Metadata-Version: 2.4
Name: rapid-doc
Version: 0.8.0
Summary: A practical tool for converting PDF to Markdown
License: Apache-2.0
Project-URL: homepage, https://github.com/RapidAI
Project-URL: repository, https://github.com/RapidAI/RapidOCR
Project-URL: issues, https://github.com/RapidAI/RapidOCR/issues
Keywords: rapid-doc, rapid_doc, onnx, convert, pdf, markdown
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Requires-Python: <3.14,>=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: boto3>=1.28.43
Requires-Dist: loguru>=0.6.0
Requires-Dist: numpy>=1.21.6
Requires-Dist: pdfminer.six==20250506
Requires-Dist: tqdm>=4.67.1
Requires-Dist: requests
Requires-Dist: pypdfium2>=4.30.0
Requires-Dist: pypdf>=5.6.0
Requires-Dist: reportlab
Requires-Dist: pdftext>=0.6.2
Requires-Dist: json-repair>=0.46.2
Requires-Dist: fast-langdetect<0.3.0,>=0.2.3
Requires-Dist: scikit-image<1.0.0,>=0.25.0
Requires-Dist: beautifulsoup4<5,>=4.13.5
Requires-Dist: pydantic<2.13,>=2.7.2
Requires-Dist: matplotlib<4,>=3.10
Requires-Dist: ftfy<7,>=6.3.1
Requires-Dist: shapely<3,>=2.0.7
Requires-Dist: tokenizers>=0.13.2
Requires-Dist: rapidocr<=3.4.3,>=3.4.0
Requires-Dist: openai>=1.63
Requires-Dist: python-docx<2,>=1.2.0
Requires-Dist: pypptx-with-oxml<2,>=1.0.3
Requires-Dist: mammoth<2,>=1.11.0
Requires-Dist: pylatexenc<3,>=2.10
Requires-Dist: lxml<7.0.0,>=4.0.0
Requires-Dist: pandas<3,>=2.3.3
Requires-Dist: openpyxl<4,>=3.1.5
Provides-Extra: test
Requires-Dist: rapid-doc[cpu]; extra == "test"
Requires-Dist: pytest; extra == "test"
Requires-Dist: fuzzywuzzy; extra == "test"
Provides-Extra: cpu
Requires-Dist: onnxruntime>=1.18.0; extra == "cpu"
Requires-Dist: openvino<=2025.4.0,>=2024.6.0; extra == "cpu"
Provides-Extra: gpu
Requires-Dist: onnxruntime-gpu<=1.23.0; extra == "gpu"
Requires-Dist: torch<3,>=2.6.0; extra == "gpu"
Requires-Dist: torchvision; extra == "gpu"
Provides-Extra: api
Requires-Dist: fastapi>=0.104.0; extra == "api"
Requires-Dist: uvicorn[standard]>=0.24.0; extra == "api"
Requires-Dist: python-multipart>=0.0.6; extra == "api"
Provides-Extra: word
Requires-Dist: pypandoc-binary; extra == "word"
Provides-Extra: html
Requires-Dist: markdown-it-py; extra == "html"
Requires-Dist: mdit-py-plugins; extra == "html"
Requires-Dist: pygments; extra == "html"
Provides-Extra: gradio
Requires-Dist: gradio!=6.0.0,!=6.0.1,!=6.0.2,!=6.1.0,!=6.2.0,!=6.3.0,!=6.4.0,!=6.5.0,!=6.5.1,!=6.6.0,!=6.7.0,<6.9.0,>=5.49.1; extra == "gradio"
Requires-Dist: gradio-pdf>=0.0.22; extra == "gradio"
Provides-Extra: core
Requires-Dist: openvino>=2024.6.0; extra == "core"
Requires-Dist: onnxruntime-gpu<=1.23.0; extra == "core"
Requires-Dist: torch<3,>=2.6.0; extra == "core"
Requires-Dist: torchvision; extra == "core"
Provides-Extra: npu
Requires-Dist: numpy<2,>=1.21.6; extra == "npu"
Requires-Dist: decorator; extra == "npu"
Requires-Dist: attrs; extra == "npu"
Requires-Dist: psutil; extra == "npu"
Requires-Dist: onnxruntime-cann==1.22.0; extra == "npu"
Requires-Dist: torch==2.5.1; extra == "npu"
Requires-Dist: torch-npu==2.5.1; extra == "npu"
Requires-Dist: torchvision; extra == "npu"
Dynamic: license-file

