Home | History | Annotate | Download | only in Lex
      1 //===- unittests/Lex/LexerTest.cpp ------ Lexer tests ---------------------===//
      2 //
      3 //                     The LLVM Compiler Infrastructure
      4 //
      5 // This file is distributed under the University of Illinois Open Source
      6 // License. See LICENSE.TXT for details.
      7 //
      8 //===----------------------------------------------------------------------===//
      9 
     10 #include "clang/Lex/Lexer.h"
     11 #include "clang/Basic/Diagnostic.h"
     12 #include "clang/Basic/DiagnosticOptions.h"
     13 #include "clang/Basic/FileManager.h"
     14 #include "clang/Basic/LangOptions.h"
     15 #include "clang/Basic/SourceManager.h"
     16 #include "clang/Basic/TargetInfo.h"
     17 #include "clang/Basic/TargetOptions.h"
     18 #include "clang/Lex/HeaderSearch.h"
     19 #include "clang/Lex/HeaderSearchOptions.h"
     20 #include "clang/Lex/ModuleLoader.h"
     21 #include "clang/Lex/Preprocessor.h"
     22 #include "clang/Lex/PreprocessorOptions.h"
     23 #include "llvm/Config/config.h"
     24 #include "gtest/gtest.h"
     25 
     26 using namespace llvm;
     27 using namespace clang;
     28 
     29 namespace {
     30 
     31 class VoidModuleLoader : public ModuleLoader {
     32   virtual ModuleLoadResult loadModule(SourceLocation ImportLoc,
     33                                       ModuleIdPath Path,
     34                                       Module::NameVisibilityKind Visibility,
     35                                       bool IsInclusionDirective) {
     36     return ModuleLoadResult();
     37   }
     38 
     39   virtual void makeModuleVisible(Module *Mod,
     40                                  Module::NameVisibilityKind Visibility,
     41                                  SourceLocation ImportLoc,
     42                                  bool Complain) { }
     43 };
     44 
     45 // The test fixture.
     46 class LexerTest : public ::testing::Test {
     47 protected:
     48   LexerTest()
     49     : FileMgr(FileMgrOpts),
     50       DiagID(new DiagnosticIDs()),
     51       Diags(DiagID, new DiagnosticOptions, new IgnoringDiagConsumer()),
     52       SourceMgr(Diags, FileMgr),
     53       TargetOpts(new TargetOptions)
     54   {
     55     TargetOpts->Triple = "x86_64-apple-darwin11.1.0";
     56     Target = TargetInfo::CreateTargetInfo(Diags, &*TargetOpts);
     57   }
     58 
     59   std::vector<Token> CheckLex(StringRef Source,
     60                               ArrayRef<tok::TokenKind> ExpectedTokens) {
     61     MemoryBuffer *buf = MemoryBuffer::getMemBuffer(Source);
     62     (void) SourceMgr.createMainFileIDForMemBuffer(buf);
     63 
     64     VoidModuleLoader ModLoader;
     65     HeaderSearch HeaderInfo(new HeaderSearchOptions, FileMgr, Diags, LangOpts,
     66                             Target.getPtr());
     67     Preprocessor PP(new PreprocessorOptions(), Diags, LangOpts, Target.getPtr(),
     68                     SourceMgr, HeaderInfo, ModLoader, /*IILookup =*/ 0,
     69                     /*OwnsHeaderSearch =*/ false,
     70                     /*DelayInitialization =*/ false);
     71     PP.EnterMainSourceFile();
     72 
     73     std::vector<Token> toks;
     74     while (1) {
     75       Token tok;
     76       PP.Lex(tok);
     77       if (tok.is(tok::eof))
     78         break;
     79       toks.push_back(tok);
     80     }
     81 
     82     EXPECT_EQ(ExpectedTokens.size(), toks.size());
     83     for (unsigned i = 0, e = ExpectedTokens.size(); i != e; ++i) {
     84       EXPECT_EQ(ExpectedTokens[i], toks[i].getKind());
     85     }
     86 
     87     return toks;
     88   }
     89 
     90   std::string getSourceText(Token Begin, Token End) {
     91     bool Invalid;
     92     StringRef Str =
     93         Lexer::getSourceText(CharSourceRange::getTokenRange(SourceRange(
     94                                     Begin.getLocation(), End.getLocation())),
     95                              SourceMgr, LangOpts, &Invalid);
     96     if (Invalid)
     97       return "<INVALID>";
     98     return Str;
     99   }
