Javadoc Search Update

This document specifies the behaviour of the Javadoc search feature in JDK 13.

Overview

The Javadoc search feature was introduced in JDK 9 with JEP 225. However, the initial specification only contained a very basic description of the search algorithm. As a consequence, selection and ranking of search results sometimes left a bit to be desired.

The purpose of this additional specification is to improve on the initial implementation by defining better rules for the selection and ranking of search results.

Definitions

The term entity is used to describe an artifact of the documented code, which includes code elements as well as additional search tags.

The term signature is used to describe how an entity is represented in Javadoc search.

The terms name and identifier are used as defined in Section 6.2 of the Java Language Specification, Java SE 12 Edition.

The terms white space and separator are used as defined in Section 3.6 and Section 3.11 of the Java Language Specification, Java SE 12 Edition.

The term camel-case is used to describe mixed case identifiers that make use of uppercase letters to mark word boundaries within the identifier.

The term query string is used to describe the characters entered in the search input box by the user.

All examples in the following sections refer to or are taken from the standard OpenJDK class libraries.

Searchable Entities

The list below describes the entities of documented code and how their signatures are built. For some elements, the signature consists of just the entity's name, while for others it contains information about containing entities or types of parameters.

Modules

The signature of a module equals the module name.

Example:

• java.base

Packages

The signature of a package consists of the package name.

If the package belongs to a named module, the module name is prepended to the signature separated by / and can be included in the search.

Example:

• java.base/java.util.concurrent

Types

The signature of a type (classes, interfaces, enums, annotation types) consists of the type name.

If the type is a nested type, names of parent types are prepended to the signature separated by . and can be included in the search.

If the type is contained in a package, the package name is prepended to the signature separated by . and can be included in the search.

Examples:

• java.lang.Object
• java.util.Map.Entry

Members

The signature of a member (methods, constructors, fields) consists of the member name prepended with its defining type, separated by ..

If the member is a method or constructor, the type names of parameters are appended to the name enclosed between ( and ) and separated by , to identify overloaded methods.

Examples:

• java.lang.Object.wait(long, int)
• java.lang.String.String(String)
• java.lang.Byte.MAX_VALUE

Search Tags

Search Tags are arbitrary searchable items defined using the @index tag anywhere in a Javadoc comment of the documented source code.

Examples:

• {@index "Java Collections Framework"}
• {@index jrt jrt}

Matching Rules

The following sections describe the special rules under which the query string is matched against the entity signatures.

Case Sensitivity

Comparison is case-insensitive unless there is at least one uppercase character in the search string.

Examples

Left Boundaries

The beginning of the query string must match a left word boundary, or a separator preceding a left word boundary.

The following are considered left word boundaries:

• The beginning of an identifier
• An uppercase letter within a camel-case identifier
• The character following a _ within an identifier

Examples:

Partial Matches

The query string matches an identifier even if characters at the end of a name identifiers are missing, as long as the query string matches the beginning of the identifier.

Examples:

Camel Case Matches

The rule for partial matches also applies to uppercase characters followed by any number of lowercase or non-letter characters within a camel-case identifier.

An uppercase character in the query string followed by zero or more lowercase or non-letter characters matches the exact same character sequence in an entity's signature, followed by zero or more lowercase or non-letter characters.

Examples:

Core vs Peripheral Matches

The part of an entity's signature that represents the entity's name is considered the core component of the signature, while any other parts of the signature are considered peripheral components.

An entity will only be included in the search result if the query string contains and matches at least part of its core component.

Examples:

Although the parameter types of a method or constructor are not considered a core component of its signature, it is possible to search for methods or constructors with specific parameter types by starting the search string with (.

Example:

Accessing Child Elements

A query string that matches a code element can be turned into a search string that matches the element's child elements by appending the separator used to connect the two levels of code elements.

Examples:

Ranking of Results

A match at the beginning of an identifier is ranked higher than a match on an uppercase character within a within a camel-case identifier and will appear higher up in the search result.

Example:

• Query string set matches types java.util.Set and java.util.HashSet but the former is ranked higher than the latter.

Partial matches are ranked by the characters that are missing in the search string. The less characters missing the higher the entity is ranked.

Examples:

• Query string java.lang.ref matches both packages java.lang.ref and java.lang.reflect but the former is ranked higher than the latter because it represents an exact match.
• Query string j.l.r also matches both packages in above example. Again, java.lang.ref is ranked higher as the query string matches it more closely (less missing characters).

White Space

White space in the query string is treated depending on context.

• White space that separates two identifiers is considered significant. A matching signature must contain white space at the corresponding position.
• White space preceding or following a separator, or at the beginning or the end of the query string, is considered optional. A matching signature may or may not contain white space at the corresponding position.

White space is matched regardless of the actual number of space characters in the query string or signature. One or more space characters in the query string matches one or more space characters in the signature.

If the query string consists entirely of white space no search is performed.

Examples:

Supported Browsers

The search feature is supported in the following browsers: