少", text)
|
||||
if match:
|
||||
return match.group(1).strip()
|
||||
|
||||
return None
|
||||
|
||||
def _extract_template_info(self, text: str) -> Optional[Dict[str, str]]:
|
||||
"""提取模板信息"""
|
||||
template_info: Dict[str, str] = {}
|
||||
|
||||
# 提取模板类型
|
||||
if "excel" in text.lower() or "xlsx" in text.lower() or "电子表格" in text:
|
||||
template_info["type"] = "xlsx"
|
||||
elif "word" in text.lower() or "docx" in text.lower() or "文档" in text:
|
||||
template_info["type"] = "docx"
|
||||
|
||||
return template_info if template_info else None
|
||||
|
||||
def _extract_target_fields(self, text: str) -> List[str]:
|
||||
"""提取目标字段 - 按分隔符切分再逐段清理"""
|
||||
fields = []
|
||||
|
||||
# 去除提取/抽取前缀
|
||||
cleaned_text = re.sub(r"^(?:提取|抽取)", "", text).strip()
|
||||
|
||||
# 按'和'、'与'、'、'分割成多段
|
||||
segments = re.split(r"[和与、]", cleaned_text)
|
||||
|
||||
# 常见前缀（这些不是字段名，需要去除）
|
||||
prefixes = ["文档中的", "文档中", "文件中的", "文件中", "内容中的", "内容中"]
|
||||
|
||||
for seg in segments:
|
||||
seg = seg.strip()
|
||||
# 去除常见前缀
|
||||
for p in prefixes:
|
||||
if seg.startswith(p):
|
||||
seg = seg[len(p):]
|
||||
break
|
||||
if seg and 2 <= len(seg) <= 20:
|
||||
fields.append(seg)
|
||||
|
||||
return list(set(fields))
|
||||
|
||||
def get_intent_history(self) -> List[Dict[str, Any]]:
|
||||
"""获取意图历史"""
|
||||
return self.intent_history
|
||||
|
||||
def clear_history(self):
|
||||
"""清空历史"""
|
||||
self.intent_history = []
|
||||
|
||||
|
||||
class DefaultIntentParser(IntentParser):
|
||||
"""默认意图解析器"""
|
||||
|
||||
async def parse(self, text: str) -> Tuple[str, Dict[str, Any]]:
|
||||
"""暂未实现"""
|
||||
raise NotImplementedError("意图解析功能暂未实现")
|
||||
# 全局单例
|
||||
intent_parser = IntentParser()
|
||||
|
||||
@@ -34,8 +34,8 @@ def setup_logging():
|
||||
# 根日志配置
|
||||
log_level = logging.DEBUG if settings.DEBUG else logging.INFO
|
||||
|
||||
# 日志目录
|
||||
log_dir = Path("data/logs")
|
||||
# 日志目录 (使用 settings.BASE_DIR 确保跨平台一致)
|
||||
log_dir = settings.BASE_DIR / "data" / "logs"
|
||||
log_dir.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
# 日志文件路径
|
||||
|
||||
@@ -223,6 +223,177 @@ class ExcelAIService:
|
||||
}
|
||||
}
|
||||
|
||||
async def analyze_excel_file_from_path(
|
||||
self,
|
||||
file_path: str,
|
||||
filename: str,
|
||||
user_prompt: str = "",
|
||||
analysis_type: str = "general",
|
||||
parse_options: Optional[Dict[str, Any]] = None
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
从文件路径分析 Excel 文件（用于从数据库加载的文档）
|
||||
|
||||
Args:
|
||||
file_path: Excel 文件路径
|
||||
filename: 文件名
|
||||
user_prompt: 用户自定义提示词
|
||||
analysis_type: 分析类型
|
||||
parse_options: 解析选项
|
||||
|
||||
Returns:
|
||||
Dict[str, Any]: 分析结果
|
||||
"""
|
||||
# 1. 解析 Excel 文件
|
||||
excel_data = None
|
||||
parse_result_metadata = None
|
||||
try:
|
||||
parse_options = parse_options or {}
|
||||
parse_result = self.parser.parse(file_path, **parse_options)
|
||||
|
||||
if not parse_result.success:
|
||||
return {
|
||||
"success": False,
|
||||
"error": parse_result.error,
|
||||
"analysis": None
|
||||
}
|
||||
|
||||
excel_data = parse_result.data
|
||||
parse_result_metadata = parse_result.metadata
|
||||
logger.info(f"Excel 解析成功: {parse_result_metadata}")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Excel 解析失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": f"Excel 解析失败: {str(e)}",
|
||||
"analysis": None
|
||||
}
|
||||
|
||||
# 2. 调用 LLM 进行分析
|
||||
try:
|
||||
if user_prompt and user_prompt.strip():
|
||||
llm_result = await self.llm_service.analyze_with_template(
|
||||
excel_data,
|
||||
user_prompt
|
||||
)
|
||||
else:
|
||||
llm_result = await self.llm_service.analyze_excel_data(
|
||||
excel_data,
|
||||
user_prompt,
|
||||
analysis_type
|
||||
)
|
||||
|
||||
logger.info(f"AI 分析完成: {llm_result['success']}")
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"excel": {
|
||||
"data": excel_data,
|
||||
"metadata": parse_result_metadata,
|
||||
"saved_path": file_path
|
||||
},
|
||||
"analysis": llm_result
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"AI 分析失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": f"AI 分析失败: {str(e)}",
|
||||
"excel": {
|
||||
"data": excel_data,
|
||||
"metadata": parse_result_metadata
|
||||
},
|
||||
"analysis": None
|
||||
}
|
||||
|
||||
async def batch_analyze_sheets_from_path(
|
||||
self,
|
||||
file_path: str,
|
||||
filename: str,
|
||||
user_prompt: str = "",
|
||||
analysis_type: str = "general"
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
从文件路径批量分析 Excel 文件的所有工作表（用于从数据库加载的文档）
|
||||
|
||||
Args:
|
||||
file_path: Excel 文件路径
|
||||
filename: 文件名
|
||||
user_prompt: 用户自定义提示词
|
||||
analysis_type: 分析类型
|
||||
|
||||
Returns:
|
||||
Dict[str, Any]: 分析结果
|
||||
"""
|
||||
# 1. 解析所有工作表
|
||||
try:
|
||||
parse_result = self.parser.parse_all_sheets(file_path)
|
||||
|
||||
if not parse_result.success:
|
||||
return {
|
||||
"success": False,
|
||||
"error": parse_result.error,
|
||||
"analysis": None
|
||||
}
|
||||
|
||||
sheets_data = parse_result.data.get("sheets", {})
|
||||
logger.info(f"Excel 解析成功，共 {len(sheets_data)} 个工作表")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Excel 解析失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": f"Excel 解析失败: {str(e)}",
|
||||
"analysis": None
|
||||
}
|
||||
|
||||
# 2. 批量分析每个工作表
|
||||
sheet_analyses = {}
|
||||
errors = {}
|
||||
|
||||
for sheet_name, sheet_data in sheets_data.items():
|
||||
try:
|
||||
if user_prompt and user_prompt.strip():
|
||||
llm_result = await self.llm_service.analyze_with_template(
|
||||
sheet_data,
|
||||
user_prompt
|
||||
)
|
||||
else:
|
||||
llm_result = await self.llm_service.analyze_excel_data(
|
||||
sheet_data,
|
||||
user_prompt,
|
||||
analysis_type
|
||||
)
|
||||
|
||||
sheet_analyses[sheet_name] = llm_result
|
||||
|
||||
if not llm_result["success"]:
|
||||
errors[sheet_name] = llm_result.get("error", "未知错误")
|
||||
|
||||
logger.info(f"工作表 '{sheet_name}' 分析完成")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"工作表 '{sheet_name}' 分析失败: {str(e)}")
|
||||
errors[sheet_name] = str(e)
|
||||
|
||||
# 3. 组合结果
|
||||
return {
|
||||
"success": len(errors) == 0,
|
||||
"excel": {
|
||||
"sheets": sheets_data,
|
||||
"metadata": parse_result.metadata,
|
||||
"saved_path": file_path
|
||||
},
|
||||
"analysis": {
|
||||
"sheets": sheet_analyses,
|
||||
"total_sheets": len(sheets_data),
|
||||
"successful": len(sheet_analyses) - len(errors),
|
||||
"errors": errors
|
||||
}
|
||||
}
|
||||
|
||||
def get_supported_analysis_types(self) -> List[str]:
|
||||
"""获取支持的分析类型"""
|
||||
return [
|
||||
|
||||
@@ -526,9 +526,10 @@ class ExcelStorageService:
|
||||
# 创建表
|
||||
model_class = self._create_table_model(table_name, columns, column_types)
|
||||
|
||||
# 创建表结构
|
||||
# 创建表结构 (使用异步方式)
|
||||
async with self.mysql_db.get_session() as session:
|
||||
model_class.__table__.create(session.bind, checkfirst=True)
|
||||
async with session.bind.begin() as conn:
|
||||
await conn.run_sync(lambda: model_class.__table__.create(checkfirst=True))
|
||||
|
||||
# 插入数据
|
||||
records = []
|
||||
|
||||
@@ -54,15 +54,21 @@ class LLMService:
|
||||
# 添加其他参数
|
||||
payload.update(kwargs)
|
||||
|
||||
import time
|
||||
_start_time = time.time()
|
||||
logger.info(f"🤖 [LLM] 正在调用 DeepSeek API... 模型: {self.model_name}")
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=60.0) as client:
|
||||
async with httpx.AsyncClient(timeout=120.0) as client:
|
||||
response = await client.post(
|
||||
f"{self.base_url}/chat/completions",
|
||||
headers=headers,
|
||||
json=payload
|
||||
)
|
||||
response.raise_for_status()
|
||||
return response.json()
|
||||
result = response.json()
|
||||
_elapsed = time.time() - _start_time
|
||||
logger.info(f"✅ [LLM] DeepSeek API 响应成功 | 模型: {self.model_name} | 耗时: {_elapsed:.2f}s | Token: {result.get('usage', {}).get('total_tokens', 'N/A')}")
|
||||
return result
|
||||
|
||||
except httpx.HTTPStatusError as e:
|
||||
error_detail = e.response.text
|
||||
@@ -78,7 +84,7 @@ class LLMService:
|
||||
pass
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"LLM API 调用异常: {str(e)}")
|
||||
logger.error(f"LLM API 调用异常: {repr(e)} - {str(e)}")
|
||||
raise
|
||||
|
||||
def extract_message_content(self, response: Dict[str, Any]) -> str:
|
||||
@@ -133,6 +139,9 @@ class LLMService:
|
||||
|
||||
payload.update(kwargs)
|
||||
|
||||
import time
|
||||
_start_time = time.time()
|
||||
logger.info(f"🤖 [LLM] 正在调用 DeepSeek API (流式) | 模型: {self.model_name}")
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=120.0) as client:
|
||||
async with client.stream(
|
||||
@@ -141,10 +150,13 @@ class LLMService:
|
||||
headers=headers,
|
||||
json=payload
|
||||
) as response:
|
||||
_elapsed = time.time() - _start_time
|
||||
logger.info(f"✅ [LLM] DeepSeek API 流式响应开始 | 模型: {self.model_name} | 耗时: {_elapsed:.2f}s")
|
||||
async for line in response.aiter_lines():
|
||||
if line.startswith("data: "):
|
||||
data = line[6:] # Remove "data: " prefix
|
||||
if data == "[DONE]":
|
||||
logger.info(f"✅ [LLM] DeepSeek API 流式响应完成")
|
||||
break
|
||||
try:
|
||||
import json as json_module
|
||||
|
||||
446
backend/app/services/multi_doc_reasoning_service.py
Normal file
446
backend/app/services/multi_doc_reasoning_service.py
Normal file
@@ -0,0 +1,446 @@
|
||||
"""
|
||||
多文档关联推理服务
|
||||
|
||||
跨文档信息关联和推理
|
||||
"""
|
||||
import logging
|
||||
import re
|
||||
from typing import Any, Dict, List, Optional, Set, Tuple
|
||||
from collections import defaultdict
|
||||
|
||||
from app.services.llm_service import llm_service
|
||||
from app.services.rag_service import rag_service
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class MultiDocReasoningService:
|
||||
"""
|
||||
多文档关联推理服务
|
||||
|
||||
功能：
|
||||
1. 实体跨文档追踪 - 追踪同一实体在不同文档中的描述
|
||||
2. 关系抽取与推理 - 抽取实体间关系并进行推理
|
||||
3. 信息补全 - 根据多个文档的信息互补填充缺失数据
|
||||
4. 冲突检测 - 检测不同文档间的信息冲突
|
||||
"""
|
||||
|
||||
def __init__(self):
|
||||
self.llm = llm_service
|
||||
|
||||
async def analyze_cross_documents(
|
||||
self,
|
||||
documents: List[Dict[str, Any]],
|
||||
query: Optional[str] = None,
|
||||
entity_types: Optional[List[str]] = None
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
跨文档分析
|
||||
|
||||
Args:
|
||||
documents: 文档列表
|
||||
query: 查询意图（可选）
|
||||
entity_types: 要追踪的实体类型列表，如 ["机构", "人物", "地点", "数量"]
|
||||
|
||||
Returns:
|
||||
跨文档分析结果
|
||||
"""
|
||||
if not documents:
|
||||
return {"success": False, "error": "没有可用的文档"}
|
||||
|
||||
entity_types = entity_types or ["机构", "数量", "时间", "地点"]
|
||||
|
||||
try:
|
||||
# 1. 提取各文档中的实体
|
||||
entities_per_doc = await self._extract_entities_from_docs(documents, entity_types)
|
||||
|
||||
# 2. 跨文档实体对齐
|
||||
aligned_entities = self._align_entities_across_docs(entities_per_doc)
|
||||
|
||||
# 3. 关系抽取
|
||||
relations = await self._extract_relations(documents)
|
||||
|
||||
# 4. 构建知识图谱
|
||||
knowledge_graph = self._build_knowledge_graph(aligned_entities, relations)
|
||||
|
||||
# 5. 信息补全
|
||||
completed_info = await self._complete_missing_info(knowledge_graph, documents)
|
||||
|
||||
# 6. 冲突检测
|
||||
conflicts = self._detect_conflicts(aligned_entities)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"entities": aligned_entities,
|
||||
"relations": relations,
|
||||
"knowledge_graph": knowledge_graph,
|
||||
"completed_info": completed_info,
|
||||
"conflicts": conflicts,
|
||||
"summary": self._generate_summary(aligned_entities, conflicts)
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"跨文档分析失败: {e}")
|
||||
return {"success": False, "error": str(e)}
|
||||
|
||||
async def _extract_entities_from_docs(
|
||||
self,
|
||||
documents: List[Dict[str, Any]],
|
||||
entity_types: List[str]
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""从各文档中提取实体"""
|
||||
entities_per_doc = []
|
||||
|
||||
for idx, doc in enumerate(documents):
|
||||
doc_id = doc.get("_id", f"doc_{idx}")
|
||||
content = doc.get("content", "")[:8000] # 限制长度
|
||||
|
||||
# 使用 LLM 提取实体
|
||||
prompt = f"""从以下文档中提取指定的实体类型信息。
|
||||
|
||||
实体类型: {', '.join(entity_types)}
|
||||
|
||||
文档内容:
|
||||
{content}
|
||||
|
||||
请按以下 JSON 格式输出（只需输出 JSON）：
|
||||
{{
|
||||
"entities": [
|
||||
{{"type": "机构", "name": "实体名称", "value": "相关数值（如有）", "context": "上下文描述"}},
|
||||
...
|
||||
]
|
||||
}}
|
||||
|
||||
只提取在文档中明确提到的实体，不要推测。"""
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个实体提取专家。请严格按JSON格式输出。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
try:
|
||||
response = await self.llm.chat(messages=messages, temperature=0.1, max_tokens=3000)
|
||||
content_response = self.llm.extract_message_content(response)
|
||||
|
||||
# 解析 JSON
|
||||
import json
|
||||
import re
|
||||
cleaned = content_response.strip()
|
||||
json_match = re.search(r'\{[\s\S]*\}', cleaned)
|
||||
if json_match:
|
||||
result = json.loads(json_match.group())
|
||||
entities = result.get("entities", [])
|
||||
entities_per_doc.append({
|
||||
"doc_id": doc_id,
|
||||
"doc_name": doc.get("metadata", {}).get("original_filename", f"文档{idx+1}"),
|
||||
"entities": entities
|
||||
})
|
||||
logger.info(f"文档 {doc_id} 提取到 {len(entities)} 个实体")
|
||||
except Exception as e:
|
||||
logger.warning(f"文档 {doc_id} 实体提取失败: {e}")
|
||||
|
||||
return entities_per_doc
|
||||
|
||||
def _align_entities_across_docs(
|
||||
self,
|
||||
entities_per_doc: List[Dict[str, Any]]
|
||||
) -> Dict[str, List[Dict[str, Any]]]:
|
||||
"""
|
||||
跨文档实体对齐
|
||||
|
||||
将同一实体在不同文档中的描述进行关联
|
||||
"""
|
||||
aligned: Dict[str, List[Dict[str, Any]]] = defaultdict(list)
|
||||
|
||||
for doc_data in entities_per_doc:
|
||||
doc_id = doc_data["doc_id"]
|
||||
doc_name = doc_data["doc_name"]
|
||||
|
||||
for entity in doc_data.get("entities", []):
|
||||
entity_name = entity.get("name", "")
|
||||
if not entity_name:
|
||||
continue
|
||||
|
||||
# 标准化实体名（去除空格和括号内容）
|
||||
normalized = self._normalize_entity_name(entity_name)
|
||||
|
||||
aligned[normalized].append({
|
||||
"original_name": entity_name,
|
||||
"type": entity.get("type", "未知"),
|
||||
"value": entity.get("value", ""),
|
||||
"context": entity.get("context", ""),
|
||||
"source_doc": doc_name,
|
||||
"source_doc_id": doc_id
|
||||
})
|
||||
|
||||
# 合并相同实体
|
||||
result = {}
|
||||
for normalized, appearances in aligned.items():
|
||||
if len(appearances) > 1:
|
||||
result[normalized] = appearances
|
||||
logger.info(f"实体对齐: {normalized} 在 {len(appearances)} 个文档中出现")
|
||||
|
||||
return result
|
||||
|
||||
def _normalize_entity_name(self, name: str) -> str:
|
||||
"""标准化实体名称"""
|
||||
# 去除空格
|
||||
name = name.strip()
|
||||
# 去除括号内容
|
||||
name = re.sub(r'[（(].*?[）)]', '', name)
|
||||
# 去除"第X名"等
|
||||
name = re.sub(r'^第\d+[名位个]', '', name)
|
||||
return name.strip()
|
||||
|
||||
async def _extract_relations(
|
||||
self,
|
||||
documents: List[Dict[str, Any]]
|
||||
) -> List[Dict[str, str]]:
|
||||
"""从文档中抽取关系"""
|
||||
relations = []
|
||||
|
||||
# 合并所有文档内容
|
||||
combined_content = "\n\n".join([
|
||||
f"【{doc.get('metadata', {}).get('original_filename', f'文档{i}')}】\n{doc.get('content', '')[:3000]}"
|
||||
for i, doc in enumerate(documents)
|
||||
])
|
||||
|
||||
prompt = f"""从以下文档内容中抽取实体之间的关系。
|
||||
|
||||
文档内容:
|
||||
{combined_content[:8000]}
|
||||
|
||||
请识别以下类型的关系：
|
||||
- 包含关系 (A包含B)
|
||||
- 隶属关系 (A隶属于B)
|
||||
- 合作关系 (A与B合作)
|
||||
- 对比关系 (A vs B)
|
||||
- 时序关系 (A先于B发生)
|
||||
|
||||
请按以下 JSON 格式输出（只需输出 JSON）：
|
||||
{{
|
||||
"relations": [
|
||||
{{"entity1": "实体1", "entity2": "实体2", "relation": "关系类型", "description": "关系描述"}},
|
||||
...
|
||||
]
|
||||
}}
|
||||
|
||||
如果没有找到明确的关系，返回空数组。"""
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个关系抽取专家。请严格按JSON格式输出。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
try:
|
||||
response = await self.llm.chat(messages=messages, temperature=0.1, max_tokens=3000)
|
||||
content_response = self.llm.extract_message_content(response)
|
||||
|
||||
import json
|
||||
import re
|
||||
cleaned = content_response.strip()
|
||||
json_match = re.search(r'\{{[\s\S]*\}}', cleaned)
|
||||
if json_match:
|
||||
result = json.loads(json_match.group())
|
||||
relations = result.get("relations", [])
|
||||
logger.info(f"抽取到 {len(relations)} 个关系")
|
||||
except Exception as e:
|
||||
logger.warning(f"关系抽取失败: {e}")
|
||||
|
||||
return relations
|
||||
|
||||
def _build_knowledge_graph(
|
||||
self,
|
||||
aligned_entities: Dict[str, List[Dict[str, Any]]],
|
||||
relations: List[Dict[str, str]]
|
||||
) -> Dict[str, Any]:
|
||||
"""构建知识图谱"""
|
||||
nodes = []
|
||||
edges = []
|
||||
node_ids = set()
|
||||
|
||||
# 添加实体节点
|
||||
for entity_name, appearances in aligned_entities.items():
|
||||
if len(appearances) < 1:
|
||||
continue
|
||||
|
||||
first_appearance = appearances[0]
|
||||
node_id = f"entity_{len(nodes)}"
|
||||
|
||||
# 收集该实体在所有文档中的值
|
||||
values = [a.get("value", "") for a in appearances if a.get("value")]
|
||||
primary_value = values[0] if values else ""
|
||||
|
||||
nodes.append({
|
||||
"id": node_id,
|
||||
"name": entity_name,
|
||||
"type": first_appearance.get("type", "未知"),
|
||||
"value": primary_value,
|
||||
"occurrence_count": len(appearances),
|
||||
"sources": [a.get("source_doc", "") for a in appearances]
|
||||
})
|
||||
node_ids.add(entity_name)
|
||||
|
||||
# 添加关系边
|
||||
for relation in relations:
|
||||
entity1 = self._normalize_entity_name(relation.get("entity1", ""))
|
||||
entity2 = self._normalize_entity_name(relation.get("entity2", ""))
|
||||
|
||||
if entity1 in node_ids and entity2 in node_ids:
|
||||
edges.append({
|
||||
"source": entity1,
|
||||
"target": entity2,
|
||||
"relation": relation.get("relation", "相关"),
|
||||
"description": relation.get("description", "")
|
||||
})
|
||||
|
||||
return {
|
||||
"nodes": nodes,
|
||||
"edges": edges,
|
||||
"stats": {
|
||||
"entity_count": len(nodes),
|
||||
"relation_count": len(edges)
|
||||
}
|
||||
}
|
||||
|
||||
async def _complete_missing_info(
|
||||
self,
|
||||
knowledge_graph: Dict[str, Any],
|
||||
documents: List[Dict[str, Any]]
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""根据多个文档补全信息"""
|
||||
completed = []
|
||||
|
||||
for node in knowledge_graph.get("nodes", []):
|
||||
if not node.get("value") and node.get("occurrence_count", 0) > 1:
|
||||
# 实体在多个文档中出现但没有数值，尝试从 RAG 检索补充
|
||||
query = f"{node['name']} 数值 数据"
|
||||
results = rag_service.retrieve(query, top_k=3, min_score=0.3)
|
||||
|
||||
if results:
|
||||
completed.append({
|
||||
"entity": node["name"],
|
||||
"type": node.get("type", "未知"),
|
||||
"source": "rag_inference",
|
||||
"context": results[0].get("content", "")[:200],
|
||||
"confidence": results[0].get("score", 0)
|
||||
})
|
||||
|
||||
return completed
|
||||
|
||||
def _detect_conflicts(
|
||||
self,
|
||||
aligned_entities: Dict[str, List[Dict[str, Any]]]
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""检测不同文档间的信息冲突"""
|
||||
conflicts = []
|
||||
|
||||
for entity_name, appearances in aligned_entities.items():
|
||||
if len(appearances) < 2:
|
||||
continue
|
||||
|
||||
# 检查数值冲突
|
||||
values = {}
|
||||
for appearance in appearances:
|
||||
val = appearance.get("value", "")
|
||||
if val:
|
||||
source = appearance.get("source_doc", "未知来源")
|
||||
values[source] = val
|
||||
|
||||
if len(values) > 1:
|
||||
unique_values = set(values.values())
|
||||
if len(unique_values) > 1:
|
||||
conflicts.append({
|
||||
"entity": entity_name,
|
||||
"type": "value_conflict",
|
||||
"details": values,
|
||||
"description": f"实体 '{entity_name}' 在不同文档中有不同数值: {values}"
|
||||
})
|
||||
|
||||
return conflicts
|
||||
|
||||
def _generate_summary(
|
||||
self,
|
||||
aligned_entities: Dict[str, List[Dict[str, Any]]],
|
||||
conflicts: List[Dict[str, Any]]
|
||||
) -> str:
|
||||
"""生成摘要"""
|
||||
summary_parts = []
|
||||
|
||||
total_entities = sum(len(appearances) for appearances in aligned_entities.values())
|
||||
multi_doc_entities = sum(1 for appearances in aligned_entities.values() if len(appearances) > 1)
|
||||
|
||||
summary_parts.append(f"跨文档分析完成：发现 {total_entities} 个实体")
|
||||
summary_parts.append(f"其中 {multi_doc_entities} 个实体在多个文档中被提及")
|
||||
|
||||
if conflicts:
|
||||
summary_parts.append(f"检测到 {len(conflicts)} 个潜在冲突")
|
||||
|
||||
return "; ".join(summary_parts)
|
||||
|
||||
async def answer_cross_doc_question(
|
||||
self,
|
||||
question: str,
|
||||
documents: List[Dict[str, Any]]
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
跨文档问答
|
||||
|
||||
Args:
|
||||
question: 问题
|
||||
documents: 文档列表
|
||||
|
||||
Returns:
|
||||
答案结果
|
||||
"""
|
||||
# 先进行跨文档分析
|
||||
analysis_result = await self.analyze_cross_documents(documents, query=question)
|
||||
|
||||
# 构建上下文
|
||||
context_parts = []
|
||||
|
||||
# 添加实体信息
|
||||
for entity_name, appearances in analysis_result.get("entities", {}).items():
|
||||
contexts = [f"{a.get('source_doc')}: {a.get('context', '')}" for a in appearances[:2]]
|
||||
if contexts:
|
||||
context_parts.append(f"【{entity_name}】{' | '.join(contexts)}")
|
||||
|
||||
# 添加关系信息
|
||||
for relation in analysis_result.get("relations", [])[:5]:
|
||||
context_parts.append(f"【关系】{relation.get('entity1')} {relation.get('relation')} {relation.get('entity2')}: {relation.get('description', '')}")
|
||||
|
||||
context_text = "\n\n".join(context_parts) if context_parts else "未找到相关实体和关系"
|
||||
|
||||
# 使用 LLM 生成答案
|
||||
prompt = f"""基于以下跨文档分析结果，回答用户问题。
|
||||
|
||||
问题: {question}
|
||||
|
||||
分析结果:
|
||||
{context_text}
|
||||
|
||||
请直接回答问题，如果分析结果中没有相关信息，请说明"根据提供的文档无法回答该问题"。"""
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个基于文档的问答助手。请根据提供的信息回答问题。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
try:
|
||||
response = await self.llm.chat(messages=messages, temperature=0.2, max_tokens=2000)
|
||||
answer = self.llm.extract_message_content(response)
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
"question": question,
|
||||
"answer": answer,
|
||||
"supporting_entities": list(analysis_result.get("entities", {}).keys())[:10],
|
||||
"relations_count": len(analysis_result.get("relations", []))
|
||||
}
|
||||
except Exception as e:
|
||||
logger.error(f"跨文档问答失败: {e}")
|
||||
return {"success": False, "error": str(e)}
|
||||
|
||||
|
||||
# 全局单例
|
||||
multi_doc_reasoning_service = MultiDocReasoningService()
|
||||
403
backend/app/services/pdf_converter_service.py
Normal file
403
backend/app/services/pdf_converter_service.py
Normal file
@@ -0,0 +1,403 @@
|
||||
"""
|
||||
PDF 转换服务
|
||||
|
||||
支持将 Word(docx)、Excel(xlsx)、Txt、Markdown(md) 格式转换为 PDF
|
||||
策略：所有格式先转为 Markdown，再通过 Markdown 转 PDF
|
||||
"""
|
||||
import io
|
||||
import logging
|
||||
import platform
|
||||
from pathlib import Path
|
||||
from typing import List, Tuple
|
||||
|
||||
from reportlab.lib.pagesizes import A4
|
||||
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
|
||||
from reportlab.lib.enums import TA_LEFT, TA_CENTER, TA_JUSTIFY
|
||||
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle
|
||||
from reportlab.pdfbase import pdfmetrics
|
||||
from reportlab.pdfbase.ttfonts import TTFont
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class PDFConverterService:
|
||||
"""PDF 转换服务"""
|
||||
|
||||
def __init__(self):
|
||||
self.supported_formats = ["docx", "xlsx", "txt", "md"]
|
||||
self._font_name = None
|
||||
self._styles = None
|
||||
self._page_width = None
|
||||
self._page_height = None
|
||||
self._setup_fonts()
|
||||
|
||||
def _setup_fonts(self):
|
||||
"""设置字体"""
|
||||
try:
|
||||
self._page_width, self._page_height = A4
|
||||
|
||||
# 查找中文字体
|
||||
font_path = self._find_chinese_font()
|
||||
if font_path:
|
||||
try:
|
||||
font = TTFont('ChineseFont', font_path)
|
||||
pdfmetrics.registerFont(font)
|
||||
from reportlab.pdfbase.pdfmetrics import registerFontFamily
|
||||
registerFontFamily('ChineseFont', normal='ChineseFont')
|
||||
self._font_name = 'ChineseFont'
|
||||
logger.info(f"成功注册中文字体: {font_path}")
|
||||
except Exception as e:
|
||||
logger.warning(f"字体注册失败: {e}, 使用Helvetica")
|
||||
self._font_name = 'Helvetica'
|
||||
else:
|
||||
self._font_name = 'Helvetica'
|
||||
logger.warning("未找到中文字体，使用 Helvetica（不支持中文）")
|
||||
|
||||
# 创建样式
|
||||
styles = getSampleStyleSheet()
|
||||
|
||||
styles.add(ParagraphStyle(
|
||||
name='ChineseTitle',
|
||||
fontName=self._font_name,
|
||||
fontSize=16,
|
||||
leading=22,
|
||||
alignment=TA_CENTER,
|
||||
spaceAfter=12,
|
||||
))
|
||||
|
||||
styles.add(ParagraphStyle(
|
||||
name='ChineseHeading',
|
||||
fontName=self._font_name,
|
||||
fontSize=14,
|
||||
leading=20,
|
||||
spaceBefore=10,
|
||||
spaceAfter=8,
|
||||
))
|
||||
|
||||
styles.add(ParagraphStyle(
|
||||
name='ChineseBody',
|
||||
fontName=self._font_name,
|
||||
fontSize=10,
|
||||
leading=14,
|
||||
alignment=TA_JUSTIFY,
|
||||
spaceAfter=6,
|
||||
))
|
||||
|
||||
styles.add(ParagraphStyle(
|
||||
name='ChineseCode',
|
||||
fontName='Courier',
|
||||
fontSize=9,
|
||||
leading=12,
|
||||
))
|
||||
|
||||
self._styles = styles
|
||||
logger.info("PDF服务初始化完成")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"PDF服务初始化失败: {e}")
|
||||
raise
|
||||
|
||||
def _find_chinese_font(self) -> str:
|
||||
"""查找中文字体"""
|
||||
system = platform.system()
|
||||
|
||||
if system == "Windows":
|
||||
fonts = [
|
||||
"C:/Windows/Fonts/simhei.ttf",
|
||||
"C:/Windows/Fonts/simsun.ttc",
|
||||
"C:/Windows/Fonts/msyh.ttc",
|
||||
"C:/Windows/Fonts/simsun.ttf",
|
||||
]
|
||||
elif system == "Darwin":
|
||||
fonts = [
|
||||
"/System/Library/Fonts/STHeiti Light.ttc",
|
||||
"/System/Library/Fonts/PingFang.ttc",
|
||||
"/Library/Fonts/Arial Unicode.ttf",
|
||||
]
|
||||
else:
|
||||
fonts = [
|
||||
"/usr/share/fonts/truetype/wqy/wqy-microhei.ttc",
|
||||
"/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc",
|
||||
]
|
||||
|
||||
for font in fonts:
|
||||
if Path(font).exists():
|
||||
return font
|
||||
return None
|
||||
|
||||
def _sanitize_text(self, text: str) -> str:
|
||||
"""清理文本"""
|
||||
if not text:
|
||||
return ""
|
||||
return text.replace('\x00', '')
|
||||
|
||||
async def convert_to_pdf(
|
||||
self,
|
||||
file_content: bytes,
|
||||
source_format: str,
|
||||
filename: str = "document"
|
||||
) -> Tuple[bytes, str]:
|
||||
"""将文档转换为 PDF"""
|
||||
try:
|
||||
if source_format.lower() not in self.supported_formats:
|
||||
return b"", f"不支持的格式: {source_format}"
|
||||
|
||||
# 第一步：转换为 Markdown
|
||||
markdown_content, error = await self._convert_to_markdown(file_content, source_format, filename)
|
||||
if error:
|
||||
return b"", error
|
||||
|
||||
# 第二步：Markdown 转 PDF
|
||||
return await self._convert_markdown_to_pdf(markdown_content, filename)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"PDF转换失败: {e}")
|
||||
import traceback
|
||||
logger.error(f"详细错误: {traceback.format_exc()}")
|
||||
return b"", f"转换失败: {str(e)}"
|
||||
|
||||
async def _convert_to_markdown(
|
||||
self,
|
||||
file_content: bytes,
|
||||
source_format: str,
|
||||
filename: str
|
||||
) -> Tuple[str, str]:
|
||||
