We Keep Coding

Codegeneration at runtime from a string to console exe is not working in C# .NET6

I have some code which must be able to generated a console application at runtime (Codegeneration with System.CodeDom). I did this already a lot, but in NET 6 now I am struggling with that and the new API. In the code below I try to compile simply from a string. See below the static class with method Start() which then should generates the application.
The compilations seems fine, no errors at the end. But when starting the generated AppCodegenerated.exe, it shows some reference exception with System.Runtime.
Please help, any Idea? Already researched a lot but could not find any useful solution..
//-
I used the Visual Studio 2022 / NET 6 and theses Nuget's:
using Basic.Reference.Assemblies;
using Microsoft.CodeAnalysis;
using Microsoft.CodeAnalysis.CSharp;
using Microsoft.CodeAnalysis.Text;
using System.Text;
namespace CompilerSimplified
{
public static class Compiler
{
public static bool Start()
{
string FileName = "AppCodegenerated";
string ExePath = AppDomain.CurrentDomain.BaseDirectory + #"\" + FileName + ".exe";
string code = #"using System; Console.WriteLine(""Hello.""); Console.ReadLine(); ";
// ------- References -------------
// .net platform references
List<MetadataReference> References = new List<MetadataReference>();
foreach (var item in ReferenceAssemblies.Net60) // ReferenceAssemblies from Nuget: Basic.Reference.Assemblies;
References.Add(item);
// or tried this: loop manually through system platform
//string[] fileEntries = Directory.GetFiles(#"C:\Program Files\dotnet\packs\Microsoft.NETCore.App.Ref\6.0.0\ref\net6.0\", "*.dll");
//foreach (string fileName in fileEntries)
// references.Add(MetadataReference.CreateFromFile(fileName));MetadataReference.CreateFromFile(fileName));
// ------- References END -------------
// delete existing file
if (File.Exists(ExePath))
File.Delete(ExePath);
// compiler options
CSharpCompilationOptions DefaultCompilationOptions =
new CSharpCompilationOptions(outputKind: OutputKind.ConsoleApplication, platform: Platform.AnyCpu)
.WithOverflowChecks(true).WithOptimizationLevel(OptimizationLevel.Release);
// encode soucre code
string sourceCode = SourceText.From(code, Encoding.UTF8).ToString();
// CSharp options
var parsedSyntaxTree = Parse(sourceCode, "", CSharpParseOptions.Default.WithLanguageVersion(LanguageVersion.CSharp10));
// compilation
var compilation = CSharpCompilation.Create(FileName, new SyntaxTree[] { parsedSyntaxTree }, references: References, DefaultCompilationOptions);
var result = compilation.Emit(ExePath);
// return
if (result.Success)
return true;
else
return false;
}
private static SyntaxTree Parse(string text, string filename = "", CSharpParseOptions options = null)
{
var stringText = SourceText.From(text, Encoding.UTF8);
return SyntaxFactory.ParseSyntaxTree(stringText, options, filename);
}
}
}
Above code runs fine without error and exports the AppCodegenerated.exe into the project /bin folder.
Execution of this generated AppCodegenerated.exe shows following on the output console:
Unhandled exception: System.IO.FileNotFoundException:
The file or assembly "System.Runtime, Version = 6.0.0.0, Culture = neutral,
PublicKeyToken = b03f5f7f11d50a3a" or a dependency on it was not found.
The system can not find the stated file.

It is not possible to codegenerate directly a console application like the initial approach above. One possible solution is to generate first a dll (what I mentioned above in the example code is working fine), and from there include that .dll into a .exe, from where the functionality can run.

C# for scripting (csx) location of script file

In F# it's rather easy with predefined identifier __SOURCE_DIRECTORY__
https://stackoverflow.com/a/4861029/2583080
However this identifier does not work in C# scripting (csx files or C# Interactive).
> __SOURCE_DIRECTORY__
(1,1): error CS0103: The name '__SOURCE_DIRECTORY__' does not exist in the current context
Getting current directory in more traditional way will not work either.
Directory.GetCurrentDirectory()
Returns: C:\Users\$USER_NAME$\
new Uri(System.Reflection.Assembly.GetExecutingAssembly().CodeBase).LocalPath;
Returns: C:\Program Files (x86)\Microsoft Visual Studio\2017\Professional\Common7\IDE\CommonExtensions\Microsoft\ManagedLanguages\VBCSharp\InteractiveComponents\

In C# you can take advantage of caller information attributes (available since C# 5 / VS2012). Just declare a method like this:
string GetCurrentFileName([System.Runtime.CompilerServices.CallerFilePath] string fileName = null)
{
return fileName;
}
And call it without specifying the optional parameter:
string scriptPath = GetCurrentFileName(); // /path/to/your/script.csx

