category.java.design.xml Maven / Gradle / Ivy
The newest version!
<?xml version="1.0" encoding="UTF-8"?>
<ruleset name="Design"
xmlns="http://pmd.sourceforge.net/ruleset/2.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://pmd.sourceforge.net/ruleset/2.0.0 https://pmd.sourceforge.io/ruleset_2_0_0.xsd">
<description>
Rules that help you discover design issues.
</description>
<rule name="AbstractClassWithoutAnyMethod"
language="java"
since="4.2"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
message="No abstract method which means that the keyword is most likely used to prevent instantiation. Use a private or protected constructor instead."
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#abstractclasswithoutanymethod">
<description>
If an abstract class does not provide any methods, it may be acting as a simple data container
that is not meant to be instantiated. In this case, it is probably better to use a private or
protected constructor in order to prevent instantiation than make the class misleadingly abstract.
</description>
<priority>1</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//ClassDeclaration
[@Abstract = true() and @Interface = false()]
[ClassBody[not(ConstructorDeclaration | MethodDeclaration)]]
[not(pmd-java:hasAnnotation('com.google.auto.value.AutoValue')
or pmd-java:hasAnnotation('lombok.AllArgsConstructor')
or pmd-java:hasAnnotation('lombok.NoArgsConstructor')
or pmd-java:hasAnnotation('lombok.RequiredArgsConstructor'))
]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public abstract class Example {
String field;
int otherField;
}
]]>
</example>
</rule>
<rule name="AvoidCatchingGenericException"
since="4.2.6"
language="java"
message="Avoid catching generic exceptions such as NullPointerException, RuntimeException, Exception in try-catch block"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoidcatchinggenericexception">
<description>
Avoid catching generic exceptions such as NullPointerException, RuntimeException, Exception in try-catch block.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//CatchParameter//ClassType[
pmd-java:typeIsExactly('java.lang.NullPointerException') or
pmd-java:typeIsExactly('java.lang.Exception') or
pmd-java:typeIsExactly('java.lang.RuntimeException')]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
package com.igate.primitive;
public class PrimitiveType {
public void downCastPrimitiveType() {
try {
System.out.println(" i [" + i + "]");
} catch(Exception e) {
e.printStackTrace();
} catch(RuntimeException e) {
e.printStackTrace();
} catch(NullPointerException e) {
e.printStackTrace();
}
}
}
]]>
</example>
</rule>
<rule name="AvoidDeeplyNestedIfStmts"
language="java"
since="1.0"
message="Deeply nested if..then statements are hard to read"
class="net.sourceforge.pmd.lang.java.rule.design.AvoidDeeplyNestedIfStmtsRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoiddeeplynestedifstmts">
<description>
Avoid creating deeply nested if-then statements since they are harder to read and error-prone to maintain.
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
public void bar(int x, int y, int z) {
if (x>y) {
if (y>z) {
if (z==x) {
// !! too deep
}
}
}
}
}
]]>
</example>
</rule>
<rule name="AvoidRethrowingException"
language="java"
since="3.8"
message="A catch statement that catches an exception only to rethrow it should be avoided."
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoidrethrowingexception">
<description>
Catch blocks that merely rethrow a caught exception only add to code size and runtime complexity.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//CatchClause[
CatchParameter/VariableId/@Name
= Block[@Size = 1]/ThrowStatement/VariableAccess/@Name]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public void bar() {
try {
// do something
} catch (SomeException se) {
throw se;
}
}
]]>
</example>
</rule>
<rule name="AvoidThrowingNewInstanceOfSameException"
since="4.2.5"
language="java"
message="A catch statement that catches an exception only to wrap it in a new instance of the same type of exception and throw it should be avoided"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoidthrowingnewinstanceofsameexception">
<description>
Catch blocks that merely rethrow a caught exception wrapped inside a new instance of the same type only add to
code size and runtime complexity.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//CatchClause
[count(Block/*) = 1]
[CatchParameter/ClassType/@SimpleName = Block/ThrowStatement/ConstructorCall/ClassType/@SimpleName]
[Block/ThrowStatement/ConstructorCall/ArgumentList/@Size = 1]
/Block/ThrowStatement/ConstructorCall
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public void bar() {
try {
// do something
} catch (SomeException se) {
// harmless comment
throw new SomeException(se);
}
}
]]>
</example>
</rule>
<rule name="AvoidThrowingNullPointerException"
language="java"
since="1.8"
message="Avoid throwing null pointer exceptions."
class="net.sourceforge.pmd.lang.java.rule.design.AvoidThrowingNullPointerExceptionRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoidthrowingnullpointerexception">
<description>
<![CDATA[
Avoid throwing NullPointerExceptions manually. These are confusing because most people will assume that the
virtual machine threw it. To avoid a method being called with a null parameter, you may consider
using an IllegalArgumentException instead, making it clearly seen as a programmer-initiated exception.
However, there are better ways to handle this:
>*Effective Java, 3rd Edition, Item 72: Favor the use of standard exceptions*
>
>Arguably, every erroneous method invocation boils down to an illegal argument or state,
but other exceptions are standardly used for certain kinds of illegal arguments and states.
