import.cpp 99 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444
  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/cpp/import.h"
  5. #include <algorithm>
  6. #include <memory>
  7. #include <optional>
  8. #include <string>
  9. #include <tuple>
  10. #include <utility>
  11. #include "clang/AST/ASTContext.h"
  12. #include "clang/AST/RecordLayout.h"
  13. #include "clang/AST/UnresolvedSet.h"
  14. #include "clang/AST/VTableBuilder.h"
  15. #include "clang/Basic/FileManager.h"
  16. #include "clang/Frontend/ASTUnit.h"
  17. #include "clang/Frontend/CompilerInstance.h"
  18. #include "clang/Frontend/CompilerInvocation.h"
  19. #include "clang/Frontend/TextDiagnostic.h"
  20. #include "clang/Lex/PreprocessorOptions.h"
  21. #include "clang/Sema/Lookup.h"
  22. #include "clang/Sema/Overload.h"
  23. #include "common/check.h"
  24. #include "common/ostream.h"
  25. #include "common/raw_string_ostream.h"
  26. #include "llvm/ADT/IntrusiveRefCntPtr.h"
  27. #include "llvm/ADT/StringMap.h"
  28. #include "llvm/ADT/StringRef.h"
  29. #include "llvm/Support/raw_ostream.h"
  30. #include "toolchain/base/kind_switch.h"
  31. #include "toolchain/check/call.h"
  32. #include "toolchain/check/class.h"
  33. #include "toolchain/check/context.h"
  34. #include "toolchain/check/control_flow.h"
  35. #include "toolchain/check/convert.h"
  36. #include "toolchain/check/cpp/access.h"
  37. #include "toolchain/check/cpp/custom_type_mapping.h"
  38. #include "toolchain/check/cpp/macros.h"
  39. #include "toolchain/check/cpp/thunk.h"
  40. #include "toolchain/check/diagnostic_helpers.h"
  41. #include "toolchain/check/eval.h"
  42. #include "toolchain/check/function.h"
  43. #include "toolchain/check/import.h"
  44. #include "toolchain/check/inst.h"
  45. #include "toolchain/check/literal.h"
  46. #include "toolchain/check/member_access.h"
  47. #include "toolchain/check/name_lookup.h"
  48. #include "toolchain/check/operator.h"
  49. #include "toolchain/check/pattern.h"
  50. #include "toolchain/check/pattern_match.h"
  51. #include "toolchain/check/type.h"
  52. #include "toolchain/check/type_completion.h"
  53. #include "toolchain/diagnostics/diagnostic.h"
  54. #include "toolchain/diagnostics/diagnostic_emitter.h"
  55. #include "toolchain/diagnostics/format_providers.h"
  56. #include "toolchain/parse/node_ids.h"
  57. #include "toolchain/sem_ir/clang_decl.h"
  58. #include "toolchain/sem_ir/class.h"
  59. #include "toolchain/sem_ir/cpp_overload_set.h"
  60. #include "toolchain/sem_ir/function.h"
  61. #include "toolchain/sem_ir/ids.h"
  62. #include "toolchain/sem_ir/inst.h"
  63. #include "toolchain/sem_ir/name_scope.h"
  64. #include "toolchain/sem_ir/typed_insts.h"
  65. namespace Carbon::Check {
  66. // Add a line marker directive pointing at the location of the `import Cpp`
  67. // declaration in the Carbon source file. This will cause Clang's diagnostics
  68. // machinery to track and report the location in Carbon code where the import
  69. // was written.
  70. static auto GenerateLineMarker(Context& context, llvm::raw_ostream& out,
  71. int line) {
  72. out << "# " << line << " \""
  73. << FormatEscaped(context.tokens().source().filename()) << "\"\n";
  74. }
  75. // Generates C++ file contents to #include all requested imports.
  76. static auto GenerateCppIncludesHeaderCode(
  77. Context& context, llvm::ArrayRef<Parse::Tree::PackagingNames> imports)
  78. -> std::string {
  79. std::string code;
  80. llvm::raw_string_ostream code_stream(code);
  81. for (const Parse::Tree::PackagingNames& import : imports) {
  82. if (import.inline_body_id.has_value()) {
  83. // Expand `import Cpp inline "code";` directly into the specified code.
  84. auto code_token = context.parse_tree().node_token(import.inline_body_id);
  85. // Compute the line number on which the C++ code starts. Usually the code
  86. // is specified as a block string literal and starts on the line after the
  87. // start of the string token.
  88. // TODO: Determine if this is a block string literal without calling
  89. // `GetTokenText`, which re-lexes the string.
  90. int line = context.tokens().GetLineNumber(code_token);
  91. if (context.tokens().GetTokenText(code_token).contains('\n')) {
  92. ++line;
  93. }
  94. GenerateLineMarker(context, code_stream, line);
  95. code_stream << context.string_literal_values().Get(
  96. context.tokens().GetStringLiteralValue(code_token))
  97. << "\n";
  98. // TODO: Inject a clang pragma here to produce an error if there are
  99. // unclosed scopes at the end of this inline C++ fragment.
  100. } else if (import.library_id.has_value()) {
  101. // Translate `import Cpp library "foo.h";` into `#include "foo.h"`.
  102. GenerateLineMarker(context, code_stream,
  103. context.tokens().GetLineNumber(
  104. context.parse_tree().node_token(import.node_id)));
  105. auto name = context.string_literal_values().Get(import.library_id);
  106. if (name.starts_with('<') && name.ends_with('>')) {
  107. code_stream << "#include <"
  108. << FormatEscaped(name.drop_front().drop_back()) << ">\n";
  109. } else {
  110. code_stream << "#include \"" << FormatEscaped(name) << "\"\n";
  111. }
  112. }
  113. }
  114. // Inject a declaration of placement operator new, because the code we
  115. // generate in thunks depends on it for placement new expressions. Clang has
  116. // special-case logic for lowering a new-expression using this, so a
  117. // definition is not required.
  118. // TODO: This is a hack. We should be able to directly generate Clang AST to
  119. // construct objects in-place without this.
  120. // TODO: Once we can rely on libc++ being available, consider including
  121. // `<__new/placement_new_delete.h>` instead.
  122. code_stream << R"(# 1 "<carbon-internal>"
  123. #undef constexpr
  124. #if __cplusplus > 202302L
  125. constexpr
  126. #endif
  127. #undef void
  128. #undef operator
  129. #undef new
  130. void* operator new(__SIZE_TYPE__, void*)
  131. #if __cplusplus < 201103L
  132. #undef throw
  133. throw()
  134. #else
  135. #undef noexcept
  136. noexcept
  137. #endif
  138. ;
  139. )";
  140. return code;
  141. }
  142. // Adds the name to the scope with the given `access_kind` and `inst_id`.
  143. // `inst_id` must have a value.
  144. static auto AddNameToScope(Context& context, SemIR::NameScopeId scope_id,
  145. SemIR::NameId name_id, SemIR::AccessKind access_kind,
  146. SemIR::InstId inst_id) -> void {
  147. CARBON_CHECK(inst_id.has_value());
  148. context.name_scopes().Get(scope_id).AddRequired(
  149. {.name_id = name_id,
  150. .result = SemIR::ScopeLookupResult::MakeFound(inst_id, access_kind)});
  151. }
  152. // Maps a Clang name to a Carbon `NameId`.
  153. static auto AddIdentifierName(Context& context, llvm::StringRef name)
  154. -> SemIR::NameId {
  155. return SemIR::NameId::ForIdentifier(context.identifiers().Add(name));
  156. }
  157. // Adds the given source location and an `ImportIRInst` referring to it in
  158. // `ImportIRId::Cpp`.
  159. static auto AddImportIRInst(SemIR::File& file,
  160. clang::SourceLocation clang_source_loc)
  161. -> SemIR::ImportIRInstId {
  162. SemIR::ClangSourceLocId clang_source_loc_id =
  163. file.clang_source_locs().Add(clang_source_loc);
  164. return file.import_ir_insts().Add(SemIR::ImportIRInst(clang_source_loc_id));
  165. }
  166. namespace {
  167. // Used to convert Clang diagnostics to Carbon diagnostics.
  168. //
  169. // Handling of Clang notes is a little subtle: as far as Clang is concerned,
  170. // notes are separate diagnostics, not connected to the error or warning that
  171. // precedes them. But in Carbon's diagnostics system, notes are part of the
  172. // enclosing diagnostic. To handle this, we buffer Clang diagnostics until we
  173. // reach a point where we know we're not in the middle of a diagnostic, and then
  174. // emit a diagnostic along with all of its notes. This is triggered when adding
  175. // or removing a Carbon context note, which could otherwise get attached to the
  176. // wrong C++ diagnostics, and at the end of the Carbon program.
  177. class CarbonClangDiagnosticConsumer : public clang::DiagnosticConsumer {
  178. public:
  179. // Creates an instance with the location that triggers calling Clang. The
  180. // `context` is not stored here, and the diagnostics consumer is expected to
  181. // outlive it.
  182. explicit CarbonClangDiagnosticConsumer(
  183. Context& context, std::shared_ptr<clang::CompilerInvocation> invocation)
  184. : sem_ir_(&context.sem_ir()),
  185. emitter_(&context.emitter()),
  186. invocation_(std::move(invocation)) {
  187. emitter_->AddFlushFn([this] { EmitDiagnostics(); });
  188. }
  189. ~CarbonClangDiagnosticConsumer() override {
  190. // Do not inspect `emitter_` here; it's typically destroyed before the
  191. // consumer is.
  192. // TODO: If Clang produces diagnostics after check finishes, they'll get
  193. // added to the list of pending diagnostics and never emitted.
  194. CARBON_CHECK(diagnostic_infos_.empty(),
  195. "Missing flush before destroying diagnostic consumer");
  196. }
  197. // Generates a Carbon warning for each Clang warning and a Carbon error for
  198. // each Clang error or fatal.
  199. auto HandleDiagnostic(clang::DiagnosticsEngine::Level diag_level,
  200. const clang::Diagnostic& info) -> void override {
  201. DiagnosticConsumer::HandleDiagnostic(diag_level, info);
  202. SemIR::ImportIRInstId clang_import_ir_inst_id =
  203. AddImportIRInst(*sem_ir_, info.getLocation());
  204. llvm::SmallString<256> message;
  205. info.FormatDiagnostic(message);
  206. // Render a code snippet including any highlighted ranges and fixit hints.
  207. // TODO: Also include the #include stack and macro expansion stack in the
  208. // diagnostic output in some way.
  209. RawStringOstream snippet_stream;
  210. if (!info.hasSourceManager()) {
  211. // If we don't have a source manager, this is an error from early in the
  212. // frontend. Don't produce a snippet.
  213. CARBON_CHECK(info.getLocation().isInvalid());
  214. } else {
  215. CodeContextRenderer(snippet_stream, invocation_->getLangOpts(),
  216. invocation_->getDiagnosticOpts())
  217. .emitDiagnostic(
  218. clang::FullSourceLoc(info.getLocation(), info.getSourceManager()),
  219. diag_level, message, info.getRanges(), info.getFixItHints());
  220. }
  221. diagnostic_infos_.push_back({.level = diag_level,
  222. .import_ir_inst_id = clang_import_ir_inst_id,
  223. .message = message.str().str(),
  224. .snippet = snippet_stream.TakeStr()});
  225. }
  226. // Returns the diagnostic to use for a given Clang diagnostic level.
  227. static auto GetDiagnostic(clang::DiagnosticsEngine::Level level)
  228. -> const Diagnostics::DiagnosticBase<std::string>& {
  229. switch (level) {
  230. case clang::DiagnosticsEngine::Ignored: {
  231. CARBON_FATAL("Emitting an ignored diagnostic");
  232. break;
  233. }
  234. case clang::DiagnosticsEngine::Note: {
  235. CARBON_DIAGNOSTIC(CppInteropParseNote, Note, "{0}", std::string);
  236. return CppInteropParseNote;
  237. }
  238. case clang::DiagnosticsEngine::Remark:
  239. case clang::DiagnosticsEngine::Warning: {
  240. // TODO: Add a distinct Remark level to Carbon diagnostics, and stop
  241. // mapping remarks to warnings.
  242. CARBON_DIAGNOSTIC(CppInteropParseWarning, Warning, "{0}", std::string);
  243. return CppInteropParseWarning;
  244. }
  245. case clang::DiagnosticsEngine::Error:
  246. case clang::DiagnosticsEngine::Fatal: {
  247. CARBON_DIAGNOSTIC(CppInteropParseError, Error, "{0}", std::string);
  248. return CppInteropParseError;
  249. }
  250. }
  251. }
  252. // Outputs Carbon diagnostics based on the collected Clang diagnostics. Must
  253. // be called after the AST is set in the context.
  254. auto EmitDiagnostics() -> void {
  255. CARBON_CHECK(sem_ir_->clang_ast_unit(),
  256. "Attempted to emit diagnostics before the AST Unit is loaded");
  257. for (size_t i = 0; i != diagnostic_infos_.size(); ++i) {
  258. const ClangDiagnosticInfo& info = diagnostic_infos_[i];
  259. auto builder = emitter_->Build(SemIR::LocId(info.import_ir_inst_id),
  260. GetDiagnostic(info.level), info.message);
  261. builder.OverrideSnippet(info.snippet);
  262. for (; i + 1 < diagnostic_infos_.size() &&
  263. diagnostic_infos_[i + 1].level == clang::DiagnosticsEngine::Note;
  264. ++i) {
  265. const ClangDiagnosticInfo& note_info = diagnostic_infos_[i + 1];
  266. builder
  267. .Note(SemIR::LocId(note_info.import_ir_inst_id),
  268. GetDiagnostic(note_info.level), note_info.message)
  269. .OverrideSnippet(note_info.snippet);
  270. }
  271. // TODO: This will apply all current Carbon annotation functions. We
  272. // should instead track how Clang's context notes and Carbon's annotation
  273. // functions are interleaved, and interleave the notes in the same order.
  274. builder.Emit();
  275. }
  276. diagnostic_infos_.clear();
  277. }
  278. private:
  279. // A diagnostics renderer based on clang's TextDiagnostic that captures just
  280. // the code context (the snippet).
