import_cpp.cpp 79 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978
  1. // Part of the Carbon Language project, under the Apache License v2.0 with LLVM
  2. // Exceptions. See /LICENSE for license information.
  3. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
  4. #include "toolchain/check/import_cpp.h"
  5. #include <memory>
  6. #include <optional>
  7. #include <string>
  8. #include <tuple>
  9. #include <utility>
  10. #include "clang/AST/ASTContext.h"
  11. #include "clang/AST/RecordLayout.h"
  12. #include "clang/Basic/FileManager.h"
  13. #include "clang/Frontend/ASTUnit.h"
  14. #include "clang/Frontend/CompilerInstance.h"
  15. #include "clang/Frontend/CompilerInvocation.h"
  16. #include "clang/Frontend/TextDiagnostic.h"
  17. #include "clang/Lex/PreprocessorOptions.h"
  18. #include "clang/Sema/Lookup.h"
  19. #include "common/check.h"
  20. #include "common/ostream.h"
  21. #include "common/raw_string_ostream.h"
  22. #include "llvm/ADT/IntrusiveRefCntPtr.h"
  23. #include "llvm/ADT/StringRef.h"
  24. #include "llvm/Support/raw_ostream.h"
  25. #include "toolchain/base/kind_switch.h"
  26. #include "toolchain/check/class.h"
  27. #include "toolchain/check/context.h"
  28. #include "toolchain/check/convert.h"
  29. #include "toolchain/check/cpp_custom_type_mapping.h"
  30. #include "toolchain/check/cpp_thunk.h"
  31. #include "toolchain/check/diagnostic_helpers.h"
  32. #include "toolchain/check/eval.h"
  33. #include "toolchain/check/function.h"
  34. #include "toolchain/check/import.h"
  35. #include "toolchain/check/inst.h"
  36. #include "toolchain/check/literal.h"
  37. #include "toolchain/check/pattern.h"
  38. #include "toolchain/check/pattern_match.h"
  39. #include "toolchain/check/type.h"
  40. #include "toolchain/check/type_completion.h"
  41. #include "toolchain/diagnostics/diagnostic.h"
  42. #include "toolchain/diagnostics/diagnostic_emitter.h"
  43. #include "toolchain/diagnostics/format_providers.h"
  44. #include "toolchain/parse/node_ids.h"
  45. #include "toolchain/sem_ir/clang_decl.h"
  46. #include "toolchain/sem_ir/class.h"
  47. #include "toolchain/sem_ir/function.h"
  48. #include "toolchain/sem_ir/ids.h"
  49. #include "toolchain/sem_ir/inst.h"
  50. #include "toolchain/sem_ir/name_scope.h"
  51. #include "toolchain/sem_ir/typed_insts.h"
  52. namespace Carbon::Check {
  53. // Add a line marker directive pointing at the location of the `import Cpp`
  54. // declaration in the Carbon source file. This will cause Clang's diagnostics
  55. // machinery to track and report the location in Carbon code where the import
  56. // was written.
  57. static auto GenerateLineMarker(Context& context, llvm::raw_ostream& out,
  58. int line) {
  59. out << "# " << line << " \""
  60. << FormatEscaped(context.tokens().source().filename()) << "\"\n";
  61. }
  62. // Generates C++ file contents to #include all requested imports.
  63. static auto GenerateCppIncludesHeaderCode(
  64. Context& context, llvm::ArrayRef<Parse::Tree::PackagingNames> imports)
  65. -> std::string {
  66. std::string code;
  67. llvm::raw_string_ostream code_stream(code);
  68. for (const Parse::Tree::PackagingNames& import : imports) {
  69. if (import.inline_body_id.has_value()) {
  70. // Expand `import Cpp inline "code";` directly into the specified code.
  71. auto code_token = context.parse_tree().node_token(import.inline_body_id);
  72. // Compute the line number on which the C++ code starts. Usually the code
  73. // is specified as a block string literal and starts on the line after the
  74. // start of the string token.
  75. // TODO: Determine if this is a block string literal without calling
  76. // `GetTokenText`, which re-lexes the string.
  77. int line = context.tokens().GetLineNumber(code_token);
  78. if (context.tokens().GetTokenText(code_token).contains('\n')) {
  79. ++line;
  80. }
  81. GenerateLineMarker(context, code_stream, line);
  82. code_stream << context.string_literal_values().Get(
  83. context.tokens().GetStringLiteralValue(code_token))
  84. << "\n";
  85. // TODO: Inject a clang pragma here to produce an error if there are
  86. // unclosed scopes at the end of this inline C++ fragment.
  87. } else {
  88. // Translate `import Cpp library "foo.h";` into `#include "foo.h"`.
  89. GenerateLineMarker(context, code_stream,
  90. context.tokens().GetLineNumber(
  91. context.parse_tree().node_token(import.node_id)));
  92. code_stream << "#include \""
  93. << FormatEscaped(
  94. context.string_literal_values().Get(import.library_id))
  95. << "\"\n";
  96. }
  97. }
  98. // Inject a declaration of placement operator new, because the code we
  99. // generate in thunks depends on it for placement new expressions. Clang has
  100. // special-case logic for lowering a new-expression using this, so a
  101. // definition is not required.
  102. // TODO: This is a hack. We should be able to directly generate Clang AST to
  103. // construct objects in-place without this.
  104. // TODO: Once we can rely on libc++ being available, consider including
  105. // `<__new/placement_new_delete.h>` instead.
  106. code_stream << R"(# 1 "<carbon-internal>"
  107. #undef constexpr
  108. #if __cplusplus > 202302L
  109. constexpr
  110. #endif
  111. #undef void
  112. #undef operator
  113. #undef new
  114. void* operator new(__SIZE_TYPE__, void*)
  115. #if __cplusplus < 201103L
  116. #undef throw
  117. throw()
  118. #else
  119. #undef noexcept
  120. noexcept
  121. #endif
  122. ;
  123. )";
  124. return code;
  125. }
  126. // Adds the name to the scope with the given `access_kind` and `inst_id`.
  127. // `inst_id` must have a value.
  128. static auto AddNameToScope(Context& context, SemIR::NameScopeId scope_id,
  129. SemIR::NameId name_id, SemIR::AccessKind access_kind,
  130. SemIR::InstId inst_id) -> void {
  131. CARBON_CHECK(inst_id.has_value());
  132. context.name_scopes().Get(scope_id).AddRequired(
  133. {.name_id = name_id,
  134. .result = SemIR::ScopeLookupResult::MakeFound(inst_id, access_kind)});
  135. }
  136. // Maps a Clang name to a Carbon `NameId`.
  137. static auto AddIdentifierName(Context& context, llvm::StringRef name)
  138. -> SemIR::NameId {
  139. return SemIR::NameId::ForIdentifier(context.identifiers().Add(name));
  140. }
  141. // Adds the given source location and an `ImportIRInst` referring to it in
  142. // `ImportIRId::Cpp`.
  143. static auto AddImportIRInst(SemIR::File& file,
  144. clang::SourceLocation clang_source_loc)
  145. -> SemIR::ImportIRInstId {
  146. SemIR::ClangSourceLocId clang_source_loc_id =
  147. file.clang_source_locs().Add(clang_source_loc);
  148. return file.import_ir_insts().Add(SemIR::ImportIRInst(clang_source_loc_id));
  149. }
  150. namespace {
  151. // Used to convert Clang diagnostics to Carbon diagnostics.
  152. //
  153. // Handling of Clang notes is a little subtle: as far as Clang is concerned,
  154. // notes are separate diagnostics, not connected to the error or warning that
  155. // precedes them. But in Carbon's diagnostics system, notes are part of the
  156. // enclosing diagnostic. To handle this, we buffer Clang diagnostics until we
  157. // reach a point where we know we're not in the middle of a diagnostic, and then
  158. // emit a diagnostic along with all of its notes. This is triggered when adding
  159. // or removing a Carbon context note, which could otherwise get attached to the
  160. // wrong C++ diagnostics, and at the end of the Carbon program.
  161. class CarbonClangDiagnosticConsumer : public clang::DiagnosticConsumer {
  162. public:
  163. // Creates an instance with the location that triggers calling Clang. The
  164. // `context` is not stored here, and the diagnostics consumer is expected to
  165. // outlive it.
  166. explicit CarbonClangDiagnosticConsumer(
  167. Context& context, std::shared_ptr<clang::CompilerInvocation> invocation)
  168. : sem_ir_(&context.sem_ir()),
  169. emitter_(&context.emitter()),
  170. invocation_(std::move(invocation)) {
  171. emitter_->AddFlushFn([this] { EmitDiagnostics(); });
  172. }
  173. ~CarbonClangDiagnosticConsumer() override {
  174. // Do not inspect `emitter_` here; it's typically destroyed before the
  175. // consumer is.
  176. // TODO: If Clang produces diagnostics after check finishes, they'll get
  177. // added to the list of pending diagnostics and never emitted.
  178. CARBON_CHECK(diagnostic_infos_.empty(),
  179. "Missing flush before destroying diagnostic consumer");
  180. }
  181. // Generates a Carbon warning for each Clang warning and a Carbon error for
  182. // each Clang error or fatal.
  183. auto HandleDiagnostic(clang::DiagnosticsEngine::Level diag_level,
  184. const clang::Diagnostic& info) -> void override {
  185. DiagnosticConsumer::HandleDiagnostic(diag_level, info);
  186. SemIR::ImportIRInstId clang_import_ir_inst_id =
  187. AddImportIRInst(*sem_ir_, info.getLocation());
  188. llvm::SmallString<256> message;
  189. info.FormatDiagnostic(message);
  190. // Render a code snippet including any highlighted ranges and fixit hints.
  191. // TODO: Also include the #include stack and macro expansion stack in the
  192. // diagnostic output in some way.
  193. RawStringOstream snippet_stream;
  194. if (!info.hasSourceManager()) {
  195. // If we don't have a source manager, this is an error from early in the
  196. // frontend. Don't produce a snippet.
  197. CARBON_CHECK(info.getLocation().isInvalid());
  198. } else {
  199. CodeContextRenderer(snippet_stream, invocation_->getLangOpts(),
  200. invocation_->getDiagnosticOpts())
  201. .emitDiagnostic(
  202. clang::FullSourceLoc(info.getLocation(), info.getSourceManager()),
  203. diag_level, message, info.getRanges(), info.getFixItHints());
  204. }
  205. diagnostic_infos_.push_back({.level = diag_level,
  206. .import_ir_inst_id = clang_import_ir_inst_id,
  207. .message = message.str().str(),
  208. .snippet = snippet_stream.TakeStr()});
  209. }
  210. // Returns the diagnostic to use for a given Clang diagnostic level.
  211. static auto GetDiagnostic(clang::DiagnosticsEngine::Level level)
  212. -> const Diagnostics::DiagnosticBase<std::string>& {
  213. switch (level) {
  214. case clang::DiagnosticsEngine::Ignored: {
  215. CARBON_FATAL("Emitting an ignored diagnostic");
  216. break;
  217. }
  218. case clang::DiagnosticsEngine::Note: {
  219. CARBON_DIAGNOSTIC(CppInteropParseNote, Note, "{0}", std::string);
  220. return CppInteropParseNote;
  221. }
  222. case clang::DiagnosticsEngine::Remark:
  223. case clang::DiagnosticsEngine::Warning: {
  224. // TODO: Add a distinct Remark level to Carbon diagnostics, and stop
  225. // mapping remarks to warnings.