If a caller passes null in some parameter for which null values are prohibited, convention dictates that
NullPointerException be thrown rather than IllegalArgumentException.
To implement that, you are encouraged to use `java.util.Objects.requireNonNull()`
(introduced in Java 1.7). This method is designed primarily for doing parameter
validation in methods and constructors with multiple parameters.
Your parameter validation could thus look like the following:
```
public class Foo {
private String exampleValue;
void setExampleValue(String exampleValue) {
// check, throw and assignment in a single standard call
this.exampleValue = Objects.requireNonNull(exampleValue, "exampleValue must not be null!");
}
}
```
]]>
</description>
<priority>1</priority>
<example>
<![CDATA[
public class Foo {
void bar() {
throw new NullPointerException();
}
}
]]>
</example>
</rule>
<rule name="AvoidThrowingRawExceptionTypes"
language="java"
since="1.8"
message="Avoid throwing raw exception types."
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoidthrowingrawexceptiontypes">
<description>
Avoid throwing certain exception types. Rather than throw a raw RuntimeException, Throwable,
Exception, or Error, use a subclassed exception or error instead.
</description>
<priority>1</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//ThrowStatement//ConstructorCall
/ClassType[
pmd-java:typeIsExactly('java.lang.Throwable')
or
pmd-java:typeIsExactly('java.lang.Exception')
or
pmd-java:typeIsExactly('java.lang.Error')
or
pmd-java:typeIsExactly('java.lang.RuntimeException')
]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Foo {
public void bar() throws Exception {
throw new Exception();
}
}
]]>
</example>
</rule>
<rule name="AvoidUncheckedExceptionsInSignatures"
since="6.13.0"
language="java"
message="A method or constructor should not explicitly declare unchecked exceptions in its ''throws'' clause"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#avoiduncheckedexceptionsinsignatures">
<description>
Reports unchecked exceptions in the `throws` clause of a method or constructor.
Java doesn't force the caller to handle an unchecked exception,
so it's unnecessary except for documentation. A better practice is to document the
exceptional cases with a `@throws` Javadoc tag, which allows being more descriptive.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//ThrowsList/ClassType[pmd-java:typeIs('java.lang.RuntimeException')]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public void foo() throws RuntimeException {
}
]]>
</example>
</rule>
<rule name="ClassWithOnlyPrivateConstructorsShouldBeFinal"
language="java"
since="4.1"
class="net.sourceforge.pmd.lang.java.rule.design.ClassWithOnlyPrivateConstructorsShouldBeFinalRule"
message="This class has only private constructors and may be final"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#classwithonlyprivateconstructorsshouldbefinal">
<description>
Reports classes that may be made final because they cannot be extended from outside
their compilation unit anyway. This is because all their constructors are private,
so a subclass could not call the super constructor.
</description>
<priority>1</priority>
<example>
<![CDATA[
public class Foo { //Should be final
private Foo() { }
}
]]>
</example>
</rule>
<rule name="CollapsibleIfStatements"
language="java"
since="3.1"
message="This if statement could be combined with its parent"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#collapsibleifstatements">
<description><![CDATA[
Reports nested 'if' statements that can be merged together by joining their
conditions with a boolean `&&` operator in between.
]]></description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//IfStatement[@Else = false()]/IfStatement[@Else = false()]
|
//IfStatement[@Else = false()]/Block[count(*) = 1]/IfStatement[@Else = false()]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
class Foo {
void bar() {
if (x) { // original implementation
if (y) {
// do stuff
}
}
}
void bar() {
if (x && y) { // clearer implementation
// do stuff
}
}
}
]]>
</example>
</rule>
<rule name="CouplingBetweenObjects"
language="java"
since="1.04"
message="A value of {0} may denote a high amount of coupling within the class (threshold: {1})"
class="net.sourceforge.pmd.lang.java.rule.design.CouplingBetweenObjectsRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#couplingbetweenobjects">
<description>
This rule counts the number of unique attributes, local variables, and return types within an object.
A number higher than the specified threshold can indicate a high degree of coupling.
</description>
<priority>3</priority>
<example>
<![CDATA[
import com.Blah;
import org.Bar;
import org.Bardo;
public class Foo {
private Blah var1;
private Bar var2;
//followed by many imports of unique objects
ObjectC doWork() {
Bardo var55;
ObjectA var44;
ObjectZ var93;
return something();
}
}
]]>
</example>
</rule>
<rule name="CognitiveComplexity"
language="java"
message="The {0} ''{1}'' has a cognitive complexity of {2}, current threshold is {3}"
since="6.35.0"
class="net.sourceforge.pmd.lang.java.rule.design.CognitiveComplexityRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#cognitivecomplexity">
<description><![CDATA[
Methods that are highly complex are difficult to read and more costly to maintain. If you include too much decisional
logic within a single method, you make its behavior hard to understand and more difficult to modify.
Cognitive complexity is a measure of how difficult it is for humans to read and understand a method. Code that contains
a break in the control flow is more complex, whereas the use of language shorthands doesn't increase the level of
complexity. Nested control flows can make a method more difficult to understand, with each additional nesting of the
control flow leading to an increase in cognitive complexity.
