String lexicographic range query in Elasticsearch

Question

Is it possible to query Elasticsearch giving it a range of strings?

Something that I'd imagine like:

Sample Mapping:

     {
      "resource" : {
        "properties" : {
          "Title" : {
            "type" : "string"
          },
          "type_" : {
            "index" : "not_analyzed",
            "type" : "string"
          },
           "Summary" : {
            "format" : "dateOptionalTime",
            "type" : "date"
          }
        }
      }
    }

Sample Query:

{
  "size" : 10,
  "query" : {
    "filtered" : {
      "query" : {
        "bool" : {
          "should" : [ {
            "text" : {
              "Title" : {
                "query" : "AAA",
                "type" : "phrase_prefix"
              }
            }
          }, {
            "range" : {
              "Title" : {
                "from" : "BBB",
                "to" : "CCC",
                "include_lower" : true,
                "include_upper" : true
              }
            }
          } ],
          "minimum_number_should_match" : 1
        }
      },
      "filter" : {
        "and" : {
          "filters" : [{
            "or" : {
              "filters" : [ {
                "term" : {
                  "type_" : "personType"
                }
              } ]
            }
          } ]
        }
      }
    }
  }
}

Data Indexed: Resources with Titles 'AAA', 'BBB', 'CCC', 'DDD'

Result Resource with Title 'AAA' (range didn't select 'BBB' and 'CCC')

Any help appreciated

javanna · Accepted Answer · 2012-09-20 14:05:35Z

2

Yes, it is possible. The RangeQuery provided by the elasticsearch query DSL internally uses the lucene TermRangeQuery for field of type string. On the other hand, it uses the lucene NumericRangeQuery for field of type number/date.

answered Sep 20, 2012 at 14:05

javanna

60.4k14 gold badges147 silver badges125 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

Queequeg Over a year ago

Well theres something wrong with my ES, because if I index some resorurces with "name" being dates, then a range query works. If, on the other hand, the "name" are just strings, then the range does not work, always 0 results. How come?

javanna Over a year ago

It probably depends on the way you are indexing data. Have you tried lowercasing the terms in your query?

Queequeg Over a year ago

well, I have a very basic index mapping because the resources I am indexing have very varied information/fields inside e.g. some have 'name', other have 'title' etc. So I leave it to ES to figure things out and it worked perfectly. But I cant make a range search with strings, even when I have indexed only the names "a","b", "c" and I query with a rage <"a","c"> - no results

javanna Over a year ago

Weird, I just tried what you described and it worked. I'd suggest to check again your data, maybe using the analyze API to see what you're actually indexing.

Queequeg · Accepted Answer · 2012-09-21 09:29:23Z

1

By default ElasticSearch keeps all indexed data lowercase. It does not lowercase the queries, though. Great.

To wrap up: after sending the queries (from and to values) all in lowercase the range searching works like a charm.

answered Sep 21, 2012 at 9:29

Queequeg

2,9248 gold badges43 silver badges69 bronze badges

1 Comment

javanna Over a year ago

I tried to tell you it was about lowercasing ;) Some queries are actually analyzed, which means that the query goes into the same analysis process as the document field. The Range Query is not one of them. You are relying on the default mapping (which uses the lucene StandardAnalyzer underneath) for the title field, but the query is used as it is, so lowercasing it manually is a good choice.

Collectives™ on Stack Overflow

String lexicographic range query in Elasticsearch

2 Answers 2

4 Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

4 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related