Full text queries

Full text queries allow you to find results based on analyzed text fields.¹ For example, by a word or phrase within a longer description.

Unlike term-level queries, the search terms you use in a full-text query are analyzed. This means what you search for is tokenized first (broken up into separate words), singular / plural variants determined, synonyms applied, and so on.

Details

Below are the various kinds of full text queries. These are sorted with the most commonly used at the top, and cover their usual usage. Each one is linked to Elasticsearch's own documentation to provide greater details. (In most cases there are many more options for each kind of query than what's documented here.)

Match

Match queries return results where the asset's value for that attribute matches some part of what you're searching.

Java

Build the query and request

IndexSearchRequest index = client.assets.select() // (1)
    .where(Asset.NAME.match("tmp")) // (1)
    .toRequest();

1. You can search across all assets using the select() method of the assets member on any client.

2. Chain a where() onto the select, with the static constant representing a field of the type you want to search to start a query, in this case the NAME of an Asset. Adding the match() predicate creates a match query.

   Equivalent query through Elastic

   Query byMatch =  MatchQuery.of(m -> m
       .field("name")
       .query("tmp")
     ._toQuery();

Python

Build the query

from pyatlan.model.assets import Asset
from pyatlan.model.fluent_search import FluentSearch

index = (FluentSearch()  # (1)
         .where(Asset.NAME.match("tmp"))  # (2)
        ).to_request()

1. You can search across all assets using a FluentSearch() object.
2. Use the class variable representing a field of the type you want to search to start a query, in this case the NAME of an Asset. Adding the match() predicate creates a match query.

Kotlin

Build the query and request

val index = client.assets.select() // (1)
    .where(Asset.NAME.match("tmp")) // (1)
    .toRequest()

1. You can search across all assets using the select() method of the assets member on any client.

2. Chain a where() onto the select, with the static constant representing a field of the type you want to search to start a query, in this case the NAME of an Asset. Adding the match() predicate creates a match query.

   Equivalent query through Elastic

   val byMatch =  MatchQuery.of(m -> m
       .field("name")
       .query("tmp")
     ._toQuery()

Raw REST API

POST /api/meta/search/indexsearch

{
  "dsl": { // (1)
    "query": { // (2)
      "match": { // (3)
        "name": { // (4)
          "query": "tmp" // (5)
        }
      }
    }
  }
}

1. Queries must be within the dsl object in the API...
2. ...and within that the query object.
3. For a match query, there needs to be a match object embedded within the query object.
4. Within this object should be a key with the name of the Elasticsearch field (Atlan attribute) to match against.
5. The value for this field (attribute) to match against should be given through the query property.

Match phrase

Unlike a regular match query, which performs a full-text search and may return results with individual words appearing in any order, a match_phrase query ensures that the terms appear in the exact order specified.

For example, to search for assets where the description field contains the exact phrase "data pipeline":

Java

Coming soon

Python

Build the query

from pyatlan.model.assets import Asset
from pyatlan.model.fluent_search import FluentSearch

index = (
    FluentSearch()  # (1)
    .where(Asset.DESCRIPTION.match_phrase("data pipeline")) # (2)
).to_request()

1. You can search across all assets using a FluentSearch() object.
2. Use the class variable representing a field of the type you want to search to start a query, in this case the DESCRIPTION of an Asset. Adding the match_phrase() predicate creates a match phrase query.

Kotlin

Coming soon

Raw REST API

POST /api/meta/search/indexsearch

{
  "dsl": { // (1)
    "query": { // (2)
      "match_phrase": { // (3)
        "description": { // (4)
          "query": "data pipeline" // (5)
        }
      }
    }
  }
}

1. Queries must be within the dsl object in the API...
2. ...and within that the query object.
3. For a match phrase query, there needs to be a match_phrase object embedded within the query object.
4. Within this object should be a key with the name of the Elasticsearch field (Atlan attribute) to match against.
5. The value for this field (attribute) to match against should be given through the query property.

Footnotes

1. This page is a summary of the details in the Elasticsearch Guide's Full text queries ↩