In csx, you are can add ExecutionContext as a parameter and access FunctionDirectory from it like so:
using System;
using Microsoft.Azure.WebJobs;
public static void Run(TimerInfo myTimer, ExecutionContext executionContext, ILogger log) {
var dir = executionContext.FunctionDirectory;
log.LogInformation($"Directory: {dir}");
}
ExecutionContext.FunctionDirectory will return the directory the contains the function's function.json file. It doesn't include the trailing .
At this time this seems to be the best documentation for ExecutionContext.
I am trying to find the answer to this question myself, and this was my previous answer.
In csx, the following helper method will return the directory "of the source file that contains the caller".
using System.IO;
...
public static string CallerDirectory([System.Runtime.CompilerServices.CallerFilePath] string fileName = null)
{
return Path.GetDirectoryName(fileName);
}
To call it, don't specify the fileName parameter.
var dir = CallerDirectory();

Alexa Skill Signature validation failing

I can't get the signature verification working, like it's described here. I'm using BouncyCastle.NetCore 1.8.1.3 and the project is a .NETCoreApp 1.0.
I'm developing on macOS 10.12.1, running dotnet core 1.0.4 and my server is running Ubuntu 16.04.2-x64 running the release version, build as netcore1.0 and ubuntu16.04-x64 app.
The rest of the system runs without problems, except the signature verification.
My validation always returns false.
Here is my service for validating the signature and body:
using System;
using System.IO;
using System.Text;
using System.Threading.Tasks;
namespace MySkill.Main.Services
{
public class CertificationValidationService : ICertificationValidationService
{
private const string Algorithm = "SHA1withRSA";
public async Task<bool> IsValidSiganture(Stream body, Stream certData, string signature)
{
var pemReader = new Org.BouncyCastle.OpenSsl.PemReader(new StreamReader(certData));
var cert = (Org.BouncyCastle.X509.X509Certificate)pemReader.ReadObject();
using (var sr = new StreamReader(body))
{
var content = await sr.ReadToEndAsync();
var result = CheckRequestSignature(Encoding.UTF8.GetBytes(content), signature, cert);
return result;
}
}
private static bool CheckRequestSignature(byte[] bodyData, string signature, Org.BouncyCastle.X509.X509Certificate cert)
{
byte[] sig = Convert.FromBase64String(signature);
var pubKey = (Org.BouncyCastle.Crypto.Parameters.RsaKeyParameters)cert.GetPublicKey();
var signer = Org.BouncyCastle.Security.SignerUtilities.GetSigner(Algorithm);
signer.Init(false, pubKey);
signer.BlockUpdate(bodyData, 0, bodyData.Length);
return signer.VerifySignature(sig);
}
}
}
Does anybody have a tip, what i'm doing wrong or has used a different framework or apis?

From the code you shared it looks correct. Without seeing the rest of the code I am not sure if the Signature you got is correct. You can take a look at our C# code that passed certification at https://github.com/sophtron/Alexa-Skill-Sophtron and compare against your own to see if there are any differences.
I tried RSACryptoServiceProvider from .net library for signature validation and it had never worked. So I had to switch to BouncyCastle.1.8.1 too and it worked for me.

Where do I put the dlls for edge.js to access

I have been using edge.js to call a C# function from within my Node.js app, however when I go to execute the C# code I get for example:
Metadata file 'System.Collections.Generic.dll' could not be found
Metadata file 'System.Text.dll' could not be found
...
My code is this below, basically wanting to run a SSIS package using a stored procedure which I am calling from C#. Basically all my referenced dll's can't be found? Where should I put the dlls for edge to find them?
var executeSQL = edge.func(function() {
/*
#r "System.Data.dll"
#r "System.Collections.Generic.dll"
#r "System.Linq.dll"
#r "System.Text.dll"
using System.Linq;
using System.Text;
using System.Data;
using System.Collections.Generic;
using System.Threading.Tasks;
public class StartUp
{
public async Task<object> Invoke(object input)
{
string result = string.Empty;
string packagePath = #"\SSISDB\test\package.dtsx";
string spName = "storedProcName";
using (var conn = new System.Data.SqlClient.SqlConnection("connectionString"))
using (var command = new System.Data.SqlClient.SqlCommand(spName, conn)
{
CommandType = System.Data.CommandType.StoredProcedure
})
{
conn.Open();
command.Parameters.AddWithValue("#PackagePath", packagePath);
command.ExecuteNonQuery();
Console.WriteLine("Finished");
};
return null;
}
}
*/
});
I know I can do this without C# and just use a module within node like mssql to execute the stored procedure but this was just an example test to get used to using edge.js

