clang/unittests/Tooling/Syntax/TokensTest.cpp

   1 //===- TokensTest.cpp -----------------------------------------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #include "clang/Tooling/Syntax/Tokens.h"
  10 #include "clang/AST/ASTConsumer.h"
  11 #include "clang/AST/Expr.h"
  12 #include "clang/Basic/Diagnostic.h"
  13 #include "clang/Basic/DiagnosticIDs.h"
  14 #include "clang/Basic/DiagnosticOptions.h"
  15 #include "clang/Basic/FileManager.h"
  16 #include "clang/Basic/FileSystemOptions.h"
  17 #include "clang/Basic/LLVM.h"
  18 #include "clang/Basic/LangOptions.h"
  19 #include "clang/Basic/SourceLocation.h"
  20 #include "clang/Basic/SourceManager.h"
  21 #include "clang/Basic/TokenKinds.def"
  22 #include "clang/Basic/TokenKinds.h"
  23 #include "clang/Frontend/CompilerInstance.h"
  24 #include "clang/Frontend/FrontendAction.h"
  25 #include "clang/Frontend/Utils.h"
  26 #include "clang/Lex/Lexer.h"
  27 #include "clang/Lex/PreprocessorOptions.h"
  28 #include "clang/Lex/Token.h"
  29 #include "clang/Tooling/Tooling.h"
  30 #include "llvm/ADT/ArrayRef.h"
  31 #include "llvm/ADT/IntrusiveRefCntPtr.h"
  32 #include "llvm/ADT/None.h"
  33 #include "llvm/ADT/Optional.h"
  34 #include "llvm/ADT/STLExtras.h"
  35 #include "llvm/ADT/StringRef.h"
  36 #include "llvm/Support/FormatVariadic.h"
  37 #include "llvm/Support/MemoryBuffer.h"
  38 #include "llvm/Support/VirtualFileSystem.h"
  39 #include "llvm/Support/raw_os_ostream.h"
  40 #include "llvm/Support/raw_ostream.h"
  41 #include "llvm/Testing/Support/Annotations.h"
  42 #include "llvm/Testing/Support/SupportHelpers.h"
  43 #include "gmock/gmock.h"
  44 #include <cassert>
  45 #include <cstdlib>
  46 #include <gmock/gmock.h>
  47 #include <gtest/gtest.h>
  48 #include <memory>
  49 #include <ostream>
  50 #include <string>
  51
  52 using namespace clang;
  53 using namespace clang::syntax;
  54
  55 using llvm::ValueIs;
  56 using ::testing::_;
  57 using ::testing::AllOf;
  58 using ::testing::Contains;
  59 using ::testing::ElementsAre;
  60 using ::testing::Field;
  61 using ::testing::IsEmpty;
  62 using ::testing::Matcher;
  63 using ::testing::Not;
  64 using ::testing::Pointee;
  65 using ::testing::StartsWith;
  66
  67 namespace {
  68 // Checks the passed ArrayRef<T> has the same begin() and end() iterators as the
  69 // argument.
  70 MATCHER_P(SameRange, A, "") {
  71   return A.begin() == arg.begin() && A.end() == arg.end();
  72 }
  73
  74 Matcher<TokenBuffer::Expansion>
  75 IsExpansion(Matcher<llvm::ArrayRef<syntax::Token>> Spelled,
  76             Matcher<llvm::ArrayRef<syntax::Token>> Expanded) {
  77   return AllOf(Field(&TokenBuffer::Expansion::Spelled, Spelled),
  78                Field(&TokenBuffer::Expansion::Expanded, Expanded));
  79 }
  80 // Matchers for syntax::Token.
  81 MATCHER_P(Kind, K, "") { return arg.kind() == K; }
  82 MATCHER_P2(HasText, Text, SourceMgr, "") {
  83   return arg.text(*SourceMgr) == Text;
  84 }
  85 /// Checks the start and end location of a token are equal to SourceRng.
  86 MATCHER_P(RangeIs, SourceRng, "") {
  87   return arg.location() == SourceRng.first &&
  88          arg.endLocation() == SourceRng.second;
  89 }
  90
  91 class TokenCollectorTest : public ::testing::Test {
  92 public:
  93   /// Run the clang frontend, collect the preprocessed tokens from the frontend
  94   /// invocation and store them in this->Buffer.
  95   /// This also clears SourceManager before running the compiler.
  96   void recordTokens(llvm::StringRef Code) {
  97     class RecordTokens : public ASTFrontendAction {
  98     public:
  99       explicit RecordTokens(TokenBuffer &Result) : Result(Result) {}
 100
 101       bool BeginSourceFileAction(CompilerInstance &CI) override {
 102         assert(!Collector && "expected only a single call to BeginSourceFile");
 103         Collector.emplace(CI.getPreprocessor());
 104         return true;
 105       }
 106       void EndSourceFileAction() override {
 107         assert(Collector && "BeginSourceFileAction was never called");
 108         Result = std::move(*Collector).consume();
 109         Result.indexExpandedTokens();
 110       }
 111
 112       std::unique_ptr<ASTConsumer>
 113       CreateASTConsumer(CompilerInstance &CI, StringRef InFile) override {
 114         return std::make_unique<ASTConsumer>();
 115       }
 116
 117     private:
 118       TokenBuffer &Result;
 119       llvm::Optional<TokenCollector> Collector;
 120     };
 121
 122     constexpr const char *FileName = "./input.cpp";
 123     FS->addFile(FileName, time_t(), llvm::MemoryBuffer::getMemBufferCopy(""));
 124     // Prepare to run a compiler.
 125     if (!Diags->getClient())
 126       Diags->setClient(new IgnoringDiagConsumer);
 127     std::vector<const char *> Args = {"tok-test", "-std=c++03", "-fsyntax-only",
 128                                       FileName};
 129     auto CI = createInvocationFromCommandLine(Args, Diags, FS);
 130     assert(CI);
 131     CI->getFrontendOpts().DisableFree = false;
 132     CI->getPreprocessorOpts().addRemappedFile(
 133         FileName, llvm::MemoryBuffer::getMemBufferCopy(Code).release());
 134     CompilerInstance Compiler;
 135     Compiler.setInvocation(std::move(CI));
 136     Compiler.setDiagnostics(Diags.get());
 137     Compiler.setFileManager(FileMgr.get());
 138     Compiler.setSourceManager(SourceMgr.get());
 139
 140     this->Buffer = TokenBuffer(*SourceMgr);
 141     RecordTokens Recorder(this->Buffer);
 142     ASSERT_TRUE(Compiler.ExecuteAction(Recorder))
 143         << "failed to run the frontend";
 144   }
 145
 146   /// Record the tokens and return a test dump of the resulting buffer.
