Swift: lock built swiftmodule traps in main

This should cover `-merge-modules` mode.

Dumping of the configuration to the target files was moved to a
separate pair of header/source files, as now it is also done in
`SwiftOutputRewrite.cpp`.

Author: redsun82

swift/extractor/BUILD.bazel

@@ -2,14 +2,10 @@ load("//swift:rules.bzl", "swift_cc_binary")
 
 swift_cc_binary(
     name = "extractor",
-    srcs = [
-        "SwiftOutputRewrite.cpp",
-        "SwiftOutputRewrite.h",
-        "SwiftExtractor.cpp",
-        "SwiftExtractor.h",
-        "SwiftExtractorConfiguration.h",
-        "main.cpp",
-    ],
+    srcs = glob([
+        "*.h",
+        "*.cpp",
+    ]),
     visibility = ["//swift:__pkg__"],
     deps = [
         "//swift/extractor/infra",

swift/extractor/SwiftExtractor.cpp

@@ -14,7 +14,7 @@
 #include "swift/extractor/trap/generated/TrapClasses.h"
 #include "swift/extractor/trap/TrapDomain.h"
 #include "swift/extractor/visitors/SwiftVisitor.h"
-#include "swift/extractor/infra/TargetFile.h"
+#include "swift/extractor/TargetTrapFile.h"
 
 using namespace codeql;
 using namespace std::string_literals;
@@ -80,20 +80,6 @@ static llvm::SmallVector<swift::Decl*> getTopLevelDecls(swift::ModuleDecl& modul
   return ret;
 }
 
-static void dumpArgs(TargetFile& out, const SwiftExtractorConfiguration& config) {
-  out << "/* extractor-args:\n";
-  for (const auto& opt : config.frontendOptions) {
-    out << "  " << std::quoted(opt) << " \\\n";
-  }
-  out << "\n*/\n";
-
-  out << "/* swift-frontend-args:\n";
-  for (const auto& opt : config.patchedFrontendOptions) {
-    out << "  " << std::quoted(opt) << " \\\n";
-  }
-  out << "\n*/\n";
-}
-
 static void extractDeclarations(const SwiftExtractorConfiguration& config,
                                 swift::CompilerInstance& compiler,
                                 swift::ModuleDecl& module,
@@ -103,12 +89,11 @@ static void extractDeclarations(const SwiftExtractorConfiguration& config,
   // The extractor can be called several times from different processes with
   // the same input file(s). Using `TargetFile` the first process will win, and the following
   // will just skip the work
-  auto trapTarget = TargetFile::create(filename + ".trap", config.trapDir, config.getTempTrapDir());
+  auto trapTarget = createTargetTrapFile(config, filename);
   if (!trapTarget) {
     // another process arrived first, nothing to do for us
     return;
   }
-  dumpArgs(*trapTarget, config);
   TrapDomain trap{*trapTarget};
 
   // TODO: remove this and recreate it with IPA when we have that
@@ -172,14 +157,6 @@ void codeql::extractSwiftFiles(const SwiftExtractorConfiguration& config,
   auto inputFiles = collectInputFilenames(compiler);
   auto modules = collectModules(compiler);
 
-  // we want to make sure any following extractor run will not try to extract things from
-  // the swiftmodule files we are creating in this run, as those things will already have been
-  // extracted from source with more information. We do this by creating empty trap files.
-  // TargetFile semantics will ensure any following run trying to extract that swiftmodule will just
-  // skip doing it
-  for (const auto& output : config.outputSwiftModules) {
-    TargetFile::create(output + ".trap", config.trapDir, config.getTempTrapDir());
-  }
   for (auto& module : modules) {
     bool isFromSourceFile = false;
     for (auto file : module->getFiles()) {

swift/extractor/SwiftExtractorConfiguration.h

@@ -3,6 +3,8 @@
 #include <string>
 #include <vector>
 
+#include "swift/extractor/infra/TargetFile.h"
+
 namespace codeql {
 struct SwiftExtractorConfiguration {
   // The location for storing TRAP files to be imported by CodeQL engine.
@@ -32,9 +34,6 @@ struct SwiftExtractorConfiguration {
   // A temporary directory that contains build artifacts generated by the extractor during the
   // overall extraction process.
   std::string getTempArtifactDir() const { return scratchDir + "/swift-extraction-artifacts"; }
-
-  // Output swiftmodule files. This also includes possible locations where XCode internally moves
-  // modules
-  std::vector<std::string> outputSwiftModules;
 };
+
 }  // namespace codeql

swift/extractor/SwiftOutputRewrite.cpp

@@ -1,5 +1,6 @@
 #include "SwiftOutputRewrite.h"
 #include "swift/extractor/SwiftExtractorConfiguration.h"
+#include "swift/extractor/TargetTrapFile.h"
 