# RapidDoc – 高速文档解析系统

## 😺 项目介绍

**RapidDoc 是一个轻量级、专注于文档解析的开源框架，支持 **OCR、版面分析、公式识别、表格识别和阅读顺序恢复** 等多种功能，支持将复杂 PDF 文档转换为 Markdown、JSON、WORD、HTML 格式。**

**支持docx/doc、pptx/ppt、xlsx/xls的原生解析（不使用模型）。**

**框架基于 [Mineru](https://github.com/opendatalab/MinerU) 二次开发，移除 VLM，专注于 Pipeline 产线下的高效文档解析，在 CPU 上也能保持不错的解析速度。**

**本项目所使用的默认模型主要来源于 [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR) 的 [PP-StructureV3](https://www.paddleocr.ai/main/version3.x/pipeline_usage/PP-StructureV3.html) 系列（OCR、版面分析、公式识别、阅读顺序恢复，以及部分表格识别模型），并已全部转换为 ONNX 格式，支持在 CPU/GPU 上高效推理。**

**同时支持自定义OCR、公式、表格模型，需实现 CustomBaseModel 的 batch_predict 方法，目前内置 [PaddleOCRVL](https://www.paddleocr.ai/main/version3.x/pipeline_usage/PaddleOCR-VL.html) 系列模型的集成。**

**KittyDoc 已经成为 RapidAI 开源家族成员**

---

> ✨如果该项目对您有帮助，您的star是我不断优化的动力！！！
>
> - [github点击前往](https://github.com/RapidAI/RapidDoc)
> - [gitee点击前往](https://gitee.com/hzkitty/KittyDoc)

## 👏 项目特点

- **OCR 识别**
  - 使用 [RapidOCR](https://github.com/RapidAI/RapidOCR) 支持多种推理引擎
  - CPU 下默认使用 OpenVINO，GPU 下默认使用 torch
  
- **版面识别**
  - 模型使用 `PP-DocLayout` 系列 ONNX 模型（v2、plus-L、L、M、S）
    - **PP-DocLayoutV2**：自带阅读顺序，效果最好，默认使用
    - **PP-DocLayoutV3**：自带阅读顺序，支持异形框
    - **PP-DocLayout_plus-L**：效果好运行稳定
    - **PP-DocLayout-L**：速度快，效果也不错
    - **PP-DocLayout-S**：速度极快，存在部分漏检

- **公式识别**
  - 使用 `PP-FormulaNet_plus` 系列 ONNX 模型（L、M、S）
    - **PP-FormulaNet_plus-L**：速度慢，支持onnx  
    - **PP-FormulaNet_plus-M**：默认使用，支持onnx和torch    
    - **PP-FormulaNet_plus-S**：速度最快，支持onnx，复杂公式精度不够
  - 支持配置只识别行间公式
  - cuda环境，默认使用torch推理，公式模型onnx gpu推理会报错，暂时无人解决 [PaddleOCR/issues/15125](https://github.com/PaddlePaddle/PaddleOCR/issues/15125), [PaddleX/issues/4238](https://github.com/PaddlePaddle/PaddleX/issues/4238), [Paddle2ONNX/issues/1593](https://github.com/PaddlePaddle/Paddle2ONNX/issues/1593)

- **表格识别**
  - 基于 [rapid_table_self](rapid_doc/model/table/rapid_table_self) 增强，在原有基础上增强为多模型串联方案：  
    - **表格分类**（区分有线/无线表格）
    - **[有线表格识别UNET](https://github.com/RapidAI/TableStructureRec)** + SLANET_plus/UNITABLE（作为无线表格识别）

- **阅读顺序恢复**
  - PP-DocLayoutV2和PP-DocLayoutV3使用版面模型自带的阅读顺序
  - 其余版面模型，使用 PP-StructureV3 阅读顺序恢复算法，基于xycut算法和版面的结果

- **推理方式**
  - 所有模型通过 ONNXRuntime 推理，OCR可配置其他推理引擎
  - 除了 OCR 和 PP-DocLayout-M/S 模型，OpenVINO推理会报错，暂时难以解决。[PaddleOCR/issues/16277](https://github.com/PaddlePaddle/PaddleOCR/issues/16277)
---

