htmloutputdev.cc

将pdf文档转换为高质量的html文档

  pages->clear(); 
  if(!noframes)
  {
    if (fContentsFrame)
	{
      if (complexMode)
		fprintf(fContentsFrame,"<A href=\"%s-%d.html\"",str->getCString(),pageNum);
      else 
		fprintf(fContentsFrame,"<A href=\"%ss.html#%d\"",str->getCString(),pageNum);
      fprintf(fContentsFrame," target=\"contents\" >Page %d</a><br>\n",pageNum);
    }
  }

  pages->pageWidth=static_cast<int>(state->getPageWidth());
  pages->pageHeight=static_cast<int>(state->getPageHeight());

  delete str;
} 


void HtmlOutputDev::endPage() {
  pages->conv();
  pages->coalesce();
  pages->dump(page, pageNum);
  
  // I don't yet know what to do in the case when there are pages of different
  // sizes and we want complex output: running ghostscript many times 
  // seems very inefficient. So for now I'll just use last page's size
  maxPageWidth = pages->pageWidth;
  maxPageHeight = pages->pageHeight;
  
  //if(!noframes&&!xml) fputs("<br>\n", fContentsFrame);
  if(!stout && !globalParams->getErrQuiet()) printf("Page-%d\n",(pageNum));
}

void HtmlOutputDev::updateFont(GfxState *state) {
  pages->updateFont(state);
}

void HtmlOutputDev::beginString(GfxState *state, GString *s) {
  pages->beginString(state, s);
}

void HtmlOutputDev::endString(GfxState *state) {
  pages->endString();
}

void HtmlOutputDev::drawChar(GfxState *state, double x, double y,
	      double dx, double dy,
	      double originX, double originY,
	      CharCode code, int nBytes, Unicode *u, int uLen) 
{
  if ( !showHidden && (state->getRender() & 3) == 3) {
    return;
  }
  pages->addChar(state, x, y, dx, dy, originX, originY, u, uLen);
}

void HtmlOutputDev::drawImageMask(GfxState *state, Object *ref, Stream *str,
			      int width, int height, GBool invert,
			      GBool inlineImg) {

  int i, j;
  
  if (ignore||complexMode) {
    OutputDev::drawImageMask(state, ref, str, width, height, invert, inlineImg);
    return;
  }
  
  FILE *f1;
  int c;
  
  int x0, y0;			// top left corner of image
  int w0, h0, w1, h1;		// size of image
  double xt, yt, wt, ht;
  GBool rotate, xFlip, yFlip;
  GBool dither;
  int x, y;
  int ix, iy;
  int px1, px2, qx, dx;
  int py1, py2, qy, dy;
  Gulong pixel;
  int nComps, nVals, nBits;
  double r1, g1, b1;
 
  // get image position and size
  state->transform(0, 0, &xt, &yt);
  state->transformDelta(1, 1, &wt, &ht);
  if (wt > 0) {
    x0 = xoutRound(xt);
    w0 = xoutRound(wt);
  } else {
    x0 = xoutRound(xt + wt);
    w0 = xoutRound(-wt);
  }
  if (ht > 0) {
    y0 = xoutRound(yt);
    h0 = xoutRound(ht);
  } else {
    y0 = xoutRound(yt + ht);
    h0 = xoutRound(-ht);
  }
  state->transformDelta(1, 0, &xt, &yt);
  rotate = fabs(xt) < fabs(yt);
  if (rotate) {
    w1 = h0;
    h1 = w0;
    xFlip = ht < 0;
    yFlip = wt > 0;
  } else {
    w1 = w0;
    h1 = h0;
    xFlip = wt < 0;
    yFlip = ht > 0;
  }

  // dump JPEG file
  if (dumpJPEG  && str->getKind() == strDCT) {
    GString *fName=new GString(Docname);
    fName->append("-");
    GString *pgNum=GString::fromInt(pageNum);
    GString *imgnum=GString::fromInt(imgNum);
    // open the image file
    fName->append(pgNum)->append("_")->append(imgnum)->append(".jpg");
    ++imgNum;
    if (!(f1 = fopen(fName->getCString(), "wb"))) {
      error(-1, "Couldn't open image file '%s'", fName->getCString());
      return;
    }