 147   std::string collectAndDump(llvm::StringRef Code) {
 148     recordTokens(Code);
 149     return Buffer.dumpForTests();
 150   }
 151
 152   // Adds a file to the test VFS.
 153   void addFile(llvm::StringRef Path, llvm::StringRef Contents) {
 154     if (!FS->addFile(Path, time_t(),
 155                      llvm::MemoryBuffer::getMemBufferCopy(Contents))) {
 156       ADD_FAILURE() << "could not add a file to VFS: " << Path;
 157     }
 158   }
 159
 160   /// Add a new file, run syntax::tokenize() on the range if any, run it on the
 161   /// whole file otherwise and return the results.
 162   std::vector<syntax::Token> tokenize(llvm::StringRef Text) {
 163     llvm::Annotations Annot(Text);
 164     auto FID = SourceMgr->createFileID(
 165         llvm::MemoryBuffer::getMemBufferCopy(Annot.code()));
 166     // FIXME: pass proper LangOptions.
 167     if (Annot.ranges().empty())
 168       return syntax::tokenize(FID, *SourceMgr, LangOptions());
 169     return syntax::tokenize(
 170         syntax::FileRange(FID, Annot.range().Begin, Annot.range().End),
 171         *SourceMgr, LangOptions());
 172   }
 173
 174   // Specialized versions of matchers that hide the SourceManager from clients.
 175   Matcher<syntax::Token> HasText(std::string Text) const {
 176     return ::HasText(Text, SourceMgr.get());
 177   }
 178   Matcher<syntax::Token> RangeIs(llvm::Annotations::Range R) const {
 179     std::pair<SourceLocation, SourceLocation> Ls;
 180     Ls.first = SourceMgr->getLocForStartOfFile(SourceMgr->getMainFileID())
 181                    .getLocWithOffset(R.Begin);
 182     Ls.second = SourceMgr->getLocForStartOfFile(SourceMgr->getMainFileID())
 183                     .getLocWithOffset(R.End);
 184     return ::RangeIs(Ls);
 185   }
 186
 187   /// Finds a subrange in O(n * m).
 188   template <class T, class U, class Eq>
 189   llvm::ArrayRef<T> findSubrange(llvm::ArrayRef<U> Subrange,
 190                                  llvm::ArrayRef<T> Range, Eq F) {
 191     assert(Subrange.size() >= 1);
 192     if (Range.size() < Subrange.size())
 193       return llvm::makeArrayRef(Range.end(), Range.end());
 194     for (auto Begin = Range.begin(), Last = Range.end() - Subrange.size();
 195          Begin <= Last; ++Begin) {
 196       auto It = Begin;
 197       for (auto ItSub = Subrange.begin(); ItSub != Subrange.end();
 198            ++ItSub, ++It) {
 199         if (!F(*ItSub, *It))
 200           goto continue_outer;
 201       }
 202       return llvm::makeArrayRef(Begin, It);
 203     continue_outer:;
 204     }
 205     return llvm::makeArrayRef(Range.end(), Range.end());
 206   }
 207
 208   /// Finds a subrange in \p Tokens that match the tokens specified in \p Query.
 209   /// The match should be unique. \p Query is a whitespace-separated list of
 210   /// tokens to search for.
 211   llvm::ArrayRef<syntax::Token>
 212   findTokenRange(llvm::StringRef Query, llvm::ArrayRef<syntax::Token> Tokens) {
 213     llvm::SmallVector<llvm::StringRef, 8> QueryTokens;
 214     Query.split(QueryTokens, ' ', /*MaxSplit=*/-1, /*KeepEmpty=*/false);
 215     if (QueryTokens.empty()) {
 216       ADD_FAILURE() << "will not look for an empty list of tokens";
 217       std::abort();
 218     }
 219     // An equality test for search.
 220     auto TextMatches = [this](llvm::StringRef Q, const syntax::Token &T) {
 221       return Q == T.text(*SourceMgr);
 222     };
 223     // Find a match.
 224     auto Found =
 225         findSubrange(llvm::makeArrayRef(QueryTokens), Tokens, TextMatches);
 226     if (Found.begin() == Tokens.end()) {
 227       ADD_FAILURE() << "could not find the subrange for " << Query;
 228       std::abort();
 229     }
 230     // Check that the match is unique.
 231     if (findSubrange(llvm::makeArrayRef(QueryTokens),
 232                      llvm::makeArrayRef(Found.end(), Tokens.end()), TextMatches)
 233             .begin() != Tokens.end()) {
 234       ADD_FAILURE() << "match is not unique for " << Query;
 235       std::abort();
 236     }
 237     return Found;
 238   };
 239
 240   // Specialized versions of findTokenRange for expanded and spelled tokens.
 241   llvm::ArrayRef<syntax::Token> findExpanded(llvm::StringRef Query) {
 242     return findTokenRange(Query, Buffer.expandedTokens());
 243   }
 244   llvm::ArrayRef<syntax::Token> findSpelled(llvm::StringRef Query,
 245                                             FileID File = FileID()) {
 246     if (!File.isValid())
 247       File = SourceMgr->getMainFileID();
 248     return findTokenRange(Query, Buffer.spelledTokens(File));
 249   }
 250
 251   // Data fields.
 252   llvm::IntrusiveRefCntPtr<DiagnosticsEngine> Diags =
 253       new DiagnosticsEngine(new DiagnosticIDs, new DiagnosticOptions);
 254   IntrusiveRefCntPtr<llvm::vfs::InMemoryFileSystem> FS =
 255       new llvm::vfs::InMemoryFileSystem;
 256   llvm::IntrusiveRefCntPtr<FileManager> FileMgr =
 257       new FileManager(FileSystemOptions(), FS);
 258   llvm::IntrusiveRefCntPtr<SourceManager> SourceMgr =
 259       new SourceManager(*Diags, *FileMgr);
 260   /// Contains last result of calling recordTokens().
