Lint Review helps automate a tedious part of code review - enforcing coding standards. By using the github API Lint Review runs a repository's configured linters and updates pull requests with line comments where lint errors would be introduced.
Lint Review requires:
- Python 2.7 (It will probably work in 2.6, but I've only tested 2.7)
- RabbitMQ (or any other Message broker that is compatible with Celery)
- A publically addressable hostname/IP that either github or your github:enterprise can reach.
- A github account with read/write access to the repositories you want linted. This account is used to post comments on pull reviews.
Lint Review runs as two processes. A web process handles accepting webhooks from github, and a celery process handles cloning repositories and running lint tools. You'll also need to have rabbitmq-server running.
If you don't want to go through the trouble of setting up your own installation of lint-review, stickler-ci.com provides a hosted version of lint-review featuring all the linters installed, and an easier to use YAML config file.
You install Lint Review by cloning the repository and installing dependecies,
or by using docker. If you are not using docker, it is recommended that you use
virtualenv
to save shaving yaks down the road.
git clone git://github.com/markstory/lint-review.git
cd lint-review
virtualenv env
source env/bin/activate
pip install .
In addition to installing the dependencies for lint-review you will also need to
install the various lint tools you want to use. Lint-review requires that lint
tools be installed relative to the source code. pep8
and flake8
will be
installed as dependencies for lint-review, but any other tools will need to be
installed using their respective installers. To install all the tools run the following:
# Install ruby tools
bundle install --path ./bundle
# Install javascript tools.
npm install
# Install PHP tools
php composer.phar install
Once the dependencies are installed you should configure the repositories you want to review.
To use aws, you can use packer to generate an Amazon Machine Image to run the application
Once you have your AWS credentials loaded you can run:
packer build packer.json
Which will generate an AMI with the application and lint checkers installed. This can then be configured using cloud-init by writing the configuration to /etc/environment
To use docker, you'll need to install both docker, docker-compose and possibly docker toolbox depending on your operating system. Once you have the docker installed, you can boot up lint-review into docker using:
docker-compose up -d broker worker web
Edit docker-composer.yml and customise your configuration by setting keys under environment for the
web and worker processes. For the most basic installation you'll need to set at least GITHUB_USERNAME
and GITHUB_PASSWORD
(or GITHUB_OAUTH_TOKEN
), and LINTREVIEW_SERVER_NAME
.
Lint review is configured through a settings file. Both the web app and celery process share the same configuration file, so configuration is easier to manage and share.
-
Copy the
settings.sample.py
tosettings.py
-
Edit the required configuration options, or set the correct environment variables.
-
Set the
LINTREVIEW_SETTINGS
environment variable to the path of your configuration files. In *nix system this can be done via:export LINTREVIEW_SETTINGS='/path/to/settings.py'
-
You can skip setting
LINTREVIEW_SETTINGS
if you're running lintreview from a directory containing yoursettings.py
file.
You can also have per install configuration files by defining the LINTRC_DEFAULTS
config
option in your settings file. This file should be a .lintrc
config file. It will be merged
with each projects .lintrc
before running tools. This gives you an easy way to have
global configuration for tools.
Once you've configured the server processes, it is time to setup some repositories to be checked.
Before Lint Review can check pull requests on a repository webhooks will need to be installed. You can install webhooks by running the built-in command line tool:
source env/bin/activate
lintreview register mark awesome-stuff
Or, if you're using Docker:
docker-compose run web lintreview register mark awesome-stuff
The above register webhooks for the given user & repository. You can use the
--user
and --password
options to provide the repository admin credentials
if the user lint-review runs as does not have admin access to the repository.
You can also use the cli tool to remove webhooks:
source env/bin/activate
lintreview unregister mark awesome-stuff
Warning The current web server name will be registered with github. Make sure it is configured properly before registering hooks, or you'll need to remove any registered hooks and start over.
Lint Review use hidden ini files to configure the tools used on each project. The
.lintrc
file defines the various linting tools and any arguments for each one. Lint
tools must be tools Lint Review knows about. See lint tools for available
tools. A sample .lintrc
file would look like.
[files]
ignore = generated/*
vendor/*
[tools]
linters = pep8, jshint
[tool_pep8]
ignore = W2,E401
[tool_jshint]
config = path/to/jshint.json
The [tools]
section is required, and linters
should be a list of
linters your project uses. Each tool can also have a section prefixed with tool_
to define additional configuration options for each tool. The documentation for
each tool outlines which options are supported.
The [files]
section is optional and allows you to define ignore patterns. These patterns
are used to find and exclude files when doing a review. Ignore patterns use glob expressions
to find files. The patterns start at the reviewed repository root. If you need to ignore mulitple
patterns separate them with new lines.
After setting up configuration you'll need to start up both processes:
source env/bin/activate
gunicorn -c settings.py lintreview.web:app
celery -A lintreview.tasks worker
Now when ever a pull request is opened or updated for a registered repository new jobs will be spun up and lint will be checked and commented on.
Uses the flake8 module to check code.
Options
ignore
Set which pep8 error codes you wish to ignore.exclude
Exclude files or directories which match these comma separated patterns (default: .svn, CVS, .bzr, .hg, .git)filename
When parsing directories, only check filenames matching these comma separated patterns (default:*.py
)select
Select errors and warnings (e.g. E,W6)max-line-length
Set maximum allowed line length (default: 79)format
Set the error format [default|pylint|]max-complexity
McCabe complexity thresholdsnippet
Interacts with flake8-snippets allowing you to trigger errors on specific snippets you want to disallow.
These options are passed into flake8 as cli options.
Uses the pep8 module to check code.
Options
ignore
Set which pep8 error codes you wish to ignore.
Uses the phpcs PEAR library to do style checks on PHP, Javascript and or CSS files.
Options
standard
The coding standard to use. By default thePSR2
standard is used. You can use any of the built-in standards or provide your own inside your project directory.extensions
The extensions to check. By default only.php
files will be checked.tab_width
The number of spaces to convert tabs into, this is useful for projects using tabs for indentation.
Uses the JSCS npm module to check javascript files. Before you can use this linter you'll need to install nodejs and the jscs npm package:
cd path/to/lintreview
npm install jscs
Options
config
Provide a path to the json config file for JSCS.preset
Set which JSCS preset you want to use. Unused ifconfig
is set.
Uses the jshint npm module to check javascript files. Before you can use this linter you'll need to install nodejs and the jshint npm package:
cd path/to/lintreview
npm install jshint
Options
config
Provide a path to the json config file for jshint.
Uses the eslint npm module to check javascript files. Before you can use this linter you'll need to install the eslint npm package:
cd path/to/lintreview
npm install eslint
Options
config
Provide a path to the json config file for eslint.
Uses the jsonlint script from demjson python module to check javascript object notation files.
Options
- None currently supported
Uses the csslint npm module to check css files. Before
you can use this linter you'll need to install nodejs and the csslint
npm package:
cd path/to/lintreview
npm install
Both warnings and errors will be turned into code review comments. If you don't want code review comments for specific rules, you should ignore them.
Options
ignore
A comma separated list of rule ids to ignore.
Uses the sass-lint npm module to check scss and sass files.
you can use this linter you'll need to install nodejs and the sass-lint
npm package:
cd path/to/lintreview
npm install
Options
ignore
A comma separated list of files to ignore.config
Project relative path to the sass-lint config file you want applied.
Uses the rubocop gem to check ruby files. You'll need to install it to use it:
gem install rubocop
Options
display_cop_names
Set totrue
to pass display cop names in offense messages.
.rubocop.yml
files will be respected, as described
here.
Uses the puppet-lint gem to check puppet manifests against the puppetlabs style guide.
You'll need to install it to use it:
gem install puppet-lint
Options
No options. .puppet-lintrc
files will be respected, to allow each project to disable
checks. A list of checks can be found by running "puppet-lint --help"
Uses the Foodcritic gem to check Chef files. You'll need to install Foodcritic:
gem install foodcritic
Options
path
If your cookbooks aren't stored in the root, use this to set the path that foodcritic runs against. Example:path = cookbooks
Uses the yamllint module to check yaml and yml files.
Options
config
Provide a path to the yaml config file for yamlhint.
Uses shellcheck to lint shell scripts.
Options
shell
Select which shell to use. Options are: bash, sh, ksh or zsh. Default:sh
exclude
String of checks to ignore. Example: SC2154,SC2069
Uses ansible-lint to lint Ansible plays.
Options
ignore
Set which ansible-lint error codes you wish to ignore.
Uses go-lint to lint go code.
Options
min_confidence
Set the confidence level of a problem before it is reported.