lldb/source/Plugins/ExpressionParser/Clang/ClangExpressionParser.cpp

   1 //===-- ClangExpressionParser.cpp -----------------------------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #include "clang/AST/ASTContext.h"
  10 #include "clang/AST/ASTDiagnostic.h"
  11 #include "clang/AST/ExternalASTSource.h"
  12 #include "clang/AST/PrettyPrinter.h"
  13 #include "clang/Basic/Builtins.h"
  14 #include "clang/Basic/DarwinSDKInfo.h"
  15 #include "clang/Basic/DiagnosticIDs.h"
  16 #include "clang/Basic/SourceLocation.h"
  17 #include "clang/Basic/TargetInfo.h"
  18 #include "clang/Basic/Version.h"
  19 #include "clang/CodeGen/CodeGenAction.h"
  20 #include "clang/CodeGen/ModuleBuilder.h"
  21 #include "clang/Edit/Commit.h"
  22 #include "clang/Edit/EditedSource.h"
  23 #include "clang/Edit/EditsReceiver.h"
  24 #include "clang/Frontend/CompilerInstance.h"
  25 #include "clang/Frontend/CompilerInvocation.h"
  26 #include "clang/Frontend/FrontendActions.h"
  27 #include "clang/Frontend/FrontendDiagnostic.h"
  28 #include "clang/Frontend/FrontendPluginRegistry.h"
  29 #include "clang/Frontend/TextDiagnostic.h"
  30 #include "clang/Frontend/TextDiagnosticBuffer.h"
  31 #include "clang/Frontend/TextDiagnosticPrinter.h"
  32 #include "clang/Lex/Preprocessor.h"
  33 #include "clang/Parse/ParseAST.h"
  34 #include "clang/Rewrite/Core/Rewriter.h"
  35 #include "clang/Rewrite/Frontend/FrontendActions.h"
  36 #include "clang/Sema/CodeCompleteConsumer.h"
  37 #include "clang/Sema/Sema.h"
  38 #include "clang/Sema/SemaConsumer.h"
  39
  40 #include "llvm/ADT/StringRef.h"
  41 #include "llvm/ExecutionEngine/ExecutionEngine.h"
  42 #include "llvm/Support/CrashRecoveryContext.h"
  43 #include "llvm/Support/Debug.h"
  44 #include "llvm/Support/Error.h"
  45 #include "llvm/Support/FileSystem.h"
  46 #include "llvm/Support/TargetSelect.h"
  47 #include "llvm/TargetParser/Triple.h"
  48
  49 #include "llvm/IR/LLVMContext.h"
  50 #include "llvm/IR/Module.h"
  51 #include "llvm/Support/DynamicLibrary.h"
  52 #include "llvm/Support/ErrorHandling.h"
  53 #include "llvm/Support/MemoryBuffer.h"
  54 #include "llvm/Support/Signals.h"
  55 #include "llvm/TargetParser/Host.h"
  56
  57 #include "ClangDiagnostic.h"
  58 #include "ClangExpressionParser.h"
  59 #include "ClangUserExpression.h"
  60
  61 #include "ASTUtils.h"
  62 #include "ClangASTSource.h"
  63 #include "ClangDiagnostic.h"
  64 #include "ClangExpressionDeclMap.h"
  65 #include "ClangExpressionHelper.h"
  66 #include "ClangExpressionParser.h"
  67 #include "ClangHost.h"
  68 #include "ClangModulesDeclVendor.h"
  69 #include "ClangPersistentVariables.h"
  70 #include "IRDynamicChecks.h"
  71 #include "IRForTarget.h"
  72 #include "ModuleDependencyCollector.h"
  73
  74 #include "Plugins/TypeSystem/Clang/TypeSystemClang.h"
  75 #include "lldb/Core/Debugger.h"
  76 #include "lldb/Core/Disassembler.h"
  77 #include "lldb/Core/Module.h"
  78 #include "lldb/Expression/IRExecutionUnit.h"
  79 #include "lldb/Expression/IRInterpreter.h"
  80 #include "lldb/Host/File.h"
  81 #include "lldb/Host/HostInfo.h"
  82 #include "lldb/Symbol/SymbolVendor.h"
  83 #include "lldb/Target/ExecutionContext.h"
  84 #include "lldb/Target/ExecutionContextScope.h"
  85 #include "lldb/Target/Language.h"
  86 #include "lldb/Target/Process.h"
  87 #include "lldb/Target/Target.h"
  88 #include "lldb/Target/ThreadPlanCallFunction.h"
  89 #include "lldb/Utility/DataBufferHeap.h"
  90 #include "lldb/Utility/LLDBAssert.h"
  91 #include "lldb/Utility/LLDBLog.h"
  92 #include "lldb/Utility/Log.h"
  93 #include "lldb/Utility/Stream.h"
  94 #include "lldb/Utility/StreamString.h"
  95 #include "lldb/Utility/StringList.h"
  96
  97 #include "Plugins/LanguageRuntime/ObjC/ObjCLanguageRuntime.h"
  98 #include "Plugins/Platform/MacOSX/PlatformDarwin.h"
  99 #include "lldb/Utility/XcodeSDK.h"
 100
 101 #include <cctype>
 102 #include <memory>
 103 #include <optional>
 104
 105 using namespace clang;
 106 using namespace llvm;
 107 using namespace lldb_private;
 108
 109 //===----------------------------------------------------------------------===//
 110 // Utility Methods for Clang
 111 //===----------------------------------------------------------------------===//
 112
 113 class ClangExpressionParser::LLDBPreprocessorCallbacks : public PPCallbacks {
 114   ClangModulesDeclVendor &m_decl_vendor;
 115   ClangPersistentVariables &m_persistent_vars;
 116   clang::SourceManager &m_source_mgr;
 117   StreamString m_error_stream;
 118   bool m_has_errors = false;
 119
 120 public:
 121   LLDBPreprocessorCallbacks(ClangModulesDeclVendor &decl_vendor,
 122                             ClangPersistentVariables &persistent_vars,
 123                             clang::SourceManager &source_mgr)
 124       : m_decl_vendor(decl_vendor), m_persistent_vars(persistent_vars),
 125         m_source_mgr(source_mgr) {}
 126
 127   void moduleImport(SourceLocation import_location, clang::ModuleIdPath path,
 128                     const clang::Module * /*null*/) override {
 129     // Ignore modules that are imported in the wrapper code as these are not
 130     // loaded by the user.
 131     llvm::StringRef filename =
 132         m_source_mgr.getPresumedLoc(import_location).getFilename();
 133     if (filename == ClangExpressionSourceCode::g_prefix_file_name)
 134       return;
 135
 136     SourceModule module;
 137
 138     for (const std::pair<IdentifierInfo *, SourceLocation> &component : path)
 139       module.path.push_back(ConstString(component.first->getName()));
 140
 141     StreamString error_stream;
 142
 143     ClangModulesDeclVendor::ModuleVector exported_modules;
 144     if (!m_decl_vendor.AddModule(module, &exported_modules, m_error_stream))
 145       m_has_errors = true;
 146
 147     for (ClangModulesDeclVendor::ModuleID module : exported_modules)
 148       m_persistent_vars.AddHandLoadedClangModule(module);
 149   }
 150
 151   bool hasErrors() { return m_has_errors; }
 152
 153   llvm::StringRef getErrorString() { return m_error_stream.GetString(); }
 154 };
 155
 156 static void AddAllFixIts(ClangDiagnostic *diag, const clang::Diagnostic &Info) {
 157   for (auto &fix_it : Info.getFixItHints()) {
 158     if (fix_it.isNull())
 159       continue;
 160     diag->AddFixitHint(fix_it);
 161   }
 162 }
 163
 164 class ClangDiagnosticManagerAdapter : public clang::DiagnosticConsumer {
 165 public:
 166   ClangDiagnosticManagerAdapter(DiagnosticOptions &opts, StringRef filename)
 167       : m_filename(filename) {
 168     DiagnosticOptions *options = new DiagnosticOptions(opts);
 169     options->ShowPresumedLoc = true;
 170     options->ShowLevel = false;
 171     m_os = std::make_shared<llvm::raw_string_ostream>(m_output);
 172     m_passthrough =
 173         std::make_shared<clang::TextDiagnosticPrinter>(*m_os, options);
 174   }
 175
 176   void ResetManager(DiagnosticManager *manager = nullptr) {
 177     m_manager = manager;
 178   }
 179
 180   /// Returns the last error ClangDiagnostic message that the
 181   /// DiagnosticManager received or a nullptr.
 182   ClangDiagnostic *MaybeGetLastClangDiag() const {
 183     if (m_manager->Diagnostics().empty())
 184       return nullptr;
 185     auto &diags = m_manager->Diagnostics();
 186     for (auto it = diags.rbegin(); it != diags.rend(); it++) {
 187       lldb_private::Diagnostic *diag = it->get();
 188       if (ClangDiagnostic *clang_diag = dyn_cast<ClangDiagnostic>(diag)) {
 189         if (clang_diag->GetSeverity() == lldb::eSeverityWarning)
 190           return nullptr;
 191         if (clang_diag->GetSeverity() == lldb::eSeverityError)
 192           return clang_diag;
 193       }
 194     }
 195     return nullptr;
 196   }
 197
 198   void HandleDiagnostic(DiagnosticsEngine::Level DiagLevel,
 199                         const clang::Diagnostic &Info) override {
 200     if (!m_manager) {
 201       // We have no DiagnosticManager before/after parsing but we still could
 202       // receive diagnostics (e.g., by the ASTImporter failing to copy decls
 203       // when we move the expression result ot the ScratchASTContext). Let's at
 204       // least log these diagnostics until we find a way to properly render
 205       // them and display them to the user.