  281. class CodeContextRenderer : public clang::TextDiagnostic {
  282. protected:
  283. using TextDiagnostic::TextDiagnostic;
  284. void emitDiagnosticMessage(
  285. clang::FullSourceLoc /*loc*/, clang::PresumedLoc /*ploc*/,
  286. clang::DiagnosticsEngine::Level /*level*/, llvm::StringRef /*message*/,
  287. llvm::ArrayRef<clang::CharSourceRange> /*ranges*/,
  288. clang::DiagOrStoredDiag /*info*/) override {}
  289. void emitDiagnosticLoc(
  290. clang::FullSourceLoc /*loc*/, clang::PresumedLoc /*ploc*/,
  291. clang::DiagnosticsEngine::Level /*level*/,
  292. llvm::ArrayRef<clang::CharSourceRange> /*ranges*/) override {}
  293. // emitCodeContext is inherited from clang::TextDiagnostic.
  294. void emitIncludeLocation(clang::FullSourceLoc /*loc*/,
  295. clang::PresumedLoc /*ploc*/) override {}
  296. void emitImportLocation(clang::FullSourceLoc /*loc*/,
  297. clang::PresumedLoc /*ploc*/,
  298. llvm::StringRef /*module_name*/) override {}
  299. void emitBuildingModuleLocation(clang::FullSourceLoc /*loc*/,
  300. clang::PresumedLoc /*ploc*/,
  301. llvm::StringRef /*module_name*/) override {}
  302. // beginDiagnostic and endDiagnostic are inherited from
  303. // clang::TextDiagnostic in case it wants to do any setup / teardown work.
  304. };
  305. // Information on a Clang diagnostic that can be converted to a Carbon
  306. // diagnostic.
  307. struct ClangDiagnosticInfo {
  308. // The Clang diagnostic level.
  309. clang::DiagnosticsEngine::Level level;
  310. // The ID of the ImportIR instruction referring to the Clang source
  311. // location.
  312. SemIR::ImportIRInstId import_ir_inst_id;
  313. // The Clang diagnostic textual message.
  314. std::string message;
  315. // The code snippet produced by clang.
  316. std::string snippet;
  317. };
  318. // The Carbon file that this C++ compilation is attached to.
  319. SemIR::File* sem_ir_;
  320. // The diagnostic emitter that we're emitting diagnostics into.
  321. DiagnosticEmitterBase* emitter_;
  322. // The compiler invocation that is producing the diagnostics.
  323. std::shared_ptr<clang::CompilerInvocation> invocation_;
  324. // Collects the information for all Clang diagnostics to be converted to
  325. // Carbon diagnostics after the context has been initialized with the Clang
  326. // AST.
  327. llvm::SmallVector<ClangDiagnosticInfo> diagnostic_infos_;
  328. };
  329. // A wrapper around a clang::CompilerInvocation that allows us to make a shallow
  330. // copy of most of the invocation and only make a deep copy of the parts that we
  331. // want to change.
  332. //
  333. // clang::CowCompilerInvocation almost allows this, but doesn't derive from
  334. // CompilerInvocation or support shallow copies from a CompilerInvocation, so is
  335. // not useful to us as we can't build an ASTUnit from it.
  336. class ShallowCopyCompilerInvocation : public clang::CompilerInvocation {
  337. public:
  338. explicit ShallowCopyCompilerInvocation(
  339. const clang::CompilerInvocation& invocation) {
  340. shallow_copy_assign(invocation);
  341. // The preprocessor options are modified to hold a replacement includes
  342. // buffer, so make our own version of those options.
  343. PPOpts = std::make_shared<clang::PreprocessorOptions>(*PPOpts);
  344. }
  345. };
  346. } // namespace
  347. // Returns an AST for the C++ imports and a bool that represents whether
  348. // compilation errors where encountered or the generated AST is null due to an
  349. // error. Sets the AST in the context's `sem_ir`.
  350. // TODO: Consider to always have a (non-null) AST.
  351. static auto GenerateAst(
  352. Context& context, llvm::ArrayRef<Parse::Tree::PackagingNames> imports,
  353. llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs,
  354. std::shared_ptr<clang::CompilerInvocation> base_invocation)
  355. -> std::pair<std::unique_ptr<clang::ASTUnit>, bool> {
  356. auto invocation =
  357. std::make_shared<ShallowCopyCompilerInvocation>(*base_invocation);
  358. // Build a diagnostics engine.
  359. llvm::IntrusiveRefCntPtr<clang::DiagnosticsEngine> diags(
  360. clang::CompilerInstance::createDiagnostics(
  361. *fs, invocation->getDiagnosticOpts(),
  362. new CarbonClangDiagnosticConsumer(context, invocation),
  363. /*ShouldOwnClient=*/true));
  364. // Extract the input from the frontend invocation and make sure it makes
  365. // sense.
  366. const auto& inputs = invocation->getFrontendOpts().Inputs;
  367. CARBON_CHECK(inputs.size() == 1 &&
  368. inputs[0].getKind().getLanguage() == clang::Language::CXX &&
  369. inputs[0].getKind().getFormat() == clang::InputKind::Source);
  370. llvm::StringRef file_name = inputs[0].getFile();
  371. // Remap the imports file name to the corresponding `#include`s.
  372. // TODO: Modify the frontend options to specify this memory buffer as input
  373. // instead of remapping the file.
  374. std::string includes = GenerateCppIncludesHeaderCode(context, imports);
  375. auto includes_buffer =
  376. llvm::MemoryBuffer::getMemBufferCopy(includes, file_name);
  377. invocation->getPreprocessorOpts().addRemappedFile(file_name,
  378. includes_buffer.release());
  379. clang::DiagnosticErrorTrap trap(*diags);
  380. // Create the AST unit.
  381. auto ast = clang::ASTUnit::LoadFromCompilerInvocation(
  382. invocation, std::make_shared<clang::PCHContainerOperations>(), nullptr,
  383. diags, new clang::FileManager(invocation->getFileSystemOpts(), fs));
  384. // Attach the AST to SemIR. This needs to be done before we can emit any
  385. // diagnostics, so their locations can be properly interpreted by our
  386. // diagnostics machinery.
  387. context.sem_ir().set_clang_ast_unit(ast.get());
  388. // Emit any diagnostics we queued up while building the AST.
  389. context.emitter().Flush();
  390. return {std::move(ast), !ast || trap.hasErrorOccurred()};
  391. }
  392. // Adds a namespace for the `Cpp` import and returns its `NameScopeId`.
  393. static auto AddNamespace(Context& context, PackageNameId cpp_package_id,
  394. llvm::ArrayRef<Parse::Tree::PackagingNames> imports)
  395. -> SemIR::NameScopeId {
  396. return AddImportNamespaceToScope(
  397. context,
  398. GetSingletonType(context, SemIR::NamespaceType::TypeInstId),
  399. SemIR::NameId::ForPackageName(cpp_package_id),
  400. SemIR::NameScopeId::Package,
  401. /*diagnose_duplicate_namespace=*/false,
  402. [&]() {
  403. return AddInst<SemIR::ImportCppDecl>(
  404. context,
  405. context.parse_tree().As<Parse::ImportDeclId>(
  406. imports.front().node_id),
  407. {});
  408. })
  409. .add_result.name_scope_id;
  410. }
  411. auto ImportCppFiles(Context& context,
  412. llvm::ArrayRef<Parse::Tree::PackagingNames> imports,
  413. llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem> fs,
  414. std::shared_ptr<clang::CompilerInvocation> invocation)
  415. -> std::unique_ptr<clang::ASTUnit> {
  416. if (imports.empty()) {
  417. return nullptr;
  418. }
  419. CARBON_CHECK(!context.sem_ir().clang_ast_unit());
  420. PackageNameId package_id = imports.front().package_id;
  421. CARBON_CHECK(
  422. llvm::all_of(imports, [&](const Parse::Tree::PackagingNames& import) {
  423. return import.package_id == package_id;
  424. }));
  425. auto name_scope_id = AddNamespace(context, package_id, imports);
  426. auto [generated_ast, ast_has_error] =
  427. GenerateAst(context, imports, fs, std::move(invocation));
  428. SemIR::NameScope& name_scope = context.name_scopes().Get(name_scope_id);
  429. name_scope.set_is_closed_import(true);
  430. name_scope.set_clang_decl_context_id(context.clang_decls().Add(
  431. {.key = SemIR::ClangDeclKey(
  432. generated_ast->getASTContext().getTranslationUnitDecl()),
  433. .inst_id = name_scope.inst_id()}));
  434. if (ast_has_error) {
  435. name_scope.set_has_error();
  436. }
  437. return std::move(generated_ast);
  438. }
  439. // Returns the Clang `DeclContext` for the given name scope. Return the
  440. // translation unit decl if no scope is provided.
  441. static auto GetDeclContext(Context& context, SemIR::NameScopeId scope_id)
  442. -> clang::DeclContext* {
  443. if (!scope_id.has_value()) {
  444. return context.ast_context().getTranslationUnitDecl();
  445. }
  446. auto scope_clang_decl_context_id =
  447. context.name_scopes().Get(scope_id).clang_decl_context_id();
  448. return dyn_cast<clang::DeclContext>(
  449. context.clang_decls().Get(scope_clang_decl_context_id).key.decl);
  450. }
  451. // Returns true if the given Clang declaration is the implicit injected class
  452. // name within the class.
  453. static auto IsDeclInjectedClassName(Context& context,
  454. SemIR::NameScopeId scope_id,
  455. SemIR::NameId name_id,
  456. const clang::NamedDecl* named_decl)
  457. -> bool {
  458. if (!named_decl->isImplicit()) {
  459. return false;
  460. }
  461. const auto* record_decl = dyn_cast<clang::CXXRecordDecl>(named_decl);
  462. if (!record_decl) {
  463. return false;
  464. }
  465. const SemIR::ClangDecl& clang_decl = context.clang_decls().Get(
  466. context.name_scopes().Get(scope_id).clang_decl_context_id());
  467. const auto* scope_record_decl =
  468. cast<clang::CXXRecordDecl>(clang_decl.key.decl);
  469. const clang::ASTContext& ast_context = context.ast_context();
  470. CARBON_CHECK(ast_context.getCanonicalTagType(scope_record_decl) ==
  471. ast_context.getCanonicalTagType(record_decl));
  472. auto class_decl = context.insts().GetAs<SemIR::ClassDecl>(clang_decl.inst_id);
  473. CARBON_CHECK(name_id == context.classes().Get(class_decl.class_id).name_id);
  474. return true;
  475. }
  476. // Returns a Clang DeclarationName for the given `NameId`.
  477. static auto GetDeclarationName(Context& context, SemIR::NameId name_id)
  478. -> std::optional<clang::DeclarationName> {
  479. std::optional<llvm::StringRef> name =
  480. context.names().GetAsStringIfIdentifier(name_id);
  481. if (!name) {
  482. // Special names never exist in C++ code.
  483. return std::nullopt;
  484. }
  485. return clang::DeclarationName(
  486. context.clang_sema().getPreprocessor().getIdentifierInfo(*name));
  487. }
  488. // Performs a qualified name lookup of the declaration name in the given scope.
  489. // Returns the lookup result if lookup was successful.
  490. static auto ClangLookup(Context& context, SemIR::NameScopeId scope_id,
  491. clang::DeclarationName name)
  492. -> std::optional<clang::LookupResult> {
  493. clang::Sema& sema = context.clang_sema();
  494. // TODO: Map the LocId of the lookup to a clang SourceLocation and provide it
  495. // here so that clang's diagnostics can point into the carbon code that uses
  496. // the name.
  497. clang::LookupResult lookup(
  498. sema, clang::DeclarationNameInfo(name, clang::SourceLocation()),
  499. clang::Sema::LookupNameKind::LookupOrdinaryName);
  500. bool found =
  501. sema.LookupQualifiedName(lookup, GetDeclContext(context, scope_id));
  502. if (!found) {
  503. return std::nullopt;
  504. }
  505. return lookup;
  506. }
  507. // Looks up the given name in the Clang AST in a specific scope. Returns the
  508. // lookup result if lookup was successful.
  509. static auto ClangLookupName(Context& context, SemIR::NameScopeId scope_id,
  510. SemIR::NameId name_id)
  511. -> std::optional<clang::LookupResult> {
  512. auto declaration_name = GetDeclarationName(context, name_id);
  513. if (!declaration_name) {
  514. return std::nullopt;
  515. }
  516. return ClangLookup(context, scope_id, *declaration_name);
  517. }
  518. // Returns whether `decl` already mapped to an instruction.
  519. static auto IsClangDeclImported(Context& context, SemIR::ClangDeclKey key)
  520. -> bool {
  521. return context.clang_decls().Lookup(key).has_value();
  522. }
  523. // If `decl` already mapped to an instruction, returns that instruction.
  524. // Otherwise returns `None`.
  525. static auto LookupClangDeclInstId(Context& context, SemIR::ClangDeclKey key)
  526. -> SemIR::InstId {
  527. const auto& clang_decls = context.clang_decls();
  528. if (auto context_clang_decl_id = clang_decls.Lookup(key);
  529. context_clang_decl_id.has_value()) {
  530. return clang_decls.Get(context_clang_decl_id).inst_id;
  531. }
  532. return SemIR::InstId::None;
  533. }
  534. // Returns the parent of the given declaration. Skips declaration types we
  535. // ignore.
  536. static auto GetParentDecl(clang::Decl* clang_decl) -> clang::Decl* {
  537. auto* parent_dc = clang_decl->getDeclContext();
  538. while (!parent_dc->isLookupContext()) {
  539. parent_dc = parent_dc->getParent();
  540. }
  541. return cast<clang::Decl>(parent_dc);
  542. }
  543. // Returns the given declaration's parent scope. Assumes the parent declaration
  544. // was already imported.
  545. static auto GetParentNameScopeId(Context& context, clang::Decl* clang_decl)
  546. -> SemIR::NameScopeId {
  547. auto* parent_decl = GetParentDecl(clang_decl);
  548. if (auto* tag_decl = dyn_cast<clang::TagDecl>(parent_decl)) {
  549. auto class_inst_id =
  550. LookupClangDeclInstId(context, SemIR::ClangDeclKey(tag_decl));
  551. CARBON_CHECK(class_inst_id.has_value());
  552. return context.classes()
  553. .Get(context.insts().GetAs<SemIR::ClassDecl>(class_inst_id).class_id)
  554. .scope_id;
  555. }
  556. if (isa<clang::NamespaceDecl, clang::TranslationUnitDecl>(parent_decl)) {
  557. auto namespace_inst_id = LookupClangDeclInstId(
  558. context, SemIR::ClangDeclKey::ForNonFunctionDecl(parent_decl));
  559. CARBON_CHECK(namespace_inst_id.has_value());
  560. return context.insts()
  561. .GetAs<SemIR::Namespace>(namespace_inst_id)
  562. .name_scope_id;
  563. }
  564. CARBON_FATAL("Unexpected kind of parent {0}", parent_decl->getDeclKindName());
  565. }
  566. // Imports a namespace declaration from Clang to Carbon. If successful, returns
  567. // the new Carbon namespace declaration `InstId`. If the declaration was already
  568. // imported, returns the mapped instruction.