  226. CARBON_DIAGNOSTIC(CppInteropParseWarning, Warning, "{0}", std::string);
  227. return CppInteropParseWarning;
  228. }
  229. case clang::DiagnosticsEngine::Error:
  230. case clang::DiagnosticsEngine::Fatal: {
  231. CARBON_DIAGNOSTIC(CppInteropParseError, Error, "{0}", std::string);
  232. return CppInteropParseError;
  233. }
  234. }
  235. }
  236. // Outputs Carbon diagnostics based on the collected Clang diagnostics. Must
  237. // be called after the AST is set in the context.
  238. auto EmitDiagnostics() -> void {
  239. CARBON_CHECK(sem_ir_->clang_ast_unit(),
  240. "Attempted to emit diagnostics before the AST Unit is loaded");
  241. for (size_t i = 0; i != diagnostic_infos_.size(); ++i) {
  242. const ClangDiagnosticInfo& info = diagnostic_infos_[i];
  243. auto builder = emitter_->Build(SemIR::LocId(info.import_ir_inst_id),
  244. GetDiagnostic(info.level), info.message);
  245. builder.OverrideSnippet(info.snippet);
  246. for (; i + 1 < diagnostic_infos_.size() &&
  247. diagnostic_infos_[i + 1].level == clang::DiagnosticsEngine::Note;
  248. ++i) {
  249. const ClangDiagnosticInfo& note_info = diagnostic_infos_[i + 1];
  250. builder
  251. .Note(SemIR::LocId(note_info.import_ir_inst_id),
  252. GetDiagnostic(note_info.level), note_info.message)
  253. .OverrideSnippet(note_info.snippet);
  254. }
  255. // TODO: This will apply all current Carbon annotation functions. We
  256. // should instead track how Clang's context notes and Carbon's annotation
  257. // functions are interleaved, and interleave the notes in the same order.
  258. builder.Emit();
  259. }
  260. diagnostic_infos_.clear();
  261. }
  262. private:
  263. // A diagnostics renderer based on clang's TextDiagnostic that captures just
  264. // the code context (the snippet).
  265. class CodeContextRenderer : public clang::TextDiagnostic {
  266. public:
  267. using TextDiagnostic::TextDiagnostic;
  268. void emitDiagnosticMessage(
  269. clang::FullSourceLoc /*loc*/, clang::PresumedLoc /*ploc*/,
  270. clang::DiagnosticsEngine::Level /*level*/, llvm::StringRef /*message*/,
  271. llvm::ArrayRef<clang::CharSourceRange> /*ranges*/,
  272. clang::DiagOrStoredDiag /*info*/) override {}
  273. void emitDiagnosticLoc(
  274. clang::FullSourceLoc /*loc*/, clang::PresumedLoc /*ploc*/,
  275. clang::DiagnosticsEngine::Level /*level*/,
  276. llvm::ArrayRef<clang::CharSourceRange> /*ranges*/) override {}
  277. // emitCodeContext is inherited from clang::TextDiagnostic.
  278. void emitIncludeLocation(clang::FullSourceLoc /*loc*/,
  279. clang::PresumedLoc /*ploc*/) override {}
  280. void emitImportLocation(clang::FullSourceLoc /*loc*/,
  281. clang::PresumedLoc /*ploc*/,
  282. llvm::StringRef /*module_name*/) override {}
  283. void emitBuildingModuleLocation(clang::FullSourceLoc /*loc*/,
  284. clang::PresumedLoc /*ploc*/,
  285. llvm::StringRef /*module_name*/) override {}
  286. // beginDiagnostic and endDiagnostic are inherited from
  287. // clang::TextDiagnostic in case it wants to do any setup / teardown work.
  288. };
  289. // Information on a Clang diagnostic that can be converted to a Carbon
  290. // diagnostic.
  291. struct ClangDiagnosticInfo {
  292. // The Clang diagnostic level.
  293. clang::DiagnosticsEngine::Level level;
  294. // The ID of the ImportIR instruction referring to the Clang source
  295. // location.
  296. SemIR::ImportIRInstId import_ir_inst_id;
  297. // The Clang diagnostic textual message.
  298. std::string message;
  299. // The code snippet produced by clang.
  300. std::string snippet;
  301. };
  302. // The Carbon file that this C++ compilation is attached to.
  303. SemIR::File* sem_ir_;
  304. // The diagnostic emitter that we're emitting diagnostics into.
  305. DiagnosticEmitterBase* emitter_;
  306. // The compiler invocation that is producing the diagnostics.
  307. std::shared_ptr<clang::CompilerInvocation> invocation_;
  308. // Collects the information for all Clang diagnostics to be converted to
  309. // Carbon diagnostics after the context has been initialized with the Clang
  310. // AST.
  311. llvm::SmallVector<ClangDiagnosticInfo> diagnostic_infos_;
  312. };
  313. // A wrapper around a clang::CompilerInvocation that allows us to make a shallow
  314. // copy of most of the invocation and only make a deep copy of the parts that we
  315. // want to change.
  316. //
  317. // clang::CowCompilerInvocation almost allows this, but doesn't derive from
  318. // CompilerInvocation or support shallow copies from a CompilerInvocation, so is
  319. // not useful to us as we can't build an ASTUnit from it.
  320. class ShallowCopyCompilerInvocation : public clang::CompilerInvocation {
  321. public:
  322. explicit ShallowCopyCompilerInvocation(
  323. const clang::CompilerInvocation& invocation) {
  324. shallow_copy_assign(invocation);
  325. // The preprocessor options are modified to hold a replacement includes
  326. // buffer, so make our own version of those options.
  327. PPOpts = std::make_shared<clang::PreprocessorOptions>(*PPOpts);
  328. }
  329. };
  330. } // namespace
  331. // Returns an AST for the C++ imports and a bool that represents whether
  332. // compilation errors where encountered or the generated AST is null due to an
  333. // error. Sets the AST in the context's `sem_ir`.
  334. // TODO: Consider to always have a (non-null) AST.
  335. static auto GenerateAst(
  336. Context& context, llvm::ArrayRef<Parse::Tree::PackagingNames> imports,
  337. llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs,
  338. std::shared_ptr<clang::CompilerInvocation> base_invocation)
  339. -> std::pair<std::unique_ptr<clang::ASTUnit>, bool> {
  340. auto invocation =
  341. std::make_shared<ShallowCopyCompilerInvocation>(*base_invocation);
  342. // Build a diagnostics engine.
  343. llvm::IntrusiveRefCntPtr<clang::DiagnosticsEngine> diags(
  344. clang::CompilerInstance::createDiagnostics(
  345. *fs, invocation->getDiagnosticOpts(),
  346. new CarbonClangDiagnosticConsumer(context, invocation),
  347. /*ShouldOwnClient=*/true));
  348. // Extract the input from the frontend invocation and make sure it makes
  349. // sense.
  350. const auto& inputs = invocation->getFrontendOpts().Inputs;
  351. CARBON_CHECK(inputs.size() == 1 &&
  352. inputs[0].getKind().getLanguage() == clang::Language::CXX &&
  353. inputs[0].getKind().getFormat() == clang::InputKind::Source);
  354. llvm::StringRef file_name = inputs[0].getFile();
  355. // Remap the imports file name to the corresponding `#include`s.
  356. // TODO: Modify the frontend options to specify this memory buffer as input
  357. // instead of remapping the file.
  358. std::string includes = GenerateCppIncludesHeaderCode(context, imports);
  359. auto includes_buffer =
  360. llvm::MemoryBuffer::getMemBufferCopy(includes, file_name);
  361. invocation->getPreprocessorOpts().addRemappedFile(file_name,
  362. includes_buffer.release());
  363. clang::DiagnosticErrorTrap trap(*diags);
  364. // Create the AST unit.
  365. auto ast = clang::ASTUnit::LoadFromCompilerInvocation(
  366. invocation, std::make_shared<clang::PCHContainerOperations>(), nullptr,
  367. diags, new clang::FileManager(invocation->getFileSystemOpts(), fs));
  368. // Attach the AST to SemIR. This needs to be done before we can emit any
  369. // diagnostics, so their locations can be properly interpreted by our
  370. // diagnostics machinery.
  371. context.sem_ir().set_clang_ast_unit(ast.get());
  372. // Emit any diagnostics we queued up while building the AST.
  373. context.emitter().Flush();
  374. return {std::move(ast), !ast || trap.hasErrorOccurred()};
  375. }
  376. // Adds a namespace for the `Cpp` import and returns its `NameScopeId`.
  377. static auto AddNamespace(Context& context, PackageNameId cpp_package_id,
  378. llvm::ArrayRef<Parse::Tree::PackagingNames> imports)
  379. -> SemIR::NameScopeId {
  380. auto& import_cpps = context.sem_ir().import_cpps();
  381. import_cpps.Reserve(imports.size());
  382. for (const Parse::Tree::PackagingNames& import : imports) {
  383. import_cpps.Add({.node_id = context.parse_tree().As<Parse::ImportDeclId>(
  384. import.node_id),
  385. .library_id = import.library_id});
  386. }
  387. return AddImportNamespaceToScope(
  388. context,
  389. GetSingletonType(context, SemIR::NamespaceType::TypeInstId),
  390. SemIR::NameId::ForPackageName(cpp_package_id),
  391. SemIR::NameScopeId::Package,
  392. /*diagnose_duplicate_namespace=*/false,
  393. [&]() {
  394. return AddInst<SemIR::ImportCppDecl>(
  395. context,
  396. context.parse_tree().As<Parse::ImportDeclId>(
  397. imports.front().node_id),
  398. {});
  399. })
  400. .add_result.name_scope_id;
  401. }
  402. auto ImportCppFiles(Context& context,
  403. llvm::ArrayRef<Parse::Tree::PackagingNames> imports,
  404. llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs,
  405. std::shared_ptr<clang::CompilerInvocation> invocation)
  406. -> std::unique_ptr<clang::ASTUnit> {
  407. if (imports.empty()) {
  408. return nullptr;
  409. }
  410. CARBON_CHECK(!context.sem_ir().clang_ast_unit());
  411. PackageNameId package_id = imports.front().package_id;
  412. CARBON_CHECK(
  413. llvm::all_of(imports, [&](const Parse::Tree::PackagingNames& import) {
  414. return import.package_id == package_id;
  415. }));
  416. auto name_scope_id = AddNamespace(context, package_id, imports);
  417. auto [generated_ast, ast_has_error] =
  418. GenerateAst(context, imports, fs, std::move(invocation));
  419. SemIR::NameScope& name_scope = context.name_scopes().Get(name_scope_id);
  420. name_scope.set_is_closed_import(true);
  421. name_scope.set_clang_decl_context_id(context.sem_ir().clang_decls().Add(
  422. {.decl = generated_ast->getASTContext().getTranslationUnitDecl(),
  423. .inst_id = name_scope.inst_id()}));
  424. if (ast_has_error) {
  425. name_scope.set_has_error();
  426. }
  427. return std::move(generated_ast);
  428. }
  429. // Looks up the given name in the Clang AST in a specific scope. Returns the
  430. // lookup result if lookup was successful.
  431. static auto ClangLookupName(Context& context, SemIR::NameScopeId scope_id,
  432. SemIR::NameId name_id)
  433. -> std::optional<clang::LookupResult> {
  434. std::optional<llvm::StringRef> name =
  435. context.names().GetAsStringIfIdentifier(name_id);
  436. if (!name) {
  437. // Special names never exist in C++ code.
