Files
libreoffice/canvas/source/opengl/ogl_canvascustomsprite.cxx

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

262 lines
9.4 KiB
C++
Raw Normal View History

/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
/*
* This file is part of the LibreOffice project.
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/.
*/
#include <sal/config.h>
#include <epoxy/gl.h>
#include <basegfx/matrix/b2dhommatrix.hxx>
#include <basegfx/point/b2dpoint.hxx>
#include <basegfx/polygon/b2dpolygontriangulator.hxx>
#include <basegfx/utils/canvastools.hxx>
#include <canvas/canvastools.hxx>
#include <canvas/verifyinput.hxx>
#include <tools/diagnose_ex.h>
#include "ogl_canvascustomsprite.hxx"
#include "ogl_canvastools.hxx"
#include "ogl_tools.hxx"
using namespace ::com::sun::star;
namespace oglcanvas
{
CanvasCustomSprite::CanvasCustomSprite( const css::geometry::RealSize2D& rSpriteSize,
const SpriteCanvasRef& rRefDevice,
SpriteDeviceHelper& rDeviceHelper ) :
mpSpriteCanvas( rRefDevice ),
maSize(rSpriteSize),
mxClip(),
maTransformation(),
maPosition(),
mfAlpha(0.0),
mfPriority(0.0)
{
ENSURE_OR_THROW( rRefDevice.get(),
"CanvasCustomSprite::CanvasCustomSprite(): Invalid sprite canvas" );
::canvas::tools::setIdentityAffineMatrix2D(maTransformation);
maCanvasHelper.init( *rRefDevice,
rDeviceHelper );
}
void CanvasCustomSprite::disposeThis()
{
::osl::MutexGuard aGuard( m_aMutex );
mpSpriteCanvas.clear();
// forward to parent
CanvasCustomSpriteBaseT::disposeThis();
}
void SAL_CALL CanvasCustomSprite::setAlpha( double alpha )
{
canvas::tools::verifyRange( alpha, 0.0, 1.0 );
::osl::MutexGuard aGuard( m_aMutex );
mfAlpha = alpha;
}
void SAL_CALL CanvasCustomSprite::move( const geometry::RealPoint2D& aNewPos,
const rendering::ViewState& viewState,
const rendering::RenderState& renderState )
{
canvas::tools::verifyArgs(aNewPos, viewState, renderState,
OSL_THIS_FUNC,
static_cast< ::cppu::OWeakObject* >(this));
::osl::MutexGuard aGuard( m_aMutex );
::basegfx::B2DHomMatrix aTransform;
::canvas::tools::mergeViewAndRenderTransform(aTransform,
viewState,
renderState);
// convert position to device pixel
maPosition = ::basegfx::unotools::b2DPointFromRealPoint2D(aNewPos);
maPosition *= aTransform;
}
void SAL_CALL CanvasCustomSprite::transform( const geometry::AffineMatrix2D& aTransformation )
{
::osl::MutexGuard aGuard( m_aMutex );
maTransformation = aTransformation;
}
void SAL_CALL CanvasCustomSprite::clip( const uno::Reference< rendering::XPolyPolygon2D >& xClip )
{
mxClip = xClip;
}
void SAL_CALL CanvasCustomSprite::setPriority( double nPriority )
{
::osl::MutexGuard aGuard( m_aMutex );
mfPriority = nPriority;
}
void SAL_CALL CanvasCustomSprite::show()
{
::osl::MutexGuard aGuard( m_aMutex );
if( mpSpriteCanvas.is() )
mpSpriteCanvas->show(this);
}
void SAL_CALL CanvasCustomSprite::hide()
{
::osl::MutexGuard aGuard( m_aMutex );
if( mpSpriteCanvas.is() )
mpSpriteCanvas->hide(this);
}
uno::Reference< rendering::XCanvas > SAL_CALL CanvasCustomSprite::getContentCanvas()
{
return this;
}
bool CanvasCustomSprite::renderSprite() const
{
if( ::basegfx::fTools::equalZero( mfAlpha ) )
return true;
TransformationPreserver aPreserver1;
const ::basegfx::B2IVector aSpriteSizePixel(
::canvas::tools::roundUp( maSize.Width ),
::canvas::tools::roundUp( maSize.Height ));
// translate sprite to output position
glTranslated(maPosition.getX(), maPosition.getY(), 0);
{
TransformationPreserver aPreserver2;
// apply sprite content transformation matrix
double aGLTransform[] =
{
maTransformation.m00, maTransformation.m10, 0, 0,
maTransformation.m01, maTransformation.m11, 0, 0,
0, 0, 1, 0,
maTransformation.m02, maTransformation.m12, 0, 1
};
glMultMatrixd(aGLTransform);
IBufferContextSharedPtr pBufferContext;
if( mfAlpha != 1.0 || mxClip.is() )
{
// drafts. Need to render to temp surface before, and then
// composite that to screen
// TODO(P3): buffer texture