Information about Cognitive complexity can be found in the original paper here:
<https://www.sonarsource.com/docs/CognitiveComplexity.pdf>
By default, this rule reports methods with a complexity of 15 or more. Reported methods should be broken down into less
complex components.
]]></description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
// Has a cognitive complexity of 0
public void createAccount() {
Account account = new Account("PMD");
// save account
}
// Has a cognitive complexity of 1
public Boolean setPhoneNumberIfNotExisting(Account a, String phone) {
if (a.phone == null) { // +1
a.phone = phone;
return true;
}
return false;
}
// Has a cognitive complexity of 4
public void updateContacts(List<Contact> contacts) {
List<Contact> contactsToUpdate = new ArrayList<Contact>();
for (Contact contact : contacts) { // +1
if (contact.department.equals("Finance")) { // +2 (nesting = 1)
contact.title = "Finance Specialist";
contactsToUpdate.add(contact);
} else if (contact.department.equals("Sales")) { // +1
contact.title = "Sales Specialist";
contactsToUpdate.add(contact);
}
}
// save contacts
}
}
]]>
</example>
</rule>
<rule name="CyclomaticComplexity"
language="java"
message="The {0} ''{1}'' has a{2} cyclomatic complexity of {3}."
since="1.03"
class="net.sourceforge.pmd.lang.java.rule.design.CyclomaticComplexityRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#cyclomaticcomplexity">
<description><![CDATA[
The complexity of methods directly affects maintenance costs and readability. Concentrating too much decisional logic
in a single method makes its behaviour hard to read and change.
Cyclomatic complexity assesses the complexity of a method by counting the number of decision points in a method,
plus one for the method entry. Decision points are places where the control flow jumps to another place in the
program. As such, they include all control flow statements, such as `if`, `while`, `for`, and `case`. For more
details on the calculation, see the documentation {% jdoc java::lang.java.metrics.JavaMetrics#CYCLO %}.
Generally, numbers ranging from 1-4 denote low complexity, 5-7 denote moderate complexity, 8-10 denote
high complexity, and 11+ is very high complexity. By default, this rule reports methods with a complexity >= 10.
Additionally, classes with many methods of moderate complexity get reported as well once the total of their
methods' complexities reaches 80, even if none of the methods was directly reported.
Reported methods should be broken down into several smaller methods. Reported classes should probably be broken down
into subcomponents.]]>
</description>
<priority>3</priority>
<example>
<![CDATA[
class Foo {
void baseCyclo() { // Cyclo = 1
highCyclo();
}
void highCyclo() { // Cyclo = 10: reported!
int x = 0, y = 2;
boolean a = false, b = true;
if (a && (y == 1 ? b : true)) { // +3
if (y == x) { // +1
while (true) { // +1
if (x++ < 20) { // +1
break; // +1
}
}
} else if (y == t && !d) { // +2
x = a ? y : x; // +1
} else {
x = 2;
}
}
}
}
]]>
</example>
</rule>
<rule name="DataClass"
language="java"
since="6.0.0"
message="The class ''{0}'' is suspected to be a Data Class (WOC={1}, NOPA={2}, NOAM={3}, WMC={4})"
class="net.sourceforge.pmd.lang.java.rule.design.DataClassRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#dataclass">
<description>
Data Classes are simple data holders, which reveal most of their state, and
without complex functionality. The lack of functionality may indicate that
their behaviour is defined elsewhere, which is a sign of poor data-behaviour
proximity. By directly exposing their internals, Data Classes break encapsulation,
and therefore reduce the system's maintainability and understandability. Moreover,
classes tend to strongly rely on their data representation, which makes for a brittle
design.
Refactoring a Data Class should focus on restoring a good data-behaviour proximity. In
most cases, that means moving the operations defined on the data back into the class.
In some other cases it may make sense to remove entirely the class and move the data
into the former client classes.
The rule uses metrics to implement its detection strategy. The violation message
gives information about the values of these metrics:
* WMC: a class complexity measure for a class, see {% jdoc java::lang.java.metrics.JavaMetrics#WEIGHED_METHOD_COUNT %}
* WOC: a 'non-triviality' measure for a class, see {% jdoc java::lang.java.metrics.JavaMetrics#WEIGHT_OF_CLASS %}
* NOPA: number of public attributes, see {% jdoc java::lang.java.metrics.JavaMetrics#NUMBER_OF_PUBLIC_FIELDS %}
* NOAM: number of public accessor methods, see {% jdoc java::lang.java.metrics.JavaMetrics#NUMBER_OF_ACCESSORS %}
The rule identifies a god class by looking for classes which have all of the following properties:
* High NOPA + NOAM
* Low WOC
* Low WMC
</description>
<priority>3</priority>
<example>
<![CDATA[
public class DataClass {
// class exposes public attributes
public String name = "";
public int bar = 0;
public int na = 0;
private int bee = 0;
// and private ones through getters
public void setBee(int n) {
bee = n;
}
}
]]>
</example>
</rule>
<rule name="DoNotExtendJavaLangError"
language="java"
since="4.0"
message="Exceptions should not extend java.lang.Error"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#donotextendjavalangerror">
<description>
Errors are system exceptions. Do not extend them.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//ClassDeclaration/ExtendsList/ClassType[pmd-java:typeIs('java.lang.Error')]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Foo extends Error { }
]]>
</example>
</rule>
<rule name="ExceptionAsFlowControl"
language="java"
since="1.8"
message="Exception thrown at line {0} is caught in this block."
class="net.sourceforge.pmd.lang.java.rule.design.ExceptionAsFlowControlRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#exceptionasflowcontrol">
<description>
This rule reports exceptions thrown and caught in an enclosing try statement.