  438. return std::nullopt;
  439. }
  440. clang::ASTUnit* ast = context.sem_ir().clang_ast_unit();
  441. CARBON_CHECK(ast);
  442. clang::Sema& sema = ast->getSema();
  443. // TODO: Map the LocId of the lookup to a clang SourceLocation and provide it
  444. // here so that clang's diagnostics can point into the carbon code that uses
  445. // the name.
  446. clang::LookupResult lookup(
  447. sema,
  448. clang::DeclarationNameInfo(
  449. clang::DeclarationName(
  450. sema.getPreprocessor().getIdentifierInfo(*name)),
  451. clang::SourceLocation()),
  452. clang::Sema::LookupNameKind::LookupOrdinaryName);
  453. auto scope_clang_decl_context_id =
  454. context.name_scopes().Get(scope_id).clang_decl_context_id();
  455. bool found = sema.LookupQualifiedName(
  456. lookup, dyn_cast<clang::DeclContext>(context.sem_ir()
  457. .clang_decls()
  458. .Get(scope_clang_decl_context_id)
  459. .decl));
  460. if (!found) {
  461. return std::nullopt;
  462. }
  463. return lookup;
  464. }
  465. // Looks up for constructors in the class scope and returns the lookup result.
  466. static auto ClangConstructorLookup(const Context& context,
  467. SemIR::NameScopeId scope_id)
  468. -> clang::DeclContextLookupResult {
  469. const SemIR::NameScope& scope = context.sem_ir().name_scopes().Get(scope_id);
  470. clang::Sema& sema = context.sem_ir().clang_ast_unit()->getSema();
  471. clang::Decl* decl =
  472. context.sem_ir().clang_decls().Get(scope.clang_decl_context_id()).decl;
  473. return sema.LookupConstructors(cast<clang::CXXRecordDecl>(decl));
  474. }
  475. // Returns true if the given Clang declaration is the implicit injected class
  476. // name within the class.
  477. static auto IsDeclInjectedClassName(const Context& context,
  478. SemIR::NameScopeId scope_id,
  479. SemIR::NameId name_id,
  480. const clang::NamedDecl* named_decl)
  481. -> bool {
  482. if (!named_decl->isImplicit()) {
  483. return false;
  484. }
  485. const auto* record_decl = dyn_cast<clang::CXXRecordDecl>(named_decl);
  486. if (!record_decl) {
  487. return false;
  488. }
  489. const SemIR::ClangDecl& clang_decl = context.sem_ir().clang_decls().Get(
  490. context.sem_ir().name_scopes().Get(scope_id).clang_decl_context_id());
  491. const auto* scope_record_decl = cast<clang::CXXRecordDecl>(clang_decl.decl);
  492. const clang::ASTContext& ast_context =
  493. context.sem_ir().clang_ast_unit()->getASTContext();
  494. CARBON_CHECK(
  495. ast_context.getCanonicalType(
  496. ast_context.getRecordType(scope_record_decl)) ==
  497. ast_context.getCanonicalType(ast_context.getRecordType(record_decl)));
  498. auto class_decl =
  499. context.sem_ir().insts().GetAs<SemIR::ClassDecl>(clang_decl.inst_id);
  500. CARBON_CHECK(name_id ==
  501. context.sem_ir().classes().Get(class_decl.class_id).name_id);
  502. return true;
  503. }
  504. // Looks up the given name in the Clang AST in a specific scope, and returns the
  505. // found declaration and its access. If the found declaration is the injected
  506. // class name, looks up constructors instead. If not found, returns `nullopt`.
  507. // If there's not a single result, returns `nullptr` and default access.
  508. // Otherwise, returns the single declaration and its access.
  509. static auto ClangLookup(Context& context, SemIR::LocId loc_id,
  510. SemIR::NameScopeId scope_id, SemIR::NameId name_id)
  511. -> std::optional<std::tuple<clang::NamedDecl*, clang::AccessSpecifier>> {
  512. auto lookup = ClangLookupName(context, scope_id, name_id);
  513. if (!lookup) {
  514. return std::nullopt;
  515. }
  516. std::tuple<clang::NamedDecl*, clang::AccessSpecifier> result{
  517. nullptr, clang::AccessSpecifier::AS_none};
  518. // Access checks are performed separately by the Carbon name lookup logic.
  519. lookup->suppressAccessDiagnostics();
  520. if (!lookup->isSingleResult()) {
  521. // Clang will diagnose ambiguous lookup results for us.
  522. if (!lookup->isAmbiguous()) {
  523. context.TODO(loc_id,
  524. llvm::formatv("Unsupported: Lookup succeeded but couldn't "
  525. "find a single result; LookupResultKind: {0}",
  526. static_cast<int>(lookup->getResultKind())));
  527. }
  528. return result;
  529. }
  530. if (!IsDeclInjectedClassName(context, scope_id, name_id,
  531. lookup->getFoundDecl())) {
  532. result = {lookup->getFoundDecl(), lookup->begin().getAccess()};
  533. return result;
  534. }
  535. clang::DeclContextLookupResult constructors_lookup =
  536. ClangConstructorLookup(context, scope_id);
  537. llvm::SmallVector<clang::CXXConstructorDecl*> constructors;
  538. for (clang::Decl* decl : constructors_lookup) {
  539. auto* constructor = cast<clang::CXXConstructorDecl>(decl);
  540. if (constructor->isDeleted() || constructor->isCopyOrMoveConstructor()) {
  541. continue;
  542. }
  543. constructors.push_back(constructor);
  544. }
  545. if (constructors.size() != 1) {
  546. context.TODO(
  547. loc_id,
  548. llvm::formatv("Unsupported: Constructors lookup succeeded but couldn't "
  549. "find a single result; Found {0} constructors",
  550. constructors.size()));
  551. return result;
  552. }
  553. result = {constructors[0], constructors[0]->getAccess()};
  554. return result;
  555. }
  556. // Returns whether `decl` already mapped to an instruction.
  557. static auto IsClangDeclImported(const Context& context, clang::Decl* decl)
  558. -> bool {
  559. return context.sem_ir()
  560. .clang_decls()
  561. .Lookup(decl->getCanonicalDecl())
  562. .has_value();
  563. }
  564. // If `decl` already mapped to an instruction, returns that instruction.
  565. // Otherwise returns `None`.
  566. static auto LookupClangDeclInstId(const Context& context, clang::Decl* decl)
  567. -> SemIR::InstId {
  568. const auto& clang_decls = context.sem_ir().clang_decls();
  569. if (auto context_clang_decl_id = clang_decls.Lookup(decl->getCanonicalDecl());
  570. context_clang_decl_id.has_value()) {
  571. return clang_decls.Get(context_clang_decl_id).inst_id;
  572. }
  573. return SemIR::InstId::None;
  574. }
  575. // Returns the parent of the given declaration. Skips declaration types we
  576. // ignore.
  577. static auto GetParentDecl(clang::Decl* clang_decl) -> clang::Decl* {
  578. auto* parent_dc = clang_decl->getDeclContext();
  579. while (!parent_dc->isLookupContext()) {
  580. parent_dc = parent_dc->getParent();
  581. }
  582. return cast<clang::Decl>(parent_dc);
  583. }
  584. // Returns the given declaration's parent scope. Assumes the parent declaration
  585. // was already imported.
  586. static auto GetParentNameScopeId(Context& context, clang::Decl* clang_decl)
  587. -> SemIR::NameScopeId {
  588. SemIR::InstId parent_inst_id =
  589. LookupClangDeclInstId(context, GetParentDecl(clang_decl));
  590. CARBON_CHECK(parent_inst_id.has_value());
  591. CARBON_KIND_SWITCH(context.insts().Get(parent_inst_id)) {
  592. case CARBON_KIND(SemIR::ClassDecl class_decl): {
  593. return context.classes().Get(class_decl.class_id).scope_id;
  594. }
  595. case CARBON_KIND(SemIR::InterfaceDecl interface_decl): {
  596. return context.interfaces().Get(interface_decl.interface_id).scope_id;
  597. }
  598. case CARBON_KIND(SemIR::Namespace namespace_inst): {
  599. return namespace_inst.name_scope_id;
  600. }
  601. default: {
  602. CARBON_FATAL("Unexpected parent instruction kind");
  603. }
  604. }
  605. }
  606. // Imports a namespace declaration from Clang to Carbon. If successful, returns
  607. // the new Carbon namespace declaration `InstId`. If the declaration was already
  608. // imported, returns the mapped instruction.
  609. static auto ImportNamespaceDecl(Context& context,
  610. clang::NamespaceDecl* clang_decl)
  611. -> SemIR::InstId {
  612. // Check if the declaration is already mapped.
  613. if (SemIR::InstId existing_inst_id =
  614. LookupClangDeclInstId(context, clang_decl);
  615. existing_inst_id.has_value()) {
  616. return existing_inst_id;
  617. }
  618. auto result = AddImportNamespace(
  619. context, GetSingletonType(context, SemIR::NamespaceType::TypeInstId),
  620. AddIdentifierName(context, clang_decl->getName()),
  621. GetParentNameScopeId(context, clang_decl),
  622. /*import_id=*/SemIR::InstId::None);
  623. context.name_scopes()
  624. .Get(result.name_scope_id)
  625. .set_clang_decl_context_id(context.sem_ir().clang_decls().Add(
  626. {.decl = clang_decl->getCanonicalDecl(), .inst_id = result.inst_id}));
  627. return result.inst_id;
  628. }
  629. static auto ImportTypeAndDependencies(Context& context, SemIR::LocId loc_id,
  630. clang::QualType type) -> TypeExpr;
  631. // Creates a class declaration for the given class name in the given scope.
  632. // Returns the `InstId` for the declaration.
  633. static auto BuildClassDecl(Context& context,
  634. SemIR::ImportIRInstId import_ir_inst_id,
  635. SemIR::NameScopeId parent_scope_id,
  636. SemIR::NameId name_id)
  637. -> std::tuple<SemIR::ClassId, SemIR::TypeInstId> {
  638. // Add the class declaration.
  639. auto class_decl = SemIR::ClassDecl{.type_id = SemIR::TypeType::TypeId,
  640. .class_id = SemIR::ClassId::None,
  641. .decl_block_id = SemIR::InstBlockId::None};
  642. auto class_decl_id = AddPlaceholderInstInNoBlock(
  643. context,
  644. SemIR::LocIdAndInst::UncheckedLoc(import_ir_inst_id, class_decl));
  645. context.imports().push_back(class_decl_id);
  646. SemIR::Class class_info = {
  647. {.name_id = name_id,
  648. .parent_scope_id = parent_scope_id,
  649. .generic_id = SemIR::GenericId::None,
  650. .first_param_node_id = Parse::NodeId::None,
  651. .last_param_node_id = Parse::NodeId::None,
  652. .pattern_block_id = SemIR::InstBlockId::None,
  653. .implicit_param_patterns_id = SemIR::InstBlockId::None,
  654. .param_patterns_id = SemIR::InstBlockId::None,
  655. .is_extern = false,
  656. .extern_library_id = SemIR::LibraryNameId::None,
  657. .non_owning_decl_id = SemIR::InstId::None,
  658. .first_owning_decl_id = class_decl_id},
  659. {// `.self_type_id` depends on the ClassType, so is set below.
  660. .self_type_id = SemIR::TypeId::None,
  661. // TODO: Support Dynamic classes.
  662. // TODO: Support Final classes.
  663. .inheritance_kind = SemIR::Class::Base}};
  664. class_decl.class_id = context.classes().Add(class_info);
  665. // Write the class ID into the ClassDecl.