  569. static auto ImportNamespaceDecl(Context& context,
  570. clang::NamespaceDecl* clang_decl)
  571. -> SemIR::InstId {
  572. auto key = SemIR::ClangDeclKey(clang_decl);
  573. // Check if the declaration is already mapped.
  574. if (SemIR::InstId existing_inst_id = LookupClangDeclInstId(context, key);
  575. existing_inst_id.has_value()) {
  576. return existing_inst_id;
  577. }
  578. auto result = AddImportNamespace(
  579. context, GetSingletonType(context, SemIR::NamespaceType::TypeInstId),
  580. AddIdentifierName(context, clang_decl->getName()),
  581. GetParentNameScopeId(context, clang_decl),
  582. /*import_id=*/SemIR::InstId::None);
  583. context.name_scopes()
  584. .Get(result.name_scope_id)
  585. .set_clang_decl_context_id(
  586. context.clang_decls().Add({.key = key, .inst_id = result.inst_id}));
  587. return result.inst_id;
  588. }
  589. static auto ImportTypeAndDependencies(Context& context, SemIR::LocId loc_id,
  590. clang::QualType type) -> TypeExpr;
  591. // Creates a class declaration for the given class name in the given scope.
  592. // Returns the `InstId` for the declaration.
  593. static auto BuildClassDecl(Context& context,
  594. SemIR::ImportIRInstId import_ir_inst_id,
  595. SemIR::NameScopeId parent_scope_id,
  596. SemIR::NameId name_id)
  597. -> std::tuple<SemIR::ClassId, SemIR::TypeInstId> {
  598. // Add the class declaration.
  599. auto class_decl = SemIR::ClassDecl{.type_id = SemIR::TypeType::TypeId,
  600. .class_id = SemIR::ClassId::None,
  601. .decl_block_id = SemIR::InstBlockId::None};
  602. auto class_decl_id = AddPlaceholderImportedInstInNoBlock(
  603. context,
  604. MakeImportedLocIdAndInst(context, import_ir_inst_id, class_decl));
  605. SemIR::Class class_info = {
  606. {.name_id = name_id,
  607. .parent_scope_id = parent_scope_id,
  608. .generic_id = SemIR::GenericId::None,
  609. .first_param_node_id = Parse::NodeId::None,
  610. .last_param_node_id = Parse::NodeId::None,
  611. .pattern_block_id = SemIR::InstBlockId::None,
  612. .implicit_param_patterns_id = SemIR::InstBlockId::None,
  613. .param_patterns_id = SemIR::InstBlockId::None,
  614. .is_extern = false,
  615. .extern_library_id = SemIR::LibraryNameId::None,
  616. .non_owning_decl_id = SemIR::InstId::None,
  617. .first_owning_decl_id = class_decl_id},
  618. {// `.self_type_id` depends on the ClassType, so is set below.
  619. .self_type_id = SemIR::TypeId::None,
  620. // TODO: Support Dynamic classes.
  621. // TODO: Support Final classes.
  622. .inheritance_kind = SemIR::Class::Base}};
  623. class_decl.class_id = context.classes().Add(class_info);
  624. // Write the class ID into the ClassDecl.
  625. ReplaceInstBeforeConstantUse(context, class_decl_id, class_decl);
  626. SetClassSelfType(context, class_decl.class_id);
  627. return {class_decl.class_id, context.types().GetAsTypeInstId(class_decl_id)};
  628. }
  629. // Imports a tag declaration from Clang to Carbon. This covers classes (which
  630. // includes structs and unions) as well as enums. If successful, returns the new
  631. // Carbon class declaration `InstId`.
  632. static auto ImportTagDecl(Context& context, clang::TagDecl* clang_decl)
  633. -> SemIR::InstId {
  634. auto import_ir_inst_id =
  635. AddImportIRInst(context.sem_ir(), clang_decl->getLocation());
  636. auto [class_id, class_inst_id] = BuildClassDecl(
  637. context, import_ir_inst_id, GetParentNameScopeId(context, clang_decl),
  638. AddIdentifierName(context, clang_decl->getName()));
  639. // TODO: The caller does the same lookup. Avoid doing it twice.
  640. auto key = SemIR::ClangDeclKey(clang_decl);
  641. auto clang_decl_id =
  642. context.clang_decls().Add({.key = key, .inst_id = class_inst_id});
  643. // Name lookup into the Carbon class looks in the C++ class definition.
  644. auto& class_info = context.classes().Get(class_id);
  645. class_info.scope_id = context.name_scopes().Add(
  646. class_inst_id, SemIR::NameId::None, class_info.parent_scope_id);
  647. context.name_scopes()
  648. .Get(class_info.scope_id)
  649. .set_clang_decl_context_id(clang_decl_id);
  650. return class_inst_id;
  651. }
  652. // Determines the Carbon inheritance kind to use for a C++ class definition.
  653. static auto GetInheritanceKind(clang::CXXRecordDecl* class_def)
  654. -> SemIR::Class::InheritanceKind {
  655. if (class_def->isUnion()) {
  656. // Treat all unions as final classes to match their C++ semantics. While we
  657. // could support this, the author of a C++ union has no way to mark their
  658. // type as `final` to prevent it, and so we assume the intent was to
  659. // disallow inheritance.
  660. return SemIR::Class::Final;
  661. }
  662. if (class_def->hasAttr<clang::FinalAttr>()) {
  663. // The class is final in C++; don't allow Carbon types to derive from it.
  664. // Note that such a type might also be abstract in C++; we treat final as
  665. // taking precedence.
  666. //
  667. // We could also treat classes with a final destructor as being final, as
  668. // Clang does when determining whether a class is "effectively final", but
  669. // to keep our rules simpler we do not.
  670. return SemIR::Class::Final;
  671. }
  672. if (class_def->isAbstract()) {
  673. // If the class has any abstract members, it's abstract.
  674. return SemIR::Class::Abstract;
  675. }
  676. // Allow inheritance from any other C++ class type.
  677. return SemIR::Class::Base;
  678. }
  679. // Checks that the specified finished class definition is valid and builds and
  680. // returns a corresponding complete type witness instruction.
  681. static auto ImportClassObjectRepr(Context& context, SemIR::ClassId class_id,
  682. SemIR::ImportIRInstId import_ir_inst_id,
  683. SemIR::TypeInstId class_type_inst_id,
  684. const clang::CXXRecordDecl* clang_def)
  685. -> SemIR::TypeInstId {
  686. if (clang_def->isInvalidDecl()) {
  687. // Clang already diagnosed this error.
  688. return SemIR::ErrorInst::TypeInstId;
  689. }
  690. // For now, if the class is empty, produce an empty struct as the object
  691. // representation. This allows our tests to continue to pass while we don't
  692. // properly support initializing imported C++ classes.
  693. // TODO: Remove this.
  694. if (clang_def->isEmpty() && !clang_def->getNumBases()) {
  695. return context.types().GetAsTypeInstId(AddInst(
  696. context,
  697. MakeImportedLocIdAndInst(
  698. context, import_ir_inst_id,
  699. SemIR::StructType{.type_id = SemIR::TypeType::TypeId,
  700. .fields_id = SemIR::StructTypeFieldsId::Empty})));
  701. }
  702. const auto& clang_layout =
  703. context.ast_context().getASTRecordLayout(clang_def);
  704. llvm::SmallVector<uint64_t> layout;
  705. llvm::SmallVector<SemIR::StructTypeField> fields;
  706. static_assert(SemIR::CustomLayoutId::SizeIndex == 0);
  707. layout.push_back(clang_layout.getSize().getQuantity());
  708. static_assert(SemIR::CustomLayoutId::AlignIndex == 1);
  709. layout.push_back(clang_layout.getAlignment().getQuantity());
  710. static_assert(SemIR::CustomLayoutId::FirstFieldIndex == 2);
  711. // TODO: Import vptr(s).
  712. // The kind of base class we've picked so far. These are ordered in increasing
  713. // preference order.
  714. enum class BaseKind {
  715. None,
  716. Empty,
  717. NonEmpty,
  718. Polymorphic,
  719. };
  720. BaseKind base_kind = BaseKind::None;
  721. // Import bases.
  722. for (const auto& base : clang_def->bases()) {
  723. CARBON_CHECK(!base.isVirtual(),
  724. "Should not import definition for class with a virtual base");
  725. auto [base_type_inst_id, base_type_id] =
  726. ImportTypeAndDependencies(context, import_ir_inst_id, base.getType());
  727. if (!base_type_id.has_value()) {
  728. // TODO: If the base class's type can't be mapped, skip it.
  729. continue;
  730. }
  731. auto base_decl_id = AddInst(
  732. context,
  733. MakeImportedLocIdAndInst(
  734. context, import_ir_inst_id,
  735. SemIR::BaseDecl{.type_id = GetUnboundElementType(
  736. context, class_type_inst_id, base_type_inst_id),
  737. .base_type_inst_id = base_type_inst_id,
  738. .index = SemIR::ElementIndex(fields.size())}));
  739. auto* base_class = base.getType()->getAsCXXRecordDecl();
  740. CARBON_CHECK(base_class, "Base class {0} is not a class",
  741. base.getType().getAsString());
  742. // If there's a unique "best" base class, treat it as a Carbon base class
  743. // too.
  744. // TODO: Improve handling for the case where the class has multiple base
  745. // classes.
  746. BaseKind kind = base_class->isPolymorphic() ? BaseKind::Polymorphic
  747. : base_class->isEmpty() ? BaseKind::Empty
  748. : BaseKind::NonEmpty;
  749. auto& class_info = context.classes().Get(class_id);
  750. if (kind > base_kind) {
  751. // This base is better than the previous best.
  752. class_info.base_id = base_decl_id;
  753. base_kind = kind;
  754. } else if (kind == base_kind) {
  755. // Multiple base classes of this kind: no unique best.
  756. class_info.base_id = SemIR::InstId::None;
  757. }
  758. auto base_offset = base.isVirtual()
  759. ? clang_layout.getVBaseClassOffset(base_class)
  760. : clang_layout.getBaseClassOffset(base_class);
  761. layout.push_back(base_offset.getQuantity());
  762. fields.push_back(
  763. {.name_id = SemIR::NameId::Base, .type_inst_id = base_type_inst_id});
  764. }
  765. // Import fields.
  766. for (auto* decl : clang_def->decls()) {
  767. auto* field = dyn_cast<clang::FieldDecl>(decl);
  768. // Track the chain of fields from the class to this field. This chain is
  769. // only one element long unless the field is a member of an anonymous struct
  770. // or union.
  771. clang::NamedDecl* single_field_chain[1] = {field};
  772. llvm::ArrayRef<clang::NamedDecl*> chain = single_field_chain;
  773. // If this isn't a field, it might be an indirect field in an anonymous
  774. // struct or union.
  775. if (!field) {
  776. auto* indirect_field = dyn_cast<clang::IndirectFieldDecl>(decl);
  777. if (!indirect_field) {
  778. continue;
  779. }
  780. chain = indirect_field->chain();
  781. field = indirect_field->getAnonField();
  782. }
  783. if (field->isBitField()) {
  784. // TODO: Add a representation for named bitfield members.
  785. continue;
  786. }
  787. if (field->isAnonymousStructOrUnion()) {
  788. // Fields within an anonymous structure or union will be added via their
  789. // IndirectFieldDecls.
  790. continue;
  791. }
  792. auto field_name_id = AddIdentifierName(context, field->getName());
  793. auto [field_type_inst_id, field_type_id] =
  794. ImportTypeAndDependencies(context, import_ir_inst_id, field->getType());
  795. if (!field_type_inst_id.has_value()) {
  796. // TODO: For now, just skip over fields whose types we can't map.
  797. continue;
  798. }
  799. // Create a field now, as we know the index to use.
  800. // TODO: Consider doing this lazily instead.
  801. auto field_decl_id = AddInst(
  802. context, MakeImportedLocIdAndInst(
  803. context, import_ir_inst_id,
  804. SemIR::FieldDecl{
  805. .type_id = GetUnboundElementType(
  806. context, class_type_inst_id, field_type_inst_id),
  807. .name_id = field_name_id,
  808. .index = SemIR::ElementIndex(fields.size())}));
  809. // The imported SemIR::FieldDecl represents the original declaration `decl`,
  810. // which is either the field or the indirect field declaration.
  811. auto key = SemIR::ClangDeclKey::ForNonFunctionDecl(decl);
  812. context.clang_decls().Add({.key = key, .inst_id = field_decl_id});
  813. // Compute the offset to the field that appears directly in the class.
  814. uint64_t offset = clang_layout.getFieldOffset(
  815. cast<clang::FieldDecl>(chain.front())->getFieldIndex());
  816. // If this is an indirect field, walk the path and accumulate the offset to
  817. // the named field.
  818. for (auto* inner_decl : chain.drop_front()) {
  819. auto* inner_field = cast<clang::FieldDecl>(inner_decl);
  820. const auto& inner_layout =
  821. context.ast_context().getASTRecordLayout(inner_field->getParent());
  822. offset += inner_layout.getFieldOffset(inner_field->getFieldIndex());
  823. }
  824. layout.push_back(
  825. context.ast_context().toCharUnitsFromBits(offset).getQuantity());
  826. fields.push_back(
  827. {.name_id = field_name_id, .type_inst_id = field_type_inst_id});
  828. }
  829. // TODO: Add a field to prevent tail padding reuse if necessary.
  830. return AddTypeInst(context,
  831. MakeImportedLocIdAndInst<SemIR::CustomLayoutType>(
  832. context, import_ir_inst_id,
  833. {.type_id = SemIR::TypeType::TypeId,
  834. .fields_id = context.struct_type_fields().Add(fields),
  835. .layout_id = context.custom_layouts().Add(layout)}));
  836. }
  837. // Creates a Carbon class definition based on the information in the given Clang
  838. // class declaration, which is assumed to be for a class definition.
  839. static auto BuildClassDefinition(Context& context,
  840. SemIR::ImportIRInstId import_ir_inst_id,
  841. SemIR::ClassId class_id,
  842. SemIR::TypeInstId class_inst_id,
  843. clang::CXXRecordDecl* clang_def) -> void {
  844. auto& class_info = context.classes().Get(class_id);
  845. CARBON_CHECK(!class_info.has_definition_started());
  846. class_info.definition_id = class_inst_id;
  847. context.inst_block_stack().Push();
  848. class_info.inheritance_kind = GetInheritanceKind(clang_def);
  849. // Compute the class's object representation.