pBufferContext = maCanvasHelper.getDeviceHelper()->createBufferContext(aSpriteSizePixel);
pBufferContext->startBufferRendering();
}
// this ends up in pBufferContext, if that one's "current"
if( !maCanvasHelper.renderRecordedActions() )
return false;
if( pBufferContext )
{
// content ended up in background buffer - compose to
// screen now. Calls below switches us back to window
// context, and binds to generated, dynamic texture
pBufferContext->endBufferRendering();
GLuint nTexture = pBufferContext->getTextureId();
glBindTexture(GL_TEXTURE_2D, nTexture);
glEnable(GL_TEXTURE_2D);
glTexParameteri(GL_TEXTURE_2D,
GL_TEXTURE_MIN_FILTER,
GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D,
GL_TEXTURE_MAG_FILTER,
GL_NEAREST);
glEnable(GL_BLEND);
glBlendFunc(GL_SRC_ALPHA,
GL_ONE_MINUS_SRC_ALPHA);
// blend against fixed vertex color; texture alpha is multiplied in
glColor4f(1,1,1,mfAlpha);
if( mxClip.is() )
{
const double fWidth=maSize.Width;
const double fHeight=maSize.Height;
// TODO(P3): buffer triangulation
Support buffering SystemDependent GraphicData (III) This change is for speedup of fat line drawing when using X11. This is a long-term problem which never really progressed, but is avoided using Cairo in the future. Still - if used, speedup using current state and buffering possibilities. Two speedup steps will be used: (1) The tesselation is no longer done using trapezoids. That works (but was done wrong leaving artifacts) but is not fast and done every time. It is even not done with FatLines and more than 1000 points. New version will use triangulation. Dspite using the existing triangulator (that works but is slow) extend the FatLine geometry creator to directly create triangles. This is also necessary since for buffering that data a transformation-invariant version is needed (in device coordinates the data changes all the time when scrolling). Trapezoids are by definition *not* transformation-invariant (e.g. rotation) (2) Buffer that triangulation - with the needed care and watch. It is e.g. necessary to react on 'hairlines' since these change their logical LineWidth view-dependent (zoom). In those cases, the buffered data *has* to be removed due to the base for buffering is the created FatLine geometry based on one stable logical LineWidth Also took the time to adapt B2DPolygonTriangulator to use an own data type (B2DTriangle) and a vector of these for better understandability and security. Adapted all usages as needed. Change-Id: Iedb2932b094a8786fd9c32d0d0ab1ca603a1a7b2 Reviewed-on: https://gerrit.libreoffice.org/60818 Tested-by: Jenkins Reviewed-by: Armin Le Grand <Armin.Le.Grand@cib.de>
2018-09-15 13:48:12 +02:00
const ::basegfx::triangulator::B2DTriangleVector rTriangulatedPolygon(
::basegfx::triangulator::triangulate(
::basegfx::unotools::b2DPolyPolygonFromXPolyPolygon2D(mxClip)));
glBegin(GL_TRIANGLES);
Support buffering SystemDependent GraphicData (III) This change is for speedup of fat line drawing when using X11. This is a long-term problem which never really progressed, but is avoided using Cairo in the future. Still - if used, speedup using current state and buffering possibilities. Two speedup steps will be used: (1) The tesselation is no longer done using trapezoids. That works (but was done wrong leaving artifacts) but is not fast and done every time. It is even not done with FatLines and more than 1000 points. New version will use triangulation. Dspite using the existing triangulator (that works but is slow) extend the FatLine geometry creator to directly create triangles. This is also necessary since for buffering that data a transformation-invariant version is needed (in device coordinates the data changes all the time when scrolling). Trapezoids are by definition *not* transformation-invariant (e.g. rotation) (2) Buffer that triangulation - with the