 #include <llvm/ADT/SmallString.h>
 #include <llvm/Support/FileSystem.h>
@@ -323,15 +324,18 @@ std::vector<std::string> collectVFSFiles(const SwiftExtractorConfiguration& conf
 
   return overlays;
 }
-std::vector<std::string> getOutputSwiftModules(
-    const std::unordered_map<std::string, std::string>& remapping) {
-  std::vector<std::string> ret;
+
+void lockOutputSwiftModuleTraps(const SwiftExtractorConfiguration& config,
+                                const std::unordered_map<std::string, std::string>& remapping) {
   for (const auto& [oldPath, newPath] : remapping) {
     if (llvm::StringRef(oldPath).endswith(".swiftmodule")) {
-      ret.push_back(oldPath);
+      if (auto target = createTargetTrapFile(config, oldPath)) {
+        *target << "// trap file deliberately empty\n"
+                   "// this swiftmodule was created during the build, so its entities must have"
+                   " been extracted directly from source files";
+      }
     }
   }
-  return ret;
 }
 
 }  // namespace codeql

swift/extractor/SwiftOutputRewrite.h

@@ -29,7 +29,7 @@ void storeRemappingForVFS(const SwiftExtractorConfiguration& config,
 // This is separate from storeRemappingForVFS as we also collect files produced by other processes.
 std::vector<std::string> collectVFSFiles(const SwiftExtractorConfiguration& config);
 
-// Returns a list of output remapped swift module files
-std::vector<std::string> getOutputSwiftModules(
-    const std::unordered_map<std::string, std::string>& remapping);
+// Creates empty trap files for output swiftmodule files
+void lockOutputSwiftModuleTraps(const SwiftExtractorConfiguration& config,
+                                const std::unordered_map<std::string, std::string>& remapping);
 }  // namespace codeql

swift/extractor/TargetTrapFile.cpp

@@ -0,0 +1,23 @@
+#include "swift/extractor/TargetTrapFile.h"
+#include <iomanip>
+namespace codeql {
+std::optional<TargetFile> createTargetTrapFile(const SwiftExtractorConfiguration& configuration,
+                                               std::string_view target) {
+  std::string trap{target};
+  trap += ".trap";
+  auto ret = TargetFile::create(trap, configuration.trapDir, configuration.getTempTrapDir());
+  if (ret) {
+    *ret << "/* extractor-args:\n";
+    for (const auto& opt : configuration.frontendOptions) {
+      *ret << "  " << std::quoted(opt) << " \\\n";
+    }
+    *ret << "\n*/\n"
+            "/* swift-frontend-args:\n";
+    for (const auto& opt : configuration.patchedFrontendOptions) {
+      *ret << "  " << std::quoted(opt) << " \\\n";
+    }
+    *ret << "\n*/\n";
+  }
+  return ret;
+}
+}  // namespace codeql

swift/extractor/TargetTrapFile.h

@@ -0,0 +1,11 @@
+#pragma once
+
+#include "swift/extractor/infra/TargetFile.h"
+#include "swift/extractor/SwiftExtractorConfiguration.h"
+
+namespace codeql {
+
+std::optional<TargetFile> createTargetTrapFile(const SwiftExtractorConfiguration& configuration,
+                                               std::string_view target);
+
+}  // namespace codeql

swift/extractor/main.cpp

@@ -68,7 +68,7 @@ int main(int argc, char** argv) {
       codeql::rewriteOutputsInPlace(configuration, configuration.patchedFrontendOptions);
   codeql::ensureDirectoriesForNewPathsExist(remapping);
   codeql::storeRemappingForVFS(configuration, remapping);
-  configuration.outputSwiftModules = codeql::getOutputSwiftModules(remapping);
+  codeql::lockOutputSwiftModuleTraps(configuration, remapping);
 
   std::vector<const char*> args;
   for (auto& arg : configuration.patchedFrontendOptions) {