  850. auto object_repr_id = ImportClassObjectRepr(
  851. context, class_id, import_ir_inst_id, class_inst_id, clang_def);
  852. class_info.complete_type_witness_id = AddInst(
  853. context,
  854. MakeImportedLocIdAndInst<SemIR::CompleteTypeWitness>(
  855. context, import_ir_inst_id,
  856. {.type_id = GetSingletonType(context, SemIR::WitnessType::TypeInstId),
  857. .object_repr_type_inst_id = object_repr_id}));
  858. class_info.body_block_id = context.inst_block_stack().Pop();
  859. }
  860. // Computes and returns the Carbon type to use as the object representation of
  861. // the given C++ enum type. This is a builtin int type matching the enum's
  862. // representation.
  863. static auto ImportEnumObjectRepresentation(
  864. Context& context, SemIR::ImportIRInstId import_ir_inst_id,
  865. clang::EnumDecl* enum_decl) -> SemIR::TypeInstId {
  866. auto int_type = enum_decl->getIntegerType();
  867. CARBON_CHECK(!int_type.isNull(), "incomplete enum type {0}",
  868. enum_decl->getNameAsString());
  869. auto int_kind = int_type->isSignedIntegerType() ? SemIR::IntKind::Signed
  870. : SemIR::IntKind::Unsigned;
  871. auto bit_width_id = GetOrAddInst(
  872. context, MakeImportedLocIdAndInst<SemIR::IntValue>(
  873. context, import_ir_inst_id,
  874. {.type_id = GetSingletonType(
  875. context, SemIR::IntLiteralType::TypeInstId),
  876. .int_id = context.ints().AddUnsigned(llvm::APInt(
  877. 64, context.ast_context().getIntWidth(int_type)))}));
  878. return context.types().GetAsTypeInstId(
  879. GetOrAddInst(context, SemIR::LocIdAndInst::NoLoc(SemIR::IntType{
  880. .type_id = SemIR::TypeType::TypeId,
  881. .int_kind = int_kind,
  882. .bit_width_id = bit_width_id})));
  883. }
  884. // Creates a Carbon class definition based on the information in the given Clang
  885. // enum declaration.
  886. static auto BuildEnumDefinition(Context& context,
  887. SemIR::ImportIRInstId import_ir_inst_id,
  888. SemIR::ClassId class_id,
  889. SemIR::TypeInstId class_inst_id,
  890. clang::EnumDecl* enum_decl) -> void {
  891. auto& class_info = context.classes().Get(class_id);
  892. CARBON_CHECK(!class_info.has_definition_started());
  893. class_info.definition_id = class_inst_id;
  894. context.inst_block_stack().Push();
  895. // Don't allow inheritance from C++ enums, to match the behavior in C++.
  896. class_info.inheritance_kind = SemIR::Class::Final;
  897. // Compute the enum type's object representation. An enum is an adapter for
  898. // the corresponding builtin integer type.
  899. auto object_repr_id =
  900. ImportEnumObjectRepresentation(context, import_ir_inst_id, enum_decl);
  901. class_info.adapt_id = AddInst(
  902. context, MakeImportedLocIdAndInst(
  903. context, import_ir_inst_id,
  904. SemIR::AdaptDecl{.adapted_type_inst_id = object_repr_id}));
  905. class_info.complete_type_witness_id = AddInst(
  906. context,
  907. MakeImportedLocIdAndInst<SemIR::CompleteTypeWitness>(
  908. context, import_ir_inst_id,
  909. {.type_id = GetSingletonType(context, SemIR::WitnessType::TypeInstId),
  910. .object_repr_type_inst_id = object_repr_id}));
  911. class_info.body_block_id = context.inst_block_stack().Pop();
  912. }
  913. // Imports an enumerator declaration from Clang to Carbon.
  914. static auto ImportEnumConstantDecl(Context& context,
  915. clang::EnumConstantDecl* enumerator_decl)
  916. -> SemIR::InstId {
  917. auto key = SemIR::ClangDeclKey(enumerator_decl);
  918. CARBON_CHECK(!IsClangDeclImported(context, key));
  919. // Find the enclosing enum type.
  920. auto enum_key = SemIR::ClangDeclKey(
  921. cast<clang::EnumDecl>(enumerator_decl->getDeclContext()));
  922. auto type_inst_id = LookupClangDeclInstId(context, enum_key);
  923. auto type_id = context.types().GetTypeIdForTypeInstId(type_inst_id);
  924. // Build a corresponding IntValue.
  925. auto int_id = context.ints().Add(enumerator_decl->getInitVal());
  926. auto import_ir_inst_id =
  927. AddImportIRInst(context.sem_ir(), enumerator_decl->getLocation());
  928. auto inst_id = AddInstInNoBlock(
  929. context,
  930. MakeImportedLocIdAndInst<SemIR::IntValue>(
  931. context, import_ir_inst_id, {.type_id = type_id, .int_id = int_id}));
  932. context.imports().push_back(inst_id);
  933. context.clang_decls().Add({.key = key, .inst_id = inst_id});
  934. return inst_id;
  935. }
  936. // Mark the given `key` as failed in `clang_decls`.
  937. static auto MarkFailedDecl(Context& context, SemIR::ClangDeclKey key) {
  938. context.clang_decls().Add({.key = key, .inst_id = SemIR::ErrorInst::InstId});
  939. }
  940. // Creates an integer type of the given size.
  941. static auto MakeIntType(Context& context, IntId size_id, bool is_signed)
  942. -> TypeExpr {
  943. auto type_inst_id = MakeIntTypeLiteral(
  944. context, Parse::NodeId::None,
  945. is_signed ? SemIR::IntKind::Signed : SemIR::IntKind::Unsigned, size_id);
  946. return ExprAsType(context, Parse::NodeId::None, type_inst_id);
  947. }
  948. // Maps a C++ builtin integer type to a Carbon type.
  949. // TODO: Handle integer types that map to named aliases.
  950. static auto MapBuiltinIntegerType(Context& context, SemIR::LocId loc_id,
  951. clang::QualType qual_type,
  952. const clang::BuiltinType& type) -> TypeExpr {
  953. clang::ASTContext& ast_context = context.ast_context();
  954. unsigned width = ast_context.getIntWidth(qual_type);
  955. bool is_signed = type.isSignedInteger();
  956. auto int_n_type = ast_context.getIntTypeForBitwidth(width, is_signed);
  957. if (clang::ASTContext::hasSameType(qual_type, int_n_type)) {
  958. TypeExpr type_expr =
  959. MakeIntType(context, context.ints().Add(width), is_signed);
  960. // Try to make sure integer types of 32 or 64 bits are complete so we can
  961. // check against them when deciding whether we need to generate a thunk.
  962. if (width == 32 || width == 64) {
  963. SemIR::TypeId type_id = type_expr.type_id;
  964. if (!context.types().IsComplete(type_id)) {
  965. TryToCompleteType(context, type_id, loc_id);
  966. }
  967. }
  968. return type_expr;
  969. }
  970. if (clang::ASTContext::hasSameType(qual_type, ast_context.CharTy)) {
  971. return ExprAsType(context, Parse::NodeId::None,
  972. MakeCharTypeLiteral(context, Parse::NodeId::None));
  973. }
  974. if (clang::ASTContext::hasSameType(qual_type, ast_context.LongTy) &&
  975. width == 32) {
  976. return ExprAsType(context, Parse::NodeId::None,
  977. LookupNameInCore(context, Parse::NodeId::None,
  978. {"CppCompat", "Long32"}));
  979. }
  980. return TypeExpr::None;
  981. }
  982. static auto MapNullptrType(Context& context, SemIR::LocId loc_id) -> TypeExpr {
  983. return ExprAsType(
  984. context, loc_id,
  985. LookupNameInCore(context, loc_id, {"CppCompat", "NullptrT"}));
  986. }
  987. // Maps a C++ builtin type to a Carbon type.
  988. // TODO: Support more builtin types.
  989. static auto MapBuiltinType(Context& context, SemIR::LocId loc_id,
  990. clang::QualType qual_type,
  991. const clang::BuiltinType& type) -> TypeExpr {
  992. clang::ASTContext& ast_context = context.ast_context();
  993. if (type.isBooleanType()) {
  994. CARBON_CHECK(ast_context.hasSameType(qual_type, ast_context.BoolTy));
  995. return ExprAsType(context, Parse::NodeId::None,
  996. context.types().GetInstId(GetSingletonType(
  997. context, SemIR::BoolType::TypeInstId)));
  998. }
  999. if (type.isInteger()) {
  1000. return MapBuiltinIntegerType(context, loc_id, qual_type, type);
  1001. }
  1002. if (type.isFloatingPoint()) {
  1003. if (type.isFloat16Type() || type.isFloat32Type() || type.isDoubleType() ||
  1004. type.isFloat128Type()) {
  1005. return ExprAsType(
  1006. context, Parse::NodeId::None,
  1007. MakeFloatTypeLiteral(
  1008. context, Parse::NodeId::None,
  1009. context.ints().Add(ast_context.getTypeSize(qual_type))));
  1010. }
  1011. // TODO: Handle floating-point types that map to named aliases.
  1012. } else if (type.isVoidType()) {
  1013. return ExprAsType(context, Parse::NodeId::None,
  1014. SemIR::CppVoidType::TypeInstId);
  1015. } else if (type.isNullPtrType()) {
  1016. return MapNullptrType(context, loc_id);
  1017. }
  1018. return TypeExpr::None;
  1019. }
  1020. // Determines whether record_decl is a C++ class that has a custom mapping into
  1021. // Carbon, and if so, returns the corresponding Carbon type. Otherwise returns
  1022. // None.
  1023. static auto LookupCustomRecordType(Context& context,
  1024. const clang::CXXRecordDecl* record_decl)
  1025. -> TypeExpr {
  1026. switch (GetCustomCppTypeMapping(record_decl)) {
  1027. case CustomCppTypeMapping::None:
  1028. return TypeExpr::None;
  1029. case CustomCppTypeMapping::Str:
  1030. return MakeStringType(
  1031. context,
  1032. AddImportIRInst(context.sem_ir(), record_decl->getLocation()));
  1033. }
  1034. }
  1035. // Maps a C++ tag type (class, struct, union, enum) to a Carbon type.
  1036. static auto MapTagType(Context& context, const clang::TagType& type)
  1037. -> TypeExpr {
  1038. auto* tag_decl = type.getDecl();
  1039. CARBON_CHECK(tag_decl);
  1040. // Check if the declaration is already mapped.
  1041. auto key = SemIR::ClangDeclKey(tag_decl);
  1042. SemIR::InstId tag_inst_id = LookupClangDeclInstId(context, key);
  1043. if (!tag_inst_id.has_value()) {
  1044. if (auto* record_decl = dyn_cast<clang::CXXRecordDecl>(tag_decl)) {
  1045. auto custom_type = LookupCustomRecordType(context, record_decl);
  1046. if (custom_type.inst_id.has_value()) {
  1047. context.clang_decls().Add({.key = key, .inst_id = custom_type.inst_id});
  1048. return custom_type;
  1049. }
  1050. }
  1051. tag_inst_id = ImportTagDecl(context, tag_decl);
  1052. }
  1053. SemIR::TypeInstId record_type_inst_id =
  1054. context.types().GetAsTypeInstId(tag_inst_id);
  1055. return {
  1056. .inst_id = record_type_inst_id,
  1057. .type_id = context.types().GetTypeIdForTypeInstId(record_type_inst_id)};
  1058. }
  1059. // Maps a C++ type that is not a wrapper type such as a pointer to a Carbon
  1060. // type.
  1061. // TODO: Support more types.
  1062. static auto MapNonWrapperType(Context& context, SemIR::LocId loc_id,
  1063. clang::QualType type) -> TypeExpr {
  1064. if (const auto* builtin_type = type->getAs<clang::BuiltinType>()) {
  1065. return MapBuiltinType(context, loc_id, type, *builtin_type);
  1066. }
  1067. if (const auto* tag_type = type->getAs<clang::TagType>()) {
  1068. return MapTagType(context, *tag_type);
  1069. }
  1070. CARBON_CHECK(!type.hasQualifiers() && !type->isPointerType(),
  1071. "Should not see wrapper types here");
  1072. return TypeExpr::None;
  1073. }
  1074. // Maps a qualified C++ type to a Carbon type.
  1075. static auto MapQualifiedType(Context& context, clang::QualType type,
  1076. TypeExpr type_expr) -> TypeExpr {
  1077. auto quals = type.getQualifiers();
  1078. if (quals.hasConst()) {
  1079. auto type_id = GetConstType(context, type_expr.inst_id);
  1080. type_expr = TypeExpr::ForUnsugared(context, type_id);
  1081. quals.removeConst();
  1082. }
  1083. // TODO: Support other qualifiers.
  1084. if (!quals.empty()) {
  1085. return TypeExpr::None;
  1086. }
  1087. return type_expr;
  1088. }
  1089. // Returns true if the type has the `_Nonnull` attribute.
  1090. static auto IsClangTypeNonNull(clang::QualType type) -> bool {
  1091. auto nullability = type->getNullability();
  1092. return nullability.has_value() &&
  1093. *nullability == clang::NullabilityKind::NonNull;
  1094. }
  1095. // Like `clang::QualType::getUnqualifiedType()`, retrieves the unqualified
  1096. // variant of the given type, but preserves `_Nonnull`.
  1097. static auto ClangGetUnqualifiedTypePreserveNonNull(
  1098. Context& context, clang::QualType original_type) -> clang::QualType {
  1099. clang::QualType type = original_type.getUnqualifiedType();
  1100. // Preserve non-nullability.
  1101. if (IsClangTypeNonNull(original_type) && !IsClangTypeNonNull(type)) {
  1102. type = context.ast_context().getAttributedType(
  1103. clang::NullabilityKind::NonNull, type, type);
  1104. }
  1105. return type;
  1106. }
  1107. // Returns the type `Core.Optional(T)`, where `T` is described by
  1108. // `inner_type_inst_id`.
  1109. static auto MakeOptionalType(Context& context, SemIR::LocId loc_id,
  1110. SemIR::InstId inner_type_inst_id) -> TypeExpr {
  1111. auto fn_inst_id = LookupNameInCore(context, loc_id, "Optional");
  1112. auto call_id = PerformCall(context, loc_id, fn_inst_id, {inner_type_inst_id});
  1113. return ExprAsType(context, loc_id, call_id);
  1114. }
  1115. // Maps a C++ pointer type to a Carbon pointer type.
