deb-python-neutronclient/neutronclient/cliff_sphinxext.py
Akihiro Motoki f2a4d5253c doc: autogenerate neutron CLI reference
This replaces the doc-migration work. We can avoid the maintenance
cost by generating CLI reference for neutron CLI like this.

This patch introduces a new sphinx directive to render CLI
global options to achieve the goal. This can be migrated to
cliff sphinxext later.

Change-Id: I7089df3f7fc6475ebdd1e71221492baf12f18226
2017-07-10 11:26:25 +00:00

386 lines
13 KiB
Python

# Copyright (C) 2017, Red Hat, Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import argparse
import fnmatch
import re
from docutils import nodes
from docutils.parsers import rst
from docutils.parsers.rst import directives
from docutils import statemachine
from oslo_utils import importutils
from cliff import commandmanager
def _indent(text):
"""Indent by four spaces."""
prefix = ' ' * 4
def prefixed_lines():
for line in text.splitlines(True):
yield (prefix + line if line.strip() else line)
return ''.join(prefixed_lines())
def _format_description(parser):
"""Get parser description.
We parse this as reStructuredText, allowing users to embed rich
information in their help messages if they so choose.
"""
for line in statemachine.string2lines(
parser.description, tab_width=4, convert_whitespace=True):
yield line
def _format_usage(parser):
"""Get usage without a prefix."""
fmt = argparse.HelpFormatter(parser.prog)
optionals = parser._get_optional_actions()
positionals = parser._get_positional_actions()
groups = parser._mutually_exclusive_groups
# hacked variant of the regex used by the actual argparse module. Unlike
# that version, this one attempts to group long and short opts with their
# optional arguments ensuring that, for example, '---format <FORMAT>'
# becomes ['--format <FORMAT>'] and not ['--format', '<FORMAT>'].
# Yes, they really do use regexes to break apart and rewrap their help
# string. Don't ask me why.
part_regexp = r'\(.*?\)+|\[.*?\]+|(?:(?:-\w|--\w+)(?:\s+<\w+>)?)|\S+'
opt_usage = fmt._format_actions_usage(optionals, groups)
pos_usage = fmt._format_actions_usage(positionals, groups)
opt_parts = re.findall(part_regexp, opt_usage)
pos_parts = re.findall(part_regexp, pos_usage)
parts = opt_parts + pos_parts
if len(' '.join([parser.prog] + parts)) < 72:
return [' '.join([parser.prog] + parts)]
return [parser.prog] + [_indent(x) for x in parts]
def _format_epilog(parser):
"""Get parser epilog.
We parse this as reStructuredText, allowing users to embed rich
information in their help messages if they so choose.
"""
for line in statemachine.string2lines(
parser.epilog, tab_width=4, convert_whitespace=True):
yield line
def _format_positional_action(action):
"""Format a positional action."""
if action.help == argparse.SUPPRESS:
return
# NOTE(stephenfin): We strip all types of brackets from 'metavar' because
# the 'option' directive dictates that only option argument names should be
# surrounded by angle brackets
yield '.. option:: {}'.format(
(action.metavar or action.dest).strip('<>[]() '))
if action.help:
yield ''
for line in statemachine.string2lines(
action.help, tab_width=4, convert_whitespace=True):
yield _indent(line)
def _format_optional_action(action):
"""Format an optional action."""
if action.help == argparse.SUPPRESS:
return
if action.nargs == 0:
yield '.. option:: {}'.format(', '.join(action.option_strings))
else:
# TODO(stephenfin): At some point, we may wish to provide more
# information about the options themselves, for example, if nargs is
# specified
option_strings = [' '.join(
[x, action.metavar or '<{}>'.format(action.dest.upper())])
for x in action.option_strings]
yield '.. option:: {}'.format(', '.join(option_strings))
if action.help:
yield ''
for line in statemachine.string2lines(
action.help, tab_width=4, convert_whitespace=True):
yield _indent(line)
def _format_parser(parser):
"""Format the output of an argparse 'ArgumentParser' object.
Given the following parser::
>>> import argparse
>>> parser = argparse.ArgumentParser(prog='hello-world', \
description='This is my description.',
epilog='This is my epilog')
>>> parser.add_argument('name', help='User name', metavar='<name>')
>>> parser.add_argument('--language', action='store', dest='lang', \
help='Greeting language')
Returns the following::
This is my description.
.. program:: hello-world
.. code:: shell
hello-world [-h] [--language LANG] <name>
.. option:: name
User name
.. option:: --language LANG
Greeting language
.. option:: -h, --help
Show this help message and exit
This is my epilog.
"""
if parser.description:
for line in _format_description(parser):
yield line
yield ''
yield '.. program:: {}'.format(parser.prog)
yield '.. code-block:: shell'
yield ''
for line in _format_usage(parser):
yield _indent(line)
yield ''
# In argparse, all arguments and parameters are known as "actions".
# Optional actions are what would be known as flags or options in other
# libraries, while positional actions would generally be known as
# arguments. We present these slightly differently.
for action in parser._get_optional_actions():
for line in _format_optional_action(action):
yield line
yield ''
for action in parser._get_positional_actions():
for line in _format_positional_action(action):
yield line
yield ''
if parser.epilog:
for line in _format_epilog(parser):
yield line
yield ''
class AutoprogramCliffDirective(rst.Directive):
"""Auto-document a subclass of `cliff.command.Command`."""
has_content = False
required_arguments = 1
option_spec = {
'command': directives.unchanged,
'ignored': directives.unchanged,
'application': directives.unchanged,
}
