Deserialize json array stream one item at a time

In order to read the JSON incrementally, you'll need to use a JsonTextReader in combination with a StreamReader. But, you don't necessarily have to read all the JSON manually from the reader. You should be able to leverage the Linq-To-JSON API to load each large object from the reader so that you can work with it more easily.

For a simple example, say I had a JSON file that looked like this:

[
  {
    "name": "foo",
    "id": 1
  },
  {
    "name": "bar",
    "id": 2
  },
  {
    "name": "baz",
    "id": 3
  }
]

Code to read it incrementally from the file might look something like the following. (In your case you would replace the FileStream with your response stream.)

using (FileStream fs = new FileStream(@"C:\temp\data.json", FileMode.Open, FileAccess.Read))
using (StreamReader sr = new StreamReader(fs))
using (JsonTextReader reader = new JsonTextReader(sr))
{
    while (reader.Read())
    {
        if (reader.TokenType == JsonToken.StartObject)
        {
            // Load each object from the stream and do something with it
            JObject obj = JObject.Load(reader);
            Console.WriteLine(obj["id"] + " - " + obj["name"]);
        }
    }
}

Output of the above would look like this:

1 - foo
2 - bar
3 - baz

I have simplified one of the samples/tests of my parser/deserializer to answer this question's use case more straightforwardly.

Here's for the test data:

https://github.com/ysharplanguage/FastJsonParser/tree/master/JsonTest/TestData

(cf. fathers.json.txt)

And here's for the sample code:

    using System;
    using System.Collections.Generic;
    using System.IO;
    using System.Linq;
    using System.Text;

    // Our stuff
    using System.Text.Json;

//...

    public class FathersData
    {
        public Father[] fathers { get; set; }
    }

    public class Someone
    {
        public string name { get; set; }
    }

    public class Father : Someone
    {
        public int id { get; set; }
        public bool married { get; set; }
        // Lists...
        public List<Son> sons { get; set; }
        // ... or arrays for collections, that's fine:
        public Daughter[] daughters { get; set; }
    }

    public class Child : Someone
    {
        public int age { get; set; }
    }

    public class Son : Child
    {
    }

    public class Daughter : Child
    {
        public string maidenName { get; set; }
    }

//...

    static void FilteredFatherStreamTestSimplified()
    {
        // Get our parser:
        var parser = new JsonParser();

        // (Note this will be invoked thanks to the "filters" dictionary below)
        Func<object, object> filteredFatherStreamCallback = obj =>
        {
            Father father = (obj as Father);
            // Output only the individual fathers that the filters decided to keep (i.e., when obj.Type equals typeof(Father)),
            // but don't output (even once) the resulting array (i.e., when obj.Type equals typeof(Father[])):
            if (father != null)
            {
                Console.WriteLine("\t\tId : {0}\t\tName : {1}", father.id, father.name);
            }
            // Do not project the filtered data in any specific way otherwise,
            // just return it deserialized as-is:
            return obj;
        };

        // Prepare our filter, and thus:
        // 1) we want only the last five (5) fathers (array index in the resulting "Father[]" >= 29,995),
        // (assuming we somehow have prior knowledge that the total count is 30,000)
        // and for each of them,
        // 2) we're interested in deserializing them with only their "id" and "name" properties
        var filters = 
            new Dictionary<Type, Func<Type, object, object, int, Func<object, object>>>
            {
                // We don't care about anything but these 2 properties:
                {
                    typeof(Father), // Note the type
                    (type, obj, key, index) =>
                        ((key as string) == "id" || (key as string) == "name") ?
                        filteredFatherStreamCallback :
                        JsonParser.Skip
                },
                // We want to pick only the last 5 fathers from the source:
                {
                    typeof(Father[]), // Note the type
                    (type, obj, key, index) =>
                        (index >= 29995) ?
                        filteredFatherStreamCallback :
                        JsonParser.Skip
                }
            };

        // Read, parse, and deserialize fathers.json.txt in a streamed fashion,
        // and using the above filters, along with the callback we've set up:
        using (var reader = new System.IO.StreamReader(FATHERS_TEST_FILE_PATH))
        {
            FathersData data = parser.Parse<FathersData>(reader, filters);

            System.Diagnostics.Debug.Assert
            (
                (data != null) &&
                (data.fathers != null) &&
                (data.fathers.Length == 5)
            );
            foreach (var i in Enumerable.Range(29995, 5))
                System.Diagnostics.Debug.Assert
                (
                    (data.fathers[i - 29995].id == i) &&
                    !String.IsNullOrEmpty(data.fathers[i - 29995].name)
                );
        }
        Console.ReadKey();
    }

The rest of the bits is available here:

https://github.com/ysharplanguage/FastJsonParser

'HTH,