AzureSearch

Use AzureSearch service with .Net SDK

In Azure Portal

Login to azure portal and find “Azure Search” item from the marketplace.

Create the search service, navigate to it and locate the menu option “Keys”. Two keys are necessary to connect and perform queries. The first key needed is the “Primary admin key”. This is used to create a search index instance in the .Net application.

The second key needed is the “Query key”. This allow to retrieve data from the search index service.

In .Net application

In the .Net application the following NuGet packages are required:

• Microsoft.Azure.Search
• Microsoft.Azure.Search.Common
• Microsoft.Azure.Search.Data
• Microsoft.Azure.Search.Service
• Microsoft.Rest.ClientRuntime
• Microsoft.Rest.ClientRuntime.Azure

It is necessary to create a model type for the search index. The model used in this example is called "People" and contains the index fields and the attributes that allow to query documents.

public class People
{
    [System.ComponentModel.DataAnnotations.Key]
    [IsFilterable]
    [JsonProperty("id")]
    public string Id { get; set; }

    [IsSearchable]
    [JsonProperty("first_name")]
    public string FirstName { get; set; }

    [IsSearchable, IsFilterable, IsSortable]
    [JsonProperty("last_name")]
    public string LastName { get; set; }

    [IsSearchable]
    public string Email { get; set; }

    [IsSearchable, IsFilterable, IsSortable, IsFacetable]
    public string Gender { get; set; }

    [IsSearchable]
    [JsonProperty("ip_address")]
    public string IpAddress { get; set; }
}

JsonProperty attributes are used to map a sample Json file. This property names will be also used as the document field names when the search index is created. Note that there are other attributes such as IsSearchable, IsFilterable, IsSortable, IsFacetable used in the property fields. Those denote how the fields are going to be used for search results. For example, “IsSearchable” authorises the field to be full-text searchable. More information about these attributes can be found in Setting attibutes

Now that the initial set up is ready in the application, it is time to create and perform queries on the index.

Create search index

The first step consists in creating a search service client instance

SearchServiceClient serviceClient =
  new SearchServiceClient(<your_search_service_name>, 
  new SearchCredentials(<your_search_admin_apikey>)
);

As a safety measure it is always useful to check whether or not the index has being already created. For this example the search index is called "people"

if (serviceClient.Indexes.Exists("people"))
{
  serviceClient.Indexes.Delete("people");
}

var definition = new Index()
{
  Name = indexName,
  Fields = FieldBuilder.BuildForType<People>()
};

serviceClient.Indexes.Create(definition);

Next step is to get an instance of the index previously created.

ISearchIndexClient indexClient = serviceClient.Indexes.GetClient(“people”);

It is also required to generate the document objects to populate the index. These are created as a collection of IndexAction items, and then added to an IndexBatch.

List<IndexAction<People>> actions = new List<IndexAction<People>>();

actions.Add(
    IndexAction.MergeOrUpload(
        new People()
        {
            Id = p.Id,
            FirstName = p.FirstName,
            LastName = p.LastName,
            Email = p.Email,
            Gender = p.Gender,
            IpAddress = p.IpAddress
        }),
        
        ...
        
);

var batch = IndexBatch.New(actions);

The action used in this example is "MergeOrUpload" which updates the document if this exist already or creates a new one if does not. More information about Index actions can be found in Index actions

Finally when the index batch is ready, it is time to populate the search index.

indexClient.Documents.Index(batch);

Run Queries on index

A search index client instance is required to run queries. Using the index name "people" the following line shows how to create the index client instance:

ISearchIndexClient indexClientForQueries = AzureSearchService.CreateSearchIndexClient(“people”);

It is also necessary to define and instance of SearchParameters to perform some queries. These might contain options such as Filter, OrderBy, Select, Top and more. However, these are optional. The following example shows how to query documents (people items) applying a filter where gender = Female, then ordering the results by lastName and finally getting the first 10 results. Results will only contain First name, Last name and email fields.

SearchParameters  parameters = new SearchParameters()
    {
        Filter = "gender eq 'Female'",
        OrderBy = new[] { "last_name asc" },
        Select = new[] { "first_name", "last_name", "email" }
        Top = 10
    };

DocumentSearchResult<People>  searchResults = indexClient.Documents.Search<People>("*", parameters);

foreach (SearchResult<People> result in searchResults.Results)
{
    var singleDocument = result.Document;
    ...
}

Another way to query documents can be done without parameters but only the search text.

DocumentSearchResult<People>  searchResults = indexClient.Documents.Search<People>(" phil* ", new SearchParameters());

Previous query will perform a full-text search in the index returning all documents that contain the word "phil". Same query run in the Azure portal will look like this:

You will need an active Azure subscription to run the console application. After source code is downloaded, locate App.config and replace the following keys with the ones you generated

  <appSettings>
      <add key="SearchServiceName" value="your_index_name"/>
      <add key="SearchServiceAdminApiKey" value="your_search_index_admin_key"/>
      <add key="SearchServiceQueryApiKey" value="your_search_index_query_key"/>
  </appSettings>

Using the Azure Search SDK for .Net allow us to perform operations with Lucene indexes proficiently.