Credo is a static code analysis tool for the Elixir language with a focus on teaching and code consistency.
It implements its own style guide.
credo
can show you refactoring opportunities in your code, complex and duplicated code fragments, warn you about common mistakes, show inconsistencies in your naming scheme and - if needed - help you enforce a desired coding style.
If you are a Rubyist it is best described as an opinionated mix between Inch and Rubocop.
The easiest way to add Credo to your project is by using Mix.
Add :credo
as a dependency to your project's mix.exs
:
defp deps do
[
{:credo, "~> 0.9.1", only: [:dev, :test], runtime: false}
]
end
And run:
$ mix deps.get
To run credo in the current project, just type:
$ mix credo
This will run the code analysis and suggest places to edit your code.
If you want to enforce a style guide and need a more traditional linting experience, try
$ mix credo --strict
and continue reading in the Configuration section.
If you want the list of issues grouped by files instead of suggestions just type:
$ mix credo list
You will get output like this:
┃ Refactoring opportunities
┃
┃ [R] ↗ If/else blocks should not have a negated condition in `if`.
┃ lib/phoenix/channel.ex:26 (Phoenix.Channel.subscribe)
┃ [R] → Function is too complex (max ABC is 15, was 43).
┃ lib/phoenix/router.ex:563:8 (Phoenix.Router.add_resources)
┃ [R] → Function is too complex (max ABC is 15, was 16).
┃ lib/phoenix/router/socket.ex:12:12 (Phoenix.Router.Socket.channel)
┃
Now you might want to know more about that particular entry, just copy the filename+line-number combo into the command:
$ mix credo lib/phoenix/channel.ex:26
┃ Phoenix.Channel
┃
┃ [R] Category: refactor
┃ ↗ Priority: medium
┃
┃ If/else blocks should not have a negated condition in `if`.
┃ lib/phoenix/channel.ex:26 (Phoenix.Channel.subscribe)
┃
┃ __ CODE IN QUESTION
┃
┃ if !Socket.authenticated?(socket, channel, topic) do
┃
┃ __ WHY IT MATTERS
┃
┃ An `if` block with a negated condition should not contain an else block.
┃
┃ So while this is fine:
┃
┃ if !allowed? do
┃ raise "Not allowed!"
┃ end
┃
┃ The code in this example ...
┃
┃ if !allowed? do
┃ raise "Not allowed!"
┃ else
┃ proceed_as_planned
┃ end
┃
┃ ... should be refactored to look like this:
┃
┃ if allowed? do
┃ proceed_as_planned
┃ else
┃ raise "Not allowed!"
┃ end
┃
┃ The reason for this is not a technical but a human one. It is easier to wrap
┃ your head around a positive condition and then thinking "and else we do ...".
┃
┃ In the above example raising the error in case something is not allowed
┃ might seem so important to put it first. But when you revisit this code a
┃ while later or have to introduce a colleague to it, you might be surprised
┃ how much clearer things get when the "happy path" comes first.
Credo is configured via a file called .credo.exs
. This file can live in your project's config/
or root folder, both is fine.
This also works for umbrella projects, where you can have individual .credo.exs
files for each app or a global one in the umbrella's config/
or root folder.
# config/.credo.exs
%{
configs: [
%{
name: "default",
files: %{
included: ["lib/", "src/", "web/", "apps/"],
excluded: []
},
checks: [
{Credo.Check.Consistency.TabsOrSpaces},
# For some checks, like AliasUsage, you can only customize the priority
# Priority values are: `low, normal, high, higher`
{Credo.Check.Design.AliasUsage, priority: :low},
# For others you can also set parameters
{Credo.Check.Readability.MaxLineLength, priority: :low, max_length: 80},
# You can also customize the exit_status of each check.
# If you don't want TODO comments to cause `mix credo` to fail, just
# set this value to 0 (zero).
{Credo.Check.Design.TagTODO, exit_status: 2},
# To deactivate a check:
# Put `false` as second element:
{Credo.Check.Design.TagFIXME, false},
# ... several checks omitted for readability ...
]
}
]
}
You can use mix credo gen.config
to generate a complete example configuration.
Users of Credo can now disable individual lines or files for all or just specific checks.
defp do_stuff() do
# credo:disable-for-next-line
IO.inspect {:we_want_this_inspect_in_production!}
end
There are four config comments:
# credo:disable-for-this-file
- to disable for the entire file# credo:disable-for-next-line
- to disable for the next line# credo:disable-for-previous-line
- to disable for the previous line# credo:disable-for-lines:<count>
- to disable for the given number of lines (negative for previous lines)
Each of these can also take the name of the check you want to disable:
defp my_fun() do
# credo:disable-for-next-line Credo.Check.Warning.IoInspect
IO.inspect {:we_want_this_inspect_in_production!}
end
Lastly, you can put a regular expression (/.+/
) instead of a check name to disable multiple checks (or if you do not want to type out the checks):
defp my_fun() do
# credo:disable-for-next-line /\.Warning\./
IO.inspect {:we_want_this_inspect_in_production!}
end
Credo fails with an exit status != 0 if it shows any issues. This enables shell based pipeline workflows (e.g. on CI systems) which test Credo compliance.
The exit status of each check is used to construct a bit map of the types of issues which were encountered by or-ing them together to produce the final result:
use Bitwise
issues
|> Enum.map(&(&1.exit_status))
|> Enum.reduce(0, &(&1 ||| &2))
This way you can reason about the encountered issues right from the exit status.