 206       Log *log = GetLog(LLDBLog::Expressions);
 207       if (log) {
 208         llvm::SmallVector<char, 32> diag_str;
 209         Info.FormatDiagnostic(diag_str);
 210         diag_str.push_back('\0');
 211         const char *plain_diag = diag_str.data();
 212         LLDB_LOG(log, "Received diagnostic outside parsing: {0}", plain_diag);
 213       }
 214       return;
 215     }
 216
 217     // Update error/warning counters.
 218     DiagnosticConsumer::HandleDiagnostic(DiagLevel, Info);
 219
 220     // Render diagnostic message to m_output.
 221     m_output.clear();
 222     m_passthrough->HandleDiagnostic(DiagLevel, Info);
 223
 224     DiagnosticDetail detail;
 225     switch (DiagLevel) {
 226     case DiagnosticsEngine::Level::Fatal:
 227     case DiagnosticsEngine::Level::Error:
 228       detail.severity = lldb::eSeverityError;
 229       break;
 230     case DiagnosticsEngine::Level::Warning:
 231       detail.severity = lldb::eSeverityWarning;
 232       break;
 233     case DiagnosticsEngine::Level::Remark:
 234     case DiagnosticsEngine::Level::Ignored:
 235       detail.severity = lldb::eSeverityInfo;
 236       break;
 237     case DiagnosticsEngine::Level::Note:
 238       // 'note:' diagnostics for errors and warnings can also contain Fix-Its.
 239       // We add these Fix-Its to the last error diagnostic to make sure
 240       // that we later have all Fix-Its related to an 'error' diagnostic when
 241       // we apply them to the user expression.
 242       auto *clang_diag = MaybeGetLastClangDiag();
 243       // If we don't have a previous diagnostic there is nothing to do.
 244       // If the previous diagnostic already has its own Fix-Its, assume that
 245       // the 'note:' Fix-It is just an alternative way to solve the issue and
 246       // ignore these Fix-Its.
 247       if (!clang_diag || clang_diag->HasFixIts())
 248         break;
 249       // Ignore all Fix-Its that are not associated with an error.
 250       if (clang_diag->GetSeverity() != lldb::eSeverityError)
 251         break;
 252       AddAllFixIts(clang_diag, Info);
 253       break;
 254     }
 255       // ClangDiagnostic messages are expected to have no whitespace/newlines
 256       // around them.
 257       std::string stripped_output =
 258           std::string(llvm::StringRef(m_output).trim());
 259
 260       // Translate the source location.
 261       if (Info.hasSourceManager()) {
 262         DiagnosticDetail::SourceLocation loc;
 263         clang::SourceManager &sm = Info.getSourceManager();
 264         const clang::SourceLocation sloc = Info.getLocation();
 265         if (sloc.isValid()) {
 266           const clang::FullSourceLoc fsloc(sloc, sm);
 267           clang::PresumedLoc PLoc = fsloc.getPresumedLoc(true);
 268           StringRef filename =
 269               PLoc.isValid() ? PLoc.getFilename() : StringRef{};
 270           loc.file = FileSpec(filename);
 271           loc.line = fsloc.getSpellingLineNumber();
 272           loc.column = fsloc.getSpellingColumnNumber();
 273           loc.in_user_input = filename == m_filename;
 274           loc.hidden = filename.starts_with("<lldb wrapper ");
 275
 276           // Find the range of the primary location.
 277           for (const auto &range : Info.getRanges()) {
 278             if (range.getBegin() == sloc) {
 279               // FIXME: This is probably not handling wide characters correctly.
 280               unsigned end_col = sm.getSpellingColumnNumber(range.getEnd());
 281               if (end_col > loc.column)
 282                 loc.length = end_col - loc.column;
 283               break;
 284             }
 285           }
 286           detail.source_location = loc;
 287         }
 288       }
 289       llvm::SmallString<0> msg;
 290       Info.FormatDiagnostic(msg);
 291       detail.message = msg.str();
 292       detail.rendered = stripped_output;
 293       auto new_diagnostic =
 294           std::make_unique<ClangDiagnostic>(detail, Info.getID());
 295
 296       // Don't store away warning fixits, since the compiler doesn't have
 297       // enough context in an expression for the warning to be useful.
 298       // FIXME: Should we try to filter out FixIts that apply to our generated
 299       // code, and not the user's expression?
 300       if (detail.severity == lldb::eSeverityError)
 301         AddAllFixIts(new_diagnostic.get(), Info);
 302
 303       m_manager->AddDiagnostic(std::move(new_diagnostic));
 304   }
 305
 306   void BeginSourceFile(const LangOptions &LO, const Preprocessor *PP) override {
 307     m_passthrough->BeginSourceFile(LO, PP);
 308   }
 309
 310   void EndSourceFile() override { m_passthrough->EndSourceFile(); }
 311
 312 private:
 313   DiagnosticManager *m_manager = nullptr;
 314   std::shared_ptr<clang::TextDiagnosticPrinter> m_passthrough;
 315   /// Output stream of m_passthrough.
 316   std::shared_ptr<llvm::raw_string_ostream> m_os;
 317   /// Output string filled by m_os.
 318   std::string m_output;
 319   StringRef m_filename;
 320 };
 321
 322 /// Returns true if the SDK for the specified triple supports
 323 /// builtin modules in system headers. This is used to decide
 324 /// whether to pass -fbuiltin-headers-in-system-modules to
 325 /// the compiler instance when compiling the `std` module.
 326 static llvm::Expected<bool>
 327 sdkSupportsBuiltinModules(lldb_private::Target &target) {
 328   auto arch_spec = target.GetArchitecture();
 329   auto const &triple = arch_spec.GetTriple();
 330   auto module_sp = target.GetExecutableModule();
 331   if (!module_sp)
 332     return llvm::createStringError("Executable module not found.");
 333
 334   // Get SDK path that the target was compiled against.
 335   auto platform_sp = target.GetPlatform();
 336   if (!platform_sp)
 337     return llvm::createStringError("No Platform plugin found on target.");
 338
 339   auto sdk_or_err = platform_sp->GetSDKPathFromDebugInfo(*module_sp);
 340   if (!sdk_or_err)
 341     return sdk_or_err.takeError();
 342
 343   // Use the SDK path from debug-info to find a local matching SDK directory.
 344   auto sdk_path_or_err =
 345       HostInfo::GetSDKRoot(HostInfo::SDKOptions{std::move(sdk_or_err->first)});
 346   if (!sdk_path_or_err)
 347     return sdk_path_or_err.takeError();
 348
 349   auto VFS = FileSystem::Instance().GetVirtualFileSystem();
 350   if (!VFS)
 351     return llvm::createStringError("No virtual filesystem available.");
 352
 353   // Extract SDK version from the /path/to/some.sdk/SDKSettings.json
 354   auto parsed_or_err = clang::parseDarwinSDKInfo(*VFS, *sdk_path_or_err);
 355   if (!parsed_or_err)
 356     return parsed_or_err.takeError();
 357
 358   auto maybe_sdk = *parsed_or_err;
 359   if (!maybe_sdk)
 360     return llvm::createStringError("Couldn't find Darwin SDK info.");
 361
 362   return XcodeSDK::SDKSupportsBuiltinModules(triple, maybe_sdk->getVersion());
 363 }
 364
 365 static void SetupModuleHeaderPaths(CompilerInstance *compiler,
 366                                    std::vector<std::string> include_directories,
 367                                    lldb::TargetSP target_sp) {
 368   Log *log = GetLog(LLDBLog::Expressions);
 369
 370   HeaderSearchOptions &search_opts = compiler->getHeaderSearchOpts();
 371
 372   for (const std::string &dir : include_directories) {
 373     search_opts.AddPath(dir, frontend::System, false, true);
 374     LLDB_LOG(log, "Added user include dir: {0}", dir);
 375   }
 376
 377   llvm::SmallString<128> module_cache;
 378   const auto &props = ModuleList::GetGlobalModuleListProperties();
 379   props.GetClangModulesCachePath().GetPath(module_cache);
 380   search_opts.ModuleCachePath = std::string(module_cache.str());
 381   LLDB_LOG(log, "Using module cache path: {0}", module_cache.c_str());
 382
 383   search_opts.ResourceDir = GetClangResourceDir().GetPath();
 384
 385   search_opts.ImplicitModuleMaps = true;
 386 }
 387
 388 /// Iff the given identifier is a C++ keyword, remove it from the
 389 /// identifier table (i.e., make the token a normal identifier).
 390 static void RemoveCppKeyword(IdentifierTable &idents, llvm::StringRef token) {
 391   // FIXME: 'using' is used by LLDB for local variables, so we can't remove
 392   // this keyword without breaking this functionality.
 393   if (token == "using")
 394     return;
 395   // GCC's '__null' is used by LLDB to define NULL/Nil/nil.
 396   if (token == "__null")
 397     return;
 398
 399   LangOptions cpp_lang_opts;
 400   cpp_lang_opts.CPlusPlus = true;
 401   cpp_lang_opts.CPlusPlus11 = true;
 402   cpp_lang_opts.CPlusPlus20 = true;
 403
 404   clang::IdentifierInfo &ii = idents.get(token);
 405   // The identifier has to be a C++-exclusive keyword. if not, then there is
 406   // nothing to do.
