zcov: / lib/Lex/TokenConcatenation.cpp


Files: 1 Branches Taken: 60.3% 79 / 131
Generated: 2010-02-10 01:31 Branches Executed: 80.2% 105 / 131
Line Coverage: 76.2% 80 / 105


Programs: 1 Runs 2897


       1                 : //===--- TokenConcatenation.cpp - Token Concatenation Avoidance -----------===//
       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                 : // This file implements the TokenConcatenation class.
      11                 : //
      12                 : //===----------------------------------------------------------------------===//
      13                 : 
      14                 : #include "clang/Lex/TokenConcatenation.h"
      15                 : #include "clang/Lex/Preprocessor.h"
      16                 : using namespace clang;
      17                 : 
      18                 : 
      19                 : /// StartsWithL - Return true if the spelling of this token starts with 'L'.
      20                1: bool TokenConcatenation::StartsWithL(const Token &Tok) const {
                        1: branch 1 taken
                        0: branch 2 not taken
      21                1:   if (!Tok.needsCleaning()) {
      22                1:     SourceManager &SM = PP.getSourceManager();
      23                1:     return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation())) == 'L';
      24                 :   }
      25                 : 
                        0: branch 1 not taken
                        0: branch 2 not taken
      26                0:   if (Tok.getLength() < 256) {
      27                 :     char Buffer[256];
      28                0:     const char *TokPtr = Buffer;
      29                0:     PP.getSpelling(Tok, TokPtr);
      30                0:     return TokPtr[0] == 'L';
      31                 :   }
      32                 : 
      33                0:   return PP.getSpelling(Tok)[0] == 'L';
      34                 : }
      35                 : 
      36                 : /// IsIdentifierL - Return true if the spelling of this token is literally
      37                 : /// 'L'.
      38                1: bool TokenConcatenation::IsIdentifierL(const Token &Tok) const {
                        1: branch 1 taken
                        0: branch 2 not taken
      39                1:   if (!Tok.needsCleaning()) {
                        0: branch 1 not taken
                        1: branch 2 taken
      40                1:     if (Tok.getLength() != 1)
      41                0:       return false;
      42                1:     SourceManager &SM = PP.getSourceManager();
      43                1:     return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation())) == 'L';
      44                 :   }
      45                 : 
                        0: branch 1 not taken
                        0: branch 2 not taken
      46                0:   if (Tok.getLength() < 256) {
      47                 :     char Buffer[256];
      48                0:     const char *TokPtr = Buffer;
                        0: branch 1 not taken
                        0: branch 2 not taken
      49                0:     if (PP.getSpelling(Tok, TokPtr) != 1)
      50                0:       return false;
      51                0:     return TokPtr[0] == 'L';
      52                 :   }
      53                 : 
      54                0:   return PP.getSpelling(Tok) == "L";
      55                 : }
      56                 : 
      57              204: TokenConcatenation::TokenConcatenation(Preprocessor &pp) : PP(pp) {
      58              204:   memset(TokenInfo, 0, sizeof(TokenInfo));
      59                 : 
      60                 :   // These tokens have custom code in AvoidConcat.
      61              204:   TokenInfo[tok::identifier      ] |= aci_custom;
      62              204:   TokenInfo[tok::numeric_constant] |= aci_custom_firstchar;
      63              204:   TokenInfo[tok::period          ] |= aci_custom_firstchar;
      64              204:   TokenInfo[tok::amp             ] |= aci_custom_firstchar;
      65              204:   TokenInfo[tok::plus            ] |= aci_custom_firstchar;
      66              204:   TokenInfo[tok::minus           ] |= aci_custom_firstchar;
      67              204:   TokenInfo[tok::slash           ] |= aci_custom_firstchar;
      68              204:   TokenInfo[tok::less            ] |= aci_custom_firstchar;
      69              204:   TokenInfo[tok::greater         ] |= aci_custom_firstchar;
      70              204:   TokenInfo[tok::pipe            ] |= aci_custom_firstchar;
      71              204:   TokenInfo[tok::percent         ] |= aci_custom_firstchar;
      72              204:   TokenInfo[tok::colon           ] |= aci_custom_firstchar;
      73              204:   TokenInfo[tok::hash            ] |= aci_custom_firstchar;
      74              204:   TokenInfo[tok::arrow           ] |= aci_custom_firstchar;
      75                 : 
      76                 :   // These tokens change behavior if followed by an '='.
