Add nullable reference type support to schema builder

docs/cli/index.js

@@ -7,18 +7,6 @@ const clrTypeOptions = [{
   name: 'assembly',
   required: false,
   summary: 'The name of or path to an assembly to load (multiple space-separated values accepted).'
-}, {
-  name: 'enums-as-ints',
-  required: false,
-  summary: 'Whether enums should be represented as integers.'
-}, {
-  name: 'nullable-references',
-  required: false,
-  summary: 'Whether reference types should be nullable.'
-}, {
-  name: 'temporal-behavior',
-  required: false,
-  summary: 'Whether timestamps should be represented with "string" schemas (ISO 8601) or "long" schemas (timestamp logical types). Options are iso8601, epochmilliseconds, and epochmicroseconds.'
 }, {
   abbreviation: 't',
   name: 'type',
@@ -66,7 +54,19 @@ module.exports = [{
     body: `$ dotnet avro create --assembly ./out/Example.Models.dll --type Example.Models.ExampleModel
 {"name":"Example.Models.ExampleModel",type":"record",fields:[{"name":"Text","type":"string"}]}`
   }],
-  options: [...clrTypeOptions]
+  options: [...clrTypeOptions, {
+    name: 'enums-as-ints',
+    required: false,
+    summary: 'Whether enums should be represented as integers.'
+  }, {
+    name: 'nullable-references',
+    required: false,
+    summary: 'Which reference types should be represented with nullable union schemas. Options are annotated (use nullable annotations if available), none, and all.'
+  }, {
+    name: 'temporal-behavior',
+    required: false,
+    summary: 'Whether timestamps should be represented with "string" schemas (ISO 8601) or "long" schemas (timestamp logical types). Options are iso8601, epochmilliseconds, and epochmicroseconds.'
+  }]
 }, {
   name: 'generate',
   summary: 'Generates C# code for a schema from the Schema Registry.',
@@ -98,7 +98,11 @@ namespace Example.Models
     body: `PS C:\\> Get-Content .\\example-model.avsc | dotnet avro generate | Out-File .\\ExampleModel.cs`,
     language: 'powershell'
   }],
-  options: [...schemaResolutionOptions]
+  options: [...schemaResolutionOptions, {
+    name: 'nullable-references',
+    required: false,
+    summary: 'Whether reference types selected for nullable record fields should be annotated as nullable.'
+  }]
 }, {
   name: 'registry-get',
   summary: 'Retrieve a schema from the Schema Registry.',

src/Chr.Avro.Cli/Cli/CreateSchemaVerb.cs

@@ -36,8 +36,8 @@ public class CreateSchemaVerb : Verb, IClrTypeOptions
         [Option("enums-as-integers", HelpText = "Whether enums should be represented with \"int\" or \"long\" schemas.")]
         public bool EnumsAsIntegers { get; set; }
 
-        [Option("nullable-references", HelpText = "Whether reference types should be represented with nullable union schemas.")]
-        public bool NullableReferences { get; set; }
+        [Option("nullable-references", HelpText = "Which reference types should be represented with nullable union schemas. Options are annotated (use nullable annotations if available), none, and all.", Default = NullableReferenceTypeBehavior.Annotated)]
+        public NullableReferenceTypeBehavior NullableReferences { get; set; }
 
         [Option("temporal-behavior", HelpText = "Whether timestamps should be represented with \"string\" schemas (ISO 8601) or \"long\" schemas (timestamp logical types). Options are iso8601, epochmilliseconds, and epochmicroseconds.")]
         public TemporalBehavior TemporalBehavior { get; set; }
@@ -61,7 +61,7 @@ protected Schema CreateSchema()
 
             var builder = new SchemaBuilder(
                 enumBehavior: EnumsAsIntegers ? EnumBehavior.Integral : EnumBehavior.Symbolic,
-                nullableReferenceTypeBehavior: NullableReferences ? NullableReferenceTypeBehavior.All : NullableReferenceTypeBehavior.None,
+                nullableReferenceTypeBehavior: NullableReferences,
                 temporalBehavior: TemporalBehavior);
 
             try

src/Chr.Avro.Cli/Cli/GenerateCodeVerb.cs

@@ -46,12 +46,16 @@ public class GenerateCodeVerb : Verb, ISchemaResolutionOptions
         [Option('s', "subject", SetName = BySubjectSet, HelpText = "If an ID is not specified, the subject of the schema.")]
         public string SchemaSubject { get; set; }
 
