Optimizely Search and Navigation - Part 2 - Filter Tips

Introduction

Continuing from Part 1 – Search Tips, today I will share the next part – filter tips.

The platform versions used for this article are Optimizely CMS 12.27.x, Optimizely Customized Commerce 14.21.x, and EpiServer.Find 16.1.x.

How to Add a New Custom Filter

In the built-in Optimizely Search & Navigation, the default filters are as follows:

• And Filter
• Bool Filter
• Exists Filter
• Geo Distance Filter
• Geo Distance Range Filter
• Geo Polygon Filter
• Has Child Filter
• Ids Filter
• Kilometers Filter
• Nested Filter
• Not Filter
• Or Filter
• Prefix Filter
• Query Filter
• Range Filter
• Script Filter
• Term Filter

These filters generate corresponding body text when sending requests to Elasticsearch through the Search & Navigation framework using JSON converters.

What if a filter exists in Elasticsearch but not in Search & Navigation, or an existing filter lacks required parameters? Can you create a new filter?

Yes, you can add new custom filters in Search & Navigation. Here's how to build a new Regular Expression filter.

Step 1: Create a New DTO Based on Filter Class

[JsonConverter(typeof(RegularExpressionFilterConverter))]
public class RegularExpressionFilter : Filter
{
    [JsonIgnore]
    public string Field { get; set; }

    [JsonProperty("value")]
    public string Value { get; set; }

    public RegularExpressionFilter(string field, string value)
    { 
        Field = field;
        Value = value;
    }
}

Step 2: Create a New JSON Converter for the New Filter

internal class RegularExpressionFilterConverter : CustomWriteConverterBase<RegularExpressionFilter>
{
    public override void WriteJson(JsonWriter writer, object value, JsonSerializer serializer)
    {
        if (value is not RegularExpressionFilter)
        {
            writer.WriteNull();
            return;
        }
        var regularExpressionFilter = (RegularExpressionFilter)value;
        writer.WriteStartObject();
        writer.WritePropertyName("regexp");
        writer.WriteStartObject();
        writer.WritePropertyName(regularExpressionFilter.Field);
        serializer.Serialize(writer, regularExpressionFilter.Value);
        writer.WriteEndObject();
        writer.WriteEndObject();
    }
}

Step 3: Create a New Filter Method for Your Search

public static DelegateFilterBuilder MatchRegularExpression(this string value, string input)
{ 
   return new DelegateFilterBuilder((string field) => new RegularExpressionFilter(field, input));
}

Step 4: Apply the New Filter to Your Search

if (!string.IsNullOrEmpty(filterOptions.Q)){
    query = query.Filter(x => x.Name.MatchRegularExpression(filterOptions.Q));
}

The keyword for search in this example could be text within regular expression syntax. You can find syntax for ElasticSearch here.

How to Apply AND/OR for a Group of Sub Filters

Consider the following example:

You have a product content type with these properties:

[Display(Name = "On sale", GroupName = SystemTabNames.Content, Order = 50)]
public virtual bool OnSale { get; set; }

[Display(Name = "New arrival", GroupName = SystemTabNames.Content, Order = 55)]
public virtual bool NewArrival { get; set; }

[Display(Name = "Best seller", GroupName = SystemTabNames.Content, Order = 58)]
public virtual bool BestSeller { get; set; }

To find products that are on sale and/or new arrivals, you can:

Use AND/OR Operator in FilterExpression:

• AND operator

query = query.Filter(x => x.OnSale.Match(true) & x.NewArrival.Match(true));

• OR operator

query = query.Filter(x => x.OnSale.Match(true) | x.NewArrival.Match(true));

Use AND/OR Filter:

• AND filter

var andFilter = new AndFilter();
andFilter.Filters.Add(new TermFilter(searchClient.GetFullFieldName("OnSale", typeof(bool)), true));
andFilter.Filters.Add(new TermFilter(searchClient.GetFullFieldName("NewArrival", typeof(bool)), true));
query = query.Filter(andFilter);

• OR Filter

var orFilter = new OrFilter();
orFilter.Filters.Add(new TermFilter(searchClient.GetFullFieldName("OnSale", typeof(bool)), true));
orFilter.Filters.Add(new TermFilter(searchClient.GetFullFieldName("NewArrival", typeof(bool)), true));
query = query.Filter(orFilter);

Note that the field name understood by Elasticsearch is not the same as the field name in the Optimizely Content Type Model. Use the following method to get the indexed field name:

public static string GetFullFieldName(this IClient searchClient, string fieldName, Type type)
{
    if (type != null)
        return fieldName + searchClient.Conventions.FieldNameConvention.GetFieldName(Expression.Variable(type, fieldName));
    return fieldName;
}

Pros and Cons:

• Using AND/OR Operator: Short and easy to use but not suitable for dynamically building filters.
• Using AND/OR Filter: Longer code but allows for flexible filter building based on field names and input values.

How to Sort Search Results Based on a Set of Conditions

Search & Navigation allows sorting based on one or more fields:

query = query.OrderBy(x => x.Name).ThenByDescending(x => x.StartPublish);

Example Requirement: Display best seller products at the top, followed by new arrivals, and then on sale products.

You can achieve this by using boost matching:

query = query.BoostMatching(x => (x as GenericProduct).NewArrival.Match(true), 2);
query = query.BoostMatching(x => (x as GenericProduct).OnSale.Match(true), 3);
query = query.BoostMatching(x => (x as GenericProduct).BestSeller.Match(true), 4);

Issue: If a product is both on sale and a new arrival, it appears at the top even if it is not a best seller. The score of this product (5) is higher than that of a best seller (4).

Solution: Ensure best sellers are always on top by setting Boost value as following rule: Next boost value = total of all previous boost values + 1:

query = query.BoostMatching(x => (x as GenericProduct).NewArrival.Match(true), 2);
query = query.BoostMatching(x => (x as GenericProduct).OnSale.Match(true), 3);
query = query.BoostMatching(x => (x as GenericProduct).BestSeller.Match(true), 6);

Conclusion

These tips are based on my experience with Search & Navigation. I hope they help you implement similar features.

Enjoy coding!