      77              204:   TokenInfo[tok::amp         ] |= aci_avoid_equal;           // &=
      78              204:   TokenInfo[tok::plus        ] |= aci_avoid_equal;           // +=
      79              204:   TokenInfo[tok::minus       ] |= aci_avoid_equal;           // -=
      80              204:   TokenInfo[tok::slash       ] |= aci_avoid_equal;           // /=
      81              204:   TokenInfo[tok::less        ] |= aci_avoid_equal;           // <=
      82              204:   TokenInfo[tok::greater     ] |= aci_avoid_equal;           // >=
      83              204:   TokenInfo[tok::pipe        ] |= aci_avoid_equal;           // |=
      84              204:   TokenInfo[tok::percent     ] |= aci_avoid_equal;           // %=
      85              204:   TokenInfo[tok::star        ] |= aci_avoid_equal;           // *=
      86              204:   TokenInfo[tok::exclaim     ] |= aci_avoid_equal;           // !=
      87              204:   TokenInfo[tok::lessless    ] |= aci_avoid_equal;           // <<=
      88              204:   TokenInfo[tok::greaterequal] |= aci_avoid_equal;           // >>=
      89              204:   TokenInfo[tok::caret       ] |= aci_avoid_equal;           // ^=
      90              204:   TokenInfo[tok::equal       ] |= aci_avoid_equal;           // ==
      91              204: }
      92                 : 
      93                 : /// GetFirstChar - Get the first character of the token \arg Tok,
      94                 : /// avoiding calls to getSpelling where possible.
      95              998: static char GetFirstChar(Preprocessor &PP, const Token &Tok) {
                       14: branch 1 taken
                      984: branch 2 taken
      96              998:   if (IdentifierInfo *II = Tok.getIdentifierInfo()) {
      97                 :     // Avoid spelling identifiers, the most common form of token.
      98               14:     return II->getNameStart()[0];
                      984: branch 1 taken
                        0: branch 2 not taken
      99              984:   } else if (!Tok.needsCleaning()) {
                      437: branch 1 taken
                      547: branch 2 taken
                      437: branch 4 taken
                        0: branch 5 not taken
                      437: branch 6 taken
                      547: branch 7 taken
     100              984:     if (Tok.isLiteral() && Tok.getLiteralData()) {
     101              437:       return *Tok.getLiteralData();
     102                 :     } else {
     103              547:       SourceManager &SM = PP.getSourceManager();
     104              547:       return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation()));
     105                 :     }
                        0: branch 1 not taken
                        0: branch 2 not taken
     106                0:   } else if (Tok.getLength() < 256) {
     107                 :     char Buffer[256];
     108                0:     const char *TokPtr = Buffer;
     109                0:     PP.getSpelling(Tok, TokPtr);
     110                0:     return TokPtr[0];
     111                 :   } else {
     112                0:     return PP.getSpelling(Tok)[0];
     113                 :   }
     114                 : }
     115                 : 
     116                 : /// AvoidConcat - If printing PrevTok immediately followed by Tok would cause
     117                 : /// the two individual tokens to be lexed as a single token, return true
     118                 : /// (which causes a space to be printed between them).  This allows the output
     119                 : /// of -E mode to be lexed to the same token stream as lexing the input
     120                 : /// directly would.
     121                 : ///
     122                 : /// This code must conservatively return true if it doesn't want to be 100%
     123                 : /// accurate.  This will cause the output to include extra space characters,
     124                 : /// but the resulting output won't have incorrect concatenations going on.
     125                 : /// Examples include "..", which we print with a space between, because we
     126                 : /// don't want to track enough to tell "x.." from "...".
     127                 : bool TokenConcatenation::AvoidConcat(const Token &PrevTok,
     128             4621:                                      const Token &Tok) const {
     129                 :   // First, check to see if the tokens were directly adjacent in the original
     130                 :   // source.  If they were, it must be okay to stick them together: if there
     131                 :   // were an issue, the tokens would have been lexed differently.
