pydoc2markdown 0.4.1

Convert Python docstrings to Markdown documentation.

PyDoc2Markdown

Convert Python docstrings into clean, structured Markdown documentation.

Table of Contents

• Why PyDoc2Markdown?
• Features
• Requirements
• Installation
• Quick Start
  • CLI Usage
  • Library Usage
• CLI Reference
• Configuration
• Library API
• Supported Docstring Formats
• Example Output
• Documentation
• License

Why PyDoc2Markdown?

Documentation generators like Sphinx are powerful but require significant configuration — themes, conf.py, extensions, and often custom builders to get clean output. pdoc and mkdocstrings are easier but still depend on a full documentation framework.

PyDoc2Markdown takes a different approach: zero configuration, zero framework dependencies, pure Markdown output. Point it at your Python project and get structured Markdown files that work anywhere — GitHub, GitLab, MkDocs, or any Markdown renderer.

• No conf.py — works out of the box
• No framework lock-in — generates plain .md files
• Minimal dependencies — Jinja2 + docstring-parser

Features

• Docstring parsing — Extract Google, NumPy, and reStructuredText style docstrings with structured params, returns, and raises.
• Markdown generation — Produce beautiful Markdown files with customizable Jinja2 templates.
• Auto-generated index & TOC — Each module gets a Table of Contents; an index.md with package grouping is created automatically.
• Package grouping — Output files are organized into subdirectories matching the package structure.
• Built-in themes — Choose between default (detailed) and minimal themes, or supply your own template.
• CLI & API — Use via command line or import as a Python library.
• Recursive processing — Scan entire packages in one command.
• Type-aware — Respects type hints and annotations.
• Advanced constructs — Supports property, classmethod, staticmethod, dataclass, Enum, TypedDict, Protocol, ABC, Pydantic, and __all__.

Requirements

• Python >= 3.10
• Dependencies: Jinja2, docstring-parser
• Optional: watchdog (for --watch mode)

Installation

# Base installation
pip install pydoc2markdown

# With file watcher support
pip install pydoc2markdown[watch]

Quick Start

CLI Usage

# Generate docs for a single file
pydoc2markdown my_module.py -o docs

# Recursively process a package
pydoc2markdown src/my_package --recursive -o docs

# Use a built-in theme
pydoc2markdown src/my_package --recursive --theme minimal -o docs

# Use a custom template
pydoc2markdown src/my_package --recursive --template custom.md.j2 -o docs

# Single combined file
pydoc2markdown src/my_package --recursive --single-file -o docs/README.md

# Watch mode — auto-regenerate on changes
pydoc2markdown src/my_package --recursive --watch -o docs

# Enable verbose logging
pydoc2markdown src/my_package --recursive -vv -o docs

Library Usage

from pathlib import Path
from pydoc2markdown import DocstringParser, MarkdownGenerator

parser = DocstringParser()
modules = parser.parse(Path("src/my_package"), recursive=True)

generator = MarkdownGenerator(theme="default")
generator.generate(modules, output_dir=Path("docs"))

CLI Reference

Flag	Default	Description
source	(required)	Path to a Python file or directory to process
-o, --output	docs / value from pyproject.toml	Output directory (or file path when --single-file is used)
--recursive	False / value from pyproject.toml	Recursively process subdirectories
--theme	default / value from pyproject.toml	Built-in theme: default (detailed) or minimal
--template	None	Path to a custom Jinja2 template for Markdown generation
--single-file	False	Generate a single combined Markdown file instead of separate files
--watch	False	Watch source files and regenerate docs on change
-v, --verbose	0	Increase verbosity (-v = INFO, -vv = DEBUG)
--version	—	Show version and exit

Configuration priority: CLI flags > [tool.pydoc2markdown] in pyproject.toml > built-in defaults.

Configuration

You can set default values in your pyproject.toml under the [tool.pydoc2markdown] section:

[tool.pydoc2markdown]
output = "docs"
theme = "default"
recursive = true

Any values set here serve as defaults and can be overridden by CLI flags.

Library API

DocstringParser

from pathlib import Path
from pydoc2markdown import DocstringParser

parser = DocstringParser()
modules = parser.parse(Path("src/my_package"), recursive=True)

MarkdownGenerator

from pathlib import Path
from pydoc2markdown import DocstringParser, MarkdownGenerator

# Parse modules
parser = DocstringParser()
modules = parser.parse(Path("src/my_package"), recursive=True)

# Default theme, separate files
gen = MarkdownGenerator(theme="default")
gen.generate(modules, output_dir=Path("docs"))

# Minimal theme
gen_min = MarkdownGenerator(theme="minimal")
gen_min.generate(modules, output_dir=Path("docs_minimal"))

# Custom template
gen_tmpl = MarkdownGenerator(template_path=Path("my_template.md.j2"))
gen_tmpl.generate(modules, output_dir=Path("docs_custom"))

# Single combined file
gen.generate_single_file(modules, output_path=Path("docs/README.md"))

# Markdown string for a single module
md_string = gen.generate_string(modules[0])

Supported Docstring Formats

PyDoc2Markdown uses docstring-parser and supports:

Style	Support	Notes
Google	✅ Full	Args, Returns, Raises, Attributes, Examples
NumPy	✅ Full	Parameters, Returns, Raises, Attributes, Examples
reStructuredText (reST)	⚠️ Partial	Basic structure via docstring-parser; :param:, :returns:, :raises: fields

Example Output

Given this source file:

from pydantic import BaseModel, Field


class User(BaseModel):
    """A user in the system."""

    id: int = Field(description="Unique identifier")
    email: str = Field(default="", description="Email address")
    is_active: bool = True


class UserService:
    """Service for managing users."""

    def get_user(self, user_id: int) -> User:
        """Fetch a user by ID.

        Args:
            user_id: The user's unique identifier.

        Returns:
            The requested User instance.

        Raises:
            ValueError: If the user does not exist.
        """
        ...

Running pydoc2markdown src/users.py -o docs produces:

# users

## Table of Contents

- [Classes](#classes)
  - [`User`](#user)
  - [`UserService`](#userservice)

## Classes

### `User` *(Pydantic)*

A user in the system.

#### Pydantic Fields

| Name | Type | Default | Description |
|------|------|---------|-------------|
| `id` | `int` | *required* | Unique identifier |
| `email` | `str` | `""` | Email address |
| `is_active` | `bool` | `True` | - |

### `UserService`

Service for managing users.

#### Methods

##### `get_user`

Fetch a user by ID.

**Parameters:**

| Name | Type | Default | Description |
|------|------|---------|-------------|
| `user_id` | `int` | *required* | The user's unique identifier. |

**Returns:** `User`

**Raises:**

- `ValueError` — If the user does not exist.

For a more complete example, see the examples/ directory or the pre-built documentation in examples/docs/.

Documentation

• Contributing Guide
• Project Structure

License

MIT License — see LICENSE for details.