rosapps/smartpdf/poppler/poppler/Lexer.h

   1 //========================================================================
   2 //
   3 // Lexer.h
   4 //
   5 // Copyright 1996-2003 Glyph & Cog, LLC
   6 // Copyright 2006 Krzysztof Kowalczyk (http://blog.kowalczyk.info)
   7 //
   8 //========================================================================
   9
  10 #ifndef LEXER_H
  11 #define LEXER_H
  12
  13 #ifdef USE_GCC_PRAGMAS
  14 #pragma interface
  15 #endif
  16
  17 #include <assert.h>
  18 #include "Object.h"
  19 #include "Stream.h"
  20
  21 class XRef;
  22
  23 #define tokBufSize 128          // size of token buffer
  24
  25 //------------------------------------------------------------------------
  26 // Lexer
  27 //------------------------------------------------------------------------
  28
  29 class Lexer {
  30 public:
  31
  32   // Construct a lexer for a single stream.  Deletes the stream when
  33   // lexer is deleted.
  34   Lexer(XRef *xrefA, Stream *str);
  35
  36   // Construct a lexer for a stream or array of streams (assumes obj
  37   // is either a stream or array of streams).
  38   Lexer(XRef *xrefA, Object *obj);
  39
  40   // Destructor.
  41   ~Lexer();
  42
  43   // Get the next object from the input stream.
  44   Object *getObj(Object *obj, int objNum = -1);
  45
  46   // Skip to the beginning of the next line in the input stream.
  47   void skipToNextLine();
  48
  49   // Skip over one character.
  50   void skipChar();
  51
  52   Stream *getStream();
  53
  54   // Get current position in file.  This is only used for error
  55   // messages, so it returns an int instead of a Guint.
  56   int getPos();
  57
  58   // Set position in file.
  59   void setPos(Guint pos, int dir = 0);
  60
  61   // Returns true if <c> is a whitespace character.
  62   static GBool isSpace(int c);
  63
  64 private:
  65
  66   // we really want lookChar() and getChar() to be inlined
  67   int lookChar() {
  68     if (LOOK_VALUE_NOT_CACHED != lookCharLastValueCached) {
  69       return lookCharLastValueCached;
  70     }
  71     lookCharLastValueCached = getChar();
  72     return lookCharLastValueCached;
  73   }
  74
  75   int getChar() {
  76     int       c;
  77     GBool     hasMoreData;
  78
  79     if (LOOK_VALUE_NOT_CACHED != lookCharLastValueCached) {
  80       c = lookCharLastValueCached;
  81       lookCharLastValueCached = LOOK_VALUE_NOT_CACHED;
  82       assert( (c >= LOOK_VALUE_NOT_CACHED) && (c < 256));
  83       return c;
  84     }
  85
  86     while (curStrHasGetBuf) {
  87       if (bufLeft > 0) {
  88         c = *bufCurPos++ & 0xff;
  89         bufLeft--;
  90         return c;
  91       }
  92       hasMoreData = fillBuf();
  93       if (!hasMoreData)
  94         nextStream();
  95     }
  96
  97     c = EOF;
  98     while (!curStr.isNone() && (c = curStr.streamGetChar()) == EOF) {
  99       nextStream();
 100     }
 101     return c;
 102   }
 103
 104   void  nextStream();
 105   GBool fillBuf();
 106
 107   Array *   streams;                // array of input streams
 108   int       strPtr;                 // index of current stream
 109   Object    curStr;                 // current stream
 110   GBool     freeArray;              // should lexer free the streams array?
 111   char      tokBuf[tokBufSize];     // temporary token buffer
 112   XRef *    xref;
 113
 114   // often (e.g. ~30% on PDF Refernce 1.6 pdf file from Adobe site) getChar
 115   // is called right after lookChar. In order to avoid expensive re-doing
 116   // getChar() of underlying stream, we cache the last value found by
 117   // lookChar() in lookCharLastValueCached. A special value
 118   // LOOK_VALUE_NOT_CACHED that should never be part of stream indicates
 119   // that no value was cached
 120   static const int LOOK_VALUE_NOT_CACHED = -3;
 121   int       lookCharLastValueCached;
 122
 123   GBool     curStrHasGetBuf;     // does current stream support GetBuf() ?
 124   char *    buf;
 125   char *    bufCurPos;
 126   int       bufSize;
 127   int       bufLeft;
 128 };
 129
 130 #endif