"""将各种格式转换为 Markdown"""
|
||||
converters = {
|
||||
"docx": self._convert_docx_to_markdown,
|
||||
"xlsx": self._convert_xlsx_to_markdown,
|
||||
"txt": self._convert_txt_to_markdown,
|
||||
"md": self._convert_md_to_markdown,
|
||||
}
|
||||
return await converters[source_format.lower()](file_content, filename)
|
||||
|
||||
async def _convert_txt_to_markdown(self, file_content: bytes, filename: str) -> Tuple[str, str]:
|
||||
"""Txt 转 Markdown"""
|
||||
try:
|
||||
text = self._decode_content(file_content)
|
||||
text = self._sanitize_text(text)
|
||||
return f"# {filename}\n\n{text}", ""
|
||||
except Exception as e:
|
||||
logger.error(f"Txt转Markdown失败: {e}")
|
||||
return "", f"文本文件处理失败: {str(e)}"
|
||||
|
||||
async def _convert_md_to_markdown(self, file_content: bytes, filename: str) -> Tuple[str, str]:
|
||||
"""Markdown 原样返回"""
|
||||
try:
|
||||
content = self._decode_content(file_content)
|
||||
content = self._sanitize_text(content)
|
||||
return f"# {filename}\n\n{content}", ""
|
||||
except Exception as e:
|
||||
logger.error(f"Markdown处理失败: {e}")
|
||||
return "", f"Markdown处理失败: {str(e)}"
|
||||
|
||||
async def _convert_docx_to_markdown(self, file_content: bytes, filename: str) -> Tuple[str, str]:
|
||||
"""Word 转 Markdown - 使用 zipfile 直接解析，更加健壮"""
|
||||
try:
|
||||
import zipfile
|
||||
import re
|
||||
|
||||
lines = [f"# {filename}", ""]
|
||||
|
||||
# 直接使用 zipfile 解析 DOCX，避免 python-docx 的严格验证
|
||||
try:
|
||||
with zipfile.ZipFile(io.BytesIO(file_content), 'r') as zf:
|
||||
# 读取主文档内容
|
||||
xml_content = zf.read('word/document.xml').decode('utf-8')
|
||||
except zipfile.BadZipFile:
|
||||
return "", "文件不是有效的 DOCX 格式"
|
||||
except KeyError:
|
||||
return "", "DOCX 文件损坏：找不到 document.xml"
|
||||
|
||||
# 简单的 XML 解析 - 提取文本段落
|
||||
# 移除 XML 标签，提取纯文本
|
||||
xml_content = re.sub(r'<w:br[^>]*>', '\n', xml_content)
|
||||
xml_content = re.sub(r'</w:p>', '\n', xml_content)
|
||||
xml_content = re.sub(r'<[^>]+>', '', xml_content)
|
||||
xml_content = re.sub(r'\n\s*\n', '\n\n', xml_content)
|
||||
|
||||
# 解码 HTML 实体
|
||||
xml_content = xml_content.replace('&', '&')
|
||||
xml_content = xml_content.replace('<', '<')
|
||||
xml_content = xml_content.replace('>', '>')
|
||||
xml_content = xml_content.replace('"', '"')
|
||||
xml_content = xml_content.replace(''', "'")
|
||||
|
||||
# 清理空白
|
||||
lines_text = [line.strip() for line in xml_content.split('\n') if line.strip()]
|
||||
|
||||
# 生成 Markdown
|
||||
for text in lines_text[:500]: # 限制最多500行
|
||||
if text:
|
||||
lines.append(text)
|
||||
|
||||
return '\n'.join(lines), ""
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Word转Markdown失败: {e}")
|
||||
import traceback
|
||||
logger.error(traceback.format_exc())
|
||||
return "", f"Word文档处理失败: {str(e)}"
|
||||
for table in doc.tables:
|
||||
lines.append("")
|
||||
for row in table.rows:
|
||||
row_data = [cell.text.strip() for cell in row.cells]
|
||||
lines.append("| " + " | ".join(row_data) + " |")
|
||||
# 表头分隔符
|
||||
if table.rows:
|
||||
lines.append("| " + " | ".join(["---"] * len(table.rows[0].cells)) + " |")
|
||||
|
||||
return "\n".join(lines), ""
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Word转Markdown失败: {e}")
|
||||
return "", f"Word文档处理失败: {str(e)}"
|
||||
|
||||
async def _convert_xlsx_to_markdown(self, file_content: bytes, filename: str) -> Tuple[str, str]:
|
||||
"""Excel 转 Markdown"""
|
||||
try:
|
||||
import openpyxl
|
||||
|
||||
wb = openpyxl.load_workbook(io.BytesIO(file_content))
|
||||
lines = [f"# {filename} - Excel数据", ""]
|
||||
|
||||
for sheet_name in wb.sheetnames[:10]:
|
||||
ws = wb[sheet_name]
|
||||
lines.append(f"## 工作表: {sheet_name}")
|
||||
lines.append("")
|
||||
|
||||
for row_idx, row in enumerate(ws.iter_rows(max_row=50, values_only=True)):
|
||||
row_data = [str(cell) if cell is not None else "" for cell in row]
|
||||
if not any(row_data):
|
||||
continue
|
||||
lines.append("| " + " | ".join(row_data) + " |")
|
||||
if row_idx == 0:
|
||||
lines.append("| " + " | ".join(["---"] * len(row_data)) + " |")
|
||||
|
||||
lines.append("")
|
||||
|
||||
return "\n".join(lines), ""
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Excel转Markdown失败: {e}")
|
||||
return "", f"Excel处理失败: {str(e)}"
|
||||
|
||||
async def _convert_markdown_to_pdf(self, markdown_content: str, filename: str) -> Tuple[bytes, str]:
|
||||
"""Markdown 转 PDF"""
|
||||
try:
|
||||
logger.info(f"Markdown转PDF开始 - filename={filename}, 字体={self._font_name}")
|
||||
logger.info(f"styles['ChineseTitle'].fontName={self._styles['ChineseTitle'].fontName}")
|
||||
|
||||
buffer = io.BytesIO()
|
||||
story = []
|
||||
|
||||
safe_filename = self._sanitize_text(filename)
|
||||
logger.info(f"safe_filename={repr(safe_filename[:50])}")
|
||||
|
||||
story.append(Paragraph(text=safe_filename, style=self._styles['ChineseTitle']))
|
||||
story.append(Spacer(1, 12))
|
||||
|
||||
in_code = False
|
||||
for line in markdown_content.split('\n'):
|
||||
line = line.strip()
|
||||
|
||||
if line.startswith('```'):
|
||||
in_code = not in_code
|
||||
story.append(Spacer(1, 6))
|
||||
continue
|
||||
|
||||
if in_code:
|
||||
story.append(Paragraph(text=self._sanitize_text(line), style=self._styles['ChineseCode']))
|
||||
continue
|
||||
|
||||
if not line:
|
||||
story.append(Spacer(1, 6))
|
||||
continue
|
||||
|
||||
# 标题处理
|
||||
if line.startswith('# '):
|
||||
story.append(Paragraph(text=self._sanitize_text(line[2:]), style=self._styles['ChineseHeading']))
|
||||
elif line.startswith('## '):
|
||||
story.append(Paragraph(text=self._sanitize_text(line[3:]), style=self._styles['ChineseHeading']))
|
||||
elif line.startswith('### '):
|
||||
story.append(Paragraph(text=self._sanitize_text(line[4:]), style=self._styles['ChineseHeading']))
|
||||
elif line.startswith('#### '):
|
||||
story.append(Paragraph(text=self._sanitize_text(line[5:]), style=self._styles['ChineseHeading']))
|
||||
elif line.startswith('- ') or line.startswith('* '):
|
||||
story.append(Paragraph(text="• " + self._sanitize_text(line[2:]), style=self._styles['ChineseBody']))
|
||||
# 表格处理
|
||||
elif line.startswith('|'):
|
||||
# 跳过 markdown 表格分隔符
|
||||
if set(line.replace('|', '').replace('-', '').replace(':', '').replace(' ', '')) == set():
|
||||
continue
|
||||
# 解析并创建表格
|
||||
table_lines = []
|
||||
for _ in range(50): # 最多50行
|
||||
if line.startswith('|'):
|
||||
row = [cell.strip() for cell in line.split('|')[1:-1]]
|
||||
if not any(row) or set(''.join(row).replace('-', '').replace(':', '').replace(' ', '')) == set():
|
||||
break
|
||||
table_lines.append(row)
|
||||
try:
|
||||
line = next(markdown_content.split('\n').__iter__()).strip()
|
||||
except StopIteration:
|
||||
break
|
||||
else:
|
||||
break
|
||||
|
||||
if table_lines:
|
||||
# 创建表格
|
||||
t = Table(table_lines, colWidths=[100] * len(table_lines[0]))
|
||||
t.setStyle(TableStyle([
|
||||
('FONTNAME', (0, 0), (-1, -1), self._font_name),
|
||||
('FONTSIZE', (0, 0), (-1, -1), 9),
|
||||
('GRID', (0, 0), (-1, -1), 0.5, '#999999'),
|
||||
('BACKGROUND', (0, 0), (-1, 0), '#4472C4'),
|
||||
('TEXTCOLOR', (0, 0), (-1, 0), '#FFFFFF'),
|
||||
]))
|
||||
story.append(t)
|
||||
story.append(Spacer(1, 6))
|
||||
else:
|
||||
story.append(Paragraph(text=self._sanitize_text(line), style=self._styles['ChineseBody']))
|
||||
|
||||
logger.info(f"准备构建PDF，story长度={len(story)}")
|
||||
|
||||
pdf_doc = SimpleDocTemplate(
|
||||
buffer,
|
||||
pagesize=(self._page_width, self._page_height),
|
||||
rightMargin=72,
|
||||
leftMargin=72,
|
||||
topMargin=72,
|
||||
bottomMargin=72
|
||||
)
|
||||
logger.info("调用pdf_doc.build()")
|
||||
pdf_doc.build(story)
|
||||
logger.info("pdf_doc.build()完成")
|
||||
|
||||
result = buffer.getvalue()
|
||||
buffer.close()
|
||||
return result, ""
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Markdown转PDF失败: {e}")
|
||||
import traceback
|
||||
logger.error(f"详细错误: {traceback.format_exc()}")
|
||||
return b"", f"Markdown转PDF失败: {str(e)}"
|
||||
|
||||
def _decode_content(self, file_content: bytes) -> str:
|
||||
"""解码文件内容"""
|
||||
encodings = ['utf-8', 'gbk', 'gb2312', 'gb18030', 'latin-1']
|
||||
for enc in encodings:
|
||||
try:
|
||||
return file_content.decode(enc)
|
||||
except (UnicodeDecodeError, LookupError):
|
||||
continue
|
||||
return file_content.decode('utf-8', errors='replace')
|
||||
|
||||
def get_supported_formats(self) -> List[str]:
|
||||
"""获取支持的格式"""
|
||||
return self.supported_formats
|
||||
|
||||
|
||||
# 全局单例
|
||||
pdf_converter_service = PDFConverterService()
|
||||
@@ -2,21 +2,32 @@
|
||||
RAG 服务模块 - 检索增强生成
|
||||
|
||||
使用 sentence-transformers + Faiss 实现向量检索
|
||||
支持 BM25 关键词检索 + 向量检索混合融合
|
||||
"""
|
||||
import json
|
||||
import logging
|
||||
import os
|
||||
import pickle
|
||||
from typing import Any, Dict, List, Optional
|
||||
import re
|
||||
import math
|
||||
from typing import Any, Dict, List, Optional, Tuple
|
||||
from collections import Counter, defaultdict
|
||||
|
||||
import faiss
|
||||
import numpy as np
|
||||
from sentence_transformers import SentenceTransformer
|
||||
|
||||
from app.config import settings
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# 尝试导入 sentence-transformers
|
||||
try:
|
||||
from sentence_transformers import SentenceTransformer
|
||||
SENTENCE_TRANSFORMERS_AVAILABLE = True
|
||||
except ImportError as e:
|
||||
logger.warning(f"sentence-transformers 导入失败: {e}")
|
||||
SENTENCE_TRANSFORMERS_AVAILABLE = False
|
||||
SentenceTransformer = None
|
||||
|
||||
|
||||
class SimpleDocument:
|
||||
"""简化文档对象"""
|
||||
@@ -25,20 +36,156 @@ class SimpleDocument:
|
||||
self.metadata = metadata
|
||||
|
||||
|
||||
class BM25:
|
||||
"""
|
||||
BM25 关键词检索算法
|
||||
|
||||
一种基于词频和文档频率的信息检索算法，比纯向量搜索更适合关键词精确匹配
|
||||
"""
|
||||
|
||||
def __init__(self, k1: float = 1.5, b: float = 0.75):
|
||||
self.k1 = k1 # 词频饱和参数
|
||||
self.b = b # 文档长度归一化参数
|
||||
self.documents: List[str] = []
|
||||
self.doc_ids: List[str] = []
|
||||
self.avg_doc_length = 0
|
||||
self.doc_freqs: Dict[str, int] = {} # 词 -> 包含该词的文档数
|
||||
self.idf: Dict[str, float] = {} # 词 -> IDF 值
|
||||
self.doc_lengths: List[int] = []
|
||||
self.doc_term_freqs: List[Dict[str, int]] = [] # 每个文档的词频
|
||||
|
||||
def _tokenize(self, text: str) -> List[str]:
|
||||
"""分词（简单的中文分词）"""
|
||||
if not text:
|
||||
return []
|
||||
# 简单分词：按标点和空格分割
|
||||
tokens = re.findall(r'[\u4e00-\u9fff]+|[a-zA-Z0-9]+', text.lower())
|
||||
# 过滤单字符
|
||||
return [t for t in tokens if len(t) > 1]
|
||||
|
||||
def fit(self, documents: List[str], doc_ids: List[str]):
|
||||
"""
|
||||
构建 BM25 索引
|
||||
|
||||
Args:
|
||||
documents: 文档内容列表
|
||||
doc_ids: 文档 ID 列表
|
||||
"""
|
||||
self.documents = documents
|
||||
self.doc_ids = doc_ids
|
||||
n = len(documents)
|
||||
|
||||
# 统计文档频率
|
||||
self.doc_freqs = defaultdict(int)
|
||||
self.doc_lengths = []
|
||||
self.doc_term_freqs = []
|
||||
|
||||
for doc in documents:
|
||||
tokens = self._tokenize(doc)
|
||||
self.doc_lengths.append(len(tokens))
|
||||
doc_tf = Counter(tokens)
|
||||
self.doc_term_freqs.append(doc_tf)
|
||||
|
||||
for term in doc_tf:
|
||||
self.doc_freqs[term] += 1
|
||||
|
||||
# 计算平均文档长度
|
||||
self.avg_doc_length = sum(self.doc_lengths) / n if n > 0 else 0
|
||||
|
||||
# 计算 IDF
|
||||
for term, df in self.doc_freqs.items():
|
||||
# IDF = log((n - df + 0.5) / (df + 0.5))
|
||||
self.idf[term] = math.log((n - df + 0.5) / (df + 0.5) + 1)
|
||||
|
||||
logger.info(f"BM25 索引构建完成: {n} 个文档, {len(self.idf)} 个词项")
|
||||
|
||||
def search(self, query: str, top_k: int = 10) -> List[Tuple[int, float]]:
|
||||
"""
|
||||
搜索相关文档
|
||||
|
||||
Args:
|
||||
query: 查询文本
|
||||
top_k: 返回前 k 个结果
|
||||
|
||||
Returns:
|
||||
[(文档索引, BM25分数), ...]
|
||||
"""
|
||||
if not self.documents:
|
||||
return []
|
||||
|
||||
query_tokens = self._tokenize(query)
|
||||
if not query_tokens:
|
||||
return []
|
||||
|
||||
scores = []
|
||||
n = len(self.documents)
|
||||
|
||||
for idx in range(n):
|
||||
score = self._calculate_score(query_tokens, idx)
|
||||
scores.append((idx, score))
|
||||
|
||||
# 按分数降序排序
|
||||
scores.sort(key=lambda x: x[1], reverse=True)
|
||||
|
||||
return scores[:top_k]
|
||||
|
||||
def _calculate_score(self, query_tokens: List[str], doc_idx: int) -> float:
|
||||
"""计算单个文档的 BM25 分数"""
|
||||
doc_tf = self.doc_term_freqs[doc_idx]
|
||||
doc_len = self.doc_lengths[doc_idx]
|
||||
score = 0.0
|
||||
|
||||
for term in query_tokens:
|
||||
if term not in self.idf:
|
||||
continue
|
||||
|
||||
tf = doc_tf.get(term, 0)
|
||||
idf = self.idf[term]
|
||||
|
||||
# BM25 公式
|
||||
numerator = tf * (self.k1 + 1)
|
||||
denominator = tf + self.k1 * (1 - self.b + self.b * doc_len / self.avg_doc_length)
|
||||
|
||||
score += idf * numerator / denominator
|
||||
|
||||
return score
|
||||
|
||||
def get_scores(self, query: str) -> List[float]:
|
||||
"""获取所有文档的 BM25 分数"""
|
||||
if not self.documents:
|
||||
return []
|
||||
|
||||
query_tokens = self._tokenize(query)
|
||||
if not query_tokens:
|
||||
return [0.0] * len(self.documents)
|
||||
|
||||
return [self._calculate_score(query_tokens, idx) for idx in range(len(self.documents))]
|
||||
|
||||
|
||||
class RAGService:
|
||||
"""RAG 检索增强服务"""
|
||||
|
||||
# 默认分块参数 - 增大块大小减少embedding次数
|
||||
DEFAULT_CHUNK_SIZE = 1000 # 每个文本块的大小（字符数），增大以提升速度
|
||||
DEFAULT_CHUNK_OVERLAP = 100 # 块之间的重叠（字符数）
|
||||
|
||||
def __init__(self):
|
||||
self.embedding_model: Optional[SentenceTransformer] = None
|
||||
self.embedding_model = None
|
||||
self.index: Optional[faiss.Index] = None
|
||||
self.documents: List[Dict[str, Any]] = []
|
||||
self.doc_ids: List[str] = []
|
||||
self._dimension: int = 0
|
||||
self._dimension: int = 384 # 默认维度
|
||||
self._initialized = False
|
||||
self._persist_dir = settings.FAISS_INDEX_DIR
|
||||
# 临时禁用 RAG API 调用，仅记录日志
|
||||
self._disabled = True
|
||||
logger.info("RAG 服务已禁用（_disabled=True），仅记录索引操作日志")
|
||||
# BM25 索引
|
||||
self.bm25: Optional[BM25] = None
|
||||
self._bm25_enabled = True # 始终启用 BM25
|
||||
# 检查是否可用
|
||||
self._disabled = not SENTENCE_TRANSFORMERS_AVAILABLE
|
||||
if self._disabled:
|
||||
logger.warning("RAG 服务已禁用（sentence-transformers 不可用），将使用 BM25 关键词检索")
|
||||
else:
|
||||
logger.info("RAG 服务已启用（向量检索 + BM25 混合检索）")
|
||||
|
||||
def _init_embeddings(self):
|
||||
"""初始化嵌入模型"""
|
||||
@@ -88,6 +235,63 @@ class RAGService:
|
||||
norms = np.where(norms == 0, 1, norms)
|
||||
return vectors / norms
|
||||
|
||||
def _split_into_chunks(self, text: str, chunk_size: int = None, overlap: int = None) -> List[str]:
|
||||
"""
|
||||
将长文本分割成块
|
||||
|
||||
Args:
|
||||
text: 待分割的文本
|
||||
chunk_size: 每个块的大小（字符数）
|
||||
overlap: 块之间的重叠字符数
|
||||
|
||||
Returns:
|
||||
文本块列表
|
||||
"""
|
||||
if chunk_size is None:
|
||||
chunk_size = self.DEFAULT_CHUNK_SIZE
|
||||
if overlap is None:
|
||||
overlap = self.DEFAULT_CHUNK_OVERLAP
|
||||
|
||||
if len(text) <= chunk_size:
|
||||
return [text] if text.strip() else []
|
||||
|
||||
chunks = []
|
||||
start = 0
|
||||
text_len = len(text)
|
||||
|
||||
while start < text_len:
|
||||
# 计算当前块的结束位置
|
||||
end = start + chunk_size
|
||||
|
||||
# 如果不是最后一块，尝试在句子边界处切割
|
||||
if end < text_len:
|
||||
# 向前查找最后一个句号、逗号、换行或分号
|
||||
cut_positions = []
|
||||
for i in range(end, max(start, end - 100), -1):
|
||||
if text[i] in '。；，,\n、':
|
||||
cut_positions.append(i + 1)
|
||||
break
|
||||
|
||||
if cut_positions:
|
||||
end = cut_positions[0]
|
||||
else:
|
||||
# 如果没找到句子边界，尝试向后查找
|
||||
for i in range(end, min(text_len, end + 50)):
|
||||
if text[i] in '。；，,\n、':
|
||||
end = i + 1
|
||||
break
|
||||
|
||||
chunk = text[start:end].strip()
|
||||
if chunk:
|
||||
chunks.append(chunk)
|
||||
|
||||
# 移动起始位置（考虑重叠）
|
||||
start = end - overlap
|
||||
if start <= 0:
|
||||
start = end
|
||||
|
||||
return chunks
|
||||
|
||||
def index_field(
|
||||
self,
|
||||
table_name: str,
|
||||
@@ -124,9 +328,20 @@ class RAGService:
|
||||
self,
|
||||
doc_id: str,
|
||||
content: str,
|
||||
metadata: Optional[Dict[str, Any]] = None
|
||||
metadata: Optional[Dict[str, Any]] = None,
|
||||
chunk_size: int = None,
|
||||
chunk_overlap: int = None
|
||||
):
|
||||
"""将文档内容索引到向量数据库"""
|
||||
"""
|
||||
将文档内容索引到向量数据库（自动分块）
|
||||
|
||||
Args:
|
||||
doc_id: 文档唯一标识
|
||||
content: 文档内容
|
||||
metadata: 文档元数据
|
||||
chunk_size: 文本块大小（字符数），默认500
|
||||
chunk_overlap: 块之间的重叠字符数，默认50
|
||||
"""
|
||||
if self._disabled:
|
||||
logger.info(f"[RAG DISABLED] 文档索引操作已跳过: {doc_id}")
|
||||
return
|
||||
@@ -139,18 +354,134 @@ class RAGService:
|
||||
logger.debug(f"文档跳过索引 (无嵌入模型): {doc_id}")
|
||||
return
|
||||
|
||||
doc = SimpleDocument(
|
||||
page_content=content,
|
||||
metadata=metadata or {"doc_id": doc_id}
|
||||
)
|
||||
self._add_documents([doc], [doc_id])
|
||||
logger.debug(f"已索引文档: {doc_id}")
|
||||
# 分割文档为小块
|
||||
if chunk_size is None:
|
||||
chunk_size = self.DEFAULT_CHUNK_SIZE
|
||||
if chunk_overlap is None:
|
||||
chunk_overlap = self.DEFAULT_CHUNK_OVERLAP
|
||||
|
||||
chunks = self._split_into_chunks(content, chunk_size, chunk_overlap)
|
||||
|
||||
if not chunks:
|
||||
logger.warning(f"文档内容为空，跳过索引: {doc_id}")
|
||||
return
|
||||
|
||||
# 为每个块创建文档对象
|
||||
documents = []
|
||||
chunk_ids = []
|
||||
|
||||
for i, chunk in enumerate(chunks):
|
||||
chunk_id = f"{doc_id}_chunk_{i}"
|
||||
chunk_metadata = metadata.copy() if metadata else {}
|
||||
chunk_metadata.update({
|
||||
"chunk_index": i,
|
||||
"total_chunks": len(chunks),
|
||||
"doc_id": doc_id
|
||||
})
|
||||
|
||||
documents.append(SimpleDocument(
|
||||
page_content=chunk,
|
||||
metadata=chunk_metadata
|
||||
))
|
||||
chunk_ids.append(chunk_id)
|
||||
|
||||
# 批量添加文档
|
||||
self._add_documents(documents, chunk_ids)
|
||||
logger.info(f"已索引文档 {doc_id}，共 {len(chunks)} 个块")
|
||||
|
||||
async def index_document_content_async(
|
||||
self,
|
||||
doc_id: str,
|
||||
content: str,
|
||||
metadata: Optional[Dict[str, Any]] = None,
|
||||
chunk_size: int = None,
|
||||
chunk_overlap: int = None
|
||||
):
|
||||
"""
|
||||
异步将文档内容索引到向量数据库（自动分块）
|
||||
|
||||
使用 asyncio.to_thread 避免阻塞事件循环
|
||||
"""
|
||||
import asyncio
|
||||
|
||||
if self._disabled:
|
||||
logger.info(f"[RAG DISABLED] 文档索引操作已跳过: {doc_id}")
|
||||
return
|
||||
|
||||
if not self._initialized:
|
||||
self._init_vector_store()
|
||||
|
||||
if self.embedding_model is None:
|
||||
logger.debug(f"文档跳过索引 (无嵌入模型): {doc_id}")
|
||||
return
|
||||
|
||||
# 分割文档为小块
|
||||
if chunk_size is None:
|
||||
chunk_size = self.DEFAULT_CHUNK_SIZE
|
||||
if chunk_overlap is None:
|
||||
chunk_overlap = self.DEFAULT_CHUNK_OVERLAP
|
||||
|
||||
chunks = self._split_into_chunks(content, chunk_size, chunk_overlap)
|
||||
|
||||
if not chunks:
|
||||
logger.warning(f"文档内容为空，跳过索引: {doc_id}")
|
||||
return
|
||||
|
||||
# 为每个块创建文档对象
|
||||
documents = []
|
||||
chunk_ids = []
|
||||
|
||||
for i, chunk in enumerate(chunks):
|
||||
chunk_id = f"{doc_id}_chunk_{i}"
|
||||
chunk_metadata = metadata.copy() if metadata else {}
|
||||
chunk_metadata.update({
|
||||
"chunk_index": i,
|
||||
"total_chunks": len(chunks),
|
||||
"doc_id": doc_id
|
||||
})
|
||||
|
||||
documents.append(SimpleDocument(
|
||||
page_content=chunk,
|
||||
metadata=chunk_metadata
|
||||
))
|
||||
chunk_ids.append(chunk_id)
|
||||
|
||||
# 使用线程池执行 CPU 密集型的 embedding 计算
|
||||
def _sync_add():
|
||||
self._add_documents(documents, chunk_ids)
|
||||
|
||||
await asyncio.to_thread(_sync_add)
|
||||
logger.info(f"已异步索引文档 {doc_id}，共 {len(chunks)} 个块")
|
||||
|
||||
def _add_documents(self, documents: List[SimpleDocument], doc_ids: List[str]):
|
||||
"""批量添加文档到向量索引"""
|
||||
if not documents:
|
||||
return
|
||||
|
||||
# 总是将文档存储在内存中（用于 BM25 和关键词搜索）
|
||||
for doc, did in zip(documents, doc_ids):
|
||||
self.documents.append({"id": did, "content": doc.page_content, "metadata": doc.metadata})
|
||||
self.doc_ids.append(did)
|
||||
|
||||
# 构建 BM25 索引
|
||||
if self._bm25_enabled and documents:
|
||||
bm25_texts = [doc.page_content for doc in documents]
|
||||
if self.bm25 is None:
|
||||
self.bm25 = BM25()
|
||||
self.bm25.fit(bm25_texts, doc_ids)
|
||||
else:
|
||||
# 增量添加：重新构建（BM25 不支持增量）
|
||||
all_texts = [d["content"] for d in self.documents]
|
||||
all_ids = self.doc_ids.copy()
|
||||
self.bm25 = BM25()
|
||||
self.bm25.fit(all_texts, all_ids)
|
||||
logger.debug(f"BM25 索引更新: {len(documents)} 个文档")
|
||||
|
||||
# 如果没有嵌入模型，跳过向量索引
|
||||
if self.embedding_model is None:
|
||||
logger.debug(f"文档跳过向量索引 (无嵌入模型): {len(documents)} 个文档")
|
||||
return
|
||||
|
||||
texts = [doc.page_content for doc in documents]
|
||||
embeddings = self.embedding_model.encode(texts, convert_to_numpy=True)
|
||||
embeddings = self._normalize_vectors(embeddings).astype('float32')
|
||||
@@ -162,12 +493,18 @@ class RAGService:
|
||||
id_array = np.array(id_list, dtype='int64')
|
||||
self.index.add_with_ids(embeddings, id_array)
|
||||
|
||||
for doc, did in zip(documents, doc_ids):
|
||||
self.documents.append({"id": did, "content": doc.page_content, "metadata": doc.metadata})
|
||||
self.doc_ids.append(did)
|
||||
def retrieve(self, query: str, top_k: int = 5, min_score: float = 0.3) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
根据查询检索相关文档块（混合检索：向量 + BM25）
|
||||
|
||||
def retrieve(self, query: str, top_k: int = 5) -> List[Dict[str, Any]]:
|
||||
"""根据查询检索相关文档"""
|
||||
Args:
|
||||
query: 查询文本
|
||||
top_k: 返回的最大结果数
|
||||
min_score: 最低相似度分数阈值
|
||||
|
||||
Returns:
|
||||
相关文档块列表，每项包含 content, metadata, score, doc_id, chunk_index
|
||||
"""
|
||||
if self._disabled:
|
||||
logger.info(f"[RAG DISABLED] 检索操作已跳过: query={query}, top_k={top_k}")
|
||||
return []
|
||||
@@ -175,28 +512,241 @@ class RAGService:
|
||||
if not self._initialized:
|
||||
self._init_vector_store()
|
||||
|
||||
if self.index is None or self.index.ntotal == 0:
|
||||
# 获取向量检索结果
|
||||
vector_results = self._vector_search(query, top_k * 2, min_score)
|
||||
|
||||
# 获取 BM25 检索结果
|
||||
bm25_results = self._bm25_search(query, top_k * 2)
|
||||
|
||||
# 混合融合
|
||||
hybrid_results = self._hybrid_fusion(vector_results, bm25_results, top_k)
|
||||
|
||||
if hybrid_results:
|
||||
logger.info(f"混合检索到 {len(hybrid_results)} 条相关文档块 (向量:{len(vector_results)}, BM25:{len(bm25_results)})")
|
||||
return hybrid_results
|
||||
|
||||
# 降级：只使用 BM25
|
||||
if bm25_results:
|
||||
logger.info(f"降级到 BM25 检索: {len(bm25_results)} 条")
|
||||
return bm25_results
|
||||
|
||||
# 降级：使用关键词搜索
|
||||
logger.info("降级到关键词搜索")
|
||||
return self._keyword_search(query, top_k)
|
||||
|
||||
def _vector_search(self, query: str, top_k: int, min_score: float) -> List[Dict[str, Any]]:
|
||||
"""向量检索"""
|
||||
if self.index is None or self.index.ntotal == 0 or self.embedding_model is None:
|
||||
return []
|
||||
|
||||
query_embedding = self.embedding_model.encode([query], convert_to_numpy=True)
|
||||
query_embedding = self._normalize_vectors(query_embedding).astype('float32')
|
||||
try:
|
||||
query_embedding = self.embedding_model.encode([query], convert_to_numpy=True)
|
||||
query_embedding = self._normalize_vectors(query_embedding).astype('float32')
|
||||
|
||||
scores, indices = self.index.search(query_embedding, min(top_k, self.index.ntotal))
|
||||
scores, indices = self.index.search(query_embedding, min(top_k * 2, self.index.ntotal))
|
||||
|
||||
results = []
|
||||
for score, idx in zip(scores[0], indices[0]):
|
||||
if idx < 0:
|
||||
continue
|
||||
doc = self.documents[idx]
|
||||
results.append({
|
||||
"content": doc["content"],
|
||||
"metadata": doc["metadata"],
|
||||
"score": float(score),
|
||||
"doc_id": doc["id"]
|
||||
results = []
|
||||
for score, idx in zip(scores[0], indices[0]):
|
||||
if idx < 0:
|
||||
continue
|
||||
if score < min_score:
|
||||
continue
|
||||
doc = self.documents[idx]
|
||||
results.append({
|
||||
"content": doc["content"],
|
||||
"metadata": doc["metadata"],
|
||||
"score": float(score),
|
||||
"doc_id": doc["id"],
|
||||
"chunk_index": doc["metadata"].get("chunk_index", 0),
|
||||
"search_type": "vector"
|
||||
})
|
||||
|
||||
return results
|
||||
except Exception as e:
|
||||
logger.warning(f"向量检索失败: {e}")
|
||||
return []
|
||||
|
||||
def _bm25_search(self, query: str, top_k: int) -> List[Dict[str, Any]]:
|
||||
"""BM25 检索"""
|
||||
if not self.bm25 or not self.documents:
|
||||
return []
|
||||
|
||||
try:
|
||||
bm25_scores = self.bm25.get_scores(query)
|
||||
if not bm25_scores:
|
||||
return []
|
||||
|
||||
# 归一化 BM25 分数到 [0, 1]
|
||||
max_score = max(bm25_scores) if bm25_scores else 1
|
||||
min_score_bm = min(bm25_scores) if bm25_scores else 0
|
||||
score_range = max_score - min_score_bm if max_score != min_score_bm else 1
|
||||
|
||||
results = []
|
||||
for idx, score in enumerate(bm25_scores):
|
||||
if score <= 0:
|
||||
continue
|
||||
# 归一化
|
||||
normalized_score = (score - min_score_bm) / score_range if score_range > 0 else 0
|
||||
doc = self.documents[idx]
|
||||
results.append({
|
||||
"content": doc["content"],
|
||||
"metadata": doc["metadata"],
|
||||
"score": float(normalized_score),
|
||||
"doc_id": doc["id"],
|
||||
"chunk_index": doc["metadata"].get("chunk_index", 0),
|
||||
"search_type": "bm25"
|
||||
})
|
||||
|
||||
# 按分数降序
|
||||
results.sort(key=lambda x: x["score"], reverse=True)
|
||||
return results[:top_k]
|
||||
|
||||
except Exception as e:
|
||||
logger.warning(f"BM25 检索失败: {e}")
|
||||
return []
|
||||
|
||||
def _hybrid_fusion(
|
||||
self,
|
||||
vector_results: List[Dict[str, Any]],
|
||||
bm25_results: List[Dict[str, Any]],
|
||||
top_k: int
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
混合融合向量和 BM25 检索结果
|
||||
|
||||
使用 RRFR (Reciprocal Rank Fusion) 算法:
|
||||
Score = weight_vector * (1 / rank_vector) + weight_bm25 * (1 / rank_bm25)
|
||||
|
||||
Args:
|
||||
vector_results: 向量检索结果
|
||||
bm25_results: BM25 检索结果
|
||||
top_k: 返回数量
|
||||
|
||||
Returns:
|
||||
融合后的结果
|
||||
"""
|
||||
if not vector_results and not bm25_results:
|
||||
return []
|
||||
|
||||
# 融合权重
|
||||
weight_vector = 0.6
|
||||
weight_bm25 = 0.4
|
||||
|
||||
# 构建文档分数映射
|
||||
doc_scores: Dict[str, Dict[str, float]] = {}
|
||||
|
||||
# 添加向量检索结果
|
||||
for rank, result in enumerate(vector_results):
|
||||
doc_id = result["doc_id"]
|
||||
if doc_id not in doc_scores:
|
||||
doc_scores[doc_id] = {"vector": 0, "bm25": 0, "content": result["content"], "metadata": result["metadata"]}
|
||||
# 使用倒数排名 (Reciprocal Rank)
|
||||
doc_scores[doc_id]["vector"] = weight_vector / (rank + 1)
|
||||
|
||||
# 添加 BM25 检索结果
|
||||
for rank, result in enumerate(bm25_results):
|
||||
doc_id = result["doc_id"]
|
||||
if doc_id not in doc_scores:
|
||||
doc_scores[doc_id] = {"vector": 0, "bm25": 0, "content": result["content"], "metadata": result["metadata"]}
|
||||
doc_scores[doc_id]["bm25"] = weight_bm25 / (rank + 1)
|
||||
|
||||
# 计算融合分数
|
||||
fused_results = []
|
||||
for doc_id, scores in doc_scores.items():
|
||||
fused_score = scores["vector"] + scores["bm25"]
|
||||
# 使用向量检索结果的原始分数作为参考
|