 407   if (!ii.isCPlusPlusKeyword(cpp_lang_opts))
 408     return;
 409   // If the token is already an identifier, then there is nothing to do.
 410   if (ii.getTokenID() == clang::tok::identifier)
 411     return;
 412   // Otherwise the token is a C++ keyword, so turn it back into a normal
 413   // identifier.
 414   ii.revertTokenIDToIdentifier();
 415 }
 416
 417 /// Remove all C++ keywords from the given identifier table.
 418 static void RemoveAllCppKeywords(IdentifierTable &idents) {
 419 #define KEYWORD(NAME, FLAGS) RemoveCppKeyword(idents, llvm::StringRef(#NAME));
 420 #include "clang/Basic/TokenKinds.def"
 421 }
 422
 423 /// Configures Clang diagnostics for the expression parser.
 424 static void SetupDefaultClangDiagnostics(CompilerInstance &compiler) {
 425   // List of Clang warning groups that are not useful when parsing expressions.
 426   const std::vector<const char *> groupsToIgnore = {
 427       "unused-value",
 428       "odr",
 429       "unused-getter-return-value",
 430   };
 431   for (const char *group : groupsToIgnore) {
 432     compiler.getDiagnostics().setSeverityForGroup(
 433         clang::diag::Flavor::WarningOrError, group,
 434         clang::diag::Severity::Ignored, SourceLocation());
 435   }
 436 }
 437
 438 /// Returns a string representing current ABI.
 439 ///
 440 /// \param[in] target_arch
 441 ///     The target architecture.
 442 ///
 443 /// \return
 444 ///     A string representing target ABI for the current architecture.
 445 static std::string GetClangTargetABI(const ArchSpec &target_arch) {
 446   if (target_arch.IsMIPS()) {
 447     switch (target_arch.GetFlags() & ArchSpec::eMIPSABI_mask) {
 448     case ArchSpec::eMIPSABI_N64:
 449       return "n64";
 450     case ArchSpec::eMIPSABI_N32:
 451       return "n32";
 452     case ArchSpec::eMIPSABI_O32:
 453       return "o32";
 454     default:
 455       return {};
 456     }
 457   }
 458
 459   if (target_arch.GetTriple().isRISCV64()) {
 460     switch (target_arch.GetFlags() & ArchSpec::eRISCV_float_abi_mask) {
 461     case ArchSpec::eRISCV_float_abi_soft:
 462       return "lp64";
 463     case ArchSpec::eRISCV_float_abi_single:
 464       return "lp64f";
 465     case ArchSpec::eRISCV_float_abi_double:
 466       return "lp64d";
 467     case ArchSpec::eRISCV_float_abi_quad:
 468       return "lp64q";
 469     default:
 470       return {};
 471     }
 472   }
 473
 474   if (target_arch.GetTriple().isRISCV32()) {
 475     switch (target_arch.GetFlags() & ArchSpec::eRISCV_float_abi_mask) {
 476     case ArchSpec::eRISCV_float_abi_soft:
 477       return "ilp32";
 478     case ArchSpec::eRISCV_float_abi_single:
 479       return "ilp32f";
 480     case ArchSpec::eRISCV_float_abi_double:
 481       return "ilp32d";
 482     case ArchSpec::eRISCV_float_abi_soft | ArchSpec::eRISCV_rve:
 483       return "ilp32e";
 484     default:
 485       return {};
 486     }
 487   }
 488
 489   if (target_arch.GetTriple().isLoongArch64()) {
 490     switch (target_arch.GetFlags() & ArchSpec::eLoongArch_abi_mask) {
 491     case ArchSpec::eLoongArch_abi_soft_float:
 492       return "lp64s";
 493     case ArchSpec::eLoongArch_abi_single_float:
 494       return "lp64f";
 495     case ArchSpec::eLoongArch_abi_double_float:
 496       return "lp64d";
 497     default:
 498       return {};
 499     }
 500   }
 501
 502   return {};
 503 }
 504
 505 static void SetupTargetOpts(CompilerInstance &compiler,
 506                             lldb_private::Target const &target) {
 507   Log *log = GetLog(LLDBLog::Expressions);
 508   ArchSpec target_arch = target.GetArchitecture();
 509
 510   const auto target_machine = target_arch.GetMachine();
 511   if (target_arch.IsValid()) {
 512     std::string triple = target_arch.GetTriple().str();
 513     compiler.getTargetOpts().Triple = triple;
 514     LLDB_LOGF(log, "Using %s as the target triple",
 515               compiler.getTargetOpts().Triple.c_str());
 516   } else {
 517     // If we get here we don't have a valid target and just have to guess.
 518     // Sometimes this will be ok to just use the host target triple (when we
 519     // evaluate say "2+3", but other expressions like breakpoint conditions and
 520     // other things that _are_ target specific really shouldn't just be using
 521     // the host triple. In such a case the language runtime should expose an
 522     // overridden options set (3), below.
 523     compiler.getTargetOpts().Triple = llvm::sys::getDefaultTargetTriple();
 524     LLDB_LOGF(log, "Using default target triple of %s",
 525               compiler.getTargetOpts().Triple.c_str());
 526   }
 527   // Now add some special fixes for known architectures: Any arm32 iOS
 528   // environment, but not on arm64
 529   if (compiler.getTargetOpts().Triple.find("arm64") == std::string::npos &&
 530       compiler.getTargetOpts().Triple.find("arm") != std::string::npos &&
 531       compiler.getTargetOpts().Triple.find("ios") != std::string::npos) {
 532     compiler.getTargetOpts().ABI = "apcs-gnu";
 533   }
 534   // Supported subsets of x86
 535   if (target_machine == llvm::Triple::x86 ||
 536       target_machine == llvm::Triple::x86_64) {
 537     compiler.getTargetOpts().FeaturesAsWritten.push_back("+sse");
 538     compiler.getTargetOpts().FeaturesAsWritten.push_back("+sse2");
 539   }
 540
 541   // Set the target CPU to generate code for. This will be empty for any CPU
 542   // that doesn't really need to make a special
 543   // CPU string.
 544   compiler.getTargetOpts().CPU = target_arch.GetClangTargetCPU();
 545
 546   // Set the target ABI
 547   if (std::string abi = GetClangTargetABI(target_arch); !abi.empty())
 548     compiler.getTargetOpts().ABI = std::move(abi);
 549
 550   if ((target_machine == llvm::Triple::riscv64 &&
 551        compiler.getTargetOpts().ABI == "lp64f") ||
 552       (target_machine == llvm::Triple::riscv32 &&
 553        compiler.getTargetOpts().ABI == "ilp32f"))
 554     compiler.getTargetOpts().FeaturesAsWritten.emplace_back("+f");
 555
 556   if ((target_machine == llvm::Triple::riscv64 &&
 557        compiler.getTargetOpts().ABI == "lp64d") ||
 558       (target_machine == llvm::Triple::riscv32 &&
 559        compiler.getTargetOpts().ABI == "ilp32d"))
 560     compiler.getTargetOpts().FeaturesAsWritten.emplace_back("+d");
 561
 562   if ((target_machine == llvm::Triple::loongarch64 &&
 563        compiler.getTargetOpts().ABI == "lp64f"))
 564     compiler.getTargetOpts().FeaturesAsWritten.emplace_back("+f");
 565
 566   if ((target_machine == llvm::Triple::loongarch64 &&
 567        compiler.getTargetOpts().ABI == "lp64d"))
 568     compiler.getTargetOpts().FeaturesAsWritten.emplace_back("+d");
 569 }
 570
 571 static void SetupLangOpts(CompilerInstance &compiler,
 572                           ExecutionContextScope &exe_scope,
 573                           const Expression &expr) {
 574   Log *log = GetLog(LLDBLog::Expressions);
 575
 576   // If the expression is being evaluated in the context of an existing stack
 577   // frame, we introspect to see if the language runtime is available.
 578
 579   lldb::StackFrameSP frame_sp = exe_scope.CalculateStackFrame();
 580   lldb::ProcessSP process_sp = exe_scope.CalculateProcess();
 581
 582   // Defaults to lldb::eLanguageTypeUnknown.
 583   lldb::LanguageType frame_lang = expr.Language().AsLanguageType();
 584
 585   // Make sure the user hasn't provided a preferred execution language with
 586   // `expression --language X -- ...`
 587   if (frame_sp && frame_lang == lldb::eLanguageTypeUnknown)
 588     frame_lang = frame_sp->GetLanguage().AsLanguageType();
 589
 590   if (process_sp && frame_lang != lldb::eLanguageTypeUnknown) {
 591     LLDB_LOGF(log, "Frame has language of type %s",
 592               lldb_private::Language::GetNameForLanguageType(frame_lang));
 593   }
 594
 595   lldb::LanguageType language = expr.Language().AsLanguageType();
 596   LangOptions &lang_opts = compiler.getLangOpts();
 597
 598   // FIXME: should this switch on frame_lang?
 599   switch (language) {
 600   case lldb::eLanguageTypeC:
 601   case lldb::eLanguageTypeC89:
 602   case lldb::eLanguageTypeC99:
 603   case lldb::eLanguageTypeC11:
 604     // FIXME: the following language option is a temporary workaround,
 605     // to "ask for C, get C++."
