This article provides instructions on creating a simple resource that will be represented by a JSON object. Complex resources, as well as resources that need to be represented in other formats, will be covered in other articles. This article will also focus on creating a resource based on a database table. Since the SRP HTTP Framework ships with a sample CONTACTS database table, we'll use it for our demonstration purposes. All of our examples will assume the following API is being called:

GET /contacts/1

Resources that are not based on data in a table can still use some of the principles presented below, but there will need to be a little more effort. We will provide additional articles demonstrating how non-database resources can be created.

The Resource Object

We've already explained that a resource can be represented in any digital format, but the conventional format used by RESTful APIs is to use JSON. In our documentation we will often make reference to the resource object, which is short-hand for the "resource as represented by a serialized JSON object". Please note that at times we will also use the term resource as a synonym for resource object, even though a resource can mean much more than that.

Have it Your Way

There are three basic methods for creating a resource within the SRP HTTP Framework although no method completely excludes the use of features from the other methods. You will choose the method that best suits your needs. We will list the pros and cons of each method and provide sample code that creates the same resource using each method.

Method 1: Using the GetDatabaseItem Service

For simple database related resources, the GetDatabaseItem service (a member of the HTTP_Resource_Services module) can be used with minimal code. Just get the Key ID from the prepopulated EndpointSegment variable and pass in the name of the database table:

API contacts.ID.GET

   KeyID   = EndpointSegment

   HTTP_Resource_Services('GetDatabaseItem', 'CONTACTS', '', KeyID)

end api

This produces the following resource object:

{
   "address":"6649 N Blue Gum St",
   "birthdate":"",
   "city":"New Orleans",
   "company":"Benton, John B Jr",
   "county":"Orleans",
   "email":"jbutt@gmail.com",
   "first_name":"James",
   "last_name":"Butt",
   "notes":"",
   "picture":"\\WebAppData\\ContactPictures\\1.jpeg",
   "state":"LA",
   "url":"http://www.bentonjohnbjr.com",
   "zip":"70116",
   "phone":[
      {
         "phone_number":"(504) 621-8927",
         "phone_type":"Phone 1"
      },
      {
         "phone_number":"(504) 845-1427",
         "phone_type":"Phone 2"
      }
   ]
}

One of the drawbacks of calling the GetDatabaseItem service is that property names are formatted with underscores (i.e., how they appear in the dictionary) rather than as camel case (which is the conventional format for JSON objects). Another drawback is that the GetDatabaseItem service attempts to create a resource object from all column data (both physical and calculated). This might be undesirable if some of the data is meaningless to the client. Consider the picture property in the above resource object. It references an image file stored locally on the server, which has no value to the client. This can also be problematic if a calculated column encounters a runtime error or is dependent upon information that only exists within an OpenInsight desktop session. To avoid these problems, we will take advantage of the optional ColumnNames and ItemArrayLabel arguments. To keep our sample code simple and concise, we will limit our resource to just the FIRST_NAME, LAST_NAME, ADDRESS, CITY, STATE, and ZIP database columns:

API contacts.ID.GET

    KeyID           = EndpointSegment

    ColumnNames     = 'FIRST_NAME' : @FM : 'LAST_NAME' : @FM : 'ADDRESS' : @FM : 'CITY' : @FM : 'STATE' : @FM : 'ZIP'
    ItemArrayLabel  = 'firstName' : @FM : 'lastName' : @FM : 'address' : @FM : 'city' : @FM : 'state' : @FM : 'zipCode'
    HTTP_Resource_Services('GetDatabaseItem', 'CONTACTS', '', KeyID, ColumnNames, ItemArrayLabel)

end api

Our resource object now appears like this:

{
    "address": "6649 N Blue Gum St",
    "city": "New Orleans",
    "firstName": "James",
    "lastName": "Butt",
    "state": "LA",
    "zipCode": "70116"
}

Pros:

Cons:

Method 2: Using the GetObject Service

GetDatabaseItem is an example of a high-level service. That is, it relies upon a simple interface and default behavior. Like other high-level services, GetDatabaseItem is built on top of slightly lower-level services. The most important one of these is GetObject (which is another member of the HTTP_Resource_Services module). Let's look at an example of an API that uses the GetObject service to produce the same resource object as the GetDatabaseItem service:

API contacts.ID.GET

    KeyID           = EndpointSegment

    ColumnNames     = 'FIRST_NAME' : @FM : 'LAST_NAME' : @FM : 'ADDRESS' : @FM : 'CITY' : @FM : 'STATE' : @FM : 'ZIP'
    PropertyNames   = 'firstName' : @FM : 'lastName' : @FM : 'address' : @FM : 'city' : @FM : 'state' : @FM : 'zipCode'
    // Create a JSON object in memory.
    objResource     = HTTP_Resource_Services('GetObject', 'CONTACTS', KeyID, ColumnNames, PropertyNames)
    If Error_Services('NoError') then
        // Serialize the JSON object.
        jsonResource    = HTTP_Resource_Services('GetSerializedResource', objResource)
        // Set the response body with the serialized JSON object and set the Content-Type response header.
        HTTP_Services('SetResponseBody', jsonResource, False$, 'application/hal+json')
    end else
        // There is an error condition so call the SetResponseError service.
        HTTP_Services('SetResponseError', '', '', 500, Error_Services('GetMessage'), FullEndpointURL)
    end

end api

In many ways, GetObject and GetDatabaseItem are the same and share some of the same arguments. GetObect differs from GetDatabaseItem in the following ways:

Since GetObject only returns a handle to the JSON object, it is the responsibility of the calling process to serialize the resource and update the HTTP response (e.g., Body, Content-Type Header, Status Code, etc.). The reason developers might choose to work with the GetObject service instead of the GetDatabaseItem service (aside from the latter being deprecated) is because developers might require a little more control and ability to customize the resource object. For this reason, developers have access to several other useful services (aka companion services) such as AddProperty, AddSubProperty, AddSubResource, AddLInkRelation, AddEmbeddedResources, and AddFormAction.

Pros:

Cons:

There is also a GetObjects service. This is a wrapper around the GetObject service. It provides the developer with a Filter argument so multiple rows from a database table can be selected and thus converted into object handles. These object handles are returned as an @FM delimited list to the calling routine. Look at the final code sample in the How do I add a sub-resource to a resource? article to see how useful this can be.

Method 3: Using the SRP_Json Utility Function

When complete control at the lowest level is required, you'll want to use SRP_JSON to create your resource. As noted above, the handle returned by the GetObject service is compatible with the SRP_JSON function and vice-versa. Therefore, a developer can choose to start with either method and continue to use SRP_JSON member services and the higher level GetObject companion services (e.g., AddProperty, AddSubProperty, etc.) at will. One unique feature of SRP_JSON is its ability to interrogate the resource object using services like GetValue. This is useful when a resource object is generated elsewhere. Examples of how this can work will be documented in another article. For now, here is an example of how our simple resource can be created primarily using SRP_JSON:

API contacts.ID.GET

    KeyID           = EndpointSegment

    // Create a JSON object in memory.
    If SRP_JSON(objResource, 'New') then
        ContactRow  = Database_Services('ReadDataRow', 'CONTACTS', KeyID)
        If Error_Services('NoError') then
            SRP_JSON(objResource, 'SetValue', 'firstName', ContactRow<CONTACTS_FIRST_NAME$>, 'String')
            SRP_JSON(objResource, 'SetValue', 'lastName', ContactRow<CONTACTS_LAST_NAME$>, 'String')
            SRP_JSON(objResource, 'SetValue', 'address', ContactRow<CONTACTS_ADDRESS$>, 'String')
            SRP_JSON(objResource, 'SetValue', 'city', ContactRow<CONTACTS_CITY$>, 'String')
            SRP_JSON(objResource, 'SetValue', 'state', ContactRow<CONTACTS_STATE$>, 'String')
            SRP_JSON(objResource, 'SetValue', 'zipCode', ContactRow<CONTACTS_ZIP$>, 'String')
            // Serialize the JSON object and release the object from memory.
            jsonResource    = SRP_JSON(objResource, 'Stringify', 'Fast')
            SRP_JSON(objResource, 'Release')
            // Set the response body with the serialized JSON object and set the Content-Type response header.
            HTTP_Services('SetResponseBody', jsonResource, False$, 'application/hal+json')
        end else
            // There is an error reading the CONTACTS row. Probably a non-existent row.
            HTTP_Services('SetResponseError', '', '', 404, 'Contact ' : KeyID : ' does not exist.', FullEndpointURL)
        end
    end else
        // There is an error condition so call the SetResponseError service.
        HTTP_Services('SetResponseError', '', '', 500, 'Unable to create JSON object', FullEndpointURL)
    end

end api

There are a few items about this sample code that need to be pointed out:

Pros:

Cons: