Home | History | Annotate | Download | only in Basic
      1 //===--- SourceLocation.h - Compact identifier for Source Files -*- C++ -*-===//
      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 /// \file
     11 /// \brief Defines the clang::SourceLocation class and associated facilities.
     12 ///
     13 //===----------------------------------------------------------------------===//
     14 
     15 #ifndef LLVM_CLANG_SOURCELOCATION_H
     16 #define LLVM_CLANG_SOURCELOCATION_H
     17 
     18 #include "clang/Basic/LLVM.h"
     19 #include "llvm/Support/PointerLikeTypeTraits.h"
     20 #include "llvm/Support/Compiler.h"
     21 #include <utility>
     22 #include <functional>
     23 #include <cassert>
     24 
     25 namespace llvm {
     26   class MemoryBuffer;
     27   template <typename T> struct DenseMapInfo;
     28   template <typename T> struct isPodLike;
     29 }
     30 
     31 namespace clang {
     32 
     33 class SourceManager;
     34 
     35 /// \brief An opaque identifier used by SourceManager which refers to a
     36 /// source file (MemoryBuffer) along with its \#include path and \#line data.
     37 ///
     38 class FileID {
     39   /// \brief A mostly-opaque identifier, where 0 is "invalid", >0 is
     40   /// this module, and <-1 is something loaded from another module.
     41   int ID;
     42 public:
     43   FileID() : ID(0) {}
     44 
     45   bool isInvalid() const { return ID == 0; }
     46 
     47   bool operator==(const FileID &RHS) const { return ID == RHS.ID; }
     48   bool operator<(const FileID &RHS) const { return ID < RHS.ID; }
     49   bool operator<=(const FileID &RHS) const { return ID <= RHS.ID; }
     50   bool operator!=(const FileID &RHS) const { return !(*this == RHS); }
     51   bool operator>(const FileID &RHS) const { return RHS < *this; }
     52   bool operator>=(const FileID &RHS) const { return RHS <= *this; }
     53 
     54   static FileID getSentinel() { return get(-1); }
     55   unsigned getHashValue() const { return static_cast<unsigned>(ID); }
     56 
     57 private:
     58   friend class SourceManager;
     59   friend class ASTWriter;
     60   friend class ASTReader;
     61 
     62   static FileID get(int V) {
     63     FileID F;
     64     F.ID = V;
     65     return F;
     66   }
     67   int getOpaqueValue() const { return ID; }
     68 };
     69 
     70 
     71 /// \brief Encodes a location in the source. The SourceManager can decode this
     72 /// to get at the full include stack, line and column information.
     73 ///
     74 /// Technically, a source location is simply an offset into the manager's view
     75 /// of the input source, which is all input buffers (including macro
     76 /// expansions) concatenated in an effectively arbitrary order. The manager
     77 /// actually maintains two blocks of input buffers. One, starting at offset
     78 /// 0 and growing upwards, contains all buffers from this module. The other,
     79 /// starting at the highest possible offset and growing downwards, contains
     80 /// buffers of loaded modules.
     81 ///
     82 /// In addition, one bit of SourceLocation is used for quick access to the
     83 /// information whether the location is in a file or a macro expansion.
     84 ///
     85 /// It is important that this type remains small. It is currently 32 bits wide.
     86 class SourceLocation {
     87   unsigned ID;
     88   friend class SourceManager;
     89   friend class ASTReader;
     90   friend class ASTWriter;
     91   enum {
     92     MacroIDBit = 1U << 31
     93   };
     94 public:
     95 
     96   SourceLocation() : ID(0) {}
     97 
     98   bool isFileID() const  { return (ID & MacroIDBit) == 0; }
     99   bool isMacroID() const { return (ID & MacroIDBit) != 0; }
    100 
    101   /// \brief Return true if this is a valid SourceLocation object.
    102   ///
    103   /// Invalid SourceLocations are often used when events have no corresponding
    104   /// location in the source (e.g. a diagnostic is required for a command line
    105   /// option).
    106   bool isValid() const { return ID != 0; }
    107   bool isInvalid() const { return ID == 0; }
    108 
    109 private:
    110   /// \brief Return the offset into the manager's global input view.