    // initialize stream
    str = ((DCTStream *)str)->getRawStream();
    str->reset();

    // copy the stream
    while ((c = str->getChar()) != EOF)
      fputc(c, f1);

    fclose(f1);
   
  if (pgNum) delete pgNum;
  if (imgnum) delete imgnum;
  if (fName) delete fName;
  }
  else {
    OutputDev::drawImageMask(state, ref, str, width, height, invert, inlineImg);
  }
}

void HtmlOutputDev::drawImage(GfxState *state, Object *ref, Stream *str,
			  int width, int height, GfxImageColorMap *colorMap,
			  int *maskColors, GBool inlineImg) {

  int i, j;
   
  if (ignore||complexMode) {
    OutputDev::drawImage(state, ref, str, width, height, colorMap, 
			 maskColors, inlineImg);
    return;
  }

  FILE *f1;
  ImageStream *imgStr;
  Guchar pixBuf[4];
  GfxColor color;
  int c;
  
  int x0, y0;			// top left corner of image
  int w0, h0, w1, h1;		// size of image
  double xt, yt, wt, ht;
  GBool rotate, xFlip, yFlip;
  GBool dither;
  int x, y;
  int ix, iy;
  int px1, px2, qx, dx;
  int py1, py2, qy, dy;
  Gulong pixel;
  int nComps, nVals, nBits;
  double r1, g1, b1;
 
  // get image position and size
  state->transform(0, 0, &xt, &yt);
  state->transformDelta(1, 1, &wt, &ht);
  if (wt > 0) {
    x0 = xoutRound(xt);
    w0 = xoutRound(wt);
  } else {
    x0 = xoutRound(xt + wt);
    w0 = xoutRound(-wt);
  }
  if (ht > 0) {
    y0 = xoutRound(yt);
    h0 = xoutRound(ht);
  } else {
    y0 = xoutRound(yt + ht);
    h0 = xoutRound(-ht);
  }
  state->transformDelta(1, 0, &xt, &yt);
  rotate = fabs(xt) < fabs(yt);
  if (rotate) {
    w1 = h0;
    h1 = w0;
    xFlip = ht < 0;
    yFlip = wt > 0;
  } else {
    w1 = w0;
    h1 = h0;
    xFlip = wt < 0;
    yFlip = ht > 0;
  }

   
  /*if( !globalParams->getErrQuiet() )
    printf("image stream of kind %d\n", str->getKind());*/
  // dump JPEG file
  if (dumpJPEG && str->getKind() == strDCT) {
    GString *fName=new GString(Docname);
    fName->append("-");
    GString *pgNum= GString::fromInt(pageNum);
    GString *imgnum= GString::fromInt(imgNum);  
    
    // open the image file
    fName->append(pgNum)->append("_")->append(imgnum)->append(".jpg");
    ++imgNum;
    
    if (!(f1 = fopen(fName->getCString(), "wb"))) {
      error(-1, "Couldn't open image file '%s'", fName->getCString());
      return;
    }

    // initialize stream
    str = ((DCTStream *)str)->getRawStream();
    str->reset();

    // copy the stream
    while ((c = str->getChar()) != EOF)
      fputc(c, f1);
    
    fclose(f1);
  
    delete fName;
    delete pgNum;
    delete imgnum;
  }
  else {
    OutputDev::drawImage(state, ref, str, width, height, colorMap,
			 maskColors, inlineImg);
  }
}



void HtmlOutputDev::drawLink(Link* link,Catalog *cat){
  double _x1,_y1,_x2,_y2,w;
  int x1,y1,x2,y2;
  
  link->getBorder(&_x1,&_y1,&_x2,&_y2,&w);
  cvtUserToDev(_x1,_y1,&x1,&y1);
  
  cvtUserToDev(_x2,_y2,&x2,&y2); 


