Skip to main content
Version: Next

Complexity Rule Set

This rule set contains rules that report complex code.

CognitiveComplexMethod

Complex methods are hard to understand and read. It might not be obvious what side-effects a complex method has. Prefer splitting up complex methods into smaller methods that are in turn easier to understand. Smaller methods can also be named much clearer which leads to improved readability of the code.

This rule measures and restricts the complexity of the method through the Cognitive Complexity metric of Sonasource. Which improves McCabe's Cyclomatic Complexity (see CyclomaticComplexMethod) considering the programmer's mental model.

Similar to cyclomatic complexity, it is a mathematical model that increases +1 complexity for flow control statements, but increases additional complexity when the statements are deeply nested.

The statements that increase the complexity or the nesting level are as follows.

  • Complexity Increments - if, when, for, while, do while, catch, labeled break, labeled continue, labeled return, recursion call, &&, ||
  • Nesting Level Increments - if, when, for, while, do while, catch, nested function
  • Additional Complexity Increments by Nesting Level - if, when, for, while, do while, catch

Active by default: No

Debt: 20min

Configuration options:

  • allowedComplexity (default: 15)

    Maximum Cognitive Complexity allowed for a method.

ComplexCondition

Complex conditions make it hard to understand which cases lead to the condition being true or false. To improve readability and understanding of complex conditions consider extracting them into well-named functions or variables and call those instead.

Active by default: Yes - Since v1.0.0

Debt: 20min

Configuration options:

  • threshold (default: 4)

    the number of conditions which will trigger the rule

Noncompliant Code:

val str = "foo"
val isFoo = if (str.startsWith("foo") && !str.endsWith("foo") && !str.endsWith("bar") && !str.endsWith("_")) {
// ...
}

Compliant Code:

val str = "foo"
val isFoo = if (str.startsWith("foo") && hasCorrectEnding()) {
// ...
}

fun hasCorrectEnding() = return !str.endsWith("foo") && !str.endsWith("bar") && !str.endsWith("_")

ComplexInterface

Complex interfaces which contain too many functions and/or properties indicate that this interface is handling too many things at once. Interfaces should follow the single-responsibility principle to also encourage implementations of this interface to not handle too many things at once.

Large interfaces should be split into smaller interfaces which have a clear responsibility and are easier to understand and implement.

Active by default: No

Debt: 20min

Configuration options:

  • allowedDefinitions (default: 10)

    The amount of allowed definitions in an interface.

  • includeStaticDeclarations (default: false)

    whether static declarations should be included

  • includePrivateDeclarations (default: false)

    whether private declarations should be included

  • ignoreOverloaded (default: false)

    ignore overloaded methods - only count once

CyclomaticComplexMethod

Complex methods are hard to understand and read. It might not be obvious what side-effects a complex method has. Prefer splitting up complex methods into smaller methods that are in turn easier to understand. Smaller methods can also be named much clearer which leads to improved readability of the code.

This rule uses McCabe's Cyclomatic Complexity (MCC) metric to measure the number of linearly independent paths through a function's source code (https://www.ndepend.com/docs/code-metrics#CC). The higher the number of independent paths, the more complex a method is. Complex methods use too many of the following statements. Each one of them adds one to the complexity count.

  • Conditional statements - if, else if, when
  • Jump statements - continue, break
  • Loops - for, while, do-while, forEach
  • Operators &&, ||, ?:
  • Exceptions - catch, use
  • Scope Functions - let, run, with, apply, and also -> Reference

Active by default: Yes - Since v1.0.0

Debt: 20min

Aliases: ComplexMethod

Configuration options:

  • allowedComplexity (default: 14)

    The maximum allowed McCabe's Cyclomatic Complexity (MCC) for a method.

  • ignoreSingleWhenExpression (default: false)

    Ignores a complex method if it only contains a single when expression.

  • ignoreSimpleWhenEntries (default: false)

    Whether to ignore simple (braceless) when entries.

  • ignoreNestingFunctions (default: false)

    Whether to ignore functions which are often used instead of an if or for statement.

