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