||||
vector_score = next((r["score"] for r in vector_results if r["doc_id"] == doc_id), 0.5)
|
||||
fused_results.append({
|
||||
"content": scores["content"],
|
||||
"metadata": scores["metadata"],
|
||||
"score": fused_score,
|
||||
"doc_id": doc_id,
|
||||
"vector_score": vector_score,
|
||||
"bm25_score": scores["bm25"],
|
||||
"search_type": "hybrid"
|
||||
})
|
||||
|
||||
logger.debug(f"检索到 {len(results)} 条相关文档")
|
||||
return results
|
||||
# 按融合分数降序排序
|
||||
fused_results.sort(key=lambda x: x["score"], reverse=True)
|
||||
|
||||
logger.info(f"RRF 混合融合: {len(fused_results)} 个文档参与融合, 向量检索命中:{len(vector_results)}, BM25命中:{len(bm25_results)}")
|
||||
|
||||
return fused_results[:top_k]
|
||||
|
||||
def _keyword_search(self, query: str, top_k: int = 5) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
关键词搜索后备方案
|
||||
|
||||
Args:
|
||||
query: 查询文本
|
||||
top_k: 返回的最大结果数
|
||||
|
||||
Returns:
|
||||
相关文档块列表
|
||||
"""
|
||||
if not self.documents:
|
||||
return []
|
||||
|
||||
# 提取查询关键词
|
||||
keywords = []
|
||||
for char in query:
|
||||
if '\u4e00' <= char <= '\u9fff': # 中文字符
|
||||
keywords.append(char)
|
||||
# 添加英文单词
|
||||
import re
|
||||
english_words = re.findall(r'[a-zA-Z]+', query)
|
||||
keywords.extend(english_words)
|
||||
|
||||
if not keywords:
|
||||
return []
|
||||
|
||||
results = []
|
||||
for doc in self.documents:
|
||||
content = doc["content"]
|
||||
# 计算关键词匹配分数
|
||||
score = 0
|
||||
matched_keywords = 0
|
||||
for kw in keywords:
|
||||
if kw in content:
|
||||
score += 1
|
||||
matched_keywords += 1
|
||||
|
||||
if matched_keywords > 0:
|
||||
# 归一化分数
|
||||
score = score / max(len(keywords), 1)
|
||||
results.append({
|
||||
"content": content,
|
||||
"metadata": doc["metadata"],
|
||||
"score": score,
|
||||
"doc_id": doc["id"],
|
||||
"chunk_index": doc["metadata"].get("chunk_index", 0)
|
||||
})
|
||||
|
||||
# 按分数排序
|
||||
results.sort(key=lambda x: x["score"], reverse=True)
|
||||
|
||||
logger.debug(f"关键词搜索返回 {len(results[:top_k])} 条结果")
|
||||
return results[:top_k]
|
||||
|
||||
def retrieve_by_doc_id(self, doc_id: str, top_k: int = 10) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
获取指定文档的所有块
|
||||
|
||||
Args:
|
||||
doc_id: 文档ID
|
||||
top_k: 返回的最大结果数
|
||||
|
||||
Returns:
|
||||
该文档的所有块
|
||||
"""
|
||||
# 获取属于该文档的所有块
|
||||
doc_chunks = [d for d in self.documents if d["metadata"].get("doc_id") == doc_id]
|
||||
|
||||
# 按 chunk_index 排序
|
||||
doc_chunks.sort(key=lambda x: x["metadata"].get("chunk_index", 0))
|
||||
|
||||
# 返回指定数量
|
||||
return doc_chunks[:top_k]
|
||||
|
||||
def retrieve_by_table(self, table_name: str, top_k: int = 5) -> List[Dict[str, Any]]:
|
||||
"""检索指定表的字段"""
|
||||
|
||||
@@ -300,13 +300,15 @@ class TableRAGService:
|
||||
filename: str,
|
||||
sheet_name: Optional[str] = None,
|
||||
header_row: int = 0,
|
||||
sample_size: int = 10
|
||||
sample_size: int = 10,
|
||||
skip_rag_index: bool = False
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
为 Excel 表构建完整的 RAG 索引
|
||||
|
||||
流程：
|
||||
1. 读取 Excel 获取字段信息
|
||||
2. 如果 skip_rag_index=True，跳过 RAG 索引，直接存 MySQL
|
||||
2. AI 生成每个字段的语义描述
|
||||
3. 将字段描述存入向量数据库
|
||||
|
||||
@@ -367,6 +369,20 @@ class TableRAGService:
|
||||
results["field_count"] = len(df.columns)
|
||||
logger.info(f"表名: {table_name}, 字段数: {len(df.columns)}")
|
||||
|
||||
# 跳过 RAG 索引时直接存 MySQL
|
||||
if skip_rag_index:
|
||||
logger.info(f"跳过 RAG 索引，直接存储到 MySQL")
|
||||
store_result = await self.excel_storage.store_excel(
|
||||
file_path=file_path,
|
||||
filename=filename,
|
||||
sheet_name=sheet_name,
|
||||
header_row=header_row
|
||||
)
|
||||
results["mysql_table"] = store_result.get("table_name") if store_result.get("success") else None
|
||||
results["row_count"] = store_result.get("row_count", len(df))
|
||||
results["indexed_count"] = 0
|
||||
return results
|
||||
|
||||
# 3. 初始化 RAG (如果需要)
|
||||
if not self.rag._initialized:
|
||||
self.rag._init_vector_store()
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
353
backend/app/services/txt_ai_service.py
Normal file
353
backend/app/services/txt_ai_service.py
Normal file
@@ -0,0 +1,353 @@
|
||||
"""
|
||||
TXT 文档 AI 分析服务
|
||||
|
||||
使用 LLM 对 TXT 文本文件进行深度分析，提取结构化数据并生成可视化图表
|
||||
"""
|
||||
import logging
|
||||
import re
|
||||
from typing import Any, Dict, List, Optional
|
||||
|
||||
from app.services.llm_service import llm_service
|
||||
from app.services.visualization_service import visualization_service
|
||||
from app.core.document_parser.txt_parser import TxtParser
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class TxtAIService:
|
||||
"""TXT 文档 AI 分析服务"""
|
||||
|
||||
def __init__(self):
|
||||
self.parser = TxtParser()
|
||||
self.llm = llm_service
|
||||
|
||||
async def analyze_txt_with_ai(
|
||||
self,
|
||||
content: str,
|
||||
filename: str = "",
|
||||
user_hint: str = "",
|
||||
analysis_type: str = "structured"
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
使用 AI 解析 TXT 文本文件
|
||||
|
||||
Args:
|
||||
content: 文本内容
|
||||
filename: 文件名（可选）
|
||||
user_hint: 用户提示词
|
||||
analysis_type: 分析类型 - "structured"（默认，提取结构化数据）或 "charts"（生成图表）
|
||||
|
||||
Returns:
|
||||
Dict: 包含结构化数据的分析结果
|
||||
"""
|
||||
try:
|
||||
if not content or not content.strip():
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档内容为空"
|
||||
}
|
||||
|
||||
# 根据分析类型选择处理方式
|
||||
if analysis_type == "charts":
|
||||
return await self.generate_charts(content, filename, user_hint)
|
||||
|
||||
# 默认：提取结构化数据
|
||||
return await self._extract_structured_data(content, filename, user_hint)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"TXT AI 分析失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e)
|
||||
}
|
||||
|
||||
async def _extract_structured_data(
|
||||
self,
|
||||
content: str,
|
||||
filename: str = "",
|
||||
user_hint: str = ""
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
从文本中提取结构化数据
|
||||
|
||||
Args:
|
||||
content: 文本内容
|
||||
filename: 文件名
|
||||
user_hint: 用户提示词
|
||||
|
||||
Returns:
|
||||
结构化数据
|
||||
"""
|
||||
try:
|
||||
# 截断内容避免超出 token 限制
|
||||
max_content_len = 8000
|
||||
text_preview = content[:max_content_len] if len(content) > max_content_len else content
|
||||
|
||||
prompt = f"""你是一个专业的数据提取专家。请从以下文本中提取结构化数据。
|
||||
|
||||
【用户需求】
|
||||
{user_hint if user_hint else "请提取文档中的所有结构化数据，包括表格数据、键值对、列表项等。"}
|
||||
|
||||
【文档内容】（{"前" + str(max_content_len) + "字符，仅显示部分" if len(content) > max_content_len else "全文"}）
|
||||
{text_preview}
|
||||
|
||||
请按照以下 JSON 格式输出：
|
||||
{{
|
||||
"type": "structured_text",
|
||||
"tables": [{{"headers": [...], "rows": [...]}}],
|
||||
"key_values": {{"键1": "值1", "键2": "值2", ...}},
|
||||
"list_items": ["项1", "项2", ...],
|
||||
"summary": "文档内容摘要"
|
||||
}}
|
||||
|
||||
重点：
|
||||
- 如果文档包含表格数据（制表符、空格对齐等），提取到 tables 中
|
||||
- 如果文档包含键值对（如 名称: 张三），提取到 key_values 中
|
||||
- 如果文档包含列表项，提取到 list_items 中
|
||||
- 如果无法提取到结构化数据，至少提供一个详细的摘要
|
||||
"""
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个专业的数据提取助手。请严格按JSON格式输出。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
response = await self.llm.chat(
|
||||
messages=messages,
|
||||
temperature=0.1,
|
||||
max_tokens=8000
|
||||
)
|
||||
|
||||
content_text = self.llm.extract_message_content(response)
|
||||
result = self._parse_json_response(content_text)
|
||||
|
||||
if result:
|
||||
logger.info(f"TXT 结构化数据提取成功: type={result.get('type')}")
|
||||
return {
|
||||
"success": True,
|
||||
"type": result.get("type", "structured_text"),
|
||||
"tables": result.get("tables", []),
|
||||
"key_values": result.get("key_values", {}),
|
||||
"list_items": result.get("list_items", []),
|
||||
"summary": result.get("summary", "")
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": True,
|
||||
"type": "text",
|
||||
"summary": text_preview[:500],
|
||||
"raw_text_preview": text_preview[:500]
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"TXT 结构化数据提取失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e)
|
||||
}
|
||||
|
||||
async def generate_charts(
|
||||
self,
|
||||
content: str,
|
||||
filename: str = "",
|
||||
user_hint: str = ""
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
从文本中提取数据并生成可视化图表
|
||||
|
||||
Args:
|
||||
content: 文本内容
|
||||
filename: 文件名
|
||||
user_hint: 用户提示词
|
||||
|
||||
Returns:
|
||||
包含图表数据和统计信息的结果
|
||||
"""
|
||||
try:
|
||||
# 截断内容避免超出 token 限制
|
||||
max_content_len = 8000
|
||||
text_preview = content[:max_content_len] if len(content) > max_content_len else content
|
||||
|
||||
# 使用 LLM 提取可用于图表的数据
|
||||
prompt = f"""你是一个专业的数据可视化助手。请从以下文本中提取可用于可视化的数据。
|
||||
|
||||
文档标题：{filename}
|
||||
|
||||
文档内容：
|
||||
{text_preview}
|
||||
|
||||
请完成以下任务：
|
||||
1. 识别文本中的表格数据（制表符分隔、空格对齐的表格等）
|
||||
2. 识别文本中的关键统计数据（百分比、数量、趋势等）
|
||||
3. 识别可用于比较的分类数据
|
||||
|
||||
请用 JSON 格式返回以下结构的数据（如果没有表格数据，返回空结构）：
|
||||
{{
|
||||
"tables": [
|
||||
{{
|
||||
"description": "表格的描述",
|
||||
"columns": ["列名1", "列名2", ...],
|
||||
"rows": [
|
||||
["值1", "值2", ...],
|
||||
["值1", "值2", ...]
|
||||
]
|
||||
}}
|
||||
],
|
||||
"key_statistics": [
|
||||
{{
|
||||
"name": "指标名称",
|
||||
"value": "数值",
|
||||
"trend": "增长/下降/持平",
|
||||
"description": "指标说明"
|
||||
}}
|
||||
],
|
||||
"chart_suggestions": [
|
||||
{{
|
||||
"chart_type": "bar/line/pie",
|
||||
"title": "图表标题",
|
||||
"data_source": "数据来源说明"
|
||||
}}
|
||||
]
|
||||
}}
|
||||
|
||||
如果没有表格数据，返回空结构：{{"tables": [], "key_statistics": [], "chart_suggestions": []}}
|
||||
请确保返回的是合法的 JSON 格式。"""
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": "你是一个专业的数据可视化助手，擅长从文本中提取数据并生成图表。"},
|
||||
{"role": "user", "content": prompt}
|
||||
]
|
||||
|
||||
response = await self.llm.chat(
|
||||
messages=messages,
|
||||
temperature=0.1,
|
||||
max_tokens=8000
|
||||
)
|
||||
|
||||
content_text = self.llm.extract_message_content(response)
|
||||
chart_data = self._parse_json_response(content_text)
|
||||
|
||||
if not chart_data:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "无法从文本中提取有效的数据结构"
|
||||
}
|
||||
|
||||
# 检查是否有表格数据
|
||||
tables = chart_data.get("tables", [])
|
||||
key_statistics = chart_data.get("key_statistics", [])
|
||||
|
||||
if not tables:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档中没有可用于图表的表格数据",
|
||||
"key_statistics": key_statistics,
|
||||
"chart_suggestions": chart_data.get("chart_suggestions", [])
|
||||
}
|
||||
|
||||
# 使用第一个表格生成图表
|
||||
first_table = tables[0]
|
||||
columns = first_table.get("columns", [])
|
||||
rows = first_table.get("rows", [])
|
||||
|
||||
if not columns or not rows:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "表格数据为空"
|
||||
}
|
||||
|
||||
# 转换为 visualization_service 需要的格式
|
||||
viz_data = {
|
||||
"columns": columns,
|
||||
"rows": rows
|
||||
}
|
||||
|
||||
# 生成可视化图表
|
||||
logger.info(f"开始生成图表，列数: {len(columns)}, 行数: {len(rows)}")
|
||||
vis_result = visualization_service.analyze_and_visualize(viz_data)
|
||||
|
||||
if vis_result.get("success"):
|
||||
return {
|
||||
"success": True,
|
||||
"charts": vis_result.get("charts", {}),
|
||||
"statistics": vis_result.get("statistics", {}),
|
||||
"distributions": vis_result.get("distributions", {}),
|
||||
"row_count": vis_result.get("row_count", 0),
|
||||
"column_count": vis_result.get("column_count", 0),
|
||||
"key_statistics": key_statistics,
|
||||
"chart_suggestions": chart_data.get("chart_suggestions", []),
|
||||
"table_description": first_table.get("description", "")
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": vis_result.get("error", "可视化生成失败"),
|
||||
"key_statistics": key_statistics
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"TXT 图表生成失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e)
|
||||
}
|
||||
|
||||
def _parse_json_response(self, content: str) -> Optional[Dict]:
|
||||
"""解析 JSON 响应，处理各种格式问题"""
|
||||
if not content:
|
||||
return None
|
||||
|
||||
import json
|
||||
|
||||
# 清理 markdown 标记
|
||||
cleaned = content.strip()
|
||||
cleaned = re.sub(r'^```json\s*', '', cleaned, flags=re.MULTILINE)
|
||||
cleaned = re.sub(r'^```\s*', '', cleaned, flags=re.MULTILINE)
|
||||
cleaned = cleaned.strip()
|
||||
|
||||
# 找到 JSON 开始位置
|
||||
json_start = -1
|
||||
for i, c in enumerate(cleaned):
|
||||
if c == '{':
|
||||
json_start = i
|
||||
break
|
||||
|
||||
if json_start == -1:
|
||||
logger.warning("无法找到 JSON 开始位置")
|
||||
return None
|
||||
|
||||
json_text = cleaned[json_start:]
|
||||
|
||||
# 尝试直接解析
|
||||
try:
|
||||
return json.loads(json_text)
|
||||
except json.JSONDecodeError:
|
||||
pass
|
||||
|
||||
# 尝试修复并解析
|
||||
try:
|
||||
# 找到闭合括号
|
||||
depth = 0
|
||||
end_pos = -1
|
||||
for i, c in enumerate(json_text):
|
||||
if c == '{':
|
||||
depth += 1
|
||||
elif c == '}':
|
||||
depth -= 1
|
||||
if depth == 0:
|
||||
end_pos = i + 1
|
||||
break
|
||||
|
||||
if end_pos > 0:
|
||||
fixed = json_text[:end_pos]
|
||||
# 移除末尾逗号
|
||||
fixed = re.sub(r',\s*([}]])', r'\1', fixed)
|
||||
return json.loads(fixed)
|
||||
except Exception as e:
|
||||
logger.warning(f"JSON 修复失败: {e}")
|
||||
|
||||
return None
|
||||
|
||||
|
||||
# 全局单例
|
||||
txt_ai_service = TxtAIService()
|
||||
@@ -53,7 +53,11 @@ class VisualizationService:
|
||||
}
|
||||
|
||||
# 转换为 DataFrame
|
||||
df = pd.DataFrame(rows, columns=columns)
|
||||
# 过滤掉行数与列数不匹配的数据
|
||||
valid_rows = [row for row in rows if len(row) == len(columns)]
|
||||
if len(valid_rows) < len(rows):
|
||||
logger.warning(f"过滤了 {len(rows) - len(valid_rows)} 行无效数据（列数不匹配）")
|
||||
df = pd.DataFrame(valid_rows, columns=columns)
|
||||
|
||||
# 根据列类型分类
|
||||
numeric_columns = df.select_dtypes(include=[np.number]).columns.tolist()
|
||||
@@ -141,18 +145,18 @@ class VisualizationService:
|
||||
charts = {}
|
||||
|
||||
# 1. 数值型列的直方图
|
||||
charts["histograms"] = []
|
||||
charts["numeric_charts"] = []
|
||||
for col in numeric_columns[:5]: # 限制最多 5 个数值列
|
||||
chart_data = self._create_histogram(df[col], col)
|
||||
if chart_data:
|
||||
charts["histograms"].append(chart_data)
|
||||
charts["numeric_charts"].append(chart_data)
|
||||
|
||||
# 2. 分类型列的条形图
|
||||
charts["bar_charts"] = []
|
||||
charts["categorical_charts"] = []
|
||||
for col in categorical_columns[:5]: # 限制最多 5 个分类型列
|
||||
chart_data = self._create_bar_chart(df[col], col)
|
||||
if chart_data:
|
||||
charts["bar_charts"].append(chart_data)
|
||||
charts["categorical_charts"].append(chart_data)
|
||||
|
||||
# 3. 数值型列的箱线图
|
||||
charts["box_plots"] = []
|
||||
|
||||
@@ -8,6 +8,7 @@ from typing import Dict, Any, List, Optional
|
||||
import json
|
||||
|
||||
from app.services.llm_service import llm_service
|
||||
from app.services.visualization_service import visualization_service
|
||||
from app.core.document_parser.docx_parser import DocxParser
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
@@ -183,7 +184,7 @@ class WordAIService:
|
||||
response = await self.llm.chat(
|
||||
messages=messages,
|
||||
temperature=0.1,
|
||||
max_tokens=50000
|
||||
max_tokens=8000
|
||||
)
|
||||
|
||||
content = self.llm.extract_message_content(response)
|
||||
@@ -192,13 +193,15 @@ class WordAIService:
|
||||
result = self._parse_json_response(content)
|
||||
|
||||
if result:
|
||||
logger.info(f"AI 表格提取成功: {len(result.get('rows', []))} 行数据")
|
||||
logger.info(f"AI 表格提取成功: {len(result.get('rows', []))} 行数据, key_values={len(result.get('key_values', {}))}, list_items={len(result.get('list_items', []))}")
|
||||
return {
|
||||
"success": True,
|
||||
"type": "table_data",
|
||||
"headers": result.get("headers", []),
|
||||
"rows": result.get("rows", []),
|
||||
"description": result.get("description", "")
|
||||
"description": result.get("description", ""),
|
||||
"key_values": result.get("key_values", {}),
|
||||
"list_items": result.get("list_items", [])
|
||||
}
|
||||
else:
|
||||
# 如果 AI 返回格式不对，尝试直接解析表格
|
||||
@@ -273,7 +276,7 @@ class WordAIService:
|
||||
response = await self.llm.chat(
|
||||
messages=messages,
|
||||
temperature=0.1,
|
||||
max_tokens=50000
|
||||
max_tokens=8000
|
||||
)
|
||||
|
||||
content = self.llm.extract_message_content(response)
|
||||
@@ -632,6 +635,281 @@ class WordAIService:
|
||||
|
||||
return values
|
||||
|
||||
async def generate_charts(
|
||||
self,
|
||||
file_path: str,
|
||||
user_hint: str = ""
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
使用 AI 解析 Word 文档并生成可视化图表
|
||||
|
||||
从 Word 文档中提取表格数据，然后生成统计图表
|
||||
|
||||
Args:
|
||||
file_path: Word 文件路径
|
||||
user_hint: 用户提示词，指定要提取的内容类型
|
||||
|
||||
Returns:
|
||||
Dict: 包含图表数据和统计信息的结果
|
||||
"""
|
||||
try:
|
||||
# 1. 先用基础解析器提取原始内容
|
||||
parse_result = self.parser.parse(file_path)
|
||||
|
||||
if not parse_result.success:
|
||||
return {
|
||||
"success": False,
|
||||
"error": parse_result.error,
|
||||
"structured_data": None
|
||||
}
|
||||
|
||||
# 2. 获取原始数据
|
||||
raw_data = parse_result.data
|
||||
paragraphs = raw_data.get("paragraphs", [])
|
||||
tables = raw_data.get("tables", [])
|
||||
content = raw_data.get("content", "")
|
||||
|
||||
logger.info(f"Word 基础解析完成: {len(paragraphs)} 个段落, {len(tables)} 个表格")
|
||||
|
||||
# 3. 优先处理表格数据
|
||||
if tables and len(tables) > 0:
|
||||
structured_data = await self._extract_tables_with_ai(
|
||||
tables, paragraphs, 0, user_hint, parse_result.metadata
|
||||
)
|
||||
elif paragraphs and len(paragraphs) > 0:
|
||||
structured_data = await self._extract_from_text_with_ai(
|
||||
paragraphs, content, 0, [], user_hint
|
||||
)
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档内容为空",
|
||||
"structured_data": None
|
||||
}
|
||||
|
||||
# 4. 检查是否有表格数据用于可视化
|
||||
if not structured_data.get("success"):
|
||||
return {
|
||||
"success": False,
|
||||
"error": structured_data.get("error", "解析失败"),
|
||||
"structured_data": None
|
||||
}
|
||||
|
||||
parse_type = structured_data.get("type", "")
|
||||
|
||||
# 5. 提取可用于图表的数据
|
||||
chart_data = None
|
||||
|
||||
if parse_type == "table_data":
|
||||
headers = structured_data.get("headers", [])
|
||||
rows = structured_data.get("rows", [])
|
||||
if headers and rows:
|
||||
chart_data = {
|
||||
"columns": headers,
|
||||
"rows": rows
|
||||
}
|
||||
elif parse_type == "structured_text":
|
||||
tables = structured_data.get("tables", [])
|
||||
if tables and len(tables) > 0:
|
||||
first_table = tables[0]
|
||||
headers = first_table.get("headers", [])
|
||||
rows = first_table.get("rows", [])
|
||||
if headers and rows:
|
||||
chart_data = {
|
||||
"columns": headers,
|
||||
"rows": rows
|
||||
}
|
||||
|
||||
# 6. 生成可视化图表
|
||||
if chart_data:
|
||||
logger.info(f"开始生成图表，列数: {len(chart_data['columns'])}, 行数: {len(chart_data['rows'])}")
|
||||
vis_result = visualization_service.analyze_and_visualize(chart_data)
|
||||
|
||||
if vis_result.get("success"):
|
||||
return {
|
||||
"success": True,
|
||||
"charts": vis_result.get("charts", {}),
|
||||
"statistics": vis_result.get("statistics", {}),
|
||||
"distributions": vis_result.get("distributions", {}),
|
||||
"structured_data": structured_data,
|
||||
"row_count": vis_result.get("row_count", 0),
|
||||
"column_count": vis_result.get("column_count", 0)
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": vis_result.get("error", "可视化生成失败"),
|
||||
"structured_data": structured_data
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档中没有可用于图表的表格数据",
|
||||
"structured_data": structured_data
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Word 文档图表生成失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
"structured_data": None
|
||||
}
|
||||
|
||||
|
||||
async def parse_word_with_ai_from_db(
|
||||
self,
|
||||
content: str,
|
||||
tables: List[Dict],
|
||||
filename: str = "",
|
||||
user_hint: str = ""
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
使用 AI 解析从数据库读取的 Word 文档内容，提取结构化数据
|
||||
|
||||
Args:
|
||||
content: 文档文本内容
|
||||
tables: 表格数据列表
|
||||
filename: 文件名
|
||||
user_hint: 用户提示词
|
||||
|
||||
Returns:
|
||||
Dict: 包含结构化数据的解析结果
|
||||
"""
|
||||
try:
|
||||
# 解析段落
|
||||
paragraphs = [p.strip() for p in content.split('\n') if p.strip()]
|
||||
|
||||
logger.info(f"从数据库解析 Word: {len(paragraphs)} 个段落, {len(tables)} 个表格")
|
||||
|
||||
# 优先处理表格数据
|
||||
if tables and len(tables) > 0:
|
||||
structured_data = await self._extract_tables_with_ai(
|
||||
tables, paragraphs, 0, user_hint, {"filename": filename}
|
||||
)
|
||||
elif paragraphs and len(paragraphs) > 0:
|
||||
structured_data = await self._extract_from_text_with_ai(
|
||||
paragraphs, content, 0, [], user_hint
|
||||
)
|
||||
else:
|
||||
structured_data = {
|
||||
"success": True,
|
||||
"type": "empty",
|
||||
"message": "文档内容为空"
|
||||
}
|
||||
|
||||
return structured_data
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"从数据库解析 Word 文档失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e)
|
||||
}
|
||||
|
||||
async def generate_charts_from_db(
|
||||
self,
|
||||
content: str,
|
||||
tables: List[Dict],
|
||||
filename: str = "",
|
||||
user_hint: str = ""
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
使用 AI 解析从数据库读取的 Word 文档并生成可视化图表
|
||||
|
||||
Args:
|
||||
content: 文档文本内容
|
||||
tables: 表格数据列表
|
||||
filename: 文件名
|
||||
user_hint: 用户提示词
|
||||
|
||||
Returns:
|
||||
Dict: 包含图表数据和统计信息的结果
|
||||
"""
|
||||
try:
|
||||
# 解析段落
|
||||
paragraphs = [p.strip() for p in content.split('\n') if p.strip()]
|
||||
|
||||
logger.info(f"从数据库生成 Word 图表: {len(paragraphs)} 个段落, {len(tables)} 个表格")
|
||||
|
||||
# 优先处理表格数据
|
||||
if tables and len(tables) > 0:
|
||||
structured_data = await self._extract_tables_with_ai(
|
||||
tables, paragraphs, 0, user_hint, {"filename": filename}
|
||||
)
|
||||
elif paragraphs and len(paragraphs) > 0:
|
||||
structured_data = await self._extract_from_text_with_ai(
|
||||
paragraphs, content, 0, [], user_hint
|
||||
)
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档内容为空"
|
||||
}
|
||||
|
||||
# 提取可用于图表的数据
|
||||
chart_data = None
|
||||
logger.info(f"准备提取图表数据，structured_data type: {structured_data.get('type')}, keys: {list(structured_data.keys())}")
|
||||
|
||||
if structured_data.get("type") == "table_data":
|
||||
headers = structured_data.get("headers", [])
|
||||
rows = structured_data.get("rows", [])
|
||||
logger.info(f"table_data类型: headers数量={len(headers)}, rows数量={len(rows)}")
|
||||
if headers and rows:
|
||||
chart_data = {
|
||||
"columns": headers,
|
||||
"rows": rows
|
||||
}
|
||||
elif structured_data.get("type") == "structured_text":
|
||||
tables_data = structured_data.get("tables", [])
|
||||
logger.info(f"structured_text类型: tables数量={len(tables_data)}")
|
||||
if tables_data and len(tables_data) > 0:
|
||||
first_table = tables_data[0]
|
||||
headers = first_table.get("headers", [])
|
||||
rows = first_table.get("rows", [])
|
||||
logger.info(f"第一个表格: headers={headers[:5]}, rows数量={len(rows)}")
|
||||
if headers and rows:
|
||||
chart_data = {
|
||||
"columns": headers,
|
||||
"rows": rows
|
||||
}
|
||||
else:
|
||||
logger.warning(f"无法识别的structured_data类型: {structured_data.get('type')}")
|
||||
|
||||
# 生成可视化图表
|
||||
if chart_data:
|
||||
logger.info(f"开始生成图表，列数: {len(chart_data['columns'])}, 行数: {len(chart_data['rows'])}")
|
||||
vis_result = visualization_service.analyze_and_visualize(chart_data)
|
||||
|
||||
if vis_result.get("success"):
|
||||
return {
|
||||
"success": True,
|
||||
"charts": vis_result.get("charts", {}),
|
||||
"statistics": vis_result.get("statistics", {}),
|
||||
"distributions": vis_result.get("distributions", {}),
|
||||
"structured_data": structured_data,
|
||||
"row_count": vis_result.get("row_count", 0),
|
||||
"column_count": vis_result.get("column_count", 0)
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": vis_result.get("error", "可视化生成失败"),
|
||||
"structured_data": structured_data
|
||||
}
|
||||
else:
|
||||
return {
|
||||
"success": False,
|
||||
"error": "文档中没有可用于图表的表格数据",
|
||||
"structured_data": structured_data
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"从数据库生成 Word 图表失败: {str(e)}")
|
||||
return {
|
||||
"success": False,
|
||||
"error": str(e)
|
||||
}
|
||||
|
||||
|
||||
# 全局单例
|
||||
word_ai_service = WordAIService()
|
||||
|
||||
@@ -39,6 +39,11 @@ openpyxl==3.1.2
|
||||
python-docx==0.8.11
|
||||
markdown-it-py==3.0.0
|
||||
chardet==5.2.0
|
||||
Pillow>=10.0.0
|
||||
pytesseract>=0.3.10
|
||||
|
||||
# ==================== PDF 生成 ====================
|
||||
reportlab>=4.0.0
|
||||
|
||||
# ==================== AI / LLM ====================
|
||||
httpx==0.25.2
|
||||
|
||||
203
docker-compose.yml
Normal file
203
docker-compose.yml
Normal file
@@ -0,0 +1,203 @@
|
||||
# ============================================================
|
||||
# FilesReadSystem Docker Compose
|
||||
# 全栈 AI 文档理解与数据融合系统
|
||||
# ============================================================
|
||||
version: "3.8"
|
||||
|
||||
services:
|
||||
# ==================== 数据库服务 ====================
|
||||
|
||||
mongodb:
|
||||
image: mongo:7.0
|
||||
container_name: filesread_mongodb
|
||||
restart: unless-stopped
|
||||
ports:
|
||||
- "27017:27017"
|
||||
environment:
|
||||
MONGO_INITDB_ROOT_USERNAME: ${MONGO_ROOT_USER:-admin}
|
||||
MONGO_INITDB_ROOT_PASSWORD: ${MONGO_ROOT_PASSWORD:-20060825fhy}
|
||||
MONGO_INITDB_DATABASE: ${MONGODB_DB_NAME:-document_system}
|
||||
volumes:
|
||||
- mongodb_data:/data/db
|
||||
networks:
|
||||
- filesread_network
|
||||
healthcheck:
|
||||
test: ["CMD", "mongosh", "--eval", "db.adminCommand('ping')", "--quiet"]
|
||||
interval: 10s
|
||||
timeout: 5s
|
||||
retries: 5
|
||||
start_period: 30s
|
||||
|
||||
mysql:
|
||||
image: mysql:8.0
|
||||
container_name: filesread_mysql
|
||||
restart: unless-stopped
|
||||
ports:
|
||||
- "3306:3306"
|
||||
environment:
|
||||
MYSQL_ROOT_PASSWORD: ${MYSQL_PASSWORD:-123456}
|
||||
MYSQL_DATABASE: ${MYSQL_DATABASE:-document}
|
||||
volumes:
|
||||
- mysql_data:/var/lib/mysql
|
||||
networks:
|
||||
- filesread_network
|
||||
healthcheck:
|
||||
test: ["CMD", "mysqladmin", "ping", "-h", "localhost", "-u", "root", "-p${MYSQL_PASSWORD:-123456}"]
|
||||
interval: 10s
|
||||
timeout: 5s
|
||||
retries: 5
|
||||
start_period: 30s
|
||||
|
||||
redis:
|
||||
image: redis:7-alpine
|
||||
container_name: filesread_redis
|
||||
restart: unless-stopped
|
||||
ports:
|
||||
- "6379:6379"
|
||||
volumes:
|
||||
- redis_data:/data
|
||||
networks:
|
||||
- filesread_network
|
||||
command: redis-server --appendonly yes --requirepass ${REDIS_PASSWORD:-}
|
||||
healthcheck:
|
||||
test: ["CMD", "redis-cli", "ping"]
|
||||
interval: 10s
|
||||
timeout: 5s
|
||||
retries: 5
|
||||
|
||||
# ==================== 应用服务 ====================
|
||||
|
||||
backend:
|
||||
build:
|
||||
context: ./backend
|
||||
dockerfile: Dockerfile
|
||||
container_name: filesread_backend
|
||||
restart: unless-stopped
|
||||
ports:
|
||||
- "8000:8000"
|
||||
environment:
|
||||
# 应用配置
|
||||
APP_NAME: FilesReadSystem
|
||||
DEBUG: ${DEBUG:-false}
|
||||
API_V1_STR: /api/v1
|
||||
|
||||
# MongoDB 配置 (使用 docker-compose 服务名)
|
||||
MONGODB_URL: mongodb://${MONGO_ROOT_USER:-admin}:${MONGO_ROOT_PASSWORD:-20060825fhy}@mongodb:27017/admin