 606     // For now, the expression parser must use C++ anytime the language is a C
 607     // family language, because the expression parser uses features of C++ to
 608     // capture values.
 609     lang_opts.CPlusPlus = true;
 610     break;
 611   case lldb::eLanguageTypeObjC:
 612     lang_opts.ObjC = true;
 613     // FIXME: the following language option is a temporary workaround,
 614     // to "ask for ObjC, get ObjC++" (see comment above).
 615     lang_opts.CPlusPlus = true;
 616
 617     // Clang now sets as default C++14 as the default standard (with
 618     // GNU extensions), so we do the same here to avoid mismatches that
 619     // cause compiler error when evaluating expressions (e.g. nullptr not found
 620     // as it's a C++11 feature). Currently lldb evaluates C++14 as C++11 (see
 621     // two lines below) so we decide to be consistent with that, but this could
 622     // be re-evaluated in the future.
 623     lang_opts.CPlusPlus11 = true;
 624     break;
 625   case lldb::eLanguageTypeC_plus_plus_20:
 626     lang_opts.CPlusPlus20 = true;
 627     [[fallthrough]];
 628   case lldb::eLanguageTypeC_plus_plus_17:
 629     // FIXME: add a separate case for CPlusPlus14. Currently folded into C++17
 630     // because C++14 is the default standard for Clang but enabling CPlusPlus14
 631     // expression evaluatino doesn't pass the test-suite cleanly.
 632     lang_opts.CPlusPlus14 = true;
 633     lang_opts.CPlusPlus17 = true;
 634     [[fallthrough]];
 635   case lldb::eLanguageTypeC_plus_plus:
 636   case lldb::eLanguageTypeC_plus_plus_11:
 637   case lldb::eLanguageTypeC_plus_plus_14:
 638     lang_opts.CPlusPlus11 = true;
 639     compiler.getHeaderSearchOpts().UseLibcxx = true;
 640     [[fallthrough]];
 641   case lldb::eLanguageTypeC_plus_plus_03:
 642     lang_opts.CPlusPlus = true;
 643     if (process_sp
 644         // We're stopped in a frame without debug-info. The user probably
 645         // intends to make global queries (which should include Objective-C).
 646         && !(frame_sp && frame_sp->HasDebugInformation()))
 647       lang_opts.ObjC =
 648           process_sp->GetLanguageRuntime(lldb::eLanguageTypeObjC) != nullptr;
 649     break;
 650   case lldb::eLanguageTypeObjC_plus_plus:
 651   case lldb::eLanguageTypeUnknown:
 652   default:
 653     lang_opts.ObjC = true;
 654     lang_opts.CPlusPlus = true;
 655     lang_opts.CPlusPlus11 = true;
 656     compiler.getHeaderSearchOpts().UseLibcxx = true;
 657     break;
 658   }
 659
 660   lang_opts.Bool = true;
 661   lang_opts.WChar = true;
 662   lang_opts.Blocks = true;
 663   lang_opts.DebuggerSupport =
 664       true; // Features specifically for debugger clients
 665   if (expr.DesiredResultType() == Expression::eResultTypeId)
 666     lang_opts.DebuggerCastResultToId = true;
 667
 668   lang_opts.CharIsSigned =
 669       ArchSpec(compiler.getTargetOpts().Triple.c_str()).CharIsSignedByDefault();
 670
 671   // Spell checking is a nice feature, but it ends up completing a lot of types
 672   // that we didn't strictly speaking need to complete. As a result, we spend a
 673   // long time parsing and importing debug information.
 674   lang_opts.SpellChecking = false;
 675
 676   if (process_sp && lang_opts.ObjC) {
 677     if (auto *runtime = ObjCLanguageRuntime::Get(*process_sp)) {
 678       switch (runtime->GetRuntimeVersion()) {
 679       case ObjCLanguageRuntime::ObjCRuntimeVersions::eAppleObjC_V2:
 680         lang_opts.ObjCRuntime.set(ObjCRuntime::MacOSX, VersionTuple(10, 7));
 681         break;
 682       case ObjCLanguageRuntime::ObjCRuntimeVersions::eObjC_VersionUnknown:
 683       case ObjCLanguageRuntime::ObjCRuntimeVersions::eAppleObjC_V1:
 684         lang_opts.ObjCRuntime.set(ObjCRuntime::FragileMacOSX,
 685                                   VersionTuple(10, 7));
 686         break;
 687       case ObjCLanguageRuntime::ObjCRuntimeVersions::eGNUstep_libobjc2:
 688         lang_opts.ObjCRuntime.set(ObjCRuntime::GNUstep, VersionTuple(2, 0));
 689         break;
 690       }
 691
 692       if (runtime->HasNewLiteralsAndIndexing())
 693         lang_opts.DebuggerObjCLiteral = true;
 694     }
 695   }
 696
 697   lang_opts.ThreadsafeStatics = false;
 698   lang_opts.AccessControl = false; // Debuggers get universal access
 699   lang_opts.DollarIdents = true;   // $ indicates a persistent variable name
 700   // We enable all builtin functions beside the builtins from libc/libm (e.g.
 701   // 'fopen'). Those libc functions are already correctly handled by LLDB, and
 702   // additionally enabling them as expandable builtins is breaking Clang.
 703   lang_opts.NoBuiltin = true;
 704 }
 705
 706 static void SetupImportStdModuleLangOpts(CompilerInstance &compiler,
 707                                          lldb_private::Target &target) {
 708   Log *log = GetLog(LLDBLog::Expressions);
 709   LangOptions &lang_opts = compiler.getLangOpts();
 710   lang_opts.Modules = true;
 711   // We want to implicitly build modules.
 712   lang_opts.ImplicitModules = true;
 713   // To automatically import all submodules when we import 'std'.
 714   lang_opts.ModulesLocalVisibility = false;
 715
 716   // We use the @import statements, so we need this:
 717   // FIXME: We could use the modules-ts, but that currently doesn't work.
 718   lang_opts.ObjC = true;
 719
 720   // Options we need to parse libc++ code successfully.
 721   // FIXME: We should ask the driver for the appropriate default flags.
 722   lang_opts.GNUMode = true;
 723   lang_opts.GNUKeywords = true;
 724   lang_opts.CPlusPlus11 = true;
 725
 726   if (auto supported_or_err = sdkSupportsBuiltinModules(target))
 727     lang_opts.BuiltinHeadersInSystemModules = !*supported_or_err;
 728   else
 729     LLDB_LOG_ERROR(log, supported_or_err.takeError(),
 730                    "Failed to determine BuiltinHeadersInSystemModules when "
 731                    "setting up import-std-module: {0}");
 732
 733   // The Darwin libc expects this macro to be set.
 734   lang_opts.GNUCVersion = 40201;
 735 }
 736
 737 //===----------------------------------------------------------------------===//
 738 // Implementation of ClangExpressionParser
 739 //===----------------------------------------------------------------------===//
 740
 741 ClangExpressionParser::ClangExpressionParser(
 742     ExecutionContextScope *exe_scope, Expression &expr,
 743     bool generate_debug_info, std::vector<std::string> include_directories,
 744     std::string filename)
 745     : ExpressionParser(exe_scope, expr, generate_debug_info), m_compiler(),
 746       m_pp_callbacks(nullptr),
 747       m_include_directories(std::move(include_directories)),
 748       m_filename(std::move(filename)) {
 749   Log *log = GetLog(LLDBLog::Expressions);
 750
 751   // We can't compile expressions without a target.  So if the exe_scope is
 752   // null or doesn't have a target, then we just need to get out of here.  I'll
 753   // lldbassert and not make any of the compiler objects since
 754   // I can't return errors directly from the constructor.  Further calls will
 755   // check if the compiler was made and
 756   // bag out if it wasn't.
 757
 758   if (!exe_scope) {
 759     lldbassert(exe_scope &&
 760                "Can't make an expression parser with a null scope.");
 761     return;
 762   }
 763
 764   lldb::TargetSP target_sp;
 765   target_sp = exe_scope->CalculateTarget();
 766   if (!target_sp) {
 767     lldbassert(target_sp.get() &&
 768                "Can't make an expression parser with a null target.");
 769     return;
 770   }
 771
 772   // 1. Create a new compiler instance.
 773   m_compiler = std::make_unique<CompilerInstance>();
 774
 775   // Make sure clang uses the same VFS as LLDB.
 776   m_compiler->createFileManager(FileSystem::Instance().GetVirtualFileSystem());
 777
 778   // 2. Configure the compiler with a set of default options that are
 779   // appropriate for most situations.
 780   SetupTargetOpts(*m_compiler, *target_sp);
 781
 782   // 3. Create and install the target on the compiler.
 783   m_compiler->createDiagnostics(m_compiler->getVirtualFileSystem());
 784   // Limit the number of error diagnostics we emit.
 785   // A value of 0 means no limit for both LLDB and Clang.