  GString* _dest=getLinkDest(link,cat);
  HtmlLink t((double) x1,(double) y2,(double) x2,(double) y1,_dest);
  pages->AddLink(t);
  delete _dest;
}

GString* HtmlOutputDev::getLinkDest(Link *link,Catalog* catalog){
  char *p;
  switch(link->getAction()->getKind()) 
  {
      case actionGoTo:
	  { 
	  GString* file=basename(Docname);
	  int page=1;
	  LinkGoTo *ha=(LinkGoTo *)link->getAction();
	  LinkDest *dest=NULL;
	  if (ha->getDest()==NULL) 
	      dest=catalog->findDest(ha->getNamedDest());
	  else 
	      dest=ha->getDest()->copy();
	  if (dest){ 
	      if (dest->isPageRef()){
		  Ref pageref=dest->getPageRef();
		  page=catalog->findPage(pageref.num,pageref.gen);
	      }
	      else {
		  page=dest->getPageNum();
	      }

	      delete dest;

	      GString *str=GString::fromInt(page);
	      /* 		complex 	simple
	       	frames		file-4.html	files.html#4
		noframes	file.html#4	file.html#4
	       */
	      if (noframes)
	      {
		  file->append(".html#");
		  file->append(str);
	      }
	      else
	      {
	      	if( complexMode ) 
		{
		    file->append("-");
		    file->append(str);
		    file->append(".html");
		}
		else
		{
		    file->append("s.html#");
		    file->append(str);
		}
	      }

	      if (printCommands) printf(" link to page %d ",page);
	      delete str;
	      return file;
	  }
	  else 
	  {
	      return new GString();
	  }
	  }
      case actionGoToR:
	  {
	  LinkGoToR *ha=(LinkGoToR *) link->getAction();
	  LinkDest *dest=NULL;
	  int page=1;
	  GString *file=new GString();
	  if (ha->getFileName()){
	      delete file;
	      file=new GString(ha->getFileName()->getCString());
	  }
	  if (ha->getDest()!=NULL)  dest=ha->getDest()->copy();
	  if (dest&&file){
	      if (!(dest->isPageRef()))  page=dest->getPageNum();
	      delete dest;

	      if (printCommands) printf(" link to page %d ",page);
	      if (printHtml){
		  p=file->getCString()+file->getLength()-4;
		  if (!strcmp(p, ".pdf") || !strcmp(p, ".PDF")){
		      file->del(file->getLength()-4,4);
		      file->append(".html");
		  }
		  file->append('#');
		  file->append(GString::fromInt(page));
	      }
	  }
	  if (printCommands) printf("filename %s\n",file->getCString());
	  return file;
	  }
      case actionURI:
	  { 
	  LinkURI *ha=(LinkURI *) link->getAction();
	  GString* file=new GString(ha->getURI()->getCString());
	  // printf("uri : %s\n",file->getCString());
	  return file;
	  }
      case actionLaunch:
	  {
	  LinkLaunch *ha=(LinkLaunch *) link->getAction();
	  GString* file=new GString(ha->getFileName()->getCString());
	  if (printHtml) { 
	      p=file->getCString()+file->getLength()-4;
	      if (!strcmp(p, ".pdf") || !strcmp(p, ".PDF")){
		  file->del(file->getLength()-4,4);
		  file->append(".html");
	      }
	      if (printCommands) printf("filename %s",file->getCString());
    
	      return file;      
  
	  }
	  }
      default:
	  return new GString();
  }
}

void HtmlOutputDev::dumpMetaVars(FILE *file)
{
  GString *var;

  for(int i = 0; i < glMetaVars->getLength(); i++)
  {
     HtmlMetaVar *t = (HtmlMetaVar*)glMetaVars->get(i); 
     var = t->toString(); 
     fprintf(file, "%s\n", var->getCString());
     delete var;
  }
}

