mkgendocs 0.2

A Python package for automatically generating documentation from Google docstrings in markdown.

mkgendocs

A Python package for automatically generating documentation pages in markdown for Python source files by parsing Google style docstring. The markdown output makes it ideal to combine with mkdocs.

Contrary to other packages, this does not use the inspect package to access signatures and docstrings. Instead, it uses python Abstract Syntax Trees (AST) to get this from the source files without executing the code -through the ast package. It also makes use of `astor (AST observe/rewrite) to convert function or class signatures back to source code strings.

Installation

Install mkgendocs from PyPI

pip install mkgendocs

Usage

gendocs --config mkgendocs.yml

A sources directory is created with the documentation that was automatically generated. Any examples in a "examples" directory are automatically copied over to the documentation, the module level docstrings are copied

Configuration Example

sources_dir: docs/sources
templates_dir: docs/templates
docstring_template: docstring_template.md

pages:
  - page: "train/model.md"
    source: "tensorx/train/model.py"
    methods:
      - Model:
          - train
          - set_optimizer
  - page: "layers/core.md"
    source: 'tensorx/layers.py'
    classes:
      - Linear
      - Module
  - page: "math.md"
    source: 'tensorx/math.py'
    functions:
      - sparse_multiply_dense

• sources_dir: directory where the resulting markdown files are created
• templates_dir: directory where template files can be stored. All the folders and files are copied to the sources directory. Any markdown files are used as templates with the tag {{autogenerated}} in the template files being replaced by the generated documentation.
• pages: list of pages to be automatically generated from the respective source files and templates:
  • page: path for page template / sources dir for the resulting page;
  • source: path to source file from which the page is to be genrated;
  • methods: list of class to method name dictionaries;
  • classes: list of classes to be fully documented;
  • functions: list of functions to be documented.