 261   TokenBuffer Buffer = TokenBuffer(*SourceMgr);
 262 };
 263
 264 TEST_F(TokenCollectorTest, RawMode) {
 265   EXPECT_THAT(tokenize("int main() {}"),
 266               ElementsAre(Kind(tok::kw_int),
 267                           AllOf(HasText("main"), Kind(tok::identifier)),
 268                           Kind(tok::l_paren), Kind(tok::r_paren),
 269                           Kind(tok::l_brace), Kind(tok::r_brace)));
 270   // Comments are ignored for now.
 271   EXPECT_THAT(tokenize("/* foo */int a; // more comments"),
 272               ElementsAre(Kind(tok::kw_int),
 273                           AllOf(HasText("a"), Kind(tok::identifier)),
 274                           Kind(tok::semi)));
 275   EXPECT_THAT(tokenize("int [[main() {]]}"),
 276               ElementsAre(AllOf(HasText("main"), Kind(tok::identifier)),
 277                           Kind(tok::l_paren), Kind(tok::r_paren),
 278                           Kind(tok::l_brace)));
 279   EXPECT_THAT(tokenize("int [[main() {   ]]}"),
 280               ElementsAre(AllOf(HasText("main"), Kind(tok::identifier)),
 281                           Kind(tok::l_paren), Kind(tok::r_paren),
 282                           Kind(tok::l_brace)));
 283   // First token is partially parsed, last token is fully included even though
 284   // only a part of it is contained in the range.
 285   EXPECT_THAT(tokenize("int m[[ain() {ret]]urn 0;}"),
 286               ElementsAre(AllOf(HasText("ain"), Kind(tok::identifier)),
 287                           Kind(tok::l_paren), Kind(tok::r_paren),
 288                           Kind(tok::l_brace), Kind(tok::kw_return)));
 289 }
 290
 291 TEST_F(TokenCollectorTest, Basic) {
 292   std::pair</*Input*/ std::string, /*Expected*/ std::string> TestCases[] = {
 293       {"int main() {}",
 294        R"(expanded tokens:
 295   int main ( ) { }
 296 file './input.cpp'
 297   spelled tokens:
 298     int main ( ) { }
 299   no mappings.
 300 )"},
 301       // All kinds of whitespace are ignored.
 302       {"\t\n  int\t\n  main\t\n  (\t\n  )\t\n{\t\n  }\t\n",
 303        R"(expanded tokens:
 304   int main ( ) { }
 305 file './input.cpp'
 306   spelled tokens:
 307     int main ( ) { }
 308   no mappings.
 309 )"},
 310       // Annotation tokens are ignored.
 311       {R"cpp(
 312         #pragma GCC visibility push (public)
 313         #pragma GCC visibility pop
 314       )cpp",
 315        R"(expanded tokens:
 316   <empty>
 317 file './input.cpp'
 318   spelled tokens:
 319     # pragma GCC visibility push ( public ) # pragma GCC visibility pop
 320   mappings:
 321     ['#'_0, '<eof>'_13) => ['<eof>'_0, '<eof>'_0)
 322 )"},
 323       // Empty files should not crash.
 324       {R"cpp()cpp", R"(expanded tokens:
 325   <empty>
 326 file './input.cpp'
 327   spelled tokens:
 328     <empty>
 329   no mappings.
 330 )"},
 331       // Should not crash on errors inside '#define' directives. Error is that
 332       // stringification (#B) does not refer to a macro parameter.
 333       {
 334           R"cpp(
 335 a
 336 #define MACRO() A #B
 337 )cpp",
 338           R"(expanded tokens:
 339   a
 340 file './input.cpp'
 341   spelled tokens:
 342     a # define MACRO ( ) A # B
 343   mappings:
 344     ['#'_1, '<eof>'_9) => ['<eof>'_1, '<eof>'_1)
 345 )"}};
 346   for (auto &Test : TestCases)
 347     EXPECT_EQ(collectAndDump(Test.first), Test.second)
 348         << collectAndDump(Test.first);
 349 }
 350
 351 TEST_F(TokenCollectorTest, Locations) {
 352   // Check locations of the tokens.
 353   llvm::Annotations Code(R"cpp(
 354     $r1[[int]] $r2[[a]] $r3[[=]] $r4[["foo bar baz"]] $r5[[;]]
 355   )cpp");
 356   recordTokens(Code.code());
 357   // Check expanded tokens.
 358   EXPECT_THAT(
 359       Buffer.expandedTokens(),
 360       ElementsAre(AllOf(Kind(tok::kw_int), RangeIs(Code.range("r1"))),
 361                   AllOf(Kind(tok::identifier), RangeIs(Code.range("r2"))),
 362                   AllOf(Kind(tok::equal), RangeIs(Code.range("r3"))),
 363                   AllOf(Kind(tok::string_literal), RangeIs(Code.range("r4"))),
 364                   AllOf(Kind(tok::semi), RangeIs(Code.range("r5"))),
 365                   Kind(tok::eof)));
 366   // Check spelled tokens.
 367   EXPECT_THAT(
 368       Buffer.spelledTokens(SourceMgr->getMainFileID()),
 369       ElementsAre(AllOf(Kind(tok::kw_int), RangeIs(Code.range("r1"))),
 370                   AllOf(Kind(tok::identifier), RangeIs(Code.range("r2"))),
 371                   AllOf(Kind(tok::equal), RangeIs(Code.range("r3"))),
 372                   AllOf(Kind(tok::string_literal), RangeIs(Code.range("r4"))),
 373                   AllOf(Kind(tok::semi), RangeIs(Code.range("r5")))));
 374
 375   auto StartLoc = SourceMgr->getLocForStartOfFile(SourceMgr->getMainFileID());
 376   for (auto &R : Code.ranges()) {
 377     EXPECT_THAT(Buffer.spelledTokenAt(StartLoc.getLocWithOffset(R.Begin)),
 378                 Pointee(RangeIs(R)));
 379   }
 380 }
 381
 382 TEST_F(TokenCollectorTest, MacroDirectives) {
 383   // Macro directives are not stored anywhere at the moment.