    111   unsigned getOffset() const {
    112     return ID & ~MacroIDBit;
    113   }
    114 
    115   static SourceLocation getFileLoc(unsigned ID) {
    116     assert((ID & MacroIDBit) == 0 && "Ran out of source locations!");
    117     SourceLocation L;
    118     L.ID = ID;
    119     return L;
    120   }
    121 
    122   static SourceLocation getMacroLoc(unsigned ID) {
    123     assert((ID & MacroIDBit) == 0 && "Ran out of source locations!");
    124     SourceLocation L;
    125     L.ID = MacroIDBit | ID;
    126     return L;
    127   }
    128 public:
    129 
    130   /// \brief Return a source location with the specified offset from this
    131   /// SourceLocation.
    132   SourceLocation getLocWithOffset(int Offset) const {
    133     assert(((getOffset()+Offset) & MacroIDBit) == 0 && "offset overflow");
    134     SourceLocation L;
    135     L.ID = ID+Offset;
    136     return L;
    137   }
    138 
    139   /// \brief When a SourceLocation itself cannot be used, this returns
    140   /// an (opaque) 32-bit integer encoding for it.
    141   ///
    142   /// This should only be passed to SourceLocation::getFromRawEncoding, it
    143   /// should not be inspected directly.
    144   unsigned getRawEncoding() const { return ID; }
    145 
    146   /// \brief Turn a raw encoding of a SourceLocation object into
    147   /// a real SourceLocation.
    148   ///
    149   /// \see getRawEncoding.
    150   static SourceLocation getFromRawEncoding(unsigned Encoding) {
    151     SourceLocation X;
    152     X.ID = Encoding;
    153     return X;
    154   }
    155 
    156   /// \brief When a SourceLocation itself cannot be used, this returns
    157   /// an (opaque) pointer encoding for it.
    158   ///
    159   /// This should only be passed to SourceLocation::getFromPtrEncoding, it
    160   /// should not be inspected directly.
    161   void* getPtrEncoding() const {
    162     // Double cast to avoid a warning "cast to pointer from integer of different
    163     // size".
    164     return (void*)(uintptr_t)getRawEncoding();
    165   }
    166 
    167   /// getFromPtrEncoding - Turn a pointer encoding of a SourceLocation object
    168   /// into a real SourceLocation.
    169   static SourceLocation getFromPtrEncoding(const void *Encoding) {
    170     return getFromRawEncoding((unsigned)(uintptr_t)Encoding);
    171   }
    172 
    173   void print(raw_ostream &OS, const SourceManager &SM) const;
    174   void dump(const SourceManager &SM) const;
    175 };
    176 
    177 inline bool operator==(const SourceLocation &LHS, const SourceLocation &RHS) {
    178   return LHS.getRawEncoding() == RHS.getRawEncoding();
    179 }
    180 
    181 inline bool operator!=(const SourceLocation &LHS, const SourceLocation &RHS) {
    182   return !(LHS == RHS);
    183 }
    184 
    185 inline bool operator<(const SourceLocation &LHS, const SourceLocation &RHS) {
    186   return LHS.getRawEncoding() < RHS.getRawEncoding();
    187 }
    188 
    189 /// \brief A trival tuple used to represent a source range.
    190 class SourceRange {
    191   SourceLocation B;
    192   SourceLocation E;
    193 public:
    194   SourceRange(): B(SourceLocation()), E(SourceLocation()) {}
    195   SourceRange(SourceLocation loc) : B(loc), E(loc) {}
    196   SourceRange(SourceLocation begin, SourceLocation end) : B(begin), E(end) {}
    197 
    198   SourceLocation getBegin() const { return B; }
    199   SourceLocation getEnd() const { return E; }
    200 
    201   void setBegin(SourceLocation b) { B = b; }
    202   void setEnd(SourceLocation e) { E = e; }
    203 
    204   bool isValid() const { return B.isValid() && E.isValid(); }
    205   bool isInvalid() const { return !isValid(); }
    206 
    207   bool operator==(const SourceRange &X) const {
    208     return B == X.B && E == X.E;
    209   }
    210 
    211   bool operator!=(const SourceRange &X) const {
    212     return B != X.B || E != X.E;
    213   }
    214 };
    215 
    216 /// \brief Represents a character-granular source range.
    217 ///
    218 /// The underlying SourceRange can either specify the starting/ending character
    219 /// of the range, or it can specify the start or the range and the start of the
    220 /// last token of the range (a "token range").  In the token range case, the
    221 /// size of the last token must be measured to determine the actual end of the
    222 /// range.