|
||||
MONGODB_DB_NAME: ${MONGODB_DB_NAME:-document_system}
|
||||
|
||||
# MySQL 配置
|
||||
MYSQL_HOST: mysql
|
||||
MYSQL_PORT: 3306
|
||||
MYSQL_USER: root
|
||||
MYSQL_PASSWORD: ${MYSQL_PASSWORD:-123456}
|
||||
MYSQL_DATABASE: ${MYSQL_DATABASE:-document}
|
||||
MYSQL_CHARSET: utf8mb4
|
||||
|
||||
# Redis 配置
|
||||
REDIS_URL: redis://:${REDIS_PASSWORD:-}@redis:6379/0
|
||||
|
||||
# LLM AI 配置
|
||||
LLM_API_KEY: ${LLM_API_KEY}
|
||||
LLM_BASE_URL: ${LLM_BASE_URL:-https://api.deepseek.com}
|
||||
LLM_MODEL_NAME: ${LLM_MODEL_NAME:-deepseek-chat}
|
||||
|
||||
# Supabase 配置
|
||||
SUPABASE_URL: ${SUPABASE_URL}
|
||||
SUPABASE_ANON_KEY: ${SUPABASE_ANON_KEY}
|
||||
SUPABASE_SERVICE_KEY: ${SUPABASE_SERVICE_KEY}
|
||||
|
||||
# Embedding / RAG 配置
|
||||
EMBEDDING_MODEL: ${EMBEDDING_MODEL:-all-MiniLM-L6-v2}
|
||||
FAISS_INDEX_DIR: /app/data/faiss
|
||||
|
||||
# 文件路径配置
|
||||
UPLOAD_DIR: /app/data/uploads
|
||||
MAX_UPLOAD_SIZE: 104857600
|
||||
|
||||
# Celery 配置
|
||||
CELERY_BROKER_URL: redis://:${REDIS_PASSWORD:-}@redis:6379/1
|
||||
CELERY_RESULT_BACKEND: redis://:${REDIS_PASSWORD:-}@redis:6379/2
|
||||
volumes:
|
||||
- backend_data:/app/data
|
||||
networks:
|
||||
- filesread_network
|
||||
depends_on:
|
||||
mongodb:
|
||||
condition: service_healthy
|
||||
mysql:
|
||||
condition: service_healthy
|
||||
redis:
|
||||
condition: service_healthy
|
||||
healthcheck:
|
||||
test: ["CMD", "python", "-c", "import httpx; httpx.get('http://localhost:8000/health')"]
|
||||
interval: 30s
|
||||
timeout: 10s
|
||||
retries: 3
|
||||
start_period: 40s
|
||||
|
||||
celery_worker:
|
||||
build:
|
||||
context: ./backend
|
||||
dockerfile: Dockerfile
|
||||
container_name: filesread_celery
|
||||
restart: unless-stopped
|
||||
command: celery -A app.celery_app worker --loglevel=info --prefetch-multiplier=1
|
||||
environment:
|
||||
# Celery 配置
|
||||
CELERY_BROKER_URL: redis://:${REDIS_PASSWORD:-}@redis:6379/1
|
||||
CELERY_RESULT_BACKEND: redis://:${REDIS_PASSWORD:-}@redis:6379/2
|
||||
|
||||
# 复用后端的数据库配置
|
||||
MONGODB_URL: mongodb://${MONGO_ROOT_USER:-admin}:${MONGO_ROOT_PASSWORD:-20060825fhy}@mongodb:27017/admin
|
||||
MONGODB_DB_NAME: ${MONGODB_DB_NAME:-document_system}
|
||||
MYSQL_HOST: mysql
|
||||
MYSQL_PORT: 3306
|
||||
MYSQL_USER: root
|
||||
MYSQL_PASSWORD: ${MYSQL_PASSWORD:-123456}
|
||||
MYSQL_DATABASE: ${MYSQL_DATABASE:-document}
|
||||
REDIS_URL: redis://:${REDIS_PASSWORD:-}@redis:6379/0
|
||||
|
||||
# LLM 配置
|
||||
LLM_API_KEY: ${LLM_API_KEY}
|
||||
LLM_BASE_URL: ${LLM_BASE_URL:-https://api.deepseek.com}
|
||||
LLM_MODEL_NAME: ${LLM_MODEL_NAME:-deepseek-chat}
|
||||
|
||||
# Embedding 配置
|
||||
EMBEDDING_MODEL: ${EMBEDDING_MODEL:-all-MiniLM-L6-v2}
|
||||
FAISS_INDEX_DIR: /app/data/faiss
|
||||
volumes:
|
||||
- backend_data:/app/data
|
||||
networks:
|
||||
- filesread_network
|
||||
depends_on:
|
||||
- redis
|
||||
- mongodb
|
||||
- mysql
|
||||
|
||||
frontend:
|
||||
build:
|
||||
context: ./frontend
|
||||
dockerfile: Dockerfile
|
||||
container_name: filesread_frontend
|
||||
restart: unless-stopped
|
||||
ports:
|
||||
- "80:80"
|
||||
environment:
|
||||
VITE_APP_ID: ${VITE_APP_ID:-}
|
||||
VITE_SUPABASE_URL: ${SUPABASE_URL}
|
||||
VITE_SUPABASE_ANON_KEY: ${SUPABASE_ANON_KEY}
|
||||
VITE_BACKEND_API_URL: /api/v1
|
||||
networks:
|
||||
- filesread_network
|
||||
depends_on:
|
||||
- backend
|
||||
|
||||
networks:
|
||||
filesread_network:
|
||||
driver: bridge
|
||||
|
||||
volumes:
|
||||
mongodb_data:
|
||||
mysql_data:
|
||||
redis_data:
|
||||
backend_data:
|
||||
169
docs/architecture.drawio
Normal file
169
docs/architecture.drawio
Normal file
@@ -0,0 +1,169 @@
|
||||
<mxfile host="app.diagrams.net" modified="2026-04-16T14:00:00.000Z" agent="Claude" version="24.0.0">
|
||||
<diagram name="系统架构图" id="architecture">
|
||||
<mxGraphModel dx="1200" dy="800" grid="1" gridSize="10" guides="1" tooltips="1" connect="1" arrows="1" fold="1" page="1" pageScale="1" pageWidth="1920" pageHeight="1080" math="0" shadow="0">
|
||||
<root>
|
||||
<mxCell id="0" />
|
||||
<mxCell id="1" parent="0" />
|
||||
|
||||
<!-- 用户访问层 -->
|
||||
<mxCell id="layer1" value="用户访问层" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=16;fontStyle=1;fontColor=#1a1a2e;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="20" width="120" height="30" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="browser" value="浏览器
(Browser)" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e3f2fd;strokeColor=#1976d2;fontColor=#0d47a1;" vertex="1" parent="1">
|
||||
<mxGeometry x="860" y="60" width="120" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 前端展示层 -->
|
||||
<mxCell id="layer2" value="前端展示层" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=16;fontStyle=1;fontColor=#1a1a2e;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="140" width="120" height="30" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="frontend_box" value="" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#f3e5f5;strokeColor=#7b1fa2;strokeWidth=2;" vertex="1" parent="1">
|
||||
<mxGeometry x="200" y="180" width="1520" height="140" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="frontend_title" value="React 18 + TypeScript + Vite + shadcn/ui" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=14;fontStyle=1;fontColor=#4a148c;" vertex="1" parent="1">
|
||||
<mxGeometry x="760" y="185" width="280" height="25" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="dashboard" value="Dashboard
首页仪表盘" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ce93d8;strokeColor=#8e24aa;fontColor=#fff;" vertex="1" parent="1">
|
||||
<mxGeometry x="240" y="220" width="120" height="80" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="documents" value="Documents
文档管理" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ce93d8;strokeColor=#8e24aa;fontColor=#fff;" vertex="1" parent="1">
|
||||
<mxGeometry x="400" y="220" width="120" height="80" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="template" value="TemplateFill
智能填表" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ce93d8;strokeColor=#8e24aa;fontColor=#fff;" vertex="1" parent="1">
|
||||
<mxGeometry x="560" y="220" width="120" height="80" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="instruction" value="Instruction
指令助手" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ce93d8;strokeColor=#8e24aa;fontColor=#fff;" vertex="1" parent="1">
|
||||
<mxGeometry x="720" y="220" width="120" height="80" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="taskhistory" value="TaskHistory
任务历史" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ce93d8;strokeColor=#8e24aa;fontColor=#fff;" vertex="1" parent="1">
|
||||
<mxGeometry x="880" y="220" width="120" height="80" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="frontend_libs" value="Recharts + Lucide Icons + React Router" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=11;fontColor=#6a1b9a;" vertex="1" parent="1">
|
||||
<mxGeometry x="1040" y="250" width="280" height="25" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 连接线：浏览器到前端 -->
|
||||
<mxCell id="conn1" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#1976d2;strokeWidth=2;" edge="1" parent="1" source="browser" target="frontend_box">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 后端服务层 -->
|
||||
<mxCell id="layer3" value="后端服务层" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=16;fontStyle=1;fontColor=#1a1a2e;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="350" width="120" height="30" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="backend_box" value="" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e8f5e9;strokeColor=#388e3c;strokeWidth=2;" vertex="1" parent="1">
|
||||
<mxGeometry x="200" y="390" width="1520" height="180" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="backend_title" value="FastAPI + Uvicorn + Celery" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=14;fontStyle=1;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="395" width="200" height="25" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="upload" value="文档上传
/upload/*" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#81c784;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="240" y="430" width="140" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="ai" value="AI分析
/ai/*" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#81c784;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="420" y="430" width="140" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="rag" value="RAG检索
/rag/*" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#81c784;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="600" y="430" width="140" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="template_api" value="模板填充
/templates/*" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#81c784;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="780" y="430" width="140" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="instruction_api" value="指令解析
/instruction/*" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#81c784;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="960" y="430" width="140" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="visualization" value="可视化
/visualization/*" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#81c784;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="1140" y="430" width="140" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="celery" value="Celery
任务调度" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#a5d6a7;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="1320" y="430" width="120" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="logging" value="监控日志" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#a5d6a7;strokeColor=#2e7d32;fontColor=#1b5e20;" vertex="1" parent="1">
|
||||
<mxGeometry x="1480" y="430" width="100" height="60" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 连接线：前端到后端 -->
|
||||
<mxCell id="conn2" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#388e3c;strokeWidth=2;dashed=1;dashPattern=8 8;" edge="1" parent="1" source="frontend_box" target="backend_box">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- AI服务层 -->
|
||||
<mxCell id="layer4" value="AI服务层" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=16;fontStyle=1;fontColor=#1a1a2e;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="600" width="120" height="30" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="ai_box" value="" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#fff3e0;strokeColor=#f57c00;strokeWidth=2;" vertex="1" parent="1">
|
||||
<mxGeometry x="300" y="640" width="1320" height="120" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="llm_title" value="LLMService - 大模型服务" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=14;fontStyle=1;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="820" y="645" width="200" height="25" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="minimax" value="MiniMax-Text-01" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#ffcc80;strokeColor=#ef6c00;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="400" y="680" width="150" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="deepseek" value="DeepSeek-chat" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#ffcc80;strokeColor=#ef6c00;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="600" y="680" width="150" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="excel_ai" value="ExcelAIService" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ffe0b2;strokeColor=#f57c00;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="820" y="680" width="130" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="word_ai" value="WordAIService" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ffe0b2;strokeColor=#f57c00;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="980" y="680" width="130" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="md_ai" value="MarkdownAIService" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ffe0b2;strokeColor=#f57c00;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="1140" y="680" width="130" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="txt_ai" value="TxtAIService" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#ffe0b2;strokeColor=#f57c00;fontColor=#e65100;" vertex="1" parent="1">
|
||||
<mxGeometry x="1300" y="680" width="130" height="50" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 连接线：后端到AI -->
|
||||
<mxCell id="conn3" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#f57c00;strokeWidth=2;dashed=1;dashPattern=8 8;" edge="1" parent="1" source="backend_box" target="ai_box">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 数据存储层 -->
|
||||
<mxCell id="layer5" value="数据存储层" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=16;fontStyle=1;fontColor=#1a1a2e;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="790" width="120" height="30" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="mongodb" value="MongoDB
文档数据库

• 原始文档内容
• 元数据信息
• 文档标签
• 处理状态" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e0e0e0;strokeColor=#616161;fontColor=#212121;align=left;spacingLeft=10;" vertex="1" parent="1">
|
||||
<mxGeometry x="240" y="830" width="200" height="160" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="mysql" value="MySQL
关系数据库

• Excel表格数据
• 结构化数据
• 字段描述
• RAG索引" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e0e0e0;strokeColor=#616161;fontColor=#212121;align=left;spacingLeft=10;" vertex="1" parent="1">
|
||||
<mxGeometry x="520" y="830" width="200" height="160" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="redis" value="Redis
缓存/队列

• 会话缓存
• 任务队列
• Celery broker
• 临时数据" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e0e0e0;strokeColor=#616161;fontColor=#212121;align=left;spacingLeft=10;" vertex="1" parent="1">
|
||||
<mxGeometry x="800" y="830" width="200" height="160" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="faiss" value="FAISS
向量数据库

• 文档向量索引
• 语义相似度
• RAG检索
• sentence-transformers" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e0e0e0;strokeColor=#616161;fontColor=#212121;align=left;spacingLeft=10;" vertex="1" parent="1">
|
||||
<mxGeometry x="1080" y="830" width="240" height="160" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 连接线：AI到存储 -->
|
||||
<mxCell id="conn4" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#616161;strokeWidth=2;dashed=1;dashPattern=8 8;" edge="1" parent="1" source="ai_box" target="mongodb">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="conn5" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#616161;strokeWidth=2;dashed=1;dashPattern=8 8;" edge="1" parent="1" source="ai_box" target="mysql">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="conn6" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#616161;strokeWidth=2;dashed=1;dashPattern=8 8;" edge="1" parent="1" source="ai_box" target="redis">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="conn7" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;strokeColor=#616161;strokeWidth=2;dashed=1;dashPattern=8 8;" edge="1" parent="1" source="ai_box" target="faiss">
|
||||
<mxGeometry relative="1" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
<!-- 标注 -->
|
||||
<mxCell id="arrow1" value="HTTP/HTTPS
WebSocket" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=10;fontColor=#1976d2;" vertex="1" parent="1">
|
||||
<mxGeometry x="1020" y="130" width="80" height="30" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="arrow2" value="API调用" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=10;fontColor=#388e3c;" vertex="1" parent="1">
|
||||
<mxGeometry x="1020" y="570" width="60" height="20" as="geometry" />
|
||||
</mxCell>
|
||||
<mxCell id="arrow3" value="数据读写" style="text;html=1;strokeColor=none;fillColor=none;align=center;verticalAlign=middle;whiteSpace=wrap;rounded=0;fontSize=10;fontColor=#616161;" vertex="1" parent="1">
|
||||
<mxGeometry x="1020" y="770" width="60" height="20" as="geometry" />
|
||||
</mxCell>
|
||||
|
||||
</root>
|
||||
</mxGraphModel>
|
||||
</diagram>
|
||||
</mxfile>
|
||||
36
frontend/Dockerfile
Normal file
36
frontend/Dockerfile
Normal file
@@ -0,0 +1,36 @@
|
||||
# ============================================================
|
||||
# FilesReadSystem Frontend - React + Vite
|
||||
# 多阶段构建: Node 构建 -> Nginx 运行
|
||||
# ============================================================
|
||||
|
||||
# === 阶段1: 构建阶段 ===
|
||||
FROM node:20-alpine AS builder
|
||||
|
||||
WORKDIR /app
|
||||
|
||||
# 复制 package 文件和锁文件
|
||||
COPY package.json pnpm-lock.yaml* ./
|
||||
|
||||
# 安装 pnpm 并安装依赖
|
||||
RUN npm install -g pnpm && \
|
||||
pnpm install --frozen-lockfile
|
||||
|
||||
# 复制源码
|
||||
COPY . .
|
||||
|
||||
# 构建生产版本
|
||||
RUN pnpm build
|
||||
|
||||
# === 阶段2: 运行阶段 ===
|
||||
FROM nginx:alpine
|
||||
|
||||
# 复制 nginx 配置
|
||||
COPY nginx.conf /etc/nginx/conf.d/default.conf
|
||||
|
||||
# 复制构建产物
|
||||
COPY --from=builder /app/dist /usr/share/nginx/html
|
||||
|
||||
# 暴露端口
|
||||
EXPOSE 80
|
||||
|
||||
CMD ["nginx", "-g", "daemon off;"]
|
||||
47
frontend/nginx.conf
Normal file
47
frontend/nginx.conf
Normal file
@@ -0,0 +1,47 @@
|
||||
# ============================================================
|
||||
# FilesReadSystem Nginx 配置
|
||||
# 反向代理 API 请求到后端
|
||||
# ============================================================
|
||||
|
||||
server {
|
||||
listen 80;
|
||||
server_name localhost;
|
||||
|
||||
# 前端静态文件
|
||||
root /usr/share/nginx/html;
|
||||
index index.html;
|
||||
|
||||
# SPA 支持 - 所有请求都尝试返回 index.html
|
||||
location / {
|
||||
try_files $uri $uri/ /index.html;
|
||||
}
|
||||
|
||||
# 静态资源缓存
|
||||
location ~* \.(js|css|png|jpg|jpeg|gif|ico|svg|woff|woff2|ttf|eot)$ {
|
||||
expires 1y;
|
||||
add_header Cache-Control "public, immutable";
|
||||
}
|
||||
|
||||
# API 反向代理到后端
|
||||
location /api/ {
|
||||
proxy_pass http://backend:8000/api/;
|
||||
proxy_http_version 1.1;
|
||||
proxy_set_header Host $host;
|
||||
proxy_set_header X-Real-IP $remote_addr;
|
||||
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
|
||||
proxy_set_header X-Forwarded-Proto $scheme;
|
||||
|
||||
# 超时设置
|
||||
proxy_connect_timeout 60s;
|
||||
proxy_send_timeout 60s;
|
||||
proxy_read_timeout 60s;
|
||||
}
|
||||
|
||||
# 文件上传代理
|
||||
location /uploads/ {
|
||||
proxy_pass http://backend:8000/uploads/;
|
||||
proxy_set_header Host $host;
|
||||
proxy_set_header X-Real-IP $remote_addr;
|
||||
client_max_body_size 100M;
|
||||
}
|
||||
}
|
||||
@@ -8,7 +8,8 @@ import {
|
||||
Menu,
|
||||
ChevronRight,
|
||||
Sparkles,
|
||||
Clock
|
||||
Clock,
|
||||
FileDown
|
||||
} from 'lucide-react';
|
||||
import { Button } from '@/components/ui/button';
|
||||
import { cn } from '@/lib/utils';
|
||||
@@ -19,6 +20,7 @@ const navItems = [
|
||||
{ name: '文档中心', path: '/documents', icon: FileText },
|
||||
{ name: '智能填表', path: '/form-fill', icon: TableProperties },
|
||||
{ name: '智能助手', path: '/assistant', icon: MessageSquareCode },
|
||||
{ name: '文档转PDF', path: '/pdf-converter', icon: FileDown },
|
||||
{ name: '任务历史', path: '/task-history', icon: Clock },
|
||||
];
|
||||
|
||||
@@ -32,7 +34,7 @@ const MainLayout: React.FC = () => {
|
||||
<FileText size={24} />
|
||||
</div>
|
||||
<div className="flex flex-col">
|
||||
<span className="font-bold text-lg tracking-tight text-sidebar-foreground">智联文档</span>
|
||||
<span className="font-bold text-lg tracking-tight text-sidebar-foreground">表易智融</span>
|
||||
<span className="text-xs text-muted-foreground">多源数据融合平台</span>
|
||||
</div>
|
||||
</div>
|
||||
@@ -66,7 +68,7 @@ const MainLayout: React.FC = () => {
|
||||
<Sparkles size={20} className="text-primary" />
|
||||
</div>
|
||||
<div className="flex flex-col overflow-hidden">
|
||||
<span className="font-semibold text-sm truncate">智联文档</span>
|
||||
<span className="font-semibold text-sm truncate">表易智融</span>
|
||||
<span className="text-[10px] uppercase tracking-wider text-muted-foreground">多源数据融合</span>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
@@ -250,6 +250,98 @@ export interface AIExcelAnalyzeResult {
|
||||
error?: string;
|
||||
}
|
||||
|
||||
// ==================== Word/TXT AI 分析类型 ====================
|
||||
|
||||
export type WordAnalysisType = 'structured' | 'charts';
|
||||
export type TxtAnalysisType = 'structured' | 'charts';
|
||||
|
||||
export interface WordAIStructuredResult {
|
||||
success: boolean;
|
||||
result?: {
|
||||
success?: boolean;
|
||||
type?: string;
|
||||
headers?: string[];
|
||||
rows?: string[][];
|
||||
key_values?: Record<string, string>;
|
||||
list_items?: string[];
|
||||
summary?: string;
|
||||
error?: string;
|
||||
};
|
||||
error?: string;
|
||||
}
|
||||
|
||||
export interface WordAIChartsResult {
|
||||
success: boolean;
|
||||
result?: {
|
||||
success?: boolean;
|
||||
charts?: {
|
||||
histograms?: Array<any>;
|
||||
bar_charts?: Array<any>;
|
||||
box_plots?: Array<any>;
|
||||
correlation?: any;
|
||||
};
|
||||
statistics?: {
|
||||
numeric?: Record<string, any>;
|
||||
categorical?: Record<string, any>;
|
||||
};
|
||||
distributions?: Record<string, any>;
|
||||
row_count?: number;
|
||||
column_count?: number;
|
||||
error?: string;
|
||||
};
|
||||
error?: string;
|
||||
}
|
||||
|
||||
export interface TxtAIStructuredResult {
|
||||
success: boolean;
|
||||
result?: {
|
||||
success?: boolean;
|
||||
type?: string;
|
||||
tables?: Array<{
|
||||
headers?: string[];
|
||||
rows?: string[][];
|
||||
}>;
|
||||
key_values?: Record<string, string>;
|
||||
list_items?: string[];
|
||||
summary?: string;
|
||||
error?: string;
|
||||
};
|
||||
error?: string;
|
||||
}
|
||||
|
||||
export interface TxtAIChartsResult {
|
||||
success: boolean;
|
||||
result?: {
|
||||
success?: boolean;
|
||||
charts?: {
|
||||
histograms?: Array<any>;
|
||||
bar_charts?: Array<any>;
|
||||
box_plots?: Array<any>;
|
||||
correlation?: any;
|
||||
};
|
||||
statistics?: {
|
||||
numeric?: Record<string, any>;
|
||||
categorical?: Record<string, any>;
|
||||
};
|
||||
distributions?: Record<string, any>;
|
||||
row_count?: number;
|
||||
column_count?: number;
|
||||
key_statistics?: Array<{
|
||||
name?: string;
|
||||
value?: string;
|
||||
trend?: string;
|
||||
description?: string;
|
||||
}>;
|
||||
chart_suggestions?: Array<{
|
||||
chart_type?: string;
|
||||
title?: string;
|
||||
data_source?: string;
|
||||
}>;
|
||||
error?: string;
|
||||
};
|
||||
error?: string;
|
||||
}
|
||||
|
||||
// ==================== API 封装 ====================
|
||||
|
||||
export const backendApi = {
|
||||
@@ -781,7 +873,8 @@ export const backendApi = {
|
||||
async exportFilledTemplate(
|
||||
templateId: string,
|
||||
filledData: Record<string, any>,
|
||||
format: 'xlsx' | 'docx' = 'xlsx'
|
||||
format: 'xlsx' | 'docx' = 'xlsx',
|
||||
filledFilePath?: string
|
||||
): Promise<Blob> {
|
||||
const url = `${BACKEND_BASE_URL}/templates/export`;
|
||||
|
||||
@@ -793,6 +886,7 @@ export const backendApi = {
|
||||
template_id: templateId,
|
||||
filled_data: filledData,
|
||||
format,
|
||||
...(filledFilePath && { filled_file_path: filledFilePath }),
|
||||
}),
|
||||
});
|
||||
|
||||
@@ -964,6 +1058,215 @@ export const backendApi = {
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
// ==================== 智能指令 API ====================
|
||||
|
||||
/**
|
||||
* 智能对话（支持多轮对话的指令执行）
|
||||
*/
|
||||
async instructionChat(
|
||||
instruction: string,
|
||||
docIds?: string[],
|
||||
context?: Record<string, any>
|
||||
): Promise<{
|
||||
success: boolean;
|
||||
intent: string;
|
||||
result: Record<string, any>;
|
||||
message: string;
|
||||
hint?: string;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/instruction/chat`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({ instruction, doc_ids: docIds, context }),
|
||||
});
|
||||
|
||||
if (!response.ok) {
|
||||
const error = await response.json();
|
||||
throw new Error(error.detail || '对话处理失败');
|
||||
}
|
||||
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('对话处理失败:', error);
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
/**
|
||||
* 获取支持的指令类型列表
|
||||
*/
|
||||
async getSupportedIntents(): Promise<{
|
||||
intents: Array<{
|
||||
intent: string;
|
||||
name: string;
|
||||
examples: string[];
|
||||
params: string[];
|
||||
}>;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/instruction/intents`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url);
|
||||
if (!response.ok) throw new Error('获取指令列表失败');
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('获取指令列表失败:', error);
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
/**
|
||||
* 执行指令（同步模式）
|
||||
*/
|
||||
async executeInstruction(
|
||||
instruction: string,
|
||||
docIds?: string[],
|
||||
context?: Record<string, any>
|
||||
): Promise<{
|
||||
success: boolean;
|
||||
intent: string;
|
||||
result: Record<string, any>;
|
||||
message: string;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/instruction/execute`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({ instruction, doc_ids: docIds, context }),
|
||||
});
|
||||
|
||||
if (!response.ok) {
|
||||
const error = await response.json();
|
||||
throw new Error(error.detail || '指令执行失败');
|
||||
}
|
||||
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('指令执行失败:', error);
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
// ==================== PDF 转换 API ====================
|
||||
|
||||
/**
|
||||
* 将文件转换为 PDF
|
||||
*/
|
||||
/**
|
||||
* PDF转换并直接下载（使用XHR，支持IDM拦截）
|
||||
*/
|
||||
async convertAndDownloadPdf(file: File): Promise<void> {
|
||||
return new Promise((resolve, reject) => {
|
||||
const xhr = new XMLHttpRequest();
|
||||
xhr.open('POST', `${BACKEND_BASE_URL}/pdf/convert`);
|
||||
|
||||
xhr.onload = function() {
|
||||
if (xhr.status >= 200 && xhr.status < 300) {
|
||||
// 创建 blob 并触发下载
|
||||
const blob = xhr.response;
|
||||
const url = URL.createObjectURL(blob);
|
||||
const a = document.createElement('a');
|
||||
a.href = url;
|
||||
a.download = `${file.name.replace(/\.[^.]+$/, '')}.pdf`;
|
||||
document.body.appendChild(a);
|
||||
a.click();
|
||||
document.body.removeChild(a);
|
||||
URL.revokeObjectURL(url);
|
||||
resolve();
|
||||
} else {
|
||||
reject(new Error(`转换失败: ${xhr.status}`));
|
||||
}
|
||||
};
|
||||
|
||||
xhr.onerror = function() {
|
||||
reject(new Error('网络错误'));
|
||||
};
|
||||
|
||||
const formData = new FormData();
|
||||
formData.append('file', file);
|
||||
xhr.responseType = 'blob';
|
||||
xhr.send(formData);
|
||||
});
|
||||
},
|
||||
|
||||
/**
|
||||
* PDF转换（返回Blob）
|
||||
*/
|
||||
async convertToPdf(file: File): Promise<Blob> {
|
||||
return new Promise((resolve, reject) => {
|
||||
const xhr = new XMLHttpRequest();
|
||||
xhr.open('POST', `${BACKEND_BASE_URL}/pdf/convert`);
|
||||
|
||||
xhr.onload = function() {
|
||||
if (xhr.status >= 200 && xhr.status < 300) {
