By default, search results in Bluge are sorted by the DocumentMatch’s score, in descending order. However, users can specify their own sort order.

Preparing the Index

In order to sort on the value of a field, that field must have been indexed with the correction options. Here we see a keyword field named style indexed with the additional Sortable() property:

doc.AddField(bluge.NewKeywordField("style", style).Sortable())

Custom Sort

First, create your SearchRequest as usual:

req := bluge.NewTopNSearch(10, query)

Then, add a custom sort order:


The SortBy method takes a slice of strings describing the custom sort order. In this simple form, strings are interpreted as field names, and the prefix - indictates descending order. A special value of _score is also recognized as the DocumentMatch’s score value.

This can be extended to sort by multiple field values. Here we sort by name ascending, score descending, then finally by document id.

req.SortBy([]string{"name", "-_score", "_id"})

Advanced Sorting

The slice of strings format we just saw is convenient, but does not allow for the full capabilities. A more advanced form can be used:


The SortByCustom method lets us specify a search.SortOrder. The search.SortOrder is a slice of *search.Sort values, which we create using the search.SortBy() helper method. This method takes a single argument which is the extendable search.TextValueSource interface. To refer to a field value indexed with the Sortable option, we can use the search.Field() helper, which takes the field name as it’s argument.

The *search.Sort value defaults to ascending order, with missing values last. These can be flipped using the Desc() and MissingFirst() methods respectively.

While this form seems quite complex, the power comes from the fact that search.TextValueSource can be implemented by your application, to allow sorting by custom functions, which can themselves refer to indexed values.