[Gdal-dev] IRasterIO in Dataset

[Fu Chen]

Frank,
Yeah, this code is more safe. Thanks.
But I find the ecw supersample is very slow indeed when my little browser to set the zoom ratio more than 1:1. Is there any reason? 

----- Original Message ----- 
From: "Frank Warmerdam" <warmerdam at pobox.com>
To: <gdal-dev at remotesensing.org>
Sent: Monday, October 27, 2003 10:05 PM
Subject: Re: [Gdal-dev] IRasterIO in Dataset


> 
> Fu Chen,
> 
> I have completed implementation of a "safe" IRasterIO() for ECWDataset
> and checked it in.  This slightly more than doubles the speed of reading
> large 3 band files with a single dataset level rasterio.
> 
> Best regards,
> 
> -- 
> ---------------------------------------+--------------------------------------
> I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
> light and sound - activate the windows | http://pobox.com/~warmerdam
> and watch the world go round - Rush    | Geospatial Programmer for Rent
> 
> 
> /************************************************************************/
> /*                             IRasterIO()                              */
> /************************************************************************/
> 
> CPLErr ECWDataset::IRasterIO( GDALRWFlag eRWFlag,
>                               int nXOff, int nYOff, int nXSize, int nYSize,
>                               void * pData, int nBufXSize, int nBufYSize,
>                               GDALDataType eBufType,
>                               int nBandCount, int *panBandMap,
>                               int nPixelSpace, int nLineSpace, int nBandSpace)
> 
> {
> /* -------------------------------------------------------------------- */
> /*      If we are supersampling we need to fall into the general        */
> /*      purpose logic.  We also use the general logic if we are in      */
> /*      some cases unlikely to benefit from interleaved access.         */
> /*                                                                      */
> /*      The one case we would like to handle better here is the         */
> /*      nBufYSize == 1 case (requesting a scanline at a time).  We      */
> /*      should eventually have some logic similiar to the band by       */
> /*      band case where we post a big window for the view, and allow    */
> /*      sequential reads.                                               */
> /* -------------------------------------------------------------------- */
>     if( nXSize < nBufXSize || nYSize < nBufYSize || nYSize == 1
>         || nBandCount > 100 || nBandCount == 1 || nBufYSize == 1 )
>     {
>         return
>             GDALDataset::IRasterIO( eRWFlag, nXOff, nYOff, nXSize, nYSize,
>                                     pData, nBufXSize, nBufYSize,
>                                     eBufType,
>                                     nBandCount, panBandMap,
>                                     nPixelSpace, nLineSpace, nBandSpace);
>     }
> 
>     CPLDebug( "ECWDataset",
>               "RasterIO(%d,%d,%d,%d -> %dx%d) - doing interleaved read.",
>               nXOff, nYOff, nXSize, nYSize, nBufXSize, nBufYSize );
> 
> /* -------------------------------------------------------------------- */
> /*      Setup view.                                                     */
> /* -------------------------------------------------------------------- */
>     UINT32 anBandIndices[100];
>     int    i;
>     NCSError     eNCSErr;
> 
>     for( i = 0; i < nBandCount; i++ )
>         anBandIndices[i] = panBandMap[i] - 1;
> 
>     eNCSErr = NCScbmSetFileView( hFileView,
>                                  nBandCount, anBandIndices,
>                                  nXOff, nYOff,
>                                  nXOff + nXSize - 1,
>                                  nYOff + nYSize - 1,
>                                  nBufXSize, nBufYSize );
> 
>     if( eNCSErr != NCS_SUCCESS )
>     {
>         CPLError( CE_Failure, CPLE_AppDefined,
>                   "%s", NCSGetErrorText(eNCSErr) );
> 
>         return CE_Failure;
>     }
> 
> /* -------------------------------------------------------------------- */
> /*      Setup working scanline, and the pointers into it.               */
> /* -------------------------------------------------------------------- */
>     GByte *pabyBILScanline = (GByte *) CPLMalloc(nBufXSize * nBandCount);
>     GByte **papabyBIL = (GByte **) CPLMalloc(nBandCount * sizeof(void*));
> 
>     for( i = 0; i < nBandCount; i++ )
>         papabyBIL[i] = pabyBILScanline + i * nBufXSize;
> 
> /* -------------------------------------------------------------------- */
> /*      Read back all the data for the requested view.                  */
> /* -------------------------------------------------------------------- */
>     for( int iScanline = 0; iScanline < nBufYSize; iScanline++ )
>     {
>         NCSEcwReadStatus  eRStatus;
>         int  iX;
> 
>         eRStatus = NCScbmReadViewLineBIL( hFileView, papabyBIL );
>         if( eRStatus != NCSECW_READ_OK )
>         {
>             CPLFree( pabyBILScanline );
>             CPLError( CE_Failure, CPLE_AppDefined,
>                       "NCScbmReadViewLineBIL failed." );
>             return CE_Failure;
>         }
> 
> 
>         for( i = 0; i < nBandCount; i++ )
>         {
>             GByte *pabyThisLine = ((GByte *)pData) + nLineSpace * iScanline +
> nBandSpace * i;
> 
>             for( iX = 0; iX < nBufXSize; iX++ )
>                 pabyThisLine[iX * nPixelSpace] =
>                     pabyBILScanline[i * nBufXSize + iX];
>         }
>     }
> 
>     CPLFree( papabyBIL );
>     CPLFree( pabyBILScanline );
> 
>     return CE_None;
> }
> 
> 
> _______________________________________________
> Gdal-dev mailing list
> Gdal-dev at remotesensing.org
> http://remotesensing.org/mailman/listinfo/gdal-dev