  1116. static auto MapPointerType(Context& context, SemIR::LocId loc_id,
  1117. clang::QualType type, TypeExpr pointee_type_expr)
  1118. -> TypeExpr {
  1119. CARBON_CHECK(type->isPointerType());
  1120. bool optional =
  1121. !IsClangTypeNonNull(type) &&
  1122. // If the type was produced by C++ template substitution, then we assume
  1123. // it was deduced from a Carbon pointer type, so it's non-null.
  1124. !type->getAs<clang::SubstTemplateTypeParmType>();
  1125. TypeExpr pointer_type_expr = TypeExpr::ForUnsugared(
  1126. context, GetPointerType(context, pointee_type_expr.inst_id));
  1127. if (optional) {
  1128. pointer_type_expr =
  1129. MakeOptionalType(context, loc_id, pointer_type_expr.inst_id);
  1130. }
  1131. return pointer_type_expr;
  1132. }
  1133. // Maps a C++ reference type to a Carbon type. We map all references to
  1134. // pointers for now. Note that when mapping function parameters and return
  1135. // types, a different rule is used; see MapParameterType for details.
  1136. // TODO: Revisit this and decide what we really want to do here.
  1137. static auto MapReferenceType(Context& context, clang::QualType type,
  1138. TypeExpr referenced_type_expr) -> TypeExpr {
  1139. CARBON_CHECK(type->isReferenceType());
  1140. SemIR::TypeId pointer_type_id =
  1141. GetPointerType(context, referenced_type_expr.inst_id);
  1142. pointer_type_id =
  1143. GetConstType(context, context.types().GetInstId(pointer_type_id));
  1144. return TypeExpr::ForUnsugared(context, pointer_type_id);
  1145. }
  1146. // Maps a C++ type to a Carbon type. `type` should not be canonicalized because
  1147. // we check for pointer nullability and nullability will be lost by
  1148. // canonicalization.
  1149. static auto MapType(Context& context, SemIR::LocId loc_id, clang::QualType type)
  1150. -> TypeExpr {
  1151. // Unwrap any type modifiers and wrappers.
  1152. llvm::SmallVector<clang::QualType> wrapper_types;
  1153. while (true) {
  1154. clang::QualType orig_type = type;
  1155. if (type.hasQualifiers()) {
  1156. type = ClangGetUnqualifiedTypePreserveNonNull(context, type);
  1157. } else if (type->isPointerType()) {
  1158. type = type->getPointeeType();
  1159. } else if (type->isReferenceType()) {
  1160. type = type.getNonReferenceType();
  1161. } else {
  1162. break;
  1163. }
  1164. wrapper_types.push_back(orig_type);
  1165. }
  1166. auto mapped = MapNonWrapperType(context, loc_id, type);
  1167. for (auto wrapper : llvm::reverse(wrapper_types)) {
  1168. if (!mapped.inst_id.has_value() ||
  1169. mapped.type_id == SemIR::ErrorInst::TypeId) {
  1170. break;
  1171. }
  1172. if (wrapper.hasQualifiers()) {
  1173. mapped = MapQualifiedType(context, wrapper, mapped);
  1174. } else if (wrapper->isPointerType()) {
  1175. mapped = MapPointerType(context, loc_id, wrapper, mapped);
  1176. } else if (wrapper->isReferenceType()) {
  1177. mapped = MapReferenceType(context, wrapper, mapped);
  1178. } else {
  1179. CARBON_FATAL("Unexpected wrapper type {0}", wrapper.getAsString());
  1180. }
  1181. }
  1182. return mapped;
  1183. }
  1184. namespace {
  1185. // Information about how to map a C++ parameter type into Carbon.
  1186. struct ParameterTypeInfo {
  1187. // The type to use for the Carbon parameter.
  1188. TypeExpr type;
  1189. // Whether to build a `ref` pattern.
  1190. bool want_ref_pattern;
  1191. };
  1192. } // namespace
  1193. // Given the type of a C++ function parameter, returns information about the
  1194. // type to use for the corresponding Carbon parameter.
  1195. //
  1196. // Note that if the parameter has a type for which `IsSimpleAbiType` returns
  1197. // true, we must produce a parameter type that has the same calling convention
  1198. // as the C++ type.
  1199. static auto MapParameterType(Context& context, SemIR::LocId loc_id,
  1200. clang::QualType param_type) -> ParameterTypeInfo {
  1201. ParameterTypeInfo info = {.type = TypeExpr::None, .want_ref_pattern = false};
  1202. // Perform some custom mapping for parameters of reference type:
  1203. //
  1204. // * `T& x` -> `ref x: T`.
  1205. // * `const T& x` -> `x: T`.
  1206. // * `T&& x` -> `x: T`.
  1207. //
  1208. // TODO: For the `&&` mapping, we allow an rvalue reference to bind to a
  1209. // durable reference expression. This should not be allowed.
  1210. if (param_type->isReferenceType()) {
  1211. clang::QualType pointee_type = param_type->getPointeeType();
  1212. if (param_type->isLValueReferenceType()) {
  1213. if (pointee_type.isConstQualified()) {
  1214. // TODO: Consider only doing this if `const` is the only qualifier. For
  1215. // now, any other qualifier will fail when mapping the type.
  1216. auto split_type = pointee_type.getSplitUnqualifiedType();
  1217. split_type.Quals.removeConst();
  1218. pointee_type = context.ast_context().getQualifiedType(split_type);
  1219. } else {
  1220. // The reference will map to a `ref` pattern.
  1221. info.want_ref_pattern = true;
  1222. }
  1223. }
  1224. param_type = pointee_type;
  1225. }
  1226. info.type = MapType(context, loc_id, param_type);
  1227. return info;
  1228. }
  1229. // Returns a block for the implicit parameters of the given function
  1230. // declaration. Because function templates are not yet supported, this currently
  1231. // only contains the `self` parameter. On error, produces a diagnostic and
  1232. // returns None.
  1233. static auto MakeImplicitParamPatternsBlockId(
  1234. Context& context, SemIR::LocId loc_id,
  1235. const clang::FunctionDecl& clang_decl) -> SemIR::InstBlockId {
  1236. const auto* method_decl = dyn_cast<clang::CXXMethodDecl>(&clang_decl);
  1237. if (!method_decl || method_decl->isStatic() ||
  1238. isa<clang::CXXConstructorDecl>(clang_decl)) {
  1239. return SemIR::InstBlockId::Empty;
  1240. }
  1241. // Build a `self` parameter from the object parameter.
  1242. BeginSubpattern(context);
  1243. clang::QualType param_type =
  1244. method_decl->getFunctionObjectParameterReferenceType();
  1245. auto param_info = MapParameterType(context, loc_id, param_type);
  1246. auto [type_inst_id, type_id] = param_info.type;
  1247. SemIR::ExprRegionId type_expr_region_id =
  1248. EndSubpatternAsExpr(context, type_inst_id);
  1249. if (!type_id.has_value()) {
  1250. context.TODO(loc_id,
  1251. llvm::formatv("Unsupported: object parameter type: {0}",
  1252. param_type.getAsString()));
  1253. return SemIR::InstBlockId::None;
  1254. }
  1255. // TODO: Fill in a location once available.
  1256. auto pattern_id = AddParamPattern(context, loc_id, SemIR::NameId::SelfValue,
  1257. type_expr_region_id, type_id,
  1258. param_info.want_ref_pattern);
  1259. return context.inst_blocks().Add({pattern_id});
  1260. }
  1261. // Returns a block id for the explicit parameters of the given function
  1262. // declaration. If the function declaration has no parameters, it returns
  1263. // `SemIR::InstBlockId::Empty`. In the case of an unsupported parameter type, it
  1264. // produces an error and returns `SemIR::InstBlockId::None`.
  1265. // TODO: Consider refactoring to extract and reuse more logic from
  1266. // `HandleAnyBindingPattern()`.
  1267. static auto MakeParamPatternsBlockId(Context& context, SemIR::LocId loc_id,
  1268. const clang::FunctionDecl& clang_decl,
  1269. int num_params) -> SemIR::InstBlockId {
  1270. if (clang_decl.parameters().empty() || num_params == 0) {
  1271. return SemIR::InstBlockId::Empty;
  1272. }
  1273. llvm::SmallVector<SemIR::InstId> params;
  1274. params.reserve(num_params);
  1275. CARBON_CHECK(
  1276. static_cast<int>(clang_decl.getNumNonObjectParams()) >= num_params,
  1277. "varargs functions are not supported");
  1278. const auto* function_type =
  1279. clang_decl.getType()->castAs<clang::FunctionProtoType>();
  1280. for (int i : llvm::seq(num_params)) {
  1281. const auto* param = clang_decl.getNonObjectParameter(i);
  1282. clang::QualType orig_param_type = function_type->getParamType(
  1283. clang_decl.hasCXXExplicitFunctionObjectParameter() + i);
  1284. // The parameter type is decayed but hasn't necessarily had its qualifiers
  1285. // removed.
  1286. // TODO: The presence of qualifiers here is probably a Clang bug.
  1287. clang::QualType param_type =
  1288. ClangGetUnqualifiedTypePreserveNonNull(context, orig_param_type);
  1289. // Mark the start of a region of insts, needed for the type expression
  1290. // created later with the call of `EndSubpatternAsExpr()`.
  1291. BeginSubpattern(context);
  1292. auto param_info = MapParameterType(context, loc_id, param_type);
  1293. auto [orig_type_inst_id, type_id] = param_info.type;
  1294. // Type expression of the binding pattern - a single-entry/single-exit
  1295. // region that allows control flow in the type expression e.g. fn F(x: if C
  1296. // then i32 else i64).
  1297. SemIR::ExprRegionId type_expr_region_id =
  1298. EndSubpatternAsExpr(context, orig_type_inst_id);
  1299. if (!type_id.has_value()) {
  1300. context.TODO(loc_id, llvm::formatv("Unsupported: parameter type: {0}",
  1301. orig_param_type.getAsString()));
  1302. return SemIR::InstBlockId::None;
  1303. }
  1304. llvm::StringRef param_name = param->getName();
  1305. SemIR::NameId name_id =
  1306. param_name.empty()
  1307. // Translate an unnamed parameter to an underscore to
  1308. // match Carbon's naming of unnamed/unused function params.
  1309. ? SemIR::NameId::Underscore
  1310. : AddIdentifierName(context, param_name);
  1311. SemIR::LocId param_loc_id =
  1312. AddImportIRInst(context.sem_ir(), param->getLocation());
  1313. // TODO: Add template support.
  1314. SemIR::InstId pattern_id =
  1315. AddParamPattern(context, param_loc_id, name_id, type_expr_region_id,
  1316. type_id, param_info.want_ref_pattern);
  1317. params.push_back(pattern_id);
  1318. }
  1319. return context.inst_blocks().Add(params);
  1320. }
  1321. // Returns the return `TypeExpr` of the given function declaration. In case of
  1322. // an unsupported return type, returns `SemIR::ErrorInst::InstId`. Constructors
  1323. // are treated as returning a class instance.
  1324. // TODO: Support more return types.
  1325. static auto GetReturnTypeExpr(Context& context, SemIR::LocId loc_id,
  1326. clang::FunctionDecl* clang_decl) -> TypeExpr {
  1327. clang::QualType orig_ret_type = clang_decl->getReturnType();
  1328. if (!orig_ret_type->isVoidType()) {
  1329. // TODO: We should eventually map reference returns to non-pointer types
  1330. // here. We should return by `ref` for `T&` return types once `ref` return
  1331. // is implemented.
  1332. auto [orig_type_inst_id, type_id] = MapType(context, loc_id, orig_ret_type);
  1333. if (!orig_type_inst_id.has_value()) {
  1334. context.TODO(loc_id, llvm::formatv("Unsupported: return type: {0}",
  1335. orig_ret_type.getAsString()));
  1336. return {.inst_id = SemIR::ErrorInst::TypeInstId,
  1337. .type_id = SemIR::ErrorInst::TypeId};
  1338. }
  1339. return {orig_type_inst_id, type_id};
  1340. }
  1341. auto* ctor = dyn_cast<clang::CXXConstructorDecl>(clang_decl);
  1342. if (!ctor) {
  1343. // void.
  1344. return TypeExpr::None;
  1345. }
  1346. // TODO: Make this a `PartialType`.
  1347. SemIR::TypeInstId record_type_inst_id = context.types().GetAsTypeInstId(
  1348. LookupClangDeclInstId(context, SemIR::ClangDeclKey(ctor->getParent())));
  1349. return {
  1350. .inst_id = record_type_inst_id,
  1351. .type_id = context.types().GetTypeIdForTypeInstId(record_type_inst_id)};
  1352. }
  1353. // Returns the return pattern of the given function declaration. In case of an
  1354. // unsupported return type, it produces a diagnostic and returns
  1355. // `SemIR::ErrorInst::InstId`. Constructors are treated as returning a class
  1356. // instance.
  1357. static auto GetReturnPattern(Context& context, SemIR::LocId loc_id,
  1358. clang::FunctionDecl* clang_decl) -> SemIR::InstId {
  1359. auto [type_inst_id, type_id] = GetReturnTypeExpr(context, loc_id, clang_decl);
  1360. if (!type_inst_id.has_value()) {
  1361. // void.
  1362. return SemIR::InstId::None;
  1363. }
  1364. if (type_inst_id == SemIR::ErrorInst::TypeInstId) {
  1365. return SemIR::ErrorInst::InstId;
  1366. }
  1367. auto pattern_type_id = GetPatternType(context, type_id);
  1368. clang::SourceLocation return_type_loc =
  1369. clang_decl->getReturnTypeSourceRange().getBegin();
  1370. if (return_type_loc.isInvalid()) {
  1371. // TODO: While `getReturnTypeSourceRange()` should work, it seems broken for
  1372. // trailing return type. See
  1373. // https://github.com/llvm/llvm-project/issues/162649. Until this is fixed,
  1374. // we fallback to `getTypeSpecStartLoc()`.
  1375. return_type_loc = clang_decl->getTypeSpecStartLoc();
  1376. }
  1377. SemIR::ImportIRInstId return_type_import_ir_inst_id =
  1378. AddImportIRInst(context.sem_ir(), return_type_loc);
  1379. SemIR::InstId return_slot_pattern_id = AddPatternInst(
  1380. context, MakeImportedLocIdAndInst(
  1381. context, return_type_import_ir_inst_id,
  1382. SemIR::ReturnSlotPattern({.type_id = pattern_type_id,
  1383. .type_inst_id = type_inst_id})));
  1384. SemIR::InstId param_pattern_id = AddPatternInst(
  1385. context,
  1386. MakeImportedLocIdAndInst(
  1387. context, return_type_import_ir_inst_id,
  1388. SemIR::OutParamPattern({.type_id = pattern_type_id,
  1389. .subpattern_id = return_slot_pattern_id,
  1390. .index = SemIR::CallParamIndex::None})));
  1391. return param_pattern_id;
  1392. }
  1393. namespace {
  1394. // Represents the parameter patterns block id, the return slot pattern id and
  1395. // the call parameters block id for a function declaration.