## 基准测试结果

### 1. OmniDocBench

以下是RapidDoc在 OmniDocBench 上的评估结果。

Pipeline 模型使用 PP-DocLayoutV2、PP-OCRv5-mobile、PP-FormulaNet_plus-M、UNET_SLANET_PLUS。
<table style="width:100%; border-collapse: collapse;">
    <caption>Comprehensive evaluation of document parsing on OmniDocBench (v1.5)</caption>
    <thead>
        <tr>
            <th>Model Type</th>
            <th>Methods</th>
            <th>Size</th>
            <th>Overall&#x2191;</th>
            <th>Text<sup>Edit</sup>&#x2193;</th>
            <th>Formula<sup>CDM</sup>&#x2191;</th>
            <th>Table<sup>TEDS</sup>&#x2191;</th>
            <th>Table<sup>TEDS-S</sup>&#x2191;</th>
            <th>Read Order<sup>Edit</sup>&#x2193;</th>
        </tr>
    </thead>
    <tbody>
        <tr>
            <td rowspan="16"><strong>Specialized</strong><br><strong>VLMs</strong></td>
            <td>PaddleOCR-VL</td>
            <td>0.9B</td>
            <td><strong>92.86</strong></td>
            <td><strong>0.035</strong></td>
            <td><strong>91.22</strong></td>
            <td><strong>90.89</strong></td>
            <td><strong>94.76</strong></td>
            <td><strong>0.043</strong></td>
        </tr>
            <td>MinerU2.5</td>
            <td>1.2B</td>
            <td><ins>90.67</ins></td>
            <td><ins>0.047</ins></td>
            <td><ins>88.46</ins></td>
            <td><ins>88.22</ins></td>
            <td><ins>92.38</ins></td>
            <td><ins>0.044</ins></td>
        </tr>
        <tr>
            <td>MonkeyOCR-pro-3B</td>
            <td>3B</td>
            <td>88.85</td>
            <td>0.075</td>
            <td>87.25</td>
            <td>86.78</td>
            <td>90.63</td>
            <td>0.128</td>
        </tr>
        <tr>
            <td>OCRVerse</td>
            <td>4B</td>
            <td>88.56</td>
            <td>0.058</td>
            <td>86.91</td>
            <td>84.55</td>
            <td>88.45</td>
            <td>0.071</td>
        </tr>
        <tr>
            <td>dots.ocr</td>
            <td>3B</td>
            <td>88.41</td>
            <td>0.048</td>
            <td>83.22</td>
            <td>86.78</td>
            <td>90.62</td>
            <td>0.053</td>
        </tr>
        <tr>
            <td>MonkeyOCR-3B</td>
            <td>3B</td>
            <td>87.13</td>
            <td>0.075</td>
            <td>87.45</td>
            <td>81.39</td>
            <td>85.92</td>
            <td>0.129</td>
        </tr>
        <tr>
            <td>Deepseek-OCR</td>
            <td>3B</td>
            <td>87.01</td>
            <td>0.073</td>
            <td>83.37</td>
            <td>84.97</td>
            <td>88.80</td>
            <td>0.086</td>
        </tr>
        <tr>
            <td>MonkeyOCR-pro-1.2B</td>
            <td>1.2B</td>
            <td>86.96</td>
            <td>0.084</td>
            <td>85.02</td>
            <td>84.24</td>
            <td>89.02</td>
            <td>0.130</td>
        </tr>
        <tr>
            <td>Nanonets-OCR-s</td>
            <td>3B</td>
            <td>85.59</td>
            <td>0.093</td>
            <td>85.90</td>
            <td>80.14</td>
            <td>85.57</td>
            <td>0.108</td>
        </tr>
        <tr>
            <td>MinerU2-VLM</td>
            <td>0.9B</td>
            <td>85.56</td>
            <td>0.078</td>
            <td>80.95</td>
            <td>83.54</td>
            <td>87.66</td>
            <td>0.086</td>
        </tr>
        <tr>
            <td>olmOCR</td>
            <td>7B</td>
            <td>81.79</td>
            <td>0.096</td>
            <td>86.04</td>
            <td>68.92</td>
            <td>74.77</td>
            <td>0.121</td>
        </tr>
        <tr>
            <td>Dolphin-1.5</td>
            <td>0.3B</td>
            <td>83.21</td>
