Update serialization code to use new signatures

Packages.Data.props

@@ -26,7 +26,7 @@
   <!-- All csproj dependencies. -->
   <ItemGroup>
     <PackageReference Update="NUnit" Version="3.13.2" />
-    <PackageReference Update="System.Net.ClientModel" Version="1.0.0-beta.1" />
+    <PackageReference Update="System.ClientModel" Version="1.0.0-alpha.20231214.13" />
     <PackageReference Update="Azure.Core" Version="1.36.0" />
     <PackageReference Update="Azure.Core.Experimental" Version="0.1.0-preview.18" />
     <PackageReference Update="Azure.ResourceManager" Version="1.9.0" />

eng/DownloadSharedSource.ps1

@@ -36,7 +36,7 @@ $files = @('AsyncLockWithValue.cs', 'CallerShouldAuditAttribute.cs', 'ClientDiag
     'OperationInternalBase.cs', 'OperationInternal.cs', 'OperationInternalOfT.cs', 'TaskExtensions.cs', 'Argument.cs', 'Multipart/MultipartFormDataContent.cs',
     'Multipart/MultipartContent.cs', 'AzureKeyCredentialPolicy.cs', 'AppContextSwitchHelper.cs',
     'OperationPoller.cs', 'FixedDelayWithNoJitterStrategy.cs', 'SequentialDelayStrategy.cs',
-    'ForwardsClientCallsAttribute.cs', 'AsyncLockWithValue.cs', 'VoidValue.cs', 'AzureResourceProviderNamespaceAttribute.cs',
+    'ForwardsClientCallsAttribute.cs', 'VoidValue.cs', 'AzureResourceProviderNamespaceAttribute.cs',
     'ChangeTrackingDictionary.cs',
     'ChangeTrackingList.cs',
     'FormUrlEncodedContent.cs',

eng/Generate.ps1

@@ -65,7 +65,8 @@ function Add-TestServer-Swagger ([string]$testName, [string]$projectSuffix, [str
     }
     $inputFile = Join-Path $testServerSwaggerPath "$testName.json"
     $inputReadme = Join-Path $projectDirectory "readme.md"
-    Add-Swagger "$testName$projectSuffix" $projectDirectory "--require=$configurationPath --try-require=$inputReadme --input-file=$inputFile $additionalArgs --clear-output-folder=true"
+    # TODO -- remove the flag when the feature is generally available
+    Add-Swagger "$testName$projectSuffix" $projectDirectory "--require=$configurationPath --try-require=$inputReadme --input-file=$inputFile $additionalArgs --clear-output-folder=true --use-model-reader-writer=true"
 }
 
 function Add-CadlRanch-TypeSpec([string]$testName, [string]$projectPrefix, [string]$cadlRanchProjectsDirectory, [string]$outputProjectDir="") {

src/AutoRest.CSharp/AutoRest.CSharp.csproj

@@ -28,7 +28,7 @@
     <PackageReference Include="nunit" />
     <!-- Any stable version of System.Memory should work here. -->
     <PackageReference Include="System.Memory" />
-    <PackageReference Include="System.Net.ClientModel" />
+    <PackageReference Include="System.ClientModel" />
   </ItemGroup>
 
   <!-- Enable SourceLink  -->

src/AutoRest.CSharp/Common/AutoRest/Plugins/CSharpProj.cs

@@ -34,6 +34,7 @@ internal class CSharpProj : IPlugin
         private string _coreCsProjContent = @"
   <ItemGroup>
     <PackageReference Include=""Azure.Core"" />
+    <PackageReference Include=""System.ClientModel"" />
   </ItemGroup>";
 
         private string _armCsProjContent = @"

src/AutoRest.CSharp/Common/AutoRest/Plugins/GeneratedCodeWorkspace.cs

@@ -2,12 +2,12 @@
 // Licensed under the MIT License. See License.txt in the project root for license information.
 