The comment from stuartd was correct in the sense to put the dlls under the same directory as the script (which I had tried) but I was still having the same issue. I solved my problem by having my C# code as a separate file and then referenced that file as below as part of the executeSSIS function. payload is just the object that gets passed from my node.js script to my C# script. Doing it this way solved my issue.
var payload = {
filePath: 'C:/temp/xlsx/' + req.file.filename,
path: req.packageName,
server: req.server
};
var executeSSIS = edge.func({
source: __dirname + '/cs/Program.cs',
references: [
__dirname + '/cs/System.Data.dll'
]
});
executeSSIS(payload);

Unity Cloud Build: post export method

Problem:
I can't seem to figure out the right signature for Unity cloud build's post export method. According to the documentation:
The fully-qualified name of a public static method you want us to call
after we finish the Unity build process (but before Xcode). For
example: ClassName.CoolMethod or NameSpace.ClassName.CoolMethod. No
trailing parenthesis, and it can't have the same name as your
Pre-Export method! This method must accept a string parameter, which
will receive the path to the exported Unity player (or Xcode project
in the case of iOS).
Here is my code:
public static void OnPostprocessDevBuildIOS(string ExportPath)
{
var projPath = ExportPath + "/Unity-iPhone.xcodeproj/project.pbxproj";
var proj = new PBXProject();
var nativeTarget =
proj.TargetGuidByName(PBXProject.GetUnityTargetName());
var testTarget =
proj.TargetGuidByName(PBXProject.GetUnityTestTargetName());
string[] buildTargets = {nativeTarget, testTarget};
proj.ReadFromString(File.ReadAllText(projPath));
proj.SetBuildProperty(buildTargets, "ENABLE_BITCODE", "NO");
File.WriteAllText(projPath, proj.WriteToString());
}
and here is the error:
I've tried multiple test method signatures and can't seem to get anything to work. I've even tried just a method that logs out the path.
Additional Information:
Unity Version: 5.3.1f
Unity Cloud Build: 5.3.1f
Target: iOS 8.0+
Also, my cloud build settings script is located in the editor folder as required.
Ok so I got the the bitCode disabling post process to work with the following code, but only when I build manually. When I build from cloud build, with no error the app freezes at the splash screen. When I build from my local machine, the app runs just fine.
[PostProcessBuild]
public static void OnPostprocessBuild(BuildTarget buildTarget, string path)
{
if (buildTarget == BuildTarget.iOS)
{
string projPath = path + "/Unity-iPhone.xcodeproj/project.pbxproj";
PBXProject proj = new PBXProject();
proj.ReadFromString(File.ReadAllText(projPath));
string nativeTarget = proj.TargetGuidByName(PBXProject.GetUnityTargetName());
string testTarget = proj.TargetGuidByName(PBXProject.GetUnityTestTargetName());
string[] buildTargets = new string[]{nativeTarget, testTarget};
proj.SetBuildProperty(buildTargets, "ENABLE_BITCODE", "NO");
File.WriteAllText(projPath, proj.WriteToString());
}
}

I too had the same issue "splash screen stuck" right after launch....
I solved this issue. Please use the below code.
Tested in Unity 5.4.1p2 and Xcode 7.3.
using UnityEngine;
using System.Collections;
using UnityEditor;
using System.IO;
using UnityEditor.Callbacks;
#if UNITY_IOS
using UnityEditor.iOS.Xcode;
#endif
public class Postprocessor : AssetPostprocessor
{
#if UNITY_IOS
[PostProcessBuild]
public static void OnPostprocessBuild(BuildTarget buildTarget, string path)
{
if (buildTarget == BuildTarget.iOS)
{
string projPath = path + "/Unity-iPhone.xcodeproj/project.pbxproj";
PBXProject proj = new PBXProject();
proj.ReadFromString(File.ReadAllText(projPath));
string target = proj.TargetGuidByName("Unity-iPhone");
proj.SetBuildProperty(target, "ENABLE_BITCODE", "false");
File.WriteAllText(projPath, proj.WriteToString());
// Add url schema to plist file
string plistPath = path + "/Info.plist";
PlistDocument plist = new PlistDocument();
plist.ReadFromString(File.ReadAllText(plistPath));
// Get root
PlistElementDict rootDict = plist.root;
rootDict.SetBoolean("UIRequiresFullScreen",true);
plist.WriteToFile(plistPath);
}
}
#endif
}

In fact OnPostprocessBuild is always called, so you don't have to put anything in post export method field, which is designed for more specific methods.