  666. ReplaceInstBeforeConstantUse(context, class_decl_id, class_decl);
  667. SetClassSelfType(context, class_decl.class_id);
  668. return {class_decl.class_id, context.types().GetAsTypeInstId(class_decl_id)};
  669. }
  670. // Imports a tag declaration from Clang to Carbon. This covers classes (which
  671. // includes structs and unions) as well as enums. If successful, returns the new
  672. // Carbon class declaration `InstId`.
  673. static auto ImportTagDecl(Context& context, clang::TagDecl* clang_decl)
  674. -> SemIR::InstId {
  675. auto import_ir_inst_id =
  676. AddImportIRInst(context.sem_ir(), clang_decl->getLocation());
  677. auto [class_id, class_inst_id] = BuildClassDecl(
  678. context, import_ir_inst_id, GetParentNameScopeId(context, clang_decl),
  679. AddIdentifierName(context, clang_decl->getName()));
  680. // TODO: The caller does the same lookup. Avoid doing it twice.
  681. auto clang_decl_id = context.sem_ir().clang_decls().Add(
  682. {.decl = clang_decl->getCanonicalDecl(), .inst_id = class_inst_id});
  683. // Name lookup into the Carbon class looks in the C++ class definition.
  684. auto& class_info = context.classes().Get(class_id);
  685. class_info.scope_id = context.name_scopes().Add(
  686. class_inst_id, SemIR::NameId::None, class_info.parent_scope_id);
  687. context.name_scopes()
  688. .Get(class_info.scope_id)
  689. .set_clang_decl_context_id(clang_decl_id);
  690. return class_inst_id;
  691. }
  692. // Determines the Carbon inheritance kind to use for a C++ class definition.
  693. static auto GetInheritanceKind(clang::CXXRecordDecl* class_def)
  694. -> SemIR::Class::InheritanceKind {
  695. if (class_def->isUnion()) {
  696. // Treat all unions as final classes to match their C++ semantics. While we
  697. // could support this, the author of a C++ union has no way to mark their
  698. // type as `final` to prevent it, and so we assume the intent was to
  699. // disallow inheritance.
  700. return SemIR::Class::Final;
  701. }
  702. if (class_def->hasAttr<clang::FinalAttr>()) {
  703. // The class is final in C++; don't allow Carbon types to derive from it.
  704. // Note that such a type might also be abstract in C++; we treat final as
  705. // taking precedence.
  706. //
  707. // We could also treat classes with a final destructor as being final, as
  708. // Clang does when determining whether a class is "effectively final", but
  709. // to keep our rules simpler we do not.
  710. return SemIR::Class::Final;
  711. }
  712. if (class_def->isAbstract()) {
  713. // If the class has any abstract members, it's abstract.
  714. return SemIR::Class::Abstract;
  715. }
  716. // Allow inheritance from any other C++ class type.
  717. return SemIR::Class::Base;
  718. }
  719. // Checks that the specified finished class definition is valid and builds and
  720. // returns a corresponding complete type witness instruction.
  721. static auto ImportClassObjectRepr(Context& context, SemIR::ClassId class_id,
  722. SemIR::ImportIRInstId import_ir_inst_id,
  723. SemIR::TypeInstId class_type_inst_id,
  724. const clang::CXXRecordDecl* clang_def)
  725. -> SemIR::TypeInstId {
  726. // For now, if the class is empty, produce an empty struct as the object
  727. // representation. This allows our tests to continue to pass while we don't
  728. // properly support initializing imported C++ classes.
  729. // TODO: Remove this.
  730. if (clang_def->isEmpty() && !clang_def->getNumBases()) {
  731. return context.types().GetAsTypeInstId(AddInst(
  732. context,
  733. MakeImportedLocIdAndInst(
  734. context, import_ir_inst_id,
  735. SemIR::StructType{.type_id = SemIR::TypeType::TypeId,
  736. .fields_id = SemIR::StructTypeFieldsId::Empty})));
  737. }
  738. const auto& clang_layout =
  739. context.ast_context().getASTRecordLayout(clang_def);
  740. llvm::SmallVector<uint64_t> layout;
  741. llvm::SmallVector<SemIR::StructTypeField> fields;
  742. static_assert(SemIR::CustomLayoutId::SizeIndex == 0);
  743. layout.push_back(clang_layout.getSize().getQuantity());
  744. static_assert(SemIR::CustomLayoutId::AlignIndex == 1);
  745. layout.push_back(clang_layout.getAlignment().getQuantity());
  746. static_assert(SemIR::CustomLayoutId::FirstFieldIndex == 2);
  747. // TODO: Import vptr(s).
  748. // Import bases.
  749. for (const auto& base : clang_def->bases()) {
  750. CARBON_CHECK(!base.isVirtual(),
  751. "Should not import definition for class with a virtual base");
  752. auto [base_type_inst_id, base_type_id] =
  753. ImportTypeAndDependencies(context, import_ir_inst_id, base.getType());
  754. if (!base_type_id.has_value()) {
  755. // TODO: If the base class's type can't be mapped, skip it.
  756. continue;
  757. }
  758. auto base_decl_id = AddInst(
  759. context,
  760. MakeImportedLocIdAndInst(
  761. context, import_ir_inst_id,
  762. SemIR::BaseDecl{.type_id = GetUnboundElementType(
  763. context, class_type_inst_id, base_type_inst_id),
  764. .base_type_inst_id = base_type_inst_id,
  765. .index = SemIR::ElementIndex(fields.size())}));
  766. // If there's exactly one base class, treat it as a Carbon base class too.
  767. // TODO: Improve handling for the case where the class has multiple base
  768. // classes.
  769. if (clang_def->getNumBases() == 1) {
  770. auto& class_info = context.classes().Get(class_id);
  771. CARBON_CHECK(!class_info.base_id.has_value());
  772. class_info.base_id = base_decl_id;
  773. }
  774. auto* base_class = base.getType()->getAsCXXRecordDecl();
  775. CARBON_CHECK(base_class, "Base class {0} is not a class",
  776. base.getType().getAsString());
  777. auto base_offset = base.isVirtual()
  778. ? clang_layout.getVBaseClassOffset(base_class)
  779. : clang_layout.getBaseClassOffset(base_class);
  780. layout.push_back(base_offset.getQuantity());
  781. fields.push_back(
  782. {.name_id = SemIR::NameId::Base, .type_inst_id = base_type_inst_id});
  783. }
  784. // Import fields.
  785. for (auto* decl : clang_def->decls()) {
  786. auto* field = dyn_cast<clang::FieldDecl>(decl);
  787. // Track the chain of fields from the class to this field. This chain is
  788. // only one element long unless the field is a member of an anonymous struct
  789. // or union.
  790. clang::NamedDecl* single_field_chain[1] = {field};
  791. llvm::ArrayRef<clang::NamedDecl*> chain = single_field_chain;
  792. // If this isn't a field, it might be an indirect field in an anonymous
  793. // struct or union.
  794. if (!field) {
  795. auto* indirect_field = dyn_cast<clang::IndirectFieldDecl>(decl);
  796. if (!indirect_field) {
  797. continue;
  798. }
  799. chain = indirect_field->chain();
  800. field = indirect_field->getAnonField();
  801. }
  802. if (field->isBitField()) {
  803. // TODO: Add a representation for named bitfield members.
  804. continue;
  805. }
  806. if (field->isAnonymousStructOrUnion()) {
  807. // Fields within an anonymous structure or union will be added via their
  808. // IndirectFieldDecls.
  809. continue;
  810. }
  811. auto field_name_id = AddIdentifierName(context, field->getName());
  812. auto [field_type_inst_id, field_type_id] =
  813. ImportTypeAndDependencies(context, import_ir_inst_id, field->getType());
  814. if (!field_type_inst_id.has_value()) {
  815. // TODO: For now, just skip over fields whose types we can't map.
  816. continue;
  817. }
  818. // Create a field now, as we know the index to use.
  819. // TODO: Consider doing this lazily instead.
  820. auto field_decl_id = AddInst(
  821. context, MakeImportedLocIdAndInst(
  822. context, import_ir_inst_id,
  823. SemIR::FieldDecl{
  824. .type_id = GetUnboundElementType(
  825. context, class_type_inst_id, field_type_inst_id),
  826. .name_id = field_name_id,
  827. .index = SemIR::ElementIndex(fields.size())}));
  828. context.sem_ir().clang_decls().Add(
  829. {.decl = decl->getCanonicalDecl(), .inst_id = field_decl_id});
  830. // Compute the offset to the field that appears directly in the class.
  831. uint64_t offset = clang_layout.getFieldOffset(
  832. cast<clang::FieldDecl>(chain.front())->getFieldIndex());
  833. // If this is an indirect field, walk the path and accumulate the offset to
  834. // the named field.
  835. for (auto* inner_decl : chain.drop_front()) {
  836. auto* inner_field = cast<clang::FieldDecl>(inner_decl);
  837. const auto& inner_layout =
  838. context.ast_context().getASTRecordLayout(inner_field->getParent());
  839. offset += inner_layout.getFieldOffset(inner_field->getFieldIndex());
  840. }
  841. layout.push_back(
  842. context.ast_context().toCharUnitsFromBits(offset).getQuantity());
  843. fields.push_back(
  844. {.name_id = field_name_id, .type_inst_id = field_type_inst_id});
  845. }
  846. // TODO: Add a field to prevent tail padding reuse if necessary.
  847. return AddTypeInst<SemIR::CustomLayoutType>(
  848. context, import_ir_inst_id,
  849. {.type_id = SemIR::TypeType::TypeId,
  850. .fields_id = context.struct_type_fields().Add(fields),
  851. .layout_id = context.custom_layouts().Add(layout)});
  852. }
  853. // Creates a Carbon class definition based on the information in the given Clang
  854. // class declaration, which is assumed to be for a class definition.
  855. static auto BuildClassDefinition(Context& context,
  856. SemIR::ImportIRInstId import_ir_inst_id,
  857. SemIR::ClassId class_id,
  858. SemIR::TypeInstId class_inst_id,
  859. clang::CXXRecordDecl* clang_def) -> void {
  860. auto& class_info = context.classes().Get(class_id);
  861. CARBON_CHECK(!class_info.has_definition_started());
  862. class_info.definition_id = class_inst_id;
  863. context.inst_block_stack().Push();
  864. class_info.inheritance_kind = GetInheritanceKind(clang_def);
  865. // Compute the class's object representation.
  866. auto object_repr_id = ImportClassObjectRepr(
  867. context, class_id, import_ir_inst_id, class_inst_id, clang_def);
  868. class_info.complete_type_witness_id = AddInst<SemIR::CompleteTypeWitness>(
  869. context, import_ir_inst_id,
  870. {.type_id = GetSingletonType(context, SemIR::WitnessType::TypeInstId),
  871. .object_repr_type_inst_id = object_repr_id});
  872. class_info.body_block_id = context.inst_block_stack().Pop();
  873. }
  874. // Computes and returns the Carbon type to use as the object representation of
  875. // the given C++ enum type. This is a builtin int type matching the enum's
  876. // representation.