            <td>0.092</td>
            <td>80.78</td>
            <td>78.06</td>
            <td>84.10</td>
            <td>0.080</td>
        </tr>
        <tr>
            <td>POINTS-Reader</td>
            <td>3B</td>
            <td>80.98</td>
            <td>0.134</td>
            <td>79.20</td>
            <td>77.13</td>
            <td>81.66</td>
            <td>0.145</td>
        </tr>
        <tr>
            <td>Mistral OCR</td>
            <td>-</td>
            <td>78.83</td>
            <td>0.164</td>
            <td>82.84</td>
            <td>70.03</td>
            <td>78.04</td>
            <td>0.144</td>
        </tr>
        <tr>
            <td>OCRFlux</td>
            <td>3B</td>
            <td>74.82</td>
            <td>0.193</td>
            <td>68.03</td>
            <td>75.75</td>
            <td>80.23</td>
            <td>0.202</td>
        </tr>
        <tr>
            <td>Dolphin</td>
            <td>0.3B</td>
            <td>74.67</td>
            <td>0.125</td>
            <td>67.85</td>
            <td>68.70</td>
            <td>77.77</td>
            <td>0.124</td>
        </tr>
        <tr>
            <td rowspan="6"><strong>General</strong><br><strong>VLMs</strong></td>
            <td>Qwen3-VL-235B-A22B-Instruct</td>
            <td>235B</td>
            <td>89.15</td>
            <td>0.069</td>
            <td>88.14</td>
            <td>86.21</td>
            <td>90.55</td>
            <td>0.068</td>
        </tr>
            <td>Gemini-2.5 Pro</td>
            <td>-</td>
            <td>88.03</td>
            <td>0.075</td>
            <td>85.82</td>
            <td>85.71</td>
            <td>90.29</td>
            <td>0.097</td>
        </tr>
        <tr>
            <td>Qwen2.5-VL</td>
            <td>72B</td>
            <td>87.02</td>
            <td>0.094</td>
            <td>88.27</td>
            <td>82.15</td>
            <td>86.22</td>
            <td>0.102</td>
        </tr>
        <tr>
            <td>InternVL3.5</td>
            <td>241B</td>
            <td>82.67</td>
            <td>0.142</td>
            <td>87.23</td>
            <td>75.00</td>
            <td>81.28</td>
            <td>0.125</td>
        </tr>
        <tr>
            <td>InternVL3</td>
            <td>78B</td>
            <td>80.33</td>
            <td>0.131</td>
            <td>83.42</td>
            <td>70.64</td>
            <td>77.74</td>
            <td>0.113</td>
        </tr>
        <tr>
            <td>GPT-4o</td>
            <td>-</td>
            <td>75.02</td>
            <td>0.217</td>
            <td>79.70</td>
            <td>67.07</td>
            <td>76.09</td>
            <td>0.148</td>
        </tr>
        <tr>
            <td rowspan="4"><strong>Pipeline</strong><br><strong>Tools</strong></td>
            <td><strong>RapidDoc</strong></td>
            <td>-</td>
            <td>87.81</td>
            <td>0.065</td>
            <td>89.348</td>
            <td>80.59</td>
            <td>87.90</td>
            <td>0.053</td>
        </tr>
        <tr>
            <td>PP-StructureV3</td>
            <td>-</td>
            <td>86.73</td>
            <td>0.073</td>
            <td>85.79</td>
            <td>81.68</td>
            <td>89.48</td>
            <td>0.073</td>
        </tr>
        <tr>
            <td>Mineru2-pipeline</td>
            <td>-</td>
            <td>75.51</td>
            <td>0.209</td>
            <td>76.55</td>
            <td>70.90</td>
            <td>79.11</td>
            <td>0.225</td>
        </tr>
        <tr>
            <td>Marker-1.8.2</td>
            <td>-</td>
            <td>71.30</td>
            <td>0.206</td>
            <td>76.66</td>
            <td>57.88</td>
            <td>71.17</td>
            <td>0.250</td>
        </tr>
    </tbody>
</table>