    100 
    101   FileSystemOptions FileMgrOpts;
    102   FileManager FileMgr;
    103   IntrusiveRefCntPtr<DiagnosticIDs> DiagID;
    104   DiagnosticsEngine Diags;
    105   SourceManager SourceMgr;
    106   LangOptions LangOpts;
    107   IntrusiveRefCntPtr<TargetOptions> TargetOpts;
    108   IntrusiveRefCntPtr<TargetInfo> Target;
    109 };
    110 
    111 TEST_F(LexerTest, GetSourceTextExpandsToMaximumInMacroArgument) {
    112   std::vector<tok::TokenKind> ExpectedTokens;
    113   ExpectedTokens.push_back(tok::identifier);
    114   ExpectedTokens.push_back(tok::l_paren);
    115   ExpectedTokens.push_back(tok::identifier);
    116   ExpectedTokens.push_back(tok::r_paren);
    117 
    118   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    119                                      "M(f(M(i)))",
    120                                      ExpectedTokens);
    121 
    122   EXPECT_EQ("M(i)", getSourceText(toks[2], toks[2]));
    123 }
    124 
    125 TEST_F(LexerTest, GetSourceTextExpandsToMaximumInMacroArgumentForEndOfMacro) {
    126   std::vector<tok::TokenKind> ExpectedTokens;
    127   ExpectedTokens.push_back(tok::identifier);
    128   ExpectedTokens.push_back(tok::identifier);
    129 
    130   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    131                                      "M(M(i) c)",
    132                                      ExpectedTokens);
    133 
    134   EXPECT_EQ("M(i)", getSourceText(toks[0], toks[0]));
    135 }
    136 
    137 TEST_F(LexerTest, GetSourceTextExpandsInMacroArgumentForBeginOfMacro) {
    138   std::vector<tok::TokenKind> ExpectedTokens;
    139   ExpectedTokens.push_back(tok::identifier);
    140   ExpectedTokens.push_back(tok::identifier);
    141   ExpectedTokens.push_back(tok::identifier);
    142 
    143   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    144                                      "M(c c M(i))",
    145                                      ExpectedTokens);
    146 
    147   EXPECT_EQ("c M(i)", getSourceText(toks[1], toks[2]));
    148 }
    149 
    150 TEST_F(LexerTest, GetSourceTextExpandsInMacroArgumentForEndOfMacro) {
    151   std::vector<tok::TokenKind> ExpectedTokens;
    152   ExpectedTokens.push_back(tok::identifier);
    153   ExpectedTokens.push_back(tok::identifier);
    154   ExpectedTokens.push_back(tok::identifier);
    155 
    156   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    157                                      "M(M(i) c c)",
    158                                      ExpectedTokens);
    159 
    160   EXPECT_EQ("M(i) c", getSourceText(toks[0], toks[1]));
    161 }
    162 
    163 TEST_F(LexerTest, GetSourceTextInSeparateFnMacros) {
    164   std::vector<tok::TokenKind> ExpectedTokens;
    165   ExpectedTokens.push_back(tok::identifier);
    166   ExpectedTokens.push_back(tok::identifier);
    167   ExpectedTokens.push_back(tok::identifier);
    168   ExpectedTokens.push_back(tok::identifier);
    169 
    170   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    171                                      "M(c M(i)) M(M(i) c)",
    172                                      ExpectedTokens);
    173 
    174   EXPECT_EQ("<INVALID>", getSourceText(toks[1], toks[2]));
    175 }
    176 
    177 TEST_F(LexerTest, GetSourceTextWorksAcrossTokenPastes) {
    178   std::vector<tok::TokenKind> ExpectedTokens;
    179   ExpectedTokens.push_back(tok::identifier);
    180   ExpectedTokens.push_back(tok::l_paren);
    181   ExpectedTokens.push_back(tok::identifier);
    182   ExpectedTokens.push_back(tok::r_paren);
    183 
    184   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    185                                      "#define C(x) M(x##c)\n"
    186                                      "M(f(C(i)))",
    187                                      ExpectedTokens);
    188 
    189   EXPECT_EQ("C(i)", getSourceText(toks[2], toks[2]));
    190 }
    191 
    192 TEST_F(LexerTest, GetSourceTextExpandsAcrossMultipleMacroCalls) {
    193   std::vector<tok::TokenKind> ExpectedTokens;
    194   ExpectedTokens.push_back(tok::identifier);
    195   ExpectedTokens.push_back(tok::l_paren);
