Elasticsearch highlighting boolean values instead of search phrase

Question

I have ran into a problem with elasticsearch highlighting. I am using elasticsearch-web plugin from Rivetlogic to integrate Elasticsearch into Liferay portal. It works just fine, but when I use highlighter on some documents, the highlighted words are wrong. These problem doesn't seem to be connected with Rivetlogic plugin itself, I was able to simulate it through Sense addon with plain elasticsearch query as well.

An example query:

POST /liferay_company_20155/com_liferay_portlet_documentlibrary_model_DLFileEntry/_search
{
   "query": {
      "query_string": {
         "query": "+(+(companyId:20155) +((+(entryClassName:com.liferay.portal.model.User) +(status:0)) (+(entryClassName:com.liferay.portlet.bookmarks.model.BookmarksEntry) +(status:0)) (+(entryClassName:com.liferay.portlet.bookmarks.model.BookmarksFolder) +(status:0)) (+(entryClassName:com.liferay.portlet.blogs.model.BlogsEntry) +(status:0)) (+(entryClassName:com.liferay.portlet.documentlibrary.model.DLFileEntry) +(status:0) +(hidden:false)) (+(entryClassName:com.liferay.portlet.documentlibrary.model.DLFolder) +(status:0) +(hidden:false)) (+(entryClassName:com.liferay.portlet.journal.model.JournalArticle) +(status:0) +(head:true)) (+(entryClassName:com.liferay.portlet.journal.model.JournalFolder) +(status:0)) (+(entryClassName:com.liferay.portlet.messageboards.model.MBMessage) +(status:0) +(discussion:false)) (+(entryClassName:com.liferay.portlet.wiki.model.WikiPage) +(status:0)))) +(assetCategoryTitles:*zkouska* assetCategoryTitles_cs_CZ:*zkouska* assetTagNames:*zkouska* comments:zkouska content:zkouska description:zkouska properties:zkouska title:zkouska url:zkouska userName:*zkouska* -stagingGroup:true city:zkouska country:zkouska emailAddress:*zkouska* firstName:zkouska fullName:zkouska lastName:zkouska middleName:zkouska region:zkouska screenName:zkouska street:zkouska zip:zkouska ddmContent:zkouska extension:zkouska fileEntryTypeId:zkouska path:*zkouska* classPK:zkouska content_cs_CZ:zkouska description_cs_CZ:zkouska entryClassPK:zkouska title_cs_CZ:zkouska type:zkouska articleId:zkouska)"
      }
   },
      "highlight": {
          "pre_tags" : ["<tag1>"],
        "post_tags" : ["</tag1>"],
         "fields": {
            "content": {}
         }
      }
}

The result highlight looks like this:

"highlight": 
{
"content": [
  " logické\n1 nebo <tag1>0</tag1> (<tag1>true</tag1> nebo <tag1>false</tag1>).\n\nfunction ALTERNATIV(P:real): Boolean;\nvar X: real;\nbegin\n\nX",
  " pouze na změnu <tag1>FALSE</tag1> na <tag1>TRUE</tag1>, případně na\npřekročení mezní hodnoty směrem nahoru). Protože C",
  " (metoda Test), a to buď z hodnoty nula (<tag1>FALSE</tag1>) na\nhodnotu různou od nuly (<tag1>TRUE</tag1>), nebo obráceně",
  "\n\ndetekci změny pouze z hodnoty <tag1>FALSE</tag1> na hodnotu <tag1>TRUE</tag1>, DetectDOWN detekuje opačnou\nzměnu. DetectALL",
  " článek. Nechť X(ui) = x, Y (ui) = y, T =\n〈<tag1>0</tag1>,∞), I = O = R, I je vstupní abeceda, O je výstupní abeceda"
]
}

Notice, that all the values "FALSE", "TRUE" and "0" are highlighted. The queried word "zkouska" is also highlighted.

Is there any problem with the query?

Any help is appriciated.

keety keety · Accepted Answer · 2015-08-21T18:24:31

Setting require_field_match to true should fix this.

Example:

POST /liferay_company_20155/com_liferay_portlet_documentlibrary_model_DLFileEntry/_search
{
   "query": {
      "query_string": {
         "query": "+(+(companyId:20155) +((+(entryClassName:com.liferay.portal.model.User) +(status:0)) (+(entryClassName:com.liferay.portlet.bookmarks.model.BookmarksEntry) +(status:0)) (+(entryClassName:com.liferay.portlet.bookmarks.model.BookmarksFolder) +(status:0)) (+(entryClassName:com.liferay.portlet.blogs.model.BlogsEntry) +(status:0)) (+(entryClassName:com.liferay.portlet.documentlibrary.model.DLFileEntry) +(status:0) +(hidden:false)) (+(entryClassName:com.liferay.portlet.documentlibrary.model.DLFolder) +(status:0) +(hidden:false)) (+(entryClassName:com.liferay.portlet.journal.model.JournalArticle) +(status:0) +(head:true)) (+(entryClassName:com.liferay.portlet.journal.model.JournalFolder) +(status:0)) (+(entryClassName:com.liferay.portlet.messageboards.model.MBMessage) +(status:0) +(discussion:false)) (+(entryClassName:com.liferay.portlet.wiki.model.WikiPage) +(status:0)))) +(assetCategoryTitles:*zkouska* assetCategoryTitles_cs_CZ:*zkouska* assetTagNames:*zkouska* comments:zkouska content:zkouska description:zkouska properties:zkouska title:zkouska url:zkouska userName:*zkouska* -stagingGroup:true city:zkouska country:zkouska emailAddress:*zkouska* firstName:zkouska fullName:zkouska lastName:zkouska middleName:zkouska region:zkouska screenName:zkouska street:zkouska zip:zkouska ddmContent:zkouska extension:zkouska fileEntryTypeId:zkouska path:*zkouska* classPK:zkouska content_cs_CZ:zkouska description_cs_CZ:zkouska entryClassPK:zkouska title_cs_CZ:zkouska type:zkouska articleId:zkouska)"
      }
   },
      "highlight": {
         "require_field_match" : true,
          "pre_tags" : ["<tag1>"],
        "post_tags" : ["</tag1>"],
         "fields": {
            "content": {}
         }
      }
}

Also on an unrelated note looking at the response it looks like the mapping for status, hidden fields seems to be string probably you want it to be boolean.

Elasticsearch highlighting boolean values instead of search phrase

1 Answers