  1396. struct FunctionParamsInsts {
  1397. SemIR::InstBlockId implicit_param_patterns_id;
  1398. SemIR::InstBlockId param_patterns_id;
  1399. SemIR::InstId return_slot_pattern_id;
  1400. SemIR::InstBlockId call_params_id;
  1401. };
  1402. } // namespace
  1403. // Creates a block containing the parameter pattern instructions for the
  1404. // explicit parameters, a parameter pattern instruction for the return type and
  1405. // a block containing the call parameters of the function. Emits a callee
  1406. // pattern-match for the explicit parameter patterns and the return slot pattern
  1407. // to create the Call parameters instructions block. Currently the implicit
  1408. // parameter patterns are not taken into account. Returns the parameter patterns
  1409. // block id, the return slot pattern id, and the call parameters block id.
  1410. // Produces a diagnostic and returns `std::nullopt` if the function declaration
  1411. // has an unsupported parameter type.
  1412. static auto CreateFunctionParamsInsts(Context& context, SemIR::LocId loc_id,
  1413. clang::FunctionDecl* clang_decl,
  1414. int num_params)
  1415. -> std::optional<FunctionParamsInsts> {
  1416. if (isa<clang::CXXDestructorDecl>(clang_decl)) {
  1417. context.TODO(loc_id, "Unsupported: Destructor");
  1418. return std::nullopt;
  1419. }
  1420. auto implicit_param_patterns_id =
  1421. MakeImplicitParamPatternsBlockId(context, loc_id, *clang_decl);
  1422. if (!implicit_param_patterns_id.has_value()) {
  1423. return std::nullopt;
  1424. }
  1425. auto param_patterns_id =
  1426. MakeParamPatternsBlockId(context, loc_id, *clang_decl, num_params);
  1427. if (!param_patterns_id.has_value()) {
  1428. return std::nullopt;
  1429. }
  1430. auto return_slot_pattern_id = GetReturnPattern(context, loc_id, clang_decl);
  1431. if (SemIR::ErrorInst::InstId == return_slot_pattern_id) {
  1432. return std::nullopt;
  1433. }
  1434. auto call_params_id =
  1435. CalleePatternMatch(context, implicit_param_patterns_id, param_patterns_id,
  1436. return_slot_pattern_id);
  1437. return {{.implicit_param_patterns_id = implicit_param_patterns_id,
  1438. .param_patterns_id = param_patterns_id,
  1439. .return_slot_pattern_id = return_slot_pattern_id,
  1440. .call_params_id = call_params_id}};
  1441. }
  1442. // Returns the Carbon function name for the given function.
  1443. static auto GetFunctionName(Context& context, clang::FunctionDecl* clang_decl)
  1444. -> SemIR::NameId {
  1445. switch (clang_decl->getDeclName().getNameKind()) {
  1446. case clang::DeclarationName::CXXConstructorName: {
  1447. auto key = SemIR::ClangDeclKey(
  1448. cast<clang::CXXConstructorDecl>(clang_decl)->getParent());
  1449. return context.classes()
  1450. .Get(context.insts()
  1451. .GetAs<SemIR::ClassDecl>(LookupClangDeclInstId(context, key))
  1452. .class_id)
  1453. .name_id;
  1454. }
  1455. case clang::DeclarationName::CXXOperatorName: {
  1456. return SemIR::NameId::CppOperator;
  1457. }
  1458. default: {
  1459. return AddIdentifierName(context, clang_decl->getName());
  1460. }
  1461. }
  1462. }
  1463. // Creates a `FunctionDecl` and a `Function` without C++ thunk information.
  1464. // Returns std::nullopt on failure. The given Clang declaration is assumed to:
  1465. // * Have not been imported before.
  1466. // * Be of supported type (ignoring parameters).
  1467. static auto ImportFunction(Context& context, SemIR::LocId loc_id,
  1468. clang::FunctionDecl* clang_decl, int num_params)
  1469. -> std::optional<SemIR::FunctionId> {
  1470. context.scope_stack().PushForDeclName();
  1471. context.inst_block_stack().Push();
  1472. context.pattern_block_stack().Push();
  1473. auto function_params_insts =
  1474. CreateFunctionParamsInsts(context, loc_id, clang_decl, num_params);
  1475. auto pattern_block_id = context.pattern_block_stack().Pop();
  1476. auto decl_block_id = context.inst_block_stack().Pop();
  1477. context.scope_stack().Pop();
  1478. if (!function_params_insts.has_value()) {
  1479. return std::nullopt;
  1480. }
  1481. auto function_decl = SemIR::FunctionDecl{
  1482. SemIR::TypeId::None, SemIR::FunctionId::None, decl_block_id};
  1483. auto decl_id = AddPlaceholderImportedInstInNoBlock(
  1484. context, SemIR::LocIdAndInst::NoLoc(function_decl));
  1485. auto virtual_modifier = SemIR::Function::VirtualModifier::None;
  1486. int32_t virtual_index = -1;
  1487. if (auto* method_decl = dyn_cast<clang::CXXMethodDecl>(clang_decl)) {
  1488. if (method_decl->size_overridden_methods()) {
  1489. virtual_modifier = SemIR::Function::VirtualModifier::Override;
  1490. } else if (method_decl->isVirtual()) {
  1491. virtual_modifier = SemIR::Function::VirtualModifier::Virtual;
  1492. }
  1493. if (virtual_modifier != SemIR::Function::VirtualModifier::None) {
  1494. // TODO: Add support for Microsoft/non-Itanium vtables.
  1495. virtual_index = dyn_cast<clang::ItaniumVTableContext>(
  1496. context.ast_context().getVTableContext())
  1497. ->getMethodVTableIndex(method_decl);
  1498. }
  1499. }
  1500. auto function_info = SemIR::Function{
  1501. {.name_id = GetFunctionName(context, clang_decl),
  1502. .parent_scope_id = GetParentNameScopeId(context, clang_decl),
  1503. .generic_id = SemIR::GenericId::None,
  1504. .first_param_node_id = Parse::NodeId::None,
  1505. .last_param_node_id = Parse::NodeId::None,
  1506. .pattern_block_id = pattern_block_id,
  1507. .implicit_param_patterns_id =
  1508. function_params_insts->implicit_param_patterns_id,
  1509. .param_patterns_id = function_params_insts->param_patterns_id,
  1510. .is_extern = false,
  1511. .extern_library_id = SemIR::LibraryNameId::None,
  1512. .non_owning_decl_id = SemIR::InstId::None,
  1513. .first_owning_decl_id = decl_id,
  1514. .definition_id = SemIR::InstId::None},
  1515. {.call_params_id = function_params_insts->call_params_id,
  1516. .return_slot_pattern_id = function_params_insts->return_slot_pattern_id,
  1517. .virtual_modifier = virtual_modifier,
  1518. .virtual_index = virtual_index,
  1519. .self_param_id = FindSelfPattern(
  1520. context, function_params_insts->implicit_param_patterns_id),
  1521. .clang_decl_id = context.clang_decls().Add(
  1522. {.key = SemIR::ClangDeclKey::ForFunctionDecl(clang_decl, num_params),
  1523. .inst_id = decl_id})}};
  1524. function_decl.function_id = context.functions().Add(function_info);
  1525. function_decl.type_id = GetFunctionType(context, function_decl.function_id,
  1526. SemIR::SpecificId::None);
  1527. ReplaceInstBeforeConstantUse(context, decl_id, function_decl);
  1528. return function_decl.function_id;
  1529. }
  1530. // Imports a C++ function, returning a corresponding Carbon function.
  1531. // `num_params` specifies how many parameters the corresponding Carbon function
  1532. // should have, which may be fewer than the number of parameters that the C++
  1533. // function has if default arguments are available for the trailing parameters.
  1534. static auto ImportFunctionDecl(Context& context, SemIR::LocId loc_id,
  1535. clang::FunctionDecl* clang_decl, int num_params)
  1536. -> SemIR::InstId {
  1537. auto key = SemIR::ClangDeclKey::ForFunctionDecl(clang_decl, num_params);
  1538. // Check if the declaration is already mapped.
  1539. if (SemIR::InstId existing_inst_id = LookupClangDeclInstId(context, key);
  1540. existing_inst_id.has_value()) {
  1541. return existing_inst_id;
  1542. }
  1543. if (clang_decl->isVariadic()) {
  1544. context.TODO(loc_id, "Unsupported: Variadic function");
  1545. MarkFailedDecl(context, key);
  1546. return SemIR::ErrorInst::InstId;
  1547. }
  1548. if (clang_decl->getTemplatedKind() ==
  1549. clang::FunctionDecl::TK_FunctionTemplate) {
  1550. context.TODO(loc_id, "Unsupported: Template function");
  1551. MarkFailedDecl(context, key);
  1552. return SemIR::ErrorInst::InstId;
  1553. }
  1554. CARBON_CHECK(clang_decl->getFunctionType()->isFunctionProtoType(),
  1555. "Not Prototype function (non-C++ code)");
  1556. auto function_id = ImportFunction(context, loc_id, clang_decl, num_params);
  1557. if (!function_id) {
  1558. MarkFailedDecl(context, key);
  1559. return SemIR::ErrorInst::InstId;
  1560. }
  1561. SemIR::Function& function_info = context.functions().Get(*function_id);
  1562. if (IsCppThunkRequired(context, function_info)) {
  1563. Diagnostics::AnnotationScope annotate_diagnostics(
  1564. &context.emitter(), [&](auto& builder) {
  1565. CARBON_DIAGNOSTIC(InCppThunk, Note,
  1566. "in thunk for C++ function used here");
  1567. builder.Note(loc_id, InCppThunk);
  1568. });
  1569. if (clang::FunctionDecl* thunk_clang_decl =
  1570. BuildCppThunk(context, function_info)) {
  1571. if (auto thunk_function_id =
  1572. ImportFunction(context, loc_id, thunk_clang_decl,
  1573. thunk_clang_decl->getNumParams())) {
  1574. SemIR::InstId thunk_function_decl_id =
  1575. context.functions().Get(*thunk_function_id).first_owning_decl_id;
  1576. function_info.SetHasCppThunk(thunk_function_decl_id);
  1577. }
  1578. }
  1579. }
  1580. return function_info.first_owning_decl_id;
  1581. }
  1582. namespace {
  1583. // An item to be imported in an import worklist.
  1584. // TODO: If worklists ever become particularly large, consider changing this
  1585. // to use a `PointerIntPair`.
  1586. struct ImportItem {
  1587. // A declaration that we want to import.
  1588. SemIR::ClangDeclKey decl_key;
  1589. // Whether we have added `decl`'s dependencies to the worklist.
  1590. bool added_dependencies;
  1591. };
  1592. // A worklist of declarations to import.
  1593. using ImportWorklist = llvm::SmallVector<ImportItem>;
  1594. } // namespace
  1595. // Adds the given declaration to our list of declarations to import.
  1596. static auto AddDependentDecl(Context& context, SemIR::ClangDeclKey decl,
  1597. ImportWorklist& worklist) -> void {
  1598. if (!IsClangDeclImported(context, decl)) {
  1599. worklist.push_back({.decl_key = decl, .added_dependencies = false});
  1600. }
  1601. }
  1602. // Finds all decls that need to be imported before importing the given type and
  1603. // adds them to the given set.
  1604. static auto AddDependentUnimportedTypeDecls(Context& context,
  1605. clang::QualType type,
  1606. ImportWorklist& worklist) -> void {
  1607. while (true) {
  1608. if (type->isPointerType() || type->isReferenceType()) {
  1609. type = type->getPointeeType();
  1610. } else if (const clang::ArrayType* array_type =
  1611. type->getAsArrayTypeUnsafe()) {
  1612. type = array_type->getElementType();
  1613. } else {
  1614. break;
  1615. }
  1616. }
  1617. if (const auto* tag_type = type->getAs<clang::TagType>()) {
  1618. AddDependentDecl(context, SemIR::ClangDeclKey(tag_type->getDecl()),
  1619. worklist);
  1620. }
  1621. }
  1622. // Finds all decls that need to be imported before importing the given function
  1623. // and adds them to the given set.
  1624. static auto AddDependentUnimportedFunctionDecls(
  1625. Context& context, const clang::FunctionDecl& clang_decl, int num_params,
  1626. ImportWorklist& worklist) -> void {
  1627. const auto* function_type =
  1628. clang_decl.getType()->castAs<clang::FunctionProtoType>();
  1629. for (int i : llvm::seq(clang_decl.hasCXXExplicitFunctionObjectParameter() +
  1630. num_params)) {
  1631. AddDependentUnimportedTypeDecls(context, function_type->getParamType(i),
  1632. worklist);
  1633. }
  1634. AddDependentUnimportedTypeDecls(context, clang_decl.getReturnType(),
  1635. worklist);
  1636. }
  1637. // Finds all decls that need to be imported before importing the given
  1638. // declaration and adds them to the given set.
  1639. static auto AddDependentUnimportedDecls(Context& context,
  1640. SemIR::ClangDeclKey key,
  1641. ImportWorklist& worklist) -> void {
  1642. clang::Decl* clang_decl = key.decl;
  1643. if (auto* clang_function_decl = clang_decl->getAsFunction()) {
  1644. AddDependentUnimportedFunctionDecls(context, *clang_function_decl,
  1645. key.num_params, worklist);
  1646. } else if (auto* type_decl = dyn_cast<clang::TypeDecl>(clang_decl)) {
  1647. if (!isa<clang::TagDecl>(clang_decl)) {
  1648. AddDependentUnimportedTypeDecls(
  1649. context, type_decl->getASTContext().getTypeDeclType(type_decl),
  1650. worklist);
  1651. }
  1652. }
  1653. auto* parent = GetParentDecl(clang_decl);
  1654. if (llvm::isa_and_nonnull<clang::TagDecl, clang::NamespaceDecl,
  1655. clang::TranslationUnitDecl>(parent)) {
  1656. AddDependentDecl(context, SemIR::ClangDeclKey::ForNonFunctionDecl(parent),
  1657. worklist);
  1658. }
  1659. }
  1660. static auto ImportVarDecl(Context& context, SemIR::LocId loc_id,
  1661. clang::VarDecl* var_decl) -> SemIR::InstId {
  1662. if (SemIR::InstId existing_inst_id =
  1663. LookupClangDeclInstId(context, SemIR::ClangDeclKey(var_decl));
  1664. existing_inst_id.has_value()) {
  1665. return existing_inst_id;
  1666. }
  1667. // Extract type and name.