## 🛠️ 安装RapidDoc

#### 使用pip安装
```bash
pip install rapid-doc[cpu] -i https://mirrors.aliyun.com/pypi/simple
或
pip install rapid-doc[gpu] -i https://mirrors.aliyun.com/pypi/simple
```

#### 通过源码安装
```bash
# 克隆仓库
git clone https://github.com/RapidAI/RapidDoc.git
cd RapidDoc

# 安装依赖
pip install -e .[cpu] -i https://mirrors.aliyun.com/pypi/simple
或
pip install -e .[gpu] -i https://mirrors.aliyun.com/pypi/simple
```
#### 使用gpu推理
```python
# rapid-doc[gpu] 默认安装 onnxruntime-gpu 最新版
# 需要确定onnxruntime-gpu与GPU对应，参考 https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements

# 在 Python 中指定 GPU（必须在导入 rapid_doc 之前设置）
import os
# 使用默认 GPU（cuda:0）
os.environ['MINERU_DEVICE_MODE'] = "cuda"
# 或指定 GPU 编号，例如使用第二块 GPU（cuda:1）
os.environ['MINERU_DEVICE_MODE'] = "cuda:1"
```
#### 使用PaddleOCRVL系列推理
vl模型的部署，参考[官方文档](https://www.paddleocr.ai/main/version3.x/pipeline_usage/PaddleOCR-VL.html#31-vlm) 
```python
import os
os.environ['PADDLEOCRVL_VERSION'] = "v1.5"
os.environ['PADDLEOCRVL_VL_REC_BACKEND'] = "vllm-server"
os.environ['PADDLEOCRVL_VL_VL_REC_SERVER_URL'] = "http://localhost:8118/v1"

from rapid_doc.model.layout.rapid_layout_self import ModelType as LayoutModelType
from rapid_doc.model.custom.paddleocr_vl.paddleocr_vl import PaddleOCRVLTableModel, PaddleOCRVLOCRModel, PaddleOCRVLFormulaModel
layout_config = {
    "model_type": LayoutModelType.PP_DOCLAYOUTV3,
}
ocr_config = {
    "custom_model": PaddleOCRVLOCRModel()
}
formula_config = {
    "custom_model": PaddleOCRVLFormulaModel()
}
table_config = {
    "custom_model": PaddleOCRVLTableModel()
}
```

#### 使用docker部署RapidDoc
RapidDoc提供了便捷的docker部署方式，这有助于快速搭建环境并解决一些棘手的环境兼容问题。

您可以在文档中获取 [Docker部署说明](docker/README.md)，镜像已推送至 [Docker Hub](https://hub.docker.com/r/hzkitty/rapid-doc)。

---
### 📋 使用

```python
import os
from pathlib import Path
from rapid_doc import RapidDoc
__dir__ = Path(__file__).resolve().parent.parent
output_dir = os.path.join(__dir__, "output")

doc_path_list = [
    __dir__ / "demo/pdfs/示例1-论文模板.pdf",
    __dir__ / "demo/docx/test.docx",
]
engine = RapidDoc()
outputs = engine(doc_path_list, output_dir=output_dir)
for output in outputs:
    print(output.markdown)
```
---

## 在线体验

### 基于Gradio的在线demo
基于gradio开发的webui，界面简洁，仅包含核心解析功能，免登录

- [![ModelScope](https://img.shields.io/badge/Demo_on_ModelScope-purple?logo=data:image/svg+xml;base64,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&labelColor=white)](https://www.modelscope.cn/studios/RapidAI/RapidDoc)

---

## 📋 使用示例

- [代码示例](./demo/demo.py)

- [参数介绍](./docs/analyze_param.md)

- [FastAPI 示例](./docker/README_API.md)
---

## 模型下载
不指定模型路径，初次运行时，会自动下载
- [RapidDoc 模型集（版面/公式/表格）](https://www.modelscope.cn/models/RapidAI/RapidDoc)  
- [RapidOCR 模型](https://www.modelscope.cn/models/RapidAI/RapidOCR)  
- [部分表格模型RapidTable](https://www.modelscope.cn/models/RapidAI/RapidTable) 

---

## 📌 TODO

- [x] 跨页表格合并
- [x] 复选框识别，使用opencv（默认关闭、opencv识别存在误检）
- [x] 提供 fastapi，支持cpu和gpu版本的docker镜像构建
- [x] 文本型pdf，表格非OCR文本提取
- [x] 文本型pdf，使用pypdfium2提取文本框bbox
- [x] 文本型pdf，支持0/90/270度三个方向的表格解析
- [x] 文本型pdf，使用pypdfium2提取原始图片（默认截图会导致清晰度降低和图片边界可能丢失部分）
- [x] 表格内公式提取，表格内图片提取
- [x] 优化阅读顺序，支持多栏、竖排等复杂版面恢复
- [x] 公式支持torch推理，可用GPU加速
- [x] 版面、表格模型支持openvino
- [x] markdown转docx、html
- [x] 支持 PP-DocLayoutV2 版面识别+阅读顺序
- [x] OmniDocBench评测
- [x] 支持自定义的ocr、table、公式。支持PaddleOCR-VL系列
- [x] 支持docx/doc、pptx/ppt、xlsx/xls的原生解析（不使用模型）


## 🙏 致谢

- [MinerU](https://github.com/opendatalab/MinerU)
- [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR)
- [RapidOCR](https://github.com/RapidAI/RapidOCR)

## ⚖️ 开源许可

基于 [MinerU](https://github.com/opendatalab/MinerU) 改造而来，已**移除原项目中的 YOLO 模型**，并替换为 **PP-StructureV3 系列 ONNX 模型**。  
由于已移除 AGPL 授权的 YOLO 模型部分，本项目整体不再受 AGPL 约束。

该项目采用 [Apache 2.0 license](LICENSE) 开源许可证。