  877. static auto ImportEnumObjectRepresentation(
  878. Context& context, SemIR::ImportIRInstId import_ir_inst_id,
  879. clang::EnumDecl* enum_decl) -> SemIR::TypeInstId {
  880. auto int_type = enum_decl->getIntegerType();
  881. CARBON_CHECK(!int_type.isNull(), "incomplete enum type {0}",
  882. enum_decl->getNameAsString());
  883. auto int_kind = int_type->isSignedIntegerType() ? SemIR::IntKind::Signed
  884. : SemIR::IntKind::Unsigned;
  885. auto bit_width_id = GetOrAddInst<SemIR::IntValue>(
  886. context, import_ir_inst_id,
  887. {.type_id = GetSingletonType(context, SemIR::IntLiteralType::TypeInstId),
  888. .int_id = context.ints().AddUnsigned(
  889. llvm::APInt(64, context.ast_context().getIntWidth(int_type)))});
  890. return context.types().GetAsTypeInstId(
  891. GetOrAddInst(context, SemIR::LocIdAndInst::NoLoc(SemIR::IntType{
  892. .type_id = SemIR::TypeType::TypeId,
  893. .int_kind = int_kind,
  894. .bit_width_id = bit_width_id})));
  895. }
  896. // Creates a Carbon class definition based on the information in the given Clang
  897. // enum declaration.
  898. static auto BuildEnumDefinition(Context& context,
  899. SemIR::ImportIRInstId import_ir_inst_id,
  900. SemIR::ClassId class_id,
  901. SemIR::TypeInstId class_inst_id,
  902. clang::EnumDecl* enum_decl) -> void {
  903. auto& class_info = context.classes().Get(class_id);
  904. CARBON_CHECK(!class_info.has_definition_started());
  905. class_info.definition_id = class_inst_id;
  906. context.inst_block_stack().Push();
  907. // Don't allow inheritance from C++ enums, to match the behavior in C++.
  908. class_info.inheritance_kind = SemIR::Class::Final;
  909. // Compute the enum type's object representation. An enum is an adapter for
  910. // the corresponding builtin integer type.
  911. auto object_repr_id =
  912. ImportEnumObjectRepresentation(context, import_ir_inst_id, enum_decl);
  913. class_info.adapt_id = AddInst(
  914. context, SemIR::LocIdAndInst::UncheckedLoc(
  915. import_ir_inst_id,
  916. SemIR::AdaptDecl{.adapted_type_inst_id = object_repr_id}));
  917. class_info.complete_type_witness_id = AddInst<SemIR::CompleteTypeWitness>(
  918. context, import_ir_inst_id,
  919. {.type_id = GetSingletonType(context, SemIR::WitnessType::TypeInstId),
  920. .object_repr_type_inst_id = object_repr_id});
  921. class_info.body_block_id = context.inst_block_stack().Pop();
  922. }
  923. auto ImportClassDefinitionForClangDecl(Context& context, SemIR::LocId loc_id,
  924. SemIR::ClassId class_id,
  925. SemIR::ClangDeclId clang_decl_id)
  926. -> bool {
  927. clang::ASTUnit* ast = context.sem_ir().clang_ast_unit();
  928. CARBON_CHECK(ast);
  929. auto* clang_decl = cast<clang::TagDecl>(
  930. context.sem_ir().clang_decls().Get(clang_decl_id).decl);
  931. auto class_inst_id = context.types().GetAsTypeInstId(
  932. context.classes().Get(class_id).first_owning_decl_id);
  933. // TODO: Map loc_id into a clang location and use it for diagnostics if
  934. // instantiation fails, instead of annotating the diagnostic with another
  935. // location.
  936. clang::SourceLocation loc = clang_decl->getLocation();
  937. Diagnostics::AnnotationScope annotate_diagnostics(
  938. &context.emitter(), [&](auto& builder) {
  939. CARBON_DIAGNOSTIC(InCppTypeCompletion, Note,
  940. "while completing C++ type {0}", SemIR::TypeId);
  941. builder.Note(loc_id, InCppTypeCompletion,
  942. context.classes().Get(class_id).self_type_id);
  943. });
  944. // Ask Clang whether the type is complete. This triggers template
  945. // instantiation if necessary.
  946. clang::DiagnosticErrorTrap trap(ast->getDiagnostics());
  947. if (!ast->getSema().isCompleteType(
  948. loc, context.ast_context().getTypeDeclType(clang_decl))) {
  949. // Type is incomplete. Nothing more to do, but tell the caller if we
  950. // produced an error.
  951. return !trap.hasErrorOccurred();
  952. }
  953. auto import_ir_inst_id =
  954. context.insts().GetCanonicalLocId(class_inst_id).import_ir_inst_id();
  955. if (auto* class_decl = dyn_cast<clang::CXXRecordDecl>(clang_decl)) {
  956. auto* class_def = class_decl->getDefinition();
  957. CARBON_CHECK(class_def, "Complete type has no definition");
  958. if (class_def->getNumVBases()) {
  959. // TODO: Handle virtual bases. We don't actually know where they go in the
  960. // layout. We may also want to use a different size in the layout for
  961. // `partial C`, excluding the virtual base. It's also not entirely safe to
  962. // just skip over the virtual base, as the type we would construct would
  963. // have a misleading size. For now, treat a C++ class with vbases as
  964. // incomplete in Carbon.
  965. context.TODO(loc_id, "class with virtual bases");
  966. return false;
  967. }
  968. BuildClassDefinition(context, import_ir_inst_id, class_id, class_inst_id,
  969. class_def);
  970. } else if (auto* enum_decl = dyn_cast<clang::EnumDecl>(clang_decl)) {
  971. BuildEnumDefinition(context, import_ir_inst_id, class_id, class_inst_id,
  972. enum_decl);
  973. }
  974. return true;
  975. }
  976. // Imports an enumerator declaration from Clang to Carbon.
  977. static auto ImportEnumConstantDecl(Context& context,
  978. clang::EnumConstantDecl* enumerator_decl)
  979. -> SemIR::InstId {
  980. CARBON_CHECK(!IsClangDeclImported(context, enumerator_decl));
  981. // Find the enclosing enum type.
  982. auto type_inst_id = LookupClangDeclInstId(
  983. context, cast<clang::EnumDecl>(enumerator_decl->getDeclContext()));
  984. auto type_id = context.types().GetTypeIdForTypeInstId(type_inst_id);
  985. // Build a corresponding IntValue.
  986. auto int_id = context.ints().Add(enumerator_decl->getInitVal());
  987. auto loc_id =
  988. AddImportIRInst(context.sem_ir(), enumerator_decl->getLocation());
  989. auto inst_id = AddInstInNoBlock<SemIR::IntValue>(
  990. context, loc_id, {.type_id = type_id, .int_id = int_id});
  991. context.imports().push_back(inst_id);
  992. context.sem_ir().clang_decls().Add(
  993. {.decl = enumerator_decl->getCanonicalDecl(), .inst_id = inst_id});
  994. return inst_id;
  995. }
  996. // Mark the given `Decl` as failed in `clang_decls`.
  997. static auto MarkFailedDecl(Context& context, clang::Decl* clang_decl) {
  998. context.sem_ir().clang_decls().Add({.decl = clang_decl->getCanonicalDecl(),
  999. .inst_id = SemIR::ErrorInst::InstId});
  1000. }
  1001. // Creates an integer type of the given size.
  1002. static auto MakeIntType(Context& context, IntId size_id, bool is_signed)
  1003. -> TypeExpr {
  1004. auto type_inst_id = MakeIntTypeLiteral(
  1005. context, Parse::NodeId::None,
  1006. is_signed ? SemIR::IntKind::Signed : SemIR::IntKind::Unsigned, size_id);
  1007. return ExprAsType(context, Parse::NodeId::None, type_inst_id);
  1008. }
  1009. // Maps a C++ builtin integer type to a Carbon type.
  1010. // TODO: Handle integer types that map to named aliases.
  1011. static auto MapBuiltinIntegerType(Context& context, SemIR::LocId loc_id,
  1012. clang::QualType qual_type,
  1013. const clang::BuiltinType& type) -> TypeExpr {
  1014. clang::ASTContext& ast_context = context.ast_context();
  1015. unsigned width = ast_context.getIntWidth(qual_type);
  1016. bool is_signed = type.isSignedInteger();
  1017. auto int_n_type = ast_context.getIntTypeForBitwidth(width, is_signed);
  1018. if (ast_context.hasSameType(qual_type, int_n_type)) {
  1019. TypeExpr type_expr =
  1020. MakeIntType(context, context.ints().Add(width), is_signed);
  1021. // Try to make sure integer types of 32 or 64 bits are complete so we can
  1022. // check against them when deciding whether we need to generate a thunk.
  1023. if (width == 32 || width == 64) {
  1024. SemIR::TypeId type_id = type_expr.type_id;
  1025. if (!context.types().IsComplete(type_id)) {
  1026. TryToCompleteType(context, type_id, loc_id);
  1027. }
  1028. }
  1029. return type_expr;
  1030. }
  1031. if (ast_context.hasSameType(qual_type, ast_context.CharTy)) {
  1032. return ExprAsType(context, Parse::NodeId::None,
  1033. MakeCharTypeLiteral(context, Parse::NodeId::None));
  1034. }
  1035. return TypeExpr::None;
  1036. }
  1037. // Maps a C++ builtin type to a Carbon type.
  1038. // TODO: Support more builtin types.
  1039. static auto MapBuiltinType(Context& context, SemIR::LocId loc_id,
  1040. clang::QualType qual_type,
  1041. const clang::BuiltinType& type) -> TypeExpr {
  1042. clang::ASTContext& ast_context = context.ast_context();
  1043. if (type.isBooleanType()) {
  1044. CARBON_CHECK(ast_context.hasSameType(qual_type, ast_context.BoolTy));
  1045. return ExprAsType(context, Parse::NodeId::None,
  1046. context.types().GetInstId(GetSingletonType(
  1047. context, SemIR::BoolType::TypeInstId)));
  1048. }
  1049. if (type.isInteger()) {
  1050. return MapBuiltinIntegerType(context, loc_id, qual_type, type);
  1051. }
  1052. if (type.isFloatingPoint()) {
  1053. if (type.isFloat16Type() || type.isFloat32Type() || type.isDoubleType() ||
  1054. type.isFloat128Type()) {
  1055. return ExprAsType(
  1056. context, Parse::NodeId::None,
  1057. MakeFloatTypeLiteral(
  1058. context, Parse::NodeId::None,
  1059. context.ints().Add(ast_context.getTypeSize(qual_type))));
  1060. }
  1061. // TODO: Handle floating-point types that map to named aliases.
  1062. }
  1063. return TypeExpr::None;
  1064. }
  1065. // Determines whether record_decl is a C++ class that has a custom mapping into
  1066. // Carbon, and if so, returns the corresponding Carbon type. Otherwise returns
  1067. // None.
  1068. static auto LookupCustomRecordType(Context& context,
  1069. const clang::CXXRecordDecl* record_decl)
  1070. -> TypeExpr {
  1071. switch (GetCustomCppTypeMapping(record_decl)) {
  1072. case CustomCppTypeMapping::None:
  1073. return TypeExpr::None;
  1074. case CustomCppTypeMapping::Str:
  1075. return MakeStringType(
  1076. context,
  1077. AddImportIRInst(context.sem_ir(), record_decl->getLocation()));
  1078. }
  1079. }
  1080. // Maps a C++ tag type (class, struct, union, enum) to a Carbon type.
  1081. static auto MapTagType(Context& context, const clang::TagType& type)
  1082. -> TypeExpr {
  1083. auto* tag_decl = type.getDecl();
  1084. CARBON_CHECK(tag_decl);
  1085. // Check if the declaration is already mapped.