 786   m_compiler->getDiagnostics().setErrorLimit(target_sp->GetExprErrorLimit());
 787
 788   if (auto *target_info = TargetInfo::CreateTargetInfo(
 789           m_compiler->getDiagnostics(),
 790           m_compiler->getInvocation().TargetOpts)) {
 791     if (log) {
 792       LLDB_LOGF(log, "Target datalayout string: '%s'",
 793                 target_info->getDataLayoutString());
 794       LLDB_LOGF(log, "Target ABI: '%s'", target_info->getABI().str().c_str());
 795       LLDB_LOGF(log, "Target vector alignment: %d",
 796                 target_info->getMaxVectorAlign());
 797     }
 798     m_compiler->setTarget(target_info);
 799   } else {
 800     if (log)
 801       LLDB_LOGF(log, "Failed to create TargetInfo for '%s'",
 802                 m_compiler->getTargetOpts().Triple.c_str());
 803
 804     lldbassert(false && "Failed to create TargetInfo.");
 805   }
 806
 807   // 4. Set language options.
 808   SetupLangOpts(*m_compiler, *exe_scope, expr);
 809   auto *clang_expr = dyn_cast<ClangUserExpression>(&m_expr);
 810   if (clang_expr && clang_expr->DidImportCxxModules()) {
 811     LLDB_LOG(log, "Adding lang options for importing C++ modules");
 812     SetupImportStdModuleLangOpts(*m_compiler, *target_sp);
 813     SetupModuleHeaderPaths(m_compiler.get(), m_include_directories, target_sp);
 814   }
 815
 816   // Set CodeGen options
 817   m_compiler->getCodeGenOpts().EmitDeclMetadata = true;
 818   m_compiler->getCodeGenOpts().InstrumentFunctions = false;
 819   m_compiler->getCodeGenOpts().setFramePointer(
 820       CodeGenOptions::FramePointerKind::All);
 821   if (generate_debug_info)
 822     m_compiler->getCodeGenOpts().setDebugInfo(codegenoptions::FullDebugInfo);
 823   else
 824     m_compiler->getCodeGenOpts().setDebugInfo(codegenoptions::NoDebugInfo);
 825
 826   // Disable some warnings.
 827   SetupDefaultClangDiagnostics(*m_compiler);
 828
 829   // Inform the target of the language options
 830   //
 831   // FIXME: We shouldn't need to do this, the target should be immutable once
 832   // created. This complexity should be lifted elsewhere.
 833   m_compiler->getTarget().adjust(m_compiler->getDiagnostics(),
 834                                  m_compiler->getLangOpts());
 835
 836   // 5. Set up the diagnostic buffer for reporting errors
 837   auto diag_mgr = new ClangDiagnosticManagerAdapter(
 838       m_compiler->getDiagnostics().getDiagnosticOptions(),
 839       clang_expr ? clang_expr->GetFilename() : StringRef());
 840   m_compiler->getDiagnostics().setClient(diag_mgr);
 841
 842   // 6. Set up the source management objects inside the compiler
 843   m_compiler->createFileManager();
 844   if (!m_compiler->hasSourceManager())
 845     m_compiler->createSourceManager(m_compiler->getFileManager());
 846   m_compiler->createPreprocessor(TU_Complete);
 847
 848   switch (expr.Language().AsLanguageType()) {
 849   case lldb::eLanguageTypeC:
 850   case lldb::eLanguageTypeC89:
 851   case lldb::eLanguageTypeC99:
 852   case lldb::eLanguageTypeC11:
 853   case lldb::eLanguageTypeObjC:
 854     // This is not a C++ expression but we enabled C++ as explained above.
 855     // Remove all C++ keywords from the PP so that the user can still use
 856     // variables that have C++ keywords as names (e.g. 'int template;').
 857     RemoveAllCppKeywords(m_compiler->getPreprocessor().getIdentifierTable());
 858     break;
 859   default:
 860     break;
 861   }
 862
 863   if (auto *clang_persistent_vars = llvm::cast<ClangPersistentVariables>(
 864           target_sp->GetPersistentExpressionStateForLanguage(
 865               lldb::eLanguageTypeC))) {
 866     if (std::shared_ptr<ClangModulesDeclVendor> decl_vendor =
 867             clang_persistent_vars->GetClangModulesDeclVendor()) {
 868       std::unique_ptr<PPCallbacks> pp_callbacks(
 869           new LLDBPreprocessorCallbacks(*decl_vendor, *clang_persistent_vars,
 870                                         m_compiler->getSourceManager()));
 871       m_pp_callbacks =
 872           static_cast<LLDBPreprocessorCallbacks *>(pp_callbacks.get());
 873       m_compiler->getPreprocessor().addPPCallbacks(std::move(pp_callbacks));
 874     }
 875   }
 876
 877   // 7. Most of this we get from the CompilerInstance, but we also want to give
 878   // the context an ExternalASTSource.
 879
 880   auto &PP = m_compiler->getPreprocessor();
 881   auto &builtin_context = PP.getBuiltinInfo();
 882   builtin_context.initializeBuiltins(PP.getIdentifierTable(),
 883                                      m_compiler->getLangOpts());
 884
 885   m_compiler->createASTContext();
 886   clang::ASTContext &ast_context = m_compiler->getASTContext();
 887
 888   m_ast_context = std::make_shared<TypeSystemClang>(
 889       "Expression ASTContext for '" + m_filename + "'", ast_context);
 890
 891   std::string module_name("$__lldb_module");
 892
 893   m_llvm_context = std::make_unique<LLVMContext>();
 894   m_code_generator.reset(CreateLLVMCodeGen(
 895       m_compiler->getDiagnostics(), module_name,
 896       &m_compiler->getVirtualFileSystem(), m_compiler->getHeaderSearchOpts(),
 897       m_compiler->getPreprocessorOpts(), m_compiler->getCodeGenOpts(),
 898       *m_llvm_context));
 899 }
 900
 901 ClangExpressionParser::~ClangExpressionParser() = default;
 902
 903 namespace {
 904
 905 /// \class CodeComplete
 906 ///
 907 /// A code completion consumer for the clang Sema that is responsible for
 908 /// creating the completion suggestions when a user requests completion
 909 /// of an incomplete `expr` invocation.
 910 class CodeComplete : public CodeCompleteConsumer {
 911   CodeCompletionTUInfo m_info;
 912
 913   std::string m_expr;
 914   unsigned m_position = 0;
 915   /// The printing policy we use when printing declarations for our completion
 916   /// descriptions.
 917   clang::PrintingPolicy m_desc_policy;
 918
 919   struct CompletionWithPriority {
 920     CompletionResult::Completion completion;
 921     /// See CodeCompletionResult::Priority;
 922     unsigned Priority;
 923
 924     /// Establishes a deterministic order in a list of CompletionWithPriority.
 925     /// The order returned here is the order in which the completions are
 926     /// displayed to the user.
 927     bool operator<(const CompletionWithPriority &o) const {
 928       // High priority results should come first.
 929       if (Priority != o.Priority)
 930         return Priority > o.Priority;
 931
 932       // Identical priority, so just make sure it's a deterministic order.
 933       return completion.GetUniqueKey() < o.completion.GetUniqueKey();
 934     }
 935   };
 936
 937   /// The stored completions.
 938   /// Warning: These are in a non-deterministic order until they are sorted
 939   /// and returned back to the caller.
 940   std::vector<CompletionWithPriority> m_completions;
 941
 942   /// Returns true if the given character can be used in an identifier.
 943   /// This also returns true for numbers because for completion we usually
 944   /// just iterate backwards over iterators.
 945   ///
 946   /// Note: lldb uses '$' in its internal identifiers, so we also allow this.
 947   static bool IsIdChar(char c) {
 948     return c == '_' || std::isalnum(c) || c == '$';
 949   }
 950
 951   /// Returns true if the given character is used to separate arguments
 952   /// in the command line of lldb.
 953   static bool IsTokenSeparator(char c) { return c == ' ' || c == '\t'; }
 954
 955   /// Drops all tokens in front of the expression that are unrelated for
 956   /// the completion of the cmd line. 'unrelated' means here that the token
 957   /// is not interested for the lldb completion API result.
 958   StringRef dropUnrelatedFrontTokens(StringRef cmd) const {
 959     if (cmd.empty())
 960       return cmd;
 961
 962     // If we are at the start of a word, then all tokens are unrelated to
 963     // the current completion logic.
 964     if (IsTokenSeparator(cmd.back()))
 965       return StringRef();
 966
 967     // Remove all previous tokens from the string as they are unrelated
 968     // to completing the current token.
 969     StringRef to_remove = cmd;
 970     while (!to_remove.empty() && !IsTokenSeparator(to_remove.back())) {
 971       to_remove = to_remove.drop_back();
 972     }
 973     cmd = cmd.drop_front(to_remove.size());
 974
 975     return cmd;
 976   }
 977
 978   /// Removes the last identifier token from the given cmd line.
 979   StringRef removeLastToken(StringRef cmd) const {
 980     while (!cmd.empty() && IsIdChar(cmd.back())) {
 981       cmd = cmd.drop_back();
 982     }
 983     return cmd;
 984   }
 985
 986   /// Attempts to merge the given completion from the given position into the
 987   /// existing command. Returns the completion string that can be returned to
 988   /// the lldb completion API.
 989   std::string mergeCompletion(StringRef existing, unsigned pos,
 990                               StringRef completion) const {
 991     StringRef existing_command = existing.substr(0, pos);
 992     // We rewrite the last token with the completion, so let's drop that
 993     // token from the command.
 994     existing_command = removeLastToken(existing_command);
 995     // We also should remove all previous tokens from the command as they
 996     // would otherwise be added to the completion that already has the
 997     // completion.
 998     existing_command = dropUnrelatedFrontTokens(existing_command);
 999     return existing_command.str() + completion.str();
1000   }
1001
1002 public:
1003   /// Constructs a CodeComplete consumer that can be attached to a Sema.
