Skip to main content

Create Python API documentation in Markdown format

Project description


insipired by the Keras Documentation

Pydocmd uses MkDocs and extended Markdown syntax to generate beautiful Python API documentation.


  • [x] Support + suffix to include documented members of a module/class
  • [ ] Expand and link cross-references (eg. #SomeClass)
  • [ ] Parse, format and link types listed in parameter/member/raise/return type docstrings (eg. someattr (int): This is...)


pip install pydoc-markdown
pip install git+  # latest development version


Pydocmd can generate plain Markdown files from Python modules using the pydocmd simple command. Specify one or more module names on the command-line. Supports the + syntax to include members of the module (or ++ to include members of the members, etc.)

pydocmd simple mypackage+ mypackage.mymodule+ >

Alternatively, pydocmd wraps the MkDocs command-line interface and generates the markdown pages beforehand. Simply use pydocmd build to build the documentation, or pydocmd serve to serve the documentation on a local HTTP server. The pydocmd gh-deploy from MkDocs is also supported.

A configuration file pydocmd.yml is required to use pydocmd in this mode. Below is an example configuration. To get started, create docs/ directory and a file pydocmd.yml inside of it. Copy the configuration below and adjust it to your needs, then run pydocmd build from the docs/ directory.

site_name: "My Documentation"

# This tells pydocmd which pages to generate from which Python modules,
# functions and classes. At the first level is the page name, below that
# is a tree of Python member names (modules, classes, etc.) that should be
# documented. Higher indentation leads to smaller header size.
- baz/
  - foobar.baz:
    - foobar.baz.CoolClass+     # (+ to include members)
    - foobar.baz.some_function
- baz/
  - foobar.more++               # (++ to include members, and their members)

# MkDocs pages configuration. The `<<` operator is sugar added by pydocmd
# that allows you to use an external Markdown file (eg. your project's README)
# in the documentation. The path must be relative to current working directory.
- Home: << ../
- foobar.baz:
  - Cool Stuff: baz/

# These options all show off their default values. You don't have to add
# them to your configuration if you're fine with the default.
docs_dir: sources
gens_dir: _build/pydocmd     # This will end up as the MkDocs 'docs_dir'
site_dir: _build/site
theme:    readthedocs
loader:   pydocmd.loader.PythonLoader
preprocessor: pydocmd.preprocessor.Preprocessor

# Whether to output headers as markdown or HTML.  Used to workaround
#  The default is
# to generate HTML with unique and meaningful id tags, which can't be done with
# markdown.
# Note: if using the simple generator mode, this will default to 'markdown'
# instead of 'html'.
headers: html

# Additional search path for your Python module. If you use Pydocmd from a
# subdirectory of your project (eg. docs/), you may want to add the parent
# directory here.
- ..



Symbols in the same namespace may be referenced by using a hash-symbol (#) directly followed by the symbols' name, including relative references. Note that using parentheses for function names is encouraged and will be ignored and automatically added when converting docstrings. Examples: #ClassName.member or #mod.function().

For absolute references for modules or members in names that are not available in the current global namespace, #::mod.member must be used (note the two preceeding two double-colons).

For long reference names where only some part of the name should be displayed, the syntax can be used, where X is the number of elements to keep. If X is omitted, it will be assumed 1. Example: results in only name being displayed.

In order to append additional characters that are not included in the actual reference name, another hash-symbol can be used, like #Signal#s.


Sections can be generated with the Markdown # <Title> syntax. It is important to add a whitespace after the hash-symbol (#), as otherwise it would represent a cross-reference. Some special sections alter the rendered result of their content, including

  • Arguments (1)
  • Parameters (1)
  • Attributes (1)
  • Members (1)
  • Raises (2)
  • Returns (2)

(1): Lines beginning with <ident> [(<type>[, ...])]: are treated as argument/parameter or attribute/member declarations. Types listed inside the parenthesis (optional) are cross-linked, if possible. For attribute/member declarations, the identifier is typed in a monospace font.

(2): Lines beginning with <type>[, ...]: are treated as raise/return type declarations and the type names are cross-linked, if possible.

Lines following a name's description are considered part of the most recent documentation unless separated by another declaration or an empty line. <type> placeholders can also be tuples in the form (<type>[, ...]).

Code Blocks

GitHub-style Markdown code-blocks with language annotations can be used.

>>> for i in range(100):


v2.0.5 (2018-11-15)

  • Now copies all files from the docs_dir (to include images etc.) (see #56)
  • Fix error with delayed imports changing dictionary size during iteration (see #57)
  • Add headers option which can be of value 'html' or 'markdown' (see #55)
  • Default headers option to 'markdown' in simple mode (see #59)

v2.0.4 (2018-07-24)

  • Add -c key=value argument for generate and simple command
  • Add filter=["docstring"] option (#43)
  • Fix regex for detecting cross-references (#44)
  • Handle classes that don't define __init__() (PR#51)
  • Add support for reStructuredText Markup (eg. :class:`MyClass`) (PR#46, #1)
  • Handle @property functions (PR#50)


  • Fix #41, #36, #31
  • Merged #39


  • Fix #25 -- Text is incorrectly rendered as code
  • Fix #26 -- Broken links for URLs with fragment identifiers
  • No longer transforms titles in a docstring that are indented (eg. to avoid an indented code block with a # comment to be corrupted)


  • Support additional_search_path key in configuration
  • Render headers as HTML <hX> tags rather than Markdown tags, so we assign a proper ID to them
  • Fix #21 -- AttributeError: 'module' object has no attribute 'signature'
  • Now requires the six module
  • FIx #22 -- No blank space after header does not render codeblocks


  • Complete overhaul of pydoc-markdown employing MkDocs and the Markdown module.

Copyright © 2017-2018 Niklas Rosenstein

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for pydoc-markdown, version 2.0.5
Filename, size File type Python version Upload date Hashes
Filename, size pydoc-markdown-2.0.5.tar.gz (14.3 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page