 384   std::string Code = R"cpp(
 385     #define FOO a
 386     #include "unresolved_file.h"
 387     #undef FOO
 388     #ifdef X
 389     #else
 390     #endif
 391     #ifndef Y
 392     #endif
 393     #if 1
 394     #elif 2
 395     #else
 396     #endif
 397     #pragma once
 398     #pragma something lalala
 399
 400     int a;
 401   )cpp";
 402   std::string Expected =
 403       "expanded tokens:\n"
 404       "  int a ;\n"
 405       "file './input.cpp'\n"
 406       "  spelled tokens:\n"
 407       "    # define FOO a # include \"unresolved_file.h\" # undef FOO "
 408       "# ifdef X # else # endif # ifndef Y # endif # if 1 # elif 2 # else "
 409       "# endif # pragma once # pragma something lalala int a ;\n"
 410       "  mappings:\n"
 411       "    ['#'_0, 'int'_39) => ['int'_0, 'int'_0)\n";
 412   EXPECT_EQ(collectAndDump(Code), Expected);
 413 }
 414
 415 TEST_F(TokenCollectorTest, MacroReplacements) {
 416   std::pair</*Input*/ std::string, /*Expected*/ std::string> TestCases[] = {
 417       // A simple object-like macro.
 418       {R"cpp(
 419     #define INT int const
 420     INT a;
 421   )cpp",
 422        R"(expanded tokens:
 423   int const a ;
 424 file './input.cpp'
 425   spelled tokens:
 426     # define INT int const INT a ;
 427   mappings:
 428     ['#'_0, 'INT'_5) => ['int'_0, 'int'_0)
 429     ['INT'_5, 'a'_6) => ['int'_0, 'a'_2)
 430 )"},
 431       // A simple function-like macro.
 432       {R"cpp(
 433     #define INT(a) const int
 434     INT(10+10) a;
 435   )cpp",
 436        R"(expanded tokens:
 437   const int a ;
 438 file './input.cpp'
 439   spelled tokens:
 440     # define INT ( a ) const int INT ( 10 + 10 ) a ;
 441   mappings:
 442     ['#'_0, 'INT'_8) => ['const'_0, 'const'_0)
 443     ['INT'_8, 'a'_14) => ['const'_0, 'a'_2)
 444 )"},
 445       // Recursive macro replacements.
 446       {R"cpp(
 447     #define ID(X) X
 448     #define INT int const
 449     ID(ID(INT)) a;
 450   )cpp",
 451        R"(expanded tokens:
 452   int const a ;
 453 file './input.cpp'
 454   spelled tokens:
 455     # define ID ( X ) X # define INT int const ID ( ID ( INT ) ) a ;
 456   mappings:
 457     ['#'_0, 'ID'_12) => ['int'_0, 'int'_0)
 458     ['ID'_12, 'a'_19) => ['int'_0, 'a'_2)
 459 )"},
 460       // A little more complicated recursive macro replacements.
 461       {R"cpp(
 462     #define ADD(X, Y) X+Y
 463     #define MULT(X, Y) X*Y
 464
 465     int a = ADD(MULT(1,2), MULT(3,ADD(4,5)));
 466   )cpp",
 467        "expanded tokens:\n"
 468        "  int a = 1 * 2 + 3 * 4 + 5 ;\n"
 469        "file './input.cpp'\n"
 470        "  spelled tokens:\n"
 471        "    # define ADD ( X , Y ) X + Y # define MULT ( X , Y ) X * Y int "
 472        "a = ADD ( MULT ( 1 , 2 ) , MULT ( 3 , ADD ( 4 , 5 ) ) ) ;\n"
 473        "  mappings:\n"
 474        "    ['#'_0, 'int'_22) => ['int'_0, 'int'_0)\n"
 475        "    ['ADD'_25, ';'_46) => ['1'_3, ';'_12)\n"},
 476       // Empty macro replacement.
 477       // FIXME: the #define directives should not be glued together.
 478       {R"cpp(
 479     #define EMPTY
 480     #define EMPTY_FUNC(X)
 481     EMPTY
 482     EMPTY_FUNC(1+2+3)
 483     )cpp",
 484        R"(expanded tokens:
 485   <empty>
 486 file './input.cpp'
 487   spelled tokens:
 488     # define EMPTY # define EMPTY_FUNC ( X ) EMPTY EMPTY_FUNC ( 1 + 2 + 3 )
 489   mappings:
 490     ['#'_0, 'EMPTY'_9) => ['<eof>'_0, '<eof>'_0)
 491     ['EMPTY'_9, 'EMPTY_FUNC'_10) => ['<eof>'_0, '<eof>'_0)
 492     ['EMPTY_FUNC'_10, '<eof>'_18) => ['<eof>'_0, '<eof>'_0)
 493 )"},
 494       // File ends with a macro replacement.
 495       {R"cpp(
 496     #define FOO 10+10;
 497     int a = FOO
 498     )cpp",
 499        R"(expanded tokens:
 500   int a = 10 + 10 ;
 501 file './input.cpp'
 502   spelled tokens:
 503     # define FOO 10 + 10 ; int a = FOO
 504   mappings:
 505     ['#'_0, 'int'_7) => ['int'_0, 'int'_0)
 506     ['FOO'_10, '<eof>'_11) => ['10'_3, '<eof>'_7)
 507 )"},
 508       {R"cpp(
 509          #define NUM 42
 510          #define ID(a) a
 511          #define M 1 + ID
 512          M(NUM)
 513        )cpp",
 514        R"(expanded tokens:
 515   1 + 42
 516 file './input.cpp'
 517   spelled tokens:
 518     # define NUM 42 # define ID ( a ) a # define M 1 + ID M ( NUM )
 519   mappings:
 520     ['#'_0, 'M'_17) => ['1'_0, '1'_0)
 521     ['M'_17, '<eof>'_21) => ['1'_0, '<eof>'_3)
 522 )"},
 523   };
 524
 525   for (auto &Test : TestCases) {
 526     std::string Dump = collectAndDump(Test.first);
 527     EXPECT_EQ(Test.second, Dump) << Dump;
 528   }
 529 }
 530
 531 TEST_F(TokenCollectorTest, SpecialTokens) {
 532   // Tokens coming from concatenations.
