About full-text search client

📘
Note
The Full-text Search Service applies to Optimizely Content Management System (CMS 11) and is not supported on Episerver Digital Experience Platform (DXP) and other cloud service deployments. It is a legacy feature that will be removed in future versions.
This section is not supported for ASP.NET Core-based applications.

The Full-Text Search Client (FTS Client) is a Framework component offering an API for adding, updating, and removing searchable items in a full-text search index. The component itself acts as a REST client, communicating over HTTP using Atom Syndication Format with extensions described in this document. The Optimizely FTS Client API is shipped together with the Optimizely FTS Service, a WCF REST service on top of an unmodified version of the open-source search engine, Lucene.NET. Optimizely FTS is an attempt to index any kind of data including CMS page data. Content in files is indexed, and the installed Ifilters will decide which files are included.

Assemblies and namespaces

The EPiServer.Search assembly contains the following namespaces:

EPiServer.Search contains core classes for the EPiServer FTS Client API, most notably the IndexRequestItem, IndexResponseItem, and SearchHandler classes.
EPiServer.Search.Configuration contains configuration classes for the Optimizely FTS Client API.
EPiServer.Search.Filter contains classes to support providers for filtering search results.
EPiServer.Search.Queries contains classes for querying the index service. Built-in queries are found in the EPiServer.Search.Queries.Lucene namespace.
EPiServer.Search.Data contains classes for integration with the Dynamic Data Store which is used for the index request queue.

Update the index

You can modify the index as follows:

Creating an IndexRequestItem specifying the unique (within the index) item ID and IndexAction (add, update, remove).
Populating it with searchable data by assigning properties.
Calling the SearchHandler.Instance.UpdateIndex method, passing the IndexRequestItem.

The Optimizely FTS Client requests are asynchronous, meaning that the request is immediately enqueued and later dequeued in a separate thread in a configurable time interval or when a queue flush is explicitly requested. The queue items are added to an Atom feed which is sent to the FTS service. If the request fails, the batch is kept and sent again at the next time interval until it succeeds. Because the items need to be indexed in the same order they are added, updated, or removed, any failing batch blocks subsequent batches. The feed itself also contains a feed attribute extension Version that contains the name and current version of the Optimizely FTS Client so that the current indexing service can handle different client versions.

The SearchHandler.Instance.UpdateIndex has the following overloads:

void UpdateIndex(IndexRequestItem item) – Updates the default configured indexing service.
void UpdateIndex(IndexRequestItem item, string namedIndexingService) – Updates the configured namedIndexingService. The Episerver FTS Client can thus be configured to use multiple services.

IndexItemBase

The IndexItemBase is an abstract class inherited by IndexRequestItem and IndexResponseItem with the following public members:

Property	Type	Comment
`AccessControlList`	`Collection<string>`	A list of groups and users that have access to this item. See AccessControlList.
`Authors`	`Collection<string>`	A list of authors to associate with this item.
`BoostFactor`	`float`	The boost factor is used to explicitly set a higher rank of this item when searching.
`Categories`	`Collection<string>`	A list of categories to associate with this item.
`Created`	`DateTime`	The date when this item was created.
`Culture`	`string`	The culture for this item.
`DataUri`	`Uri`	A Uri to where the indexing service should fetch external data that is not part of the feed, such as files.
`DisplayText`	`string`	The indexing service should return the part of the searchable text when getting search results.
`Id`	`string`	The unique identified for this item within this named index.
`ItemStatus`	`ItemStatus`	The status (Approved, Pending, Removed) for this item.
`Metadata`	`string`	Additional searchable data of “unlimited” size that is not supposed to be returned or displayed together with the search results.
`ItemType`	`string`	The type of this item. For example, `EPiServer.Community.Blog`, `EPiServer.Core.PageData`.
`Modified`	`DateTime`	The date when this item was last modified.
`NamedIndex`	`string`	Specifies the named index to update when multiple indexes are configured in the indexing service. See Named indexes and Multi search in the About full-text search server topic.
`PublicationEnd`	`DateTime?`	The expiration date for this item, after which the item is returned in searches.
`PublicationStart`	`DateTime?`	The start date for when this item is included in searches.
`ReferenceId`	`string`	A reference to a “parent” item in the index and is used when having searchable content in sub-items (for example, Comments) where matches in the sub-item should result in a search hit for the parent item. See the ReferenceId section in the About full-text search server topic.
`Title`	`string`	The title of the item.
`Uri`	`Uri`	The Uri to use when linking the search result item to the content source.
`VirtualPathNodes`	`Collection<string>`	A list of nodes that build a path under which the item can be found. Used together with trailing wildcard queries. For example, if item1 has a path of A/B and item2 has a path of A, a query like A* would result in two results (_item1and_item2) while a query like A/B* would result in one result (_item1_). See VirtualPath in the About full-text search server topic. Note: Any white spaces within a node are automatically removed.