GBool HtmlOutputDev::dumpDocOutline(Catalog* catalog)
{ 
	FILE * output;
	GBool bClose = gFalse;

	if (!ok || xml)
    	return gFalse;
  
	Object *outlines = catalog->getOutline();
  	if (!outlines->isDict())
    	return gFalse;
  
	if (!complexMode && !xml)
  	{
		output = page;
  	}
  	else if (complexMode && !xml)
	{
		if (noframes)
		{
			output = page; 
			fputs("<hr>\n", output);
		}
		else
		{
			GString *str = basename(Docname);
			str->append("-outline.html");
			output = fopen(str->getCString(), "w");
			if (output == NULL)
				return gFalse;
			delete str;
			bClose = gTrue;
     		fputs("<HTML>\n<HEAD>\n<TITLE>Document Outline</TITLE>\n</HEAD>\n<BODY>\n", output);
		}
	}
 
  	GBool done = newOutlineLevel(output, outlines, catalog);
  	if (done && !complexMode)
    	fputs("<hr>\n", output);
	
	if (bClose)
	{
		fputs("</BODY>\n</HTML>\n", output);
		fclose(output);
	}
  	return done;
}

GBool HtmlOutputDev::newOutlineLevel(FILE *output, Object *node, Catalog* catalog, int level)
{
  Object curr, next;
  GBool atLeastOne = gFalse;
  
  if (node->dictLookup("First", &curr)->isDict()) {
    if (level == 1)
	{
		fputs("<A name=\"outline\"></a>", output);
		fputs("<h1>Document Outline</h1>\n", output);
	}
    fputs("<ul>",output);
    do {
      // get title, give up if not found
      Object title;
      if (curr.dictLookup("Title", &title)->isNull()) {
		title.free();
		break;
      }
      GString *titleStr = new GString(title.getString());
      title.free();

      // get corresponding link
      // Note: some code duplicated from HtmlOutputDev::getLinkDest().
      GString *linkName = NULL;;
      Object dest;
      if (!curr.dictLookup("Dest", &dest)->isNull()) {
		LinkGoTo *link = new LinkGoTo(&dest);
		LinkDest *linkdest=NULL;
		if (link->getDest()==NULL) 
	  		linkdest=catalog->findDest(link->getNamedDest());
		else 
	  		linkdest=link->getDest()->copy();
		delete link;
		if (linkdest) { 
	  		int page;
	  		if (linkdest->isPageRef()) {
	    		Ref pageref=linkdest->getPageRef();
	    		page=catalog->findPage(pageref.num,pageref.gen);
	  		} else {
	    		page=linkdest->getPageNum();
	  		}
	  		delete linkdest;

			/* 			complex 	simple
			frames		file-4.html	files.html#4
			noframes	file.html#4	file.html#4
	   		*/
	  		linkName=basename(Docname);
	  		GString *str=GString::fromInt(page);
	  		if (noframes) {
	    		linkName->append(".html#");
				linkName->append(str);
	  		} else {
    			if( complexMode ) {
	   		   		linkName->append("-");
	      			linkName->append(str);
	      			linkName->append(".html");
	    		} else {
	      			linkName->append("s.html#");
	      			linkName->append(str);
	    		}
	  		}
			delete str;
		}
      }
      dest.free();

      fputs("<li>",output);
      if (linkName)
		fprintf(output,"<A href=\"%s\">", linkName->getCString());
      fputs(titleStr->getCString(),output);
      if (linkName) {
		fputs("</A>",output);
		delete linkName;
      }
      fputs("\n",output);
      delete titleStr;
      atLeastOne = gTrue;

      newOutlineLevel(output, &curr, catalog, level+1);
      curr.dictLookup("Next", &next);
      curr.free();
      curr = next;
    } while(curr.isDict());
    fputs("</ul>",output);
  }
  curr.free();

  return atLeastOne;
}

void HtmlOutputDev::updateCharSpace(GfxState *state)
{
	pages->updateCharSpace(state);
}