 533   recordTokens(R"cpp(
 534     #define CONCAT(a, b) a ## b
 535     int a = CONCAT(1, 2);
 536   )cpp");
 537   EXPECT_THAT(std::vector<syntax::Token>(Buffer.expandedTokens()),
 538               Contains(HasText("12")));
 539   // Multi-line tokens with slashes at the end.
 540   recordTokens("i\\\nn\\\nt");
 541   EXPECT_THAT(Buffer.expandedTokens(),
 542               ElementsAre(AllOf(Kind(tok::kw_int), HasText("i\\\nn\\\nt")),
 543                           Kind(tok::eof)));
 544   // FIXME: test tokens with digraphs and UCN identifiers.
 545 }
 546
 547 TEST_F(TokenCollectorTest, LateBoundTokens) {
 548   // The parser eventually breaks the first '>>' into two tokens ('>' and '>'),
 549   // but we choose to record them as a single token (for now).
 550   llvm::Annotations Code(R"cpp(
 551     template <class T>
 552     struct foo { int a; };
 553     int bar = foo<foo<int$br[[>>]]().a;
 554     int baz = 10 $op[[>>]] 2;
 555   )cpp");
 556   recordTokens(Code.code());
 557   EXPECT_THAT(std::vector<syntax::Token>(Buffer.expandedTokens()),
 558               AllOf(Contains(AllOf(Kind(tok::greatergreater),
 559                                    RangeIs(Code.range("br")))),
 560                     Contains(AllOf(Kind(tok::greatergreater),
 561                                    RangeIs(Code.range("op"))))));
 562 }
 563
 564 TEST_F(TokenCollectorTest, DelayedParsing) {
 565   llvm::StringLiteral Code = R"cpp(
 566     struct Foo {
 567       int method() {
 568         // Parser will visit method bodies and initializers multiple times, but
 569         // TokenBuffer should only record the first walk over the tokens;
 570         return 100;
 571       }
 572       int a = 10;
 573
 574       struct Subclass {
 575         void foo() {
 576           Foo().method();
 577         }
 578       };
 579     };
 580   )cpp";
 581   std::string ExpectedTokens =
 582       "expanded tokens:\n"
 583       "  struct Foo { int method ( ) { return 100 ; } int a = 10 ; struct "
 584       "Subclass { void foo ( ) { Foo ( ) . method ( ) ; } } ; } ;\n";
 585   EXPECT_THAT(collectAndDump(Code), StartsWith(ExpectedTokens));
 586 }
 587
 588 TEST_F(TokenCollectorTest, MultiFile) {
 589   addFile("./foo.h", R"cpp(
 590     #define ADD(X, Y) X+Y
 591     int a = 100;
 592     #include "bar.h"
 593   )cpp");
 594   addFile("./bar.h", R"cpp(
 595     int b = ADD(1, 2);
 596     #define MULT(X, Y) X*Y
 597   )cpp");
 598   llvm::StringLiteral Code = R"cpp(
 599     #include "foo.h"
 600     int c = ADD(1, MULT(2,3));
 601   )cpp";
 602
 603   std::string Expected = R"(expanded tokens:
 604   int a = 100 ; int b = 1 + 2 ; int c = 1 + 2 * 3 ;
 605 file './input.cpp'
 606   spelled tokens:
 607     # include "foo.h" int c = ADD ( 1 , MULT ( 2 , 3 ) ) ;
 608   mappings:
 609     ['#'_0, 'int'_3) => ['int'_12, 'int'_12)
 610     ['ADD'_6, ';'_17) => ['1'_15, ';'_20)
 611 file './foo.h'
 612   spelled tokens:
 613     # define ADD ( X , Y ) X + Y int a = 100 ; # include "bar.h"
 614   mappings:
 615     ['#'_0, 'int'_11) => ['int'_0, 'int'_0)
 616     ['#'_16, '<eof>'_19) => ['int'_5, 'int'_5)
 617 file './bar.h'
 618   spelled tokens:
 619     int b = ADD ( 1 , 2 ) ; # define MULT ( X , Y ) X * Y
 620   mappings:
 621     ['ADD'_3, ';'_9) => ['1'_8, ';'_11)
 622     ['#'_10, '<eof>'_21) => ['int'_12, 'int'_12)
 623 )";
 624
 625   EXPECT_EQ(Expected, collectAndDump(Code))
 626       << "input: " << Code << "\nresults: " << collectAndDump(Code);
 627 }
 628
 629 class TokenBufferTest : public TokenCollectorTest {};
 630
 631 TEST_F(TokenBufferTest, SpelledByExpanded) {
 632   recordTokens(R"cpp(
 633     a1 a2 a3 b1 b2
 634   )cpp");
 635
 636   // Sanity check: expanded and spelled tokens are stored separately.
 637   EXPECT_THAT(findExpanded("a1 a2"), Not(SameRange(findSpelled("a1 a2"))));
 638   // Searching for subranges of expanded tokens should give the corresponding
 639   // spelled ones.
 640   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2 a3 b1 b2")),
 641               ValueIs(SameRange(findSpelled("a1 a2 a3 b1 b2"))));
 642   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2 a3")),
 643               ValueIs(SameRange(findSpelled("a1 a2 a3"))));
 644   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("b1 b2")),
 645               ValueIs(SameRange(findSpelled("b1 b2"))));
 646
 647   // Test search on simple macro expansions.
 648   recordTokens(R"cpp(
 649     #define A a1 a2 a3
 650     #define B b1 b2
 651
 652     A split B
 653   )cpp");
 654   // Ranges going across expansion boundaries.
 655   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2 a3 split b1 b2")),
 656               ValueIs(SameRange(findSpelled("A split B"))));
 657   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2 a3")),
 658               ValueIs(SameRange(findSpelled("A split").drop_back())));
 659   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("b1 b2")),
 660               ValueIs(SameRange(findSpelled("split B").drop_front())));
 661   // Ranges not fully covering macro invocations should fail.
 662   EXPECT_EQ(Buffer.spelledForExpanded(findExpanded("a1 a2")), llvm::None);
 663   EXPECT_EQ(Buffer.spelledForExpanded(findExpanded("b2")), llvm::None);
 664   EXPECT_EQ(Buffer.spelledForExpanded(findExpanded("a2 a3 split b1 b2")),
 665             llvm::None);
 666
 667   // Recursive macro invocations.