  1086. SemIR::InstId tag_inst_id = LookupClangDeclInstId(context, tag_decl);
  1087. if (!tag_inst_id.has_value()) {
  1088. if (auto* record_decl = dyn_cast<clang::CXXRecordDecl>(tag_decl)) {
  1089. auto custom_type = LookupCustomRecordType(context, record_decl);
  1090. if (custom_type.inst_id.has_value()) {
  1091. context.sem_ir().clang_decls().Add(
  1092. {.decl = record_decl, .inst_id = custom_type.inst_id});
  1093. return custom_type;
  1094. }
  1095. }
  1096. tag_inst_id = ImportTagDecl(context, tag_decl);
  1097. }
  1098. SemIR::TypeInstId record_type_inst_id =
  1099. context.types().GetAsTypeInstId(tag_inst_id);
  1100. return {
  1101. .inst_id = record_type_inst_id,
  1102. .type_id = context.types().GetTypeIdForTypeInstId(record_type_inst_id)};
  1103. }
  1104. // Maps a C++ type that is not a wrapper type such as a pointer to a Carbon
  1105. // type.
  1106. // TODO: Support more types.
  1107. static auto MapNonWrapperType(Context& context, SemIR::LocId loc_id,
  1108. clang::QualType type) -> TypeExpr {
  1109. if (const auto* builtin_type = type->getAs<clang::BuiltinType>()) {
  1110. return MapBuiltinType(context, loc_id, type, *builtin_type);
  1111. }
  1112. if (const auto* tag_type = type->getAs<clang::TagType>()) {
  1113. return MapTagType(context, *tag_type);
  1114. }
  1115. CARBON_CHECK(!type.hasQualifiers() && !type->isPointerType(),
  1116. "Should not see wrapper types here");
  1117. return TypeExpr::None;
  1118. }
  1119. // Maps a qualified C++ type to a Carbon type.
  1120. static auto MapQualifiedType(Context& context, clang::QualType type,
  1121. TypeExpr type_expr) -> TypeExpr {
  1122. auto quals = type.getQualifiers();
  1123. if (quals.hasConst()) {
  1124. auto type_id = GetConstType(context, type_expr.inst_id);
  1125. type_expr = {.inst_id = context.types().GetInstId(type_id),
  1126. .type_id = type_id};
  1127. quals.removeConst();
  1128. }
  1129. // TODO: Support other qualifiers.
  1130. if (!quals.empty()) {
  1131. return TypeExpr::None;
  1132. }
  1133. return type_expr;
  1134. }
  1135. // Maps a C++ pointer type to a Carbon pointer type.
  1136. static auto MapPointerType(Context& context, clang::QualType type,
  1137. TypeExpr pointee_type_expr) -> TypeExpr {
  1138. CARBON_CHECK(type->isPointerType());
  1139. if (auto nullability = type->getNullability();
  1140. !nullability.has_value() ||
  1141. *nullability != clang::NullabilityKind::NonNull) {
  1142. // TODO: Support nullable pointers.
  1143. return TypeExpr::None;
  1144. }
  1145. SemIR::TypeId pointer_type_id =
  1146. GetPointerType(context, pointee_type_expr.inst_id);
  1147. return {.inst_id = context.types().GetInstId(pointer_type_id),
  1148. .type_id = pointer_type_id};
  1149. }
  1150. // Maps a C++ type to a Carbon type. `type` should not be canonicalized because
  1151. // we check for pointer nullability and nullability will be lost by
  1152. // canonicalization.
  1153. static auto MapType(Context& context, SemIR::LocId loc_id, clang::QualType type)
  1154. -> TypeExpr {
  1155. // Unwrap any type modifiers and wrappers.
  1156. llvm::SmallVector<clang::QualType> wrapper_types;
  1157. while (true) {
  1158. clang::QualType orig_type = type;
  1159. if (type.hasQualifiers()) {
  1160. type = type.getUnqualifiedType();
  1161. } else if (type->isPointerType()) {
  1162. type = type->getPointeeType();
  1163. } else {
  1164. break;
  1165. }
  1166. wrapper_types.push_back(orig_type);
  1167. }
  1168. auto mapped = MapNonWrapperType(context, loc_id, type);
  1169. for (auto wrapper : llvm::reverse(wrapper_types)) {
  1170. if (!mapped.inst_id.has_value() ||
  1171. mapped.type_id == SemIR::ErrorInst::TypeId) {
  1172. break;
  1173. }
  1174. if (wrapper.hasQualifiers()) {
  1175. mapped = MapQualifiedType(context, wrapper, mapped);
  1176. } else if (wrapper->isPointerType()) {
  1177. mapped = MapPointerType(context, wrapper, mapped);
  1178. } else {
  1179. CARBON_FATAL("Unexpected wrapper type {0}", wrapper.getAsString());
  1180. }
  1181. }
  1182. return mapped;
  1183. }
  1184. // Returns a block for the implicit parameters of the given function
  1185. // declaration. Because function templates are not yet supported, this currently
  1186. // only contains the `self` parameter. On error, produces a diagnostic and
  1187. // returns None.
  1188. static auto MakeImplicitParamPatternsBlockId(
  1189. Context& context, SemIR::LocId loc_id,
  1190. const clang::FunctionDecl& clang_decl) -> SemIR::InstBlockId {
  1191. const auto* method_decl = dyn_cast<clang::CXXMethodDecl>(&clang_decl);
  1192. if (!method_decl || method_decl->isStatic() ||
  1193. isa<clang::CXXConstructorDecl>(clang_decl)) {
  1194. return SemIR::InstBlockId::Empty;
  1195. }
  1196. // Build a `self` parameter from the object parameter.
  1197. BeginSubpattern(context);
  1198. // Perform some special-case mapping for the object parameter:
  1199. //
  1200. // - If it's a const reference to T, produce a by-value `self: T` parameter.
  1201. // - If it's a non-const reference to T, produce an `addr self: T*`
  1202. // parameter.
  1203. // - Otherwise, map it directly, which will currently fail for `&&`-qualified
  1204. // methods.
  1205. //
  1206. // TODO: Some of this mapping should be performed for all parameters.
  1207. clang::QualType param_type =
  1208. method_decl->getFunctionObjectParameterReferenceType();
  1209. bool addr_self = false;
  1210. if (param_type->isLValueReferenceType()) {
  1211. param_type = param_type.getNonReferenceType();
  1212. if (param_type.isConstQualified()) {
  1213. // TODO: Consider only doing this if `const` is the only qualifier. For
  1214. // now, any other qualifier will fail when mapping the type.
  1215. auto split_type = param_type.getSplitUnqualifiedType();
  1216. split_type.Quals.removeConst();
  1217. param_type = method_decl->getASTContext().getQualifiedType(split_type);
  1218. } else {
  1219. addr_self = true;
  1220. }
  1221. }
  1222. auto [type_inst_id, type_id] = MapType(context, loc_id, param_type);
  1223. SemIR::ExprRegionId type_expr_region_id =
  1224. EndSubpatternAsExpr(context, type_inst_id);
  1225. if (!type_id.has_value()) {
  1226. context.TODO(loc_id,
  1227. llvm::formatv("Unsupported: object parameter type: {0}",
  1228. param_type.getAsString()));
  1229. return SemIR::InstBlockId::None;
  1230. }
  1231. // TODO: Fill in a location once available.
  1232. auto pattern_id =
  1233. addr_self ? AddAddrSelfParamPattern(context, SemIR::LocId::None,
  1234. type_expr_region_id, type_inst_id)
  1235. : AddSelfParamPattern(context, SemIR::LocId::None,
  1236. type_expr_region_id, type_id);
  1237. return context.inst_blocks().Add({pattern_id});
  1238. }
  1239. // Returns a block id for the explicit parameters of the given function
  1240. // declaration. If the function declaration has no parameters, it returns
  1241. // `SemIR::InstBlockId::Empty`. In the case of an unsupported parameter type, it
  1242. // produces an error and returns `SemIR::InstBlockId::None`.
  1243. // TODO: Consider refactoring to extract and reuse more logic from
  1244. // `HandleAnyBindingPattern()`.
  1245. static auto MakeParamPatternsBlockId(Context& context, SemIR::LocId loc_id,
  1246. const clang::FunctionDecl& clang_decl)
  1247. -> SemIR::InstBlockId {
  1248. if (clang_decl.parameters().empty()) {
  1249. return SemIR::InstBlockId::Empty;
  1250. }
  1251. llvm::SmallVector<SemIR::InstId> params;
  1252. params.reserve(clang_decl.getNumNonObjectParams());
  1253. for (unsigned i : llvm::seq(clang_decl.getNumNonObjectParams())) {
  1254. const auto* param = clang_decl.getNonObjectParameter(i);
  1255. // TODO: Get the parameter type from the function, not from the
  1256. // `ParmVarDecl`. The type of the `ParmVarDecl` is the type within the
  1257. // function, and isn't in general the same as the type that's exposed to
  1258. // callers. In particular, the parameter type exposed to callers will never
  1259. // be cv-qualified.
  1260. clang::QualType param_type = param->getType();
  1261. // Mark the start of a region of insts, needed for the type expression
  1262. // created later with the call of `EndSubpatternAsExpr()`.
  1263. BeginSubpattern(context);
  1264. auto [type_inst_id, type_id] = MapType(context, loc_id, param_type);
  1265. // Type expression of the binding pattern - a single-entry/single-exit
  1266. // region that allows control flow in the type expression e.g. fn F(x: if C
  1267. // then i32 else i64).
  1268. SemIR::ExprRegionId type_expr_region_id =
  1269. EndSubpatternAsExpr(context, type_inst_id);
  1270. if (!type_id.has_value()) {
  1271. context.TODO(loc_id, llvm::formatv("Unsupported: parameter type: {0}",
  1272. param_type.getAsString()));
  1273. return SemIR::InstBlockId::None;
  1274. }
  1275. llvm::StringRef param_name = param->getName();
  1276. SemIR::NameId name_id =
  1277. param_name.empty()
  1278. // Translate an unnamed parameter to an underscore to
  1279. // match Carbon's naming of unnamed/unused function params.
  1280. ? SemIR::NameId::Underscore
  1281. : AddIdentifierName(context, param_name);
  1282. // TODO: Fix this once templates are supported.
  1283. bool is_template = false;
  1284. // TODO: Fix this once generics are supported.
  1285. bool is_generic = false;
  1286. SemIR::InstId binding_pattern_id =
  1287. // TODO: Fill in a location once available.
  1288. AddBindingPattern(context, SemIR::LocId::None, name_id, type_id,
  1289. type_expr_region_id, is_generic, is_template)
  1290. .pattern_id;
  1291. SemIR::InstId var_pattern_id = AddPatternInst(
  1292. context,
  1293. // TODO: Fill in a location once available.
  1294. SemIR::LocIdAndInst::NoLoc(SemIR::ValueParamPattern(
  1295. {.type_id = context.insts().Get(binding_pattern_id).type_id(),
  1296. .subpattern_id = binding_pattern_id,
  1297. .index = SemIR::CallParamIndex::None})));
  1298. params.push_back(var_pattern_id);
  1299. }
  1300. return context.inst_blocks().Add(params);
  1301. }
  1302. // Returns the return `TypeExpr` of the given function declaration. In case of
  1303. // an unsupported return type, returns `SemIR::ErrorInst::InstId`. Constructors
  1304. // are treated as returning a class instance.
  1305. // TODO: Support more return types.
  1306. static auto GetReturnTypeExpr(Context& context, SemIR::LocId loc_id,
  1307. clang::FunctionDecl* clang_decl) -> TypeExpr {
  1308. clang::QualType ret_type = clang_decl->getReturnType();
  1309. if (!ret_type->isVoidType()) {
  1310. TypeExpr mapped_type = MapType(context, loc_id, ret_type);
  1311. if (!mapped_type.inst_id.has_value()) {
  1312. context.TODO(loc_id, llvm::formatv("Unsupported: return type: {0}",
  1313. ret_type.getAsString()));
  1314. return {.inst_id = SemIR::ErrorInst::TypeInstId,
  1315. .type_id = SemIR::ErrorInst::TypeId};
  1316. }
  1317. return mapped_type;
  1318. }
  1319. if (!isa<clang::CXXConstructorDecl>(clang_decl)) {
  1320. // void.
