Create Chart from Native JavaScript

[istduncan]

Hello.
Has anyone found a way to trick the Chart wizard into creating a dataset using native JavaScript, along with some of the bonded values from a table? My table contains only detail rows, and I am trying to create a chart from the subcategories. I have the data, but it did not come in a usable format for the chart wizard to manipulate. Is there a way to build a table or format the data using Native JavaScript to create the pie chart?

[hvbtup]

I'm not using BIRT charts, but I think they just use BIRT datasets. So why don't you create a separate scripted dataset that does what you want?
In the layout, you'd use a hidden list before/above your chart, which is bound to your original JDBC dataset.
Then, either there or in the JDBC dataset you use JS events to fill a JS object structure that serves as the basis for the scripted dataset.
This hidden list item forces the JDBC dataset to execute and thus fills the data for your scripted dataset before the scripted dataset is used by the chart item.

[istduncan]

@hvbtup, are there specific JS functions I should use to build my dataset via script and present it in the chart wizard?

[hvbtup]

How you store your data is your decision. There's nothing special.

[istduncan]

I've built an object just using the following
const data = [ {column1: "data1", column2: 1}, {column1: "data2", column2: 2} ];

I need to know how it should be configured in the series and the category. I added what I created to JS's onCreate, hoping to call it from there, but it doesn't make a difference. Do you know any examples I could look at?

[hvbtup]

First of all, you probably want to use a report variable to store your data and later retrieve it in the scripted data set.
A report variable must be explicitily created in the BIRT designer (technically, this is not quite true).
You access it with vars["MyGlobalVariableName"].
In your trivial example, to store the values, you could use something like this:

vars["data"] = [ {column1: "data1", column2: 1}, {column1: "data2", column2: 2} ];

In a real world example, you would probably initialize it with an empty array and then use vars["data"].push(...); to add more rows.

Your scripted DS (or maybe more than one) would use vars["data"] to retrieve one row at a time.
You'll need an additional helper variable to store the current index (or maybe a closure, but I'm not sure that would work).
I usually create a report variable "XXXhelper" where XXX is the name of the scripted DS for this.
Initialize it in the open event of the scripted DS XXX:
vars["XXXhelper"] = { index: 0; };
In the fetch event, IIRC, one needs to return true or false to flag if a row was found or not.
It would look like this, for example:

var helper = vars["XXXhelper"];
var data = vars["data"];
if (helper.index < data.length) {
    var rec = data[helper.index++];
    row["column1"] = rec.column1;
    row["column2"] = rec.column2;
    return true;
} else {
    return false;
}

There's usually no need for logic in the close event.

I suggest you first use a table item bound to the scripted DS to demonstrate that it's working as expected, before you use the DS for a chart item.

[istduncan]

Thank you!!!!
I have one more small issue. I cannot pull in the column-bound variables from the other table. Is there a specific way they must be pulled?

[hvbtup]

You can either use the same technique to fill a different part of your JS object structure.
Or, if there is a master-detail relationship between the data sets, you would probably nest the two lists/tables and access the outer row with row._outer["OuterColumn"].

[istduncan]

The data is now coming from the Workday ERP, as shown in the screenshot. I thought there was, but I could be wrong. a way to reference another DS in the script?

[hvbtup]

Data Sets are fully independent from anything except their parameters.
Maybe you could try a Joint Data Set to combine data from two different Data Sets. Personally, I never needed them.

The layout is a tree structure. A layout item (node) can access the bound data along its branch to the tree root.
E.g. if I have this tree (the numbers are hypothetical item IDs):

body (1)
   list (2) bound to DS A
      detail
          table (3) bound to DS B
              detail row (4)
                  cell (5)
                      dynamic text item (6)

Then, in the expression (and the visibility expression, and onCreate event, filters etc) for the dynamic text item (6), you can access the data from DS B using row["columnXFromB"] and the data from DS A using row._outer["columnYFromA"]. You can also access "columns" that you manually add to the binding in (2) or (4).
row searches the branch upwards to the root until it finds an item with binding. _outer continues searching upwards until it finds the next item with binding.
So row._outer._outer would work if the whole branch would contain at least three items with binding.

One exception: In the dataset parameter binding, of course row cannot access this item's binding, because the item itself is still to be constructed, and you cannot use the data (which you search for) for searching this very same data.
E.g. in the DS parameter binding of (3), row means the current row from (2).

[istduncan]

Thank you again!!!
I took what you gave me and just created a variable for each value and added them to the custom DS.