The SearchMessageService component includes logic for determining, when a response is received, whether to send a JMS message. The purpose of this logic is to prevent sending of messages that will distort the Search reporting results. For example, if a query has been initiated by a Web spider or by the Search Testing feature in ATG Merchandising rather than by a site visitor, you typically will not want to take this query into account in reports.
SearchMessageService suppresses messages in the following situations:
If the request is determined to come from browser type that is considered a Web spider
If the request comes from a specified IP address or user account
Detecting Web Spiders
Web spiders (also called robots) crawl the Web and create indexes for Web search services such as Google. This activity is generally benign, but it can skew reporting results. For example, if a spider issues a search query, that query will be reflected in any search reports you run. This is generally undesirable, since you typically want the reports to reflect only queries issued by actual site visitors. Therefore, the SearchMessageService has a mechanism for determining whether a search query is being issued by a spider, and if it is, suppressing the firing of search events.
To enable this mechanism, you set the SearchMessageService component’s spiderlikeTypes property to an array of the atg.servlet.BrowserType components that you consider to be spiders. When a search query is issued, the SearchMessageService examines the userAgent property of the search request (e.g., the QueryRequest or BrowseRequest component). SearchMessageService compares the value of the userAgent property with the values of the patterns properties of the spiderlikeTypes components, and if it finds a match, suppresses the events.
Typically, the search request’s userAgent property is set to the value of the User-Agent property of the HTTP request. You can override this value by explicitly setting the userAgent property in the properties file of the search request component. This is what happens in the ATG Merchandising Search Testing environment. When you use Search Testing, the userAgent property of the QueryRequest component is set to SearchTesting. By default, one of the spiderlikeTypes components is /atg/dynamo/servletpipeline/BrowserTypes/Robot, and one of the entries in this component’s patterns array is SearchTesting, so events are not fired.
Filtering by IP Address or User Account
The SearchMessageService has a messageFilter property that points to a component of class atg.search.events.MessageFilter. The MessageFilter component determines whether to suppress Search messages, based on the IP address or the user account associated with the request. By default, the SearchMessageService component’s messageFilter property is set to /atg/search/events/MessageFilter.
To configure this component, set the following properties:
Property | Description |
|---|---|
| A array of the IP addresses that Search messages should not be fired for. Note that the entries can include wildcards (e.g., 10.1.4.*). |
| A list of usernames that Search messages should not be fired for. |
