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

//------------------------------------------------------------------------
// TextFlow
//------------------------------------------------------------------------

class TextFlow {
public:

  TextFlow(TextPage *pageA, TextBlock *blk);
  ~TextFlow();

  // Add a block to the end of this flow.
  void addBlock(TextBlock *blk);

  // Returns true if <blk> fits below <prevBlk> in the flow, i.e., (1)
  // it uses a font no larger than the last block added to the flow,
  // and (2) it fits within the flow's [priMin, priMax] along the
  // primary axis.
  GBool blockFits(TextBlock *blk, TextBlock *prevBlk);

  // Get the head of the linked list of TextBlocks.
  TextBlock *getBlocks() { return blocks; }

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

private:

  TextPage *page;		// the parent page
  double xMin, xMax;		// bounding box x coordinates
  double yMin, yMax;		// bounding box y coordinates
  double priMin, priMax;	// whitespace bounding box along primary axis
  TextBlock *blocks;		// blocks in flow
  TextBlock *lastBlk;		// last block in this flow
  TextFlow *next;

  friend class TextWordList;
  friend class TextPage;
};

#if TEXTOUT_WORD_LIST

//------------------------------------------------------------------------
// TextWordList
//------------------------------------------------------------------------

class TextWordList {
public:

  // Build a flat word list, in content stream order (if
  // text->rawOrder is true), physical layout order (if <physLayout>
  // is true and text->rawOrder is false), or reading order (if both
  // flags are false).
  TextWordList(TextPage *text, GBool physLayout);

  ~TextWordList();

  // Return the number of words on the list.
  int getLength();

  // Return the <idx>th word from the list.
  TextWord *get(int idx);

private:

  GList *words;			// [TextWord]
};

#endif // TEXTOUT_WORD_LIST

//------------------------------------------------------------------------
// TextPage
//------------------------------------------------------------------------

class TextPage {
public:

  // Constructor.
  TextPage(GBool rawOrderA);

  // Destructor.
  ~TextPage();

  // Start a new page.
  void startPage(GfxState *state);

  // End the current page.
  void endPage();

  // Update the current font.
  void updateFont(GfxState *state);

  // Begin a new word.
  void beginWord(GfxState *state, double x0, double y0);

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

  // End the current word, sorting it into the list of words.
  void endWord();

  // Add a word, sorting it into the list of words.
  void addWord(TextWord *word);

  // Add a (potential) underline.
  void addUnderline(double x0, double y0, double x1, double y1);

  // Add a hyperlink.
  void addLink(int xMin, int yMin, int xMax, int yMax, Link *link);

  // Coalesce strings that look like parts of the same line.
  void coalesce(GBool physLayout, GBool doHTML);

  // Find a string.  If <startAtTop> is true, starts looking at the
  // top of the page; else if <startAtLast> is true, starts looking
  // immediately after the last find result; else starts looking at
  // <xMin>,<yMin>.  If <stopAtBottom> is true, stops looking at the
  // bottom of the page; else if <stopAtLast> is true, stops looking
  // just before the last find result; else stops looking at
  // <xMax>,<yMax>.
  GBool findText(Unicode *s, int len,
		 GBool startAtTop, GBool stopAtBottom,
		 GBool startAtLast, GBool stopAtLast,
		 GBool caseSensitive, GBool backward,
		 double *xMin, double *yMin,
		 double *xMax, double *yMax);

  // Get the text which is inside the specified rectangle.
  GString *getText(double xMin, double yMin,
		   double xMax, double yMax);

  // Find a string by character position and length.  If found, sets
  // the text bounding rectangle and returns true; otherwise returns
  // false.
  GBool findCharRange(int pos, int length,
		      double *xMin, double *yMin,
		      double *xMax, double *yMax);

  // Dump contents of page to a file.
  void dump(void *outputStream, TextOutputFunc outputFunc,
	    GBool physLayout);

  // Get the head of the linked list of TextFlows.
  TextFlow *getFlows() { return flows; }

#if TEXTOUT_WORD_LIST
  // Build a flat word list, in content stream order (if
  // this->rawOrder is true), physical layout order (if <physLayout>
  // is true and this->rawOrder is false), or reading order (if both
  // flags are false).
  TextWordList *makeWordList(GBool physLayout);
#endif

private:

  void clear();
  void assignColumns(TextLineFrag *frags, int nFrags, int rot);
  int dumpFragment(Unicode *text, int len, UnicodeMap *uMap, GString *s);

  GBool rawOrder;		// keep text in content stream order

  double pageWidth, pageHeight;	// width and height of current page
  TextWord *curWord;		// currently active string
  int charPos;			// next character position (within content
				//   stream)
  TextFontInfo *curFont;	// current font
  double curFontSize;		// current font size
  int nest;			// current nesting level (for Type 3 fonts)
  int nTinyChars;		// number of "tiny" chars seen so far
  GBool lastCharOverlap;	// set if the last added char overlapped the
				//   previous char

