textoutputdev.h

这是一个做pdf阅读器的源代码文件,是大家学习阅读器资料的很好参考

//========================================================================
//
// TextOutputDev.h
//
// Copyright 1997-2003 Glyph & Cog, LLC
//
//========================================================================

#ifndef TEXTOUTPUTDEV_H
#define TEXTOUTPUTDEV_H

#include <aconf.h>

#ifdef USE_GCC_PRAGMAS
#pragma interface
#endif

#include <stdio.h>
#include "gtypes.h"
#include "GfxFont.h"
#include "OutputDev.h"

class GString;
class GList;
class GfxFont;
class GfxState;
class UnicodeMap;
class Link;

class TextWord;
class TextPool;
class TextLine;
class TextLineFrag;
class TextBlock;
class TextFlow;
class TextWordList;
class TextPage;

//------------------------------------------------------------------------

typedef void (*TextOutputFunc)(void *stream, char *text, int len);

//------------------------------------------------------------------------
// TextFontInfo
//------------------------------------------------------------------------

class TextFontInfo {
public:

  TextFontInfo(GfxState *state);
  ~TextFontInfo();

  GBool matches(GfxState *state);

#if TEXTOUT_WORD_LIST
  // Get the font name (which may be NULL).
  GString *getFontName() { return fontName; }

  // Get font descriptor flags.
  GBool isFixedWidth() { return flags & fontFixedWidth; }
  GBool isSerif() { return flags & fontSerif; }
  GBool isSymbolic() { return flags & fontSymbolic; }
  GBool isItalic() { return flags & fontItalic; }
  GBool isBold() { return flags & fontBold; }
#endif

private:

  GfxFont *gfxFont;
#if TEXTOUT_WORD_LIST
  GString *fontName;
  int flags;
#endif

  friend class TextWord;
  friend class TextPage;
};

//------------------------------------------------------------------------
// TextWord
//------------------------------------------------------------------------

class TextWord {
public:

  // Constructor.
  TextWord(GfxState *state, int rotA, double x0, double y0,
	   int charPosA, TextFontInfo *fontA, double fontSize);

  // Destructor.
  ~TextWord();

  // Add a character to the word.
  void addChar(GfxState *state, double x, double y,
	       double dx, double dy, Unicode u);

  // Merge <word> onto the end of <this>.
  void merge(TextWord *word);

  // Compares <this> to <word>, returning -1 (<), 0 (=), or +1 (>),
  // based on a primary-axis comparison, e.g., x ordering if rot=0.
  int primaryCmp(TextWord *word);

  // Return the distance along the primary axis between <this> and
  // <word>.
  double primaryDelta(TextWord *word);

  static int cmpYX(const void *p1, const void *p2);

  // Get the TextFontInfo object associated with this word.
  TextFontInfo *getFontInfo() { return font; }

  // Get the next TextWord on the linked list.
  TextWord *getNext() { return next; }

#if TEXTOUT_WORD_LIST
  int getLength() { return len; }
  Unicode getChar(int idx) { return text[idx]; }
  GString *getText();
  GString *getFontName() { return font->fontName; }
  void getColor(double *r, double *g, double *b)
    { *r = colorR; *g = colorG; *b = colorB; }
  void getBBox(double *xMinA, double *yMinA, double *xMaxA, double *yMaxA)
    { *xMinA = xMin; *yMinA = yMin; *xMaxA = xMax; *yMaxA = yMax; }
  void getCharBBox(int charIdx, double *xMinA, double *yMinA,
		   double *xMaxA, double *yMaxA);
  double getFontSize() { return fontSize; }
  int getRotation() { return rot; }
  int getCharPos() { return charPos; }
  int getCharLen() { return charLen; }
  GBool getSpaceAfter() { return spaceAfter; }
#endif

  GBool isUnderlined() { return underlined; }
  Link *getLink() { return link; }

private:

  int rot;			// rotation, multiple of 90 degrees
				//   (0, 1, 2, or 3)
  double xMin, xMax;		// bounding box x coordinates
  double yMin, yMax;		// bounding box y coordinates
  double base;			// baseline x or y coordinate
  Unicode *text;		// the text
  double *edge;			// "near" edge x or y coord of each char
				//   (plus one extra entry for the last char)
  int len;			// length of text and edge arrays
  int size;			// size of text and edge arrays
  int charPos;                  // character position (within content stream)
  int charLen;                  // number of content stream characters in
                                //   this word
  TextFontInfo *font;		// font information
  double fontSize;		// font size
  GBool spaceAfter;		// set if there is a space between this
				//   word and the next word on the line
  TextWord *next;		// next word in line

#if TEXTOUT_WORD_LIST
  double colorR,		// word color
         colorG,
         colorB;
#endif