    196   ExpectedTokens.push_back(tok::identifier);
    197   ExpectedTokens.push_back(tok::r_paren);
    198 
    199   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    200                                      "f(M(M(i)))",
    201                                      ExpectedTokens);
    202   EXPECT_EQ("M(M(i))", getSourceText(toks[2], toks[2]));
    203 }
    204 
    205 TEST_F(LexerTest, GetSourceTextInMiddleOfMacroArgument) {
    206   std::vector<tok::TokenKind> ExpectedTokens;
    207   ExpectedTokens.push_back(tok::identifier);
    208   ExpectedTokens.push_back(tok::l_paren);
    209   ExpectedTokens.push_back(tok::identifier);
    210   ExpectedTokens.push_back(tok::r_paren);
    211 
    212   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    213                                      "M(f(i))",
    214                                      ExpectedTokens);
    215   EXPECT_EQ("i", getSourceText(toks[2], toks[2]));
    216 }
    217 
    218 TEST_F(LexerTest, GetSourceTextExpandsAroundDifferentMacroCalls) {
    219   std::vector<tok::TokenKind> ExpectedTokens;
    220   ExpectedTokens.push_back(tok::identifier);
    221   ExpectedTokens.push_back(tok::l_paren);
    222   ExpectedTokens.push_back(tok::identifier);
    223   ExpectedTokens.push_back(tok::r_paren);
    224 
    225   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    226                                      "#define C(x) x\n"
    227                                      "f(C(M(i)))",
    228                                      ExpectedTokens);
    229   EXPECT_EQ("C(M(i))", getSourceText(toks[2], toks[2]));
    230 }
    231 
    232 TEST_F(LexerTest, GetSourceTextOnlyExpandsIfFirstTokenInMacro) {
    233   std::vector<tok::TokenKind> ExpectedTokens;
    234   ExpectedTokens.push_back(tok::identifier);
    235   ExpectedTokens.push_back(tok::l_paren);
    236   ExpectedTokens.push_back(tok::identifier);
    237   ExpectedTokens.push_back(tok::identifier);
    238   ExpectedTokens.push_back(tok::r_paren);
    239 
    240   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    241                                      "#define C(x) c x\n"
    242                                      "f(C(M(i)))",
    243                                      ExpectedTokens);
    244   EXPECT_EQ("M(i)", getSourceText(toks[3], toks[3]));
    245 }
    246 
    247 TEST_F(LexerTest, GetSourceTextExpandsRecursively) {
    248   std::vector<tok::TokenKind> ExpectedTokens;
    249   ExpectedTokens.push_back(tok::identifier);
    250   ExpectedTokens.push_back(tok::identifier);
    251   ExpectedTokens.push_back(tok::l_paren);
    252   ExpectedTokens.push_back(tok::identifier);
    253   ExpectedTokens.push_back(tok::r_paren);
    254 
    255   std::vector<Token> toks = CheckLex("#define M(x) x\n"
    256                                      "#define C(x) c M(x)\n"
    257                                      "C(f(M(i)))",
    258                                      ExpectedTokens);
    259   EXPECT_EQ("M(i)", getSourceText(toks[3], toks[3]));
    260 }
    261 
    262 TEST_F(LexerTest, LexAPI) {
    263   std::vector<tok::TokenKind> ExpectedTokens;
    264   ExpectedTokens.push_back(tok::l_square);
    265   ExpectedTokens.push_back(tok::identifier);
    266   ExpectedTokens.push_back(tok::r_square);
    267   ExpectedTokens.push_back(tok::l_square);
    268   ExpectedTokens.push_back(tok::identifier);
    269   ExpectedTokens.push_back(tok::r_square);
    270   ExpectedTokens.push_back(tok::identifier);
    271   ExpectedTokens.push_back(tok::identifier);
    272   ExpectedTokens.push_back(tok::identifier);
    273   ExpectedTokens.push_back(tok::identifier);
    274 
    275   std::vector<Token> toks = CheckLex("#define M(x) [x]\n"
    276                                      "#define N(x) x\n"
    277                                      "#define INN(x) x\n"
    278                                      "#define NOF1 INN(val)\n"
    279                                      "#define NOF2 val\n"
    280                                      "M(foo) N([bar])\n"
    281                                      "N(INN(val)) N(NOF1) N(NOF2) N(val)",