needed care and watch. It is e.g. necessary to react on 'hairlines' since these change their logical LineWidth view-dependent (zoom). In those cases, the buffered data *has* to be removed due to the base for buffering is the created FatLine geometry based on one stable logical LineWidth Also took the time to adapt B2DPolygonTriangulator to use an own data type (B2DTriangle) and a vector of these for better understandability and security. Adapted all usages as needed. Change-Id: Iedb2932b094a8786fd9c32d0d0ab1ca603a1a7b2 Reviewed-on: https://gerrit.libreoffice.org/60818 Tested-by: Jenkins Reviewed-by: Armin Le Grand <Armin.Le.Grand@cib.de>
2018-09-15 13:48:12 +02:00
for( size_t i=0; i<rTriangulatedPolygon.size(); i++ )
{
Support buffering SystemDependent GraphicData (III) This change is for speedup of fat line drawing when using X11. This is a long-term problem which never really progressed, but is avoided using Cairo in the future. Still - if used, speedup using current state and buffering possibilities. Two speedup steps will be used: (1) The tesselation is no longer done using trapezoids. That works (but was done wrong leaving artifacts) but is not fast and done every time. It is even not done with FatLines and more than 1000 points. New version will use triangulation. Dspite using the existing triangulator (that works but is slow) extend the FatLine geometry creator to directly create triangles. This is also necessary since for buffering that data a transformation-invariant version is needed (in device coordinates the data changes all the time when scrolling). Trapezoids are by definition *not* transformation-invariant (e.g. rotation) (2) Buffer that triangulation - with the needed care and watch. It is e.g. necessary to react on 'hairlines' since these change their logical LineWidth view-dependent (zoom). In those cases, the buffered data *has* to be removed due to the base for buffering is the created FatLine geometry based on one stable logical LineWidth Also took the time to adapt B2DPolygonTriangulator to use an own data type (B2DTriangle) and a vector of these for better understandability and security. Adapted all usages as needed. Change-Id: Iedb2932b094a8786fd9c32d0d0ab1ca603a1a7b2 Reviewed-on: https://gerrit.libreoffice.org/60818 Tested-by: Jenkins Reviewed-by: Armin Le Grand <Armin.Le.Grand@cib.de>
2018-09-15 13:48:12 +02:00
const::basegfx::triangulator::B2DTriangle& rCandidate(rTriangulatedPolygon[i]);
glTexCoord2f(
rCandidate.getA().getX()/fWidth,
rCandidate.getA().getY()/fHeight);
glVertex2d(
rCandidate.getA().getX(),
rCandidate.getA().getY());
glTexCoord2f(
rCandidate.getB().getX()/fWidth,
rCandidate.getB().getY()/fHeight);
glVertex2d(
rCandidate.getB().getX(),
rCandidate.getB().getY());
glTexCoord2f(
rCandidate.getC().getX()/fWidth,
rCandidate.getC().getY()/fHeight);
glVertex2d(
rCandidate.getC().getX(),
rCandidate.getC().getY());
}
glEnd();
}
else
{
const double fWidth=maSize.Width/aSpriteSizePixel.getX();
const double fHeight=maSize.Height/aSpriteSizePixel.getY();
glBegin(GL_TRIANGLE_STRIP);
glTexCoord2f(0,0); glVertex2d(0,0);
glTexCoord2f(0,fHeight); glVertex2d(0, aSpriteSizePixel.getY());
glTexCoord2f(fWidth,0); glVertex2d(aSpriteSizePixel.getX(),0);
glTexCoord2f(fWidth,fHeight); glVertex2d(aSpriteSizePixel.getX(),aSpriteSizePixel.getY());
glEnd();
}
glBindTexture(GL_TEXTURE_2D, 0);
glDisable(GL_TEXTURE_2D);
}
}
glColor4f(1,0,0,1);
glBegin(GL_LINE_STRIP);
glVertex2d(-2,-2);
glVertex2d(-2,maSize.Height+4);
glVertex2d(maSize.Width+4,maSize.Height+4);
glVertex2d(maSize.Width+4,-2);
glVertex2d(-2,-2);
glVertex2d(maSize.Width+4,maSize.Height+4);
glEnd();
std::vector<double> aVec;
aVec.push_back(mfAlpha);
aVec.push_back(mfPriority);
aVec.push_back(maCanvasHelper.getRecordedActionCount());
renderOSD( aVec, 10 );
return true;
}
}
/* vim:set shiftwidth=4 softtabstop=4 expandtab: */