Table of Contents
Filter files may be used to include or exclude bug reports for particular classes and methods. This chapter explains how to use filter files.
Conceptually, a filter matches bug instances against a set of criteria. By defining a filter, you can select bug instances for special treatment; for example, to exclude or include them in a report.
A filter file is an XML document with a top-level FindBugsFilter
element
which has some number of Match
elements as children. Each Match
element represents a predicate which is applied to generated bug instances.
Usually, a filter will be used to exclude bug instances. For example:
$
findbugs -textui -excludemyExcludeFilter.xml
myApp.jar
However, a filter could also be used to select bug instances to specifically report:
$
findbugs -textui -includemyIncludeFilter.xml
myApp.jar
Match
elements contain children, which are conjuncts of the predicate.
In other words, each of the children must be true for the predicate to be true.
<Bug>
This element specifies a particular bug pattern or patterns to match.
The pattern
attribute is a comma-separated list of
bug pattern types. You can find the bug pattern types for particular
warnings by looking at the output produced by the -xml
output option (the type
attribute of BugInstance
elements), or from the bug
descriptions document.
For more coarse-grained matching, use code
attribute. It takes
a comma-separated list of bug abbreviations. For most-coarse grained matching use
category
attriute, that takes a comma separated list of bug category names:
CORRECTNESS
, MT_CORRECTNESS
,
BAD_PRACTICICE
, PERFORMANCE
, STYLE
.
If more than one of the attributes mentioned above are specified on the same
<Bug>
element, all bug patterns that match either one of specified
pattern names, or abreviations, or categories will be matched.
As a backwards compatibility measure, <BugPattern>
and
<BugCode>
elements may be used instead of
<Bug>
element. Each of these uses a
name
attribute for specifying accepted values list. Support for these
elements may be removed in a future release.
<Confidence>
This element matches warnings with a particular bug confidence.
The value
attribute should be an integer value:
1 to match high-confidence warnings, 2 to match normal-confidence warnings,
or 3 to match low-confidence warnings. <Confidence> replaced
<Priority> in 2.0.0 release.
<Priority>
Same as <Confidence>
, exists for backward compatibility.
<Rank>
This element matches warnings with a particular bug rank.
The value
attribute should be an integer value
between 1 and 20, where 1 to 4 are scariest, 5 to 9 scary, 10 to 14 troubling,
and 15 to 20 of concern bugs.
<Package>
This element matches warnings associated with classes within the package specified
using name
attribute. Nested packages are not included (along the
lines of Java import statement). However matching multiple packages can be achieved
easily using regex name match.
<Class>
This element matches warnings associated with a particular class. The
name
attribute is used to specify the exact or regex match pattern
for the class name. The role
attribute is the class role.
As a backward compatibility measure, instead of element of this type, you can use
class
attribute on a Match
element to specify
exact an class name or classregex
attribute to specify a regular
expression to match the class name against.
If the Match
element contains neither a Class
element,
nor a class
/ classregex
attribute, the predicate will apply
to all classes. Such predicate is likely to match more bug instances than you want, unless it is
refined further down with apropriate method or field predicates.
<Source>
This element matches warnings associated with a particular source file. The
name
attribute is used to specify the exact or regex match pattern
for the source file name.
<Method>
This element specifies a method. The name
is used to specify
the exact or regex match pattern for the method name.
The params
attribute is a comma-separated list
of the types of the method's parameters. The returns
attribute is
the method's return type. The role
attribute is
the method role. In params
and returns
, class names
must be fully qualified. (E.g., "java.lang.String" instead of just
"String".) If one of the latter attributes is specified the other is required for creating a method signature.
Note that you can provide either name
attribute or params
and returns
attributes or all three of them. This way you can provide various kinds of
name and signature based matches.
<Field>
This element specifies a field. The name
attribute is is used to specify
the exact or regex match pattern for the field name. You can also filter fields according to their signature -
use type
attribute to specify fully qualified type of the field. You can specify eiter or both
of these attributes in order to perform name / signature based matches. The role
attribute is
the field role.
<Local>
This element specifies a local variable. The name
attribute is is used to specify
the exact or regex match pattern for the local variable name. Local variables are variables defined within a method.
<Or>
This element combines Match
clauses as disjuncts. I.e., you can put two
Method
elements in an Or
clause in order to match either method.
<And>
This element combines Match
clauses which both must evaluate to true. I.e., you can put
Bug
and Confidence
elements in an And
clause in order
to match specific bugs with given confidence only.
<Not>
This element inverts the included child Match
. I.e., you can put a
Bug
element in a Not
clause in order to match any bug
excluding the given one.
If the name
attribute of Class
, Source
,
Method
or Field
starts with the ~ character
the rest of attribute content is interpreted as
a Java regular expression that is matched against the names of the Java element in question.
Note that the pattern is matched against whole element name and therefore .* clauses need to be used at pattern beginning and/or end to perform substring matching.
See java.util.regex.Pattern
documentation for pattern syntax.
Match
clauses can only match information that is actually contained in the
bug instances. Every bug instance has a class, so in general, excluding
bugs by class will work.
Some bug instances have two (or more) classes. For example, the DE (dropped exception)
bugs report both the class containing the method where the dropped exception
happens, and the class which represents the type of the dropped exception.
Only the first (primary) class is matched against Match
clauses.