    282                                      ExpectedTokens);
    283 
    284   SourceLocation lsqrLoc = toks[0].getLocation();
    285   SourceLocation idLoc = toks[1].getLocation();
    286   SourceLocation rsqrLoc = toks[2].getLocation();
    287   std::pair<SourceLocation,SourceLocation>
    288     macroPair = SourceMgr.getExpansionRange(lsqrLoc);
    289   SourceRange macroRange = SourceRange(macroPair.first, macroPair.second);
    290 
    291   SourceLocation Loc;
    292   EXPECT_TRUE(Lexer::isAtStartOfMacroExpansion(lsqrLoc, SourceMgr, LangOpts, &Loc));
    293   EXPECT_EQ(Loc, macroRange.getBegin());
    294   EXPECT_FALSE(Lexer::isAtStartOfMacroExpansion(idLoc, SourceMgr, LangOpts));
    295   EXPECT_FALSE(Lexer::isAtEndOfMacroExpansion(idLoc, SourceMgr, LangOpts));
    296   EXPECT_TRUE(Lexer::isAtEndOfMacroExpansion(rsqrLoc, SourceMgr, LangOpts, &Loc));
    297   EXPECT_EQ(Loc, macroRange.getEnd());
    298 
    299   CharSourceRange range = Lexer::makeFileCharRange(
    300            CharSourceRange::getTokenRange(lsqrLoc, idLoc), SourceMgr, LangOpts);
    301   EXPECT_TRUE(range.isInvalid());
    302   range = Lexer::makeFileCharRange(CharSourceRange::getTokenRange(idLoc, rsqrLoc),
    303                                    SourceMgr, LangOpts);
    304   EXPECT_TRUE(range.isInvalid());
    305   range = Lexer::makeFileCharRange(CharSourceRange::getTokenRange(lsqrLoc, rsqrLoc),
    306                                    SourceMgr, LangOpts);
    307   EXPECT_TRUE(!range.isTokenRange());
    308   EXPECT_EQ(range.getAsRange(),
    309             SourceRange(macroRange.getBegin(),
    310                         macroRange.getEnd().getLocWithOffset(1)));
    311 
    312   StringRef text = Lexer::getSourceText(
    313                                CharSourceRange::getTokenRange(lsqrLoc, rsqrLoc),
    314                                SourceMgr, LangOpts);
    315   EXPECT_EQ(text, "M(foo)");
    316 
    317   SourceLocation macroLsqrLoc = toks[3].getLocation();
    318   SourceLocation macroIdLoc = toks[4].getLocation();
    319   SourceLocation macroRsqrLoc = toks[5].getLocation();
    320   SourceLocation fileLsqrLoc = SourceMgr.getSpellingLoc(macroLsqrLoc);
    321   SourceLocation fileIdLoc = SourceMgr.getSpellingLoc(macroIdLoc);
    322   SourceLocation fileRsqrLoc = SourceMgr.getSpellingLoc(macroRsqrLoc);
    323 
    324   range = Lexer::makeFileCharRange(
    325       CharSourceRange::getTokenRange(macroLsqrLoc, macroIdLoc),
    326       SourceMgr, LangOpts);
    327   EXPECT_EQ(SourceRange(fileLsqrLoc, fileIdLoc.getLocWithOffset(3)),
    328             range.getAsRange());
    329 
    330   range = Lexer::makeFileCharRange(CharSourceRange::getTokenRange(macroIdLoc, macroRsqrLoc),
    331                                    SourceMgr, LangOpts);
    332   EXPECT_EQ(SourceRange(fileIdLoc, fileRsqrLoc.getLocWithOffset(1)),
    333             range.getAsRange());
    334 
    335   macroPair = SourceMgr.getExpansionRange(macroLsqrLoc);
    336   range = Lexer::makeFileCharRange(
    337                      CharSourceRange::getTokenRange(macroLsqrLoc, macroRsqrLoc),
    338                      SourceMgr, LangOpts);
    339   EXPECT_EQ(SourceRange(macroPair.first, macroPair.second.getLocWithOffset(1)),
    340             range.getAsRange());
    341 
    342   text = Lexer::getSourceText(
    343           CharSourceRange::getTokenRange(SourceRange(macroLsqrLoc, macroIdLoc)),
    344           SourceMgr, LangOpts);
    345   EXPECT_EQ(text, "[bar");
    346 
    347 
    348   SourceLocation idLoc1 = toks[6].getLocation();
    349   SourceLocation idLoc2 = toks[7].getLocation();
    350   SourceLocation idLoc3 = toks[8].getLocation();
    351   SourceLocation idLoc4 = toks[9].getLocation();
    352   EXPECT_EQ("INN", Lexer::getImmediateMacroName(idLoc1, SourceMgr, LangOpts));
    353   EXPECT_EQ("INN", Lexer::getImmediateMacroName(idLoc2, SourceMgr, LangOpts));
    354   EXPECT_EQ("NOF2", Lexer::getImmediateMacroName(idLoc3, SourceMgr, LangOpts));
    355   EXPECT_EQ("N", Lexer::getImmediateMacroName(idLoc4, SourceMgr, LangOpts));
    356 }
    357 
    358 } // anonymous namespace
    359