+        [Option("nullable-references", HelpText = "Whether reference types selected for nullable record fields should be annotated as nullable.")]
+        public bool NullableReferences { get; set; }
+
         [Option('v', "version", SetName = BySubjectSet, HelpText = "The version of the schema.")]
         public int? SchemaVersion { get; set; }
 
         protected override async Task Run()
         {
-            var generator = new CSharpCodeGenerator();
+            var generator = new CSharpCodeGenerator(
+                enableNullableReferenceTypes: NullableReferences);
             var reader = new JsonSchemaReader();
             var schema = reader.Read(await ((ISchemaResolutionOptions)this).ResolveSchema());
 

src/Chr.Avro.Codegen/Codegen/CSharpCodeGenerator.cs

@@ -16,6 +16,20 @@ namespace Chr.Avro.Codegen
     /// </summary>
     public class CSharpCodeGenerator : ICodeGenerator
     {
+        private readonly bool enableNullableReferenceTypes;
+
+        /// <summary>
+        /// Initializes a new instance of the <see cref="CSharpCodeGenerator" /> class.
+        /// </summary>
+        /// <param name="enableNullableReferenceTypes">
+        /// Whether reference types selected for nullable record fields should be annotated as
+        /// nullable.
+        /// </param>
+        public CSharpCodeGenerator(bool enableNullableReferenceTypes = true)
+        {
+            this.enableNullableReferenceTypes = enableNullableReferenceTypes;
+        }
+
         /// <summary>
         /// Generates a class declaration for a record schema.
         /// </summary>
@@ -304,7 +318,7 @@ protected virtual TypeSyntax GetPropertyType(Schema schema, bool nullable = fals
                     throw new UnsupportedSchemaException(schema, $"{schema.GetType()} is not recognized by the code generator.");
             }
 
-            if (nullable && value)
+            if (nullable && (enableNullableReferenceTypes || value))
             {
                 type = SyntaxFactory.NullableType(type);
             }

src/Chr.Avro.Fixtures/Fixtures/NullableMemberClass.cs

@@ -0,0 +1,56 @@
+#pragma warning disable CS8618
+
+namespace Chr.Avro.Fixtures
+{
+    using System;
+    using System.Collections.Generic;
+
+    public class NullableMemberClass
+    {
+        public Guid ObliviousGuidProperty { get; set; }
+
+        public Guid? ObliviousNullableGuidProperty { get; set; }
+
+        public string ObliviousStringProperty { get; set; }
+
+        public string[] ObliviousArrayOfStringsProperty { get; set; }
+
+        public List<string> ObliviousListOfStringsProperty { get; set; }
+
+        public Dictionary<string, object> ObliviousDictionaryOfObjectsProperty { get; set; }
+
+#nullable enable
+        public Guid GuidProperty { get; set; }
+
+        public Guid? NullableGuidProperty { get; set; }
+
+        public string StringProperty { get; set; }
+
+        public string? NullableStringProperty { get; set; }
+
+        public string[] ArrayOfStringsProperty { get; set; }
+
+        public string?[] ArrayOfNullableStringsProperty { get; set; }
+
+        public string[]? NullableArrayOfStringsProperty { get; set; }
+
+        public string?[]? NullableArrayOfNullableStringsProperty { get; set; }
+
+        public List<string> ListOfStringsProperty { get; set; }
+
+        public List<string?> ListOfNullableStringsProperty { get; set; }
+
+        public List<string>? NullableListOfStringsProperty { get; set; }
+
+        public List<string?>? NullableListOfNullableStringsProperty { get; set; }
+
+        public Dictionary<string, object> DictionaryOfObjectsProperty { get; set; }
+
+        public Dictionary<string, object?> DictionaryOfNullableObjectsProperty { get; set; }
+
+        public Dictionary<string, object>? NullableDictionaryOfObjectsProperty { get; set; }
+
+        public Dictionary<string, object?>? NullableDictionaryOfNullableObjectsProperty { get; set; }
+#nullable disable
+    }
+}

src/Chr.Avro/Abstract/NullableReferenceTypeBehavior.cs

@@ -14,5 +14,11 @@ public enum NullableReferenceTypeBehavior
         /// Match .NET’s nullable semantics, assuming reference types are always nullable.
         /// </summary>
         All,
+
+        /// <summary>
+        /// Inspect nullable reference type metadata to infer nullability. For types where metadata
+        /// is not present, behavior will be identical to <see cref="None" />.
+        /// </summary>
+        Annotated,
     }
 }

src/Chr.Avro/Abstract/RecordSchemaBuilderCase.cs

@@ -12,6 +12,8 @@ namespace Chr.Avro.Abstract
     /// </summary>
     public class RecordSchemaBuilderCase : SchemaBuilderCase, ISchemaBuilderCase
     {
+        private readonly NullabilityInfoContext nullabilityContext;
+
         /// <summary>
         /// Initializes a new instance of the <see cref="RecordSchemaBuilderCase" /> class.
         /// </summary>
@@ -32,6 +34,8 @@ public RecordSchemaBuilderCase(
             MemberVisibility = memberVisibility;
             NullableReferenceTypeBehavior = nullableReferenceTypeBehavior;
             SchemaBuilder = schemaBuilder ?? throw new ArgumentNullException(nameof(schemaBuilder), "Schema builder cannot be null.");
+
+            nullabilityContext = new NullabilityInfoContext();
         }
 
         /// <summary>
@@ -73,7 +77,7 @@ public virtual SchemaBuilderCaseResult BuildSchema(Type type, SchemaBuilderConte
 
                 if (!type.IsValueType && NullableReferenceTypeBehavior == NullableReferenceTypeBehavior.All)
                 {
-                    schema = new UnionSchema(new[] { new NullSchema(), schema });
+                    schema = MakeNullableSchema(schema);
                 }
 
                 try
@@ -89,8 +93,8 @@ public virtual SchemaBuilderCaseResult BuildSchema(Type type, SchemaBuilderConte
                 {
                     var memberType = member switch
                     {
-                        FieldInfo fieldMember => fieldMember.FieldType,
-                        PropertyInfo propertyMember => propertyMember.PropertyType,
+                        FieldInfo fieldInfo => fieldInfo.FieldType,
+                        PropertyInfo propertyInfo => propertyInfo.PropertyType,
                         _ => default,
                     };
 
@@ -101,6 +105,21 @@ public virtual SchemaBuilderCaseResult BuildSchema(Type type, SchemaBuilderConte
 
                     var field = new RecordField(member.Name, SchemaBuilder.BuildSchema(memberType, context));
 