                     2599: branch 2 taken
                     2022: branch 3 taken
                     2580: branch 6 taken
                       19: branch 7 taken
                     2528: branch 13 taken
                       52: branch 14 taken
                     2528: branch 15 taken
                     2093: branch 16 taken
     132             4621:   if (PrevTok.getLocation().isFileID() && Tok.getLocation().isFileID() &&
     133                 :       PrevTok.getLocation().getFileLocWithOffset(PrevTok.getLength()) ==
     134                 :         Tok.getLocation())
     135             2528:     return false;
     136                 : 
     137             2093:   tok::TokenKind PrevKind = PrevTok.getKind();
                      324: branch 1 taken
                     1769: branch 2 taken
     138             2093:   if (PrevTok.getIdentifierInfo())  // Language keyword or named operator.
     139              324:     PrevKind = tok::identifier;
     140                 : 
     141                 :   // Look up information on when we should avoid concatenation with prevtok.
     142             2093:   unsigned ConcatInfo = TokenInfo[PrevKind];
     143                 : 
     144                 :   // If prevtok never causes a problem for anything after it, return quickly.
                      746: branch 0 taken
                     1347: branch 1 taken
     145             2093:   if (ConcatInfo == 0) return false;
     146                 : 
                      469: branch 0 taken
                      878: branch 1 taken
     147             1347:   if (ConcatInfo & aci_avoid_equal) {
     148                 :     // If the next token is '=' or '==', avoid concatenation.
                      467: branch 1 taken
                        2: branch 2 taken
                        0: branch 4 not taken
                      467: branch 5 taken
                        2: branch 6 taken
                      467: branch 7 taken
     149              469:     if (Tok.is(tok::equal) || Tok.is(tok::equalequal))
     150                2:       return true;
     151              467:     ConcatInfo &= ~aci_avoid_equal;
     152                 :   }
     153                 : 
                       33: branch 0 taken
                     1312: branch 1 taken
     154             1345:   if (ConcatInfo == 0) return false;
     155                 : 
     156                 :   // Basic algorithm: we look at the first character of the second token, and
     157                 :   // determine whether it, if appended to the first token, would form (or
     158                 :   // would contribute) to a larger token if concatenated.
     159             1312:   char FirstChar = 0;
                      988: branch 0 taken
                      324: branch 1 taken
     160             1312:   if (ConcatInfo & aci_custom) {
     161                 :     // If the token does not need to know the first character, don't get it.
     162                 :   } else {
     163              988:     FirstChar = GetFirstChar(PP, Tok);
     164                 :   }
     165                 : 
                        0: branch 0 not taken
                      324: branch 1 taken
                      536: branch 2 taken
                        6: branch 3 taken
                        0: branch 4 not taken
                        7: branch 5 taken
                      425: branch 6 taken
                        0: branch 7 not taken
                        0: branch 8 not taken
                        1: branch 9 taken
                        0: branch 10 not taken
                        1: branch 11 taken
                        9: branch 12 taken
                        3: branch 13 taken
                        0: branch 14 not taken
     166             1312:   switch (PrevKind) {
     167                0:   default: assert(0 && "InitAvoidConcatTokenInfo built wrong");
     168                 :   case tok::identifier:   // id+id or id+number or id+L"foo".
     169                 :     // id+'.'... will not append.
                       10: branch 1 taken
                      314: branch 2 taken
     170              324:     if (Tok.is(tok::numeric_constant))
     171               10:       return GetFirstChar(PP, Tok) != '.';
     172                 : 
                      314: branch 1 taken
                        0: branch 2 not taken
                        0: branch 4 not taken
                      314: branch 5 taken
                        0: branch 6 not taken
                      314: branch 7 taken
     173              314:     if (Tok.getIdentifierInfo() || Tok.is(tok::wide_string_literal) /* ||
     174                 :      Tok.is(tok::wide_char_literal)*/)
     175                0:       return true;
     176                 : 
     177                 :     // If this isn't identifier + string, we're done.
                      314: branch 1 taken
                        0: branch 2 not taken
                      313: branch 4 taken
                        1: branch 5 taken
                      313: branch 6 taken
                        1: branch 7 taken
     178              314:     if (Tok.isNot(tok::char_constant) && Tok.isNot(tok::string_literal))
     179              313:       return false;
     180                 : 
     181                 :     // FIXME: need a wide_char_constant!