  TextPool *pools[4];		// a "pool" of TextWords for each rotation
  TextFlow *flows;		// linked list of flows
  TextBlock **blocks;		// array of blocks, in yx order
  int nBlocks;			// number of blocks
  int primaryRot;		// primary rotation
  GBool primaryLR;		// primary direction (true means L-to-R,
				//   false means R-to-L)
  TextWord *rawWords;		// list of words, in raw order (only if
				//   rawOrder is set)
  TextWord *rawLastWord;	// last word on rawWords list

  GList *fonts;			// all font info objects used on this
				//   page [TextFontInfo]

  double lastFindXMin,		// coordinates of the last "find" result
         lastFindYMin;
  GBool haveLastFind;

  GList *underlines;		// [TextUnderline]
  GList *links;			// [TextLink]

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

//------------------------------------------------------------------------
// TextOutputDev
//------------------------------------------------------------------------

class TextOutputDev: public OutputDev {
public:

  // Open a text output file.  If <fileName> is NULL, no file is
  // written (this is useful, e.g., for searching text).  If
  // <physLayoutA> is true, the original physical layout of the text
  // is maintained.  If <rawOrder> is true, the text is kept in
  // content stream order.
  TextOutputDev(char *fileName, GBool physLayoutA,
		GBool rawOrderA, GBool append);

  // Create a TextOutputDev which will write to a generic stream.  If
  // <physLayoutA> is true, the original physical layout of the text
  // is maintained.  If <rawOrder> is true, the text is kept in
  // content stream order.
  TextOutputDev(TextOutputFunc func, void *stream,
		GBool physLayoutA, GBool rawOrderA);

  // Destructor.
  virtual ~TextOutputDev();

  // Check if file was successfully created.
  virtual GBool isOk() { return ok; }

  //---- get info about output device

  // Does this device use upside-down coordinates?
  // (Upside-down means (0,0) is the top left corner of the page.)
  virtual GBool upsideDown() { return gTrue; }

  // Does this device use drawChar() or drawString()?
  virtual GBool useDrawChar() { return gTrue; }

  // Does this device use beginType3Char/endType3Char?  Otherwise,
  // text in Type 3 fonts will be drawn with drawChar/drawString.
  virtual GBool interpretType3Chars() { return gFalse; }

  // Does this device need non-text content?
  virtual GBool needNonText() { return gFalse; }

  //----- initialization and control

  // Start a page.
  virtual void startPage(int pageNum, GfxState *state);

  // End a page.
  virtual void endPage();

  //----- update text state
  virtual void updateFont(GfxState *state);

  //----- text drawing
  virtual void beginString(GfxState *state, GString *s);
  virtual void endString(GfxState *state);
  virtual void drawChar(GfxState *state, double x, double y,
			double dx, double dy,
			double originX, double originY,
			CharCode c, int nBytes, Unicode *u, int uLen);

  //----- path painting
  virtual void stroke(GfxState *state);
  virtual void fill(GfxState *state);
  virtual void eoFill(GfxState *state);

  //----- link borders
  virtual void processLink(Link *link, Catalog *catalog);

  //----- special access

  // Find a string.  If <startAtTop> is true, starts looking at the
  // top of the page; else if <startAtLast> is true, starts looking
  // immediately after the last find result; else starts looking at
  // <xMin>,<yMin>.  If <stopAtBottom> is true, stops looking at the
  // bottom of the page; else if <stopAtLast> is true, stops looking
  // just before the last find result; else stops looking at
  // <xMax>,<yMax>.
  GBool findText(Unicode *s, int len,
		 GBool startAtTop, GBool stopAtBottom,
		 GBool startAtLast, GBool stopAtLast,
		 GBool caseSensitive, GBool backward,
		 double *xMin, double *yMin,
		 double *xMax, double *yMax);

  // Get the text which is inside the specified rectangle.
  GString *getText(double xMin, double yMin,
		   double xMax, double yMax);

  // Find a string by character position and length.  If found, sets
  // the text bounding rectangle and returns true; otherwise returns
  // false.
  GBool findCharRange(int pos, int length,
		      double *xMin, double *yMin,
		      double *xMax, double *yMax);

#if TEXTOUT_WORD_LIST
  // Build a flat word list, in content stream order (if
  // this->rawOrder is true), physical layout order (if
  // this->physLayout is true and this->rawOrder is false), or reading
  // order (if both flags are false).
  TextWordList *makeWordList();
#endif

  // Returns the TextPage object for the last rasterized page,
  // transferring ownership to the caller.
  TextPage *takeText();

  // Turn extra processing for HTML conversion on or off.
  void enableHTMLExtras(GBool doHTMLA) { doHTML = doHTMLA; }

private:

  TextOutputFunc outputFunc;	// output function
  void *outputStream;		// output stream
  GBool needClose;		// need to close the output file?
				//   (only if outputStream is a FILE*)
  TextPage *text;		// text for the current page
  GBool physLayout;		// maintain original physical layout when
				//   dumping text
  GBool rawOrder;		// keep text in content stream order
  GBool doHTML;			// extra processing for HTML conversion
  GBool ok;			// set up ok?
};

#endif