+                    if (NullableReferenceTypeBehavior == NullableReferenceTypeBehavior.Annotated)
+                    {
+                        var nullabilityInfo = member switch
+                        {
+                            FieldInfo fieldInfo => nullabilityContext.Create(fieldInfo),
+                            PropertyInfo propertyInfo => nullabilityContext.Create(propertyInfo),
+                            _ => default,
+                        };
+
+                        if (nullabilityInfo != null)
+                        {
+                            field.Type = ApplyNullabilityInfo(field.Type, nullabilityInfo);
+                        }
+                    }
+
                     if (member.GetAttribute<DefaultValueAttribute>() is DefaultValueAttribute defaultAttribute)
                     {
                         field.Default = new ObjectDefaultValue<object>(defaultAttribute.Value, field.Type);
@@ -116,5 +135,127 @@ public virtual SchemaBuilderCaseResult BuildSchema(Type type, SchemaBuilderConte
                 return SchemaBuilderCaseResult.FromException(new UnsupportedTypeException(type, $"{nameof(RecordSchemaBuilderCase)} cannot be applied to array or primitive types."));
             }
         }
+
+        /// <summary>
+        /// Makes a schema and any of its children nullable based on a type member's nullability.
+        /// </summary>
+        /// <param name="schema">
+        /// A <see cref="Schema" /> object to apply nullability info to.
+        /// </param>
+        /// <param name="nullabilityInfo">
+        /// A <see cref="NullabilityInfo" /> object for the member that <paramref name="schema" />
+        /// represents.
+        /// </param>
+        private static Schema ApplyNullabilityInfo(Schema schema, NullabilityInfo nullabilityInfo)
+        {
+            if (schema is ArraySchema arraySchema)
+            {
+                if (nullabilityInfo.ElementType != null)
+                {
+                    // if the type is an array, this is easy; recurse with the element type info:
+                    schema = new ArraySchema(ApplyNullabilityInfo(
+                        arraySchema.Item,
+                        nullabilityInfo.ElementType))
+                    {
+                        LogicalType = schema.LogicalType,
+                    };
+                }
+                else
+                {
+                    // otherwise, if the type is generic, try to map one of its type arguments to
+                    // the IEnumerable<T> type argument and recurse with that argument type info:
+                    var genericType = nullabilityInfo.Type
+                        .GetGenericTypeDefinition();
+
+                    if (genericType.GetEnumerableType() is Type genericItemType)
+                    {
+                        var infoIndex = genericType
+                            .GetGenericArguments()
+                            .ToList()
+                            .FindIndex(candidate => candidate == genericItemType);
+
+                        if (infoIndex >= 0)
+                        {
+                            schema = new ArraySchema(ApplyNullabilityInfo(
+                                arraySchema.Item,
+                                nullabilityInfo.GenericTypeArguments[infoIndex]));
+                        }
+                    }
+                }
+            }
+            else if (schema is MapSchema mapSchema)
+            {
+                // if the type is generic, use the same trick as for IEnumerable<T>:
+                var genericType = nullabilityInfo.Type
+                    .GetGenericTypeDefinition();
+
+                if (genericType.GetDictionaryTypes() is (_, Type genericValueType))
+                {
+                    var infoIndex = genericType
+                        .GetGenericArguments()
+                        .ToList()
+                        .FindIndex(candidate => candidate == genericValueType);
+
+                    if (infoIndex >= 0)
+                    {
+                        schema = new MapSchema(ApplyNullabilityInfo(
+                            mapSchema.Value,
+                            nullabilityInfo.GenericTypeArguments[infoIndex]));
+                    }
+                }
+            }
+
+            // check the top level last (this also handles UnionSchema, which is the only other
+            // schema type that can have children):
+            if (nullabilityInfo.ReadState == NullabilityState.Nullable)
+            {
+                schema = MakeNullableSchema(schema);
+            }
+
+            return schema;
+        }
+
+        /// <summary>
+        /// Ensures that a schema is nullable (a <see cref="UnionSchema" /> containing a
+        /// <see cref="NullSchema" />).
+        /// </summary>
+        /// <param name="schema">
+        /// A schema to make nullable if not already.
+        /// </param>
+        /// <returns>
+        /// A <see cref="UnionSchema" /> containing a <see cref="NullSchema" />. If
+        /// <paramref name="schema" /> is already a <see cref="UnionSchema" /> containing a
+        /// <see cref="NullSchema" />, it will be returned as is. If <paramref name="schema" />
+        /// is a <see cref="UnionSchema" /> but does not contain a <see cref="NullSchema" />, a new
+        /// <see cref="UnionSchema" /> will be returned with a new <see cref="NullSchema" /> as the
+        /// first member. If <paramref name="schema" /> is not a <see cref="UnionSchema" />, a new
+        /// <see cref="UnionSchema" /> will be returned comprising a <see cref="NullSchema" /> and
+        /// <paramref name="schema" />.
+        /// </returns>
+        private static Schema MakeNullableSchema(Schema schema)
+        {
+            if (schema is UnionSchema unionSchema)
+            {
+                // if a null schema is already present, all good:
+                if (unionSchema.Schemas.Any(schema => schema is NullSchema))
+                {
+                    return schema;
+                }
+                else
+                {
+                    // ordinarily, null would come first in a union, but since default values for
+                    // record fields correspond to the first schema in the union, we append to
+                    // avoid invalidating any default values:
+                    return new UnionSchema(unionSchema.Schemas.Concat(new[] { new NullSchema() }))
+                    {
+                        LogicalType = unionSchema.LogicalType,
+                    };
+                }
+            }
+            else
+            {
+                return new UnionSchema(new[] { new NullSchema(), schema });
+            }
+        }
     }
 }