|
||||
resolve(xhr.response);
|
||||
} else {
|
||||
reject(new Error(`转换失败: ${xhr.status}`));
|
||||
}
|
||||
};
|
||||
|
||||
xhr.onerror = function() {
|
||||
reject(new Error('网络错误'));
|
||||
};
|
||||
|
||||
const formData = new FormData();
|
||||
formData.append('file', file);
|
||||
xhr.responseType = 'blob';
|
||||
xhr.send(formData);
|
||||
});
|
||||
},
|
||||
|
||||
/**
|
||||
* 批量将文件转换为 PDF
|
||||
*/
|
||||
async batchConvertToPdf(files: File[]): Promise<Blob> {
|
||||
const formData = new FormData();
|
||||
files.forEach(file => formData.append('files', file));
|
||||
|
||||
const url = `${BACKEND_BASE_URL}/pdf/convert/batch`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
method: 'POST',
|
||||
body: formData,
|
||||
});
|
||||
|
||||
if (!response.ok) {
|
||||
const error = await response.json();
|
||||
throw new Error(error.detail || '批量PDF转换失败');
|
||||
}
|
||||
|
||||
return await response.blob();
|
||||
} catch (error) {
|
||||
console.error('批量PDF转换失败:', error);
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
/**
|
||||
* 获取支持的 PDF 转换格式
|
||||
*/
|
||||
async getPdfSupportedFormats(): Promise<{
|
||||
success: boolean;
|
||||
formats: string[];
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/pdf/formats`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url);
|
||||
if (!response.ok) throw new Error('获取支持的格式失败');
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('获取支持的格式失败:', error);
|
||||
return { success: false, formats: ['docx', 'xlsx', 'txt', 'md'] };
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
// ==================== AI 分析 API ====================
|
||||
@@ -998,11 +1301,19 @@ export const aiApi = {
|
||||
* 上传并使用 AI 分析 Excel 文件
|
||||
*/
|
||||
async analyzeExcel(
|
||||
file: File,
|
||||
options: AIAnalyzeOptions = {}
|
||||
file: File | null,
|
||||
options: AIAnalyzeOptions = {},
|
||||
docId: string | null = null
|
||||
): Promise<AIExcelAnalyzeResult> {
|
||||
const formData = new FormData();
|
||||
formData.append('file', file);
|
||||
|
||||
if (docId) {
|
||||
formData.append('doc_id', docId);
|
||||
} else if (file) {
|
||||
formData.append('file', file);
|
||||
} else {
|
||||
throw new Error('必须提供文件或文档ID');
|
||||
}
|
||||
|
||||
const params = new URLSearchParams();
|
||||
if (options.userPrompt) {
|
||||
@@ -1079,7 +1390,9 @@ export const aiApi = {
|
||||
try {
|
||||
const response = await fetch(url);
|
||||
if (!response.ok) throw new Error('获取分析类型失败');
|
||||
return await response.json();
|
||||
const data = await response.json();
|
||||
// 转换后端返回格式 {excel_types: [], markdown_types: []} 为前端期望的 {types: []}
|
||||
return { types: data.excel_types || [] };
|
||||
} catch (error) {
|
||||
console.error('获取分析类型失败:', error);
|
||||
throw error;
|
||||
@@ -1090,15 +1403,21 @@ export const aiApi = {
|
||||
* 上传并使用 AI 分析 Markdown 文件
|
||||
*/
|
||||
async analyzeMarkdown(
|
||||
file: File,
|
||||
file: File | null,
|
||||
options: {
|
||||
docId?: string;
|
||||
analysisType?: MarkdownAnalysisType;
|
||||
userPrompt?: string;
|
||||
sectionNumber?: string;
|
||||
} = {}
|
||||
): Promise<AIMarkdownAnalyzeResult> {
|
||||
const formData = new FormData();
|
||||
formData.append('file', file);
|
||||
if (file) {
|
||||
formData.append('file', file);
|
||||
}
|
||||
if (options.docId) {
|
||||
formData.append('doc_id', options.docId);
|
||||
}
|
||||
|
||||
const params = new URLSearchParams();
|
||||
if (options.analysisType) {
|
||||
@@ -1240,28 +1559,31 @@ export const aiApi = {
|
||||
},
|
||||
|
||||
/**
|
||||
* 上传并使用 AI 分析 TXT 文本文件，提取结构化数据
|
||||
* 上传并使用 AI 分析 TXT 文本文件，提取结构化数据或生成图表
|
||||
*/
|
||||
async analyzeTxt(
|
||||
file: File
|
||||
file: File | null,
|
||||
docId: string | null = null,
|
||||
analysisType: TxtAnalysisType = 'structured'
|
||||
): Promise<{
|
||||
success: boolean;
|
||||
filename?: string;
|
||||
structured_data?: {
|
||||
table?: {
|
||||
columns?: string[];
|
||||
rows?: string[][];
|
||||
};
|
||||
summary?: string;
|
||||
key_value_pairs?: Array<{ key: string; value: string }>;
|
||||
numeric_data?: Array<{ name: string; value: number; unit?: string }>;
|
||||
};
|
||||
analysis_type?: string;
|
||||
result?: any;
|
||||
error?: string;
|
||||
}> {
|
||||
const formData = new FormData();
|
||||
formData.append('file', file);
|
||||
if (file) {
|
||||
formData.append('file', file);
|
||||
}
|
||||
if (docId) {
|
||||
formData.append('doc_id', docId);
|
||||
}
|
||||
|
||||
const url = `${BACKEND_BASE_URL}/ai/analyze/txt`;
|
||||
const params = new URLSearchParams();
|
||||
params.append('analysis_type', analysisType);
|
||||
|
||||
const url = `${BACKEND_BASE_URL}/ai/analyze/txt?${params.toString()}`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
@@ -1383,28 +1705,35 @@ export const aiApi = {
|
||||
// ==================== Word AI 解析 ====================
|
||||
|
||||
/**
|
||||
* 使用 AI 解析 Word 文档，提取结构化数据
|
||||
* 使用 AI 解析 Word 文档，提取结构化数据或生成图表
|
||||
*/
|
||||
async analyzeWordWithAI(
|
||||
file: File,
|
||||
userHint: string = ''
|
||||
file: File | null,
|
||||
docId: string | null = null,
|
||||
userHint: string = '',
|
||||
analysisType: WordAnalysisType = 'structured'
|
||||
): Promise<{
|
||||
success: boolean;
|
||||
type?: string;
|
||||
headers?: string[];
|
||||
rows?: string[][];
|
||||
key_values?: Record<string, string>;
|
||||
list_items?: string[];
|
||||
summary?: string;
|
||||
filename?: string;
|
||||
analysis_type?: string;
|
||||
result?: any;
|
||||
error?: string;
|
||||
}> {
|
||||
const formData = new FormData();
|
||||
formData.append('file', file);
|
||||
if (file) {
|
||||
formData.append('file', file);
|
||||
}
|
||||
if (docId) {
|
||||
formData.append('doc_id', docId);
|
||||
}
|
||||
if (userHint) {
|
||||
formData.append('user_hint', userHint);
|
||||
}
|
||||
|
||||
const url = `${BACKEND_BASE_URL}/ai/analyze/word`;
|
||||
const params = new URLSearchParams();
|
||||
params.append('analysis_type', analysisType);
|
||||
|
||||
const url = `${BACKEND_BASE_URL}/ai/analyze/word?${params.toString()}`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
@@ -1459,4 +1788,137 @@ export const aiApi = {
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
// ==================== 智能指令 ====================
|
||||
|
||||
/**
|
||||
* 识别自然语言指令的意图
|
||||
*/
|
||||
async recognizeIntent(
|
||||
instruction: string,
|
||||
docIds?: string[]
|
||||
): Promise<{
|
||||
success: boolean;
|
||||
intent: string;
|
||||
params: Record<string, any>;
|
||||
message: string;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/instruction/recognize`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({ instruction, doc_ids: docIds }),
|
||||
});
|
||||
|
||||
if (!response.ok) {
|
||||
const error = await response.json();
|
||||
throw new Error(error.detail || '意图识别失败');
|
||||
}
|
||||
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('意图识别失败:', error);
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
/**
|
||||
* 执行自然语言指令
|
||||
*/
|
||||
async executeInstruction(
|
||||
instruction: string,
|
||||
docIds?: string[],
|
||||
context?: Record<string, any>
|
||||
): Promise<{
|
||||
success: boolean;
|
||||
intent: string;
|
||||
result: Record<string, any>;
|
||||
message: string;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/instruction/execute`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, {
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({ instruction, doc_ids: docIds, context }),
|
||||
});
|
||||
|
||||
if (!response.ok) {
|
||||
const error = await response.json();
|
||||
throw new Error(error.detail || '指令执行失败');
|
||||
}
|
||||
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('指令执行失败:', error);
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
// ==================== 对话历史 API ====================
|
||||
|
||||
/**
|
||||
* 获取对话历史
|
||||
*/
|
||||
async getConversationHistory(conversationId: string, limit: number = 20): Promise<{
|
||||
success: boolean;
|
||||
messages: Array<{
|
||||
role: string;
|
||||
content: string;
|
||||
intent?: string;
|
||||
created_at: string;
|
||||
}>;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/conversation/${conversationId}/history?limit=${limit}`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url);
|
||||
if (!response.ok) throw new Error('获取对话历史失败');
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('获取对话历史失败:', error);
|
||||
return { success: false, messages: [] };
|
||||
}
|
||||
},
|
||||
|
||||
/**
|
||||
* 删除对话历史
|
||||
*/
|
||||
async deleteConversation(conversationId: string): Promise<{
|
||||
success: boolean;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/conversation/${conversationId}`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url, { method: 'DELETE' });
|
||||
if (!response.ok) throw new Error('删除对话历史失败');
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('删除对话历史失败:', error);
|
||||
return { success: false };
|
||||
}
|
||||
},
|
||||
|
||||
/**
|
||||
* 获取会话列表
|
||||
*/
|
||||
async listConversations(limit: number = 50): Promise<{
|
||||
success: boolean;
|
||||
conversations: Array<any>;
|
||||
}> {
|
||||
const url = `${BACKEND_BASE_URL}/conversation/all?limit=${limit}`;
|
||||
|
||||
try {
|
||||
const response = await fetch(url);
|
||||
if (!response.ok) throw new Error('获取会话列表失败');
|
||||
return await response.json();
|
||||
} catch (error) {
|
||||
console.error('获取会话列表失败:', error);
|
||||
return { success: false, conversations: [] };
|
||||
}
|
||||
},
|
||||
|
||||
};
|
||||
|
||||
@@ -41,7 +41,7 @@ const Assistant: React.FC = () => {
|
||||
{
|
||||
id: '1',
|
||||
role: 'assistant',
|
||||
content: '您好！我是智联文档 AI 助手。您可以告诉我您想对文档进行的操作，例如：\n- "帮我列出最近上传的所有 docx 文档"\n- "从 2026 财报文档中提取出关键的利润数据"\n- "帮我创建一个汇总各部门报销单的填表任务"\n\n请问有什么我可以帮您的？',
|
||||
content: '您好！我是表易智融 AI 助手。您可以告诉我您想对文档进行的操作，例如：\n- "帮我列出最近上传的所有 docx 文档"\n- "从 2026 财报文档中提取出关键的利润数据"\n- "帮我创建一个汇总各部门报销单的填表任务"\n\n请问有什么我可以帮您的？',
|
||||
created_at: new Date().toISOString()
|
||||
}
|
||||
]);
|
||||
|
||||
@@ -15,12 +15,14 @@ import {
|
||||
Sparkles,
|
||||
Database,
|
||||
FileSpreadsheet,
|
||||
RefreshCcw
|
||||
RefreshCcw,
|
||||
Trash2
|
||||
} from 'lucide-react';
|
||||
import { backendApi } from '@/db/backend-api';
|
||||
import { formatDistanceToNow } from 'date-fns';
|
||||
import { zhCN } from 'date-fns/locale';
|
||||
import { cn } from '@/lib/utils';
|
||||
import { toast } from 'sonner';
|
||||
|
||||
type DocumentItem = {
|
||||
doc_id: string;
|
||||
@@ -87,7 +89,7 @@ const Dashboard: React.FC = () => {
|
||||
<section className="flex flex-col md:flex-row md:items-center justify-between gap-4">
|
||||
<div className="space-y-1">
|
||||
<h1 className="text-3xl font-extrabold tracking-tight">
|
||||
欢迎使用 <span className="text-primary">智联文档</span> 系统 👋
|
||||
欢迎使用 <span className="text-primary">表易智融</span> 系统 👋
|
||||
</h1>
|
||||
<p className="text-muted-foreground">基于大语言模型的文档理解与多源数据融合系统</p>
|
||||
</div>
|
||||
@@ -108,7 +110,7 @@ const Dashboard: React.FC = () => {
|
||||
<div className="grid grid-cols-1 md:grid-cols-3 gap-6">
|
||||
{[
|
||||
{ label: '已上传文档', value: stats.docs, icon: FileText, color: 'bg-blue-500', trend: '非结构化文档', link: '/documents' },
|
||||
{ label: 'Excel 文件', value: stats.excelFiles, icon: FileSpreadsheet, color: 'bg-emerald-500', trend: '结构化数据', link: '/excel-parse' },
|
||||
{ label: 'Excel 文件', value: stats.excelFiles, icon: FileSpreadsheet, color: 'bg-emerald-500', trend: '结构化数据', link: '/documents' },
|
||||
{ label: '填表任务', value: stats.tasks, icon: TableProperties, color: 'bg-indigo-500', trend: '待实现', link: '/form-fill' }
|
||||
].map((stat, i) => (
|
||||
<Card key={i} className="border-none shadow-md overflow-hidden group hover:shadow-xl transition-all duration-300">
|
||||
@@ -164,8 +166,30 @@ const Dashboard: React.FC = () => {
|
||||
{doc.doc_type.toUpperCase()} • {formatDistanceToNow(new Date(doc.created_at), { addSuffix: true, locale: zhCN })}
|
||||
</p>
|
||||
</div>
|
||||
<div className="px-2 py-1 rounded-full text-[10px] font-bold uppercase tracking-wider bg-muted">
|
||||
{doc.doc_type}
|
||||
<div className="flex items-center gap-2">
|
||||
<div className="px-2 py-1 rounded-full text-[10px] font-bold uppercase tracking-wider bg-muted">
|
||||
{doc.doc_type}
|
||||
</div>
|
||||
<Button
|
||||
variant="ghost"
|
||||
size="icon"
|
||||
className="opacity-0 group-hover:opacity-100 text-destructive hover:bg-destructive/10 transition-opacity"
|
||||
onClick={async (e) => {
|
||||
e.stopPropagation();
|
||||
if (!confirm(`确定要删除 "${doc.original_filename}" 吗？`)) return;
|
||||
try {
|
||||
const result = await backendApi.deleteDocument(doc.doc_id);
|
||||
if (result.success) {
|
||||
setRecentDocs(prev => prev.filter(d => d.doc_id !== doc.doc_id));
|
||||
toast.success('文档已删除');
|
||||
}
|
||||
} catch (err: any) {
|
||||
toast.error(err.message || '删除失败');
|
||||
}
|
||||
}}
|
||||
>
|
||||
<Trash2 size={16} />
|
||||
</Button>
|
||||
</div>
|
||||
</div>
|
||||
))}
|
||||
@@ -197,7 +221,7 @@ const Dashboard: React.FC = () => {
|
||||
<div className="grid grid-cols-1 sm:grid-cols-2 gap-4">
|
||||
{[
|
||||
{ title: '上传文档', desc: '支持 docx/md/txt', icon: FileText, link: '/documents', color: 'bg-blue-500' },
|
||||
{ title: '解析 Excel', desc: '上传并分析数据', icon: FileSpreadsheet, link: '/excel-parse', color: 'bg-emerald-500' },
|
||||
{ title: '解析 Excel', desc: '上传并分析数据', icon: FileSpreadsheet, link: '/documents', color: 'bg-emerald-500' },
|
||||
{ title: '智能填表', desc: '自动填写表格模板', icon: TableProperties, link: '/form-fill', color: 'bg-indigo-500' },
|
||||
{ title: 'AI 助手', desc: '自然语言交互', icon: MessageSquareCode, link: '/assistant', color: 'bg-amber-500' }
|
||||
].map((item, i) => (
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
@@ -1,22 +1,10 @@
|
||||
import React, { useState, useRef, useEffect } from 'react';
|
||||
import {
|
||||
Send,
|
||||
Bot,
|
||||
User,
|
||||
Sparkles,
|
||||
Trash2,
|
||||
RefreshCcw,
|
||||
FileText,
|
||||
TableProperties,
|
||||
ChevronRight,
|
||||
ArrowRight,
|
||||
Loader2
|
||||
} from 'lucide-react';
|
||||
import { Send, Bot, User, Sparkles, Trash2, FileText, TableProperties, ArrowRight, Search, MessageSquare } from 'lucide-react';
|
||||
import { Button } from '@/components/ui/button';
|
||||
import { Input } from '@/components/ui/input';
|
||||
import { Card, CardContent, CardHeader, CardTitle } from '@/components/ui/card';
|
||||
import { ScrollArea } from '@/components/ui/scroll-area';
|
||||
import { Badge } from '@/components/ui/badge';
|
||||
import { Markdown } from '@/components/ui/markdown';
|
||||
import { backendApi } from '@/db/backend-api';
|
||||
import { toast } from 'sonner';
|
||||
import { cn } from '@/lib/utils';
|
||||
@@ -26,14 +14,30 @@ type ChatMessage = {
|
||||
role: 'user' | 'assistant';
|
||||
content: string;
|
||||
created_at: string;
|
||||
intent?: string;
|
||||
result?: any;
|
||||
};
|
||||
|
||||
const InstructionChat: React.FC = () => {
|
||||
const [messages, setMessages] = useState<ChatMessage[]>([]);
|
||||
const [input, setInput] = useState('');
|
||||
const [loading, setLoading] = useState(false);
|
||||
const [currentDocIds, setCurrentDocIds] = useState<string[]>([]);
|
||||
const [conversationId, setConversationId] = useState<string>('');
|
||||
const scrollAreaRef = useRef<HTMLDivElement>(null);
|
||||
|
||||
// 初始化会话ID
|
||||
useEffect(() => {
|
||||
const storedId = localStorage.getItem('chat_conversation_id');
|
||||
if (storedId) {
|
||||
setConversationId(storedId);
|
||||
} else {
|
||||
const newId = `conv_${Date.now()}_${Math.random().toString(36).substring(7)}`;
|
||||
setConversationId(newId);
|
||||
localStorage.setItem('chat_conversation_id', newId);
|
||||
}
|
||||
}, []);
|
||||
|
||||
useEffect(() => {
|
||||
// Initial welcome message
|
||||
if (messages.length === 0) {
|
||||
@@ -41,29 +45,49 @@ const InstructionChat: React.FC = () => {
|
||||
{
|
||||
id: 'welcome',
|
||||
role: 'assistant',
|
||||
content: `您好！我是智联文档 AI 助手。
|
||||
content: `您好！我是表易智融 AI 助手。
|
||||
|
||||
我可以帮您完成以下操作：
|
||||
**📄 文档智能操作**
|
||||
- "提取文档中的医院数量和床位数"
|
||||
- "帮我找出所有机构的名称"
|
||||
|
||||
📄 **文档管理**
|
||||
- "帮我列出最近上传的所有文档"
|
||||
- "删除三天前的 docx 文档"
|
||||
**📊 数据填表**
|
||||
- "根据这些数据填表"
|
||||
- "将提取的信息填写到Excel模板"
|
||||
|
||||
📊 **Excel 分析**
|
||||
- "分析一下最近上传的 Excel 文件"
|
||||
- "帮我统计销售报表中的数据"
|
||||
**📝 内容处理**
|
||||
- "总结一下这份文档"
|
||||
- "对比这两个文档的差异"
|
||||
|
||||
📝 **智能填表**
|
||||
- "根据员工信息表创建一个考勤汇总表"
|
||||
- "用财务文档填充报销模板"
|
||||
**🔍 智能问答**
|
||||
- "文档里说了些什么？"
|
||||
- "有多少家医院？"
|
||||
|
||||
请告诉我您想做什么？`,
|
||||
created_at: new Date().toISOString()
|
||||
}
|
||||
]);
|
||||
|
||||
// 获取已上传的文档ID列表
|
||||
loadDocuments();
|
||||
}
|
||||
}, []);
|
||||
|
||||
const loadDocuments = async () => {
|
||||
try {
|
||||
const result = await backendApi.getDocuments(undefined, 50);
|
||||
if (result.success && result.documents) {
|
||||
const docIds = result.documents.map((d: any) => d.doc_id);
|
||||
setCurrentDocIds(docIds);
|
||||
if (docIds.length > 0) {
|
||||
console.log(`已加载 ${docIds.length} 个文档`);
|
||||
}
|
||||
}
|
||||
} catch (err) {
|
||||
console.error('获取文档列表失败:', err);
|
||||
}
|
||||
};
|
||||
|
||||
useEffect(() => {
|
||||
// Scroll to bottom
|
||||
if (scrollAreaRef.current) {
|
||||
@@ -89,95 +113,157 @@ const InstructionChat: React.FC = () => {
|
||||
setLoading(true);
|
||||
|
||||
try {
|
||||
// TODO: 后端对话接口，暂用模拟响应
|
||||
await new Promise(resolve => setTimeout(resolve, 1500));
|
||||
// 使用真实的智能指令 API
|
||||
const response = await backendApi.instructionChat(
|
||||
input.trim(),
|
||||
currentDocIds.length > 0 ? currentDocIds : undefined,
|
||||
{ conversation_id: conversationId }
|
||||
);
|
||||
|
||||
// 简单的命令解析演示
|
||||
const userInput = userMessage.content.toLowerCase();
|
||||
let response = '';
|
||||
// 根据意图类型生成友好响应
|
||||
let responseContent = '';
|
||||
const resultData = response.result;
|
||||
|
||||
if (userInput.includes('列出') || userInput.includes('列表')) {
|
||||
const result = await backendApi.getDocuments(undefined, 10);
|
||||
if (result.success && result.documents && result.documents.length > 0) {
|
||||
response = `已为您找到 ${result.documents.length} 个文档：\n\n`;
|
||||
result.documents.slice(0, 5).forEach((doc: any, idx: number) => {
|
||||
response += `${idx + 1}. **${doc.original_filename}** (${doc.doc_type.toUpperCase()})\n`;
|
||||
response += ` - 大小: ${(doc.file_size / 1024).toFixed(1)} KB\n`;
|
||||
response += ` - 时间: ${new Date(doc.created_at).toLocaleDateString()}\n\n`;
|
||||
});
|
||||
if (result.documents.length > 5) {
|
||||
response += `...还有 ${result.documents.length - 5} 个文档`;
|
||||
switch (response.intent) {
|
||||
case 'extract':
|
||||
// 信息提取结果
|
||||
const extracted = resultData?.extracted_data || {};
|
||||
const keys = Object.keys(extracted);
|
||||
if (keys.length > 0) {
|
||||
responseContent = `✅ 已提取到 ${keys.length} 个字段的数据：\n\n`;
|
||||
for (const [key, value] of Object.entries(extracted)) {
|
||||
const values = Array.isArray(value) ? value : [value];
|
||||
const displayValues = values.length > 10 ? values.slice(0, 10).join(', ') + ` ...（共${values.length}条）` : values.join(', ');
|
||||
responseContent += `**${key}**: ${displayValues}\n`;
|
||||
}
|
||||
responseContent += `\n💡 可直接使用以上数据，或说"填入表格"继续填表操作。`;
|
||||
} else {
|
||||
responseContent = resultData?.message || '未能从文档中提取到相关数据。请尝试更明确的字段名称。';
|
||||
}
|
||||
} else {
|
||||
response = '暂未找到已上传的文档，您可以先上传一些文档试试。';
|
||||
}
|
||||
} else if (userInput.includes('分析') || userInput.includes('excel') || userInput.includes('报表')) {
|
||||
response = `好的，我可以帮您分析 Excel 文件。
|
||||
break;
|
||||
|
||||
请告诉我：
|
||||
1. 您想分析哪个 Excel 文件？
|
||||
2. 需要什么样的分析？（数据摘要/统计分析/图表生成）
|
||||
case 'fill_table':
|
||||
// 填表结果
|
||||
const filled = resultData?.result?.filled_data || {};
|
||||
const filledKeys = Object.keys(filled);
|
||||
if (filledKeys.length > 0) {
|
||||
responseContent = `✅ 填表完成！成功填写 ${filledKeys.length} 个字段：\n\n`;
|
||||
for (const [key, value] of Object.entries(filled)) {
|
||||
const values = Array.isArray(value) ? value : [value];
|
||||
const displayValues = values.length > 10 ? values.slice(0, 10).join(', ') + ` ...（共${values.length}条）` : values.join(', ');
|
||||
responseContent += `**${key}**: ${displayValues}\n`;
|
||||
}
|
||||
responseContent += `\n📋 请到【智能填表】页面查看或导出结果。`;
|
||||
} else {
|
||||
responseContent = resultData?.message || '填表未能提取到数据。请检查模板表头和数据源内容。';
|
||||
}
|
||||
break;
|
||||
|
||||
或者您可以直接告诉我您想从数据中了解什么，我来为您生成分析。`;
|
||||
} else if (userInput.includes('填表') || userInput.includes('模板')) {
|
||||
response = `好的，要进行智能填表，我需要：
|
||||
case 'summarize':
|
||||
// 摘要结果
|
||||
if (resultData?.action_needed === 'provide_document' || resultData?.action_needed === 'upload_document') {
|
||||
responseContent = `📋 ${resultData.message}\n\n${resultData.suggestion || ''}`;
|
||||
} else if (resultData?.ai_summary) {
|
||||
// AI 生成的摘要
|
||||
responseContent = `📄 **${resultData.filename}** 摘要分析：\n\n${resultData.ai_summary}`;
|
||||
} else {
|
||||
responseContent = resultData?.message || '未能生成摘要。请确保已上传文档。';
|
||||
}
|
||||
break;
|
||||
|
||||
1. **上传表格模板** - 您要填写的表格模板文件（Excel 或 Word 格式）
|
||||
2. **选择数据源** - 包含要填写内容的源文档
|
||||
case 'question':
|
||||
// 问答结果
|
||||
if (resultData?.answer) {
|
||||
responseContent = `**问题**: ${resultData.question}\n\n**答案**: ${resultData.answer}`;
|
||||
} else if (resultData?.context_preview) {
|
||||
responseContent = `**问题**: ${resultData.question}\n\n**相关上下文**：\n${resultData.context_preview}`;
|
||||
} else {
|
||||
responseContent = resultData?.message || '请先上传文档，我才能回答您的问题。';
|
||||
}
|
||||
break;
|
||||
|
||||
您可以去【智能填表】页面完成这些操作，或者告诉我您具体想填什么类型的表格，我来指导您操作。`;
|
||||
} else if (userInput.includes('删除')) {
|
||||
response = `要删除文档，请告诉我：
|
||||
case 'search':
|
||||
// 搜索结果
|
||||
const searchResults = resultData?.results || [];
|
||||
if (searchResults.length > 0) {
|
||||
responseContent = `🔍 找到 ${searchResults.length} 条相关内容：\n\n`;
|
||||
searchResults.slice(0, 5).forEach((r: any, idx: number) => {
|
||||
responseContent += `**${idx + 1}.** ${r.content?.substring(0, 100)}...\n\n`;
|
||||
});
|
||||
} else {
|
||||
responseContent = '未找到相关内容。请尝试其他关键词。';
|
||||
}
|
||||
break;
|
||||
|
||||
- 要删除的文件名是什么？
|
||||
- 或者您可以到【文档中心】页面手动选择并删除文档
|
||||
case 'compare':
|
||||
// 对比结果
|
||||
const comparison = resultData?.comparison || [];
|
||||
if (comparison.length > 0) {
|
||||
responseContent = `📊 对比了 ${comparison.length} 个文档：\n\n`;
|
||||
comparison.forEach((c: any) => {
|
||||
responseContent += `- **${c.filename}**: ${c.doc_type}, ${c.content_length} 字\n`;
|
||||
});
|
||||
} else {
|
||||
responseContent = '需要至少2个文档才能进行对比。';
|
||||
}
|
||||
break;
|
||||
|
||||
⚠️ 删除操作不可恢复，请确认后再操作。`;
|
||||
} else if (userInput.includes('帮助') || userInput.includes('help')) {
|
||||
response = `**我可以帮您完成以下操作：**
|
||||
case 'edit':
|
||||
// 文档编辑结果
|
||||
if (resultData?.edited_content) {
|
||||
responseContent = `✏️ **${resultData.original_filename}** 编辑完成：\n\n${resultData.edited_content.substring(0, 500)}${resultData.edited_content.length > 500 ? '\n\n...(内容已截断)' : ''}`;
|
||||
} else {
|
||||
responseContent = resultData?.message || '编辑完成。';
|
||||
}
|
||||
break;
|
||||
|
||||
📄 **文档管理**
|
||||
- 列出/搜索已上传的文档
|
||||
- 查看文档详情和元数据
|
||||
- 删除不需要的文档
|
||||
case 'transform':
|
||||
// 格式转换结果
|
||||
if (resultData?.excel_data) {
|
||||
responseContent = `🔄 格式转换完成！\n\n已转换为 **Excel** 格式，共 **${resultData.excel_data.length}** 行数据。\n\n${resultData.message || ''}`;
|
||||
} else if (resultData?.content) {
|
||||
responseContent = `🔄 格式转换完成！\n\n目标格式: **${resultData.target_format?.toUpperCase()}**\n\n${resultData.message || ''}`;
|
||||
} else {
|
||||
responseContent = resultData?.message || '格式转换完成。';
|
||||
}
|
||||
break;
|
||||
|
||||
📊 **Excel 处理**
|
||||
- 分析 Excel 文件内容
|
||||
- 生成数据统计和图表
|
||||
- 导出处理后的数据
|
||||
case 'unknown':
|
||||
// 检查是否需要用户上传文档
|
||||
if (resultData?.suggestion) {
|
||||
responseContent = resultData.suggestion;
|
||||
} else if (resultData?.message && resultData.message !== '无法理解该指令，请尝试更明确的描述') {
|
||||
responseContent = resultData.message;
|
||||
} else {
|
||||
responseContent = `我理解您想要： "${input.trim()}"\n\n请尝试以下操作：\n\n1. **提取数据**: "提取医院数量和床位数"\n2. **填表**: "根据这些数据填表"\n3. **总结**: "总结这份文档"\n4. **问答**: "文档里说了什么？"\n5. **搜索**: "搜索相关内容"`;
|
||||
}
|
||||
break;
|
||||
|
||||
📝 **智能填表**
|
||||
- 上传表格模板
|
||||
- 从文档中提取信息填入模板
|
||||
- 导出填写完成的表格
|
||||
|
||||
📋 **任务历史**
|
||||
- 查看历史处理任务
|
||||
- 重新执行或导出结果
|
||||
|
||||
请直接告诉我您想做什么！`;
|
||||
} else {
|
||||
response = `我理解您想要： "${input.trim()}"
|
||||
|
||||
目前我还在学习如何更好地理解您的需求。您可以尝试：
|
||||
|
||||
1. **上传文档** - 去【文档中心】上传 docx/md/txt 文件
|
||||
2. **分析 Excel** - 去【Excel解析】上传并分析 Excel 文件
|
||||
3. **智能填表** - 去【智能填表】创建填表任务
|
||||
|
||||
或者您可以更具体地描述您想做的事情，我会尽力帮助您！`;
|
||||
default:
|
||||
responseContent = response.message || resultData?.message || '已完成您的请求。';
|
||||
}
|
||||
|
||||
const assistantMessage: ChatMessage = {
|
||||
id: Math.random().toString(36).substring(7),
|
||||
role: 'assistant',
|
||||
content: response,
|
||||
created_at: new Date().toISOString()
|
||||
content: responseContent,
|
||||
created_at: new Date().toISOString(),
|
||||
intent: response.intent,
|
||||
result: resultData
|
||||
};
|
||||
|
||||
setMessages(prev => [...prev, assistantMessage]);
|
||||
} catch (err: any) {
|
||||
toast.error('请求失败，请重试');
|
||||
console.error('指令执行失败:', err);
|
||||
toast.error(err.message || '请求失败，请重试');
|
||||
|
||||
const errorMessage: ChatMessage = {
|
||||
id: Math.random().toString(36).substring(7),
|
||||
role: 'assistant',
|
||||
content: `抱歉，处理您的请求时遇到了问题：${err.message}\n\n请稍后重试，或尝试更简单的指令。`,
|
||||
created_at: new Date().toISOString()
|
||||
};
|
||||
setMessages(prev => [...prev, errorMessage]);
|
||||
} finally {
|
||||
setLoading(false);
|
||||
}
|
||||
@@ -189,10 +275,10 @@ const InstructionChat: React.FC = () => {
|
||||
};
|
||||
|
||||
const quickActions = [
|
||||
{ label: '列出所有文档', icon: FileText, action: () => setInput('列出所有已上传的文档') },
|
||||
{ label: '分析 Excel 数据', icon: TableProperties, action: () => setInput('分析一下 Excel 文件') },
|
||||
{ label: '智能填表', icon: Sparkles, action: () => setInput('我想进行智能填表') },
|
||||
{ label: '帮助', icon: Sparkles, action: () => setInput('帮助') }
|
||||
{ label: '提取医院数量', icon: Search, action: () => setInput('提取文档中的医院数量和床位数') },
|
||||
{ label: '智能填表', icon: TableProperties, action: () => setInput('根据这些数据填表') },
|
||||
{ label: '总结文档', icon: MessageSquare, action: () => setInput('总结一下这份文档') },
|
||||
{ label: '智能问答', icon: Bot, action: () => setInput('文档里说了些什么？') }
|
||||
];
|
||||
|
||||
return (
|
||||
@@ -241,9 +327,11 @@ const InstructionChat: React.FC = () => {
|
||||
? "bg-primary text-primary-foreground shadow-xl shadow-primary/20 rounded-tr-none"
|
||||
: "bg-white border border-border/50 shadow-md rounded-tl-none"
|
||||
)}>
|
||||
<p className="text-sm leading-relaxed whitespace-pre-wrap font-medium">
|
||||
{m.content}
