Suggester
OpenGrok has a suggester support which tries to autocomplete the user queries.
The ChronicleMap dependency does not work out of the box with Java 9+; as a result, promoting terms based on the previous searches won't work. To solve this issue, following parameters are needed to add to java invocation:
--add-exports java.base/jdk.internal.ref=ALL-UNNAMED
--add-exports java.base/sun.nio.ch=ALL-UNNAMED
--add-exports jdk.unsupported/sun.misc=ALL-UNNAMED
--add-exports jdk.compiler/com.sun.tools.javac.file=ALL-UNNAMED
--add-opens jdk.compiler/com.sun.tools.javac=ALL-UNNAMED
--add-opens java.base/java.lang=ALL-UNNAMED
--add-opens java.base/java.lang.reflect=ALL-UNNAMED
--add-opens java.base/java.io=ALL-UNNAMED
--add-opens java.base/java.util=ALL-UNNAMED
If you use Tomcat as a servlet container, these options can be specified by adding the following to the setenv.sh file:
JDK_JAVA_OPTIONS="--add-exports java.base/jdk.internal.ref=ALL-UNNAMED --add-exports java.base/sun.nio.ch=ALL-UNNAMED --add-exports jdk.unsupported/sun.misc=ALL-UNNAMED --add-exports jdk.compiler/com.sun.tools.javac.file=ALL-UNNAMED --add-opens jdk.compiler/com.sun.tools.javac=ALL-UNNAMED --add-opens java.base/java.lang=ALL-UNNAMED --add-opens java.base/java.lang.reflect=ALL-UNNAMED --add-opens java.base/java.io=ALL-UNNAMED --add-opens java.base/java.util=ALL-UNNAMED"Note: | specifies the caret position in the input field.
If the user types a prefix, then suggester suggests the terms with the same prefix. Example for the prefix t|:
- the
- this
- to
- …
If the user specifies a term with a specified fuzzy factor, then the suggester suggests similar terms. Example for the query car|~1:
- can
- bar
- char
- …
If the user types a phrase query, then the suggester tries to suggest the terms that follow. Example for the query "public c|":
- class
- com
- collection
- …
Note: it is possible to use a sloppy factor ~ as well.
If the user types a range query, then the suggester tries to suggest the terms that exist and satisfy the query. Example for the query [a TO b|]:
- by
- basis
- boolean
- …
Note: suggestions work also for the lower term.
If the user types a regexp query, then the suggester suggests the terms that would be accepted by this query. Example for the query /[aA].*|/:
- a
- and
- apache
- …
If the user types a wilcard query, then the suggester suggests the terms that would be accepted by this query.
Example for the query *e:
- the
- file
- use
- …
Suggester configuration is bundled in suggesterConfig property of the configuration. In read-only configuration it looks for example like this:
<void property="suggesterConfig">
<void property="minChars">
<int>3</int>
</void>
</void>Configurable properties of the suggester configuration:
Specifies if the suggester is enabled. Some users might not want the suggester functionality.
- Property:
enabled - Type:
boolean - Default value:
true
Specifies how many results the suggester should return at maximum.
- Property:
maxResults. - Type:
int - Default value:
10
Specifies the minimum number of characters that are needed for the suggester to start looking for suggestions. More characters mean less possible candidates. It can significantly improve performance of the suggester.
- Property:
minChars - Type:
int - Default value:
0
Specifies a set of projects for which the suggester should be enabled.
- Property:
allowedProjects - Type: set of strings
- Default value:
null(all allowed)
Specifies how many projects can be selected at the same time and the suggestions will work.
- Property:
maxProjects - Type:
int - Default value:
Integer.MAX_VALUE(unlimited)
Specifies the fields for which the suggester should be enabled. For instance, it might be desired that the suggestions should only work for the \textit{full} field. OpenGrok uses some fields to store private data and it is not desired to create suggester data structures for those fields. Therefore, only the main search fields are specified by default.
- Property:
allowedFields - Type: set of strings
- Default value:
[full, defs, refs, path, hist, type]
Specifies if the suggester should support complex queries. If set to false then only simple prefix lookups by using the WFST data structure will be performed.
- Property:
allowComplexQueries - Type:
boolean - Default value:
true
Specifies if the most popular completion should be enabled.
If set to false then it slightly increases the performance and there is no need for WFST rebuilds.
- Property:
allowMostPopular - Type:
boolean - Default value:
true
Specifies if the scores should be displayed next to the suggestions
- Property:
showScores - Type:
boolean - Default value:
false
Specifies if the suggestions should show in which project the term was found. If there are multiple projects then showing all the names is not feasible; therefore, only the number of projects will be shown.
- Property:
showProjects - Type:
boolean - Default value:
true
Specifies if the time it took the suggester to find the suggestions should be displayed.
- Property:
showTime - Type:
boolean - Default value:
false
Specifies how often should the suggester rebuild the WFST data structures. The value must be in the Unix cron format
- Property:
rebuildCronConfig - Type:
string - Default value:
0 0 * * *(every day at midnight)
Specifies after how much time the suggester should kill the threads that build the suggester data structures. Slow machines should specify more time. This option is mainly here to prevent the suggester to hang in the initialization.
- Property:
buildTerminationTime - Type:
int - Default value:
1800(30 minutes)
Specifies a time threshold for suggestions in milliseconds. If the computation exceeds this time, it will be stopped and partial results will be returned.
- Property:
timeThreshold - Type:
int - Default value:
2000(2 seconds)
Specifies size of the thread pool used for rebuilding suggester data in terms of percentage of online CPUs in the system.
- Property:
rebuildThreadPoolSizeInNcpuPercent - Type:
int - Default value:
80(80%)