 using System;
+using System.ClientModel;
 using System.Collections.Generic;
 using System.Collections.Immutable;
 using System.Diagnostics;
 using System.IO;
 using System.Linq;
-using System.Net.ClientModel;
 using System.Threading;
 using System.Threading.Tasks;
 using AutoRest.CSharp.Common.AutoRest.Plugins;

src/AutoRest.CSharp/Common/AutoRest/Plugins/NewProjectScaffolding.cs

@@ -305,6 +305,7 @@ private string GetSrcCsproj()
 
   <ItemGroup>
     <PackageReference Include=""Azure.Core"" />
+    <PackageReference Include=""System.ClientModel"" />
     <PackageReference Include=""System.Text.Json"" />
   </ItemGroup>
 
@@ -322,7 +323,7 @@ private string GetSrcCsproj()
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include=""System.Net.ClientModel"" Version=""1.0.0-beta.1"" />
+    <PackageReference Include=""System.ClientModel"" Version=""1.0.0-alpha.20231116.3"" />
     <PackageReference Include=""System.Text.Json"" Version=""4.7.2"" />
   </ItemGroup>
 

src/AutoRest.CSharp/Common/Decorator/DefaultDerivedSchema.cs

@@ -4,15 +4,14 @@
 using System;
 using System.Collections.Generic;
 using System.Linq;
-using System.Text;
+using AutoRest.CSharp.Common.Input;
 using AutoRest.CSharp.Input;
-using AutoRest.CSharp.Output.Models.Types;
 
 namespace AutoRest.CSharp.Common.Decorator
 {
     internal static class DefaultDerivedSchema
     {
-        public const string DefaultDerivedExtension = "x-ms-autorest-defaultDerivedSchema";
+        private const string DefaultDerivedExtension = "x-ms-autorest-defaultDerivedSchema";
 
         public static void AddDefaultDerivedSchemas(CodeModel codeModel)
         {
@@ -52,8 +51,9 @@ private static void BuildInternalDefaultDerivedType(ObjectSchema schema, Diction
                 throw new InvalidOperationException($"Found a child poly {schema.Language.Default.Name} that we weren't able to determine its base poly from {string.Join(',', schema.Parents?.Immediate.Select(p => p.Name) ?? Array.Empty<string>())}");
 
             //Since the unknown type is used for deserialization only we don't need to create if its an input only model
+            // TODO -- remove this condition completely when remove the UseModelReaderWriter flag
             var hasXCsharpUsageOutput = !actualBaseSchema.Extensions?.Usage?.Contains("output", StringComparison.OrdinalIgnoreCase);
-            if (!actualBaseSchema.Usage.Contains(SchemaContext.Output) &&
+            if (!Configuration.UseModelReaderWriter && !actualBaseSchema.Usage.Contains(SchemaContext.Output) &&
                 !actualBaseSchema.Usage.Contains(SchemaContext.Exception) &&
                 (!hasXCsharpUsageOutput.HasValue ||
                 hasXCsharpUsageOutput.Value))
@@ -96,6 +96,7 @@ private static void BuildInternalDefaultDerivedType(ObjectSchema schema, Diction
                         },
                         DiscriminatorValue = "Unknown",
                         SerializationFormats = { KnownMediaType.Json },
+                        IsUnknownDiscriminatorModel = true
                     };
 
                     if (actualBaseSchema.Parents is not null)

src/AutoRest.CSharp/Common/Generation/Writers/CodeWriter.cs

@@ -8,6 +8,7 @@
 using System.Linq;
 using System.Text;
 using AutoRest.CSharp.Common.Input;
+using AutoRest.CSharp.Common.Output.Expressions.ValueExpressions;
 using AutoRest.CSharp.Generation.Types;
 using AutoRest.CSharp.Utilities;
 using Microsoft.CodeAnalysis.CSharp;
@@ -175,6 +176,9 @@ public CodeWriter Append(FormattableString formattableString)
                     case CodeWriterDeclaration declaration:
                         Append(declaration);
                         break;
+                    case ValueExpression expression:
+                        this.WriteValueExpression(expression);
+                        break;
                     case var _ when isLiteralFormat:
                         Literal(argument);
                         break;

