docapi 0.1.3

DocAPI is a Python package that automatically generates API documentation using LLM.

[ English | 中文 ]

DocAPI is a Python package that uses LLM to automatically generate API documentation.

Features

• The Flask framework supports automatic scanning of the routing structure of API services;

• Supports a variety of mainstream commercial and open source models at home and abroad;

• Supports automatic document generation and partial document update;

• Supports Chinese and English documents;

• Supports web page deployment to display API documentation.

Installation

pip install -U docapi

or

pip install -U docapi -i https://pypi.org/simple

GitHub source code installation

pip install git+https://github.com/Shulin-Zhang/docapi

Usage

Automatically scan the routing structure. This is only valid for flask projects and must be used in the environment of api projects.

OpenAI:

export OPENAI_API_KEY=api_key

export OPENAI_API_MODEL=gpt-4o-mini

# Generate documents
docapi generate server.py --lang en

# Update documents
docapi update server.py --lang en

# Start web service
docapi serve

Azure OpenAI:

export AZURE_OPENAI_API_KEY=api_key

export AZURE_OPENAI_ENDPOINT=endpoint

export OPENAI_API_VERSION=version

export AZURE_OPENAI_MODEL=gpt-4o-mini

# 生成文档
docapi generate server.py --template <template_path>

# 更新文档
docapi update server.py --template <template_path>

# 启动web服务
docapi serve docs --ip 0.0.0.0 --port 9000

Qianwen, Open source deployment:

export OPENAI_API_KEY=api_key

export OPENAI_API_BASE=api_base_url

export OPENAI_API_MODEL=model_name

# Generate documents
docapi generate server.py --lang en

# Update documents
docapi update server.py --lang en

# Start web service
docapi serve

Baidu Qianfan:

export QIANFAN_ACCESS_KEY=access_key

export QIANFAN_SECRET_KEY=secret_key

export QIANFAN_MODEL=ERNIE-3.5-8K

# Generate documents
docapi generate server.py --lang en

# Update documents
docapi update server.py --lang en

# Start web service
docapi serve

ZhipuAI:

export ZHIPUAI_API_KEY=api_key

export ZHIPUAI_MODEL=glm-4-flash

# Generate documents
docapi generate server.py

# Update documents
docapi update server.py

# Start web service
docapi serve

Code calls

import os
from docapi import DocAPI, llm_builder, prompt
from docapi.scanner import flask_scanner

os.environ['OPENAI_API_KEY'] = "api_key"
os.environ['OPENAI_API_BASE'] = "api_base"
os.environ['OPENAI_API_MODEL'] = "model_name"

docapi = DocAPI.build(lang="en")

# template = '''<document template>
# '''

# docapi = DocAPI(llm=llm_builder.build_llm(), scanner=flask_scanner, prompt=prompt.doc_en, template=template)

docapi.generate("flask_project/flask_server.py", "docs")

# docapi.update("flask_project/flask_server.py", "docs")

# docapi.serve("docs", ip="127.0.0.1", port=8080)

Changelog

• [2024-11-17] Support Zhipu AI, Baidu Qianfan model, optimize document structure, and add javascript code examples; Remove the execution mode that uses the configuration file.

• [2024-11-20] Support custom document templates.

• [2024-11-24] Support multi-threaded acceleration requests.

Supported Models

• OpenAI

• AzureOpenAI

• Tongyi Qianwen

• Zhipu AI

• Baidu Qianfan

• Open source model

Supported API Frameworks

• Flask

Automatic scanning is only valid for the Flask framework and is recommended for use on Flask services.

API Web Page

TODO

• Supports large models such as Wenxin Yiyan and Zhipu AI.

• Supports automatic scanning of frameworks such as Django.

• Supports online web page display of documents.

• Supports custom document templates.

• Multithreading accelerates requests.

• Import to postman.