It is possible. I just haven't had the time to answer yet. I will look at it in a few hours when I'm back at my computer ;)

Anyways, can you provide some further information about the editor? Eg. an example of the JSON your editor is generating? That would help me give a better answer.

That helps a bit :D

It is however worth mentioning that the way you're saving your data is a bit wrong. In the JSON you have added properties directly to the object representing the control, whereas they should ideally be added to the value object.

So rather than saving tabelsData to $scope.control.tabelsData, you should ideally save it to $scope.control.value.tabelsData.

Working with the existing editor / JSON value
My package has a converter system that lets you add new models for your custom editors. Since your data isn't saved as the value object, we can't utilize this system.

Instead we can work directly with the JObject's and JArray's representing the JSON. You can see this little example on how to read the different values:

using System;
using Newtonsoft.Json.Linq;
using Skybrud.Umbraco.GridData;
using Skybrud.Umbraco.GridData.Extensions.Json;

namespace ConsoleApplication1 {

    public class Hest {

        public static void Test(GridDataModel model) {

            foreach (GridControl control in model.GetAllControls("paragraphGroup")) {

                // Get the value of "mainHeading"
                string mainHeading = control.JObject.GetString("mainHeading");

                // Get the value of "heading"
                string heading = control.JObject.GetString("heading");

                // Write to the console (or add to index)
                if (!String.IsNullOrWhiteSpace(mainHeading)) { Console.WriteLine(mainHeading); }
                if (!String.IsNullOrWhiteSpace(heading)) { Console.WriteLine(heading); }

                // Get the array representing "tabelsData"
                JArray tabelsData = control.JObject.GetArray("tabelsData");

                // Make sure to check whether the array is null
                if (tabelsData != null) {

                    // Iterate through the objects in the array
                    foreach (JObject obj in tabelsData) {

                        // get the "content" property
                        string content = obj.GetString("content");

                        // Write to the console (or add to index)
                        if (!String.IsNullOrWhiteSpace(content)) { Console.WriteLine(content); }

                    }

                }

            }

        }

    }

}

My example simply writes the values to the console, but you could just as easily add them to the search index ;)

Using the build-in converter system

Creating a custom converter has a few steps, so it may take a bit more work to get implemented. A benefit is that it will make your code and views cleaner.

This also assumes that you have fixed your editor to save your data to $scope.control.value instead of just $scope.control.

First you need to implement a model representing your JSON data. I have done this for you with the ParagraphGroupControlValue and ParagraphGroupTableData classes. I haven't added all properties though.

ParagraphGroupControlValue.cs:

using Newtonsoft.Json.Linq;
using Skybrud.Umbraco.GridData;
using Skybrud.Umbraco.GridData.Extensions.Json;
using Skybrud.Umbraco.GridData.Interfaces;
using Skybrud.Umbraco.GridData.Json;

namespace ConsoleApplication1 {

    public class ParagraphGroupControlValue : GridJsonObject, IGridControlValue {

        #region Properties

        public GridControl Control { get; private set; }

        public string MainHeading { get; private set; }

        public string Heading { get; private set; }

        public ParagraphGroupTableData[] TableData { get; private set; }

        #endregion

        #region Constructors

        protected ParagraphGroupControlValue(JObject obj) : base(obj) { }

        #endregion

        #region Static methods

        public static ParagraphGroupControlValue Parse(GridControl control, JObject obj) {
            if (obj == null) return null;
            return new ParagraphGroupControlValue(obj) {
                Control = control,
                MainHeading = obj.GetString("mainHeading"),
                Heading = obj.GetString("heading"),
                TableData = obj.GetArray("tabelsData", ParagraphGroupTableData.Parse) ?? new ParagraphGroupTableData[0]
            };
        }

        #endregion

    }

}

ParagraphGroupTableData.cs:

using Newtonsoft.Json.Linq;
using Skybrud.Umbraco.GridData.Extensions.Json;
using Skybrud.Umbraco.GridData.Json;

namespace ConsoleApplication1 {

    public class ParagraphGroupTableData : GridJsonObject {

        #region Properties

        public string Heading { get; private set; }
        public string ImagePath { get; private set; }
        public string Id { get; private set; }
        public string Content { get; private set; }
        public bool RevertOrder { get; private set; }

        #endregion

        #region Constructors

        protected ParagraphGroupTableData(JObject obj) : base(obj) { }

        #endregion

        #region Static methods

        public static ParagraphGroupTableData Parse(JObject obj) {
            if (obj == null) return null;
            return new ParagraphGroupTableData(obj) {
                Heading = obj.GetString("heading"),
                ImagePath = obj.GetString("imagePath"),
                Id = obj.GetString("id"),
                Content = obj.GetString("content"),
                RevertOrder = obj.GetBoolean("revertOrder")
            };
        }

        #endregion

    }

}

The first class represents the object stored in the value property, while the second class represents each object in the tabelsData array.