1004   ///
1005   /// \param[out] expr
1006   ///    The whole expression string that we are currently parsing. This
1007   ///    string needs to be equal to the input the user typed, and NOT the
1008   ///    final code that Clang is parsing.
1009   /// \param[out] position
1010   ///    The character position of the user cursor in the `expr` parameter.
1011   ///
1012   CodeComplete(clang::LangOptions ops, std::string expr, unsigned position)
1013       : CodeCompleteConsumer(CodeCompleteOptions()),
1014         m_info(std::make_shared<GlobalCodeCompletionAllocator>()), m_expr(expr),
1015         m_position(position), m_desc_policy(ops) {
1016
1017     // Ensure that the printing policy is producing a description that is as
1018     // short as possible.
1019     m_desc_policy.SuppressScope = true;
1020     m_desc_policy.SuppressTagKeyword = true;
1021     m_desc_policy.FullyQualifiedName = false;
1022     m_desc_policy.TerseOutput = true;
1023     m_desc_policy.IncludeNewlines = false;
1024     m_desc_policy.UseVoidForZeroParams = false;
1025     m_desc_policy.Bool = true;
1026   }
1027
1028   /// \name Code-completion filtering
1029   /// Check if the result should be filtered out.
1030   bool isResultFilteredOut(StringRef Filter,
1031                            CodeCompletionResult Result) override {
1032     // This code is mostly copied from CodeCompleteConsumer.
1033     switch (Result.Kind) {
1034     case CodeCompletionResult::RK_Declaration:
1035       return !(
1036           Result.Declaration->getIdentifier() &&
1037           Result.Declaration->getIdentifier()->getName().starts_with(Filter));
1038     case CodeCompletionResult::RK_Keyword:
1039       return !StringRef(Result.Keyword).starts_with(Filter);
1040     case CodeCompletionResult::RK_Macro:
1041       return !Result.Macro->getName().starts_with(Filter);
1042     case CodeCompletionResult::RK_Pattern:
1043       return !StringRef(Result.Pattern->getAsString()).starts_with(Filter);
1044     }
1045     // If we trigger this assert or the above switch yields a warning, then
1046     // CodeCompletionResult has been enhanced with more kinds of completion
1047     // results. Expand the switch above in this case.
1048     assert(false && "Unknown completion result type?");
1049     // If we reach this, then we should just ignore whatever kind of unknown
1050     // result we got back. We probably can't turn it into any kind of useful
1051     // completion suggestion with the existing code.
1052     return true;
1053   }
1054
1055 private:
1056   /// Generate the completion strings for the given CodeCompletionResult.
1057   /// Note that this function has to process results that could come in
1058   /// non-deterministic order, so this function should have no side effects.
1059   /// To make this easier to enforce, this function and all its parameters
1060   /// should always be const-qualified.
1061   /// \return Returns std::nullopt if no completion should be provided for the
1062   ///         given CodeCompletionResult.
1063   std::optional<CompletionWithPriority>
1064   getCompletionForResult(const CodeCompletionResult &R) const {
1065     std::string ToInsert;
1066     std::string Description;
1067     // Handle the different completion kinds that come from the Sema.
1068     switch (R.Kind) {
1069     case CodeCompletionResult::RK_Declaration: {
1070       const NamedDecl *D = R.Declaration;
1071       ToInsert = R.Declaration->getNameAsString();
1072       // If we have a function decl that has no arguments we want to
1073       // complete the empty parantheses for the user. If the function has
1074       // arguments, we at least complete the opening bracket.
1075       if (const FunctionDecl *F = dyn_cast<FunctionDecl>(D)) {
1076         if (F->getNumParams() == 0)
1077           ToInsert += "()";
1078         else
1079           ToInsert += "(";
1080         raw_string_ostream OS(Description);
1081         F->print(OS, m_desc_policy, false);
1082       } else if (const VarDecl *V = dyn_cast<VarDecl>(D)) {
1083         Description = V->getType().getAsString(m_desc_policy);
1084       } else if (const FieldDecl *F = dyn_cast<FieldDecl>(D)) {
1085         Description = F->getType().getAsString(m_desc_policy);
1086       } else if (const NamespaceDecl *N = dyn_cast<NamespaceDecl>(D)) {
1087         // If we try to complete a namespace, then we can directly append
1088         // the '::'.
1089         if (!N->isAnonymousNamespace())
1090           ToInsert += "::";
1091       }
1092       break;
1093     }
1094     case CodeCompletionResult::RK_Keyword:
1095       ToInsert = R.Keyword;
1096       break;
1097     case CodeCompletionResult::RK_Macro:
1098       ToInsert = R.Macro->getName().str();
1099       break;
1100     case CodeCompletionResult::RK_Pattern:
1101       ToInsert = R.Pattern->getTypedText();
1102       break;
1103     }
1104     // We also filter some internal lldb identifiers here. The user
1105     // shouldn't see these.
1106     if (llvm::StringRef(ToInsert).starts_with("$__lldb_"))
1107       return std::nullopt;
1108     if (ToInsert.empty())
1109       return std::nullopt;
1110     // Merge the suggested Token into the existing command line to comply
1111     // with the kind of result the lldb API expects.
1112     std::string CompletionSuggestion =
1113         mergeCompletion(m_expr, m_position, ToInsert);
1114
1115     CompletionResult::Completion completion(CompletionSuggestion, Description,
1116                                             CompletionMode::Normal);
1117     return {{completion, R.Priority}};
1118   }
1119
1120 public:
1121   /// Adds the completions to the given CompletionRequest.
1122   void GetCompletions(CompletionRequest &request) {
1123     // Bring m_completions into a deterministic order and pass it on to the
1124     // CompletionRequest.
1125     llvm::sort(m_completions);
1126
1127     for (const CompletionWithPriority &C : m_completions)
1128       request.AddCompletion(C.completion.GetCompletion(),
1129                             C.completion.GetDescription(),
1130                             C.completion.GetMode());
1131   }
1132
1133   /// \name Code-completion callbacks
1134   /// Process the finalized code-completion results.
1135   void ProcessCodeCompleteResults(Sema &SemaRef, CodeCompletionContext Context,
1136                                   CodeCompletionResult *Results,
1137                                   unsigned NumResults) override {
1138
1139     // The Sema put the incomplete token we try to complete in here during
1140     // lexing, so we need to retrieve it here to know what we are completing.
1141     StringRef Filter = SemaRef.getPreprocessor().getCodeCompletionFilter();
1142
1143     // Iterate over all the results. Filter out results we don't want and
1144     // process the rest.
1145     for (unsigned I = 0; I != NumResults; ++I) {
1146       // Filter the results with the information from the Sema.
1147       if (!Filter.empty() && isResultFilteredOut(Filter, Results[I]))
1148         continue;
1149
1150       CodeCompletionResult &R = Results[I];
1151       std::optional<CompletionWithPriority> CompletionAndPriority =
1152           getCompletionForResult(R);
1153       if (!CompletionAndPriority)
1154         continue;
1155       m_completions.push_back(*CompletionAndPriority);
1156     }
1157   }
1158
1159   /// \param S the semantic-analyzer object for which code-completion is being
1160   /// done.
1161   ///
1162   /// \param CurrentArg the index of the current argument.
1163   ///
1164   /// \param Candidates an array of overload candidates.
1165   ///
1166   /// \param NumCandidates the number of overload candidates
1167   void ProcessOverloadCandidates(Sema &S, unsigned CurrentArg,
1168                                  OverloadCandidate *Candidates,
1169                                  unsigned NumCandidates,
1170                                  SourceLocation OpenParLoc,
1171                                  bool Braced) override {
1172     // At the moment we don't filter out any overloaded candidates.
1173   }
1174
1175   CodeCompletionAllocator &getAllocator() override {
1176     return m_info.getAllocator();
1177   }
1178
1179   CodeCompletionTUInfo &getCodeCompletionTUInfo() override { return m_info; }
1180 };
1181 } // namespace
1182
1183 bool ClangExpressionParser::Complete(CompletionRequest &request, unsigned line,
1184                                      unsigned pos, unsigned typed_pos) {
1185   DiagnosticManager mgr;
1186   // We need the raw user expression here because that's what the CodeComplete
1187   // class uses to provide completion suggestions.
1188   // However, the `Text` method only gives us the transformed expression here.
1189   // To actually get the raw user input here, we have to cast our expression to
1190   // the LLVMUserExpression which exposes the right API. This should never fail
1191   // as we always have a ClangUserExpression whenever we call this.
1192   ClangUserExpression *llvm_expr = cast<ClangUserExpression>(&m_expr);
1193   CodeComplete CC(m_compiler->getLangOpts(), llvm_expr->GetUserText(),
1194                   typed_pos);
1195   // We don't need a code generator for parsing.
1196   m_code_generator.reset();
1197   // Start parsing the expression with our custom code completion consumer.
