AI-enabled documentation generation

[thorwhalen]

This discussion is to accumulate ideas and resources for generating documentation with AI, and discussing these.

Quick-links

• Resources

[thorwhalen]

Resources

Libraries

• Code Narrator (JS).
  • Post - Code-Narrator: Automating Documentation with GPT-4.
  • Code narrator - howto create a howto guide - YouTube

[thorwhalen]

Tools (chatGPT research)

Table

Here is a feature comparison table for various AI-enabled Python code documentation tools. This table includes information on features, pricing, licensing, and more to help you decide which tool best fits your needs.

Feature/Tool	DocuWriter.ai	Cody by Sourcegraph	CodePal	Docify AI	Swimm	Doxygen
Documentation Types	General, API, Library, Database, End-User, Testing Strategy, Deployment	General, contextual help, chat-based interaction	General, API, End-User, Technical, Maintenance	General, Inline comments, Translated comments, Custom template comments	Overview documentation, code snippets, pull requests, branches documentation	General, API, Class, Function
Code Refactoring	Yes, including intelligent code simplification and maintainability boosts	No	No	No	No	No
Language Support	Multi-language support (32 languages)	Primarily English	Over 30 programming languages, including Python, Java, JavaScript	Supports 9 programming languages, including Python, Java, JavaScript, C#, Ruby, PHP, Go, C, C++	Language agnostic, supports all programming languages	Supports many languages, primarily C++, C, Java, etc.
IDE Integration	VSCode	VSCode, JetBrains	VSCode	VSCode, Eclipse, IntelliJ	VSCode, other editors through extensions	No direct IDE integration
Pricing	Starts at $29/month (Starter), $49/month (Professional), $129/month (Enterprise)	Open-source, free	Subscription-based, pricing details not readily available	Free trial available, pricing not readily available	Subscription-based, pricing details not readily available	Free, open-source
Free Trial	Yes	N/A	Yes	Yes	Yes	N/A
Licensing	Subscription-based	Open-source	Subscription-based	Subscription-based	Subscription-based	Open-source
Additional Features	Code language converter, UML diagram generator, Zapier integration, automatic Git documentation	Uses advanced search and codebase context to help understand and fix code faster	AI-generated documentation, automated testing	Proprietary AI models for privacy, inline comments, high performance on large codebases	Auto-docs for entire repositories, snippets-to-doc, PR-to-doc, branch-to-doc, chat-to-doc	Extracts documentation from annotated source code, multiple export formats (HTML, PDF, LaTeX)
More Information	DocuWriter.ai	Cody by Sourcegraph	CodePal	Docify AI	Swimm	Doxygen

Key Takeaways:

• DocuWriter.ai offers a comprehensive set of features including multiple types of documentation, code refactoring, language conversion, and UML diagram generation. It integrates with VSCode and starts at $29/month oai_citation:1,DocuWriter.ai - #1 AI Code documentation tools oai_citation:2,DocuWriter.ai Pricing: Cost and Pricing plans oai_citation:3,DocuWriter.ai Software Reviews, Demo & Pricing - 2024.
• Cody by Sourcegraph is an open-source tool providing contextual help and chat-based interaction to assist with code understanding and fixing. It integrates with multiple IDEs and is free to use oai_citation:4,GitHub - sourcegraph/cody: Type less, code more: Cody is an AI code assistant that uses advanced search and codebase context to help you write and fix code..
• CodePal supports over 30 programming languages and provides various documentation types, although detailed pricing information is not readily available oai_citation:5,Python Code Documentation - CodePal.
• Docify AI focuses on generating docstrings, inline comments, and translating comments. It supports multiple IDEs and offers a free trial oai_citation:6,Docify AI – Code Comment & Documentation Tool.
• Swimm generates documentation through static code analysis and supports comprehensive documentation creation for repositories, code snippets, pull requests, and branches. It also provides a free trial oai_citation:7,Code Documentation Generators: 6 Great Tools to Use - Swimm.
• Doxygen is a well-known open-source tool that generates documentation from annotated source code and supports multiple export formats oai_citation:8,Code Documentation Generators: 6 Great Tools to Use - Swimm.

Each tool has unique strengths, making them suitable for different use cases depending on your specific needs in documentation, integration, and budget.

List

Here is a list of existing AI-enabled tools for Python code documentation that you can use or build upon for your project:

1. DocuWriter.ai:

   • Features: Generates comments, refactors code, converts code between languages, and creates various types of documentation, including general, API, library, database, end-user, testing strategy, and deployment documentation. It integrates with VSCode.
   • Use Case: Comprehensive tool for generating and maintaining detailed documentation across multiple programming languages.
   • Link: DocuWriter.ai

2. Cody by Sourcegraph:

   • Features: AI assistant for writing, fixing, and understanding code using advanced search to pull context from both local and remote codebases. It works within IDEs like VSCode and JetBrains.
   • Use Case: Ideal for integrating documentation generation directly into the development workflow and leveraging contextual understanding of the codebase.
   • Link: Cody by Sourcegraph

3. CodePal:

   • Features: Uses AI to generate code documentation by analyzing codebases, with a focus on providing accurate, readable documentation that improves code quality and developer collaboration.
   • Use Case: Suitable for generating comprehensive documentation for Python and other programming languages, ensuring clarity and consistency across the codebase.
   • Link: CodePal

4. Docify AI:

   • Features: Automated generation of docstrings, inline comments, and translation of comments to multiple languages. Supports multiple IDEs including VSCode, Eclipse, and IntelliJ.
   • Use Case: Best for developers needing quick and efficient inline comments and documentation for legacy or new codebases.
   • Link: Docify AI

5. Swimm:

   • Features: Generates documentation through static code analysis and relational ranking algorithms. Offers features like snippets-to-doc, PR-to-doc, and branch-to-doc to create documentation from specific code snippets, pull requests, or branches.
   • Use Case: Useful for creating and maintaining effective documentation automatically, integrating closely with the software development lifecycle.
   • Link: Swimm

6. Doxygen:

   • Features: Extracts documentation from annotated source code in various formats such as HTML, PDF, and LaTeX.
   • Use Case: Widely used for generating documentation from C++, C, Java, and other languages, providing detailed and structured documentation.
   • Link: Doxygen

These tools offer a range of features that can significantly reduce the time and effort required for documenting Python code. You can use these tools directly or as inspiration for developing your AI-enabled documentation tool, integrating the best features that suit your specific needs and workflow.

[thorwhalen]

Custom GPTs for documentation generation

• DocStringGPT: Custom GPT I made to try out some AI-enabled documentation generation things. By @thorwhalen, experimental. The part I worked on the most was to get it to generate actual files that I could download. That part works on chatGPT desktop, but the download links weren't rendered correctly on Safari. Given this, I looked into a new thing I saw in custom chatGPTs: Actions. See my question (and answer)
• Also tried Python Docstring Writer before doing mine, experimenting with making it do downloadable files.