So, for example, if you want to suppress IC (initialization circularity)
reports for classes "com.foobar.A" and "com.foobar.B", you would use
two Match
clauses:
<Match> <Class name="com.foobar.A" /> <Bug code="IC" /> </Match> <Match> <Class name="com.foobar.B" /> <Bug code="IC" /> </Match>
By explicitly matching both classes, you ensure that the IC bug instance will be matched regardless of which class involved in the circularity happens to be listed first in the bug instance. (Of course, this approach might accidentally supress circularities involving "com.foobar.A" or "com.foobar.B" and a third class.)
Many kinds of bugs report what method they occur in. For those bug instances,
you can put Method
clauses in the Match
element and they should work
as expected.
1. Match all bug reports for a class.
<Match> <Class name="com.foobar.MyClass" /> </Match>
2. Match certain tests from a class by specifying their abbreviations.
<Match> <Class name="com.foobar.MyClass"/ > <Bug code="DE,UrF,SIC" /> </Match>
3. Match certain tests from all classes by specifying their abbreviations.
<Match> <Bug code="DE,UrF,SIC" /> </Match>
4. Match certain tests from all classes by specifying their category.
<Match> <Bug category="PERFORMANCE" /> </Match>
5. Match bug types from specified methods of a class by their abbreviations.
<Match> <Class name="com.foobar.MyClass" /> <Or> <Method name="frob" params="int,java.lang.String" returns="void" /> <Method name="blat" params="" returns="boolean" /> </Or> <Bug code="DC" /> </Match>
6. Match a particular bug pattern in a particular method.
<!-- A method with an open stream false positive. --> <Match> <Class name="com.foobar.MyClass" /> <Method name="writeDataToFile" /> <Bug pattern="OS_OPEN_STREAM" /> </Match>
7. Match a particular bug pattern with a given priority in a particular method.
<!-- A method with a dead local store false positive (medium priority). --> <Match> <Class name="com.foobar.MyClass" /> <Method name="someMethod" /> <Bug pattern="DLS_DEAD_LOCAL_STORE" /> <Priority value="2" /> </Match>
8. Match minor bugs introduced by AspectJ compiler (you are probably not interested in these unless you are an AspectJ developer).
<Match> <Class name="~.*\$AjcClosure\d+" /> <Bug pattern="DLS_DEAD_LOCAL_STORE" /> <Method name="run" /> </Match> <Match> <Bug pattern="UUF_UNUSED_FIELD" /> <Field name="~ajc\$.*" /> </Match>
9. Match bugs in specific parts of the code base
<!-- match unused fields warnings in Messages classes in all packages --> <Match> <Class name="~.*\.Messages" /> <Bug code="UUF" /> </Match> <!-- match mutable statics warnings in all internal packages --> <Match> <Package name="~.*\.internal" /> <Bug code="MS" /> </Match> <!-- match anonymoous inner classes warnings in ui package hierarchy --> <Match> <Package name="~com\.foobar\.fooproject\.ui.*" /> <Bug pattern="SIC_INNER_SHOULD_BE_STATIC_ANON" /> </Match>
10. Match bugs on fields or methods with specific signatures
<!-- match System.exit(...) usage warnings in void main(String[]) methods in all classes --> <Match> <Method returns="void" name="main" params="java.lang.String[]" /> <Bug pattern="DM_EXIT" /> </Match> <!-- match UuF warnings on fields of type com.foobar.DebugInfo on all classes --> <Match> <Field type="com.foobar.DebugInfo" /> <Bug code="UuF" /> </Match>
11. Match bugs using the Not filter operator
<!-- ignore all bugs in test classes, except for those bugs specifically relating to JUnit tests --> <!-- i.e. filter bug if ( classIsJUnitTest && ! bugIsRelatedToJUnit ) --> <Match> <!-- the Match filter is equivalent to a logical 'And' --> <Class name="~.*\.*Test" /> <!-- test classes are suffixed by 'Test' --> <Not> <Bug code="IJU" /> <!-- 'IJU' is the code for bugs related to JUnit test code --> </Not> </Match>
12. Full exclusion filter file to match all classes generated from Groovy source files.
<?xml version="1.0" encoding="UTF-8"?> <FindBugsFilter> <Match> <Source name="~.*\.groovy" /> </Match> </FindBugsFilter>
<FindBugsFilter> <Match> <Class name="com.foobar.ClassNotToBeAnalyzed" /> </Match> <Match> <Class name="com.foobar.ClassWithSomeBugsMatched" /> <Bug code="DE,UrF,SIC" /> </Match> <!-- Match all XYZ violations. --> <Match> <Bug code="XYZ" /> </Match> <!-- Match all doublecheck violations in these methods of "AnotherClass". --> <Match> <Class name="com.foobar.AnotherClass" /> <Or> <Method name="nonOverloadedMethod" /> <Method name="frob" params="int,java.lang.String" returns="void" /> <Method name="blat" params="" returns="boolean" /> </Or> <Bug code="DC" /> </Match> <!-- A method with a dead local store false positive (medium priority). --> <Match> <Class name="com.foobar.MyClass" /> <Method name="someMethod" /> <Bug pattern="DLS_DEAD_LOCAL_STORE" /> <Priority value="2" /> </Match> <!-- All bugs in test classes, except for JUnit-specific bugs --> <Match> <Class name="~.*\.*Test" /> <Not> <Bug code="IJU" /> </Not> </Match> </FindBugsFilter>