    223 class CharSourceRange {
    224   SourceRange Range;
    225   bool IsTokenRange;
    226 public:
    227   CharSourceRange() : IsTokenRange(false) {}
    228   CharSourceRange(SourceRange R, bool ITR) : Range(R),IsTokenRange(ITR){}
    229 
    230   static CharSourceRange getTokenRange(SourceRange R) {
    231     CharSourceRange Result;
    232     Result.Range = R;
    233     Result.IsTokenRange = true;
    234     return Result;
    235   }
    236 
    237   static CharSourceRange getCharRange(SourceRange R) {
    238     CharSourceRange Result;
    239     Result.Range = R;
    240     Result.IsTokenRange = false;
    241     return Result;
    242   }
    243 
    244   static CharSourceRange getTokenRange(SourceLocation B, SourceLocation E) {
    245     return getTokenRange(SourceRange(B, E));
    246   }
    247   static CharSourceRange getCharRange(SourceLocation B, SourceLocation E) {
    248     return getCharRange(SourceRange(B, E));
    249   }
    250 
    251   /// \brief Return true if the end of this range specifies the start of
    252   /// the last token.  Return false if the end of this range specifies the last
    253   /// character in the range.
    254   bool isTokenRange() const { return IsTokenRange; }
    255   bool isCharRange() const { return !IsTokenRange; }
    256 
    257   SourceLocation getBegin() const { return Range.getBegin(); }
    258   SourceLocation getEnd() const { return Range.getEnd(); }
    259   const SourceRange &getAsRange() const { return Range; }
    260 
    261   void setBegin(SourceLocation b) { Range.setBegin(b); }
    262   void setEnd(SourceLocation e) { Range.setEnd(e); }
    263 
    264   bool isValid() const { return Range.isValid(); }
    265   bool isInvalid() const { return !isValid(); }
    266 };
    267 
    268 /// \brief A SourceLocation and its associated SourceManager.
    269 ///
    270 /// This is useful for argument passing to functions that expect both objects.
    271 class FullSourceLoc : public SourceLocation {
    272   const SourceManager *SrcMgr;
    273 public:
    274   /// \brief Creates a FullSourceLoc where isValid() returns \c false.
    275   explicit FullSourceLoc() : SrcMgr(0) {}
    276 
    277   explicit FullSourceLoc(SourceLocation Loc, const SourceManager &SM)
    278     : SourceLocation(Loc), SrcMgr(&SM) {}
    279 
    280   /// \pre This FullSourceLoc has an associated SourceManager.
    281   const SourceManager &getManager() const {
    282     assert(SrcMgr && "SourceManager is NULL.");
    283     return *SrcMgr;
    284   }
    285 
    286   FileID getFileID() const;
    287 
    288   FullSourceLoc getExpansionLoc() const;
    289   FullSourceLoc getSpellingLoc() const;
    290 
    291   unsigned getExpansionLineNumber(bool *Invalid = 0) const;
    292   unsigned getExpansionColumnNumber(bool *Invalid = 0) const;
    293 
    294   unsigned getSpellingLineNumber(bool *Invalid = 0) const;
    295   unsigned getSpellingColumnNumber(bool *Invalid = 0) const;
    296 
    297   const char *getCharacterData(bool *Invalid = 0) const;
    298 
    299   const llvm::MemoryBuffer* getBuffer(bool *Invalid = 0) const;
    300 
    301   /// \brief Return a StringRef to the source buffer data for the
    302   /// specified FileID.
    303   StringRef getBufferData(bool *Invalid = 0) const;
    304 
    305   /// \brief Decompose the specified location into a raw FileID + Offset pair.
    306   ///
    307   /// The first element is the FileID, the second is the offset from the
    308   /// start of the buffer of the location.
    309   std::pair<FileID, unsigned> getDecomposedLoc() const;
    310 
    311   bool isInSystemHeader() const;
    312 
    313   /// \brief Determines the order of 2 source locations in the translation unit.
    314   ///
    315   /// \returns true if this source location comes before 'Loc', false otherwise.
    316   bool isBeforeInTranslationUnitThan(SourceLocation Loc) const;
    317 
    318   /// \brief Determines the order of 2 source locations in the translation unit.
    319   ///
    320   /// \returns true if this source location comes before 'Loc', false otherwise.
    321   bool isBeforeInTranslationUnitThan(FullSourceLoc Loc) const {
    322     assert(Loc.isValid());
    323     assert(SrcMgr == Loc.SrcMgr && "Loc comes from another SourceManager!");
    324     return isBeforeInTranslationUnitThan((SourceLocation)Loc);
    325   }
    326 
    327   /// \brief Comparison function class, useful for sorting FullSourceLocs.