|
||||
</p>
|
||||
{m.role === 'assistant' ? (
|
||||
<Markdown content={m.content} className="text-sm leading-relaxed prose prose-sm max-w-none" />
|
||||
) : (
|
||||
<p className="text-sm leading-relaxed whitespace-pre-wrap font-medium">{m.content}</p>
|
||||
)}
|
||||
<span className={cn(
|
||||
"text-[10px] block opacity-50 font-bold tracking-widest",
|
||||
m.role === 'user' ? "text-right" : "text-left"
|
||||
|
||||
446
frontend/src/pages/PdfConverter.tsx
Normal file
446
frontend/src/pages/PdfConverter.tsx
Normal file
@@ -0,0 +1,446 @@
|
||||
/**
|
||||
* PDF 转换页面
|
||||
* 支持将 Word、Excel、Txt、Markdown 格式转换为 PDF
|
||||
*/
|
||||
import React, { useState, useCallback } from 'react';
|
||||
import { useDropzone } from 'react-dropzone';
|
||||
import {
|
||||
FileText,
|
||||
Upload,
|
||||
Download,
|
||||
FileSpreadsheet,
|
||||
File as FileIcon,
|
||||
Loader2,
|
||||
CheckCircle,
|
||||
AlertCircle,
|
||||
Trash2,
|
||||
FileDown,
|
||||
X,
|
||||
Copy
|
||||
} from 'lucide-react';
|
||||
import { Button } from '@/components/ui/button';
|
||||
import { Card, CardContent, CardHeader, CardTitle, CardDescription } from '@/components/ui/card';
|
||||
import { Badge } from '@/components/ui/badge';
|
||||
import { Label } from '@/components/ui/label';
|
||||
import { toast } from 'sonner';
|
||||
import { cn } from '@/lib/utils';
|
||||
import { backendApi } from '@/db/backend-api';
|
||||
|
||||
type FileState = {
|
||||
file: File;
|
||||
status: 'pending' | 'converting' | 'success' | 'failed';
|
||||
progress: number;
|
||||
pdfBlob?: Blob;
|
||||
error?: string;
|
||||
};
|
||||
|
||||
const SUPPORTED_FORMATS = [
|
||||
{ ext: 'docx', name: 'Word 文档', icon: FileText, color: 'blue' },
|
||||
{ ext: 'xlsx', name: 'Excel 表格', icon: FileSpreadsheet, color: 'emerald' },
|
||||
{ ext: 'txt', name: '文本文件', icon: FileIcon, color: 'gray' },
|
||||
{ ext: 'md', name: 'Markdown', icon: FileText, color: 'purple' },
|
||||
];
|
||||
|
||||
const PdfConverter: React.FC = () => {
|
||||
const [files, setFiles] = useState<FileState[]>([]);
|
||||
const [converting, setConverting] = useState(false);
|
||||
const [convertedCount, setConvertedCount] = useState(0);
|
||||
|
||||
const onDrop = useCallback((acceptedFiles: File[]) => {
|
||||
const newFiles: FileState[] = acceptedFiles.map(file => ({
|
||||
file,
|
||||
status: 'pending',
|
||||
progress: 0,
|
||||
}));
|
||||
setFiles(prev => [...prev, ...newFiles]);
|
||||
}, []);
|
||||
|
||||
const { getRootProps, getInputProps, isDragActive } = useDropzone({
|
||||
onDrop,
|
||||
accept: {
|
||||
'application/vnd.openxmlformats-officedocument.wordprocessingml.document': ['.docx'],
|
||||
'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet': ['.xlsx'],
|
||||
'application/vnd.ms-excel': ['.xls'],
|
||||
'text/markdown': ['.md'],
|
||||
'text/plain': ['.txt'],
|
||||
},
|
||||
multiple: true,
|
||||
});
|
||||
|
||||
const handleConvert = async () => {
|
||||
if (files.length === 0) {
|
||||
toast.error('请先上传文件');
|
||||
return;
|
||||
}
|
||||
|
||||
setConverting(true);
|
||||
setConvertedCount(0);
|
||||
|
||||
const pendingFiles = files.filter(f => f.status === 'pending' || f.status === 'failed');
|
||||
let successCount = 0;
|
||||
|
||||
for (let i = 0; i < pendingFiles.length; i++) {
|
||||
const fileState = pendingFiles[i];
|
||||
const fileIndex = files.findIndex(f => f.file === fileState.file);
|
||||
|
||||
// 更新状态为转换中
|
||||
setFiles(prev => prev.map((f, idx) =>
|
||||
idx === fileIndex ? { ...f, status: 'converting', progress: 10 } : f
|
||||
));
|
||||
|
||||
try {
|
||||
// 获取 PDF blob
|
||||
const pdfBlob = await backendApi.convertToPdf(fileState.file);
|
||||
|
||||
// 触发下载
|
||||
const url = URL.createObjectURL(pdfBlob);
|
||||
const a = document.createElement('a');
|
||||
a.href = url;
|
||||
a.download = `${fileState.file.name.replace(/\.[^.]+$/, '')}.pdf`;
|
||||
document.body.appendChild(a);
|
||||
a.click();
|
||||
document.body.removeChild(a);
|
||||
URL.revokeObjectURL(url);
|
||||
|
||||
// 保存 blob 以便批量下载
|
||||
setFiles(prev => prev.map((f, idx) =>
|
||||
idx === fileIndex ? { ...f, status: 'success', progress: 100, pdfBlob } : f
|
||||
));
|
||||
successCount++;
|
||||
setConvertedCount(successCount);
|
||||
toast.success(`${fileState.file.name} 下载已开始`);
|
||||
} catch (error: any) {
|
||||
setFiles(prev => prev.map((f, idx) =>
|
||||
idx === fileIndex ? { ...f, status: 'failed', error: error.message || '转换失败' } : f
|
||||
));
|
||||
}
|
||||
}
|
||||
|
||||
setConverting(false);
|
||||
toast.success(`转换完成：${successCount}/${pendingFiles.length} 个文件`);
|
||||
};
|
||||
|
||||
const handleDownload = (fileState: FileState) => {
|
||||
if (!fileState.pdfBlob) return;
|
||||
|
||||
const url = URL.createObjectURL(fileState.pdfBlob);
|
||||
const link = document.createElement('a');
|
||||
link.href = url;
|
||||
link.download = `${fileState.file.name.replace(/\.[^.]+$/, '')}.pdf`;
|
||||
document.body.appendChild(link);
|
||||
link.click();
|
||||
document.body.removeChild(link);
|
||||
URL.revokeObjectURL(url);
|
||||
};
|
||||
|
||||
const handleDownloadAll = async () => {
|
||||
const successFiles = files.filter(f => f.status === 'success' && f.pdfBlob);
|
||||
|
||||
if (successFiles.length === 0) {
|
||||
toast.error('没有可下载的文件');
|
||||
return;
|
||||
}
|
||||
|
||||
if (successFiles.length === 1) {
|
||||
handleDownload(successFiles[0]);
|
||||
return;
|
||||
}
|
||||
|
||||
// 多个文件，下载 ZIP
|
||||
try {
|
||||
const zipBlob = await backendApi.batchConvertToPdf(
|
||||
successFiles.map(f => f.file)
|
||||
);
|
||||
const url = URL.createObjectURL(zipBlob);
|
||||
const link = document.createElement('a');
|
||||
link.href = url;
|
||||
link.download = 'converted_pdfs.zip';
|
||||
document.body.appendChild(link);
|
||||
link.click();
|
||||
document.body.removeChild(link);
|
||||
URL.revokeObjectURL(url);
|
||||
toast.success('ZIP 下载开始');
|
||||
} catch (error: any) {
|
||||
toast.error(error.message || '下载失败');
|
||||
}
|
||||
};
|
||||
|
||||
const handleRemove = (index: number) => {
|
||||
setFiles(prev => prev.filter((_, i) => i !== index));
|
||||
};
|
||||
|
||||
const handleClear = () => {
|
||||
setFiles([]);
|
||||
setConvertedCount(0);
|
||||
};
|
||||
|
||||
const getFileIcon = (filename: string) => {
|
||||
const ext = filename.split('.').pop()?.toLowerCase();
|
||||
const format = SUPPORTED_FORMATS.find(f => f.ext === ext);
|
||||
if (!format) return FileIcon;
|
||||
return format.icon;
|
||||
};
|
||||
|
||||
const getFileColor = (filename: string) => {
|
||||
const ext = filename.split('.').pop()?.toLowerCase();
|
||||
const format = SUPPORTED_FORMATS.find(f => f.ext === ext);
|
||||
return format?.color || 'gray';
|
||||
};
|
||||
|
||||
const colorClasses: Record<string, string> = {
|
||||
blue: 'bg-blue-500/10 text-blue-500',
|
||||
emerald: 'bg-emerald-500/10 text-emerald-500',
|
||||
purple: 'bg-purple-500/10 text-purple-500',
|
||||
gray: 'bg-gray-500/10 text-gray-500',
|
||||
};
|
||||
|
||||
return (
|
||||
<div className="space-y-8 pb-10">
|
||||
<section className="flex flex-col md:flex-row md:items-center justify-between gap-4">
|
||||
<div className="space-y-1">
|
||||
<h1 className="text-3xl font-extrabold tracking-tight">文档转 PDF</h1>
|
||||
<p className="text-muted-foreground">将 Word、Excel、文本、Markdown 文件转换为 PDF 格式</p>
|
||||
</div>
|
||||
{files.length > 0 && (
|
||||
<div className="flex gap-2">
|
||||
<Button variant="outline" onClick={handleClear}>
|
||||
<Trash2 size={18} className="mr-2" />
|
||||
清空
|
||||
</Button>
|
||||
<Button onClick={handleDownloadAll} disabled={files.filter(f => f.status === 'success').length === 0}>
|
||||
<Download size={18} className="mr-2" />
|
||||
打包下载 ({files.filter(f => f.status === 'success').length})
|
||||
</Button>
|
||||
</div>
|
||||
)}
|
||||
</section>
|
||||
|
||||
<div className="grid grid-cols-1 lg:grid-cols-3 gap-6">
|
||||
{/* 左侧：上传区域 */}
|
||||
<div className="lg:col-span-1 space-y-6">
|
||||
{/* 上传卡片 */}
|
||||
<Card className="border-none shadow-md">
|
||||
<CardHeader className="pb-4">
|
||||
<CardTitle className="flex items-center gap-2">
|
||||
<Upload className="text-primary" size={20} />
|
||||
上传文件
|
||||
</CardTitle>
|
||||
<CardDescription>拖拽或点击上传要转换的文件</CardDescription>
|
||||
</CardHeader>
|
||||
<CardContent className="space-y-4">
|
||||
<div
|
||||
{...getRootProps()}
|
||||
className={cn(
|
||||
"border-2 border-dashed rounded-2xl p-8 transition-all duration-300 flex flex-col items-center justify-center text-center cursor-pointer group",
|
||||
isDragActive ? "border-primary bg-primary/5" : "border-muted-foreground/20 hover:border-primary/50 hover:bg-primary/5",
|
||||
converting && "opacity-50 pointer-events-none"
|
||||
)}
|
||||
>
|
||||
<input {...getInputProps()} />
|
||||
<div className="w-14 h-14 rounded-xl bg-primary/10 text-primary flex items-center justify-center mb-4 group-hover:scale-110 transition-transform">
|
||||
{converting ? <Loader2 className="animate-spin" size={28} /> : <Upload size={28} />}
|
||||
</div>
|
||||
<p className="font-semibold text-sm">
|
||||
{isDragActive ? '释放以开始上传' : '点击或拖拽文件到这里'}
|
||||
</p>
|
||||
<div className="mt-4 flex flex-wrap justify-center gap-2">
|
||||
{SUPPORTED_FORMATS.map(format => (
|
||||
<Badge key={format.ext} variant="outline" className={cn("text-xs", colorClasses[format.color])}>
|
||||
{format.name}
|
||||
</Badge>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 转换按钮 */}
|
||||
{files.length > 0 && (
|
||||
<Button
|
||||
onClick={handleConvert}
|
||||
disabled={converting || files.filter(f => f.status === 'pending' || f.status === 'failed').length === 0}
|
||||
className="w-full bg-gradient-to-r from-primary to-purple-600 hover:from-primary/90 hover:to-purple-600/90"
|
||||
>
|
||||
{converting ? (
|
||||
<>
|
||||
<Loader2 className="mr-2 animate-spin" size={16} />
|
||||
转换中... ({convertedCount}/{files.filter(f => f.status === 'pending' || f.status === 'failed').length})
|
||||
</>
|
||||
) : (
|
||||
<>
|
||||
<FileDown className="mr-2" size={16} />
|
||||
开始转换 ({files.filter(f => f.status === 'pending' || f.status === 'failed').length})
|
||||
</>
|
||||
)}
|
||||
</Button>
|
||||
)}
|
||||
</CardContent>
|
||||
</Card>
|
||||
|
||||
{/* 格式说明 */}
|
||||
<Card className="border-none shadow-md">
|
||||
<CardHeader className="pb-4">
|
||||
<CardTitle className="flex items-center gap-2">
|
||||
<FileText className="text-primary" size={20} />
|
||||
支持的格式
|
||||
</CardTitle>
|
||||
</CardHeader>
|
||||
<CardContent>
|
||||
<div className="space-y-3">
|
||||
{SUPPORTED_FORMATS.map(format => {
|
||||
const Icon = format.icon;
|
||||
return (
|
||||
<div key={format.ext} className="flex items-center gap-3 p-2 rounded-lg hover:bg-muted/30 transition-colors">
|
||||
<div className={cn("w-8 h-8 rounded flex items-center justify-center", colorClasses[format.color])}>
|
||||
<Icon size={16} />
|
||||
</div>
|
||||
<div className="flex-1">
|
||||
<p className="text-sm font-medium">.{format.ext.toUpperCase()}</p>
|
||||
<p className="text-xs text-muted-foreground">{format.name}</p>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
})}
|
||||
</div>
|
||||
</CardContent>
|
||||
</Card>
|
||||
</div>
|
||||
|
||||
{/* 右侧：文件列表 */}
|
||||
<div className="lg:col-span-2 space-y-6">
|
||||
<Card className="border-none shadow-md">
|
||||
<CardHeader>
|
||||
<div className="flex items-center justify-between">
|
||||
<div className="space-y-1">
|
||||
<CardTitle className="flex items-center gap-2">
|
||||
<FileIcon className="text-primary" size={20} />
|
||||
文件列表
|
||||
</CardTitle>
|
||||
<CardDescription>
|
||||
共 {files.length} 个文件，已转换 {files.filter(f => f.status === 'success').length} 个
|
||||
</CardDescription>
|
||||
</div>
|
||||
</div>
|
||||
</CardHeader>
|
||||
<CardContent>
|
||||
{files.length === 0 ? (
|
||||
<div className="text-center py-12 text-muted-foreground">
|
||||
<FileIcon size={48} className="mx-auto mb-4 opacity-30" />
|
||||
<p>暂无文件，上传文件开始转换</p>
|
||||
</div>
|
||||
) : (
|
||||
<div className="space-y-3">
|
||||
{files.map((fileState, index) => {
|
||||
const Icon = getFileIcon(fileState.file.name);
|
||||
const color = getFileColor(fileState.file.name);
|
||||
|
||||
return (
|
||||
<div
|
||||
key={index}
|
||||
className="flex items-center gap-4 p-4 rounded-xl border bg-card hover:bg-muted/30 transition-colors"
|
||||
>
|
||||
<div className={cn("w-10 h-10 rounded-lg flex items-center justify-center shrink-0", colorClasses[color])}>
|
||||
<Icon size={20} />
|
||||
</div>
|
||||
<div className="flex-1 min-w-0">
|
||||
<p className="font-semibold truncate">{fileState.file.name}</p>
|
||||
<div className="flex items-center gap-2">
|
||||
<span className="text-xs text-muted-foreground">
|
||||
{(fileState.file.size / 1024).toFixed(1)} KB
|
||||
</span>
|
||||
{fileState.status === 'pending' && (
|
||||
<Badge variant="secondary" className="text-xs">待转换</Badge>
|
||||
)}
|
||||
{fileState.status === 'converting' && (
|
||||
<Badge variant="default" className="text-xs bg-blue-500">转换中</Badge>
|
||||
)}
|
||||
{fileState.status === 'success' && (
|
||||
<Badge variant="default" className="text-xs bg-emerald-500">已转换</Badge>
|
||||
)}
|
||||
{fileState.status === 'failed' && (
|
||||
<Badge variant="destructive" className="text-xs">失败</Badge>
|
||||
)}
|
||||
</div>
|
||||
{fileState.status === 'converting' && (
|
||||
<div className="mt-1 h-1 bg-muted rounded-full overflow-hidden">
|
||||
<div
|
||||
className="h-full bg-primary transition-all duration-300"
|
||||
style={{ width: `${fileState.progress}%` }}
|
||||
/>
|
||||
</div>
|
||||
)}
|
||||
{fileState.error && (
|
||||
<p className="text-xs text-destructive mt-1">{fileState.error}</p>
|
||||
)}
|
||||
</div>
|
||||
<div className="flex items-center gap-2 shrink-0">
|
||||
{fileState.status === 'success' && (
|
||||
<>
|
||||
<Button variant="ghost" size="icon" onClick={() => handleDownload(fileState)}>
|
||||
<Download size={18} className="text-emerald-500" />
|
||||
</Button>
|
||||
<Button
|
||||
variant="ghost"
|
||||
size="icon"
|
||||
onClick={() => {
|
||||
// 复制下载链接到剪贴板
|
||||
if (fileState.pdfBlob) {
|
||||
const url = URL.createObjectURL(fileState.pdfBlob);
|
||||
navigator.clipboard.writeText(url);
|
||||
toast.success('链接已复制');
|
||||
}
|
||||
}}
|
||||
>
|
||||
<Copy size={18} />
|
||||
</Button>
|
||||
</>
|
||||
)}
|
||||
{(fileState.status === 'pending' || fileState.status === 'failed') && (
|
||||
<Button
|
||||
variant="ghost"
|
||||
size="icon"
|
||||
onClick={() => handleRemove(index)}
|
||||
className="text-destructive hover:bg-destructive/10"
|
||||
>
|
||||
<X size={18} />
|
||||
</Button>
|
||||
)}
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
})}
|
||||
</div>
|
||||
)}
|
||||
</CardContent>
|
||||
</Card>
|
||||
|
||||
{/* 使用说明 */}
|
||||
<Card className="border-none shadow-md bg-gradient-to-br from-primary/5 to-purple-500/5">
|
||||
<CardHeader className="pb-4">
|
||||
<CardTitle className="flex items-center gap-2">
|
||||
<FileText className="text-primary" size={20} />
|
||||
使用说明
|
||||
</CardTitle>
|
||||
</CardHeader>
|
||||
<CardContent>
|
||||
<div className="space-y-3 text-sm text-muted-foreground">
|
||||
<div className="flex gap-3">
|
||||
<div className="w-6 h-6 rounded-full bg-primary/10 text-primary flex items-center justify-center shrink-0 text-xs font-bold">1</div>
|
||||
<p>上传要转换的文件，支持 Word(.docx)、Excel(.xlsx)、文本(.txt)、Markdown(.md) 格式</p>
|
||||
</div>
|
||||
<div className="flex gap-3">
|
||||
<div className="w-6 h-6 rounded-full bg-primary/10 text-primary flex items-center justify-center shrink-0 text-xs font-bold">2</div>
|
||||
<p>点击「开始转换」按钮，系统将自动将文件转换为 PDF 格式</p>
|
||||
</div>
|
||||
<div className="flex gap-3">
|
||||
<div className="w-6 h-6 rounded-full bg-primary/10 text-primary flex items-center justify-center shrink-0 text-xs font-bold">3</div>
|
||||
<p>转换完成后，点击下载按钮获取 PDF 文件，或使用「打包下载」一次性下载所有文件</p>
|
||||
</div>
|
||||
</div>
|
||||
</CardContent>
|
||||
</Card>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
};
|
||||
|
||||
export default PdfConverter;
|
||||
@@ -245,37 +245,32 @@ const TemplateFill: React.FC = () => {
|
||||
};
|
||||
|
||||
const handleExport = async () => {
|
||||
if (!templateFile || !filledResult) {
|
||||
console.error('handleExport 失败: templateFile=', templateFile, 'filledResult=', filledResult);
|
||||
toast.error('数据不完整，无法导出');
|
||||
return;
|
||||
}
|
||||
|
||||
console.log('=== handleExport 调试 ===');
|
||||
console.log('templateFile:', templateFile);
|
||||
console.log('templateId:', templateId);
|
||||
console.log('filledResult:', filledResult);
|
||||
console.log('filledResult.filled_data:', filledResult.filled_data);
|
||||
console.log('=========================');
|
||||
|
||||
const ext = templateFile.name.split('.').pop()?.toLowerCase();
|
||||
if (!templateFile || !filledResult) return;
|
||||
|
||||
try {
|
||||
// 使用新的 fillAndExportTemplate 直接填充原始模板
|
||||
const blob = await backendApi.fillAndExportTemplate(
|
||||
templateId || '',
|
||||
const ext = templateFile.name.split('.').pop()?.toLowerCase();
|
||||
const exportFormat = (ext === 'docx') ? 'docx' : 'xlsx';
|
||||
// 对于 Word 模板，如果已有填写后的文件（已填入表格单元格），传递其路径以便直接下载
|
||||
const filledFilePath = (ext === 'docx' && filledResult.filled_file_path)
|
||||
? filledResult.filled_file_path
|
||||
: undefined;
|
||||
const blob = await backendApi.exportFilledTemplate(
|
||||
templateId || 'temp',
|
||||
filledResult.filled_data || {},
|
||||
ext === 'docx' ? 'docx' : 'xlsx'
|
||||
exportFormat,
|
||||
filledFilePath
|
||||
);
|
||||
const ext_match = templateFile.name.match(/\.([^.])+$/);
|
||||
const baseName = ext_match ? templateFile.name.replace(ext_match[0], '') : templateFile.name;
|
||||
const downloadName = `filled_${baseName}.${exportFormat}`;
|
||||
const url = URL.createObjectURL(blob);
|
||||
const a = document.createElement('a');
|
||||
a.href = url;
|
||||
a.download = `filled_${templateFile.name}`;
|
||||
a.download = downloadName;
|
||||
a.click();
|
||||
URL.revokeObjectURL(url);
|
||||
toast.success('导出成功');
|
||||
} catch (err: any) {
|
||||
console.error('导出失败:', err);
|
||||
toast.error('导出失败: ' + (err.message || '未知错误'));
|
||||
}
|
||||
};
|
||||
@@ -561,7 +556,7 @@ const TemplateFill: React.FC = () => {
|
||||
</div>
|
||||
<h3 className="text-xl font-bold mb-2">AI 正在智能分析并填表</h3>
|
||||
<p className="text-muted-foreground text-center max-w-md">
|
||||
系统正在从 {sourceFiles.length || sourceFilePaths.length} 份文档中检索相关信息...
|
||||
系统正在从 {sourceFiles.length || sourceFilePaths.length || sourceDocIds.length || 0} 份文档中检索相关信息...
|
||||
</p>
|
||||
</CardContent>
|
||||
</Card>
|
||||
@@ -577,7 +572,7 @@ const TemplateFill: React.FC = () => {
|
||||
填表完成
|
||||
</CardTitle>
|
||||
<CardDescription>
|
||||
系统已根据 {sourceFiles.length || sourceFilePaths.length} 份文档自动完成表格填写
|
||||
系统已根据 {filledResult.source_doc_count || sourceFiles.length || sourceFilePaths.length || sourceDocIds.length} 份文档自动完成表格填写
|
||||
</CardDescription>
|
||||
</CardHeader>
|
||||
<CardContent>
|
||||
@@ -641,6 +636,16 @@ const TemplateFill: React.FC = () => {
|
||||
<div className="text-muted-foreground text-xs mt-1">
|
||||
来源: {detail.source} | 置信度: {detail.confidence ? (detail.confidence * 100).toFixed(0) + '%' : 'N/A'}
|
||||
</div>
|
||||
{detail.warning && (
|
||||
<div className="mt-2 p-2 bg-yellow-50 border border-yellow-200 rounded-lg text-yellow-700 text-xs">
|
||||
⚠️ {detail.warning}
|
||||
</div>
|
||||
)}
|
||||
{detail.values && detail.values.length > 1 && !detail.warning && (
|
||||
<div className="mt-2 text-xs text-muted-foreground">
|
||||
多值: {detail.values.join(', ')}
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
</div>
|
||||
))}
|
||||
|
||||
@@ -4,6 +4,7 @@ import Documents from '@/pages/Documents';
|
||||
import TemplateFill from '@/pages/TemplateFill';
|
||||
import InstructionChat from '@/pages/InstructionChat';
|
||||
import TaskHistory from '@/pages/TaskHistory';
|
||||
import PdfConverter from '@/pages/PdfConverter';
|
||||
import MainLayout from '@/components/layouts/MainLayout';
|
||||
|
||||
export const routes = [
|
||||
@@ -31,6 +32,10 @@ export const routes = [
|
||||
path: '/task-history',
|
||||
element: <TaskHistory />,
|
||||
},
|
||||
{
|
||||
path: '/pdf-converter',
|
||||
element: <PdfConverter />,
|
||||
},
|
||||
],
|
||||
},
|
||||
{
|
||||
|
||||
@@ -23,7 +23,6 @@
|
||||
"noUnusedParameters": true,
|
||||
"noFallthroughCasesInSwitch": true,
|
||||
"noUncheckedSideEffectImports": true,
|
||||
"baseUrl": ".",
|
||||
"paths": {
|
||||
"@/*": ["./src/*"]
|
||||
},
|
||||
|
||||
@@ -1,854 +0,0 @@
|
||||
diff --git a/backend/app/api/endpoints/templates.py b/backend/app/api/endpoints/templates.py
|
||||
index 572d56e..706f281 100644
|
||||
--- a/backend/app/api/endpoints/templates.py
|
||||
+++ b/backend/app/api/endpoints/templates.py
|
||||
@@ -13,7 +13,7 @@ import pandas as pd
|
||||
from pydantic import BaseModel
|
||||
|
||||
from app.services.template_fill_service import template_fill_service, TemplateField
|
||||
-from app.services.excel_storage_service import excel_storage_service
|
||||
+from app.services.file_service import file_service
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@@ -28,13 +28,15 @@ class TemplateFieldRequest(BaseModel):
|
||||
name: str
|
||||
field_type: str = "text"
|
||||
required: bool = True
|
||||
+ hint: str = ""
|
||||
|
||||
|
||||
class FillRequest(BaseModel):
|
||||
"""填写请求"""
|
||||
template_id: str
|
||||
template_fields: List[TemplateFieldRequest]
|
||||
- source_doc_ids: Optional[List[str]] = None
|
||||
+ source_doc_ids: Optional[List[str]] = None # MongoDB 文档 ID 列表
|
||||
+ source_file_paths: Optional[List[str]] = None # 源文档文件路径列表
|
||||
user_hint: Optional[str] = None
|
||||
|
||||
|
||||
@@ -71,7 +73,6 @@ async def upload_template(
|
||||
|
||||
try:
|
||||
# 保存文件
|
||||
- from app.services.file_service import file_service
|
||||
content = await file.read()
|
||||
saved_path = file_service.save_uploaded_file(
|
||||
content,
|
||||
@@ -87,7 +88,7 @@ async def upload_template(
|
||||
|
||||
return {
|
||||
"success": True,
|
||||
- "template_id": saved_path, # 使用文件路径作为ID
|
||||
+ "template_id": saved_path,
|
||||
"filename": file.filename,
|
||||
"file_type": file_ext,
|
||||
"fields": [
|
||||
@@ -95,7 +96,8 @@ async def upload_template(
|
||||
"cell": f.cell,
|
||||
"name": f.name,
|
||||
"field_type": f.field_type,
|
||||
- "required": f.required
|
||||
+ "required": f.required,
|
||||
+ "hint": f.hint
|
||||
}
|
||||
for f in template_fields
|
||||
],
|
||||
@@ -135,7 +137,8 @@ async def extract_template_fields(
|
||||
"cell": f.cell,
|
||||
"name": f.name,
|
||||
"field_type": f.field_type,
|
||||
- "required": f.required
|
||||
+ "required": f.required,
|
||||
+ "hint": f.hint
|
||||
}
|
||||
for f in fields
|
||||
]
|
||||
@@ -153,7 +156,7 @@ async def fill_template(
|
||||
"""
|
||||
执行表格填写
|
||||
|
||||
- 根据提供的字段定义，从已上传的文档中检索信息并填写
|
||||
+ 根据提供的字段定义，从源文档中检索信息并填写
|
||||
|
||||
Args:
|
||||
request: 填写请求
|
||||
@@ -168,7 +171,8 @@ async def fill_template(
|
||||
cell=f.cell,
|
||||
name=f.name,
|
||||
field_type=f.field_type,
|
||||
- required=f.required
|
||||
+ required=f.required,
|
||||
+ hint=f.hint
|
||||
)
|
||||
for f in request.template_fields
|
||||
]
|
||||
@@ -177,6 +181,7 @@ async def fill_template(
|
||||
result = await template_fill_service.fill_template(
|
||||
template_fields=fields,
|
||||
source_doc_ids=request.source_doc_ids,
|
||||
+ source_file_paths=request.source_file_paths,
|
||||
user_hint=request.user_hint
|
||||
)
|
||||
|
||||
@@ -194,6 +199,8 @@ async def export_filled_template(
|
||||
"""
|
||||
导出填写后的表格
|
||||
|
||||
+ 支持 Excel (.xlsx) 和 Word (.docx) 格式
|
||||
+
|
||||
Args:
|
||||
request: 导出请求
|
||||
|
||||
@@ -201,25 +208,124 @@ async def export_filled_template(
|
||||
文件流
|
||||
"""
|
||||
try:
|
||||
- # 创建 DataFrame
|
||||
- df = pd.DataFrame([request.filled_data])
|
||||
+ if request.format == "xlsx":
|
||||
+ return await _export_to_excel(request.filled_data, request.template_id)
|
||||
+ elif request.format == "docx":
|
||||
+ return await _export_to_word(request.filled_data, request.template_id)
|
||||
+ else:
|
||||
+ raise HTTPException(
|
||||
+ status_code=400,
|
||||
+ detail=f"不支持的导出格式: {request.format}，仅支持 xlsx/docx"
|
||||
+ )
|
||||
|
||||
- # 导出为 Excel
|
||||
- output = io.BytesIO()
|
||||
- with pd.ExcelWriter(output, engine='openpyxl') as writer:
|
||||
- df.to_excel(writer, index=False, sheet_name='填写结果')
|
||||
+ except HTTPException:
|
||||
+ raise
|
||||
+ except Exception as e:
|
||||
+ logger.error(f"导出失败: {str(e)}")
|
||||
+ raise HTTPException(status_code=500, detail=f"导出失败: {str(e)}")
|
||||
|
||||
- output.seek(0)
|
||||
|
||||
- # 生成文件名
|
||||
- filename = f"filled_template.{request.format}"
|
||||
+async def _export_to_excel(filled_data: dict, template_id: str) -> StreamingResponse:
|
||||
+ """导出为 Excel 格式"""
|
||||
+ # 将字典转换为单行 DataFrame
|
||||
+ df = pd.DataFrame([filled_data])
|
||||
|
||||
- return StreamingResponse(
|
||||
- io.BytesIO(output.getvalue()),
|
||||
- media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
|
||||
- headers={"Content-Disposition": f"attachment; filename={filename}"}
|
||||
- )
|
||||
+ output = io.BytesIO()
|
||||
+ with pd.ExcelWriter(output, engine='openpyxl') as writer:
|
||||
+ df.to_excel(writer, index=False, sheet_name='填写结果')
|
||||
|
||||
- except Exception as e:
|
||||
- logger.error(f"导出失败: {str(e)}")
|
||||
- raise HTTPException(status_code=500, detail=f"导出失败: {str(e)}")
|
||||
+ output.seek(0)
|
||||
+
|
||||
+ filename = f"filled_template.xlsx"
|
||||
+
|
||||
+ return StreamingResponse(
|
||||
+ io.BytesIO(output.getvalue()),
|
||||
+ media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
|
||||
+ headers={"Content-Disposition": f"attachment; filename={filename}"}
|
||||
+ )
|
||||
+
|
||||
+
|
||||
+async def _export_to_word(filled_data: dict, template_id: str) -> StreamingResponse:
|
||||
+ """导出为 Word 格式"""
|
||||
+ from docx import Document
|
||||
+ from docx.shared import Pt, RGBColor
|
||||
+ from docx.enum.text import WD_ALIGN_PARAGRAPH
|
||||
+
|
||||
+ doc = Document()
|
||||
+
|
||||
+ # 添加标题
|
||||
+ title = doc.add_heading('填写结果', level=1)
|
||||
+ title.alignment = WD_ALIGN_PARAGRAPH.CENTER
|
||||
+
|
||||
+ # 添加填写时间和模板信息
|
||||
+ from datetime import datetime
|
||||
+ info_para = doc.add_paragraph()
|
||||
+ info_para.add_run(f"模板ID: {template_id}\n").bold = True
|
||||
+ info_para.add_run(f"导出时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
|
||||
+
|
||||
+ doc.add_paragraph() # 空行
|
||||
+
|
||||
+ # 添加字段表格
|
||||
+ table = doc.add_table(rows=1, cols=3)
|
||||
+ table.style = 'Light Grid Accent 1'
|
||||
+
|
||||
+ # 表头
|
||||
+ header_cells = table.rows[0].cells
|
||||
+ header_cells[0].text = '字段名'
|
||||
+ header_cells[1].text = '填写值'
|
||||
+ header_cells[2].text = '状态'