  • nestingFunctions (default: ['also', 'apply', 'forEach', 'isNotNull', 'ifNull', 'let', 'run', 'use', 'with'])

    Comma separated list of function names which add complexity.

LabeledExpression

This rule reports labeled expressions. Expressions with labels generally increase complexity and worsen the maintainability of the code. Refactor the violating code to not use labels instead. Labeled expressions referencing an outer class with a label from an inner class are allowed, because there is no way to get the instance of an outer class from an inner class in Kotlin.

Active by default: No

Debt: 20min

Configuration options:

  • ignoredLabels (default: [])

    allows to provide a list of label names which should be ignored by this rule

Noncompliant Code:

val range = listOf<String>("foo", "bar")
loop@ for (r in range) {
if (r == "bar") break@loop
println(r)
}

class Outer {
inner class Inner {
fun f() {
val i = this@Inner // referencing itself, use `this instead
}
}
}

Compliant Code:

val range = listOf<String>("foo", "bar")
for (r in range) {
if (r == "bar") break
println(r)
}

class Outer {
inner class Inner {
fun f() {
val outer = this@Outer
}
fun Int.extend() {
val inner = this@Inner // this would reference Int and not Inner
}
}
}

LargeClass

This rule reports large classes. Classes should generally have one responsibility. Large classes can indicate that the class does instead handle multiple responsibilities. Instead of doing many things at once prefer to split up large classes into smaller classes. These smaller classes are then easier to understand and handle less things.

Active by default: Yes - Since v1.0.0

Debt: 20min

Configuration options:

  • allowedLines (default: 600)

    The maximum number of lines allowed per class.

LongMethod

Methods should have one responsibility. Long methods can indicate that a method handles too many cases at once. Prefer smaller methods with clear names that describe their functionality clearly.

Extract parts of the functionality of long methods into separate, smaller methods.

Active by default: Yes - Since v1.0.0

Debt: 20min

Configuration options:

  • allowedLines (default: 60)

    number of lines in a method that are allowed at maximum

LongParameterList

Reports functions and constructors which have more parameters than a certain threshold.

Active by default: Yes - Since v1.0.0

Debt: 20min

Configuration options:

  • threshold (default: 6)

    Deprecated: Use functionThreshold and constructorThreshold instead

    number of parameters required to trigger the rule

  • functionThreshold (default: 6)

    number of function parameters required to trigger the rule

  • constructorThreshold (default: 7)

    number of constructor parameters required to trigger the rule

  • ignoreDefaultParameters (default: false)

    ignore parameters that have a default value

  • ignoreDataClasses (default: true)

    ignore long constructor parameters list for data classes

  • ignoreAnnotatedParameter (default: [])

    ignore the annotated parameters for the count (e.g. fun foo(@Value bar: Int) would not be counted

MethodOverloading

This rule reports methods which are overloaded often. Method overloading tightly couples these methods together which might make the code harder to understand.

Refactor these methods and try to use optional parameters instead to prevent some of the overloading.

Active by default: No

Debt: 20min

Configuration options:

  • allowedOverloads (default: 6)

    The allowed number of overloads for a method.

NamedArguments

Reports function invocations which have more arguments than a certain threshold and are all not named. Calls with too many arguments are more difficult to understand so a named arguments help.

Active by default: No

Requires Type Resolution

Debt: 5min

Configuration options:

  • allowedArguments (default: 3)

    The allowed number of arguments for a function.

  • ignoreArgumentsMatchingNames (default: false)

    ignores when argument values are the same as the parameter names

Noncompliant Code:

fun sum(a: Int, b: Int, c: Int, d: Int) {
}
sum(1, 2, 3, 4)

Compliant Code:

fun sum(a: Int, b: Int, c: Int, d: Int) {
}
sum(a = 1, b = 2, c = 3, d = 4)

NestedBlockDepth

This rule reports excessive nesting depth in functions. Excessively nested code becomes harder to read and increases its hidden complexity. It might become harder to understand edge-cases of the function.

