This code adds search functionality to your EpiServer project based on Azure Cognitive Search service
Once you create Azure Cognitive Search service you'll get service name and admin keys. You will need to provide them during initialization process described in the next section.
NOTE: service name is without search.windows.net suffix.
For basic search initialization it's enough to add code like this:
In your startup class.
Step 1.
public void ConfigureServices(IServiceCollection services)
{
// (...)
services.AddEpiServerAzureSearch<ContentDocument, DefaultDocumentBuilder>("yourservicename", "YOURADMINKEY");
// (...)
}You need to pass your service name and the admin key.
Step 2.
public void Configure(IApplicationBuilder app, IWebHostEnvironment env)
{
// (...)
app.UseEpiServerAzureSearch<ContentDocument>();
// (...)
}This will add scheduled job, called: [Search] Index content to your EpiServer installation which is responsible for full index of content and will register event handlers so content is updated on publish events.
This package allows to configure EpiServer built in search features (in admin panel) to make use of Azure Search Index.
Basic usage of this is to add class in your code which you will mark with [SearchProvider] attribute and extend ContentSearchProviderBase class.
[SearchProvider]
public class PageSearchProvider : ContentSearchProviderBase<ContentDocument>
{
public PageSearchProvider(IAzureSearchService azureSearchService, IContentLanguageAccessor contentLanguageAccessor) : base(azureSearchService, contentLanguageAccessor)
{
}
public override string Area => "CMS/pages";
public override string Category => "Find pages";
}You can adjust Area and Category values to your needs - depending which search (pages/blocks/media) it should be attached to
It is highly probable that you'll want to extend built in ContentDocument class. You'll have to do it in case when, for example, you want to add new field to be indexed. This is how you do it:
Let's define new model. Only restriction here is that it inherits from SearchDocument but in this example we'll extend built in ContentDocument:
public class MyCustomDocument : ContentDocument
{
public int[] Categories { get; set; }
}Now, we have to create logic which will be responsible for converting your EpiServer content do your new document type MyCustomDocument.
In order to do so, we'll create MyCustomDocumentBuilder class:
public class MyCustomDocumentBuilder : DefaultDocumentBuilder<MyCustomDocument>
{
public MyCustomDocumentBuilder(IUrlResolver urlResolver,
IContentLoader contentLoader,
IContentExtractorController contentExtractorController,
IContentTypeRepository contentTypeRepository)
: base(urlResolver, contentLoader, contentExtractorController, contentTypeRepository)
{
}
public override MyCustomDocument Build(IContent content)
{
var document = base.Build(content);
if (content is PageData pageData)
{
document.Categories = pageData.Category.ToArray();
}
return document;
}
public override MyCustomDocument Build(PageData pageData)
{
return Build(pageData as IContent);
}
}Now, we have to setup search to make use of new document and document builder. In order to do so, just set your types as a generic parameters in startup class.
Step 1.
public void ConfigureServices(IServiceCollection services)
{
// (...)
services.AddEpiServerAzureSearch<MyCustomDocument, MyCustomDocumentBuilder>("yourservicename", "YOURADMINKEY");
// (...)
}Step 2.
public void Configure(IApplicationBuilder app, IWebHostEnvironment env)
{
// (...)
app.UseEpiServerAzureSearch<MyCustomDocument>();
// (...)
}Content document has this field:
[IsSearchable]
public string[] ContentBody { get; set; }It is meant to store all content which is supposed to be searchable - article body, intro, section page summary etc.
By default, if you don't do anything, no content will be extracted there. There are two options you can do it:
You can mark any field with this attribute and its content will be added as a content to ContentBody field.
public class ArticlePage : PageData
{
[Forte.EpiServer.AzureSearch.Model.Indexable]
public virtual XhtmlString Body { get; set; }
[Forte.EpiServer.AzureSearch.Model.Indexable]
public virtual string Summary { get; set; }
}This attribute does further extraction for properties of types XhtmlString, 'ContentArea', any block type (derivative of BlockData) and 'ContentReference'(as long as it's referencing block and not page). On all other field types, ToString() will be called. If you want to do something more sophisticated, like order content extraction texts results, use IContentExtractor instead.
You can create class which will implement IContentExtractor and decide what and when should end up in ContentBody. Inject and use 'Forte.EpiServer.AzureSearch.ContentExtractor.XhtmlStringExtractor' for extracting plain text from Xhtml based properties types (XHtmlString and ContentArea):
public class ArticlePageContentExtractor : IContentExtractor
{
private readonly XhtmlStringExtractor _xhtmlStringExtractor;
public ArticlePageContentExtractor(XhtmlStringExtractor xhtmlStringExtractor)
{
_xhtmlStringExtractor = xhtmlStringExtractor;
}
public bool CanExtract(IContentData content)
{
return content is ArticlePage;
}
public ContentExtractionResult Extract(IContentData content, IContentExtractorController extractor)
{
var article = (ArticlePage) content;
var articleBody = _xhtmlStringExtractor.GetPlainTextContent(article.Body, extractor);
var articleSummary = article.Summary;
return new ContentExtractionResult(new[] {articleBody, articleSummary}, null);
}
}Note, that once created, such class has to be registered in your startup class, for example:
public void ConfigureServices(IServiceCollection services)
{
// (...)
services.AddTransient<IContentExtractor, ArticlePageContentExtractor>();
// (...)
}