[SwiftCaching] Create standalone reproducer from swift caching build

Add a new option `-gen-reproducer` that when swift caching is used,
create a standalone reproducer that can be used to reproduce the
`swift-frontend` invocation.

Author: cachemeifyoucan

include/swift/AST/DiagnosticsFrontend.def

@@ -537,6 +537,11 @@ GROUPED_ERROR(error_load_input_from_cas, CompilationCaching, none, "failed to lo
 
 GROUPED_ERROR(error_wrong_input_num_for_input_file_key, CompilationCaching, none, "-input-file-key only support one input file", ())
 
+ERROR(error_gen_reproducer_not_caching, none, "-gen-reproducer only supports swift caching (-cache-compile-job)", ())
+ERROR(error_cannot_create_reproducer_dir, none, "failed to create reproducer director '%0': %1", (StringRef, StringRef))
+
+NOTE(note_reproducer, none, "reproducer is available at: %0", (StringRef))
+
 // Dependency Verifier Diagnostics
 ERROR(missing_member_dependency,none,
       "expected "

include/swift/Frontend/CompileJobCacheKey.h

@@ -51,6 +51,11 @@ llvm::Error printCompileJobCacheKey(llvm::cas::ObjectStore &CAS,
                                     llvm::cas::ObjectRef Key,
                                     llvm::raw_ostream &os);
 
+/// Iterating through command-line options in cache key.
+llvm::Error iterateCommandLine(llvm::cas::ObjectStore &CAS,
+                               llvm::cas::ObjectRef Key,
+                               std::function<llvm::Error(StringRef)> Callback);
+
 } // namespace swift
 
 #endif

include/swift/Frontend/FrontendOptions.h

@@ -292,6 +292,12 @@ class FrontendOptions {
   /// by the Clang importer as part of semantic analysis.
   bool ModuleHasBridgingHeader = false;
 
+  /// Generate reproducer.
+  bool GenReproducer = false;
+
+  /// Directory to generate reproducer.
+  std::string GenReproducerDir;
+
   /// Indicates whether or not the frontend should print statistics upon
   /// termination.
   bool PrintStats = false;

include/swift/Option/FrontendOptions.td

@@ -1533,6 +1533,12 @@ def enable_address_dependencies : Flag<["-"], "enable-address-dependencies">,
 def disable_address_dependencies : Flag<["-"], "disable-address-dependencies">,
   HelpText<"Disable enforcement of lifetime dependencies on addressable values.">;
 
+def gen_reproducer : Flag<["-"], "gen-reproducer">,
+                     HelpText<"Generate a reproducer for current compilation.">;
+def gen_reproducer_dir
+    : Separate<["-"], "gen-reproducer-dir">,
+      HelpText<"Path to directory where reproducers write to.">;
+
 } // end let Flags = [FrontendOption, NoDriverOption, HelpHidden]
 
 def disable_experimental_parser_round_trip : Flag<["-"],

lib/Frontend/ArgsToFrontendOptionsConverter.cpp

@@ -164,6 +164,9 @@ bool ArgsToFrontendOptionsConverter::convert(
   Opts.ParallelDependencyScan = Args.hasFlag(OPT_parallel_scan,
                                              OPT_no_parallel_scan,
                                              true);
+  Opts.GenReproducer |= Args.hasArg(OPT_gen_reproducer);
+  Opts.GenReproducerDir = Args.getLastArgValue(OPT_gen_reproducer_dir);
+
   if (const Arg *A = Args.getLastArg(OPT_dependency_scan_cache_path)) {
     Opts.SerializedDependencyScannerCachePath = A->getValue();
   }

lib/Frontend/CompileJobCacheKey.cpp

@@ -105,17 +105,23 @@ swift::createCompileJobCacheKeyForOutput(llvm::cas::ObjectStore &CAS,
   return CAS.storeFromString({BaseKey}, OS.str());
 }
 
+static llvm::Error validateCacheKeyNode(llvm::cas::ObjectProxy Proxy) {
+  if (Proxy.getData().size() != sizeof(uint32_t))
+    return llvm::createStringError("incorrect size for cache key node");
+  if (Proxy.getNumReferences() != 1)
+    return llvm::createStringError("incorrect child number for cache key node");
+
+  return llvm::Error::success();
+}
+
 llvm::Error swift::printCompileJobCacheKey(llvm::cas::ObjectStore &CAS,
                                            llvm::cas::ObjectRef Key,
                                            llvm::raw_ostream &OS) {
   auto Proxy = CAS.getProxy(Key);
   if (!Proxy)
     return Proxy.takeError();
-
-  if (Proxy->getData().size() != sizeof(uint32_t))
-    return llvm::createStringError("incorrect size for cache key node");
-  if (Proxy->getNumReferences() != 1)
-    return llvm::createStringError("incorrect child number for cache key node");
+  if (auto Err = validateCacheKeyNode(*Proxy))
+    return Err;
 
   uint32_t InputIndex = llvm::support::endian::read<uint32_t>(
       Proxy->getData().data(), llvm::endianness::little);
@@ -153,3 +159,41 @@ llvm::Error swift::printCompileJobCacheKey(llvm::cas::ObjectStore &CAS,
 
   return llvm::Error::success();
 }
+
+llvm::Error
+swift::iterateCommandLine(llvm::cas::ObjectStore &CAS, llvm::cas::ObjectRef Key,
+                          std::function<llvm::Error(StringRef)> Callback) {
+  auto Proxy = CAS.getProxy(Key);
+  if (!Proxy)
+    return Proxy.takeError();
+
+  if (auto Err = validateCacheKeyNode(*Proxy))
+    return Err;
+
+  auto Base = Proxy->getReference(0);
+  llvm::cas::TreeSchema Schema(CAS);
+  auto Tree = Schema.load(Base);
+  if (!Tree)
+    return Tree.takeError();
+
+  std::string BaseStr;
+  llvm::raw_string_ostream BaseOS(BaseStr);
+  return Tree->forEachEntry(
+      [&](const llvm::cas::NamedTreeEntry &Entry) -> llvm::Error {
+        auto Ref = Entry.getRef();
+        auto DataProxy = CAS.getProxy(Ref);
+        if (!DataProxy)
+          return DataProxy.takeError();
+
+        if (Entry.getName() != "command-line")
+          return llvm::Error::success();
+
+        StringRef Line, Remain = DataProxy->getData();
+        while (!Remain.empty()) {
+          std::tie(Line, Remain) = Remain.split(0);
+          if (auto Err = Callback(Line))
+            return Err;
+        }
+        return llvm::Error::success();
+      });
+}