The next step is to implement a custom converter - this is done by implementing the IGridConverter interface. For your editor, this could look like:

using Newtonsoft.Json.Linq;
using Skybrud.Umbraco.GridData;
using Skybrud.Umbraco.GridData.Extensions.Json;
using Skybrud.Umbraco.GridData.Interfaces;
using Skybrud.Umbraco.GridData.Rendering;

namespace ConsoleApplication1 {

    public class ParagraphGroupGridConverter : IGridConverter {

        public virtual bool ConvertControlValue(GridControl control, JToken token, out IGridControlValue value) {

            value = null;

            switch (control.Editor.Alias) {

                case "paragraphGroup":
                    value = ParagraphGroupControlValue.Parse(control, token as JObject);
                    break;

            }

            return value != null;

        }

        public virtual bool ConvertEditorConfig(GridEditor editor, JToken token, out IGridEditorConfig config) {
            config = null;
            return false;
        }

        public virtual bool GetControlWrapper(GridControl control, out GridControlWrapper wrapper) {

            wrapper = null;

            switch (control.Editor.Alias) {

                case "paragraphGroup":
                    wrapper = control.GetControlWrapper<ParagraphGroupControlValue>();
                    break;

            }

            return wrapper != null;

        }

    }

}

To quickly summarize the class, the ConvertControlValue method makes sure the value is properly converted - eg. so that you can call control.GetValue<ParagraphGroupControlValue>().

The ConvertEditorConfig method convert the configuration of the editor. Not sure whether you're using the configuration for anything, the converter does nothing at that point.

Finally the GetControlWrapper method tells my package how it should initialize a wrapper for the class. The wrapper is something that can be used in your views, so it isn't really necessary for indexing.

Since the both the model and the converter are now in place, the last step is to tell my package about the converter. The converted should be added during Umbraco startup, which can be done similar to:

using Skybrud.Umbraco.GridData;
using Umbraco.Core;

namespace ConsoleApplication1 {

    public class Startup : IApplicationEventHandler {

        public void OnApplicationStarted(UmbracoApplicationBase umbraco, ApplicationContext context) {
            GridContext.Current.Converters.Add(new ParagraphGroupGridConverter());
        }

        public void OnApplicationStarting(UmbracoApplicationBase umbraco, ApplicationContext context) { }

        public void OnApplicationInitialized(UmbracoApplicationBase umbraco, ApplicationContext context) { }

    }

}

So rather than the first code example in the top of this post, the same class could now look like:

using System;
using Skybrud.Umbraco.GridData;

namespace ConsoleApplication1 {

    public class Hest {

        public static void Test(GridDataModel model) {

            foreach (GridControl control in model.GetAllControls("paragraphGroup")) {

                // Get the strongly typed value
                ParagraphGroupControlValue value = control.GetValue<ParagraphGroupControlValue>();

                // Write to the console (or add to index)
                if (!String.IsNullOrWhiteSpace(value.MainHeading)) { Console.WriteLine(value.MainHeading); }
                if (!String.IsNullOrWhiteSpace(value.Heading)) { Console.WriteLine(value.Heading); }

                // Iterate through the table data in the array
                foreach (ParagraphGroupTableData data in value.TableData) {

                    // Write to the console (or add to index)
                    if (!String.IsNullOrWhiteSpace(data.Content)) { Console.WriteLine(data.Content); }

                }

            }

        }

    }

}

I know that working with the grid can be a bit complex, so let me know if you have any further questions ;)

1. The way I understand you, you have some kind of grid picker on your regular pages? Then when these pages are being indexed, you could make lookups in the content cache for the selected grid pages. How to accomplish this depends a lot on your solution and current code, so I might not be able to help much with this one.

2. If you have a look at my example, I'm adding the combined search value fro the grid into a new field e.Fields["content"] = combined.ToString();. So to add a new field, you could jsut do something like e.Fields["myNewField"] = "Hello World";.

This is a bit out of the scope of my package, and again, it depends a lot on your solution.

If the addresses are static (hardcoded?) you can just add new field(s) for these when the relevant nodes are being indexed.