  GBool underlined;
  Link *link;

  friend class TextPool;
  friend class TextLine;
  friend class TextBlock;
  friend class TextFlow;
  friend class TextWordList;
  friend class TextPage;
};

//------------------------------------------------------------------------
// TextPool
//------------------------------------------------------------------------

class TextPool {
public:

  TextPool();
  ~TextPool();

  TextWord *getPool(int baseIdx) { return pool[baseIdx - minBaseIdx]; }
  void setPool(int baseIdx, TextWord *p) { pool[baseIdx - minBaseIdx] = p; }

  int getBaseIdx(double base);

  void addWord(TextWord *word);

private:

  int minBaseIdx;		// min baseline bucket index
  int maxBaseIdx;		// max baseline bucket index
  TextWord **pool;		// array of linked lists, one for each
				//   baseline value (multiple of 4 pts)
  TextWord *cursor;		// pointer to last-accessed word
  int cursorBaseIdx;		// baseline bucket index of last-accessed word

  friend class TextBlock;
  friend class TextPage;
};

//------------------------------------------------------------------------
// TextLine
//------------------------------------------------------------------------

class TextLine {
public:

  TextLine(TextBlock *blkA, int rotA, double baseA);
  ~TextLine();

  void addWord(TextWord *word);

  // Return the distance along the primary axis between <this> and
  // <line>.
  double primaryDelta(TextLine *line);

  // Compares <this> to <line>, returning -1 (<), 0 (=), or +1 (>),
  // based on a primary-axis comparison, e.g., x ordering if rot=0.
  int primaryCmp(TextLine *line);

  // Compares <this> to <line>, returning -1 (<), 0 (=), or +1 (>),
  // based on a secondary-axis comparison of the baselines, e.g., y
  // ordering if rot=0.
  int secondaryCmp(TextLine *line);

  int cmpYX(TextLine *line);

  static int cmpXY(const void *p1, const void *p2);

  void coalesce(UnicodeMap *uMap);

  // Get the head of the linked list of TextWords.
  TextWord *getWords() { return words; }

  // Get the next TextLine on the linked list.
  TextLine *getNext() { return next; }

  // Returns true if the last char of the line is a hyphen.
  GBool isHyphenated() { return hyphenated; }

private:

  TextBlock *blk;		// parent block
  int rot;			// text rotation
  double xMin, xMax;		// bounding box x coordinates
  double yMin, yMax;		// bounding box y coordinates
  double base;			// baseline x or y coordinate
  TextWord *words;		// words in this line
  TextWord *lastWord;		// last word in this line
  Unicode *text;		// Unicode text of the line, including
				//   spaces between words
  double *edge;			// "near" edge x or y coord of each char
				//   (plus one extra entry for the last char)
  int *col;			// starting column number of each Unicode char
  int len;			// number of Unicode chars
  int convertedLen;		// total number of converted characters
  GBool hyphenated;		// set if last char is a hyphen
  TextLine *next;		// next line in block

  friend class TextLineFrag;
  friend class TextBlock;
  friend class TextFlow;
  friend class TextWordList;
  friend class TextPage;
};

//------------------------------------------------------------------------
// TextBlock
//------------------------------------------------------------------------

class TextBlock {
public:

  TextBlock(TextPage *pageA, int rotA);
  ~TextBlock();

  void addWord(TextWord *word);

  void coalesce(UnicodeMap *uMap);

  // Update this block's priMin and priMax values, looking at <blk>.
  void updatePriMinMax(TextBlock *blk);

  static int cmpXYPrimaryRot(const void *p1, const void *p2);

  static int cmpYXPrimaryRot(const void *p1, const void *p2);

  int primaryCmp(TextBlock *blk);

  double secondaryDelta(TextBlock *blk);

  // Returns true if <this> is below <blk>, relative to the page's
  // primary rotation.
  GBool isBelow(TextBlock *blk);

  // Get the head of the linked list of TextLines.
  TextLine *getLines() { return lines; }

  // Get the next TextBlock on the linked list.
  TextBlock *getNext() { return next; }

private:

  TextPage *page;		// the parent page
  int rot;			// text rotation
  double xMin, xMax;		// bounding box x coordinates
  double yMin, yMax;		// bounding box y coordinates
  double priMin, priMax;	// whitespace bounding box along primary axis

  TextPool *pool;		// pool of words (used only until lines
				//   are built)
  TextLine *lines;		// linked list of lines
  TextLine *curLine;		// most recently added line
  int nLines;			// number of lines
  int charCount;		// number of characters in the block
  int col;			// starting column
  int nColumns;			// number of columns in the block

  TextBlock *next;
  TextBlock *stackNext;

  friend class TextLine;
  friend class TextLineFrag;
  friend class TextFlow;
  friend class TextWordList;
  friend class TextPage;
};