Guid type causes KeyNotFoundException in calls to CreateDeserializerOnly<T> or CreateGenericDeserializerOnly

[sokhalsa]

When attempting to call AvroSerializer.CreateDeserializerOnly or AvroSerializer.CreateGenericDeserializerOnly with a writer schema for a type that contains a Guid, a KeyNotFoundException ("The given key was not present in the dictionary.") error is thrown in JsonSchemaBuilder.CreatePrimitiveTypeSchema because "System.Guid" is not a key in the PrimitiveRuntimeType dictionary.

The stack trace is:

at System.Collections.Generic.Dictionary2.get_Item(TKey key) at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.CreatePrimitiveTypeSchema(String type, Dictionary2 attributes) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 410
at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.ParsePrimitiveTypeFromString(String token) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 382
at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.Parse(JToken token, NamedSchema parent, Dictionary2 namedSchemas) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 105 at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.ParseRecordField(JObject field, NamedSchema parent, Dictionary2 namedSchemas, Int32 position) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 349
at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.<>c__DisplayClass3.b__2(JToken field, Int32 index) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 318
at System.Linq.Enumerable.d__72.MoveNext() at System.Collections.Generic.List1..ctor(IEnumerable1 collection) at System.Linq.Enumerable.ToList[TSource](IEnumerable1 source)
at Microsoft.Hadoop.Avro.Schema.JsonExtensions.ReadArrayProperty[T](JToken token, String propertyName, Boolean isOptional, Func3 func) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonExtensions.cs:line 195 at Microsoft.Hadoop.Avro.Schema.JsonExtensions.OptionalArrayProperty[T](JToken token, String propertyName, Func3 func) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonExtensions.cs:line 78
at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.ParseRecordType(JObject record, NamedSchema parent, Dictionary2 namedSchemas) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 309 at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.ParseJsonObject(JObject token, NamedSchema parent, Dictionary2 namedSchemas) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 140
at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.Parse(JToken token, NamedSchema parent, Dictionary`2 namedSchemas) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 88
at Microsoft.Hadoop.Avro.Schema.JsonSchemaBuilder.BuildSchema(String schema) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\Schema\JsonSchemaBuilder.cs:line 71
at Microsoft.Hadoop.Avro.AvroSerializer.CreateForCore[T](String writerSchema, AvroSerializerSettings settings) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\AvroSerializer.Static.cs:line 177
at Microsoft.Hadoop.Avro.AvroSerializer.CreateDeserializerOnly[T](String writerSchema, AvroSerializerSettings settings) in d:\Git\WindowsAzure\azure-sdk-for-net\src\HDInsight\Microsoft.Hadoop.Avro\AvroSerializer.Static.cs:line 98

[sokhalsa]

It looks like it has to do with multiple Guid members in a class causing JsonSchemaBuilder to the schema incorrectly.

Given the following class:

[System.Runtime.Serialization.DataContract]
    public class ClassOfGuid
    {
        [System.Runtime.Serialization.DataMember]
        public Guid PrimitiveGuid;

        [System.Runtime.Serialization.DataMember]
        public Guid AnotherPrimitiveGuid;
    }

A call to AvroSerializer.CreateDeserializerOnly will fail with a KeyNotFoundException. The following test case demonstrates that:

    [TestMethod]
    [TestCategory("CheckIn")]
    public void JsonSchemaBuilder_ClassOfGuidWriterSchema()
    {
        var settings = new Microsoft.Hadoop.Avro.AvroSerializerSettings();
        Microsoft.Hadoop.Avro.IAvroSerializer<ClassOfGuid> serializer = Microsoft.Hadoop.Avro.AvroSerializer.Create<ClassOfGuid>(settings);
        Microsoft.Hadoop.Avro.AvroSerializer.CreateDeserializerOnly<ClassOfGuid>(serializer.WriterSchema.ToString(), settings);
    }

[markcowl]

@sokhalsa Thank you for the feedback. Assigned to the appropriate team

[jholovacs]

Was anything ever done with this? It's still open 2 years later.

[shahabhijeet]

please reopen if this is still an issue.

[IvanDeev]

Looks like it wasn't fixed. Reproduces in Microsoft.Hadoop.Avro2 1.2.1 and microsoft.hadoop.avro-core 1.1.19. And I don't see Microsoft.Hadoop.Avro in azure-sdk-for-net now.