go/revive

mirror of https://github.com/mgechev/revive.git synced 2024-11-28 08:49:11 +02:00

Minko Gechev 6a62ee9f02 Update the list of contributors

2019-03-01 11:45:22 -08:00

23 KiB

Raw Blame History

revive

Fast, configurable, extensible, flexible, and beautiful linter for Go. Drop-in replacement of golint. Revive provides a framework for development of custom rules, and lets you define a strict preset for enhancing your development & code review processes.

Logo by Georgi Serev

Here's how revive is different from golint:

Allows to enable or disable rules using a configuration file.
Allows to configure the linting rules with a TOML file.
2x faster running the same rules as golint.
Provides functionality for disabling a specific rule or the entire linter for a file or a range of lines.
- golint allows this only for generated files.
Optional type checking. Most rules in golint do not require type checking. If you disable them in the config file, revive will run over 6x faster than golint.
Provides multiple formatters which let us customize the output.
Allows to customize the return code for the entire linter or based on the failure of only some rules.
Everyone can extend it easily with custom rules or formatters.
Revive provides more rules compared to golint.

Who uses Revive

tidb - TiDB is a distributed HTAP database compatible with the MySQL protocol
ferret - Declarative web scraping
gopass - The slightly more awesome standard unix password manager for teams
gitea - Git with a cup of tea, painless self-hosted git service
excelize - Go library for reading and writing Microsoft Excel™ (XLSX) files
aurora - aurora is a web-based Beanstalk queue server console written in Go

Open a PR to add your project.

revive

Usage

Since the default behavior of revive is compatible with golint, without providing any additional flags, the only difference you'd notice is faster execution.

Text Editors

Support for VSCode in vscode-go.
Support for Atom via linter-revive.
Support for vim via w0rp/ale:

call ale#linter#Define('go', {
\   'name': 'revive',
\   'output_stream': 'both',
\   'executable': 'revive',
\   'read_buffer': 0,
\   'command': 'revive %t',
\   'callback': 'ale#handlers#unix#HandleAsWarning',
\})

Installation

go get -u github.com/mgechev/revive

Command Line Flags

revive accepts three command line parameters:

-config [PATH] - path to config file in TOML format.
-exclude [PATTERN] - pattern for files/directories/packages to be excluded for linting. You can specify the files you want to exclude for linting either as package name (i.e. github.com/mgechev/revive), list them as individual files (i.e. file.go), directories (i.e. ./foo/...), or any combination of the three.
-formatter [NAME] - formatter to be used for the output. The currently available formatters are:
- default - will output the failures the same way that golint does.
- json - outputs the failures in JSON format.
- ndjson - outputs the failures as stream in newline delimited JSON (NDJSON) format.
- friendly - outputs the failures when found. Shows summary of all the failures.
- stylish - formats the failures in a table. Keep in mind that it doesn't stream the output so it might be perceived as slower compared to others.
- checkstyle - outputs the failures in XML format compatible with that of Java's Checkstyle.

Sample Invocations

revive -config revive.toml -exclude file1.go -exclude file2.go -formatter friendly github.com/mgechev/revive package/...

The command above will use the configuration from revive.toml
revive will ignore file1.go and file2.go
The output will be formatted with the friendly formatter
The linter will analyze github.com/mgechev/revive and the files in package

Comment Annotations

Using comments, you can disable the linter for the entire file or only range of lines:

//revive:disable

func Public() {}
//revive:enable

The snippet above, will disable revive between the revive:disable and revive:enable comments. If you skip revive:enable, the linter will be disabled for the rest of the file.

With revive:disable-next-line and revive:disable-line you can disable revive on a particular code line.

You can do the same on a rule level. In case you want to disable only a particular rule, you can use:

//revive:disable:unexported-return
func Public() private {
  return private
}
//revive:enable:unexported-return

This way, revive will not warn you for that you're returning an object of an unexported type, from an exported function.