 668   recordTokens(R"cpp(
 669     #define ID(x) x
 670     #define B b1 b2
 671
 672     ID(ID(ID(a1) a2 a3)) split ID(B)
 673   )cpp");
 674
 675   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("b1 b2")),
 676               ValueIs(SameRange(findSpelled("( B").drop_front())));
 677   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2 a3 split b1 b2")),
 678               ValueIs(SameRange(findSpelled(
 679                   "ID ( ID ( ID ( a1 ) a2 a3 ) ) split ID ( B )"))));
 680   // Mixed ranges with expanded and spelled tokens.
 681   EXPECT_THAT(
 682       Buffer.spelledForExpanded(findExpanded("a1 a2 a3 split")),
 683       ValueIs(SameRange(findSpelled("ID ( ID ( ID ( a1 ) a2 a3 ) ) split"))));
 684   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("split b1 b2")),
 685               ValueIs(SameRange(findSpelled("split ID ( B )"))));
 686   // Macro arguments
 687   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1")),
 688               ValueIs(SameRange(findSpelled("a1"))));
 689   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a2")),
 690               ValueIs(SameRange(findSpelled("a2"))));
 691   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a3")),
 692               ValueIs(SameRange(findSpelled("a3"))));
 693   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2")),
 694               ValueIs(SameRange(findSpelled("ID ( a1 ) a2"))));
 695   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a1 a2 a3")),
 696               ValueIs(SameRange(findSpelled("ID ( a1 ) a2 a3"))));
 697
 698   // Empty macro expansions.
 699   recordTokens(R"cpp(
 700     #define EMPTY
 701     #define ID(X) X
 702
 703     EMPTY EMPTY ID(1 2 3) EMPTY EMPTY split1
 704     EMPTY EMPTY ID(4 5 6) split2
 705     ID(7 8 9) EMPTY EMPTY
 706   )cpp");
 707   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("1 2 3")),
 708               ValueIs(SameRange(findSpelled("1 2 3"))));
 709   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("4 5 6")),
 710               ValueIs(SameRange(findSpelled("4 5 6"))));
 711   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("7 8 9")),
 712               ValueIs(SameRange(findSpelled("7 8 9"))));
 713
 714   // Empty mappings coming from various directives.
 715   recordTokens(R"cpp(
 716     #define ID(X) X
 717     ID(1)
 718     #pragma lalala
 719     not_mapped
 720   )cpp");
 721   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("not_mapped")),
 722               ValueIs(SameRange(findSpelled("not_mapped"))));
 723
 724   // Multiple macro arguments
 725   recordTokens(R"cpp(
 726     #define ID(X) X
 727     #define ID2(X, Y) X Y
 728
 729     ID2(ID(a1), ID(a2) a3) ID2(a4, a5 a6 a7)
 730   )cpp");
 731   // Should fail, spans multiple arguments.
 732   EXPECT_EQ(Buffer.spelledForExpanded(findExpanded("a1 a2")), llvm::None);
 733   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a2 a3")),
 734               ValueIs(SameRange(findSpelled("ID ( a2 ) a3"))));
 735   EXPECT_THAT(
 736       Buffer.spelledForExpanded(findExpanded("a1 a2 a3")),
 737       ValueIs(SameRange(findSpelled("ID2 ( ID ( a1 ) , ID ( a2 ) a3 )"))));
 738   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a5 a6")),
 739               ValueIs(SameRange(findSpelled("a5 a6"))));
 740   EXPECT_THAT(Buffer.spelledForExpanded(findExpanded("a4 a5 a6 a7")),
 741               ValueIs(SameRange(findSpelled("ID2 ( a4 , a5 a6 a7 )"))));
 742   // Should fail, spans multiple invocations.
 743   EXPECT_EQ(Buffer.spelledForExpanded(findExpanded("a1 a2 a3 a4")), llvm::None);
 744 }
 745
 746 TEST_F(TokenBufferTest, ExpandedTokensForRange) {
 747   recordTokens(R"cpp(
 748     #define SIGN(X) X##_washere
 749     A SIGN(B) C SIGN(D) E SIGN(F) G
 750   )cpp");
 751
 752   SourceRange R(findExpanded("C").front().location(),
 753                 findExpanded("F_washere").front().location());
 754   // Sanity check: expanded and spelled tokens are stored separately.
 755   EXPECT_THAT(Buffer.expandedTokens(R),
 756               SameRange(findExpanded("C D_washere E F_washere")));
 757   EXPECT_THAT(Buffer.expandedTokens(SourceRange()), testing::IsEmpty());
 758 }
 759
 760 TEST_F(TokenBufferTest, ExpansionsOverlapping) {
 761   // Object-like macro expansions.
 762   recordTokens(R"cpp(
 763     #define FOO 3+4
 764     int a = FOO 1;
 765     int b = FOO 2;
 766   )cpp");
 767
 768   llvm::ArrayRef<syntax::Token> Foo1 = findSpelled("FOO 1");
 769   EXPECT_THAT(
 770       Buffer.expansionStartingAt(Foo1.data()),
 771       ValueIs(IsExpansion(SameRange(Foo1.drop_back()),
 772                           SameRange(findExpanded("3 + 4 1").drop_back()))));
 773   EXPECT_THAT(
 774       Buffer.expansionsOverlapping(Foo1),
 775       ElementsAre(IsExpansion(SameRange(Foo1.drop_back()),
 776                               SameRange(findExpanded("3 + 4 1").drop_back()))));
 777
 778   llvm::ArrayRef<syntax::Token> Foo2 = findSpelled("FOO 2");
 779   EXPECT_THAT(
 780       Buffer.expansionStartingAt(Foo2.data()),
 781       ValueIs(IsExpansion(SameRange(Foo2.drop_back()),
 782                           SameRange(findExpanded("3 + 4 2").drop_back()))));
 783   EXPECT_THAT(Buffer.expansionsOverlapping(
 784                   llvm::makeArrayRef(Foo1.begin(), Foo2.end())),
 785               ElementsAre(IsExpansion(SameRange(Foo1.drop_back()), _),
 786                           IsExpansion(SameRange(Foo2.drop_back()), _)));
 787
 788   // Function-like macro expansions.