def _load_command(self, manager, command_name):
"""Load a command using an instance of a `CommandManager`."""
try:
# find_command expects the value of argv so split to emulate that
return manager.find_command(command_name.split())[0]
except ValueError:
raise self.error('"{}" is not a valid command in the "{}" '
'namespace'.format(
command_name, manager.namespace))
def _generate_nodes(self, title, command_name, command_class,
ignored_opts):
"""Generate the relevant Sphinx nodes.
This is a little funky. Parts of this use raw docutils nodes while
other parts use reStructuredText and nested parsing. The reason for
this is simple: it avoids us having to reinvent the wheel. While raw
docutils nodes are helpful for the simpler elements of the output,
they don't provide an easy way to use Sphinx's own directives, such as
the 'option' directive. Refer to [1] for more information.
[1] http://www.sphinx-doc.org/en/stable/extdev/markupapi.html
:param title: Title of command
:param command_name: Name of command, as used on the command line
:param command_class: Subclass of :py:class:`cliff.command.Command`
:param prefix: Prefix to apply before command, if any
:param ignored_opts: A list of options to exclude from output, if any
:returns: A list of nested docutil nodes
"""
command = command_class(None, None)
parser = command.get_parser(command_name)
ignored_opts = ignored_opts or []
# Drop the automatically-added help action
for action in list(parser._actions):
for option_string in action.option_strings:
if option_string in ignored_opts:
del parser._actions[parser._actions.index(action)]
break
section = nodes.section(
'',
nodes.title(text=title),
ids=[nodes.make_id(title)],
names=[nodes.fully_normalize_name(title)])
source_name = '<{}>'.format(command.__class__.__name__)
result = statemachine.ViewList()
for line in _format_parser(parser):
result.append(line, source_name)
self.state.nested_parse(result, 0, section)
return [section]
def run(self):
self.env = self.state.document.settings.env
command_pattern = self.options.get('command')
application_name = (self.options.get('application')
or self.env.config.autoprogram_cliff_application)
global_ignored = self.env.config.autoprogram_cliff_ignored
local_ignored = self.options.get('ignored', '')
local_ignored = [x.strip() for x in local_ignored.split(',')
if x.strip()]
ignored_opts = list(set(global_ignored + local_ignored))
# TODO(sfinucan): We should probably add this wildcarding functionality
# to the CommandManager itself to allow things like "show me the
# commands like 'foo *'"
manager = commandmanager.CommandManager(self.arguments[0])
if command_pattern:
commands = [x for x in manager.commands
if fnmatch.fnmatch(x, command_pattern)]
else:
commands = manager.commands.keys()
output = []
for command_name in sorted(commands):
command_class = self._load_command(manager, command_name)
title = command_name
if application_name:
command_name = ' '.join([application_name, command_name])
output.extend(self._generate_nodes(
title, command_name, command_class, ignored_opts))
return output
class CliffAppDirective(rst.Directive):
"""Auto-document a `cliff.app.App`."""
has_content = False
required_arguments = 1
option_spec = {
'arguments': directives.unchanged,
'ignored': directives.unchanged,
'application': directives.unchanged,
}
def _generate_nodes(self, title, app, app_name, ignored_opts):
"""Generate the relevant Sphinx nodes.
This is a little funky. Parts of this use raw docutils nodes while
other parts use reStructuredText and nested parsing. The reason for
this is simple: it avoids us having to reinvent the wheel. While raw
docutils nodes are helpful for the simpler elements of the output,
they don't provide an easy way to use Sphinx's own directives, such as
the 'option' directive. Refer to [1] for more information.
[1] http://www.sphinx-doc.org/en/stable/extdev/markupapi.html
:param title: Title of command
:param app: Subclass of :py:class`cliff.app.App`
:param app_name: The name of the cliff application.
This is used as the command name.
:param ignored_opts: A list of options to exclude from output, if any
:returns: A list of docutil nodes
"""
parser = app.parser
ignored_opts = ignored_opts or []
# Drop the automatically-added help action
for action in list(parser._actions):
for option_string in action.option_strings:
if option_string in ignored_opts:
del parser._actions[parser._actions.index(action)]
break
parser.prog = app_name
source_name = '<{}>'.format(app.__class__.__name__)
result = statemachine.ViewList()
for line in _format_parser(parser):
result.append(line, source_name)
section = nodes.section()
self.state.nested_parse(result, 0, section)
return section.children
def run(self):
self.env = self.state.document.settings.env
cliff_app_class = importutils.import_class(self.arguments[0])
app_arguments = self.options.get('arguments', '').split()
cliff_app = cliff_app_class(*app_arguments)
application_name = (self.options.get('application')
or self.env.config.autoprogram_cliff_application)
global_ignored = self.env.config.autoprogram_cliff_ignored
local_ignored = self.options.get('ignored', '')
local_ignored = [x.strip() for x in local_ignored.split(',')
if x.strip()]
ignored_opts = list(set(global_ignored + local_ignored))
output = []
title = application_name
output.extend(self._generate_nodes(
title, cliff_app, application_name, ignored_opts))
return output
def setup(app):
app.add_directive('autoprogram-cliff', AutoprogramCliffDirective)
app.add_config_value('autoprogram_cliff_application', '', True)
app.add_config_value('autoprogram_cliff_ignored', ['--help'], True)
app.add_directive('cliff-app', CliffAppDirective)