Prefer extracting the nested code into well-named functions to make it easier to understand.

Active by default: Yes - Since v1.0.0

Debt: 20min

Configuration options:

  • allowedDepth (default: 4)

    The maximum allowed nested block depth for a function

NestedScopeFunctions

Although the scope functions are a way of making the code more concise, avoid overusing them: it can decrease your code readability and lead to errors. Avoid nesting scope functions and be careful when chaining them: it's easy to get confused about the current context object and the value of this or it.

Reference

Active by default: No

Requires Type Resolution

Debt: 5min

Configuration options:

  • allowedDepth (default: 1)

    The maximum allowed depth for nested scope functions.

  • functions (default: ['kotlin.apply', 'kotlin.run', 'kotlin.with', 'kotlin.let', 'kotlin.also'])

    Set of scope function names which add complexity. Function names have to be fully qualified. For example 'kotlin.apply'.

Noncompliant Code:

// Try to figure out, what changed, without knowing the details
first.apply {
second.apply {
b = a
c = b
}
}

Compliant Code:

// 'a' is a property of current class
// 'b' is a property of class 'first'
// 'c' is a property of class 'second'
first.b = this.a
second.c = first.b

ReplaceSafeCallChainWithRun

Chains of safe calls on non-nullable types are redundant and can be removed by enclosing the redundant safe calls in a run {} block. This improves code coverage and reduces cyclomatic complexity as redundant null checks are removed.

This rule only checks from the end of a chain and works backwards, so it won't recommend inserting run blocks in the middle of a safe call chain as that is likely to make the code more difficult to understand.

The rule will check for every opportunity to replace a safe call when it sits at the end of a chain, even if there's only one, as that will still improve code coverage and reduce cyclomatic complexity.

Active by default: No

Requires Type Resolution

Debt: 10min

Noncompliant Code:

val x = System.getenv()
?.getValue("HOME")
?.toLowerCase()
?.split("/") ?: emptyList()

Compliant Code:

val x = getenv()?.run {
getValue("HOME")
.toLowerCase()
.split("/")
} ?: emptyList()

StringLiteralDuplication

This rule detects and reports duplicated String literals. Repeatedly typing out the same String literal across the codebase makes it harder to change and maintain.

Instead, prefer extracting the String literal into a property or constant.

Active by default: No

Debt: 5min

Configuration options:

  • allowedDuplications (default: 2)

    The maximum allowed amount of duplications.

  • ignoreAnnotation (default: true)

    if values in Annotations should be ignored

  • excludeStringsWithLessThan5Characters (default: true)

    if short strings should be excluded

  • ignoreStringsRegex (default: '$^')

    RegEx of Strings that should be ignored

Noncompliant Code:

class Foo {

val s1 = "lorem"
fun bar(s: String = "lorem") {
s1.equals("lorem")
}
}

Compliant Code:

class Foo {
val lorem = "lorem"
val s1 = lorem
fun bar(s: String = lorem) {
s1.equals(lorem)
}
}

TooManyFunctions

This rule reports files, classes, interfaces, objects and enums which contain too many functions. Each element can be configured with different thresholds.

Too many functions indicate a violation of the single responsibility principle. Prefer extracting functionality which clearly belongs together in separate parts of the code.

Active by default: Yes - Since v1.0.0

Debt: 20min

Configuration options:

  • allowedFunctionsPerFile (default: 11)

    The maximum allowed functions per file

  • allowedFunctionsPerClass (default: 11)

    The maximum allowed functions per class

  • allowedFunctionsPerInterface (default: 11)

    The maximum allowed functions per interface

  • allowedFunctionsPerObject (default: 11)

    The maximum allowed function per object

  • allowedFunctionsPerEnum (default: 11)

    The maximum allowed functions in enums

  • ignoreDeprecated (default: false)

    ignore deprecated functions

  • ignorePrivate (default: false)

    ignore private functions

  • ignoreOverridden (default: false)

    ignore overridden functions