Use wwJsonSerializer to create Sample Data

I frequently create demos for various applications and components and quite frequently I create objects on the fly using EMPTY objects using ADDPROPERTY(). The code that does is a bit verbose and looks something like this:

*** Note objects are serialized as lower case
loCustomer = CREATEOBJECT("EMPTY")
 
*** Recommend you assign your own ids for easier querying
ADDPROPERTY(loCustomer,"_id",SYS(2015))
ADDPROPERTY(loCustomer,"FirstName","Markus")
ADDPROPERTY(loCustomer,"LastName","Egger")
ADDPROPERTY(loCustomer,"Company","EPS Software")
ADDPROPERTY(loCustomer,"Entered", DATETIME())
 
loAddress = CREATEOBJECT("EMPTY")
ADDPROPERTY(loAddress,"Street","32 Kaiea")
ADDPROPERTY(loAddress,"City","Paia")
ADDPROPERTY(loCustomer,"Address",loAddress)
 
loOrders = CREATEOBJECT("Collection")
ADDPROPERTY(loCustomer,"Orders",loOrders)
 
loOrder = CREATEOBJECT("Empty")
ADDPROPERTY(loOrder,"Date",DATETIME())
ADDPROPERTY(loOrder,"OrderId",SUBSTR(SYS(2015),2))
ADDPROPERTY(loOrder,"OrderTotal",120.00)
loOrders.Add(loOrder)
 
loOrder = CREATEOBJECT("Empty")
ADDPROPERTY(loOrder,"Date",DATETIME())
ADDPROPERTY(loOrder,"OrderId",SUBSTR(SYS(2015),2))
ADDPROPERTY(loOrder,"OrderTotal",120.00)
loOrders.Add(loOrder)

While this works, it’s pretty verbose and a little hard to read with all the nesting and relationships. It’s still a lot less code than explicitly creating a class and then assigning properties but even so, it’s hard to see the relationship between the nested objects. The more nesting there is the more nasty this sort of code gets regardless of whether you use AddProperty() or CREATEOBJECT() with assignments.

Given that Json Serialization is readily available now, it’s actually pretty easy to replace this code with something that’s a little easier to visualize.

Take the following code which produces a very similar object by using a JSON string serialized to an object:

DO wwJsonSerializer
 
*** Create a sample object
TEXT TO lcJson TEXTMERGE NOSHOW
{
    _id: "<< loMongo.GenerateId() >>",
    FirstName: "Rick",
    LastName: "Strahl",
    Company: "West Wind",
    Entered: "<<TTOC(DateTime(),3)>>Z",
    Address: {
        Street: "32 Kaiea",
        City: "Paia"
    },
    Orders: [
        { OrderId: "ar431211", OrderTotal: 125.44 },
        { OrderId: "fe134341", OrderTotal: 95.12 }
    ]
}

ENDTEXT

loSer = CREATEOBJECT("wwJsonSerializer")
loCustomer = loSer.DeserializeJson(lcJson)
 
? loCustomer.LastName
? loCustomer.Entered
? loCustomer.Address.City
 
loOrder = loCustomer.Orders[1]
? loOrder.OrderTotal

The data is created in string format and embedded inside of a TEXTMERGE statement to produce a JSON string. The JSON string is then sent through a JSON deserializer which in turn produces a FoxPro object (created from EMPTY objects and Collections).

Note that you can also inject data into this structure because of the TEXTMERGE clause that allows merging FoxPro expressions. again this works well to essentially ‘script’ your object definition.

Caveats

I use this mostly for demos and other scenarios where I need to create some static data quickly and for that it works very well because I can control the content exactly with the values I push into the object.

However, keep in mind that JSON requires data to be encoded to a certain degree. Strings need to escape extended characters, quotes and a host of other things. If you’re using Web Connection you can use the JsonString() and JsonDate() functions in the wwUtils library.

TEXT TO lcJson TEXTMERGE NOSHOW
{   
    _id: "<< loMongo.GenerateId()" >>,
    FirstName: << JsonString(lcFirstName) >>,
    LastName: <<JsonString(lcLastName) >>,
    Entered: << JsonDate(DateTime()>>,
    …
}

Although this gets a little more verbose as well, it’s still easier to read than the pure FoxPro object syntax.

Visualize

Anyway – something to keep in mind. When you need more expressive syntax to construct objects – and especially complex objects – a JSON Serializer might just do the trick and provide you complex structure in a more visual way.