This use of exceptions as a form of `goto` statement is discouraged, as that may
hide actual exceptions, and obscures control flow, especially when debugging.
To fix a violation, add the necessary validation or use an alternate control structure.
</description>
<priority>3</priority>
<example>
<![CDATA[
public void bar() {
try {
try {
} catch (Exception e) {
throw new WrapperException(e);
// this is essentially a GOTO to the WrapperException catch block
}
} catch (WrapperException e) {
// do some more stuff
}
}
]]>
</example>
</rule>
<rule name="ExcessiveImports"
language="java"
since="1.04"
message="A high number of imports can indicate a high degree of coupling within an object."
class="net.sourceforge.pmd.lang.java.rule.design.ExcessiveImportsRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#excessiveimports">
<description>
A high number of imports can indicate a high degree of coupling within an object. This rule
counts the number of unique imports and reports a violation if the count is above the
user-specified threshold.
</description>
<priority>3</priority>
<example>
<![CDATA[
import blah.blah.Baz;
import blah.blah.Bif;
// 28 others from the same package elided
public class Foo {
public void doWork() {}
}
]]>
</example>
</rule>
<rule name="ExcessiveParameterList"
language="java"
since="0.9"
message="Avoid long parameter lists."
class="net.sourceforge.pmd.lang.java.rule.design.ExcessiveParameterListRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#excessiveparameterlist">
<description>
Methods with numerous parameters are a challenge to maintain, especially if most of them share the
same datatype. These situations usually denote the need for new objects to wrap the numerous parameters.
</description>
<priority>3</priority>
<example>
<![CDATA[
public void addPerson( // too many arguments liable to be mixed up
int birthYear, int birthMonth, int birthDate, int height, int weight, int ssn) {
. . . .
}
public void addPerson( // preferred approach
Date birthdate, BodyMeasurements measurements, int ssn) {
. . . .
}
]]>
</example>
</rule>
<rule name="ExcessivePublicCount"
language="java"
since="1.04"
message="This class has a bunch of public methods and attributes"
class="net.sourceforge.pmd.lang.java.rule.design.ExcessivePublicCountRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#excessivepubliccount">
<description>
Classes with large numbers of public methods and attributes require disproportionate testing efforts
since combinational side effects grow rapidly and increase risk. Refactoring these classes into
smaller ones not only increases testability and reliability but also allows new variations to be
developed easily.
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
public String value;
public Bar something;
public Variable var;
// [... more more public attributes ...]
public void doWork() {}
public void doMoreWork() {}
public void doWorkAgain() {}
// [... more more public methods ...]
}
]]>
</example>
</rule>
<rule name="FinalFieldCouldBeStatic"
language="java"
since="1.1"
message="This final field could be made static"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#finalfieldcouldbestatic">
<description>
If a final field is assigned to a compile-time constant, it could be made static, thus saving overhead
in each object at runtime.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//FieldDeclaration
[pmd-java:modifiers() = 'final']
[not(pmd-java:modifiers() = 'static')]
[not(./ancestor::ClassDeclaration[1][pmd-java:hasAnnotation('lombok.experimental.UtilityClass')])]
[not(.//Annotation[pmd-java:typeIs('lombok.Builder.Default')])]
/VariableDeclarator[*[last()][@CompileTimeConstant = true()
or self::NullLiteral
or self::VariableAccess[@Name = //FieldDeclaration[pmd-java:modifiers() = 'static']/VariableDeclarator/VariableId/@Name]
or self::FieldAccess
or self::ArrayAllocation/ArrayType/ArrayDimensions/ArrayDimExpr/NumericLiteral[@IntLiteral = true()][@Image = "0"]]]
/VariableId
[not(@Name = //MethodDeclaration[not(pmd-java:modifiers() = 'static')]
//SynchronizedStatement/(VariableAccess|FieldAccess[ThisExpression])/@Name)]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Foo {
public final int BAR = 42; // this could be static and save some space
}
]]>
</example>
</rule>
<rule name="GodClass"
language="java"
since="5.0"
message="Possible God Class (WMC={0}, ATFD={2}, TCC={1})"
class="net.sourceforge.pmd.lang.java.rule.design.GodClassRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#godclass">
<description>
The God Class rule detects the God Class design flaw using metrics. God classes do too many things,
are very big and overly complex. They should be split apart to be more object-oriented.
The rule uses the detection strategy described in "Object-Oriented Metrics in Practice".
The violations are reported against the entire class.