  1668. clang::QualType var_type = var_decl->getType();
  1669. SemIR::TypeId var_type_id = MapType(context, loc_id, var_type).type_id;
  1670. if (!var_type_id.has_value()) {
  1671. context.TODO(loc_id, llvm::formatv("Unsupported: var type: {0}",
  1672. var_type.getAsString()));
  1673. return SemIR::ErrorInst::InstId;
  1674. }
  1675. SemIR::NameId var_name_id = AddIdentifierName(context, var_decl->getName());
  1676. SemIR::VarStorage var_storage{.type_id = var_type_id,
  1677. .pattern_id = SemIR::InstId::None};
  1678. // We can't use the convenience for `AddPlaceholderInstInNoBlock()` with typed
  1679. // nodes because it doesn't support insts with cleanup.
  1680. SemIR::InstId var_storage_inst_id =
  1681. AddPlaceholderImportedInstInNoBlock(context, {loc_id, var_storage});
  1682. auto clang_decl_id = context.clang_decls().Add(
  1683. {.key = SemIR::ClangDeclKey(var_decl), .inst_id = var_storage_inst_id});
  1684. // Entity name referring to a Clang decl for mangling.
  1685. SemIR::EntityNameId entity_name_id =
  1686. context.entity_names().AddSymbolicBindingName(
  1687. var_name_id, GetParentNameScopeId(context, var_decl),
  1688. SemIR::CompileTimeBindIndex::None, false);
  1689. context.cpp_global_names().Add({.key = {.entity_name_id = entity_name_id},
  1690. .clang_decl_id = clang_decl_id});
  1691. // Create `RefBindingPattern` and `VarPattern` in a `NameBindingDecl`.
  1692. context.pattern_block_stack().Push();
  1693. SemIR::TypeId pattern_type_id = GetPatternType(context, var_type_id);
  1694. SemIR::InstId binding_pattern_inst_id =
  1695. AddPatternInst<SemIR::RefBindingPattern>(
  1696. context, loc_id,
  1697. {.type_id = pattern_type_id, .entity_name_id = entity_name_id});
  1698. var_storage.pattern_id = AddPatternInst<SemIR::VarPattern>(
  1699. context, Parse::VariablePatternId::None,
  1700. {.type_id = pattern_type_id, .subpattern_id = binding_pattern_inst_id});
  1701. context.imports().push_back(AddInstInNoBlock<SemIR::NameBindingDecl>(
  1702. context, loc_id,
  1703. {.pattern_block_id = context.pattern_block_stack().Pop()}));
  1704. // Finalize the `VarStorage` instruction.
  1705. ReplaceInstBeforeConstantUse(context, var_storage_inst_id, var_storage);
  1706. return var_storage_inst_id;
  1707. }
  1708. // Imports a declaration from Clang to Carbon. Returns the instruction for the
  1709. // new Carbon declaration, which will be an ErrorInst on failure. Assumes all
  1710. // dependencies have already been imported.
  1711. static auto ImportDeclAfterDependencies(Context& context, SemIR::LocId loc_id,
  1712. SemIR::ClangDeclKey key)
  1713. -> SemIR::InstId {
  1714. clang::Decl* clang_decl = key.decl;
  1715. if (auto* clang_function_decl = clang_decl->getAsFunction()) {
  1716. return ImportFunctionDecl(context, loc_id, clang_function_decl,
  1717. key.num_params);
  1718. }
  1719. if (auto* clang_namespace_decl = dyn_cast<clang::NamespaceDecl>(clang_decl)) {
  1720. return ImportNamespaceDecl(context, clang_namespace_decl);
  1721. }
  1722. if (auto* type_decl = dyn_cast<clang::TypeDecl>(clang_decl)) {
  1723. auto type = clang_decl->getASTContext().getTypeDeclType(type_decl);
  1724. auto type_inst_id = MapType(context, loc_id, type).inst_id;
  1725. if (!type_inst_id.has_value()) {
  1726. context.TODO(AddImportIRInst(context.sem_ir(), type_decl->getLocation()),
  1727. llvm::formatv("Unsupported: Type declaration: {0}",
  1728. type.getAsString()));
  1729. return SemIR::ErrorInst::InstId;
  1730. }
  1731. context.clang_decls().Add({.key = key, .inst_id = type_inst_id});
  1732. return type_inst_id;
  1733. }
  1734. if (isa<clang::FieldDecl, clang::IndirectFieldDecl>(clang_decl)) {
  1735. // Usable fields get imported as a side effect of importing the class.
  1736. if (SemIR::InstId existing_inst_id = LookupClangDeclInstId(context, key);
  1737. existing_inst_id.has_value()) {
  1738. return existing_inst_id;
  1739. }
  1740. context.TODO(AddImportIRInst(context.sem_ir(), clang_decl->getLocation()),
  1741. "Unsupported: field declaration has unhandled type or kind");
  1742. return SemIR::ErrorInst::InstId;
  1743. }
  1744. if (auto* enum_const_decl = dyn_cast<clang::EnumConstantDecl>(clang_decl)) {
  1745. return ImportEnumConstantDecl(context, enum_const_decl);
  1746. }
  1747. if (auto* var_decl = dyn_cast<clang::VarDecl>(clang_decl)) {
  1748. return ImportVarDecl(context, loc_id, var_decl);
  1749. }
  1750. context.TODO(AddImportIRInst(context.sem_ir(), clang_decl->getLocation()),
  1751. llvm::formatv("Unsupported: Declaration type {0}",
  1752. clang_decl->getDeclKindName()));
  1753. return SemIR::ErrorInst::InstId;
  1754. }
  1755. // Attempts to import a set of declarations. Returns `false` if an error was
  1756. // produced, `true` otherwise.
  1757. static auto ImportDeclSet(Context& context, SemIR::LocId loc_id,
  1758. ImportWorklist& worklist) -> bool {
  1759. // Walk the dependency graph in depth-first order, and import declarations
  1760. // once we've imported all of their dependencies.
  1761. while (!worklist.empty()) {
  1762. auto& item = worklist.back();
  1763. if (!item.added_dependencies) {
  1764. // Skip items we've already imported. We checked this when initially
  1765. // adding the item to the worklist, but it might have been added to the
  1766. // worklist twice before the first time we visited it. For example, this
  1767. // happens for `fn F(a: Cpp.T, b: Cpp.T)`.
  1768. if (IsClangDeclImported(context, item.decl_key)) {
  1769. worklist.pop_back();
  1770. continue;
  1771. }
  1772. // First time visiting this declaration (preorder): add its dependencies
  1773. // to the work list.
  1774. item.added_dependencies = true;
  1775. AddDependentUnimportedDecls(context, item.decl_key, worklist);
  1776. } else {
  1777. // Second time visiting this declaration (postorder): its dependencies are
  1778. // already imported, so we can import it now.
  1779. auto decl_key = worklist.pop_back_val().decl_key;
  1780. auto inst_id = ImportDeclAfterDependencies(context, loc_id, decl_key);
  1781. CARBON_CHECK(inst_id.has_value());
  1782. if (inst_id == SemIR::ErrorInst::InstId) {
  1783. return false;
  1784. }
  1785. CARBON_CHECK(IsClangDeclImported(context, decl_key));
  1786. }
  1787. }
  1788. return true;
  1789. }
  1790. // Imports a declaration from Clang to Carbon. If successful, returns the
  1791. // instruction for the new Carbon declaration. All unimported dependencies are
  1792. // imported first.
  1793. static auto ImportDeclAndDependencies(Context& context, SemIR::LocId loc_id,
  1794. SemIR::ClangDeclKey key)
  1795. -> SemIR::InstId {
  1796. // Collect dependencies by walking the dependency graph in depth-first order.
  1797. ImportWorklist worklist;
  1798. AddDependentDecl(context, key, worklist);
  1799. if (!ImportDeclSet(context, loc_id, worklist)) {
  1800. return SemIR::ErrorInst::InstId;
  1801. }
  1802. return LookupClangDeclInstId(context, key);
  1803. }
  1804. // Imports a type from Clang to Carbon. If successful, returns the imported
  1805. // TypeId. All unimported dependencies are imported first.
  1806. static auto ImportTypeAndDependencies(Context& context, SemIR::LocId loc_id,
  1807. clang::QualType type) -> TypeExpr {
  1808. // Collect dependencies by walking the dependency graph in depth-first order.
  1809. ImportWorklist worklist;
  1810. AddDependentUnimportedTypeDecls(context, type, worklist);
  1811. if (!ImportDeclSet(context, loc_id, worklist)) {
  1812. return {.inst_id = SemIR::ErrorInst::TypeInstId,
  1813. .type_id = SemIR::ErrorInst::TypeId};
  1814. }
  1815. return MapType(context, loc_id, type);
  1816. }
  1817. auto ImportCppFunctionDecl(Context& context, SemIR::LocId loc_id,
  1818. clang::FunctionDecl* clang_decl, int num_params)
  1819. -> SemIR::InstId {
  1820. return ImportDeclAndDependencies(
  1821. context, loc_id,
  1822. SemIR::ClangDeclKey::ForFunctionDecl(clang_decl, num_params));
  1823. }
  1824. // Imports a Clang declaration into Carbon and adds that name into the
  1825. // `NameScope`.
  1826. static auto ImportNameDeclIntoScope(Context& context, SemIR::LocId loc_id,
  1827. SemIR::NameScopeId scope_id,
  1828. SemIR::NameId name_id,
  1829. SemIR::ClangDeclKey key,
  1830. SemIR::AccessKind access_kind)
  1831. -> SemIR::ScopeLookupResult {
  1832. SemIR::InstId inst_id = ImportDeclAndDependencies(context, loc_id, key);
  1833. if (!inst_id.has_value()) {
  1834. return SemIR::ScopeLookupResult::MakeNotFound();
  1835. }
  1836. AddNameToScope(context, scope_id, name_id, access_kind, inst_id);
  1837. return SemIR::ScopeLookupResult::MakeWrappedLookupResult(inst_id,
  1838. access_kind);
  1839. }
  1840. // Returns true if the scope is the top `Cpp` scope.
  1841. static auto IsTopCppScope(Context& context, SemIR::NameScopeId scope_id)
  1842. -> bool {
  1843. const SemIR::NameScope& name_scope = context.name_scopes().Get(scope_id);
  1844. CARBON_CHECK(name_scope.is_cpp_scope());
  1845. return name_scope.parent_scope_id() == SemIR::NameScopeId::Package;
  1846. }
  1847. // For a builtin name like `Cpp.long`, returns the associated type.
  1848. static auto LookupBuiltinName(Context& context, SemIR::LocId loc_id,
  1849. SemIR::NameScopeId scope_id,
  1850. SemIR::NameId name_id) -> SemIR::InstId {
  1851. if (!IsTopCppScope(context, scope_id)) {
  1852. return SemIR::InstId::None;
  1853. }
  1854. auto name = context.names().GetAsStringIfIdentifier(name_id);
  1855. if (!name) {
  1856. return SemIR::InstId::None;
  1857. }
  1858. const clang::ASTContext& ast_context = context.ast_context();
  1859. // List of types based on
  1860. // https://github.com/carbon-language/carbon-lang/blob/trunk/proposals/p5448.md#details
  1861. auto builtin_type =
  1862. llvm::StringSwitch<clang::QualType>(*name)
  1863. .Case("signed_char", ast_context.SignedCharTy)
  1864. .Case("short", ast_context.ShortTy)
  1865. .Case("int", ast_context.IntTy)
  1866. .Case("long", ast_context.LongTy)
  1867. .Case("long_long", ast_context.LongLongTy)
  1868. .Case("unsigned_char", ast_context.UnsignedCharTy)
  1869. .Case("unsigned_short", ast_context.UnsignedShortTy)
  1870. .Case("unsigned_int", ast_context.UnsignedIntTy)
  1871. .Case("unsigned_long", ast_context.UnsignedLongTy)
  1872. .Case("unsigned_long_long", ast_context.UnsignedLongLongTy)
  1873. .Case("float", ast_context.FloatTy)
  1874. .Case("double", ast_context.DoubleTy)
  1875. .Case("long_double", ast_context.LongDoubleTy)
  1876. .Case("void", ast_context.VoidTy)
  1877. .Default(clang::QualType());
  1878. if (builtin_type.isNull()) {
  1879. if (*name == "nullptr") {
  1880. // Map `Cpp.nullptr` to an uninitialized value of type `Core.CppNullptrT`.
  1881. auto type_id = MapNullptrType(context, loc_id).type_id;
  1882. return GetOrAddInst<SemIR::UninitializedValue>(
  1883. context, SemIR::LocId::None, {.type_id = type_id});
  1884. }
  1885. return SemIR::InstId::None;
  1886. }
  1887. SemIR::InstId inst_id =
  1888. MapNonWrapperType(context, loc_id, builtin_type).inst_id;
  1889. if (!inst_id.has_value()) {
  1890. context.TODO(loc_id, llvm::formatv("Unsupported: builtin type: {0}",
  1891. builtin_type.getAsString()));
  1892. return SemIR::ErrorInst::InstId;
  1893. }
  1894. return inst_id;
  1895. }
  1896. auto ImportCppOverloadSet(
  1897. Context& context, SemIR::LocId loc_id, SemIR::NameScopeId scope_id,
  1898. SemIR::NameId name_id, clang::CXXRecordDecl* naming_class,
  1899. clang::UnresolvedSet<4>&& overload_set,
  1900. clang::OverloadCandidateSet::OperatorRewriteInfo operator_rewrite_info)
  1901. -> SemIR::InstId {
  1902. SemIR::CppOverloadSetId overload_set_id = context.cpp_overload_sets().Add(
  1903. SemIR::CppOverloadSet{.name_id = name_id,
  1904. .parent_scope_id = scope_id,
  1905. .naming_class = naming_class,
  1906. .candidate_functions = std::move(overload_set),
  1907. .operator_rewrite_info = operator_rewrite_info});
  1908. auto overload_set_inst_id = AddInstInNoBlock<SemIR::CppOverloadSetValue>(
  1909. context, loc_id,
  1910. {.type_id = GetCppOverloadSetType(context, overload_set_id,
  1911. SemIR::SpecificId::None),
  1912. .overload_set_id = overload_set_id});
  1913. context.imports().push_back(overload_set_inst_id);
  1914. return overload_set_inst_id;
  1915. }
  1916. // Gets the best access for an overloaded function set. This is the access that
  1917. // we use for the overload set as a whole. More fine-grained checking is done
  1918. // after overload resolution.
