reStructuredText linter

restructuredtext, restructured, text, rest, rst, lint
pip install restructuredtext-lint==1.4.0



Build Status

reStructuredText linter

This was created out of frustration with PyPI; it sucks finding out your reST is invalid after uploading it. It is being developed in junction with a Sublime Text linter.

Getting Started

Install the module with: pip install restructuredtext_lint

import restructuredtext_lint
errors = restructuredtext_lint.lint("""
Hello World

# `errors` will be list of system messages
# [<system_message: <paragraph...><literal_block...>>]
errors[0].message  # Title underline too short.

CLI Utility

For your convenience, we present a CLI utility rst-lint (also available as restructuredtext-lint).

$ rst-lint --help
usage: rst-lint [-h] [--version] [--format {text,json}] [--encoding ENCODING]
                [--level {debug,info,warning,error,severe}]
                [--rst-prolog RST_PROLOG]
                path [path ...]

Lint reStructuredText files. Returns 0 if all files pass linting, 1 for an
internal error, and 2 if linting failed.

positional arguments:
  path                  File/folder to lint

optional arguments:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
  --format {text,json}  Format of the output (default: "text")
  --encoding ENCODING   Encoding of the input file (e.g. "utf-8")
  --level {debug,info,warning,error,severe}
                        Minimum error level to report (default: "warning")
  --rst-prolog RST_PROLOG
                        reStructuredText content to prepend to all files
                        (useful for substitutions)

$ rst-lint README.rst
WARNING README.rst:2 Title underline too short.

Other tools

restructuredtext-lint is also integrated in other tools. A list can be found and updated in our wiki

PyPI issues

While a document may lint cleanly locally, there can be issues when submitted it to PyPI. Here are some common problems:

  • Usage of non-builtin lexers (e.g. bibtex) will pass locally but not be recognized/parsable on PyPI
    • This is due to PyPI not having a non-builtin lexer installed
    • Please avoid non-builtin lexers to avoid complications
    • For more information, see #27
  • Relative hyperlinks will not work (e.g. ./UNLICENSE)
    • According to Stack Overflow, hyperlinks must use a scheme (e.g. http, https) and that scheme must be whitelisted
    • Please use absolute hyperlinks (e.g.


restructuredtext-lint exposes a lint and lint_file function

restructuredtext_lint.lint(content, filepath=None, rst_prolog=None)

Lint reStructuredText and return errors

  • content String - reStructuredText to be linted
  • filepath String - Optional path to file, this will be returned as the source
  • rst_prolog String - Optional content to prepend to content, line numbers will be offset to ignore this


  • errors List - List of errors
    • Each error is a class from docutils with the following attrs
      • line Integer|None - Line where the error occurred
        • On rare occasions, this will be None (e.g. anonymous link mismatch)
      • source String - filepath provided in parameters
      • level Integer - Level of the warning
        • Levels represent 'info': 1, 'warning': 2, 'error': 3, 'severe': 4
      • type String - Noun describing the error level
        • Levels can be 'INFO', 'WARNING', 'ERROR', or 'SEVERE'
      • message String - Error message
      • full_message String - Error message and source lines where the error occurred
    • It should be noted that level, type, message, and full_message are custom attrs added onto the original system_message

restructuredtext_lint.lint_file(filepath, encoding=None, *args, **kwargs)

Lint a reStructuredText file and return errors

Returns: Same structure as restructuredtext_lint.lint


Under the hood, we leverage docutils for parsing reStructuredText documents. docutils supports adding new directives and roles via register_directive and register_role.


Unfortunately due to customizations in Sphinx's parser we cannot include all of its directives/roles (see #29). However, we can include some of them as one-offs. Here is an example of adding a directive from Sphinx.



.. highlight:: python

    Hello World!

# Load in our dependencies
from docutils.parsers.rst.directives import register_directive
from sphinx.directives.code import Highlight
import restructuredtext_lint

# Load our new directive
register_directive('highlight', Highlight)

# Lint our README
errors = restructuredtext_lint.lint_file('docs/sphinx/README.rst')
print errors[0].message # Error in "highlight" directive: no content permitted.


Here is an example of all invalid properties

rst = """
Some content.

Hello World
Some more content!
errors = restructuredtext_lint.lint(rst, '')
errors[0].line  # 5
errors[0].source  #
errors[0].level  # 2
errors[0].type  # WARNING
errors[0].message  # Title underline too short.
errors[0].full_message  # Title underline too short.
                        # Hello World
                        # =======


In lieu of a formal styleguide, take care to maintain the existing coding style. Add unit tests for any new or changed functionality. Test via nosetests.


Support this project and others by twolfson via donations.


As of Nov 22 2013, Todd Wolfson has released this repository and its contents to the public domain.

It has been released under the UNLICENSE.