Remove unused toolchain components from WebAssembly SDK (#80)

Currently, Wasm SDK contains whole host toolchain package contents, but some components are unnecessary and duplicated with the installed host toolchain.  This PR removes unused components as Linux SDK generator does, and in addition, removes lldb-related components.

The total reduction in `.artifactbundle` for Linux host is: 2.7GB -> 1.3GB ( -1.1GB by platform libraries and host tools, -300MB by lldb components)

Author: kateinoigakukun

Sources/SwiftSDKGenerator/Generator/SwiftSDKGenerator+Unpack.swift

@@ -13,7 +13,7 @@
 import GeneratorEngine
 import struct SystemPackage.FilePath
 
-private let unusedDarwinPlatforms = [
+let unusedDarwinPlatforms = [
   "watchsimulator",
   "iphonesimulator",
   "appletvsimulator",
@@ -22,7 +22,7 @@ private let unusedDarwinPlatforms = [
   "appletvos",
 ]
 
-private let unusedHostBinaries = [
+let unusedHostBinaries = [
   "clangd",
   "docc",
   "dsymutil",
@@ -31,28 +31,42 @@ private let unusedHostBinaries = [
   "swift-package-collection",
 ]
 
+let unusedHostLibraries = [
+  "sourcekitd.framework",
+  "libsourcekitdInProc.so",
+]
+
 extension SwiftSDKGenerator {
   func unpackHostSwift(hostSwiftPackagePath: FilePath) async throws {
     logGenerationStep("Unpacking and copying Swift binaries for the host triple...")
     let pathsConfiguration = self.pathsConfiguration
 
     try await inTemporaryDirectory { fileSystem, tmpDir in
       try await fileSystem.unpack(file: hostSwiftPackagePath, into: tmpDir)
-      // Remove libraries for platforms we don't intend cross-compiling to
-      for platform in unusedDarwinPlatforms {
-        try await fileSystem.removeRecursively(at: tmpDir.appending("usr/lib/swift/\(platform)"))
-        try await fileSystem.removeRecursively(at: tmpDir.appending("usr/lib/swift_static/\(platform)"))
-      }
-      try await fileSystem.removeRecursively(at: tmpDir.appending("usr/lib/sourcekitd.framework"))
-
-      for binary in unusedHostBinaries {
-        try await fileSystem.removeRecursively(at: tmpDir.appending("usr/bin/\(binary)"))
-      }
-
+      try await self.removeToolchainComponents(tmpDir)
       try await fileSystem.rsync(from: tmpDir.appending("usr"), to: pathsConfiguration.toolchainDirPath)
     }
   }
 
+  func removeToolchainComponents(
+    _ packagePath: FilePath,
+    platforms: [String] = unusedDarwinPlatforms,
+    libraries: [String] = unusedHostLibraries,
+    binaries: [String] = unusedHostBinaries
+  ) async throws {
+    // Remove libraries for platforms we don't intend cross-compiling to
+    for platform in platforms {
+      try self.removeRecursively(at: packagePath.appending("usr/lib/swift/\(platform)"))
+      try self.removeRecursively(at: packagePath.appending("usr/lib/swift_static/\(platform)"))
+    }
+    for binary in binaries {
+      try self.removeRecursively(at: packagePath.appending("usr/bin/\(binary)"))
+    }
+    for library in libraries {
+      try self.removeRecursively(at: packagePath.appending("usr/lib/\(library)"))
+    }
+  }
+
   func unpackTargetSwiftPackage(targetSwiftPackagePath: FilePath, relativePathToRoot: [FilePath.Component], sdkDirPath: FilePath) async throws {
     logGenerationStep("Unpacking Swift distribution for the target triple...")
 

Sources/SwiftSDKGenerator/Generator/SwiftSDKGenerator.swift

@@ -199,6 +199,10 @@ public actor SwiftSDKGenerator {
     return result
   }
 
+  func contentsOfDirectory(at path: FilePath) throws -> [String] {
+    try self.fileManager.contentsOfDirectory(atPath: path.string)
+  }
+
   func copy(from source: FilePath, to destination: FilePath) throws {
     try self.removeRecursively(at: destination)
     try self.fileManager.copyItem(atPath: source.string, toPath: destination.string)

Sources/SwiftSDKGenerator/SwiftSDKRecipes/WebAssemblyRecipe.swift

@@ -49,6 +49,22 @@ public struct WebAssemblyRecipe: SwiftSDKRecipe {
     let pathsConfiguration = generator.pathsConfiguration
     logGenerationStep("Copying Swift binaries for the host triple...")
     try await generator.rsync(from: self.hostSwiftPackagePath.appending("usr"), to: pathsConfiguration.toolchainDirPath)
+
+    logGenerationStep("Removing unused toolchain components...")
+    let liblldbNames: [String] = try await {
+      let libDirPath = pathsConfiguration.toolchainDirPath.appending("usr/lib")
+      return try await generator.contentsOfDirectory(at: libDirPath).filter { dirEntry in
+        // liblldb is version suffixed: liblldb.so.17.0.0
+        dirEntry.hasPrefix("liblldb")
+      }
+    }()
+    try await generator.removeToolchainComponents(
+      pathsConfiguration.toolchainDirPath,
+      platforms: unusedDarwinPlatforms + ["linux", "embedded"],
+      libraries: unusedHostLibraries + liblldbNames,
+      binaries: unusedHostBinaries + ["lldb", "lldb-argdumper", "lldb-server"]
+    )
+
     try await self.copyTargetSwift(from: self.targetSwiftPackagePath.appending("usr/lib"), generator: generator)
 
     let autolinkExtractPath = generator.pathsConfiguration.toolchainBinDirPath.appending("swift-autolink-extract")