1198   ParseInternal(mgr, &CC, line, pos);
1199   CC.GetCompletions(request);
1200   return true;
1201 }
1202
1203 unsigned ClangExpressionParser::Parse(DiagnosticManager &diagnostic_manager) {
1204   return ParseInternal(diagnostic_manager);
1205 }
1206
1207 unsigned
1208 ClangExpressionParser::ParseInternal(DiagnosticManager &diagnostic_manager,
1209                                      CodeCompleteConsumer *completion_consumer,
1210                                      unsigned completion_line,
1211                                      unsigned completion_column) {
1212   ClangDiagnosticManagerAdapter *adapter =
1213       static_cast<ClangDiagnosticManagerAdapter *>(
1214           m_compiler->getDiagnostics().getClient());
1215
1216   adapter->ResetManager(&diagnostic_manager);
1217
1218   const char *expr_text = m_expr.Text();
1219
1220   clang::SourceManager &source_mgr = m_compiler->getSourceManager();
1221   bool created_main_file = false;
1222
1223   // Clang wants to do completion on a real file known by Clang's file manager,
1224   // so we have to create one to make this work.
1225   // TODO: We probably could also simulate to Clang's file manager that there
1226   // is a real file that contains our code.
1227   bool should_create_file = completion_consumer != nullptr;
1228
1229   // We also want a real file on disk if we generate full debug info.
1230   should_create_file |= m_compiler->getCodeGenOpts().getDebugInfo() ==
1231                         codegenoptions::FullDebugInfo;
1232
1233   if (should_create_file) {
1234     int temp_fd = -1;
1235     llvm::SmallString<128> result_path;
1236     if (FileSpec tmpdir_file_spec = HostInfo::GetProcessTempDir()) {
1237       tmpdir_file_spec.AppendPathComponent("lldb-%%%%%%.expr");
1238       std::string temp_source_path = tmpdir_file_spec.GetPath();
1239       llvm::sys::fs::createUniqueFile(temp_source_path, temp_fd, result_path);
1240     } else {
1241       llvm::sys::fs::createTemporaryFile("lldb", "expr", temp_fd, result_path);
1242     }
1243
1244     if (temp_fd != -1) {
1245       lldb_private::NativeFile file(temp_fd, File::eOpenOptionWriteOnly, true);
1246       const size_t expr_text_len = strlen(expr_text);
1247       size_t bytes_written = expr_text_len;
1248       if (file.Write(expr_text, bytes_written).Success()) {
1249         if (bytes_written == expr_text_len) {
1250           file.Close();
1251           if (auto fileEntry = m_compiler->getFileManager().getOptionalFileRef(
1252                   result_path)) {
1253             source_mgr.setMainFileID(source_mgr.createFileID(
1254                 *fileEntry, SourceLocation(), SrcMgr::C_User));
1255             created_main_file = true;
1256           }
1257         }
1258       }
1259     }
1260   }
1261
1262   if (!created_main_file) {
1263     std::unique_ptr<MemoryBuffer> memory_buffer =
1264         MemoryBuffer::getMemBufferCopy(expr_text, m_filename);
1265     source_mgr.setMainFileID(source_mgr.createFileID(std::move(memory_buffer)));
1266   }
1267
1268   adapter->BeginSourceFile(m_compiler->getLangOpts(),
1269                            &m_compiler->getPreprocessor());
1270
1271   ClangExpressionHelper *type_system_helper =
1272       dyn_cast<ClangExpressionHelper>(m_expr.GetTypeSystemHelper());
1273
1274   // If we want to parse for code completion, we need to attach our code
1275   // completion consumer to the Sema and specify a completion position.
1276   // While parsing the Sema will call this consumer with the provided
1277   // completion suggestions.
1278   if (completion_consumer) {
1279     auto main_file =
1280         source_mgr.getFileEntryRefForID(source_mgr.getMainFileID());
1281     auto &PP = m_compiler->getPreprocessor();
1282     // Lines and columns start at 1 in Clang, but code completion positions are
1283     // indexed from 0, so we need to add 1 to the line and column here.
1284     ++completion_line;
1285     ++completion_column;
1286     PP.SetCodeCompletionPoint(*main_file, completion_line, completion_column);
1287   }
1288
1289   ASTConsumer *ast_transformer =
1290       type_system_helper->ASTTransformer(m_code_generator.get());
1291
1292   std::unique_ptr<clang::ASTConsumer> Consumer;
1293   if (ast_transformer) {
1294     Consumer = std::make_unique<ASTConsumerForwarder>(ast_transformer);
1295   } else if (m_code_generator) {
1296     Consumer = std::make_unique<ASTConsumerForwarder>(m_code_generator.get());
1297   } else {
1298     Consumer = std::make_unique<ASTConsumer>();
1299   }
1300
1301   clang::ASTContext &ast_context = m_compiler->getASTContext();
1302
1303   m_compiler->setSema(new Sema(m_compiler->getPreprocessor(), ast_context,
1304                                *Consumer, TU_Complete, completion_consumer));
1305   m_compiler->setASTConsumer(std::move(Consumer));
1306
1307   if (ast_context.getLangOpts().Modules) {
1308     m_compiler->createASTReader();
1309     m_ast_context->setSema(&m_compiler->getSema());
1310   }
1311
1312   ClangExpressionDeclMap *decl_map = type_system_helper->DeclMap();
1313   if (decl_map) {
1314     decl_map->InstallCodeGenerator(&m_compiler->getASTConsumer());
1315     decl_map->InstallDiagnosticManager(diagnostic_manager);
1316
1317     clang::ExternalASTSource *ast_source = decl_map->CreateProxy();
1318
1319     auto *ast_source_wrapper = new ExternalASTSourceWrapper(ast_source);
1320
1321     if (ast_context.getExternalSource()) {
1322       auto *module_wrapper =
1323           new ExternalASTSourceWrapper(ast_context.getExternalSource());
1324
1325       auto *multiplexer =
1326           new SemaSourceWithPriorities(module_wrapper, ast_source_wrapper);
1327
1328       ast_context.setExternalSource(multiplexer);
1329     } else {
1330       ast_context.setExternalSource(ast_source);
1331     }
1332     m_compiler->getSema().addExternalSource(ast_source_wrapper);
1333     decl_map->InstallASTContext(*m_ast_context);
1334   }
1335
1336   // Check that the ASTReader is properly attached to ASTContext and Sema.
1337   if (ast_context.getLangOpts().Modules) {
1338     assert(m_compiler->getASTContext().getExternalSource() &&
1339            "ASTContext doesn't know about the ASTReader?");
1340     assert(m_compiler->getSema().getExternalSource() &&
1341            "Sema doesn't know about the ASTReader?");
1342   }
1343
1344   {
1345     llvm::CrashRecoveryContextCleanupRegistrar<Sema> CleanupSema(
1346         &m_compiler->getSema());
1347     ParseAST(m_compiler->getSema(), false, false);
1348   }
1349
1350   // Make sure we have no pointer to the Sema we are about to destroy.
1351   if (ast_context.getLangOpts().Modules)
1352     m_ast_context->setSema(nullptr);
1353   // Destroy the Sema. This is necessary because we want to emulate the
1354   // original behavior of ParseAST (which also destroys the Sema after parsing).
1355   m_compiler->setSema(nullptr);
1356
1357   adapter->EndSourceFile();
1358
1359   unsigned num_errors = adapter->getNumErrors();
1360
1361   if (m_pp_callbacks && m_pp_callbacks->hasErrors()) {
1362     num_errors++;
1363     diagnostic_manager.PutString(lldb::eSeverityError,
1364                                  "while importing modules:");
1365     diagnostic_manager.AppendMessageToDiagnostic(
1366         m_pp_callbacks->getErrorString());
1367   }
1368
1369   if (!num_errors) {
1370     type_system_helper->CommitPersistentDecls();
1371   }
1372
1373   adapter->ResetManager();
1374
1375   return num_errors;
1376 }
1377
1378 /// Applies the given Fix-It hint to the given commit.
1379 static void ApplyFixIt(const FixItHint &fixit, clang::edit::Commit &commit) {
1380   // This is cobbed from clang::Rewrite::FixItRewriter.
1381   if (fixit.CodeToInsert.empty()) {
1382     if (fixit.InsertFromRange.isValid()) {
1383       commit.insertFromRange(fixit.RemoveRange.getBegin(),
1384                              fixit.InsertFromRange, /*afterToken=*/false,
1385                              fixit.BeforePreviousInsertions);
1386       return;
1387     }
1388     commit.remove(fixit.RemoveRange);
1389     return;
1390   }
1391   if (fixit.RemoveRange.isTokenRange() ||
1392       fixit.RemoveRange.getBegin() != fixit.RemoveRange.getEnd()) {
1393     commit.replace(fixit.RemoveRange, fixit.CodeToInsert);
1394     return;
1395   }
1396   commit.insert(fixit.RemoveRange.getBegin(), fixit.CodeToInsert,
1397                 /*afterToken=*/false, fixit.BeforePreviousInsertions);
1398 }
1399
1400 bool ClangExpressionParser::RewriteExpression(
1401     DiagnosticManager &diagnostic_manager) {
1402   clang::SourceManager &source_manager = m_compiler->getSourceManager();
1403   clang::edit::EditedSource editor(source_manager, m_compiler->getLangOpts(),
1404                                    nullptr);
1405   clang::edit::Commit commit(editor);
1406   clang::Rewriter rewriter(source_manager, m_compiler->getLangOpts());
1407
1408   class RewritesReceiver : public edit::EditsReceiver {
1409     Rewriter &rewrite;
1410
1411   public:
1412     RewritesReceiver(Rewriter &in_rewrite) : rewrite(in_rewrite) {}
1413
1414     void insert(SourceLocation loc, StringRef text) override {
1415       rewrite.InsertText(loc, text);
1416     }
1417     void replace(CharSourceRange range, StringRef text) override {
1418       rewrite.ReplaceText(range.getBegin(), rewrite.getRangeSize(range), text);
1419     }
1420   };
1421
1422   RewritesReceiver rewrites_receiver(rewriter);
1423
1424   const DiagnosticList &diagnostics = diagnostic_manager.Diagnostics();
1425   size_t num_diags = diagnostics.size();
1426   if (num_diags == 0)
1427     return false;
1428
1429   for (const auto &diag : diagnostic_manager.Diagnostics()) {
1430     const auto *diagnostic = llvm::dyn_cast<ClangDiagnostic>(diag.get());
1431     if (!diagnostic)
1432       continue;
1433     if (!diagnostic->HasFixIts())
1434       continue;
1435     for (const FixItHint &fixit : diagnostic->FixIts())
1436       ApplyFixIt(fixit, commit);
1437   }
1438
1439   // FIXME - do we want to try to propagate specific errors here?