  1919. static auto GetOverloadSetAccess(const clang::UnresolvedSet<4>& overload_set)
  1920. -> SemIR::AccessKind {
  1921. SemIR::AccessKind access_kind = SemIR::AccessKind::Private;
  1922. for (clang::DeclAccessPair overload : overload_set.pairs()) {
  1923. access_kind = std::min(access_kind, MapCppAccess(overload));
  1924. if (access_kind == SemIR::AccessKind::Public) {
  1925. break;
  1926. }
  1927. }
  1928. return access_kind;
  1929. }
  1930. // Imports an overload set from Clang to Carbon and adds the name into the
  1931. // `NameScope`.
  1932. static auto ImportOverloadSetIntoScope(Context& context, SemIR::LocId loc_id,
  1933. SemIR::NameScopeId scope_id,
  1934. SemIR::NameId name_id,
  1935. clang::CXXRecordDecl* naming_class,
  1936. clang::UnresolvedSet<4>&& overload_set)
  1937. -> SemIR::ScopeLookupResult {
  1938. SemIR::AccessKind access_kind = GetOverloadSetAccess(overload_set);
  1939. SemIR::InstId inst_id = ImportCppOverloadSet(
  1940. context, loc_id, scope_id, name_id, naming_class, std::move(overload_set),
  1941. /*operator_rewrite_info=*/{});
  1942. AddNameToScope(context, scope_id, name_id, access_kind, inst_id);
  1943. return SemIR::ScopeLookupResult::MakeWrappedLookupResult(inst_id,
  1944. access_kind);
  1945. }
  1946. // Imports the constructors for a given class name. The found constructors are
  1947. // imported as part of an overload set into the scope. Currently copy/move
  1948. // constructors are not imported.
  1949. static auto ImportConstructorsIntoScope(Context& context, SemIR::LocId loc_id,
  1950. SemIR::NameScopeId scope_id,
  1951. SemIR::NameId name_id)
  1952. -> SemIR::ScopeLookupResult {
  1953. auto* naming_class =
  1954. cast<clang::CXXRecordDecl>(GetDeclContext(context, scope_id));
  1955. clang::DeclContextLookupResult constructors_lookup =
  1956. context.clang_sema().LookupConstructors(naming_class);
  1957. clang::UnresolvedSet<4> overload_set;
  1958. for (auto* decl : constructors_lookup) {
  1959. auto info = clang::getConstructorInfo(decl);
  1960. if (!info.Constructor || info.Constructor->isCopyOrMoveConstructor()) {
  1961. continue;
  1962. }
  1963. overload_set.addDecl(info.FoundDecl, info.FoundDecl.getAccess());
  1964. }
  1965. if (overload_set.empty()) {
  1966. return SemIR::ScopeLookupResult::MakeNotFound();
  1967. }
  1968. return ImportOverloadSetIntoScope(context, loc_id, scope_id, name_id,
  1969. naming_class, std::move(overload_set));
  1970. }
  1971. // Attempts to import a builtin name from Clang to Carbon and adds the name into
  1972. // the scope.
  1973. static auto ImportBuiltinNameIntoScope(Context& context, SemIR::LocId loc_id,
  1974. SemIR::NameScopeId scope_id,
  1975. SemIR::NameId name_id)
  1976. -> SemIR::ScopeLookupResult {
  1977. SemIR::InstId builtin_inst_id =
  1978. LookupBuiltinName(context, loc_id, scope_id, name_id);
  1979. if (builtin_inst_id.has_value()) {
  1980. AddNameToScope(context, scope_id, name_id, SemIR::AccessKind::Public,
  1981. builtin_inst_id);
  1982. return SemIR::ScopeLookupResult::MakeWrappedLookupResult(
  1983. builtin_inst_id, SemIR::AccessKind::Public);
  1984. }
  1985. return SemIR::ScopeLookupResult::MakeNotFound();
  1986. }
  1987. // Checks if the name scope is a class that is not complete.
  1988. static auto IsIncompleteClass(Context& context, SemIR::NameScopeId scope_id)
  1989. -> bool {
  1990. auto class_decl = context.insts().TryGetAs<SemIR::ClassDecl>(
  1991. context.name_scopes().Get(scope_id).inst_id());
  1992. return class_decl.has_value() &&
  1993. !context.types().IsComplete(
  1994. context.classes().Get(class_decl->class_id).self_type_id);
  1995. }
  1996. // Maps a Clang constant expression to a Carbon constant. Currently supports
  1997. // only integer constants.
  1998. // TODO: Add support for the other constant types for which a C++ to Carbon type
  1999. // mapping exists.
  2000. static auto MapConstant(Context& context, SemIR::LocId loc_id,
  2001. clang::Expr* expr) -> SemIR::InstId {
  2002. CARBON_CHECK(expr, "empty expression");
  2003. auto* integer_literal = dyn_cast<clang::IntegerLiteral>(expr);
  2004. if (!integer_literal) {
  2005. context.TODO(
  2006. loc_id, "Unsupported: constant type: " + expr->getType().getAsString());
  2007. return SemIR::ErrorInst::InstId;
  2008. }
  2009. SemIR::TypeId type_id =
  2010. MapType(context, loc_id, integer_literal->getType()).type_id;
  2011. if (!type_id.has_value()) {
  2012. CARBON_DIAGNOSTIC(InCppConstantMapping, Error, "invalid integer type");
  2013. context.emitter().Emit(loc_id, InCppConstantMapping);
  2014. return SemIR::ErrorInst::InstId;
  2015. }
  2016. auto int_id = context.ints().Add(integer_literal->getValue().getSExtValue());
  2017. auto inst_id = AddInstInNoBlock<SemIR::IntValue>(
  2018. context, loc_id, {.type_id = type_id, .int_id = int_id});
  2019. context.imports().push_back(inst_id);
  2020. return inst_id;
  2021. }
  2022. // Imports a macro definition into the scope. Currently supports only simple
  2023. // object-like macros that expand to a constant integer value.
  2024. // TODO: Add support for other macro types and non-integer literal values.
  2025. static auto ImportMacro(Context& context, SemIR::LocId loc_id,
  2026. SemIR::NameScopeId scope_id, SemIR::NameId name_id,
  2027. clang::MacroInfo* macro_info)
  2028. -> SemIR::ScopeLookupResult {
  2029. clang::Expr* macro_expr =
  2030. TryEvaluateMacroToConstant(context, loc_id, name_id, macro_info);
  2031. if (!macro_expr) {
  2032. return SemIR::ScopeLookupResult::MakeNotFound();
  2033. }
  2034. auto inst_id = MapConstant(context, loc_id, macro_expr);
  2035. if (inst_id == SemIR::ErrorInst::InstId) {
  2036. return SemIR::ScopeLookupResult::MakeNotFound();
  2037. }
  2038. AddNameToScope(context, scope_id, name_id, SemIR::AccessKind::Public,
  2039. inst_id);
  2040. return SemIR::ScopeLookupResult::MakeWrappedLookupResult(
  2041. inst_id, SemIR::AccessKind::Public);
  2042. }
  2043. // Looks up a macro definition in the top-level `Cpp` scope. Returns nullptr if
  2044. // the macro is not found or the scope is not the top-level `Cpp` scope.
  2045. static auto LookupMacro(Context& context, SemIR::NameScopeId scope_id,
  2046. SemIR::NameId name_id) -> clang::MacroInfo* {
  2047. auto name_str_opt = context.names().GetAsStringIfIdentifier(name_id);
  2048. if (!name_str_opt || !IsTopCppScope(context, scope_id)) {
  2049. return nullptr;
  2050. }
  2051. clang::Preprocessor& preprocessor = context.clang_sema().getPreprocessor();
  2052. // TODO: Do the identifier lookup only once, rather than both here and in
  2053. // ClangLookupName.
  2054. clang::IdentifierInfo* identifier_info =
  2055. preprocessor.getIdentifierInfo(*name_str_opt);
  2056. if (!identifier_info) {
  2057. return nullptr;
  2058. }
  2059. clang::MacroInfo* macro_info = preprocessor.getMacroInfo(identifier_info);
  2060. if (macro_info && !macro_info->isUsedForHeaderGuard() &&
  2061. !macro_info->isFunctionLike() && !macro_info->isBuiltinMacro()) {
  2062. return macro_info;
  2063. }
  2064. return nullptr;
  2065. }
  2066. auto ImportNameFromCpp(Context& context, SemIR::LocId loc_id,
  2067. SemIR::NameScopeId scope_id, SemIR::NameId name_id)
  2068. -> SemIR::ScopeLookupResult {
  2069. Diagnostics::AnnotationScope annotate_diagnostics(
  2070. &context.emitter(), [&](auto& builder) {
  2071. CARBON_DIAGNOSTIC(InCppNameLookup, Note,
  2072. "in `Cpp` name lookup for `{0}`", SemIR::NameId);
  2073. builder.Note(loc_id, InCppNameLookup, name_id);
  2074. });
  2075. if (IsIncompleteClass(context, scope_id)) {
  2076. return SemIR::ScopeLookupResult::MakeError();
  2077. }
  2078. if (clang::MacroInfo* macro_info = LookupMacro(context, scope_id, name_id)) {
  2079. return ImportMacro(context, loc_id, scope_id, name_id, macro_info);
  2080. }
  2081. auto lookup = ClangLookupName(context, scope_id, name_id);
  2082. if (!lookup) {
  2083. return ImportBuiltinNameIntoScope(context, loc_id, scope_id, name_id);
  2084. }
  2085. // Access checks are performed separately by the Carbon name lookup logic.
  2086. lookup->suppressAccessDiagnostics();
  2087. if (lookup->isOverloadedResult() ||
  2088. (lookup->isSingleResult() &&
  2089. lookup->getFoundDecl()->isFunctionOrFunctionTemplate())) {
  2090. clang::UnresolvedSet<4> overload_set;
  2091. overload_set.append(lookup->begin(), lookup->end());
  2092. return ImportOverloadSetIntoScope(context, loc_id, scope_id, name_id,
  2093. lookup->getNamingClass(),
  2094. std::move(overload_set));
  2095. }
  2096. if (!lookup->isSingleResult()) {
  2097. // Clang will diagnose ambiguous lookup results for us.
  2098. if (!lookup->isAmbiguous()) {
  2099. context.TODO(loc_id,
  2100. llvm::formatv("Unsupported: Lookup succeeded but couldn't "
  2101. "find a single result; LookupResultKind: {0}",
  2102. static_cast<int>(lookup->getResultKind())));
  2103. }
  2104. context.name_scopes().AddRequiredName(scope_id, name_id,
  2105. SemIR::ErrorInst::InstId);
  2106. return SemIR::ScopeLookupResult::MakeError();
  2107. }
  2108. if (IsDeclInjectedClassName(context, scope_id, name_id,
  2109. lookup->getFoundDecl())) {
  2110. return ImportConstructorsIntoScope(context, loc_id, scope_id, name_id);
  2111. }
  2112. auto key = SemIR::ClangDeclKey::ForNonFunctionDecl(lookup->getFoundDecl());
  2113. return ImportNameDeclIntoScope(context, loc_id, scope_id, name_id, key,
  2114. MapCppAccess(lookup->begin().getPair()));
  2115. }
  2116. auto ImportClassDefinitionForClangDecl(Context& context, SemIR::LocId loc_id,
  2117. SemIR::ClassId class_id,
  2118. SemIR::ClangDeclId clang_decl_id)
  2119. -> bool {
  2120. clang::ASTUnit* ast = context.sem_ir().clang_ast_unit();
  2121. CARBON_CHECK(ast);
  2122. auto* clang_decl =
  2123. cast<clang::TagDecl>(context.clang_decls().Get(clang_decl_id).key.decl);
  2124. auto class_inst_id = context.types().GetAsTypeInstId(
  2125. context.classes().Get(class_id).first_owning_decl_id);
  2126. // TODO: Map loc_id into a clang location and use it for diagnostics if
  2127. // instantiation fails, instead of annotating the diagnostic with another
  2128. // location.
  2129. clang::SourceLocation loc = clang_decl->getLocation();
  2130. Diagnostics::AnnotationScope annotate_diagnostics(
  2131. &context.emitter(), [&](auto& builder) {
  2132. CARBON_DIAGNOSTIC(InCppTypeCompletion, Note,
  2133. "while completing C++ type {0}", SemIR::TypeId);
  2134. builder.Note(loc_id, InCppTypeCompletion,
  2135. context.classes().Get(class_id).self_type_id);
  2136. });
  2137. // Ask Clang whether the type is complete. This triggers template
  2138. // instantiation if necessary.
  2139. clang::DiagnosticErrorTrap trap(ast->getDiagnostics());
  2140. if (!ast->getSema().isCompleteType(
  2141. loc, context.ast_context().getCanonicalTagType(clang_decl))) {
  2142. // Type is incomplete. Nothing more to do, but tell the caller if we
  2143. // produced an error.
  2144. return !trap.hasErrorOccurred();
  2145. }
  2146. auto import_ir_inst_id =
  2147. context.insts().GetCanonicalLocId(class_inst_id).import_ir_inst_id();
  2148. if (auto* class_decl = dyn_cast<clang::CXXRecordDecl>(clang_decl)) {
  2149. auto* class_def = class_decl->getDefinition();
  2150. CARBON_CHECK(class_def, "Complete type has no definition");
  2151. if (class_def->getNumVBases()) {
  2152. // TODO: Handle virtual bases. We don't actually know where they go in the
  2153. // layout. We may also want to use a different size in the layout for
  2154. // `partial C`, excluding the virtual base. It's also not entirely safe to
  2155. // just skip over the virtual base, as the type we would construct would
  2156. // have a misleading size. For now, treat a C++ class with vbases as
  2157. // incomplete in Carbon.
  2158. context.TODO(loc_id, "class with virtual bases");
  2159. return false;
  2160. }
  2161. BuildClassDefinition(context, import_ir_inst_id, class_id, class_inst_id,
  2162. class_def);
  2163. } else if (auto* enum_decl = dyn_cast<clang::EnumDecl>(clang_decl)) {
  2164. BuildEnumDefinition(context, import_ir_inst_id, class_id, class_inst_id,
  2165. enum_decl);
  2166. }
  2167. return true;
  2168. }
  2169. } // namespace Carbon::Check