|
||||
+
|
||||
+ for field_name, field_value in filled_data.items():
|
||||
+ row_cells = table.add_row().cells
|
||||
+ row_cells[0].text = field_name
|
||||
+ row_cells[1].text = str(field_value) if field_value else ''
|
||||
+ row_cells[2].text = '已填写' if field_value else '为空'
|
||||
+
|
||||
+ # 保存到 BytesIO
|
||||
+ output = io.BytesIO()
|
||||
+ doc.save(output)
|
||||
+ output.seek(0)
|
||||
+
|
||||
+ filename = f"filled_template.docx"
|
||||
+
|
||||
+ return StreamingResponse(
|
||||
+ io.BytesIO(output.getvalue()),
|
||||
+ media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
|
||||
+ headers={"Content-Disposition": f"attachment; filename={filename}"}
|
||||
+ )
|
||||
+
|
||||
+
|
||||
+@router.post("/export/excel")
|
||||
+async def export_to_excel(
|
||||
+ filled_data: dict,
|
||||
+ template_id: str = Query(..., description="模板ID")
|
||||
+):
|
||||
+ """
|
||||
+ 专门导出为 Excel 格式
|
||||
+
|
||||
+ Args:
|
||||
+ filled_data: 填写数据
|
||||
+ template_id: 模板ID
|
||||
+
|
||||
+ Returns:
|
||||
+ Excel 文件流
|
||||
+ """
|
||||
+ return await _export_to_excel(filled_data, template_id)
|
||||
+
|
||||
+
|
||||
+@router.post("/export/word")
|
||||
+async def export_to_word(
|
||||
+ filled_data: dict,
|
||||
+ template_id: str = Query(..., description="模板ID")
|
||||
+):
|
||||
+ """
|
||||
+ 专门导出为 Word 格式
|
||||
+
|
||||
+ Args:
|
||||
+ filled_data: 填写数据
|
||||
+ template_id: 模板ID
|
||||
+
|
||||
+ Returns:
|
||||
+ Word 文件流
|
||||
+ """
|
||||
+ return await _export_to_word(filled_data, template_id)
|
||||
diff --git a/backend/app/core/document_parser/docx_parser.py b/backend/app/core/document_parser/docx_parser.py
|
||||
index 75e79da..03c341d 100644
|
||||
--- a/backend/app/core/document_parser/docx_parser.py
|
||||
+++ b/backend/app/core/document_parser/docx_parser.py
|
||||
@@ -161,3 +161,133 @@ class DocxParser(BaseParser):
|
||||
fields[field_name] = match.group(1)
|
||||
|
||||
return fields
|
||||
+
|
||||
+ def parse_tables_for_template(
|
||||
+ self,
|
||||
+ file_path: str
|
||||
+ ) -> Dict[str, Any]:
|
||||
+ """
|
||||
+ 解析 Word 文档中的表格，提取模板字段
|
||||
+
|
||||
+ 专门用于比赛场景：解析表格模板，识别需要填写的字段
|
||||
+
|
||||
+ Args:
|
||||
+ file_path: Word 文件路径
|
||||
+
|
||||
+ Returns:
|
||||
+ 包含表格字段信息的字典
|
||||
+ """
|
||||
+ from docx import Document
|
||||
+ from docx.table import Table
|
||||
+ from docx.oxml.ns import qn
|
||||
+
|
||||
+ doc = Document(file_path)
|
||||
+
|
||||
+ template_info = {
|
||||
+ "tables": [],
|
||||
+ "fields": [],
|
||||
+ "field_count": 0
|
||||
+ }
|
||||
+
|
||||
+ for table_idx, table in enumerate(doc.tables):
|
||||
+ table_info = {
|
||||
+ "table_index": table_idx,
|
||||
+ "rows": [],
|
||||
+ "headers": [],
|
||||
+ "data_rows": [],
|
||||
+ "field_hints": {} # 字段名称 -> 提示词/描述
|
||||
+ }
|
||||
+
|
||||
+ # 提取表头（第一行）
|
||||
+ if table.rows:
|
||||
+ header_cells = [cell.text.strip() for cell in table.rows[0].cells]
|
||||
+ table_info["headers"] = header_cells
|
||||
+
|
||||
+ # 提取数据行
|
||||
+ for row_idx, row in enumerate(table.rows[1:], 1):
|
||||
+ row_data = [cell.text.strip() for cell in row.cells]
|
||||
+ table_info["data_rows"].append(row_data)
|
||||
+ table_info["rows"].append({
|
||||
+ "row_index": row_idx,
|
||||
+ "cells": row_data
|
||||
+ })
|
||||
+
|
||||
+ # 尝试从第二列/第三列提取提示词
|
||||
+ # 比赛模板通常格式为：字段名 | 提示词 | 填写值
|
||||
+ if len(table.rows[0].cells) >= 2:
|
||||
+ for row_idx, row in enumerate(table.rows[1:], 1):
|
||||
+ cells = [cell.text.strip() for cell in row.cells]
|
||||
+ if len(cells) >= 2 and cells[0]:
|
||||
+ # 第一列是字段名
|
||||
+ field_name = cells[0]
|
||||
+ # 第二列可能是提示词或描述
|
||||
+ hint = cells[1] if len(cells) > 1 else ""
|
||||
+ table_info["field_hints"][field_name] = hint
|
||||
+
|
||||
+ template_info["fields"].append({
|
||||
+ "table_index": table_idx,
|
||||
+ "row_index": row_idx,
|
||||
+ "field_name": field_name,
|
||||
+ "hint": hint,
|
||||
+ "expected_value": cells[2] if len(cells) > 2 else ""
|
||||
+ })
|
||||
+
|
||||
+ template_info["tables"].append(table_info)
|
||||
+
|
||||
+ template_info["field_count"] = len(template_info["fields"])
|
||||
+ return template_info
|
||||
+
|
||||
+ def extract_template_fields_from_docx(
|
||||
+ self,
|
||||
+ file_path: str
|
||||
+ ) -> List[Dict[str, Any]]:
|
||||
+ """
|
||||
+ 从 Word 文档中提取模板字段定义
|
||||
+
|
||||
+ 适用于比赛评分表格：表格第一列是字段名，第二列是提示词/填写示例
|
||||
+
|
||||
+ Args:
|
||||
+ file_path: Word 文件路径
|
||||
+
|
||||
+ Returns:
|
||||
+ 字段定义列表
|
||||
+ """
|
||||
+ template_info = self.parse_tables_for_template(file_path)
|
||||
+
|
||||
+ fields = []
|
||||
+ for field in template_info["fields"]:
|
||||
+ fields.append({
|
||||
+ "cell": f"T{field['table_index']}R{field['row_index']}", # TableXRowY 格式
|
||||
+ "name": field["field_name"],
|
||||
+ "hint": field["hint"],
|
||||
+ "table_index": field["table_index"],
|
||||
+ "row_index": field["row_index"],
|
||||
+ "field_type": self._infer_field_type_from_hint(field["hint"]),
|
||||
+ "required": True
|
||||
+ })
|
||||
+
|
||||
+ return fields
|
||||
+
|
||||
+ def _infer_field_type_from_hint(self, hint: str) -> str:
|
||||
+ """
|
||||
+ 从提示词推断字段类型
|
||||
+
|
||||
+ Args:
|
||||
+ hint: 字段提示词
|
||||
+
|
||||
+ Returns:
|
||||
+ 字段类型 (text/number/date)
|
||||
+ """
|
||||
+ hint_lower = hint.lower()
|
||||
+
|
||||
+ # 日期关键词
|
||||
+ date_keywords = ["年", "月", "日", "日期", "时间", "出生"]
|
||||
+ if any(kw in hint for kw in date_keywords):
|
||||
+ return "date"
|
||||
+
|
||||
+ # 数字关键词
|
||||
+ number_keywords = ["数量", "金额", "人数", "面积", "增长", "比率", "%", "率"]
|
||||
+ if any(kw in hint_lower for kw in number_keywords):
|
||||
+ return "number"
|
||||
+
|
||||
+ return "text"
|
||||
diff --git a/backend/app/services/template_fill_service.py b/backend/app/services/template_fill_service.py
|
||||
index 2612354..94930fb 100644
|
||||
--- a/backend/app/services/template_fill_service.py
|
||||
+++ b/backend/app/services/template_fill_service.py
|
||||
@@ -4,13 +4,12 @@
|
||||
从非结构化文档中检索信息并填写到表格模板
|
||||
"""
|
||||
import logging
|
||||
-from dataclasses import dataclass
|
||||
+from dataclasses import dataclass, field
|
||||
from typing import Any, Dict, List, Optional
|
||||
|
||||
from app.core.database import mongodb
|
||||
-from app.services.rag_service import rag_service
|
||||
from app.services.llm_service import llm_service
|
||||
-from app.services.excel_storage_service import excel_storage_service
|
||||
+from app.core.document_parser import ParserFactory
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@@ -22,6 +21,17 @@ class TemplateField:
|
||||
name: str # 字段名称
|
||||
field_type: str = "text" # 字段类型: text/number/date
|
||||
required: bool = True
|
||||
+ hint: str = "" # 字段提示词
|
||||
+
|
||||
+
|
||||
+@dataclass
|
||||
+class SourceDocument:
|
||||
+ """源文档"""
|
||||
+ doc_id: str
|
||||
+ filename: str
|
||||
+ doc_type: str
|
||||
+ content: str = ""
|
||||
+ structured_data: Dict[str, Any] = field(default_factory=dict)
|
||||
|
||||
|
||||
@dataclass
|
||||
@@ -38,12 +48,12 @@ class TemplateFillService:
|
||||
|
||||
def __init__(self):
|
||||
self.llm = llm_service
|
||||
- self.rag = rag_service
|
||||
|
||||
async def fill_template(
|
||||
self,
|
||||
template_fields: List[TemplateField],
|
||||
source_doc_ids: Optional[List[str]] = None,
|
||||
+ source_file_paths: Optional[List[str]] = None,
|
||||
user_hint: Optional[str] = None
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
@@ -51,7 +61,8 @@ class TemplateFillService:
|
||||
|
||||
Args:
|
||||
template_fields: 模板字段列表
|
||||
- source_doc_ids: 源文档ID列表，不指定则从所有文档检索
|
||||
+ source_doc_ids: 源文档 MongoDB ID 列表
|
||||
+ source_file_paths: 源文档文件路径列表
|
||||
user_hint: 用户提示（如"请从合同文档中提取"）
|
||||
|
||||
Returns:
|
||||
@@ -60,28 +71,23 @@ class TemplateFillService:
|
||||
filled_data = {}
|
||||
fill_details = []
|
||||
|
||||
+ # 1. 加载源文档内容
|
||||
+ source_docs = await self._load_source_documents(source_doc_ids, source_file_paths)
|
||||
+
|
||||
+ if not source_docs:
|
||||
+ logger.warning("没有找到源文档，填表结果将全部为空")
|
||||
+
|
||||
+ # 2. 对每个字段进行提取
|
||||
for field in template_fields:
|
||||
try:
|
||||
- # 1. 从 RAG 检索相关上下文
|
||||
- rag_results = await self._retrieve_context(field.name, user_hint)
|
||||
-
|
||||
- if not rag_results:
|
||||
- # 如果没有检索到结果，尝试直接询问 LLM
|
||||
- result = FillResult(
|
||||
- field=field.name,
|
||||
- value="",
|
||||
- source="未找到相关数据",
|
||||
- confidence=0.0
|
||||
- )
|
||||
- else:
|
||||
- # 2. 构建 Prompt 让 LLM 提取信息
|
||||
- result = await self._extract_field_value(
|
||||
- field=field,
|
||||
- rag_context=rag_results,
|
||||
- user_hint=user_hint
|
||||
- )
|
||||
-
|
||||
- # 3. 存储结果
|
||||
+ # 从源文档中提取字段值
|
||||
+ result = await self._extract_field_value(
|
||||
+ field=field,
|
||||
+ source_docs=source_docs,
|
||||
+ user_hint=user_hint
|
||||
+ )
|
||||
+
|
||||
+ # 存储结果
|
||||
filled_data[field.name] = result.value
|
||||
fill_details.append({
|
||||
"field": field.name,
|
||||
@@ -107,75 +113,113 @@ class TemplateFillService:
|
||||
return {
|
||||
"success": True,
|
||||
"filled_data": filled_data,
|
||||
- "fill_details": fill_details
|
||||
+ "fill_details": fill_details,
|
||||
+ "source_doc_count": len(source_docs)
|
||||
}
|
||||
|
||||
- async def _retrieve_context(
|
||||
+ async def _load_source_documents(
|
||||
self,
|
||||
- field_name: str,
|
||||
- user_hint: Optional[str] = None
|
||||
- ) -> List[Dict[str, Any]]:
|
||||
+ source_doc_ids: Optional[List[str]] = None,
|
||||
+ source_file_paths: Optional[List[str]] = None
|
||||
+ ) -> List[SourceDocument]:
|
||||
"""
|
||||
- 从 RAG 检索相关上下文
|
||||
+ 加载源文档内容
|
||||
|
||||
Args:
|
||||
- field_name: 字段名称
|
||||
- user_hint: 用户提示
|
||||
+ source_doc_ids: MongoDB 文档 ID 列表
|
||||
+ source_file_paths: 源文档文件路径列表
|
||||
|
||||
Returns:
|
||||
- 检索结果列表
|
||||
+ 源文档列表
|
||||
"""
|
||||
- # 构建查询文本
|
||||
- query = field_name
|
||||
- if user_hint:
|
||||
- query = f"{user_hint} {field_name}"
|
||||
-
|
||||
- # 检索相关文档片段
|
||||
- results = self.rag.retrieve(query=query, top_k=5)
|
||||
-
|
||||
- return results
|
||||
+ source_docs = []
|
||||
+
|
||||
+ # 1. 从 MongoDB 加载文档
|
||||
+ if source_doc_ids:
|
||||
+ for doc_id in source_doc_ids:
|
||||
+ try:
|
||||
+ doc = await mongodb.get_document(doc_id)
|
||||
+ if doc:
|
||||
+ source_docs.append(SourceDocument(
|
||||
+ doc_id=doc_id,
|
||||
+ filename=doc.get("metadata", {}).get("original_filename", "unknown"),
|
||||
+ doc_type=doc.get("doc_type", "unknown"),
|
||||
+ content=doc.get("content", ""),
|
||||
+ structured_data=doc.get("structured_data", {})
|
||||
+ ))
|
||||
+ logger.info(f"从MongoDB加载文档: {doc_id}")
|
||||
+ except Exception as e:
|
||||
+ logger.error(f"从MongoDB加载文档失败 {doc_id}: {str(e)}")
|
||||
+
|
||||
+ # 2. 从文件路径加载文档
|
||||
+ if source_file_paths:
|
||||
+ for file_path in source_file_paths:
|
||||
+ try:
|
||||
+ parser = ParserFactory.get_parser(file_path)
|
||||
+ result = parser.parse(file_path)
|
||||
+ if result.success:
|
||||
+ source_docs.append(SourceDocument(
|
||||
+ doc_id=file_path,
|
||||
+ filename=result.metadata.get("filename", file_path.split("/")[-1]),
|
||||
+ doc_type=result.metadata.get("extension", "unknown").replace(".", ""),
|
||||
+ content=result.data.get("content", ""),
|
||||
+ structured_data=result.data.get("structured_data", {})
|
||||
+ ))
|
||||
+ logger.info(f"从文件加载文档: {file_path}")
|
||||
+ except Exception as e:
|
||||
+ logger.error(f"从文件加载文档失败 {file_path}: {str(e)}")
|
||||
+
|
||||
+ return source_docs
|
||||
|
||||
async def _extract_field_value(
|
||||
self,
|
||||
field: TemplateField,
|
||||
- rag_context: List[Dict[str, Any]],
|
||||
+ source_docs: List[SourceDocument],
|
||||
user_hint: Optional[str] = None
|
||||
) -> FillResult:
|
||||
"""
|
||||
- 使用 LLM 从上下文中提取字段值
|
||||
+ 使用 LLM 从源文档中提取字段值
|
||||
|
||||
Args:
|
||||
field: 字段定义
|
||||
- rag_context: RAG 检索到的上下文
|
||||
+ source_docs: 源文档列表
|
||||
user_hint: 用户提示
|
||||
|
||||
Returns:
|
||||
提取结果
|
||||
"""
|
||||
+ if not source_docs:
|
||||
+ return FillResult(
|
||||
+ field=field.name,
|
||||
+ value="",
|
||||
+ source="无源文档",
|
||||
+ confidence=0.0
|
||||
+ )
|
||||
+
|
||||
# 构建上下文文本
|
||||
- context_text = "\n\n".join([
|
||||
- f"【文档 {i+1}】\n{doc['content']}"
|
||||
- for i, doc in enumerate(rag_context)
|
||||
- ])
|
||||
+ context_text = self._build_context_text(source_docs, max_length=8000)
|
||||
+
|
||||
+ # 构建提示词
|
||||
+ hint_text = field.hint if field.hint else f"请提取{field.name}的信息"
|
||||
+ if user_hint:
|
||||
+ hint_text = f"{user_hint}。{hint_text}"
|
||||
|
||||
- # 构建 Prompt
|
||||
- prompt = f"""你是一个数据提取专家。请根据以下文档内容，提取指定字段的信息。
|
||||
+ prompt = f"""你是一个专业的数据提取专家。请根据以下文档内容，提取指定字段的信息。
|
||||
|
||||
需要提取的字段：
|
||||
- 字段名称：{field.name}
|
||||
- 字段类型：{field.field_type}
|
||||
+- 填写提示：{hint_text}
|
||||
- 是否必填：{'是' if field.required else '否'}
|
||||
|
||||
-{'用户提示：' + user_hint if user_hint else ''}
|
||||
-
|
||||
参考文档内容：
|
||||
{context_text}
|
||||
|
||||
请严格按照以下 JSON 格式输出，不要添加任何解释：
|
||||
{{
|
||||
"value": "提取到的值，如果没有找到则填写空字符串",
|
||||
- "source": "数据来源的文档描述",
|
||||
- "confidence": 0.0到1.0之间的置信度
|
||||
+ "source": "数据来源的文档描述（如：来自xxx文档）",
|
||||
+ "confidence": 0.0到1.0之间的置信度，表示对提取结果的信心程度"
|
||||
}}
|
||||
"""
|
||||
|
||||
@@ -226,6 +270,54 @@ class TemplateFillService:
|
||||
confidence=0.0
|
||||
)
|
||||
|
||||
+ def _build_context_text(self, source_docs: List[SourceDocument], max_length: int = 8000) -> str:
|
||||
+ """
|
||||
+ 构建上下文文本
|
||||
+
|
||||
+ Args:
|
||||
+ source_docs: 源文档列表
|
||||
+ max_length: 最大字符数
|
||||
+
|
||||
+ Returns:
|
||||
+ 上下文文本
|
||||
+ """
|
||||
+ contexts = []
|
||||
+ total_length = 0
|
||||
+
|
||||
+ for doc in source_docs:
|
||||
+ # 优先使用结构化数据（表格），其次使用文本内容
|
||||
+ doc_content = ""
|
||||
+
|
||||
+ if doc.structured_data and doc.structured_data.get("tables"):
|
||||
+ # 如果有表格数据，优先使用
|
||||
+ tables = doc.structured_data.get("tables", [])
|
||||
+ for table in tables:
|
||||
+ if isinstance(table, dict):
|
||||
+ rows = table.get("rows", [])
|
||||
+ if rows:
|
||||
+ doc_content += f"\n【文档: {doc.filename} 表格数据】\n"
|
||||
+ for row in rows[:20]: # 限制每表最多20行
|
||||
+ if isinstance(row, list):
|
||||
+ doc_content += " | ".join(str(cell) for cell in row) + "\n"
|
||||
+ elif isinstance(row, dict):
|
||||
+ doc_content += " | ".join(str(v) for v in row.values()) + "\n"
|
||||
+ elif doc.content:
|
||||
+ doc_content = doc.content[:5000] # 限制文本长度
|
||||
+
|
||||
+ if doc_content:
|
||||
+ doc_context = f"【文档: {doc.filename} ({doc.doc_type})】\n{doc_content}"
|
||||
+ if total_length + len(doc_context) <= max_length:
|
||||
+ contexts.append(doc_context)
|
||||
+ total_length += len(doc_context)
|
||||
+ else:
|
||||
+ # 如果超出长度，截断
|
||||
+ remaining = max_length - total_length
|
||||
+ if remaining > 100:
|
||||
+ contexts.append(doc_context[:remaining])
|
||||
+ break
|
||||
+
|
||||
+ return "\n\n".join(contexts) if contexts else "（源文档内容为空）"
|
||||
+
|
||||
async def get_template_fields_from_file(
|
||||
self,
|
||||
file_path: str,
|
||||
@@ -236,7 +328,7 @@ class TemplateFillService:
|
||||
|
||||
Args:
|
||||
file_path: 模板文件路径
|
||||
- file_type: 文件类型
|
||||
+ file_type: 文件类型 (xlsx/xls/docx)
|
||||
|
||||
Returns:
|
||||
字段列表
|
||||
@@ -245,43 +337,108 @@ class TemplateFillService:
|
||||
|
||||
try:
|
||||
if file_type in ["xlsx", "xls"]:
|
||||
- # 从 Excel 读取表头
|
||||
- import pandas as pd
|
||||
- df = pd.read_excel(file_path, nrows=5)
|
||||
+ fields = await self._get_template_fields_from_excel(file_path)
|
||||
+ elif file_type == "docx":
|
||||
+ fields = await self._get_template_fields_from_docx(file_path)
|
||||
|
||||
- for idx, col in enumerate(df.columns):
|
||||
- # 获取单元格位置 (A, B, C, ...)
|
||||
- cell = self._column_to_cell(idx)
|
||||
+ except Exception as e:
|
||||
+ logger.error(f"提取模板字段失败: {str(e)}")
|
||||
|
||||
- fields.append(TemplateField(
|
||||
- cell=cell,
|
||||
- name=str(col),
|
||||
- field_type=self._infer_field_type(df[col]),
|
||||
- required=True
|
||||
- ))
|
||||
+ return fields
|
||||
|
||||
- elif file_type == "docx":
|
||||
- # 从 Word 表格读取
|
||||
- from docx import Document
|
||||
- doc = Document(file_path)
|
||||
-
|
||||
- for table_idx, table in enumerate(doc.tables):
|
||||
- for row_idx, row in enumerate(table.rows):
|
||||
- for col_idx, cell in enumerate(row.cells):
|
||||
- cell_text = cell.text.strip()
|
||||
- if cell_text:
|
||||
- fields.append(TemplateField(
|
||||
- cell=self._column_to_cell(col_idx),
|
||||
- name=cell_text,
|
||||
- field_type="text",
|
||||
- required=True
|
||||
- ))
|
||||
+ async def _get_template_fields_from_excel(self, file_path: str) -> List[TemplateField]:
|
||||
+ """从 Excel 模板提取字段"""
|
||||
+ fields = []
|
||||
+
|
||||
+ try:
|
||||
+ import pandas as pd
|
||||
+ df = pd.read_excel(file_path, nrows=5)
|
||||
+
|
||||
+ for idx, col in enumerate(df.columns):
|
||||
+ cell = self._column_to_cell(idx)
|
||||
+ col_str = str(col)
|
||||
+
|
||||
+ fields.append(TemplateField(
|
||||
+ cell=cell,
|
||||
+ name=col_str,
|
||||
+ field_type=self._infer_field_type_from_value(df[col].iloc[0] if len(df) > 0 else ""),
|
||||
+ required=True,
|
||||
+ hint=""
|
||||
+ ))
|
||||
|
||||
except Exception as e:
|
||||
- logger.error(f"提取模板字段失败: {str(e)}")
|
||||
+ logger.error(f"从Excel提取字段失败: {str(e)}")
|
||||
|
||||
return fields
|
||||
|
||||
+ async def _get_template_fields_from_docx(self, file_path: str) -> List[TemplateField]:
|
||||
+ """从 Word 模板提取字段"""
|
||||
+ fields = []
|
||||
+
|
||||
+ try:
|
||||
+ from docx import Document
|
||||
+
|
||||
+ doc = Document(file_path)
|
||||
+
|
||||
+ for table_idx, table in enumerate(doc.tables):
|
||||
+ for row_idx, row in enumerate(table.rows):
|
||||
+ cells = [cell.text.strip() for cell in row.cells]
|
||||
+
|
||||
+ # 假设第一列是字段名
|
||||
+ if cells and cells[0]:
|
||||
+ field_name = cells[0]
|
||||
+ hint = cells[1] if len(cells) > 1 else ""
|
||||
+
|
||||
+ # 跳过空行或标题行
|
||||
+ if field_name and field_name not in ["", "字段名", "名称", "项目"]:
|
||||
+ fields.append(TemplateField(
|
||||
+ cell=f"T{table_idx}R{row_idx}",
|
||||
+ name=field_name,
|
||||
+ field_type=self._infer_field_type_from_hint(hint),
|
||||
+ required=True,
|
||||
+ hint=hint
|
||||
+ ))
|
||||
+
|
||||
+ except Exception as e:
|
||||
+ logger.error(f"从Word提取字段失败: {str(e)}")
|
||||
+
|
||||
+ return fields
|
||||
+
|
||||
+ def _infer_field_type_from_hint(self, hint: str) -> str:
|
||||
+ """从提示词推断字段类型"""
|
||||
+ hint_lower = hint.lower()
|
||||
+
|
||||
+ date_keywords = ["年", "月", "日", "日期", "时间", "出生"]
|
||||
+ if any(kw in hint for kw in date_keywords):
|
||||
+ return "date"
|
||||
+
|
||||
+ number_keywords = ["数量", "金额", "人数", "面积", "增长", "比率", "%", "率", "总计", "合计"]
|
||||
+ if any(kw in hint_lower for kw in number_keywords):
|
||||
+ return "number"
|
||||
+
|
||||
+ return "text"
|
||||
+
|
||||
+ def _infer_field_type_from_value(self, value: Any) -> str:
|
||||
+ """从示例值推断字段类型"""
|
||||
+ if value is None or value == "":
|
||||
+ return "text"
|
||||
+
|
||||
+ value_str = str(value)
|
||||
+
|
||||
+ # 检查日期模式
|
||||
+ import re
|
||||
+ if re.search(r'\d{4}[年/-]\d{1,2}[月/-]\d{1,2}', value_str):
|
||||
+ return "date"
|
||||
+
|
||||
+ # 检查数值
|
||||
+ try:
|
||||
+ float(value_str.replace(',', '').replace('%', ''))
|
||||
+ return "number"
|
||||
+ except ValueError:
|
||||
+ pass
|
||||
+
|
||||
+ return "text"
|
||||
+
|
||||
def _column_to_cell(self, col_idx: int) -> str:
|
||||
"""将列索引转换为单元格列名 (0 -> A, 1 -> B, ...)"""
|
||||
result = ""
|
||||
@@ -290,17 +447,6 @@ class TemplateFillService:
|
||||
col_idx = col_idx // 26 - 1
|
||||
return result
|
||||
|
||||
- def _infer_field_type(self, series) -> str:
|
||||
- """推断字段类型"""
|
||||
- import pandas as pd
|
||||
-
|
||||
- if pd.api.types.is_numeric_dtype(series):
|
||||
- return "number"
|
||||
- elif pd.api.types.is_datetime64_any_dtype(series):
|
||||
- return "date"
|
||||
- else:
|
||||
- return "text"
|
||||
-
|
||||
|
||||
# ==================== 全局单例 ====================
|
||||
|
||||
@@ -1,53 +0,0 @@
|
||||
diff --git a/frontend/src/db/backend-api.ts b/frontend/src/db/backend-api.ts
|
||||
index 8944353..94ac852 100644
|
||||
--- a/frontend/src/db/backend-api.ts
|
||||
+++ b/frontend/src/db/backend-api.ts
|
||||
@@ -92,6 +92,7 @@ export interface TemplateField {
|
||||
name: string;
|
||||
field_type: string;
|
||||
required: boolean;
|
||||
+ hint?: string;
|
||||
}
|
||||
|
||||
// 表格填写结果
|
||||
@@ -625,7 +626,10 @@ export const backendApi = {
|
||||
*/
|
||||
async fillTemplate(
|
||||
templateId: string,
|
||||
- templateFields: TemplateField[]
|
||||
+ templateFields: TemplateField[],
|
||||
+ sourceDocIds?: string[],
|
||||
+ sourceFilePaths?: string[],
|
||||
+ userHint?: string
|
||||
): Promise<FillResult> {
|
||||
const url = `${BACKEND_BASE_URL}/templates/fill`;
|
||||
|
||||
@@ -636,6 +640,9 @@ export const backendApi = {
|
||||
body: JSON.stringify({
|
||||
template_id: templateId,
|
||||
template_fields: templateFields,
|
||||
+ source_doc_ids: sourceDocIds || [],
|
||||
+ source_file_paths: sourceFilePaths || [],
|
||||
+ user_hint: userHint || null,
|
||||
}),
|
||||
});
|
||||
|
||||
diff --git a/frontend/src/pages/TemplateFill.tsx b/frontend/src/pages/TemplateFill.tsx
|
||||
index 8c330a9..f9a4a39 100644
|
||||
--- a/frontend/src/pages/TemplateFill.tsx
|
||||
+++ b/frontend/src/pages/TemplateFill.tsx
|
||||
@@ -128,8 +128,12 @@ const TemplateFill: React.FC = () => {
|
||||
setStep('filling');
|
||||
|
||||
try {
|
||||
- // 调用后端填表接口
|
||||
- const result = await backendApi.fillTemplate('temp-template-id', templateFields);
|
||||
+ // 调用后端填表接口，传递选中的文档ID
|
||||
+ const result = await backendApi.fillTemplate(
|
||||
+ 'temp-template-id',
|
||||
+ templateFields,
|
||||
+ selectedDocs // 传递源文档ID列表
|
||||
+ );
|
||||
setFilledResult(result);
|
||||
setStep('preview');
|
||||
toast.success('表格填写完成');
|
||||
@@ -1,221 +0,0 @@
|
||||
diff --git "a/\346\257\224\350\265\233\345\244\207\350\265\233\350\247\204\345\210\222.md" "b/\346\257\224\350\265\233\345\244\207\350\265\233\350\247\204\345\210\222.md"
|
||||
index bcb48fd..440a12d 100644
|
||||
--- "a/\346\257\224\350\265\233\345\244\207\350\265\233\350\247\204\345\210\222.md"
|
||||
+++ "b/\346\257\224\350\265\233\345\244\207\350\265\233\350\247\204\345\210\222.md"
|
||||
@@ -50,7 +50,7 @@
|
||||
| `prompt_service.py` | ✅ 已完成 | Prompt 模板管理 |
|
||||
| `text_analysis_service.py` | ✅ 已完成 | 文本分析 |
|
||||
| `chart_generator_service.py` | ✅ 已完成 | 图表生成服务 |
|
||||
-| `template_fill_service.py` | ❌ 未完成 | 模板填写服务 |
|
||||
+| `template_fill_service.py` | ✅ 已完成 | 模板填写服务，支持直接读取源文档进行填表 |
|
||||
|
||||
### 2.2 API 接口 (`backend/app/api/endpoints/`)
|
||||
|
||||
@@ -61,7 +61,7 @@
|
||||
| `ai_analyze.py` | `/api/v1/analyze/*` | ✅ AI 分析（Excel、Markdown、流式） |
|
||||
| `rag.py` | `/api/v1/rag/*` | ⚠️ RAG 检索（当前返回空） |
|
||||
| `tasks.py` | `/api/v1/tasks/*` | ✅ 异步任务状态查询 |
|
||||
-| `templates.py` | `/api/v1/templates/*` | ✅ 模板管理 |
|
||||
+| `templates.py` | `/api/v1/templates/*` | ✅ 模板管理 (含 Word 导出) |
|
||||
| `visualization.py` | `/api/v1/visualization/*` | ✅ 可视化图表 |
|
||||
| `health.py` | `/api/v1/health` | ✅ 健康检查 |
|
||||
|
||||
@@ -78,8 +78,8 @@
|
||||
|------|----------|------|
|
||||
| Excel (.xlsx/.xls) | ✅ 已完成 | pandas + XML 回退解析 |
|
||||
| Markdown (.md) | ✅ 已完成 | 正则 + AI 分章节 |
|
||||
-| Word (.docx) | ❌ 未完成 | 尚未实现 |
|
||||
-| Text (.txt) | ❌ 未完成 | 尚未实现 |
|
||||
+| Word (.docx) | ✅ 已完成 | python-docx 解析，支持表格提取和字段识别 |
|
||||
+| Text (.txt) | ✅ 已完成 | chardet 编码检测，支持文本清洗和结构化提取 |
|
||||
|
||||
---
|
||||
|
||||
@@ -87,7 +87,7 @@
|
||||
|
||||
### 3.1 模板填写模块（最优先）
|
||||
|
||||
-**这是比赛的核心评测功能，必须完成。**
|
||||
+**当前状态**：✅ 已完成
|
||||
|
||||
```
|
||||
用户上传模板表格(Word/Excel)
|
||||
@@ -103,30 +103,34 @@ AI 根据字段提示词从源数据中提取信息
|
||||
返回填写完成的表格
|
||||
```
|
||||
|
||||
-**需要实现**：
|
||||
-- [ ] `template_fill_service.py` - 模板填写核心服务
|
||||
-- [ ] Word 模板解析 (`docx_parser.py` 需新建)
|
||||
-- [ ] Text 模板解析 (`txt_parser.py` 需新建)
|
||||
-- [ ] 模板字段识别与提示词提取
|
||||
-- [ ] 多文档数据聚合与冲突处理
|
||||
-- [ ] 结果导出为 Word/Excel
|
||||
+**已完成实现**：
|
||||
+- [x] `template_fill_service.py` - 模板填写核心服务
|
||||
+- [x] Word 模板解析 (`docx_parser.py` - parse_tables_for_template, extract_template_fields_from_docx)
|
||||
+- [x] Text 模板解析 (`txt_parser.py` - 已完成)
|
||||
+- [x] 模板字段识别与提示词提取
|
||||
+- [x] 多文档数据聚合与冲突处理
|
||||
+- [x] 结果导出为 Word/Excel
|
||||
|
||||
### 3.2 Word 文档解析
|
||||
|
||||
-**当前状态**：仅有框架，尚未实现具体解析逻辑
|
||||
+**当前状态**：✅ 已完成
|
||||
|
||||
-**需要实现**：
|
||||
-- [ ] `docx_parser.py` - Word 文档解析器
|
||||
-- [ ] 提取段落文本
|
||||
-- [ ] 提取表格内容
|
||||
-- [ ] 提取关键信息（标题、列表等）
|
||||
+**已实现功能**：
|
||||
+- [x] `docx_parser.py` - Word 文档解析器
|
||||
+- [x] 提取段落文本
|
||||
+- [x] 提取表格内容
|
||||
+- [x] 提取关键信息（标题、列表等）
|
||||
+- [x] 表格模板字段提取 (`parse_tables_for_template`, `extract_template_fields_from_docx`)
|
||||
+- [x] 字段类型推断 (`_infer_field_type_from_hint`)
|
||||
|
||||
### 3.3 Text 文档解析
|
||||
|
||||
-**需要实现**：
|
||||
-- [ ] `txt_parser.py` - 文本文件解析器
|
||||
-- [ ] 编码自动检测
|
||||
-- [ ] 文本清洗
|
||||
+**当前状态**：✅ 已完成
|
||||
+
|
||||
+**已实现功能**：
|
||||
+- [x] `txt_parser.py` - 文本文件解析器
|
||||
+- [x] 编码自动检测 (chardet)
|
||||
+- [x] 文本清洗
|
||||
|
||||
### 3.4 文档模板匹配（已有框架）
|
||||
|
||||
@@ -215,5 +219,122 @@ docs/test/
|
||||
|
||||
---
|
||||
|
||||
-*文档版本: v1.0*
|
||||
-*最后更新: 2026-04-08*
|
||||
\ No newline at end of file
|
||||
+*文档版本: v1.1*
|
||||
+*最后更新: 2026-04-08*
|
||||
+
|
||||
+---
|
||||
+
|
||||
+## 八、技术实现细节
|
||||
+
|
||||
+### 8.1 模板填表流程（已实现）
|
||||
+
|
||||
+#### 流程图
|
||||
+```
|
||||
+┌─────────────┐ ┌─────────────┐ ┌─────────────┐
|
||||
+│ 上传模板 │ ──► │ 选择数据源 │ ──► │ AI 智能填表 │
|
||||
+└─────────────┘ └─────────────┘ └─────────────┘
|
||||
+ │
|
||||
+ ▼
|
||||
+ ┌─────────────┐
|
||||
+ │ 导出结果 │
|
||||
+ └─────────────┘
|
||||
+```
|
||||
+
|
||||
+#### 核心组件
|
||||
+
|
||||
+| 组件 | 文件 | 说明 |
|
||||
+|------|------|------|
|
||||
+| 模板上传 | `templates.py` `/templates/upload` | 接收模板文件，提取字段 |
|
||||
+| 字段提取 | `template_fill_service.py` | 从 Word/Excel 表格提取字段定义 |
|
||||
+| 文档解析 | `docx_parser.py`, `xlsx_parser.py`, `txt_parser.py` | 解析源文档内容 |
|
||||
+| 智能填表 | `template_fill_service.py` `fill_template()` | 使用 LLM 从源文档提取信息 |
|
||||
+| 结果导出 | `templates.py` `/templates/export` | 导出为 Excel 或 Word |
|
||||
+
|
||||
+### 8.2 源文档加载方式
|
||||
+
|
||||
+模板填表服务支持两种方式加载源文档：
|
||||
+
|
||||
+1. **通过 MongoDB 文档 ID**：`source_doc_ids`
|
||||
+ - 文档已上传并存入 MongoDB
|
||||
+ - 服务直接查询 MongoDB 获取文档内容
|
||||
+
|
||||
+2. **通过文件路径**：`source_file_paths`
|
||||
+ - 直接读取本地文件
|
||||
+ - 使用对应的解析器解析内容
|
||||
+
|
||||
+### 8.3 Word 表格模板解析
|
||||
+
|
||||
+比赛评分表格通常是 Word 格式，`docx_parser.py` 提供了专门的解析方法：
|
||||
+
|
||||
+```python
|
||||
+# 提取表格模板字段
|
||||
+fields = docx_parser.extract_template_fields_from_docx(file_path)
|
||||
+
|
||||
+# 返回格式
|
||||
+# [
|
||||
+# {
|
||||
+# "cell": "T0R1", # 表格0，行1
|
||||
+# "name": "字段名",
|
||||
+# "hint": "提示词",
|
||||
+# "field_type": "text/number/date",
|
||||
+# "required": True
|
||||
+# },
|
||||
+# ...