src/AutoRest.CSharp/Common/Generation/Writers/CodeWriterExtensions.Methods.cs

@@ -271,8 +271,12 @@ public static void WriteValueExpression(this CodeWriter writer, ValueExpression
             switch (expression)
             {
                 case CastExpression cast:
+                    // wrap the cast expression with parenthesis, so that it would not cause ambiguity for leading recursive calls
+                    // if the parenthesis are not needed, the roslyn reducer will remove it.
+                    writer.AppendRaw("(");
                     writer.Append($"({cast.Type})");
                     writer.WriteValueExpression(cast.Inner);
+                    writer.AppendRaw(")");
                     break;
                 case CollectionInitializerExpression(var items):
                     writer.AppendRaw("{ ");
@@ -570,9 +574,12 @@ public static void WriteValueExpression(this CodeWriter writer, ValueExpression
                     writer.AppendRawIf(op, operandOnTheLeft);
                     break;
                 case BinaryOperatorExpression(var op, var left, var right):
+                    // we should always write parenthesis around this expression since some or the logic operator has lower priority, and we might get trouble when there is a chain of binary operator expression, for instance (a || b) && c.
+                    writer.AppendRaw("(");
                     writer.WriteValueExpression(left);
                     writer.AppendRaw(" ").AppendRaw(op).AppendRaw(" ");
                     writer.WriteValueExpression(right);
+                    writer.AppendRaw(")");
                     break;
                 case TernaryConditionalOperator ternary:
                     writer.WriteValueExpression(ternary.Condition);
@@ -585,8 +592,9 @@ public static void WriteValueExpression(this CodeWriter writer, ValueExpression
                     writer.Append($"{parameterName}: ");
                     writer.WriteValueExpression(value);
                     break;
-                case ParameterReference parameterReference:
-                    writer.Append($"{parameterReference.Parameter.Name:I}");
+                case ParameterReference(var parameter):
+                    writer.AppendRawIf("ref ", parameter.IsRef);
+                    writer.Append($"{parameter.Name:I}");
                     break;
                 case FormattableStringToExpression formattableString:
                     writer.Append(formattableString.Value);
@@ -639,6 +647,10 @@ public static void WriteValueExpression(this CodeWriter writer, ValueExpression
                     writer.WriteValueExpression(index);
                     writer.AppendRaw("]");
                     break;
+                case DeclarationExpression(var variable, var isOut):
+                    writer.AppendRawIf("out ", isOut);
+                    writer.Append($"{variable.Type} {variable.Declaration:D}");
+                    break;
             }
 
             static void WriteArguments(CodeWriter writer, IEnumerable<ValueExpression> arguments)

src/AutoRest.CSharp/Common/Generation/Writers/CodeWriterExtensions.cs

@@ -255,7 +255,7 @@ private static IDisposable WriteMethodDeclarationNoScope(this CodeWriter writer,
                     writer.AppendRaw(isBase ? ": base(" : ": this(");
                     foreach (var argument in arguments)
                     {
-                        writer.WriteValueExpression(new FormattableStringToExpression(argument));
+                        writer.WriteValueExpression(argument);
                         writer.AppendRaw(", ");
                     }
                     writer.RemoveTrailingComma();
@@ -321,6 +321,8 @@ public static void WriteParameter(this CodeWriter writer, Parameter clientParame
                 writer.AppendRaw("]");
             }
 
+            writer.AppendRawIf("ref ", clientParameter.IsRef);
+
             writer.Append($"{clientParameter.Type} {clientParameter.Name:D}");
             if (clientParameter.DefaultValue != null)
             {

