[jira] [Reopened] (OAK-3407) Configuration to pass suggestion fields through analyzer

["Tommaso Teofili (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/OAK-3407?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tommaso Teofili reopened OAK-3407:
----------------------------------

current ACL check with 
{code}
Collection<String> indexedFields = MultiFields.getIndexedFields(searcher.getIndexReader());
                        QueryParser qp = new MultiFieldQueryParser(Version.LUCENE_47, indexedFields.toArray(new String[indexedFields.size()]), indexNode.getDefinition().getAnalyzer());
{code}

is prone to runtime errors when there're lots of indexed fields.

{nocode}
org.apache.lucene.queryparser.classic.ParseException: Cannot parse ...: too many boolean clauses
{nocde}

While that was introduced to solve the issue when _:suggest_ field has to be used to perform ACL check, it was actually suboptimal.
Matching should be done using the correct analyzer on the _:suggest_ field, depending on the {{defintion.isSuggestAnalyzed}} property.

The fix should be the following:
{code}
QueryParser qp =  new QueryParser(Version.LUCENE_47, FieldNames.SUGGEST, 
                                indexNode.getDefinition().isSuggestAnalyzed() ? indexNode.getDefinition().getAnalyzer() :
                                SuggestHelper.getAnalyzer());
{code}

> Configuration to pass suggestion fields through analyzer
> --------------------------------------------------------
>
>                 Key: OAK-3407
>                 URL: https://issues.apache.org/jira/browse/OAK-3407
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: lucene
>            Reporter: Vikas Saurabh
>            Assignee: Tommaso Teofili
>            Priority: Minor
>             Fix For: 1.3.12
>
>         Attachments: OAK-3407.0.patch
>
>
> Currently, suggested phrases are fairly long. e.g. form {{SuggestTest}} with nodes containing {{in 2015 my fox is red, like mike's fox and john's fox}} and {{in 2015 a red fox is still a fox}} a suggest query for {{in 201}} the result would contain 2 suggestions (both the phrase independently).
> We should probably allow passing suggestion field thorough analyzer so that word level suggestions are possible. Moreover, that would also for suggestion to be available from middle of phrases e.g. suggestions for {{ite}} should give suggestion as {{item}} which, in turn, can search documents containing {{this is the most important item in the whole agenda}}.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)