This is the fourth page to this blog. Page 3 was a blog on how to parse the XML response. At the time, I didn't know how to parse json. Honestly, I really didn't understand JSON vs XML. Concurrent with this project, I needed to convert a FedEx Rate and Time request. A little background. Currently I have a function in Foxpro that gets from Fedex the rate for shipping a package. I use a SOAP request which is fairly simple. I run a web page on FedEx's web site passing in the URL everything about the shipment. It returns an XML stream that I parse. FedEx is discontinuing this service as of June 2026. Instead, they are requiring an OAUTH authentication and it is only in JSON. No XML. So I was forced to learn to parse JSON. To do this, I asked ChatGPT to explain the difference and give me a simple parsing routine to extract one value from the stream in Foxpro 6.0 code. In looking at this simple parsing routing, I understood the principles and I realized it would be very easy to provide an option to parse this project in JSON.

What I also realized is that each entity (In QBO, an entitiy is a type of transaction such as Invoice or Bill) would require a separate parsing routine as you will see.

Lets get started.  I called this program DBFIT1JSON.prg  It was my first attempt.
So this first piece of code is a revised version to the MAKECALL.PRG at the bottom.

 As you can see, different parsers are called depending on the entity being parsed.  But the good news is that they are easily graftable.

Here is the first part of the PRG to parse a request for a stream for the invoice and credit memo entities.  Note, this also includes the child records.  You are going to see a createobject function in this program.  This creates a problem.  If the com object fails to load, it will hang foxpro.  On the first call, it seems to open fine.  But we will make successive calls and what was happening is that on the second call, the com object was busy or non responsive and hung Foxpro so I moved this line of code back to the GETINVOICE.prg and run it only once instead of each successive parsing routine.  Once created, it can be reused over and over again so that worked.

 Now we are ready to loop through the stream

That is the JSON Parser.
Some notes.  It is important to be able to visualize what the json looks like with it's indentations.  The indentations tell you the depth needed.  This example doesn't get that deep.  I will create another one of these parsing routines for parsing a report stream.  One trick to visualizing it is to suspend the program at the beginning of the program.  Then do _cliptext=lcresponse.  This will place the entire stream into the clipboard.  Then go to Chatgpt and ask them to indent the JSON.  Make sure your lcresponse is just a few transactions.  Chatgpt will tell you the stream is too long.  Then copy and past this into word.  Change the font to courier (Fixed width font) so you can see the indent structure.  Another trick is to set step on in the program and try to watch some lines like loScript.Eval("data.QueryResponse."+mentity+"["+allt(STR(lnInv,3))+"].DocNumber")  Start with this loScript.Eval("data.QueryResponse.Invoice[0].DocNumber")  This should give you the first invoice #.  the [0] is the incrementer

Next note.  Unlike the XML parser, there is no need to create the parse.dbf and figure out the levels.  This is much more elegant and actually simpler by far.

Well that is it.  Extracting is one thing, writing is another.  Intuit only accepts JSON to write so the process is to construct a JSON.  My next project.
If you found this helpful, send me an email Craigs@sobelsoftware.com

Craig Sobel