Default values for the checks are based on their category:
consistency: 1
design: 2
readability: 4
refactor: 8
warning: 16
So an exit status of 12 tells you that you have only Readability Issues and Refactoring Opportunities, but e.g. no Warnings.
suggest
is the default command of Credo. It suggests issues to fix in your code, but it cuts the list to a digestable count. If you want to see the full list, use the --all
switch.
Example usage:
$ mix credo # display standard report
$ mix credo suggest # same thing, since it's the default command
$ mix credo --all --format=json # include low priority issues, output as JSON
$ mix credo suggest --help # more options
list
also suggests issues, but it groups them by file and does NOT cut the list to a certain count.
Example usage:
$ mix credo list # show issues grouped by file
$ mix credo list --format=oneline # show issues grouped by file, one issue per line
$ mix credo list --format=oneline -a # same thing, include low priority issues
$ mix credo list --help # more options
explain
allows you to dig deeper into an issue, by showing you details about the issue and the reasoning by it being reported. To be convenient, you can just copy-paste the filename:line_number:column
string from the report behind the Credo command to check it out.
Credits: This is inspired by how you can snap the info from failed tests behind mix test
.
Example usage:
$ mix credo lib/my_app/server.ex:10:24 # show explanation for the issue
$ mix credo explain lib/my_app/server.ex:10:24 # same thing
There are no additional options.
categories
shows you all issue categories and explains their semantics.
There are no additional options.
Use --format
to format the output in one of the following formats:
To run only a subset of checks, you can use the --checks
or the -c
alias (also aliased as --only
).
# to only run the Readability checks, use:
$ mix credo --only readability
# to only run Readability checks and Warnings, use:
$ mix credo --only readability,warning
The second example illustrates how the command takes a comma-separated list. All commands matching any of the passed items will be run.
You can use partial names to quickly run checks. mix credo --only todo
will show all # TODO
comments since todo
will match Credo.Check.Design.TagTODO. mix credo --only inspect
will show you all calls to IO.inspect
since it matches Credo.Check.Warning.IoInspect.
To ignore selected checks, you can use the --ignore-checks
or the -i
alias (also aliased as --ignore
).
# to ignore all Readability checks, use:
$ mix credo --ignore readability
# to ignore all Readability checks and the DuplicatedCode check, use:
$ mix credo --ignore readability,duplicated
The second example illustrates how the command takes a comma-separated list. All commands matching any of the passed items will be ignored.
You can use partial names to quickly exclude checks. mix credo --ignore nameredec
will exclude all checks for variables/parameters having the same name as declared functions by matching Credo.Check.Warning.NameRedeclarationBy...
You can also use Credo to parse source that has been piped directly into it.
This is especially useful when integrating with external editors. You can use this feature by passing the --read-from-stdin
option as follows:
$ echo 'IO.puts("hello world");' | mix credo --format=flycheck --read-from-stdin
# stdin:1: C: There is no whitespace around parentheses/brackets most of the time, but here there is.
Notice the origin if the source is coming annotated as stdin
, you can change this annotation by passing it along after option like so:
$ echo 'IO.puts("hello world");' | mix credo --format=flycheck --read-from-stdin /path/representing/the_current/source.ex
# /path/representing/the_current/source.ex:1: C: There is no whitespace around parentheses/brackets most of the time, but here there is.
Do note with the passed option as filename is a stub that is just used to prefix the error and so certain editors can annotate the original file.
If you do not want or are not allowed to include Credo in the current project you can also install it as an archive. For this, you also need to install bunt:
git clone [email protected]:rrrene/bunt.git
cd bunt
mix archive.build
mix archive.install
cd -
git clone [email protected]:rrrene/credo.git
cd credo
mix deps.get
mix archive.build
mix archive.install
Important: You have to install bunt
as well:
git clone https://github.com/rrrene/bunt
cd bunt
mix archive.build
mix archive.install
You will now be able to invoke credo as usual through Mix with mix credo
. This option is especially handy so credo can be used by external editors.
Use the --verbose
switch to include the code snippets in question in the output.
Use the --all-priorities
switch to include low priority issues in the output (aliased as --strict
).
Like any code linter, Credo reports issues. Contrary to many other linters these issues are not created equal. Each issue is assigned a priority, based on a base priority set by the config and a dynamic component based on violation severity and location in the source code.
These priorities hint at the importance of each issue and are displayed in the command-line interface using arrows: ↑ ↗ → ↘ ↓
By default, only issues with a positive priority are part of the report (↑ ↗ →).
These checks take a look at your code and ensure a consistent coding style. Using tabs or spaces? Both is fine, just don't mix them or Credo will tell you.
Readability checks do not concern themselves with the technical correctness of your code, but how easy it is to digest.
The Refactor checks show you opportunities to avoid future problems and technical debt.
While refactor checks show you possible problems, these checks try to highlight possibilities, like - potentially intended - duplicated code or TODO:
and FIXME
annotations.
These checks warn you about things that are potentially dangerous, like a missed call to IEx.pry
or a call to String.downcase
without saving the result.
Some IDEs and editors are able to run mix credo
in the background and mark and issues directly in the editor view.
- IntelliJ Elixir - Elixir plugin for JetBrains IDEs (IntelliJ IDEA, Rubymine, PHPStorm, PyCharm, etc)
- Fork it!
- Create your feature branch (
git checkout -b my-new-feature
) - Commit your changes (
git commit -am 'Add some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create new Pull Request
René Föhring (@rrrene)
Credo is released under the MIT License. See the LICENSE file for further details.