src/AutoRest.CSharp/Common/Generation/Writers/LongRunningOperationWriter.cs

@@ -218,7 +218,8 @@ private static void WriteCreateResultBody(CodeWriter writer, LongRunningOperatio
                 var scopeName = operation.Diagnostics.ScopeName;
                 var nextLinkName = pagingResponse.NextLinkPropertyName;
                 var itemName = pagingResponse.ItemPropertyName;
-                FormattableString returnValue = $"{typeof(GeneratorPageableHelpers)}.{nameof(GeneratorPageableHelpers.CreateAsyncPageable)}({responseVariable}, _nextPageFunc, Product.DeserializeProduct, _clientDiagnostics, _pipeline, {scopeName:L}, {itemName:L}, {nextLinkName:L}, cancellationToken)";
+                // TODO -- why we have a hard-coded Product.DeserializeProduct here???
+                FormattableString returnValue = $"{typeof(GeneratorPageableHelpers)}.{nameof(GeneratorPageableHelpers.CreateAsyncPageable)}({responseVariable}, _nextPageFunc, e => Product.DeserializeProduct(e), _clientDiagnostics, _pipeline, {scopeName:L}, {itemName:L}, {nextLinkName:L}, cancellationToken)";
                 WriteCreateResultReturnValue(writer, resultType, returnValue, async);
             }
             else if (operation.ResultSerialization != null)

src/AutoRest.CSharp/Common/Generation/Writers/ModelWriter.cs

@@ -6,6 +6,7 @@
 using System.ComponentModel;
 using System.Globalization;
 using System.Linq;
+using AutoRest.CSharp.Common.Output.Models.Types;
 using AutoRest.CSharp.Generation.Types;
 using AutoRest.CSharp.Output.Models;
 using AutoRest.CSharp.Output.Models.Types;
@@ -69,6 +70,8 @@ private void WriteObjectSchema(CodeWriter writer, ObjectType schema)
                 writer.Line();
                 using (writer.Scope())
                 {
+                    WritePrivateRawDataField(writer, schema);
+
                     WriteConstructor(writer, schema);
 
                     WriteProperties(writer, schema);
@@ -78,15 +81,33 @@ private void WriteObjectSchema(CodeWriter writer, ObjectType schema)
 
         protected virtual void WriteProperties(CodeWriter writer, ObjectType schema)
         {
+            var rawDataField = (schema as SerializableObjectType)?.RawDataField;
             foreach (var property in schema.Properties)
             {
+                if (property == rawDataField)
+                    continue;
+
                 WriteProperty(writer, property);
 
                 if (property.FlattenedProperty != null)
                     WriteProperty(writer, property.FlattenedProperty);
             }
         }
 
+        // TODO -- this is workaround because we are declaring fields and properties in different way, and this raw data field and AdditionalProperties property could be converted from each other.
+        private void WritePrivateRawDataField(CodeWriter writer, ObjectType schema)
+        {
+            if ((schema as SerializableObjectType)?.RawDataField is not { } rawDataField)
+                return;
+
+            writer.WriteXmlDocumentationSummary($"{rawDataField.Description}");
+            writer.Append($"{rawDataField.Declaration.Accessibility} ")
+                .AppendRawIf("readonly ", schema.IsStruct)
+                .Line($"{rawDataField.Declaration.Type} {rawDataField.Declaration.Name};");
+
+            writer.Line();
+        }
+
         private void WriteFieldModifiers(CodeWriter writer, FieldModifiers modifiers)
         {
             writer.AppendRawIf("public ", modifiers.HasFlag(FieldModifiers.Public))

src/AutoRest.CSharp/Common/Generation/Writers/PageableMethodsWriterExtensions.cs

@@ -243,9 +243,13 @@ private static FormattableString GetValueFactory(CSharpType? pageItemType)
                 return $"e => {BinaryDataType}.{nameof(BinaryData.FromString)}(e.{nameof(JsonElement.GetRawText)}())";
             }
 
-            if (!pageItemType.IsFrameworkType && pageItemType.Implementation is SerializableObjectType { JsonSerialization: { }, IncludeDeserializer: true } type)
+            if (!pageItemType.IsFrameworkType && pageItemType.Implementation is SerializableObjectType { JsonSerialization: { } } type)
             {
-                return $"{type.Type}.Deserialize{type.Declaration.Name}";
+                // TODO -- we no longer need this once we remove the UseModelReaderWriter flag
+                if (Configuration.UseModelReaderWriter)
+                    return $"e => {type.Type}.Deserialize{type.Declaration.Name}(e)";
+                else
+                    return $"{type.Type}.Deserialize{type.Declaration.Name}";
             }
 
             var deserializeImplementation = JsonCodeWriterExtensions.GetDeserializeValueFormattable($"e", pageItemType);