     182                 : 
     183                 :     // If the string was a wide string L"foo" or wide char L'f', it would
     184                 :     // concat with the previous identifier into fooL"bar".  Avoid this.
                        0: branch 1 not taken
                        1: branch 2 taken
     185                1:     if (StartsWithL(Tok))
     186                0:       return true;
     187                 : 
     188                 :     // Otherwise, this is a narrow character or string.  If the *identifier*
     189                 :     // is a literal 'L', avoid pasting L "foo" -> L"foo".
     190                1:     return IsIdentifierL(PrevTok);
     191                 :   case tok::numeric_constant:
     192                 :     return isalnum(FirstChar) || Tok.is(tok::numeric_constant) ||
                      536: branch 1 taken
                        0: branch 2 not taken
                      536: branch 4 taken
                        0: branch 5 not taken
                      532: branch 6 taken
                        4: branch 7 taken
                      322: branch 8 taken
                      210: branch 9 taken
                        0: branch 10 not taken
                      322: branch 11 taken
     193              536:     FirstChar == '+' || FirstChar == '-' || FirstChar == '.';
     194                 :   case tok::period:          // ..., .*, .1234
     195                 :     return FirstChar == '.' || isdigit(FirstChar) ||
                        3: branch 0 taken
                        3: branch 1 taken
                        3: branch 2 taken
                        0: branch 3 not taken
                        0: branch 5 not taken
                        3: branch 6 taken
                        3: branch 7 taken
                        3: branch 8 taken
     196                6:     (PP.getLangOptions().CPlusPlus && FirstChar == '*');
     197                 :   case tok::amp:             // &&
     198                0:     return FirstChar == '&';
     199                 :   case tok::plus:            // ++
     200                7:     return FirstChar == '+';
     201                 :   case tok::minus:           // --, ->, ->*
                      424: branch 0 taken
                        1: branch 1 taken
                        0: branch 2 not taken
                      424: branch 3 taken
     202              425:     return FirstChar == '-' || FirstChar == '>';
     203                 :   case tok::slash:           //, /*, //
                        0: branch 0 not taken
                        0: branch 1 not taken
                        0: branch 2 not taken
                        0: branch 3 not taken
     204                0:     return FirstChar == '*' || FirstChar == '/';
     205                 :   case tok::less:            // <<, <<=, <:, <%
                        0: branch 0 not taken
                        0: branch 1 not taken
                        0: branch 2 not taken
                        0: branch 3 not taken
                        0: branch 4 not taken
                        0: branch 5 not taken
     206                0:     return FirstChar == '<' || FirstChar == ':' || FirstChar == '%';
     207                 :   case tok::greater:         // >>, >>=
     208                1:     return FirstChar == '>';
     209                 :   case tok::pipe:            // ||
     210                0:     return FirstChar == '|';
     211                 :   case tok::percent:         // %>, %:
                        1: branch 0 taken
                        0: branch 1 not taken
                        0: branch 2 not taken
                        1: branch 3 taken
     212                1:     return FirstChar == '>' || FirstChar == ':';
     213                 :   case tok::colon:           // ::, :>
     214                 :     return FirstChar == '>' ||
                        9: branch 0 taken
                        0: branch 1 not taken
                        0: branch 3 not taken
                        9: branch 4 taken
                        9: branch 5 taken
                        9: branch 6 taken
     215                9:     (PP.getLangOptions().CPlusPlus && FirstChar == ':');
     216                 :   case tok::hash:            // ##, #@, %:%:
                        3: branch 0 taken
                        0: branch 1 not taken
                        3: branch 2 taken
                        0: branch 3 not taken
                        0: branch 4 not taken
                        3: branch 5 taken
     217                3:     return FirstChar == '#' || FirstChar == '@' || FirstChar == '%';
     218                 :   case tok::arrow:           // ->*
                        0: branch 1 not taken
                        0: branch 2 not taken
                        0: branch 3 not taken
                        0: branch 4 not taken
     219                0:     return PP.getLangOptions().CPlusPlus && FirstChar == '*';
     220                 :   }
     221                 : }

Generated: 2010-02-10 01:31 by zcov