 789   recordTokens(R"cpp(
 790     #define ID(X) X
 791     int a = ID(1+2+3);
 792     int b = ID(ID(2+3+4));
 793   )cpp");
 794
 795   llvm::ArrayRef<syntax::Token> ID1 = findSpelled("ID ( 1 + 2 + 3 )");
 796   EXPECT_THAT(Buffer.expansionStartingAt(&ID1.front()),
 797               ValueIs(IsExpansion(SameRange(ID1),
 798                                   SameRange(findExpanded("1 + 2 + 3")))));
 799   // Only the first spelled token should be found.
 800   for (const auto &T : ID1.drop_front())
 801     EXPECT_EQ(Buffer.expansionStartingAt(&T), llvm::None);
 802
 803   llvm::ArrayRef<syntax::Token> ID2 = findSpelled("ID ( ID ( 2 + 3 + 4 ) )");
 804   EXPECT_THAT(Buffer.expansionStartingAt(&ID2.front()),
 805               ValueIs(IsExpansion(SameRange(ID2),
 806                                   SameRange(findExpanded("2 + 3 + 4")))));
 807   // Only the first spelled token should be found.
 808   for (const auto &T : ID2.drop_front())
 809     EXPECT_EQ(Buffer.expansionStartingAt(&T), llvm::None);
 810
 811   EXPECT_THAT(Buffer.expansionsOverlapping(llvm::makeArrayRef(
 812                   findSpelled("1 + 2").data(), findSpelled("4").data())),
 813               ElementsAre(IsExpansion(SameRange(ID1), _),
 814                           IsExpansion(SameRange(ID2), _)));
 815
 816   // PP directives.
 817   recordTokens(R"cpp(
 818 #define FOO 1
 819 int a = FOO;
 820 #pragma once
 821 int b = 1;
 822   )cpp");
 823
 824   llvm::ArrayRef<syntax::Token> DefineFoo = findSpelled("# define FOO 1");
 825   EXPECT_THAT(
 826       Buffer.expansionStartingAt(&DefineFoo.front()),
 827       ValueIs(IsExpansion(SameRange(DefineFoo),
 828                           SameRange(findExpanded("int a").take_front(0)))));
 829   // Only the first spelled token should be found.
 830   for (const auto &T : DefineFoo.drop_front())
 831     EXPECT_EQ(Buffer.expansionStartingAt(&T), llvm::None);
 832
 833   llvm::ArrayRef<syntax::Token> PragmaOnce = findSpelled("# pragma once");
 834   EXPECT_THAT(
 835       Buffer.expansionStartingAt(&PragmaOnce.front()),
 836       ValueIs(IsExpansion(SameRange(PragmaOnce),
 837                           SameRange(findExpanded("int b").take_front(0)))));
 838   // Only the first spelled token should be found.
 839   for (const auto &T : PragmaOnce.drop_front())
 840     EXPECT_EQ(Buffer.expansionStartingAt(&T), llvm::None);
 841
 842   EXPECT_THAT(
 843       Buffer.expansionsOverlapping(findSpelled("FOO ; # pragma")),
 844       ElementsAre(IsExpansion(SameRange(findSpelled("FOO ;").drop_back()), _),
 845                   IsExpansion(SameRange(PragmaOnce), _)));
 846 }
 847
 848 TEST_F(TokenBufferTest, TokensToFileRange) {
 849   addFile("./foo.h", "token_from_header");
 850   llvm::Annotations Code(R"cpp(
 851     #define FOO token_from_expansion
 852     #include "./foo.h"
 853     $all[[$i[[int]] a = FOO;]]
 854   )cpp");
 855   recordTokens(Code.code());
 856
 857   auto &SM = *SourceMgr;
 858
 859   // Two simple examples.
 860   auto Int = findExpanded("int").front();
 861   auto Semi = findExpanded(";").front();
 862   EXPECT_EQ(Int.range(SM), FileRange(SM.getMainFileID(), Code.range("i").Begin,
 863                                      Code.range("i").End));
 864   EXPECT_EQ(syntax::Token::range(SM, Int, Semi),
 865             FileRange(SM.getMainFileID(), Code.range("all").Begin,
 866                       Code.range("all").End));
 867   // We don't test assertion failures because death tests are slow.
 868 }
 869
 870 TEST_F(TokenBufferTest, MacroExpansions) {
 871   llvm::Annotations Code(R"cpp(
 872     #define FOO B
 873     #define FOO2 BA
 874     #define CALL(X) int X
 875     #define G CALL(FOO2)
 876     int B;
 877     $macro[[FOO]];
 878     $macro[[CALL]](A);
 879     $macro[[G]];
 880   )cpp");
 881   recordTokens(Code.code());
 882   auto &SM = *SourceMgr;
 883   auto Expansions = Buffer.macroExpansions(SM.getMainFileID());
 884   std::vector<FileRange> ExpectedMacroRanges;
 885   for (auto Range : Code.ranges("macro"))
 886     ExpectedMacroRanges.push_back(
 887         FileRange(SM.getMainFileID(), Range.Begin, Range.End));
 888   std::vector<FileRange> ActualMacroRanges;
 889   for (auto Expansion : Expansions)
 890     ActualMacroRanges.push_back(Expansion->range(SM));
 891   EXPECT_EQ(ExpectedMacroRanges, ActualMacroRanges);
 892 }
 893
 894 TEST_F(TokenBufferTest, Touching) {
 895   llvm::Annotations Code("^i^nt^ ^a^b^=^1;^");
 896   recordTokens(Code.code());
 897
 898   auto Touching = [&](int Index) {
 899     SourceLocation Loc = SourceMgr->getComposedLoc(SourceMgr->getMainFileID(),
 900                                                    Code.points()[Index]);
 901     return spelledTokensTouching(Loc, Buffer);
 902   };
 903   auto Identifier = [&](int Index) {
 904     SourceLocation Loc = SourceMgr->getComposedLoc(SourceMgr->getMainFileID(),
 905                                                    Code.points()[Index]);
 906     const syntax::Token *Tok = spelledIdentifierTouching(Loc, Buffer);
 907     return Tok ? Tok->text(*SourceMgr) : "";
 908   };
 909
 910   EXPECT_THAT(Touching(0), SameRange(findSpelled("int")));
 911   EXPECT_EQ(Identifier(0), "");
 912   EXPECT_THAT(Touching(1), SameRange(findSpelled("int")));
 913   EXPECT_EQ(Identifier(1), "");
 914   EXPECT_THAT(Touching(2), SameRange(findSpelled("int")));
 915   EXPECT_EQ(Identifier(2), "");
 916
 917   EXPECT_THAT(Touching(3), SameRange(findSpelled("ab")));
 918   EXPECT_EQ(Identifier(3), "ab");
 919   EXPECT_THAT(Touching(4), SameRange(findSpelled("ab")));
 920   EXPECT_EQ(Identifier(4), "ab");
 921
 922   EXPECT_THAT(Touching(5), SameRange(findSpelled("ab =")));
 923   EXPECT_EQ(Identifier(5), "ab");
 924
 925   EXPECT_THAT(Touching(6), SameRange(findSpelled("= 1")));
 926   EXPECT_EQ(Identifier(6), "");
 927
 928   EXPECT_THAT(Touching(7), SameRange(findSpelled(";")));
 929   EXPECT_EQ(Identifier(7), "");
 930
 931   ASSERT_EQ(Code.points().size(), 8u);
 932 }
 933
 934 TEST_F(TokenBufferTest, ExpandedBySpelled) {
 935   recordTokens(R"cpp(
 936     a1 a2 a3 b1 b2
 937   )cpp");
 938   // Sanity check: expanded and spelled tokens are stored separately.