The rule uses metrics to implement its detection strategy. The violation message
gives information about the values of these metrics:
* WMC: a class complexity measure, see {% jdoc java::lang.java.metrics.JavaMetrics#WEIGHED_METHOD_COUNT %}
* ATFD: a measure of how much data external data the class uses, see {% jdoc java::lang.java.metrics.JavaMetrics#ACCESS_TO_FOREIGN_DATA %}
* TCC: a measure of how tightly related the methods are, see {% jdoc java::lang.java.metrics.JavaMetrics#TIGHT_CLASS_COHESION %}
The rule identifies a god class by looking for classes which have all of the following properties:
* High WMC
* High ATFD
* Low TCC
See also the reference:
Michele Lanza and Radu Marinescu. *Object-Oriented Metrics in Practice:
Using Software Metrics to Characterize, Evaluate, and Improve the Design
of Object-Oriented Systems.* Springer, Berlin, 1 edition, October 2006. Page 80.
</description>
<priority>3</priority>
</rule>
<rule name="ImmutableField"
language="java"
since="2.0"
message="Field ''{0}'' may be declared final"
class="net.sourceforge.pmd.lang.java.rule.design.ImmutableFieldRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#immutablefield">
<description>
Reports non-final fields whose value never changes once object initialization ends,
and hence may be marked final.
Note that this rule does not enforce that the field value be deeply immutable itself.
An object can still have mutable state, even if all its member fields are declared final.
This is referred to as shallow immutability. For more information on mutability,
see *Effective Java, 3rd Edition, Item 17: Minimize mutability*.
Limitations: We can only check private fields for now.
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
private int x; // could be final
public Foo() {
x = 7;
}
public void foo() {
int a = x + 2;
}
}
]]>
</example>
</rule>
<rule name="InvalidJavaBean"
language="java"
since="6.52.0"
message="The bean ''{0}'' is missing a getter for property ''{1}''."
class="net.sourceforge.pmd.lang.java.rule.design.InvalidJavaBeanRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#invalidjavabean">
<description>
Identifies beans, that don't follow the [JavaBeans API specification](https://download.oracle.com/otndocs/jcp/7224-javabeans-1.01-fr-spec-oth-JSpec/).
Each non-static field should have both a getter and a setter method. If the field is just used internally and is not
a bean property, then the field should be marked as `transient`.
The rule verifies that the type of the field is the same as the result type of the getter. And that this type matches
the type used in the setter.
The rule also checks, that there is a no-arg or default constructor available.
Optionally the rule also verifies, that the bean implements `java.io.Serializable`. While this is a requirement for the
original JavaBeans specification, frameworks nowadays don't strictly require this anymore.
In order to avoid many false positives in classes that are not beans, the rule needs to be explicitly
enabled by configuring the property `packages`.
</description>
<priority>3</priority>
<example>
<![CDATA[
package org.example.beans;
public class MyBean { // <-- bean is not serializable, missing "implements Serializable"
private String label; // <-- missing setter for property "label"
public String getLabel() {
return label;
}
}
]]>
</example>
</rule>
<rule name="LawOfDemeter"
language="java"
since="5.0"
message="Potential violation of the law of Demeter ({0})"
class="net.sourceforge.pmd.lang.java.rule.design.LawOfDemeterRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#lawofdemeter">
<description>
The law of Demeter is a simple rule that says "only talk to friends". It forbids
fetching data from "too far away", for some definition of distance, in order to
reduce coupling between classes or objects of different levels of abstraction.
The rule uses a notion of "degree", that quantifies how "far" an object is.
Expressions with too high degree can only be used in certain ways. The degree of
an expression is defined inductively:
- The degree of `this` is 0
- The degree of a method parameter is 1
- The degree of a new object created in a method is 1
- The degree of a static variable is 1
- The degree of a field access expression like `expr.field` is the degree of `expr` plus 1
- The degree of a "getter expression" like `expr.getFoo()` is the degree of `expr` plus 1
- The degree of a "transformation expression" like `expr.withFoo("")` is the degree of `expr`
- The degree of a variable is the maximum degree of all the assignments that reach it
Intuitively, the more you call getters, the more the degree increases. Eventually
the degree reaches the report threshold (property `trustRadius`) and the expression
is reported. The details of the calculation are more involved and make room for common
patterns, like usage of collections (objects that are in a list or array have the
same degree as their container), the builder pattern, and getters that do not appear
to break a boundary of abstraction.
Be aware that this rule is prone to many false-positives and low-priority warnings.
You can increase the `trustRadius` property to reduce them drastically. The default
`trustRadius` of 1 corresponds to the original law of Demeter (you're only allowed
one getter call on untrusted values). Given some `trustRadius` value:
- expressions of degree lower or equal to `trustRadius` are not reported
- expressions of degree exactly `trustRadius + 1` are reported, unless they are only returned
from the current method, or passed as argument to another method. Without this exception it
would not be possible to extract any information from e.g. method parameters.
- values of degree strictly greater than `trustRadius + 1` are not reported. The
intuition is that to obtain a value of degree `n > 1` then you must use an expression
of degree `n - 1`, so if you have `n > trustRadius + 1`, there you're using some value
of degree `trustRadius + 1` that will be reported.
See also the references:
* Andrew Hunt, David Thomas, and Ward Cunningham. The Pragmatic Programmer. From Journeyman to Master. Addison-Wesley Longman, Amsterdam, October 1999.;
* K.J. Lieberherr and I.M. Holland. Assuring good style for object-oriented programs. Software, IEEE, 6(5):38–48, 1989.;
* <http://www.ccs.neu.edu/home/lieber/LoD.html>
* <http://en.wikipedia.org/wiki/Law_of_Demeter>
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
/**
* This example will result in one violation.
*/
public void example(Bar b) { // b has degree 1
// `b.getC()` has degree 2, it's breaking a boundary of abstraction and so is reported.
b.getC().doIt();
// To respect the law of Demeter, Bar should encapsulate its
// C member more properly, eg by exposing a method like this:
b.callDoItOnC();
// a constructor call, not a method call.
D d = new D();
// this method call is ok, because we have create the new
// instance of D locally.
d.doSomethingElse();
}
}
]]>
</example>
</rule>
<rule name="LogicInversion"
language="java"
since="5.0"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
message="Use opposite operator instead of the logic complement operator."
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#logicinversion">
<description>
Use opposite operator instead of negating the whole expression with a logic complement operator.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//UnaryExpression[@Operator='!']/InfixExpression[@Operator = ('==', '!=', '<', '>', '<=', '>=')]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public boolean bar(int a, int b) {
if (!(a == b)) { // use !=
return false;
}
if (!(a < b)) { // use >=
return false;
}
return true;
}
]]>
</example>
</rule>
<rule name="LoosePackageCoupling"
language="java"
since="5.0"
message="Use of ''{0}'' outside of package hierarchy ''{1}'' is not recommended; use recommended classes instead"
class="net.sourceforge.pmd.lang.java.rule.design.LoosePackageCouplingRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#loosepackagecoupling">
<description>
Avoid using classes from the configured package hierarchy outside of the package hierarchy,
except when using one of the configured allowed classes.
</description>
<priority>3</priority>
<example>
<![CDATA[
package some.package;
import some.other.package.subpackage.subsubpackage.DontUseThisClass;
public class Bar {
DontUseThisClass boo = new DontUseThisClass();
}
]]>
</example>
</rule>
<rule name="NcssCount"
language="java"
message="The {0} ''{1}'' has a NCSS line count of {2}."
since="6.0.0"
class="net.sourceforge.pmd.lang.java.rule.design.NcssCountRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#ncsscount">
<description>
This rule uses the NCSS (Non-Commenting Source Statements) metric to determine the number of lines
of code in a class, method or constructor. NCSS ignores comments, blank lines, and only counts actual
statements. For more details on the calculation, see the documentation
{% jdoc java::lang.java.metrics.JavaMetrics#NCSS %}.
</description>
<priority>3</priority>
<example>
<![CDATA[
import java.util.Collections; // +0
import java.io.IOException; // +0
class Foo { // +1, total Ncss = 12
public void bigMethod() // +1
throws IOException {
int x = 0, y = 2; // +1
boolean a = false, b = true; // +1
if (a || b) { // +1
try { // +1
do { // +1
x += 2; // +1
} while (x < 12);
System.exit(0); // +1
} catch (IOException ioe) { // +1
throw new PatheticFailException(ioe); // +1
}
} else {
assert false; // +1
}
}
}
]]>
</example>
</rule>
<rule name="NPathComplexity"
language="java"
since="3.9"
message="The {0} ''{1}'' has an NPath complexity of {2}, current threshold is {3}"
class="net.sourceforge.pmd.lang.java.rule.design.NPathComplexityRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#npathcomplexity">
<description>
The NPath complexity of a method is the number of acyclic execution paths through that method.
While cyclomatic complexity counts the number of decision points in a method, NPath counts the number of
full paths from the beginning to the end of the block of the method. That metric grows exponentially, as
it multiplies the complexity of statements in the same block. For more details on the calculation, see the
documentation {% jdoc java::lang.java.metrics.JavaMetrics#NPATH %}.
A threshold of 200 is generally considered the point where measures should be taken to reduce
complexity and increase readability.
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
public static void bar() { // Ncss = 252: reported!
boolean a, b = true;
try { // 2 * 2 + 2 = 6
if (true) { // 2
List buz = new ArrayList();
}
for(int i = 0; i < 19; i++) { // * 2
List buz = new ArrayList();
}
} catch(Exception e) {
if (true) { // 2
e.printStackTrace();
}
}
while (j++ < 20) { // * 2
List buz = new ArrayList();
}
switch(j) { // * 7
case 1:
case 2: break;
case 3: j = 5; break;
case 4: if (b && a) { bar(); } break;
default: break;
}
do { // * 3
List buz = new ArrayList();
} while (a && j++ < 30);
}
}
]]>
</example>
</rule>
<rule name="SignatureDeclareThrowsException"
language="java"
since="1.2"
message="A method/constructor should not explicitly throw java.lang.Exception"
class="net.sourceforge.pmd.lang.java.rule.design.SignatureDeclareThrowsExceptionRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#signaturedeclarethrowsexception">
<description>
A method/constructor shouldn't explicitly throw the generic java.lang.Exception, since it
is unclear which exceptions that can be thrown from the methods. It might be
difficult to document and understand such vague interfaces. Use either a class
derived from RuntimeException or a checked exception.
