Which array item failed deserialization? – System.Text.Json
A recipe for capturing the JSON of the particular array item that failed deserialization with System.Text.Json
.
When deserializing a response from an external, third-party API it’s fairly common to receive a “wrapper” DTO which consists of an array of “item” DTOs and possibly some additional properties. Depending on the API, the list of items could be large and the object graph of each item could be complex.
Assume you get well-formed JSON back from the API (so no mismatched curlies or anything like that) but deserialization of one the items in the array fails for some reason (perhaps an unsupported enum value or a string where a number should go). By default, you’ll still get a JsonException
back from System.Text.Json
with something along the lines of
The JSON value could not be converted to Some.Namespace.Containing.Dtos.EntityStatus. Path: $.status | LineNumber: 8 | BytePositionInLine: 25.
Not very helpful for production support / debugging. The approach I took to solving this was to introduce a custom JsonConverter
that captures the raw item JSON and attempts to run it through JsonSerializer.Deserialize<T>()
. If deserialization fails (i.e. a JsonException
is thrown) then we throw a custom exception (JsonItemDeserializationException
) containing the raw JSON that failed to deserialize along with the underlying JsonException
as the inner exception. Implementation of CapturingJsonConverter<T>
is given below:
public class CapturingJsonConverter<T> : JsonConverter<T>
{
public override T Read(ref Utf8JsonReader reader, Type typeToConvert, JsonSerializerOptions options)
{
// Capture the JSON from the reader so that we can include it in the thrown exception if deserialization fails.
if (!JsonDocument.TryParseValue(ref reader, out var document))
{
throw new JsonException();
}
var json = document.ToJson();
// Create a copy of the JsonSerializerOptions except with this converter removed
// to avoid infinite recursion when we deserialize
var clonedOptions = options.ShallowClone();
clonedOptions.Converters.Clear();
foreach (var converter in options.Converters.Where(c => c.GetType() != GetType()))
{
clonedOptions.Converters.Add(converter);
}
try
{
return JsonSerializer.Deserialize<T>(json, clonedOptions);
}
catch (JsonException ex)
{
throw new JsonItemDeserializationException(typeof(T), json, ex);
}
}
public override void Write(Utf8JsonWriter writer, T value, JsonSerializerOptions options)
{
throw new InvalidOperationException();
}
}
public class JsonItemDeserializationException : Exception
{
public JsonItemDeserializationException(Type itemType, string json, JsonException jsonException) : base($"Deserialization of {itemType.Name} JSON item failed", jsonException)
{
ItemType = itemType;
Json = json;
}
public Type ItemType { get; }
public string Json { get; }
public JsonException JsonException => InnerException as JsonException;
}
The code is all pretty self-explanatory - the only slightly tricky bit is we need to create a copy of the JsonSerializerOptions
with this converter removed. If we didn’t do that, we’d cause infinite recursion (because the call to Serializer.Deserialize
would end up using the same converter from the options).
To use the converter, simply add it to the options e.g.
private static JsonSerializerOptions BuildJsonSerializerOptions()
{
return new JsonSerializerOptions
{
Converters =
{
new JsonStringEnumConverter(),
new CapturingJsonConverter<MyEntityDto>()
}
};
}
And at the call site use JsonSerializer.Deserialize
as usual, but be prepared for both JsonException
and JsonItemDeserializationException
.
try
{
var wrapper = JsonSerializer.Deserialize<ResponseWrapperDto>(responseJson, serializerOptions);
return wrapper.Items;
}
catch (JsonItemDeserializationException ex)
{
Log
.ForContext("ItemJson", ex.Json)
.Error(ex, "Failed to deserialize item of type {ItemType}", ex.ItemType);
throw;
}
catch (JsonException)
{
// Malformed JSON? In practice would probably omit this catch block
throw;
}
The implementation makes use of a couple of extension methods. Those are given for completeness:
public static class JsonDocumentExtensions
{
public static string ToJson(this JsonDocument document)
{
using var stream = new MemoryStream();
using Utf8JsonWriter writer = new Utf8JsonWriter(stream, new JsonWriterOptions {Indented = true});
document.WriteTo(writer);
writer.Flush();
return Encoding.UTF8.GetString(stream.GetBuffer());
}
}
public static class JsonSerializerOptionsExtensions
{
public static JsonSerializerOptions ShallowClone(this JsonSerializerOptions options)
{
var cloned = new JsonSerializerOptions
{
AllowTrailingCommas = options.AllowTrailingCommas,
DefaultBufferSize = options.DefaultBufferSize,
Encoder = options.Encoder,
DictionaryKeyPolicy = options.DictionaryKeyPolicy,
IgnoreNullValues = options.IgnoreNullValues,
IgnoreReadOnlyProperties = options.IgnoreReadOnlyProperties,
MaxDepth = options.MaxDepth,
PropertyNamingPolicy = options.PropertyNamingPolicy,
PropertyNameCaseInsensitive = options.PropertyNameCaseInsensitive,
ReadCommentHandling = options.ReadCommentHandling,
WriteIndented = options.WriteIndented
};
foreach (var converter in options.Converters)
{
cloned.Converters.Add(converter);
}
return cloned;
}
}
It’s probably worth calling out not to use this in performance-critical code (by reading the Utf8JsonReader
into a JsonDocument
we’re no-doubt removing a bunch of the low-level-string handling memory optimizations).