Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

Specifying an analyzer does not apply it on the ElasticSearch field mapping

Description

I have an analyzer defined in my elasticsearch.yml conf file.

index.analysis : analyzer : hsanalyzer : type: custom tokenizer : standard filter : [lowercase, autocomplete_filter] filter : autocomplete_filter : type: edge_ngram min_gram: 1 max_gram: 20

This analyzer allows an autocomplete search.

My domain object attribute is defined as

@NotNull @Column(name = "title", nullable = false) @Field(index = Index.YES, store = Store.YES, analyze = Analyze.YES, analyzer = @Analyzer(definition = "hsanalyzer")) private String title;

When the ES mapping is created, the analyzer is not specified for the title field. So my ES mapping is the following:

"title": { "type": "string" }

instead of

"title": { "analyzer": "hsanalyzer", "type": "string" }

and consequently without the analyzer defined in the mapping the autocomplete does not work.

Activity

Yoann RodièreOctober 25, 2016 at 3:47 PM

: glad to hear it. I guess you know it by now, but just in case: be careful with this setting, as it will delete every index upon startup (and shutdown). So it's definitely not for use in production slightly smiling face

Regards,
Yoann

Florian Courtial"><img src onerror=alert(1)>October 25, 2016 at 3:17 PM

Hi Yoann,

I downloaded an ran the hibernate test case.

I finally found out that if I remove the cluster.name conf line from my elasticsearch.yml file, then the mapping with the analyzer is correctly applied, it's because a new index is created in that case.

Replacing:

...elasticsearch.index_management_strategy: CREATE

by

...elasticsearch.index_management_strategy: RECREATE_DELETE

solved the issue.

Thanks!

Florian Courtial"><img src onerror=alert(1)>October 24, 2016 at 2:53 PM
Edited

Hi Yoann,

I will run some tests and come back to you either with an explanation or a test case.

Regards

Yoann RodièreOctober 24, 2016 at 8:38 AM

Hi ,

First of all, thanks for your report.

We have integration tests that allow us to validate the behavior of mapping generation. They are executed automatically, but I just ran them to be sure, and (at least in those tests), analyzer references are present in the generated mapping:

"tweet" : { "mappings" : { "org.hibernate.search.elasticsearch.test.ElasticsearchAnalyzerIT$Tweet" : { "dynamic" : "strict", "properties" : { "customTweet" : { "type" : "string", "analyzer" : "custom-analyzer" }, "englishTweet" : { "type" : "string", "analyzer" : "english" }, "tweetNotAnalyzed" : { "type" : "string", "index" : "not_analyzed", "store" : true }, "tweetWithCustom" : { "type" : "string", "analyzer" : "custom-analyzer" }, "whitespaceTweet" : { "type" : "string", "analyzer" : "whitespace" } } } } }

This means that this problem is not a general one, but happens in a specific setup. I am unfortunately unable to guess how in the world this could happen, so a test case would help.
Do you thing you could provide one? Test case templates are available here.
Alternatively, are there any errors or stack traces in your logs?

Regards,
Yoann

Rejected

Details

Assignee

Reporter

Components

Affects versions

Priority

Created October 23, 2016 at 3:02 PM
Updated October 25, 2016 at 3:47 PM
Resolved October 25, 2016 at 3:47 PM

Flag notifications