Configuration

revive can be configured with a TOML file. Here's a sample configuration with explanation for the individual properties:

# When set to false, ignores files with "GENERATED" header, similar to golint
ignoreGeneratedHeader = true

# Sets the default severity to "warning"
severity = "warning"

# Sets the default failure confidence. This means that linting errors
# with less than 0.8 confidence will be ignored.
confidence = 0.8

# Sets the error code for failures with severity "error"
errorCode = 0

# Sets the error code for failures with severity "warning"
warningCode = 0

# Configuration of the `cyclomatic` rule. Here we specify that
# the rule should fail if it detects code with higher complexity than 10.
[rule.cyclomatic]
  arguments = [10]

# Sets the severity of the `package-comments` rule to "error".
[rule.package-comments]
  severity = "error"

Default Configuration

The default configuration of revive can be found at defaults.toml. This will enable all rules available in golint and use their default configuration (i.e. the way they are hardcoded in golint).

revive -config defaults.toml github.com/mgechev/revive

This will use the configuration file defaults.toml, the default formatter, and will run linting over the github.com/mgechev/revive package.

Custom Configuration

revive -config config.toml -formatter friendly github.com/mgechev/revive

This will use config.toml, the friendly formatter, and will run linting over the github.com/mgechev/revive package.

Recommended Configuration

The following snippet contains the recommended revive configuration that you can use in your project:

ignoreGeneratedHeader = false
severity = "warning"
confidence = 0.8
errorCode = 0
warningCode = 0

[rule.blank-imports]
[rule.context-as-argument]
[rule.context-keys-type]
[rule.dot-imports]
[rule.error-return]
[rule.error-strings]
[rule.error-naming]
[rule.exported]
[rule.if-return]
[rule.increment-decrement]
[rule.var-naming]
[rule.var-declaration]
[rule.package-comments]
[rule.range]
[rule.receiver-naming]
[rule.time-naming]
[rule.unexported-return]
[rule.indent-error-flow]
[rule.errorf]
[rule.empty-block]
[rule.superfluous-else]
[rule.unused-parameter]
[rule.unreachable-code]
[rule.redefines-builtin-id]

Available Rules

List of all available rules. The rules ported from golint are left unchanged and indicated in the golint column.

Name	Config	Description	`golint`	Typed
`context-keys-type`	n/a	Disallows the usage of basic types in `context.WithValue`.	yes	yes
`time-naming`	n/a	Conventions around the naming of time variables.	yes	yes
`var-declaration`	n/a	Reduces redundancies around variable declaration.	yes	yes
`unexported-return`	n/a	Warns when a public return is from unexported type.	yes	yes
`errorf`	n/a	Should replace `errors.New(fmt.Sprintf())` with `fmt.Errorf()`	yes	yes
`blank-imports`	n/a	Disallows blank imports	yes	no
`context-as-argument`	n/a	`context.Context` should be the first argument of a function.	yes	no
`dot-imports`	n/a	Forbids `.` imports.	yes	no
`error-return`	n/a	The error return parameter should be last.	yes	no
`error-strings`	n/a	Conventions around error strings.	yes	no
`error-naming`	n/a	Naming of error variables.	yes	no
`exported`	n/a	Naming and commenting conventions on exported symbols.	yes	no
`if-return`	n/a	Redundant if when returning an error.	yes	no
`increment-decrement`	n/a	Use `i++` and `i--` instead of `i += 1` and `i -= 1`.	yes	no
`var-naming`	whitelist & blacklist of initialisms	Naming rules.	yes	no
`package-comments`	n/a	Package commenting conventions.	yes	no
`range`	n/a	Prevents redundant variables when iterating over a collection.	yes	no
`receiver-naming`	n/a	Conventions around the naming of receivers.	yes	no
`indent-error-flow`	n/a	Prevents redundant else statements.	yes	no
`argument-limit`	int	Specifies the maximum number of arguments a function can receive	no	no
`cyclomatic`	int	Sets restriction for maximum Cyclomatic complexity.	no	no
`max-public-structs`	int	The maximum number of public structs in a file.	no	no
`file-header`	string	Header which each file should have.	no	no
`empty-block`	n/a	Warns on empty code blocks	no	no
`superfluous-else`	n/a	Prevents redundant else statements (extends `indent-error-flow`)	no	no
`confusing-naming`	n/a	Warns on methods with names that differ only by capitalization	no	no
`get-return`	n/a	Warns on getters that do not yield any result	no	no
`modifies-parameter`	n/a	Warns on assignments to function parameters	no	no
`confusing-results`	n/a	Suggests to name potentially confusing function results	no	no
`deep-exit`	n/a	Looks for program exits in funcs other than `main()` or `init()`	no	no
`unused-parameter`	n/a	Suggests to rename or remove unused function parameters	no	no
`unreachable-code`	n/a	Warns on unreachable code	no	no
`add-constant`	map	Suggests using constant for magic numbers and string literals	no	no
`flag-parameter`	n/a	Warns on boolean parameters that create a control coupling	no	no
`unnecessary-stmt`	n/a	Suggests removing or simplifying unnecessary statements	no	no
`struct-tag`	n/a	Checks common struct tags like `json`,`xml`,`yaml`	no	no
`modifies-value-receiver`	n/a	Warns on assignments to value-passed method receivers	no	yes
`constant-logical-expr`	n/a	Warns on constant logical expressions	no	no
`bool-literal-in-expr`	n/a	Suggests removing Boolean literals from logic expressions	no	no
`redefines-builtin-id`	n/a	Warns on redefinitions of builtin identifiers	no	no
`function-result-limit`	int	Specifies the maximum number of results a function can return	no	no
`imports-blacklist`	[]string	Disallows importing the specified packages	no	no
`range-val-in-closure`	n/a	Warns if range value is used in a closure dispatched as goroutine	no	no
`waitgroup-by-value`	n/a	Warns on functions taking sync.WaitGroup as a by-value parameter	no	no
`atomic`	n/a	Check for common mistaken usages of the `sync/atomic` package	no	no
`empty-lines`	n/a	Warns when there are heading or trailing newlines in a block	no	no
`line-length-limit`	int	Specifies the maximum number of characters in a line	no	no
`call-to-gc`	n/a	Warns on explicit call to the garbage collector	no	no