  1321. return TypeExpr::None;
  1322. }
  1323. // TODO: Make this a `PartialType`.
  1324. SemIR::TypeInstId record_type_inst_id = context.types().GetAsTypeInstId(
  1325. context.sem_ir()
  1326. .clang_decls()
  1327. .Get(context.sem_ir().clang_decls().Lookup(
  1328. cast<clang::Decl>(clang_decl->getParent())))
  1329. .inst_id);
  1330. return {
  1331. .inst_id = record_type_inst_id,
  1332. .type_id = context.types().GetTypeIdForTypeInstId(record_type_inst_id)};
  1333. }
  1334. // Returns the return pattern of the given function declaration. In case of an
  1335. // unsupported return type, it produces a diagnostic and returns
  1336. // `SemIR::ErrorInst::InstId`. Constructors are treated as returning a class
  1337. // instance.
  1338. static auto GetReturnPattern(Context& context, SemIR::LocId loc_id,
  1339. clang::FunctionDecl* clang_decl) -> SemIR::InstId {
  1340. auto [type_inst_id, type_id] = GetReturnTypeExpr(context, loc_id, clang_decl);
  1341. if (!type_inst_id.has_value()) {
  1342. // void.
  1343. return SemIR::InstId::None;
  1344. }
  1345. auto pattern_type_id = GetPatternType(context, type_id);
  1346. SemIR::InstId return_slot_pattern_id = AddPatternInst(
  1347. // TODO: Fill in a location for the return type once available.
  1348. context,
  1349. SemIR::LocIdAndInst::NoLoc(SemIR::ReturnSlotPattern(
  1350. {.type_id = pattern_type_id, .type_inst_id = type_inst_id})));
  1351. SemIR::InstId param_pattern_id = AddPatternInst(
  1352. // TODO: Fill in a location for the return type once available.
  1353. context, SemIR::LocIdAndInst::NoLoc(SemIR::OutParamPattern(
  1354. {.type_id = pattern_type_id,
  1355. .subpattern_id = return_slot_pattern_id,
  1356. .index = SemIR::CallParamIndex::None})));
  1357. return param_pattern_id;
  1358. }
  1359. namespace {
  1360. // Represents the parameter patterns block id, the return slot pattern id and
  1361. // the call parameters block id for a function declaration.
  1362. struct FunctionParamsInsts {
  1363. SemIR::InstBlockId implicit_param_patterns_id;
  1364. SemIR::InstBlockId param_patterns_id;
  1365. SemIR::InstId return_slot_pattern_id;
  1366. SemIR::InstBlockId call_params_id;
  1367. };
  1368. } // namespace
  1369. // Creates a block containing the parameter pattern instructions for the
  1370. // explicit parameters, a parameter pattern instruction for the return type and
  1371. // a block containing the call parameters of the function. Emits a callee
  1372. // pattern-match for the explicit parameter patterns and the return slot pattern
  1373. // to create the Call parameters instructions block. Currently the implicit
  1374. // parameter patterns are not taken into account. Returns the parameter patterns
  1375. // block id, the return slot pattern id, and the call parameters block id.
  1376. // Produces a diagnostic and returns `std::nullopt` if the function declaration
  1377. // has an unsupported parameter type.
  1378. static auto CreateFunctionParamsInsts(Context& context, SemIR::LocId loc_id,
  1379. clang::FunctionDecl* clang_decl)
  1380. -> std::optional<FunctionParamsInsts> {
  1381. if (isa<clang::CXXDestructorDecl>(clang_decl)) {
  1382. context.TODO(loc_id, "Unsupported: Destructor");
  1383. return std::nullopt;
  1384. }
  1385. auto implicit_param_patterns_id =
  1386. MakeImplicitParamPatternsBlockId(context, loc_id, *clang_decl);
  1387. if (!implicit_param_patterns_id.has_value()) {
  1388. return std::nullopt;
  1389. }
  1390. auto param_patterns_id =
  1391. MakeParamPatternsBlockId(context, loc_id, *clang_decl);
  1392. if (!param_patterns_id.has_value()) {
  1393. return std::nullopt;
  1394. }
  1395. auto return_slot_pattern_id = GetReturnPattern(context, loc_id, clang_decl);
  1396. if (SemIR::ErrorInst::InstId == return_slot_pattern_id) {
  1397. return std::nullopt;
  1398. }
  1399. auto call_params_id =
  1400. CalleePatternMatch(context, implicit_param_patterns_id, param_patterns_id,
  1401. return_slot_pattern_id);
  1402. return {{.implicit_param_patterns_id = implicit_param_patterns_id,
  1403. .param_patterns_id = param_patterns_id,
  1404. .return_slot_pattern_id = return_slot_pattern_id,
  1405. .call_params_id = call_params_id}};
  1406. }
  1407. // Creates a `FunctionDecl` and a `Function` without C++ thunk information.
  1408. // Returns std::nullopt on failure. The given Clang declaration is assumed to:
  1409. // * Have not been imported before.
  1410. // * Be of supported type (ignoring parameters).
  1411. static auto ImportFunction(Context& context, SemIR::LocId loc_id,
  1412. clang::FunctionDecl* clang_decl)
  1413. -> std::optional<SemIR::FunctionId> {
  1414. context.scope_stack().PushForDeclName();
  1415. context.inst_block_stack().Push();
  1416. context.pattern_block_stack().Push();
  1417. auto function_params_insts =
  1418. CreateFunctionParamsInsts(context, loc_id, clang_decl);
  1419. auto pattern_block_id = context.pattern_block_stack().Pop();
  1420. auto decl_block_id = context.inst_block_stack().Pop();
  1421. context.scope_stack().Pop();
  1422. if (!function_params_insts.has_value()) {
  1423. return std::nullopt;
  1424. }
  1425. auto function_decl = SemIR::FunctionDecl{
  1426. SemIR::TypeId::None, SemIR::FunctionId::None, decl_block_id};
  1427. auto decl_id =
  1428. AddPlaceholderInstInNoBlock(context, Parse::NodeId::None, function_decl);
  1429. context.imports().push_back(decl_id);
  1430. SemIR::NameId function_name_id =
  1431. isa<clang::CXXConstructorDecl>(clang_decl)
  1432. ? context.classes()
  1433. .Get(context.insts()
  1434. .GetAs<SemIR::ClassDecl>(LookupClangDeclInstId(
  1435. context,
  1436. cast<clang::Decl>(clang_decl->getParent())))
  1437. .class_id)
  1438. .name_id
  1439. : AddIdentifierName(context, clang_decl->getName());
  1440. auto function_info = SemIR::Function{
  1441. {.name_id = function_name_id,
  1442. .parent_scope_id = GetParentNameScopeId(context, clang_decl),
  1443. .generic_id = SemIR::GenericId::None,
  1444. .first_param_node_id = Parse::NodeId::None,
  1445. .last_param_node_id = Parse::NodeId::None,
  1446. .pattern_block_id = pattern_block_id,
  1447. .implicit_param_patterns_id =
  1448. function_params_insts->implicit_param_patterns_id,
  1449. .param_patterns_id = function_params_insts->param_patterns_id,
  1450. .is_extern = false,
  1451. .extern_library_id = SemIR::LibraryNameId::None,
  1452. .non_owning_decl_id = SemIR::InstId::None,
  1453. .first_owning_decl_id = decl_id,
  1454. .definition_id = SemIR::InstId::None},
  1455. {.call_params_id = function_params_insts->call_params_id,
  1456. .return_slot_pattern_id = function_params_insts->return_slot_pattern_id,
  1457. .virtual_modifier = SemIR::FunctionFields::VirtualModifier::None,
  1458. .self_param_id = FindSelfPattern(
  1459. context, function_params_insts->implicit_param_patterns_id),
  1460. .clang_decl_id = context.sem_ir().clang_decls().Add(
  1461. {.decl = clang_decl, .inst_id = decl_id})}};
  1462. function_decl.function_id = context.functions().Add(function_info);
  1463. function_decl.type_id = GetFunctionType(context, function_decl.function_id,
  1464. SemIR::SpecificId::None);
  1465. ReplaceInstBeforeConstantUse(context, decl_id, function_decl);
  1466. return function_decl.function_id;
  1467. }
  1468. // Imports a function declaration from Clang to Carbon. If successful, returns
  1469. // the new Carbon function declaration `InstId`. If the declaration was already
  1470. // imported, returns the mapped instruction.
  1471. static auto ImportFunctionDecl(Context& context, SemIR::LocId loc_id,
  1472. clang::FunctionDecl* clang_decl)
  1473. -> SemIR::InstId {
  1474. // Check if the declaration is already mapped.
  1475. if (SemIR::InstId existing_inst_id =
  1476. LookupClangDeclInstId(context, clang_decl);
  1477. existing_inst_id.has_value()) {
  1478. return existing_inst_id;
  1479. }
  1480. if (clang_decl->isVariadic()) {
  1481. context.TODO(loc_id, "Unsupported: Variadic function");
  1482. MarkFailedDecl(context, clang_decl);
  1483. return SemIR::ErrorInst::InstId;
  1484. }
  1485. if (clang_decl->getTemplatedKind() ==
  1486. clang::FunctionDecl::TK_FunctionTemplate) {
  1487. context.TODO(loc_id, "Unsupported: Template function");
  1488. MarkFailedDecl(context, clang_decl);
  1489. return SemIR::ErrorInst::InstId;
  1490. }
  1491. if (auto* method_decl = dyn_cast<clang::CXXMethodDecl>(clang_decl)) {
  1492. if (method_decl->isVirtual()) {
  1493. context.TODO(loc_id, "Unsupported: Virtual function");
  1494. MarkFailedDecl(context, clang_decl);
  1495. return SemIR::ErrorInst::InstId;
  1496. }
  1497. }
  1498. CARBON_CHECK(clang_decl->getFunctionType()->isFunctionProtoType(),
  1499. "Not Prototype function (non-C++ code)");
  1500. auto function_id = ImportFunction(context, loc_id, clang_decl);
  1501. if (!function_id) {
  1502. MarkFailedDecl(context, clang_decl);
  1503. return SemIR::ErrorInst::InstId;
  1504. }
  1505. SemIR::Function& function_info = context.functions().Get(*function_id);
  1506. if (IsCppThunkRequired(context, function_info)) {
  1507. clang::FunctionDecl* thunk_clang_decl =
  1508. BuildCppThunk(context, function_info);
  1509. if (thunk_clang_decl) {
  1510. SemIR::FunctionId thunk_function_id =
  1511. *ImportFunction(context, loc_id, thunk_clang_decl);
  1512. SemIR::InstId thunk_function_decl_id =
  1513. context.functions().Get(thunk_function_id).first_owning_decl_id;
  1514. function_info.SetHasCppThunk(thunk_function_decl_id);
  1515. }
  1516. }
  1517. return function_info.first_owning_decl_id;
  1518. }
  1519. namespace {
  1520. // An item to be imported in an import worklist.
  1521. // TODO: If worklists ever become particularly large, consider changing this
  1522. // to use a `PointerIntPair`.
  1523. struct ImportItem {
  1524. // A declaration that we want to import.
  1525. clang::Decl* decl;
  1526. // Whether we have added `decl`'s dependencies to the worklist.