</description>
<priority>3</priority>
<example>
<![CDATA[
public void foo() throws Exception {
}
]]>
</example>
</rule>
<rule name="SimplifiedTernary"
language="java"
since="5.4.0"
message="This conditional expression can be simplified with || or &&"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#simplifiedternary">
<description>
<![CDATA[
Reports ternary expression with the form `condition ? literalBoolean : foo`
or `condition ? foo : literalBoolean`.
These expressions can be simplified as follows:
* `condition ? true : expr` simplifies to `condition || expr`
* `condition ? false : expr` simplifies to `!condition && expr`
* `condition ? expr : true` simplifies to `!condition || expr`
* `condition ? expr : false` simplifies to `condition && expr`
]]>
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//ConditionalExpression[BooleanLiteral and not(NullLiteral)]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Foo {
public boolean test() {
return condition ? true : something(); // can be as simple as return condition || something();
}
public void test2() {
final boolean value = condition ? false : something(); // can be as simple as value = !condition && something();
}
public boolean test3() {
return condition ? something() : true; // can be as simple as return !condition || something();
}
public void test4() {
final boolean otherValue = condition ? something() : false; // can be as simple as condition && something();
}
public boolean test5() {
return condition ? true : false; // can be as simple as return condition;
}
}
]]>
</example>
</rule>
<rule name="SimplifyBooleanExpressions"
language="java"
since="1.05"
message="Avoid unnecessary comparisons in boolean expressions"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#simplifybooleanexpressions">
<description>
Avoid unnecessary comparisons in boolean expressions, they serve no purpose and impacts readability.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//InfixExpression[@Operator = ("==", "!=")]/BooleanLiteral
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Bar {
// can be simplified to
// bar = isFoo();
private boolean bar = (isFoo() == true);
public isFoo() { return false;}
}
]]>
</example>
</rule>
<rule name="SimplifyBooleanReturns"
language="java"
since="0.9"
message="This if statement can be replaced by `{0}`"
class="net.sourceforge.pmd.lang.java.rule.design.SimplifyBooleanReturnsRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#simplifybooleanreturns">
<description>
Avoid unnecessary if-then-else statements when returning a boolean. The result of
the conditional test can be returned instead.
</description>
<priority>3</priority>
<example>
<![CDATA[
public boolean isBarEqualTo(int x) {
if (bar == x) { // this bit of code...
return true;
} else {
return false;
}
}
public boolean isBarEqualTo(int x) {
return bar == x; // can be replaced with this
}
]]>
</example>
</rule>
<rule name="SimplifyConditional"
language="java"
since="3.1"
message="No need to check for null before an instanceof"
class="net.sourceforge.pmd.lang.java.rule.design.SimplifyConditionalRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#simplifyconditional">
<description>
No need to check for null before an instanceof; the instanceof keyword returns false when given a null argument.
</description>
<priority>3</priority>
<example>
<![CDATA[
class Foo {
void bar(Object x) {
if (x != null && x instanceof Bar) {
// just drop the "x != null" check
}
}
}
]]>
</example>
</rule>
<rule name="SingularField"
language="java"
since="3.1"
message="Perhaps ''{0}'' could be replaced by a local variable."
class="net.sourceforge.pmd.lang.java.rule.design.SingularFieldRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#singularfield">
<description>
Reports fields which may be converted to a local variable. This is so because
in every method where the field is used, it is assigned before it is first read.
Hence, the value that the field had before the method call may not be observed,
so it might as well not be stored in the enclosing object.
Limitations:
* We can only check private fields for now.
* The rule is not aware of threading, so it may cause false positives in concurrent code.
Such FPs are best handled by suppression (see also the `ignoredAnnotations` property).
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
private int x; // this will be reported
public int foo(int y) {
x = y + 5; // assigned before any read
return x;
}
public int fooOk(int y) {
int z = y + 5; // might as well be a local like here
return z;
}
}
]]>
</example>
</rule>
<rule name="SwitchDensity"
language="java"
since="1.02"
message="A high ratio of statements to labels in a switch statement. Consider refactoring."
class="net.sourceforge.pmd.lang.java.rule.design.SwitchDensityRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#switchdensity">
<description>
A high ratio of statements to labels in a switch statement implies that the switch statement
is overloaded. Consider moving the statements into new methods or creating subclasses based
on the switch variable.
</description>
<priority>3</priority>
<example>
<![CDATA[
public class Foo {
public void bar(int x) {
switch (x) {
case 1: {
// lots of statements
break;
} case 2: {
// lots of statements
break;
}
}
}
}
]]>
</example>
</rule>
<rule name="TooManyFields"
language="java"
since="3.0"
message="Too many fields"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#toomanyfields">
<description>
Classes that have too many fields can become unwieldy and could be redesigned to have fewer fields,
possibly through grouping related fields in new objects. For example, a class with individual
city/state/zip fields could park them within a single Address field.