    328   struct BeforeThanCompare : public std::binary_function<FullSourceLoc,
    329                                                          FullSourceLoc, bool> {
    330     bool operator()(const FullSourceLoc& lhs, const FullSourceLoc& rhs) const {
    331       return lhs.isBeforeInTranslationUnitThan(rhs);
    332     }
    333   };
    334 
    335   /// \brief Prints information about this FullSourceLoc to stderr.
    336   ///
    337   /// This is useful for debugging.
    338   LLVM_ATTRIBUTE_USED void dump() const;
    339 
    340   friend inline bool
    341   operator==(const FullSourceLoc &LHS, const FullSourceLoc &RHS) {
    342     return LHS.getRawEncoding() == RHS.getRawEncoding() &&
    343           LHS.SrcMgr == RHS.SrcMgr;
    344   }
    345 
    346   friend inline bool
    347   operator!=(const FullSourceLoc &LHS, const FullSourceLoc &RHS) {
    348     return !(LHS == RHS);
    349   }
    350 
    351 };
    352 
    353 /// \brief Represents an unpacked "presumed" location which can be presented
    354 /// to the user.
    355 ///
    356 /// A 'presumed' location can be modified by \#line and GNU line marker
    357 /// directives and is always the expansion point of a normal location.
    358 ///
    359 /// You can get a PresumedLoc from a SourceLocation with SourceManager.
    360 class PresumedLoc {
    361   const char *Filename;
    362   unsigned Line, Col;
    363   SourceLocation IncludeLoc;
    364 public:
    365   PresumedLoc() : Filename(0) {}
    366   PresumedLoc(const char *FN, unsigned Ln, unsigned Co, SourceLocation IL)
    367     : Filename(FN), Line(Ln), Col(Co), IncludeLoc(IL) {
    368   }
    369 
    370   /// \brief Return true if this object is invalid or uninitialized.
    371   ///
    372   /// This occurs when created with invalid source locations or when walking
    373   /// off the top of a \#include stack.
    374   bool isInvalid() const { return Filename == 0; }
    375   bool isValid() const { return Filename != 0; }
    376 
    377   /// \brief Return the presumed filename of this location.
    378   ///
    379   /// This can be affected by \#line etc.
    380   const char *getFilename() const { return Filename; }
    381 
    382   /// \brief Return the presumed line number of this location.
    383   ///
    384   /// This can be affected by \#line etc.
    385   unsigned getLine() const { return Line; }
    386 
    387   /// \brief Return the presumed column number of this location.
    388   ///
    389   /// This cannot be affected by \#line, but is packaged here for convenience.
    390   unsigned getColumn() const { return Col; }
    391 
    392   /// \brief Return the presumed include location of this location.
    393   ///
    394   /// This can be affected by GNU linemarker directives.
    395   SourceLocation getIncludeLoc() const { return IncludeLoc; }
    396 };
    397 
    398 
    399 }  // end namespace clang
    400 
    401 namespace llvm {
    402   /// Define DenseMapInfo so that FileID's can be used as keys in DenseMap and
    403   /// DenseSets.
    404   template <>
    405   struct DenseMapInfo<clang::FileID> {
    406     static inline clang::FileID getEmptyKey() {
    407       return clang::FileID();
    408     }
    409     static inline clang::FileID getTombstoneKey() {
    410       return clang::FileID::getSentinel();
    411     }
    412 
    413     static unsigned getHashValue(clang::FileID S) {
    414       return S.getHashValue();
    415     }
    416 
    417     static bool isEqual(clang::FileID LHS, clang::FileID RHS) {
    418       return LHS == RHS;
    419     }
    420   };
    421 
    422   template <>
    423   struct isPodLike<clang::SourceLocation> { static const bool value = true; };
    424   template <>
    425   struct isPodLike<clang::FileID> { static const bool value = true; };
    426 
    427   // Teach SmallPtrSet how to handle SourceLocation.
    428   template<>
    429   class PointerLikeTypeTraits<clang::SourceLocation> {
    430   public:
    431     static inline void *getAsVoidPointer(clang::SourceLocation L) {
    432       return L.getPtrEncoding();
    433     }
    434     static inline clang::SourceLocation getFromVoidPointer(void *P) {
    435       return clang::SourceLocation::getFromRawEncoding((unsigned)(uintptr_t)P);
    436     }
    437     enum { NumLowBitsAvailable = 0 };
    438   };
    439 
    440 }  // end namespace llvm
    441 
    442 #endif
    443