 939   EXPECT_THAT(findExpanded("a1 a2"), Not(SameRange(findSpelled("a1 a2"))));
 940   // Searching for subranges of expanded tokens should give the corresponding
 941   // spelled ones.
 942   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("a1 a2 a3 b1 b2")),
 943               ElementsAre(SameRange(findExpanded("a1 a2 a3 b1 b2"))));
 944   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("a1 a2 a3")),
 945               ElementsAre(SameRange(findExpanded("a1 a2 a3"))));
 946   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("b1 b2")),
 947               ElementsAre(SameRange(findExpanded("b1 b2"))));
 948
 949   // Test search on simple macro expansions.
 950   recordTokens(R"cpp(
 951     #define A a1 a2 a3
 952     #define B b1 b2
 953
 954     A split B
 955   )cpp");
 956   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("A split B")),
 957               ElementsAre(SameRange(findExpanded("a1 a2 a3 split b1 b2"))));
 958   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("A split").drop_back()),
 959               ElementsAre(SameRange(findExpanded("a1 a2 a3"))));
 960   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("split B").drop_front()),
 961               ElementsAre(SameRange(findExpanded("b1 b2"))));
 962
 963   // Ranges not fully covering macro expansions should fail.
 964   recordTokens(R"cpp(
 965     #define ID(x) x
 966
 967     ID(a)
 968   )cpp");
 969   // Spelled don't cover entire mapping (missing ID token) -> empty result
 970   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("( a )")), IsEmpty());
 971   // Spelled don't cover entire mapping (missing ) token) -> empty result
 972   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( a")), IsEmpty());
 973
 974   // Recursive macro invocations.
 975   recordTokens(R"cpp(
 976     #define ID(x) x
 977     #define B b1 b2
 978
 979     ID(ID(ID(a1) a2 a3)) split ID(B)
 980   )cpp");
 981
 982   EXPECT_THAT(
 983       Buffer.expandedForSpelled(findSpelled("ID ( ID ( ID ( a1 ) a2 a3 ) )")),
 984       ElementsAre(SameRange(findExpanded("a1 a2 a3"))));
 985   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( B )")),
 986               ElementsAre(SameRange(findExpanded("b1 b2"))));
 987   EXPECT_THAT(Buffer.expandedForSpelled(
 988                   findSpelled("ID ( ID ( ID ( a1 ) a2 a3 ) ) split ID ( B )")),
 989               ElementsAre(SameRange(findExpanded("a1 a2 a3 split b1 b2"))));
 990   // FIXME: these should succeed, but we do not support macro arguments yet.
 991   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("a1")), IsEmpty());
 992   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( a1 ) a2")),
 993               IsEmpty());
 994
 995   // Empty macro expansions.
 996   recordTokens(R"cpp(
 997     #define EMPTY
 998     #define ID(X) X
 999
1000     EMPTY EMPTY ID(1 2 3) EMPTY EMPTY split1
1001     EMPTY EMPTY ID(4 5 6) split2
1002     ID(7 8 9) EMPTY EMPTY
1003   )cpp");
1004   // Covered by empty expansions on one of both of the sides.
1005   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( 1 2 3 )")),
1006               ElementsAre(SameRange(findExpanded("1 2 3"))));
1007   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( 4 5 6 )")),
1008               ElementsAre(SameRange(findExpanded("4 5 6"))));
1009   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( 7 8 9 )")),
1010               ElementsAre(SameRange(findExpanded("7 8 9"))));
1011   // Including the empty macro expansions on the side.
1012   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("EMPTY ID ( 1 2 3 )")),
1013               ElementsAre(SameRange(findExpanded("1 2 3"))));
1014   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("ID ( 1 2 3 ) EMPTY")),
1015               ElementsAre(SameRange(findExpanded("1 2 3"))));
1016   EXPECT_THAT(
1017       Buffer.expandedForSpelled(findSpelled("EMPTY ID ( 1 2 3 ) EMPTY")),
1018       ElementsAre(SameRange(findExpanded("1 2 3"))));
1019
1020   // Empty mappings coming from various directives.
1021   recordTokens(R"cpp(
1022     #define ID(X) X
1023     ID(1)
1024     #pragma lalala
1025     not_mapped
1026   )cpp");
1027   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("# define ID ( X ) X")),
1028               IsEmpty());
1029   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("# pragma lalala")),
1030               IsEmpty());
1031
1032   // Empty macro expansion.
1033   recordTokens(R"cpp(
1034     #define EMPTY
1035     EMPTY int a = 100;
1036   )cpp");
1037   EXPECT_THAT(Buffer.expandedForSpelled(findSpelled("EMPTY int").drop_back()),
1038               IsEmpty());
1039 }
1040
1041 TEST_F(TokenCollectorTest, Pragmas) {
1042   // Tokens coming from concatenations.
1043   recordTokens(R"cpp(
1044     void foo() {
1045       #pragma unroll 4
1046       for(int i=0;i<4;++i);
1047     }
1048   )cpp");
1049 }
1050 } // namespace