</description>
<priority>3</priority>
<properties>
<property name="maxfields" type="Integer" description="Max allowable fields" min="1" max="1000" value="15"/>
<property name="xpath">
<value>
<![CDATA[
//ClassDeclaration/ClassBody
[count(FieldDeclaration
[not(pmd-java:modifiers() = 'final')]
[not(pmd-java:modifiers() = 'static')]
) > $maxfields]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Person { // too many separate fields
int birthYear;
int birthMonth;
int birthDate;
float height;
float weight;
}
public class Person { // this is more manageable
Date birthDate;
BodyMeasurements measurements;
}
]]>
</example>
</rule>
<rule name="TooManyMethods"
language="java"
since="4.2"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
message="This class has too many methods, consider refactoring it."
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#toomanymethods">
<description>
A class with too many methods is probably a good suspect for refactoring, in order to reduce its
complexity and find a way to have more fine grained objects.
</description>
<priority>3</priority>
<properties>
<property name="maxmethods" type="Integer" description="The method count reporting threshold" min="1" max="1000" value="10"/>
<property name="xpath">
<value>
<![CDATA[
//ClassDeclaration/ClassBody
[
count(MethodDeclaration[
not (
(starts-with(@Name,'get') or starts-with(@Name,'set') or starts-with(@Name,'is'))
and
count(Block/*) <= 1
)
]) > $maxmethods
]
]]>
</value>
</property>
</properties>
</rule>
<rule name="UselessOverridingMethod"
language="java"
since="3.3"
message="Overriding method merely calls super"
class="net.sourceforge.pmd.lang.java.rule.design.UselessOverridingMethodRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#uselessoverridingmethod">
<description>
The overriding method merely calls the same method defined in a superclass.
</description>
<priority>3</priority>
<example>
<![CDATA[
public void foo(String bar) {
super.foo(bar); // why bother overriding?
}
public String foo() {
return super.foo(); // why bother overriding?
}
@Id
public Long getId() {
return super.getId(); // OK if 'ignoreAnnotations' is false, which is the default behavior
}
]]>
</example>
</rule>
<rule name="UseObjectForClearerAPI"
language="java"
since="4.2.6"
message="Rather than using a lot of String arguments, consider using a container object for those values."
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#useobjectforclearerapi">
<description>
When you write a public method, you should be thinking in terms of an API. If your method is public, it means other class
will use it, therefore, you want (or need) to offer a comprehensive and evolutive API. If you pass a lot of information
as a simple series of Strings, you may think of using an Object to represent all those information. You'll get a simpler
API (such as doWork(Workload workload), rather than a tedious series of Strings) and more importantly, if you need at some
point to pass extra data, you'll be able to do so by simply modifying or extending Workload without any modification to
your API.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//MethodDeclaration[pmd-java:modifiers() = 'public']
[count(FormalParameters/FormalParameter[pmd-java:typeIs('java.lang.String')]) > 3]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class MyClass {
public void connect(String username,
String pssd,
String databaseName,
String databaseAdress)
// Instead of those parameters object
// would ensure a cleaner API and permit
// to add extra data transparently (no code change):
// void connect(UserData data);
{
}
}
]]>
</example>
</rule>
<rule name="UseUtilityClass"
language="java"
since="0.3"
message="All methods are static. Consider using a utility class instead. Alternatively, you could add a private constructor or make the class abstract to silence this warning."
class="net.sourceforge.pmd.lang.java.rule.design.UseUtilityClassRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#useutilityclass">
<description>
For classes that only have static methods, consider making them utility classes.
Note that this doesn't apply to abstract classes, since their subclasses may
well include non-static methods. Also, if you want this class to be a utility class,
remember to add a private constructor to prevent instantiation.
(Note, that this use was known before PMD 5.1.0 as UseSingleton).
</description>
<priority>3</priority>
<example>
<![CDATA[
public class MaybeAUtility {
public static void foo() {}
public static void bar() {}
}
]]>
</example>
</rule>
<rule name="MutableStaticState"
language="java"
since="6.35.0"
message="Do not use non-final non-private static fields"
class="net.sourceforge.pmd.lang.rule.xpath.XPathRule"
externalInfoUrl="${pmd.website.baseurl}/pmd_rules_java_design.html#mutablestaticstate">
<description>
Non-private static fields should be made constants (or immutable references) by
declaring them final.
Non-private non-final static fields break encapsulation and can lead to hard to find
bugs, since these fields can be modified from anywhere within the program.
Callers can trivially access and modify non-private non-final static fields. Neither
accesses nor modifications can be guarded against, and newly set values cannot
be validated.
If you are using this rule, then you don't need this
rule {% rule java/errorprone/AssignmentToNonFinalStatic %}.
</description>
<priority>3</priority>
<properties>
<property name="xpath">
<value>
<![CDATA[
//FieldDeclaration[pmd-java:modifiers() = "static"][not(pmd-java:modifiers() = ("private", "final"))]
]]>
</value>
</property>
</properties>
<example>
<![CDATA[
public class Greeter { public static Foo foo = new Foo(); ... } // avoid this
public class Greeter { public static final Foo FOO = new Foo(); ... } // use this instead
]]>
</example>
</rule>
</ruleset>