1440   if (!commit.isCommitable())
1441     return false;
1442   else if (!editor.commit(commit))
1443     return false;
1444
1445   // Now play all the edits, and stash the result in the diagnostic manager.
1446   editor.applyRewrites(rewrites_receiver);
1447   RewriteBuffer &main_file_buffer =
1448       rewriter.getEditBuffer(source_manager.getMainFileID());
1449
1450   std::string fixed_expression;
1451   llvm::raw_string_ostream out_stream(fixed_expression);
1452
1453   main_file_buffer.write(out_stream);
1454   diagnostic_manager.SetFixedExpression(fixed_expression);
1455
1456   return true;
1457 }
1458
1459 static bool FindFunctionInModule(ConstString &mangled_name,
1460                                  llvm::Module *module, const char *orig_name) {
1461   for (const auto &func : module->getFunctionList()) {
1462     const StringRef &name = func.getName();
1463     if (name.contains(orig_name)) {
1464       mangled_name.SetString(name);
1465       return true;
1466     }
1467   }
1468
1469   return false;
1470 }
1471
1472 lldb_private::Status ClangExpressionParser::DoPrepareForExecution(
1473     lldb::addr_t &func_addr, lldb::addr_t &func_end,
1474     lldb::IRExecutionUnitSP &execution_unit_sp, ExecutionContext &exe_ctx,
1475     bool &can_interpret, ExecutionPolicy execution_policy) {
1476   func_addr = LLDB_INVALID_ADDRESS;
1477   func_end = LLDB_INVALID_ADDRESS;
1478   Log *log = GetLog(LLDBLog::Expressions);
1479
1480   lldb_private::Status err;
1481
1482   std::unique_ptr<llvm::Module> llvm_module_up(
1483       m_code_generator->ReleaseModule());
1484
1485   if (!llvm_module_up) {
1486     err = Status::FromErrorString("IR doesn't contain a module");
1487     return err;
1488   }
1489
1490   ConstString function_name;
1491
1492   if (execution_policy != eExecutionPolicyTopLevel) {
1493     // Find the actual name of the function (it's often mangled somehow)
1494
1495     if (!FindFunctionInModule(function_name, llvm_module_up.get(),
1496                               m_expr.FunctionName())) {
1497       err = Status::FromErrorStringWithFormat(
1498           "Couldn't find %s() in the module", m_expr.FunctionName());
1499       return err;
1500     } else {
1501       LLDB_LOGF(log, "Found function %s for %s", function_name.AsCString(),
1502                 m_expr.FunctionName());
1503     }
1504   }
1505
1506   SymbolContext sc;
1507
1508   if (lldb::StackFrameSP frame_sp = exe_ctx.GetFrameSP()) {
1509     sc = frame_sp->GetSymbolContext(lldb::eSymbolContextEverything);
1510   } else if (lldb::TargetSP target_sp = exe_ctx.GetTargetSP()) {
1511     sc.target_sp = target_sp;
1512   }
1513
1514   LLVMUserExpression::IRPasses custom_passes;
1515   {
1516     auto lang = m_expr.Language();
1517     LLDB_LOGF(log, "%s - Current expression language is %s\n", __FUNCTION__,
1518               lang.GetDescription().data());
1519     lldb::ProcessSP process_sp = exe_ctx.GetProcessSP();
1520     if (process_sp && lang != lldb::eLanguageTypeUnknown) {
1521       auto runtime = process_sp->GetLanguageRuntime(lang.AsLanguageType());
1522       if (runtime)
1523         runtime->GetIRPasses(custom_passes);
1524     }
1525   }
1526
1527   if (custom_passes.EarlyPasses) {
1528     LLDB_LOGF(log,
1529               "%s - Running Early IR Passes from LanguageRuntime on "
1530               "expression module '%s'",
1531               __FUNCTION__, m_expr.FunctionName());
1532
1533     custom_passes.EarlyPasses->run(*llvm_module_up);
1534   }
1535
1536   execution_unit_sp = std::make_shared<IRExecutionUnit>(
1537       m_llvm_context, // handed off here
1538       llvm_module_up, // handed off here
1539       function_name, exe_ctx.GetTargetSP(), sc,
1540       m_compiler->getTargetOpts().Features);
1541
1542   ClangExpressionHelper *type_system_helper =
1543       dyn_cast<ClangExpressionHelper>(m_expr.GetTypeSystemHelper());
1544   ClangExpressionDeclMap *decl_map =
1545       type_system_helper->DeclMap(); // result can be NULL
1546
1547   if (decl_map) {
1548     StreamString error_stream;
1549     IRForTarget ir_for_target(decl_map, m_expr.NeedsVariableResolution(),
1550                               *execution_unit_sp, error_stream,
1551                               function_name.AsCString());
1552
1553     if (!ir_for_target.runOnModule(*execution_unit_sp->GetModule())) {
1554       err = Status(error_stream.GetString().str());
1555       return err;
1556     }
1557
1558     Process *process = exe_ctx.GetProcessPtr();
1559
1560     if (execution_policy != eExecutionPolicyAlways &&
1561         execution_policy != eExecutionPolicyTopLevel) {
1562       lldb_private::Status interpret_error;
1563
1564       bool interpret_function_calls =
1565           !process ? false : process->CanInterpretFunctionCalls();
1566       can_interpret = IRInterpreter::CanInterpret(
1567           *execution_unit_sp->GetModule(), *execution_unit_sp->GetFunction(),
1568           interpret_error, interpret_function_calls);
1569
1570       if (!can_interpret && execution_policy == eExecutionPolicyNever) {
1571         err = Status::FromErrorStringWithFormat(
1572             "Can't evaluate the expression without a running target due to: %s",
1573             interpret_error.AsCString());
1574         return err;
1575       }
1576     }
1577
1578     if (!process && execution_policy == eExecutionPolicyAlways) {
1579       err = Status::FromErrorString(
1580           "Expression needed to run in the target, but the "
1581           "target can't be run");
1582       return err;
1583     }
1584
1585     if (!process && execution_policy == eExecutionPolicyTopLevel) {
1586       err = Status::FromErrorString(
1587           "Top-level code needs to be inserted into a runnable "
1588           "target, but the target can't be run");
1589       return err;
1590     }
1591
1592     if (execution_policy == eExecutionPolicyAlways ||
1593         (execution_policy != eExecutionPolicyTopLevel && !can_interpret)) {
1594       if (m_expr.NeedsValidation() && process) {
1595         if (!process->GetDynamicCheckers()) {
1596           ClangDynamicCheckerFunctions *dynamic_checkers =
1597               new ClangDynamicCheckerFunctions();
1598
1599           DiagnosticManager install_diags;
1600           if (Error Err = dynamic_checkers->Install(install_diags, exe_ctx))
1601             return Status::FromError(install_diags.GetAsError(
1602                 lldb::eExpressionSetupError, "couldn't install checkers:"));
1603
1604           process->SetDynamicCheckers(dynamic_checkers);
1605
1606           LLDB_LOGF(log, "== [ClangExpressionParser::PrepareForExecution] "
1607                          "Finished installing dynamic checkers ==");
1608         }
1609
1610         if (auto *checker_funcs = llvm::dyn_cast<ClangDynamicCheckerFunctions>(
1611                 process->GetDynamicCheckers())) {
1612           IRDynamicChecks ir_dynamic_checks(*checker_funcs,
1613                                             function_name.AsCString());
1614
1615           llvm::Module *module = execution_unit_sp->GetModule();
1616           if (!module || !ir_dynamic_checks.runOnModule(*module)) {
1617             err = Status::FromErrorString(
1618                 "Couldn't add dynamic checks to the expression");
1619             return err;
1620           }
1621
1622           if (custom_passes.LatePasses) {
1623             LLDB_LOGF(log,
1624                       "%s - Running Late IR Passes from LanguageRuntime on "
1625                       "expression module '%s'",
1626                       __FUNCTION__, m_expr.FunctionName());
1627
1628             custom_passes.LatePasses->run(*module);
1629           }
1630         }
1631       }
1632     }
1633
1634     if (execution_policy == eExecutionPolicyAlways ||
1635         execution_policy == eExecutionPolicyTopLevel || !can_interpret) {
1636       execution_unit_sp->GetRunnableInfo(err, func_addr, func_end);
1637     }
1638   } else {
1639     execution_unit_sp->GetRunnableInfo(err, func_addr, func_end);
1640   }
1641
1642   return err;
1643 }