  1527. bool added_dependencies;
  1528. };
  1529. // A worklist of declarations to import.
  1530. using ImportWorklist = llvm::SmallVector<ImportItem>;
  1531. } // namespace
  1532. // Adds the given declaration to our list of declarations to import.
  1533. static auto AddDependentDecl(const Context& context, clang::Decl* decl,
  1534. ImportWorklist& worklist) -> void {
  1535. if (!IsClangDeclImported(context, decl)) {
  1536. worklist.push_back({.decl = decl, .added_dependencies = false});
  1537. }
  1538. }
  1539. // Finds all decls that need to be imported before importing the given type and
  1540. // adds them to the given set.
  1541. static auto AddDependentUnimportedTypeDecls(const Context& context,
  1542. clang::QualType type,
  1543. ImportWorklist& worklist) -> void {
  1544. while (true) {
  1545. if (type->isPointerType() || type->isReferenceType()) {
  1546. type = type->getPointeeType();
  1547. } else if (const clang::ArrayType* array_type =
  1548. type->getAsArrayTypeUnsafe()) {
  1549. type = array_type->getElementType();
  1550. } else {
  1551. break;
  1552. }
  1553. }
  1554. if (const auto* tag_type = type->getAs<clang::TagType>()) {
  1555. AddDependentDecl(context, tag_type->getDecl(), worklist);
  1556. }
  1557. }
  1558. // Finds all decls that need to be imported before importing the given function
  1559. // and adds them to the given set.
  1560. static auto AddDependentUnimportedFunctionDecls(
  1561. const Context& context, const clang::FunctionDecl& clang_decl,
  1562. ImportWorklist& worklist) -> void {
  1563. for (const auto* param : clang_decl.parameters()) {
  1564. AddDependentUnimportedTypeDecls(context, param->getType(), worklist);
  1565. }
  1566. AddDependentUnimportedTypeDecls(context, clang_decl.getReturnType(),
  1567. worklist);
  1568. }
  1569. // Finds all decls that need to be imported before importing the given
  1570. // declaration and adds them to the given set.
  1571. static auto AddDependentUnimportedDecls(const Context& context,
  1572. clang::Decl* clang_decl,
  1573. ImportWorklist& worklist) -> void {
  1574. if (auto* clang_function_decl = clang_decl->getAsFunction()) {
  1575. AddDependentUnimportedFunctionDecls(context, *clang_function_decl,
  1576. worklist);
  1577. } else if (auto* type_decl = dyn_cast<clang::TypeDecl>(clang_decl)) {
  1578. if (!isa<clang::TagDecl>(clang_decl)) {
  1579. AddDependentUnimportedTypeDecls(
  1580. context, type_decl->getASTContext().getTypeDeclType(type_decl),
  1581. worklist);
  1582. }
  1583. }
  1584. if (!isa<clang::TranslationUnitDecl>(clang_decl)) {
  1585. AddDependentDecl(context, GetParentDecl(clang_decl), worklist);
  1586. }
  1587. }
  1588. // Imports a declaration from Clang to Carbon. Returns the instruction for the
  1589. // new Carbon declaration, which will be an ErrorInst on failure. Assumes all
  1590. // dependencies have already been imported.
  1591. static auto ImportDeclAfterDependencies(Context& context, SemIR::LocId loc_id,
  1592. clang::Decl* clang_decl)
  1593. -> SemIR::InstId {
  1594. if (auto* clang_function_decl = clang_decl->getAsFunction()) {
  1595. return ImportFunctionDecl(context, loc_id, clang_function_decl);
  1596. }
  1597. if (auto* clang_namespace_decl = dyn_cast<clang::NamespaceDecl>(clang_decl)) {
  1598. return ImportNamespaceDecl(context, clang_namespace_decl);
  1599. }
  1600. if (auto* type_decl = dyn_cast<clang::TypeDecl>(clang_decl)) {
  1601. auto type = clang_decl->getASTContext().getTypeDeclType(type_decl);
  1602. auto type_inst_id = MapType(context, loc_id, type).inst_id;
  1603. if (!type_inst_id.has_value()) {
  1604. context.TODO(AddImportIRInst(context.sem_ir(), type_decl->getLocation()),
  1605. llvm::formatv("Unsupported: Type declaration: {0}",
  1606. type.getAsString()));
  1607. return SemIR::ErrorInst::InstId;
  1608. }
  1609. context.sem_ir().clang_decls().Add(
  1610. {.decl = clang_decl, .inst_id = type_inst_id});
  1611. return type_inst_id;
  1612. }
  1613. if (isa<clang::FieldDecl, clang::IndirectFieldDecl>(clang_decl)) {
  1614. // Usable fields get imported as a side effect of importing the class.
  1615. if (SemIR::InstId existing_inst_id =
  1616. LookupClangDeclInstId(context, clang_decl);
  1617. existing_inst_id.has_value()) {
  1618. return existing_inst_id;
  1619. }
  1620. context.TODO(AddImportIRInst(context.sem_ir(), clang_decl->getLocation()),
  1621. "Unsupported: field declaration has unhandled type or kind");
  1622. return SemIR::ErrorInst::InstId;
  1623. }
  1624. if (auto* enum_const_decl = dyn_cast<clang::EnumConstantDecl>(clang_decl)) {
  1625. return ImportEnumConstantDecl(context, enum_const_decl);
  1626. }
  1627. context.TODO(AddImportIRInst(context.sem_ir(), clang_decl->getLocation()),
  1628. llvm::formatv("Unsupported: Declaration type {0}",
  1629. clang_decl->getDeclKindName()));
  1630. return SemIR::ErrorInst::InstId;
  1631. }
  1632. // Attempts to import a set of declarations. Returns `false` if an error was
  1633. // produced, `true` otherwise.
  1634. static auto ImportDeclSet(Context& context, SemIR::LocId loc_id,
  1635. ImportWorklist& worklist) -> bool {
  1636. // Walk the dependency graph in depth-first order, and import declarations
  1637. // once we've imported all of their dependencies.
  1638. while (!worklist.empty()) {
  1639. auto& item = worklist.back();
  1640. if (!item.added_dependencies) {
  1641. // Skip items we've already imported. We checked this when initially
  1642. // adding the item to the worklist, but it might have been added to the
  1643. // worklist twice before the first time we visited it. For example, this
  1644. // happens for `fn F(a: Cpp.T, b: Cpp.T)`.
  1645. if (IsClangDeclImported(context, item.decl)) {
  1646. worklist.pop_back();
  1647. continue;
  1648. }
  1649. // First time visiting this declaration (preorder): add its dependencies
  1650. // to the work list.
  1651. item.added_dependencies = true;
  1652. AddDependentUnimportedDecls(context, item.decl, worklist);
  1653. } else {
  1654. // Second time visiting this declaration (postorder): its dependencies are
  1655. // already imported, so we can import it now.
  1656. auto* decl = worklist.pop_back_val().decl;
  1657. auto inst_id = ImportDeclAfterDependencies(context, loc_id, decl);
  1658. CARBON_CHECK(inst_id.has_value());
  1659. if (inst_id == SemIR::ErrorInst::InstId) {
  1660. return false;
  1661. }
  1662. CARBON_CHECK(IsClangDeclImported(context, decl));
  1663. }
  1664. }
  1665. return true;
  1666. }
  1667. // Imports a declaration from Clang to Carbon. If successful, returns the
  1668. // instruction for the new Carbon declaration. All unimported dependencies are
  1669. // imported first.
  1670. static auto ImportDeclAndDependencies(Context& context, SemIR::LocId loc_id,
  1671. clang::Decl* clang_decl)
  1672. -> SemIR::InstId {
  1673. // Collect dependencies by walking the dependency graph in depth-first order.
  1674. ImportWorklist worklist;
  1675. AddDependentDecl(context, clang_decl, worklist);
  1676. if (!ImportDeclSet(context, loc_id, worklist)) {
  1677. return SemIR::ErrorInst::InstId;
  1678. }
  1679. return LookupClangDeclInstId(context, clang_decl);
  1680. }
  1681. // Imports a type from Clang to Carbon. If successful, returns the imported
  1682. // TypeId. All unimported dependencies are imported first.
  1683. static auto ImportTypeAndDependencies(Context& context, SemIR::LocId loc_id,
  1684. clang::QualType type) -> TypeExpr {
  1685. // Collect dependencies by walking the dependency graph in depth-first order.
  1686. ImportWorklist worklist;
  1687. AddDependentUnimportedTypeDecls(context, type, worklist);
  1688. if (!ImportDeclSet(context, loc_id, worklist)) {
  1689. return {.inst_id = SemIR::ErrorInst::TypeInstId,
  1690. .type_id = SemIR::ErrorInst::TypeId};
  1691. }
  1692. return MapType(context, loc_id, type);
  1693. }
  1694. // Maps `clang::AccessSpecifier` to `SemIR::AccessKind`.
  1695. static auto MapAccess(clang::AccessSpecifier access_specifier)
  1696. -> SemIR::AccessKind {
  1697. switch (access_specifier) {
  1698. case clang::AS_public:
  1699. case clang::AS_none:
  1700. return SemIR::AccessKind::Public;
  1701. case clang::AS_protected:
  1702. return SemIR::AccessKind::Protected;
  1703. case clang::AS_private:
  1704. return SemIR::AccessKind::Private;
  1705. }
  1706. }
  1707. // Imports a `clang::NamedDecl` into Carbon and adds that name into the
  1708. // `NameScope`.
  1709. static auto ImportNameDeclIntoScope(Context& context, SemIR::LocId loc_id,
  1710. SemIR::NameScopeId scope_id,
  1711. SemIR::NameId name_id,
  1712. clang::NamedDecl* clang_decl,
  1713. clang::AccessSpecifier access)
  1714. -> SemIR::ScopeLookupResult {
  1715. SemIR::InstId inst_id =
  1716. ImportDeclAndDependencies(context, loc_id, clang_decl);
  1717. if (!inst_id.has_value()) {
  1718. return SemIR::ScopeLookupResult::MakeNotFound();
  1719. }
  1720. SemIR::AccessKind access_kind = MapAccess(access);
  1721. AddNameToScope(context, scope_id, name_id, access_kind, inst_id);
  1722. return SemIR::ScopeLookupResult::MakeWrappedLookupResult(inst_id,
  1723. access_kind);
  1724. }
  1725. auto ImportNameFromCpp(Context& context, SemIR::LocId loc_id,
  1726. SemIR::NameScopeId scope_id, SemIR::NameId name_id)
  1727. -> SemIR::ScopeLookupResult {
  1728. Diagnostics::AnnotationScope annotate_diagnostics(
  1729. &context.emitter(), [&](auto& builder) {
  1730. CARBON_DIAGNOSTIC(InCppNameLookup, Note,
  1731. "in `Cpp` name lookup for `{0}`", SemIR::NameId);
  1732. builder.Note(loc_id, InCppNameLookup, name_id);
  1733. });
  1734. auto decl_and_access = ClangLookup(context, loc_id, scope_id, name_id);
  1735. if (!decl_and_access) {
  1736. return SemIR::ScopeLookupResult::MakeNotFound();
  1737. }
  1738. auto [decl, access] = *decl_and_access;
  1739. if (!decl) {
  1740. context.name_scopes().AddRequiredName(scope_id, name_id,
  1741. SemIR::ErrorInst::InstId);
  1742. return SemIR::ScopeLookupResult::MakeError();
  1743. }
  1744. return ImportNameDeclIntoScope(context, loc_id, scope_id, name_id, decl,
  1745. access);
  1746. }
  1747. } // namespace Carbon::Check