|
||||
+# ]
|
||||
+```
|
||||
+
|
||||
+### 8.4 字段类型推断
|
||||
+
|
||||
+系统支持从提示词自动推断字段类型：
|
||||
+
|
||||
+| 关键词 | 推断类型 | 示例 |
|
||||
+|--------|----------|------|
|
||||
+| 年、月、日、日期、时间、出生 | date | 出生日期 |
|
||||
+| 数量、金额、比率、%、率、合计 | number | 增长比率 |
|
||||
+| 其他 | text | 姓名、地址 |
|
||||
+
|
||||
+### 8.5 API 接口
|
||||
+
|
||||
+#### POST `/api/v1/templates/fill`
|
||||
+
|
||||
+填写请求：
|
||||
+```json
|
||||
+{
|
||||
+ "template_id": "模板ID",
|
||||
+ "template_fields": [
|
||||
+ {"cell": "A1", "name": "姓名", "field_type": "text", "required": true, "hint": "提取人员姓名"}
|
||||
+ ],
|
||||
+ "source_doc_ids": ["mongodb_doc_id_1", "mongodb_doc_id_2"],
|
||||
+ "source_file_paths": [],
|
||||
+ "user_hint": "请从合同文档中提取"
|
||||
+}
|
||||
+```
|
||||
+
|
||||
+响应：
|
||||
+```json
|
||||
+{
|
||||
+ "success": true,
|
||||
+ "filled_data": {"姓名": "张三"},
|
||||
+ "fill_details": [
|
||||
+ {
|
||||
+ "field": "姓名",
|
||||
+ "cell": "A1",
|
||||
+ "value": "张三",
|
||||
+ "source": "来自：合同文档.docx",
|
||||
+ "confidence": 0.95
|
||||
+ }
|
||||
+ ],
|
||||
+ "source_doc_count": 2
|
||||
+}
|
||||
+```
|
||||
+
|
||||
+#### POST `/api/v1/templates/export`
|
||||
+
|
||||
+导出请求：
|
||||
+```json
|
||||
+{
|
||||
+ "template_id": "模板ID",
|
||||
+ "filled_data": {"姓名": "张三", "金额": "10000"},
|
||||
+ "format": "xlsx" // 或 "docx"
|
||||
+}
|
||||
+```
|
||||
\ No newline at end of file
|
||||
@@ -1,59 +0,0 @@
|
||||
RAG 服务临时禁用说明
|
||||
========================
|
||||
日期: 2026-04-08
|
||||
|
||||
修改内容:
|
||||
----------
|
||||
应需求，RAG 向量检索功能已临时禁用，具体如下:
|
||||
|
||||
1. 修改文件: backend/app/services/rag_service.py
|
||||
|
||||
2. 关键变更:
|
||||
- 在 RAGService.__init__ 中添加 self._disabled = True 标志
|
||||
- index_field() - 添加 _disabled 检查，跳过实际索引操作并记录日志
|
||||
- index_document_content() - 添加 _disabled 检查，跳过实际索引操作并记录日志
|
||||
- retrieve() - 添加 _disabled 检查，返回空列表并记录日志
|
||||
- get_vector_count() - 添加 _disabled 检查，返回 0 并记录日志
|
||||
- clear() - 添加 _disabled 检查，跳过实际清空操作并记录日志
|
||||
|
||||
3. 行为变更:
|
||||
- 所有 RAG 索引构建操作会被记录到日志 ([RAG DISABLED] 前缀)
|
||||
- 所有 RAG 检索操作返回空结果
|
||||
- 向量计数始终返回 0
|
||||
- 实际向量数据库操作被跳过
|
||||
|
||||
4. 恢复方式:
|
||||
- 将 RAGService.__init__ 中的 self._disabled = True 改为 self._disabled = False
|
||||
- 重新启动服务即可恢复 RAG 功能
|
||||
|
||||
目的:
|
||||
------
|
||||
保留 RAG 索引构建功能的前端界面和代码结构，暂不实际调用向量数据库 API，
|
||||
待后续需要时再启用。
|
||||
|
||||
影响范围:
|
||||
---------
|
||||
- /api/v1/rag/search - RAG 搜索接口 (返回空结果)
|
||||
- /api/v1/rag/status - RAG 状态接口 (返回 vector_count=0)
|
||||
- /api/v1/rag/rebuild - RAG 重建接口 (仅记录日志)
|
||||
- Excel/文档上传时的 RAG 索引构建 (仅记录日志)
|
||||
|
||||
========================
|
||||
后续补充 (2026-04-08):
|
||||
========================
|
||||
修改文件: backend/app/services/table_rag_service.py
|
||||
|
||||
关键变更:
|
||||
- 在 TableRAGService.__init__ 中添加 self._disabled = True 标志
|
||||
- build_table_rag_index() - RAG 索引部分被跳过，仅记录日志
|
||||
- index_document_table() - RAG 索引部分被跳过，仅记录日志
|
||||
|
||||
行为变更:
|
||||
- Excel 上传时，MySQL 存储仍然正常进行
|
||||
- AI 字段描述仍然正常生成（调用 LLM）
|
||||
- 只有向量数据库索引操作被跳过
|
||||
|
||||
恢复方式:
|
||||
- 将 TableRAGService.__init__ 中的 self._disabled = True 改为 self._disabled = False
|
||||
- 或将 rag_service.py 中的 self._disabled = True 改为 self._disabled = False
|
||||
- 两者需同时改为 False 才能完全恢复 RAG 功能
|
||||
@@ -1,144 +0,0 @@
|
||||
# 模板填表功能变更日志
|
||||
|
||||
**变更日期**: 2026-04-08
|
||||
**变更类型**: 功能完善
|
||||
**变更内容**: Word 表格解析和模板填表功能
|
||||
|
||||
---
|
||||
|
||||
## 变更概述
|
||||
|
||||
本次变更完善了 Word 表格解析、表格模板构建和填写功能，实现了从源文档（MongoDB/文件）读取数据并智能填表的核心流程。
|
||||
|
||||
### 涉及文件
|
||||
|
||||
| 文件 | 变更行数 | 说明 |
|
||||
|------|----------|------|
|
||||
| backend/app/api/endpoints/templates.py | +156 | API 端点完善，添加 Word 导出 |
|
||||
| backend/app/core/document_parser/docx_parser.py | +130 | Word 表格解析增强 |
|
||||
| backend/app/services/template_fill_service.py | +340 | 核心填表服务重写 |
|
||||
| frontend/src/db/backend-api.ts | +9 | 前端 API 更新 |
|
||||
| frontend/src/pages/TemplateFill.tsx | +8 | 前端页面更新 |
|
||||
| 比赛备赛规划.md | +169 | 文档更新 |
|
||||
|
||||
---
|
||||
|
||||
## 详细变更
|
||||
|
||||
### 1. backend/app/core/document_parser/docx_parser.py
|
||||
|
||||
**新增方法**:
|
||||
|
||||
- `parse_tables_for_template(file_path)` - 解析 Word 文档中的表格，提取模板字段
|
||||
- `extract_template_fields_from_docx(file_path)` - 从 Word 文档提取模板字段定义
|
||||
- `_infer_field_type_from_hint(hint)` - 从提示词推断字段类型
|
||||
|
||||
**功能说明**:
|
||||
- 专门用于比赛场景：解析表格模板，识别需要填写的字段
|
||||
- 支持从表格第一列提取字段名，第二列提取提示词/描述
|
||||
- 自动推断字段类型（text/number/date）
|
||||
|
||||
### 2. backend/app/services/template_fill_service.py
|
||||
|
||||
**重构内容**:
|
||||
|
||||
- 不再依赖 RAG 服务，直接从 MongoDB 或文件读取源文档
|
||||
- 新增 `SourceDocument` 数据类
|
||||
- 完善 `fill_template()` 方法，支持 `source_doc_ids` 和 `source_file_paths`
|
||||
- 新增 `_load_source_documents()` - 加载源文档内容
|
||||
- 新增 `_extract_field_value()` - 使用 LLM 提取字段值
|
||||
- 新增 `_build_context_text()` - 构建上下文（优先使用表格数据）
|
||||
- 完善 `_get_template_fields_from_docx()` - Word 模板字段提取
|
||||
|
||||
**核心流程**:
|
||||
```
|
||||
1. 加载源文档（MongoDB 或文件）
|
||||
2. 对每个字段调用 LLM 提取值
|
||||
3. 返回填写结果
|
||||
```
|
||||
|
||||
### 3. backend/app/api/endpoints/templates.py
|
||||
|
||||
**新增内容**:
|
||||
|
||||
- `FillRequest` 添加 `source_doc_ids`, `source_file_paths`, `user_hint` 字段
|
||||
- `ExportRequest` 添加 `format` 字段
|
||||
- `_export_to_word()` - 导出为 Word 格式
|
||||
- `/templates/export/excel` - 专门导出 Excel
|
||||
- `/templates/export/word` - 专门导出 Word
|
||||
|
||||
### 4. frontend/src/db/backend-api.ts
|
||||
|
||||
**更新内容**:
|
||||
|
||||
- `TemplateField` 接口添加 `hint` 字段
|
||||
- `fillTemplate()` 方法添加 `sourceDocIds`, `sourceFilePaths`, `userHint` 参数
|
||||
|
||||
### 5. frontend/src/pages/TemplateFill.tsx
|
||||
|
||||
**更新内容**:
|
||||
|
||||
- `handleFillTemplate()` 传递 `selectedDocs` 作为 `sourceDocIds` 参数
|
||||
|
||||
---
|
||||
|
||||
## API 接口变更
|
||||
|
||||
### POST /api/v1/templates/fill
|
||||
|
||||
**请求体**:
|
||||
```json
|
||||
{
|
||||
"template_id": "模板ID",
|
||||
"template_fields": [
|
||||
{
|
||||
"cell": "A1",
|
||||
"name": "姓名",
|
||||
"field_type": "text",
|
||||
"required": true,
|
||||
"hint": "提取人员姓名"
|
||||
}
|
||||
],
|
||||
"source_doc_ids": ["mongodb_doc_id"],
|
||||
"source_file_paths": [],
|
||||
"user_hint": "请从xxx文档中提取"
|
||||
}
|
||||
```
|
||||
|
||||
**响应**:
|
||||
```json
|
||||
{
|
||||
"success": true,
|
||||
"filled_data": {"姓名": "张三"},
|
||||
"fill_details": [...],
|
||||
"source_doc_count": 1
|
||||
}
|
||||
```
|
||||
|
||||
### POST /api/v1/templates/export
|
||||
|
||||
**新增支持 format=dicx**，可导出为 Word 格式
|
||||
|
||||
---
|
||||
|
||||
## 技术细节
|
||||
|
||||
### 字段类型推断
|
||||
|
||||
| 关键词 | 推断类型 |
|
||||
|--------|----------|
|
||||
| 年、月、日、日期、时间、出生 | date |
|
||||
| 数量、金额、比率、%、率、合计 | number |
|
||||
| 其他 | text |
|
||||
|
||||
### 上下文构建
|
||||
|
||||
源文档内容构建优先级：
|
||||
1. 结构化数据（表格数据）
|
||||
2. 原始文本内容（限制 5000 字符）
|
||||
|
||||
---
|
||||
|
||||
## 相关文档
|
||||
|
||||
- [比赛备赛规划.md](../比赛备赛规划.md) - 已更新功能状态和技术实现细节
|
||||
20
package.json
20
package.json
@@ -1,20 +0,0 @@
|
||||
{
|
||||
"name": "filesreadsystem",
|
||||
"version": "1.0.0",
|
||||
"description": "",
|
||||
"main": "index.js",
|
||||
"directories": {
|
||||
"doc": "docs"
|
||||
},
|
||||
"scripts": {
|
||||
"test": "echo \"Error: no test specified\" && exit 1"
|
||||
},
|
||||
"repository": {
|
||||
"type": "git",
|
||||
"url": "https://gitea.kronecker.cc/OurCodesAreAllRight/FilesReadSystem.git"
|
||||
},
|
||||
"keywords": [],
|
||||
"author": "",
|
||||
"license": "ISC",
|
||||
"type": "commonjs"
|
||||
}
|
||||
BIN
屏幕截图 2026-04-18 002609.png
Normal file
BIN
屏幕截图 2026-04-18 002609.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 552 KiB |
558
比赛备赛规划.md
558
比赛备赛规划.md
@@ -1,558 +0,0 @@
|
||||
# 比赛备赛规划文档
|
||||
|
||||
## 一、赛题核心理解
|
||||
|
||||
### 1.1 赛题名称
|
||||
**A23 - 基于大语言模型的文档理解与多源数据融合**
|
||||
参赛院校：金陵科技学院
|
||||
|
||||
### 1.2 核心任务
|
||||
1. **文档解析**：解析 docx/md/xlsx/txt 四种格式的源数据文档
|
||||
2. **模板填写**：根据模板表格要求，从源文档中提取数据填写到 Word/Excel 模板
|
||||
3. **准确率与速度**：准确率优先，速度作为辅助评分因素
|
||||
|
||||
### 1.3 评分规则
|
||||
| 要素 | 说明 |
|
||||
|------|------|
|
||||
| 准确率 | 填写结果与样例表格对比的正确率 |
|
||||
| 响应时间 | 从导入文档到得到结果的时间 ≤ 90s × 文档数量 |
|
||||
| 评测方式 | 赛方提供空表格模板 + 样例表格（人工填写），系统自动填写后对比 |
|
||||
|
||||
### 1.4 关键Q&A摘录
|
||||
|
||||
| 问题 | 解答要点 |
|
||||
|------|----------|
|
||||
| Q2: 模板与文档的关系 | 前2个表格只涉及1份文档；第3-4个涉及多份文档；第5个涉及大部分文档（从易到难） |
|
||||
| Q5: 响应时间定义 | 从导入文档到最终得到结果的时间 ≤ 90s × 文档数量 |
|
||||
| Q7: 需要读取哪些文件 | 每个模板只读取指定的数据文件，不需要读取全部 |
|
||||
| Q10: 部署方式 | 不要求部署到服务器，本地部署即可 |
|
||||
| Q14: 模板匹配 | 模板已指定数据文件，不需要算法匹配 |
|
||||
| Q16: 数据库存储 | 可跳过，不强制要求 |
|
||||
| Q20: 创新点 | 不用管，随意发挥 |
|
||||
| Q21: 填写依据 | 按照测试表格模板给的提示词进行填写 |
|
||||
|
||||
---
|
||||
|
||||
## 二、已完成功能清单
|
||||
|
||||
### 2.1 后端服务 (`backend/app/services/`)
|
||||
|
||||
| 服务文件 | 功能状态 | 说明 |
|
||||
|----------|----------|------|
|
||||
| `file_service.py` | ✅ 已完成 | 文件上传、保存、类型识别 |
|
||||
| `excel_storage_service.py` | ✅ 已完成 | Excel 存储到 MySQL，支持 XML 回退解析 |
|
||||
| `table_rag_service.py` | ⚠️ 已禁用 | RAG 索引构建（当前禁用，仅记录日志） |
|
||||
| `llm_service.py` | ✅ 已完成 | LLM 调用、流式输出、多模型支持 |
|
||||
| `markdown_ai_service.py` | ✅ 已完成 | Markdown AI 分析、分章节提取、流式输出、图表生成 |
|
||||
| `excel_ai_service.py` | ✅ 已完成 | Excel AI 分析 |
|
||||
| `visualization_service.py` | ✅ 已完成 | 图表生成（matplotlib） |
|
||||
| `rag_service.py` | ⚠️ 已禁用 | FAISS 向量检索（当前禁用） |
|
||||
| `prompt_service.py` | ✅ 已完成 | Prompt 模板管理 |
|
||||
| `text_analysis_service.py` | ✅ 已完成 | 文本分析 |
|
||||
| `chart_generator_service.py` | ✅ 已完成 | 图表生成服务 |
|
||||
| `template_fill_service.py` | ✅ 已完成 | 模板填写服务，支持多行提取、直接从结构化数据提取、JSON容错、Word文档表格处理 |
|
||||
|
||||
### 2.2 API 接口 (`backend/app/api/endpoints/`)
|
||||
|
||||
| 接口文件 | 路由 | 功能状态 |
|
||||
|----------|------|----------|
|
||||
| `upload.py` | `/api/v1/upload/document` | ✅ 文档上传与解析 |
|
||||
| `documents.py` | `/api/v1/documents/*` | ✅ 文档管理（列表、删除、搜索） |
|
||||
| `ai_analyze.py` | `/api/v1/analyze/*` | ✅ AI 分析（Excel、Markdown、流式） |
|
||||
| `rag.py` | `/api/v1/rag/*` | ⚠️ RAG 检索（当前返回空） |
|
||||
| `tasks.py` | `/api/v1/tasks/*` | ✅ 异步任务状态查询 |
|
||||
| `templates.py` | `/api/v1/templates/*` | ✅ 模板管理（含多行导出、Word导出、Word结构化字段解析） |
|
||||
| `visualization.py` | `/api/v1/visualization/*` | ✅ 可视化图表 |
|
||||
| `health.py` | `/api/v1/health` | ✅ 健康检查 |
|
||||
|
||||
### 2.3 前端页面 (`frontend/src/pages/`)
|
||||
|
||||
| 页面文件 | 功能 | 状态 |
|
||||
|----------|------|------|
|
||||
| `Documents.tsx` | 主文档管理页面 | ✅ 已完成 |
|
||||
| `TemplateFill.tsx` | 智能填表页面 | ✅ 已完成 |
|
||||
| `ExcelParse.tsx` | Excel 解析页面 | ✅ 已完成 |
|
||||
|
||||
### 2.4 文档解析能力
|
||||
|
||||
| 格式 | 解析状态 | 说明 |
|
||||
|------|----------|------|
|
||||
| Excel (.xlsx/.xls) | ✅ 已完成 | pandas + XML 回退解析，支持多sheet |
|
||||
| Markdown (.md) | ✅ 已完成 | 正则 + AI 分章节 |
|
||||
| Word (.docx) | ✅ 已完成 | python-docx 解析，支持表格提取和字段识别 |
|
||||
| Text (.txt) | ✅ 已完成 | chardet 编码检测，支持文本清洗和结构化提取 |
|
||||
|
||||
---
|
||||
|
||||
## 三、核心功能实现详情
|
||||
|
||||
### 3.1 模板填写模块（✅ 已完成）
|
||||
|
||||
**核心流程**：
|
||||
```
|
||||
上传模板表格(Word/Excel)
|
||||
↓
|
||||
解析模板，提取需要填写的字段和提示词
|
||||
↓
|
||||
根据源文档ID列表读取源数据（MongoDB或文件）
|
||||
↓
|
||||
优先从结构化数据直接提取（Excel rows）
|
||||
↓
|
||||
无法直接提取时使用 LLM 从文本中提取
|
||||
↓
|
||||
将提取的数据填入原始模板对应位置（保持模板格式）
|
||||
↓
|
||||
导出填写完成的表格（Excel/Word）
|
||||
```
|
||||
|
||||
**关键特性**：
|
||||
- **原始模板填充**：直接打开原始模板文件，填充数据到原表格/单元格
|
||||
- **多行数据支持**：每个字段可提取多个值，导出时自动扩展行数
|
||||
- **结构化数据优先**：直接从 Excel rows 提取，无需 LLM
|
||||
- **JSON 容错**：支持 LLM 返回的损坏/截断 JSON
|
||||
- **Markdown 清理**：自动清理 LLM 返回的 markdown 格式
|
||||
|
||||
### 3.2 Word 文档解析（✅ 已完成）
|
||||
|
||||
**已实现功能**：
|
||||
- `docx_parser.py` - Word 文档解析器
|
||||
- 提取段落文本
|
||||
- 提取表格内容（支持比赛表格格式：字段名 | 提示词 | 填写值）
|
||||
- `parse_tables_for_template()` - 解析表格模板，提取字段
|
||||
- `extract_template_fields_from_docx()` - 提取模板字段定义
|
||||
- `_infer_field_type_from_hint()` - 从提示词推断字段类型
|
||||
- **API 端点**：`/api/v1/templates/parse-word-structure` - 上传 Word 文档，提取结构化字段并存入 MongoDB
|
||||
- **API 端点**：`/api/v1/templates/word-fields/{doc_id}` - 获取已存文档的模板字段信息
|
||||
|
||||
### 3.3 Text 文档解析（✅ 已完成）
|
||||
|
||||
**已实现功能**：
|
||||
- `txt_parser.py` - 文本文件解析器
|
||||
- 编码自动检测 (chardet)
|
||||
- 文本清洗（去除控制字符、规范化空白）
|
||||
- 结构化数据提取（邮箱、URL、电话、日期、金额）
|
||||
|
||||
---
|
||||
|
||||
## 四、参赛材料准备
|
||||
|
||||
### 4.1 必交材料
|
||||
|
||||
| 材料 | 要求 | 当前状态 | 行动项 |
|
||||
|------|------|----------|--------|
|
||||
| 项目概要介绍 | PPT 格式 | ❌ 待制作 | 制作 PPT |
|
||||
| 项目简介 PPT | - | ❌ 待制作 | 制作 PPT |
|
||||
| 项目详细方案 | 文档 | ⚠️ 部分完成 | 完善文档 |
|
||||
| 项目演示视频 | - | ❌ 待制作 | 录制演示视频 |
|
||||
| 训练素材说明 | 来源说明 | ⚠️ 已有素材 | 整理素材文档 |
|
||||
| 关键模块设计文档 | 概要设计 | ⚠️ 已有部分 | 完善文档 |
|
||||
| 可运行 Demo | 核心代码 | ✅ 已完成 | 打包可运行版本 |
|
||||
|
||||
### 4.2 Demo 提交要求
|
||||
|
||||
根据 Q&A：
|
||||
- 可以只提交核心代码，不需要完整运行环境
|
||||
- 现场答辩可使用自带笔记本电脑
|
||||
- 需要提供部署和运行说明（README）
|
||||
|
||||
---
|
||||
|
||||
## 五、测试验证计划
|
||||
|
||||
### 5.1 使用现有测试数据
|
||||
|
||||
```
|
||||
docs/test/
|
||||
├── 2023年文化和旅游发展统计公报.md
|
||||
├── 2024年卫生健康事业发展统计公报.md
|
||||
├── 第三次全国工业普查主要数据公报.md
|
||||
```
|
||||
|
||||
### 5.2 模板填写测试流程
|
||||
|
||||
1. 准备一个 Word/Excel 模板表格
|
||||
2. 指定源数据文档
|
||||
3. 上传模板和文档
|
||||
4. 执行模板填写
|
||||
5. 检查填写结果准确率
|
||||
6. 记录响应时间
|
||||
|
||||
### 5.3 性能目标
|
||||
|
||||
| 指标 | 目标 | 当前状态 |
|
||||
|------|------|----------|
|
||||
| 信息提取准确率 | ≥80% | 需测试验证 |
|
||||
| 单次响应时间 | ≤90s × 文档数 | 需测试验证 |
|
||||
|
||||
---
|
||||
|
||||
## 六、工作计划（建议）
|
||||
|
||||
### 第一优先级：端到端测试
|
||||
- 使用真实测试数据进行准确率测试
|
||||
- 验证多行数据导出是否正确
|
||||
- 测试 Word 模板解析是否正常
|
||||
|
||||
### 第二优先级：Demo 打包与文档
|
||||
- 制作项目演示 PPT
|
||||
- 录制演示视频
|
||||
- 完善 README 部署文档
|
||||
|
||||
### 第三优先级：优化
|
||||
- 优化响应时间
|
||||
- 完善错误处理
|
||||
- 增加更多测试用例
|
||||
|
||||
---
|
||||
|
||||
## 七、注意事项
|
||||
|
||||
1. **创新点**：根据 Q&A，不必纠结创新点数量限制
|
||||
2. **数据库**：不强制要求数据库存储，可跳过
|
||||
3. **部署**：本地部署即可，不需要公网服务器
|
||||
4. **评测数据**：初赛仅使用目前提供的数据
|
||||
5. **RAG 功能**：当前已临时禁用，不影响核心评测功能（因为使用直接文件读取）
|
||||
|
||||
---
|
||||
|
||||
*文档版本: v1.5*
|
||||
*最后更新: 2026-04-09*
|
||||
|
||||
---
|
||||
|
||||
## 八、技术实现细节
|
||||
|
||||
### 8.1 模板填表流程
|
||||
|
||||
#### 流程图
|
||||
```
|
||||
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
|
||||
│ 上传模板 │ ──► │ 选择数据源 │ ──► │ 智能填表 │
|
||||
└─────────────┘ └─────────────┘ └─────────────┘
|
||||
│
|
||||
┌─────────────────────────┼─────────────────────────┐
|
||||
│ │ │
|
||||
▼ ▼ ▼
|
||||
┌───────────────┐ ┌───────────────┐ ┌───────────────┐
|
||||
│ 结构化数据提取 │ │ LLM 提取 │ │ 导出结果 │
|
||||
│ (直接读rows) │ │ (文本理解) │ │ (Excel/Word) │
|
||||
└───────────────┘ └───────────────┘ └───────────────┘
|
||||
```
|
||||
|
||||
#### 核心组件
|
||||
|
||||
| 组件 | 文件 | 说明 |
|
||||
|------|------|------|
|
||||
| 模板上传 | `templates.py` `/templates/upload` | 接收模板文件，提取字段 |
|
||||
| 字段提取 | `template_fill_service.py` | 从 Word/Excel 表格提取字段定义 |
|
||||
| 文档解析 | `docx_parser.py`, `xlsx_parser.py`, `txt_parser.py` | 解析源文档内容 |
|
||||
| 智能填表 | `template_fill_service.py` `fill_template()` | 结构化提取 + LLM 提取 |
|
||||
| 多行支持 | `template_fill_service.py` `FillResult` | values 数组支持 |
|
||||
| JSON 容错 | `template_fill_service.py` `_fix_json()` | 修复损坏的 JSON |
|
||||
| 结果导出 | `templates.py` `/templates/export` | 多行 Excel + Word 导出 |
|
||||
|
||||
### 8.2 源文档加载方式
|
||||
|
||||
模板填表服务支持两种方式加载源文档：
|
||||
|
||||
1. **通过 MongoDB 文档 ID**：`source_doc_ids`
|
||||
- 文档已上传并存入 MongoDB
|
||||
- 服务直接查询 MongoDB 获取文档内容
|
||||
|
||||
2. **通过文件路径**：`source_file_paths`
|
||||
- 直接读取本地文件
|
||||
- 使用对应的解析器解析内容
|
||||
|
||||
### 8.3 Word 表格模板解析
|
||||
|
||||
比赛评分表格通常是 Word 格式，`docx_parser.py` 提供了专门的解析方法：
|
||||
|
||||
```python
|
||||
# 提取表格模板字段
|
||||
from docx_parser import DocxParser
|
||||
parser = DocxParser()
|
||||
fields = parser.extract_template_fields_from_docx(file_path)
|
||||
|
||||
# 返回格式
|
||||
# [
|
||||
# {
|
||||
# "cell": "T0R1", # 表格0，行1
|
||||
# "name": "字段名",
|
||||
# "hint": "提示词",
|
||||
# "field_type": "text/number/date",
|
||||
# "required": True
|
||||
# },
|
||||
# ...
|
||||
# ]
|
||||
```
|
||||
|
||||
### 8.4 字段类型推断
|
||||
|
||||
系统支持从提示词自动推断字段类型：
|
||||
|
||||
| 关键词 | 推断类型 | 示例 |
|
||||
|--------|----------|------|
|
||||
| 年、月、日、日期、时间、出生 | date | 出生日期 |
|
||||
| 数量、金额、比率、%、率、合计 | number | 增长比率 |
|
||||
| 其他 | text | 姓名、地址 |
|
||||
|
||||
### 8.5 API 接口
|
||||
|
||||
#### POST `/api/v1/templates/upload`
|
||||
|
||||
上传模板文件，提取字段定义。
|
||||
|
||||
**响应**：
|
||||
```json
|
||||
{
|
||||
"success": true,
|
||||
"template_id": "/path/to/saved/template.docx",
|
||||
"filename": "模板.docx",
|
||||
"file_type": "docx",
|
||||
"fields": [
|
||||
{"cell": "A1", "name": "姓名", "field_type": "text", "required": true, "hint": "提取人员姓名"}
|
||||
],
|
||||
"field_count": 1
|
||||
}
|
||||
```
|
||||
|
||||
#### POST `/api/v1/templates/fill`
|
||||
|
||||
填写请求：
|
||||
```json
|
||||
{
|
||||
"template_id": "模板ID",
|
||||
"template_fields": [
|
||||
{"cell": "A1", "name": "姓名", "field_type": "text", "required": true, "hint": "提取人员姓名"}
|
||||
],
|
||||
"source_doc_ids": ["mongodb_doc_id_1", "mongodb_doc_id_2"],
|
||||
"source_file_paths": [],
|
||||
"user_hint": "请从xxx文档中提取"
|
||||
}
|
||||
```
|
||||
|
||||
**响应（含多行支持）**：
|
||||
```json
|
||||
{
|
||||
"success": true,
|
||||
"filled_data": {
|
||||
"姓名": ["张三", "李四", "王五"],
|
||||
"年龄": ["25", "30", "28"]
|
||||
},
|
||||
"fill_details": [
|
||||
{
|
||||
"field": "姓名",
|
||||
"cell": "A1",
|
||||
"values": ["张三", "李四", "王五"],
|
||||
"value": "张三",
|
||||
"source": "结构化数据直接提取",
|
||||
"confidence": 1.0
|
||||
}
|
||||
],
|
||||
"source_doc_count": 2,
|
||||
"max_rows": 3
|
||||
}
|
||||
```
|
||||
|
||||
#### POST `/api/v1/templates/export`
|
||||
|
||||
导出请求（创建新文件）：
|
||||
```json
|
||||
{
|
||||
"template_id": "模板ID",
|
||||
"filled_data": {"姓名": ["张三", "李四"], "金额": ["10000", "20000"]},
|
||||
"format": "xlsx"
|
||||
}
|
||||
```
|
||||
|
||||
#### POST `/api/v1/templates/fill-and-export`
|
||||
|
||||
**填充原始模板并导出**（推荐用于比赛）
|
||||
|
||||
直接打开原始模板文件，将数据填入模板的表格/单元格中，然后导出。**保持原始模板格式不变**。
|
||||
|
||||
**请求**：
|
||||
```json
|
||||
{
|
||||
"template_path": "/path/to/original/template.docx",
|
||||
"filled_data": {
|
||||
"姓名": ["张三", "李四", "王五"],
|
||||
"年龄": ["25", "30", "28"]
|
||||
},
|
||||
"format": "docx"
|
||||
}
|
||||
```
|
||||
|
||||
**响应**：填充后的 Word/Excel 文件（文件流）
|
||||
|
||||
**特点**：
|
||||
- 打开原始模板文件
|
||||
- 根据表头行匹配字段名到列索引
|
||||
- 将数据填入对应列的单元格
|
||||
- 多行数据自动扩展表格行数
|
||||
- 保持原始模板格式和样式
|
||||
|
||||
#### POST `/api/v1/templates/parse-word-structure`
|
||||
|
||||
**上传 Word 文档并提取结构化字段**（比赛专用）
|
||||
|
||||
上传 Word 文档，从表格模板中提取字段定义（字段名、提示词、字段类型）并存入 MongoDB。
|
||||
|
||||
**请求**：multipart/form-data
|
||||
- file: Word 文件
|
||||
|
||||
**响应**：
|
||||
```json
|
||||
{
|
||||
"success": true,
|
||||
"doc_id": "mongodb_doc_id",
|
||||
"filename": "模板.docx",
|
||||
"file_path": "/path/to/saved/template.docx",
|
||||
"field_count": 5,
|
||||
"fields": [
|
||||
{
|
||||
"cell": "T0R1",
|
||||
"name": "字段名",
|
||||
"hint": "提示词",
|
||||
"field_type": "text",
|
||||
"required": true
|
||||
}
|
||||
],
|
||||
"tables": [...],
|
||||
"metadata": {
|
||||
"paragraph_count": 10,
|
||||
"table_count": 1,
|
||||
"word_count": 500,
|
||||
"has_tables": true
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
#### GET `/api/v1/templates/word-fields/{doc_id}`
|
||||
|
||||
**获取 Word 文档模板字段信息**
|
||||
|
||||
根据 doc_id 获取已上传的 Word 文档的模板字段信息。
|
||||
|
||||
**响应**：
|
||||
```json
|
||||
{
|
||||
"success": true,
|
||||
"doc_id": "mongodb_doc_id",
|
||||
"filename": "模板.docx",
|
||||
"fields": [...],
|
||||
"tables": [...],
|
||||
"field_count": 5,
|
||||
"metadata": {...}
|
||||
}
|
||||
```
|
||||
|
||||
### 8.6 多行数据处理
|
||||
|
||||
**FillResult 数据结构**：
|
||||
```python
|
||||
@dataclass
|
||||
class FillResult:
|
||||
field: str
|
||||
values: List[Any] = None # 支持多个值（数组）
|
||||
value: Any = "" # 保留兼容（第一个值）
|
||||
source: str = "" # 来源文档
|
||||
confidence: float = 1.0 # 置信度
|
||||
```
|
||||
|
||||
**导出逻辑**：
|
||||
- 计算所有字段的最大行数
|
||||
- 遍历每一行，取对应索引的值
|
||||
- 不足的行填空字符串
|
||||
|
||||
### 8.7 JSON 容错处理
|
||||
|
||||
当 LLM 返回的 JSON 损坏或被截断时，系统会：
|
||||
|
||||
1. 清理 markdown 代码块标记（```json, ```）
|
||||
2. 尝试配对括号找到完整的 JSON
|
||||
3. 移除末尾多余的逗号
|
||||
4. 使用正则表达式提取 values 数组
|
||||
5. 备选方案：直接提取所有引号内的字符串
|
||||
|
||||
### 8.8 结构化数据优先提取
|
||||
|
||||
对于 Excel 等有 `rows` 结构的文档，系统会：
|
||||
|
||||
1. 直接从 `structured_data.rows` 中查找匹配列
|
||||
2. 使用模糊匹配（字段名包含或被包含）
|
||||
3. 提取该列的所有行值
|
||||
4. 无需调用 LLM，速度更快，准确率更高
|
||||
|
||||
```python
|
||||
# 内部逻辑
|
||||
if structured.get("rows"):
|
||||
columns = structured.get("columns", [])
|
||||
values = _extract_column_values(rows, columns, field_name)
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 九、依赖说明
|
||||
|
||||
### Python 依赖
|
||||
|
||||
```
|
||||
# requirements.txt 中需要包含
|
||||
fastapi>=0.104.0
|
||||
uvicorn>=0.24.0
|
||||
motor>=3.3.0 # MongoDB 异步驱动
|
||||
sqlalchemy>=2.0.0 # MySQL ORM
|
||||
pandas>=2.0.0 # Excel 处理
|
||||
openpyxl>=3.1.0 # Excel 写入
|
||||
python-docx>=0.8.0 # Word 处理
|
||||
chardet>=4.0.0 # 编码检测
|
||||
httpx>=0.25.0 # HTTP 客户端
|
||||
```
|
||||
|
||||
### 前端依赖
|
||||
|
||||
```
|
||||
# package.json 中需要包含
|
||||
react>=18.0.0
|
||||
react-dropzone>=14.0.0
|
||||
lucide-react>=0.300.0
|
||||
sonner>=1.0.0 # toast 通知
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 十、启动说明
|
||||
|
||||
### 后端启动
|
||||
|
||||
```bash
|
||||
cd backend
|
||||
.\venv\Scripts\Activate.ps1 # 或 Activate.bat
|
||||
pip install -r requirements.txt # 确保依赖完整
|
||||
.\venv\Scripts\python.exe -m uvicorn app.main:app --host 127.0.0.1 --port 8000 --reload
|
||||
```
|
||||
|
||||
### 前端启动
|
||||
|
||||
```bash
|
||||
cd frontend
|
||||
npm install
|
||||
npm run dev
|
||||
```
|
||||
|
||||
### 环境变量
|
||||
|
||||
在 `backend/.env` 中配置：
|
||||
```
|
||||
MONGODB_URL=mongodb://localhost:27017
|
||||
MONGODB_DB_NAME=document_system
|
||||
MYSQL_HOST=localhost
|
||||
MYSQL_PORT=3306
|
||||
MYSQL_USER=root
|
||||
MYSQL_PASSWORD=your_password
|
||||
MYSQL_DATABASE=document_system
|
||||
LLM_API_KEY=your_api_key
|
||||
LLM_BASE_URL=https://api.minimax.chat
|
||||
LLM_MODEL_NAME=MiniMax-Text-01
|
||||
```
|
||||
Reference in New Issue
Block a user