Home > Articles > Data > SQL Server

  • Print
  • + Share This
This chapter is from the book

To get started accessing SQL Server via HTTP, you should set up an IIS virtual directory using the Configure IIS Support menu option in the SQLXML program folder. Of course, you can retrieve XML data from SQL Server without setting up a virtual directory (e.g., by using ADO or OLE DB); I'm referring exclusively to retrieving XML data from SQL Server via HTTP.

Configuring a virtual directory allows you to work with SQL Server's XML features via HTTP. You use a virtual directory to establish a link between a SQL Server database and a segment of a URL. It provides a navigation path from the root directory on your Web server to a database on your SQL Server.

SQL Server's ability to publish data over HTTP is made possible through SQLISAPI, an Internet Server API (ISAPI) extension that ships with the product. SQLISAPI uses SQLOLEDB, SQL Server's native OLE DB provider, to access the database associated with a virtual directory and return results to the client.

Client applications have four methods of requesting data from SQL Server over HTTP. These can be broken down into two broad types: those more suitable for private intranet access because of security concerns, and those safe to use on the public Internet.

Private Intranet

  1. Post an XML query template to SQLISAPI.

  2. Send a SELECT…FOR XML query string in a URL.

Public Internet

  1. Specify a server-side XML schema in a virtual root.

  2. Specify a server-side XML query template in a virtual root.

Due to their open-ended nature, methods 1 and 2 could pose security risks over the public Internet but are perfectly valid on corporate or private intranets. Normally, Web applications use server-side schemas and query templates to make XML data accessible to the outside world in a controlled fashion.

Configuring a Virtual Directory

Load the Configure IIS Support utility in the SQLXML folder under Start | Programs. You should see the IIS servers configured on the current machine. Click the plus sign to the left of your server name to expand it. (If your server isn't listed—for example, if it's a remote server—right-click the IIS Virtual Directory Manager node and select Connect to connect to your server.) To add a new virtual directory, right-click the Default Web Site node and select New | Virtual Directory. You should then see the New Virtual Directory Properties dialog.

Specifying a Virtual Directory Name and Path

The Virtual Directory Name entry box is where you specify the name of the new virtual directory. This is the name that users will include in a URL to access the data exposed by the virtual directory, so it's important to make it descriptive. A common convention is to name virtual directories after the databases they reference. To work through the rest of the examples in the chapter, specify Northwind as the name of the new virtual directory.

Though Local Path will sometimes not be used, it's required nonetheless. In a normal ASP or HTML application, this would be the path where the source files themselves reside. In SQLISAPI applications, this folder does not necessarily need to contain anything, but it must exist nevertheless. On NTFS partitions, you must also make sure that users have at least read access to this folder in order to use the virtual directory. You configure which user account will be used to access the application (and thus will need access to the folder) in the dialog's Security page.

Click the Security tab to select the authentication mode you'd like to use. You can use a specific user account, Windows Integrated Authentication, or Basic (clear text) Authentication. Select the option that matches your usage scenario most closely; Windows Integrated Authentication will likely be the best choice for working through the demos in this chapter.

Next, click the Data Source page tab. This is where you set the SQL Server and the database that the virtual directory references. Select your SQL Server from the list and specify Northwind as the database name.

Go to the Virtual Names table and set up two virtual names, templates and schemas. Create two folders under Northwind named Templates and Schemas so that each of these virtual names can have its own local folder. Set the type for schemas to schema and the type for templates to template. Each of these provides a navigation path from a URL to the files in its local folder. We'll use them later.

The last dialog page we're concerned with is the Settings page. Click it, then make sure every checkbox on it is checked. We want to allow all of these options so that we may test them later in the chapter. The subsections below provide brief descriptions of each of the options on the Settings page.

Allow sql=… or template=… or URL queries

When this option is enabled, you can execute queries posted to a URL (via an HTTP GET or POST command) as sql= or template= parameters. URL queries allow users to specify a complete Transact-SQL query via a URL. Special characters are replaced with placeholders, but, essentially, the query is sent to the server as is, and its results are returned over HTTP. Note that this option allows users to execute arbitrary queries against the virtual root and database, so you shouldn't enable it for anything but intranet use. Go ahead and enable it for now so that we can try it out later.

Selecting this option disables the Allow template=… containing updategrams only option because you can always post XML templates with updategrams when this option is selected. The Allow template=… containing updategrams only option permits XML templates (that contain only updategrams) to be posted to a URL. Since this disallows SQL and XPath queries from existing in a template, it provides some limited security.

Template queries are by far the most popular method of retrieving XML data from SQL Server over HTTP. XML documents that store query templates—generic parameterized queries with placeholders for parameters—reside on the server and provide a controlled access to the underlying data. The results from template queries are returned over HTTP to the user.

Allow XPath

When Allow XPath is enabled, users can use a subset of the XPath language to retrieve data from SQL Server based on an annotated schema. Annotated schemas are stored on a Web server as XML documents and map XML elements and attributes to the data in the database referenced by a virtual directory. XPath queries allow the user to specify the data defined in an annotated schema to return.

Allow POST

HTTP supports the notion of sending data to a Web server via its POST command. When Allow POST is enabled, you can post a query template (usually implemented as a hidden form field on a Web page) to a Web server via HTTP. This causes the query to be executed and returns the results back to the client.

As I mentioned earlier, the open-endedness of this usually limits its use to private intranets. Malicious users could form their own templates and post them over HTTP to retrieve data to which they aren't supposed to have access or, worse yet, make changes to it.

Run on the client

This option specifies that XML formatting (e.g., FOR XML) is to be done on the client side. Enabling this option allows you to offload to the client the work of translating a rowset into XML for HTTP queries.

Expose runtime errors as HTTP error

This option controls whether query errors in an XML template are returned in the HTTP header or as part of the generated XML document. When this option is enabled and a query in a template fails, HTTP error 512 is returned and error descriptions are returned in the HTTP header. When it's disabled and a template query fails, the HTTP success code, 200, is returned, and the error descriptions are returned as processing instructions inside the XML document.

Enable all the options on the Settings page except the last two described above and click OK to create your new virtual directory.


TIP: A handy option on the Advanced tab is Disable caching of mapping schemas. Normally, mapping schemas are cached in memory the first time they're used and accessed from the cache thereafter. While developing a mapping schema, you'll likely want to disable this so that the schema will be reloaded each time you test it.

  • + Share This
  • 🔖 Save To Your Account