IndexRequestItem

The IndexRequestItem class inherits the IndexItemBase and holds the searchable content to be indexed.

Property	Type	Comment
`IndexAction`	`IndexAction`	Enum for Add, Update, Remove.
`AutoUpdateVirtualPath`	`bool`	Boolean value indicates if the query should affect other items whose virtual path nodes begin with the same nodes as the ones provided in the requested item.

IndexResponseItem

The IndexResponseItem class inherits the IndexItemBase and holds the content returned by the indexing service.

Property	Type	Comment
`Score`	`float`	The score set by the indexing service for this IndexResponseItem.

Query the index

Querying the index is done by passing an object of type IQueryExpression to SearchHandler.Instance.GetSearchResults. The IQueryExpression interface defines the property string QueryExpression which returns the textual representation of the expression to be parsed by the indexing service.

The SearchHandler.Instance.GetSearchResults has two overloads:

SearchResults GetSearchResults(IQueryExpression queryExpression, int page, int pageSize) searches the default configured namedIndexingService and the default namedIndex in the service for matches for the passed queryExpression.
SearchResults GetSearchResults(IQueryExpression queryExpression, string namedIndexingService, List<string> namedIndexes, int page, int pageSize) connects to the past namedIndexingService and searches the list of passed namedIndexes by performing a multi-search. See Named indexes and Multi search in the Full-text search topic.

The page and pageSizeSearchResults parameter together with the TotalHits enables paging of search results, see Paging in the Full-text search topic.

Query enums

The EPiServer.Search.Queries.Lucene namespace contains the following enums that is used by many of the query expressions:

Enum	Comment
`Field`	The Field enum holds values for separately searchable fields in the indexing service by queries taking a Field in the constructor. For example, finding occurrences for hello world in the Field. Title and/or sv in the `Field.Culture`.
`LuceneOperator`	The Operator enum holds values for AND, OR, and NOT.

Implemented queries

The EPiServer.Search.Queries.Lucene namespace contains the following query expression implementations compatible with the Optimizely FTS Service:

Class	Constructor	Comment
`AccessControlQuery`	`LuceneOperator` `innerOperator`	Empty constructor. `AccessControlList` returns a `List<string>` where users and groups that should have read access to the index document can be added. Any group or user added to the query must match at least one group or user in the `IndexItemBase.AccessControlList`. The default `innerOperator` is `OR`.
`CategoryQuery`	`LuceneOperator` `innerOperator`	Queries the index for occurrences with an exact match for the category literal in the query.
`CreatedDateRangeQuery`	`DateTime start`, `DateTime end`, `bool inclusive`	Inherits `RangeQuery`. Match occurrences where the `Created` field lies within the start and end parameters.
`FieldQuery`	`string queryExpression`, `Field field`	Queries the index with the passed `queryExpression` (typically a word or a phrase) for the passed `Field`. If no `Field` is passed the `Field.Default` is used. Example: `FieldQuery q = new FieldQuery("\" looking for something\"");` returns occurrences where any searchable field contains the phrase looking for something.
`FuzzyQuery`	`string word`, `Field field`, `float similarityFactor`	Queries the index for occurrences of similar words for the passed word. The higher `similarityFactor` the more similar the word needs to be to create an index hit. For example, a field containing you may reconsider and this is a reconsideration match for a certain `similarityFactor`.
`GroupQuery`	`LuceneOperator` `innerOperator`	Group queries do not query result in any query expression in itself but is used to group queries with an operator between then. The `QueryExpressions` property returns a `List<IQueryExpression>` where query expressions can be added with the `innerOperator` between them. `GroupQuery` is an `IQueryExpression` which lets you combine groups and expressions with their respective inner operators recursively.
`ItemStatusQuery`	`ItemStatus status`	Inherits `GroupQuery`. Match occurrences where the `ItemStatus` value matches any of passed `ItemStatus` flags. You can pass several `ItemStatus` values to the constructor by using bit operands, for example, new `ItemStatusQuery(ItemStatus.ApprovedItemStatus.Pending)`. Occurrences that match ANY of the passed `ItemStatus` values are accepted as matches. You should not add additional queries to the `ItemStatusQuery` `QueryExpressions` list. If you want to combine an `ItemStatusQuery` with another query, you should combine them using a `GroupQuery`.
`ModifiedDateRangeQuery`	`DateTime start`, `DateTime end`, `bool inclusive`	Inherits `RangeQuery`. Match occurrences where the modified field lies within the start and end parameters.
`ProximityQuery`	`String phrase`, `Field field`, `int distance`	Queries the index for occurrences where the words in the passed phrase exist with the passed distance from each other. For example, a document with the phrase Optimizely for the win would match the `ProximityQuery` phrase Optimizely win with the passed distance of 2.
`RangeQuery`	`String start`, `string end`, `Field field`, `bool inclusive`	Queries the index for occurrences within the passed literal start and end. For example, 20010202, 20030303, `Field.Modified`, true match index documents with the `Field.Modified` set between (and included) the start and end.
`TermBoostQuery`	`String phrase`, `Field field`, `float boostFactor`	Queries the index with a boost factor for the field, giving occurrences for the phrase in the field higher scores/relevance.
`VirtualPathQuery`	()	Empty constructor. The property `VirtualPathNodes` returns a `List<string>` where path nodes can be added in the order they appear from the root. The resulting path matches any index items with a path that starts with the query path.

SearchResults

The SearchResults class holds data from an indexing service response and is assigned internally when getting IndexResponseItems back from the service:

Property	Type	Comment
`IndexResponseItems`	`List<IndexResponseItem>`	Gets a list of `IndexResponseItems` returned by the indexing service
`TotalHits`	`int`	Gets the number of the total matching items in the indexing service.
`Version`	`string`	Gets the current indexing service name and version.

Filter search results

The Optimizely FTS Client lets you plug in one or many filter providers to explicitly include or exclude specific items from the results by overriding the SearchResultFilter.SearchResultFilterProvider.Filter(IndexResponseItem item) method and adding a provider to the configuration. The SearchResultFilter return type is an Enum with values for Include, Exclude and NotHandled. Where the IndexResponseItem is not handled by this provider, the system should forward it to the next configured provider. If the IndexResponseItem is not handled by any configured provider, the configured defaultInclude behavior is used.

Atom and extensions

The standard Atom format is extended with both attributes and elements with the default EPiServer.Search.IndexingService namespace to map the data in the IndexRequestItem and IndexResponseItem as shown in the following table:

Property	Name	Extension
`IndexItemBase.AccessControlList`	ACL	Element Extension
`IndexItemBase.Authors`	Authors	No
`IndexItemBase.BoostFactor`	BoostFactor	Attribute extension
`IndexItemBase.Categories`	Categories	No
`IndexItemBase.Created`	PublishDate	No
`IndexItemBase.Culture`	Culture	Attribute extension
`IndexItemBase.DataUri`	DataUri	Attribute extension
`IndexItemBase.DisplayText`	Content	No
`IndexItemBase.Id`	Id	No
`IndexItemBase.ItemType`	Type	Attribute extension
`IndexItemBase.Metadata`	Metadata	Element extension
`IndexItemBase.Modified`	LastUpdatedTime	No
`IndexItemBase.NamedIndex`	NamedIndex	Attribute extension
`IndexItemBase.ReferenceId`	ReferenceId	Attribute extension
`IndexItemBase.Title`	Title	No
`IndexItemBase.Uri`	BaseUri	No
`IndexItemBase.Version`	Version	Attribute extension
`IndexItemBase.VirtualPath`	VirtualPath	Element extension
`IndexRequestItem.IndexAction`	IndexAction	Attribute extension
`IndexRequestItem.AutoUpdateVirtualPath`	AutoUpdateVirtualPath	Attribute extension
`IndexResponseItem.Score`	Score	Attribute extension

Example of a typical update index request

Request

POST http://localhost.:8072/IndexingService/update/?accesskey=accesskey1 
  HTTP/1.1
  Content-Type: f
  Host: localhost.:8072
  Content-Length: 773
  Expect: 100-continue
<?xml version="1.0" encoding="utf-8"?>
  <feed p1:Version="EPiServer.Search v.6.1.28.0" 
        xmlns:p1="EPiServer.Search.IndexingService" 
        xmlns="http://www.w3.org/2005/Atom">
    <title></title>
    <id>uuid:6d85bc21-020e-4058-994c-090061f9d89c;id=12</id>
    <updated>2010-04-26T20:02:02Z</updated>
    <entry p1:IndexAction="add" 
           p1:BoostFactor="1" 
           p1:Type="" p1:Culture="" 
           p1:NamedIndex="testindex2" 
           p1:ReferenceId="">
      <id>1</id>
      <title></title>
      <published>2010-04-26T22:02:02+02:00</published>
      <updated>2010-04-26T22:02:02+02:00</updated>
      <content></content>
      <p1:Metadata xmlns:p1="EPiServer.Search.IndexingService"></p1:Metadata>
      <ACL xmlns="EPiServer.Search.IndexingService"></ACL>
      <VirtualPath xmlns="EPiServer.Search.IndexingService"></VirtualPath>
    </entry>
  </feed>

Response

HTTP/1.1 200 OK
Content-Length: 0
Server: Microsoft-HTTPAPI/1.0
Date: Mon, 26 Apr 2010 20:02:02 GMT

Example of a typical get search results response

Request

GET http://localhost.:8072/IndexingService/search/?q=EPISERVER_SEARCH_ID%3a(1)&namedindexes=&offset=0&limit=20&format=xml&accesskey=accesskey1 
  HTTP/1.1
  Content-Type: application/xml
  Host: localhost.:8072

Response

HTTP/1.1 200 OK
Content-Length: 638
Content-Type: application/xml; charset=utf-8
Server: Microsoft-HTTPAPI/1.0
Date: Mon, 26 Apr 2010 20:02:02 GMT

Example of a typical reset index request

Request

POST http://localhost.:8072/IndexingService/reset/?namedindex=default&accesskey=accesskey1 
  HTTP/1.1
  Content-Type: application/xml
  Host: localhost.:8072
  Content-Length: 0

Response

HTTP/1.1 200 OK
Content-Length: 0
Server: Microsoft-HTTPAPI/1.0
Date: Mon, 26 Apr 2010 20:02:00 GMT

Example of a typical get named indexes request

Request

GET http://localhost.:8072/IndexingService/namedindexes/?accesskey=accesskey1 
  HTTP/1.1
  Content-Type: application/xml
  Host: localhost.:8072
  Connection: Keep-Alive

Response

HTTP/1.1 200 OK
Content-Length: 1048
Content-Type: application/xml; charset=utf-8
Server: Microsoft-HTTPAPI/1.0
Date: Mon, 26 Apr 2010 20:02:00 GMT