Configurable rules

Here you can find how you can configure some of the existing rules:

`var-naming`

This rule accepts two slices of strings, a whitelist and a blacklist of initialisms. By default the rule behaves exactly as the alternative in golint but optionally, you can relax it (see golint/lint/issues/89)

[rule.var-naming]
  arguments = [["ID"], ["VM"]]

This way, revive will not warn for identifier called customId but will warn that customVm should be called customVM.

Available Formatters

This section lists all the available formatters and provides a screenshot for each one.

Friendly

Stylish

Default

The default formatter produces the same output as golint.

Plain

The plain formatter produces the same output as the default formatter and appends URL to the rule description.

Unix

The unix formatter produces the same output as the default formatter but surrounds the rules in [].

Extensibility

The tool can be extended with custom rules or formatters. This section contains additional information on how to implement such.

To extend the linter with a custom rule or a formatter you'll have to push it to this repository or fork it. This is due to the limited -buildmode=plugin support which works only on Linux (with known issues).

Custom Rule

Each rule needs to implement the lint.Rule interface:

type Rule interface {
	Name() string
	Apply(*File, Arguments) []Failure
}

The Arguments type is an alias of the type []interface{}. The arguments of the rule are passed from the configuration file.

Example

Let's suppose we have developed a rule called BanStructNameRule which disallow us to name a structure with given identifier. We can set the banned identifier by using the TOML configuration file:

[rule.ban-struct-name]
  arguments = ["Foo"]

With the snippet above we:

Enable the rule with name ban-struct-name. The Name() method of our rule should return a string which matches ban-struct-name.
Configure the rule with the argument Foo. The list of arguments will be passed to Apply(*File, Arguments) together with the target file we're linting currently.

A sample rule implementation can be found here.

Custom Formatter

Each formatter needs to implement the following interface:

type Formatter interface {
	Format(<-chan Failure, RulesConfig) (string, error)
	Name() string
}

The Format method accepts a channel of Failure instances and the configuration of the enabled rules. The Name() method should return a string different from the names of the already existing rules. This string is used when specifying the formatter when invoking the revive CLI tool.

For a sample formatter, take a look at this file.

Speed Comparison

Compared to golint, revive performs better because it lints the files for each individual rule into a separate goroutine. Here's a basic performance benchmark on MacBook Pro Early 2013 run on kubernetes:

golint

time golint kubernetes/... > /dev/null

real    0m54.837s
user    0m57.844s
sys     0m9.146s

revive

# no type checking
time revive -config untyped.toml kubernetes/... > /dev/null

real    0m8.471s
user    0m40.721s
sys     0m3.262s

Keep in mind that if you use rules which require type checking, the performance may drop to 2x faster than golint:

# type checking enabled
time revive kubernetes/... > /dev/null

real    0m26.211s
user    2m6.708s
sys     0m17.192s

Currently, type checking is enabled by default. If you want to run the linter without type checking, remove all typed rules from the configuration file.